Comparativegrammar Sanskrit Greek Latin Latvian German

HONn
19
Wisi
0 vLGL9Cl0
OLNOHOL JO ALISHSAINN
qustepipoh
127, Zho >
vf
mw}
Digitizedby the Internet /

in 2007 with funding
‘
Microsoft Corpor: ‘
ar, + |
e "7
t a
Wise . i
= Rs
vs
DT
Ry
he.
wears " =|
wy , \ aan
x
rs
TM et
aa . 5, (ae
i AY
fs y {7 -- st i
a. put 1
5) Mes oe
| http://www.archive.org/details/comparativegramm01 nme
‘
-,
Bite ea
f
f
r
a4
eta
COMPARATIVE GRAMMAR
or Hk
SANSKRIT, ZEND,
GREEK, LATIN, LITHUANIAN, GOTHIC, GERMAN,
AND SCLAVONIC LANGUAGES.
BY
PROFESSOR F. BOPP.
TRANSLATED FROM THE GERMAN
BY
EDWARD B. EASTWICK, F-.R.S., F.S.A.
VOL. I. 6} 3
FOURTH EDITION.
WILLIAMS AND NORGATE:

14, HENRIETTA STREET, COVENT GARDEN, LONDON;
AND
20, SOUTH FREDERICK STREET, EDINBURGH.

1885.
PREFACE TO THE SECOND EDITION,
the adoption of the Work as a Lecture Book at Oxford, and

by the extensive use which Rawlinson and other eminent
scholars have made of it in their researches. _
It remains to be added, that while the Notes and Preface
made by Professor Wilson, the former Editor, have been re-
tained, [ must be myself held responsible for the errors and
defects, whatever they may be, of the present edition.
EDWARD B. EASTWICK,
HaILEYsuRY COLLEGE,
February, 1854.
PREFACE
TO THE
FIRST EDITION.
——g —__
Te study of Comparative Philology has of late years bean

cultivated in Germany, especially, with remarkable ability
and proportionate success. The labours of Grom, Port,
Bopp, and other distinguished Scholars, have given a new
character to this department of literature; and have sub-
stituted for the vague conjectures suggested by external
and often accidental coincidences, elementary principles,
based upon the prevailing analogies of articulate sounds and the
grammatical structure of language.
But although the fact that a material advance has been
made in the study of Comparative Philology is generally
known, and some of the particulars have been communi-
cated to the English public through a few works on Clas-
sical Literature, or in the pages of periodical criticism;
yet the full extent of the progress which has been effected,
and the steps by which it has been attained, are imper-
fectly appreciated in this country. The study of the
German language is yet far from being extensively pur-
sued; and the results which the German Philologers have
developed, and the reasonings which have led to them.
being accessible to these only who can consult the origina!
writers, are withheld from many individuals of education
and learning to whom the affinities of cultivated speech
are objects of interest and inquiry. Translations of the
works, in which the information they would gladly seek
a
ii PREFACE.
for, is conveyed, are necessary to bring within their reach

the materials that have been accumulated by German in-
dustry and erudition, for the illustration of the history of
human speech.
Influenced by these considerations, Lord Francis Earrton
was some time since induced to propose the translation
of a work which occupies a prominent place in the litera-
ture of Comparative Philology on the Continent—tlie
Vergleichende Grammatik of Professor Bopp of Berlin. Jn
this work a new and remarkable class of affinities has
been systematically and elaborately investigated. Taking
as his standard the Sanskrit language, Professor Borp has
traced the analogies which associate with it and with each
other—the Zend, Greek, Latin, Gothic, German, and Scla-
vonic tongues: and whatever may be thought of some
of his arguments, he may be considered to have established
beyond reasonable question a near relationship between
the languages of nations separated by the intervention of
centuries, and the distance of half the globe, by differences
of physical formation and social institutions,—between the
forms of speech current among the dark-complexioned
natives of India and the fair-skinned races of ancient and
modern Europe;—a relationship of which no_ suspicion
existed fifty years ago, and which has been satisfactorily
established only within a recent period, during which the
Sanskrit language has been carefully studied, and the princi-
ples of alphabetical and syllabic modulation upon which its
grammatical changes are founded, have been applied to its
kindred forms ot speech by the Philologers of Germany.
As the Vergleichende Grammatik of Professor Bopp is
especially dedicated to a comprehensive comparison of lan-
guages, and exhibits, in some detail, the principles of the
Sanskrit as the ground-work and connecting bond of the
comparison, it was regarded as likely to offer most in-
terest to the Philologers of this country, and to be one of
PREFACE, ll
the most acceptable of its class to English students: it

was therefore selected as the subject of translation. The
execution of the work was, however, opposed by two con-
siderations—the extent of the original, and the copiousness
of the illustrations derived from the languages of the East,
the Sanskrit and the Zend. A complete translation de-
manded more time than was compatible with Lord F.
Ecerton’s other occupations; and as he professed not a fa-
miliarity with Oriental Literature, he was reluctant to
render himself responsible for the correctness with which
the orientalisms of the text required to be represented.
This difficulty was, perhaps, rather over-rated, as the
Grammar itself supplies all the knowledge that is needed,
and the examples drawn from the Sanskrit and Zend
speak for themselves as intelligibly as those derived from
Gothic and Sclavonic. In order, however, that the publication
might not be prevented by any embarrassment on this account,
I offered my services in revising this portion of the work;
and have hence the satisfaction of contributing, however
humbly, to the execution of a task which I consider likely to
give a beneficial impulse to the study of Comparative Philology
in Great Britain.
The difficulty arising from the extent of the original
work, and the consequent labour and time requisite for its
translation, was of a more serious description. This, how-
ever, has been overcome by the ready co-operation of a
gentleman, who adds a competent knowledge of German
to eminent acquirements as an Oriental Scholar. Having
mastered several of the spoken dialects of Western India,
and made himself acquainted with the sacred language of
the Parsees during the period of his service under the
Presidency of Bombay, Lieutenant Eastwick devoted part -
of a furlough, rendered necessary by failing health, to a
residence in Germany, where he acquired the additional
qualifications enabling him to take a share in the transla-
a2
iv PREFACE.
tion of the Vergleichende Grammatik. He has accord-

ingly translated all those portions of the Comparative Gram-
mar, the rendering of which was incompatible with the
leisure of the Noble Lord with whom the design originated,
who has borne a share in its execution, and who has taken
a warm and liberal interest in its completion.
The Vergleichende Grammatik, originally published in
separate Parts, has not yet reached its termination. In
his first plan the author comprised the affinities of Sanskrit,
Zend, Greek, Latin, Gothic, and its Teutonic descendants.
To these, after the conclusion of the First Part, he added
the Sclavonic. He has since extended his researches to the
analogies of the Celtic and the Malay-Polynesian dialects,
but has not yet incorporated the results with his general
Grammar. The subjects already treated of are quite suf-
ficient for the establishment of the principles of the com-
parison, and it is not proposed to follow him in his subse-
quent investigations. The first portions of the present
Grammar comprise the doctrine of euphonic alphabetical
changes, the comparative inflexions of Substantives and
Adjectives, and the affinities of the Cardinal and Ordinal
Numerals. The succeeding Parts contain the comparative
formation and origin of the Pronouns and the Verbs: the
latter subject is yet unfinished. ‘The part of the translation
now offered to the public stops with the chapter on the
Numerals, put the remainder is completed, and will be
published without delay.
With respect to the translation, I may venture to affirm,
although pretending to a very slender acquaintance with
German, that it has been made with great scrupulous-
ness and care, and that it has required no ordinary pains
to render in English, with fidelity and perspicuity, the not
unfrequently difficult and obscure style of the original.
H. H. WILSON,
Ocetoler, 1845.
THE AUTHOR’S PREFACE.
ee
I conremPate in this work a description of the compara-

tive organization of the languages enumerated in the title
page, comprehending all the features of their relationship,
and an inquiry into their physical and mechanical laws, and
the origin of the forms which distinguish their grammatical
relations. One point alone I shall leave untouched, the secret
of the roots, or the foundation of the nomenclature of the
primary ideas. I shall not investigate, for example, why the
root i signifies “go” and not “stand”; why the combina-
tion of sounds stha or sta signifies “stand” and not “go.”
I shall attempt, apart from this, to follow ont as it were
the language in its stages of being and march of develop-
ment; yet in such a manner that those who are predeter-
mined not to recognise, as explained, that which they main-
tain to be inexplicable, may perhaps find less to offend them
in this work than the avowal of such a general plan might
lead them to expect. In the majority of cases the primary
signification, and, with it, the primary source of the gramma-
tical forms, spontaneously present themselves to observation
in consequence of the extension of our horizon of language,
and of the confronting of sisters of the same lingual stock
separated for ages, but bearing indubitable features of their
family connection. In the treatment, indeed, of our European
tungues a new epoch could not fail to open upon us in the
discovery of another region in the world of language, namely
the Sanskrit,* of which it has been demonstrated, that, in its
* Sanskrita signifies “adorned, completed, perfect ”; in respect to lan-

guage, “classic”; and isthus adapted to denote the entire family or
race.” It is compounded of the elements sam, “‘ with,” and hkrita
(nom. kritas, krité, kritam), “‘ made,” with the insertion of a euphonic s
(§§. 18. 96.).
vi PREFACE.
grammatical constitution, it stands in the most intimate relation

to the Greek, the Latin, the Germanic, &c.; so that it has-
afforded, for the first time, a firm foundation for the com-
prehension of the grammatical connection between the two
languages called the Classical, as well as of the relation of
these two to the German, the Lithuanian, and Sclavonic.
Who could have dreamed a century ago that a language
would be brought to us from the far East, which should
accompany, pari passt, nay, sometimes surpass, the Greek
in all those perfections of form which have been hitherto
considered the exclusive property of the latter, and be
adapted throughout to adjust the perennial strife between
the Greek dialects, by enabling us to determine where each
of them has preserved the purest and the oldest forms?
The relations of the ancient Indian languages to their
European kindred are, in part, so palpable as to be obvious
to every one who casts a glance at them, even from a dis-
tance: in part, however, so concealed, so deeply implicated in
the most secret passages of the organization of the language,
that we are compelled to consider every language subjected
to a comparison with it, as also the language itself, from new
stations of observation, and to employ the highest powers of
grammatical science and method in order to recognise and
iHustrate the original unity of the different grammars. The
Semitic languages are of a more compact nature, and,
putting out of sight lexicographical and syntactical features,
extremely meagre in contrivance; they had little to part
with, and of necessity have handed down to succeeding ages
what they were endowed with at starting. The tricon-
sonantal fabric of their roots (§. 107.), which distinguishes this
race from others, was already of itself sufficient to designate the
parentage of every individual of the family. The family bond,
on the other hand, which embraces the Indo-European race
of languages, is not indeed Jess universal, but, in most of its
bearings, of a quality infinitely more refined. The members
of this race inherited, from the period of their earliest youth,
PREFACE. vii
endowments of exceeding richness, and, with the capability

(§. 108.), the methods, also, of a system of unlimited com-
position and agglutination. Possessing much, they were able
to bear the loss of much, and yet to retain their local life; and
by multiplied losses, alterations, suppressions of sounds, con-
versions and displacements, the members of the common
family are become scarcely recognisable to each other.
It is at least a fact, that the relation of the Greek to the
Latin, the most obvious and palpable, though never quite
overlooked, has been, down to our time, grossly misunder-
stood; and that the Roman tongue, which, in a grammatical
point of view, is associated with nothing but itself, or with
what is of its own family, is even now usually regarded as
a mixed language, because, in fact, it contains much which
sounds heterogeneous to the Greek, although the elements
from which these forms arose are not foreign to the Greek
and other sister languages, as I have endeavoured partly
to demonstrate in my “ System of Conjugation.”*
The close relationship between the Classical and Germanic
languages has, with the exception of mere comparative lists
of words, copious indeed, but destitute of principle and
critical judgment, remained, down to the period of the appear-
ance of the Asiatic intermediary, almost entirely unobserved,
although the acquaintance of philologists with the Gothic dates
now from a century and a half; and that language is so perfect
in its Grammar and so clear in its affinities, that had it been
earlier submitted to a rigorous and systematit process of com-
parison and anatomical investigation, the pervading relation
* Frankfort.a. M. 1816. A translation of my English revision of

this treatise ( “‘ Analytical Comparison of the Sanskrit, Greek, Latin and
Teutonic Languages,” in the ‘Annals of Oriental Literature,”? London
1820.) by Dr. Pacht, is to be found in the second and third number of the
second annual issue of Seebode’s new Record of Philology and Peda-
gogicalscience. Grimm’s masterly German Grammar was unfortunately
unknown to me when I wrote the English revision, and I could then
make use only of Hickes and Fulda for the old German dialects,
Vili PREFACE.
of itself, and, with it, of the entire Germanic stock, to the

Greek and Roman, would necessarily have long since been
unveiled, tracked through all its variations, and by this time
been understood and recognised by every philologer.* For
what is more inaportant, or can be more earnestly desired by
the cultivator of the classical languages, than their comparison
with our mother tongue in her oldest and most perfect form ?
Since the Sanskrit has appeared above our horizon, that element
can no longer be excluded from a really profound investigation
of any province of language related to it; a fact, however,
which sometimes escapes the notice of the most approved
and circumspect labourers in this department.t We need
* Rask has been the first to supply a comprehensive view of the close
relatiouship between the Germanic and the Classical Languages, in his
meritorious prize treatise ‘‘On the Thracian Tribe of Languages,”’ com-
pleted in 1814 and published in 1818, from which Vater gives an extract
in his Comparative Tables. It cannot be alleged as a reproach against
him that he did not profit by the Asiatic intermediary not then exten-
sively known; but his deficiency in this respect shews itself the more
sensibly, as we see throughout that he was in a condition to use it with
intelligence. Under that deficiency, however, he almost everywhere
halts halfway towards the truth. We have to thank him for the
suggestion of the law of displacement of consonants, more acutely
considered and fundamentally developed by Grimm (§. 87., and see
Vater, §. 12.). ;
+ We refer the reader to the very weighty judgment of W. von. Hum-
boldt onthe indispensable necessity of the Sanskrit for the history and
philosophy of language (Indische Bibl. I. 133). We may here borrow,
also, from Grimm’s preface to the second edition of his admirable
Grammar, some words which are worthy of consideration (I. vi.): “ As
the too exalted position of the Latin and Greek serves not for all
questions in German Grammar, where some words are of simpler and
deeper sound, so however, according to A. W. Schlegel’s excellent re-
mark, the far more perfect Indian Grammar may, in these cases, supply
the requisite corrections. The dialect which history demonstrates to be
the oldest and least corrupted must, in the end, present the most pro-
found rules for the general exposition of the race, and thus lead us on to
the reformation, without the entire subversion of the rules hitherto
discovered, of the more recent modes of speech,”
~ PREFACE. ix
not fear that that practical and profound research in utrdgque

lingué, which is of most importance to the philologer can
suffer prejudice by extension over too many languages;
for the variety vanishes when the real identity is recog-
nised and explained, and the false light of discrepancy is
excluded. It is one thing, also, to learn a language,
another to teach one, i.e. to describe its mechanism and
organization. The learner may confine himself within the
narrowest limits, and forbear to look beyond the language
to be studied: the teacher’s glance, on the contrary, must pass
beyond the confined limits of one or two members of a family,
and he must summon around him the representatives of the
entire race, in order to infuse life, order, and organic mutual
dependency into the mass of the languages spread before him.
To attempt this appears to me the main requirement of the
present period, and past centuries have been accumulating
_ materials for the task.
The Zend Grammar could only be recovered by the process
of a severe regular etymology, calculated to bring back the
unknown to the known, the much to the little; for this re-
markable language, which in many respects reaches beyond,
and is an improvement on, the Sanskrit, and makes its theory
more attainable, would appear to be no longer intelligible to
the disciples of Zoroaster. Rask, who had the cpportunity to
satisfy himself on this head, says expressly (V. d. Hagen,
p- 33) that its forgotten lore has yet to be rediscovered. I
am also able, I believe, to demonstrate that the Pehlvi trans-
lator (tom. IT. pp. 476, et seq.) of the Zend Vocabulary, edited
by Anquetil, has frequently and entirely failed in conveying
the grammatical sense of the Zend words which he translates.
The work abounds with singular mistakes; and the distorted
relation of Anquetil’s French translation to the Zend expres-
sions is usually to be ascribed to the mistakes in the Pehlvi
interpretations of the Zend original. Almost all the oblique
cases, by degrees, come to take rank as nominatives: the
numbers, too, are sometimes mistaken. Further, we find forms
x PREFACE.
of cases produced by the Pehlvi translator as verbal persons,

and next these also confounded with each other, or translated
by abstract nouns.* Anquetil makes, as far as I know, no
* J give the Zend expressions according to the system of representation

explained in §.30., annexing the original characters, which are exhibited
in type for the first time in this book, and which were lately cut at the
order of the Royal Society of Literature by R. Gotzig, according to the
exemplar of the lithographed M.S. of M. Burnouf. I give the Pehlvi
words exactly according to Anquetil (II. 435): G¢5.wgas ahmdhim,
“nav,” P. rouman (cf. p. 502, roman, “nos”), A. “je,” “mot;” rsddgreras
ahubya, “ bonis” (with dual termination, 3. 215), P. avaéhk, A. “bon,”
“* excellent ;” pOw7As aété, “hi,” “i,” P. varman, “is,” A. “lué ;”
Gewgu anhém, “I was,” or also “TI am,” P. djanounad, “he is,” A. “il
est ;” aorgurgy anheus, “mundi,” P. akhé, A. “le monde ;” FyHgy srs
avaéshanm, “ horum,” P. varmouschan, “it,” A. “eux ;” 5905257055 baraiti,
“fert,” P, dadrouneschné, “the carrying” (eschné, in Pehlvi, forms abstract
substantives), A. “il porte,” “il execute,” “porter ;” 35s bis, “twice,”
P. dou, “two,” A. “deux ;” bérétebié Uy 355902575, baratibyé, ‘ ferenti-
bus?” (unquestionably a plural dative and ablative), P. dadrouneschné, “the
carrying,” A. “porter;” we t4, “tus.” BP. tou, “tun.” Ai” AMAUEO
ticha, “eaque,” (neut. 3. 231), P. zakedj, A. “ce ;” Lousy, jatd, “the
smitten” (cf. Sansk. hatas from han), P. maitouned, “he smites,” A. ‘il
Srappe ;” POA JAS, janat, “he smote,” P. maitouneschné, “the smiting,”
A. “frapper ;” wheres zanthra, “per genitorem,” P. zarhounad, “ gi-
gnit,” A, “il engendre,” shoes stri, “femina,” P. vahad, A. “ femelle ;”
¢ S39 8trim, “ feminam,” P. vahad, A. “femelle ;” Gy Zwsposs stdranm,
“ stellarum,” P. setaran, A. “les étoiles ;” sa pasgas0d fra-détdi, “to
the given,” or “especially given,” P. feraz Acheschné (nomen actionis),
A. “donner abondamment ;” GFyYjwGwasp gaéthananm, “mundorum,”
P. guehan (cf. w>)> A. “Te monde ;” assG peo. gatiimcha, “ locum-
que,” P. gah, A. “lieu ;” sds J nar, “of the man,” P. guebna hamat
advak, A. “un homme ;” assy nara, “two men,” P. guebna hamat dou,
A. “deux hommes ;” Gy J299575a0 f ndirihananm, “feminarum,” P. nai-
vik hamat sé, A. “trois (ou plusieurs) femmes ;” Gy907d thryanm,
“trium,” P. sevin, A. “ troisiéme ;” assGegasQ vahmémcha, “ precla-
rumque,” P. néaeschné, “ adoratio,” A. “je fais néaesch ;” sasgash vah-
mii, “preclaro,” P. néaesch, honam, “ adorationem facio,” A, “je bénis
et
PREFACE. xi
remark on the age of the Vocabulary to which I advert; while

he ascribes to another, in which the Pehlvi is interpreted
through the Persian, an antiquity of four centuries. The
et fais néaesch.” I do not insist on translating the adjective asguQ valma

by “preclarus,” but I am certain of this, that vahmén and vahmdi are
nothing else than the accusative and dative of the base vahma; and that
savgash vahmdi could be the first person of a verb is not to be thought pos-
sible tora moment. Anquetil, however, in the interlinear version of the be-
ginning ofthe V. S. attempted by him, gives two other evident dativescom-
pounded with the particle ass cha, “and,” as the first person singular of the
present, viz. asgssaw
7Sbasyans esnaothrdi-cha, A570 A599.59059.559.509
Srasastayaé-cha (see }. 164.), by “placere cupio,” “‘vota facio. One sees
then, from the example here adduced, the number of which I could with
ease greatly increase, that the Pehlvi Translator of the said Vocabulary
has, no more than Anquetil, any grammatical acquaintance with the Zend
language, and that both regarded it rather in the light of an idiom, poor
in inflexions; so that,as in the Pehlviand Modern Persian, the grammatical
power of the members of a sentence would be to be gathered rather from
their position than from their terminations. And Anquetil expressly
says (II. 415.): “‘ Za construction dans la langue Zende, semblable en cela
aux autres ididmes de T Orient, est astreinte a peu de regles(!). La for-
mation des tems des Verbes y est & peu prés la méme que dans le Persan,
plus trainante cependant, parce qu'elle est accompagnée de toutes les
voyelles (!). How stands it, then, with the Sanskrit translation of the
Jzeschne made from the Pehlvi more than three centuries before that of
Axcquetil, This question will, without doubt, be very soon answered by
M. E. Burnouf, who has already supplied, and admirably illustrated
(Nouv. Journ, Asiat., T. III. p. 321), two passages from the work in a
very interesting extract from its Commentary on the V. S. These pas-
sages are, however, too short to permit of our grounding on them over-
bold influences as to the whole; moreover, their contents are of such a
nature that the inflexionless Pehlvi language could follow the Zend ori-
ginal almost verbatim. The one passage signifies, ‘I call upon, I mag-
nify the excellent pure spell, and the excellent man, the pure and the
strict, strong like Dami (? cf. Sansk. upamdna, “ similarity ;” and V. S.,
p- 423, daéméis drujé) Izet.” It is, however, very surprising, and of evil
omen, that Neriosengh, or his Pehlvi predecessor, takes the feminine
genitive dahmayds as a plural genitive, since this expression is evidently,
as Burnouf rightly remarks, only an epithet of dfritéis. I abstain from
speaking of the dubious expression déméis upamanahé, and content my-
self
xii PREFACE.
one in question cannot therefore be ascribed to any very late

period. The necessity, indeed, of interpretation for the Zend
must have been felt much sooner than for the Pehlvi, which
remained much longer current among the Parsee tribes. It
was therefore an admirable problem which had for its solution
the bringing to light, in India, and, so to say, under the very
eye of the Sanskrit, a sister language, no longer understood,
and obscured by the rubbish of ages ;—a problem of which the
solution indeed has not hitherto been fully obtained, but beyond
doubt will be. The first contribution to the knowledge of
this language which can be relied on—that of Rask —namely,
his treatise “ On the age and authenticity of the Zend Language
and the Zend-Avesta,” published in 1826, and made generally
accessible by V. d. Hagen’s translation, deserves high honour
as a first attempt. The Zend has to thank this able man
(whose premature death we deeply deplore) for the more
natural appearance which it has derived from his rectification
of the value of its written characters. Of three words of
different declensions he gives us the singular inflections, though
with some sensible deficiencies, and those, too, just in the places
where the Zend forms are of most interest, and where are some
which display that independence of the Sanskrit which Rask
claims, perhaps in too high a degree, for the Zend ; a language
we are, however, unwilling to receive as a mere dialect of
the Sanskrit, and to which we are compelled to ascribe an
independent existence, resembling that of the Latin as com-
pared with the Greek, or the Old Northern with the Gothic.
For the rest, I refer the reader to my review of Rask’s and
Bohlen’s treatises on the Zend in the Annual of Scientific
Criticism for December 1831, as also to an earlier work
(March 1831) on the able labours of E. Burnouf in this newlv-
self with having pointed out the possibility of another view of the cm-
siruction, different from that which has been very profoundly discussed
by Burnouf, and which is based on Neriosengh. The second pussize
sigaifies, “I call upon and magnify the stars, the moon, the swi, the
eternal, self-created lights!
PREFACE. xiii
opened field. My observations, derived from the original texts

edited by Burnouf in Paris, and by Olshausen in Hamburgh,
already extend themselves, in these publications, over all parts
of the Zend Grammar; and nothing therefore has remained for
me here, but further to establish, to complete, and to adjust the
farticulars in such a manner that the reader may be conducted
on a course parallel with that of the known languages, with the
greatest facility towards an acquaintance with the newly-disco-
yered sister tongue. In order to obviate the difficulty and the
labour which attend the introduction of the learner to the Zend
and Sanskrit—difficulty sufficient to deter many, and to harass
any one—I have appended to the original characters the pronun-
ciation, laid down on a consistent method, or in places where, for
reasons of space, one character alone is given, it is the Roman.
This method is also perhaps the best for the gradual introduc-
tion of the reader to the knowledge of the original characters.
As in this work the languages it embraces are treated for
their own sakes, i.e. as objects and not means of knowledge,
and as I aim rather at giving a physiology of them than an
introduction to their practical use, it has been in my power to
omit many particulars which contribute nothing to the
character of the whole; and I have gained thereby more
space for the discussion of matters more important, and more
intimately incorporated with the vital spirit of the language.
By this process, and by the strict observance of a method which
brings under one view all points mutually dependent and
mutually explanatory, I have, as I flatter myself, succeeded in
assembling under one group, and in a reasonable space, the
leading incidents of many richly-endowed languages or grand
dialects of an extinct original stock. Special care has been
bestowed throughout on the German. This care was indispen-
sable to one who, following Grimm’s admirable work, aimed
at applying to it the correction and adjustment that had become
necessary in his theory of relations, the discovery of new affinities,
or the more precise definition of those discovered, and to catch,
with greater truth, at every step of grammatical progress, the
xiv PREFACE,
monitory voices of the Asiatic as well as the European sisterhood.

It was necessary, also, to set aside many false appearances of afli-
nity; as, for example, to deprive the 7 in the Lithuanian geri of
its supposed connection with the of Gothic, Greek, and Latin
forms, such as gédai, aya6or, boni (see p. 251, Note f, and com-
pare Grimm I. 827.11); and to disconnect the Latin is of lupis
(lupibus) from the Greek is of AdKots (AvKot-or). As concerns
the method followed in treating the subject of Germanic
grammar, it is that of deducing all from the Gothic as the
guiding star of the German, and explaining the latter simulta-
neously with the older languagesand the Lithuanian. At the close
of each lecture on the cases, a tabular view is given of the results
obtained, in which every thing naturally depends on the. most
accurate distinction of the terminations from the base, which
ought not, as usually happens, to be put forward capriciously,
so that a portion of the base is drawn into the inflection, by
which the division becomes not merely useless, but injurious,
as productive of positive error. Where there is no real
termination none should be appended for appearance sake: thus,
for example, we give, §. 148, p. 164, the nominatives ydpa,
terra, giba, &c., as without inflection cf. §. 137. The division
gib-a would lead us to adopt the erroneous notion that a is the
termination, whereas it is only the abbreviation of the 6 (from the
old 4, §. 69.) of the theme.* In certain instances it is extraordi-
* The simple maxim laid down elsewhere by me, and deducible only
from the Sanskrit, that the Gothic 6 is the long of a, and thereby when
shortened nothing but a, as the latter lengthened can only become 4, ex-
tends its influence over the whole grammar and construction of words, and
explains, for example, how from dags, “day ” (theme DAG A),may be de-
rived, without change of vowel, dégs (DOGA), “daily”; for this deriva-
tion is absolutely the same as when in Sanskrit rdjata, “argenteus,” comes
from rajata, “ argentum,’”’on which more hereafter. Generally speaking,
and with few exceptions, the Indian system of vowels, pure from consonantal
and other altering influences, is of extraordinary importance for the eluci-
dation of the German grammar: on it principally rests my own theory of
vowel change, which differs materially from that of Grimm, and which I
explain by mechanical laws, with some modifications of my earlier defini-
tions
PREFACE, xv
narily difficult in languages not now thoroughly understood to

hit on the right divisions, and to distinguish apparent termina-
tions from true. I have never attempted to conceal these difficul-
ties from the reader, but always to remove them from his path.
The High German, especially in its oldest period (from the
eighth to the eleventh century), I have only mentioned in the
general description of forms when it contributes something of
importance. The juxta-position of it in its three main periods
with the Gothic, grammatically explained at the close of each
chapter, is sufficient, with a reference also to the treatise on
sounds intended to prepare and facilitate my whole Grammar,
after the model of my Sanskrit Grammar. Wherever, in
addition, explanatory remarks are necessary, they are
given. The second part will thus begin with the com-
parative view of the Germanic declensions, and I shall then
‘proceed to the adjectives, in order to describe their formations
of gender and degrees of comparison ; from these to the pronouns.
As the peculiarities of inflection of the latter must have, for
the most part, already been discussed in the doctrine of the
universal formation of the cases, inasmuch as they are inti-
mately connected and mutually illustrative, what will remain to
be said on their behalf will claim the less space, and the main
compass of the second division will remain for the verb. To
the formation and comparison of words it is my intention to
devote a separate work, which may be considered as a completion
of its antecedent. In this latter the particles, conjunctions,
and original prepositions, will find their place, being, I consider,
partly offshoots of pronominal roots, and partly naked roots of
tions, while with Grimm it has a dynamic signification. A comparison

with the Greek and Latin vocalism, without a steady reference to the
£anskrit, is, in my opinion, for the German more confusing than enlight-
ening, as the Gothic is generally more original in its vocal system, and at
least more consistent than the Greek and Latin, which latter spends its
whole wealth of vowels, although not without pervading rules, in merely
responding to a solitary Indian a (septimus for septamas, quatuor for
chatvdr-as téacap-es, momordi fr mamarda).
3
xvi PREFACE.
this class of words,* and which will, therefore, be treated in

this point of view among the pronominal adjectives.+ It is
likely that a chasm in our literature, very prejudicial to inquiries
of this kind, may be shortly filled up by a work ready for the
press, and earnestly looked for by all friends of German and
general philology, the Old High German Treasury of Graff.
What we may expect from a work founded on a comprehensive
examination of the MS. treasures of libraries national and
foreign, as well as on a correction of printed materials, may be
gathered from a survey of the amount contributed to knowledge
in a specimen of the work, small, but happily selected, “ The
Old High German Prepositions.”
* [ refer the reader preliminarily to my two last treatises (Berlin, Ferd.
Diimmler) ‘‘ On Certain Demonstrative Bases, and their connection with
various Prepositions and Conjunctions,” and “ On the Influence of Pro-
nouns on the Formation of Words.” Compare, also, C. Gottl. Schmidt's
excellent tract “Quest. Gramm. de Prepositionibus Grecis,” and the
review of the same, distinguished by acute observations, by A. Benary,
in the Berlin Annual (May 1830). If we take the adverbs of place in
their relation to the prepositions—and a near relation does exist—we shall
find in close connection with the subject a remarkable treatise of the
minister W. von Humboldt, “On the Affinity of the Adverbs of Place to
the Prepositions in certain Languages.’ The Zend has many grammatical
rules which were established without these discoveries, and have since
been demonstrated by evidence of facts. Among them it was a satisfaction
to me to find a word, used in Sanskrit only as a preposition (ava, ‘ from,’’)
in the Zend a perfect and declinable pronoun (§.172.). Next we find
sa-cha, “isque,” which in Sans\rit is only a pronoun, in its Zend shape
spas ha-cha (§. 53.), often used as a preposition to signify “out of” ;
the particle as cha, “ and,” loses itself, like the cognate que in absque, in
the general signification.
* Remark.— What in §. 68. is said of the rise of the u or o out of the
older a is so far to be corrected according to my later conviction, that
nothing but a retroactive influence is to be ascribed to the liquids; and
the u and the 0, in forms like plinfemu (mo), plintyu, are to be exempted
frum the influence of the antecedent consonants.”
+ The arrangement thus aunounced, as intended, has undergone, as will
be seen, considerable modification.— Editor.
F. BOPP.
Bern, 18383.
- COMPARATIVE GRAMMAR,
CHARACTERS AND SOUNDS.

Sanskrit writing distinguishes the long from their cor-
responding short vowels by particular characters, slightly
differing from these latter in form. We distinguish the long
vowels, and the diphthongs z e and Sf 0, which spring from
4 and w united with an antecedent a, by a circumflex. The
simple vowels are, first, the three, original and common to all
languages, a, 7, u, short and long; secondly, a vowel r, pecu-
liar to the Sanskrit, which I distinguish by +, and its long
sound by 7. The short r (=) is pronounced like the con-
sonant r with a scarcely-distinguishable i, and in European
texts is usually written rz; the long 7 (=) is scarcely to
be distinguished from the union of an r with a long 7. Both
vowels appear to me to be of later origin; and r presents
itself generally as a shortening of the syllable ar by sup-
pression of the a. The long 7 (=) is of much rarer occur-
rence. In declension it stands only for a lengthening of the r,
where, according to the laws of the formation of cases, a short
vowel at the end of the inflective base must be lengthened;
and in the conjugation and formation of words, those roots to
which grammarians assign a terminating @ 7 almost always
substitute for this unoriginal vowel we ar, § ir, 3 ér, or,
after labials, Sx dr. The last simple vowel of the Sanskrit
writing belongs more to the grammarians _[G. Ed. p. 2.]
than to the language: it is in character, as well as in pro-
nunciation, an union of an @/ with qr (wz), or, when
lengthened, with & 7 (=). We require no representative
for this vowel, and shal] not further advert to it.
2, Sanskrit possesses two kinds of diphthongs. In the one,
B
2 CHARACTERS AND SOUNDS,
_ a short a united with a following ¢ becomes z é (equivalent

to the French ai), and with « becomes Wt 6 (equivalent to
the French au); so that neither of the united elements is
heard, but both melt into a third sound. In the second kind,
a long. @ with a following i becomes @ ai, and with wu,
St au, as in the German words waise, baum; so that the
two elements form indeed one syllable, but are both audible.
In order, however, to fix the observation on the greater
weight of the a in this diphthong, we write di for @, and dw
for wt. That in v é and St é a short, in @ di and 8 du,
a long a is bound up, linfer from this, that where, in order
to avoid a hiatus, the last element of a diphthong merges
into its corresponding semi-vowel, out of z é and ®t 6 pro-
ceed the sounds wq ay and ¥@ av (with short a), but out
of 24i and wt du proceed by and dv. If, according to
the rules of combination, a concluding wt 4, with an ¥ 4%
2 7%, or tu, & & Of a following word, be contracted, like the
short a, into @ é and at d, but not into @ di and @ du,
this, in my view, is to be understood as if the long a, before
its combination with the initial vowel of the following word,
had shortened itself. This should the less surprise us, as the
long a before a dissimilar vowel of an appended inflexion or
a suffix entirely disappears; and, for example, get dadd
with sa ws makes neither géta daddus, nor ata dadés,
but egq dadus. The opinion [ have already expressed on
[G. Ed. p.3.] this point I have since found confirmed
by the Zend ; in which sw di always stands in the place
of the Sanskrit @ di, and gus do or >a du for St du. In
support, also, of my theory, appears the fact, that a con-
cluding a (short or long) with a following z é or Wt 4, be-
comes @ di and @i du; of which it is to be understood, that
the short a contained in é and 6 merges with the antecedent
a into a long a, which then, with the 7 of the diphthong é,
becomes di, and with the uw of 6, becomes du. For example,
aaaa mamditat, from AR VAR mama état, is to be understood
CHARACTERS AND SOUNDS. 3
as if the diphthong z é united its first element @ with the

preceding a into 4, and with this, further united its last
element (7) into & 4i. [Compare § 688, p. 917.]
3. Among the simple vowels the old Indian alphabet is
deficient in the designation of the Greek epsilon and omicron
(e€ and ©) whose sounds, if they existed when the Sanskrit was
a living language, yet could only have evolved theniselves,
subsequently to the fixing of its written character, out of the
short a; for an alphabet which lends itself to the subtlest
gradations of sound would assuredly not have neglected the
difference between a, é, and 6, if the sounds had been forth-
coming.* It is important here to observe, that in the oldest
Germanic dialect, namely, the Gothic, the sounds and charac-
ters of the short e and o are also wanting, and that either
a, i, or u corresponds, in that dialect, to our German short e.
For example, faltha, “ich falte,” “I fold ;’ giba, “ich gebe,”
“I give.” In the Zend the Sanskrit # a remains usually » a,
or has changed itself, according to certain ([G. Ed. p.4.]
tules, into ¢ é. Thus, for example, before a concluding m
we always find ¢ é; compare the accusative §e7 Gd puthré-m
“filium”™ with Waa putra-m ; and its genitive pws? Gd
puthra-hé with Tam@ putra-sya. In Greek the Sanskrit wa
becomes 4, €, or o, without presenting any certain rules for the
choice on each occasion between these three vowels; but the
prevailing practice is, that in the terminations of nominal
bases the Greek o answers to the Indian W a, except in the
vocatives, where an ¢ is substituted. In the Latin, besides
a, é,and 06, u also is employed, in the terminations of nouns
of the second declension and of the first person plural, as also
in some adverbial suffixes, to replace the Sanskrit ¥ a
4, As in the Greek the short Sanskrit a is oftener replaced
by ¢ or o than by a short a,so the long = 4 is oftener re-
* Grimm, Vol. i. p. 594; with whom I entirely concur in this matter;
having long abandoned a contrary opinion, which | maintained in 1819
in the Annals of Oriental Literature.
B2
ee
4 CHARACTERS AND SOUNDS.
presented by 7 or w than by a long alpha: and though in the

Doric the long a has maintained itself in places where the
ordinary dialect employs an 7, no similar trace of the long @
for w is to be found. wuatfa dadhdmi, “I place,” becomes
riOnut; zetfa daddmi, “I give,” ddwur; the dual termi-
nation #T@_ tdm answers to Ty, and only in the imperative
to Twv: on the other hand, the "Tq dm of the genitive plural
is always represented by wv. Never, if we except pecu-
liarities of dialect, does either 7 or w stand for the Indian
diphthongs @ é or Wt 6, formed by ¥ 7 or an ¥ ~ following
a long. 4: for the first, the Greek substitutes e: or o: (because
for @ a, and also for a, € and o are the substitutes), and for
the last, ev or ov, Thus, zufa émi, “I go,’ becomes etm;
ae patés, “thou mayest fall,” nintois; ae véda, “ I know,”
oida; mt go, mas. fem. “a bullock or heifer,” Pod-s. From
this dropping of the ¢ or ~ in the Indian diphthongs é and 6 it
[G. Ed. p. 5.] may happen that a, e, or 0, answer to these
diphthongs; thus, wate ékataras, “one of two,” becomes
EKATEPOS 5 aq dévri,* “brother-in-law,” Latin, levir (nom. 241
dévd, accus. tata devar-am), becomes dazjp (from daFijp, dai-
Fne); ta_ déva-s, “God,” @eds; and the o in Bods, Boi, stands
for Bou-ds, Bov-i, the wu of which must have passed into F, and
certainly did so at first, as is proved by the transition into the
Latin bovis, bovi, and the Indian af gavi (locative) from gé-i.
5. In Latin we sometimes find the long e, which, however,
may be shortened by the influence of the following conso-
nant, arising from the mixturé of a and i, as in the above-
mentioned word /évir, and in the subjunctive amémus; cf.
araaa kdmayéma, from kdmaya-tma.
6. If we inquire after the greater or less relative weight
of the vowels of different quality, I have discovered, by
* The original has devr, but, as observed in p. 1, in European texts it

is usual to write ri for @Y; and the absence ofany sign for the vowel sound
is calculated to cause embarrassment: it seems advisable, therefore, to
expreas Y by ri.— Editor.
various but sure appearances, which I shall further illustrate

in my treatise on Forms, that in Sanskrit B a and w 4 are
graver than the corresponding quantity of the vowel 7; and
this discovery is of the utmost importance for every Treatise
on special as well as comparative Grammar. It leads us, in
particular, to important discoveries with respect to the Ger-
manic modification of vowels. In Latin, also, the i may be
considered as lighter than a, and generally takes the place of
the latter when a root with an original a would otherwise be
burthened with a reduplication of sound. Hence, forexample,
abjicio for abjacio, tetigi for tetagi. I am compelled by this
view to retract an earlier conjecture, that the 7 in tetigi was
produced by a virtue of assimilation in the termination 7. I
have also to relieve myself from my former theory, that the e
in words like inermis, imberbis, instead of _—_[G. Ed. p. 6.]
inarmis, imbarbis, springs from a retrospective power of
assimilation in the following i, after the fashion of the modi-
fication of the vowel in German (Grimm, p- 80), and must
place it in the same class with the e in such forms as abjectus
and tubicen. The Latin radical a, for instance, is subject to
a double alteration, when the root is burthened with ante-
cedent les or words: it becomes 7 in open syllables, but
e if the vowel is pressed upon by a following consonant un-
attended by a vowel. Hence we have tubicen, abjectus, in
contrast to tubicinis, abjicio; and inermis, imberbis, not inirmis,
imbirbis: on the contrary, inimicus, insipidus, not inemicus,
insepidus. In connection with this stands the transition of the
first or second declension into the third. As usis the masculine
form for a, we ought to say inermus, imberbus; but inermis,
imberbis, and other such forms, owe their origin to the les-er
t
weight of the 7. With the displacement of the accent, where
=e”
it ovcurs, this change of the vowel has nothing to do; but the
F
removal of the accent and the weakening of the vowel are
nearly related, and are both occasioned by the composition.
In the Lithuanian we find similar appearances ;_as, for ex-
e,
Tae
sagas
ye
6 CITARACTERS AND SOUNDS.
ample, pénas, “lord,” at the end of conapounds, is weakened

into ponis, as rdtponis, “ councillor,” Germ. rathsherr.” (See
p. 1305, Note *).
7. Sanskrit Grammar gives no certain indication of the
relative weight of the w~ with regard to the other original
vowels, The w is a vowel too decided and full of character to
allow of its being exchanged in this language, in relief of its
weight, for any other letter. It is the most obstinate of all,
and admits of no exclusion from a terminating syllable, in
cases where a and ¢ admit suppression. Nor will it retire
[G. Ed. p.7.] from a reduplicated syllable in cases where
a allows itself to be weakened down to 7. Thus in Latin we
have pupugi, tutudi; while a, in cases of repetition, is re-
duced to ¢ or é (tetigi, fefelli, &c.) In the Gothic, also, the
wu may boast of its pertinacity: it remains firm as the ter-
minating vowel of nominal bases where @ and 7 have under-
gone suppression, and in no single case has it been extin-
guished or transmuted. No power, however, exists which
will not yield at last to time; and thus in the High German,
whose oldest records are nearly four centuries younger than
Ulphilas, the w has, in many cases, given way, or become in
declension similar to 7. (See also §§ 490, 584.)
8. If, in the matter of the relative dignity of the vowels, we
cast a glance at another race of languages, we find in Arabic
the u taking precedence in nobility, as having its place in the
ncminative, while the declension is governed by the change
of the terminating vowel ; 7, on the contrary, shews itself to
be the weakest vowel, by having its place in the genitive, the
most dependent case of the Arabic, and one which cannot be
separated from the governing word. J, also, is continually
used in cases where the grammatical relation is expressed by
a preposition, Compare, also, in the plural, the dna of the
nominative with the termination éna of the oblique cases. A
stands between the strong w and the weaker 7; and under
the threefold change of vowels has its place in the accusative,
which admits of more freedom than the genitive. In the

oblique cases, however, of nouns, and in the two-fold change
of vowels, it stands opposed to the w of the nominative, and
in the dependent subjunctive of the verb to the w of the
independent indicative.
9. Between the vowels and the consonants, or at the close of
the list of vowels, are commonly placed two signs, the sounds
of which are rather to be considered as ap- [G. Ed. p 8.]
pendages to, or modifications of, the preceding vowels, than as
independent sounds, and take, also, no place in the alphabet of
the Native Grammarians, inasmuch as they are considered
neither asconsonants nor vowels, but rather as complements
to the latter. The first, which we distinguish by 2, is called
Anuswara, “echo;” and is;in fact,a thick nasal echo, which I
think is best represented by the nasal x at the end ofa French
syllable. The weakness of its expression is discernible in the
fact that it does not, like a consonant, impede the euphonic
influence of an i or u on a following s, (see Sanskrit Gram-
mar, R. 101°). It has its place before semi-vowels (a y,
ir, & 1, % x), sibilants, and h; and we might thence term
it the nasal of the two last lists of consonants, and assign its
alphabetical place between them. A concluding 4 m, fol-
lowed by a consonant of the said two lists, passes into Anu-
swara; for example, WeATH tasydm, “in this,’ becomes wei
tasy4n, with the French nasal pronunciation of the , if such
a word as wat rdtrdu, “in the night,” come after. In con-
nection with the 4 s of a verbal termination, a radical q »
also passes into Anuswara; as, éfa hazisi, “thou killest,”
from ¢q fan. Great confusion, however, has arisen from
the circumstance that the Indian copyists allow themselves to
express the unaltered concluding & m,as well as all the nasal
alterations, and, in the middle of words, each of the six nasal
sounds (the proper AnuswaAra included), by Anuswara.* Ihave
* The practice is not unauthorized by rule. A final 4 is convertible to
Anuswara before any consonant (Pan. 8. 3.23); and a media 4 or @ is
convertible
endeavoured, in my Grammar, to remedy this confusion in the

simple theory of Anuswara. My predecessors in the treatment
of Sanskrit Grammar make no distinction between the real
and the supposititious Anuswara. Colebrooke gives it, in
[G. Ed. p.9.] general, the pronunciation of », and calls it
“a shortening of the nasal consonants at the end ofa syllable,”
which leads to the error, that each of the nasal characters,
even
the concluding am, may be abbreviated into Anuswara.
Forster expresses it by the in the English word plinth;
Carey and Yates by the English combination ng; Wilkins
by m. All substitute it for the concluding 4 of grammatical
terminations: and as they give rules for the transition of the
Anuswara into mA or 7 , the necessary consequence occurs, that
we must write abhavan or abhavang,“1 was;” duntan or dan-
tang, “a tooth ;” not abhavam, dantam. Colebrooke, on the
other hand, expressing a Sanskrit inscription in Roman letters
(Asiatic Transactions, Vol. VIL.) gives the proper termina-
tion m, and before ¢, by a euphonic rule, n; but he maintains
the original m before sibilants and half vowels where Anu-
swara is due ; as vidwishdm srimad, for fafgat vidwishdi. On
the other hand, F. von Schlegel and Frank write n, for the
value of Anuswara, in the place of m in several grammatical
terminations, The first, for example, gives danan, “a gift,” for
dinam; the second, ahan for aham, “I.” A. W. von Schlegel
gives rightly m instead ofa spurious or representative Anu-
swaraat the end of words; and makes,for example,the infinitive
termination in tum, not in tunor tung. He, nevertheless, on this
important point of grammar, retains the erroneous opinion,
that the Anuswara is a variable nasal, which, before vowels,
must of necessity pass into m(Preface to the Bhag. Gita, p. xv.);
while the direct converse is the fact, that the concluding m is
convertible to Anuswiara before any consonant except a semi-vowel or a

nasal. (Ib. 8.3.24.) Such are the rules, In practice, the mutation of the
final 4 is constant: that of the medial nasal is more variable, and in ge-
neral the change occurs before the semi-vowels and sibilants.—Zdiécr.
the variable nasal, which, under certain conditions, passes into

the proper Anuswara; but before vowels is necessarily re-
tained, both in writing and pronunciation. [G. Ed. p. 10.]
That Von Schlegel also still continues the original 4 m at the
end of words as an euphonic alteration of the dead sound of
Anuswara appears from his mode of printing Sanskrit text, in
which he makes no division between a concluding mand
the commencing vowel of the following word ; while he does
make a division after 4 m, and thereby shews that he admits
a division after terminating letters which remain unaffected
by the influence of the letters which follow. If, however, we
write aq Wadta tdn abravit, “he said to them,” we must
also write wa WHAtA, tdm abravit, “he said to her 7” not
amadia, tdmabravit, for the 4 of ata tdm is original, and not,
as Von Schlegel thinks, begotten out of Anuswara. The conjec-
ture of C. Lassen (Ind. Bibl. Book IIL. p. 39), that the Anuswara
is to be understood, not as an after sound (Nachlaut), not as an
echo (Nachha’l), but as a sound which regulates itself by that
which follows—as it were the term Nacilaut, with the accent
on laut*—appears to me highly improbable. Schlegel’s xasalis
mutabilis would indeed bejustified by this view, and the imputa-
tion of error removed from the Indian Grammarians, to whom
we willingly concede a knowledge of the value of the Sanskrit
signs of sound, and whom we are unwilling to censure for de-
signating a half sound as mutable, in a language whose termi-
* This seems intended for an explanation, for Lassen has nothing like
it. Ihave not found an etymological explanation of the term in any
grammatical commentary; but it may be doubted if the explanation of
the text, or that given by Lassen, be correct. Anuswara may indeed be
termed sequens sonus ; but by that is to be understood the final or closing
sound ofa syllable. Any other nasal may be used as the initial letter of
a syllable; but the nasal Anuswara is exclusively an “after” sound, or
final It is not even capable of blending, as it were, with a following
vowel, like a final m or m, as in tén- or témabravit. It is the legitimate
representative of either of the other nasals when those are absolutely
terminal,
nating sounds are almost always governed by the following -

words. It is true the half sound owes its being to the muta-
bility of aconcluding m, but is not mutable itself, since it never
has an independent existence ofits own at the end of any word:
in the middle, however, of a radical syllable, as eq dans,
fa hizs, it is susceptible of expulsion, but not of alteration.
[G. Ed. p. 11.) That the Indian Grammarians, however,
consider the m and not the 7% as the original but mutable
letter in grammatical terminations, like BA am, ATA bhydm,
&c., appears from the fact that they always write these
terminations, where they give them separate, with the labial
nasal, and not with Anuswara. If it be objected that this is
of no importance, as dependent on the caprice of the editor
or copyist, we can adduce as a decisive proof of the just
views of the Indian Grammarians in this respect, that when
they range the declensions of words in the order of their
terminating letters, the Pronouns gga idam, and fay kim,
in which they consider the m as primitive, are treated when
the turn comes of the labial nasal m, and together with
WMA prasdm, “quiet,” from the root wR sam. (Laghu-
Kaunuudi, p. 46.)
10. The deadened nasal, which is expressed in the Lithuanian
by particular signs over the vowel which it follows, appears
to be identical with the Sanskrit Anuswara ; and we write it
in the same manner with 7. At the end of words it stands
for the remainder of an ancient m, in the accusative singular
for example; and the deadening of m before s into % presents
terminal, and in pronunciation retains their respective sounds, according

to the initial consonant of the following word. Again, with regard to its
relation to thesemi-vowels and sibilants, it may be regarded asappropriate
to them merely in as far as neither of the other nasals is so considered.
In this sense Anuswara may be termed a subsidiary or supplemental sound,
being prefixed with most propriety to those letters which, not being classed
under either of the five series of sounds, have no rightful claim to the
nasals severally comprehended within each respective series.— Editor.
CHARACTERS AND SOUNDS. il
a remarkable accordance with the Sanskrit rule of euphony

before mentioned. From laupsin-u, “1 praise,” therefore
comes Jaupsizsu, “I shall praise ;* as in Sanskrit ¢enfa
hansydmi, “ I shall kill,” from the root ea han. In the
Prakrit, not only the 4 m, but the 4 x, at the end of words,
has always fallen into Anuswara, without regard to the follow-
ing letters Thus we read in Chezy’s edition of the Sakun-
tala, p. 70, wad, which is certainly to be pronounced, not
bhaaram, but bhaavaz, for wrtq bhagavan; [G. Ed. p. 12.]
au kudhan, for ean kutham.*
11..The second of the signs before mentioned is named
Visarga, which signifies abandonment. It expresses a breath-
ing, which is never primitive, but only appears at the end
of words in the character of an euphonic alteration of
asandtr. These two letters (s, r) are very mutable
at the end of words, and are changed into Visarga before a
pause or the deadened letters of the guttural and labial
classes (§. 12.). We write this sign A to distinguish it from
the true z kh
12. The proper consonants are classed in the Sanskrit
alphabet according to the organs used in their pronunciation;
and form, in this division, five classes. A sixth is formed by
the semi-vowels, and a seventh by the sibilants and the
=~. In the first five ranks of these consonants the single
letters are so arranged, that the first are the surd or hard
consonants, the thin (tenues), and their aspirates; next, the
sonant or soft, the medials, and their aspirates, each class
being completed by its nasal. The nasals belong, like the
vowels and semi-vowels, to the sonants; the sibilants to the
surd or hard. Every thin and every medial letter has its cor-
responding aspirate. The aspirates are pronounced, like their
* No native scholar would read these as bhaavah or kudhan,as the

text affirms, but bha-avam, kudham, agreeably to the final 4 represented
by Anuswara.— Editor.
respective non-aspirates, with a clearly audible 4; thus, for

example, q ih, not like the English th; WH ph, not f or ;
and @ kh, not like the Greek y.* In an etymological point
of view it is important to observe that the aspirates of
different organs are easily exchanged with each other;
thus, ax bhar, we dhar, (4 bhri, ¥ dhyi, §. 1.) “to bear,” “to
hold,” are perhaps originally identical. yaa dhimacs,
[G. Ed.p.13.] “smoke,” is, in Latin, fumu-s. In Greek,
6avw, as well as déve, is related to za han, from wa dhan,
“to: kill.” The Gothic thliuhan is the German fliehen, Old
‘High German eliuhan.
13. The first class is that of the gutturals, and includes the
letters @ k, @ kh ™ gy, X gh En. The nasal of this class
is pronouneed like the German n before gutturals, as in the
words sinken, enge, so as to prepare for the following gut-
tural. In the middle of words it is only found before
gutturals ; and, at the end, supplies the place of & m when
the following word begins with a guttural.t We write it
without the distinctive sign, as its guttural nature is easily
recognised by the following consonant. The aspirates of
this class are not of frequent use, either at the beginning or
end of words. In some Greek words we find y in the place
of a kh: compare ovu€, ovuy-os, with nakha, “ anail ;” xovyn,
Kovyxos, with sankha, “shell;” yaive, xavea, with khan, “to
* The original here adds—‘‘ We designate the aspirate by a comma,

as t', db’? The use of such a mark is, however, unsightly, and appears
likely to cause occasional perplexity and doubt. It seems therefore pre-
ferable to adhere to the usual mode of expressing the aspirated letters,
as dh, bh, and the like. It is only necessary to remember that th and ph
are the letters ¢ and p with an aspiration, and not the ¢h and f of the
English alphabet.— Editor.
+ A careful examination will perhaps shew that the several nasals of
the Sanskrit alphabet are mere modifications of one sound, according to
the manner in which that is affected by a succeeding letter; and that the
modifications prevail equally in most languages, although it has not been
thought necessary to provide them with distinct symbols.— Editor.
dig.” As regards the sonant aspirates, the ¥ gh of gharma,

“heat” (in Greek @épuy), has passed into the aspiration of
another organ; @¥ laghu, “ light,” has laid aside the gut-
tural in the Latin levis, and, in virtue of the 7, changed the
u into. The guttural has kept its place in the German
leieht, the English light, and the Old High German Jihti.
14. The second class is that of the palatals ; and includes
the sounds ch and j, with their aspirates and nasal. We write
= ch, & chh, 4 j,* Wjh* An. This class is an offshoot
from the preceding, and to be considered as a softening of it.
It is only found before vowels and weak consonants (semi-
vowels and nasals); and before strong consonants, and at the
end of a word, generally retires into the class from which
it springs.. Thus, for example, the base [G. Ed. p. 14.]
ara vich, “speech,” “ voice” (cf. vox), makes, in the unin-
flected nominative, arq vdk ; in the instrumental and locative
plurals, afta vdg-bhis, arg vdkshu. In the cognate lan-
guages we have to look for, in the place of the letters of this
class, first, gutturals; next, labials, on account of their mutual
affinity ; thirdly, the sounds of ¢, as, according to pronun-
ciation, the first element of the palatals is a ¢ or d; fourthly,
sibilants, as being the last element in the letters of this class.
Compare watfa pachdmi, “I cook,” (inf. paktum, part. pass.
_ pakta), with coquo, xénw (néntw, nétrw, nécow); AT chatur,
“four,” nom. @atta chatwdras, with guatuor, téttapes, té0-
capes, Gothic fidedr, Lithuanian ketturi; waa panchan, “five”
(nom. accus. pancha), with quinque, névre, néure, Gothic fimf,
Lithuanian penki; waa rdjan, “king,” with rez, regis;
Wad rdjata, nom. rdjatam, “silver” (from rdj, “to shine”),
with argentum, &pyupos ; wa jdnu, “knee,” with genu, "yovu.
With regard to the aspirates of this class, the chh,as an initial
letter in some words, answers to sc, ox; fata chhind-
* The original has g and g;

J but the appropriate symbols in a
are 7 and its aspirate.
—
mas, “we cleave,” faafa_chhinadmi, “I cleave,’ answers

to the Latin scindo; arat chhdyd, “ shadow,” to the Greek
oxida. As the terminating letter of a root chk answers, in
we prachh,“ to ask,” to the Gothic A in frah, “I or he asked,”
and to the German and Latin g in frage, rogo, in case that
the latter, as I suspect, is a modification of progo. The nasal
of this class, for which we require no distinctive sign, as it
only precedes palatals, deviates but slightly from the sound
of the guttural , and is pronounced nearly, like 7).
15. The third class is called that of the linguals or cerebrals,
and embraces a peculiar kind of sounds of ¢, together with its
[G. Ed, p.15.] nasal; a kind not original, but which has
developed itself from the ordinary class of ¢ sounds. We dis-
tinguish them by a point under the letter, thus, @ ¢, % th,
ad, dh, O 2. In the Prakrit this class has obtained great
supremacy, and has frequently supplanted the ordinary ¢.
We there find, for example, wg b/ddu, for waa bhavatu, “let
it be;” and yea padhama, for Waa prathama, “ the first.”
With regard to the nasal, the substitution of q for 4 is
nearly universal. The Indian Grammarians approach the
Prakrit nearer than the Sanskrit, when at the beginning of
roots they use the same substitution. The practice, also,
which we have condemned (§. 9.), of using Anuswara for
am, at the end of words, is more Prakrit than Sanskrit,
At the beginning of words these letters are seldom found in
Sanskrit, but they are found as terminations to a certain
number of roots; for example, "2 at, “to go,” They are
pronounced by bending back the tongue against the roof of
the mouth, by which a hollow sound is expressed, as if from
the head.* The nasal of this class has sometimes overstepped
the limits of its usual laws: it is found before vowels, which
* Here, also, it may be doubted if similar modifications of the dental

sounds are not discoverable in languages which do not express them by
separate symbols. The ¢ of the Italian tutto is the Sanskrit @—Zaditor.
is not the case with the nasals of the preceding classes; yet
never at the beginning of words.
16. The fourth class embraces the dentals, or the sounds
which properly answer to the common d and ¢, together
with the common #, which belongs to them, #t @ th = d,
a dh, q n. Of the aspirates of this organ, we have to re-
mark, that q th, in an etymological respect, never—at least
in no instance of which we are aware—is represented in
Greek by 6, but always like the natural ¢, by r. On the other
hand, y dh does correspond to 6, which also sometimes re-
presents ¢ d. Thus the imperative ending fy dhi, in Greek
becomes 6; wy madhu, “honey,” “wine,” is wé6v; gurfa
dadhdmi, “1 place,” TéOnus; efeaz duhitar [G. Ed. p.16.]
(efeq duhitri, §.1.), “daughter,” @vyarnp; am dwdr, f. and
dwara, neut. (nom. dwéram), “door,” 6ipa; % déva, Lithuan.
diewas, “God,” Qeds. With regard to the hard aspirate, com-
pare the terminations te and tov with ¥ tha and a thas, the
former in the plural, the second in the dual of the present
and future; orjow with wranfa sthdsydmi, “I shall stand”;
éctéov with weg asthi, “bone”; in the Latin, rota with
tT ratha, “carriage”; and in the Gothic, the ending ¢, in
the second person singular of the preterite, with tha; for
example, vais-t, “thou knewest,” with aq vét-tha. From
the beginning of words in the Sanskrit this aspirate is nearly
excluded.
7. The interchange of d and / is well known. Upon it,
among other instances, is founded the relation of lacryma to
daxpv, Saxpuza. In Sanskrit, also, an apparently original
~d often corresponds to the / of cognate European lan-
guages; for example, ata dip, “to light,” étq dipa, “lamp,”
becomes Adumw, Aaynas; *#e déha, “body,” Gothic Jeik.
On this relation also rests, as I have shewn elsewhere, the
relation of our lf, Gothic Jif, in elf, zwilf, Gothic tralif,to
ear. dasan, 8éxa. As also the second consonant has under-
gone alteration,and has migrated from the gutturals into the
labials; and as, moreover, the number “ten,” taken alone, is,
in Gothic, tathun, in German zehn, its origin from lif was
deeply concealed; and even the Lithuanian /ika, which accom-
panies the simple numbers in their compounded forms from
eleven to twenty, remained long under my notice without
result. The fact, however, that one and the same word may,
in the course of time, assume various forms for various objects,
proved, as it is, by numberless examples, requires no further
[G. Ed. p.17.] support. With respect to the affinity of Aikos
in 7Aikos, &c., and of the Gothic Jeiks in hvéleiks, “like to
whom?” togg drisa, Prakrit fea disa, “like,” I refer the reader
to my Treatise on the Pronoun and its influence (Berlin, pub-
lished by Diimmler); and only remark, in addition, that by
this analogy of Alkos, leiks, I was first led to that of lif to déxa;
while the Lithuanian /éka had not yet attracted my observation.
18. The labial class comes next, namely, 4 p, % ph, = 8,
3 bh, ¥ m. The hard aspirate ph is among the rarer letters;
the most usual words in which it occurs are, W@ phala,
“fruit,” ta phéna, “foam,” and the forms which come
from the root Wa phull, “to burst, blow, bloom.” The
sonant aspirate 4 bh belongs, together with ¥ dh, to the most
frequent of the aspirates. In the Greek and Latin, ¢ and f
are the letters which most frequently correspond to this
% bh, especially at the beginning of words; for example,
4 bhri, “to bear,” fero, pépw; wW bhi, “to be,” fu-i, pi-o.
a bh is also often represented by 6 in Latin, especially in
the middle of words. The f of fero becomes 6 in certain
compounds which-rank as simple words with a derivable
suffix, as ber, brum, brium, in words like saluber, candelabrum,
manubrium. Thus the f of fu appears as 6 in the forms
amabam, amabo, which I have recognised as compounds, and
which will be hereafter explained. The dative and ablative
termination plural wa bhyas, becomes bus in Latin, The
nasal of this class, a m, is subject, at the end of a word, to
several alterations, and only remains fast. before a pause, a
vowel, or letters of its own class: it otherwise governs itself

according to the nature of the following letters, and may pass,
in this manner, into any of the four preceding nasals, and
weakens itself into the softened nasalsound _[G. Ed. p.18.]
of the proper Anuswara, if followed by a semi-vowel, a sibilant,
or @h. M has also a full right to the name of a mutable
nasal. It is, however, not beseeming, when, in editions of a
text otherwise conspicuous for accuracy, we find 3, though
protected in its original condition by a pause, or by the
following letters, written as Anuswara.
19. The semi-vowels follow next: ayitinagiar. We
distinguish y by the sound of our German j, or the English
y in the word year. As the Latin j in English has the sound
of a softened g, so in Prakrit q y often passes into 4 j,
and in Greek, upon this exchange of sound rests the relation
of Cevyvupt, Cuyos, &e. to the root TZ yuj, “to bind,” and that
of the verbs in af» to the Indian verbs in waifa aydmi ; for
C is ds, but the sound dsch is not to be looked for in the Greek.
The relation of the Persian ..,\4> javdn, “young, to the
Sanskrit Theme gaq yuvan, Lat. juvenis, belongs to this
place. By v we here designate the sound of the German w
and English v. After consonants, as TA twdm, “thee,”
this letter takes the pronunciation of the English w. The
occasional hardening of the v into a guttural deserves mention
here; thus, in Latin, vic-si (vizi), victum, spring from viv ; and
in facio I recognise the Sanskrit causal wraatfa bhdv-ayd-mi,
“I make to be,” from the root = bhi. The connection be-
tween fac-tus and fio is practically demonstrated. Refer back,
in the Old and Modern Greek, to the occasional hardening
of the Digamma into y (cf. C. G. Schmidt in the Berlin
Jahrbuch, 1831, p. 613.). The voice cannot dwell on 4 v or
a y; and these two letters are therefore, as in the Semitic
languages, excluded from the end of words: _[G. Ed. p.19.]
therefore the word fea div, “Heaven.” forms its nomina-
tive, which ought to be div (divs being forbidden, see §. 94.).
Cc
from #t dyé. Nominal bases in y do not exist. <r at the end

of a word is subject to many alterations, and is interchange-
able with q s. In places where the concluding s, by favour
of the following letter, is retained, t r becomes @ s; and, on
the other hand, remaius unaltered in places where 4 s be-
comes Tt 7, namely, before vowels and sunant consonants.
20. The semi-vowels, by reason of their tractable and fluent
nature, are easily interchanged. For instance, in the more
recent Sanskrit works e l often stands for Tr* We often,
also, find in the cognate “European languages lfor gv. On
this interchange is founded the relation of the Latin suffix lent
(e.g. opulens), and of the Gothic laud(a)-s}+ (see §. 116.).i
hvélauds, “ quantus,” svalauds, “tantus,” samalauds, “ Ste so
much,” to the Sanskrit at vant (in the strong case, §. 119.),
in words like waaa dhanavant, “endowed with wealth,”
ara tdvant, “so much,” grat ydvant, “how much.” On
the cliange between v and r is founded, as I believe, the re-
lation of the Old High German pir-u-més, “ we are” (sing. pim,
watfa bhav-d-mi), to wataa bhav-d-mas ; as also that of serir-
-u-més, “we shriek,” to wrqaTaa Srdv-ayd-mas, “ we make
to hear” (§.109.); as also that of triusu, “I fall,” from the
[G. Ed. p.20.] root ¢trus, to the Sanskrit oe, dhwans, “ to
fall ;’ + and of the Cretan tpé “thee” from 7Fé, to the Sanskrit
twa. “The semi-vowel / is also exchanged with the nasals;
thus, waa anya-s, “ the other,” becomes alius in Latin, and
* It is scarcely correct to say “often,” as the instances are rare: nor

are they restricted to recent works. Menu has aélika for asrtka,—Ed.
+ Grimm (iii. p. 46) assumes an adjective lauds, “ great;” which, as
far as the Gothic at least is concerned, might be dispensed with, as it is
of the greatest antiquity as a suffix, and does not appear alone as an
adjective, even in the oldest periods.
{ Dh, according to §. 16., = the Greek 9; and to the $, according
to §. 87., corresponds the old High German ¢. The wu of trus, from the
old a, may be produced by the influence of the r, or of the dropped
nasal,
waita anfara-s, “the other,” alter; az vad, “to speak,”

answers to the Gothie lath-én, “ called,” « invited,” ga-lathon,
“called together”: wr dhma, “to blow,” answers to flare,
(§. 109.) Compare, also, balbus with BapBaive.
21. The last class embraces the sibilants and h: ¥ 5 ¥ sh,
as,and¢h. The first sibilant is spoken with a slight aspi-
ration, and usually written by the English sh.* It belongs to
the palatal class, and thence supplies the place of the third or
proper @ s when a hard palatal | ch or & chh follows; for
instance, way ati rdmas charati, instead of THA Att rdmas
charati, “Ramas goes.” In its origin, ys appears to have
sprung from £; and in Greek and Latin we find « and ¢ regu-
larly corresponding to the Sanskrit y s. The Gothic substi-
tutes h in pursuance of the law of change of sound; but the
Lithuanian stands the nearest to the Sanskrit with reference
to this letter, and has in its stead a sibilant compound sz, pro-
nounced like sh. Compare decem, déxa, Gothic taihun, Lithuan.
dészimtis, with eq dasun (nom. ty dasa); canis, ciwv, Gothic
hunds, Lithuan. szue (gen. szuns), with "Wa Swan (nom. vat Sud,
gen. Tae sunas, kuvos), “dog ;” daxpu, lacrima, aszara, f. with
asru n. “tear ;” equus (=ecvus), Lith. aszwa f. “ mare,”
with #7 aswa (nom. "wa aswas), “horse ;” szaka f. with
grat sakha “bough.” The Lith. szwenta-s, “ holy,” answers
to the Zend spyyeds spénta (§. 50.). At the end of a word,
and in the middle before strong consonants, y $ is not al-
lowed, although admitted as an euphonic substitute for a con-
cluding @ s before an initial hard palatal. Otherwise §
usually falls back into the soundfrom which _(G. Ed. p. 21.]
it appears to have originated, namely, k. In some roots,
however, gt s passes into z ¢ ; for instance, gq dris, “ seeing,”
and fay vis, “a man of the third caste,” form, in the unin-
flected nominative, gq drii, faz vit. The second sibilant,
% sh, is pronounced like our sch, or sh in English, and
* More usually ¢; the sh is reserved for the eerebral sibilant.— Editor.

c2 ~
belongs to the lingual class. It often steps, according to

certain rules into the place of @ s; thus, for instance, after
@ k, a s never follows, but only ¥ sh; and the &, x, in Greek
and Latin, are regularly represented by @ ksh. Compare
afeya dukshina, with dex-ter, de&ios, Lithuanian désziné, “ the
right hand.” Of the vowels, i, u, and vi, short or long, are
averse from &s, to which a and 4 alone are inclined. After the
first-named vowels, @ s passes into ¥ sh; for instance, watfa
tandshi, instead of walfa tandsi (extendis). As an initial, 4 sh
is extremely rare: the Indian grammarians, however, write
the roots which, under certain circumstances, change as into
4 sh, from the first with a q sh. A word which really be-
gins with ¥ sh is wq shagh, “ six;” to which the Lith. szeszi, a
plural nominative, answers most nearly, while other cognate
languages indicate an original ordinary s. At the end of a
word, and in the middle before other strong consonants, such
as Zt, Z th, 4 sh is not permitted, but in most roots passes
into @ k, but with some into z ¢: the number six, mentioned
above, becomes, in the uninflected nominative, 42 shat.
22. The third sibilant is the ordinary s of all languages, but
which, at the end of Sanskrit words, holds a very insecure po-
sition, and by certain rules is subjected to transmutation into
T & a sh, tr, 3 ah or k Visarga (§. 11.), and uw; and only re-
mains unaltered before ¢ and th. We write, for example, way
atfa snus tarati, “the son passes over,” but atfa 44: tarati
[G. Ed. p.22.] sénuli, wart ACfa sinus charali (it), WaT
wafa sénur bhavati (est). This sensitiveness against a con-
cluding @ s can only have arisen in the later period of the
language, after its division; as in the cognate languages the
concluding s remains unaltered, or where it has been changed
for r does not return into its original form. Thus, in the
decree against Timotheus (Maittaire, §. 383-4.) p everywhere
stands for ¢: Tipdceop 6 MiAyorop—naparyivdpevop—Avupaive-
Ta Tap akoap TOV véwy, &e.* The Sanskrit could not endure
——
* Cf, Hartung, p. 106.

r before ¢. The Latin protects the s usually at the end of

words; but in the classical period generally sacrifices it, when
between two vowels, to the r; for instance, genus, generis, for
genesis; a contrast to forms found in Varro and Festus, such
as plusima, fuedesum, meliosem, majosibus, in which the s
evinces its original existence in the history of the language
(see §. 127.). The accusative form arbosem, recorded by
Festus, is more startling, for here 7 is the original form, if,
as I can hardly doubt, arbor, arbus, is related to the word of
such frequent occurrence in the Zend-Avesta, »s2s»7> urvara,
“tree.” This expression is not wanting in the Sanskrit,
(sat urvard;) but it signifies, according to Wilson, “ fruitful
land,” and “land” in general.
23. h belongs to the le.ters which, in Sanskrit, are never
admitted at the end of words, nor in the middle before strong
consonants. In these places it passes, by certain rules, into
z4¢d,a@k,ormg. In Greek we often find y in the place
of the Sanskrit = 4: compare xetmav, hiems, with fea hima,
“snow,” “rime ;” yaipw with eurfa hrish- [G. Ed. p. 23.)
ydmi, gaudeo; ynv with €a hansa, “ goose ;* y6és, heri, with
wa hyas, “ yesterday ;" oyos with vah, “ to transport.”
We rae ‘nd K, ¢, tieh: isos anole cor, Gothic hoses
with ¢¢ hrid (n. ¢eu hridaya), “heart.” We sometimes, but
rarely, find the spiritus asper substituted for h; for instance,
aipéw, gufa hardmi, “I take away.” The Lithuanian ex-
hibits sometimes sz for h; for instance, asz, “ I,” for weR
aham, szirdis f. “heart,” for ¢¢ hrid. This letter stands
sometimes in Sanskrit for a mutilation of other aspirated
consonants, of which the aspiration alone has been sup-
pressed ; thus, instead of the imperative ending fa dhi, we
generally find hi; on which account the grammarians accept
fz hi, and not fy dhi, as the original ending, and assume that
hi passes into dhi, for euphonic reasons, after consonants.
The root We grah, “to take,” is written in the Vedas 7
grabh, and answers thus more nearly to the German greifen,
and the Persian yiriftan.
We give here a general view of the Sanserit characters,

with their respective values.
VOWELS.
wa wd, £4 $4 Fu Be Bri, Hz
ANUSWARA AND VISARGA,
* Mn, : ah.
CONSONANTS.
Gutturals......... Hh, @kh, 1g, Bgh, Fn.
Palatals .........- Gch, Bchh, Aj, Mh, An.
Linguals .........24 tth, 3d, @dh, Un.
Dentals' si 2. Sy . at at, ed, adh, an.
Labials 3% eee ie tes x es Up, wa ph, q b, a bh, am.
Semi-Vowels....... ay, tr, GL av.

Sibilants and Aspirates, ys, Ush, Ws, Eh
[G. Ed. p. 24.] The vowel characters given above are
found only at the beginning of words; and in the middle or
end of a word are supplied in the following manner: ¥ a is
left unexpressed, but is contained in every consonant which
is not distinguished by a sign of rest (,) or connected with
another vowel. @k is thus read ka; and k by itself, or the
absence of the a, is expressed by @. i, & 7, are expressed
by f, %, and the first of these two is placed before, the second
after, the consonant to which it relates; for instance, fa ki,
athki. For tu, %d,% ri,
¥]77,the signs , a, ez, are placed
under their consonants ; as, & hu, & ki, @ hei, Fihyi. For
zw éand @ di, ~‘ and ~ are placed over their consonants; as,
a hé, Hkdi: Wd and W du are written by omission of the ,
which is here only a fulcrum; as, #1 ié, @t kdu. The con-
sonants without vowels, instead of appearing in their entire
shapes, and with the sign of rest, are usually written so that
their distinctive sign is connected with the following conso-
nant; for instance, for 4 4, a, we have 3 %, %; and thus
matsya is written Ap, not AAA 5 for H + 3 we have Ri;
and for @ + W we have Lf
25. The Sanskrit letters are divided into hard or surd, and’
soft orsonant. Surd are, all the tenues, with their correspond-
ing aspirates ; and in fact, according to the order given above,
the first two letters in each of the first five rows, also the three
sibilants. Soft are, the medials, with their aspirates, the @,
the nasals, semi-vowels, and all vowels. Another division also
appears to us convenient—that of the consonants into strong
and weak; in which the nasals and semi-vowels come under
the denomination of weak; the remaining consonants under
that of the strong. The weak consonants and vowels exercise
no influence, as initial letters of inflections and suffixes, in
the formation of words, on the terminating (GG. Ed. p. 25.]
letters of a root; while they themselves are compelled to
accommodate themselves to a following strong consonant.
26. With regard to the vowels, it is of consequence to
direct the observation to two affections of them, of frequent
occurrence in the development of forms of Sanskrit; of which
the one is called Guna, or virtue; the other Vriddhi, increase
0 augmentation. My predecessors in grammatical inquiry
have given no information as to the essence, but have only
expounded the effects of these vowel alterations; and it was
only in my critical labours upon Grimm’s German Grammar*
that I came upon the trace of the true nature and distinctive
qualities of these affections, as also of the law by which Guna
is usually produced and governed, and at the same time of its
hitherto undetected existence in the Greek and Germanic,
and, most conspicuously, in the Gothic. My views in this
particular have since derived remarkable confirmation from
the Zend, with relation to which I refer to §. 2., in which, as
I flatter myself, I have dealt successfully with an apparent
contradiction to my explanation. Guna consists in prefixing
short a, and Vriddhi in prefixing a long one: in both, how-
ever, the a melts into a diphthong with the primitive vowel.
* Berlin Journal, Feb. 1827, p. 254.

according to certain euphonic laws. ¥ i, namely, and %, melt

with the wa of Guna into e@é; Su, Kd, into BH 6. These
diphthongs, however, dissolve again before vowels into wa ay
and ¥Wq av; ¥Y vi and 77 become, in virtue of the action of
Guna, BT ar; by that of Vriddhi, sit dr. As in Greek the
[G. Ed. p.26.] short Sanskrit a is frequently replaced by
€; so we find the Guna here, when a radical « or v is prolonged
by prefixing ane. As in the Sanskrit the root ¥i, “to go,”
forms, by the Guna modification, wfa émi (from a-imi), “I
go,” in contrast to imas, “we go;” thus in Greek also we
have ety in contrast to muev. As the root yy budh, in several
tenses in the three numbers, rises, in virtue of Guna, into
ara bédh (from baudh), for instance, #ratfa bédhdmi, “1 know;”
so in the Greek* the root guy (€pvyov), in the present be-
comes gev'yw. In the Gothic, in the strong form of Grimm’s
8th and 9th conjugations, the radical vowel, strengthened by
a in the singular of the preterite, stands in the same con-
trast to the i and u of the plural, as is the case in the corre-
sponding tense of the Sanskrit. Compare baug, “I bent,” in
contrast to bugum, “ we bent,” with the Sanskrit form of the
same signification, singular gitt bubhdja, plural qaifaa
bubhujima, of the root 4 bhuj; compare vait, “I know,” in
contrast with vi/um, “ we know,” with the Sanskrit forms of
the same signification, 4 véda (from vaida), fafea vidima,
from the root faz vid, “ know,” which, like the correspond-
ing Gothic and Greek root, employs the terminations of the
preterite with a present signification.
27. We have, however, the Sanskrit Guna in yet another
form in the Gothic—a form which I have but lately dis-
covered, but of which the historical connection with the
Sanskrit modification appears to me not the less certain. I
once thought that I had accounted in a different manner for
the relation existing between biuga, “I bend,” and its root
* Regarding Greek o: as Guna of 1, see §, 491.; and as to Guna in Old

Sclavonic and Lithuanian, see §§. 255.) 9, 741., 746.
bug, and I conceived myself bound to ascribe generally, in the

present tense, to the prevalent 7 of terminations a retro-active
influence. It now, however, seems to me indisputable that
Grimm’s 8th and 9th conjugations of the ([G. Ed. p.27.]
first class correspond to my first Sanskrit conjugation (r. 326.);
so that the Guna a of the special tenses has been weakened
to i, while the monosyllabic preterite maintains the Guna
vowel in the more important shape of a; just as in the 10th,
11th, and 12th conjugations, according to Grimm’s division,
the radical a, which has remained in the preterite singular,
is, in the present and other tenses, weakened toi; so that, for
instance, af, “1” and “he eat,” corresponds to the root WE
ad, “to eat;” but in the present, ita stands in place of the
form "fa admi, “I eat.”* ‘
28. The Zend possesses, besides the Sanskrit Guna, which
has remained everywhere where it stands in Sanskrit, a
vowel application peculiar to itself, which likewise consists
in 4s a, and which was first observed by M. E. Burnouf.t
The vowels which admit this addition in the interior, but
not at the end of words, are, first, the short s i, > u, L Oo;
2dly, the Guna diphthongs » é and 44. The two latter
are the most usually befriended by this addition, and » é
takes it in all cases where the opportunity occurs, both as an
initial letter, and even at the end of words wherever the
dependent particle as cha, “ and,” is appended to it; hence,
for example, po2sasy nairé, “homini,” po7Gaws dthre, “igni”; but
asqsroastasy naraécha, “hominique,” assras7Gus dthraécha, “ igni-
que.” Also where an é stands in two consecutive syllables, an
a is placed before each. Hence, for instance, Uysssro.sgor025
aétnéibyé, from wre élébhyas. The only case in which, ex-
* It would be difficult to adduce a better instance of the phonetic defi-

ciencies of our English alphabet than this sentence, in which I am forced
to translate the present and past tenses of essen by the samecharacters.
What foreign student could guess or remember that the one is pronounced
eet, the other ett? The preterite “ate” is obsolete.— Translator.
+ N. Journ. Asiat., T. III. p.327.
cepting at the actual end of the word, » é remains without

the preceding 1s a, is when it is produced by the influeuce of
a yo y, out of wa or wd. We say, indeed, Ladssroupy
[G. Ed. p.28.] yaéibyd, “ quibus,” from ae yébhyas; but
Not W979 Ayaésé, but posyJ. dyésé, “T glorify,” from
the Sanskrit root, which has been lost, for the verb ay yas,
from which comes aya yasas, “glory.” Yet we find, for
sro C_ yézi, “if” (cf. afe yadi), sometimes, though perhaps
erroneously, also sas,c_ yaézi. The addition of the sa
before 4 6 is just as unlimited, but the occasion is far less
frequent. Examples of it are, bchas aéz6, “ strength,” from
waa das; robase%es kérénadt, “he made,” from a kri, ac-
cording to the fifth class, for waatq akrinét ; rob mradt
“he spoke,” from sata abrét, which would be the regular
form, instead of waatt abravit (Gramm. Crit. r. 352.).. We
also find ¢4x% mradn, “I spoke,” for wat abrdm, which
would be the form used were, in the Sanskrit adjunct
tenses, as in the Greek, a mere nasal, and not Wa am, the
suffix of the first person. The vowels s i and > uw are
much more sparing in their attraction of the » a now in
question: they refuse it always at the beginning of words,
and in the middle before two consonants; and if transferred
from the end of a word to its middle, by an adventitious ter-
mination or word, they do not acquire the capacity of being
wedded to an asa. We say, for example, ¢¢65 imém,
“this” (accus.), not §¢Gsas aimém; 306.55 mithwana,
“a pair,” not AJWG 55G maithwana ; 3955752590 guiriby6,
“montibus,” not L935s057sa59 gairaibyé. The > w also, ac-
cording to set rules, very frequently abstains from the »s a;
for instance, by, urund, (anime,) not byras uraund, from
js»? urvan; on the contrary, »7,.@ fauruna, “ young,”
from weu taruna. Where, however, the Sanskrit ¥ u is
replaced by & o (§. 32.), an » a is placed before it, as well
at the beginning as before two consonants; and in this case
\ o stands in this respect in the same category as » é¢and
(G. Ed. p. 29.] 4 6. Compare pbs? raoch, “ light,” with
Baruch ; Eye waspbass Saochantanm (lucentium) with

Wants suchyatdm ; wpGbas aocta, “ he spoke,” with sa ukta,
which I form, by theory, after the analogy of wfeya akshipia
(Gram. Crit. r. 339.), leaving out the augment.
29. In the Vriddhi modification, the vowels = i, $7, melt
with the preceding
= 4 into
2 di; J u, & d@, into Wt du;
Bri, Yrt, into wx dr. The simple vowel w a, as also the
diphthongs z e and #0, which would produce the same
effect by Guna as by Vriddhi—for a+a, like 4+ a, makes d;
a+é like 4+6 makes di; a+4é, like 4+6, makes du—are
capable of only one higher modification, and reserve this one
for cases where grammatical Jaws demand the highest step,
namely, Vriddhi, and remain in the cases of Guna unaltered,
unless extraordinary grounds of exception occur. It may be
convenient here to give a connected summary of the results
produced by Guna and Vriddhi.
Primitive Vowels, 3a, m4,3i, $% Fu Bd, Bri
SS ae ee oes eee BG 4 SMG, MS, Arar,
Vriddhi. .... aid, ... Udi, Udi, Ddu, du, wee*
Primitive Vowels, & 772, eé 24, M6 Bau.
| Sia Wt ar, --- a ye,- ; ae
Vriddhi. Somnie.= BI dr, 2 ai, se. BN du,
30. We now proceed to the exposition of the Zend writing,

which, like the Semitic, proceeds from right to left, and
towards the comprehension of which Rask has contributed
valuable corrections, which give the language an appearance
more natural and more in consonance with the Sanskrit than
it assumed in the hands of former commentators, Anquetil’s
pronunciation having admitted much that was heterogeneous,
especiallyin the vowels. We follow the order of the Sanskrit
* According to original Grammars the Guna letters are a, ¢,0; the

Vriddhi, 4, ai, au; the two first, a and 4, being severally substituted for
the vowel sounds of ri, Zi, in combination with the semi-vowels r and /,
as ar, al, Gr, al.—Editor
eae
ee
eS
eee
alphabet in giving the corresponding value of each letter in

[G. Ed. p.30.] the Zend. The Sanskrit short Wa has two,
or rather three, representatives; the first is », which An-
quetil pronounces as a or e, but Rask, certainly with truth,
limits to a. The second is ¢g, which Rask pronounces like
the short @ of the Danish, or like the short German @, as in
Hinde, or as a in cane in English, and e in the French aprés.
I consider this ¢ as the shortest vowel, and write it e. We
often find it inserted between two consonants which form a
double consonant in the Sanskrit; for instance, assrg/rsgasg
dadarésa (pret. redupl.), for the Sanskrit zegt dadarsa, “he”
or “TI saw;” sws¢ £959 dadémahi (V. S. p. 102), “ we give,”
for the Véda form eafa dadmasi. This shortest @ is also
always appended to an originally terminating r. Thus, for in-
stance, EArsqogysas antaré, “ between,” E/aspoauss ddtaré, “ giver,”
“ creator,” eAasre hvaré, “sun,” stand for the corresponding
Sanskrit forms wet antar, eat datar, WT swar, “ heaven.”
It is worthy also of remark, that always before a final
§ m, and generally before a final yn, and frequently before
an intermediate vowelless ys n, the older ® a@ becomes ¢ é.
Compare, for instance, §¢7G>9 puthre-m, “ filium™ with Waa
putra-m ; yew anh-én, “they were,” with waa dsan, Hoa;
FEMywer hént-ém, ‘the existing one,” with waw sant-am,
pre-sentem, ab-sentem. This retro-active influence of the
nasal reminds us of the shortening power of the Latin ter-
mination m; as, for instance, stém, stémus (Sanskrit fava
tishthéy-am, fata tishthéma).
31. Anquetil entirely refuses to admit into his alphabet a
letter differing but little from the ¢ é above discussed, but
yet distinct from it by rule in practice, namely, ¢, which
Rask teaches us to pronounce like a long Danish e&. We find
this letter usually in connection with a following > u, and
this vowel appears to admit, with the excep- _—_[G. Ed. p. 31.]
tion of the long .w 4, no vowel but this ¢ before it. We write
this ¢ e without the diacritic sign, inasmuch as we represent
the 7», like the Sanskrit z, by é& Eu >¢ corresponds etymo-
logically to the Sanskrit #t 6, or diphthong formed by Wa

and gu; thus, for example, the nominal bases in u, which
in the Sanskrit genitive, by the influence of Guna, i.e. by the
prefixing of a short a, make é-s, form, in Zend, Ore Us.
Compare, for instance, »>¢wa5d paseus with uyita pasés,
from pasu, “pecus.” And yet the Sanskrit é does not uni-
versally become eu in Zend, but often remains as it is, and
specially in cases where it arises out of the termination as,
by the solution of the s into u. According to its pronuncia- _
tion, >¢ eu would appear to be a diphthong, and to form
but one syllable, as in our German words heute, Leute, &c.
The long a (4) is written ww.
32. Short and long i are represented, as are long and
short u, by special characters, si, 5% > u 9 #: Anquetil,
however, gives to the short i the pronunciation e, and to the
short u (>) that of o; while, according to Rask, only UL is
pronounced as short o.* This short o frequently holds the
etymological place of the Sanskrit 3 wu, and never corresponds
to any other Sanskrit vowel. For the diphthong ¥f du, in
particular, we have generally the Zend gus do: we yet find,
sometimes, also >.u du; for instance, «».we gdus, “ bos,” is
more frequent than »sguse gdos, for the Sanscrit ata gdus.
33. The Sanskrit diphthong é, formed out of a+, is re-
presented by », which, especially as a terminating letter, is
also written ee and which we, as in Sanskrit, represent by é.
We must here, however, observe, that the Sanskrit z é is not
always preserved as » é in the Zend, but is sometimes re-
placed by sh éi, which appears to prevail particularly after
a preceding 3} y, especially at the end of (G. Ed. p.32.]
words. The Vriddhi diphthong 2 di (out of 4+i) is always
represented by sav di; 6, either by the equivalent b—for
which we often find ba substituted by the neglect of copy-
ists—or by the above-mentioned >E eu, which, according to
rule, before a terminating 2 s replaces the Indian #t 6;
* But see §. 447. Note.

so that a termination in asl és* is unheard of in the Zend.

For the Vriddhi diphthong W du (out of 4+u) we gene-
rally fiud do, for which there is a special character EMS 5
more rarely >» du. It would appear that saw di, gus ad,
>as du, and the sls éi which replaces » é, should be pro-
nounced as diphthongs, i.e. as monosyllables,
34. Anuswara and Visarga do not exist in Zend, unless we
admit the nasal specified in § 61. as answering to the sound
of the Sanskrit Anuswara. We proceed meanwhile, for the
present, to the proper consonants. The first letter of the
Sanskrit guttural class has divided itself into two characters
bearing reference to different functions, 5 and @; of which
the first, which we represent by 4, only appears before vowels
and » v; the other, which we write c, precedes especially
consonants, excepting » v. Compare, for instance, by ké,
335 kd, goasg kat, (quis, que, quid), roclesasw» hakéret, “ once,”
sosbsy karéiti, “ he made,” 28»3 kva, “ where,” with at ké,
at kd, fora, kim, wea_sakrit, muta kardti, and Rkwa: on the
other hand, 76a csathra, “king,” with wa kshatra ;
seGsw hicti, “ pouring out” (V. S. p. 198), with fafa sikti
(from faq sich). In what manner the pronunciation of this
@ ¢ differs from that of the 9 & can indeed hardly be de-
fined with certainty: it is probably softer, weaker than that
of the 9k, which latter is fenced in by no strong consonants.
Rask selects for it the character q, without observing that this
letter prefers only to precede consonants, and in this position
[G. Ed. p.83.] always corresponds to the Sanskrit @ &.
Burnouf considers @ as an aspirate, and writes pwsG Gas
takhmahé. He writes, on the other hand, the letter yo, which
Rask treats as an aspirate, with g. Burnouf has not yet given
his reason, which I think, however, I can guess, namely, that
6 cis found before 7, which, according to Burnouf’s just
ad as és, according to Burnouf, occurs occasionally as the termination

of the oe singular of the u-bases for the more common 9>E eus ;
C.J» MONYASC AS bazaés, * brachii.”
CHARACTERS AND SOUNDS, 31
remark, generally confers an aspirate upon a preceding con-

sonant. Iconsider this reason, however, as insufficient; and
think that @ c stands before r, because, as we have before
remarked, all consonants, v excepted, only admit before them
that modification of the & sound which is expressed by &.
It would be impossible for 7 r, and the other letters of simi-
lar agency, to convey aspiration to the preceding hard gut-
tural if cf kh be not extant in Zend; so that, for instance,
the root @q khan, “ to dig,” sounds jas kan in Zend. There
are, however, some words in which @ kh is represented by
6. From atkhara, “ass,” we find the accusative Gores
carém; and we find, also, the @ kh of af@ sakhi, “ friend,”
replaced by c; the accusative, for instance, waraa sukhdyam
transformed into ¢sa0u@ase hacdim. It may therefore remain a
question whether 5 k or @ c, in respect of their sounds, have
the better right to be referred to @ kh; but this much is
certain, that @ k before vowels and before q v is only repre-
sented by 9 in Zend; before other consonants only by @;
which latter we shall, till better advised, continue to render
by c.
35. Anquetil ascribes to @ the value of yw, and to both
the pronunciation kh; while Rask considers the latter alone,
by reason of the aspiration stroke which he recognises, as
aspirated, and compares it to the Spanish z and the Arabic
a and our German ch. Burnoufrenders [G. Ed. p. 34.]
yw by g; and observes (1. c. p. 345) that the Sanskrit syllable
%@ swa becomes ga in Zend, namely, in 4A swapna, “ sleep,”
written, according to Burnouf, gafna, and in & swa (suus),
“his.” We are inclined to add to these examples, Ase WO
khanha, (nom.) accus. Gey khanhrém, from eat swasd,
“sister”(soror); ANA swasdram (sororem); and Lychasys kha-
réno, “ splendour,” as related to wt swar, “ heaven,” and gt
sur, “to shine.” We must, however, at the same time, remark,
that & sw does not universally become © kh, and that @ swa
in particular, in an isolated position and with a possessive
signification, much oftener appears in the shape of »s»w hva,
bP CHARACTERS AND SOUNDS.
or that of w»~v0w hava. We render wp by kh, aud support our

view of its aspiration more on the fact, that in modern Persian
it corresponds frequently to re our ch, than on the circum-
stance that Rask has marked it as aspirated. This modern
Persian ~ is pronounced, indeed, at present, without aspira-
tion, like an Italian c before a, 0, u; but its value in Arabic,
and the choice of this letter, so powerfully aspirated in the
Arabic to designate a specia] guttural sound, in true Persian
words, seems to indicate an intrinsic stronger or milder aspi-
ration. As yw kh is derived from the Sanskrit & swa, it was
not applied to replace the @ k before letters, which would
without it produce an aspiration. It may also be here conve-
nient to remember that either u or v (4) accompanies the
Persian ~ when the latter replaces at the beginning of a
[G. Ed. p.35.] word the Sanskrit @ sw. It is true that , v
is no longer sounded before long vowels, but it must originally
have had its influence on the pronunciation, and cannot have
been introduced into writing entirely without object, and for
the mere employment of the copyist. Compare \)> khudd,
“God,” with @za swadatta, “self-given ;” for which, in Zend,
we have, under a more regular participial form (see Gramm.
Crit. r. 608), ASPAUGASYD khaddta*; which Anquetil, or his
Parsi teacher, always understands in the sense of, “given
through God,” deceived, probably, by the resemblance of
sound to \\>khudd; while Neriosengh properly translates it
by qurea swayandatta. The Persian \)> khudd is, however,
as Burnouf correctly assumes, actually related to the Zend
ASAUSAS khadata, so as to have its name based in the idea,
“created by itself,” while in its form it has been mutilated of
one syllable. In Sanskrit we find both ey swabhi, “ self-
existent,” and also the more common War swayambhii, as
appellations of Brahma and Vishnu. That, however, as has
often been maintained, our word “God” is really related to
* This word comes frum the root dhd, “to place,” not from dé, “ to
give,” see §. 637.
IQ khudé, and that its primal signification has thus been dis-
covered through the Zend, we are forced still to doubt. We
will here only call to mind that the Germanic forms, especially
in the older dialects, in general approximate much more to
the Sanskrit than to the modern Persian. ®@ sw, in par-
ticular, in the Gothic, either remains unaltered, or becomes
si (§. 20.). The pronominal syllable @ swa exhibits itself in
the Gothic as a pronominal adverb, sva (so) “ thus ;” and with
an instrumental form, své (wie) “how.” The neuter sub-
stantive svés (Theme srésa) means Eigenthum, “property,” as
in Sanskrit the neuter @ swa. I know of no certain form in
which a Germanic g or k corresponds to a Sanskrit & sw or a
Persian ¢ kh. To return, however,tothe ([G. Ed. p. 36.]
Persian khu=® sw: compare ,.,i8> hkhuftan, “ to sleep,”
with wa swap; W\5> kh(w)db, “sleep,” with era swdpa ;
ew! kh(w)dndan, “ to sing,” with wa swan, “ to sound ;”
»\> ki(w)thar, “sister,” with a swasri, Gothic svistar;
Mw > khur-shid, “sun,” Zend chan hvaré, with St swar,
“heaven.” In some words ¢ kh corresponds to a Sanskrit &
before r, in which position the Zend loves an aspiration; in
the modern Persian, however, a vowel intrudes between the
guttural and the r; thus, ...da<)> khirdm-idan, “ to proceed
with pomp,” corresponds to the Sanskrit #A_kram, “ to go,”
“to step ;” and ow }> khiridan, “to buy,” to the Sanskrit
equivalent root mt kri. The Persian ¢ kh answers to the
Sanskrit aspirated @ kh, in the word > khar, “ass”
(Sanskrit et khara).
_ 36. The guttural 7, and its aspirate q, are represented by
eg and 9 gh. The Sanskrit q gh has, however, sometimes
dismissed the aspiration in Zend; at least sGeye garéma,
“heat” (6épun and Wérme), answers to the Sanskrit 9
gharma: on the other hand, the 449 ghna in ssyos6ehb veré-
thraghna, “victorious,” corresponds to the Sanskrit g ghna at
- the end of compounds; for instance, in WR satru-ghna, “enemy
slayer.” The Zend asjoasloghh vérethraghna properly signi-
D
fies, like the word so often used in the same sense prs6e%ely
verethra-zan, “ killer of Vritra,” and proves a connection be-
tween the Zendish and Indian mythologies, which, however,
in consequence of the obscuration of meanings in Zend, and
the oblivion of the old Myths, now only exists in affinities of
speech. “Killer of Vritra” is one of the most usual titles of
honour of the prince of the lesser gods, or Indra, who, from
his slaughter of the demon Vritra, of the race of the Da-
[G. Ed. p.87.] nawas, bears this name.
We shall discuss the nasals apart in §. 60.
37. Of the Sanskrit palatals the Zend has only the tenuis;
namely qs ch (=~), and the media, namely yj (=): the
aspirates are wanting, which is not surprising, as they are of
rare occurrence in the Sanskrit. The following are exam-
ples: sposastsys charaiti, “he goes,” Sanskrit =f charati;
Uses Gasschathwaré, “four” (nom. plur. masc.) Sansk. SAITR
chatwadras, AAT chatwaré ; bubs adjé, “ strength,” Sansk,
WAT djas, Wrst dj6. It is, however, to be observed, that,
wiile the Sanskrit ch remains, by rule, unaltered in Zend, the
sonantj is often replaced by other letters; and first, by ¢ z;
for instance, spay zdta, “ born,” Sansk. ata jdta; secondly,
by e sh; for instance, >yeeb shénu, “knee,” Sansk. a1q jdnu.
38. The modification of the sounds of ¢, peculiar to the
Sanskrit, contained in the third row of consonants, is wanting
in the Zend. We pass, therefore, to the ordinary sounds of
that letter, the dentals. These are, et (%), G th (q), 9d
‘t) edh (4), together with a ¢ (re), peculiar to the Zend,
of which more hereafter. The ~@¢ is like the guttural which
we represent by k (g), in this respect, that its position is
almost limited to one preceding vowels. Before 7 + and
wf w, and sometimes before yy y, in order to gratify the
affection of the latter for an aspirate, the aspirated G th
steps in. Thus, for instance, Gyros thwanm signifies “thee,”
while the nominative is written ¢ ~ tém, and the genitive
asa tava; and the word Aseaw dar, “ fire,” nom. sepa
dtars, makes, after rejection of the a which preceded r, po7Saus

dthré, “igni,” noxI6xs dihrat, “ab igne,” &c. If, however,
the ¢ be protected by a preceding consonant, excepting 2,
the succeeding semi-vowel is thereby de- [G. Ed. p. 38.]
prived of its retro-active power. We find, for instance,
astossasly vastra, not 3766 vasthra, “ garment,” “vest ;”
but we have ws OyG manthra, “ speech,” not sowg mantra,
from the root ys§ man. At the end of a word, and, which
rarely occurs, before strong consonants, (§. 25.) at the begin-
ning also, and middle of a word, the Sanskrit ¢ (7) is re-
presented by a special letter, namely, by > which we, with
Burnouf, write ¢, but formerly wrote with a simple ¢ undotted
below, because no change is possible with @ or G. Rask
represents it by th, because he recognises the sign of aspira-
tion. I am unable, however, to assent to the universal
validity of this sign of Rask’s, and I incline to rejecting the
aspirate, as in Sanskrit, from the end of words. We should
also remember that the diphthong é is written »~ as well
as Ww; the last, which prevails at the end of words, with
a stroke similar to that which distinguishes our nS from ¢.
Before consonants, for instance,in the word dugr0as990
tkaéshé, the sounding of th would be more precarious than
that of ¢, in case this th did not somewhat partake of a sibi-
lant sound. I think, however, that mt has merely a
feebler pronunciation than ~ ¢, and is, so to say, the last
breathing of ¢; as, in Sanskrit, s and r, at the end of words,
are diluted to Visarga (S. 11.); and as @ #, in Prakrit, and
also in Greek, is, at the end of words, altogether suppressed.
39. 4 is the ordinary d @, and @- according to Rask’s
just remarg, its aspirate dh. This represents the Sanskrit
4 dh, for instance, in the imperative ending fy The
Zend, moreover, favours @ dh for_4 d in the middle of
words between two vowels. We find, for instance, sp.
dita, “ given,” but s§.%>234 dadhdmi, Sanskrit zetfa daddmi,
“I give”; and spwe sag mazda-dhéia, [G. Ed. p. 39. |
D2
“given by Ormusd,” “created”; serpy yédhi, “if,” San-

skrit afe yadi ; »e@wd pddha, “foot,” Sansk. ate pdda.
40. The labial class embraces the letters o p, \ fi Ss
and the nasal of this organ ¢ m, of which more hereafter.
®% p answers to the Sanskrit Up, and is transformed into
\ f by the retro-active aspirative power of a following 7 r,
as s, and yn; whence, for instance, the preposition 9 pra
(pro, mpé) becomes, in Zend, 193 fra; and the primitive
words das ap, “ water” (aqua, and perhaps agpdés ), d¢/eg
kerép, “ body,” form in the nominative, aasd.us dfs, weg
kéréfs; on the other hand, in the accusative, Ged dpém,
Gede%9 kérépém, or Gedhwe9 kehrpém. In regard to the
power which resides in n of aspirating a p, compare > das
t fnu, “burning,” from the root dxs~ tap, with the deriva-
tive from the same root spspyjwdaupaw didpayéiti, “he
shines” (See Vendidad Sade, p. 333), and the plural ass
csafna, “nights,” with the ablative singular 5 rsd 05.055
csapardt (Vendidad Sade, p. 330), in which, even in the root,
the interchange between n and r is observable, as the same
takes place in the Sanskrit between weq ahan and Bet
ahar, “day.” (Gramm. Crit. r. 228. annot.) Originally—
i.e. standing for itself, and not proceeding from the o p
by the influence described—S f is of very rare occurrence.
In some instances known to me it corresponds to the San-
skrit 4 bh, which, however, for the most part, in the Zend
has rejected the aspiration. In Anquetil’s Vocabulary we
find ndfo, “navel,” which in Sanskrit is written arfa ndbhi;
and in the fem. actus. plural, of frequent occurrence in the
Zend-Avesta, wsio_edow hufédhris, we recognise the San-
skrit grg subhadra “ very fortunate,” “very excellent,”
also a title of Vishnu.
41. We come now to the semi-vowels, and must, in order
to follow the order of the Sanskrit alphabet, discuss y in the
(G. Ed. p.40.] next place, by which we express the sound
of the German and Italian j, the English consonantal y. This
semi-vowel is written at the beginning of words by yy or

£, and in the middle by the duplication of the u 99, as in
the Old High German we find w expressed. This semi-vowel,
and the vowels which correspond to it, si and 3 7, introduce
into the preceding syllable an s%; an interesting pheno-
menon, first observed by Burnouf (1. c. pp. 340, 341),and which
in its principle is connected with the German vowel modifi-
cation (§. 73.). We are obliged to ascribe a similar influence
also to the diphthong » é where it stands at the end ofa
word. Frequent occasion for this presents itself in the dat.
sing. and the third pers. pres. of the middle verb. For in-
stance, rosasy nairé, ““ homini,” for whsy naré, is frequent;
but asi70asAas 4 naraécha, “ hominique,” is an exception. The
vowels after which, by the attractive power of the letters
mentioned, an + i is placed, are sa, wd, >u, 9 %, » 4, 4 é,
as to which we must also observe, that u, in the case of a
succeeding i, is lengthened. Examples are: »)j@3s§ mai-
dhya (axa madhya) “ middle”; ays nairya, “man” ;
Sosasmass bavaiti, “ he is”; spswe a6 dadhditi, “ he gives” ;
SpsHIIwd.woaw dipayéiti, “he shines” ; spsbyekes kerénditi,
“he makes”; 59.590 stiidhi, “‘ praise,” instead of s@>os9
studhi, from the root >~ss stu (aq); 3997590 tuirya, “ the
fourth,” from ‘at chatur, with the cha suppressed *;
259379 ww Ghuirya, an adjective, derived from 1s?>ws ahura.
With regard to the influence of y) y we must observe, that
it does not mix up an si with a vowel immediately pre-
ceding, but only with one separated from it by one conso-
nart; for if there be two, unless the first be wn the retro-
active power of y, i, or ¢, is neutralized; thus spews asti,
Not 0955 aisti, stands for “he is”; on the other hand we
have SPysa235 bavainti, Sansk. wafet bhavanti, “ they are.”
Several other consonants also resist simply —_[G. Ed. p. 41.]
this power of attraction ; thus we have SIwasy dakhyu, not
* Or more immediately from the Sanskrit ordinal way ¢uryya or qua

turiya, “ fourth.”—Editor. 2
>33w5a5g daikhyu, “land,” “ province”; and the é of the

personal terminations 5§ mi and sw hi, or sx shi, obtain
no influence over the preceding syllable. In the same man-
ner, in the first person plural, swas¢ mahi, not swss¢ maihi,
corresponds to the Veda termination af@ masi; and in the
genitive of the stems, or inflective bases, in a a, pours
a-hé, not powrsas aihé, stands for wet a-sya.
42. ys y sometimes also exerts that disturbing influence
on a following » a or ww d, which is equivalent to the in-
sertion of a vowel, or of %, and consequently effects their
transmutation into ~» é*; thus the bases of nouns in
* The expression of the text is ‘‘iufsert umlautenden Einfluss.” It is

hardly possible to render into English without circumlocution certain
terms which the philologers of Germany have invented and adopted to
express the various modifications of the Indo-Germanic vowel; such as,
Ablaut, Auflaut, Inlaut, Umlaut. Whether these terms have in them-
selves the virtue of suggesting to a Teutonic ear the particular modification
of the vowel to which they are respectively applied may be doubted; but
if to the student and the teacher they answer the purpose of a memoria
technica, their use is fully justified by the necessity of the case, and the
practice of a language which possesses a singular and inexhaustible power
of progress and adaptation to exigencies. In cur language, it seems to us
that the uncouthness of such compounds as Upsound, Offsound, and In-
sound, could hardly be compensated by any advantage to be derived from
their use ; and we therefore purpose, in the course of this work, where any
of these terms occur in the original, to retain them in their German shape.
Of these terms, Ablaut and Umlaut are those which chiefly, if not alone
are used by our author. Jniaut is, we believe, merely the Sanskrit Guna.
The meaning of the two former, and their distinction from each other,
may best be explained by the following extract from our author’s excel-
lent work the Vocalismus, p. 10.
‘I designate,” he says, “‘ by the term Abdlaut, a change of the root
vowel, which is distinguished from the Umilaut by the fact that it is not
produced by the influence of the vowel of the termination ;for Umlaut is
a mere affection, disturbance (T'riibung) of the primary sound, through
which that sound becomes more homogeneous with the vowel of the ter-
mination; while in the Ablaut, without any recognised external cause, it
makes. room for another, and, in general, totally different sound ; as in
Gothic, nima, ‘I take’; nam, ‘I took.’ I say, without any recognised ex-
ternal
ayy ya form, in the genitive, wwryy yé-hé, instead of

powsrn ya-lé; and, with the verb, the old Sanskrit q ya
or atyé of the fourth and tenth classes, in the present
singular becomes wyy yé. Compare s¢r))sdwow did-
pryémi, sxryssdwow didpayéhi, spsreyywdwosw Aldpayéiti, |
with the Sanskrit sravqatfa d/dpaydmi, arataafa dtdpayasi,
_ ‘SataTaafa didpayati. In the last syllable, 49) ya before § m,
according to rule, becomes 37; and after the same
analogy, ¢s» vam becomes ¢; wm. We find, therefore,
for instance, ¢3s9@ tdirim, “quartum, from S90 tili-
rya; and GpKs7S thrishiim, “tertiam partem,” & IxpP7Gasgs
chathrushiim, “quartam partem,’ from ass thrishva,
asrrear7Gasgs chathrushva. This appearance is to be thus
understood, that the antecedent semi-vowel, after the suppres-
sion of the a, passes into its corresponding vowel, which,
however, according to the rule of §. 64., must be a long one.
The 3 y*, after its influence has transformed » a into
» ¢, is often itself suppressed ; thus we find Grass970.5400)
fradaéiaém, “1 shewed,” from mega prddésayam, which
ternal cause ; because I think I can shew that the Ablaut aiso is produced
by the particular quality and condition of the termination. Whether,
however, we seek for the radical vowel in the present or the preteriie, the
change is equally one quite different from that of the Indian Guna or
Vriddhi, and in this respect, that it is a positive change ; while in Sanskrit
the root vowel is not in fact changed, but only receives an increment, and
that increment always one and the same, with which it diphthongizes it-
self,as in Greek,« and v with ¢, Necre, Gevyo. In respect of signification,
likewise, there is a difference between the Indian Guna and Vriddhi and
Germanic Ablaut, for the Ablaut has acquired for itselfa significatory
power for grammatical purposes, even if, as I conjecture, it did not origi-
nally possess such: the contrast between the present and the past seems
to rest upon it, and there are indications that the latter is expressed by this
change. In Sanskrit, Guna and Vriddhi present no indication of this sig-
nificatory power, but, merely in the character of diphthongizing modifica-
tions, accompany those inflections which do signify grammatical relations.”
Farther illustrations of these latter remarks are to be found in the
Note 4, which Professor Bopp has appended to the above passage of the
Vocalismus. — Trans.
* Cf. p. 963, Note.
according to the rule of the tenth class, would be formed

from fey dis. The genitive termination @& sya appears
everywhere reduced into pow’ hé. The semi-vowels yg
and » v are generally suppressed after preceding conso-
[G. Ed. p.42.] nants*; and thus, also, the imperative
ending @ swa gives up its w. ;
43. In Sanskrit, q y is sometimes, for euphony, inter-
posed between two vowels (Gram. Crit. rr. 271. 310. 311.);
but this does not uniformly occur. In Zend, the interposi-
tion of y between > u, 9 %, and a following » é seems to
amount to a law. Thus the Sanskrit qa bruvé, “I say”
(from mand z, Gram. Crit. r. 55.), becomes, in Zend, 109397
mriyé (§. 63.); and the neuter form # dwé, “ two,” after the
vocalization of the w into u, takes the form woss>4 duyé.
44. We have already remarked (§. 30.) with respect to
7 r, that at the end of a word an ¢é is always appended to
it; for instance, E7asyosg dataré, “Creator,” “Giver”;
chs»w hvaré, “Sun,” instead of Assy ditar ; As»w dwar.
In the middle of a word, where an w h is not introduced.
according to §. 48., the union of 7 r with a following con-
sonant is mostly avoided ; so, indeed, that to the originally
vowelless 7 an é is appended: thence, for instance, ssehsgasg
dadarésa, from #e§ dadarsa, “vidi,” “ vidit”; or the r is
transposed, in the same manner as is usual in the Sanskrit for
the avoidance of the union of t r with two following con-
sonants. (Gram. Crit. r. 34°.) Hence, for instance, »»17Gas
dthrava, “priests” (nominative), accus. GEyasnas7 Gay dthra-
vaném, from the theme prsmAsqoass dtarvan, which in the weak
cases (§. 129.) contracts itself into pra dthurun or pPrasGass
dthaurun. (§. 28.) To this, also, pertains the fact that poly-
syllabic stems (or uninflected bases) in As ar, at the be-
ginning of compounded forms, transpose this syllable into
as? ra; and thus wsJ6us dthra, “ fire,” stands instead of
* Bu’ see § 721,

CHARACTERS AND SOUNDS. 4}
2G dthar.* The combinations 537 ry, (G. Ed. p.43.]

»/> urv, are only permitted where a vowel follows, and the
combination 9s ars only as a termination, and in the middle
of a word before ~ ¢; for instance, ays 90 tiirya, “the
fourth”; asyy2saslo vairya, “strong”; poms urvan, “soul”
as»ase haurva, “whole” (?); s/o. dtars, “ fire” (nomina-
tive) ; sohs nars, “of a man”; ase.379 karsta, “ploughed” ;
but 415>7Gases chathrus, “four times,” for a3) Gases chathurs,
since here no a precedes the rs. ;
45. It is worthy of remark, that in the Zend the / is want-
ing, as in Chinese the r, while, nevertheless, it exists in the
modern Persian, and shews itself in words which are not of
Semitic origin. The Sanskrit | v has three representatives
in the Zend, b, », and w. The two first are so far distin-
guished from each otherin their use, that 6 corresponds to the
Sanskrit v only at the beginning, and » only in the middle
of words; for instance, gl vuém, “ we,” = Fay vayam,
as»aso tava (tui)= awa tava. This distinction, as Rask justly
assumes, is only graphic. os; which I, with Burnouf, ren-
der by w, most frequently occurs after G th, so that » never
accompanies an antecedent Gth. On the other hand we find »
much oftener than of after the aspirated medials of this class.
Perhaps the law here obtains that the @_dh, which, accord-
ing to §. 39., stands for 4d (z) sis only followed by », while
an original @ dh, corresponding to a Sanskrit | dh, only
appears in conjunction with w. Thus gu»easg dadhvdo,
“having created,” “given,” from the root 14 dé, answers
to the Sanskrit nom. eatq dadwan; while the accusative,
of frequent occurrence in the Vendidad, ¢¢ywot
ey adhwaném,
seems to be identical with the Sanskrit wearaa adhwinam,
“viam.” (Vend. Olsh. p. 18.) After other consonants than
_* By Staimme, the author liere evidently means the crude derivative

words which serve as Stems or Bases to inflected words, or those in com-
bination with inflectional termmations; thus éthra for dthar, forms
athrava, dthravaném, not dtharva, dtharvanam, &c.—Editor.
t The root corresponds to the Sanskrit dhd, see §. 637.
42 CHARACTERS AND SOUNDS
G th and @. dh, & w appears not to be admitted, but only

» v; on the other hand, oS w much prevails between two i's
or 5 @ and ys y, in which position » v is not allowed.
[G. Ed. p.44.] Thus we read in the Vendidad (Qlsh. p. 23),
the nominatives 53%, driwis, “ beggar,” (?) and assossasg
daiwis, “a worshipper of Daéva.” 29505559 daiwis however,
as derived from daéva through the suffix s i, seems to me
dubicus, and I prefer the variation 209570259 daévis. Or is it
between é and i also that of w only can be allowed? Another
instance is, bsdoosas aiwyéd, “ aguis,” as dative and ablative
plural; an interesting form which long remained a mystery
to me, but which I am now in condition to explain. It springs
from the root os ap, “ water” in such a manner, that after
suppression of the p,* the Sanskrit termination wa bhyas,
which elsewhere, in the Zend, appears only as Loss by,
has weakened itself to Lysov wyé, and, according to §. 41. has
introduced an s i into the base. Another instance in which
4% bh has weakened itself in the Zend into a semi-vowel,
and obtained the form of w in virtue of its position between
two 5 i's, is the very common preposition sows, aiwi, for
which, however, 5355 aibi is sometimes substituted. It may
be appropriate here to remark that 4 bh appears in the
Zend, in other company, in the enfeebled shape of » »,
We find, namely, the base a4 ubha, “ both,” not only in the
shape as> wba, but also in that of asmbas aova (§. 28.), the
neuter dual form of which I think I recognise in the Vend. S.
p. 88., where poyyedss pomVEGs Yyy3925,C. pordas aové yasnd
améshé spénté, can hardly signify any thing else than “ambos
+ venerans Amschaspantos” (non conniventes Sanctos, see Nalus,
vv. 25, 26.) Anquetil interprets (T. 3, p. 472.) ové, by “tous
deux.” We have still another position to mention, in which
[G. Ed. p.45.] the semi-vowel of w appears, namely,
before 7 r, in which connection the softer w is more appro-
* Compare, in this respect, way abhra, “ cloud,” for way ab-bhra,

““water-bearing,’’and the Zend aspg7e:
su d-bértta, nom. “‘ water-bearer.’’
+ Burnouf readsadi (i.e, “over” )and makesyasne, signify “‘reverence.”?
priate than the harder » v. The only example of this case

is the feminine .w/o3.9 Suwrd, “ sword,” “dagger,” in which
we believe we recognise the Sanskrit yp subhra, “shining,”*
As to the pronunciation of the of w, I think, with Burnouf,
that it accords with the English w, which also is akin to the
Sanskrit ¥ v after consonants. Rask reverses the powers,
pronouncing the Zend of as the English v, and the letters
b and » as the English w.
46. I have not detected in the v and w a power of at-
traction similar to that which belongs to the yy y, as de-
scribed in §. 41., unless the term 1s»/as haurva, “all,” which
often occurs, as well as asduvs vispa, is derived from the
Sanskrit #4 sarwa, “all.” I have, however, already else-
where ascribed to the corresponding vowel > u a power of
attraction, howbeit sparingly exerted; in virtue of which,
for instance, the base pasmAasoas dtarvan, “ priests,” in the
weak cases (see §. 129.), after that poh wan has contracted
itself into y> un, by the influence of this u, also converts the
a of the preceding syllable into u; hence, for instance, in
the dative, wpa dtauruné for wpa dtaruné. The
Sanskrit wea taruna, “young,” is, in Zend, aspreo turuna
or 25)>/>0¢ tauruna (§. 28.); and 4a vasu, “thing,” “riches,”
[G. Ed. p.46.] has, by the influence of the concluding u,
converted itself into »whb véhu.
47. Burnouf was the first to remark on the fact, pecu-
liar to the Zend, that the semi-vowels are fond of commnu-
nicating an aspiration to a preceding consonant; and we
(§. 40.) have ascribed a similar influence to » sand sm
and find ourselves compelled to assign the same also to the
* The accusative gress suwranm, appears in Olshausen, p. 13, with

._ the variation egrhs sufranm. (§. 40.) Thez ‘ve often find the instru-
mental AIS Se 39suwrya, for which, however, we must read BCRswrsss
suwraya, if Suwrya be not derivable from a Theme shBss suwri, after
the analogy of arent sundari, from Wet sundara, (Gramm, Crit. r. 270.)
labial nasal, by which, for instance, the feminine participle

waat jagmushi has changed itself to SRO IGQNY, jaghmisht.
The dental medial is free from this influence, for we find
as»g dva, “two,” 1963>7g drucs, “a demon,” (accus. ¢¢y>74
drujém,) not 3@>79_dhrucs, §¢a270_ dhrujém. The guttural
medial is, however, exposed to this influence, as in the
abovementioned instance of jaghmdshi. We have, on the
other hand, adduced, in §. 38., a limitation of this appearance.
The aspirating virtue of the yy y is less potent than that of
the 7 r and oS w, and we find y often preceded by the un-
aspirated ¢; for instance, in asysess bifya, “the second,”
IS S057C thritya, “the third”: on the other hand, we have
r99Seheg meréthyu, “ death,” Sansk. aq mrityu.
48. In connection with the above rule stands the pheno-
menon, that before r, when followed by any consonant not
a sibilant, an h is usually placed; for instance 2597.96
mahrka, “death,” from the root 2s¢ mar (q mri,) “to die’ ;
GedhvE9 kehrpem, or Gergles kérépem, “the body ™ (nom.
awd¢%eg kérefs) ; asyJoreh velrka, or asgeFeh véréka, “ wolf,”
(qa vrika.) The semi-vowel y also, which only appears be-
fore vowels, sometimes attracts an w h; thus, MI JWAWS.
thwahya, “through thee,” corresponds to the Sanskrit mat
twauyd; and the word asyywas0 csahya (nom. Ly serasases
[G. Ed. p.47.] csahyé adduced by Rask, stands for MY JAS
csaya and comes from the root 543 csi, “to rule,” (fey Ashi.)
49. We come now to the sibilants. The first, a palatal,
pronounced in Sanskrit with a gentle aspiration, y, which
we express by § in Sanskrit, and ¢ in Zend, is written » in
the latter. Its exact pronunciation is scarcely ascertain-
able. Anquetil assigns it that of the ordinary s. It in
general ovcurs in those positions in which the Sanskrit in
corresponding words has its J $; thus, for instance, dasa,
‘ten,” sata, “ hundred,” pasu, “beast,” are common to both
languages. In this respect s9 s has spread itself wider in
Zend than in Sauskrit; that before several consonants,
.
namely, @ ¢, a k, and ym as well at the beginning as in

the middle of words—in the latter place, however, only
after a a, a d, and » an—it corresponds to the Sanskrit
dental or ordinary sa. Compare Lusoss staré, “the stars,”
with wie sféras; sebawoss stdémi, “I praise,” with efa
stiumi; sess asti, “he is,” with wf asti; FyVpwas
astanm, “ossium,” with wfeq asthi; 24239 shanda,
“ shoulder,” (?) with err skandha; x snd, “ to purify,”
with qq snd, “to bathe.” We might infer from this cir-
cumstance that s » was pronounced as a simple s, yet it
may have to do with a dialectical preference for the sound
sh, as happens with the German s in the Suabian dialect,
and pretty universally at the beginning of words before ¢
and p. It is further to be remarked, that $ » occurs also
at the end of words after y av. The occasion for this pre-
sents itself in the nom. sing. masc. of bases in @ys nt.
_ 50. The semi-vowel » v is regularly hardened into d p
after s9 $; hence, for instance, ws spd, “ canis.” G¢ywdss
Spdném “ canem," asds9sb vispa, “all,” [G.Ed.p. 48.]
4303945 aspa, “ horse,” corresponding to the Sanskrit var swd.
WIAA Swanam, fara viswa, WA aswa. MP yEds9 spenta, “holy,”
is not corresponded to by a Sanskrit wet “wenta, which must
have originally been in use, and which the Lithuanian
szanta-s indicates. From the Zend 13d..s aspa, the trans-
ition is easy to the Greek frzos, which is less obvious in the
case of the Indian aswa.
51. For the Sanskrit lingual sibilant q sh, the Zend
supplies two letters, «=» and K- The first, according to
Rask, is pronounced like the ordinary s, and therefore like the
Sanskrit dental s @; while yy has the sound of ¥=sh,
and marks this by a stroke ofaspiration. We therefore write
itsh.* Rask observes that these two letters are often inter-
changed in MSS.; which he accounts for by the circumstance
_ * It isin this Translation given sh without any mark. Sh denotes the
Sansk.¥.
that «9 is used in the Pehlevi for sh, and that the Pars?
copyists have been long better acquainted with the Pehlevi
than the Zend. We find, also, in the Codex edited by Burnouf,
ay almost everywhere corresponding to ¥ sh. We recognise,
however, from the text edited by Olshausen of a part of the
Vendidad, and the variations appended, that although in ety-
mological respects «3 as well as yyy corresponds to the San-
skrit 4 sh, the principal position of » is before strong con-
sonants (§. 25.) and at the end of words; a position of much
importance in the Zend, and which requires attention in the
cases of other classes of letters. In this respect « re-
senibles, among the dentals, me £ among the gutturals @ ¢,
and among the nasals principally ys n. At the end of
words, indeed, »» s corresponds to the Sanskrit 4 s, but yet
[G. Ed. p.49.] only.after such letters as, in the middle of
a word, would, according to Rule 101(*) of my Sanskrit Gram-
mar, change an original 4 s into 4 sh; namely, after vowels
other than a and 4, and after the consonants & ¢ and 7 r.
Hence, for instance, the nominative 9.30550 paitis, “Lord,”
23>980 pasus, “ beast,” awseous dtars, “fire,” 965>7g drucs,
“dzmon,” from the theme w74 druj. On the other hand,
ws baruns, “bearing,” from Pyyrs barant.* In the
word was» csvas, “ six,” it is true a terminating ww s
stands after a; but it does not here replace a Sanskrit qs,
but the original q sh of wq shash. As evidence of the use
of »» s for | sh before strong consonants, we may adduce
the very usual superlative suffix wows ista (i.e. (0T0S),
corresponding to the Sanskrit ¥¥ ishtha. Other examples
are sspastsg karsta, “ploughed,” for gw krishta. In the
word asjsyjxss sayana “camp,” »» stands irregularly for
» §, which latter was to be anticipated from the San-
skrit yaa sayana (cf. saété, §. 54.) In the fem. numeral
* T retain here the original ¢, since the theme of the word does not
appear in use. o ¢ must otherwise have been changed for meZ.
Vrs.ussgo tisaré, “three” (Olsh. p. 26), the +s might seem

questionable, for the Sanskrit form is fawa tisras, and @ ac-.
cording to §. 53., becomes w h. The a, however, is here
in a position (after = i) in which the Sanskrit favours the
conversion of @ s into 4 sh; and on this rests the Zend
form biswse tisaré. That it does not, however, stand as
Vases tisharé, as we might expect from §. 52., is certainly
not to be ascribed to the original existence of » a, for
Vs.nrsee tisaré stands for Lasso tisré,
52. yy stands for the Sanskrit qsh be- [G. Ed. p..50]
fore vowels and the semi-vowels yj y and » v; compare
FyRpraspros aétaéshanm and »s»yyrupers aétuéshva, with
@auIE ééshdm, “ horum,” and zag étéshu, “in his’; SIMVAG
mashya, “man,” with #qA manushya. Yet yy sh does not
unite itself with an antecedent @ c; but for the Sanskrit
W ksh we find almost everywhere in Olshausen’s text, and
without variation, »0@ cs; hence, for instance, 2s7Grx.ws
csathra, “king,” Sanskrit wa kshatra, “a man of the war-
like or royal caste.” The word of frequent occurrence,
ag barre cshnaéma, and the third person connected with
it, spsrssugbasyry cshnaémayéiti, we must, on a double
ground, reject, and prefer the variation given at p. 33, since
40 s here is prolonged, as well by the preceding c as by
the following n. It is, however, worthy of remark, that the
Sanskrit q ksh in many Zend words abandons the guttural,
and appears as yy sh. For instance, efaa dakshina,
“ dexter,” becomes asysrasg dashina (Lithuan. dészin2, “the
right hand”), and wie akshi, “eye,” becomes syyas ashi,
which, however, seems only to occur at the end of possessive
compounds (Bahuvrihi).
53. w h is never, in etymological respects, the repre-
sentative of the Sanskrit ¢ h, but of the pure and dental
sibilant qs. Before vowels, semi-vowels, and m, in Zend,
this letter invariably becomes w, possibly because q sw
(S. 35.) takes the shape yo kh; while before n, and such con-
sonants as cannot unite with a preceding Ah, (§. 49.) it is to
be looked for in the shape of ws. The (G.Ed.p.81.]
roots which begin with * sp and eA sph have not yet been
detected by me in the Zend; but I am convinced that
bark)§spris, for instance, “ to touch,” could not begin other-
wise in Zend than with ds sp. Compare, for instance—
ZEND. SANSKRIT.
ssw hd, “they,” aos.
spdsw hapta, “seven,” an sapta.
ree/egn’ hakérét, “once,” aaa sakrit.
sevas ahi, “thou art,” wfa asi.
swGwrs ahmdi, “to this,” WR asmdi.
chase hvaré, “ sun,” Mat swar, “ heaven.”
as» hva, “his,” Se soswa.
The word 2s»¢sw hizva, “tongue,” from fagtjihwa, deserves
mention, because the sibilant quality of the q j is treated
as 4 s, and replaced by w h (§. 58.).
54. [ do not remember to have met with an instance of
the combination 2» hr; the Sanskrit word aga sahasra,
“thousand,” which might give occasion for it, has rejected
the sibilant in the last syllable, and taken the shape asTaugasey
hazanra. If, in the word .5549>¥ huska, “ dry,” Sansk.
Wea sushka, w replaces the Sansk. a s, we must remember
that the Latin siccus indicates a Sansk. @ s, because ¢ regu-
larly answers to q $ In many instances of Sanskrit roots
beginning with @ s, the corresponding Zend form may be
grounded on the change which is effected on an initial as by
the influence of certain prepositions. (Gram. Crit. r. 80.)
[G. Ed. p.52.] Thus I believe I have clearly ascertained
the existence of the Sanskrit participle fag siddha, “ per-
fected,” in the term of frequent occurrence in.the Vendidad
Gepvsawwyny shdistém; after the analogy of »s957s irista,
“deceased,” from 647s irith (see §. 99.) Olshausen notifies
(p. 29) as variations of Geeywswyy shdistéem— Fe~s5.ws9
sdistem, Gspxvswyy shdistim, Fsevswxy shdistim, and
Fereuswry shdistem. In all these forms, the long a pre-
sents a difficulty; for, according to §. 28., faw shidh would
give the form @ssyxy shaidh; and this, with the suffix ta,
CHARACTERS AND SOUNDS 49
aspxsuyy shaista, in the nom. and accus, neut. Fersswry

shdistém. What Anquetil (vol. If. p. 279) translates, Juste
juge du monde qui existe par votre puissance, vous qui étes la
pureté méme, quelle est la premiere chose qui plaise @ cette terre
(que nous habhitons), et la rende favorable, runs in the criginal
(Olsh., p. 29, Burnouf, p. 137), -sas» oss FyYysGrse esos
eG FHS yt) begs Gus w Gs2sbasd SF. HAUHHS Fyryse
Détaré gaéthananm astvaitinanm ashdum! kva paoirim an-
hdo zemé shéistem? “Creator mundorum existentium, pure!
ubi (quid) primum hujus terre perfectum (bonum?”™)
55. The nominative pronominal base & sya (Gramm.
Crit. r. 268), in the Veda dialect, is under the influence of
the preceding word; and we see in Rosen’s specimen, p. 6,
this pronoun, when it follows the particle = u, converted
into @ shya, after the analogy of rule 101° of my Grammar.
I have detected a similar phenomenon in the Zend pronouns ;
for we find xw hé, “jus,” “ei,” which is founded on 2
lost Sanskrit @ sé (cf. % mé, “ mei,” “ mihi,” and @ #4, “ tui,’
“tibi”), when it follows sg» yézi, “if,” taking the form
yeas sé (more correctly, perhaps, my shé); for instance;
at p. 37 of Olshausen: while on the same page we find
PW 2345.56Hy yézicha hé, (und wenn ihm,) (G. Ea.p.53.]
“and if to him.” In the following page we find a similar
phenomenon, if, as I can hardly doubt, gusrsy shdo (thus I
read it with the variation), corresponds to the Sanskrit
wat asdu(“ille,” “illa”): sspghug wie gure goss 68 46 mesby
ages, Néit zi tm 2do shdo yd (text, gus, ydo) darégha
akarsta (text, assexy2s41s adarsta), “For not this earth which
_ lies long unploughed.”
56%). An wh standing between a or 4 and a following
vowel is usually preceded by a guttural nasal (> 2); and
this appendage seems indispensable—I remember, at least,
no exception—in cases where the following vowel is a, 4,
ore. We find, for instance, ISP BU SSASE SI usazayanha,
“thou wast born”; while in the active the personal ending
sw hi of the present admits no nasal; and we find, for
E
SC
hl
—
instance, swas ahi, “thou art,” swaswGasy bacsahi, “ thou

givest,” not swv anhi, sw guess bacsanhi.
566). The termination as, which in Sanskrit only before sonant
consonants (§. 25.) and wa, dissolves its qinto 3 and contracts
the latter together with the preceding a into wt 6 (compare the
French au, from al): this ancient termination as appears in
Zend, as also in Prakrit and Pali, always under the shape of 6
On the other hand, the termination ds, which in Sanskrit
before all sonant letters entirely abandons the s, in Zend
has never allowed the concluding sibilant entirely to expire,
but everywhere preserves its fusion in the shape of bo (for
[G. Ed. p.54.]_ —u);_ and I consider myself thereby strongly
supported in a conjecture I enounced before my acquaintance
with Zend,* that in Sanskrit the suppression of a terminating
s after @ had preceded the vocalization of this sintouw. It
is remarkable that where, in Zend, as above observed, an
2 2 precedes the w h which springs out of the s of the
syllable 4s, or where, before the enclitic particle x3 cha,
the s above mentioned is changed into w s, together with
these substantial representatives of the s, its evaporation
into Go is also retained, and the sibilant thus appears in
a double form, albeit torpid and evanescent. To illustrate
this by some examples, the Sanskrit ara mds, “ luna”—
an uninflected nominative, for the s belongs to the root—
receives in Zend the form gus¢ mdo, in which o represents
the Sanskrit s; amg més-cha, “ lunaque,” gives US asssgusg
mdoscha, and aTaR mdsam, “ lunam,” Ggw;—5G mdonhém ; so
that in the two last examples the Sanskrit sibilant is repre-
sented by a vowel and a consonant. The analogy of mdonhém,
“lunam,” is followed in all similar instances; for example,
for ara dsa “ fuit,” we find asw3gus donha, and for wreaa
dsdm, “ earum,” Gyasw37Q5 donhaimt.
* Observations, rule 78 of the Latim edition of Sanskrit Grammar.

+ Burnouf is of a different opinion as to the matter in question, for in
the
CHARACTERS AND SOUNDS. 5t
57. Two sibilants remain to be mentioned, namely, ¢ and

eb, of which the former was probably pronounced like the
French z, and may therefore be replaced [G. Ed. p.55.]
by that letter. Etymologically this letter answers to the
Sanskrit ¢h for the most part, which never corresponds
to the Zend wh. Compare, for example,
SANSKRIT. ZEND.
wen cham, “ is G E625 azém.
zat hasta, “hand,” MOA Zaséa.
Wea sahasra, “ thousand,” .s73u¢a hazanra.
zfat hanti, “he strikes,” SPyysasy Zainti.
Feta vahati, “he carries,” spsasgs vazaiti.
“ bears,”
fe hi, “for,” Sg zi.
fat jihwé, “ tongue,” amg ser hizva, (8. 53.)
meq mahat, “ great,” deasg mazé (from mazas,
ace. Few zu¢s§ mazanhém.)
58. Sometimes ¢ z appears also in the place of the San-
skrit Hj so that the sibilant portion of this letter, pro-
nounced dsch, is alone represented, and the d sound sup-
pressed (see §. 53.). Thus <u» yaz, “to adore,” answers
to the Sanskrit 7H yqj ; aserbass zaésha, “ to please,” springs
from the Sanskrit root wa jush, “to please or gratify.”
Thirdly, the Zend z represents also the Sanskrit 7 g, which
is easily accounted for by the relationship between g and j.
The Indian gé, (accus. gém;) bos and terra, has, in Zend,
as also in Greek, clothed itself in two forms; the first
the Nouveau Journ. Asiatique, tom. iii. p. 342, speaking of the relation of
mdonko to mananhé, without noticing the analogies which occur in cases
of repetition, mdosh-cha, “lunaque,” urvéraosh-cha, “ arboresque,” he says,
** In mdenghé, there is perhaps this difference, that the ngh does not re-
place the Sanskrit s, for this letter has already become o in consequence
of a change of frequent occurrence which we have lately noticed.
E2
signification has maintained itself in Zend, but in Greek

has given way to the labial; and Bovs and avEuses gdos, Or
aaue gdus, correspond to the Sanskrit nom. aq gdus.
[G. Ed.p.56.] For the signification “earth” the Greek
has preserved the guttural, which in Zend is replaced by z.
The nom. gusg z40 supposes an Indian form 71a gds, for
mira gdus ; in the accusative, G ys zanm agrees, in respect
of inflection, as closely as possible with mTa gdm and yy.
59. e is of less frequent use, and was probably pro-
nounced like the French 7: we write it zh. It is observable,
that as the French j in many words corresponds to the Latin
semi-vowel 7, and derives from it its own developement, so
also sometimes, in Zend, & zh has arisen out of the San-
skrit q y. Thus, for instance, Jay yiyam, “ you,” (vos),
becomes ¢¢ebsp~y yuizhém. Sometimes, also, eo zh has
sprung from the sound of the English 7, and corresponds to
the Sanskrit 4 j, as in > yedo zhénu, Sanskrit aq janw, “ knee.”
Finally, it stands as a terminating letter in some prefixes, in
the place of the Sanskrit dental qs after i and u; thus,
SossAasseb5y nizhbaraiti, “he carries out”; GEG peo>g
duzh-tictém, “ ill spoken”: on the other hand, ¢¢pasgrg
dus-matém, “ill thought.”
60. We have still to elucidate the nasals, which we have
postponed till now, because for them a knowledge of the
system of the other sounds is indispensable. We must first
of all mention a difference from the Sanskrit, that in Zend
every organ has not its particular nasal; but that here, in
respect of n, two main distinctions are established, and that
these mainly depend on the circumstance whether 2 precedes
a vowel or a consonant. In this manner J and are so
contrasted, that the first finds its place chiefly before whole
and half vowels, and also at the end of words; the latter only
[G. Ed. p.57.] in the middle of strong consonants. We
find, for instance, SGrO93A57059 ys490» hankdrayémi, “T glorify”;
asso pancha, “ five” ; FEM PWAHIIHY IS bishyantém: on the
other hand, sy nd (nom.) “man” rosy néit, “ pot? ; seoss?ass

barayen, “they might bear”; .s5ps anya, “ the other.”
Concerning the difference between sand ys—a difference
not recognised in European alphabets—it is probable that
yy being always fenced in by strong consonants, must have
had a duller and more suppressed sound than the freer 4;
and by reason of this weak and undecided character of its
pronunciation, would appear to have applied itself more
easily to every organ of the following letter.
61. Still feebler and more undecided than yy, perhaps
an equivalent to the Indian Anuswara, we conjecture to
have been the nasal y, which is always involved with 4s a,
and which seems from its form to have been a fusion of
as and } We find this letter, which we write an, first,
before sibilants, before w A, like the Anuswara, and before
the Sepirates G th and §f; for instance, Hyg Jaw csayans,
“ regnans,” accus. FEM Wasysaswees csayantém ; rp wGssyswryss
zanhyamdna, a part of the middle future of the root sasg zan,
“to beget,” but, as it seems to me, with a passive signifi-
eation (“qui nascetur,” Vend. S. pp. 28 and 103.); Sg
maithra, “ speech,” from the root ys§ man ; >a, janfnu,
“mouth,” probably from the Sanskrit 7q jap, “to pray,”
§. 40., and with the nasal inserted. Secondly, before a
terminating § m and yn. We have here to observe that
the Sanskrit termination "mq dm is always changed to
Gy anm in Zend; for intance, § yes dadhanm, “I gave,”
Sanskrit weer adaddm ; Gye. pddhananm, “ pedum,”
Sans. wera pdddndm; and that the ter- [G. Ed. p. 58.]
mination of the third person plural, #¥ an, provided the a do
not pass into é, always appears as a double nasal sy azin.*
62. For the nasal, which, according to §. 56., is placed as
an euphonic addition before the wh, which springs from
a s, the Zend has two characters, 3 and 4, to both which
* The termination ann from dn belongs to the potential, precative, and

svbjunctive.
Anquetil assigns the sound ng.* We write them 2, in order

to avoid giving the appearance of a g preceded by a gut-
tural » to this guttural, which is only a nasal precursor of
the following wh. As to the difference in the use of these
two letters, > always follows a and do; 5, on the contrary,
comes after i and e, for which the occasion is rare. For
instance, in the relative plural nom. ww5ry9 yénhé, “qui,”
and in the fem. pron. genitives, as gusw5sas ainhdo, “ hujus,”
which often oceurs, but as often without s i,and with > ™
gw zu anhdo. What phonetic difference existed between >
and .6 we cannot venture to pronounce. Anyuetil as we
have seen, assigns the same pronunciation to each; while Rask
compares 4 with the Sanskrit palatal 3 n, and illustrates
its sound by that of the Spanish and Portuguese nh.
63. The labial nasal § m does not differ from the San-
skrit q: it must, however, be remarked, that it sometimes
takes the place of b. At least the root @ bré, “speak,” in
Zend becomes 36 mri; as G76 mraém, “I spoke,” ropa?7
mraét, “he spoke”: in a similar manner is the Indian
aw mukha, “mouth,” related to the Latin bucea ; and not
[G. Ed. p.59.] much otherwise the Latin mare to the
Sanskrit atfz vdri, “ water.” I consider, also, multus re-
lated to @g@ bahula, the Greek modus, and the Gothic filu.
64. A concluding § m operates in a double manner on
a preceding vowel. It weakens (see §. 30.) the sa to ¢@;
and, on the other hand, lengthens the vowels i and wu;
thus, for instance, ¢ 49040 paitim, “the Lord,” ¢ss taniim,
“the body,” from the bases 49430 paiti, >yse tanu. In
contradiction to this rule we find the vocative of frequent
oveurrence, G>asyxa5 ashdum, “pure.” Here, however,
>a du, as a diphthong, answers to the Sanskrit ¥? du, the
last element of which is not capable of further lengthening
* Burnouf also writes the first of these ng. I have done the same in
my reviews in the- Journal of Lit. Crit.
The form in question is a contraction of the theme

jooorgss ashavan; with an irregular conversion of the
concluding jn into § m.
65, We give here a complete summary of the Zend
characters.
Simple Vowels: wa, ¢ 2 ge; wa; st 3%; ru boyd
Diphthongs: yo, » 4 sbdi; sw di; 6, gudo, rv du.
Gutturals: 5 k (before vowels and » v), & c (principally
before consonants), yo kh (from @ sw, before vowels
and 93¥); @9 9 gh.
Palatals: w-ch, » j.
Dentals: t (before vowels and 3) y) ~ ¢ (before con-
sonants and at the end of words), @ th (before whole
and semi-vowels), 4 d, @_dh.
Labials: © p, \f (the latter before vowels, semi-vowels,
nasals, and .s s),_5 b
Semi-vowels: C., », 339 y (the two 6G. Ed. p.60.]
first initial, the last medial), 7, \ r (the last only after
4A. G, » v (the first initial, the last medial), oSm.
Sibilants and h: s 3, yy sh, 0 8, & zh (or like the
French 7), ¢ 2, & h.
Nasals : yn (before vowels, semi-vowels, and at the end
of words), wy n (before strong consonants), x an (be-
fore sibilants, w h, ¢ th, Sf, ¢ m, and jn), 2 2 (between
asa or gus do, and wh, and between a and r*), Sa
(between si or » é,and wh), § m.
Remark also the Compounds ew for was ah, and cexy for
paw st.
66. We refrain from treating specially of the Greek,
Latin, and Lithuanian systems of sounds, but must here
devote a closer consideration to the Germanic. The Gothic
a, which, according to Grimm, is always short, answers
* E.g. aslgugase hazanra, “a thousand.”

completely to the Sanskrita; and the sounds of the Greek

e and o are wanting, in their character of degeneration
from a, in Gothic as well as in Sanskrit. The ancient
a has not, however, always been retained in Gothic; but
in radical syllables, as well as in terminations, has often
been weakened toi, or has undergone suppression; often,
also, by the influence of a following liquid, has been con-
verted into u. Compare, for instance, sibun, “seven,” with
anq saptan; taihun, “ten,” with aw dasan.
°
67. We believe ourselves authorized to lay down as a

law, that &a in polysyllabic words before a terminating s
is everywhere weakened into Zz, or suppressed; but before
a terminating th generally appears asi. A concluding ¥a
in the Gothic either remains unaltered, or disappears: it
never becomes 7.
68. In the Old High German the Gothic a either remains
[G. Ed. p.61.] unaltered, or is weakened to e, or is changed
by the influence of a liquid to u= perhaps 0. According to
this, the relation of the unorganic e to the Gothic a is the
same as that of the Gothic i (§. 66.) to "a; compare, for
instance, in the genitive of the bases in a yaa vrika-syd,
Gothic vulfi-s, Old High German wolfe-s. In the dative piural
wolfu-m stands to vulfa-m in the same relation as above (§. 66.),
sibun to saptan. The precedence of a liquid has also, in Old
High German, sometimes converted this a into u or 0; com-
pare plinte-mu(mo), caeco, with the Gothic blindamma. Also
after the German 7 or y, which in Sanskrit (a y) belongs as
a semi-vowel to the same class as 7, the Old High German
seems to prefer u to a; thence plintju, without 7 also plintu,
“ceca,” as a fem. nom. sing., and neuter nom. ace. voc.
plural ; plinta “ caecam.” The u of the first person present, as
kipu, “I give,” Gothic giba, 1 ascribe to the influence of
the dropped personal letter m. Respecting the degenera-
tion of the original a@ sound to wu compare also §. 66. In
the Old High German inseparable preposition ki (our
German ge) = Gothic ya, Sanskrit @ sa or @q sam, we
have an example in which the Gothic-Sanskrit @ has_be-

come i.
69, For the Sanskrit wid, the Gothic, which has no
long a, almost always substitutes 6 (§. 4.), and this 6, in cases
of abbreviation, falls back into the short a. Thus, for instance,
in Grimm’s first fem. declension of the strong form, the
nom. and accus. sing. d is softened to a, whence giba, gihé-s
(§. 118.). Generally in the Gothic polysyllabic forms, the
concluding "td is shortened to a; and where @ stands
at the termination, ar originally succeeding consonant has
been dropped ; for instance, in the gen. plur. fem. 6 stands
for am dm. Sometimes, also, in the Gothic, é@ corresponds
to the Sanskrit 4, as in the gen. plur. masc. and neuter. In
the Old High German the Gothic 6 either | [G. Ed. p. 62.]
remains 6, as in the gen. plur., or divides itself into two
short vowels; and, according to differences of origin, into oa,
ua, or uo; of which, in the Middle High German, wo prevails;
while in the Modern High German the two divided vowels
are contracted into % For the Gothic é=wid, the Old,
Middle, and Modern High German have preserved the old 4,
except in the gen. plural.
70. For $i and $i the Gothic has i and ei; which latter,
as Grimm has sufficiently shewn, is everywhere to be
considered as long i, and also in Old and Middle High
German is so represented. We, together with Grimm, as
in the case of the other vowels, designate its prolongation
by a circumflex. In the Modern High German the long i
appears mostly as ei; compare, for instance, mein with the
Gothie genitive meina, and the Old and Middle High Ger-
man min. Sometimes a short i is substituted, as in /ich,
answering to the Gothic leiks, “like,” at the end of com-
pounds. On the longi, in wir, “nos,” Gothic veis, we can
lay no stress, as we match the dat. sing. mir also with the
Gothic mis. It is scarcely worth remarking that we usually,
in writing, designate the elongation of the i and other vowels
by the addition of an A.
71. While the original @a has undergone many altera-

tions in the Germanic languages, and has produced both
z and u, I have been able to detect no other alterations in
i and 7? than that i is as often suppressed as a; but it never
happens, unless some rare exceptions have escaped me, that
i is replaced by a heavier vowel a or u.* We may lay
|G. Ed, p.63.] it down as a rule, that final 7 has given
way in German everywhere, as it has generally in Lativ-
Compare.
SANSKRIT. GREEK. LATIN. GOTHIC.
aft pari, TEP, per, fair. (8. 82.)
suft upari, bmép, super, ufar.
afer asti, éori, est, ist.
aint santi, évti, sunt, — sind.
72. Where a concluding i occurs in Gothic and Old High
German it is always a mutilation of the German j (or y) toge-
ther with the following vowel; so that j, after the suppres-
sion of this vowel, has vocalized itself. Thus the uninflected
Gothic accus. hari, “exercitum,” is a mutilation of harya.t The
Sanskrit would require harya-m; and the Zend, after §. 42.,
meeting the Germanic half way, hari-m. Before a con-
cluding s also, in the Gothic, gi is usually suppressed ; and
the Gothic terminating syllable is, is mostly a weakening
of as, §. 67. In Old High German, and still more in Middle
and Modern High German, the Gothic i has often degene-
rated into e, which, where it occurs in the accented syllable,
is expressed in Grimm by @. We retain this character. We
have also to observe of the Gothic, that, in the old text, i
* The Sanskrit faq pitri, “father,” probably stands for atq pdéri,
yuler” ; and the European languages have adhered to the true original.
(Gramm. Crit. r. 178, Annot.)
+ In the text harja; but in order to shew more exactly the connection
with the Sanscrit gy, vide §. 68. 1.12.; and as the 7 is simply and uni-
versally pronounced y, the German 7 will be represented by y in this
translation.
CHARACTERS AND SOUNDS. a9
at the beginning of a syllable is distinguished by two dots

above, which Grimm retains.
73. Asin Zend (§. 41.), by the attractive force of i, é, or
y, an é is introduced into the antecedent syllable; so also, in
Old High German, the corresponding sounds have obtained
an assimilating power; and frequently an [G. Ed. p. 64.]
a of the preceding syllable is converted into e, without any
power of prevention on the part of either a single or double
consonant. Thus, for instance, we find from asf, “branch,”
the plural esti; from anst, “grace,” the plural ensti; and from
vallu, “I fall,’ the second and third persons vellis, vellit.
This law, however, has not prevaded the Old High German
universally: we find, for instance, arpi, “hereditas,” not
erpi; zahari, “laeryme,” not zaheri.
74. In the Middle High German, the e, which springs
from the older i, has both retained and extended the power
of modification and assimilation; inasmuch as, with few
limitations, (Grimm, p. 332,) not only every a by its retro-
spective action becomes e, but generally, also, 4, u, and o
are modified into e, ii, and 6; 6 into @, and uo into ue.
Thus the plural geste, drete, briiche, kiche, lene, gruese,
from gast, drat, bruch, koch, lén, gruoz. On the other hand,
in the Old High German, the e which has degenerated from
i or a obtains no such power; and we find in the genitive
singular of the above words, gaste-s, drate-s, &c., because
the Old High German has already, in the declension of the
masculine i class, reduced to e the i belonging to the class,
and which in Gothic remains unaltered.
5. The e produced in Old and Middle High German
by the modification of a, is retained in the Modern High
German, in cases where the trace of the original vowel is
either extinguished or scarcely felt; as, Ende, Engel, setzen,
netzen, nennen, brennen; ‘Goth. andi, aggilus, satyan, natyan,
namnyan, brannyan. Where, however, the original vowel
is distinctly opposed to the change, we place 4, short or
[G. Ed. p.65.] long, from short or long a; and in the same
relation, z from u, 6 from o, du from au; for instance, Briinde,
Pfile, Dinste, Fliige, Koche, Tone, Baume, from Brand,
Pfil, &e.
76. For tu, %é, the Gothic has u, which is generally short.
Among the few examples cited by Grimm, p. 41, of long u,
we particularize the comparative sitizd, the essential part of
which corresponds to the Sansk. wg swddu, “sweet,” (70v- ¢),
and in which the long « may stand as a compensation for
the absence of the w(v), which becomes vocalized. In Old High
German it seems to me that péam, “to dwell,” and tréén, “to
trust,” correspond to the Sanskrit roots 4bhi, “tobe.” 3 dhrié
“to stand fast’—from which comes wa dhruva, “ fast,”
“constant,” “certain” (Gramm. Crit. r. 51.)—with the Guna
form of which (§. 26.) the Goth, bauan, trauan, is connected;
cf. ufaaa bhav-itum, “to be,” ufaqy dhrav-itum, “to stand
fast.” The Middle High German continues the Gothic Old
High German @, but the Modern High German substitutes
au, whence bauen, trauen, Taube (Gothic débé).
77. As out of the Sanskrit gu, in Zend, the sound of a
short U* has developed itself (§. 32.), thus, also, the Gothic
u shes itself, in the more recent dialects, oftener in the
form of o than in its own. Thus have the Verbs in the
Old and Middle High German (Grimm’s 9th conjug.) pre-
served a radical wu in the plur. of the pret., but replaced it
by o in the passive part. Compare, for instance, bugum,
“we bend,” dugans, “bent,” with Old High German pukumés,
pokanér, Middle High German bugen, bogen. The example
adduced shews, also, the softening of the old u to e, in un-
accented syllabes, in Middle High German as in Modern
High German; so that this unaccented e may represent all
original vowels—a, i, u; and we may lay it down as a rule,
that all long and short vowels in the last syllable of poly-
* Cf. §. 447. Note.

CHARACTERS AND SOUNDS. 6l
syllabic words, are either worn away or softened down toa

mute e.
78. For the diphthongs @4 (a+i) and —_[G. kd. p. 66.]
st 6 (a + u), the Gothic has ai and au, which are also
monosyllabic, and were perhaps pronounced like & é and 6.
Compare bavaima, “edificemus” with waa bhavéma, “ simus”;
sunau-s, “of a son,” with its equivalent qata sund-s. Where
these Gothic diphthongs aiand au have maintained themselves
unaltered in value, they then appear, in writing, as é and 6,*
which must be consideredas contractions of a + i and a+u;
as in the Latin amémus, from amaimus (§. 5.); and as in
the almost solitary case of bds, the long o of which is the
result of a contraction of a+u, whose latter element appears
again before vowels in the independent shape of »v (bovis,
bovem), while the first element Gd, in its degeneration,
appears as 6 (§. 3.). Compare,
SANSKRIT. GOTHIC. OLD HIGH GERMAN.
axa charéma (eamus), _faraima, varémés.
“ata charéta (eatis), faraith, varét.
were tébhyas (his), thaim dém.
79. In like manner, in all subjunctives, and in the pro-
nominal declension in which the adjective bases in a take
part, an Old High German é corresponds to the Sanskrit
zé and Gothic ai. The Middle High [G. Ed. p.67.]
erman has shortened this é, as standing in an unaccented
terminating syllable (varen, varet). Besides this, the Middle
High German has, in common with the Old High German,
If, however, the Gothic diphthongs in question were not pronounced

like their etymological equivalents e é and wy 6, but, as Grimm con-
ceives, approximate to the Vriddhi-change (§. 26.) 2 ai and Wf du: in
such case the High German é, 6, as opposed to the Gothic ai, au, are not
merely continuations of these Gothic diphthongs: but the pronunciation
assigned by the Sanskrit to the union of a with é or u, must have been
first introduced into the Germanic, under certain conditions, in the eighth
century.
preserved the diphthong é where it stood in radical syllables

under the protection of a following u, r (out of the older s),
or h (ch), even in cases where one of these letters had been
dropped, or where u had vocalized itself into v or o.
(Grimm. pp. 90. 343). Compare,
OLD MIDDLE
GOTHIC. HIGH GERMAN. HIGH GERMAN.
aiv, “ evum,” éwin.

snaivs, “nix,” snéo, sné.
mais, “magis,” mér, mé.
laisyan, “ docere,” léran, léren.
laihv, “ commodavit,” léh léch.
In the Modern High German this é is partly preserved,
partly replaced; for instance, mér (mehr), Schné (Schnee),
Séle (Gothic satvala); but ich lieh, gedich. (Grimm. p. 983.).
s0. As the é for the Gothic ai, so the 6 for au, in the
Old and Middle High German, is favoured by certain
consonants; and those which favour the 6 are the more
numerous. They consist of the dentals (according to the
Sanskrit division, §. 16.) t, d, z, together with their nasal
and sibilant (n, s); further, the semi-vowel r; and h, which,
as a termination in Middle High German, becomes ch (See
Grimm, pp. 94. 345) The roots, which in the Gothic
admit the Guna modification of the radical wu by a, in
the preterite singular, oppose to the Gothic au, in Middle
and Old High German, a double form; namely, 6 under the
condition above mentioned, and next ou, §. 34., in the absence
[G. Ed. p.68.] of the letter which protects 6. For instance,
Old High German zéh, Middle High German zéch (trazi,
traxit) Gothic tauh, Sanskrit géte dudéha (mulxi, mulsi,) ;
but pouc, bouc, fieri, flexit, Gothic baug, Sanskrit gira
bu-bhdja. The Modern High German exhibits the Gothic
diphthong au, either, like the Middle and Old High Ger-
man, as 6, and in a more extended degree, and subject
to the modification of §. 75; or next, shortened to o,
the particulars of which will be explained under the verb,

or, thirdly, as au; for instance, daupya, “I baptize,” hlaupa,
*T run”; or, fourthly, as eu, §. 83. :
81. As Ulfilas, in proper names, represents both e and az
by ai, and likewise o and av by au (Paitrus, Galeilaia,
apaustaulus, Paulus); and as, in the next place, not every
Gothic ai and au in the cognate dialects is represented in like
manner, but in some cases the Gothic ai is replaced in Old
High German by a simple i or @, and au by u or o (§. 77.);
but in the others, ai is replaced by 4 or (§. 85.) by ei, and
au by 6 or (§. 84.) ou; therefore Grimm deduces from these
facts a double value of the diphthongs ai and au; one
with the accent on the last element (ai, aé#), another with
the accent on the a (di, Gu). We cannot, however, give im-
plicit belief to this deduction of the acute author of the
German system of sounds, and prefer assuming an equal
value in all cases of the Gothic ai and au, although we
might support Grimm’s view by the fact, that, in Sanskrit,
z é, wt 4, never replace his af and aé; but everywhere,
where occasion occurs, do replace di and du. We think, how-
ever, that the difference is rather phonetic than etymological.
As concerns the ai and au in proper names, it may be ac-
counted for, inasmuch as the Gothic was [G. Ed. p. 69.]
deficient in equivalents for these non-primitive vowels, which
have degenerated from the original wa. Could Ulfilas
have looked back into the early ages of his language, and
have recognised the original idenity of ¢ and o with his a,
he would perhaps have used the latter as their substitutes.
From his point of sight, however, he embraced the az and
au, probably because these mixed diphthongs passed with
him as weaker than the long é and 6, ejusdem generis,=
(at4). It is important here to observe, that in Greek also
at is felt as weaker than y and «@, as is proved by the fact
that a: does not attract the accent towards itself (témropas
not tumréua:. The expression of the Greek a: and av by
the Gothic ai and au requires the less justification, because

even if ai was pronounced like z 4 and au like wt 6, yet
the written character presents these diphthongs as a still
perceptible fusion of a with a following 7 or u.
82. As to the other statement, namely, that not every
Gothic ai and au produces the same effect in the younger
dialects, nor has the same foundation in the older Sanskrit,
it might be sufficient to observe upon one feature of dialect
peculiar to the Gothic, that h and r do not content them-
selves with a pure preceding i, but require it to be affected
by Guna (§. 26.); thus, ai for i, and au for uw; while other
dialects exhibit the i and u before h and r in the same
form as before every other consonant. The relation of the
Gothic to their Sanskrit equivalents,
GOTHIC SANSKRIT.
saihs, “six,” wa shash,
taihun, “ ten,” aw dasan,
faihu, “ cattle,” Wy pasu,
svaihra “ father-in-law,” YC Swasura,
taihsvé, “ dextera,” afar dakshind,
io hairté, “heart,” ez hrid (from hard §. 1.),
& bairan, “to bear,” way bhartum,
ce distairan, “to tear,” afcaa dar-i-tum,
& stairné, “star,” att tard,
is not so to be understood as though an i had been placed
after the old a, but that, by the softening down of the a to
i (§. 66.), the forms sihs, tihun, had been produced; out of
which, afterwards, the Guna power arising from A and r
had produced saths, tathun, bairan. The High German has,
however, remained at the earlier stage; for Old High Ger-
man séhs, (Anglo-Saxon, “six,”) and ¢éhan or téhun, &c., rest
upon an earlier Gothic sihs, tihun. Thus, tohtar rests on an
earlier Gothic duhtar, for the Guna form dauhtar, Sanskrit
gfent duhitar, (afeq duhitri, §. 1.) “daughter.” Where the
Sanskrit wa has preserved itself in the Gothic unaltered, that

is, not weakened to i, the occasion is absent for the de-
velopment of the diphthong ai, since it is not the a before
h and r which demands a subsequent addition, but the #
which demands a precedent one; compare ahtau, “ eight,”
with wet ashidu.*
83. The alterations tc which the simple vowels have
been subjected appear again in the simple elements of the
diphthungs, as well in the relation of the Gothic to the
Sanskrit, as in that of the younger Germanic dialects to
the Gothic. Thus the a element of the diphthong Wt 6
shews itself often in the Gothic, and in certain places in a
regular manner, as i (§. 27.); and in the same places the
a contained in @ é (a +i) becomes i, which, with the second
element of the diphthong, generates a long i (written as e?,
§. 70.). The Gothic iu has either retained that form in Old
High German, or has altered sometimes one, sometimes both
of its constituents. Thus have arisenio,é@o. [G. Ed. p.71.]
Thereis a greater distance to be passed in Otfrid’s theory of
the substitution of ia for iu, which cannot fail to surprise, as
we know that a simple u never becoms a.t In Middle
High German iu has either remained unaltered, or has been
changed to ie, which is as old as the latest Old High Ger-
man, as it is found in Notker. In Modern High German
the substitution of ie for the old iu is that which princi-
* Ahtau—ashtéu is perhaps the only case in which the Gothic au cor-

responds to the Sanskrit Vriddhi diphthong @j du ; on the other hand,
au often answers to WY 6—(a+u).
+ There is yet another ia in Old High German, namely, that which
Grimm (p. 103) very acutely represents as the result of a contraction, and
formerly dissyllabic, to which, therefore, there is no counterpart diph-
thong in Gothic. The most important case will be discussed under the
head of the verb, in preterites, such as hialt, “I held,” Gothic haihald.
After this analogy fiar, “four,” (according to Otfrid), arose out of the
Gothic fidvor, in this way, that, after the extrusion of the de, the 6 passed
into its correspanding short vowel —Grimm, p. 198.
¥
pally prevails, in which, however, the e is only visibly re-

tained, for phonetically it is absorbed by the 7 Compare
ich biete with the Gothic biuda, giesse with giuta. Besides
this form, we also find ew in place of the old iu or still
older au, in cases, namely, where e can be accounted for as
the result of a no longer perceptible modification (Grimm,
p. 523, §. 75.); compare Leute with the Gothic /audeis, Old High
German liuti, “ people”; Heu, “hay,” with Goth, havi, “grass.”
Usually, however, the Gothic has already acquired an iu in
place of this eu, and the original au (which becomes av be-
fore vowels ) is to be sought in the Sanskrit; for instance,
Neune, “nine,’ Old High German niuni, Gothic niuneis,
Sanskrit qaa navan (as theme); neu, “new,” Old High Gers
man niwi (indeclinable), Gothic nivi-s, Sanskrit 77a nava-s.
This e, however, is difficult to account for, in as far as it is
connected with the Um/aut, because it corresponds to an 7 in
Middle and Old High German; and this vowel, of itself
answering to an i or y in the following syllable, is capable of
no alteration through their power of attraction. Long u for
iu, equivalent to a transposition of the diphthong, is found in
lugen, “to lie,” trigen, “to deceive,” Middle High German
liugen, triugen.
[G. Ed. p.72.] 84. Where the a element of the Sanskrit
it d retains its existence in the Gothic, making aw the equi-
valent of 6, the Middle High German, and a part of the Old
High German authorities, have ou in the place of au,
although, as has been remarked in §. 80, under the influence
of certain consonants 6 prevails. Compare Old High German
pouc, Middle High German bouc, with the Gothic preterite
baug, “flexi.” The o of the High German ou has the same
relation to the corresponding Gothic a in au, as the Greek
o in Bots bears to the Sanskrit 8a, which undergoes a
fusion with 3 u in the #té of the cognate word mt gé.
The oldest Old High German authorities (Gl. Hrab. Ker. Is.)
have au for the ou of the later (Grimm. p, 99); and as,
nnder the conditions specified in §. 80., they also exhibit 6,
this tells in favour of Grimm’s assumption, that au in the
Gothic and oldest High German was pronounced like our
German au, and thus not like the Sanskrit = 6 (out of
a+u). In this case,in the Gothic ai, also, both the let-
ters must have been sounded, and this diphthong must be
only an etymological, and not a phonetic equivalent of the
Sanskrit z é.
85. In the Gothic diphthong ai the a alone is susceptible
of alteration, and appears in High German softened down to
e, in the cases in which the é, contracted from ai (§. 78.), does
not occur. In Modern High German, however, ei, in pro-
nunciation, = ai. Compare
OLD MIDDLE MODERN
GOTHIC. HIGH GERMAN, HIGH GERMAN. HIGH GERMAN,
haita, “‘ voco,” heizu, heize, heisse.
skaida, “ separo,” skeidu, scheide, scheide.
86. (1.) Let us now consider the consonants, preserving the
Indian arrangement, and thus examining (G. Ed. p.73.]
the gutturals first. Of these, the Gothic has merely the
tenuis and the medial (k, g); and Ulfilas, in imitation of the
Greek, places the latter as a nasal before gutturals; for in-
stance, drigkan, “to drink”; briggan, “to bring”; tuggé,
“tongue”; yuggs, “ young” ;gaggs, “a going” (subst.). For
the compound fv the old writing has a special character,
which we, like Grimm, render by gv, although g does not
appear elsewhere, and v also combines with g; so that gv
_ (=4v) plainly bears
the same relation to gv that k bears tog;
compare siggvan, “to sink,” with siggvan, “to read,” “to
sing.” JH also, in Gothic, willingly combines with v; and
for this combination, also, the original text has a special
character; compare sathvan, leihvan, with our sehen, leihen.
In respect to h by itself we have to observe that it often
appears in relations in which the dentals place their th and
the labials their f, so that in this case it takes the place of
F2
kh, which is wanting in the Gothic. In this manner is

aih related to aigum, “we have,” as bauth to budum, and gaf
to gébum. Probably the pronunciation of the Gothic h was
not in all positions the same, but in terminations, and before
t and s, if not generally before consonants, corresponded to
our ch. The High German has ch as an aspirate of the k:
for this tenuis, however, either k or ¢ stands in the older
dialects, the use of which, in Middle High German, is so dis-
tinguished, that c stands as a terminating letter, and in the
middle of words before f, and ch also stands for a double k.
(Grimm, p. 422.) This distinction reminds us of the use of
‘the Zend @ c in contrast to 9 k, as also of the no ¢ in con-
trast to @ é. (§§. 34. 38.)
(2.) The palatals and linguals are wanting in Gothic, as
‘in Greek and Latin; the dentals are, in Gothic, £ th, d,
[G. Ed. p.74.] together with their nasal n. For th the
Gothic alphabet has a special character. In the High
German z (=ts) fills the place of the aspiration of the #, so
that the breathing is replaced by the sibilation. By the side
of this z in the Old High German, the old Gothic -th also
maintains its existence.* There are two species of z, which,
in Middle High German, do not agree with each other. In
in the other, s; and this latter
the one, ¢ has the preponderance,
is written by Isidor zs, and its reduplication zss, while the
reduplication of the former he writes tz. In the Modern
High German the second species has only retained the
sibilant, but in writing is distinguished, though not uni-
versally, from s proper. Etymologically, both species of
the Old and Middle High German z fall under the same
head, and correspond to the Gothic ¢.
(3.) The labials are, in Gothic, p, f, b, with their nasal
* Our Modern High German 7h is, according to Grimm (p. 525), in-
organic, and to be rejected. “It is, neither in pronunciation nor origin,
properly aspirated, and nothing but a mere tenuis,”
CHARACTERS AND SOUNDS 69
m. The High German supplies this organ, as the Sanskrit

does all, with a double aspiration, a surd (f= & p/)) (see
§. 25.) and a sonant, which is written v, and comes nearer to
the Sanskrit 3 bh. In Modern High German we perceive
no longer any phonetic difference between f and v; but in
Middle High German v shews itself in this manner softer
than f, in that, first, at the end of words it is transformed
into f, on the same principle by which, in such a position,
the medials are converted into tenues; for instance, wolf not
wolv, but genitive wolves; second, that in the middle before
surd consonants it becomes f, hence zwelve becomes zwelfte,
funve becomes funfte, funfzic. At the beginning of words f
and v, in Middle High German, seem of equal signification,
and their use in the MSS. is precarious, ([G. Ed. p. 75.]
but v preponderates (Grimm, pp. 339. 400). It is the same
in Old High German; yet Notker uses f as the original
primarily existing breathing-sound, and v as the softer or
sonant aspiration, and therefore employs the latter in cases
where the preceding word concludes with one of those letters,
which otherwise (§. 93.) soften down a tenuis to its medial
(Grimm, pp. 135, 136); for instance, demo vater, den vater,
but not des vater but des fater. So far the rule is less
stringent (observes Grimm:), that in all cases f may stand
for v, but the converse does not hold. Many Old High
German authorities abandon altogether the initiatory v,
and write f for it constantly, namely, Kero, Otfrid, Tatian.
The aspiration of the p is sometimes, m Old High German,
also rendered by ph, but,in general, only at the beginning of
words of foreign origin, phorta, phenning; in the middle,
and at the end occasionally, in true Germanic forms, such
as wérphan, warph, wurphumés, in Tatian; limphan in Oifrid
and Tatian. According to Grimm, ph, in many cases, has
had the mere sound of f. “In monumental inscriptions,
however, which usually employ f, the ph of many words
had indisputably the sound of pf; for example, if Otfrid
_-
writes kuphar, “ cuprum,” scepheri, “ Creator,” we are not to

assume that these words were pronounced kufar, sceferi”
(p. 132). In Middle High German the initial ph of foreign
words of the Old High German has become pf (Grimm,
p. 326). Inthe middle and at the end we find pf, first, always
after m, kampf, “pugna,” tampf, “ vapor,” krempfen, “contra-
here,” in which case p is an euphonic appendage to f, in order
to facilitate a union with m. Secondly, in compounds with the
inseparable prefix ent, which, before the labial aspirates, lays
aside its f, or, as seems to me the sounder supposition, converts
that letter, by assimilation, into the labial tenuis. Hence, for
[G. Ed. p.76.] instance, enp-finden, later and more harmo-
nious emp-finden, for ent-finden. Standing alone, neverthe-
less, it appears, in Middle High German, vinden, but v does not
combine with p, for after the surd p (§. 25.) the surd aspirate
is necessary (see Grimm, p.398). Thirdly, after short vowels
the labial aspirates are apt to be preceded by their tenues, as
well in the middle as at the end of words: just as in Sanskrit
(Gramm. Crit. r. 88.) the palatal surd aspirate between a short
and another vowel or semi-vowel is preceded by its tenuis;
and, for instance, y=afa prichchhati is said for yafa prichhati.
“interrogat,” from the root wa prachh. In this light I
view the Middle High German forms kopf, kropf, tropfe,
klopfen, kripfen, kapfen (Grimm, p, 398). In the same words
we sometimes find ff, as kaffen, schuffen. Here, also, p has
assimilated itself to the following f; for f, even though it be
the aspirate of p, is not pronounced like the Sanskrit % ph,
that is, like p with a clearly perceptible h; but the sounds
p and h are compounded into a third simple sound lying
between the two, which is therefore capable of reduplication,
as in Greek ¢ unites itself with 0, while ph +th would be im-
possible.
(4.) The Sanskrit semi-vowels are represented in Gothic
by j (=y), 7, 4, v; the same in High German; only in Old
High German Manuscripts the sound of the Indo-Gothie »
CHARACTERS AND SOUNDS. Ti
(our w) is most usually represented by wu, in Middle High

German by vv: j (or y) in both is writteni. We agree with
Grimm in using j (or ») and w forall periods of the High Ger-
man. After an initial consonant in Old High German, the
semi-vowel w in most authorities is expressed by u; for in-
stance, zuelif, “ twelve,” Gothic tvalif. As in the Sanskrit and
Zend the semi-vowels y and v often arise out of the cor-
responding vowels i and u, so also in the [G. Ed. p.77.]
Germanic; for instance, Gothic suniv-é, “filiorum,” from the
base sunu, with u affected by Guaa (iu, §. 27.). More usually,
however, in the Germanic, the converse occurs, namely,
that y and v, at terminations and before consonants, have
become vocalised (see §. 73.), and have only retained their
original form before terminations beginning with a vowel;
for if, for instance, thius, “servant,” forms thivis in the
genitive, we know, from the history of the word, that this
v has not sprung from the u of the nominative, but that
thius is a mutilation of thivas (§. 116.); so that after the
lapse of the a the preceding semi-vowel has become a whole
one. In like manner is thivi, “ maid-servant,” a mutila-
tion of the base thivyé (§. 120.), whose nominative, like the
accusative, probably ‘was thivya, for which, however, in the
accusative, after the v had become vocalized, thiuya was
substituted. :
(5.) Of the Sanskrit sibilants, the Germanic has only
the last, namely, the pure dental qs. Out of this, how-
ever, springs another, peculiar, at least in use, to the
Gothic, which is written z, and had probably a softer pro-
nunciation than s. This z is most usually found between two
vowels, as an euphonic alteration of s, but sometimes also
between a vowel and », /, or n; and between liquids (J, r, n)
and a vowel, y.or n, in some words also before d ; finally,
before the guttural medial, in the single instance, azgé,
“ashes”; everywhere thus before sonants, and it must
therefore itself be considered as a sonant sibilant (§. 25.), while
s is the surd. It is remarkable, in a grammatical point af

view, that a concluding s before the enclitic particles ei and
uh, and before the passive addition a, passes into z; hence, for
instance, thizei “cujus,” from this “hujus,” thanzei “ quos,”
from thans “hos,” vileizuh “visne” from vileis “ vis,” haitaza
© yocaris,’ from haitis “vocas,” or rather from its earlier form
[G. Ed. p.78.] haitas. The root slép, “to sleep,” forms,
by a reduplication, in the preterite, saizlép, “I or he slept.
Other examples are, izvis, “vobis,” “vos,” razn “house,” talzyan,
“to teach,” marzyan, “to provoke,” fairena, “heel.” The
High German loves the softening of s into 7, especially
between two vowels (see §. 22.); but this change has not
established itself as a pervading law, and does not extend
over all parts of the Grammar. For instance, in Old High
German, the final s of several roots has changed itself into
y before the preterite terminations which commence with a
vowel; on the other hand, it has remained unaltered in the
uninflected first and third pers. sing. indicative, and also
before the vowels of the present. For example, from the
root lus, comes liusu, “I lose,” Jés, “I or he lost,” lurumés
“we lost.” While in these cases the termination takes s
under its protection, yet the s of the nominative singular,
where it has not been altogether dropped, is everywhere
softened down to r; and, on the other hand, the concluding
s of the genitive has, down to our time, remained unaltered,
and thus an organic difference has arisen between two cases
originally distinguished by a similar suffix. For instance,
- OLD MODERN
GOTHIC. HIGH GERMAN, HIGH GERMAN.
Nominative . . blind’-s, plinté-r, blinde-r.

Genitive . . . blindi-s, plinte-s, blinde-s.
87. The Germanic tongues exhibit, in respect of con-
sonants, a remarkable law of displacement, which has been
first recognised and developed with great ability by Grimm,
According to this law, the Gothic, and the other dialects
with the exception of the High German, in relaticn to the

Greek, Latin, and, with certain limits,also [G. Ed. p. 79.]
to the Sanskrit and Zend, substitute aspirates for the original
tenues, h for k, th for ¢, and f for p; tenues for medials, ¢ for
d, p for b, and k for g; finally, medials for aspirates, g for y,
d for 6, and 6 for f. The High German bears the same
regular relation to the Gothic as the latter to the Greek, and
substitutes its aspirates for the Gothic tenues and Greek
medials; its tenues for the Gothic medials and Greek aspi-
rates; and its medials for the Gothic aspirates and Greek
tenues. Yet the Gothic labial and guttural medial exhibits
itself unaltered in mostof the Old High German authorities,
_ as in the Middle and Modern High German; for instance,
Gothic biuga, “flecto,” Old High German biuga and piuka,
Middle High German biuge, Modern High German biege.
For the Gothic f, the Old High German substitutes v, espe-
cially as a first letter (§. 86.3.). In the ¢ sounds, z in High
German (=?s) replaces an aspirate. The Gothic has no
aspiration of the k, and either replaces the Greek « by the
simple aspiration fA, in which case it sometimes coincides
with the Sanskrit¢ 4, or it falls to the level of the High
German, and, in the middle or end of words, usually gives
g instead of k, the High German adhering, as regards the
beginning of words, to the Gothic practice, and participating
with that dialect in the use of the h. We give here Grimm's
* table, illustrating the law of these substitutions, p, 584.
Greek ....-.P .BF D Thi
\T K.G-Ch
Gothie......F $PB|TRATD KG
Old High German, B(V)F P'D ZT |G ChK
[G. Ed. p. 80.] EXAMPLES.*

OLD
SANSKRIT. GREEK. LATIN. GOTHIC, HIGH GERM.
wea pdda-s, mous, T00-d¢, pes, pedis, fotus, vUuoz.

waa panchan, TEUTE, quinque, Jimf, ving.
qi pirna, TAEOS, plenus, fulls, vol.
faq pitri, TATNP, pater, fadreinf, vatar.
suc upari, vnép,
ce 4
super, ufar, ubar.
KavvaBis, cannabis, hanaf.
we bhanj, frangere, brikan, préchan.
WH Shuj, frut, fructus, brikén, prichén.
aq bhrdtri Srater, bréthar, pruoder.
¥y bhri, pépe, fero, baira, piru.
wz bhri, oppts, prawa.
we kapdla, m.n., Kepady, caput, haubith, houpit.
rq twam (nom.), 7%, thu, du.
WA tam (acc.), TOV, is-tum, thana, dén.
waa trayas (n.pl.), Tpets, tres, threis, dri.
Wat antara, ETEpOS, alter, anthar, andar.
eam danta-m/(acc.), OOOVT- Gy dentem, thuntu-s, zand,
Zi dwau (n. du), Svo, duo, tvai, zuéné,
efean dakshind, degia, dextra, taihsvé, zésawa.
Be uda, vowp, unda, vaté, wazar.
afeq duhitri, Suyarnp, dauhtar, tohtar.
a art dwar, Sipa, fores, daur, tor.
" wy madhu, pébu, méto.
& swan, KUWY, canis, hunths, hund,
= a4 hridaya, Kapdia, cor, hairté, hérza.
s We aksha, OKO, oculus, augé, ouga.
wy asru, daxpu, lacrima, tagr m., zahar.
, WW pasu, a pecus, faihu, vihu.
eratie
* The Sanskrit words here stand, where the termination is not separated
from the base, or the case not indicated, in their crude or simple form
(theme) ; of the verb, we give only the bare root.
+t “ Parents.”
oLp
SANSKRIT. GREEK. LATIN. GOTHIC. HIGH GERM.
WHT swasura, exupds, socer, svaihra, suehur.
zwa_ dasan, déxa, decem, taihun, zéhan.
Bt jnd, yvOus, gnosco, kan, chan.
sitfa jati,* "yévos, genus, kuni, —chuni.
wrq janu, yovu, genu, kniu, —chniu.
wea mahat, péyados, magnus, mikils, mihil.
¥a haxsa, x7" anser, gans, _— kans.
wa hyas, x9€s, heri, gistra, _késtar.
fee lik, Aérxo, lingo, laig6, lékdm.
88. The Lithuanian has left the consonants without
displacement in their old situations, only, from its defi-
ciency in aspirates, substituting simple tenues for the
Sanskrit aspirated tenues, and medials. for the aspirated
medials. Compare,
LITHUANIAN. SANSKRIT.
rata-s, “ wheel,” tea ratha-s, “waggon.”

bisu, “I would be,” vfaarta Shavishydmi.
ka-s, “ who,” wa ha-s.
diimi, “I give,” zatfa daddmi. (G. Ed. p. 82.]
pats, “husband,” “master.” afta pati-s.
penki, “five,” way panchan.
trys, “three,” waa érayas (n. pl. m.)
keturi, “ four,” warta chatwaras (n. pl. m.)
ketwirtas, “the fourth,” area chaturtha-s.
szaké, f. “ bough,” wre sékhd.
Irregular deviations occur, as might be expected, in indi-
vidual cases. Thus, for instance, naga-s, “nail” (of the
foot or finger), not naka-s, answers to the Sanskrit waa
nakhas. The Zend stands, as we have before remarked,
in the same rank,in all essentia] respects,
as the Sanskrit,
* From jan, “to be born.”

Greek, and Latin. As, however, according to §. 47., certain

consonants convey an aspiration to the letter which precedes
them, this may occasion an accidental coincidence between
the Zend and the Gothie; and both languages may, in
like manner and in the same words, depart from the ori-
ginal tenuis. Compare,
GOTHIC. ZEND. SANSKRIT.
thir (theme), “ three,” IG thri, fa tri.
thus, “to thee,” sboSS thwéi, —-& tw8*
fra, (inseparable prep.) asd fra, W pra
Sriyd, “ T love,” 56a) dau dfrindmi,} wraifa prindmi.
ahvat, “a river,” avd dfs wy ap (theme).
[G. Ed. p.83.]. I pronounce this coincidence between the
Gothic and the Zend aspirates accidental, because the causes
of it are distinct; as, on the one ‘side, the Gothic accords no
aspirating influence to the letters v and r (truda, trauan, trim-
pen, tvai), and, in the examples given above, th and / stand,
only because, according to rule, Gothic aspirates are to be
expected in the place of original tenues; on the other
side, the Zend everywhere retains the original tenues, where
the letters named in §. 47. do not exhibit an influence, which
is unknown to the Gothic; so that, quite according to order,
in by far the majority of forms which admit of comparison,
either Gothic aspirates are met with in the place of Zend
tenues, or, according to another appointment of the Ger-
manic law of substitution, Gothic tenues in that of Zend
medials. Compare,
* Twé occurs as an uninflected genitive in Rosen’s Veda-Specimen,

p. 26, and may, like the mutilated 7 ¢é, be also used as a dative.
+ “TI bless,”’ from the Sanskrit root pri, “to love,” united with the
prep. 2.
t Ahva. The Sanskrit-Zend expression signifies “water” ; and the
Gothic form developes itself through the transition, of frequent occurrence,
of p to k, for which the law of substitution requires h (see also agua).
GOTHIC. ZEND.
thu, “ thou,” 6300 tim.
fidvér, (ind.) “ four,” Y2usesGass chathwéré (n. pl. m)
fimf, wsd pancha.
fulls, “ full,” ee peréné (n. m.)-
fadrein, “ parents,” §¢/[email protected] paitar-ém (patrem).
faths, “ master,” A9S0559 paili-s.
faihu, “ beast,” AVN PASU-s.
faryith, “he wanders,” sass charaiti.
fitu-s, “ foot,” seaud pddha (§. 39.)
fraihith, “he asks,” ssassog/d pérésaiti.
ufar, “ over,” 3153» upairi, (§. 41.)
af, “ from,” was apa.
thai, “ these,” no 1.
hvas, “ who,” by ké. [G. Ed. p. 84.]
tvai, “ two,” as»g dra.
taihun, “ ten,” 223g dasa.
tai/.své, “ right hand,” 4345439259 dashina, “ dexter.”
In the Sanskrit and Zend the sonant aspirates, not the

surd, as in Greek, (¢ / too is sonant, see §. 25.) correspond,
according to rule, to the Gothic medials: as, however, in the
Zend the bh is not found, 1s 4 answers to the Gothic 6b.
Compare,
GOTHIC. ZEND. SANSKRIT.
bairith, “he carries,” ssas/ass baraiti, faut bibharti.
bréthar, “ brother,” §¢ Aso.) brdtarém(acc.)aTaca bhrdtaram (ace.)
bai, “ both,” ass uba, sat ubhdu (n. ac. v. du.)
brikan, “ to use,” wat bhuj, “ to eat.”
bi (prep.) S55 abi, sss aiwi, Bi abhi.
midya, “middling,” 2393@3%¢ maidhya, weg madhya.
bindan, “ bind,” Cys bandh, wat bandh.
89. Violations of the law of displacement of sounds, both by
persistence in the same original sound, or the substitution of
irregular sounds, are frequent in the middle and at the end of
words. Thus, in the Old High German vatar, the t of the

Greek matyp remains; in the Gothic fadrein, “ parentes,” d is
substituted irregularly for th. The same phenomenon occurs
in the cases of the Old High German olpenta, and the Gothic
ulbandus, contrasted with the 7 of éAepavt-; thus, also, the ¢
of @qt chatur, “ quatuor,” has become d in the Gothic
Jidvér instead of th; but in High German has entirely dis-
appeared. The p of the Sanskrit root @y swap, (Latin
sopio,) “sleep,” has been preserved in the Gothic slépa, and
[G. Ed. p. 85.] the Old High German s/dfu stands in the
Gothic category, but the Sanskrit root is more faithfully
preserved in the Old High German in in-suepyu (sopio, see
§. 86. 4.)
90. Nor have the inflexions or grammatical appendages
everywhere submitted* to the law of displacement, but have,
in many instances, either remained faithful to the primary
sound, or have, at least, rejected the particular change pre-
scribed by §. 87. Thus the Old High German has, in the
third person, as well singular as plural, retained the original
t; compare hapét, “ he has,” hapént, “ they have,” with habet,
habent: the Gothic, on the contrary, says habaith, haband;
the first in accordance with the law, the last in violation of
it, for habanth. Thus, also, in the part. pres., the ¢ of the old
languages has become, under the influence of the preceding
n, not th but d; the ¢ of the part. pass., however, is changed
before the s of the nom. into th, but before vowel termina-
* It would be better to regard the phenomenon here discussed by as-

suming d as the proper character of the third person in Gothic; and
viewing the Old High German ¢ as the regular substitute for it. The
d has been retained in the Gothic passive also (bair-a-da), and the active
form bairith is derivable from dairid, in that the Gothic prefers the aspi-
rates to the medials at the end of aword, The same is the case with the
part. pass., the suffix of which is, in Gothic, da, whence, in Old High Ger-
man, in consequence of the second law for the permutation of sounds,
comes ta; so that the old form recurs again, re-introduced by a fresh cor-
ruption.
tions, by an anomalous process, into d; after
the same prin
ciple by which the th of the third person before the vowel
increment of the passive is softened to d; so that de®, in-
stead of tha, corresponds to the Greek ro, of érvmrert-o, and
to the Sanskrit @ ta, of saa abhavata. The Old High
German, on the other hand, has preserved the original ¢ in
both participles: hapéntér, hapétér, Gothic habands, genitive
habandins ; habaiths, gen. habaidis.
91. Special notice is due to the fact, that in the middle
of words under the protection of a preceding consonant,
the old consonant often remains without displacement,
sometimes because it chimes in well with the preceding sound,
sometimes because, through regard for the preceding let-
ters, alterations have been admitted other than those which
the usual practiceas to displacement would lead us to expect.
Mute consonants (§. 25.), among which, in {[G. Ed. p. 86.]
the Germanic, the h must be reckoned, where it is to be pro-
nounced like our ch, protect a succeeding original ¢ Thus,
saat ashidu, “eight,” éx7d, “octo,” is in Goth. ahtau, in Old
High German ahté: aaa naktam (adverbial accusative),
“night,” w0&, vuxrés, “nor,” “noctis,” is in Gothic nahts,
Old High German naht. The liquids, on the other hand, like
the vowels, which they approach nearest of all consonants,
affect a d or th after themselves. From these euphonic
causes, for instance, the feminine suffix fa ¢i in Sanskrit, in
Greek ouc, as moinois, which designates abstract substantives,
appears in Gothic in three forms, ti, di, and thi. The ori-
ginal form ¢i shews itself after f, into which p and b mostly
resolve themselves, and also after s and h; for instance,
anst(2)s (§. 117.), “grace,” from the root an, Old High Ger-
man unnan, “to be gracious,” with the insertion of an
euphonics: fralust(i)s, “loss,” (from Jus, pres. liusa): maht(i)s,
“strength,” (from magan): fra-gifi(i)s,* betrothment,” (from
“gib, gaf), also fragibts, perhaps erroneously, as 4 has little
* Da is an abbreviation
of dai = G. ra: Sansk. té,’see §. 466.
80 CHARACTERS AD SOUNDS.
affinity with ¢: ga-skaft(i)s, “creation,” (from skap-an). The

form di finds its place after vowels, but is able, where the
vowel of the suffix falls away, i.e. in the nom. and accus.
sing., to convert d into th, because th can, more easily than d,
dispense with a following vowel, and is a favourite letter at
the end of words and before consonants, though d also is
tolerated in such a position. Hence the root bud, “to bid,”
( pres. biuda, §. 27.) forms, in the uninflected condition of the
pret., bauth, in the plur. bud-um; and the nomina! base,
mana-sé-di, “ world,” (according to Grimm’s well-founded
interpretation, “seed, not seat, of man,”) forms in the nom.
and accus. mana-séths, mana-séth, or mana-séds, mana-séd ;
but in the dat. mana-sédai not -séthai. On the other hand,
after liquids the suffix is usually thi, and after n, di: the
dental, however, once chosen, remains afterwards in every
position, either withouta vowel or before vowels; for instance,
gabaurths, “birth,” dat. gabaurthat; gafaurds, “gathering”
[G. Ed. p.87.] (from far-yan, “to go”), gen. gafaurdais:
gakunths, “esteem,” gen. gakunthais; gamunds, “memory,”
gen. gamundais; gaqvumths, “ meeting,” dat. gagvumthai, dat.
plur. gagvumthim. From the union with m, d is excluded.
On the whole, however, the law here discussed accords re-
markably with a similar phenomenon in modern Persian,
where the original ¢ of grammatical terminations and suffixes
is maintained only after mute consonants, but after vowels
and liquids is changed into d: hence, for instance, girif-tan,
“to take,” bas-tan, “to bind,” ddsh-tan, “to have,” pukh-tan,
‘to cook”: on the other hand, dd-dan, “to give,” bur-dan, “to
bear,” 4m-dun, “to come.” I do not, therefore, hesitate to
release the Germanic suffix ¢i, and all other suffixes originally
commencing with t, from the general law of substitution of
sounds, and to assign the lot of this ¢ entirely to the controul
of the preceding letter. The Old High German, in the case
of our suffix ti, as in that of other suffixes and terminations
originally commencing with ¢, accords to the original ¢ a
CHARACTERS AND SOUNDS. 8]
far more extensive prevalence, than does the Gothic; inas-

much as it retains that letter, not only when protected by
s, h, and f, but also after vowels and liquids—after m an
euphonic f is inserted ;—and the ¢ is only after / changed
intod. Hence, for instance, ans-t, “grace,” hlouft, “course,”
mah-t, “might,” sd-t, “seed,” kipurt, “birth,” var-t, “jour-
ney,’ mun-t, “protection,” ki-wal-t, “force,” scul-t, schuld,
“ouilt,” chumft, “ arrival.”
92. The law of substitution shews the greatest perti-
nacity at the beginning of words, and I have found it every-
where observed in the relation of the Gothic to the Greek
and Latin. On the other hand, in some roots which are
either deficient or disfigured in the Old European languages,
but which are common to the Germanic and the Sanskrit,
the Gothic stands on the same footing with [G. Ed. p.88.]
the Sanskrit, especially in respect of initial medials. Thus,
aq bandh, “to bind,” is also band in Gothic, not pand; We
grah, in the Vedas wt grabh, “to take,” “seize,” is grip
(pres. greipa with Guna, §. 27.) not krip;* to mm gd and
m™ gam, “to go,” correspond gagya, “I go,” and ga-tvé,
“ street ;” ae dah, “to burn,” is, in Old High German, dah-
an (Saw), “to burn,” “to light.” I can detect, however,
no instance in which Gothic tenues corr:spond to Sanskrit
as initial letters.
93(2). We return now to the Sanskrit, in order, with rela-
tion to the most essential laws of sound, to notice one ad-
verted to in our theory of single letters; where it was said
of several concurrent consonants that they were tolerated
neither at the end of words, nor in the middle before strong
consonants, and how their places were supplied in such situa-
tions. It is besides to be observed, tliat, properly, tenues
alone can terminate a Sanskrit word; but medials, only
before sonants, (§. 25,) may either be retained, if they origi-
nally terminate an inflective base, or take the place of a tenuis
* The Latin prehendo is probably related to the Sanskrit root We grah,
through the usual mee betwcen gutturals and labials.
6
or an aspirate, if these happen to precede sonants in a

yivageeige As examples, we select gftq harit, (viridis),
‘ green,” efae véda-vid, “skilled in the Véda,” wae dhana-
labh, “acquiring wealth.” These words are, according to
§. 94., without a nominative sign. We find, also, wfe efta
asti harit, “he is green,” afet Aefaa asti vedd-vit, sf wasy
asti dhana-lap ; on the other hand, zie wafer harid asti, tefae
wafer vedavid asti, waea wfet dhana-lab asti; also, efte wafa
harid bhavati, &e. With this Sanskrit law the Middle High
[G. Ed. p.89.] German is very nearly in accordance, which
indeed tolerates aspirates at the end of words, contrary to the
custom of the Sanskrit, only with a conversion of the sonant
v into the surd f, see §. 86. 3.; but, like the Sanskrit, and
independent of the law of displacement explained in §. 87.,
supplies the place of medials at the end of words regularly
by tenues. As, for example, in the genitives tages, eides,
uibes, of which the nom. and accus. sing., deprived of the
inflexion and the terminating vowel of the base, take the
forms tac, (§. 86. 1.) eit, wip. So also as to the verb; for
instance, the roots trag, lud, grab, form, in the uninflected
Ist and 3d pers. sing. pret., truoc, luot, gruop, plur. truogen,
luoden, gruoben. Where, on the other hand, the tenuis or
aspirate (v excepted) is radical, there no alteration of sound
occurs in declension or in conjugation. For instance, wort,
gen. wortes, not wordes, as in Sansk. g¢q dadat, “the giver,”
gen. tera dadalas, not gaze dadadas, but faq vit, “knowing,”
gen. fazq vidas, from the base faq vid. In Old High —
-
German different authorities of the language are at variance

with respect to the strict observance of this law. Isidor is
in accordance with it, insomuch that he converts d at the
end into ¢, and g into c; for instance, wort, wordes; dae,
dages. The Gothic excludes only the labial medials from
terminations, but replaces them, not by tenues, but by
aspirates. Hence gaf, “I gave,” in contrast to gébum, and
the accusatives hlaif, lauf, thiuf, opposed to the nominatives
hluibs, laubs, thiubs, gen. hlaibis, &. The guttural and dental
medials (g, d) are tolerated by the Gothic in terminations;

yet even in these, in individual cases, a preference appears
for the terminating aspirates. Compare bauth, “I or he
offered,” with budum, “we offered,” from the root bud;
haitad-a “nominatur” with haitith (§. 67.) “nominat; aih,
“T have,” “he has,” with aigum, “we have.”
[G. Ed. p.90.] | 93(°)}, In a sense also opposed to that of the
above-mentioned Sanskrit law, we find, in Old High German.
yet only in Notker, an euphonic relation between terminating
and initial letters of two words which come together. (Grimm,
pp- 130, 138, 181). As in Sanskrit the tenuis appears as an
essential consonant, fit for the conclusion of a sentence, but
exchangeable, under the influence of a word following in a
sentence, for the medials; so with Notker the tenuis ranks
as a true initial; stands therefore at the beginning of a
sentence, and after strong consonants; but after vowels
and the weakest consonants the liquid is turned into a
medial. Thus, for instance, ih pin, “I am,” but ih ne bin;
ter dag, “the day,” but tes tages; mit kote, “ with God,” but
minan got, “ my God.”
94. Two consonants are no longer, in the existing con-
dition of the Sanskrit, tolerated at the end of a word, but
the latter of the two is rejected. This emasculation, which
must date from an epoch subsequent to the division of the
language, as this law is not recognised either by the Zend
or by any of the European branches of the family, has
had, in many respects, a disadvantageous operation on the
Grammar, and has mutilated many forms of antiquity re-
quired by theory. In the High German we may view, as
in some degree connected with this phenomenon, the cir-
cumstance that roots with double liquids—ll, mm, nn, rr—
in forms which are indeclinable (and before the consonants
of inflexions) reject the latter of the pair. In the case, also,
of terminations in double h ort, one is rejected. Hence,
for instance, from stihhu (pungo) er-priftu (stringy), the 1st
and 3d pers. pret. stah, ar-prat.. In Middle High German,
G2
|
ie
in declensions in ck, ff, the last is rejected; for instance,

boc, gen. bockes; grif, griffes: tz loses the t; for instance,
schaz, schatzes.
95. Between a final qn and a suc- [G. Ed. p.91.]
ceeding ¢ sound —as which the palatals also must be
reckoned, for 4 ch is equivalent to tsh—in the Sanskrit an
euphonic sibilant is interposed, from the operation of the
following ¢; and by this sibilant, is converted, §. 9., into
Anuswara ; for instance, waa Wa abhavanis tatra, (abhavan-
s-tatru), “they were there.” With this coincides the cir-
cumstance, that, in High German, between a radical n and
the ¢ of an affix, an s, in certain cases, is inserted; for in-
stance, from the root ann, “to favour,” comes, in Old High
German, an-s-t, “thou favourest,” on-s-ta or onda, “I fa-
voured,” an-s-t, “favour”; from prann comes prun-s-t,
“ardour”; from chan is derived chun-s-t, “ knowledge,” our
German Kunst, in which, as in Brunst and Gunst, (from
génnen, probably formed from the ann before noticed, and
the preposite g(e),) the euphonic s has stood fast. The Gothic
exhibits this phenomenon nowhere, perhaps, but in an-s-ts
and allbrun-s-ts ‘holocaustum.’ In Old High German we
find still an s inserted after r, in the root tarr; hence, tar-s-t,
“thou darest,” tor-s-ta, I dared.” (Cf. §. 616. 2d Note:)
96. In Sanskrit the interposed euphonic s has extended
itself further only among the prefixed prepositions, which
generally enter into most intimate and facile connection with
the following root. In this manner the euphonic s steps in
between the prepositions 44 sam, Wa ava, Uf pari, afa prati,
and certain words which begin with @ hk. With this the
Latin s between ab or ob and c, g, and p, remarkably accords*,
[G. Ed. p.92.] which s, ab retains even in an isolated posi-
tion, when the above-mentioned letters follow. To this
we also refer the cosmittere of Festus, instead of commillere
* We scarcely think it necessary to defend ourselves for dividing, with

Vossius, ob-solesco, rather than with Schneider (p. 571) obs-olesco.
(Schneider, p. 475), unless an original smitfo, for mito, is

involved in this compound. In the Greek, ¢ shews an incli-
nation for connection with 7, 6, and yu, and precedes these
letters as an euphonic link, especially after short vowels, in
cases which require no special mention. In compounds like
caxes-maAos I reckon the ¢, in opposition to the common
theory, as belonging to the base of the first member (§. 128.).
We have yet to consider a case of the interpolation of an
euphonic labial, which is common to the Old Latin and Ger-
manic, and serves to facilitate the union of the labial nasal
with a dental. The Latin places p between m and a following
t or s; the Gothic and Old High German f between m and ¢.
Thus, sumpsi, prompsi, dempsi, sumptus, promptus, demptus ;
Gothic andanum-j-ts, “acceptance”; Old High German
chum-f-t, “arrival.” In Greek we find also the interpola-
tion of an euphonic § after py, of a 3 after v, of a 6 after co,
in order to facilitate the union of yu, v, and o with p and A
(ueonpBpia, wéuBAetat, avdpds, iuadcGAn—see Buttman, p. 80);
while the Modern Persian places an euphonic d between
the vowel of a prefixed preposition and that of the following
word, as be-d-i, “to him.”
97. The Greek affords few specimens of variability at the
end of words, excepting from peculiarities of dialect, as the
substitution of p fors. The alteration of the v in the article
in old inscriptions, and in the prefixes cvv, év, and maAw,
seems analogous to the changes which, according to §. 18.,
the terminating 4 m, in Sanskrit, undergoes in all cases,
with reference to the letter which follows. [G. Ed. p. 93.]
The concluding vin Greek is also generally a derivative
from p, and corresponds to this letter, which the Greek
never admits as a termination in analogous forms of the
Sanskrit, Zend, and Latin. N frequently springs from
a final ¢; thus, for instance, yey (Doric pes) and the
dual tov answer to the Sanskrit personal terminations
Ra mas, Te thas, 7a tas. Ihave found this explanation,
which I have given elsewhere, of the origin of the v from ¢
ee
a
ee
subsequently confirmed by the Prakrit, in which, in like

manner, the concluding s of the instrumental termination
plural fra bhis has passed into the dull # (Anuswara, §. 9.),
and f@ hi is said for bhis. An operation, which has a pre-
judicial effect on many Greek terminations, and disturbs the
relation to cognate languages, is the suppression of the ¢
sound at the end of words, where, in Sanskrit, Zend, and
Latin it plays an essential part. In respect of the vowels,
it is also worthy of notice, that in Sanskrit, but not in
Zend, at the meeting of vowel terminations and com-
mencements, a hiatus is guarded against, either by the
fusion of the two vowels, or, in cases where the vowel has
a cognate semi-vowel at its command, by its transition
into this latter, provided the vowel following be unlike.
We find, for instance, wetteA astidam, “est hoc,” and wey
way asty ayam, “est hie.” For the sake of clearness, and
because the junction of two vowels might too often give
the appearance of two or more words to one, I write in my
most recent text wet ‘eq, in order, by an apostrophe
which I employ as a sign of fusion, to indicate that the
vowel which appears wanting in the 44 dam is contained
in the final vowel of the preceding word. We might,
perhaps, still better write wat ‘ea, in order directly
[G. Ed. p.94.] at the close of the first word to shew that
its final vowel has arisen out of a contraction, and that the
following word participates in it.*
98. We have now to consider the alterations in the
middle of words, i.e. those of the final letters of the
roots and nominal bases before grammatical endings, and
we find, with respect to these, most life, strength, and
consciousness in the Sanskrit; and this language is
* We cannot guide ourselves here by the original MSS., as these exhibit
no separation of words, and entire verses are written together without
interruption, as though they were only a series of senseless syllables, and
not words of independent place and meaning. As we must depart from
Indian practice, the more complete the more rational the separation.
CHARACTERS AND SOUNDS. ST
placed on the highest point of antiquity, insomuch as the

signification of every radical portion is still so strongly
felt, that while it admits of moderate changes, for the
avoiding: ofsharshness, it never, if we except some vowel
elisions, permits the radical sense to be obliterated, or
rendered irrecognisable by concessions too great, or trans-
itions too daring. Yet does the Sanskrit, more than any
of its kindred, afford a field for the conflict of unsociable
consonants, a conflict, however, which is honourably and
strenuously maintained. The Vowels and weak consonants,
(§. 25.) of grammatical endings and suffixes exert no in-
fluence over preceding consonants; but strong consonants,
if surd (§. 25.), require a tenuis, and if sonant a medial,
before them. Thus, q¢ and ¥y th allow only of 4, not
qth 1g, 4 gh preceding them; only qé, not ¥q th,
= d, 4% dh; while on the other hand, y dh allows only 7 g,
not qk, @ kh, @ gh; only & d, not q 4, | th, Y dh; only
= b,not q p, ph, 4 bh to precede it. The ([G. Ed. p. 95.|
roots and the nominal bases have to regulate their final let-
ters by this law; and the occasion frequently presents itself,
since, in comparison with the cognate languages, a far greater
proportion of the roots connect the personal terminations
immediately with the root; and also among the case termi-
nations there are many which begin with consonants (sq
bhydm, fra bhis, era bhyas, @ su). To cite instances, the
root we ad, “to eat,” forms wfa admi, “I eat”; but not
Weta adsi (for s is surd), nor wefa ad-ti, Wea ad-tha, but
wf at-si, fa at-ti, Wra at-tha: on the other hand, in the
imperative, wfe ad-dii, “eat.” The base wz pad, “ foot,”
forms, in the locative plural, aq pat-su, not Wa pad-su; on
the other hand, aga mahat, “ great,” forms, in the instru-
mental plural, agfs@ mahad-blis not aeae mahat-bliis.
. 99. The Greek and Latin, as they have come down to us,
have either altogether evaded this conflict of consonants,
or exhibit, in most cases, with regard to the first of any
two contiguous consonants, a disposition to surrender it, or
at least an indifference to its assistance towards the signi-

fication of the word, since they either abandon it altogether,
or violently alter it, i.e. convey it beyond the limits of its
proper organ. These two languages afford fewer occasions
for harsh unions of consonants than the Sanskrit, princi-
pally because, with the exception of "EZ and “IA in Greek,
and ES, FER, VEL, ED, in Latin, as éo-ti, éo-pév; éo-7é,
id-pev, io-Te, est, estis, fer-t, fer-tis, vul-t, vul-tis, no root, termi-
nated by a consonant, joins on its personal terminations, or
any of them, without the aid of a connecting vowel. The
Greek perf. pass. makes an exception, and requires euphonie
alterations, which, in part, come within the natural limits
recognised by the Sanskrit, and, in part, overstep them.
[G. Ed. p.96.] The gutturals and labials remain on the
ancient footing, and before o and 7 observe the Sanskrit law
of sound cited in §. 98.; according to which x-o(€), x-7, 2-0,
m-T, are applied to roots ending in k, yy, x, or 7, , d, because
the surd o or 7 suffers neither medials nor aspirates before
it; hence rétpim-cau, TéTpem-T0, from TPIB, tétTuK-ca, TéTUK- |
rat, from TYX. The Greek, however, diverges from the
Sanskrit in this, that ~ does not leave the consonant which
precedes it unaltered, but assimilates labials to itself, and con-
verts the guttural, tenuis and aspirate into medials. For .
TETUM-pal, TETPI-pol, TéMAEY-pou, TéTUVY-wau, we should, on |
Sanskrit principles, write (§. 98.) rérum-pou, TET pL B-ot, TeMAEK-
pou, TeTvxy-pat. The ¢ sounds carry concession too far, and
abandon the Sanskrit, or original principle, as regards the
gutturals ; inasmuch as 9, 6, and ¢ (3c), instead of passing into
t before o and 7, are extinguished before o, and before 7 and v
p become o (rémero-Ta, némer-oa, méneic-you, instead of

METEIT-THL, TETMELT-THl, MeTELO-pal, Or meTEIO-wa. The Greek
declension affords occasion for the alteration of consonants
only through the ¢ of the nominative and the dative plural
termination in oz; and here the same principle holds good as
in the case of the verb, and in the formation of words: kh and
g become, as in Sanskrit, k (E=x-s), and b and ph become p.
The ¢ sounds, on the other hand, contrary to the Sanskrit; and

in accordance with the enfeebled condition, in this respect,
of the Greek, vanish entirely. We find zov-¢ for zdér-s,
mov-ci for xot-ot, which latter naturally and originally must
have stood for 1od-c, 1o0d-ct.
100. In Latin the principal occasion for the alteration of
consonants presents itself before the s of the perfect and
the ¢ of the supine, or other verbal substantive or adjective
(participles) beginning with ¢; anditisin (G. Ed. p. 97.]
accordance with the Sanskrit law cited §. 98., and the original
condition of the language, that the sonant guttural passes,
before s and ¢, into c, the sonant labial into p, as in rec-si
(rexi), rectum from req, scripsi, scriptum from scrib. It is also
in accordance with the Sanskrit that h, as a sonant (§. 25.)
and incompatible with a tenuis, becomes c before s and f¢;
compare vec-sit (vezit), with the word of like signification
a-vik-shit. If of the two final consonants of a
root the last vanishes before the s of the perfect tense
(mulst from mule and mulg, sparsi from sparg), this accords
with the Sanskrit law of sounds, by which, of two termi-
nating consonants of a nominal base, the last vanishes
before consonants of the case terminations. D ought to
become ¢ before s; and then the form, so theoretically
created, claut-sit from claud, would accord with the Sanskrit
forms, such as watrdtt a-tdui-sit, “he tormented,” from 9
tud. Instead, howeuer, of this, the d allows itself to be
extinguished; so, however, that, in compensation, a short
vowel of the root is made long, as di-vi-si; or, which is
less frequent, the d assimilates itself to the following s, as
cessi from ced. With roots in ¢, which are rarer, assimi-
lation usually takes place, as con-cus-si from cut; on the
other hand, m?2-si, not mis-si, for mif-si, from mit or mit.
B, m, and r also afford instances of assimilation in jus-si,
pres-si, ges-si, us-si.* A third resource, for the avoidance
* Compared with the Sanskrit, in which aa ush signifies “burn”;
the sibilant must here pass for the original form.
agp CHARACTERS AND SOUNDS,
of an union, very natural, but not endurable in this weak-

ened state of the language, ts, is the suppression of the
latter of these two letters, which is also compensated by
the lengthening of a short radical vowel; thus,* sédi from
[G. Ed. p.98.] séd, r2di from vid. I believe, at least, that
these forms are not derivable from sedui, vidui, and I class
them with forms like fodi from /éd, légi, for lec-si, from /éy,
Sugi, for fuc-si, from fig. To these probably also belong cavi,
Savi, fovi, for p&vi, vivi, from cin, &e. A cavui, &e. is hardly
conceivable; cavi could never have had such an origin. I|
conjecture forms such as cau-si, fau-si, after the analogy of
cautum, fautum ; or moc-si (movi), after the analogy of vic-si,
con-nic-si. (§. 19.) Possibly a moc-si form might derive pro-
bability from the adverb moz, since the latter is probably
derived from mov, as cito is from another root of motion.
The c of fluc-si, struc-si, (fluxi, &c.) fluzum, structum, must,
in the same manner, be considered as a hardening of v;
and a flu-vo, stru-vo, be presupposed, with regard to which
it is to be remembered, that, in Sanskrit also, wv often de-
velopes itself out of gu before vowels (Gram. Crit. r. 50.);
on which principle, out of flu, stru, before vowels, we might
obtain fluv, struv, and thence before consonants fluc, struc.
Thus, also, fructus out of fruv-or for fru-or. In cases of t
preceded by consonants, the suppression of s is the rule,
and ar-si for ard-i an exception. Prandi, frendi, pandi,
verti, &c., are in contrast to ar-si and other forms, like
mulsi above mentioned, in their preserving the radical letter
in preference to the auxiliary verb; and they accord in
this with the Sanskrit rule of sound, by which the s of
waatreA atdut-sam, BE akshaip-sam, &c., for the avoidance
of hardness, is suppressed before strong consonants, and
we find, for instance, sta atdut-ta, instead of witra atdut-
sta. The perfects scidi, fidi, are rendered doubtful by
their short vowel, and in their origin probably belong
to the reduplicated preterites, their first syllable having
* Cf. §& 547., and for the whole §. cf. &. 547. 576. 579.
CHARACTERS AND SOUNDS. 9]
perished in the lapse of time: in other [G. Ed. p. 99.]

respects, fidi, sctdi, correspond to tufiidi, pupiigi, not to speak
of tetigi, the i of which latter is not original.
101. The suffixes employed in the formation of words
and beginning with ¢, for the representation of which the
supine may stand, deserve spevial consideration, in regard
to the relations of sound generated by the conflict between
t and the preceding consonant. According to the original
law observed in the Sanskrit, a radical ¢ ought to remain
unaltered before tum, and d should pass into ¢; as, 734
bhéltum, “ to cleave,” from faq bhid. According to the dege-
nerated practice of the Greek, a radical d or ¢ before ¢
would become s, Of this second gradation we find a rem-
nant in comes-tus, comes-tura, analogous to es-t, es-tis, &c.
from edo: we find, however, no comes-tum, comes-tor, but
in their place comesum, comesor. We might question whe-
ther, in comésum, the s belonged to the root or to the suf-
fix; whether the d of ed, or the t of tum, had been changed
into s. The form com-es-tus might argue the radicality of
the s; but it is hard to suppose that the language should
have jumped at once from esfus to ésus, between which two
an essus probably intervened, analogous to cessum, fissum,
quassum, &c., while the ¢ of tum, tus, &c., assimilated itself
to the preceding s. Out of essum has arisen ésum, by the
suppression of an s, probably the first; for where of a pair
of consonants the one is removed, it is generally the first,
(eiué from écpi, ro-ci from 7od-ci,) possibly because, as in
§. 100., an auxiliary verb is abandoned in preference to a
letter of the main verb. After that the language had, through
such forms as é@-sum, cd-sum, divi-sum, fis-sum, quas-sum,
habituated itself to an sin suffixes properly beginning with a
t, s might easily insinuate itself into forms where it did not
owe its origin to assimilation. Cs (x) isa [G. Ed. p. 100.]
favourite combination; hence, fic-sum, nec-sum, &c. for fic-
tum, nec-tum. The liquids, m excepted, evince special incli-
nation for a succeeding s, most of all the r; hence, ter-sum,

mer-sum, cur-sum, par-sum, ver-sum, in contrast to par-tum,
tor-tum: there are also cases in which r, by a conversion
into s, accommodates itself to ¢, as in ges-fum, us-tum,
tos-tum.* This answers to the Sanskrit obligatory conver-
sion of a concluding r into s before an initial t; as, wTTe
mcta ata bhrdtas tdraya mam, “ brother save me,” instead of
“rat bhrdtar: on the other hand, in the middle of words r
remains unaltered before ¢; hence, for instance, waa bhar-
tum, not wer bhastum, “to bear.” JL exhibits in the Latin
the forms fal-sum, pul-sum, vul-sum, in contrast to cul-tum ;
n exhibits ten-tum, can-tum, opposed to man-sum. The other
forms in n-sum, except cen-sum, have been mulcted of a
radical d, as ton-sum, pen-sum.
102. In the Germanic languages, ¢ alone gives occasion
for an euphonic conversion of a preceding radical consonant;
for instance, in the 2d pers. sing. of the strong preterit»,
where, however, the ¢ in the Old High German is retainel
only in a few verbs, which associate a present signi-
fication with the form of the preterite. In the weak pre-
terites, also, which spring from these verbs, the auxiliary ¢,
where it remains unaltered, generates the same euphonic
relations. We find in these forms the Germanic on the sam>
footing as the Greek, in this respect, that it converts radical
¢ sounds (f, th, d, and in Old and Middle High German z
also) before a superadded ¢ into s. Hence, for instance, in
[G. Ed. p.101.] Gothic maimais-t (abscidisti), for maimait-t,
fai-fals-t (plicavisti), for fai-falth-t, ana-baus-t (imperasti), for
ana-baud-t. In Old and Middle High German weis-t, “ thou
knowest,” for weiz-t. The Gothic, in forming out of the
root vit, in the weak preterite, vis-sa (“I knew”), instead of
* The obvious relationship of torreo with répooua, and qa trish from

.
tarsh, argues the derivation of the latter r from s, Upon that of uro
from 3H ush, see 3. 97.
vistu, from vitta, resembles, in respect of assimilation, the

Latin forms mentioned in §. 161., such as quas-sum for quas-
tum, from guat-tum. The Old High German, however, which
also adopts wis-sa, but from muvz makes not muos-sa, but
muo-sa, corresponds, in the latter case, to such Latin forms,
as ca-sum, clau-sum. The case is different in Old High Ger-
man with those verbs of the first weak conjugation, which,
having their syllables made long generally through two
terminating consonants in the preterite, apply the ¢ of the
auxiliary verb directly to the root. Here the transition of t¢
into s does not occur, but ¢, z, and even d, remain unaltered;
and only when another consonant precedes them ¢ and d are
extinguished, z on the contrary remains ; for instance, leit-ta,
“puxi,” ki-neiz-ta, “ ar¥.ixi,” ar-dd-ta, “ vasTavi,” walz-ta,
“votvi,” liuh-ta, “vuxi,” for livht-ta; hul-ta, “ pracavi,” for
huld-ta. Of double consonants one only is retained, and of ch
or cch only h; other consonantal combinations remain, how-
ever, undisturbed, as ran-ta, “ cucurRI,” for rann-ta; wanh-ta,
* vacILLavi,” for wanch-ta ; dah-ia, “ rexi,” for dacch-ta. The
Middle High German follows essentially the same principles,
only a simple radical ¢ gives way before the auxiliary verb,
and thus lei-te is opposed to the Old High German /eit-ta ; on
the other hand, in roots in /d and rd the d may be maintained,
and the ¢ of the auxiliary be surrendered—as dulde, “ ToLERAvt”
—unless we admit a division of dul-de, and consider the d as
a softened ¢. The change of g into c (§. 98.) is natural, but
not universal; for instance, anc-te, “ arctavi,” for ang-te ; but
against this law b remains unaltered. ([G. Ed. p. 102.]
Before the formative suffixes beginning with ¢*, both in Gothic
and High German, guttural and labial tenues and medials are
changed into their aspirates, although the tenuis accord with
a following ¢ Thus, for instance, in Gothic, vah-teéd,
* With the exception of the High German passive part. of the weaker
form, which, in the adjunction of its ¢ to the root, follows the analogy of
the pret. above described
“watch,” from vak; sauh-t(i)s, “sickness," from suk ;

mah-t(i)s, “might,” from mag; ga-skaf-t(i)s, “creation,”
from skap; fragif-t(i)s, “ betrothment,” from gid, softened
from gab; Old High German sult, malt, ki-skaft, “ creature,”
kift, “gift.” The dentals replace the aspirate th by the
sibilant (s), as is the case in Gothic before the pers. cha-
racter £ of the preterite, as th cannot be combined with ¢.
The formation of words, however, affords few examples of
this kind: under this head comes our masé, related to the
Gothic mats, “food,” and matyan, “to eat.” In Gothic, the
s of bléstreis, “ worshipper,” springs from the ¢ of blétan,
“to worship”: beist, “leaven,” comes probably from beit
(beitan, “to bite,” Grimm, ii. p. 208). The Zend accords,
in this respect, with the Germanic*, but still more with
the Greek, in that it converts its ¢ sounds into w 8, not
only before ~ é, but also before § m; for instance, 2539575
irista, “dead,” from the root 32s irith; wovass basta,
“ bound,” from Cys bandh, with the nasal excluded ; as
in Modern Persian si. bastah, from wv band; »Gx~~s
aésma, “ wood,” from =m idhma. :
103. It is a violation of one of the most natural laws of
sound, that, in Gothic, the medial g does not universally
pass into k or h (=ch), before the personal character ¢ of
(G. Ed. p.103.] the pret., but generally is retained; ana
we find, for instance, 6g-t, “thou fearest,” mag-t, “thou
canst”; and yet, before other inflections formed with ¢,
the g undergoes an euphonic transition into hf, as for in-
stance, éh-ta, “I feared,” mah-ts, “might.”
104. When in Sanskrit, according to §. 98., the aspiration
of a medial undergoes a necessary suppression, it falls back,
under certain conditions and according to special laws,
upon the initial consonant of the root, yet only upon a
medial, or throws itself onward on the initial consonant. of
* Cf. the Sclavonic and Lithuanian, §. 457.
+ No other roots in g in this person are to be found in Ulfilas,
the following suffix. We find, for instance, wrenfa b/ot-

syfmi, “I shall know,” for qyeeqfa bédh-sydmi ; aaa véda-
bhut, “knowing the vedas,” for qa budh; qe bud-dia,
“knowing,” for qt budhta ; ‘rearfa dhdk-shydmi, “¥ shall
milk,” for egenfa dvh-sydmi; ea dug-dha, “milked,” for
gen duh-ta. In Greek we find a remarkable relic of the first
part of the transposition of the aspirate,* in the necessary
suppression of the aspirate in some roots which begin with
t and end with an aspirate before o, 7, and y, letters which
admit of no union with an aspirate, and in its being thrown
back on the initial letter, by which process + becomes 6.
Hence, tpépw, Opér-ow, (Opéyrw), Ope rjp, Opéu-pa, tapi, Oar-
Tw, érapny, TéeFapu-por; Tevpos, Opvn-Tw, ETpvpyv, Optu-pa;
Tpéxw, OpéEomat; Opit, teryds, Taxts, Gacowv. In the spirit
of this transposition of the aspirate, éy obtains the spiritus as-
per when y is obliged to merge in the tenuis, (Exrds, Ew, &£15).+
* Sce J. L. Burnouf in the Asiatic Journal, III. 368; and Buttmann,
pp. 77, 78.
+ It is usual to explain this appearance by the supposition of two aspi-
rations in the root of these forms, of which one only is supposed to appear
in deference to the euphonic law which forbids the admission of two con.
secutive aspirated syllables. This one would be the last [G. Ed. p. 104.]
of the two, and the other would only shew itself when the latter had been
forced to merge in the tenuis. Opposed, however, to this explanation is
the fact, that, on account of the inconvenience of accumulated aspirates, the
language has guarded itself in the original formation of its roots against
the evil, and has never admitted an aspirated consonant at o1.ce for the
initial and final letter ofa root. In Sanskrit, the collection of w hose roots
iscomplete, there is no such instance. The forms, however, édap@nv,
rebapba, rebapbe, rebadara, tebpapba, <bpéepOnv,, present a difficulty.
These, perhaps, are eccentricities of usage, which, once habituated to the
initial aspiration by its frequent application to supply the place of the ter-
minating one, began to-assume its radicality, and extended it wider than
was legitimate. We might also say, that since ¢@ (as x4) is so favourite
a combination in Greek that it is even substituted for 76 and 86—while, ac-
cording to §. 98., an original ¢@ ought to become 76—on this ground the
tendency to aspiration of the root remained unsatisfied hy ¢era¢@nv Ke.;
and as if the @ only existed out of reference to the @, the original ter-
minating aspirate necessarily fell back on the radical initial. This theory,
which seems to me sound, would only leave re@agara to be explained.
Cg Gg
OF THE ROOTS.
(G. Ed. p.105.] 105. There are in Sanskrit, and the lan-
guages which are akin to it, two classes of roots: from the
one, which is by far the more numerous, spring verbs, and
nouns (substantives and adjectives) which stand in fraternal
connection with the verbs, not in the relation of descent from
them, not begotten by them, but sprung from the same
shoot with them. We term them, nevertheless, for the
sake of distinction, and according to prevailing custom,
Verbal Roots; and the verb, too, stands in close formal
connection with them, because from many roots each per-
son of the present is formed by simply adding the requi-
site personal termination. From the second class spring
pronouns, all original prepositions, conjunctions, and par-
ticles: we name them Pronominal Roots, because they all
express a pronominal idea, which, in the prepositions, con-
junctions, and particles, lies more or less concealed. No
simple pronouns can be carried back, either according to
their meaning or their form, to any thing more general, but
their declension-theme (or inflective base) is at the same
time their root. The Indian Grammarians, however, derive
all words, the pronouns included, from verbal roots, although
the majority of pronominal bases, even in a formal respect,
are opposed to sueh a derivation, because they, for the most
part, end with a: one, indeed, consists simply of a. Among
[G. Ed. p.106.] the verbal roots, however, there is not a
single one in d, although long a, and all other vowels, #t
du excepted, occur among the final letters of the verbal
roots. Accidental external identity takes place between the
verbal and pronominal roots; ¢.g. i signifies, as a verbal
root, “to go,” as a pronominal root, “ he,” “this.”
106. The verbal roots, like those of the pronouns, are
OF THE ROOTS. 97
monosyllabic; and the polysyllabic forms represented by

the grammarians as roots contain either a reduplicate-
syllable, as aT] jagri, “to wake,” or a preposition which has
grown up with the root, as wait ava-dhir, “ to despise ™ ;
or they have sprung from a noun, like gart kumdr, “ to
play,” which I derive from @att kumdra, “a boy.” Except
the law of their being monosyllabic, the Sanskrit roots are
subjected to no further limitation, and their one-syllableness
may present itself under all possible forms, in the shcrtest
and most extended, as well as those of a ‘middle degree.
This free state of irrestriction was necessary, as the language
was to contain within the limits of one-syllableness the
whole body of fundamental ideas. The simple vowels and
consonants were not sufficient: it was requisite to frame
roots also where several consonants, combined in inseparable
unity, became, as it were, simple sounds; e.g. eat sthd, “to
stand,” a root in which the age of the co-existence of the s
and th is supported by the unanimous testimony of all the
members of our race of languages. So also, in a
skand, “to go,” (Lat. scand-o) the age of the combination of
consonants, both in the beginning and ending of the root, is
certified by the agreement of the Latin with the Sanskrit.
The proposition, that in the earliest period of language a
simple vowel is sufficient to express a verbal idea, is sup-
ported by the remarkable coneurrence of ([G. Ed.p. 107.]
nearly all the individuals of the Sanskrit family of lan-
_ guages in expressing the idea “to go” by the root i.
107. The nature and peculiarity of the Sanskrit verbal
roots explains itself still more by comparison with those
of the Semitic languages. These require, as far as we
trace back their antiquity, three consonants, which, as I.
have already elsewhere shewn,* express the fundamental
* Trans. of the Hist. Phil. Class of the R. A. of Litt. of Berlin for the
year 1824, p. 126, &c.
ts
98 OF THE ROOTS.
idea by themselves alone, without the aid of vowels; and

although they may be momentarily compressed into one
syllable, still, in this, the combination of the middle radical
with the first’ or last cannot be recognised as original and
belonging to the root, because it is only transitory, and
chiefly depends on the mechanism of the construction of
the word. Thus, in Hebrew, kdédl, “slain,” in the fem.,
on account of the addition dh contracts itself to kta! (ktitl-
-ah); while kété, “slaying,” before the same addition, com-
presses itself in an opposite manner, and forms 4£déldh.
Neither ktd/, therefore, nor ké4tl, can be regarded as the root;
and just as little can it be looked for in tél, as the status con-
structus of the infinitive; for this is only a shortening of the
absolute form kdtél, produced by a natural tendency to pass
hastily to the word governed by the infinitive, which, as it
were, has grown to it. In the imperative k/é/ the abbrevia-
tion is not external, subject to mechanical conditions, but
rather dynamic, and occasioned by the hurry with which a
command is usually enunciated. In the Semitic languages,
in decided opposition to those of the Sanskrit family, the’
vowels belong, not to the root, but to the grammatical motion,
the secondary ideas, and the mechanism of the construction of
[G. Ed. p.108.] the word. By them, for example, is dis-
tinguished, in Arabic, katala, “ he slew,” from kutila, “he was
slain”; and in Hebrew, kété/, ‘‘ slaying,” from kdtél, “slain.”
A Semitic root is unpronounceable, because, in giving it
vowels, an advance is made to a special grammatical form, and —
it then no longer possesses the simple peculiarity of a root
raised above all grammar. But in the Sanskrit family of
languages, if its oldest state is consulted inthe languages which
have continued most pure, the root appears as a circumscribed
nucleus, which is almost unalterable, and which surrounds
itself with foreign syllables, whose origin we must investi-
gate, and whose destination is, to express the secondary
ideas of grammar which the root itself cannot express.
OF THE ROOTS. 99
' The vowel, with one or more consonants, and sometimes

without any consonant whatever, belongs to the fundamental
meaning: it can be lengthened to the highest degree, or
raised by Guna or Vriddhi; and this lengthening or raising,
and, more lately, the retention of an original a, opposed to
its weakening to z or change to u (SS. 66., 67.), belongs not to
the denoting of grammatical relations, which require to be
more clearly pointed out, but, as I imagine I can prove, only
to the mechanism, the symmetry of construction.
108. As the Semitic roots, on account of their construc-
tion, possess the most surprising capacity for indicating
the secondary ideas of grammar by the mere internal mould-
ing of the root, of which they also make extensive use, while
the Sanskrit roots, at the first grammatical movement, are
compelled to assume external additions ; so must it appear
strange, that F. von Schlegel,* while he ([G. Ed. p. 109.]
divides languages in general into two chief races, of which
the one denotes the secondary intentions of meaning by an
internal alteration of the sound of the root by inflexion, the
other always by the addition of a word, which may by
itself signify plurality, past time, what is to be in future,
or other relative ideas of that kind, allots the Sanskrit
and its sisters to the former race, and the Semitic lan-
guages to the second. “There may, indeed,” he writes,
p- 48, “arise an appearance of inflexion, when the annexed
particles are melted down with the chief word so as to be
no longer distinguishable ;but where in a language, as in
the Arabic, and in all which are connected with it, the first
and most important relations, as those of the person to
verbs, are denoted by the addition of particles which have
a meaning for themselves individually, and the tendency
to which suffixes shews itself deeply seated in the language,
- it may there be safely assumed that the same may have
* In his work on the language

and wisdom of the Indians.
H 2
106 OF THE ROOTS,
occurred in other positions, where the annexation of par-

ticles of a foreign nature no longer admits of such clear
discrimination: one may at least safely assume that the
language, on the whole, belongs to this chief race, although
in this single point, by admixture or artificial adornment,
it has adopted another and a higher character.” We must
here preliminarily observe, that, in Sanskrit and the Jan-
guages connected with it, the personal terminations of the
verls shew at least as great a similarity to isolated pro-
nouns as in Arabic. How should any language, which
expresses the pronominal relations of the verbs by syllables
annexed either at the beginning or end of the word, in the
choice of these syllables avoid, and not rather select, those
which, in their isolated state, also express the corresponding
[G. Ed. p.110.] pronominal ideas? By inflexion, F. von
Schlegel understands the internal alteration of the sound of
the root, or (p. 35) the internal modification of the root, which
he (p. 48) opposes to addition from without. But when from
do or dw, in Greek, comes d/dw-pt, 56-cw, Jo-Oyodue8a, what
are the forms pl, ow, Onodpeba, but palpable external addi-
tions to the root, which is not at all internally altered, or
only in the quantity of the vowel? If, then, by inflexion,
an internal modification of the root is to be understood,
the Sanskrit and Greek &c. have in that case—except the
reduplication, which is supplied by the elements of the root
itself—scarce any inflexion at all to shew. If, however,
Onoépe8a is an external modification of the root do, simply
because it is combined with it, touches it, with it expresses
a whole; then the idea of sea and continent may be repre-
sented as an internal modification of the sea, and vice versd.
P. 50, F. von Schlegel remarks: “In the Indian or Grecian
language every root is truly that which the name says,
and like a living germ; for since the ideas of relation are
denoted by internal alteration, freer room is given for
deve!opment, the fulness of which can be indefinitely
_ OF THE ROOTS. 101
‘extended, and is, in fact, often wondrously rich. All, how-

ever, which in this manner proceeds from the simple root,
still retains the stamp of its relationship, adheres to it, and
thus reciprocally bears and supports itself.” I find, how-
ever, the inference not established ; for from the capability
of expressing ideas of relation by internal alteration of the
root, how can the capability be deduced of surrounding the
(internally unalterable) root indefinitely, with foreign syllables
externally added? What kind of stamp of relationship is
there between ys, ow, Gyoduefa, and the ([G. Ed. p. 111.]
roots to which these significative additions are appended ?
We therefore recognise in the inflexions of the Sanskrit
family of languages no internal involutions of the rvot, but
elements of themselves significative, and the tracing of the
origin of which is the task of scientific grammar. But even
if the origin of not a single one of these inflexions could be
traced with certainty, still the principle of the formation
of grammar, by external addition, would not, for that
reason, be the less certain, because, at the first glance, in
the majority of inflexions, one discovers at least so much,
that they do not belong to the root, but have been added
from without. A. W. von Schlegel, also, who, in essential
points, assents to the above-mentioned division of lJan-
guages,* gives us to understand, with regurd to the so-called
* Nevertheless, in his work, “Observations sur la langue et la littérature

provencales,"’ p. 14, &c., he gives three classes, viz. Les langues sans aucune
structure grammaticale, les langues qui emploient des affixes, et les langues
d inflexions. Of the latter, he says: ‘‘Je pense, cependant, qu'il faut
assigner le premier rang aux la” gues a inflexions. On pourroit jes appeler
les langues organiques, parce qu’elles renferment un principe vivant de
developpement et d’accroissement, et qu’elles ont seules, si je puis m’ex-
primer ainsi, une végétation abondante et féconde. Le merveillevx
artifice de ces langues est, de former une immense variété de mots, et de
marquer la liaison des idées que ces mots désignent, moyennant un assez
petit nombre de syllabes qui, considérées séparément, n’ont point de signi-
fication
102 OF THE ROOTS.
inflexions, that they are not modifications of the root, but

foreign additions, whose characteristic lies in this, that
[G. Ed. p.112.] regarded, per se, they have no meaning.
In the Semitic, the appended grammatical syllables or in-
flexions have no meaning, at least in so far that they do not,
any more than in Sanskrit, occur isolated in a completely
similar state. In Arabic, for instance, antum, and not tum,
is said for “ye”; and in Sanskrit ma, ta, and not mi, ti, are
the declinable bases of the first and third person; and at-Ti,
“he eats,” has the same relation to TA-m, “him,” that in
Gothic JT-a, “I eat,” has to the monosyllabic AT, “I ate.”
The reason for weakening the a of the base to i is probably,
in the different cases of the two sister languages, the same,
viz. the greater extent of the form of word with i (comp.
§.6.) If, then, the division of languages made by F. von
Schlegel is untenable, on the reasons on which it is
founded, still there is much ingenuity in the thought of a
natural history or classification of languages. We prefer,
however, to present, with A. W. von Schlegel (1. ¢.), three
classes, and distinguish them as follows: first, languages
with monosyllabic roots, without the capability of composition,
and hence without organism, without grammar. This class
comprises Chinese, where ull is hitherto bare root, and the
grammatical categories, and secondary relations after the
fication, mais qui déterminent avec précision le sens du mot auqnel elles
sont jointes. En modifiant les lettres radicales, et en ajoutant aux racines
des syllabes dérivatives, on forme de mots dérivés de diverses espéces, et
des dérivés des dérivés. On compose des mots de plusieurs racines pour
exprimer les idées complexes. Ensuite on décline les substantifs, les
adjectifs, et les pronoms, par genres, par nombres, et par cas; on conjugue
les verbes par voix, par modes, par temps, par nombres, et par personnes,
en employant de méme des désinences et quelquefois des augmens qui, sé-
parément, ne signifient rien. Cette méthode procure l’avantage d’énoncer
en un seul mot l’idée principale, souvent déja trés-modifiée et trés-com-
plexe, avec tout son cortége d’idées accessoires et de relations variables.
OF THE ROOTS. : 103
main point, can only be discovered from the position of the

roots in the sentence.* Secondly, languages with mono-
syllabic roots, which are capable of combination, and obtain
their organism and grammar nearly in this way alone. The
chief principle of the formation of words, in this class,
appears to me to lie in the combination of verbal and pro-
nominal roots, which together represent, [G. Ed.p. 113.]
as it were, body and soul (Comp. §. 100.). To this class belongs
the Sanskrit fumily of languages, and moreover all other
languages, so far as they are not comprehended under 1. and
3., and have maintained themselves in a condition which
renders it possible to trace back their forms of words to the
simplest elements. Thirdly, languages with dissyllabic verbal
roots, and three necessary consonants as single vehicles of
the fundamental meaning. This class comprehends merely
the Semitic languages, and produces its grammatical forms,
not simply by combination, like the second class, but by a
mere internal modification of the roots. We here gladly
award to the Sanskrit family of languages a great superiority
over the Semitic, which we do not, however, find in the use
of inflexions as syllables per se devoid of meaning, but in
the copiousness of these grammatical additions, which are
really significative, and connected with words used isolated;
in the judicious, ingenious selection and application of them,
and the accurate and acute defining of various relations, which
hereby becomes possible; finally, in the beautiful adjustment
of these additions to a harmonious whole, which bears the
appearance of an organized body.
109°. The Indian Grammarians divide the roots accord-
ing to properties, (which extend only to the tenses which
* We find this view of the Chinese admirably elucidated in W. von

Humboldt’s talented pamphlet, “Zeftre 2 M. Abel Remusat, sur la na-
ture des formes grammaticales en général, et sur le génie de la langue
chinoise.”
104 OF THE ROOTS.
1 call the special tenses,* and to the part. pres.,) into ten
classes, all of which we have re-discovered in the Zend also,
and examples of which are given in the following paragraph.
[G. Ed. p.114.] We shall here give the characteristics of
the Sanskrit classes, and compare with them those which
correspond in the European sister languages.
(1.) The first and sixth class add "a to the root; and
we reserve the discussion of the origin of this and other
conjugational affixes for the disquisition on the verb. The
point of difference between the first class of nearly 1000
roots (almost the half of the entire number) and the sixth
class, which contains about 130 roots, lies in this, that the
former raise the vowel of the root by Guna (§. 26.), while
the latter retain it pure; eg. atufa bddhati, “he knows,”
from qy budh (1.); gefa tudati, “he vexes” (comp. tundit),
from qe tud (6.) Asa has noGuna,t no discrimination can
take place through this vowel between the classes 1. and 6.:
but nearly all the roots which belong to either, having ¥ a
as the radical vowel, are reckoned in the first class. In Greek,
e (before nasals o, §. 3.) corresponds to the affix wa; and
Acin-o-pev,t pevy-o-wev, from AID, GY (éArmov, Epvyov),
belong to the first class, because they have Guna (§. 26.);
while, e.g. O/y-o-yev, OAiP-o-uev, &c., fall under the sixth
class.|| In Latin we recognise, in the third conjugation,
* In Greek, the present (indic. imper. and optat., the form of the Greek
subjunct. is wanting in Sanskrit) and imperfect correspond to them; be-
yond which certain conjugation-signs do not extend. In German, the
present of every mood corresponds.
+ The accent here distinguishes the Ist cl. from the 6th. e.g. for pdtati
did it belong to the 6th. cl., we should have patdti.
t We give the plural, because the singular, on account of abbreviation,
makes the thing less perspicuous,
|| Sanskrit long vowels admit Guna only when they occur at the end of
the root, but in the beginning and middle remain without admixture of
the @ a; so do short vowels before double consonants.
OF THE ROOTS. 105
which I would raise to the first, the cognate of the Sanskrit

first and sixth class, since we regard the addition i as a
weakening of the old a (§.8.); and e.g. legimus has the same
relation to Aéy-o-ev, that the genitive ped-is has to 10d-d¢
where the Sanskrit has likewise a (Wea [G. Ed. p. 115.]
pad-as). In leg-u-nt, from ley-a-nti, the old a, through the
influence of the liquid, has become u (Comp. §.66.). In
German, all the primitive (strong) verbs, with the exception
of some remains of the fourth class (No. 2.), stand in clear
connection with the Sanskrit first class, which is here, for the
first time, laid down in its full extent.* The wa which
is added to the root has, in Gothict, before some personal
terminations, remained unchanged ; before others, according
to §. 67., and as in Latin, been weakened to i; so, hait-a, “I
am. called,” hait-i-s, hait-i-th, 2d pers. du. huit-a-ts ; pl. hait-
a-m, hait-i-th, hait-a-nd. The radical vowels i and u keep the
Guna addition, as in Sanskrit, only that the a which gives
the Guna is here weakened to i (§. 27.), which, with a radical
i, is aggregated into a long i (written ei, §.70.): hence keina
(=Kina, from kiina), “I germinate,” from KIN; biuga,
“T bend,” from BUG, Sanskrit wy bhuj, whence wa bhugna,
“bent.” The diphthongs ai, au, as in Sanskrit @ and W
(S. 2.), are incapable of any Guna; as are é( =, §. 69.) and
a. The Sanskrit radical vowel w a has, however, in Gothic,
experienced a threefold destiny. It has either remained
unaltered in the special tenses, and is lengthened in the
preterite, except in reduplicate roots (i.e. to d, see §. 69.)—
* I have already, in my Review of Grimm’s Grammar, expressed the

conjecture that the a of forms like haita, haitam, haitaima, &c. does not
belong to the personal termination, but is identical with the W a of the
‘Sanskrit Ist and 6th classes; but 1 was not then clear regarding the Guna
in the present in all roots with vowels capable of Guna. (See Ann. Reg.
for Crit. of Litt., Book II. pp. 282 and 259:)
+ We make frequent mention of the Gothic alone as the true starting-
point and light of German Grammar. The application to the High Ger-
man wil! hereafter present itself.
106 OF THE ROOTS.
thus, e.g. far-i-th, “he wanders,” answers to 4tfa charati

[G. Ed. p. 116.] (8. 14.), and for, “he wandered,” to watz
chachdra ; or, secondly, the old a shews itself in the special
tenses weakened to-i, but retained in the monosyllabic singu-
Jar of the preterite: so that here the stronger a (§. 8.) corre-
sponds to the weaker é in the same way that, in the first case,
the ¢6(= 14) does to the shorta. The root wq ad, “to
eat,” in Gothic, according to §. 87., forms 47’; hence, in the
present, ita; in the sing. pret., at, as-t, at. The third fate
which befalls the a of the root in Gothic is a complete
extirpation, and compensation by the weaker i, which is
treated like an original 7, existing in the Sanskrit; i.e. in the
special tenses it receives Guna by é, and in the pret. sing. by
a (§. 27.), but in the pret. pl. it is preserved pure. To this
class belongs the KIN, “to germinate,” mentioned above,
pres. keina, pret. sing. kain, pl. kin-um. The corresponding
Sanskrit root is #4 jan, “ to produce,” “to be born” (see
7 «666
§. 87.): the same relation, too, has greipa, graip, yripum,

from GRIP, “to seize,” to mt grabh (Véda form): on the
other hand, BIT, “to bite,” * (beita, bait, bitum), has an
original i, which exists in Sanskrit (comp. faz bhid, “ to
cleave”); just so, VT, “to know,” Sanskrit fag vid.
(2.) The fourth class of Sanskrit roots adds to them the
syllable aq ya, and herein agrees with the special tenses of
the passive; and from the roots which belong to it spring
chiefly neuter verbs, as e.g. ayafa nasyali, “ he perishes,”
Their number amounts altogether to about 130. The German
has preserved one unmistakeable remnant of this class, in
those strong verbs which again lay aside, in the preterite, the
syllable ya (weakened to yi), which is added to the root in the
[G. Ed. p. 117.]__ special tenses; e.g. vahs-ya(Zend JH
ucs-yann, “crescebant,” Vendidad S. p. 257), “cresco,” vahs-
yith, “ crescit,” pret. véhs.
* Occurs only with the prep. and, and with the meaning “ to scold,”
but corresponds to the Old High German root BJZ, “ to bite.”
-
OF THE ROOTS. 107
(3.) The second, third, and seventh classes add the personal
termination direct to the root; but inthe cognate European
languages, to facilitate the conjugation, these classes have
mainly passed over to the first class; e. g. ed-i-mus, not ed-mus
(as a remnant of the old construction es-f, es-tis), Gothic
it-a-m, Old High German iz-a-més not iz-més, answering
to the Sanskrit wera ad-mas. The second class, to which
wz ad belongs, leaves the root without any characteristic
addition, with Guna of the vowels capable of Guna before
light terminations, which must be hereafter explained ; hence,
e.g. wf émi, corresponding to ¥#a imas, from ¥i “to go,”
as in Greek ei to ivev. It contains not more than about
seventy roots, partly terminating in consonants, partly in
yowels. In this and the third-class, the Greek exhibits roots,
almost entirely ending in vowels, as the above mentioned
lL, A, TNO (yvG-6:), AQ, STA, CH, SY Edu), AY, &e.
To the consonants the direct combination with the conso-
nants of the termination has become too heavy, and ‘ES alone
(because of the facility of cu, cv) has remained in the San-
skrit second class, as the corresponding root in Latin, Lithua-
nian,and German. Hence, wfeq asli, éoti, Lithuan. esti, est,
Gothic and High German ist. In the Latin there fall also
to the second class, J, DA, STA, FLA, FA,and NA; and also
in-quam, whence QUA weakened to QUI, is the root, which,
in Gothic, appears as QUAT, weakened to QUIT, with the
accretion ofa JT. FER and VEL (VUL) have preserved
some persons of the ancient construction.* ([G. Ed. p. 118.]
The third class is distinguished from the second by a syllable
of reduplication in the special tenses, and has maintained
itself under this form in Greek also, and Lithuanian. In
* Five roots of the second class introduce in Sanskrit, between the con-
sonants of the root and the personal termination, an ¥ i, as Uifefa réd-i-
mi, “I weep,” from Re rud. I can, however, nolonger believe that the
i of the Latin third conjug. is connected with this ¥ i, as there is scarce
any doubt of its relationship with the # a of the very copious first class.
108 OF THE ROOTS.
Sanskrit it comprehends about twenty roots; eg. warfa

daddmi, Siw, Lithwanian dudu; ewafa dadhdmi, réOnue
(8. 16.); sratfat jajanmi, “I beget,” comp. yHyv-o-uo. The
seventh class, of about twenty-four roots, introduces, in the
special tenses, a nasal into the root, which is extended before
the light personal terminations to the syllable xa; ¢g. faafa
bhinadmi, “I cleave,” fara bhindmas, “we cleave.” The
Latin has kept the weaker form of this nasalization, but has
further added to the root the affix of the first class (p. 114 G.
Ed.); hence findo, find-i-mus. From the Greek come to be here
considered roots, like MA©, AAB, OIT, in which the inserted
nasal has been repeated further on in the word, with the pre-
fixed a, and, like the Latin find-i-mus, is connected with the
affix of the first class; thus, pav0-dv-o-uev, AnuP-dv-o-pev,
Oiryry-dev-0-jev. )
(4.) The fifth class, of about thirty roots, has nw; and the
eighth, with ten roots, which, excepting # kri, “to make,”
all terminate in q » or @ n, has uw for its characteristic addi-
tion: the u, however, of these two classes is lengthened
before the light terminations by Guna, which in the corre-
sponding Greek appended syllables, vy and v, is supplied by
lengthening the v; thus, e.g. de‘kvour, decxviuev, as in Sanskrit
sataifa dp-nd-mi, “ ad-ip-is-cor,” argra dp-nu-mas, “ adipisei-
mur.” An example of the eighth class is wa tan, “to extend,”
whence watfa tan-d-mi=Tav-0-1a, AYRA, tan-u-mas=TAv-U- Mes.
With the 3 w, v, of the eighth class, is probably connected
[G. Ed. p.119.]_ the v in some Gothic strong verbs, where,
however, it adheres so firmly to the root, that, in a German
point of view, it must be regarded asa radical. Hence it is
not dropped in the preterite, and receives, in the special
tenses, like all strong verbs, the affix of the Sanskrit first
class; e.g. saihva,* “I see,” sahv, “I saw.”
(5.) The ninth class adds at 4 to the root, which syl-
lable, before heavy terminations, instead of being shortened
* I now consider the v of saihva and similar verbs as purely euphonic,
cf. §. 86, and Latin forms like cogno, linquo, stinguo.
OF THE ROOTS. 109
to a na, replaces the heavy = @ by the lighter $7 ($ 6.\,

and is thus weakened to at né# E.g. from ¥q mrid, “to
crush,” (comp. mordeo) comes gzatta mridndmi, setae mrid-
nimas. In this is easily perceived the relationship with
Greek formations in vy (vas) vasev; e.g. Sauvyur, apyva-
pev. As Ge, and o, are originally one, formations like tésu-vo-
ev belong to this class, only that they have wandered into the
more modern w-conjugation at a remote period of antiquity;
for more lately vew would not have become vw from vn.
(6.) The tenth class adds wa aya to the root, but is dis-
tinguished from the other classes in this farther important
point, that this affix is not limited to the special tenses:
the final a of =a aya is peculiar to them, but wa ay
extends, with very few exceptions, to all the other forma-
tions of the root. All causals, and many denominatives,
follow this class, and, indeed, from every root a causal can
be formed by the addition wa ay, which is always accom-
panied by Guna of the middle vowel of the root capable of
Guna, or by Vriddhi of every radical final vowel and of a
niddle a belonging to the root; e.g. a¢ufa véd-aya-ti “he
makes to know,” from fag vid; wraafe srév-aya-ti, “he makes
to hear,” from 4 sru. We recognise, in German, the affix
Sq aya at least in two shapes: inthe one (G. Ed. p. 120.]
the first a, in the other the last, is lost, and in the latter case
y has become 7; so that I have no longer any scruple in
tracing back Grimm’s first and third conjugation of the weak
formto a common origin. According to allprobability, how-
ever, the verbs with the affix 6 also (as Old High German
manén, “to mention,” “to make to think,”) belong to this
class, regarding which we will speak further under the verb.
The Old High German gives é as the contraction of a +i,
(see §. 78.), but retains its é more firmly than the Gothic its
ai, which, in several persons, sinks into a simple a. Compare
Gothic haba, habam, haband, with Old High German hapém,
hapémes, hapént. Very remarkable, however, is the concur-.
rence of the Prakrit with the Old High German and the Latin.
110 OF THE ROOTS,
of the 2d conj. in this point, that it in like manner has

contracted the affix wa ayato zé. Compare Sanskrit arqarfa
mdnaydmt, “I honour,” Prakrit arafa mdnémi,* Old High
German, var-maném, “I despise,” Latin moneo:
[G. Ed. p. 121.] OLD
SANSKRIT. PRAKRIT. HIGH GERMAN. LATIN.
araatfa mdnaydmi = aafAmdnémi —var-maném _moneo
aTauta mdnayasi arafa mdnési manés mon's
areata mdnayati . arate manédi manét monet
RTAaTAa mdnaydmas adize mdnémha manémes monémus
araay mdnayatha atara mdnédha manét — monétis
aTaafeat mdnayanti = atarfat mdnénti manént monent
In regard to those weak verbs, which have suppressed the
first vowel of the Sanskrit wa aya, and give therefore ya as
affix, we will here further recall attention to the forms iga
(ige), which occasionally occur in Old High German and
Anglo Saxon, whose connection with wa aya is to be traced
thus, that the semi-vowel y has become hardened to g,
(comp. §. 19.), and the preceding a weakened toi. In Greek,
the cognate verbs to the Sanskrit of the tenth class are to
* I am not at present able to adduce this verb from the edited texts: it
is, however, certain, that mdnaydmi in this dialect can have no other sound
but mdnémi, The conjugation is supported by other examples of this
class, as chintémi, “I think” (from chintaydmi), nivédémi (from nivé-
daydmi). In the plural the termination mha is nothing else than the ap-
pended verb substantive (Sansk. smas, ‘‘we are”). In the third pers, pl.,
together with mdnenti the forms mdnaanti and mdnanti are also admis-
sible. The Indian Grammarians assume for the Sanscrit a root mén, “ to
honour”: more probably, however, the verb, for which this root is sup-
plied, is only a denominative from mdna, “honour” ; and this substantive
itself a derivation from man, “‘ to think,’ whence ava-man, ‘‘to despise,”
as in Old High German var-MAWN (by Otfrid, fir-MON). The root,
therefore, which is contained in varmaném is identical with the Gothic
MAN (man, “JI mean,” “I think,” pl. munum see §. 66.). To this class
belongs, also, the Latin monere, as, “to make to think” (Old High German
manén), the radical o for a of which we explain by the principle of §. 66.
(see, also, §. 3.); while the i of memin-i is a weakening Of the original a,
explained by §. 6.
OF THE ROOTS. lil
be looked for in those in aw, ew, ow; in Latin, besides the

2d conjugation compared above, most verbs of the Ist aud 4th
also belong to this affinity. We shall recur to them when
speaking of the verb.
109°. In order to adduce single examples of the mul-
tiform construction of the roots, let us examine the order
of the final letters; but we will select only such examples
as are common to the Sanskrit and several sister lan-
guages. The greatest forbearance, however, is requisite,
as an authenticated comparison of all that admits of com-
parison would easily swell to a book, which shall hereafter
be devoted to this subject.*
(1.) Roots ending with a vowel:— [G. Ed, p. 122.]
“There are, as has been already remarked (§. 105.), no
roots in Wa; but roots in wtdéare numerous. Thus 77° ga,t
“to go,” contained in the Latin navi-ga-re; also, perhaps,
in fati-gare, the first member of which belongs to fatiscor,
fessus; in Greek, Bin: answers to sama jagdmi, and rests
on the frequent interchange of gutturals and labials; Gothic
ga-thvé, “a street,” (see p. 102. G. Ed.); Zend pw gd-tu,
“a place,” (nom. ase gdtus; Old High German gd-m,
“I go,” =samnfa ja-gé-mi; not therefore, as Grimm con-
jectures (p. 868), by syncope from gungu, but, with a more
ancient and regular foundation, only with a suppression ot
the Sanskrit syllable of reduplication, introduced, therefore
from the third into the second class (see p. 117. G. Ed.), as in
Latin, da-mus answering to dido-uev. Thus, also, sfd-m,
std-s, sta-t, in like mamner, with suppressed reduplication,
corresponds to i-ory-pe (for cicrypu), and to the Sanskrit
root et sthd, which is irregularly inflected, frrsifa tishthami,
fasta tishthasi, fagfa tishthati, for tasthdmi, tasthdsi, tasthdti.
* Somewhat that pertains to this subject I have already put together

very concisely at the end of my Sanserit Glossary.
+ The attached cyphers denote the classes described in §. 1093.
112 ; OF THE ROOTS.
which will be more closely considered hereafter. The

Latin, in root and inflexion, most resembles the Old High
German: the Zend, however, in its sGaspeyse histdmi® (for
sistdmi, see §. 53.), appears in a genuine Greek dress, Ob-
serve, also, the guspeyrasGas? rathaéstdo, “ warrior,” which
occurs so often in the Zend-Avesta, properly “chariot
stander,” with o for s as the sign of the nominative.
How, then, in Old High German, comes from STA the
extended form of the root SZ4NT;, whence the pre-
sent stuntu, “I stand,” and preterite stuont, “I or he
stood”; for which the Gothic has sfanda, stéth? We
will here only preliminarily remark, that we have ob-
served in Zend also, in some roots terminating in 4, an
inclination to connect themselves with a ¢-sound. Thus
we find, from v4.9 snd, “to wash,” “to purify,” (Sansk. @t snd,
“to bathe,”) whence sndta, “ purified,” in Vend. S. p. 233,
frequently JEOINaaw js509 fra-snddhayén “ lavent” ; from ag
da, “to lay,” (Sans. yt dhd, p. 118 G. Ed.), we find prysGsasgs J
nidaithyann, “ deponant” (as Vendidad S. pp. 205 and 206,
pryVssGsasgsy WEEE w5aw huské zémé nidaithyann, “ in siccd
terra deponant™): from the same root we find the imperative
[G. Ed. p.123.] form, AsG Gags) ni-da-thama, “ depona-
mus” (Vend. S. p. 208, »s¢.0/ss G9 9 G gysous.s/s Gry 285
5G. Gags SMH AEG sewas kva naraim isritanaim
taniim barama Ahura mazda kva niddthdma, “Quo hominum
mortuorum corpus feramus, ubi deponamus”?). Of the Ger-
manic we will further remark, that the root at md, “to
measure” (cf. ué-Tpov), has connected itself with a t-sound, and
forms, in Gothic, MAT, present mita (§. 109°. 1.). at’ jnd,
“to be acquainted with,” “to know,” TNQ, GNA (gnarus)
Old High German CHNA (§.87.); whence chnd-ta, “I knew,”
annexing the auxiliary verb direct, as in Latin (g)no-vi. To
* I believe I may deduce this form from the 3d pers. pl. SOWETO SY
histénti (cf. torayrt) in the V. 8. p.183: more on this head under the verb.
OF THE ROOTS. 113
the special form, HrTatfa jéndmi, for aratfe jnd-nd-mi, may be-
long the Gothic root KANN, Old High German CHANN
(kann. chan, “I sarge see §.94., kunnum, chunnum, “ we
know,” see §.66.). wr dhmd, “to blow,” alters itself in the
special forms to wa dham, Latin FZA, according to the
second class (§. 109* 2), Old High German PLA (8§. 12. 20.),
whence pld-ta, “flavi.” As in Sanskrit, from the above-men-
tioned wa dham, comes the nominal base wrat dhamant.
“a vein”; so may the Gothic base BLOTHA (nom. acc.
bléth, “blood”) come here also under consideration. We
pass on to roots in i, and have to remark that the root
mentioned at p. 107. G. Ed., ¢i, “to go,” is not unknown
inGerman. We find it in the Gothic imperative hir-i, “come
here”; du. hir-yats; pl. hir-yith. I believe, too, that in the
i r preterite iddya, “I went,” the i alone can be as-:
sumed as the root. In Zend occurs spss aéi-ti, “he goes™
(from efa éi, according to §§. 28. 41.), Lithuan. ei-ti. fa”
sri, “to go,” with the prep. 3q uf, “to raise itself”; hence,
wfega uchchhrita,“raised,” “high”; compare cre-sco, cre-vi
(see §. 21.), Old High German SCRIT, “to step,” with the
addition ofa ¢, as in the case of mat, from at m4: perhaps
the Latin gradior, as well as cresco, might be here included,
the Guna form of the vowel, as in wafa sray-a-ti, “he
goes,” being observed. fim smi, “to smile,” Old High
German SMIL; w¥ pri, “to love,” Zend sd fri (8. 47.), Goth.
Sriys, “T love” (§. 87.), compare fra priya, “dear.” —a bhi,
“to fear,” fatfa biblé-mi, “I fear”; Lithuan. biyau; Gothic
fiya, *T hate” (fiyais, Siyaith), fiyands, “ foe”; Old High Ger-
man véém or fiém, “I hate”: the Greek @é@-o-na: answers to
the Sanskrit reduplication of bibhémi; so that, contrary to
the common rule, the aspirates have remained in the prefix,
but in the base itself have become medials, and this has left
only £ as the whole root, as in Sanskrit da-d-mus, “we give,”
for da-dd-mas, d:-8o-yes. Perhaps, also, [G. Ed. p. 124.]
e14, pérdopnou, is to be referred to the roots in i, so that an
I
114 OF THE ROOTS,
unorganic dental affix would be to be assumed. gf s

* to lie,” “to sleep,’ with irregular Guna in the middle;
hence S4-té=xel-ra. #° hré, “to be ashamed”; Old High
German HRU, “to. repent” Cleteyvy hrou, hru-umés, see
p- 115. G. Ed.). Of roots in u, % dru, “ to run,” gafa drav-
a-ti, “he runs” may furnish, through the Guna form, the.
Greek dpa-cKxw, d:-Spa-cKxw, which appears hence. to derive
its a with suppression of the digamma: the p of dpéuw, how-
ever, might pass as a hardening of the ¥v (§. 63), and
dpéu-o-pev, Spéu-e-re, &c., therefore represent most truly the
forms drav-d-mas, drav-a-tha. “ plu, “to go,” “to swim,”
“to float” (34 plava, “a ship”), Latin FLU. The Greek
mTAéw, TAS is again not to be so regarded as if the old u had:
been corrupted to ¢€ or o, but 7Aé(F)w, 7Ao(F)w. supply the place
of the Guna form in plav-é (of the middle voice), 3d pers.
plav-a-té ; the future mAévow, the v having the Guna (§. 26.),
answers toNt@ pld-shyé; Lithuan. plaukiu, “I swim,” with
a guttural added, as in Latin fluc-si from fluv (p. 98.
G. Ed.) Old High German VLUZ, “to flow,” pre-sup-
poses the Gothic FZUT' (§.87.); with the favourite dental,
addition, with which all final vowels are so commonly.
invested. a sru, “to hear,” KAY (§§. 20., 21.), Gothie
HLIU-MAN (nominative hAliuma), “ear,” as “hearer,”
with weakened Guna (§. 27.); with regard to the ki for
sr, compare, also, clunis with rat srént, f. “hip.”)
Lithaan. klausau, “I hear.” Perhaps erudio, as “to make
hear,” is to be referred to this class: the derivation from e
and rudis is little satisfactory. Anquetil introduces a Zend
erodé, célebre, (kAurds), which I have not yet found in the ori-
ginal text, but I meet with the causal form s¢xyya»a029
srdvayémi (Sansk. wraarfa srdvaydmi), “I speak,” “ recite™
(V.S. p. 38). The Old High German, scrirumés, “we have
exclaimed,” gives SCRIR as the root, and rests probably on
the form srév (§. 20.), with a thinning of the 4 to é (§. 66.)
the present and sing. preterite, however, have lost the r (scriu
t
OF THE ROOTS. 115
for scriru, screi for screir), like the Greek xA7-cw, xéxAy-Ku, &e.
The Latin clamo, however, has the same relation to WT srdv
that mare has to arft vdri, “ water” (§. 63.), and dpe to
¥4 drav, from ¥ dru, “to run.” >w hu, “to extol,” “to
glorify” (aspp pe huniita, “he celebrated,” V. S. p. 39.), is
probably the root of the Greek tivo (Uu(e)vos), which I do
not like to regard as an irregular derivative from vdw.
Upii- ~ ty purify,” PUrus. This root is the verbal
parent of the wind and fire, which are both represented
as pure. Waa pavana (with Gunaand ana ([G. Ed. p.125.]
as suffix) is “the wind,” and the corresponding Gothic FONA
(neut. nom. ace. fén, see §. 116.) is “ fire,” which in Sanskrit
is called wea pdv-a-ka, with Vriddhi and aka as suffix.
The relation of FON.A to waa pavana resembles that of the
Latin mélo from mavolo; the loss of the syllable 4 va
is replaced by the lengthening of the a (§.69.). The Greek
mvp and Old High German VIURA (nom. ace. viur), the
latter with weakened Guna (§.27.), and ra as suffix, both
fallto the root, yp. q bri, “to speak,” Zend 9% mr
(e.g. gbass mraé-m, “I spoke,” V. S. p. 123.); the Greek
pé(F)w rests on the Guna form watfa brav-i-mi, and has,
as often happens, lost the former of two initial consonants
(cf. also péw, pevw, and ruo, with g sru, “to flow”). _The
Old High German SPRAH, or SPRAHH (sprihhu, “I
speak,” sprah, “I spoke”) appears to have proceeded from
wa brav, by hardening the 4 v (see §. 19.), and prefixing an
s akin tothe p. x bhi, “to be,” Zend gs bi, Lithuan. BU
(future bisu, “I will be”), Latin FU, Greek ®Y. Pro-
bably, also, BY, in zpéc-Gv-s, xpecBirys, &e., is only
another form of this root (cf. §. 18.); so that zpé¢ would
have to be regarded as a preposition from apd (9 pra,)
essentially distinguished only by a euphonic = (cf. §. 96.).
Moreover, the base zpéc@u has a striking resemblance to
wy prabhu (excelsus, augustus), literally, “being before.”
In Old High German pim or bim corresponds to the
eisi: 19
116 OF THE ROOTS.
Sanskrit watfa bhavdmi: more exact, however, is the corre-

spondence in the plural of pir-u-més, pir-u-t, to bhav-d-mas,
“sumus,” bhav-a-tha, “ estis” (see §. 19.). To this class belongs,
also, PU, “to dwell” (pii-ta, “T dwelt”), as the Sanskrit qa
vas “to dwell,” in German VAS, WAS, has become seyn. In
Sanskrit, too, from W bhi, “to be,” comes the substantive
bhav-ana “house,” as place of being. The Gothic baua,
“T build,” may be regarded as the causal of the idea “ to
be,” like the Latin faciv (§. 19.): its conjugation answers
also to arautfa bhdvaydmi, “I make to be,” which, in Pra-
krit, may sound bhdvémi, bhdvési, bhdvéti (Gothic baua,
bauais, bauait). See p. 121 G. Ed. Sanskrit roots ending in
diphthongs (zé, 8d, & di; there are no roots in Wt du)
follow in their formations, in many respects, the analogy of
roots in at d. We abstain from adducing examples of
them, as they also offer little occasion for comparison.
(2.) Roots terminating with a consonant, We shall give
[G. Ed. p.126.] only a few examples, in which we compare
roots with the same vowel, and proceed in the order, a, i, u.
According to §. 1. we do not allow the vowel ¥7i and ¥]77
to belong to the root. Long radical vowels before a final
consonant are rare; and the majority of them are probably
not original.
The most numerous class of roots ending with a conso-
nant has a medial wa. So qq** vach, Zend yah vach
(secbas adcta, “dixit,” Vend. S. p. 124), Greek EU for FED
(§. 14.), Latin VOC, Old High German, WAH, WAG (ki-
wahu, “mentionem facio,” pret. ki-wuoh pl. ki wuogumés).
wa prachh, Zend s3¢/ed pérés, Gothic FRAH; pres watfa
prichchhdmi, sGass9¢/ed pérésdmi, fraiha for friha (see §. 82,
and §. 109°.1.); the Latin ROG (rogo, interrogo) appears to
be abbreviated from FROG. ua pat, “to fall,” “to fly,”
Zend esd pat, “to fly” (Vend. S. p. 257. Uyyasb awd RONPY
JHPIIA0B> asasw7> ppovo.sd yat frd vayé patann urvara ucsyann,
“where birds fly, trees grow”). One sees clearly from this
OF THE ROOTS. 1i7
that,in Greek, aixrw, merdéw, merdoucn, métomor, wrHm, &e-

belong to a common root IIET; Latin PET, peto, im-peto,
prepetes, penna by assimilation for pet-na.. In Gothic
FATH, or, with the vowel weakened, FI TH, might be
looked for. To the latter corresponds, according to §. 87.
Old High German VED, in véd-ara, “ feather,” aq
vad, “ to speak,” Latin VAD, contained in vas, vad-is. From
3¢ vad proceeds the abbreviated form 3 ud, to which per-
tains “YA (idw, idéw, dys). The Old High German gives
WAZ (var-wizu “maledico”), with z for d, according to §. 87.,
and the vowel of the base lengthened, as in ateurfa vddaydmi,
according to the tenth class. #@° sad,.“ to sink, with the
prep. fa ni, “to set oneself down”; Latin SED, SID, sido,
sedeo ; Greek ‘EA, ‘IZ, é£os, pa, ious; Gothic SAT
(S. 87.), sita, “I sit” (p. 116 G. Ed.).. =a? an, “to blow,”
“to breathe,” wfae anila, “wind,” Gothic 4N, usana,
“I expire,” cf. dveyos, “animus.” wa° jan, “to beget,”
Zend sass zan (S. 58.), SExy zazdmi, “I beget,” Sanskrit
water jajanmi, Greek TEN, Latin GEN (yiyvouna, yévos,
gigno, genus), Gothic KIN, “to germinate,” (p. 116 G. Ed.) ;
kuni, “gender” (S. 66.). at kar (= kri), e.g. Biita kardti,
“facit”: this root, in Zend, follows the fifth class; e.g.
spsbasyc7es kérénaditi (8. 41.), “ facit,” robasye%es kérénadt, “ fe-
cil,” Sed 9/5429 kéréniidhi, “fac”; Old High German kara-
wan or garawan, “to prepare”; Latin creo, cura (cf. @&
kuru, “ fac”), ceremonia, and with p for c (§. 14.), paro ; Greek
Kpaive, Kpa-Tos ; with 7, mpaccw, moax-cw, [G. Ed. p. 127.]
mpGy-na, where the guttural appears to be a hardening of the
qv (8.19), eg. of waft kurvanti, “ faciunt” (from kur-u-
-anti). az vah, “to drive,” “to carry,” Zend gu vaz (8. 57.),
Latin VEH, Greek cyos, “ wagon,” as bearer, carrier, for
Féyos. we svas, “to breathe,” cf. spiro, according to
§§. 50. and 22. weg"’ grah, “to take”: the original
form, occurring in the Vedas, is wm grabh. To this the
Zend form ‘belongs, according to the tenth class, and,
118 OF THE ROOTS.
indeed, so that the y bh appears before vowels as » v, but

before @tas ap. Thus we read in the Vend. S. p. 155:
Gewdcheoa Fela, Uyy roausyccheng> roshy SP] © Grawypyas
MGI SWIAS POW AWS Peswssamheoas ashdum ; yézi néit
uzvarézydt yb narém dgéréptém dgeurvayéité, kd hé asti chitha?
“Pure! si non dimittit, qui hominem captum capit (i. e. tenet),
quenam ei est pena” ?* In the European sister languages
I believe I recognise this root in three forms: the Gothic
GRIP has been already mentioned (p. 116 G. Ed.), likewise
Pprehendo (§. 92. note): by changing the medials into their te-
nues, KAETI also seems to belong to this class, Gothic HZ/F,
“to steal,” hliftus, “thief.” Finally, also, in Greek, yptmos,
ypthos, “the net,” stands quite isolated, and appears to
me to be related to the Indian qa grabh, by changing
the a intoi. wra? ds, “to sit,’ Greek ‘H= a remnant of
tne second class, terminating in a consonant to be supplied
at §. 109%. 3.; jjo-ror answers exactly to wre ds-té (middle
voice), and hence jar stands for jopor, as efi for éoui (San-
skrit asmi). urs’ bhrdj, “to shine,” Zend sees béréz (§. 58),
ort 6g Aas bareéz, whence the part. pres. wyrsse/es bérézant,
nom. m. Sywscere_s bérézans, “ splendens,” “ altus,” very fre-
quently occurs. This Zend form prepares the way for the Old
High German root PERAH, whence PERAH-TA}, nom.
perah-t, “fulgidus.” Tothis root belongs, also, our Pracht. The
Greek language gives ®AET (§. 20.) a cognate root, and thus
[G. Ed. p.128.] points to a Sanskrit short a for the long
one. The cognate root in Latin is FLAG, flagro.
chhid, “to cleave,” SCID, scind-i-mus=chhindmas (§. 14.):
ZXIZ, perhaps also ZKIA, oxidvyus, &e. belong to this
place; the form is more genuine, and the ideas, too, of
* Anquetil translates, “‘ Si celui qui a commis l Aguerefté ne reconnoit

pas sa faute quelle sera sa punition.”
+ Cf. p. 1281. Note *
{ The h (in the sense of ch) corresponding to the j, y, accords with
§. 87., but is moreover favoured by the following ¢.
OF THE ROOTS. 119
Clearing, dispersing, separating, are kindred ones. The

Gothic SKAID, “to separate,” if the relationship is
certain, has a stiffened Guna, so that ai appears to belong
to the root. According to §. 87., however, the Gothic
form should be SKAIT and the Old High German SKEIZ
for SEEID. fae” vid, “to know,” Zend sb vid, "IA;
Gothic VID, Old High German, VJZ; in the Latin VJD,
and in eidw, “I see,” the seeing is regarded as something,
which “makes to know,” and the conjugation of video is causal,
according to p. 121 G.Ed. Thus, also, another root, signify-
ing “to know,” namely Ty budh, has, in Zend, gained the
meaning “to see.”* According to the tenth class, and
with the prep. ni, VID, in Zend, signifies “to summon”
( sGrm93sQn»1»5) nivaédhayémi, “invoco,” see §. 28.) In Go-
thic, VIT receives through the prep. in the meaning “to
adore” (inveita, invait, invitum). —feg’ dis, “to shew,”
Zend 939 di’; hence Uyyasss0.54503 fradaésayé, “thou
shewest” (Vend. S. p. 123), Greek AIK, with Guna defxvuys,
according to the fifth class; Latin DIC, in dico, as it were,
**to point out,” and dicis ( dicis causa). In Gothic, the rule
laid down in §. 87. requires the form TJH, and this root,
combined with ga, signifies “to announce” (ga-teiha, ga-
tath, ga-taihum, for ga-tihum, according to §. 82.). On the
other hand, in faikus, “sign,” the law for the transposition
of letters is violated. ata! jiv, “life;” Lithuanian gywa-s,
“alive,” gywent “I live,” gywata “life;* Gothic QUIVA,
nom. guivs, “alive”; Latin VIV, as it appears from QUIV,
as bis from duis (Sansk. fea dwis), viginti from tviginti. The
Zend has dropped either the vowel or the v of this root.
Hence, e g. »»y_jva, nom. Loy jud, “ living,” (V. S. p. 189);
and Uyssupsuw hu-jitayé, “bonam vitam habentes” (1. c. p. 222),
from ses hu-jiti. From ji, the root, would become. with
Guna, jaydémi, on which rests the Greek Caw, the j having
* Vide Gram. Crit. p. 328.

120 OF THE ROOTS,
fallen out (§. 14.); but Bios also belongs to this root, and finds
a medium of comparison with #tq j’v, in the Latin vivo. Of
roots with w, ce ruch, “‘to shine,” and a rud, “to weep,”
may serve as examples; the former, in Zeud, is yb? raéch,
(§$§. 28. 32.), and follows the tenth class, e.g. 50.570 y3a5as7
[G. Ed. p.129.] raochayéiti, “splendet.” In Latin correspond
LUC, luc-s, luceo (§. 20.) and RUD: the Greek has, in both
roots, replaced the r by /, and presents, for comparison, AYK
(aupiAdKn, Avkdpws) and AYZ; to the former, Advyvos, Avy-
vévw, &c,, has the same relation that, in Zend, aug» aso
tafnu-s, “‘ burning,” has to the root dase tap (8. 40.) We
must assign Aeuxds also, with Guna, to the root AYK. The
Gothic gives LUH for LUK, according to §. 87.; whence,
with the original, or with weakened Guna (§§. 26., 27.),
spring forms like lauhméni, “lightning,” lauhatyan, “ to
lighten,” liuhath, “light.” Without Guna, and preserving
the old smooth letter, stands Jukarn (theme, Jukarna, neut.),
“lamp,” rather isolated. A root corresponding to eq rud
is wanting in Gothic, but the Old High German has for it,
quite regularly according to §. 87., RUZ, “ to weep” (riuzu,
réz for rauz, according to §. 80., ruzumés). rie bhish,
“to adorn,” is perhaps contained in the Latin or-no, with
loss of the initial letter, as amo in relation to a@raafa
kamaydmi, “I love.” .With regard to the r for ¥ sh,
advert to the relation of uro to 3¥ ush, “to burn,” ;
sev, “to honour,” ay médh, “to think”(?). The latter
cannot hitherto be quoted as a verb: it springs, however,
from awa médhas-and at médhd, “ understanding,” unless
it should be preferred to assume for these words a root
midh, which, however, the Grammarians do not exhibit.
The Gothic has, for comparison, MIT, whence mité, “I
think”: the Greek furnishes an analogous word to sév,
viz. SEB, céBw. (§. 4.)
110. From the monosyllabic roots proceed nouns, sub-
stantive and adjective, by the annexation of syllables,
OF THE ROOTS, 12]
which we should not, without examination, regard as not,

per se, significative and, as it were, supernatural mystic
beings; to a passive belief in whose undiscoverable nature
we are not willing to surrender ourselves. It is more
natural to suppose that they have or had meaning, and.
that the organism of language connects that which has a
meaning with what is likewise significative. Why should
not language denote accessory ideas, by accessory words
appended to the root? Language, which possesses both
sense and body, infuses sense and imparts form to every
word. The objectof nouns is to represent (G. Ed. p. 130.]
persons or things, to which that which the abstract root ex-
presses adheres; and hence it is most natural to look for
pronouns in the elements used in the formation of words, as
the bearers of qualities, actions, and conditions, which the
root expresses in abstracto. There appears, too, in reality,
as we shall develope in the chapter on the pronouns, a com-
plete* identity between the most important elements in
the formation of words and some pronominal bases which
are declined even in an isolated state. But it is not sur-
prising that several of the elements of verbal formation, in
the class of independent words, should not admit of more
certain explanation; for these affixes have their origin in
the most obscure and early epoch of language, and subse-
quently they have themselves lost all consciousness as to
whence they have been taken, on which account the ap-
pended suffix does not always keep equal pace with the
alterations which, in the course of time, occur in the cor-
_ responding isolated word; or it has been altered while the
other remains unchanged. Still, in individual cases, we
may remark the admirable exactitude with which the
appended grammatical syllables have maintained them-
* I direct attention preliminarily to my treatise “‘On the Influence of

Pronouns in the Formation of Words” (Berlin, by F. Diimmler).
122 OF THE ROOTS,
selves through thousands of years in an unaltered form;

I say, we may remark this from the perfect accordance
which exists between various individuals of the Sanskrit
family of languages, although these languages have been
removed, as it were, from each other's eyes since time
immemorial, and every sister dialect has, since that removal,
been left to its own fate and experience.
111. There are also pure radical words, i.¢. those of which
the theme, without suffix of derivation or personality, repre-
G. Ed. p.181.] sents the naked root, which are then united
in declension with the syllables which denote the relations of
case. Except at the end of compounds, such radical words
are, in Sanskrit, few in number, and are all feminine ab-
stracts; as, Ht bhi, “fear,” qy yudh, “ contest,” qq mud,
“joy.” In Greek and Latin the pure root is the most rare
form of the word ; but it does not always appear as an abstract
substantive. As, for instance, e.g. pAoy (pAdk-s), dm (67-¢).
vip (vir-s), leg (lec-s), pac (pac-s), duc (duc-s), pel-lic (pel-lec-s).
In German, commencing even with the Gothic, no pure
radical words exist, although, by reason of the abbrevia-~
tion of the base of the word in the singular, many words
have assumed that appearance; for from the abbreviation
of these verbal bases, which has been constantly extending
during the lapse of time, it is precisely the most modern
dialects which appear to exhibit the greatest number of
naked roots as nouns. (cf.§.116.) Naked roots seem most
generally used at the end of compounds, on account of the
clogging of the preceding part of the word. According to
this principle, in Sanskrit, every root can, in this position,
designate the agent by itself; as, eg. wafag dharma-vid.
“duty-knowing.” In Latin, the use of these compounds
is as frequent as in Sanskrit, only that, according to §. 6.
a radical a is weakened to i or e; thus, carni-fic (fec-s),
tubi-cin (cen) An example in Greek is yepwB (for -va
from vin-rw). Sanskrit roots which end with short vowels;
OF THE ROOTS. 123
as fa ji, “to conquer,” are, in compounds of this kind,

supported by the addition of a ¢, which so much the more
appears to be a simple phonetic affix without signification.
that these weakly-constructed roots appear to support them-
selves on an auxiliary ¢ before the gerundial suffix ya also.
Thus, e.g. aifaa svarga-jit, “ conquering the heaven,” fafa
vi-jit-ya, “by conquering.” InLatinI find ([G. Ed. p.132.]
interesting analogies to these formations in IT and STIT,
from the roots J and STA, the latter weakened to STI ac-
eording to §.6. Thus, com-it (com-es), “goer with”; equ-it
jegu-es), “goer on horseback”; al-it (al-es), “goer with
wings”; super-stit (-stes), “standing by.” ‘The German has
in this way supported throughout with a ¢ several roots ter-
minating with a vowel, and hence given to this letter the
character of radicalism, as above mentioned (p. 123 G. Ed.)
in MAT, from ay mé, “ to measure.”
( 124
FORMATION OF CASES.
112. The Indian Grammarians take up the declinable
_, word in its primary form, i.e. in its state when destitute
~ of all case-termination; and this bare form of the word is
given also in dictionaries. In this we follow their example;
and where we give Sanscrit and Zend nouns, they stand,
unless it is otherwise specified, or the sign of case is
Separated from the base, in their primary form. The
Indian Grammarians, however, did not arrive at their pri-
mary forms by the method of independent analysis, as it
were by an anatomical dissection or chemical decomposi-
tion of the body of language; but were guided by the
practical use of the language itself, which, at the beginning
of compounds—and the art of composition is, in Sanscrit,
just as necessary as that of conjugation or declension—
Vrequires the pure primary form; naturally with reserva-
tion of the slight changes of the adjoining limits of sound,
rendered necessary at times by the laws of euphony. As
the primary form at the beginning of compounds can re-
present every relation of case, it is, as it were, the case
general, or the most general of cases, which, in the unli-
mited use of compounds, occurs more frequently than any
other. Nevertheless, the Sanskrit language does not every-
where remain true to the strict and logical principle usually
[G. Ed. p. 184.] followed in composition; and as if to vex
the Grammarians, and put their logic to the test, it places as
the first member of the compounds in the pronouns of the
first and second person the ablative plural, and in those of the
third person the nom. and ace. sing. of the neuter, instead of
‘the true primary form. The Indian Grammarians, then, in
g FORMATION OF CASES, 125
_/this point, have applied to the cases furnished to them by

Sthe language, and take the augmented “eq asmat or
WHE as “from us, hmat or mshoiend:
; “from 226 as the soc eg in the eaten or as
the primary form, although in both pronominal forms only
aa and qyu belong to the base, which, however, does not
extend to the singular. That, however, in spite of this
error, the Indian Grammarians understand how to decline
the pronouns, and that they are not deficient in external
rules for this purpose, is a matter of course. That the
interrogative, in its declension, resembles bases in a, can-
not escape any one who holds the neuter fam kim for the
original indeclinable form of the word. Panini settles the
matter here with a very laconic rule, when he says (edit.
Cale, p. 969) faa:a: kimali kali, i. e, ka* is substituted for
kim. If this strange method were to be followed in Latin,
and the neuter guid in like manner regarded as the
theme, then, in order to get at the dative cu-i —
the analogy of fructui), one would have to say “ quidis cus,”
or “qguidi cus.” In another place (p. 825), Panini forms
from idam, “this” (which in tike manner has the honour
of passing for a base) and him, “what?” a copulative
compound; and by gefgart Sgat idankimér iski, the Gram-
marian teaches that the putative bases in (6G. Ed. p. 135.]
the formations under discussion substitute for themselves
the forms# and ké.
® | 113. The Sanskrit, and the languages akin to it, which
in this respect have still kept upon the old footing, distin-
guish, besides the two natural genders, another—the
neuter, which the Indian Grammarians call Kliva, i.e. eu-
nuch; which appears to be a peculiarity of the San-
' * He forms, namely, from kits regarded as a base, kim-as, which

in reality does not occur, and which has, for the sake of euphony, here
become kimah.
126 FORMATION OF CASES,
skrit, or most perfect family of languages. According to —

its original intention this gender had to represent inani-
’ mate nature, but it has not everywhere confined itself to
these old limits: the language imparts life to what ‘is
inanimate, and, on the other hand, (according to the view
then taken,) impairs the personality of what is by nature
‘ animate. The feminine in Sanskrit, both in the base and
in the case-terminations, loves a luxurious fullness of
- form; and where it is distinguished from the other
genders in the base or in the termination, it marks this
distinction by broader, -and more sonant vowels. The
neuter, on the other hand, prefers the greatest conciseness,
but distinguishes itself from the masculine, not in the base,
but only, in the most conspicuous cases, in the nominative
and its perfect counterpart the accusative; in the vocative ie
also, when this is the same as the nominative.
114. Number, in Sanskrit and its sister languages, is
distinguished, not by a particular affix denoting the number,
but by the /selection or modification of the case-syllable,/
so that, with the case-suffix, the number is at once known;
“@s Js bhyam, bhydm, and bhyas are cognate syllables, and,
among other relations, express that of the dative; the first
in the singular (only in the pronoun of the 2d person, qt
tubhyam, “ to thee’), the second in the dual, the third in the
plural. The dual, like the neuter, in course of time is the first
to be lost with the weakening of the vitality [G. Ed. p, 136.]
of the view taken by the senses, or is more and more straitened
in its use, and then replaced by the abstract. plural expressive
_of infinite number. The Sanskrit. possesses the dual most
fully, both in the noun and in the verb, and employs it every-
where where its use could be expected. In the Zend, which
otherwise approximates so closely to the Sanskrit, it is
found very rarely in the verb, more frequently in the
noun. The Pali has only as much left of it as the Latin,
viz, &@remnant of it in two words, which signify “two”
FORMATION OF CASES: 127 4. 4h
af am
Fons es
t wanting.
it is entirely
and “both”; in the Prakri Of ~ +
the German languages, only the eldest dialect, the Gothic, < ©?" Re >
possesses it, but merely in the verb; while, on the con- o~"
trary, in the Hebrew (speaking here of the Semitic
languages) it is retained only in the noun, in disadvan-
tageous contrast with the Arabic, which, in many other
respects also, is a more perfect language, and which main-
tains the dual in equal fulness in the verb also; while in ~
the Syriac it has been almost entirely lost in the noun as
well as in the verb.*
_- 115, The case-terminations express the reciprocal rela-
tions of nouns, i.e. the relations of the persons spoken of, to
| one another, which principally and originally referred only:
“ to space, but from space were extended ‘also: to time and
L cause. \According to their origin, they @ are,at least for the
most part, pronouns, as will be more clearly developed
hereafter, /Whence could the exponents of the relations
of space, which have grown up with the primary words
© into a whole, have better been taken, than from those
words which express personality, with their inherent secon-
dary idea of room, of that which is nearer or more distant,
? of that which is on this or that side? / |G. Ed. p. 137.]
As also in verbs the personal terminations, i. e. the pronominal
suffixes—although, in the course of time, they are no longer
recognised and felt to be that which, by their demonstrable
origin, they imply and are—are replaced, or, if we may
* | use the expression, commented on by the isolated pronouns
A prefixed to the verb; so, in the more sunken, insensible |
- state of the language, the spiritually dead case-terminations
are, in their signification of space, replaced, supported, or ex-
* Regarding the character, the natural foundation, and the finer gra-
dations in the use of the dual, and its diffusion into the different provinces
of language, we possess a talented inquiry, by W. von Humboldt, in the
Transactions of the Academy for the year 1827 ; and some which have been
published by Diimmler.
plained by prepositions, and in their personal signification by

the article.
116. Before we describe the formation of cases in the
order in which the Sanskrit Grammarians dispose them,
it appears desirable to give the different final sounds of
the nominal bases with which the case-suffixes unite them-
selves, as well as to point out the mode in which the cognate
languages are in this respect related to one another. The
three primary vowels (a, i, u) occur in Sanskrit, both short
and long, at the end of nominal bases; thus, Wa, $i, Fu;
ward, $7, %a%. To the short a, always masculine or neuter,
never feminine, a, corresponds in Zend and Lithuanian, and
also in German, where, however, even in the Gothic (in /
Grimm’s first strong declension), especially in substantives,
it is only sparingly retained: in more modern dialects it i
commonly supplanted by a more recent u ore. In Greek,
the corresponding termination is the 0 of the second declen-
sion (e.g. in Adyo-s): and o was also the termination of the
- Latin noun in ancient times; but in the classic period, al-
though sometimes retained, it was commonly changed to «
in the nom. and accus. sing. (of the second declension). An
old a, however, is still left in cola, gena, cida, at the end of
compounds, where, however, from the want of other ana-
logies, it is used in declension similarly to the feminine
[G. Ed. p.188.] originally long a, on which account the
nominative is written, not colas, genas, cidas, but cola, &c.
The Grecian masculines of the first declension in 4-s,* with the
n-s which has proceeded therefrom, must likewise, accord-
ing to their origin, be compared with the Sanskrit mas-
culine short a, to which, in regard of quality and preserva-
tion of the nominative sign, they have remained faithful,
while the o of the second declension has preserved its old
original brevity. Their identity with bases in o is excel-
lently shewn by the genitive in_ov, which does not at all
* Cf. p. 1294. 1. 20. G. Ed,

FORMATION OF CASES, 129
suit a theme in a or 4; and further, from such compounds

as LAy-s, ToudorpiBy-s, in which the vowel that has
been added to the roots HQA and TPIB supplies the place
of the Sanskrit a in similar compounds for which, in Greek, o
usually stands.
117. To the short i, which occurs in the three genders,
the same vowel corresponds in the cognate languages. In
German it is to be looked for in Grimm's fourth strong
declension, which I shall make the second; where, how-
ever, from the destructive alterations of time, it becomes
nearly as hard as the a of the first declension. In Latin,
i is interchanged with e; hence facile for facili, mare
for mari, Sanskrit aft vdri, “water.” In Greek, before
vowels the « is generally weakened to the unorganic e. The
short u also shews itself in Sanskrit in the three genders,
as in Greek v, and u in Gothic, where it distinguishes itself
from the a and i in that it is retained as well before
the s of the nominative as in the uninflected accusative.
In Latin the corresponding letter is the u of the fourth
declension. i
118. The long vowels (4, # 4) belong, in Sanskrit, prin-
cipally to the feminine (see §. 113.), are never found in thie
neuter, and occur in the masculine very rarely. In Zend
the long final a has generally been shortened in polysyllabic
words; as it has in Gothic, in which bases [G. Ed. p. 139.]
in 6 correspond (§. 69.) to the Sanskrit feminine bases in 4,
and the 6 in the uninflected nom. and accus. sing. is shortened
to a, with the exception of the monosyllabic forms sé, “she,”
“this,” Sanskrit at sé, Zend h4; hvé, “ which?’ Sanskrit
and Zend k4. The Latin, also, in the uninflected nom. and
voe., has shortened the old feminine long a; but the Lithu-
anian has, in the nom., maintained the original length. In
Greek, the Doric & approaches most nearly to the Sanskrit
feminine = 4, which the common dialect has sometimes
preserved, sometimes shortened, sometimes transformed
into 7.
R
119. The long @ appears, in Sanskrit, most frequently

as a characteristic addition in the formation of feminine
bases, thus, the feminine base agat mahati (magna)
springs from agq mahat. The same holds good in Zend.
Moreover, the feminine character 7 has been preserved
most strictly in Lithuanian, where, for example, in the
part. pres. and fut. an i is added to the old participial
suffix ant, and ésant-i, “the existing,” bé-sent-i, “that
that shall be,” correspond to the Sanskrit eat sat-¢ (for
asati or asanti), ufawat bhav-i-shyant?. In Greek and
Latin this feminine long i has become incapable of declen-
sion; and where it has still left traces, there a later un-
organic affix has become the bearer of the case-termina-
tions. This affix is, in Greek, either a or 6; in Latin, ¢,
Thus, deta corresponds to the Sanskrit wigt swidw-é,
from wg swidu, “sweet”; -zpia, -7p1d, e.g. dpynoTpa,
( Anorpis, Anorpid-os, to the Sanskrit 4 tri, e.g. afeat janitré,
“genitress,” to which the Latin genitri-c-s, genitré-c-is, cor-
responds; while in the Greek yevére:pa, and similar forma-
tions, the old feminine i is forced back a syllable. This
[G. Ed. p.140.] analogy is followed by péAava, raAawa,
Tépeiva, and substantive derivations, as Téxrava, Ad&Kawva.
In Gepatava, Aéarva, the base of the primitive is, as in the
nom. masc., shortened by a7. In Oécuva, AdKauva, it is to be
assumed that the proper primitive in v or vt has been lost,
or that these are formations of a different kind, and corre-
spond to the rather isolated word in Sanskrit great In-
drdni, as the wife of Indra, as derived from g¢ Indra, is
termed. The cases where the feminine iis solely represented
by a are essentially limited to feminine derivatives from
forms in vt, where 7 passes into o: the preceding v, however,
is replaced by v or 4, or the mere lengthening of the pre-
ceding vowel, or it is assimilated to the o: )
hence, oug-a, eic-a, edc-a, (ao-a", ito-a
for ovT-a, eVvT-a, evT-4, QVT-a, vUvT-a,
* In Doric subsequent and original ac-a.
FORMATION OF CASES, 13]
To this analogy belong, moreover, the feminine substantives,

like 6aAacoa, Baciiioca, pélicoa, which J. Grimm (IL. 328.)
very correctly, in my opinion, compares with forms like
xapi-ecca, weAtd-eooa, and explains the double ¢ by gemi-
nation or assimilation. The feminine formations by a
simple a instead of the original « are most corrupt, and,
relatively, the most recent; and herein the Greek is not
supported by any of the cognate languages. The Latin,
its twin-sister, which otherwise runs parallel to it, leaves,
in the part. pres. and other adjective bases terminating
with a consonant, the feminine undistinguished from the
masculine through all the cases, since it has no longer the
power of declining the old 2.
120. The German, too, can no longer fully decline the
old feminine 7; and the Gothic, by a foreign affix, intro-
duces it into the 6 declension, but in the singular of sub-
stantives shortens the syllable yé in the ([G. Ed p. 141.]
uninflected nominative and vocative to i,in the adjective to ya.
More commonly, however, the old bases in ¢ are introduced,
by the frequently employed affix of an n, into the so-called weak
declension; and as ¢ in Gothic is denoted by ei, so to the
Sanskrit feminine participial bases in wat anti, and to the
fem. comparative bases in $qat zyasi, correspond the forms
ndein, izein, regarding the nominative of which refer to §, 142.
121 The long u (d#) appears, in Sanskrit, rather seldom
at the end of primary forms, and is for the most part
feminine. The words most in use are qy vadhi, “a wife,”
4 bhi, “earth,” wy swaSri, “ mother-in-law ” (socrus), 4 bhri,
“eyebrow.” To the latter corresponds é¢pus, likewise with
the long v, the declension of which, however, is not different
from thatofthe short v; while in Sanskrit the long wuis distin-
guished from the short feminine u in the same way as $7
from ¥i. But few monosyllabic primary forms end, in
Sanskrit, with diphthongs, not any at all with-eé; with 2 di
(from 4+4i, see §. 2.) onlytrai, masc. “thing,” “riches”; in
K 2
132 FORMATION OF CASES.
the nom. irregularly t1a_rd-s for te rdi-s. In this is recog-

nised the Latin re-s. Still I do not believe that Latin bases
in @ should therefore be looked upon as corresponding to
the Sanskrit @ 4i; for, in the first place, the Latin é corre-
sponds elsewhere to the Sanskrit vé (from a4+i), never to
di; secondly, the connection of the é of the fifth declension
with the originally long a of the first is not to be mis-
taken (to which it bears the same relation that the Ionic
n does to the Doric @), for many words with the same mean- \
ing belong to the A and E declension; and, for example,

a suffix which is employed for the formation of abstracts
from adjectives is sounded as well ti@ as tia (planitie-s,
[G. Ed. p.142.] planitia, canitie-s, canitia); and ié-s, and ia,
in the formation of primitive and derivative words—like
effigie-s, effiyia, pauperie-s, pauperia—are clearly one and the
same suffix, identical with the Sanskrit at yd, which is used
for the same purpose, and the Greek fa, Ionic ij. Let us now
consider the objections which are opposed to the original
identity of the feminine @ and a. The most weighty is
the s in the nom, sing. and pl.: és, é-s for @, ei, as musa,
muse (musai), Kepady, kepadai. As regards the s in the
singular, it is, if the identity with the first declension be
authentic, very remarkable; and forms like species, canities,
seem to be true lingual patriarchs: for the Sanskrit, like the
Zeud, Greek, Gothic, Lithuanian, exhibits the absence of the
nominative sign in the corresponding feminine bases in a.
I have, however, never considered as original the aban-
donment of the nominative sign, and the complete equal-
ization with the primary form in qat sutd, “ daughter,” and
similar words, although it has appeared to me as losing
itself very deeply in far-distant ages. The Latin, how-
ever, in some other points of Grammar, shews greater
antiquity than the Sanskrit and Greek, as, for example
(to confine the present instance to the nominative case),
participial nominatives, like amans, legens, are better and
FORMATION OF CASES. 133
older forms than the Sanskrit and Greek, like q=_tudan,

Aéyav, TOeis, because they have preserved the nomina-
tive s together with the nasal, and therein stand on
the same footing with Zend forms, like sy»ass davans,
“being.” I cannot, therefore, find, in the retention of
the nominative sign in the fifth declension, any decisive
argument against its original identity with the first. We
will treat hereafter of the s of the nominative plural. In
the genitive singular the common form ei answers to deae
(deai), the more rare, however, and better, in és to familias.
Schneider searches, but fortunately without [G. Ed p. 143.]
success, for genitives like die-is: we require them as little,
perhaps, as a familia-is, Let dies be written with Greek
letters d:n-s, and then, perhaps, a die-is will be as little re-
qnired as a dixy-os. Although a few bases of the third de-
clension, by rejecting a consonant or an entire syllable, have
passed into the fifth declension, we will not therefore infer
that all bases in e have arisen from such an abbreviation, If
QUIET, after rejecting the ¢, could be declined according to
the fifth declension, then must there necessarily have for-
merly been a fifth, i.e. there must have been bases in @,
otherwise from QUIET could only have come QUII (quies,
quiis, like cedes); i.e. in spite of the rejection of the ¢
it must have continued in the third declension. The connec-
tion between ré-s and the abovementioned Sanskrit ¥ rdi is,
in my opinion, to be arrived at through the irregular
nominative wa rd-s; and according to this re-s would be
supported on an old d@: it answers to 7a rd-s as ré-bus to
wnaa_ré-bhyas, and as in Greek y7j-v to the Sanskrit mm
dm, “terram,” which, in the remaining cases, has 2 gé for
its base. In Lithuanian there are feminine primary forms
in e (Ruhig’s third declension) which resemble the Greek 7
in the suppression of the singular nominative sign, but in the
nominative plural in e-sapproach more closely the Latin
in é,
122, Primary forms in wt 6 are rare in Sanskrit: the

only ones known to me are gt dyé, “ heaven,” and mH gd:
the former is feminine, and properly proceeds from fey div
(a radical word from fea div, “to shine”) by the vocali-
zation of the qv, after which the vowel ¥i becomes its
semi-vowelgyy. In the accusative the 6 bases change this
diphthong into 4d. To the @ thus obtained in atq dyd-m,
[G. Ed. p. 144.] 711% gd-m, corresponds the Latin e of die-m,
the Greek y, Doric a, of y4-v, ya-v: the Latin e, however, is
rendered short by the influence of the final m: the original
language requires dié-m. In Sanskrit, also, from f¢q div,
“to shine,” are derived appellations of day; as on the other
side, in Latin, those for the heaven—divum, sub divo, sub dio
—viz. feat divd, as an adverb, “by day,” and used as a
primary form at the beginning of compounds; and also
feva divasa, masc., and g dyu, neuter (a contraction from
div), which latter signifies both “day” and “heaven.”
To @ dyu answers, after rejecting the d (as viginti for
dviginti), the Latin Ju of Ju-piter, “heavens-lord or
father”: the oblique cases Jov-is, Jov-i, Jov-em answer
better to the broader theme @t dyé, whence the dative
aa dyav-é, and the locat. afa dyav-i. The Djovis, moreover,
furnished by Varro, deserves mention, as that which keeps
most faithfully to the ancient form. The Grecian Zevg sig-
nifies, therefore, in accordance with its origin primarily,
“heaven”: I form its relation to at dyé thus, that after
dropping the ¢d the following semi-vowel q y became
¢(§. 19.). The oblique cases, on the contrary (A:ds, Axi, &c.),
belong to the Sanskrit qdyu, and must originally have
had a digamma, proceeding by the natural law of sound from
u, after which change the semi-vowel j7 must have become
a vowel. Azdc has the same relation to A:Fés, that, in Latin,
sub dio has to sub divo.
123. Let us now consider the second of the abovemen-
tioned primary forms in 4, viz. mY gd. It has several
meazings; but the most common are “bull,” as masculine,

and “cow” and “earth” as feminine. Both significations
have in Zend, as in Greek, dividcd themselves into two
forms. The Greek has preserved for the meaning “earth ~_
the old guttural. With regard to the vowel, -y7, ya follows the
example of the Indian accusative, where, as has been already
remarked, mra_gdm (yyv) stands for gé-m [G. Ed. p. 145.]
or gav-am. For the meaning “ox” the Greek has preserved
the old diphthong—{for, for wté—axu may very well be
expected, according to §. 4., ov) —but has exchanged the guttu-
ral medials for labials, as, p. 122 G. Ed. Bi@y: for amta
jaydmi. The base BOY before vowels must originally have
become BOF ; thus, in the dative, BoF-i would answer te the
Sanskrit locat. af gav-i, and the Latin dative bov-i; but in
the present state of the language the middle digamm:
between two vowels has always been dropped; and there
is not, as with the initial digamma, the medium of metre
for replacing it in the oldest writings. Only theory and
comparative grammar can decide here. The Latin has,
in the word bd-s, changed the vowels (a + u)—(which were
originally of different kinds, but have been united into a
diphthong)—into a homogeneous mass (cf. §. 4.), the nature
of whose contraction, however, discloses itself before vowel
inflexions, since the u-half of BO becomes v, and the short a
is resolved into the ferm of a short o; thus, bov-i answers
to the Sanskrit locat. 1fqgav-i. The Zend for the meaning
“earth” has changed the guttural of the word under dis-
cussion into z, and gives in the nominative gusg zdo for
aaug zds (8. 56”.), in the accusative ys zarim (§.61.): Tam
not able to adduce other cases. For the meaning “ox”
the guttural has remained in Zend, and the nominative
is then +».wW gdu-s or ~wEwe gdo-s.
124. I know only two words in Sanskrit which terminate
in Wi du—t néu, “ship,” -and 7@t gldu, “ moon”: the former
has navigated very far on the oceau of our wide province of
° 136 FORMATION OF CASES.
language, without, however, in Sanskrit, having arrived at a

secure etymological haven. I believe # ndw to be an abbre-
viation of snau (cf. péw, pedw, ruo, with @ sru, p. 125 G. ed.),
[G. Ed. p. 146.] and that it therefore proceeds from the root
at snd, “to bathe,” which originally, perhaps, may also have
meant “to swim,” and with which véw, véw, na-to, appear to
be connected. tndu would consequently be a radical word;
and in regard to the vowel would stand for nd, according to
the analogy of zat daddu (dedi, dedit) for dadd, from dadd-a.
As a, according to §. 6., is a grave vowel, the Greek cannot
represent the Sanskrit Vriddhi-diphthong ¥ du better than
by av, while wt 6 (from short a+u) is commonly repre-
sented by evor ov. Hence #4 ndu-s and vad-¢ correspond
as exactly as possible; the v of NAY, however, like that
of BOY, has maintained itself only before consonants; and
the digamma, which replaces it, is lost before vowel in-
flexions; vj-ec, va-es, are from vaF-es (Sansk. ATR név-as),
as d-es from PoF-es. The Latin has given this word a
foreign addition, and uses navi-s, navi-bus, for nau-s, nau-bus.*
As the semi-vowel v is easily hardened to a guttural
(§. 19.), we have here also, for nau, ndv-am, a sister form
in our Nachen, Old High German naccho, “ship,” gen, dat.
nacchin.
125. We pass over to the consonants: of these, n, ¢, s,
and r appear in Sanskrit most frequently at the end of
primary forms; all other consonants occur only in radical
words, which are rare, and in some nominal bases of uncer-
tain origin. We consider next the more rare or radical
consonants. Of gutturals (k, kh, g, gh) we find none at
* Thus in German an i has been added to the above-mentioned xt

g6, which, however, according to §. 117., is suppressed, together with the
case sign in Old High German; hence chuo, “cow,” gen. chuoi, where
the ¢ does not belong to the case designation, but to the here uninflected
base.
the end of the nominal bases most in use; in Greek and

Latin, on the contrary, they are of frequent occurrence;
e is in Latin both radical and derivative, [G. Ed. p. 147.]
g only radical—DUC, VORAC, EDAC, LEG. In Greek,
K, X; and +y are only radical, or occur in words of unknown
origin, as @PIK, KOPAK, ONYX (Sanskrit nakha), @AOT.
Of the palatals, ch and j in Sanskrit occur most frequently in
ara_vdch, “speech, voice” (VOC, ON); aH rdj, “ king,” the
latter only at the end of compounds; way asrij, “ blood”
(sanguis): in Zend we have y>7gdruj, f., as name of an
evil demon, probably from the Sanskrit root ¢g druh, “ to
hate.” Of the two classes of the T-sound, the first, or
lingual (z #, &c.), is not used at the end of nominal bases;
and therefore the second, dental, or proper T-class, is so
much the more frequently employed. Still ¢ d, ¥ dh, occur
only in radical words, and therefore seldom; ¥y th perhaps
only in 4 path, as the secondary theme of ufaq pathin,
“way”; nom. Ware panthds, from qa panthas, which I
think I again recognise in the Latin PONT, pons. Other
examples are, W@ ad, “eating,” at the end of compounds,
and ay yudh, f., “strife.” The letter qé is so much the
more common, that several of the most frequently employed
suffixes end with it, as that of the part. pres. in Wa a¢ or
aq ant, Greek and Latin nt. The Greek, besides 7, ex-
hibits also ¢ and 6 at the end of primary forms which are not
radical; still KOPYO and ‘OPNIO appear to me to be pro-
perly compounds, and to contain the roots OH, OE (the vowel
being dropped) as their last member; and according to this,
KOPYO would properly mean “ what is placed on the head”;
so in Sanskrit, ye sarad, “autumn,” “rainy season,”
which Grammarians explain by a suffix ad, in my opinion
means nothing but “ water giving,” and contains the root
a dd, “to give,” with 4 suppressed. ’OPNIO finds in
Greek itself no etymology: the Sanskrit offers for its expla-
nation
"tfw arani (according to the pronunciation of Ben-
gal, oroni), “wood”; and if dpv is con- ([G. Ed. p. 148.]
138 EORMATION OF CASES.
nected therewith, we may refer to 6éw, “to run,” in respect

to the 6: “bird” therefore would derive its name from its
going in the wood; while in Sanskrit, from its passage
through the air, it is called, among other names, faga
vtha-ga. Regarding the later origin of the 3 in feminine
bases in :d, an account is given in §.119.; that is to say,
patronymics in «0 may be compared with Sanskrit ones in ¢,
e.g. Rat bhaimé, “the daughter of Bhima. Probably, too,
the d in feminine patronymics in a@ is a later addition; they
spring, like those in :d, not from their masculines, but directly
from the primary word of the masculine, and, in my opinion,
stand in sisterly, not in filial connection with them. In
Latin, d appears as a more modern affix in the base PECUD,
which the Sanskrit, Zend, and Gothic terminate with u
(Sans.-Zend, pasu, Goth. fathu). In Gothic, primary forms
with a final 7-sound are chiefly limited to the part. pres.,
where the old ¢ appears changed into d, which remains
without extraneous addition: there only, however, where
the form stands substantively; otherwise, with the excep-
tion of the nominative, it is conducted by the affix an
into a more current province of declension. The more
modern German dialects under no circumstances leave the
old T-sound without a foreign addition commixed with
the base. In Lithuanian the participial suffix ant, in re-
gard of the nom. sing. ans for ants, rests exactly upon
the Latin and Zend step, which extends beyond the San-
skrit; but in most of the remaining cases the Lithuanian
cannot decline any more consonants, i.e. cannot unite
them with pure case terminations, but transports them
always, by a more modern affix, into a vowel-declension;
and, indeed, to the participial suffix ant is added the
(G. Ed. p.149.]_ syllable ia, by the influence of which
the ¢ experiences the euphonic transformation into ch
(= fsch*). The nasal of this dental T-class, viz. the
* This sound is expressed by ez, as in Mielcke’s edition of Ruhig’s
Grammar. :
proper n, belongs to those consonants which occur most

frequently at the end of nominal bases. In the German
all the words of Grimm’s weak declension like the San-
skrit, and the masculine and feminine in Latin, reject in
the nominative the n of the base, and thereby have a
vowel termination. The Lithuanian presents the same
appearance in the nominative, but in most of the oblique
cases adds to a base in en sometimes ia, sometimes a
simple 7.
126. Primary forms with a final labial, including the
nasal (m) of this organ, appear in Sanskrit only in naked
roots, as the last member of compounds, and here, too,
but seldom. In isolated use, however, we have Wy ap
(probably from the root "Tq dp, “to take in,” “to compre-
hend”), “water,” which is used only in the plural; in
Zend, however, in the singular also.* In Greek and Latin,
also, bases in p, b, ¢, are either evidently radical, or of
unknown origin, with probably radical letters at the end;
or in Latin they have suppressed, in the nominative, a
vowel belonging to the base; and so,as in [G.- Ed. p. 150.]
German, the first and fourth strong declensions, according
to Grimm, have only the appearance of a base terminating
with a consonant. Of this kind is plebs, from plebis; to
explain which it is not requisite to turn, with Voss, to
the Greek mA7G0s: one must keep to the Latin root PLE.
The derivative bis, bés, | explain like bus, bundus, bilis, bam,
* The Latin adds an a to this old consonantal base, and thus arises,
according to the frequent interchange of p with-qu (cf. quinque with
way panchan), aqua; on the other hand, am-nis rests on the form ap,
as somnus for sopnus, and cepuvds, for ceBvds, in analogy with a Sanskri‘
euphonic law (Gramm, Crit. r.58.). The Sanskrit has from the same
root another neuter, ibt§ Gpas, in which we recognise the Latin @quor,
which therefore would not proceed from @quus, but is transferred from
the waves, or the mirror of the'sea, to other things of a similar nature.
In Greek, appés appears to belong to the same origin.
bo (amabam, -bo), as from the root FU. “to be,” which, like
FER, often changes the B in its middle into F (§. 18.).
Without appealing to the cognate languages, it is difficult, in
Latin, to distinguish those bases which truly and origi-
nally terminate in a consonant from those which only ap-
pear to do so; for the declension in i has clearly operated
on the consonantal declension, and introduced an 7 into dif-
ferent places in which it is impossible it could have stood
originally. In the dative and ablative plural, the i of forms
like amantibus, vocibus, admits of being explained as a con-
junctive vowel, for facilitating the affix; it is, however
in my opinion, more correct to say that the bases VOC,
AMANT, &c., because they could not unite with bus, have,
in the present state of the Latin language, been lengthened
to VOCI, AMANTI; so that we ought to divide voci-bus,
amanti-bus, just as at §. 125. it was said of the Lithuanian,
that in most cases it extends its participial bases in ant to
anchia (euphonic for antia). This view of forms like amanti-
-bus is proved to be the more probable, in that in the geni-
tive plural also before um, as before the a of neuters, an é
frequently finds its place, without its being possible to say
that in amanti-um, amanti-a, the i would be necessary to
facilitate the annexation of the ending. On the other
hand, juveni-s, cani-s, forming the genitives canu-m, juven-
-um, remind us of older bases in n; as in Sanskrit aa
éwan, “a dog” (abbreviated ya sun), and gaq yuvan,
“ young” (abbreviated W_ yin), in Greek xiwv, abbreviated
_ [G. Ed. p. 151.] KYN, really close their theme with n. The
German resembles the Latin in this point, that for the
convenience of declension it has added an i to several nume-
rals, whose theme originally terminated with a conso-
nant; thus, in Gothic, from FIDVORI (Sanskrit wat
chatur, in the strong cases §. 129. =raqTt chatwdr) comes the
dative Hanere The themes ana saptan, ‘“seven,” Aa
navan, “ nine,” awa dasan, “ ten,” by the addition of an i
in Old High German mould themselves to SIBUNI,

NIUNI, ZEHANT; which forms, at the same time, pass as
masculine nominatives, as these cases, in Old High German,
have lost the case-suffix s. The corresponding Gothic
nominatives, if they occurred, would be sibunei-s, niunei-s,
taihunei-s. More on this point hereafter.
127. Of the semi-vowels (y, 7, 4, v), I have never
found in Sanskrit q y and = Zl at the end of bases, and
av only in the word f¢q div, before mentioned, which
contracts itself in several cases to wt dyé and q dyu. On
the other hand, < occurs very frequently, especially in
words which are formed by the suffix WW tar,* to which, in
the cognate languages, likewise correspond bases in r.
Moreover, r in Latin appears frequently as an alteration
of an original s, as, in the comparative suffix ior (San-
skrit faa fyas); and, further, as an abbreviation of ri-s,
re, as l for li-s,/e; or, in the second declension, as abbre-
viated from ru-s; as in Gothic, vair, “man,” for vair(a)s,
belongs to bases in a (§. 116.). In Greek “AA appears as a
consonantal base ;but in contrast with the ([G. Ed. p. 152.]
Sanskrit «fee salila, “water,” GA-¢ appears abbreviated
exactly in the same manner as péya-¢ from peyados.
128, Of the Sanskrit sibilants, the two first (q 3, ¥ sh),
as also the = h, are found only in radical words, and there-
fore seldom; # s, on the contrary, concludes some very
common suffixes used in the formation of words, as Wa as,
which forms principally neuters, e.g. wae téjas, “splendour,”
“strength,” from fax tij, “to sharpen.” The Greek ap-
pears to be without bases in =; this, however, proceeds
from the following reason, that this sibilant between two
* Bases in "¥{ ar inseveral cases, and in the primary

form also at the
beginning of compounds, contract the syllable Bq arto |] ri; and this
‘q Ti is regarded by the Grammarians as their proper final sound. (§. 1.)
vowels, especially in the last syllable, is usually rejected,

hence, neuters like pévos, yévog (from MENES. TENE.
with change of the e into 0), form in the genitive péveos,
yéveos, for péveros, yéveros. The ¢ of the nominative,
however, belongs, as I have already elsewhere remarked,
to the base, and not to the case designation, as neuters
have no ¢ in the nominative. In the dative plural, how-
ever, in the old epic language, the 2, as it did not stand
between two vowels, maintained itself; hence tetyeo-c1,
épeo-o1; so likewise in compounds, like caxés-maAos, TeAes-
épos, in which it would be wrong to assume the annexation
of a = to the vowel of the base. In yipas, yjpa-os, for
yijpac-os, after restoring the = of the base, the form of word
answers exactly to the Sanskrit wra_jaras, “age,” although
the Indian form is not neuter, but feminine. In Lithua-
nian, another remarkable remnant of the Sanskrit suffixes
terminating with s has been preserved, viz. in the partic.
perf., in the oblique cases of which us corresponds to the
Sanskrit 74 ush (euphonic for sa@ us) of the weakest cases
(§. 130,); still, in Lithuanian, on account of the above-
noticed incapacity for the declension of the consonants, the
old us is conducted, as in other similar cases, by the subse-
quent addition of ia, a or i, partly into the a, partly into the
[G. Ed. p. 1583] ideclension; and only the nominative and
the vocative, which is the same with it, belong, in the singular,
to the consonantal declension.
129. The Sanskrit and Zend have eight cases, viz. be-
sides those which exist in Latin, an instrumental and a
locative. These two cases exist also in Lithuanian;
Ruhig calls the former the instrumental ablative, the latter
the local ablative ;in Lithuanian, however, the proper abla-
tive—which in Sanskrit expresses the relation “ whence ?”—
is wanting. With reference to the primary form, which
in Sanskrit does not remain the same in all words, or
suffixes used in the formation of words through all the cases,

a division of the cases into strong and weak is desirable
for this language. The strong cases are the nominative,
accusative, and vocative of the three numbers, with excep-
tion of the accusative plural, which, together with all the
other cases, is weak. Where a double or triple formation
of the primary form exists, there, with surprising regu-
larity, the cases which have been designated as strong
always exhibit the fullest form of the theme, which, from
a comparison of languages, is proved to be the original
one; while the other cases exhibit a weakened form of it,
which appears also in the beginning of compounds, and
hence is represented by the native Grammarians, accord-
ing to §. 122., as the proper primary form. The pres.
part. may serve as an example: it forms the strong cases
with the suffix ant, but in the weak cases and in the be-
ginning of compounds rejects n, which is retained by the
cognate European languages, as also, for the most part,
by Zend; so that wq at is given as the suffix of this bar
ticiplein preference to wet_ant. The root az tud, “ to vex,”
e.g. exhibits in the participle mentioned the form wart tu-
-dant as the strong and original theme (cf. tundent-em),
and qem tudut as the weak theme; hence the masculine
is declined, [G. Ed. p. 154.]
STRONG CASES, WEAK CASES.
Singular : Nom. Voce. i a

Acc. gewa tudantam «ewes
ESE ER ae gent fudata.
ee eee Gen tudaté.
SS ee were tudatas.
Ee are ware tudutas.
as wafa tudati.
Dual: Nom. Ace. Voc. qeatiudantfu .......
Instr. Dat. Abl -..... gee tudadbhydm.
en Ete SS. RG aqate tudatés,
STRONG CASES. WEAK CASES.
Plural: Nom. Voc. . . awa tudantas . . . s - +

Ace. " - 2 2 + +) Weta tudatas.
Instr. - 2 ee «+ © Tafa cudadbhis.
Dat. Abl. oe ee ee) ERA tudadbhyas.
Gen. CLL |tudatdm.
Loc. see eee | Gere fudatsu.
130. Where three formations of the primary form per-
vade the declension of a word or a suffix, the weakest form
of the theme there occurs in those weak cases whose termina-
tions begin with a vowel, the middle form before those case-
suffixes which commence with a consonant. This rule makes
a division of the cases into strong, weaker or middle, and
weakest, desirable. (See Gramm. Crit. r. 185.)
131. In suffixes used in the formation of words, which in
Sanskrit separate into different forms, the Zend usually carries
the strong form through all the cases; for instance, the part.
pres. retains the nasal in most of the cases, which in Sanskrit
[G. Ed. p. 155.] proceed from the weakened theme. Words,
however, are not wanting which follow the theory of the
Sanskrit gradations of form. Thus, the Sanskrit base
aq swan, “hound,” which in the weakest cases is con-
tracted to Yq sun, appears in Zend likewise in a double
form, and presents the weak genitive sién-6 over against
the strong nominative and accusative spd, spdn-ém, San-
skrit vat swd, vaTay swdnam (§. 50.). The base ap, “ water,”
which, in Sanskrit,-in the strong cases has a long 4, but
is not used in the singular, forms in the Zend the strong
sing. nom. asday afs (§. 40.), accus. ¢gdau dpém; on the
other hand, ap-é, “ of the water,” ap-at, “from the water,” &.*
* This word occurs in the Codex of the V. S., edited by Burnouf, very
frequently, and mostly with that quantity of the initial @ which is
required by the theory; so that where that is not the case it can only
be imputed to an error in writing.
NOMINATIVE SINGULAR. 145
In the plural, where the Zend very frequently makes the

nominative and accusative the same, confusion has, for
this reason, crept in; and the weak Uypss séné, “ canes,”
is found for Lyywsd9 Spano in the nominative; and, on the
other hand, the strong Lous dpé, in the nominative as well
as in the accusative.*
132. The Greek, in the declension of xiwv, has limited the
strong form to the nom. and voc. sing.: in ([G. Ed. p. 156.]
some cognate words in p, however, in accordance with the
Sanskrit, it has given the accusative also the strong form, in
which the Gothic agrees with it. Compare zaryjp, tatépa,
matep, matpi, with faa pitd, faata pitaram, fant pitar, fafa
puri (locat.); and the Gothic bréthar, as nom., accus., and
vocat., opposed to bréthrs, “ of the brother,” brdthr, “to the
brother,” with the Sanskrit wrat bhrdfd, TATA bhrataram,
wat bhrdtar, dative urs bhratré, locat. atfa bhrdtri. |Accord-
ing to the same principle in bases in an, in Gothic, the a in
the genitive and dative sing. is weakened to i (§. 140.); while
the nominative, accus., and vocat. retain the original a; e.g.
ahma, ahmin-s, ahmin, ahman, ahma, from 4HMAN, “spirit”
(§. 140.).
133. As regards the mode of combining the final vowels
of the primary forms with case-suffizes beginning with a
vowel, we must first draw attention to a phenomenon, which
is almost limited to the Sanskrit, and the diaiects which
* I have, however, found also Goa apé m the accusative; and am

therefore in doubt, whether in this word, owing to the facile exchange of
2s a and « d, the confusion has not originated in mere graphical over-
sights. Thus, V.S. p. 21, we find: BUIVays shy asp boas
PH Wy jaw ass Vasygys UI PWEA4G Gp? vanhuis vahistdo muzda-
dhitéo ashaonis dyésé, “aquas puras, optimas, ab Ormuzdo creatas, mundas
celebro”; and YdIws ¢5d3959 vispdo apd, “omnes aquas.” On the
other hand, in the page following: assxuges spuds pugs
PHPK yw Asus guss7asy>7> imde apes-tha zma:-cha urardos-cha dyese,
“ has aquasque terrasque arboresque ceicbro.’
L
approximate most nearly to it, as Pali and Prakrit, through

which, to avoid a hiatus, and to maintain pure the vowels
of the base and of the termination, a euphonic n is introduced.
This euphonic expedient cannot, in the extent in which it
exists in Sanskrit, belong to the original state of the lan-
guage ; otherwise it would not be almost entirely lost in the
cognate European dialects, and even in the Zend. We there-
fore regard it as a peculiarity of the dialect, which, after the
period of the division of languages, became the prevailing
one in India, and has raised itself to be the universal written
language in that country. It is necessary here to remark,
that the Véda language did not use the euphonic n so univer-
sally as the common Sanskrit; and together with wat
énd, Fat ind, Tat und, occur also Wat ayd, Fat iyd, Tat uyd.
The euphonic n is most frequently employed by the neuter
[G. Ed. p.157.] gender, less so by the masculine, and most
rarely by the feminine: the latter limits its use to the plural
genitive termination wTA dm, in which place it is intro-
duced by the Zend also, although not as indispensably re-
quisite. And it is remarkable, that precisely in this place
in Old High German, and other Old German dialects, an n
has been retained before the case-suffix; thus in Old High
German, ahé-n-é, “ aquarum,” from the feminine theme 4H O
(nom. aha). Besides the use of the euphonic n, there is fur-
ther to be remarked, in Sanskrit and Zend, the attachment of
Guna to the vowels of the base (§. 26.) in certain cases, to
which also the Gothic presents analogies.
SINGULAR.
NOMINATIVE.
134. Bases, of the masculine and feminine genders, end-
ing with a vowel have, in the Sanskrit family of languages,
(under the limitation of §. 137.) s as nominative-suffix, which
in Zend, after an a preceding it, always melts into u, and is
then contracted with the a to 4 (§. 2.), while this in Sanskrit
takes place only before sonant letters (§. 25.)* Examples

are given at §. 148. I find the origin of this case-designation
in the pronominal base @ sa, “he,” “this,” fem. atsd4; and
a convincing proof of this assertion is the fact, that the said
pronoun does not extend beyond the limits of the nom. mase.
and fem., but is replaced in the nom. neuter, and in the
oblique cases of the masculine, by at ta, and feminine at
ta regarding which more hereafter.
135. The Gothic suppresses a andi be- (G. Ed. p. 158.]
fore the case-suffix s, except in monosyllabic bases, where
this suppression is impossible. Hva-s, “who?” i-s, “he,” are
used, but vulf-s, “ wolf,” gast-s, “stranger,” for vulfa-s, gasti-s
(ef hosti-s, according to §.87.). In masculine substantive
bases in ja (ya), however, the final vowel is retained, only
weakened to z (§. 66.); e.g. haryi-s, “army.” If, however,
as is generally the case, the final syllable is preceded by a
long syllable, or by more than one, the ji (yi) is contracted
to ei (= 7, §. 70.); e.g. ondei-s, “end,” raginei-s, “counsel,”
for andyi-s, raginyi-s. This contraction extends also to the
genitive, which is in like manner denoted by s. To the
Gothic nominatives in yi-s correspond the Lithuanian, like
Atpirktoyi-s, “ Saviour,” “the i of which has likewise arisen
from an elder a.t I deduce this from the majority of the
oblique eases, which agree with those of the a bases.
Where, however, in Lithuanian, a consonant precedes
the final syllable ya, which is the more common case,
there the y is changed into the vowel i, and the follow-
ing i, which had arisen from a, is suppressed: hence,
yaunikki-s, “young man,” for yaunikkyi-s from yaunikkya-s.
Hereto correspond in Gothic all adjective bases in ya,
* B a: Gait HAsuld mama, “filius meus,” GAA Aq sutas tava, “fi-

lius tuus”’ (5. 22.).
+ Through the influence of the y, in accordance with a Zend law of
euphony (§. 42.).
- = Respecting the nom. e.g. of Gothic bases in ya, see p. 1309 G. Ed.,
~ Remark. <--,
L2
as midi-s “the middle” (man), for midyi-s from midya-s,

Sanskrit awa madhva-s, The Zend also, in the vocali-
zation* of the syllable ya, presents a remarkable analogy
to the Lithuanian and Gothic in contracting the syllable
ayy ya before a final ¢ m regularly to si, as also wh va
to 9 @ (§. 42.).
136. The High German has, up to our time, preserved
the old nominative sign in the changed form of r; never-
theless, as early es in the Old High German, in pronouns
and adjectives only, with a vowel termination of the base.
[G. Ed. p. 159.] The High German is, however, in this
point, superior to the Gothic in fulness, that in its a bases—
to which belong all strong adjectives—it has not suppressed
the vowel before the case-sign, but preserved it in the form
of e, which, in Old High German—as it appears. through
the influence of the r—is long, but only in polysyllabic,
not in monosyllabic forms. Thus, e.g. plint-ér, “ coecus,”
completes the Gothic blind-s for blinda-s; as to the Gothic
i-s, “he,” corresponds i-r; Middle and New High German
e-r. The Old Northern has likewise r as the nomina-
tive sign, and, in fact, everywhere where, in Gothic, s
stands. In the other dialects the nominative character is
entirely. lost,
137. Feminine Sanskrit bases in wd, and, with very
few exceptions, polysyllables in $ 7, together with wt strié,
“wife,” like the corresponding forms of the cognate lan-
guages, have lost the old nominative sign (with the exception
of the Latin é bases, see §. 121.), and give the pure base: the
cognate languages do the same, the base having been weak-
ened by the abbreviation of the final vowel. In Gothic, 6 be-
comes a (§. 69.); only sé, “this,” and hwé “which?” remain
unshortened, on account of their being monosyllabic, as in
Zend savy hd and 15 kd; while in polysyllabic forms the
* I have used vocalization and vocalize to express the change of a semi-

vowel t® its corresponding vowel,— Trans. ,
av4 is shortened. In Zend, 57 also is shortened, even in

the monosyllabic s%ow stri, “ wife,” see V. S. par. 136, (by
Olshausen), p. 28, where we read xs%ox stri-cha, “femi-
naque’; whilst elsewhere the appended .5 cha preserves
the original length of the vowel. Here, too, the Zend nomi-
natives in » é deserve to be mentioned, which seem very
similar to the Greek in 7; as wycled peréné, “ plena,” which
in the Vendidad occurs very often in relation to gu zdo,
“earth,” without my being able to remember that I have
found another case from ryehed perené. But from the
nom. 743955 kainé,*‘maid” (Sanskrit @ar [G. Ed. p. 160.]
kanyd), which is of frequent occurrence, I find the accus.
GFyII/59 kanyarim (V. S. p. 420); this furnishes the proof
that the » é in the nominative is generated by the eupho-
nic influence of the suppressed yy y (8. 42.). In wysProwusls
brdturyé, “ cousin,” and 7997570 tiiryé, “a relation in the
fourth degree” (V. S. p. 380), the yy y has remained; on
the other hand, in w5.wyjy nydhé “ grandmother,” the
dropping of a 9) y must be again assumed. We cannot
here refrain from conjecturing that the é@ also of the Latin
fifth declension, as with very few exceptions it is everywhere
preceded by an i, is likewise produced from @ by the in-
finence of this i; so that the Latin here stands in reversed
relation to the Greek, where « rejects the combination with
y, and preserves the original a (co¢/a).
138. Bases of the masculine and feminine genders which
terminate with a consonant, lose, in Sanskrit, according to
§.94., the nominative sign s; and if two consonants termi-
nate the base, then, according to the same law, the latter of
these also is lost. Hence, faa bibhrat, for faa bibhrat-s,
“the bearer”; wea tudan, for qert tudant-s “the vexer”;
are vdk (from arq vich, £), for arey vak-sh, “speech.”
The Zend, Greek, and Latin, in preserving the nominative
sign after consonants, stand in an older position than the
Sanskrit; Zend sada af-s (for dp-s, §.40.), “ water”;
msdees k“réfs, “ body”; 21963>%5 druc-s (from the base druj),

“a demon.” The Latin and Greek, where the final conso-
nant of the base will not combine with the s of the nomi-
native, prefer abandoning a portion of the base, as yépr¢ for
xaprt-s, comes for comit-s (cf.§.6.). The Latin, Zolic, and
Lithuanian agree remarkably with the Zend in this point,
[G. Ed. p.161.] that nt, in combination with s, gives the
form ns; thus amans, 7:Oévs, Lith. sukans (§. 10.), corre-
spond to the Zend yds» whs9 srdvayans, “the speaking”
(man).
139. A final n after a short vowel is, in Sanskrit, no
favourite combination of sound, although one not prohibited.
It is expelled from the theme in the first member of a
compound, eg. tags rdja-putra, “king’s son,” for TTT
rajan-pulra ; and it is rejected in the nominative also, and
a preceding short vowel is lengthened in masculines;
e.g. trat raja, “king,” from Tat rdjan, 1.3; ATA ndma,
“name,” from ata ndman, n.; wat dhani, m., ufa dhani, n.,
from afta dhanin, “rich.” The Zend in this agrees exactly
with the Sanskrit; but from the dislike to a long a at the
end, which has been before mentioned, omits the length-
ening of the vowel; e.g. spasms ashava, “the pure” (man),
from posmasmggas ashavan, M.; asGxe9ass chashma, “ eye,” from
psExpasys chashman, n. The Latin follows the Sanskrit in
the suppression of the n in the nominative, in the mas-
culine, and feminine, but not in the neuter: sermo,
sermon-is, actio, action-is; but nomen, not nome or nomo.
The root can at the end of compounds, refrains from
rejecting the n, probably in order not to weaken still more
this weak radical syllable; thus tubi-cen, fidi-cen, os-cen (see
§. 6.). Lien-is an abbreviation of lieni-s ; hence the reten-
tion of the n is not surprising. Pecten stands rather
isolated. In Sanskrit the naked roots also follow the prin-
ciple of the rejection of n; ¥a@ “slaying,” “smiting,”
nom. @1 hd, is, however, the only root in n which I have
met with so used. Wa swan “ hound,” nom. vat sw, which,

in the weakest cases, contracts its theme to Yq sun,
is of obscure origin. The Latin has extended the base
‘aq Swan, in the — by an unorganic addition,
to cani; so Wt yuvan, “young,” has become juvent
(cf. §. 126.). As regards the opposition ([G. Ed. p. 162.]
between o and i, by which, in several words—as homo, homin-
-is, arundo, arundin-is—the nominative is distinguished from
the oblique cases, this 0 appears to me a stronger vowel,
which compensates for the loss of the n, and therefore is
substituted for the weaker i; according to the same prin-
ciple by which, in Sanskrit, the nom. wat dhani;} comes
from ufqq dhanin; and, in Lithuanian, bases in en and un
give, in the nominative, i (=uo) for e or u. Thus,
from the bases 4K MEN, “stone,” SZUN, “hound,” come the
nominatives akmu, szu; as in Sanskrit, from the primary
forms of the same signification, Wya4q asman, WA swan,
have arisen WAT asmé and Ww swd. It does not follow that
homin-is has come from homon-is,t because the old language
hemo, hemonis, for homo. hominis; but mon and min are
cognate suffixes, signifying the same, and were originally
one, and therefore may be simultaneously affixed to one and
the same word.
140. The German language also rejects a final n of the
base in the nominative and in the neuter, in the accu-
* Although its quantity in the actual condition of the language is arbi-

trary, still it appears to have been originally long, and to imply a similar
contrast to the Greek nv, «v-os; wy, ov-os. For the rest it has been
already remarked, that between short vowels also exists a difference of
gravity (9. 6.).
+ In bases in Wy an the lengthening extends to all the strong cases,
with the exception of the vocat. sing.; thus, not merely Feel raja, “rex,”
but also Tatty rijdn-am, “regem,” TaTaa rajdnas, “reges.””
t I now prefer taking thei of omin-is, &c., as the weakening of the 0
of homo. The relation resembles that of Gothic forms like ahmin-is,
ahmin, to the nom. and ace. ahma, ahman, which preserve the original vowel.
sative also, like Sanskrit. In Gothic, in the masculine

and neuter—where alone, in my opinion, the n has an
old and original position—an a always precedes the n.
There are, that is to say, only bases in an, none in in and
un; the latter termination is foreign to the Sanskrit also.
[G. Ed. p. 163.] The a, however, is weakened to i in the
genitive and dative (see §. 132.); while in Sanskrit, in these
cases, as especially in the weakest cases (§. 130.), it is entirely
dropped.* Among masculine bases. in an, in Gothic, exist
several words, in which an is the whole derivative-suffix,
and which therefore correspond to the Sansk. wrt rdj-an,
“king,” as “ruler.” Thus 4H-AN, “spirit,”as “thinker”
(ah-ya, “I think”), STAU-AN, “Judge” (stau-ya, “I judge”),
whence the nominatives aha, staua. There are also, as in
Sanskrit, some masculine formations in man; as, AHMAN,
“spirit,” nom. ahma, with which perhaps the Sansk. wrava
dtman, “soul,” nom. wrt dimd, is connected; in case this
stands for dh-man, and comes from a lost root wre dh,
“to think,’{ where it is to be remembered that also the
root q@ nah, “to bind,” has, in several places, changed its
hintot. The Gothic MJZH-MAN, nom. milh-ma, “cloud,”
appears to have sprung from the Sanskrit root mih, by the
addition of an 7, whence, remarkably enough, by the suffix
a, and by exchanging the ¢ h for q gh, arises the nomi-
nal base #y mégha, “cloud.” In Latin ming-o answers to
mih, and in Greek 6-ury-éw; the meaning is in the
three languages the same.
141. Neuter bases in an, after rejecting the n, lengthen,
in Gothic, the preceding a to 6, in the nominative, accusa-
* In case two consonants do not precede the termination Wa an;

C9: ATAAA dtman-as, not dtmn-as, but aaa ndmn-as, not ndman-as,
* nominis,”
+ Perhaps identical with the actually-occurring WE ah, **to speak,” as
man, “to think,” in Zend means also “ to speak”; whence 5G, ¢
manthra, “speech,” and in Gothic MUN-THA, nom. munths, “ mouth”
§.66..
tive, and vocative, which sound the same; ([G. Ed. p. 164.]
so that in these cases the Gothic neuter follows the theory of
the strong cases (§.129.), which the Sanskrit neuter obeys
only in the nom., accus., and vocat. plural, where, for ex-
ample, wratft chatwdr-i, “four,” with a strong theme, is
opposed to the weak cases like wafita chaturbhis (instr.),
‘maa chaturbhyas. The a, also, of neuter bases in an is
lengthened in the nominative, accusative, and vocative plural
in Sanskrit, and in Gothic; and hence wratfa némén-i,
Gothic namén-a, run parallel to one another. However, in
Gothic namn-a also exists, according to the theory of the
Sanskrit weakest cases (§. 130.), whence proceeds the plural
genitive Aram ndmn-dm, “ nominum”™; while the Gothic
namén-é has permitted itself to be led astray by the example
of the svrong cases, and would be better written namn-é or
namin-é.
142. In the feminine declension in German I can find
no original bases in n, as also in Sanskrit there exist no
feminines in an or in; but feminine bases are first formed
by the addition of the usual feminine character $7; as,
ust rajné, “queen,” from way rdjan ; ufaat dhanini, “ the
rich” (fem.), from wftaq dhanin, m.n. “rich.” Gothic fe-
minine substantive bases in n exhibit, before this consonant,
either an é (==, §. 69.) orei: these are genuine feminine
final vowels, to which the addition of an n can have been
only subsequently made. And already, at §. 120., a close
connection of bases in ein (én) with the Sanskrit in $7
and Lithuanian in 7, has been pointed out. Most substan-
tive bases in ein are feminine derivatives from masculine-
neuter adjective bases in a, under the same relation, ex-
cluding the modern n, as in Sanskrit that of aut sundari,
“the fair” (woman), from get sundaru m. n. “ beautiful”
Gothic substantive bases in ein for the most part raise
the adjective, whence they are derived, to an abstract;
ote:
* Vide p. 1083, Note.

154 “FORMATION OF CASES,
[G. Ed. p.165.] e.g. MANAGEIN, “crowd, nom. manageéi,

from the adjective base MANAG A (nominative masc. manag-s,
neut. managa-ta); MIKILEIN, nom. mikilei, “ greatness,”
from MIKILA (mikil-s, mikila-ta), “great.” As to feminine
bases in én, they have arisen from feminine bases in 6;
and I have already observed that feminine adjective bases
in én—as BLINDON, nom. blindé, gen. blindén-s—must be
derived, not from their masculine bases in an, but from the
primitive feminine bases in 6 (nom. a, Grimm’s strong adjec- —
tives). Substantive bases with the genitive feminine in 6n pre-
suppose older ones in 6; and correspond, where comparison
is made with old languages connected in their bases, to
Sanskrit feminines in 4, Greek in a, 9, Latin in a; and in
these old languages never lead to bases with a final n.
Thus, ZUGGON (pronounced tungén), nom. tuggd, answers
to the Latin lingua, and to the Sanskrit fag jihwd,
(=dschihw4, see §. 17.); and DA URON, nom. dauré, to the
Greek 6vpa; VIDOVON, nom. vidévé, “ widow,” to the San-
skrit fawat vidhavd, “the without man” (from the prep.
fa vi and wa dhava, “ man”), and the Latin vidua. It is
true that, in MITATHYON, “ measure,” nom. mitathyd, the
suffix thyén completely answers to the Latin tion, e.g. in
ACTI ON; but here in Latin, too, the on is a later addition,
as is evinced from the connection of fi-on with the Sanskrit
suffix fa ti, of the same import, and Greek oi-s (old tis),
Gothic ti, thi, di (see §.91.). And in Gothic, together with
the base MITATHYON exists one signifying the same, MJ-
TATHI, nom. mitaths. In RATHYON, nom. rathyé, “ac-
count,” a relationship with RATION, at least in respect of
the suffix, is only a seeming one; for in Gothic the word is
[G. Ed. p. 166.] to be divided thus, rath-yén ; the th belongs,
in the Gothic soil, to the root, whence the strong part. rath-
an(a)-s has been preserved. The suffix yén, of RATHYON
therefore corresponds to the Sanskrit y4; e.g. in fae vid-yd,
“knowledge.” Of the same origin is GA-RUN-YON, nom.
garunyé, “ inundation.”
143. If a few members of a great family of languages

have suffered a loss in one and the same place, this may be
accident, and may be explained on the general ground, that
-all sounds, in all languages, especially when final, are sub-
ject to abrasion; but the concurrence of so many languages
in a loss in one and the same place points to relationship,
or to the high antiquity of such a loss; and in the case
before us, refers the rejection of.an n of the base in the
nominative to a period before the migration of languages,
and to the position of the original site of the human races,
which were afterwards separated. It is surprising, there-
fore, that the Greek, in this respect, shews no agreement
with its sisters; and in its v bases, according to the measure
of the preceding vowel, abandons either merely the nomi-
native sign, or the v alone, never both together. It is a
question whether this is a remnant of the oldest period
of language, or whether the vy bases, carried away by
’ the stream of analogies in the other consonantal declensions,
and by the example of their own oblique cases, which
do not permit the remembrance of the vy to be lost, again
returned, at a comparatively later period, into the common
‘and oldest path, after they had experienced a similar
Joss to the Sanskrit, Zend, &c., by which we should be
conducted to nominative forms like evdaiuw, evdauo, Tépn,
Tépe, TaAG, TAAG? I do not venture to decide with positive-
ness on this point, but the latter view appears to be the
more probable. It here deserves to be [G. Ed. p. 167.]
remarked, that, in German, the n, which in Gothic, in
the nominative, is always suppressed, has in more modern
dialects made its way in many words from the oblique
cases again into the nominative. So early as the Old
High German this was the case; and, in fact, in femi-
nine bases in in (Gothic ein, §. 70.), which, in the nomi-
native, oppose to the Gothic ei the full base in: as
guotlihhin, “glory” (see Grimm, p. 628). In our New High
German the phenomenon is worthy of notice, that many

original n bases of the masculine gender, through a con-
fusion in the use of language, are, in the singular, treated
as if they originally terminated in na; i.e. as if they be-
longed to Grimm’s first strong declension. Hence the n
makes its appearance in the nominative, and the genitive
regains the sign s, which, indeed, in Gothic, is not want-
ing in the n bases, but in High German was withdrawn
from them more than a thousand years since. Thus,
Brunnen, Brunnens, is used instead of the Old High Ger-
man prunno, prunnin, and the Gothic brunna, brunnin-s.
In some words, together with the restored n there occurs in
the nominative, also, the ancient form with n suppressed, as
Backe or Backen, Same or Samen; but the genitive has in
these words also introduced the s of the strong declension,
Among neuters the word Herz deserves consideration.
The base is, in Old High German, HERZAN, in Middle
High German HERZEN; the nominatives are, herza,
herze; the New German suppresses, together with the
n of Herzen, the vowel also, as is done by many mas-
culine n bases; as, e.g. Bar for Bare. As this is not a
transition into the strong declension, but rather a greater
weakening of the weak nominative, the form Herzens,
therefore, in the genitive, for an uninflected Herzen, is sur-
[G. Ed. p. 168.] prising. With this assumed or newly-re-
stored inflection s would be to be compared, in Greek, the
nominative¢, as of deAdi-s, wéAa-s; and with the n of Brun-
nen for Brunne, the v of daiuwv, répnv; in case, as is ren-
dered probable by the cognate languages, these old forms
have been obtained from still older, as deApi, uéAa, daiuw, réoy
by an unorganic retrogade step into the stronger declension,*
* That, in Greek, the renunciation of a v of the base is not entirely
unknown may be here shewn by an interesting example. Several
cardinal numbers in Sanskrit conclude their base with oa viz.
panchan,
NOMINATIVE SINGULAR. - 157
144. Bases in wz ar (¥ 77, §. 1.) in Sanskrit reject the r in

the nominative, and, like those in a n, lengthen the pre-
ceding vowel ; e.g. from fanz pitar, “father,” tat bhrdtar,
“brother,” aTAT médtar, *i iother” afent duhitar, “ daughter,”
come fat pitd, wtat bhrdtd, arat mdtd, efemt duhitd. The
lengthening of the a serves, I believe, as a -compensation
for the rejected r. As to the retention, however, through
all the strong cases, excepting the vocative, of the long a
of the agent, which corresponds to Greek formations in
Tp, Twp, and to Latin in for, this takes place because, in all pro-
bability, in these words at tér, and not Wt tar, is the
original form of the suffix; and this is also supported by
the length of the suffix being retained in Greek and Latin
through all the cases—ryp, twp, tor; only [G. Ed. p. 169.|
that in Latin a final r, in polysyllabic words, shortens an
originally long vowel. Compare
SANSKRIT. GREEK. LATIN.
Nom. sing. eat ddtd, Soryp, —datar,

Ace. sing. aratty ddtdr-am, Sorjp-a, dator-em,
N. A. V.dual, erate ddtdr-au, Sorio-e, ......
Nom. Voe. pl. eratta ddtdr-as, Sorip-es, dator-es.
The Zend follows the analogy of the Sanskrit, both in the
rejection of the r in the nominative, and in the length
menehan, “five,” saptan, “seven,” ashtan with ashtau, “eight,” navan,

‘nine,’ dasan, “ten.” These numerals are, indeed, used adjectively,
when they are not governed by the gender of their substantive, but display
always a neuter form, and indeed, which is surprising, in the nominative,
accusative, and vocative sing. terminations, but in the other cases the suit-
able Plnrakendings ; e.g. Wa UATAS pancha (not panchdnas) raéjdnaz
‘‘quingue reges”; on the other hand, waa Uae panchasu rdjasu “in
quinque regibus.” To the neuter nominatiyes and accusative of the sin-
gular Gy pancha, BA sapta, Aq nava, and ey dasa—which rest on the
regular suppression of the n—answer the Greek wévre, émrd, évvéa, déxa,
with the distinction that they have become quite indeclinable, and retain
the old uninflected nominative through all the cases.
of the preceding a of the noun agent, in the same places

as in the Sanrkrit, with the exception of the nominative sin-
gular, where the long a, as always when final, is shortened;
eg »ossd paita, “father,” aso.w4 data, “giver,” “Creator ;”
ace G¢eAspssd paitar-em, G¢wso.g ddtdr-ém. In Lithua-
nian there are some interesting remains, but only of femi-
nine bases in er, which drop this letter in the nomina-
tive, but in most of the oblique cases extend the old
er base by the later addition of an i. Thus mofé, “wife,”
dukte “ daughter,” answer to the abovementioned Arm
mata, efeat duhitd; and, in the plural, moter-és, dukter-és, to
alata mdtar-as, efeata duhitar-as. In the genitive singu-
lar [ regard the form moter-s, dukter-s, as the elder and
more genuine, and moteriés, dukteriés, as corruptions be-
longing to the « bases. In the genitive plural the base
has kept clear of this unorganic i; hence, moler-d, dukter-d,
not moteri-i, dukteri-i. Besides the words just mentioned,
the base SESSER, “sister,” belongs to this place: it
answers to the Sanskrit waz swasar, nom. Wat swasd; but
distinguishes itself in the nominative from mote and dukte,
in that the e, after the analogy of bases in en, passes into @,
thus sessit.
[G. Ed. p.170.] 145. The German languages agree in their
r bases (to which but a few words belong denoting affinity)
with the Greek and Latin in this point, that, contrary to the
analogy just described, they retain the r in the nominative.
As tarnp, untip, Ouvryarnp, Sayp (Sanskrit, faz dévar, tydéuri,
nom. @at dévd), frater, soror ; so in Gothic, bréthar, svistar,
dauhtar ;in Old High German, vatar, pruodar, suéstar, tohtar.
It is a question whether this r in the nominative is a rem-
nant of the original language, or, after being anciently
suppressed, whether it has not again made its way in the
actual condition of the language from the oblique cases
into the nominative. I think the latter more probable;
for the Sanskrit, Zend, and Lithuanian are three witnesses
for the antiquity of the suppression of the r; and the

Greek words like zatijp, pytnp, cwrip, pytwp, exhibit some
thing peculiar and surprising in the consonantal declension,
in that p and ¢ not combining, they have not rather pre-
ferred giving up the base-consonant than the case-sign (as
mats, mous, &e.). It would appear that the form tng is of
later origin, for this reason, that the p having given place
to the nominative ¢, the form 7y-s, whence typ-o¢ should
come, was, by an error of language, made to correspond to
the y-¢ of the first- declension. The want of a cognate
form in Latin, as in Zend and Sanskrit, as also the, in
other respects, cognate form and similarity of meaning
with mg tdr, t0-r, typ and twp, speak at least plainly enough
for the spuriousness and comparative youth of the nouns of
agency in 77°.
146. Masculine and feminine primary forms in ¥@ as
in Sanskrit lengthen the a in the nominative singular.
They are, for the most part, compounded, and contain, as
_ the last member, a neuter substantive in Wa as, as gaaa
durmanas, “ evil-minded,” from qa_dus (G. Ed. p. 171]
(before sonant letters—§. 25.—et dur) and aqa_ manas,
“mind,” whence the nom. masc. and fem, ¢%ara_ durmands,
neut. g44a durmanas. A remarkable agreement is here
shewn by the Greek, in ducpevijs, 6,4 opposed to 7d Sucpevés.
The as of gaara durmands, however, belongs, though
‘unrecognised, to the base; and the nominative character is,
wanting, according to §. 94. In Greek, on the other hand,
the ¢ of Sucuevne has the appearance of an inflexion, because
the genitive, &c.,is not dvcpevés-os, like the Sanskrit e#aaa
durmanas-as, but dvcpevéos. If, however, what was said at
§. 128 is admitted, that the ¢ of uévos belongs to the base, and
Méveos is abbreviated from jévec-os, then in the compound
duopev7ys also, and all similar adjectives, a = belonging to
the base must be recognised, and the form ducpevécos
must lie at the bottom of the genitive ducuevéos. In the
160 FORMATION OF CASES:
nominative, therefore, either the ¢ belongs to the base, and

then the agreement. with gata durmands would be com-
plete ; or the ¢ of the base has been dropped before the case-
signs. The latter is, in my opinion, least probable; for the
former is supported by the Latin also, where the forms which
answer to the Sanskrit as bases are in the nom. mase, and
fem. in like manner without the case-sign. Thus the San-
skrit comparative suffix is $ya ¢yas—the last a but one of
which is lengthened in the strong cases, and invested with a
dull nasal (Anuswara, §. 9.)—in Latin, idr, with the s changed
into r, which so frequently happens; and the nominative in
both genders is without the case-sign: the originally long o,
however, is shortened by the influence of the final r. In the
neuter us corresponds to the Sanskrit wa as, because w is
favourable to a final s, and prevents its transition into 7;
hence gravius has the same relation to the Sanskrit 1ttra
gartyas (irregular from 7& guru, “heavy,”) as lupus to
[G. Ed. p.172.] gaa, vrikas, only that the s of the nomi-
native character in the latter belongs in the former to the
base. The final syllable dr, though short, must nevertheless
be held, in Latin, as graver than is, and hence gravior forms
a similar antithesis to gravius that in Greek duopevjs does to
Svopevés, and in Sanskrit Grae durmands to qaae durmanas.
147. In Lithuanian a sai ethes which stands quite
isolated, méni (= ménuo), “moon” and “month,” deserves
here to be mentioned: it proceeds from the primary form
MENES*, and, in regard to the suppression of the final
consonant and the transformation of the preceding vowel, has
the same relation to it that, as above (S. 139.), akmi has
* The relation of this to aTa mds, which signifies the same—from aTe
mds, ‘to measure,” without a derivative suffix—is remarkable; for the
interposed nasal syllable ne answers to the Sanskrit # na in roots of the
seventh class (see p.118); and in this respect MENES bears the same
relation to the Latin MENSTI that |. c. frafa bhinadmi does to findo,
toAKMEN, sess to SESSER : in the oblique cases, also,

the s of the base again r2-appears, but receives, as in the
er and en bases, an unorganic increase: thus the genitive
ismenesio, whence ME NESTA is the theme; as wil/ko, “lupi,”
from WILKA, nom. wilka-s.
14s. In neuters, throughout the whole Sanskrit family of
languages the nominative is identical with the accusative,
which subject is treated of at §. 152. &c. We here give a
general view of the nominative formation, and select for the
several terminations and gender of the primary forms, both
for these cases and for all others which suit our purpose, the
following examples: Sanskrit qa vrika, m. “wolf ;” @ ka,
“who?” era déna, n. “ gift >” # ta, n. “this; fagt jihwd,
f.“tongue;” @t kd, “which?” ofa pati, m. “lord,” “husband ;”
Wifa priti,f. “love ;* af<vdri, n. “ water;” afaupat bhavishyanti,
“who is about to be;” aa sunu, m.“son;” [G. Ed. p 173.]
a tanu, f. ““body;” #y madhu, n. “honey,” “ wine ;”
vadhu, £. “ wife ;* z%tgd, m. f. “bullock,” “‘cow;” aftndu, £.
“ship.” Of the consonantal declension we select only such
final consonants as occur most frequently, whether in single
words or in entire classes of words: arq_vdch, £ “speech "*;
wom bharant, in the weakened form, we#, bharat (§. 129.) m.
n. “bearing,” “receiving,” from 3 bhar (¥ bhri) cl. 1;
sea dtman, m. “soul; ama ndman, n. “name ;* uTat
blirdtar, m. “brother ;~ afeat duhitar, f. “daughter ;* erat
datar, m. “ giver ;” qa vachas, n. “ speech,” Greek, "ENE,
éxog (§§. 14. 128.), for FEDLES, Fexos. Zend, assJereh véerhka,
m. “wolf ;* 29 ka, m. “who?” sso.sg ddta, n. datum; a3
ta, n. “this;” sw»¢sw hized, f. “tongue ;” .ws kd, “ which?”
* Masculines and feminines in the consonantal declension agree in all

cases: hence an example of one of the two genders is sufficient. The
only exception is the accusative plural of words denoting relationship in
wit ar &, §. 114.), wh'ch form this case from the abbreviated theme in
Wri.
$0.50 paiti, m. (§. 41.) “Lord ;” sp das dfriti, £. “blesa-

ing ;” sss vairi, n. “ water ;” SPs eps dishyainté,
“who will be ;” >s9s0 pasu, m. “‘ tame animal ;” >gaseo tanu,
f. “ body ;” >@a§ madhu, n. “ wine;” be gé, m. f. “ bullock,”
[G. Ed. p.174.] “cow”™*; gash vach, f. “speech,” “voice “+;
Wyss barant, or pyres burént, weakened form asdys
barat, m.n. “ bearing ;” asGs925 asman, m. “heaven;’” sasGawy
ndman (also js§ nanman), n. “name;” Aseass brdtar,t
* It has been remarked at §.123 of the cognate nom. Gusg xo,
“earth,” accus. OHS zanm, that I have only met with these two cases.
The very common form Ges 2m, which is found only in the other
oblique cases, is nevertheless represented by Burnouf, in a very interesting
article in the Journal des Savans (Aug. 1832), which I only met with
after that page had been printed, as belonging to the same theme.
T agree with him on this point at present, so much the rather as I believe
I can account for the relationship of Ge ¢ zemé, “ terre,” (dat.) SGES
eémi, “in terra,” &c. to the Sanskrit 774 gavé, Mfq gavi. 1 do not doubt,
that is to say, that, in accordance with what has been remarked at §.63.
and p.114,the Zend § m is to be regarded as nothing else than the
hardening of the original v. The Indian zt gé, before vowel terminations
gav, would consequently have made itself almost unintelligible in the
meaning “earth,” in Zend, by a double alteration; first by the transition
of g to z, in which j must be assumed as the middle step—in which
e.g. Gasy_jam, “to go,” from m1 gam, has remained ; secondly, by the
hardening of the » tom. Advert, also, to the Greek 8n, for yn, in dnuqrnp ;
since 6 and 6% from 4 j (=dsch), have so divided themselves in the
sound whence they have sprung, that the Greek has retained the 7-sound,
the Zend the sibilant.
+ I cannot quote the nominative of this word; but it can only be
wS.usly vde-s, as palatals before «45 s change into GS c; and thus, from
w>4 druj, “an evil demon,” occurs very frequently the nom. 03>7g
drue-s. I have scarcely any doubt, too, that what Anquetil, in his
Vocabulary, writes vdhksch, and renders by “parler, cri,” is the nomi-
native of the said base; as Anquetil everywhere denotes 6 by kh, and
aS by sch.
t In the theme we drop, intentionally, the ¢ é required by §, 44, as it
is clear that ArsossJs brdétar, not gAasqo.s7s brdtaré, must be the base
word ; Ascoastass baratar also occurs, with as a interposed.
NOMINATIVE SINGULAR, 163-
m, “ brother;” Mse.9> 4dughdhar, f. “ daughter ;" spay

déitar, m. “giver,” “creator ;” Louw vaché, n. (8. 56°.)
“word.” It is not requisite to give here examples in Greek
and Latin: from Lithuanian and Gothic we select the bases,
Lith. WILK4A, Goth. VULFA. m. “ wolf;” Lith. K4, Goth.
HVA, m. “who?” Lith. GERA, n. “good;” TA, n. “ the ;”
Goth. DAURA, n. “gate,” .(Sanskrit, gr< dwéra, n.); THA,
n. “this;” Lith, RAN KA, f. “hand;” Goth, GI BO, f. “ gift” |
(§. 69.); HVO, f.* which 2”; Lith. PATT, m. “ Lord”*; Goth.
GASTI, m, “stranger;” J, m. “he,” n. (G. Ed.p.175.]
“it;” Lith. AWI, f. “sheep,” (Sansk. wfa avi, m. cf. ovis,
dis); Goth. ANSTT, f.*mercy;” Lith. Goth. SUNU, m. “son;
Goth. HANDU, f.“hand;” Lith. DAR KU, n. “ugly;” Goth.
FATIH, n. “beast;” Lith. SUK ANT, m.t “turning; Goth.
FIYAND, m. “foe;” Lith, AKMEN, m. “stone;” Goth.
AHMAN, wm.“ spirit;’ NAMAN, n. “name;” BR OTHAR,
m. “brother;”> DAUHTAR, Lith. DUKTER, f. “ daughter.”
SANSERIT. ZEND. GREEK. LATIN. LITHUAN. GOTHIC.
m. vrika-s, véhrké,t AdxKo-s, lupu-s, wilka-s, vulf’s.

m. ka-s, ké,t aS case + kere, hva-s.
_ # In the comp. wiess-pati-s, “landlord”; isolated pat-s, “ husband,”

‘with 7 in the nominative suppressed, as is the case in Gothic
in all bases
in. Compare the Zend sossds9shvii-paiti, “lord oftheregion.”
- t These and- other bases ending with a consonant are given only in
those cases which have remained free from a subsequent vowel addition.
} Before the enclitic particle eha, as well here as in all other forms, the
termination as, which otherwise becomes 6 (§.56>.), retains the same
form which, in Sanskrit also, WA as assumes before q cha: hence is said
aspessass eve véhrkascha, “ lupusque,” as in Sanskrit vrikascha.
And the appended cha preserves the otherwise shortened final vowel
in its original length: hence apis»wsy jikvdcha, “ linguaque,”
PSO WINIIKD bushyuiniicha, “ futuraque,” ASA O.W75 bratdcha,
“ fraterque” Even without the 4sqs at times the original length of the
final vowel is found undiminished: the principle of abbreviation, how-
ever, remains adequately proved, and I therefore observe it everywhere
in the terminations.
M2
164 FORMATION OF CASES, -
SANSKRIT. ZEND. GREEK. LATIN. LITHUAN. GOTHIC.

D, ddna-m, daté-m, dapo-v, donum, géra, daur’.
n.: ta-t, - ta-t, TO, is-tu-d, ta-i, tha-ta.
f.. gihwé, hizva,* xapa, terra, ranka, giba.
f. kd, ka, Aa ote) Yuded hvé.
m. pati-s, paiti-s, méat-s, — hosti-s, pati-s, gast’-s,
Ms th oe iy oe tives i-s, whee i-s.
f. priti-s, Gfriti-s mopti-s, siti-s, awi-s, anst'-s,
2. vari, vairi, idpt, snare, No
ae ae ae Sate See oth, i-ta.
Ef. bhavishyanti, bishyainti*® .... es Se bisenti, ....
'F m.siinu-s, jU-8, iv Ov-¢, CUu-8, suni-s, sunu-s.
Se. tanu-s, eaibba aie ddl oe ee handu-s
ile madhu, madhu, pébu, pecuy dark, faihu.
f.. vadhi-s, hae Past lt oes: OE ok
m. f. gdu-s,} gau-s,t Bov-s, bd-s, ....
f. nducs, eae vau-¢, Seteek este
f. vak, vic-s, on-s, VOC“8y ks a
m.- bharan, baran-s, gépwv, feren-s, sukan-s, fiyand-s
m. dimd’, asma,* Saiuwy, sermo’, akm®?, ahma’.
mn, ndma’, nama’, TaAav, nomen, .... namo’.
m, bhrdtd’, brata’,* matyp, frater, ..e.6 bréthar. e
f. duhitd’, dughdha,*, Ovyarnp, mater, dukté,’ dauhtar,

m, détd, ddta’,* doryp, dator, ....
n. vachas, vaché,* émos, OPUs, © ve ee
ACCUSATIVE,
SINGULAR.
149. The character of the accusative is m in Sanskrit,
‘Zend, and Latin; in Greek yv, for the sake of euphony. In
Lithuanian the old m has become still more weakened to
* See the marginal note marked ({) on the foregoing page.

+ Irregularly for 71rq_g6-s.
t Or wsguso dos, §. 33.
ACCUSATIVE SINGULAR. 165
the dull re-echoing nasal, which in Sanskrit is called Anu-

[G. Ed. p.177.] swara, and which we, in both languages,
express by 7 (§. 10.). The German languages have, so early
as the Gothic even, lost the accusative mark in substantives
entirely, but in pronouns of the 3d person, as also in adjec-
tive bases ending with a vowel which follow their declen-
sion, they have hitherto retained it; still only in the
masculine: the feminine uowhere exhibits an accusative
character, and is, like its nominative, devoid of inflexion.
The Gothic gives na instead of the old m; the High
German, with more correctness, a simple n: hence, Gothic
blind-na, “caecum,” Old High German plinta-n, Middle and
Modern High German blinde-n.
150. Primary forms terminating with a consonant prefix
to the case-sign m a short vowel, as otherwise the combi-
nation would be, in most cases, impossible: thus, in San-
skrit am, in Zend and Latin ém, appears as the accusative
termination*: of the Greek av, which must originally have
existed, the v is, in the present condition of the language,
lost: examples are given in §. 157.
151. Monosyllabic words in @ dé, and du, in Sanskrit,
like consonantal bases, give am in place of the mere m, as
the accusative termination, probably in order in this way
to become polysyllabic. Thus, wt hi, “fear,” and néu,
“ship,” form, not bhi-m and ndu-m, as the Greek v@u-v would
eesti
* From
the bases 274 druj and ash vach,I find besides 6eu2 74
drujém, § <q. vich?m, in the V.S.; also frequently ¢5y>7g drujim,
¢ vdchim: and if these forms are genuine, which I scarcely doubt,
they are to be thus explained—that the vowel which stands before m is
only a means of conjunction for appending the m; for this purpose, how-
ever, the Zend uses, besides the ¢é mentioned at §.30, not unfrequently
Si; eg for SPREE _4055 dadémahi, occurs also $0 28§59159 dadimahi,
and many similar forms; as sess 5i9> us-i-mahi, answeringto the San-
skrit SYTe_usmas (in the Védas Jyafa uémasi), “we will.”
[G. Ed. p. 178.] lead us to expect, but fray bhiy-am, aTay

ndv-am. With this agree the Greek themesiin ev, since these
give e-a, from eF-a, for ev-v; e.g. BaoiAé(F)a, for BaciAev-v:
It is, however, wrong to regard the Latin em as the true, ori--
ginally sole accusative termination, and for lupu-m, hora-m,
fruc-tum, diem, to seek out an older form lupo-em, hora-em,
fructu-em, die-em. That the simple nasal suffices to charac-
terize the accusative, and that a precursory vowel was only
added out of other necessary reasons, is proved by the history
of our entire family of languages, and would be adequately
established, without Sanskrit and Zend, by the Greek, Li-
thuanian, and Gothic. The Latin em in the accusative
third declension is of a double kind: in one case the e
belongs to the base, and stands, as in innumerable cases,
for i; so that e-m, of igne-m (Sanskrit uafraa agni-m),
corresponds to the Indian i-m, Zend é-m, Greek :-v, Li-
thuanian i-n, Gothic i-na (from ina, “him”); but in the
em of consonantal bases the e answers to the Indian a, to
which it corresponds in many other cases also.
152. The Sanskrit and Zend neuter bases in a, and
those akin to them in Greek and Latin, as well as the two
natural genders, give a nasal as the sign of the accusative,
and introduce into the nominative also this character,
which is less personal, less animated, and is hence appro-
priated to the accusative as well as to the nominative in
the neuter : hence, Sansk. waaR sayana-m, Zend G EpAsS IAs
sayané-m, “a bed”; so in Latin and Greek, donu-m, d@po-v
All other bases, with but few exceptions, in Latin, remain
in the nominative and accusative without any case charac-
ter, and give the naked base, which in Latin, however, re-
places a final i by the cognate e; thus, maré for mari corre-
[G. Ed. p.179.] sponds to the Sanskrit arfz vdri, “ water”;
the Greek, like the Sanskrit and Zend, leaves the « unchanged
—idpr-s, Spt, as in Sanskrit yfaa suchis, wife suchi. The
following are examples of neuter u bases, which supply the
place both of nominative and accusative: in Sanskrit ay

madhu, “honey,” “wine,” wy asru, “tear,” wre swddu,
“sweet”; in Zend »wbl odhu, “ wealth” (Sanskrit 7a
vasu); in Greek pé6u, daxpv, 730; in Latin pecu, genu. The
length of this u is unorganic, and has probably passed into
the nominative, accusative, and vocative from the oblique
cases, where the length is to be explained from the sup-
pressed case terminations. With regard to the fact that
final u is always long in Latin, there is perhaps a reason
always at hand for this length: in the ablative, for ex-
ample, the length of the originally short u is explicable as a
compensation for the case sign which has been dropped,
by which, too, the @ of the second declension becomes long.
The original shortness of the u of the fourth declension
is perceivable from the dat. pl. i-bus. The , in Greek
words like -yévos, jévoc, evyevés, has been already explain-
ed at §. 128. as belonging to the base: the same is the
case with the Latin e in neuters like genus, corpus,
gravius: it is the other form of the r of the oblique cases,
like gener-is, corpor-is, gravior-is (see §. 127.); and corpus
appears akin to the Sanskrit neuter of the same mean-
ing, Wa vapus, gen. Fa vapu-sh-as (see §. 19.), and
would consequently have an r too much, or the Sanskrit
has lost one.* The = also of neuter bases in T, in tetudds,
tépas, does not seem to me to be the case sign, but an
exchange with T, which is not admissible at the end, but is
either rejected (uéA:, mpayua) orexchanged [G. Ed. p. 180.]
for a cognate 2, as mpés from zpori, Sanskrit ufa pratit
* Compare,in this respect, brachium, Spayiov, with 4Tg@_ bahu-s,

“arm”; frango, piryypt, with wafsa bhanajmi, “I break,” wre
bhanjmas, “ we break.”
t With this view, which I have already developed in my treatise
“On some Demonstrative Bases, and their connection with various Pre-
positions and Conjunctions” (Berlin, by Diimmler), pp. 4—6, corre-
sponds, as to the essential points, what Hartung has since said on this
subject
In Latin it is to be regarded as inconsistent with the spirit

of the language, that most adjective bases ending with a
consonant retain the nominative sign s of the two natural
genders in the neuter, and in this gender extend it also to
the accusative, as if it belonged to the base, as capac-s felic-s,
soler(t)s, aman(t)s. In general, in Latin, in consonantal bases,
the perception of the distinction of gender is very much
blunted, as, contrary to the principle followed by the San-
skrit, Zend, Greek, and Gothic, the feminine is no longer
distinguished from the masculine.
153. In Gothic substantives, as well neuter as mascu-
line, the case sign m is wanting, and hence neuter bases
in a stand on the same footing with the ?, u, and consonantal
bases of the cognate languages in that, in the nominative
and accusative, they are devoid of all inflexion. Compare,
with regard to the form of this case, daur(a) with ery
dwéram, which has the same meaning. In Gothic there
are no neuter substantives in i; on the other hand, the
[G. Ed.p.181.] substantive bases in ya, by suppression of
the a in the nominative and accusative singular (cf. §. 135.),
gain in these cases the semblance of i bases; e.g. from the
base REIKYA, “rich” (Sanskrit wusq rédjya, likewise
neuter), comes, in the case mentioned, reiki, answering to
the Sanskrit tqH rdjya-m. The want of neuter ¢ bases
subject in his valuable work on “ On the Cases,” p. 152, &c.; where also
the p of fap and vdwp is explained as coming from T, through the inter-
vention of 3.. The Sanskrit, however, appears to attribute a different
origin to the p ofthese forms. To aa yakrit “liver” (likewise neuter),
corresponds both jecur and fap, through the common interchange between
kh and p: both owe to it their p, as jrar-os does its r. “Hsrar-os should be
Hmapt-os, Sanskrit aaAAa yakrit-as, But the Sanskrit also in this word, in
the weak cases, can give up the 7, but then irregularly substitutes Ss for a
Ht, e.g. gen. THA yakn-as for qaHTa yakanas. With regari to the
p of ddwp, compare FF udra, “ water,” in AAG, sam-udra, “sea.”
e
in German is the less surprising, that in the cognate Sanskrit,

Zend, and Greek, the corresponding termination in the neuter
is not very common. Of neuter u roots the substantive de-
clension has preserved only the single #4] HU. “ beast.” In
Lithuanian the neuter in substantives is entirely lost, and
has left traces only in pronouns and adjectives, where the
latter relate to pronouns. Adjective bases in u, in this
case, have their nominative and accusative singular in ac-
cordance with the cognate languages, without case sign;
e.g. darki, ‘“‘ugly,” corresponds as nominative and accusa-
tive neuter to the masculine nominative darki-s, accusative
darku-n. This analogy, however, is followed in Lithua-
nian, by the adjective bases in a also; and thus géra,
“good,” corresponds as nominative and accusative to the
masculine forms géra-s, géra-n,* which are provided with
the sign of the case.
[G. Ed. p.182.] 154. It is a question whether the m, as
the sign of the nominative and accusative neuter (it is ex-
cluded from the vocative in Sanskrit and Zend), was origi-
nally limited simply to the a bases, and was not joined to the
* The e of neuter forms like dide, “great,” from the base DIDYA—
nom. masc. didi-s for didya-s, as §. 135.* yaunikkis, “ youngling”—I ex-
plain through the euphonic influence of the suppressed y. As also the
feminine originally long a is changed into e by the same influence, so is
the nominative and accusative neuter in such words identical with the
nominative feminine, which is likewise, according to §. 137, devoid of in-
flexion ; and dide therefore signifies also “‘ magna,” and answers, as femi-
nine, very remarkably to the Zend nominatives explained at §. 137., as
ryeheo perené, py sous brdturyé. In this sense are to be regarded,
also, the feminine substantives in Ruhig’s third declension, as far as they
terminate in the nominative in ¢, as giesme, “song.” As no masculine
forms in is correspond to them, the discovery of the true nature of these
words becomes more difficult ;for the lost y or i has been preserved only
in the genitive plural, where giesmy-i is to be taken like rank-t from
ranka, i.e. the final vowel of the bases is suppressed before the termina-
tion, or has been melted down with it.
é and u bases also; so that, in Sanskrit, for vdri we had ori-

ginally vari-m, for madhu, madhu-m? I should not wish to
deny the original existence of such forms; for why should
the a bases alone have felt the necessity of not leaving
the nominative and accusative neuter without a sign of
relation or of personality? It is more probable that the
a bases adhered only the more firmly to the termination
once assumed, because they are by far the most numerous,
and could thus present. a stronger opposition to the de-
structive influence of time by means of the greater force
of their analogies; in the same way as the verb sub-
stantive, in like manner, on account of its frequent use, has
allowed the old inflexion to pass less into oblivion, and in
German has continued to our time several of the progeny of
the oldest period ; as, for instance, the nasal, as characteristic
of the Ist person in bi-n, Old High German pi-m Sans. watfa
bhavd-mi. In Sanskrit, one example of an m as the nomina-
tive and accusative sign of an 7 base is not wanting, although
it stands quite isolated; and indeed this form occurs in the
pronominal declension, which everywhere remains longest
true to the traditions of bygone ages. I mean the inter-
rogative form fax ki-m, “what”? from the base fa ki,
which may perhaps, in Sanskrit, have produced a ki-t,
which is contained in the Latin qui-d, and which I recog-
nise again, also, in the enclitic faq chit, weakened from far
ki-t. Otherwise i or u-bases of pronouns in the nomina-
tive accusative neuter do not occur; for Wy amu, “that”
(man), substitutes wea adas ; and ¥i, “ this,” combines with
[G. Ed. p. 183] ¢— dam (geq idam, “ this”). Concerning
the original procedure of consonantal bases in the nominative
and accusative neuters no explanation is afforded by the pro-
nominal declension, as all primary forms of pronouns termi-
nate in vowels, and, indeed, for the most part, in a.
155. Pronominal bases in a in Sanskrit give ¢,in Zend 4, as
the inflexion of the nominative and accusative neuter. The
Gothic gives,
as in the accusative masculine,na form or n,
so here ta for simple ¢; and transfers these, like other pecu-
liarities of the pronominal declension, as in the other Ger-
man dialects, also to the adjective a bases; e.g. blinda-ta,
“caecum,” midya-ta, “medium.” The High German gives,
in the older period, z instead of the Gothic ¢ (§. 87.), in
the most modern period, s. The pronominal base J (later E)
follows in German, as in Latin, the analogy of the old a
bases, and the Latin gives, as in the old ablative, d instead
of t. The Greek must abandon all 7’ sounds at the end of
words: the difference of the pronominal from the common
o declension consists, therefore, in this respect, merely in
the absence of all inflexion. From this difference, however,
and the testimony of the cognate languages, it is perceived
that ro was originally sounded tor or tod, for a tov would
have remained unaltered, as in the masculine accusative.
Perhaps we have a remnant of a neuter-inflexion 7 in or7,
so that we ought to divide ér-r:; and therefore the double 7,
in this form, would no more havea mere metrical foundation,
than the double o (§. 128.) in opec-c1. (Buttmann, p. 85.)
° 156. We find the origin of the neuter case-sign ¢ in the
pronominal base @ éa, “he,” “ this,” (Greek TO, Goth, THA,
&c.); and a convincing proof of the correctness of this ex-
planation is this, that wa ta-t “it” “this,” stands, in regard
to the base, in the same contrast with asa, “he,” at sd,
“she,” as t,as the neuter case-sign, does to [G. Ed. p. 184.]
the nominative s of masculine and feminine nouns (§. 134.).
The m of the accusative also is, I doubt not, of pronominal
origin; and it is remarkable that the compound pronouns
i-ma, “this,” and a-mu, “ that,” occur just as little as ta in the
nominative masculine and feminine; but the Sanskrit sub-
stitutes for the base amu, in the nominative masculine and
feminine singular the form asdu, the s of which, therefore,
stands in the same relation to the m of #Rq amu-m, “ illum,”
. Say amu-shya, “ illius,” and other oblique cases, as, among
172 FORMATION OF CASES.,.
the case-terminations, the sign of the masculine feminine

nominative to the m of the accusative and neuter nomina-
tive. Moreover, in Zend is used 9sa3¢4 imat, “ this,” (n.)
(nom. accus.), but not imé, “this” (m.), but ¢725 aém (from
way ayam), and ¢s¢m (from ¢4y iyam), “ this” (f.). Observe
in Greek the pronominal base MI, which occurs only in the
accusative, and, in regard to its vowel, has the same rela-
tion to ® ma (in the compounded base 3m i-ma) that faq
ki-m “what?” has to #@ ka-s “who”? The Gothic neut,
termination fa anwers, in respect to the transposition of
sound (§. 87:), to the Latin d (id, istud): this Latin d, how-
ever, seems to me a descent from the older ¢; as, e.g., the
b of ab has proceeded from the p of the cognate wy apa,
ané; and in Zend the d of ¢¢
41s" d-dém, “him,” is clearly
only a weakening of the ¢ of @ ta, » ta.t
[G. Ed. p.185.] 157. Tothe Sanskrit ta-t, mentioned above,
Zend ta-t, Greek 74, &c., corresponds a Lithuanian fai, “ the,”
as the nominative and accusative singular. I do not believe,
however, that the 2 which is here incorporated in the base 74
* The 4 of 4-d2m is the preposition corresponding to the Sansk. 4.
+ See my treatise “‘ On the Origin of the Cases” in the Trans. of the
Berlin Academy for the year 1826. AsT in Greek easily becomes 5 (but a
final = has in many parts of Grammar become v), Hartung founds on this,
in the pamphlet before mentioned, p. 154, the acute conjecture of an
original identity of neuters in vy (m) with those in ¢. We cannot, how-
ever, agree with him in this, because the m, on account of the origin
which we ascribe to this case-sign, is as little surprising in the nominative
of the neuter as in the accusative of the more animated genders; and
besides, a greater antiquity is proved to belong to the neuter m, through
the Sanskrit and Zend, than probably the v sounds can boast, which, in
Greek, stand for an older 3, as pev for pes (aa mas), and in the dual roy,
rov for Wa thas, Wa tas. What is wanting in the Greek, viz. a neuter e
inflexion s, appears, however, to be possessed by the Sanskrit; and J am

inclined to divide the form WEA adas, * that ” (nom. accus.) into a-da-s,
and to explain it as a corruption of a-da-t (cf. Gramm. Crit. Addend. to
r. 299.) ; but to regard the syllable da as weakened from fa, as in the Zend
GEgw d-de-m, “him.” We shall recur to this when treating of the
pronouns.
e
ACCUSATIVE: SINGULAR. 173
js any way connected with the neuter ¢, d, of the cognate
languages: I should rather turn to a relationship with the
é demonstrative in the Greek (otroci, éxeivoc/), and’ to the
= it, which is, in like manner, used enclitically in the
Védas—a petrified neuter, which is no longer conscious of any
gender or case; and hence, in several cases, combining with
masculine pronouns of the third person,* This ¥q it, is
consequently the sister form of the Latin id and Gothic i-ta,
‘which, in the Greek éxervoci, has, perhaps only from neces-
sity, dropped the r or 6, and which already, ere I was ac-
quainted with the Véda-dialect, I represented as a consis-
tent part of the conjunctions Yq chét (from cha+it), “if,”
and 4q nét (na+ it). ** °° [G. Ed.'p. 186.]
' The words mentioned ‘at §. 148.’ form in the accusative:
SANSKRIT. ZEND. GREEK. LATIN. LITHUAN. GOTHIC.
m. vrika-m, véhrké-m, Nixo-v, lupu-m, wilka-n, vulf’.

m. ka-m, ké-m, EE Veg ka-n, hwa-na.
n. dana-m, détém, $8@po-v, donu-m, géra, daur’.
n. ta-t, ta-t, 76, is-tu-d, _ ta-i, tha-ta.
f. jihwd-m, hizva-im, yopa-v, terram, ranka-ti, giba.
f. ka-m, ka-nm, Seiwa ee TIE hvé.t
* Examples are given by Rosen in his Véda Specimen, pp. 24, 25,
which, though short, are in the highest degree interesting for Sanskrit
and comparative Grammar; as, Wea sait, “he,” afaq tamit, “him” ;
wanton tayérit, “ of these two”; WeTST tasméit, “to him”; West
asmdit, “to this” (m). The Zend combines in the same way » ¢ or
$i with the interrogative: 70.39.59 kasé and 3959 kasi, “who”? occur
frequently. Perhaps only one of the two modes of writing is correct.
Cf. Gramm. Crit. Addend. to r. 270.
_ + One would expect hvé-na, or, with abbreviation of the base, hra-na,
which would be the same as the masculine. With regard to the lost case-
termination, it may be observed, that, in general, the feminines are less
constant in handing down the old inflexions. A charge which is incurred
by the Sanskrit in the nominative, since it gives ka for kd-s* (§. 137.), is
incurred by the Gothic (for in this manner the corruption spreads) in the
accusative also.
* Cf. §. 386. p. 544,
SANSKRIT. ZERD. GREEK. LATINe LITHUAN. GOTHIC,
m. pati-m, paiti-m, moot-v, hostem, pdti-n, gast’.
MM. sess cave cece cece cece tNMe
f. _priti-m, afritt-m, mopti-v, siti-m, Gwi-n, anst’.
n. vdri, vairi, ido, MATC, scse wue
hes i092 <38 oi0.8.8 a» 0, ded, coos tte
f. bhavishyantim,bishyaintt-m, .... 0 6005 cee e® cee
rqm.sénu-m, pasi-m, ixSb-v, pecu-m, sunu-n, sunu.
a f. tanu-m, tané-m, mitu-v, socru-m,.... handu.
men. madhu, madhu, BéOv, pecu, darki, faihu
=f. vadhd-m, Kale Pe eee me
©, m.f.g4-m,t ga-nm,t Bov-v, boveem, ..+. «eeo
f. nédv-am, coos, YOY) <0S) 50 bl et ee
f. vdch-am, vich-ém, Om-y VOC-EM, «oes coos
* The feminine participial bases in i, mentioned at §. 119., remain free

from foreign commixture only in the nominative and vocative singular:
in all other cases, to the old ¢ is further added a more modern a; and the
declension then follows RAN KA exactly; only that in some cases, through
the euphonic influence of the i, and in analogy with the Zend and the
Latin fifth declension (§.187.), the added a becomes, or may become, e:
in the latter case the z is suppressed, as 1.c. 70 Jo°9 kainé for kainyé (§, 42.).
Thus, from sukanti, ‘the turning”’ (f.), sukusi, “‘the having turned” (f.),
and suksenti, “the about to turn,” Mielcke gives the accusatives sukan-
ezen (see. p. 1388, Note) or sukanczian, sukusen, and suksenczen or suk-
senczian. And even if, according to Ruhig (by Mielcke, pp. 3, 4), the é
before a, e, 0, u is scarcely heard, it must not therefore, in this case, as
well as in those there enumerated, be the less regarded as etymologically
present, and it was originally pronounced so as to be fully audible. From
the feminine, where the i, asSanskrit grammar shews, has an original posi-
tion, this vowel appears to have made its way, in Lithuanian participial
bases, into the oblique cases of the masculine, and to be here invested with
a short masculine a. The accusative sukanti-n, ‘‘ the turning” (masc.), is
therefore to be regarded in the same light as yaunikki-n, from the theme
YAUNIKYA, i.e. it stands for sukantyi-n from sukantya-n, and hence
answers to the Zend accusatives, like ¢ gs990 tilirt-m for tiiryém (§..42.),
and to the Gothic, like hari from the base HAP YA (§. 135.).
t See §. 122.
INSTRUMENTAL, DATIVE SINGULAR. 175
. SANSKRIT. ZEND. GREEK. LATIN. LITHUAN. GOTHIC.
m. bharant-am, barént-em, _gépovt-a, ferent-em, .... fiyand.

m. dtmdn-am, asman-ém, Saipov-a, sermon-em,...+ ahman,
n. ndma’, nama’, TaAQY, nomen, e+e name.
m. bhrdtar-am, brdtar-em, matép-a, fratr-em, .... bréthar.
f. duhitar-am, dughdhar-ém, 6vyatép-a, matr-em, .... dauhtar.
m. ddtar-am, ddtdr-ém, dorie-a, datdr-em,
n. vachas, vaché,* eros, opus,
INSTRUMENTAL, DATIVE.
158. The instrumental is denoted in Sanskrit by = 4;
and this inflexion is, in my opinion, a (G. Ed. p.188.]
lengthening of the pronominal base W a, and identical with
the preposition wt d, “to,” “towards,” “up to,” which
springs from this pronoun, and appears only as a prefix.
The Zend 4 appears still more decidedly in its pronominal
nature in the compound mentioned at §. 156. Note *, Gaus
a-dém, “him,” “this,” (m.) fem. Gy_ss d-danm. As a
case-sign, 2s d generally appears abbreviated (see p. 163.
Note ft), even where this termination has been melted into
one with a preceding 2s a of the base; so that in this case
the primary form and the instrumental are completely
similar; e.g. asenbass zaésha, “ voluntarily,” asrrbasgas
azaésha, “ involuntarily,” (V. S. p. 12.) asySbassd5au skyaéthna,
“‘actione,” often occur; ays ana, “ through this™ (m.),
wee 5.0.50 paiti-béréta, “allevato."+ The long 4 appears
in the instrumental only in monosyllabic bases in os a;
thus aso khd, “proprio” V. S. p. 46.), from the base asyo
kha (Sanskrit @ swa, §. 35.). In Sanskrit a euphonic qn
is added to bases ending with short vowels in the masc,
* See §.56>.
+ Cf. Gramm. Crit. r. 638. Rem. This interesting instrumental form
was not known by Rask when he published his work on the Zend, and
it was not easy to discover it, on account of its discrepancy from the San-
skrit and the many other forms with final as a.
176 ral FORMATION OF CASES.
and neut. genders;* a final @ a, however, is, as in several

other cases, changed into v é; and the wid of the case-
suffix is shortened, as it appears to me, by the influence of
this clog of the base; as gam vriké-n-a, but wfraqat agni-
n-d, TCM vari-n-d, FAT siinu-n-d, wat madhu-n-d, from
ga vrika,’ &e. The Védas, however, exhibit further
remains of formations without the euphonic n, as WHat
swapnay-d for xa swapné-n-a from WH swapna, m. “sleep”
(see §.133.); Seat uru-y-d for TEUt uru-n-a, from FR urw,
“reat,” with a euphonic ay(8. 43.); wateat prabdhav-d, from
ware prabdhu, from arg bdhu, “ arm,” with the preposition
[G. Ed. p.189.] wpra. The Véda-form qyat swapnayd,
finds analogies in the common dialect in wat mayd,
“through me,” and wat twayd, “ through thee,” from the
bases ma and twa, the a of which in this case, as in the
loc., passes into é@ And from ufa pati, m. “Lord,” and
af@ sakhi, m. “ friend,” the common dialect forms instru-
mentals without the interposition of A 2, viz. Wat puy-d,
wear sakhy-d. Feminines never admit a euphonic n; but
d, as before some other vowel terminations, passes into
z é, that is to say, 7 is blended with it, and it is shortened
to aa; hence, fagm jihway-d (from jihwé+d). The Zend
follows in this the analogy of the Sanskrit.
159. As é in Gothic, according to §. 69., just like 4, re-
presents wt 4, so the forms thé, hvé, which Grimm (pp. 790.
and 798.) regards as instrumentals, from the demonstrative
base THA and the interrogative HVA, correspond very
remarkably to the Zend instrumentals, as avy khd from
the base sy kha. We must, however, place also své
in the class of genuine Zend instrumental forms, which
have been correctly preserved: besides své from SVA is also,
* The original has “ Stammen gen. masc. und fem.;” but genitives of
nouns in @ do not take a euphonic n, nor do feminine nouns ending in
short vowels use such an augment in the instrumental: here is no doubt
some typographic error.— Editor.
INSTRUMENTAL, DATIVE SINGULAR, 177
in respect of its base, akin to yo kid from kha (8. 35.).*

The meaning of své is “as” (&s), and the sé, which has arisen
in High German from sva or své, means both “as” and
“so,” &c. The case relations, however, which are expressed by
“as” and “so” are genuine instrumentals.t [G. Ed. p. 190.]
The Anglo-Saxon form for své is svd, in which the colouring
of the Zend so kid is most truly preserved. The Gothic
sva, “80,” is, according to its form, only the abbreviation of
své, as a is the short equivalent both of é and of d: through
this abbreviation, however, sva has become identical with
its theme, just as 2. ana in Zend is, according to §. 158,
not distinguished from its theme.
160. {As the dative in Gothic and in Old High German.
very frequently expresses the instrumental relation, and
the termination also of the dative is identical with the
Sanskrit-Zend instrumental character, shortened only, as
in polysyllabic words in Zend, it may be proper here to
describe at the same time the formation of the German
dative. In a bases it is in Gothic, as in Zend, identical
with the theme, and from VULFA comes vulfa, as asJercl
vehrka from VEHRKA. Moreover, there are some other
remarkable datives, which have preserved their due length,
and answer to the monosyllabic instrumentals thé, vé, své,
which have been already explained, viz. hvammé-h, hvar-
yammé-h, “cuique,’ and ainummé-hun, “ulli,’ for ainammé
* Grimm’s conjectures regarding the forms sva and své (IIT. 43.) ap-
pear to me untenable ; and an explanation of these forms, without the
intervention of the Sanskrit and Zend, is impossible. More regarding this
at the pronouns.
+ If“as” is regarded as “through which means, in which manner or
way,” and “so” as “through this means, in this way,” it is certain that
among the eight cases of the Sanskrit language there is none which woud
be adapted in the relative and demonstrative to express “as” and “ so.”
t The German dat. sing.is according to §. 356, Rem. 3., to be every-
where identified with the Sanskrit dative ; and so, too, the dat. pl. the m
of which apprvaches as closely to the Sansk. bhyas, Latin bus, Lith. mus,
as the instrumental termination bhic, Lith. mis.
N
hun (§. 66.).* Bases ini reject this vowel before the case-
sign; hence gast’-a for gasti-a: on the other hand, in the
u bases the termination is suppressed, and the base-vowel
receives the Guna: hence sunau, which will have been pro-
nounced originally su-nav-a ; so that, after suppressing the
termination, the vy has again returned to its original vowel
nature. The form sunav-a would answer to the Véda form
watgat pra-bdhav-d. In Zend, the bases which terminate
with 5 i and » u, both in the instrumental and before most
[G. Ed, p.191.] of the other vowel terminations, assume
Guna or notat pleasure. Thus we find in the Vend. S. p. 469,
ws»acgass bdzav-a, “brachio,” as analogous to WaTgat pra-
-bahav-d (§. 57.); on the other hand, p. 408, NBII zanthwa
from zantu, “the slaying,” “killing.” From >sy0 pansnu,
“dust,” we find, 1. c. p. 229, the form gs pansni, which
Anquetil translates by “par cette poussiere”; and if the read-
ing is correct, then pansni, in regard of the suppressed ter-
mination (compensation for which is made by lengthening
the base vowel ), would answer to the Gothic sunau.
161. Bases ending with a consonant have lost, in Ger-
man, the dative character: hence, in Gothic, fiyand, ahmin,
bréthr (§. 132.), for fiyand-a, ahmin-a, bréthr-a.t All femi-
nines, too, must be " pronounced to have lost the dative
sign, paradoxical as it may appear to assert that the Gothic
gibai, “dono,” and thizai, “ huic,” izai, “ei,” do not contain
any dative inflexion, while we formerly believed the ai of
gibai to be connected with the Sanskrit feminine dative
* Here the appended particle has preserved the original length of the
termination, as is the case in Zend in all instrumentals, if they are com-
bined with ass cha, “ and.”
+ The Old High German form fatere (for fatera), “ patri,” proceeds,
as do the genitive futere-s, and the accusative fatera-n, from a theme
FATERA, extended by a. The accusative fatera-n, however, is remark-
able, because substantives, so early as in the Gothic, have lost the accusa-
tive sign, together with the final vowel of the base. In Old High Germana
few other substantives and proper names follow the analogy of FATERA.
character 2 4i. But as we have recognised in the mas-

culine and neuter dative the Indo-Zend instrumental,
we could not, except from the most urgent necessity,
betake ourselves to the Sanskrit dative for explanation of
the Gothic feminine dative. This necessity, however,
does not exist, for, e.g., hveitai, “albae,” from HVEITO from
HVEITA, may be deduced from the instrumental rat
Swétay-4, “alba,” from gat swétd, by suppressing the ter-
mination, and changing the semi-vowel to a vowel in the
same manner as, above, sunau from sunav-a, [G. Ed. p. 192.]
or as the fem. handau, “ manui,” from handav-a. Analogous
with sunau, handau, are also the dative feminine i bases;
and, e.g., anstai, “ gratie,” has the same relation to its theme
ANSTI that handau has to HANDU.
162. In Old High German the forms diu, hviu, corre-
spond to the Gothic instrumentals thé, hvé; but authorities
differ as to the mode of writing them,* regarding which
we shall say more under the pronouns. The form hiu,
also, from a demonstrative base HJ, has been preserved in
the compound hiutu for hiu-tagu, “on this day,” “ to-day”
(see Grimm, p. 794), although the meaning is here pro-
perly locative. The Gothic has for it the dative himma-
-daga. This termination u has maintained itself also in
substantive and adjective bases masc. neut. in a and i,
although it is only sparingly used, and principally after the
preposition mit (see Graff, l.c. pp. 110, 111); mit wortu,
“with a word,” from WORT'A; mit cuatu, “ with good,” from
CUATA; mit kastu, “ with a guest,” from KASTYT. It is here
important to remark, that the instrumental in Sanskrit
very frequently expresses, per se, the sociative relation.
We cannot, however, for this reason look upon this u case
as generically different from the common dative, which, we
have already remarked, is likewise of instrumental origin
* With reference to their use with various prepositions we refer our
readers to Graff’s excellent treatise, “The Old High German Preposi-
tions,” p. 181, &e.
N2
and meaning: we rather regard the u* as a corruption

[G. Fd. p.193.] (although one of very ancient datc) of u,
just as in the neuter plural of pronouns and adjectives a u
corresponds to the short a of the Gothic and the older coguate
languages. In Lithuanian the a bases form their instru-
mental in a, which is long, and in which the final vowel
of the base has been melted down. That this %@, also, has
arisen from a long a, and thus, e.g. diewi is akin to the
Zend a»~259 daéva, “deo,” for 70254 daévd, appears to
me .the less doubtful, as also in the plural diewais answers
very surprisingly to »w5.w»70.39 daéviis, ta dévdis. More-
over, in many other parts of grammar, also, the Lithuanian
% corresponds to the Sanskrit wtd; e.g. in the plural
genitive. In feminine a bases, also, in Lithuanian, the
vowel of the base is melted down with that of the termi-
nation, but its quality is not changed; as, e.g. ranka
“manu,” from RANKA. In all other bases mi stands as
the termination, to which the plural instrumental termi-
nation mis has the same relation as, in Latin, bis to bi
(voBIS, tiBl); and, according to §. 63., I do not doubt
that in both numbers the m has arisen from bd.
163. The bases given in §. 148. form, in the instrumental
and in the Gothic, in the dative,
SANSKRIT. ZEND. LITHUANIAN. GOTHI(.
m. vriké-n-a, vehrka, wilki, vulfa.

f. jihway-d. hizvay-a, ranka, gibai.
m. paty-4, paithy-a, _—pati-mi, — gast’-a.
* Contrary to Grimm’s opinion, I cannot let the instrumental wu pass as

long, even not to notice its derivation from a short a; for, first, it ap-
pears, accordingtoNotker, in the pronominal forms diu, &c. without a cir-
caumflex (other instrumentals of the kind do not occur in his works);
secondly, like the short a, it is exchanged for o (§. 77.); hence, wio,
wéo, with wiu, wio-lih, huéo-lih, “ qualis”’ (properly, “similar to whom”);
thirdly, the length of this u cannot be deduced from the Gothic forms thé,
hvé, své, because these, in all probability, owe the retention of their long
vowel to their being monosyllabic (cf. §, 137.).
SANSERIT. ZEND. LITHUANIAN. GOTHIc.
f. prity-d, afrithy-a, awi-mi, anstai. “G

f. bhavishyanty-d, bishyainty-a, .... -e.s
m. sénu-n-d, pasv-a, sunu-mi, sunau. *>
f. tanw-d, tanv-a, ..-. handau. 2
f. vadhw-d, wa TARE wi Poo Ss
m.f.gav-d. gav-a, eet a vied
f. ndv-d, Pes taal beak
f. vdch-d, vich-a, ete oh, 2
m. bharat-d, barént-a, ~2e- fiyand,
m. dtman-t, asman-a, wea ahmin.
n. ndmn-d, ndman-a, » ees =Mnamin.
m. bhrda'r-d, brdthr-a, . 2. bréthr.
f. duhitr-d, dughdher-a, .... dauhtr.
m. . détr-d, dathr-a, PES » oh ®
n. — vachas-d, vacanh-a, Stars
164. In Sanskrit and Zend, é is the sign of the dative,

which, I have scarce any doubt, originally belongs to the
demonstrative base é, whence the nom. waq ayam (from
é+am), “this”; which, however, as it appears, is itself
only an extension of the base & a, from which arise most
of the cases of this pronoun (a-smdi, a-smdt, a-smin, &c.);
and regarding which it is to be observed, that the common
a bases, also, in Sanskrit in many cases extend this vowel
to é by the admixture of an i (§. 2.). The dative sign con-
sequently would, in its origin, be most intimately con-
nected with the case, which, as (§. 160.) was explained, de-
notes, in German, both the dative and instrumental rela-
tion, and occurs in Zend also with a dative signification.*
* E.g. Vend. 8. p. 452 spsweasg wssspsay

yawg sors Leda
63960 Liosasns6S Hadmé azizénditibis dadhditi csaitd-puthrim, “Hom
gives a splendid daughter to those who have not had offspring.” The
lithographed Codex, however, gives the form azizdnditibis as three words,
ws
[G. Ed. p.195.] We have here further to remark, that in

the pronoun of the 2d person the affix mq bhyam (from
bhi + am) in war tu-bhyam, “to thee,” stands in evident
relationship to the instrumental fa bhis in the plural.
The feminine bases in 4, @, d, and, at will also, those in i and
u, prolong in Sanskrit the dative termination zu é to z di;
with the final 4 of the base an i is blended; hence fagra
jihwéy-di from jivdi-di. On the other hand, $i and = u re-
ceive the Guna augment before z é, but not before the
broader 2 di; as Waa siéinav-é from simu. In Zend, femi-
nine @ and 7-bases, like the Sanskrit, have 4 for their termi-
nation: however, hizvdy-di is not used, but SAY JAI?¢ HY
hizvay-di, from the base hizvd, as long vowels in the penulti-
mate, in polysyllabic bases, are so frequently shortened.
Bases in $i have, in combination with the particle as cha,
preserved the Sanskrit form most truly, and exhibit, without
exception in this case, the form aywasyyas ay-aé-cha (see
§. 28.), e.g. MVOAKYIASOAS karstayaécha, “and on account
of the ploughing,” “in order to plough” (Vend. S. p. 198),
[G. Ed. p. 196.] from karste. Without cha, however, the
form w»¢ e¢ is almost the sole one that occurs, e.g.
Pepghsys khareteé, “in order to eat,’ from Soghsyo khareti.
This form, I doubt not, has arisen from » yas ay-é, by re-
jecting the semi-vowel, after which the preceding » a has
become ¢ e (8. 31.). Forms like pops afrité* or ep Nau
afrite, which sometimes occur, and are most corrupted, may
MASSE S(O IAW JANE $595 azi zdnditi bis. Such separations in the middle of a
word are, however, in this Codex, qnite common, I entertain no doubt
of the correctness of the length of the a, both of zé and ndi; and I anti-
cipate a variety azizanaitibis or—bis. Probably also csaété is to be read for
csait6. Anquetil translates: “O Hom, donnez 4 la femme, qui n’a pas
encore engendré, beaucoup d’enfans brillans.” We will return to this passage
hereafter ; and we will here further remark that, at the same page of the
Vend. S., the instr. 19.5570.45 aébis also occurs in the sense of “to them.”
* Cf, pe 286 Note te
ae! INSTRUMENTAL, DATIVE SINGULAR. 183
rest on errors in writing.* Bases in u may take Guna;

- 69. psu van-hav-é from regu vanhu, “pure”; or not,
as ~»Gx? rathv-é from >? ratu, “great,” “lord.” The
form without Guna is the more common. A euphonic 9) y
also is found interposed between the base and the termi-
nation (§. 43) e.g. PHyyjse tanu-y-é, “ corpori.”
165. Bases in # a add to the case-sign é also an Wa;
but from z é (=a +i) and a is formed Wa aya; and this,
with the a of the base, gives dya, thus zara vrikdye.
Hence may have arisen, by suppressing the final a, the
Zendian sausfveh vehrkdi, after which the preceding semi-
vowel must return to its vowel nature. It might, how-
ever, be assumed, that the Zend has never added an @ to
the dative é, and that this is a later appearance in Sanskrit,
which arose after the division of languages; for from a +é
is formed, quite regularly, di (§. 2.). The Sanskrit forms
also, from the particle # sma, which is added to pro-
nouns of the 3d person, the dative @ smdi; and thus, e.g.
aa hasmdi, “to whom”? answers to the Zend SAWEISS
kahmdi. The Sanskrit, in this case, abstains from adding
the Wa, which is elsewhere appended to the dative z é;
since @ sma, already encumbered with the preceding prin-
cipal pronoun, cannot admit any superfluity in its termi-
nation, and for this reason gives upitsradi- [G. Ed. p.197.]
cal & a before the termination ¥4 in in the locative case
also, and forms sm’-in for smén.
166. The particle & sma, mentioned in the preceding see-
tion, which introduces itself between the base and the ter-
mination, not only in the singular, but (and this, in fact,
occurs in pronouns of the two first persons) in the plural also,
ifnot separated from both—as I have first attempted to shew
* eo Sas Gfrite is undoubtedly incorrect: however, ¢ ¢ is often

found erroneously for 7 e in other forms also.
in my Sanskrit Grammar—gives to the pronominal declen-

sion the appearance of greater peculiarity than it in fact
possesses. As this particle recurs also in the cognate
European languages, and there, as I have already elsewhere
partly shewn, solves several enigmas of declension, we
will therefore here, at its first appearance, pursue all its
modifications and corruptions, as far as it is possible. In
Zend, sma, according to §. 53., has been changed to hma;
and also in Prakrit and Pali, in the plural of the two first
persons, the s has become A, and besides, by transposition
of the two consonants, the syllable hma has been altered
to mha; eg. Prakrit wee amhé, “ we” (dupes), Pali seer
amhdkam, Zend ¢¢9.»G5 ahmakem, juOv. From the Prakrit-
Pali mha we arrive at the Gothic nsa in u-nsa-ra, yudr,
u-nsi-s,* “nobis,” “nos.’ In that the Gothic has left the
sibilant unaltered, it stands on an older footing than the
Pali and Prakrit; and on the other hand, by the change
of m into n, for more facile combination with the follow-
ing s, it rests on-a more modern stage. We cannot,
therefore, any longer assume the ns of uns, “nos,” to be
[G. Ed. p. 198.] the common accusative termination, as we
have formerly done in unison with Grimmt—cf. vulfa-ns,
gasti-ns, sunu-ns—and thence allow it, as though it had be-
come a property of the base, to enter into some other cases,
and connect it with new case-terminations. To this is op-
posed, also, the 2d person, where izvis (i-zvi-s) stands in the
accusative, and yet in essentials the two persons are identical
in their declension; uns, “ nobis,” “ nos,” stands, therefore, for
unsi-s (from unsa-s), and this has s as the case-suffix, and u-nsa
(weakened from u-nsi) as the compound base. And we
* The a being changed into s, according to r. 67,

+ I. 813. “unsara appears to be derived from the accusative uns, as
also the dative unsis, which, with izwis, preservesa parallel sound to the
dative singular.” Cf, I, 813, 34,
cannot, also, any longer regard the u of unsa-ra, “ nostri,”

_&c. as the vocalized v of veis, “we,” although the i of
izvara, “vestri,” &c. can be nothing else than the vocalized
y of yus, “ your”; for in Sanskrit, also, the syllable yu of
yiyam, “ye,” (§.43.) goes through all the oblique cases,
while in the Ist person the q v of qua vayam, “we,” is
limited to the nominative, but the oblique cases combine a
base 8 a with the particle sma. This a, then, in Gothic,
through the influence of the following liquid, has become
u; hence, unsa-ra, &c. for ans-ara (§. 66.).
167. As in Zend, the Sanskrit possessive @ swa shews
itself* in very different forms in juxta-position with diffe-
rent letters, so [ believe I can point out the particle
™@ sma in Gothic at least under four forms; namely,
as nsa, zva, gka, and mma. The first has been already
discussed; the second—zva, and in a weakened form zvi—
occurs in the pronoun of the 2d person, in the place where
the Ist has nsa (nsi); and while in the cognate Asiatic
languages (Sanskrit, Zend, Pali, Prakrit), as also in Greek and
Lithuanian, the two pronouns run quite [G. Ed. p. 199.]
-parallel in the plural, since they both exhibit the interposed
particle under discussion, either in its original form, or simi-
larly modified, in Gothic a discrepancy has arisen between the
two persons, in that the syllable sma has in them been
doubly transformed. The form zva from sma rests, first,
on the not surprising change of the s into z (§. 86. 5.);
secondly, on the very common change of m and v (§. 63.).
168. From the Gothic downwards, the particle sma has
been still further corrupted in the German dialects, in the
pronoun of the 2d person, by the expulsion of the sibilant.
The Old High German i-wa-r has nearly the same relation
to the Gothic i-zva-ru that the Homeric genitive roto has
* See Ann. of Lit. Crit. March 1831, p. 376, &e.

to the Sanskrit wet tasya, which is older than the Homeric

form. Compare, without intervention of the Gothic, the
Old High German i-wa-r, i-u, i-wi-h, with the Sanskrit
yu-shmd-kam, yu-shma-bhyam, yu-shmd-n, and with the Li-
thuanian yu-st, yt-mus, yi-s: thus it would be regarded as
settled, that the w or u belongs to the base, but is not the
corrupted remainder of a far-extended intermediate pro-
noun; and it would be incorrect to divide iw-ar, iw-ih, iu,
for i-wa-r, &c. I, too, formerly entertained that erroneous
opinion. A repeated examination, and the enlarged views
since then obtained through the Zend, Prakrit, and Pali,
leave me thoroughly convinced, that the Gothic interme-
diate syllable zva has not been lost in High German, but
that one portion of it has been preserved even to our
time (e-we-r from i-zva-ra, e-u-ch from i-zvi-s, Old High
German i-wi-h): on the other hand, the wu of the base yu
(q yu), as in Gothic so also in the oldest form of the High
[G. Ed. p.200.] German, is rejected in the oblique cases,
both in the plural and in the dual* ; and the Gothic i- zva-ra,
Old High German i-wa-r, &c., stand for yu-zva-ra, yu-wa-r.
The Old Saxon, however, and Anglo-Saxon, like the Lithua-
nian, shew themselves, in respect to the preservation of the
base, more complete than the Gothic, and carry the u,
which in Anglo-Saxon has become o, through all the
oblique cases: iu-we-r, o-ve-r, “vestri,” &c. If merely
the two historical extremes of the forms here under dis-
cussion—the Sanskrit and New German forms—be con-
trasted with one another, the assertion must appear very
paradoxical, that ever and qaraq yushmdkam are connected,
and, indeed, in such wise, that the wu of ever has nothing
* So much the more remarkable is the «, which is still retained in the

North Friesian dialect (Grimm, p. 814), where, e.g. yu-nkeer, yu-nk, on
regard to the base, distinguishes itself advantageously from the Gothic
i-yqua-ra, i-ngvi-s.
in common with the u of q yu, but finds its origin in the

m of the syllable = sma.
169. The distinction of the dual and plural in theoblique
eases of the two first persons is not organic in German; for
the two plural numbers are distinguished originally only by
the case-terminations. These, however, in our pronouns
are, in Gothic, the same; and the difference between the
two plural numbers appears to lie in the base—ugka-ra,*
vOiv, UNSO-TA, Huov, igqva-ra, TPAiy, izva-ra, budv. But from
a more close analysis of the forms in the two plural num-
bers, and from the light afforded us by the cognate Asiatic
languages, it appears that the proper base is also identical
in the two plural numbers; and it is only the particle sma
combined with it which has become doubly corrupted, and
then the one form has become fixed in the dual, the other in
the plural. The former comes nearest to ([G. Ed. p. 201.)
the Prakrit-Pali form ve mha, and between u-nsa-ra and
u-gka-ra (=u-nka-ra) an intervening u-nha-ra or u-mha-ra
must be assumed. At least I do not think that the old s be-
came k at one spring, but that the latter is a hardened form
of an earlier h, which has remained in the Prakrit and Pali,
as in the singular nominative the k of ik has been developed
from the h of sgqaham. The second person gives, in
Gothic, gv (=kv §. 86. 1.) for k, while the other dialects leave
the guttural the same form in both persons: Old High Ger-
man, u-ncha-r, i-ncha-r; Old Slavonic, u-nke-r, i-rke-r;
Anglo-Saxon, u-nce-r, i-nce-r. It would consequently
appear proved that the dual and plural of the two first
persons are not organically or originally different, but be-
long, as distortions and mutilations of different kinds, to
one and the same original form; and that therefore these
two pronouns have preserved the old dual just as little as
* It must not be overlooked,

that here g before k only represents the
nasal answering to k (86. 1.).
the other pronouns and all substantive and adjective de-

clensions.
170. The fourth form in which & sma appears in Gothic
is that which I first remarked, and which [ have brought
forward already in the “Annals of Oriental Literature”
(p. 16). What I have there said, that the datives singular,
like thamma, imma, haye arisen, by assimilation, from tha-
sma, i-sma, I have since found remarkably confirmed by
the Grammar of the Old Prussian published by Vater, a
language which is nearly connected with the Lithuanian
and Gothic, since here all pronouns of the third person
have smu in the dative. Compare, e.g. antar-smu with the
Gothic anthara-mma, “to the other”: ka-smu with the
Gothic hva-mma, “to whom?” We have also shewn in
Greek, since then, a remnant of the appended pronoun
sma similar to the Gothic, and which rests on assimilation,
[G. Ed. p.202.] since we deduced the Holic forms a@-pp-es,
U-pp-es, &e., from &-cope-es, v-cpe-es, to which the common
forms 7pels, duets, have the same relation that the Old High
German de-mu has to the Gothic tha-mma, only that jets, duets,
in respect to the termination ets, are more perfect than the
/Kolic forms, since they have not lost the vowel of the particle
ope, but have contracted pe-e¢ to pets.
171. The Gothic datives in mma_ are, as follows from
§. 160., by origin, instrumentals,* although the particle sma
in Sanskrit has not made its way into these cases, and e.g.
wa téna, “through him,” not tasména, or, according to the
Zend principle (§. 158.), tasma (for tasmd), is used ;—I
say, according to the Zend principle; for though in this
* The difference between the forms thé, hvé, explained at §.159., and
the datives tha-mma, hva-mma, consists first in this, that the latter express
the case relation by the affixed particle, the former in the main base;
secondly, in this, that thamma, hvamma, for thammé, hvammé, on account
of their being polysyllabic, have not preserved the original length of
the termination (cf. §. 137.) ee
language ima has entered mv the instrutmental mascusine

and neuter, this case in the base ta could only be asgase fahma
or avg tahmd (from ta-hma-d). In the feminine, as we
can sufficiently prove, the appended pronoun really occurs in
the instrumental; and while e.g. from the masculine and neuter
base asyas ana, “this” (m.), “this” (n.), we have found the
instrumental of the same sound 41s ana not anahma, from
the demonstrative base 1s a occurs rather often the feminine
instrumental .5))¢.5 ahmy-a, from the fem. base y§s ahmi,
increased by the appended pronoun.
172. The Sanskrit appended pronoun [G. Ed. p. 203.]
#4 sma should, in the feminine, form either = smd or wt
smé: on the latter is based the Zend form ¥¢ hmé, mentioned
at §.171. But in Sanskrit the feminine form wt smi has
been preserved only in such a mutilated condition,* that be-
fore my acquaintance with the Zend I could not recognise it.
From ta-smé must come the dative ta-smy-di, the gen. and
ablative ta-smy-ds, and the locative ta-smy-dm. These forms,
by rejecting the m, have become abbreviated to wa ta-sy-di,
Weara_ta-sy-ds, wWeava ta-sy-dm; and the same is the case
with the feminine pronoun sm? in all similar compounds; so
that the forms mentioned appear to have proceeded from the
masculine and neuter genitive tasya, by the annexation of new
case-terminations. This opinion was the more to be relied
on, that in Gothie, also, the feminine forms thi-zés, “ hujus,”
* The Zend, too, has not everywhere so fully preserved the feminine
hmi, as in the instr. a-/my-a; but in the genitive, dative, and ablative
has gone even farther than the Sanskrit in the demolition of this word,
and has therein rejected not only the m but also the. The feminine
gusezu a-nh-Go (§. 56*.), “hujus,’”’ for a~hmy-do, often occurs; and for it
also guser 5.5 ainh-do, in which the i is, to use the expression, a reflec
tion of the lost yy y(§. 41.). From another demonstrative base we find
thé dative SAVY U3 ava-nh-di, and more than once the ablative
PEASY PAS ava-ni-at for ava-hmy-di, ava-hmy-ét.
thi-zai, “huic,” might be deduced from the masculine genitive

this, by the addition of the terminations ds and ai; and as, too,
in Lithuanian, the whole of the oblique cases singular of the
Ist and 2d person stand in close connection with the Sanskrit-
Zend genitives AA mama, »js3G mana, WA tava, rs» tava,
and have the same as base. After discovering the Zend fe-
[G. Ed. p.204.] minine pronominal forms in hmy-a in the
instrumental and locative—in the latter fur hmy-anm—the
above-mentioned forms in Sanskrit cannot be regarded other-
wise than as abbreviations of fa-smy-di, &c., as this is far more
suited to the nature of the thing. The Gothic forms then,
thizds, thizai, will be regarded as abbreviated, and must be di-
vided into thi-z6-s, thi-zai. The masculine and neuter appended
pronoun sma must, for instance, in Gothic give the feminine
base SMO= ett smé, as BLINDO, nom. blinda, “ceca,” from
BLINDA, m. n. (nom. blind’-s, blinda-ta). SMO, however,
by the loss of the m, as experienced by the Sanskrit in the
feminine, has become SO; but the s, on account of its posi-
tion between two vowels (according to §. 86.5.), has become z.
Therefore, thi-z6-s * has only s as case-sign, and the dative
thi-zai, like gibai in §. 161., is without case character. With
the masculine and neuter genitive thi-s, therefore, thi-zé-s, thi-
zai, have nothing in common but the demonstrative theme
THA, and the weakening of its a to i (§. 66.).
173. Gothic adjective bases in a (Grimm’s strong ad-
jectives) which follow the pronominal declension, differ
from it, however, in this point, that they do not weaken
the final a of the base before the appended pronoun to i,
but extend it to ai, and form the feminine dative from the
simple theme, according to the analogy of the substan-
tives:t hence blindai-zé-s, blindai, not blindi-zé-s, blindi-zai.
* Of. §. 356. Rem. 3. p. 501, last line but seven.

+ With respect to the extension of the a to ai, compare the gen. pl. and
Sanskrit forms, as ¢é-bhyas, “iis, téshdm, “eorum,” for ta-bhyas, ta sdm,
174. The Zend introduces our pronominal syllable sma

in the form of Amz also into the second, and probably into
the first person too: we find repeatedly, in the locative,
$605 thwa-hm’-t, instead of the Sanskrit [G. Ed. p. 205.]
fa tevay-i, and hence deduce, in the Ist person, ma-him’-é,
which we cannot quote as occurring. The Prakrit, in this
respect, follows the analogy of the Zend; and in the 2d per-
son gives the form qafe#t tuma-sm’-i, “in thee,” or, with
assimilation, waft tumammi, with qi tumé (from fuma-)
and wg état; and aafe mama-sm’-i or Raf mama-mmi, “in
me,” together with the simple 4e maé and ag mai.* Ought
not, therefore, in German also, in the singular of the two
first persons, a remnant of the pronominal syllable sma to be
looked for? The s in the Gothic mi-s, “to me,” thu-s,
“to thee,” and si-s, “to himself,” appears to me in no
other way intelligible; for in our Indo-European family of
languages there exists no s as the suffix of the instrumental
or dative. Of similar origin is the s in the plural u-nsi-s,
“nobis,” “nos,” i-zvi-s, “ vobis,” “vos”; and its appearance in
two otherwise differently denoted cases cannot therefore be
surprising, because this s is neither the dative nor accusative
character, but belongs to a syllable, which could be declined
through all cases, but is here deprived of all case-sign. In
u-nsi-s, i-zvi-s, therefore, the Sanskrit = sma is doubly con-
tained, once as the base, and next as the apparent case-suffix.
I am inclined, also, to affirm of the above-mentioned Prakrit
forms, fu-ma-sm'i, “in thee,” and ma-ma-sm’i, “in me,”
that they doubly contain the pronominal syllable sma, and
that the middle syllable has dropped a preceding s. For
there is no more favourite and facile combination in our
class of languages’ than of a pronoun with a pronoun; and
what is omitted by one dialect in this respect is often
afterwards supplied by another more modern dialect.
* See £ssai sur

le Pali, by E. Burnouf
and Lassen, pp. 173.175.
[G. Ed. p. 206.] 175. The k in the Gothic accusatives mi-k,

thu-k, si-k (me, te, se), may be deduced, as above, in u-gka-ra,
vaiv, &c., from s, by the hardening of an intervening h; so
that mi-s is altered to mi-h, and thence to mi-k; and there-
fore, in the singular, as also in the plural, the dative and ac-
cusative of the two first persons are, in their origin, identical.
In Old High German and Anglo-Saxon our particle ap-
pears in the accusative singular and plural in the same
form: Old High German mi-h “me,” di-h, “thee,” u-nsi-h,
“us,” i-wi-h, “you”; Anglo-Saxon me-c, “me,” u-si-c, “us,”
the-c, “thee,” ev-vi-c, “you”: on the other hand, in the
dative singular the old s of the syllable sma has become r
in the High German, but has disappeared in the Old Saxon
and Anglo-Saxon: Old High German mi-r, di-r; Old
Saxon mi, thi; Anglo-Saxon me, the.
176. In Lithuanian 4 sma appears in the same form
as in the middle of the .above-mentioned (§. 174.) Prakrit
forms; namely, with s dropped, as ma; and indeed, first, in
the dative and locative sing. of the pronouns of the 3d per-
son and adjectives ; and, secondly, in the genitive dual of the
two first persons: we cannot, however, refer to this the m,
which the latter in some cases have in common with the
substantive declension. The pronominal base 7'4, and the
adjective base GE RA, form, in the dative, té-mui, “ to thee,”
gerd-mui, “to the good” (shortened tém, gerdm), and in the
locative ta-mé, gera-mé; and if -mui and -mé are compared
with the corresponding cases of the substantive a bases, it
is easily seen that mui and mé have sprung from ma. The
pronouns of the two first persons form, in the genitive dual,
mu-mi, yu-mé, according to the analogy of pond, “of the
two lords.”
* We have a remnant of a more perfect form of the particle wt sma in

the locative interrogative form ka-mmé, “where”? Sansk. aay ka-smin,
“in
177. Lithuanian substantives have i for ([G. Ed. p.207.]

the dative character, but i bases have ei*; a final a before
this i passes intou; hence wilku-i. Although we must refuse
a place in the locative to the dative i of the Greek and Latin.
still this Lithuanian dative character appears connected with
the Indo-Zend é, so that only the last element of this diph-
thong, which has grown out of a+i, has been left. For
the Lithuanian has, besides the dative, also a real locative,
which, indeed, in the a bases corresponds exactly with the
Sanskrit and Zend.
178. The nominal bases, Sanskrit, Zend, and Lithuanian,
explained at §. 148., excepting the neuters ending with a
vowel and pronouns, to the full declension of which we
shall return hereafter, form in the dative:
SANSKRIT. ZEND. LITHUANIAN,
m. vrikdya, véhrkdi, wilku-i.
f. jihway-di, hizvay-di, ranka-i.
m. paty-ét paite-é?} pach-ei.
f. pritay-é, afrite-é, dwi-ei.
f. bhavishyanty-di, bishyainty-di,
m. sénav-é, pasv-é, sunu-i.
“in whom,” which, according to the common declension, would be

RA kasmé (from kasma-i). Compare the Gothic hvamma, “ towhom?”
for hvasma.
* The form dwiui, with dwiei appears to admit of being explained as
arising from the commixture of the final vowel of the a@ bases,
+ The form
qw@ patyé is, with respectto its want of Guna, irregular,
and should be Wad patayé.
t In combination with ass cha we find in V.S., p. 473. 5973995250
paithyé-cha,
and hence deduce for the instramental (p. 193 G. Ed.) the form
paithya, while, according to ‘. 47., also paitya might be expected. From *
sSasw haci, “friend,” I find in V. S., p. 162, the instrumental asyys@asey
hacaya with Guna, after the analogy of the as»asgus bdzava, mentioned
at §. 160.
oO
ze FORMATION OF CASES,
SANSKRIT. ZEND, LITHUANIAN.

et f. tanav-é, tanu-y-é,? + sgl
2 f. vadhw-di, nae a o-sialin
= m.f. gav-4, gav-é, vee e
tf f. ndv-é, 5 sie id ae
Z f. odch-é, vdch-é, ao whe
m. bharat-é, barént-é, ccee
m. dtman-é, asmain-é, <a ceke
n. ndmn-é,t ndmain-é, cece
m. bhrétr-é, brdthr-é, eieelin
f. duhitr-é, dughdhér-é,+ eke
m. ddtr-e, ddthr-é, 3 Lee
vachas-é,} vachanh-é, cere
* I give wyy> {3 tanuyé with euphonic 2, because I have found this

form frequently, which, however, cannot, for this reason, be considered eg
peculiar to the feminine ; and, instead of it, also tanvé and tanavé may be
regarded as equally correct. Cf. §. 43., where, however, it is necessary to
observe, that the insertion of a euphonic $3 y between u and é is not
every where necessary ; and, for instance, in the dative is the more rare form.
t The ¢ ¢ in wiee.9 >s dughdheré, and in the instr. wee 9 >s
dughdhéra, is placed there merely to avoid the harsh combination of three
consonauts. I deduce these forms from the plural genitive Gy” ce.9 >9
dughdhér-anm, for & g7e.9 28 dughdhr-anm.
t Respecting are ndmné, for aTaay némané, and so in the instru-
mental ATQt ndmnd, for ATHAT namané, see §.140, In Zend, in this and
similar words, I have not met with the rejection of the a in the weakest
cases (§. 130.), but examples of its retention, e.g. in the compound aocté-
-naman, whence the genitive aocté-ndmané (Vend.S. p. 4, and frequently).
I consider the initial a in this compound as the negation, without eupho-
nie n; for in all probability it means “ having untold (countless) names.”
SimiJar compounds precede, viz. EAs oass POW ISI AS Tareas
4JSGLLIAGS hazanré-ghabshahé batvare-chashmané, “of the thousand
eared, ten thousand eyed.” Cf. Anquetil I1. 82. In words in van, on
the other hand, ss a is rejected in the weakest cases, and then the
» v becomes > u or bo. Regarding the addition of the » ¢ in J5rsGawsy
ndmainé, see §, 41.
ABLATIVE SINGULAR. 195
ABLATIVE.
179. The Ablative in Sanskrit has qt [G. Ed. p. 209.]

for its character, regarding the origin of which there can no
longer be any uncertainty, as soon as the influence of
pronouns on the formation of cases has been recognised, as
we are conducted at once to the demonstrative base fa,
which already, in the neuter nominative, and accusative,
has assumed the nature of a case-sign, and which we shall
subsequently, under the verb, see receiving the function of
a personal termination. This ablative character, however,
has remained only in bases in ¥ a, which is lengthened
before it; a circumstance that induced the Indian Gramma-
rians, who have been followed by the English, to represent
‘wa dt as the ablative termination. It would therefore be
to be assumed, that in ata vrikdt the a of the base has
been melted down with the 4 of the termination.*
180. M. E. Burnouft has been the first [G. Ed. p. 210.]
to bring home the ablative character to a class of words in
Zend which had lost it in Sanskrit, and whence it can be
satisfactorily inferred that a simple ¢, and not di, is the true
ablative character. We mean the declension in u, of which
hereafter. As regards bases in a, which in Sanskrit alone
have preserved the ablative, we have to observe, that in
* I have drawn attention already, in the first (German) edition of my

Sanscrit Grammar, to the arbitrary and unfounded nature of this assump-
tion ()§. 156. and 264.); and I have deduced from the ablatives of the
pronouns of the two first persons (maz, twat) that either at with short a, or,
more correctly, a simple ¢, must be regarded as the ablative termination.
This view I supported in the Latin edition of my Grammar, on the ground
that in old Latin also a simple d appears as the suffix of the ablative. But
since thea the justness of my opinion regarding the Sanskrit ablative has
been still more emphatically confirmed by the Zend language, because the
Zend stands in a closer and more evident connection with the Sanskrit
than does the Latin. -
+ Nouveau Journal Asiatique 1829, tom. III. 311

02
196 ‘ FORMATION OF CASES,
Zend also the short vowel is lengthened, and thus roswghorel

vehrkd-t answers to Tara vrikd-t. ‘Bases in + i have éi-t
in the ablative: whence may be inferred in Sanskrit ablatives
like wit paté-t, Wtid prité-t (8. 33.), which, by adding Guna
to the final vowel, would agree with genitives in és. The
Zend-Avesta, as far as it is hitherto edited, nevertheless
offers but few examples of such ablative forms in sh di-t:
I owe the first perception of them to the word rosbog baw
dfritéit, “ benedictione,” in a passage of the Vendidad,* ex-
plained elsewhere, which recurs frequently. Examples of
masculine bases are perhaps rosb7p.nsroasdg rosbyys? rajoit
zaratustréit, “ institutione zaratustrica” (V. S. p. 86), although
otherwise sas) rajt, which I have not elsewhere met with,
is a masculine: the adjective base zaratustri, however, be-
longs to the three genders. From 333 gairi, “ moun-
[G. Ed. p.211.] tain,” occurs the ablative resbhse gardit
in the Yescht-Side.f Bases in u have robs ao-tt in the
ablativel]; and in no class of words, with the exception of
* See Gramm. Crit. add. ad r. 156.

+ What Anquetil IIT. 170. Rem. 4, writes gueréed can be nothing else
than the ablative rosso garéit, for Anquetil generally expresses
© by gu, »s by e, sl by 6e, and po by d. The nominal base 5752500 gairi,
however, is treated in Zend as if gari was the original form, and thei
which precedes the 7 was produced by the final %, as remarked by
M. Burnouf in the article quoted at p. 173, and confirmed by the genitive
wosbirs garéis, That, however, which is remarked by M. Burnouf,
], c. with respect to the genitive, and of which the Vend. S, p. 64. affords
frequent proof in the genitive wsh~o.so patdis, must also be extended to
the ablative in dit ; ‘and thei, which, according to §, 41., is adduced through
the final 5 i of the base, is dropped again before this termination.
t For this we also find ROE eut; 9. RO>E99 14956 mainyeut from
mainyu.
|| Interchanges of Go and 4 6 are particularly common, owing to the
slight difference of these letters. Thus, e.g. for bsg mraét, “he
spoke,’’ occurs very frequently Lw7¢ mraot ; the former, however, is,
as we can satisfactorily prove, the right reading; for, first, it is supported
by
-ABLATIVE SINGULAR. 197
that in a, does the ablative more frequently occur, although

these words are in number but five er six, the ablative use
of which is very frequent; e.g. robase pgs
1544 ddonhadét, “ crea-
tione,” from ddonhu, in a passage explained elsewhere*
robasew anhaé-t, “mundo,” from »w3w anhu ; robasjasyo
tanaét “corpore,” from >ys@ tanu. Bases ending with con-
sonants are just as little able to annex the (G. Ed. p. 212.)
ablative 9s ¢ without the intervention of another letter, as
the accusative is to annex m without an intermediate letter;
and they have af as their termination, numerous examples of
which occur; e.g. roasdas ap-at, “ aqua”; roas7Gaus dthr-at,
“igne”; rexsyasG essays chashman-at, “oculo”; noasyaswz
gus
ndonhan-at “naso”; roxe» 74 druj-at, “ demone” ; PONS
vis-at, “loco” (cf. vicus, according to §. 21.). Owing to
the facile interchange of the » a with ww 4, roas Gt is
Sometimes erroneously written for gos af; thus, Vendidad
S. p. 338, rospyaysbass saéchant-dt for rospyyayshass saé-
chant-at “lucente.” Bases in u sometimes follow the
by the Sanskrit form ata

w abrét, for which the irregular form watt
abrav-it is used; and secondly, it answers to the Ist pers. mraém (V.S.
p. 123) : thirdly, the Sanskrit qf é is, in Zend, never represented by bas
ao, but by L 6, before which, according to §.28., another 4s a is placed,
hence bas aé: on the other hand, bias ao represents u, in accordance with
§. 32 and §. 28. If, then, »s980 pasu formed in the ablative robasond
pasaot, this would conduct us to a Sanskrit W3t pasu t; while from the
ablatives resbp shuafritéi-t, rosbJoawrpeavtss zaratustréi-t, resbue
garéi-t, and from the analogy, in other respects, with the genitive, the
Guna form, Uy pasé-t must bededuced. Moreover, in the Vend. S.
the ablative form bas aé-t actually occurs; for at p. 102 (asguaser
ROA DW $256 pobasse uh hacha vanheaé-t mananh-at, “from pure
spirit”) occurs vanheaét, the ablative of vanhu; and the € & preceding
the @ is an error in orthography, and vanhaét is the form intended:
p. 245 occurs robawgy anhaét, “* mundo,” from anhu,
* Gramm. Crit. §. 640. ann, 2.
consonantal declension.in having ners at as the ablative ter-

mination instead of a mere ¢; just as in the genitive, besides
a simple s, they exhibit also an 6 (from as, §. 56”.), although
more rarely. Thus, for the above-mentioned Laspse
tanaot, “corpore,” occurs also tanv-at (Vend. S. p. 482).*
Feminine bases in ww 4 and 37 have roaw dt in the ablative,
as an analogous form to the feminine genitive termination
ata ds, whence, in the Zend gus do; eg. ROWIINEA 9
dahmay-dt, “ preclara,” from s0G5_4 dahmd ; roaussasias)s
urvaray-at) “ arbore,” from sshs»7> urvard; 3997S 7ass
[G. Ed. p. 213.] barethry-dt, “ genitrice,” from Jee bare-
thri.+ The feminine bases also in u, and perhaps also those
in i, may share this feminine termination nos dt; thus,
from zantu, “ begetting,” comes the ablative zanthw-dt (cf.
Gramm. Crit. §. 640. Rem. 2.). Although, then, the ablative
has been sufficiently shewn to belong to all declensions in
Zend, and the ablative relation is also, for the most part,
denoted by the actual ablative, still the genitive not un-
frequently occurs in the place of the ablative, and even
adjectives in the genitive in construction with sub-
stantives in the ablative. ‘Thus we read, Vend S. p. 479,
arsb J39A59S596 AG ROMY rosa POUUBUMAS ASAI hacha
avanhdlt visat yat mdzdayasnéis, “ex hac terrd quidem maz-
dayasnicd.”
* Burnouf writes tanavat, probably according to another Codex.

I hold both forms to be correct, the rather as in the genitive, also, both
tanv-6 and tanav-é occur ; and in general, before all terminations beginning
with a vowel, both the simple form and that with Guna are possible.
+ Vendidad Sade, p. 436: Vos ghsostsys bsJoeh sGs,c_
GeIGod ASHASeY rosy s7Gehass roawyy/sasgg asgeos Yatha vehrké
chathwaré-jangré nishdarédairydt baréthryat hacha puthrém, “‘As a wolf,
a four-footed animal, tears a child from its mother.” This sentence is
also important as an example of the intensive form (cf. Gramm. Crit.
§. 363.). The Codex, however, divides incorrectly nishdaré dairydt.
+ Regarding this form, see p. 172. Rem.
181. The Old Roman corresponds with the Zend in re-

gard to the designation of the ablative; and in those two
memorials of the language, that on the Columna rostrata, and
the S. C. de Bacchanalibus, which are the most important
inscriptions that remain, all ablatives end with d; so that
it is surprising that the ablative force of this letter could
be overlooked, and that the empty name of a paragogic d
could be held satisfactory. Bases ending with a conso-
nant use ed as ablative suffix, as in the accusative they
have em instead of a simple m: hence, forms like pre-
sent-ed dictator-ed, answer to the Zend saéchant-at dthr-at
(lucente igne); while navale-d* preda-d, inalto-d mari-d,
senatu-d, like the above-mentioned Zend forms reslue
garéi-t, “ monte,” robsjx tanaéd-t, “ corpore, &c.;
and in
Sanskrit yara vrikd-t, “lupo,” have a simple T sound to
denote the ablative. The Oscan also takes the ablative
sign d through all declensions, as appears from the remark-
able inscription of Bantia, e.g. dolu-d [G. Ed p. 214.]
mallu-d, cum preivatu-d, touta-d presenti-d.t It may be pre-
liminarily observed, that,in the 3d person of the imperative,
old Latin and Oscan forms like es-tod, es-tud—for es-to, and
therefore with a double designation of person—correspond
remarkably to similar Véda forms with which we are
hitherto acquainted only from Panini; e.g. ataata jiva-tat,
which signifies both “vivat” and “vive,” but in the latter
sense is probably only an error in the use of the language
(cf. vivito as 3d and 2d person).
182. In classical Latinity a kind of petrified ablative
form appears to be contained in the appended pronoun ~
met, which may be transferred from the Ist person to the
others also, and answers to the Sanskrit ablative mat,
“from me.” But it is possible, also, that met may have
* The
e here belongs to the base, which alternates between ¢andi.
+ SeeO. Miiller’s Etruscans, p. 36,
200 FURMATION OF CASES.
dropped an initial s, and may stand for smet, and so be-

long to the appended pronoun @ sma, explained in §. 165.
&ec., corresponding with its ablative smdt, to which it
stands in the same relation that memor (for mesmor) does
to = smri—from smar, §. 1—‘‘to remember.” The com-
bination of this syllable, then, with pronouns of the three
persons, would require no excuse, for 4 sma, as has
been shewn, unites itself to all persons, though it must
itself be regarded as a pronoun of the 3d person.* The
conjunction sed, too, is certainly nothing but the ablative
of the reflexive; and sed occurs twice in the S. C. de Bacch.
as an evident pronoun, and, in fact, governed by inter;
[G. Ed. p.215.] whence it may be assumed that inter can
be used in construction with the ablative, or also that, in the
old languages, the accusative is the same with the ablative:
the latter view is confirmed by the accusative use of ted and
med in Plautus.
+183. In Sanskrit the ablative expresses distance from a
place, the relation “ whence ;” and this is the true, original
destination of this case, to which the Latin remained
constant in the names of towns. From the relation
“whence,” however, the ablative is, in Sanskrit, trans-
ferred to the causal relation also; since that on account
of which any thing is done is regarded as the place whence
an action proceeds. In this manner the confines of the abla-
_tive and instrumental touch one another, and #a ¢éna (§. 158.)
and wea tasmdt, may both express “on account of which.”
In adverbial use the ablative spreads still further, and in
* some words denotes relations, which are otherwise foreign
to the ablative. In Greek, adverbs in ws may be looked upon
.as sister forms of the Sanskrit ablative; so that w-s, from
bases in o, would have the same relation to the Sanskrit
* The reduplication in me-mor, from me-smor, would be of the kind
used in Sanskrit, e.g. pasparsa, “he touched,” of which hereafter.
+ Cf. the Gothic ablatives in 6, adduced in §, 294, Rem. 1, p, 384,
wara 4-t, from bases in a, that, e.g. d’dwor has to eetfat dadd-ti
Thus, ou@-; may be akin to the Sanskrit ata samé-t,
“from the similar,” both in termination and in base. In
Greek, the transition of the T sounds into ¢ was requisite, if
indeed they were not to be entirely suppressed*; and in
§. 152. we have seen neuter bases in 7, in the uninflected
cases, preserve their final letter from being entirely lost by
changing it into s. We deduce, therefore, [G. Ed. p. 216.]
adverbs like 6u0-s, ovrw-s, &-s, from 6uO-7, ovTw-7, &-T OF
ou@-0, &c., and this is the only way of bringing these forma-
tions into comparison with the cognate languages; and it is
not to be believed that the Greek has created for this ad-
verbia] relation an entirely peculiar form, any more than
other case-terminations can be shewn to be peculiar to the
Greek alone. The relation in adverbs in w-s is the same as
that of Latin ablative forms like hoc modo, quo modo, raro,
perpetuo. In bases ending with a consonant, og for or might
be expected as the termination, in accordance with Zend
ablatives like ROASJASERVAKS chashman-at, “ oculo™;. but then
the ablative adverbial termination would be identical with
that of the genitive: this, and the preponderating analogy
of adverbs from o bases, may have introduced forms like
cwdpdv-we, which, with respect to their termination, may be
compared with Zend feminine ablatives like ro.wys76e1us
baréthry-dt. We must also, with reference to the irre-
gular length of this adverbial termination, advert to the
Attic genitives in ws for os.
* As, in otro, together with ovre-s, S38, dpve, and adverbs from
prepositions—¢fo, ava, xdtw, &c. It is here desirable to remark, that in
Sanskrit, also, the ablative termination occurs in adverbs from prepositions,
as SMeTa adhastat, “(from) beneath,” OCT purastat, “(from) before,”
&e. (Gram. Crit. § 652 p.279.).
+ In compounds, remains of ablative forms may exist with the original
T sound retained. We will therefore observe, that in “Adpodirn the first
member
THE GENITIVE.
[G. Ed. p.217.] 184. In no case do the different members

of the Sanskrit family of languages agree so fully as in the
genitive singular; only that in Latin the two first declen-
sions, together with the fifth, as well as the two first persons
of the pronouns, have lost their old termination, and have re-
placed it by that of the old locative. The Sanskrit termi-
nations of the genitive are 4 s, WMsya, Wa as, and WTA ds:
the three first are common to the three genders: as is
member has a genuine ablative meaning; and as the division ad¢po-diry

admits of no satisfactory explanation, one may rest satisfied with dgpod-irn.
In Sanserit, satfeat abhrdditd would mean “the female who proceeded
from acloud,” for abnrd-t must become abhrdd before itd ($. 93%.) ; and in
neuter verbs the otherwise passive participial suffix ¢a has usually a past
active ineaning. Of this usage ry, in dppod-irn, might be a remnant, and
this compound might mean, therefore, “‘She who arose, who sprang, from
foam.” ‘The only difficulty here is the short vowel of 08 for wd. As re-
gards the Sanskrit, here also the s of the ablative may in most declensions
rest on an exchange with an older ¢ (cf. p. 184 G. ed. Note); and, as the
Zend gives us every reason to expect Sanskrit ablatives like jiiwdy-dt,
prité-t, siiné-t, bhavishyanty-at, dtman-at ; so it will be most natural to
refer the existing forms jihwdy-ds, prité-s, &c., where they have an abla-
tive meaning, to the exchange of ¢ with s, which is more or Jess in vogue
according to the variety of dialects ;particularly as it is known, also, that,
vice versa, according to certain laws, @ s passes into at (Gramm. Crit.
§..100.). Consequently the identity between the genitive and ablative, in
most declensions, would be only external, and the two cases would vary
in their history ; so that, e.g. jihwdy-ds would be, in one sense, viz. in that
of lingue, independent and original; and in another, that of lingud, a
corruption of jihwdy-dt. At the time when Sanskrit and Zend were sepa-
rated from one another, the retention of the original ¢ must have been
the prevailing inclination, and, together with it, may also its change into.
s have arisen, as the Zend also uses, at times, the genitive fori with an
ablative meaning (e.g. Vend. 8. p. 177.). ;
GENITIVE SINGULAR. 203
principally confined to the consonantal bases,* and hence has

the same relation to s that, in the accusative, am has to m,
and, in the Zend ablative, at has to ft.
185. Before the genitive sign asthe ([G. Ed. p. 218.]
vowels = i and = u take Guna; and the Zend, and in a
more limited degree, also the Lithuanian and Gothic, share
this augment. All u bases, for example, in Lithuanian and
Gothic, prefix an @ to their final vowel: hence the Lithuanian
sunai-s and Gothic sunau-s correspond to the Sanskrit a
siinds (filit) from sunaus (§. 2.). In the i bases in Gothic, Guna
is restricted to the feminines ; thus anstai-s, “ gratie,” answers
to wie prité-s. Respecting Lithuanian genitives of i bases
see §.193. The High German has, from the earliest period,
dropped the genitive sign in all feminines: in consonantal
bases (§§. 125. 127.) the sign of the genitive is wanting in
the other genders also.
186. The form which the Sanskrit genitive termination
after consonants assumes, as it were of necessity (§. 94.),
viz. as for s, has in Greek, in the form os, passed over also to
the vowels « and v and diphthongs terminating in v; and
genitives like moprei-s, iy@ev-s, which would be in accordance
with §. 185. are unheard of; but dpri-os, ix6d-o¢ answer,
like 10d-ds, to Sanskrit genitives of consonantal bases, as Wa
pad-as, “ pedis,” araa_ vdch-as, “vocis.” The Latin, on the
other hand, answers more to the other sister languages,
but is without Guna: so hosti-s is like the Gothic genitive
gasti-s. In the u bases (fourth declension) the lengthening
of the u may replace the Guna, or, more correctly, this
class of words followed the Greek or consonantal principle,
and the vowel dropped before s was compensated for by
_* Besides this, it occurs only in monosyllabic bases in $4, Hi, & di, and
BW Gu ; e. g- ray-as, “rei,” ndv-as, “navis :” and in neuters in ziand ya,
which, by the assumption of an euphonic * n, assimilate to the consonantal
declension in most cases.
lengthening the u. The S.C. de Bacch. gives the genitive

senatu-os in Grecian garb. Otherwise the termination is
of consonantal bases is better derived from the Sanskrit ae
[G. Ed. p. 219.] as than from the Greek os, because the old
Sanskrit a in other places in Latin has been weakened to, i,
as frequently happens in Gothic (§§. 66. 67.).
187. With regard to the senatu-os just mentioned, it is
important to remark, that, in Zend also, the w bases, in-
stead of annexing a simple s in the genitive, as w»¢yspsasG
mainyéu-s, “of the spirit,” from mainyu, may, after the
manner of consonantal bases, add L 6 (from as, ef. p. 212,
G. Ed.), as »w3u9 danhv-d, or byaswus danhav-6, for
danheu-s “loci,” from >¥34 danhu. This kind of genitive
occurs very frequently as a substitute for the locative, as
also for the ablative (Vend. S. p. 177), more rarely with a
genuine genitive meaning.*
188. Bases in 8 a, and pronouns of the third person, of
-which only amu ends with a vowel other than a, have, in
Sanskrit, the more full genitive sign ™@ sya; hence, e.g.
Za vrika-sya, “lupi,” we ta-sya, “hujus,” &c., 2AM amu-shya,
* It might be assumed that as Baovdéos clearly stands for Basi\éFos,

Bods for BoF és, vads for vaF ds, (§. 124.), so also doreos would stand for
doreFos, and that doreos, therefore, should be compared with the Zend
genitives with Guna, as boas 35 danhav-d. The e¢, therefore, in’
aoreos would not be a corrupted v of the base, but the Guna vowel foreign
to the base ;but the v of the base, which, according to the original law of
sound, must become F before vowels, is, like all other digammas in the
actual condition of the language, suppressed. The « is certainly a very
heterogeneous vowel to the v, and the corruption of the latter to ¢, in the
middle of a word, would be a greater violation of the old relations of sound
than the rejection of a v sound between two vowels. The corruption
of ¢ to é is less surprising, and occurs also in Old High German (§. 72.).
In Greek, also, a consonant y is wanting, but cannot have been originally
deficient; and therefore the question might be mooted whether also
modews, owvdreos may not stand for pole-yds, sinape-yos.
« illius,” (§. 21.) In Zend this termination [G. Ed. p. 220.]

appears in the form of hé, (§. 42): hence, e. g. powass Jere
véhrkahé, “lupi,” wowryy/s9¢9 titiryé-hé, “ quarti,” for tuirya-hé.
189. In Greek and Latin we have already, in another
place, pointed out a remnant of the genitive termination
= sya, and, in fact, precisely in places where it might be
most expected. As bases in ¥ a correspond to the Greek
bases in o, and as cin Greek at the furthest extremity of words
between two vowels is generally dislodged, I do not enter-
tain the smallest doubt that the old epic genitive termination
in so is an abbreviation of cio; and that e. g. in roto— Tey
ta-sya, the first o belongs to the base, and only :o to the
case-sign. As regards, however, the loss of the o in Toio, the
Greek Grammar supplies us with another oto, where a & is
lost, the necessary and original existence of which no one
can doubt: éd/doc0, and the ancient position of the = in the
- second person, testify for d:do:co instead of d:doto, as for éAe-
yeco instead of éAéyou, just as the Indian weg ta-sya for
To-cio instead of toto. In the common language the ,, also,
has been dropped after the o, and the o of the termination,
which has remained, has been contracted with that of the base
to ov; hence rou from to-o. The Homeric form ao (Bopéao,
Aiveiao) belongs likewise to this place, and stands for a-zo,
and this for a-s:o (§.116.). The Latin has transposed our
= sya to jus, with the change, which is so frequent, of the
old a before the final s to u (cf. gare vrika-s, “lupu-s,” Wary
yunjmas, jungimus); hence, hu jus, cu-jus, e-jus, illius for illi-
-jus, &e. I cannot, however, believe that the i of the second
- declension is an abbreviation of oro, of which the : alone has
been retained ;* for it is clear that/upiand [G. Ed. p. 221.]
lupe from lupai rest on the same principle; and if Jupi pro-
ceeds from AvKo1o, whence can /upai be derived, as the cor-
responding Greek feminines nowhere exhibit an aio or yo?
* Hartung’s Cases, p. 211.

190. In Lithuanian the genitives of the a bases differ re-

markably from those of the other declezsions, and denote
the case by 0, in which vowel, at the same time, the final
vowel of the base is contained; thus, wilko, “lupi,” for
wilka-s. It is probable that this o (6) has arisen from a-s,
according to a contraction similar to that in the Zend (§. 56>.}.
In old Sclavonic, also, 0 occurs, answering to the Sanskrit
as; and nebo, gen. nebese, corresponds to the Sanskrit wwrae
nabhas. That, however, the Lithuanian has left the sylla-
ble as in the nominative unaltered, but in the genitive has
contracted it to o, may induce the remark, that like cor-
ruptions do not always find entrance in like places, if they
have not raised themselves to a pervading Jaw. In this
manner, in Gothic, the old a has remained in the interroga-
tive base HVA in the nominative (hvas), but in the genitive
hvi-s the weakening to i has taken place; so that here, as
in Lithuanian, only the more worthy powerful nominative
has preserved the older more powerful form, and an unor-
ganic difference has found its way into the two cases, which
ought to be similar.
191. The Gothic has no more than the Lithuanian pre-
served a remnant of the more full genitive termination sya,
and the Gothic a bases, in this case, resemble the ibases,
because a before final s has, according to §. 67., become
weakened to i; thus vulfi-s for vulfa-s; as also in Old
Saxon the corresponding declension exhibits a-s together
with e-s, although more rarely; thus, daga-s, “of the day,”
[G. Ed. p. 222.] answering to the Gothic dagi-s. The conso-
nantal, bases have, in Gothic, likewise a simple s for case-sign;
hence, ahmin-s, fiyand-s, bréthr-s (§. 132.). The older sister
dialects lead us to conjecture that originally an a, more
lately an i, preceded this s—ahmin-as, fiyand-as, bréthr-as,—
which, as in the nominative of the a bases (vulf’-s for vulfa-s),
has been suppressed. The Zend exhibits in the r roots an
agreement with the Gothic, and forms, e.g. s/s) nar-s, “ of
the man,” not nar-é, probably on account of the nature of

the r bordering on that of a vowel, and of its facile combi-
nation with s.*
192. Feminines in Sanskrit have a fuller genitive ter-
mination in bases ending with a vowel, viz. ds for simple
s (see §. 113.); and, in fact, so that the [G. Ed. p. 223.]
short-ending bases in $i and $u may use at will either
simple q s or "Ta ds; and instead of wits prité-s, Wate
tané-s, also Wtara_prity-ds, tata tanw-ds, occur. The long
_ vowels = d, 4 & @,} have always wTa_ ds; hence, fagrart
jihwiy-ds, afrerrara bhavishyanty-ds, qava vadhw-ds. This
termination "ra ds, is, in Zend, according to §. 56°,
sounded do; hence, gusyyas»¢s hizvay-do, EUSIC WIN IHV2S
bushyainty-do. In bases in siand>uI have not met
* Hence I deduce the genitives aayDaso.us)s brdtar-s, awa 4
dughdhar-s—which cannot be quoted—and the probability that the corre-
sponding Sanskrit forms are properly bhrdtur, duhitur, which cannot be
gleaned from the Sanskrit alone, on account of §. 11., and by reason of the
elsewhere occurring euphonic interchange of s and r. arate bhrdétur, and
similar forms, would therefore stand for -urs, and this apparently for ars,
through the influence of the liquids ;and, according to §. 94., they would
have lost the genitive sign. The same is the case with the numeral adverb
at chatur, ‘four times,’’ for wae chaturs ; for which the Zend, by
Bantpaline the r, gives wae. 5s ti Be (§. 44.). The Indian Gram-
marians also, in the genitives under discussion, assume the absence of the
genitive sign (Laghu-Kaumudi, p. 35). As, however, the Visarga, in
wale hrdshtu (from the theme PTET kréshtar or may kréshtri, see §. 1.),
may evidently stand as well for s asfor7; so in such doubtful cases it is
of no consequence to which side the Indian Grammarians incline, where
arguments are not found in the Sanskrit itself, or in the cognate languages,
which either confirm or refute their statements. And it is impossible, if
the Visarga, in Tq: bhrdtul, stands for vr, that the preceding uw can be
& transposition of the final letter of the base (=A 37m), for this cannot be
both retained in the form of r, and yet changed into u (cf. Colebrook,
p. 55, Rem.)
+ Only the few monosyllabic words. make an exception. (Gramm.
Crit. §. 130.)
£08 FORMATION OF CASES.
with this termination; together with ashe ddan Afritéi-s,

9g yas taneu-s, or by»paso tanv-d, bnasyasio tanav-é, I find
no easy yO sodas Gfrithy-do, gss»sase tanv-do. The cognate
European languages exhibit no stronger termination in the
feminine than in the masculine and neuter; the Gothic, how-
ever, shews a disposition to greater fulness in the feminine
genitive, inasmuch as the 6 bases preserve this vowel in con-
tradistinction to the nominative and accusative; but the
i bases, as has been shewn above, attach Guna to this vowel,
while the masculines do not strengthen it at all. Compare
gibé-s with the uninflected and base-abbreviated nominative —
and accusative giba, and anstai-s with gasti-s. Respecting
the pronominal and adjective genitives, as thi-zd-s, blindai-
z6-s, see §. 172. The Greek, also, in its feminine first declen-
sion preserves the original vowel length in words which have
weakened the nominative and accusative—ogvpas, Movons,
[G. Ed. p. 224.] opposed to cdipa, cpvpa-v, potodv.* In
Latin, also, G-s, with the original length of the base escas,
terras, &c. stands opposed to esc, esca-m. It cannot be sup-
posed that these genitives are borrowed from the Greek;
they are exactly what might be expected to belong to a
language that has s for the genitive character. That,
however, this form, which no doubt extended originally to
all a bases, gradually disappeared, leaving nothing but a
few remains, and that the language availed itself of other
helps, is in accordance with the usual fate of languages
which continually lose more and more of their old heredi-
ditary possessions.
193. The Lithuanian, in its genitive rank-ds for ranka-s,
* The Attic termination ws is, perhaps, a perfect transmission of the

Sanskrit "Tq ds; so that forms like 7éAe-os answer to wtare prity-ds,
Although the Greck as is not limited to the feminine, it is nevertheless
excluded from the neuter (doreos), and the preponderating number of «
bases are feminine.
resembles the Gothic; and in some other cases, also, re-

places the feminine @ by a long or short o. It is doubtful
how the genitives of i bases, like awiés, are to be regarded.
As they are, for the most part, feminine. and the few mas-
culines may have followed the analogy of the prevailing
gender, the division awi-és might be made; and this might
be derived, through the assimilative force of the i, from
_awi-ds. (cf. p..174, note*), which would answer to the San-
skrit genitives like hata prity-ds. If, however, it be com-
pared with atta prités, and the é of awiés be looked upon
_.as Guna of the i (§. 26.), then the reading awiés for awés is
objectionable. Ruhig, indeed, in his Glossary, frequently
_ leaves out the i, and gives ugnés, “of the fire,” for ugniés;
but in other cases, also, an i is suppressed before the e
_ generated by its influence (p. 174, note*); and, e.g., all
_ feminine bases in y@ have, iu the genitive, és for i-és or y-és,
. as giesmé-s, for giesmyés, from GIESMYA (see p. 169, note).
_ Therefore the division awié-s might alsc be made, and it
. might be assumed that the i bases have. in some cases, ex-
perienced an extension of the base, similar to those which
were explained in the note, p. 174 (cf. §. 120.). This
view appears to me the most correct,espe- [G. Ed. p. 225.]
cially as in the vocative, also, awié answers to giesme for
giesmye, Or giesmie.
/194. As regards the origin of the form through which,
in the genitive, the thing designated is personified, with
the secondary notion of the relation of space, the language
in this case returns back to the same pronoun, whence, in
§. 134., the nominative was derived./ And there is a pro-
noun for the fuller termination also, viz. = sya, which occurs
only in the Védas (cf. §. 55.), and the s of which is replaced
in the oblique cases likewise, as in the neuter, by ¢ (Gramm.
Crit. §. 268.}; so that = sya stands in the same relation to
wa fyo-m and wa tya-t that @ sa does to Wm ta-m, 7 ta-t.
It is evident, therefore, that in = sya, @ tya, the bases @sa,
@ fa, are contained, with the vowel suppressed and united
P
with the relative base @ ya. Here follows) a general view

of the genitive formation :*
SANSKRIT. ZEND. GREEK. LATINe LITHUAN, GOTH
m. vrika-sya, véhrku-hé, AUKO-10;. + ewe wilkd, ~vulfi

m. ka-sya; ka-hé, oe ee CURjuUs, ko, hvi-
f. jihwdy-ds, hizvay-do, xapi-s, terra-s, ranké-s, gibé
m. paté-s, patéi-s, see!) hosti-s, » ++ »T gast
paty-us, sexie be NOTt-OG, » aole
f. prité-s, fritéi-s, aver oer tyettine,
prity-ds, etanere PUTE-WS, ew wo
f. bhavishyanty-ds, bishyainty-do, « «++ 206 ee e . e -.3
m. séné-s, paseu-s. eobstie lash ss sunad-s, suni

% shane paso-6, ixGb-05,. we es
f. tané-s, taneu-s, oes « § 80CrHs, eee han
tanw-ds, tanv-6, niTv-0¢, 5 aii ot eee ee

f. vadhw-ds, & wie ofa oh 1k eens ef
]
‘dD
d‘pq
m.f. gé-s, geu-s; Bo(F)-0s, . bov-is, oe ee oe
f.
[‘9¢2 ndv-as, > webne v&(F)-655) sa es eoeoee ee
f. wdch-as, vach-6,t 67-06, | voc-is, ee ee ee
m. bharat-as, barént-6,§ pépovt-os, ferent-is, eee I fiya
m. dtman-as, asman-6,t daiuov-os, sermon-is, Gkmen-s, ahn
n. ndmn-as, ndman-é,t TadAaY-os, nomin-is, nan
* The meanings will be found in §. 148.

+ See §. 193,
} See p. 163. Note f.
§ And doss2us baraté also may occur, according to the analogy of
Lroasce7e_s berézaté, splendentis,” V. 8. p. 87, and passim. .The reten-
tion of the nasal in the genitive, however, as in all other cases, is the more
common form, and can be abundantly quoted. For Louse Zassbarénté,
also Leoyys.s%ss barantd, is possible, and likewise, in the other cases, the
older ss a for ¢é. In some participles, as in ssy19904009 Ssuyans (nom.),
which is of constant recurrence as the usual epithet of agriculture
(2539039556 vaistrya) ¢ é never occurs.
] Vide §, 254. p. 302, Note f.
LOCATIVE SINGULAR. 211
SANSKRIT. ZEND. GREEK. LATIN.‘ LITRUAN- GOTHIC.

bhrdtur, brdtar-s,* matp-os, fratr-is, .... bréthr-s.
duhitur, dughdhar-s,t @vyatp-ds, matr-is, _dugter-s, dauhtr-s.
détur, dadtar-s, Sorip-os, dator-is, «+++ +++
vachas-as, vachanh-6,t —éne(c)-os, Operri&y:- oeininylit ie *
: THE LOCATIVE,
- 195. This case has, in Sanskrit and Zend,§$ i for its cha-
racter, and in Greek and Latin||has received the function of
the dative, yet has not suffered its locative _[G. Ed, p. 227.]
signification to be lost; hence, Awddv:, Mapabdur, ZaAdapivi,
GPG, oikol, yauai; and, transferred to time, 77 av77 jpépg,
7)avy vukti. So in Sanskrit, feaa divasé, “in the day;” fafy
nisi, “in the night.”
196. With wa of the base preceding it, the locative ¥i
passes into zé (§. 2.), exactly as in Zend ;. but here, also,
sb Gi stands for » é@ (§. 33.); so that in this the Zend
approaches very closely to the Greek datives like ofkor;
poi, and coi, in which / has not yet become subscribed, or
been replaced by the extinction of the base vowel. To the
forms mentioned answers sbs3@5¢ maidhyéi, “in the mid-
dle.’ _One must be careful not to regard this and similar
phenomena as shewing a more intimate connexion between
Greek and Zend.
197. In Lithuanian, which language possesses a proper
locative, bases in a correspond in this case in a remark-
able manner with the Sanskrit and Zend; since they con-
# It would be better toread brdthr-6, after the analogy of dathr-é,

“ creatoris.”” (Burnouf,
“ Yacna,”’ p.363, Note).
+ The gen. ofdighdar
is probably dughdér-6 (see p. 194, Note t).
} See p. 163, Note tf.
§ Few cases admit ofbeing more abundantly
quoted in Zend than the
locative, with which, nevertheless, Rask appears to have been unacquainted
at the time of publishing
his treatise, as he does
not give it in anyor
his three paradigms.
BT now refer the Latin dative to the Sanskrit dative, rather than to
the locative ; see p. 1227 G. Ed., Note t. .
ez
tract this a with the old locative i, which appears pure -

nowhere any more, to 2; hence, diewé, “in God,” from
DIEIVA, answers to 2% dévé, po»7025g daévé. The bases
which ‘terminate with other vowels employ, however, in
Lithuanian, without exception, ye as the locative termina-
tion, without any accent upon the e, a circumstance which
must not be overlooked. This e is, perhaps, only an unor-
ganic echo, which has occasioned the change of the old loeative
i into y as, in Zend, the plural locative termination su, by
adding an a, appears, for the most part, in the form of as»
[G. Ed. p. 228.] shva, or »s»whva. To the Lithuanian ye
answers also, in old Sclavonic, a locative termination ye, for
which several declensions have the original pure i; so
that nebes-i, “in Heaven,” and imen-i, “in the name,” agree
most strictly with the Sanskrit qafa nabhas-i and arta
ndman-i, from 7a nabhas, away ndman.
198. Masculine bases in i and wu, and, optionally, feminine
bases also, have a different locative termination in San-
skrit, viz. W@W du, before which ¥*i and gu are dropped;
but in uf# pati, “lord,” and af@ sakhi, “friend,” the i has
remained in its euphonic change to q y: hence, ual paty-du,
wen sakhy-du. If we consider the vocalization of the s to u,
shewn in §. 56°., and that, in all probability, in the dual,
also, St du has proceeded from ura ds (§. 206.) ;.moreover,
the circumstance that in the Védas the genitive occurs
with a locative meaning (eferarata dakshindyds, “in dexterd,”
for efeyarata dakshindydm, Panini VII. 1. 39.); and, finally, a
the fact that, in Zend, masculines in i and u likewise em-

ploy genitive terminations with a locative signification ; we
shall be much disposed to recognise in this wt du, from
ata ds, a sort of Attic or produced genitive termination.
199. In u bases, instead of the locative the Zend usually
employs the genitive termination 4 6 (from wa as), while,
in ‘a genitive meaning, the form ~»¢ eu-s is more com-
mon; thus we read, in the Vend. S. p. 337. sgasorns a
—
SOYYIAM IIA ROME bow gy aétahmi anhvd yat astvainti, “in
—_
hoe mundo quidem eristente.” This Zend termination 6 (from

a+u) has the same relation to the Sanskrit du that a
short a has to a long a, and the two locative terminations
are distinguished only by the quantity of the first member
of the diphthong. On the other hand, we find in the
feminine base >a tanu, “ body,” very often the genuine
locative form 3» sso tanv-i ; and we do not doubt that, in
Sanskrit also, originally the u bases of the [G. Ed. p. 229.]
three genders admitted in the locative the termination i
(afra sunw-i, af tanw-i, afea madhw-i, or Rufa madhu-n-i).
Bases in 5 i employ, in the locative, the usual genitive
termination di-s; thus, in the Vend.S. p. 234, WY-wGasy sGas
3507392593299 6406 rosy ahmi namdné yat mdzdaya‘néis, “ in
hae terra quidem mazdayasnica, which Anquetil renders by
“ dans le pays des mazdeiesnans.” In pronouns, also, though
they have a locative, the genitive sometimes occurs with
a locative meaning; e.g. Vend. S. p. 46, pours wuss
ainhé visé, “in this way,” or “ place,” (cf. the feminine form
guser
as ainhdo, §. 172. Note.).
200. From the Zend and Sanskrit we have already been
compelled to acknowledge a connexion between the genitive
and locative; and as we have seen the locative replaced
by the genitive, so must we, in Latin, recognise a replacing
of the genitive by the locative. Through the formal
agreement of the corresponding Latin and Sanskrit termi-
nation, and from the circumstance that the genitive occurs
with a locative meaning only in the two first declensions
(Rome, Corinthi, humi), not in the third or in the plural (ruri
not ruris), M. Prof. Rosen was first induced to characterize
the Latin genitive of the two first declensions as borrowed
from the old locative; a view, the correctness of which I
do not doubt, and which I have already corroborated else-
where by the genitives of the two first persons, in which mei
tui, agree most surprisingly with afa may: (from mé-i, §. 2.),
“in me,” wfa fwayi (from fvé-i). Or ought, perhaps, a double
inflexion i to be assumed as the sign of both a genitive and
a locative dative? Should Rome (from Romai), Corinthi,

be on one occasion genitives and on another locatives, and
(G. Ed. p. 230.] in their different meaning be also of
different origin? And where, then, would the origin of'the
genitive Rome be found, as that of the locative has been
found already ? Should mei, tui, be compared, not with afa
mayi, ata twayi, poi, Toi, but with HA mama, Wa tava, Mov, Tov,
Goth. meina, theina? As the cases, like their substitutes the
prepositions, pass easily from one relation of space to
another, and, to use the expression, the highest .become the
lowest, nothing appears to me more probable, than that,
after the first. declension had lost its G-s, then the dative,
according to its origin a locative, necessarily became substi-
tuted for the genitive also.*. In the second declension the
form o-i, which belongs to the dative locative, corresponding
to the Greek w, om—and of which examples still remain
handed down to us (as populoi Romanoi)—has become doubly
altered: either the vowel of the base alone, or only that
[G. Ed. p.231.] of the termination, has been left, and the
first form has fixed itself in the dative, and the latter in the
* The assumption that a rejected s lies at the base of the genitives in 2,

ae (a-i) appears to me inadmissible, because in all other parts of Grammar
—numerous as the forms with a final s otherwise are—this letter has in
Roman defied all the assaults of time, and appears everywhere where the
cognate languages lead us to expect it: no terre for terras (acc. pl.), no
lupi for lupos, no ame for amas, &c. The question is not here that of an
occasional suppression of the s in old poets, before a consonant in the word
following. The genitives in e-s and @-s occurring in inscriptions (pro-
vincie-s, su@-s, see Struve, p. 7.) appear to be different modes of writing
one and the same form, which corresponds to the Greek 1-s for a-s ; and
I would not therefore derive the common genitive sue@—older form suai—
from su@s with the s dropped, The genitives in us, given by Hartung
(p. 161.) from inscriptions in Orelli (nomin-us, ewercitu-us, Castor-us, &c.),
I am not surprised at, for this reason, that generally ws is, in Latin, a
favourite termination for AA as ; hence nomin-us has the same relation
w aTaa némn-as, that nomin-i-bus has to @raera ndma’-bhyas, and
lupus, to yaa vrika-s.
genitive, which is therefore similar to the nom. plural, where,

in like manner, Romani stands for Romanoi. But the dative
is not universally represented in Latin by a locative ter-
mination; for in the pronouns of the two first persons mihi
answers to Haq ma-hyam, from ma-bhyam, and tibi to way
tu-bhyam ; as, however, the league between the dative and
locative had been once concluded, this truly dative termi-
nation occurs with a locative meaning (ii, ubi), while vice
versd, in Sanskrit, the locative very frequently supplies the
place of the dative, which latter, however, is most usually
expressed by the genitive, so that the proper dative is, for
the most part, applied to denote the causal relation.
201. Pronouns of the 3d person have, in Sanskrit, $4 in
instead of i in the locative, and the W a of the appended
pronoun @ sma is elided (see §. 165.); hence, afaq
tasmin, “in him”; wfera kasm’in, “in whom?” This n,
which seems to me to be of later origin, as it were an n
é ixév, does not extend to the two first persons, and
is wanting in Zend also in those of the third; hence,,
35 ahmi, “in this.” As to the origin of the.i signifying
the place or time of continuance,it is easily discovered as
soon as 7 is found as the root of a demonstrative; which,
however, like the true form of all other pronominal roots,
has escaped the Indian Grammarians.
202. Feminine bases ending with long simple vowels
have, in Sanskrit, a peculiar locative termination ; viz. 4Ty
dm, in which, also, the feminines im short i and u may at
will participate (cf. §. 192.); while the monosyllabic femi-
nine bases in long$ # and & %, for =m dm, admit also the
common $7; hence, fram bhiy-dm or fafa bhiy-i, “in
fear,” from wt bhé.* In Zend this termi- [G. Ed. p. 232.]
* Perhaps the termination dm is a corruption of the feminine genitive

termination ds (cf. §. 198. efaqurara, dakshindyds for dakshindyim),
where it should be observed that in Prakrit, as in Greek, a final s has
frequently become a nasal.
nation dm has become abbreviated to a (cf. §..214.); hence,

sysesy yahmy-a, “in which,” from s6s9 yahmé
(cf. §. 172.). This termination appears, however, in Zend,
to be less diffused than in Sanskrit, and not to be applicable
to feminines in si and >u. The form tanwi is clearly
more genuine than the Sanskrit ¢andu, although from the
earliest period, also, tanwdm may have existed.
203. We here give a general view of the locative, and
of the cases akin to it in Greek and Latin (see §. 148.) :
SANSKRIT. ZEND. GREEK. LATIN. LITHUAN-
m. vriké,* vehrké,* Atco, lup’z, wilke.

f. jgihway-dm, hizvay-a, xapq, — terra-i, ranko-ye.
m. paty-du,f ia oe noot-i, host’, pati-ye.
f. prit’-du,t ie mopti-t, sit'-2, awi-ye.
f. bhavishyanty-dm, bashyaintya,.... ...5 «eee
m. sin’-du, oft ixOv-1, pecu-%, sunu-ye.
f. tan’-du,§ tanw-i, mitv-t, socru-t, se us
n. madhu-n-i, Pb te péOv-t, ERS gia sn
f. vadhw-dm, Re Spe . «5 6s 5, et eee
m.f.gav-i, gav-i, Bo(F)-i, bovz, 1.2...
af. ndv-i, ee ae vOF)-° Sh SS Sere
bm. bharat-i, barént-i, épovr-t, ferent7, ....
~m. dtman-i, asmain-i, Saipov-t, sermon-i,... «
Sn. ndmn-i, nimain-i, tdAav-t, nomin-i, ....
' m. bhrdtar-i, brdthr-i?|| marp-i, fratrz, ....
f. duhitar-i, dughdhér-i? @vyatp-i, matr#z, ....
m. ddtar-i, dathr-i?\| Sornp-t, datdrz, ....
n. vachas-i, _ vacanh-i, éme(o)-t, oper-%, 1...
* See §.196. + See§.198. { Orprity-am. § Or tanw-dm.

|| The rejection of the a preceding the r in the theme seems to me more
probable than its retention, The ¢ of the termination is guaranteed by the
other consonantal declension, which in this case we can abundantly enough
exemplify. (Regarding dughdhér-i, see p. 194, Note +). That in Sanskrit
bhratar-i, duhitar-i, ddtar-i, are used instead of bhrdtri, &c. is contrary
to
VOCATIVE SINGULAR. 217
VOCATIVE.
204. The vocative in the Sanskrit family of languages

has either no case-sign at all, or is identical with the
nominative: the former is the principle, the latter the
practical corruption, and is limited in Sanskrit to mono-
syllabic bases terminating in a vowel: hence, wta_bhi-s
“fear!” as xi-s. A final a of the nominal (G. Ed. p. 234.]
bases remains, in Sanskrit and Zend, unchanged ; in Lithua-
nian it is weakened to e; and the Greek and Latin also, in
the uninflected vocative of the corresponding declension,
prefer a short e to o or u, which, under the protection of the
terminations, appears as the final letter of the base. We
must avoid seeing in Avxe, lupé, case terminations: these
forms have the same relation to ye vrika that TEVTE,
quingue, have to T= pancha; and the old a, which ap-
pears in Avxos as 0, in /upus as i, has assumed the form of
é without any letter following it. In Zend, the consonantal
bases, when they have s in the nominative, retain it in the
vocative also; thus, in the present participle we have fre-
quently found the form of the nominative in the sense of the
vocative.
205. Bases in i and wu have, in Sanskrit, Guna; neuters,
however, have also the pure vowel: on the other hand,
to the theory of the weakest cases (§. 130.), to which in other respects
the
locative belongs. As, however, bases in Wl ar (=J ri), with respect to
the rejection and lengthening of the a, have a very great agreement with
bases in an, it must here be further remarked, that these too, in the
locative, do not strictly follow the suppression of the a in the weakest
cases, which is conditionally prescribed in §, 140., but optionally retain
the a, or reject
it; so that with némn-i also ndéman-i
is used. With
brdtar-i, however, exists no bhrdtr-i, and the form pitr-i, given at §. 132.
is an oversight: the Greek warp-i may therefore, with respect to the
shortening of the base, be better compared with the dative pitr-é.
polysyllabic feminines in # and @ shorten this final vowel;

while a final wt 4, by the commixture of an i, becomes é
(§.2.). The language, however, both by producing and
shortening the final vowel, clearly aims at one and the
same end, only by opposite ways; and this end, in fact, is
a certain emphasis in the address. To the Guna form
wai 6, from a+u, correspond remarkably the Gothic and
Lithuanian; as sunau, sunai, resembling the Sanskrit
wat siiné,* Gothic feminine bases in i do not oceur in
~ [G. Ed. p. 285.]. Ulfilas in the vocative: as, however, they,
in other respects, run parallel to the u bases, the vocative
anstai, from ANST'T, might be expected as an analogous form
to handau. The Lithuanian i bases in the vocative extend
their theme in the same manner as in the genitive (§. 193.);
so that, properly, there is no vocative of this class of words,
and awie answers to zwdke, giesme (Ruhig’s third declension),
for zwakie, giesmye.{ Masculine bases, in Gothic, in i, like
the masculine and neuter a bases, have lost their final vowel
in the vocative, just as in the accusative and nominative ;
hence vulf*, daur’, gast’. In bases in n the Gothic shares
with the Latin the suppression of’ the final consonant,
which has passed over from the nominative to the voca-
tive; while only the Sanskrit and Zend again introduce
# The Zend can at will attach Guna to a final > u, or not; and we find
both Uys JI9G mainy6é and >33 JIG mainyu as the yocative of >y3 JIG
mainyu, “spirit.” On-the other hand, we have founda final » ¢only, with-
out Guna; and indeed frequently s~osasd paiti, “lord” So Vend 8.
p- 456, so.saso Uy
ywwGasy ASCEHY SEY.5.39> usihista namdnd-paiti,“Arise, lord
of the place!”” The si between the preposition and the verb serves as
a conjnuctive vowel, to assist the juncture of the words (cf. §. 150. Note).
+ It follows from this, and from §.193., that (§. 177.) I have incor-
rectly assumed e# as the termination in the dative. For dwi-ei, the division
should be made thus, dwie-i ; and this is analogous with zwdke-i, giesme-t,
for zwdkie-i, giesmye-i.
VOCATIVE SINGULAR, | 219
into the vocative the nasal which had been dropped in the
nominative. Adjectives in German, with respect to the
vocative, have departed from the old path, and’ retain
the case-sign of the nominative; hence Gothic blind’s,
“blind!” - In Old Northern, substantives also follow this
irregular use of the nominative sign. The Greek has
preserved a-tolerable number of its vocatives pure from
the nominative sign, and in some classes of words uses’
the bare base, or that abbreviation of it which the laws of
euphony or effeminacy rendered requisite ;hence, téAav op-
posed to raAas, yapiev for yapievt’ opposed to ydpiers, mat
for zai:d opposed to ais. * In guttural and labial bases the
language has not got free of the nominative sign in the voca-
tive, because xs and’ as (&, y+) are very favourite combina-
tions; to which the alphabet also has paid homage by parti-
cular letters to represent them. Still the [G. Ed. p. 236.]
vocative ava, together with ava, is remarkable, and has that
sound which might be expected from a theme @vaxt’, to
which, in its uninflected state, neither xr, nor, conveniently,
even the x, could be left.. “For the rest it is easy to imagine
(says Buttmann, p. 180), that particularly such things as are
not usually addressed, prefer, when they happen to be ad-
dressed, to retain the form of the nominative, as & mods!” *
The Latin has followed still farther the road of corruption in
the vocative which was prepared by the Greek, and employs
in its place the nominative universally, except in the mascu-
line second declension. The substautive bases mentioned in
§. 148. form, in the vocative,
* To this circumstance may also the re-introduction of the case-sign in

the neuter be owing, while the Sanskrit employs the bare base. More-
over, this fact also may have co-operated towards the Greek more easily
freeing itself in the vocative from the bare primary form, because it ap-
pears at the beginning of compounds much more rarely than in Sanskrit,
(See §. 112.)
SANSERIT. ZEND. GREEK. LATIN LITHUAN. GOTHIC.

m. vrika, vehrka, Avdxe, lupe, wilke, vulf’.
n. . déna, data, Spo-v, donu-m, .... daur’.
f. - jihwé, hizvé? xaopa, terra, ranka, giba? »
m. .paté, paiti, most, hosti-s, ...+ gast.
f. _prttd dfriti, ROPTly., Siti, iy, mew ornl wewiei
n. véri, vairi, 1806 pMAFE os afeste 9h« HGTesaem
f. bhavishyanti, bdshyainti, ...- je dine cinie oeieaetaio et
m. sund, pasu, ixOd, pecu-s,. sunadl, sunau.
f, _tané, tanu, mitv, socru-s, «+++ handau.
n. madhu, madhu, MEOu, pect, vane deve
f. .vadhu,
@ m. f.gdu-s, gau-s, Bov, __bo-s, 0 00 6 yorejeere
Ef. ndu-s, ee vau, o.'é feceiy: 430:se aE
sf. vdk, vices ? On-s; VvOC=8, s:$ duet A board
8 m. bharan, baran-s, gépwv, feren-s, sukan-s, fiyand.
“m, dtman, asman, datyov, sermo, dkmit, ahma’.
n. ndman, ndman, Ta&Aav, nomen, ..«.. namé.
m. bhrdtar, brdtaré,* md&tep, frater, ...++ bréthar.
f. duhitar, dughdharé,* Ov-yarep,mater, moté, dauhtar.
m. détar, datare,* Sorip,. dator, «4st ewe
n. vachas, vaché, énos,t opus, «>» Hee
DUAL.
NOMINATIVE, ACCUSATIVE, VOCATIVE.
206. These three cases have, in Sanskrit, in the mascu-

line and feminine, the termination Wt du, which probably
arose from "ra ds by vocalization of the s (cf. §§. 56°. and
198.), and is therefore only a stronger form of the plural
termination as. The dual, both in the cases mentioned and
in the others, prefers the broadest terminations, because
it is based on a more precise intention than the indefinite
* See §. 44. + See§.128.

NOMINATIVE, ACCUSATIVE, VOCATIVE DUAL. 221
‘plural, and needs, therefore, stronger emphasis, and more

lively personification. Compare, also, in the neuter, the
long ¢ of the dual with the short i of the plural; as
waut asruné with wafa asrini.
207. While the Prakrit and Pali have lost [G. Ed. p. 238.]
the dual, the Zend has retained it; still, however, so that
‘instead of it the plural often occurs, and in the Vend. S.,
p- 203, rosqss2asddy> edo ss Gd schénubyaschit, “and as far as
the knees,” is used with a plural termination. In the verb
‘the dual is still more rare; but here, however, it is not en-
tirely lost, and many examples of it can be quoted in the
V.S.* The Sanskrit termination Wt du occurs in the cor-
‘responding places in Zend in the form of gus do, which,
according to §. 56°., stands at the same time for the Sanskrit
‘termination wre ds, and gives an emphatic proof that the
Sanskrit dual termination ¥ du is nothing else than a cor-
ruption of ara ds, and, in fact, an occasional one which
‘appears in grammar only once or twice (see §. 198.), while
‘the example herein given by the Sanskrit has been raised
‘to a general principle by the Zend. This principle be-
comes almost irrefragable matter of fact from the conside-
ration that the Zend has even actually retained, in the
dual, the sibilant before the particle 4 cha, and uses
dos-cha, not do-cha, as might have been expected if the
dual termination # du, in Sanskrit, were the original form,
and not a corruption of "rads. Thus we read in the
Vend. S. p- 225, aspsssgussoaspg/eGas asssrgus» ru poass> sho
téi ubaé hurvdos-cha amérétat-dos-cha, “the two Haurvats and
Amertats.”+ What Anquetil, in his Voca- [G. Ed. p. 239.]
_ ™ Cf Gramm. Crit. Add. to r. 137.

+ Cf. Anquetil IT. 175. The two Genii, which Anquetil writes Khor-
dad and Amerdad, appear very frequently in the dual, also with the ter-
mination bya (§. 212.) ;and where they occur with plural terminations,
this may be ascribed to the disuse of the dual, and the possibility of
replacing
222 TG FORMATION OF CASES.
bulary (p. 456), writes. naereketdo, and renders by “deur

femmes,” can be nothing else than gusy3019.5/sawy ndirikay-do,
from the base 1099.57.52 ndirikd. The form GUNN S059 575.9
niirikaydo is, however, evidently more genuine than
95/384 ndiriké; as, according to the Sanskrit principle
(§. 213.), from a feminine base must have been formed
nairikd. From »¢aus bdzu, Rask cites the form EUSP AUS
bdzvdo, “arms,” .without remarking that it is a dual: it
clearly belongs, however, to this number, which was to be
expected referring to the arms; and ><axss bdzu forms, in
the nominative plural, dorgaus bazvéd or Urrasgauss bazavé.
Still, in the edited parts of the Zend-Avesta, examples are
wanting of bdzvdo, regarding the genuineness of which, how-
ever, I have no doubt.
208. In the Véda dialect, the termination = du occurs
frequently abbreviated to d, so that the last element of the
diphthong is suppressed. Several examples..of this abbre-
viated form occur in Rosen’s “Specimen”; as, waar
asvin-d, “ the two Aswins,” from asvin, and att nard, “ two
- [G. Ed. p.240.}] men,” which can be derived both from nar
replacing the dual in all cases by the plural. Thus we read, l, ¢, p. 211,
haurvatdt-6 and amérét-as-cha as accusative, and with the fullest and
perhaps sole correct reading of the theme. We will, however, not dwell
on this point any longer here, but only remark, that haurvatdt is very
frequently abbreviated to haurvat, and the 4 of amérédt is often found
shortened; whence, p. 104, asddanoar asehaurvatbya, asddsqoaspoehegas
amérétathya, (see §.38.)$ 28995 ANOAPERGS amérétata bya is a palpable
error. Undoubtedly, in the passage before us, for hurvdoscha, must be
read either haurvatéoscha, or haurvatdtdoscha, or haurvatatéoscha. Com-
pare 1. c. p. 91, Asawa 7base haérvatatéus-cha with the termi-
nation 39>. dus for wens dos (cf. §. 33.), but incorrectly 4 é for bs.
The two twin genii are feminine, and mean apparently, “* Entireness” and
“Immortality.”” The forms preceding them, therefore, ¢éi and ubaé, are
likewise feminine; the former for # ¢¢(§. 33.), the latter for gay ubhé
(cf. §. 28). We must also regard the dual form mentioned at §, 45, of
the so-called Amschaspants not as neuter, but as feminine,
(q nri) and from nara, but which more probably comes

from nar. In Zend the abbreviated termination from du is
likewise employed, and, in fact, more copiously than the fuller
termination ; and we rejoice to see, in the Heaven of Ormuzd
also, the twin pair called Indian, and celebrated for their
youthful beauty. _We. read,-namely, “in Vend.. S. ‘p. 313,
WOSSESN basa, asa ysdsas. aspind-cha yavané yaz
(maidhé), “ Asvinosque juvenes ‘veneramur,” which Anquetil
renders by “je fais Jzeschné & ['excellens toujours (subsistant”’).
The Sanskrit. fae asvind however, can, in Zend, give
nothing but aspind or aspina (§. 50.): the former we owe
here to the protecting particle » cha (see p. 175, Notet
G. Ed.)... The. plural. yavan-é (from yavanas), referring to
the dual aspind, is worthy of remark, however (if the read-
ing be correct),as it furnishes a new proof that, in the
received condition of the Zend, the dual was near being
lost: the verb being, for the most part, found in the plural
when referring to nouns in the dual form.
_. 209. From the Véda termination 4, and the short’ a,*
which. frequently stands for it in Zend, the transition is
easy to the Greek ¢, as this vowel, at the end of words, is a
fayourite: representative. of the old @; and, as above, in the
vocative (§. 204.), Avce stood for Ja vrika, asgPwich vehrha,
so here, also, ayvdpa, (with euphonic 5) corresponds to the
above-mentioned Véda 4a nard, and Zend sy nar-a. Al-
though, according to §. 4., w also very frequently stands for
wm 4, still we must avoid regarding Av«cw as the analogous
form to. yar vrikd, or ausJareh vehrkd (see §. 211.). That
however, the Lithuanian dual « of masculine [G. Ed. p. 241.]
bases in a (in the nominative) is connected with the Véda and
Zend dual termination spoken of, i. e, has proceeded from @, I
_ * Thus, Vendidad Sade, p. 23, asoaspoe7egas ass ase haurvata

améretdta, “the two Haurvats and Amertats”; p. 136, and frequently,
whys) »»4 dva nara, “two men.” Cf, Gramm. Crit. Add. to r. 137.
have the less doubt, because in the other declensions the Li-
thuanian dual also agrees in this case most strictly with the
Sanskrit, and the Lithuanian u or % (uo) is, in some other
places, equally the representative of an old d (see §. 162.),
compare, dimi, or didu, “I give,” with zetfa daddmi;
disu, “I will give,” with erearfa ddsydmi. And the mono-
syllabic pronominal bases also in a sound in the dual a;
thus ti=at td, ku=kd. We hold, therefore, the Véda
_ form yar vrikd, the Zend ayer vehrkd, and the Li-
thuanian wilki, as identical in principle: we are, at
least, much more inclined to this view of the matter
than to the assumption that the u of wilki is the last
portion of the Sanskrit diphthong Wt du, and that wilki
belongs to the form Yat vrikdu. In the vocative the Lithu-
anian employs a shorter u, and the accent falls on the
preceding syllable: thus wilku, opposed to wilki, in which
respect may be compared warep opposed to maryp, and §. 205.
210. Masculine and feminine bases in i and u suppress,
in Sanskrit, the dual case termination Wt du, and, in com-
pensation, lengthen the final vowel of the base in its unin-
flected form ; thus, Wat pati, from fa pati; we sind, from
wy sinu. The gus»gays bdzv-do, “arms,” (from b4zu) men-
tioned in §. 207., is advantageously distinguished from these
abbreviated forms. The curtailed form is not, however,
wanting in Zend also, and is even the one most in use.
From »93/3¢ mainyu, “ spirit,” we frequently find the dual
393 /5G mainyi : on the other hand, for 36¢e brézit, “two
[G. Ed. p.242.]_ fingers,” we meet with the shortened form
»¢¢7¢ érézu, which is identical with the theme (Vend. S.
p- 318, rge7e 23» Ava érézu.
211.. The Lithuanian, in its 7 and wu bases, rests on the
above-mentioned Sanskrit principle of the suppression of
the termination and lengthening of the final vowel: hence,
awt, “ two sheep” (fem.), answers to wat avi, from sfa avi ;
and suni, “two sons,” to aq siéni. On this principle rests
also the Greek dual of the two first declensions. If it be

not desired entirely to remove the w of Avxw from a Grecian
soil, and banish it completely to India, 1t may be allowed
to seek its origin, not in the long a of yar vrikd, but in
the short o of the base, as the first declension has a long
a in the dual, because its bases terminate with a, although
in the common dialect this letter is very frequently repre-
sented by 7. Or may it, perhaps, have happened, that, in
the dual a of the first declension an « subscribed has been
lost, and thus té for 7&@ would correspond to the Sanskrit
# t# (from t4+i or 7)? Be that as it may, still the dual
has always the quality a, because it is comprehended in the
base, and the w of Atvcw may be regarded as merely the
lengthening of the o of Av«o; for it must be assumed, that if
the Sanskrit a bases had preserved the short a in Greek, and
yaa vrika-s had become Avca-s, then the dual too would
be Avxa, and not AvKw.
212. Neuters have, in the Sanskrit dual, for the termi-
nation of the cases under discussion, not #t du, but 4, as in
the plural they have not as but short i(z). A final & a
of the base with this = @ passes into wz é (§. 2.);, hence,
Ma saté, “two hundred,” from gra sata-i: —[G. Ed. p. 243.]
other vowels interpose a euphonic n; hence, areat té/u-n-Z,
“two palates.” In Zend I can quote the neuter dual only in
the a bases; as, for example, we frequently find POP sas
saité (§. 41.), answering to the Sanskrit ya Saté; and poss
po7gwgaser duyé hazanré, “two thousand,” (§. 43.) for } age
dwé sahasré.
213. The Greek has renounced a termination distin-
guishing the neuter from the two natural genders; but
the Sanskrit’ appears to have extended the neuter ¢ men-
tioned above also to the feminine 4 bases. But the coin-
idence -of the feminine form fax jihwé, “two tongues,”
from fagtjihwd, with the neuter 214 déné, “ two gifts,” is,
as the Zend instructs us, only external, and the two forms
Q
226 Txt FORMATION OF CASES.
meet in quite different ways, and have such a relation to

one another, that in dédné, from déna+¢4, a dual termina-
tion, and, in fact, the usual one of neuters, is actually con-
tained; but in fax jihwé the masculine-feminine termina-
tion du (from ds, §. 206.) is lost, but can, however, be again
restored from the Zend form gusdIas5.s/sasy ndirikay-do, “ two
women.” I believe, that is to say, that faz jihwé has
arisen or been corrupted from fagat jihway-du* in such a
manner, that after the termination has been dropped, the
preceding semi-vowel has returned to its vowel nature, and
has become a diphthong with the 4 of the base (see §. 2. and
cf. p. 121 G.ed.). The dual jihwé, therefore, like the Gothic
singular dative gtbai (§. 161.) would have only an apparent
termination, i.e. an extension of the base which originally
accompanied the real case termination. In Zend, however,
the abbreviaged feminine dual form in » é likewise occurs
(§. 207. NoteT). and is, indeed, the prevalent one; but it is
[G. Ed. p.244.] remarkable, and a fair and powerful con-
firmation of my assertion, that even this abbreviated form
in »~ é, where the appended particle wg cha stands be-
side it, has preserved the case sign $; and, as above,
suvguspaspgegas amérétat-dos-cha, “the two Amertats,” so
we find, Vend. S. p. 58, poyeds sus yng améshes-cha
spénté, “and two Amshaspants” (“non-conniventesque sanc-
tos,” cf. wfaa amisha and Nalus V. 25, 26. and see §. 50.).f
The form ss» és is to be deduced from the full form
sx¢gusddas ay-dos; so that, after dropping the gus do, the pre- —
ceding ay must have been contracted to é, just as (p. 121
* Cf. the dual genitive and locative faeare jihway-ds.

+ The MS, has here assyes9¢G.s amésescha, but ¢ frequently occurs
in the place of yo, although, as it appears, throngh an crror. Cf. L ¢
p. 88, COED powegs bys. porbas aové yasnb amesé spénte;
and see §. 51,
_ G. Ed.) in Prakrit, ef émi has arisen from the Sanskrit

watfa aydmi, by rejecting the 4. We may support the
derivation of faz jihwé from fagat jihway-du, by this
circumstance, also, that in the Véda dialect the feminine 7
bases may lose the dual termination du, and then display the
maked base; thus, in the scholia to Panini, arret saraet
vardhi updnahdu, “ boar-leather shoes,” for arcrat vdrdhydu.
It is very remarkable, that even this Véda form, only one
example of which can be quoted, can be referred to the Zend
language. We find, frequently, yyys»¢e fevishé applied to
feminine dual substantives (e.g. Vend. S. p. 225.); and I
infer that its theme ends with a long, not a short i, from the
frequently-occurring plural accusative wyyps¢gy tevishis
(Vend. S. pp. 99, 102).*
214, To the Sanskrit-Zend feminine dual [G. Ed. p. 245.]
forms in é answer the Lithuanian in i, as ranki, from
RANKA ; so that of the diphthong z é only the last ele-
ment is left. The Lithuanian forms the accusative dual, in
contradistinction to the cognate languages, according to the
analogy of the singular, by a ringing nasal, e. g- witkun. The
Latin has preserved only in duo and ambo a remnant of the
dual corresponding to the Greek, which, however, in the
oblique cases, is replaced by plural terminations. Here fol-
lows a general view of the nominative, accusative, and voca-
tive dual (see §. 148.).
* It is perhaps a participle of the reduplicated pret., according to the

analogy of the Sanskrit afaze ténivas, fem. waet ténusht (Gramm.
Crit. §. 603.); and indeed, from the root »«5¢0 ¢av, “to beable,” it may
signify “ powerful, strong.” The ¢ ¢ for ~ é is explained by the infiu-
ence of the » v. And $C5753-0—0> utayiiti also is an adjective feminine
dual; but I am unable to quote examples of the other cases of this word,
from which to learn whether ¥é or5 iisitsfinal vowel.
qe?
SANSKRIT. ZEND. "GREEK. LITHUANIAN:

on” vrikdu, vehrkdo, rye eo ‘ec eee
es vrikd, vehrkd,* —AdKw, N. wilki, V. wilku.

, n. déné, daté, ddpw, 2 ies
wf. eevee hizvay-do, eee eee
5 jihwé, hizvé, xopa, N. ranki, V. rdnki.

‘qm. pat, paiti ? moot-e, NN, pati, V. pati.
af priti, afriti ? népti-e, N.awz, N. dwi.
on. viri-n-t, ghee idpi-e, tine
iS .
mm
a
# While consonantal bases occur in the dual both with a long and a short
_a, the a bases, contrary to the practice otherwise adopted of shortening a
final d, exhibit in the nom. acc. dual, for the most part, the original long
vowel. I deduce this, among other words, from the so-called Amshas-
pants, which, together with the feminine form noticed at §.207. Note t.,
are found alsoasmasculine ; e g. Vend. S. pp. 14. 30, 31, &c.: AUEDIEGAS
J970 YSN berzgusgru 7G assur wwpyyeday améshd spintd
hucsathré hudéonhé dyésé, “1 glorify the two Amshaspants (non conni-
ventesque sanctos) the good rulers, who created good.” If amésha spentd
and hucsathra were plural forms, the final a would be short, or at least
appear much more frequently short than long; while, on the contrary,
these repeatedly recurring expressions, if I mistake not, have everywhere
a long a, and only in the vocative a short a (Vend, 8. p. 67. Cf. §. 209.).
That the epithet huddonhé is in the plural cannot incur doubt, from the
dual nature of the Amshasp (cf. §. 208): this resembles, to a certain
degree, the use of adjective genitives referring to a substantive in the
ablative, which was mentioned in §.180. We find, also, the forms
ameshdo spéntdo (Vend. S. p. 313.), which indeed might also be feminine
plural furms, but. shew themselves only as masculine duals, in the same
meaning as the so frequent ameshé spénta, We find also, frequently,
PISSING MUS YEdI9 spénisté mainyt, “the two most holy spirits”
(p. 80), through which the dual form in @ of bases in a is likewise con-
firmed in the most unequivocal manner. The answer to the query,
Whether generally only two Amshaspants are to be assumed? whether
the genitive plural (ameshananm spéntananm), and sometimes also the
accusative plural, is only the representative of the dual, which is yery
uncertain and shaken in its use? whether under the name Amshaspants,
perhaps, we should always understand the Genii Haurvat (Khordad) and
Amertui
SANSERIT. ZEND. GREEK. LITHUANIAN.
f. bhavishyanty-du, bishyainti, .... a a
m. sind; -- past, - tyGt-e, N. suni, V. séinu.
f. tant, - - . tant, TiTv-€, Jee
m. madhi-n-i, o"s*e%s pédu-e, see
f. vadhw-du, sees ee eeive
m. f. gav-du,* eeee Bot F)-e, Sees
f. ndv-du, e'e"s"s ‘wa(F)-e, coi
f. vich-du, vach-do, eee eree
Amertat, and whether these two Genii, according to the principle of the
Sanskrit copulative compounds, have the dual termination for this reason
alone, that they are usually found together, and are, together, two?
whether, in fine, these two twin-genii are identical with the Indian
Aswinen, which were referred in §.208. to the Zend-Avesta? The reply
to all these queries lies beyond the aim of this book. We will here only
notice that, Vend. S. pp. 80 and 422, the Genii Haurvat and Amertat,
although each is in the dual, still are, together, named MO E99
PHOIMTE WAKE PII/ING gpenista mainyt mazda tevishi, &c., “the
two most holy spirits, the great, strong.” As Genii, and natural objects
of great indefinite number, where they are praised, often have the word
vispa, “all,” before them, it would be important to shew whether “all
Amshaspants” are never mentioned ; and the utter incompatibility of the
Amsh. with the word vispa would then testify the impassable duality of
these Genii. If they are identical with the celestial physicians, the Indian
Aswinen, then “ Entireness” and “Immortality” would be no unsuitable
names for them. In Panini we find (p.803) the expressions ATAtfaaa
métara-pitardu and fqATATAT pitara-mdtaré marked as peculiar to the
Védas. They signify “the parents,” but, literally, they probably mean
“two mothers two fathers,” and “two fathers two mothers.” For the
first member of the compound can here scarcely be aught but the abbre-
viated dual pitard, ma&taré; and if this is the case, we should
here hava
an analogy to the conjectured signification of haurvat-a and amérétat-a.
* Bases in #6 form the strong cases (§. 129.) from yt du ; those in
‘Sq an, and nouns of the agent in AZ Zar, lengthen in those cases,
with
the exception of the vocative singular. the last vowel but one (see
§. 144.). ‘
SANSERIT. ZEND. GREEK. LITHUANIAN,

vach-d,* vdch-a, on-e, “8008
m. bharant-du, barant-do, ...- coe
bharant-4, barant-a, pépovT-e, oe
m. dtmdn-du,t asman-do, .... adit
dtman-d, asman-a, Saiuov-e, N. V. Gkmen-u.
n. ndmn-t, aistals TAAQY-€, Pa ie
m. bhrétar-du, brdtar-do, ...- a shia
bhrdtar-4 brdtar-a, matép-e, siete
f. duhitar-du, dughdhar-do, .... eee
duhitar-4, dughdhar-a, Ovyarép-€, ~— «se
m. ddtdr-du,t datdr-ao, eee ccee
datar-d, datdr-a, doT7p-e, oma.
n. vachas-i, eteAye éne(a)-€, ove
INSTRUMENTAL, DATIVE, ABLATIVE.

215. These three cases have in the Sanskrit and Zend dual
a common termination; while in Greek the genitive has
joined itself to the dative, and borrowed its termination from
it. It is in Sanskrit s1H bhydm, which in Zend has been
abbreviated to »33s bya. Connected with the same is, first,
the termination vaR bhyam, which, in the pronoun of the two
first persons, denotes the dative singular and plural, but
in the singular of the first person has become abbreviated
to aq hyam (§. 23). This abbreviation appears, however,
[G. Ed. p.249.] to be very ancient, as the Latin agrees
* The Véda duals in &@ are as yet only cited in bases in a, m, and ar
(sq, §.1.); however, the Zend leads us to expect their extension to the
other consonantal declensions, as also the circumstance that, in other parts of
grammar, in the Védas 4 is occasionally found for du, and other diph- —
thongs; e.g. Art ndbha, as locative for anit nabhau, from arf nabhi,
navel.”
_ t See the marginal note marked (*), p. 229.
INSTRUMENTAL, DATIVE, ABLATIVE DUAL. 231
remarkably with it; and mi-hi corresponds to waA ma-hyam,

as ti-bi does to qa ftu-bhyam. In the second place, wa
_bhyas, which expresses the dative and ablative plural, is
pronounced in Zend byé (§. 56°.), in Latin bus, suppressing
the y, and with the usual change of as into us. The Li-
thuanian has mus for bus in the dative plural (§. 63.): this
more complete form has, however, remained only in the
pronoun of the two first persons, where mu-mus, “nobis,”
-mus, “ vobis,” are used as well as mu-m’s, yu-m’s; while
in all other words we find simply ms as the sign of the
dative—wilka-ms, &c. In the dual dative the Lithuanian
has only the m of the Sanskrit termination »1q_ bhydm, as
wilka-m. This m is, however, not the final letter of bhydm,
but the initial labial, 5, in a nasal form (§. 63.)*: to me, at
least, it appears improper to regard this dual termination
otherwise than that of the cognate plural case; and I have
no doubt of the identity of the m of wilka-m, Avxoww, with
that of wilka-ms (for wilka-mus), AvKois. According to this
explanation, therefore, the German plural dative corresponds
to the Lithuanian dual dative, vulfa-m, gasti-m, sunu-m.}
216. A third form related to the dual ter- [G. Ed. p. 250.]
mination »1m bhydm is fa bhis, as sign of the instru-
mental plural. This termination which is in Zend «4.5 bis,
* On the facile transition of v into m (cf. p.114) rests also, I doubt

not, the connexion of the termination Tay yuram, “ye two,” Sra
dvim, “we two,” with the common termination du, before vowels ar,
which in the pronouns spoken of has stiffened into dm, and in this form
has remained even before consonants. Whether the case is the same with
the verbal third dual person WTA tam shall be discussed hereafter.
+ Cf. Grimm, I. 828.17, where the identity of the Lithuanian-German
inflection m with the b (bh of the older languages) was first shewn. When,
however, Grimm, l.c., says of the Lithuanian that only the pronouns and
adjectives have ms in the dative plural, the substantives simply m, this is
perhaps a mistake, or the plural is named instead
of the dual; for Ruhig
gives ponams, “dominis,” akims, “oculis,” &e.
232 : FORMATION OF CASES. —
(also 3.85 bis), has in Latin fixed itself in the dative and
ablative,* which must together supply the place of the instru-
mental; while in Lithuanian, with the exchange of the
labial medial for the nasal of this organ (§. 63.), mis is the
property of the instrumental alone, so that puti-mis answers
to ufafaa pati-bhis, 033450439 paiti-bis.
217. I have already elsewhere affirmed, that the Greck
termination ¢i, giv, is to be referred to this place,t and what
is there said may be introduced here also. If giv, and not
gi, be assumed to be the elder of the two forms, we may offer
the conjecture that it has arisen from ¢rs, following the analogy
of the change of ses into ev in the Ist person plural, which
corresponds to the Sanskrit mas and Latin must; dts would
correspond to the Sanskrit bhis and Latin bis, in nobis, vobis.
Perhaps, also, there originally existed a difference between
gt and giv (which we find used indifferently for the singular
and plural), in that the former may have belonged to the
singular, the latter to the plural; and they may have had
the same relation to one another that, in Latin, bi has to
bis in tibi and vobis; and that, in Lithuanian, mi has to mis
in akimi, “through the eye,” and akimis, “through the
eyes.” It has escaped notice that the terminations ¢: and
[G. Ed. p.251.]_ gw belong principally to the dative: their
locative and instrumental use—autdgu, bend, Binpw—is ex-
plained by the fact, that the common dative also has assumed
the sign of these relations. The strict genitive use of the ter-
mination du, giv, may perhaps be altogether denied; for if pre-
positions, which are elsewhere used in construction with the
* In the Ist and 2d pronoun (no-bis, vo-bis), where bis supplies the
place of the bus which proceeds from wma bhyas.
+ Trans. Berlin Academy, 1826. Comparison of Sanskrit with its eog-
nate languages, by Prof. Bopp. Essay III. p. 81.
¢ Observe, alsc, that the Sanskrit instrumental termination Dhis has
been, in Prakrit, corrupted to fz Ain.
genitive, occur also with the case in gr, giv, we are not com-
pelled, on this account, to regard the latter as the genitive
or representative of the genitive. In general, all prepositions,
which are used in construction with the genitive, would,
according to the sense, be better used with an ablative or a
locative, if these cases were particularly represented in
Greek. The suffix Gey also, of genuine ablative signification,
expressing separation from a place, is incorrectly consi-
dered to represent the genitive termination, where the
latter, in the common dialect, has received the sign of the
lost ablative. In doce daxpuddiv miunAavto, daxpvodi would,
in Sanskrit, be rendered by wyfita asrubhis: the relation
is entirely instrumental, and is not changed because the
verb mentioned is more usually, though less suitably, used
with the genitive. The same is the case with doce da-
kpvody tépcavto. In *IAidgu KAuTa tetyea it is not requisite
to make ‘IA:dg¢s governed by te‘yea, but it may be regarded
as locative “to Ilium.” And in Od. XIL. 45. (xoAds FP aug
éctedduv bis dvdpGv muGouéver) there is no necessity to look
upon éoTedgw as the genitive, for it can be aptly rendered
_ by ossibus. I know no passages besides where a genitive
meaning could be given to forms in g: and giv. To the
accusative, likewise, the form ¢:, dw, is foreign, and accord-
ing to its origin does not suit it; nor does it appear in
the train of prepositions, which elsewhere occur with the
accusative, with the single exception of és évyyguv in Hesiod
(cf. Buttmann, p. 205). Astothe opinion ([G. Ed. p. 252.]
of the old Grammarians, that ¢:, gv, may stand also in the
nominative and vocative, and as to the impropriety of the «
subscribed before this termination in the dative singular of
the first declension, we refer the reader to what Buttmann
(p. 205) has rightly objected on this head.
. 218. The neuters in 2, mentioned in §. 128., are nearly
the only ones from bases ending with a consonant, which
occur in combination with @:, gm, in forms like dyeo-pi,
opeo-pi, o770ec-prv, which have been misunderstood, be-

cause the = dropped before vowel terminations was not
recognised as the property of the base. Of the other con-
sonants, v is the only one, and KOTYAHAON the only y
base, which occurs in combination with gv; and since N
does not combine with ® so readily as 3, it assumes an auxi-
liary vowel o—xotvAnddv-o-giv—after the analogy of com-
pound words like xvv-o-Oapo7ys. This example is followed,
without the necessity for it however, by daxpu—daxpudqu;
while vav-div, in an older point of view, resembles exactly
the Sanskrit #frq ndubhis; for in compounds, also, the
base NAY keeps free from the conjunctive vowel o, on which
account vavctabuov may be compared with Sanskrit com-
pounds like #teq ndu-stha, “ standing (being) in the ship.”
219. But to return to the Sanskrit dual termination
way bhydm, it is further to be remarked, that before it
a final w a is lengthened; hence, yarata vrikdbhydm for
garaty vrikabhyam. It hardly admits of any doubt, that
this lengthening extended to the cognate plural termina-
tion fra bhis ; and that hence, from ya vrika also vrikd-bhis
would be found. The common dialect has, however, ab-
breviated this form to Faq vrikdis, which is easily derived
from vrikdbhis by rejecting the bh; for & di is, according
[G. Ed. p.253.] to §. 2.,=d+% This opinion, which |
have before expressed,* I can now support by new arguments.
In the first place, which did not then occur to me in dis-
cussing this question, the pronouns of the two first persons
really form from their appended pronoun & sma, smd-bhis;
hence wenfiq asmdbhis, yarfra yushmdbhis ; which forms
stand in the same relation to the garfra_ vrikd-bhis,
assumed by me, that the accusatives SmTq asmdn, TATA
yushmdn, do to Tata vrikdn, “lupos.” Secondly, the opinion
* Trans, Berlin Academy, 1826. Comparison of Sanskrit with its cog-

nate languages, by Prof. Bopp. Essay III. p. 79.
which [ arrived at theoretically has, since then, been so far
practically established by the Véda dialect, that, in it, from a
final wa not 4-bhis but &bhis, has been formed, according to
the analogy of the dative and ablative, as yaraq_vrikébhyas;
hence, wafita asvébhis, “ per equos,” from ww asva. In the
common dialect the pronominal form efirg é-bhis “per hos,”
answers to this Véda form, which must properly be de-
rived from the pronominal base ¥ a, which generally plays
the chief part in the declension of ¥aq idam. Though, then, on
one side, from the pronoun ¥ a springs the form eftra é-bhis;
on the other side, from wet asma and FH yushma proceed
the forms wentfita asmabhis, gartfra yushmabhis; and though
the Véda dialect, in its substantive and adjective bases in a,
attaches itself to the former form, still no necessity hence
arises for supposing the abbreviated dis to be based on an
é-bhis,* as that could never lead to dis. Perhaps, however,
dbhis might become @bhis, either through the assimilative
force of the i of bhis, or through analogy to [G. Ed. p. 254.]
the dative &bhyas, the é of which may, in like manner, owe
its origin to the re-active influence of the q y.f
220. The Prakrit has fully followed out the path com-
menced by the Véda dialect, and changed into z é the 4 of
* From ébhis would come, after rejecting the bh, not dis, but ayis, for
é, =a+i, cannot be combined with a following i into a diphthong, or, as
it is itself already a diphthong, into a triphthong.
+ I do not regard the Véda Aaa nadyiis, for aetieg nadi-bhis, as
an abbreviation of nadi-bhis (for after rejecting the bh, from nadi+is
would be formed nadis), but as a very common instrumental, for which
an extension of the base nadi to nadya is to be assumed. On the other
hand, the Zend pronominal instrumental dis mentioned by Burnouf
(Nonuy. Journ. Asiat. ITI. 310.) may here be considered, which occurs fre-
quently in the Jzeshne, and is probably an abbreviation of 213.55.59 dibis or
S555 dibis, from a base di, the accusative of which 354 dim, “him,”
is often found with é unlengthened, contrary to §.64. The connection of
the base +
$9 di with aso ta cannot, on this account, be disputed.
asm4-bhis, yushmd-bhis, as also, in the locative plural, that of

asmdsu, yushmdsu; hence weefé amhé-hin, Tete tumhé-hin,
weed amhésu, Wea tumhésu. Moreover, in Prikrit, all other
a_ bases, as well pronouns as substantives and adjectives,
terminate the instrumental plural with ef€ e-hia ; and thus
myate kusumé-hin, “ floribus,” (from kusuma,) answers to the
Véda eqafia kusumé-bhis. Before, however, the forms in
cia é-bhis, ef é-hin, had arisen, from wife dbhis, by the
change of 4 into é, dis must have proceeded by means of
rejection and contraction from that most early form. This
form exists also in the oldest hymns of the Védas, together
with that in viva ébhis: thus, in Rosen, p. 14, aaa yajniis;
pp- 15 and 21 wary arkdis. In Zend the abbreviated form
dis is the only one that occurs, which it does, indeed, ex-
tremely often.
221. Before the dual termination 33s bya the Zend, in
[G. Ed. p.255.] its a bases, differs from the Sanskrit in the
same way as the Zend and Prakrit do before the termina-
tion fra bhis, fe hin; it employs, namely, é for d: but
from véhrké-bya, according to §§. 28. 41. comes véhrkaéibya.
Thus, in the Vendidad, 2539ss7.se
wd rasddssroa»w hvatibya
padhaéibya, “ suis pedibus,” = Saray Tenarsa swabhydm pada-
bhydm; asdds7xesI056 zastaéibya (Eva) “ manibus.” But
in this case, also, the diphthong z é is supplied by di (§. 33.) ;
e.g. asddsslss> ubdibya, “ ambobus” (Vend. S. p. 305). If in
this form the lost nasal be restore¢, and it be assumed (of
which I have no doubt) that the Greek dual termination w is an
abbreviation of the Sanskrit bhydm;* then the Homeric forms
like duor-iv are to be compared with the adds > ubdi-bya
* By rejecting the labial, as in qaqrq vrihdis from yanrfire vrihadhis,

and by contracting the qTq ydm to w, as when, in Sanskrit, for yashta,.
ishta is said, from yaj, “to sacrifice,” and n Zend ¢ sim, “hee,” for
F474 ‘yam (see, also, §. 42.).
above mentioned; where, therefore, the first « would fall to

the base, which it lengthens, the other to the termination,
The third declension, by its forms like da:udv-o1v, might give
rise to the conjecture, that or and not wy is the true termina-
tion: the latter, however, is shewn to be so from the two
first declensions, where wv and not ov is attached to the final
vowel of the base (Movca-iv, Adyo-wv). In the third, there-
fore, we explain the o before sv in the same manner as, $. 213.
before giv (xotvAndov-6-giv); viz. as a conjunctive vowel,
which has made its way from the bases which necessarily
have it, ie. from those terminating in a consonant into
those which might dispense with it (into the bases in
tand v); as, in general, in the third declension the conso-
nantal bases. have given the tone, and have shewn the way
to the vowels:and v. Itmight, however, not [G. Ed. p. 256.]
have been necessary for the conjunctive vowel o to make its
appearance between consonants and the termination, as
Samov-rv could very easily be uttered; but the o of damudvorw
comes evidently from a time when the w was still preceded
by the consonant, which the corresponding Sanskrit termi-
nation bhydm leads us to expect ; im all probability ag; thus,
Sa:uov-o-rv, from damov-o-piv.* We should have, therefore,
here a different gw from that which, in §. 217., we endea-
voured to explain from @:s, fa bhis: the nasal in the dual
(p)w stands quite regularly for its predecessor m, as, in ge-
neral, at the end of words. In order to present to our
* The conjunctive vowel o, therefore, before the dual termination w,

has an origin exactly similar to that of the possessive suffix evr, which has
been already elsewhere compared with the Sanskrit wa vant. Evr must
therefore have been originally pronounced Fevr; and the conjunctive
vowel, which the digamma made requisite or desirable before consonantal
bases, and which, from thence, has extended itselfto the whole third
declension,
has remained also after the digamma has been dropped, and
thus zrup-d-ers answers to rupoiv, from svp-o-ir: on the other hand, rupé-ess
to ripow (rupo-iv).
view still more clearly how forms quite similar take root
in the language as corruptions of preceding dissimilar
forms, let the form érurrov be considered as the first per-
son singular and third person plural; in one case from
érunrop, in the other from érumrovr.
222. If the dual termination w be explained as a con-
traction of bhydm, we shall have found, also, the origin of
the dative plural termination «v, which appears to have been
changed in this number in the pronouns of one gender as
it were by accident (ip’-tv, ty-iv, o¢’-iv, together with
opi-o1). The Greek, however, in this respect, is guided or
misled by the Sanskrit; or, more correctly, the distinction
of the plural dative of the pronouns of one gender is very
ancient, and the Sanskrit has in them wa bhyam as termi-
nation (wenaTy asma-bhyam, “nobis,” Tara yushma-bhyam,
(G. Ed. p. 257.) “wobis”), opposed to the wra_ bhyas of all
other words. From this bhyam, then, we arrive at «v quite
as easily, or more so, than from the dual termination bhydm
(cf. §. 42.). As, however, 4 bhyam, and its abbreviated form
aq hyam, according to §. 215., has also its place in the singular
dative of the pronouns of one gender, but occurs nowhere
else; as, moreover, the Latin also, in the pronouns referred
to, has maintained a genuine dative termination, and to the
common i, which is borrowed from the locative, presents in
contrast the termination bi or hi (for bhi) (§. 200.); we can,
therefore, in the singular wv also of éu’-iv, re-lv, 7-iv, iv, of'-iv,
see nothing else than an abbreviation of = bhyam, a form
which the Latin and Greek have shared in such a manner,
that the former has retained the beginning and the latter
the end. In the i both coincide.* The occasional accu-
* A short time since, Max. Schmidt, in his excellent treatise ‘‘ Com-

mentatio de Pronomine Greco et Latino” (p.77), endeavoured to con-
nect the termination w here treated of with the Sanskrit in a different
way, by designating it as the sister form of the pronominal locative ter-
mination
gative use of this termination, in Theocritus, is to be ex-

plained from its original signification being no longer felt,
and the exchange of its v with that of the accusative thereby
caused. On the other hand, we have in piv and viv real accu-
satives, and should therefore divide them pi-v, wi-v; and
not assume, with Buttmann (p. 296), a connection between
this form and the dative -iv.
223. As to the origin of the case-suffixes [G. Ed. p. 258.]
firq_ bhi-s, eam bhy-am, waTR bhy-dm, and wre bhy-as, which
begin with + bhy (from fa bhi), we must notice, first,
their connection with the preposition wf abhi, “ to,” “ to-
wards,” “against,” (whence wfiwa_abhi-tas, “at,” ef. “apud™).
However, in abhi itself bhi is clearly, in like manner, the ter-
mination, and the demonstrative Wa the theme; so that this
preposition, in respect to its termination, is to be regarded
as a sister form to the Latin fi-bi, si-bi, i-bi, u-bi ;* just as
another preposition, which springs from the pronominal
base a, viz. wfa adhi, “ over,” finds analogous forms in the
Greek locatives, like 5-01, AAo-61, ovpavd-6 (§. 16.). Related
to the suffix fu dhi is ¥ dha, which has been retained in
the common dialect only in the abbreviation ha, in 7-ha,
“here,” and in the preposition sa-ha, “with”; but in the
Véda dialect exhibits the original form and more extended
diffusion, and in the Zend, also, is found in several pro-
mination ¥¥ in (§. 201.). In this view similar forms would be con-

trasted, exclusive of the length of the Greek w, which, according to my
explanation, may pass as compensation for the a, which has been dropped.
Still I lay less stress on the difference of quantity than on this, that it is
precisely the pronouns of one gender in the Sanskrit, which exhibit in the
locative not in but the common i (§.201.), but I attach still more weight
to what has been said above in support of my opinion.
* In Prakrit the termination f¥ hin, which is connected with fay bhi
(cf. §. 217.), unites also with other pronominal bases, for the formation of
locative adverbs, as Af¢ ta-hin, “there,” fe ka-hin, “ where ?”’
nominal bases with a locative signification; eg. »ex»as

ava-dha, “here.” In the Greek, compare 9a of évOa, op-
posed to Gev, from évéev, éuédev, &e., from wa dhas, for
wa tas, in Wa a-dhas, “beneath”: in which formations
4 dh stands as a permutation of ¢, and occurs in this way,
also, in some other formations.* Th refore dha, dhi, are
to be derived from the demonstrative Lase # ta; but it is
more difficult to trace the origin of th fa bhi of wfa abhi
(Greek audi). I suspect that an initia] consonant has been
[G. Ed. p. 259.) dropped. As in Greek, also, pév is used for
opiv, and as in Sanskrit fayfa vinsati “twenty,” is clearly
an abbreviation of fiyrfa dwissati, and in Zend ww 5_s bis,
“twice,” asysoss bitya, “the second,” is used for 034 dvis,
(Sanskrit faq dwis), 539009 dvitya (Sanskrit fgata dwitiya),
so fit bhi may be identical with the pronominal base’= swa
or fe swi—whence the Greek o¢eis, opiv, piv, &c.; amd so
indeed, that after the s has been dropped, the following
semi-vowel has been strengthened or hardened, just as in
the Zend »33_s bis, »ysess bitya, and the Latin bis, bi. The
changed sibilant might also be recognised in the aspira-
tion of the ¥ bh, as, in Prakrit (§. 166.), # sma has become
tzmha; and, (which comes still closer to the case before us), in
Greek for opiy is found also Wiv. And, in Sanskrit, that 4 bh
should spring from b+/ is not entirely unknown; and in
this way is to be explained the relation of waa bhiyas,
“more,” to 7 bahw, “ much,” the a being rejected (Gramm.
Crit. r. 251. Rem.).
224. The following will serve as a general view of the
dual termination under discussion, in Sanskrit, Zend, Greek,
and Lithuanian:—
* Among others, in the 2d person plural of the middle sy diwé and

eae d)wam for % twé, eA twam.
SANSKRIT. ZEND. GREEK. LITHUANIAN.
in. vrikd-bhydm, hooce p42 ino wilka-m.

f. jihwd-bhydm, _hizvd-bya, xOpa-v, ranko-m.
- m. pati-bhydm, paiti-bya, mogi-o-tv, pati-m.
f. tanu-bhyém, tanu-bya, WtT0-0-tv, FIA
£. vag-bhydm, vdch-e-bya,* én-0-tv, Cieuns
m. bharad-bhydm, baran’-bya, gepovt-o-v, ..-.. 3S
m. dtma’-bhyém.t asma’-bya, Soupov-o-iv, .... 5
~
3
* I deduce this form principally from the base vba? raoch, “ light,”
which often occurs in the terminations beginning with_s 5, and always
interposes <e asconjunctive vowel—w55_5 gybas? raoch-e-bis, byssequbis?
raoch-e-by6. We find, also, ass seuasm 6 vi-vach-e-bis (Vend. S. p. 63.).
Bases in 7 r interpose € @; those in @ ¢, when a vowel precedes that
letter, conjoin the termination direct (asd3sroaspaupg eg amérétdtat-
bya, according to §. 38.): on the other hand, the @ ¢ of ys né is
rejected; thus, V.S. p. 9. asddssece7e_s bérézén'-bya, “ splendentibus,”
with j> contrary to §. 60. The form Gesss roads brvat-byanm,
“ superciliis,” also deserves notice, because in this solitary word the case
termination appears unreduced (§.61.). The MS., however, as often as
this word occurs, always divides the termination from the base (Vend. S.
P. 269, twice G35 roam) brvat byanm ; pp. 321 and 322, roasmhass
barvat byanm, probably for bravat byanm; so that it would seem
that 9025»; brvat isthe ablative singular of atheme 97 bri (Sansk. y
bhréi). I have not found this word in any other case: it is not likely,
however,that any thing but pass brvat or © as»4y brvant is its
theme: in the latter case it would be a participial form, and would
demonstrate,that instead of the last consonant of nt, the last but one also
maybe rejected. Or are we to regard brvat byanm as a form of that sin-
gular kind that unites with the termination
of the ablative singular that
of the dual, and thus 97. bri would still bethe theme?
+ N, in Sanskrit
and Zend,isrejected before case terminations beginning
with a consonant ; thus, in Greek, daipo-c1, andin Gothic akma’-zm.
R
SANSKRIT, ZEND, GREEK. LITHUANIAN,

m. bhrdtri-bhydm,* bhratar-é-bya, MATEP-0-1Vy www
n. vaché-bhydm,} —vachéd-bya, éné(c)-0-1"7, se
GENITIVE, LOCATIVE.
(G. Ed. p.261.] 225. These two cases, in Sanskrit, have the
common termination wWra ds, which may be connected with
the singular genitive termination. The following are
examples: Jama vrikay-ds, fagata jihway-ds (cf. §. 158.),
wee paty-ds, wWeata tanw-ds, arate vach-ds, CLEC: §bhrdtr-és,
vachas-és. In Zend this termination seems to have
disappeared, and to be replaced by the plural; likewisein
Lithuanian, where, awy-é is both dual and plural genitive. :
PLURAL.
NOMINATIVE, VOCATIVE.
226. Masculines and feminines have, in Sanskrit, Wa as

for the termination of the nominative plural, with which, as
in the cognate languages, the vocative is identical in all de-
clensions. I consider this as to be an extended form of
the singular nominative sign s; so that in this extension
of the case-suffix lies a symbolical allusion to plurality:
and the s, which is too personal for the neuter, is wanting
in that gender, in the singular and dual, as well as in
the plural. The three numbers, therefore, with regard to
their masculine-feminine termination or personal designa-
tion, are related to one another, as it were, like positive,
comparative, and superlative, and the highest degree be-
longs to the dual. In Zend wa as has, according to §. 56.
* aq ar before case terminations beginning with consonants is short-

ened to = ri (§. 127.).
+ See §, 56>.
NOMINATIVE, VOCATIVE PLURAL 243
become 6 or 391s as before the appended particles cha and

chit; the Greek exhibits es, under the restriction of §. 228. ;
the Latin és,* with unorganic length of quantity through
the influence of the s; the Lithuanian has és in bases in r
but elsewhere simple s. Thus the words efgata duhitar-as,
apse 9 4 dughdhar-as-cha, Guyarép-es, dukter-és, matr-
-es, correspond with one another.
227. Thea of the termination is melted [G. Ed. p. 262.]
down with a preceding # a of the base to 4; thus, |
vrikds, from vrika + as, corresponds to the Gothic vulfés, from
VULF4as (§.69.). In this concretion only, however, with
the vowel of the base, the Gothic has preserved the full ter-
mination; but elsewhere, both with vowel and consonantal
bases, the s alone of the old as is left, as in general the ter-
mination asin Gothic polysyllabic forms has everywhere been
weakened to is or s (cf. §§. 135. 191.): hence, sunyu-s, ahman-s,
for suniv-as, ahman-as. And @t 4, too, is contracted with
the termination as to ds; hence, fraTa jihwas, for jihwé-as.
It cannot, however, be shewn with certainty, from what
has been just said, that the Gothic gibés, from GIBO, has
_ simple s or as (contracted with the base vowel to 6=4) for
its case designation.
228. The masculine pronominal bases in a refuse, in
Sanskrit, Zend, and Gothic, the full nominative designa-
tion, and in place of it extend the base by the addition
of an i, which, according to §. 2. with the a of the
base forms x é} for which, in Zend, is used ~ é or sb 6i;
* Vide§.797.p. 1078.
WE og eehcnad “tswich cahn eesfo
wb ald OK Os
case terminations are then first conjoined, there is good ground
to assume
that in 7 ¢é, and similar forms, no case designation at all is contained, and
that the pronouns, as purely words of personality, find themselves snffici-
ently personified in this case through themselves alone ; as in the singular
sz is said for sas, in Sanskrit as in Gothic, and in Greek 6 for és; whilein
Latin, with is-te also ipse and ille are robbed of the nominative sign.
This opinion is remarkably confirmed by the fact that Wat ami (Grimm.
R2 Crit,
hence, Sanskrit 2? té, Zend pow té, Gothic thai, “ this,”

[G. Ed. p.263.] answering to the feminine form qe tds,
gusgo tdo (§. 56".), thds. To this corresponds, in Greek, tof
(Doric for oi). In Greek and Latin, however, this i, which
practically replaces the termination as (es, és), has not re-
mained in the masculine pronominal bases in 0 (=® a,
§. 116.); but all other bases of the second, as of the first declen-
sion, have, in Greek and Latin, taken example from it; hence,
AvKolt, yGpat, for Avxo-es, ywea-es, lupi (from lupoi), terre
(from terrai), for lupo-es, terra-es. The Latin fifth declension,
although in its origin identical with the first (§. 121.), has
preserved the old termination; hence, rés from ré-és, as,
in Sanskrit jihwds from jihwd-as. The Lithuanian has
fixed narrower restrictions than the Greek and Latin on
the misuse of the pronominal inflexion under discussion, or,
to speak more correctly, want of inflexion: it gives, indeed,
wilkai=Adxo1, lupi, but not rankai, but rankos. Honour,
therefore, to the Gothic! that in this respect it has not
overstepped by one hair the old Sanskrit-Zend limits; for
that the adjective a bases, as they in general follow the
pronominal declension, give also ai for és (blindui “ ceci”) is,
therefore, no violation of the old law.
229. In Zend, in consonantal bases the dual termination
gus do also (from wre ds, §. 207.) occurs with a plural signi-
fication ; thus, frequently, gusspsasly vidch-do, “voces,” gussas?
Crit. §. 271.) shews itself clearly through most of the oblique cases, as
ami-byas, “ illis,” ami-sham, “illorum,” to be the naked theme. The form
which occurs in the Zend-Avesta asps9gdsgh vispes-cha, ‘ omnesque”?
(V.S. p. 49), considered as a contraction of vispay-as-cha (cf. §. 244.),
leads to the conjecture, that to # ¢é, and similar uninflected forms, the
termination as also might attach itself; thus, WAR tay-as. In Zend, the
pronominal form in é occurs, for the most part, in the accusative plural;
and thus the abovementioned vispes-cha 1. c. stands probably as accu-
sative, although, according to Anqueti]’s inaccurate translation, it might
be regarded as the nominative.
NOMINATIVE, VOCATIVE PLURAL. 245
-raoch-Ho, “luces,” which forms cannot be regarded, perhaps,

as regular plurals of basesind@; forI believe ([G. Ed.p.264.]
I can guarantee that there exists no such base as wyawh
vachd and web? racchd. The form bw gus donhd in a
bases, as Iepeinicch véhrkdonhé, “Tupi,” and “lupos,” rests
on that in the Védas, but which only occurs in the nomina-
tive, "Tea dsas (8. 56.); e.g. SASS stémdsas, “songs of
praise,” for etata stémds, from wim stéma.*
230. Basesini
and u have, in Sanskrit, Guna; hence wag
patay-as, TATA siinaw-as, for paty-as, siinw-as. The Gothic
also has preserved this Guna, but in its weakened form i
(§. 27.), which, before u, becomes y; hence, sunyu-s, “sons,”
(for suniu-s, from sunau-s,) a form which would be unin-
telligible without the Guna theory, which has been shewn
to belong to the German. It i bases the Guna i is melted
down with that of the base to long i (written ei, §. 70.);
hence, gastei-s, anstei-s, from GASTI, ANSTI (cf. p. 105.).
The Zend employs Guna or not at pleasure ; hence by spsas0d
paity-é, or paitay-6," bys950 pase-6, or pasav-d.
231. Neuters have, in Zend, as in the cognate Euro-
pean languages, a short a for their termi- ([G. Ed. p. 265.]
nationt; perhaps the remains of the full as, which belongs
to the natural genders, after the s, which is too per-
* This form is,in myopinion,

to be so regarded,
as that,
for greater
emphasis,
the termination as has been a second time Sa 2 the
termination, which had become concrete with the base.
+ The é, which, accordingto §.41., isblended with the bade, remains
in spite of the a preceding
the y.
t Simple as this point is, I have nevertheless found it very difficult to
come to a firm conclusion regarding it, although,
from the first,
I have
directed my attention towards it. Burnouf has Already (Nouv. Journ.
Asiat, I11. 309, 310) given the plural neuter form, and instituted com-
parisons with the Gothic and Greek, &c. But from forms like hu-mata,
“ bene-cogitata,” “ hiicta,”’ “‘ bene-dicta,” it cannot be perceived what the
neuter plural termination properly is; because, setting out with the San-
skrit, we are tempted to assume that the true termination in these forms
has
sonal for the dead speechless gender, has been dropped.

[G. Ed. p. 266.] This a remains, then, in the accusative.
The masculine and feminine have, in the same case,
generally likewise as (Zend 4 6, Asgssoas ascha). The
following are examples: asys»anys ashavan-a, “pura,”
aspyyrasse% __s bérézant-a, “ splendentia;” as vach-a, “verba;”
assy nar-a, “ homines;” asceyas ast-a, “ossa.” In no-
minal bases in a the termination is melted down with
the vowel of the base: the d so produced has, however, in
the received condition of the language, according to a
has been dropped, and its loss either compensated by lengthening the final
vowel, or not. We must therefore direct our attention to bases with a
different termination than a, especially to such as terminate with a con-
sonant. The examination of this subject is, however, much embarrassed,
in that the Zend, without regard to the gender of the singular, is prone,
contrary to natural expectation, to make every noun neuter in the
plural ;an inclination which goes so far, that the numerous class of a bases
have hereby entirely lost the masculine nominative, and but sparingly
exhibit the masculine accusative. When, e.g. mashya, “human being,”
is, in the plural nominative, likewise, mashya (with cha, mashya-cha), here
I am nevertheless convinced that this plural mashya, or mashyd, is not an
abbreviation of mashydn from mashyds (§.56>.), as in no other part of
Zend Grammar ys a or ww 4 stands for “aTa_ ds: I am persuaded that this
form belongs to the neuter. The replacing, however, of the plural mas-
culine by neuters rests upon a deep internal feeling of the language;
for in the plural number it is clear that gender and personality are far in
the back ground. The personality of the individual is lost in the abstract
infinite and inanimate plurality ; and so far we can but praise the Zend
for its evitation of gender in the plural. We must blame it, however, in
this point, that it does not, in all places, bring the adjectives or pronouns
into concord with the substantives to which they refer, and that in this
respect it exhibits a downright confusion of gender, and a disorder which
has very much impeded the inquiry into this subject. Thus, e.g. vispa
anaghra-raochdo (not raoch-a), “ all lights which have had no beginning”;
tisaré (fem.) sata or thrayé (masc.) sata, “three hundred” ; chathwéré
(masc.) gata “four hundred.” In general the numbers “ three” and
“four” appear to have lost the neuter ; hence, also, thrayé csafn-a, “three
nights,” chathwdré esafn-a, “four nights”: inVend. S. p. 237, on the other
hand, stands ¢é nara yd, “those persons who’... .” JI divide thus xar-a
although
NOMNIATIVE, VOCATIVE 1 LURAL. 247
principle often quoted, been again shortened, and remains
only in monosyllabic bases and before annexed particles.
The Gothic and Zend, in this respect,stand [G. Ed. p. 267.]
very remarkably upon one and the same footing; for thé,
“hac,” is used (for thd, §. 69.), from TH 4a; hvé, “ que,” for
HV¥4a; but daura, from DAURA, as, in Zend, we té,
“hee,” 23 yd, “que,” opposed to +3945 agha, “ peccata,”
from agha. It- cannot, therefore, be said of the Gothic that
the a of the base has been dropped before that of the termi-
although the form might also belong to a theme nara, which also occurs,
but much less frequently than nar ;whence also, elsewhere, the masculine
nar-6 taé-cha, “and those persons.” From the theme wach, “ word,”
“speech,” we find frequently vdch-a (also, erroneously as it appears,
vach-a); e.g. Vend. S. p. 34, wpughuw spo spagow asa
vdcha humata hicta hvarésta, “verba bene-cogitata,
bene-dicta, bene-peracta.”
From prsmarpas ashavan, “ pure,” occurs very often the neuter plural
*shvana-a: as, however, the theme ashavan sometimes, too, although
very rarely, extends itself unorganically to ashavana, this form proves less
(though it be incorrect) that the neuter ashavan-a should be derived from
the unorganic extremely rare ashavana, than from the genuine and most
common ashavan, in the weak cases ashaunorashaon. Participial forms,
too, in nt are -very common in the neuter plural; and I have never found
any ground for assuming that the Zend, like the Pali and Old High Ger-
man, has extended the old participial theme by a vowel addition. In
the Vend. S., p.119, we find an accusative agha aiwishitér-a, “ peccata
corrumpentia(?).”’ Anquetil renders both expressions together by “ia
corruption du ceur” (11. 227.); but probably aiwi-sitéra stands for
-csitéra, and means literally “the destroying” (cf. fey Ashi, intrans. “to
be ruined”). So much is certain, that aiwi is a preposition (p. 42), and
taris the suffix used in the formation of the word (§. 144.), which is in
the strong cases ¢ér; and from this example it follows, as also from asha-
van-a, that where there are more forms of the theme than one, the Zend,
like the Sanskrit (see Gramm. Crit. r. 185. c.), forms the nominative, ac-
cusative, and vocative plural from the stronger theme. I refrain from ad-
ducing other examples for the remarkable and not to have been expected
proposition, that the Zend, in variance from the Sanskrit, forms its plural
neuters according to the principle of the Latin nomin-a, Greek radav-a,
Gothie namén-a or namn-a.
nation, for it could not be dropped, because the base-vowel

and termination have been, from the first, concrete. The old
length of quantity might, however, be weakened: this is
the fate of long vowels especially at the end of words. It
cannot, therefore, be said of the Greek 7a dGpa and the
Latin dona, that the a entirely belongs to the termination,
This a is an old inheritance of the oldest date, from the
time when the second declension, to use the expression,
terminated its bases with d. This d@ has since then be-
come, in Greek, o or e (§. 204.), in Latin, u, 0, or e, and has
maintained its ancient quality only in the plural neuter,
and the a, which has grown out of d+d, has become
shortened. This da, however, in contrast with its offspring
vd, & u, May even pass for a more weighty ending, which
unites base and termination, than if dwpo or dwpe, dond,
doné, stood as the plural neuter. |
232. Bases ini and uw may, in Zend, suppress their final
vowel before the termination, and u may be suppressed and
replaced by lengthening the base-vowel: thus we read in
the Vend. S. pp. 46 and 48, ase gara, “ hills,” from suse
gairi (see p. 196, Note ft): on the other hand, p. 313, gairés
(fem.). That which Anquetil (II. 268.) renders by “une
action qui empéche de passer le pont, le péché contre nature,”
runs in the original (p. 119), IC a5S Lasssyo 3G g7e0 1 125won
asd30dsabhsy agha andpérétha skyaothna yd naré-vaipaya,
[G. Ed. p.268.] i. e. “the sins which stop the bridge, the
actions which....”; and here it is evident that andpérétha
stands for andpéréthw-a, for pérétu means actually “bridge.”*
* Burnouf’s MS, divides thus, ana pérétha, which is following Olshau-

sen (p. 6), but with the various reading andpérétha. I have no ground
for assuming that in Zend there exists a preposition and, “ without,”’ so
that and pérétha might mean “ without a bridge”; and that pérétu would,
in the singular instrumental, form péréthwa or pérétava, I suppose, there-
fore, that pérétu may be conjoined with the preposition 4, and then the
negative an have been prefixed.
But a final u may also be retained, in the form of a semi-

vowel, either pure or with Guna: the latter form I recognise
in s»s—p.wC_ ydtava (Vend. S. p. 120; in Olshausen, p. 7),
which can only be the plural accusative of >pwC_ ydtu,
for it stands with asgas agha, “peccata; and in the same
page in Olshausen occurs a derivative of ydtu in the accu-
sative singular, viz. Fer yeprew,C_ ydtuméntém, “the magi-
cian,” “gifted with magic” (according to Anquetil, magicien).
I render, therefore, agha ydtava literally by “the sins of
sorcery” (Anquetil, “Ja magie trés mauvaise”); and in An-
quetil’s Vocabulary is (p. 467) Gg Sas,.C, yathvaim, the
regular plural genitive of our base ydiu, which means,
therefore, “of the sorceries”; while Anquetil faultily gives
it the meaning of the derivative (magiciens), and, according
to his custom, takes this oblique case for a nominative.
An example of a neuter plural form without Guna is at V.S.
p- 122, asgusew héndva “the Indies”; with hapta héndu, “the
seven Indies” (Ang. II. p. 270). It has the epithet us-astar-a
(“up-starred?”) in opposition to @sysew Fersqersasasrasy
daus-astarém héndum, “to the ill-starred(?) [G. Ed. p. 269.]
Indies.” An example, in which the suppressed termination in
a u base is replaced by lengthening the final vowel, is the very
frequently occurring p»bb véhd, “goods,” from »wbb véhu.
233. The interrogative base ki (cf. quis, quid), which in
Sanskrit forms only the singular nominative-accusative (neu-
ter) fam ki-m, but is elsewhere replaced by ka; whence, in
Zend, rousg ka-t, “what”: this base, the use of which is very
limited, forms in Zend the plural neuter »5935 ky-a*; and
* V.S.p. 341. srw wGse See ell asus POWPHAS 25995

asopeaw a0 kya atte vacha yoi hénti gdthdhva thris Gmrita (erro-
neously thris dmriita), “ What are the words which are thrice said in the
prayers (songs)?’”’ The masculine forms aé¢é and yéi can here, according
to Note at §.231., occasion tio difficulty. So also V.S. p. 85, 933 Aya
before
this form is the more important, since we still require

examples which can be relied upon, in which the i of the
base is not suppressed before the termination a (above,
gara for gairy-a), although it may with reason be conjec-
tured, that, in accordance with the abovementioned héndv-a
and ydtav-a, forms also like vairy-a or vairay-a, from vairi,
were in use. As in Gothic, neuter substantive and adjec-
tive bases in i are wanting, the numeral base ZHR/,
“three,” and the pronominal base J, “he,” are very im-
portant for the neuter cases under discussion, in which
they form thriy-a (thriya hunda, “three hundred”) and iy-a,
according to the principle of the Sanskrit monosyllabic
forms, of which the i sound has not passed into its simple
semi-vowel, but into iy; thus, in Sanskrit, feat bhiy-d, from
wt bhi.
234. The Sanskrit gives, in place of the Zend- European
neuter a, an ¥ i, perhaps as the weakening of a former a
[G. Ed. p. 270.] (§. 6.); the final vowel of the base is length-
ened, and between it and the case termination a euphonic n
is placed (§. 133.); hence evatfadand-n-i, anita vdri-n-i,* wufa
madhé-n-i.t The bases which terminate with a single con-
_ sonant—q_n and tr being excepted—prefix to it a nasal,
before the masculine Goaspas7 ratavé (Gorascoas7 25995 Kya ratavé,

‘‘ which are the lords” ?),
* According to a euphonic law (Gram. Crit. r. 84°.), an qn following
after tr and some other letters, is, under certain conditions, changed into
Wy 2. -
‘t In the Védas, the ni in a bases is frequently found suppressed ; ¢.g.
fava viswd, “omnia,” from viswa. In this way the Sanskrit is connected
with the Zend vispa, viépd-cha: but perhaps this coincidence is only exter-
nal ; for as the Sanskrit nowhere uses a neuter terminationa, faxgqviswa can-
not well be deduced from vispat+a, but can only be explained as an ab-
breviation of the d-ni, which likewise occurs in the Védas, as also US
purt, multa,” “magna,” is used for wef puriimi(Rosen’s Spec. pp. 9, 10).
and after s and n the preceding vowel is lengthened; hence

aaifa vachdn-si, arafa ndmadn-i. Into relation with this i
might be brought the neuter inflexion of gue (quai) and he-c
(haic) which stand in Latin very isolated; que is, however,
still tolerably distant from the Sanskrit @tfa kd-n-i, while it
is nearly identical with the neuter dual @ ké from kati
(§. 212.). Since, however, theantiquity of this dual termination
is supported by the Zend, the plural form kéni stands on the
other side isolated, and its age is thereby rendered doubtful ;
as, moreover, the Latin, in the verb also, has introduced a
termination originally dual into the plural*; - (G. Ed. p. 271.]
we cannot avoid recognising in the Latin plural gue a
remnant as true as possible of the Sanskrit dual & Aé.
235. We give here a general view] of the formation of
the plural nominative, and of the vocative, identical with
it and the neuter accusative:
SANSKRIT. ZEND- GREEK. LATIN. LITHUAN. GOTHIC.
m. vrikds, vehrkdonhé,t AvKot, lup’-%, __wilkai, vulfés.
m. #, té, Tol, ist’, tiet — thai.
n. ddnd-n-i, data, dea, dona, . +. daura.
f. jihwés, hizvdo, x@pa, terrae, rankos, gibés.
* The termination ¢is answers to wa thas, Greek rov from ros, not to
q tha or @ ta, Greek re. With respect to the otherwise remarkable
declension of gui, and of hic, which is akin to it, 1 would refer prelimi-
narily to my treatise “On the Influence of Pronouns in the formation of
Words” (by F. Diimmler), ‘p. 2.
+ See §. 229.
} This form belongs not to the base 7'A (=@ ta), whence, in the sin:
gular, éa-s, and nearly all the other cases; but to TA, whence, through
the influence of the i, tie has been developed (cf. p. 174, Note* and
§. 193.) ; and whence, in the dative dual and plural, tie-m, tie-ms. The
nominative plural is, however, without a case termination. The original
form TJZA corresponds to the Véda w tyu, mentioned in §.194.; while
the base Sf syu (a shya, see §. 55.) is fully declined in Lithuanian in the
form of SZJE, and in the plural nominative, likewise without inflexion,
18
252 EORMATION OF CASES.
SANSKRIT. ZEND. GREEK. LATIN. LITHUAN. GOTHIG.

ii
|
f. tds, tio, TH, is-tae, tes, thés.
m. patay-as, paity-6,* moot-es, _—host’-@8,t .... gastei-s.
f. pritay-as, dfrity-0,* mopTi-es, mess'~28,t Gwy-s, anstei-s.
n. véri-n-i, var’-a, LOpt-O, MAKinds | is -tatlliies« a
ey eres ky-a,t oeab oaew) i elie! a.
f. bhavishyanty-as, bishyainty-6,* ....- od pw e ee
m. sénav-as, pasv-6,* ix Ov-es, pecir-s, sinu-s, sunyu-s.
f. tanav-as, tanv-6,* TiTU-€S, SOCTU-S, +s a hand yu-s.
n. madhi-n-i, madhv-a, méOu-a, pec, see eee
(222d
5]
‘py
f. vadhw-as, wip Fale ey sie 00 LOSE ieee
m.f. gdv-as, geu-s,§ Bo(F)-es, . bov-68;+ swine Owls ae
is szie. From the pronominal declension the form ée (from ia) has found
its way into the declension of the adjective also: so that the base GERA,
“good,” forms several cases from GERIE; viz. dat. du. gerie-m for
gera-m, dat. pl. gerie-ms for gera-ms, and nom. pl. geri for gerai. This
geri appears to stand in most complete agreement with the Latin nomina-
tives of the corresponding declension (bonz, lup?); but the difference be-
tween the two languages is this, that the i of boni (for bono-i) belongs to
the termination, while ger? is void of termination, and stands for gerie
(analogous with ¢ée), but this latter for gerie-i (cf. yaunikkie-i.)
* See p. 163, Note f. ;
t See p. 1078. ee
a
a
t To this ky-a, from ki-a, corresponds surprisingly the Latin qui-a

(quianam, quiane), if, as I scarce doubt, it is a plural neuter, as quod isa
singular neuter (cf. Max. Schmidt “‘ De pron. Greco et Latino,” p, 34).
In the meaning “that,” guia is clearly shewn to be an accusative: the
meaning “ because ” is less apt for this case, and would be better expressed
by an instrumental or an ablative; but in the singular guod we must be
content to see the idea “‘ because” expressed by an accusative. On the
other hand, guo, among other meanings, signifies “whither,” a genuine
accusative signification in Sanskrit grammar. Without the support of
quod we might conjecture that an instrumental singular had been pre-
served in quia, after the analogy of asyyo5asd paity-a, for paiti.
§ We might expect gav-d, gavas-cha, * bovesque;’’ but we read EO
geus in the Vend. S. p. 253, L. 9, in combination with the pronominal _
nenters asco td, “illa,” sy yd, “que,” which, according to §. 231,
Note, cannot surprise us.
SANSKRIT. ZEND. ~ GREEK. LATIN. LITHUAN. GOTHIC.
niv-as, rhage 8 va(F)-es, eras

vdch-as, viach-é,* Ont-€¢, vocés,t Bey 36 Se
bharant-as, barént- 6,* pépovt-ec, ferent-@s,+ .... —fiyand-s.
‘Almdn-as, asman-é, Saiuov-es, sermon-és,t.... ahman-s.
niman-i, ndman-a, TaAav-a, nomin-d, .... namén-a.
bhrdtar-as, brdtar-é,* matép-ec, fratr-es,t .... Were ey
duhitar-as, dughdhar-é,* 6u-yatép-es, matr-es,+ dugter-és, ....
datdr-as, détar-6,* dorhp-es, dator-es,t .... eee
vachdis-i, vachanh-a§ éme(c)-a, oper-d, «1.2 eae
THE ACCUSATIVE.
[¢2¢°d
+9]
"py
236. The bases which end with a short vowel annex 4
nm in Sanskrit, and lengthen the final vowel of the base;
hence, Jara vrikdn, qty patin, TAA siniin, &e. We might
imagine this n to be related to the m of the singular ac-
cusative, as in the verb the termination wtfq dni (1st pers.
sing. imperative) has clearly proceeded from "tf# dmi. The
cognate dialects speak, however, in favour of Grimm’s acute
conjecture, that the Sanskrit n is, in the accusative plural
masculine, an abbreviation of ns,|| which has remained en-
tire in the Gothic—vulfa-ns, gasti-ns, sunu-ns,—but has been
divided in the other sister languages; since the Sanskrit,
according to §. 94., has given up the latter of the two con-
* See p. 163. Notetf

+ See Note+ in preceding page.
t The Gothic r bases annex in the plural a u, and can therefore be
contrasted no further with the cognate languages. BROTHAR becomes
BROTHRU, whence bréthryu-s, &c., according to the analogy of sunyu-s.
§ Or aswzpeuah vachenha. Thus we read Vend. S. p. 127, néménha,
which, I think, must be regarded as accusative of nimé (aaa namas,
“adoration”), and as governed by wee > 5 béréthra, “from him
who brings,” “ from him offering.”
|| The Old Prussian, too, exhibits in the ace. pl. ns, e.g. tava-ns, rarépas.
Respecting the Véda termination ar, from vis, see §.517. Remark.
sonants, and has lengthened, as it appears, in compensa-

tion for this, the final vowel of the base*; while the Greek
[G. Ed. p. 274.] AdKous has preserved the sibilant, but has
permitted the v to volatilize to v.¢ In fact, Avxo-vs has the
same relation to Auxovs that tumrovo: has to tumtover, from
[G. Ed. p. 275.] témrovrs.f For méoi-as, ixOd-as, we could
not, however, expect a ndat-ve, ix60-vs, as the Greek makes the
« and v bases in all parts similar to the bases which terminate
with a consonant, which, in Sanskrit, have as for a termi-
nation; hence wqq padas=7ddas: and even in the most
vigorous period of the language ns could not have attached itself
to a consonant preceding. This as for ns may be compared with
* Thus vrikdn for vrikans ; as, fasta vidwdns, whence the accusative
vidwdns-am, in the uninflected nominative faETy vidwdi,
(‘‘ sapiens’’).
+ As the v also passes into ¢ (riGcis for riOévs, Holic riya, pédass for
tuiar(r)s, peAavs), Hartung (1. ¢ p. 263) is correct in explaining in this
sense the « in Kolic accusative forms like vdyors, rois orparnyois, &c. As
regards, however, the feminine accusatives like peyadats, roixidats, reipats,
quoted by him, I believe that they have followed the analogy of the mas-
culines, from which they sufficiently distinguish their gender by the a
preceding the 1; we cannot, however, thence infer, that also the first and
specially feminine declension had originally accusatives in ys, as neither
has the Gothic in the corresponding declension an ns, nor does the San-
skrit exhibit an n (see §.287., and cf. Rask in Vater’s Tables of Compa-
rison, p. 62).
t It cannot be said that rémrovor proceeded from rimrovrox, a truly
monstrous form, which never existed in Greek, while the rimrovre before
us answers to all the requirements of Greek Grammar, as to that of the
whole base, since o-yr: corresponds to the Sansk anti, Zend énti, Goth. nt’;
and from the singular r: (Dor.), in the plural nothing else than vr: can be
expected, But to arrive at ovox from ovre it is not requisite to invent
first so strange a form as ovror; for that ovrs can become ovox is proved
by the circumstance that the latter has actually arisen from it, by the
very usual transition of T into =, and the not rare vocalization of the
N to Y, as also in Sanskrit, in all probability, wa us has arisen from nit
(cf. p. 172, Note *), of which more hereafter. Sut if in the dative plural,
indeed, ov-cx has arisen from ovr-os, not from opy-cx (A€over not dalpovor),
we
ACCUSATIVE PLURAL. 255
the Ionic ata, ato, for vra:, vto, a form which has extended
from the places where the vocalization of the v was necessary,
to those also where v might be added (weme/Oara:, tetpa-
gator; then, also, menavatat, kexAratat, &c. for mémavvTai,
xéxAwvtat). This comparison with the 3d person plural ap-
pears to me the more in point, as, in my opinion, the n
in the presupposed forms, like gaa vrikans, afta patins,
AvKovs, has the same object that it has in the 3d person
plural; viz. allusion to plurality by extending (nasalizing)
the syllable preceding the sign of personality. The in-
troduction of a nasal is an admixture which is least of all
foreign, and comes nearest to the mere lengthening of an
already existing vowel.
237. Feminine bases with a final vowel follow in San-
skrit the analogy of consonantal bases; but with the sup-
pression of the a,* thus s for as or ns; they may perhaps,
too, never have had 2s, for else hence would have arisen,
as in the masculine, a simple n: to the (G. Ed. p. 276.]
we must remember that the abandonment of the n before case terminations

beginning with a consonant is a very old and therefore pre-Greek pheno-
menon, which is not to be accounted for in the Greek, and wherefore no
compensation is to be required for the », which has been dropped. But
even if it were so, we must still be satisfied, if the demand for compen-
sation for a lost y remains unfulfilled in several places of grammar; for
there are two kinds of euphonic alteration in all languages: the one,
which has acquired the force ofa general law, makes its appearance under
a similar form on each similar occasion, while the other only irregularly
and occasionally shews itself.
* Monosyllabic bases only have preserved the a as the case sign in
the singular nominative (§.137.); hence, feqaa striy-as, “feminas,’’
Wa bhuvas, “terras,” from Ral st7i, * bhé. There is scarce a doubt
that this form originally extended to polysyllabic bases also ; for besides
the Greek, the Zend also partly evinces this (§.238.), as also the ciroum-
stance that in the actual condition of the Sanskrit language the accu-
sative plural shews, in general, an inclination to weaken itself, and thus
contrast itself more submissively with the imperious nominative (§. 129.).
feminine gender, too, the well-sounding Ionic a is more suit-
able than n. In general, the Sanskrit feminines in other parts
of grammar cast off the n, which is annexed by masculines
and neuters (§. 133.). Moreover, the Gothic also, in feminine
6 bases, gives no ns, but it appears that thés= ara tds (eas,
has) is a pure dowry from the ancestral house; and when the
feminine i and wu bases in Gothic, by forms like i-ns, u-ns,
assimilate themselves to the masculines, this may be regarded
as a disguise of gender, or a deviation caused by the example
of the masculines. The consonant bases follow the ex-
ample of the Indian, but have lost the a, as in the nomi-
native (§. 227.); hence, fiyand-s, ahman-s, for Siyand-as,
ahman-as.
238. Feminines with a short final vowel lengthen it, to
compensate, as it appears, for the suppression of the a;
thus wiata priti-s is formed from prity-as, and tani-s
from ents The Greek seeding Sut ote re-
spect, only a casual coincidence, through forms in ‘fs, ds,
which, however, are not restricted to the feminine, and
stand at the same time, in the nominative, for t-es, v-e¢.
The Zend, like the Greek, follows in its i and u bases the
analogy of the consonantal terminations; hence, Ly sosasd
paity-é (paity-as-cha,) brs950 pasv-6 (pasv-as-cha, or, with
Guna, paitay-6, pasav-6. In feminine bases in i, u, occur at
times also the forms 7-s, d-s, corresponding to the Sanskrit;
as, 34/530 gairt-s, “montes” (Vendidad S. p. 313.), asy¢e%
éréztt-s,“rectas,” wo a50 tafnii-s, “urentes,” 03 90¢7¢0 pérétii-s,
“pontes.”
239. Masculine bases in » a, where they are not replaced
by the neuter (§. 231. Note), have, in the accusative, an (cf. §.61.);
28, wFsiman,* “hos,” often occurs, weys9G mazistan, “ maxi-
mos’ (Vend. 8. p.65.). The sibilant is retained before the
[G. Ed. p.277.] particle a cha, and these forms can be
copiously quoted; as, xs wxegs améshans-cha, “ non-
* Cf. Védic forms in dn.

conniventesque ”; ASI QIS 96 manthrans-cha, “ sermonesque™;

ASIII YSFII7OAS aésmans-cha, “lignaque”; aspsngsytossaul vds-
tryané-chu, “agricolasque."* The form ass yspp7oasGas athau-
run-ans-cha, “ presbyterosque” (V.S. p. 65.), is remarkable, as
there is no reason elsewhere to assume a theme athauruna;
and this form would accordingly shew that consonantal bases
also could assume the inflexion ns, with an unavoidable
auxiliary vowel however; unless, indeed, we are to suppose
that, in the perverted feeling of the language, it has been
introduced by the preponderating analogy of the a bases.
More important, therefore, than this MVP) sGas athaur-
unatis-cha are the accusatives «»¢/sy nareus, “ homines,”
and argos streus, “ stellas,” which occur very frequently;
while from Ascous dtar, “fire,” we have found, not wore7Gas
éthr-eus, but 4JGus dthr-d, in which it is to be remarked
that dtar distinguishes itself from other words in r in this
point also, that it forms, in the nominative singular, not
axes dia, but «07s. dtars. But how is the termination eus
to be explained? I believe in no other way but from »
ans, by changing the n into a vowel,as in [G. Ed. p. 278.]
Asyous; after which, according to §. 31., the 1s a has be-
come ¢ e: the sibilant, however, which, after » a and yu,
is #9$, must, after > u, appear as 29 s. We actually find, too,
in the V.S. p. 311, sey ner-ans in the sense of a dative:
* I formerly thought I could, through forms of this kind, quote the

introduction of a euphonic s in Zend, according to the analogy of §.95.
But if this introduction cannot be proved by cases, inwhichno ground
exists for the assumption of an original sibilant, preserved merely by the
particle asgs cha (cf. §§. 56>. 207. 228.), then the above examples are the
more important, in order to supply a fresh proof that vs is the original
designation of masculine plural accusatives of themes terminating with a
vowel. The superlative wGens gow Gekh véréthrazanstéma (of which
hereafter) may be regarded asderived from a participial nominative. Other
cases, which might suggest occasion to assume, in Zend, a euphonic s after
n, have been nowhere met with by me.
8
YLdooesas areas w9gsg sywey rors _sgsaug ddidi at nérans

mazdé ahurd ashaoné, &c. “da quidem hominibus, magne Ahure!
puris.”
240. As a in Sanskrit occurs the most often of all letters
as the termination of masculine bases, and we cannot mis-
take, in the history of our family of languages, the disposi-
tion in the sunken state of a language to introduce, by an
unorganic addition, the more inconvenient consonantal de-
clension into that of the vowels, I cannot therefore think
that it admits of any doubt, that the New Persian plural
termination dn, which is restricted to the designation of
animate creatures, is identical with the Sanskrit "Tq dn in
the masculine plura) accusative: thus, ..,\s, marddn, “ ho-
mines,” answers to HATq martydn, “ mortales,” “ homines.”*
241. If, then, the termination wy! dn, applied to animate
beings, belongs to a living being in the old language, the
inanimate neuter will be fitted to give us information re-
garding that New Persian plural termination which is
appended to the appellations of inanimate objects. A
suffix, in the formation of words which is peculiarly
the property of the neuter, is w@ as (§. 128.), which is still
more frequently used in Zend than in Sanskrit. In the
plural, these Zend neuters form anha or énha (§§. 56%. 235.);
and with this ha is evidently connected the lengthened \»
ha in New Persian; thus, \p}9) voz-hd, “ days,” answers to
the Zend aw usb? raochanha, “lights.” Many New Per-
sian words have been compared with New German words
[G. Ed. p.279.]. and often, too, correctly; but, except
through the medium: of the Sanskrit and Zend, it could not
have been conjectured that our “ Worter” is, in respect to its
termination, related to the New Persian hd. As, however,
the High German has, from its earliest period, repeatedly
changed s into r, and a into i (later e), I have no
* Thus in Spanish the whole ploral has the termination of the Latin
accusutive.
| doubt the ir—Middle and New High German er—which
makes its appearance in the plural in many Old High Ger-
man neuters, is identical with the Sanskrit neuter suffix
| Wa as; eg. husir, “houses,” chalpir, “calves” (cf. Grimm,
pp. 622 and 631).*
242. Here follows a general view of the accusative for-
mation: .
| SANSKRIT. ZEND. GREEK. LATIN. LITHUAN. GOTHIC
vrikd-n, véhrka-n, AvKo-vus, lupd-s, wilkii-s, vulfa-ns.
dénd-n-i, data, dapa, dona, esse aura.
jihwa-s, hizvd-o, xapa-s, terra-s, ranka-s, gibé-s,
td-s, ta-o, TEC, is-ta-s, ta-s, thd-s.
pati-n, paity-6,f TOC t-A6, host’-es, .... gasti-ns.
bhiy-as, afrity-6,t MOpTi-as, Mess'-e3, «ee Pe
priti-s, Gfrili-s, TOOTI-S, gee dwy-s, _ansti-ns.)
vdri-n-i, var -a, idpi-a, MaTi-dy «eos oe
Bee kya, 2.6 #0 ~~ ores tye
bhavishyanti-s, bishyainti-s,f .... sialan [G. Ed. p.280.]
stind-n, pasv-6,t ix Od-as, pecu-s, suni-s, sunu-ns.
bhuv-as, tanv-6,f TITU-AS, Seer sue Ries
tanti-s, tanti-s, nitv-¢, socrii-s, s-.- handu-ns.
madhi-n-i, madhv-a,t péOu-a, pecu-a, + .se eT
* This ir, however, is treated in declension as if the theme originally

terminated in a, and would thus, in Sanskrit, be asa. Hence, compared
with the dative hisiru-m (from hisira-m, ¢.168.), the nom. accus. hiisir
appears an abbreviation. Bu the relation of our ir to the Sanskrit as
is nut thereby disturbed, because in general, most of the original consonantal
terminations in High German have received unorganic vowel additions,
Cf. pp. 148 and 191, G. Ed. Note. More regarding this hereafter.
+ See p.175, G. Ed. Note. f.
t This form is further confirmed by 1s» pspbssed pés6-tanva, from
peso-tanu, which signifies the hind part of the body ({.199.), but is also
used in the sense of ‘‘ blow on the hinder part of the body”; and in this
- manner it occurs in the 15th Fargard of the Vend.: assasw ROA PSA
asmpsobured SP WIaa05 Gas asgelrsnawsSbass gynainhat (ainhat?)
s2 hacha
SANSKRIT. ZEND. GREEK. LATIN. LITHUAN, § GOTHIO
f. vadhd-s, ob W Sle wats Peer eit ty
m. f. gd-s,* gdu-s, Bo(F)-as, — bov-és, suas “
_f. —ndv-as. o TAME va(F)-as, » ales és ee eco
f. vdch-as, vach-6d,t on-as; voc-és, 6 SD ee ae
m. bharat-as,t —_barént-d,f pépovr-as, ferent-s, .... fiy
m. 4tman-as, asman-6,f daipov-as, sermon-és,.... ah
n. ndmdn-i, ndman-a, TaAav-a, nomin-a, .-.. nam
m. bhrdtri-n§ brdthr-eus? matép-as, fratr-es, .... ;
f. duhitri-s,§. |dughdhér-eus? @uyarép-as, matr-es, duyter-és,...
m. détri-n,§ ddthr-eus ? Dotnp-as, «dutdr-és, .1.. «s
n. vachdns-i, vachanh-a, éme(a)-a, ~—oper-d, ses wee
THE INSTRUMENTAL.
[G. Ed. p.281.] 243, The formation of this case, and what is
connected with it, has been already explained in §§. 215—224.;
it is therefore sufficient to give here a comparison of the forms.
which correspond to one another in the cognate languages,
hacha skyaéthnd-varéza atha buvainti pésb-tanva, “ hae pro facti-peractione

‘tum sunt verbera posteriori corpori inflicta” (Anquetil, Celui qui commet
cette action sera coupable du tanafour). Inregard to the andpéretha, men-
tioned at §.232., it is further to be noticed that the G th can only be
occasioned by a os w that has been dropped (§. 47.), for the theme of the
concluding substantive is rpered pérétu, not péréthu (Vend. S. pp. 313
and 362, twice).
* Irregularly-from a theme. mq g@ (§. 122.), for TAR gav-as. The
Zend ADPAVO gdus (also ASE O gdos), which often occurs, rests on the
strengthened Sanskrit form at gdu; so that in respect of the strong and
weak cases (§. 129.), the relation in this word is distorted. In the nomi-
native, for instance, we should expect AOA O gus, and in the accusative
MOPED Jeus, rather than vice versd.
+ See p. 163, Note f.
T See §.129.
§ See §.127. Note
and §. 249. NoteJ.
INSTRUMENTAL PLURAL. 261
by which a summary view of the subject may be assisted.

As the German, in its singular dative,* is identical with
the Sanskrit-Zend instrumental, it is hence deducible that
its character m (for b see §. 215.),in the dative plural,
must rather be regarded as an abbreviation of firq bhis
_ than as belonging to the dative-ablative termination wa
bhyas; although it approaches equally near to the two old
terminations.
SANSKRIT. ZEND. GREEK. LATIN. LITHUAN- GOTH. DAT.
m.vriké-bhis, .... Geé-giv, vorbis, ...- vulfa-m.

vrikd-is, veéhrkd-is, .... o.2.- wilka-is, -....
f. jihwé-bhis, hizvd-bis, .... » +.» ranko-mis, gibé-m.
priti-bhis, friti-bis, .... » «++ Gwi-mis, ansti-m.
m. séimu-bhis, pasu-bis, .... ... + sunu-mis, sunu-m.
f. nau-bhis, .... vav-Piy, «eee seer adi..g
m. dtma’-bhis, asma’-bis, .... HS 6 or) aba ae.
n. ndma’-bhis, nima-bis, .... Chews. nse (wn dil Sentence.
n. vaché-bhis,+ vaché-bis,t dyer-piv,t ..- + (G. Ed. p. 282.]
THE DATIVE, ABLATIVE,
244. Mention has already been made of the suffix of

these two cases in §. 215. Only the s of the Latin bus has
been left in the first, second, and (according to Nonius)
occasionally, also, in the fourth declension; for the é of
lupi-s, terré-s, spect-s (for speci-bus from specu-bus), must be
allotted to the base. Jwupi-s stands for lupo-bus, as evinced
by ambo-bus, duo-bus. From o-bus (by lightening the final
vowel of the base, o, u, from an original a, §. 6.), as occurs
in the beginning of compounds (multi-plex for multu-plex
or multo-plez, of which hereafter), the language arrived at
i-bus, (parvi-bus, amici-bus, dii-bus, cf. Hartung, p. 261). In
the first declension a-bus has been retained with tolerable
® Vide §.160. Note t

+ See §§. 56>. and 128.
frequency, but the middle step é-bus is wanting; yet the

language has scarcely made the spring from a-bus at once
to 7-s, but a-bus has weakened the a of the base to 7, which,
to compensate for the bu which has been dropped, has been
lengthened; thus terri-s from terri-bus, for terra-bus, as
[G. Ed. p. 288.] malo from mdvolo. Compare,
SANSKRIT. ZEND. LATIN. LITHUANIAN,
m. vriké-bhyas, vehrkaéi-byé, lupi-s. wilka-m(u)s.*
f. jihwd-bhyas, hizvd-byé, terri-8, ranko-m(u)s.
m. pati-bhyas, paiti-byd, hosti-bus, «+ . -F
f. priti-bhyas, afriti-byé, messi-bus, awi-m(u)s.
m. bhavishyantt-bhyas,
bi shyainti-byd, .... wad
m. séinu-bhyas, pasu-byé, pecu-bus,t sunu-m(u)s,
f.. vag-bhyas, vach-e-by6, _ voc-i-bus. www
m. bharad-bhyas, barén-by4,§ —_ferent-i-bus, ....
m. dtma’-bhyas, asma’-byé, sermon-i-bus, ....
m. bhrdtri-bhyas, _. brdtar-é-byé, fratr-i-bus, ....
THE GENITIVE.
245.
The genitive plural in Sanskrit, in substantives
and adjectives, has“the termination "mm dm, in the Zend
anm, according to §. 61. The Greek wy bears the same re-
lation to the original form of the termination that éd/dwy
does toweary adaddm (§§. 4. 10.). The Latin has, as usual 2
* See §. 215.
+ The masculine é bases pass in the plural, by an unorganic increment,
into a different declension. And in the dual and dative singular, also,
PATTI had to be given up (Mielcke, p. 35, Rem. 1.).
{ I have selected the masculine base PECU, which occurs only ina
few cases, on account of its connection with >39.30) pasu, and I have car-
ried it through all the cases, and think, therefore, that I may here also"
give the original u-bus for the corruption i-bus,
§ See §. 224. Note *, p. 241.
GENITIVE PLURAL. 263
preserved the labial final nasal in its original form, but

by its influence has shortened the preceding vowel; hence,
ped-um (=pad-dm), the u of which supplies the place of a
short a, as in lupum = yam vrikam, Avco-v.* [G. Ed. p, 284.]
The German, like the Lithuanian, has dropped the final nasal.
In Gothic, however, the = 4, which has been left, shews itself
under two forms, and thereby an unorganic difference has
been introduced between the feminine genitive termi-
nation and that of the masculine-neuter; since the fuller 6
has remained only to the feminine 6 and n bases.
246. Bases ending with a vowel, with the exception,
partly necessary and partly arbitrary, of monosyllables,
place, in Sanskrit, a euphonic n between the termination and
the base, the final vowel of which, if short, is lengthened.
This interposition appears to be pristine, since the Zend
partakes of it, although in a more limited degree; for
instance, in all bases in a a and w 4: hence, Gyiyrs97eeh
vehrka-n-anm, Gypsy, jihva-n-anm. To the latter cor-
respond very remarkably the genitives (which occur in
Old High German, Old Saxon, and Anglo-Saxon, in the
* Regarding the termination isewm in consonantal bases, and, vice versd,

respecting um in places where i-um might have been expected, we refer
the reader to §. 126. In adjectives the feminine character 7 mentioned in
§. 119. may have had its effect, and may have passed over from the femi-
nine to the other genders, according to the analogy of the Lithuanian
(p. 174. Note *§. 157.): thus the i offerenti-um reminds us of the Sanskrit
feminine wtatt bharanti. The same is the case with the é of the neuter
form ferenti-a; it is bequeathed by the deceased feminine theme FE-
RENTI. On the other hand, contrary to the opinion preferred in
§, 126., we must now regard the i before bus (e.g. voc-i-bus) as a conjune
tive vowel, like the ge in the Zend vdch-e-byé. Here it is to be observed
that those consonantal bases, which admit neither i-a nor i-um, must never-
theless proceed before bus to annex ani. In the chapter upon the adjec-
tives we shall recur to the feminine character é; and then treat also of the
é for ¢ in the singular ablative of the common dialect.
corresponding class of words) in 6-n-6, e-n-a; hence, Old

High German képé-n-6, Old Saxon g?bé-n-6, Anglo-Saxon
oygaipaeed
247. We find the bases in short and long i, in Zend, if
[G. Ed. p.285.] polysyllabic, only with euphonic n: on the
other hand the monosyllabic i bases annex the termination
direct, either attaching Guna to the final vowel, or keeping it
pure; thus,thry-anm or thray-anm, “trium,” from thri; vay-anm,
“avium,” from vi. Bases in > u admit both of the annexing
the termination direct and of the insertion of the euphonie n ;
but I find from the masculine »s9.s0 pasu only pasv-anm : on
the other hand, I have found from feminine bases like >sas
tanu, “body,” >s.34 nasu, “corpse” (cf. véxus according to
§. 21.), hitherto only u-n-aim. With Guna G yrvsasd
pasav-anm would serve as a prototype for the Gothie suniv-é
with Guna weakened (§. 27.).
248. Pronouns of the third person have, in Sanskrit,
ama sdm” for “mq dm; and this may be the original and
formerly universal form of the case-suffix, so that dm
would properly be only the termination of the termination,
and the s connected with the genitive singular would be
the chief person. If this is the case, the abbreviation of
this termination in substantives and adjectives must still
be recognised as very ancient; for the Gothic, which in the
plural nominative restricts itself so rigorously to the old
limits (§. 228.), gives to the sibilant, in the genitive also,
no wider scope; hence thi-zé (§. 86. 5.) =te-shdm (for té
sdm, according to §. 21.) “horum”; thi-zo = td-sdm, “ha-
rum.” Here the a, like the 6 of the base THA, THO,
appears weakened to i (§. 66.): on the other hand, the ad-
jective a and 6 bases, which follow the pronominal de-
clension, have ai-zé, ai-zé; and blindai-zé, “cecorum” (for
blinda-zé), answers exactly to the Sanskrit War ée-shdm
* Cf. Old Prussian son, e.g. in stei-son, rwv.’”

4 ’ GENITIVE PLURAL. 265
(from tai-sdm) from the base @ ta. The High German has
changed the old sibilant to r, as in many other places;
te" hence, in Old High German, dé-ré for thi-zé and thi-zé, of
which termination only the r has remained [G. Ed. p. 286.]
to us. To the Latin, in like manner, belongs rum for sum
(§. 22.); hence, istorum, istarum.*
249. We give here a general view of the formation of
the genitive:
SANSKRIT. ZEND- GREEK. LATIN. LITHUAN. GOTHIC.
vriki-n-dm, vehrka-n-anm, NoK-wv, lupd-rum, wilk’-i, vulf’é.

- n.té-shdm, taé-shanm, T-Ov, isté-rum, t-é, thi-zé.
jihwd-n-dm, hizva-n-aim, ywpa-wv, terra-rum, rank’-t, képé-n-6.f
* This rum, however, has, like the property of the plural nominative
(§. 228.), found its way or returned from the pronominal declension
into the entire second, first, and fifth declension, which is originally iden-
tical with the latter (§§.121 and 187.). The transplanting of the rum
termination into the declensions mentioned was the easier, as aH pronouns
in the genitive plural belong to the second and first declension. Forms,
however, remain, especially in the old languages, which evince that the
language was not always equally favourable to the bringing back the ter-
mination rum (deum, socium, amphorum, drachmum, agricolum, &c.).
On the other hand, the termination rum appears also to have attempted
to fix itselfinconsonantal bases, with eas conjunctive vowel, if, at least, the
forms furnished by Varro and Charis.—boverum, Joverum, lapiderum,
regerum, nucerum (Hartung, p. 255.)—are to be regarded as correct, and
do not perhaps stand for bovo-rum, &c. ; as also,in Zend, the base g6 may
extend itselfto gava. The Latin rum and Sanskrit Tq sdém lead us to
expect the Greck cov: this is not met with, however, even in the pro-
noun ; so that the Greek, in this respect, stands in the strongest opposition
to the Latin. ._ The forms in a-wy, ¢-wy (e.g. aitd-wv, aité-ay, ayopd-ar,
ayopé-av) point, however, to a consonant that has been dropped. It isa
question, therefore, whether universally a = (cf. §. 128.), or, as the San-
skrit and Zend lead us to expect, only in pronouns a , but in other words
of the first and second declension an N has been dropped, as in peifo
from yeifova. According to this, Avxey would be to be derived from
huko-v-wv, xopdev from yepa-y-er , but ray from rocwy rdev from racey.
+ Old High German, see
§.246. .
SANSKRIT, ZEND. GREEK. LATIN. LITHUAN. GOTHIC.
f. td-sdim, 4-onhanm,* —rd-wy, ista-rum, t-d, i

‘Sm. n.tray-d-ndm, thray-anm, Tpl-Ov, tri-um, tri-d,
rf priti-n-am, dfriti-n-anm, mnopti-wv, messi-um, awi-i,
~ ™m. sini-n-dm, pasv-anm, tyOi-wv, pecu-um, sunt, suni
8 f. tani-n-dm, tanu-n-anm, mTb-wv, socru-um, .... hand
“im. f. gav-dm, gav-anm, Bo(F)av, bov-um, «+.- cess
f. nd-vdm, orn va(F)-@v, .... 7 ame +.
f. vdch-dm, vach-anm, ém-Qy, voc-um, ys: «oa
m.n.bharat-dm, barént-anm,t epdvt-wy, ferenti-um, .... Jiyan
m, dtman-dm, asman-anm, Saipov-wv, sermon-um, akmen-t, ahman
m. bhrityi-n-dm, brathr-anm,t matép-wv, fratr-um, ... vere
* This word often occurs, and corresponds to the Sanskrit wrary d-sim
“harum,” “earum” (‘. 56°.) ; from uso td, tdonhanm would be expected,
which I am unable to quote. The compound (polysyllabic) pronominal
bases shorten the last syllable but one; hence, Fy orrs aé-tanhanm
not aétdonhanm, as might be expected from BATaTA etd-sdm,
+ Or, also, Fepnyrass barantanm, as in the Vendidad Sade, p. 131,
Fy yasbasss Saochantanm, *‘lucentium :? on the other hand, also
frequently saochentanm.
} This and the following genitives from bases inar are clearly moro g-nuine
and are more nearly allied therefore to the cognate European languages than
the corresponding ones in Sanskrit, which, in this case, has shortened ar to
‘q ri, and has then treated it according to the analogy of vowels. From ds
nar frequently occurs nar-aim, with retention of the a, on account of the
~ base being monosyllabic: on the other hand, dthr.anm from dtar, “fire,”
and Gg 7as.90 tisr-anm “ trium,” fem. for the Sanskrit _tisri-n-dm
(Gramm. Crit, r. 255,). From [email protected]>)>4 dughdhar, we find the form
dughdhér-anm (cf, p. 208, G. Ed. Note t+): the Codex has, however,
dugdér-anm (p.472, L.2.). In general, in this word the readings dughdhar
and dugdar are interchanged in various passages: the former, however,
is the more common.
THE LOCATIVE PLURAL. 267
LOCATIVE.
250, The character of the plural locative [G. Ed. p. 288.]
is, in Sanskrit, # su, which is subject to be changed into y
shu (. 21.), for which, in Zend, is found yyy shu (§. 52.);
while from @ su, according to §. 53., has been formed >» hu,
The more usual form for shu and hu (for which, also, occur
shi and hii) is, however, »»ry shva, a»w hva, which leads
to a Sanskrit =swa. This appears to me to be the original
form of the termination; for nothing is more common in
Sanskrit than that the syllables 4 wa and q ya should free
themselves from their vowel, and then change the semi-vowel
into a vowel, as 3m ukia is said for vakta (see also §. 42,).
The supposition, therefore, of the Indian abbreviation of the
termination is far more probable than that of a Zend
extension of it by a lately-added a, especially as in no
other case does a similar aftergrowth admit of being esta-
blished. But if = swa is the original form of the termi-
nation, it is then identical with the reflective-possessive
base @ swa, of which more hereafter.* The same relation
which, in Latin, si-bi has to su-bi (which might be conjec-
tured from su-i), or that fi-bi has to éu-bi, Sanskrit pay
tu-bhyam, the Greek dative-loeative termination o: (civ) has
to the Sanskrit @ su.t
* Therefore, in Zend, the locative as»y49.57G thrishva, * in tribus,” is

identical with ase s%e trishva, “ the third part,” since the pronoun in
the latter compound denotes the idea of part.
+ Regarding the termination w of the pronoun of the Ist and 2d
person see §. 222. From the Holic form dypéow, quoted by Hartung
(p. 260) from Apoll., I cannot infer that w is an abbreviation of cw:
if it were so, the vy also in jpiv would not adhere so firmly. It appears to
me more suitable, therefore, to accord to the common declension an in-
fluence upon the transformation of the form of inflexion peculiar to the
pronouns without gender, but of the highest antiquity ; an influence which
has penetrated further in odio: for odiv.
[G. Ed. p. 289] 251. The bases in @ a add to that vowel,

as in many other cases, ani; but from a+i is formed zg é
(§. 2.), to which the Greek o: corresponds; hence, AvKoi-o1 =
yarq vriké-shu. Hence the « in Greek has also passed over
to the bases in a-, y-, either preserving its full value or sub-
scribed, while in Sanskrit the ¥ a remains pure; hence, faata
jihwd-su, with which the locatives of names of towns best
agree, as HAataaou, ‘OAuuniaor, ‘A@jvyo: (Buttmann, M 116.
R. 7. and Hartung, p. 461.).*
252, Like the Gothic, the Lithuanian has an unorganic
difference between the terminations which mark the case
in the masculine and feminine in the genitive plural: the
first has the sound of se, and the latter of sa, with the
original and more powerful a, which, in the masculine, has
softened into e. The ending sa is plainly from the swa,
assumed above (p. 267, 1. 7.) to be the original form, from
which it is made by rejecting the semi-vowel.
253. Here follows a general view of the Sanskrit, Zend,
and Lithuanian plural locatives, with the Greek datives :—
SANSKRIT. ZEND. LITHUAN. GREEK.
m. vriké-shu, véhrkaé-shva, wilki-se, NbKor-or.
f. jihwd-su, hizvd-hva, ranko-sa, OdAvpniaot, yopai-ot.
Sf priti-shu; dfriti-shva,t Gwi-sa, — mopti-on.
i m. sénu-shu, pasu-shva, dangii-se, iy@v-o1.
rzm.f. go-shu, ++. cee Bov-ot.
& & nau-shu, eee eevee VaU-Cl.
S
pa)
* The common termination ois, ats (o1-s, ai-s), formed by curtailing ~

ot-ot, at-ot, and so brought into agreement of sound with the third declen-
tion, is here lost, through its apparent connection with the Sanskrit cur-
tailed instrumental ending 2¥ dis (§.219.), which had before required
consideration, because the Greek dative is also used as the instrumental.
+ I have no authority for the locative of the Zend bases m #, but it can
only be analogous to that of the bases in u, which can be referred to in
copious instances.
LOCATIVE PLURAL. 269
SANSERIT. ZEND. LITHUAN. GREEK.

f. vdk-shu, vdc-sva ? Fen.) ats Cok
m.n.bharat-su, brdtar-eshva? .... pépov-or
m. dtma-su, asma-hva,* wees «=Oatipo’-or
m. bhrdtri-shu, .... see.) © TaTpa-out
n. vachas-su, vaché-hva,t sees) Omeo-ct.
* Thus, in the Vend. Sade, p. 499, as»wasyyy> ushahva, from yasyty>

ushan, and p. 500, asus G04 ddmahva, from psGaws déman.
t The a in this form is not, as is generally supposed, a conjunctive
vowel, but rests on a transposition ; as @8paxoy for dapxov, and in Sanskyit
Feanta drakshydmi, “1 willsee,” for ceanfa darkshydmi (Sansk. Gramm.
§.34>.) : thus rarpacx (compare rérpact) for rarapo: (compare réocapor),
which, by preserving the original vowel, agrees with the Sanskrit base
pitar better than warépa, warépes, &e. The same applies to the dative
apvaot, since the theme of dpyds has, as appears from the cognate word
piv, apny, appny, rejected a vowel between the p and », which again appears
in the dative plural in the form ofan a, and removed from its place.
The whole REN appears to be a transposition of Wer, Sanskrit AT nar
(yr), “a man,” for apqv properly means “‘ male sheep.”’ The a of dpvact
is therefore etymologically identical with that of avdpaex (comp. Kiihner’s
complete Greek Grammar, §.281. Rem. 2.). Itismore difficult to give any
accurate account of the aof vider: it is either the older and stronger form
for the ¢ of viéo:, or this word must have had, besides its three themes
(YIO, ‘YI, ‘YIEY), a fourth, YIAT, from which came vider, as yovacr from
TONAT, the more prevailing co-theme of ONY, which latter agrees with
aryt Iniene
the Vendidad Sade, p. 499, we find the analogous plural locatives
amwbJig uairdhva, and ss»wbd.s0G esapéhva. Anquetil translates
the former by “‘ au lever du soleil,’’ and the latter by “ @ Ja nuit.’ Itisim-
possihle to pronounce these forms aught but derivatives from themes in
was as a 6, §. 56>.) Most of the cases of the latter word, which occurs
very frequently in various forms, spring from a theme in 7s ar, and the
interchange of AsdasasGs csapar with Yodasase esaps is a similar case
to that in Sanskrit, where Wea chan, “day,” forms some cases from
WEA chas (from which wet ahé in wea ahébhis, &c.) ; and together
: with
G. Ed. p.291.] “Remark.—From the bases 1n EX, to which

The dative eco. (= weg as-su) properly belongs, this form
appears to have imparted itself to other bases terminating
[G. Ed. p. 292.] differently, in which, for this case, an ex-
tension of the original theme by eg is to be adopted; which,
in its origin, is identical with the abovementioned (§. 241.)
plural increase to the base by ir (from és and this, from as), in
Old High German forms, as hésir, “houses,” chalpir “calves,”
which are the plural themes, with which the nominative, accu-
sative, and vocative are identical, and from which, in the
dative, by the addition of the ending for that case, arises hu-
sirum, chalpirum; as, in Greek, Ktveo-o1, vexveoo!, mavTecor,
yuvaikecot, ToAiecot, and others, from the unorganically in-
creased themes KYNEZ, NEKYE3, &c., according to the ana-
logy of ENES. From the doubled = one may then be re-
jected (avaxreow, roAtect, ujveot), or the doubling of a & by it-
self be employed ; as, for example, véxv-cor, for véxu-or. This,
with the theme Wza_exists another, Bet ahar. The anomaly of the
Sanskrit ‘day’ appears, in Zend, to have passed completely over to
‘‘night,”’ as this latter word has also a theme in n, namely p9oarsnGs
esapan, of which the genitive pl. Gyros csafnanm—analogous with
META ahndm, “dierum” (§. 40. relative to ffor d p)—is found in con-
nection with the feminine numeral GyV2usseo tisranm, “ trium” (Vend.S.
p. 246); for we read, I. c. §. 163., asnanmcha (= ahndncha),
csafananmcha (read csafnanmcha), “of days and nights.” In Sanskrit,
by the suffix Wa, the form WE ahna, derivative, but equal in its meaning,
has arisen out of Bey ahan, which, however, occurs only in compounds
(as Wag ptrvéhna, “the early part of the day’’), and in the adverbial
dative wera ahndya, ‘soon,’’ “ immediately,’”’ which, therefore, it is not
necessary to deduce from the root z hnu, with the a privative. The
Zend, however, whose night-nomenclature, in this respect also, is not out-
e
stripped by the Sanskrit, produces, as it appears, by a similar mutation,
a5Vrsa0ses csafna from frsdasane csapan ; whence we find the locative
wefrase esafné, which might also be taken for the dative of {SDIAIGS
csapan,
LOCATIVE PLURAL. 271
in most important particulars, isadopted by Thiersch, §. 128.,

for the developement of the forms in ecor; only that he with-
draws from the neuter bases described in §. 128., as BEAEZ,
the = which belongs to them, and, by a supposition, proved
to be erroneous, BEAE is made the theme: and he divides
forms like oyecg: into dye-cdi instead of cyeo-gi, and, by
assimilation, derives dye-cor from dye-ogr; while, as I be-
lieve I have proved, the forms oyeo-ge and dyeco: rest on
entirely different case-suffixes (§. 218.), and have only the
base *OXES in common with one another. An assimila-
tion, however, may be remarked in -youvac-or, from -youvat-c1,
so that the first letter has assimilated itselfto the second, not
the reverse. In dézac-c: we shall leave it undecided whether
the first = be primitive, and AEDA> the theme (comp.
yeas, §. 128.), or whether it has arisen out of 7, and so
AENAT with TEPAT, KEPAT, belong to one class. If,
esapan, but that it is preceded (V.S. p. 163.) by the unequivocal adjective

locative (OG sy naémé (from IG DAY naéma, * half’’). Compare, also,
1. c. §. 149., where popdaanes 9IG.5 p95 ps905 25765 ithra, asné, ithra,
esafné, probably means “in this day,” “in this night,” with the locative
adverb 35G.s ithra, “here,” in the sense of a locative demonstrative.
To the theme asphasn csafna, the plural of the same sound ecsafna,
might also be assigned, which occurs I. c. §§. 330. 331., and in several
places elsewhere: 25 pases Uyys7e thray6 csafna, “three nights,’
asa IMIG esvas csafna, “six nights,” ashasas ISPS J
nava csafna, “nine nights,” ifhere csafna be not (as in §. 231. Note { it
was considered to be) rather to be taken for the plural of {0G csapan,
as neuter, since, as has been before observed, the Zend uses the gender of
the substantive with great laxity, especially in the plural. For the
frequently-occurring ablative s23a3G5 csapardt, however, we
cannot, assume another theme csapara, but we must, if the reasling be
correct, admit that feminine consonantal roots in the ablative adopt also
the broader ending, d¢ for at.
however, in all these forms, we allow only o: or ow to be the

case-suffix, and all that precedes it is referred to the true or un-
organic increase of the base, it can therewith not be denied that
not even to Homer himself, in forms like éreoo:, not to men-
tion unorganic forms like xivecou, did the entire eco: present
[G. Ed. p. 298.] itself as pertaining to that which marked
the case; for in the feeling of the speaker éxeco: could pre-
sent itself, during that period of the language, only as what
it ‘is, namely, as émeo-o1, while éxecos, éxect, plural éreca and
not ézeos, &e., were used in declension. But different from
what has been here adopted is the assumption of Hartung
(p. 260, ff.) and Kiihner (I. c. §. 255. R. 8.), in the most ma-
terial points following Greg. Cor. Hol. §. 35., relative to the
production of the Greek plural datives. Kiihner says (I. ¢.)
The character of the dative plural is e¢ (character of the
plural) and « or wv (character of the dative singular), there-
fore, eo:(v).” I, however, think eg not the character of num-
ber, but of the nominative plural, and connected with the
nominative singular through its 2: a union of the plural
nominative suffix with the singular dative is, to me, not to
be imagined. If it were so, how could neuter nouns, to
which eg in the nominative is quite foreign, arrive, in the
dative, at their identity of form with the natural sexes?
It further deserves to be remarked, that, in Prakrit, the
locative ending ¥ su frequently assumes an Anuswara, and
so adapts itself, by the form @ sun, for su, to the Greek,
ow, for ot.
254. After laying down the laws of the formation of a
single case, it may serve to facilitate the general survey if
examples are adduced of the most important classes of
words in their connected declension. We pass over here
from the Sanskrit, and go to the other languages in their
order, according as they have, in the particular cases,
most truly preserved their original form; and where one
or other of them has departed entirely from the original
FORMATION OF CASES, . 253
principle of formation, or by an unorganic increase to the

base has entered the province of another declension, we
there, in the place in question, exclude it from the com-
parison.
MASCULINE BASES IN a, GREEK IN 0, LATIN IN u, 0.

SINGULAR.
Nominative, Sanskrit vrika-s, Lithuanian wilka-s, Zend

vehrk-6, with cha, vehrkas-cha, Greek Adxo-s,
Latin lupu-s, Gothic vulf’-s.*
Accusative, Sanskrit, vrika-m, Lithua- [G. Ed. p. 294.]
' : nian wilka-2, Zend vehrké-m, Greek AvxKo-v,
Latin lupu-m, Gothic vulf’.
Instrumental, Sanskrit vriké-n-a, Zend vehrka, Gothic Dat.
vulfa, Lithuanian Instr. wi/ki.
Dative, Sanskrit vrikdya, Zend véhrkdi, Lithuanian
wilkui.
Ablative, Sanskrit vrikd-t, Zend vehrkd-t, Latin lup-o(d)
(see §. 181.).
Genitive, Sanskrit vrika-sya, Greek Avxo-(c):ot, Zend
vehrka-hé, Gothic vulfi-s, Lithuanian wilko.
* The meaning is, in all these languages, the same, and so is the theme
in its first origin. The connection of the Lithuan. wilkas with rrikas
rests on the very usual interchange of the semi-vowels r and 7; and this
latter goes through the whole of the European sister languages. The
Gothic vulfs shews, moreover, the equally common interchange of gut-
turals and labials, and follows the rule for the alteration of letters (Asp.
for Tennis, see §. 87-). In Latin the same thing takes place with regard
to the supply of the guttural by the corresponding labial; but Jupus is
further altered through the loss of the initial letter V, as is the Greek
AvKo-s: it may, however, be assumed, that this v is introduced into the
middle of the word in being vocalized into u. While therefore, in Li-
thuanian,in wilkas, J and & are united, they are, inGreek, separated by v.
+ M. Reimnitz, whose pamphlet, “The System of Greek Declension”
(Potsdam, 1831), had not been seen by me before I completed the preceding
Part of this book, unfolds (1.c p. 122° passim) the same views concerning
T the
Locative, Sanskrit vriké (from vrika+i), Zend véhrké

(G. Ed. p. 295.] (maidhydi, §. 196.), Lithuanian wilké, Greek
Dat. AvK@ (oikor §. 195.) Latin Gen. lup’-i.
Vocative, Sanskrit vrika, Zend vehrka, Lithuanian wilké,
Greek Ave, Latin lupe, Gothic vulf”.
DUAL.
Nom.Acc.Voe. Sanskrit vrikdu, Védic vrikd, Zend vehrka,

Lith. Nom. wilku, Voc. wilku, Greek AvKcw.
Instr. Dat.Abl. Sanskrit vrikd-bhydm, Zend vehrkaéi-bya, Greek
Dat. Gen. Avxo-v, Lithuanian Dat. ‘wilka-m
(see §. 215.).
Gen. Loe. Sansk. vrikay-ds, Zend véhrkay-6 (see Rem. 1. )
Lithuanian wilki.
PLURAL.
Nora. Voce. Sanskrit vrikds, Gothic vulfés.*

Accusative, Sanskrit vrikd-n, Zend véhrka-n, Goth. vulfa-ns,
Greek Avxo-vg (from AvdKo-vs, §. 236.), Lithu-
anian wilkis, Latin lupd-s. |
the Greek ovo and its connection with the Sanskrit a-sya which I have, with-
out being aware of his concurrence, brought forward in §. 189. I have,
however, in this respect, already stated my views in my pamphlet ‘ On
the Demonstrative and the Origin of Case” (in the Transactions of the
Hist. Phil. Class of the Academy of Science of Berlin for the year 1826,
p- 100. Here I have only further to observe, that the Greek adj dnydovos,
from the root AHMO, is, in the suffix by which it is formed, probably con-
nected with the genitive ending in the text; and is therefore remarkable
with reference to the preservation of the s, which is lost in djp010. With
regard to the origin of dyudotos from the genitive, let reference be made
to the Latin cujus, a, um; and the identity of the Sanskrit suffix of words
like aaa manushija, “man,” as a derivative from Manu, with the geni-
tive ending w shya for By sya, as in WAU amu-shya, “illius.”
* With reference to the Zend, see §. 231. Note{; and with regard to
the Greek, Latin, and Lithuanian forms Avxo, Jupi, wilkai, see §. 228.
Instrumental, Sanskrit vrikd-is* (from vrikd-bhis), Véda

vriké-bhis, Zend vehrkd-is, Lithuanian wilka-is.
Prakrit déve-hin (from déva, “God,” see
§. 220.), Greek 6e6-fiv,t Gothic Dat. Instr.
wulfa-m (§. 215).
Dat. Abl. Sanskrit vriké-bhyas, Zend [G. Ed. p. 296.]
Sere vehrkaéi-byd, Latin lupi-s (amici-bus §. 244.),
Lithuanian wilka-m(u)s (§. 215.).
Genitive, Sanskrit vrikd-n-dm, Zend véhrka-n-anm, Greek
Av«’-wv, Lithuanian nilk’-%, Gothic mulf’-é,
Latin lupé-rum (8. 248.),
* I take the liberty, in order

to separate the base and the termination, to
divide the diphthongs, as above in Avxo-vs ; therefore one must here pro-
nounce vrikdis, and in Lithuanian wi/kais, not as trisyllables, but as
dissyllables.
+ I have remarked at *. 217., but only as a conjecture, that the ending
gw in the plural is perhaps identical with the Sanskrit firq Dis, and the
thence-derived Prakrit f% Ain, and the Latin bis in nobis, vobis ; and
I will not advance more than aconjecture here, also, incomparing 66 gw
with dévé-hin. This only is certain, that with the syllable fiq bhi, which
inSanskrit,
lies atthe bottom of the case-forms firg dhis, wR bhyam,
and waTq_ bhydm, as their common root (see §. 215. passim), the Greek gu
and ¢y is also to be associated. I here willingly agree with M. Ag. Be-
nary (Berl. Ann. July 1833, p. 51.), that gw might be formed from the
ending a bhyam (§. 222.) by the contraction of gqya into i (as in jyiy, épiv,
teiv, &c. §. 222.). The third possible supposition would be the derivation
from the usual dative-ablative plural termination wrq_ bhyas ; again with
the corruption of s to vy,as in the Ist person plural
yey from pes, and in
the 2d and 3d person roy, rov from we thas, Te tas. The fourth possible
case would be the derivation from the dual termination aT bhyém
(§. 215.), and the changing this number of restricted plurality to that of
unlimited plurality. I prefer, however, to consider gw (¢:) as from one
of the multifarious terminations of the Sanskrit plural belonging to all
declensions ; therefore, from fe bhis i ae bhyas.
Ms
Locative, Sanskrit vriké-shu, Zend véhrkaé-shva, Lithn-

anian wilkise, Greek Dat. Av«or-or.
NEUTER BASES IN d, GREEK o, LATIN u, 0.

SINGULAR.
Nom. Ace. Sanskrit déna-m, Zend ddté-m, Latin donu-m,

Greek dpo-v, Lithuanian géra, Gothic daur’.
Vocative, Sanskrit dina, Zend ddta, Gothic daur’.
The rest as the masculine.
DUAL,
Nom. Acc.Voe. Sanskrit dané (from déna+?), Zend dd#é.

[G. Ed. p. 297.] PLURAL.
Nom. Acc. Voce. Sanskrit dand-n-i, Védic dand, Zend ddta, Lae
tin dona, Greek dGpa, Gothic daura.
“Remark 1—The Zend system of declension has re-

ceived some valuable additions from the treatises pub-
blished by Burnouf since the appearance of the First Part
of this book, which I must lay before my readers.* First
a dual case, viz. the genitive-locative, which I imagined
to be lost in the Zend, as I had searched for it alone in
vain, and could supply all the other dual endings in tole-
rable copiousness. M. Burnouf supplies this (Yasna, Notes
et éclaircissements, p. cxxu.) by the expressions bysho
bow gw ubéyé anhvé which are to be twice found in V. S$. tea
a
p- 312, and on both occasions are rendered by Anquetil, whose
* First, a review of this Part in the Journal des Savans, which refers
particularly to the Zend; then the First Part of the First Volume of a
Commentary on the Yagna; lastly, a disquisition in the Nouveau Journal
Asiatique, “ Sur les mots Zends et Sanscrits Vahista et Vasichta, et sur
quelques superlatifs en Zend.”
translation is in this place particularly confused, “ dans ce

monde.” This translation might lead us astray so much
_ the more easily, that boway anhvé, according to &. 187.,
might also be the singular genitive, which frequently
occurs with a locative meaning. We await the elucida-
tion which Neriosingh’s Sanskrit translation will give of
this passage; but, for the present, content ourselves with
the inferences deduced by Burnouf. bssb_ ubdy6, ac-
cording to that authority, corresponds with the Sanskrit
saata_ubhayds (amborum, in ambobus), with 6 for a, probably,
according to Burnoufs acute conjecture, through the
influence of the preceding 5b, and with the loss of the con-
cluding s. I am the more inclined to assent to Burnouf’s
opinion regarding the origin of the first 6 of byyh13> ubdyd,
as I have been so fortunate as to find another example
for the hitherto missing dual case, in which Uyyas ayé, not
Lys dyé, actually occurs ; because, that is to say, no letter
exercising the force of assimilation in question precedes
the a—I mean the form Ly yasos.as zastayé (= Sanskrit
hastayds), “in the hands,” from ass. zasta, [G.Ed.p.298.]
in a passage of the Jzeschne, which has perhaps not yet been
examined by M. Burnouf (V. S. p. 354.): sauygyas swGasg
Lysasossasg Gyi93s FEU? 74 kathd ashdi drujém dyanm zaitayé,*
which Anquetil (p. 192) translates by “Comment moi pur,
mettrai-je le main sur le Daroudj?” It appears, how-
ever, that sawyyss ashdi can as little be a nominative as
Lyyspwasy zastayé a singular accusative; and I believe
lam not wrong in the following literal translation: “Howcan
I give the (Demon) Drudj into the hands of the pure (into
the power)?”
“Remark 2.—In the instrumental singular M. Burnouf
admits the termination ana in bases in a (Yacna, p. 98.
passim), with n introduced, for the sake of euphony,
* The Codex has faultily saya; asdi and Geers drvjem.
o16- - FORMATION OF CASES,
according to the analogy of the Sanskrit wa éna (§. 158.).

He rests this, among other forms, on that of 25J5G I 7OSG
maésmana, “urind,” a word which had often attracted my
attention, and from which [, in like manner, would have
deduced instrumentals in a-n-a if I had not differed from
Burnouf in the etymology of the same, as I make its
theme terminate in n; and this word, which I remember
to have seen only in the instrumental, I derive from the
Sanskrit root fag mih, “mingere,” by a suffix wa man,
according to the analogy of prsGarghss barésman, from ¥€
vrih, “to grow,” whose instrumental as4as¢s9¢/5 baresmana,
analogous with asyasGs~wxs1G maésmana, occurs very fre-
quently. M. Burnouf appears, on the other hand, to
adopt a suffix ma in the word maésmana, in which we
think we cannot agree with him as long as we cannot
supply any cases which must indubitably belong to a
theme in a. If, further, some words, which in their theme
terminate in s.s as (\, Sanskrit wa as), adopt ana in the
instrumental form—M. Burnouf quotes, p. 100 note, »syassg
mazana, asfas9 9799 Srayana, and ass vanhana; still, in
my opinion, bases in @ may be assigned as the origin of
these forms, and they can be divided maza-na, &c., only
in as far as such forms have been already proved to belong
to undoubted bases in a. But now we prefer dividing
them mazan-a, so that the letter s, with which these themes
originally terminate, is interchanged with a nasal, just as,
[G. Ed. p. 299.] in Sanskrit, the words wRA yakrit, Wea
sakrit change their t for n in the weak cases, and may sub-
stitute waa yakan, Waa sckan; or as, in more remote
auslogy, the Greek, in the first person plural, has formed pev
from nes (#@ mas, “mus”). Besides this, M. Burnouf cites
also the interrogative instrumental .syas9 kana, “with what?”
which is the only word that brings to my mind somewhat of
conviction, and had struck my attention before, in passages
like popasgy,C wyows 294099 kana vazna yazdné, “ with
FORMATION OF CASES, ses 279
what offering shall I sacrifice?” (V.S. p, 481.) I have not,

however, ventured to draw a grammatical deduction from
_ this form, because the pronominal bases are prone to
unite with one another, and because I believed I might
assume that the same pronoun which is contained in Wa
ana and va éna forms also the last element of asyasg_ Kana,
if from this base the instrumental only had been evolved
or preserved, as has also occurred in the Sanskrit
Wa ana and vq éna in but a few cases. For the rest, ~
the Greek xeivos also appears connected with this 28J299
kana, if it is looked upon as a theme, with which the in-
strumental must agree in sound, for xeives, if not directly
of interrogative meaning, is still plainly connected with
the old interrogative base (comp. @wa kaschana, “ who-
ever.”). Under these circumstances I eannot yet admit
of any instrumentals in a-n-a, especially as also the bases
in i and u (in which the Sanskrit in the masculine and neu-
ter likewise introduces a euphonic n) in the Zend, in words
which we have noticed, have dispensed with a similar insertion
(§. 160.). In another place (Journal des Savans), M. Bur-
nouf deduces the frequently-oecurring instrumental AWS JasyaVAs
ashayd, “ with purity,” from the masculine theme asyy,s
asha ; and there would be accordingly 9K ashaya, an
instrumental form, at present standing alone in the Zend,
which I hesitate to acknowledge, although it would be
analogous to the Védic form mentioned in § 18. waa
swapnayd, if one derives this, with the Indian grammarians,
from a theme GH swapna. But if instrumental forms of
this kind, in the Védas or in the Zend, are not to be pro-
duced in other undoubted instances as in the case of
adjectives in construction with masculine or neuter sub-
stantives, nothing prevents the assumption, that the form
wat swapnayd belongs to a feminine theme at swapnd,
especially as the suffix 4 na occurs also in other abstracts
in the feminine form 1 né, and therefore qwat swapnayd
may be explained according to the analogy of qanat trish-

[G. Ed. p. 300.] nayd, “with thirst.” In every case I think
I may deduce the Zend SYM ashaya from a feminine
theme asyyyas ashd, as the Zend in general, in the substantive,
passes readily from one sex to the other; and, for example,
with a masculine base ROSNY manthra, “a speech,” occurs,
also, a feminine ss Sra manthrd.
“Remark 3.—For the genitive termination wow hé there
also exists, as Burnouf has most satisfactorily proved, a
form nearer to’ the Sanskrit sya, viz. wyyw hyd, which,
although rather rare in comparison with the more
corrupt form hé, is still sufficiently frequent in some
chapters of the Jzeschne to satisfy one perfectly of its
signification, according to the proofs given by Burnouf.
I too had remarked words with the ending sawvyyw hyd,
but in passages where Anquetil’s translation was little
adapted to bring to light the genitive nature of the same,
which, besides, was very much obscured through its usual
representative ~ow hé, and was, moreover, concealed from
me under the appearance of an instrumental form.
However, the termination hyd—for which is sometimes
found, also, avyjyo khyd—approaches so very near to the
Sanskrit sya, and agrees with it so precisely according
to rule, as far as the unorganic lengthening of the a, that
a single passage, with the accurate translation of Nerio-
singh, who, in the passages hitherto edited, follows the
original word by word, would have led us to it. Such a
passage_is given, although with a different aim, by Bur-
nouf in his Yagna (Notes, p. cxxxix.), which we here annex,
as it is interesting in other respects, also, for grammar :—
AW J3IAS9 Uyy»Prbasod ASS MPLIAS AW(OAS Wor Sees AW59.059
GEpwrers roaug AV WE 099 ewe kasnd zanthwd pata
ashahyd paourvyé kasnd kheng strencha dat adhvdném. Ne-
riosiugh translates this passage word for word, only that
he renders kasnd, “ which man?” (here properly not more
than “who,” for the idea of man is lost in the general

signification of the whole,) not by @t at ké nd, but simply,
by @t 4é, as follows: @t wan: fam gee oat a: ata
amas fet aeata kd jananéh pité punyasya prathaman*
(fags aerate aH WA Kila sadvydpd- [G. Ed. p. 301.]
ratvan kas chakré, i.e. “ boni originem quis fecit?”) kak sir-
yasya tdrakdndncha dadéu padavim (fas aT wars al Zet
kila mdrgan téshin ké daddu,i.e. “ viam ipsis quis dedit?”).
We translate from the Zend,“ Quis (qualis vir) creatione pater
est puritatis (or puri) primus? quis (qualis vir) soli stellisque
dedit viam?” The Zend expression worse tzanthwéd, for
which, in the lithographed codex, p. 351, is erroneously
given aes zanthd, is plainly the instrumental of »>eyyax«
zantu; which would correspond to the theme of a Sanskrit
infinitive, war { jantum, as the latter is feminine, and to which
I have, in another place, referred the ablative posed rere
zanthwét (Gramm. Crit. p. 253.). This form is, besides, re-
markable on this account, viz. that it is identical with the
Sanskrit instrumental gerund, which, from #4 jan, without a
conjunctive vowel and without the euphonious suppression of
the 4 n, would sound 4watjantwd. With regard, however,
to the length of the concluding a of the Zend form, which is
preserved contrary to the prevailing rule (see §§. 118, 158.
and 160. p. 191 G. Ed., where, however, wows, janthwa is to
be read for zanthwa), I do not attach any particular import-
ance to that, because in the chapter from which this _pas-
.Sage is taken a, originally short, is repeatedly to be found
lengthened. The Sanskrit waa: jananéh, with which Nerio-
singh translates the Zend instrumental case, must be con-
sidered as an ablative, as this case often enters the depart-
ment of the instrumental, and is also capable of expressing
* Perhaps the adverb Wag prathaman, “primum,” is a corruption for
Wan: prathamak, “primus,” which answers to the original, and is to be
expected from the sense.
+ Vide asto.woh Gagezanthwd, p. 1244 G. ed.
282 - FORMATION OF UASES,
the preposition “through” (for example, Nal. XIL 89.).

Considered as a genitive, #a™: jananéh would not correspond
with sos rue zanthwd, which cannot possibly be a genitive,
for the genitive of >» zantu could only be APEDO WASS
zanteus, or, also, JosSras¢ zanthwé, or Jorasco sas zantavé (see
§. 187.), but in no case WS SrpsX6 zanthwd. Add to this, also,
that sata janani is feminine, like the Zend >was zantu, and
ya punyasya, therefore, could no more pass as the epithet
of wae jananéh than, in Zend, avyywayyas ashahyd could
pass as the epithet of sodas zanthwd. I will, however, as
concerns the Zend, lay no great stress on this circumstance,
since in it the genders of the substantive are constantly
changing. M. Burnouf, who looks upon w?a¥: jananéh as a
genitive, and refers ywey punyasya to it, according to this
interpretation justly takes objection to the qwrt punyasya,
which does not agree with the gender of wafa janani, but he
confirms, however, the reading expressly by the addition of a
(G. Ed. p.302.] sic. His translation runs, “Quel est le pre-
mier pére de la creation pure? qui a montré leur route au soleil
et aux astres.” I look with anxiety for M. Burnouf's further
explanation of this passage, but expect from him rather in-
formation of value in other respects, than to find that he has
succeeded in making the forms waa: jananéh and sso Orgse
zanthwd pass for genitives. Anquetil’s traditionary inter-
pretation sounds, in this place, very strange, but does not
contradict my apprehension of sod ras zanthwé: he makes
the genitive wyywxsyyas ashahyd pass for the nominative,
and does not, therefore, throw any light on the meaning of
the termination avsyyw hyd; for, in the presumption that it
was right, IBS SUILLIS ashahyd might, perhaps, have next
becn taken for an instrumental, and perhaps have been trans-
lated “father with purity.” His translation is as follows:
“Quel est le premier pére pur® qui a engendré? qui a donné
* In other places (V.S. p.885) Anquetil renders (p. 137) the words
AWE ID
de lui méme les astres qui ne sont pasa deux faces?” The
sun is here quite left out of the question; and it must be
acknowledged, that, as far as relates to etymology, it is
very much obscured in this passage; we might identify,
with reference to the form of eys¢y kheny, this expression
with the reflective pronoun aso kha (as in kha-ddta, “ created
of itself,” which is often said of the stars, as of self-
created lights), and consider it as the epithet of assyyse poss
Siren-cha; so that it would correspond as accusative plural
to the Sanskrit atq swan. It is here to be remarked, that
in some chapters of the Jzeschne, eys ng is repeatedly
found instead of a simple nasal, and, indeed, without
regard to the organ of the following initial letter. So we
read, in the V. S. p. 391, Cw< GrwSasnyrg dushacsathreng,*
ewes Sbasssypvanrs dusskyaéthneng, Owe sPassed4 dushda-
éneng. Anquetil, indeed, renders these expressions as
singular nominatives, “ce roi mechant, qui fait le mal, attaché
@ la mauvaise loi”; but they, together with [G. Ed. p.303.]
Le zugsamedrg dushvachanhé, berau vg nurs dushmananhé,
refer to the plural Loupe 74 drégvaté, and I have no
doubt of their accusative nature: the whole passage, how-
ever, like many others in the Jzeschne, can be explained
only with the help of Neriosingh’s Sanskrit translation.
We can but regret that the in other respects highly valuable
elaborate exactitude of Burnouf’s excellent Commentary
leaves us no hope that he will come very soon to the
elucidation of this and other passages, regarding which
Iam most curious. But to return to our ewes kheng,
AVISW IPAS AVOID pata ashahyd rightly by pére de la pureté: his

translation is, however, little calculated to throw light on the connection
of the passage referred to.
* The lithographed MS. has Oys9¢/ Gan Asu3>4 dusa csathreng as
two words; the a is, however, clearly only a conjunctive vowel, to unite
the prefix y49>g dush more conveniently with the following as@Ses.
the » kh makes no difficulty in this expression, even in its

acceptation for the sun, for which, commonly, gras hvaré
is found (the Sanskrit = swar, “‘ heaven,”), as wo kh is used
very frequently for »w hv (see §. 35.); but we might here
expect to find gArsyo kharé, and may suppose that the
ey ng has arisen out of n, and this letter out of r, as
these liquids are easily interchanged, as is shewn in San-
skrit, by the connection of weq ahan, “ day,” with WET
“ night,”
ahar, and, in the Zend, that of Psd csapan,
with Asdsaue csapar (I write it thus, and not ersdasases
csapareé, designedly, see §. 44.). At all events I take ewe
kheng to be the accusative, if, indeed, it may not also be
conjectured that the base 2s»w hvar may have entirely lost
its r, and that it may be yew kheng for Ge khem, the
accusative of a base asy kha. assyyse ous stren-cha, also,
according to my opinion, is the accusative, and not, as one
might expect from the Sanskrit translation, the genitive
plural, which more frequently occurs in the form Gyvwsqoss
stdranm. Although, from this, wera stren might easily
be formed by contraction and combination with ass cha, I
nevertheless prefer acknowledging in asuyyse oss strencha, a
secondary form of «s»¢%os Sstreus, explained in §. 239.;
so that the nasal, here vocalized to u, is there retained,
but the sibilant has been removed (comp. §. 239.); espe-
cially as, in other places also, »s4 dd is found in construc-
tion with the accusative of the person, which has been
given. In the Zend expression, ¢¢jw»@s adhvdném, the
Sanskrit weataa adhwéinam cannot fail to be observed
(comp. §. 45.); but in the lithographed MS. we have in-—
stead of this, G Eprurrgas advdném, which is easily seen to be
an error. This false reading appears, nevertheless, to be an
ancient one, and widely diffused; and upon this is founded |
Anquetil’s, or rather his Parsi teacher’s, interpretation, which
is strangely at variance with Neriosingh’s exposition; “ qui
[G. Ed. p.3(4.] ne sont pas a deux faces,” so that » a is
—
taken for the well-known privative particle, »»g dva as

the number two, and the last portion finds in the Sanskrit
Sraq dnana, “countenance,” its corresponding syllable.
FEMININE BASES IN 4, GOTHIC 6 (§. 118.).
Nominative, Sanskrit dhard,* Greek yopa, Lithuanian
; ranka, Zend hizva, Gothic giba, Latin terra.
Accusative, Sansk. dhard-m, Latin ferram, Zend hizva-nm,
Greek ywpa-y, Lith. ranka-n, Goth. giba.
Instrumental, Sanskrit dharay-d, Zend hizvay-a, Gothic Dat.
| Instr. gibai (§. 161.), Lithuanian ranka.
Dative, Sansk. dhardy-di, Zend hizvay-di, Lith. ranka-i.
Ablative, Zend hizvay-dt, Latin terra(d).
Genitive, Sanskrit dhardy-4s, Zend hizvay-do, Greek
YOpars, Latin terra-s, Lithuanian ranki-s,
Gothic gibé-s.
Locative, Sanskrit dhardy-dm (§. 202.), Zend hizvay-a,
Lithuanian ranko-ye (§. 197.).
Vocative. Sanskrit dharé, Zend hizvé (?), Greek yapa,
Latin terra, Lithuanian ranka, Gothic giba (?).
DUAL,
Nom. Acc. Voc. Sanskrit dharé, Zend hizvé (§. 213.), pee
Nom. ranki, Voc. ranki.
Instr. Dat. Ab]. Sanskrit dhard-bhydm, Zend Sar Fag Greek
ee Dat. Gen. yopa-iv, Lith. Dat. ranko-m (§. 215.).
Gen. Loc. Sanskrit dharay-és. [G. Ed. p. 305.]
* Means “earth,” and is probably connected with the Greek x@pa, as
‘aspirates are easily interchanged (Buttmann, §.16. Rem.1.). The root is
Y dhri (UT dhar, §.1.), “to hold,” “carry ;’’ whence, also, yT¢t dhard,
which, bytreason of the long vowel of its root, approaches nearer the
Greek x4pa (§. 4.), although it does not signify earth.
+ Without being able to quote this case in Zend bases in 4, Istill have
no doubt of the genuineness of the above form, since I can prove by other
cognate case terminations: 1, That the@ is not shortened; and 2. also
that an iis not introduced
into the theme by the assimilative
power of the
termination ; hence, e.g. in the instr. pl. »»4_s.vjco geéndbis (VY. S.
p- 308.) from aseo génd “woman ” (yurn).
ea)co FORMATION OF CASES,
FEMININE BASES IN 7.*

SINGULAR.
Nominative, Sanskrit priti-s, Zend dfriti-s, Greek mépri-s,

Latin furri-s, Lithuanian awi-s, Gothic anst’-s.
Accusative, Sanskrit priti-m, Latin turri-m, Zend dfriti-m,
Greek népzi-v, Lithuanian géwi-n, Gothic anst’.
Instrumental, Sanskrit prity-4, Zend dfrithy-a, Gothic Dat.
Instr. anstai (without case suffix, see §. 161.).
Dative, Sanskrit pritay-é (or prity-di, §. 164.), Zend
Afrite-é.t
Ablative, Zend dfritéi-t, Latin turri-(d).
Genitive, Sanskrit prité-s (or only with the feminine
termination prity-ds), Gothic anstai-s, Zend
dfritéi-s, Greek népri-os, pice-ws, Lat turri-s.
i
Locative, Sanskrit prit-du, (or with the feminine termi-
nation only prity-dm).
Vocative, Sanskrit prité, Zend dfriti, Greek néprt.
DUAL,
Nom. Acc.Voe. Sanskrit prit?, Zend dfriti(?), Lithuanian Nom,

[G. Ed. p. 806.] awi, Voc. dwi.
#* It may be sufficient to give here the cases of a Sanskrit masculine in

¥ i, which differ from the feminine paradigma : from agni, “ fire,” comes the
instrumental singular agni-n-d—whilst from pati, “ master,” comes paty-a,
and from sakhi, “‘friend,’’ sakhy-d (see §. 158.)—and in the accus. plural
wt agni-n.
+ Differing from what is stated in §. 164. p. 196, G. Ed., it is now my
opinion that the ge in wersddas Gfriteé does not represent the as a of
the original form possaspe sodasafritayé, but is the contraction of a and y;
as, for instance, in the Prakrit faafa chintémi, from feraratia chinta-
ydmi. ¢ e is here a weaker form of é=v, and is more properly used to
represent the latter than another vowel. With regard to the Lithuanian,
see p. 218, Note fT.
Instr. Dat. Abl. Sanskrit priti-bhydm, Zend 4friti-bya, Greek

Gen. Dat. zopri-o-iv, Lithuanian Dat. dGwi-m
. (§. 215.).
Gen. Loc. Sanskrit prity-és, Zend dfrithy-é (?) (see p. 276.
Rem. 1.).
PLURAL,
Nom. Voc. Sanskrit pritay-as, Zend dfrithy-é (with cha
“and” dfrithy-a3-cha), Greek mépti-es, Latin
turr’-és," Gothic anstei-s, Lithuanian dmy-s.
Accusative, Sanskrit priti-s, Zend dfriti-s, Greek mopzi-s,
. Gothic ansti-ns, Lithuanian dwy-s.
Instrumental, Sanskrit priti-bhis, Zend dfriti-bis, Lithuanian
awi-mis, Gothic Dat. Instr. ansti-m (§. 215.).
Dat. Abl. Sanskrit priti-bhyas, Zend dfriti-byé, Latin tur-
ri-bus, Lithuanian awi-m(u)s (§. 215.).
Genitive, Sanskrit priti-n-dm, Zend 4friti-n-anm, Latin
turri-um, Greek xopri-wy, Lithuanian awi-d,
Gothic anst’-é.
Locative, Sanskrit priti-shu, Zend 4friti-shva (or dfriti-
shu), Lithuanian dwi-sa, Greek Dat. aépz7i-c1.
NEUTER BASES IN 2.
SINGULAR,
Nom. Acc.Voe.Sanskrit vari, Zend vairi, Greek iSo:, Latin

mare.
The rest like the masculine.
DUAL.
Nom. Ace.Voc. Sanskrit véri-n-i.

PLURAL.
Nom. Ace.Voc.Sanskrit vdr/-n-i, Zend [G. Ed. p.307.]

vdr-a, Greek idpr-a, Latin mari-a, Gothic
thriy-a (from THRI, “ three”).
* Vide p. 1078 G. ed. as to turré-s and similar forms.

MASCULINE BASES IN wu.

SINGULAR.
Nominative, Sanskrit sénu-s, Gothic sunu-s, Lithuanian

suni-s, Zend pasu-s, Latin pecu-s, Greek
Borpu-s.
Accusative, Sanskrit sénu-m, Latin pecu-m, Zend pasd-m,
Greek érpu-v, Lithuanian sunu-n, Gothic
sunu.
Instrumental, Sanskrit stinu-n-d (Véda prabéhav-d, from pra-
bahu, §. 158.), Zend pasv-a, Gothic Dat. Instr.
sunau.
Dative, Sanskrit sdnav-é, Zend paiv-é, Lithuanian —
sunu-i.
Ablative, Zend pasaé-t, Latin pecu-(d).
Genitive, Sanskrit séné-s (from sunau-s), Gothic sunau-s,
Lithuanian sunai-s, Zend paseu-s or pasv-6
(from pasv-as), Latin pecti-s, Greek Bérpu-os.
Locative, Sanskrit séin’-du.
Vocative Sanskrit sind (from sunau), Gothic sunau,
Lithuanian sunat, Zend pasu, Greek Bérpv.
DUAL.
Nom. Ace. Voc. Sanskrit sind, Zend pasé, Lithuanian Nom.

sunt, Voc. sinu.
Instr. Dat. Abl.Sanskrit sdnu-bhydm, Zend pasu-bya, Greek
Botpt-o-1v, Lithuanian sunu-m (§. 215.)
Gen. Loc. Sanskrit sinv-ds, Zend pasv-6 (see p, 276.
[G. Ed. p. 308.] Rem. 1.)
PLURAL.
Nom. Voc. Sanskrit sénav-as, Greek Pédzpu-es, Zend

pasv-6 (with cha, pasvas-cha), Latin peci-s, —
Gothic sunyu-s (for suniu-s, from sunau-s, —
§. 230.), Lithuanian sinu-s. |
Instrumental, Sanskrit sdnu-bhis, Zend pasu-bis, Lithuanian —
sunu-mis, Gothic Dat. Instr. swnu-m (§.215.).
Genitive. Sanskrit sénu-n-dm Zend pasv-anm, Latin

pecu-um, Greek forpi-wv, Gothic suniv-é, Li-
thuanian sun’-i.
Lovative, Sanskrit szinu-shu, Zend pasu-shva (or pasu-
-shu), Lithuanian sunii-se, Greek Dat. Bérpu-s:.
Remark.—Feminine bases in z« in Sanskrit differ in
declension from the masculine, exactly as, p. 305 G. Ed,, wifa
priti f. differs from wftq agni in.
NEUTER BASES IN uw.
SINGULAR.
Nom. Acc.Voc.Sanskrit madhu, Zend madhu, Greek péév,

Latin pecu, Gothic faihu.
DUAL.
Nom. Acc. Voc. Sanskrit madhu-n-4..

PLURAL.
Nom. Acc.Voc. Sanskrit madhi-n-i, Zend madhv-a, Greek

géGu-a, Latin pecu-a.
FEMININE BASESINéZ [G. Ed. p.309.]

SINGULAR.
Sanskrit. Zend.
Nom. ndri, “woman,” bhi-s, “fear,” niiri, “ woman.”
Accus. néri-m, bhiy-am, ndirt-m.
Instr. ndry-d, bhiy-4, niiry-a.
Dat. — ndtry-di bhiy-é, or bhiy-di, néiry-di.
Abl. —_ndrry-ds, bhiy-as or bhiy-Gs, ndiry-dt.
Gen. ndry-és, _ bhiy-as or bhiy-ds, ndiry-do.
Loc. — ndry-dm. bhiy-i or bhiy-dm, ndiry-a.
Voc. _ndri, bhi-s, niiri.
sig,
DUAL,
Sanskrit. Zend.
N.A. V. ndry-du, bhiy-du, ndiré (see §. 213, p. 227.)
I.D. Ab. ndri-bhydm, bhé-bhydm, ndiri-bya.
Loc. ndry-és, bhiy-ds, ndiry-6 ?
PLURAL.
N.V. ndry-ds, bhiy-as, ndiry-do.
Accus. ndri-s, bhiy-as, ndirt-s.
Instr. nédri-bhis, bhi-bhis, niiri-bis.
D. Abl. ndri-bhyas, — bhé-bhyas, ndiri-byé.
Gen. néri-n-dm, —_bhiy-dim,* niiri-n-anm.
Loc. _ néri-shu, bhi-shu, ndiri-shva or -shu.
“ Remark.—By the side of the declension of monosyllabic
feminine bases in 7, which may reject the terminations
peculiar to the feminine alone, may be placed the Greek
[G. Ed. p. 3101 «Zs, and aremarkable similarity of inflexion
will be observed, as Nom. bhi-s, xi-s, Gen. bhiy-as, xi-d¢, Loc,
Dat. bhiy-i, xi-i, Ace. stré-m,+ xi-v, Voc. bhé-s, Ki-s. Plural: Nom.
bhiy-as, xi-es, Gen. bhiy-dm. xi-Gv, Loc. Dat. bhi-shu, xi-ot, Ace.
bhiy-as, xt-as, Voc. bhiy-as, xi-es. I consider, however, this
coincidence as accidental, but, nevertheless, an accidental coin-
cidence of that nature, that can only occur in languages
which were originally really one: and undoubtedly the
terminations, whose common sound appears so startling,
are historically connected. As far, however, as concerns
the theme, I believe, with Ktihner (§. 287.), that the 7 of xi was
not the original concluding radical letter of the word, but that a
consonant has fallen out after the « I would rather, however,
leave the question as to this consonant undecided, than assume
* Or bhi-n-dm. Further, the longer case-terminations, which belong

to the feminine (see §. 164.), are added at will to the monosyllabic femi-
nines in i, &; for example, together with dhiyé, bhruvé, also bhiydi,
bhruvdi.
+ Or, like the other monosyllabic words in 7, with the termination am,
striy-am.
that KIF is the true theme, and that the nominative was origi-
nally «Fs; for if xids, xi‘, in the form in which they have
_ been received, be analogous to Aids, Av, from ArFos, AiFi,
still, to establish a theme KIF, a proof must be brought
similar to that which really attaches to A:Fi from its being
found in inscriptions. And besides this, that which of itself is
alone sufficient proof, the cognate Sanskrit word f¢q div,
“heaven™ (§. 122.) likewise attests a digamma. All ground
for supposing a theme KIF is, however, wanting, for the long
« could, as in the Sanskrit
wt bhi, and like the long v in é¢pis,
be also the real final letter of the base, only that the long
@ in the Sanskrit, except in compounds (for example zraHt \
gata-bhi m.£., “ void of fear,” seat m.f., “water-drinking,”
see |
Gramm. Crit. §§. 169. 170.), concludes only the feminine themes.
We will therefore seek elucidation regarding the Greek xis
in another way, through the Sanskrit, and we find this, as it
appears to me, through a like masculine base, which approxi-
mates closely to the x«i-s, as well in form as in meaning;
namely, in atz kita, Nom. @iza kita-s, “insect” “ worm,”
which would lead us to expect in the Greek xitog, Acc. xitov,
to which «ic, xiv, bear the same relation as pé-yas, néyav, to the
to be presupposed péyaAos, uéyadov. I do not consider it re-
quisite to assume a theme METAT, although the Sanskrit
meq mahat, “ great,” might support it; but eq mahat isa
participial form, and its full and original form ([G.Ed.p.311.]
(8. 129.) is’ wea mahant, Nom. masc. agrq mahdn, which
would correspond to the Greek yeyav.”
FEMININE
BASES IN iu, U.
SINGULAR.
Sanskrit. Greek.
Nom. vadhi-s, “wife,” bhri-s, “eye-brow,” éppi-s.
Accus. vadhd-m, _—bhruv-am, dgpu-v.
Instr. vadhw-d, bhruv-d, er
Dat. vadhw-di, bhruv-é (or -di),
u2
SINGULAR,
Sanskrit. Greek,
Abl. _vadhw-ds, bhruv-as (or -ds), pe
Gen. -vadhw-as, bhruv-as (or -ds), - dppv-og
Loc. vadhw-dm, bhruv-i (or -dm), dppu-i.
Voc. vadhu, bhri-s, ppv.
DUAL.
N.Ac.V. vadhw- du, bhruv-du, dppt-e.

I.D. Ab. vadhii-bhydm. bhréi-bhydm. dppv-o-tv.
G.L. vadhw-ds, bhruv-és. otk
PLURAL
N.V. vadhw-as, bhruv-as, oppv-ec.

Accus. vadhii-s, bhruv-as, dppv-as.
Instr. vadhi-bhis, bhra-bhis, } aieia
D. Abl. vadhi-bhyas, bhra-bhyas, « ab
Gen. vadhi-n-dm, bhruv-dm (or bhré-n-dm), dpi-wr.
Loc. vadhd-shu, bhri-shu, _ bppd-or.
Remark.—The identity of 4 bhré and ‘O®PY* is

[G. Ed. p.312.] sufficient proof that the length of the v is
organic (comp. §. 121.), and it is not necessary, therefore, to
suppose a theme O®PYF (comp. Kuhner §. 289.) so as to
consider éppus as coming from é¢puFs, and the long v as a
compensation for the rejected F, as perhaps éAas from pédave.
That, however, F originally stood—for example, é¢pvFos—
before the terminations now commencing with a vowel, though
at a time when the language had not a Grecian form is
shewn by the Sanskrit bhruv-as ; by which, at the same time,
the shortening of the v in this case is justified, for the Sanskrit
* The o in d¢pus is based on the peculiar disposition of the Greek to

prefix a vowel to words which originally commenced with a consonant,
to which I have already drawn attention in another place, and by which,
among other things, the relation of dvvé, évoua, to aaa nakha-s, AVA
nama, is shewn.
changes, that is to say in polysyllables, as well v as 0, before

-yowel terminations, into a simple v; but in monosyllables,
- in order to avoid commencing with two consonants, or to
gain a polysyllabic form, the semi-vowel has its corre-
sponding short vowel placed before it, and thus is formed
3q uv (iv), as well from u as from %, as, under a similar
condition, ¢q from i and ¢: hence the two opposite forms,
for example, vadhw-as (not vadhuv-as), “women,” and
bhruv-as (not bhrw-as), “the eyebrows;” as above, bhiy-as
(not bhy-as), opposed to ndry-as (ndriy-as). In the dative
plural the short v of é¢pv-c: for é@pv-c1 may be attributed to
the effeminate habit of regularly shortening the v before vowel
terminations.”
BASES IN du (#i),”
SINGULAR. ;
Sanskrit. Greek. ,
Nominative, ndu-s, vau-s.
Accusative, ndv-am, vav-v.
Genitive, nav-as, va(F)-ds.
Locative, niv-i, va F)-i.
“Vocative, nidu-s, vau-s.
DUAL. [G. Ed. p. 313.|
Nom. Acc. Voc. ndv-du, va(F)-e.
Instr. Dat. Abl. ndu-bhydm, va(F)-o-tv.
PLURAL.
Nominative, ndv-as, va(F)-es.

Accusative, ndv-as, v@(F)-as.
Genitive, ndv-dm, val Fav.
Locative, ndu-shu, Dat. vav-ci.
Vocative, ndv-as, va(F)-es.
“Remark.—I find no sufficient grounds, with Kuhner,
1. c. §. 283.) to suppose that the base of the nominatives
* I give only the cases retained in the Greek.

in aus, evs, ous, originally terminated in F, so that in the
case before us it would be requisite to suppose a theme NAF:
for even if the vocalization of F to v, in order to facilitate the
junction with a consonant following, did not surprise us—
(forms like vaFs, vaFor, could never occur) ;—still, on the other
hand, the transition of the sound v into its corresponding
semi-vowel, in order to avoid the hiatus, is far more
regular, and is required in the Sanskrit according to the
common rules of euphony. We will not therefore differ
from the Indian grammarians, by the assumption of a
theme arq_ ndv for a ndu, and mq gav for mt gé (bos); al-
though, if there were adequate reasons for it, the practice
of the Indian grammarians would not restrain us from
laying down 74 gav and @1q ndv in the Sanskrit as the true
themes, which maintained themselves in this form only
before vowel terminations, but before consonants have
allowed the » to pass into a u, according to the analogy
of the anomalous fea div, “ heaven”; whence, for example,
the instrumental plural gfiva dyu-bhis for feafra div-bhis,
which would be phonetically impossible (Gramm. Crit.
§. 208.). The Latin navis cannot compel us to lay down a
theme ndv for the Sanskrit and Greek, for the Latin base
has extended itself by an unorganic i, as swan, “ dog,” length-
ened to cani; and therefore it exhibits in its declension
nowhere u, but universally ».
[G. Ed, p. 314.] BASES TERMINATING WITH A CONSONANT.
SINGULAR.
Sanskrit. Zend. Latin. Greek

Thema, VACH, VACH, VOC, ‘On.
Nom. vttk, vde-s, voc-s, on-¢.
Accus. vach-am, vach-ém voc-em, Ona.
Instr. vach-d, vach-a, ek AE
Dative, vdch-é, vich-é. eas he
-- ee
* See Locative.
FORMATION OF CASES.
SINGULAR.
_. Ablat. _ aise vach-at, voc-e(d), tania

Gen. vich-as, ~ vach-6,t voc-is, ém-s.
Loc. vich-i, vich-i, D. voc-i, D. én-i.
Voe. vak, vie-s ? VOC-S, on-¢
DUAL.
N. Ace. V. vich-du, — vdch-do, AF ose ae

or véch-4,t — vdch-a, ae On-€
I. D. Abl. vag-bhydm, .... as ee D. G. é2-o-iv
G. L. vach-és, vach-é ? “secs eae
PLURAL.
N. V. vach-as, vach-6,f voc-es, On-€6.

Accus. vach-as, vdch-6,t voc-es, On-as.
Instr. vdg-bhis, “ie
D. Abl. vdg-bhyas, .... voc-i-bus, ogre
Gen. vach-dm, vach-anm voc-um,
Loc. viak-shu, vie-shva ? D. én-ci.
“Remark 1.—I leave the terminationsin (G. Ed. p. 315.]

the Zend which commence with 5 unnoticed, since, contrary
to my former opinion (§. 224. Note *), I look on the
¢ & in forms like 33 5 pbs? raochebis, no longer as a con-
junctive vowel; and therefore no longer attribute the said
form to a theme gb? raoch, but assume that wg sequbas7
raochebis, and similar forms, have proceeded from bases in
46(from a3§.56°.); so that I look upon the¢e as a corruption
of the 6, and to the form Lasseubs? raochebyé I place as
anterior a lost form byishybas7 raochd-byd§ Ina similar way
# Like the Genitive.
+ With cha, “and,” véchas-cha.
Tt See p. 230, Note *.
§ M. Burnonf, who has induced me, by his excellent pamphlet, cited at
p- 276, on the Vahista (in the separate impression, p. 1¢, and following), to
rectify my former views, leaves, p.18 note, the question still unde-
cided, whether forms like «555 ESNE mazebis, ws seyasg manebis,
ASS STN
[G.Ed.p.316.] I find, in the Prakrit ( Urvasi, by Lenz,

p. 40.), WUE achharéhin for WaUvE achhardhin (Sanskrit apsa-
réblas); and if this form is genuine, then the ¢ e, in forms
like wg sews? raochebis, appears to stand for» 4, as generally
many interchanges between ¢ e and » é oecur, although in
the case before us the ¢ e is very constantly written, and
» é has not yet been pointed out in its place. If it is further
considered that we often find eC ye for bic y6, “which,”
¢9 ke for by ké, “ who?” and in the pronoun of the 2d
person in the plural also eb ve for Lb v6; and, finally, in
the pronoun of the Ist person ¢y ne for \y né; then we
see the change of the L 6 with ¢e is sufficiently ascer-
tained, although it appears to be restricted to the end of
words of a monosyllabic form; and in these the practice of
writing the 4 6 is the prevailing one, while before termi-
WSS eva vachebis, a3, i$¢ bs? raochebis,

have so arisen from the bases
1grsg mazd, &e., that the w 6 (was as) is suppressed, and ge then
introduced as conjunctive vowel; or whether, before the 6-(from as) only,
the s has been rejected, and the preceding a with an epenthetic i united
with an e. In the former case I should not have been entirely wrong,
from the analogy of raoch-e-bis, to deduce forms like vdch-e-bis. I con-
sider, however, the last view as the right one, only that I prefer letting
the 6 from the pre-supposed original form, mané-Lis, raoché-bis, be changed
in its whole force into ¢ e, rather than reduce it into its elements, and
mix the first of the said elements (a) with a conjoined i: for the deri-
vation of manebis from maniibis from manabis, for manasbis, would extend
to the Sanskrit form wathiry mandébhis, which originally may have been
manarbhis (manas-bhis was never possible), But I believe that in the
Zend the form ebis really preceded the form 6bis. M. Burnouf, in his
review in the Journal des Savans (in the separate impression, pp. 30, 31),
calls attention to a form }assedoaul vdghzhbyé, for which is once
found, in the Vend. Side, pp. 69 and 70, Jassedoep.nh vdghézhéebyo,
once Yossedoo.uh vighzhébyé, and once Yosseoggauh Kee
whic
nations beginning with 5 as yet no 6 has been pointed

out; so that 6 appears to be as repugnant to a preceding 6
as favourable to a following 4, if the conjecture of Burnouf,
mentioned at p. 297, G. Ed., is well-founded. On this point
I was not yet clearly informed, when, at §§. 224. and 242., [
inconsiderately imagined I could deduce vaché-bya, vaché-bis,
from dyasly vaché (from vacha). Instead of this should be
read sddseuabh vache-bya, ag sequal vache-bis ; and besides
this, in the locative singular, sess vachahi for se wyuh
vachanhi; since the nasal to be prefixed to the h, according
to §. 56°., falls away when the vowel which follows the h
is i, which has been already indicated in the paragraph
quoted, but since then fully proved by Bur- (G. Ed. p. 317.]
nouf. Besides, there really occurs, also, in one passage (where,
unfortunately, the lithographed MS. is faulty, and is therefore
which, with the conjunctive vowel 2 é (see §. 30.) introduced in different

ways, plainly represent one and the same word, and have proceeded from
}33;eoo.uh vighzhbyé, which itself never occurs. Although these
forms, which had struck me likewise, clearly belong to a theme which
means “discourse,” and is connected with our vdch, I would still rather
not, with Burnouf, derive it from véch ; so that the nominative of this,
saul) vdes, raised to a secondary theme, would be contained therein.
We dare not, without further authority, attribute to the Zend such a
malformation, although it derives its superlatives in as¢¢qo ¢éma from
the masculine nominative, instead of from the theme. But Anguetil, in
his Glossary, gives a form vakhsenghé, “parole utile,” which we ought
probably to read wuguarcioh vacsanhé (as dative), if not with long a
wuwBauh vdesanhé. This latter form would belong to a theme
Gsctial vdcsé (véesas); from which, in the dat. abl. pl., }assed9.wh
vaghzhbyé (vaghézhbyé, &c.) might proceed for Gor weBaubh vdeshyé ;
as with WEST GG mazebis, AWLST/GE manebis, occur also 435_s¢45G
mazbis, ADE _SWANG manbis ; for the 1 s of Lwowh vdes6é must, aa
Burnouf
has shewn, in contact with b become eb zh.
impossible for me to use) the locative sorb vachahi;

that is to say, in the Vend. S. p. 173, where, for asprwasy0G
asyrowasash manahéché vachahéchd, is to be read ssw.5G
awpsseassh manahiché vachahiché. Ina Grammar, the lost
acquaintance with which is again to be restored, oversights
of this kind will, I trust, be excused in the first labourers;
and if, for example, Rask gives to the word paiti the genitive
paitdéis, while, according to §. 180. p. 196, Note }, patdis is to
be written, still the form paitdis was, in its time, instructive
in the main, and first taught me that the Sanskrit genitive
termination é-s corresponds to the form dis in the Zend.
If, too, Rask has incorporated in his scheme of declensions
the ablative paitdit (for patéit), this was indeed a new error,
but also a new advantage for the Zend Grammar in its
then state, and brought to light a new and important fact,
which I believe I was the first to discover; namely this,
that bases in i form their ablative in éit, for which the
proofs in the Zend-Avesta, as much as I have of it, are
neither numerous nor easily found. I make this remark
because M. Burnouf, as it appears to me, speaks too unfa-
vourably of such theoretic formations. As far as I am
concerned, I believe I may assert that my communications
regarding Zend Grammar are founded on careful reflec-
tion. I could not, however, perfectly conclude my con-
siderations, and I am very ready to complete and adjust
them through those of M. Burnouf. For in this book
also, in regard to Zend Grammar, one must carefully
distinguish the disquisitions given in the text from the
general comparison added at the end of each rule regarding
case. In the former I give only those Zend forms which
I have seen, and I thence deduce theoretic laws: in the
latter I seek to make the deductions from the inquiries
pursued in the text evident in one select example. I am
perfectly sure of the prevailing majority of the forms
given in the tables, and can produce abundant examples
of them. I have marked some as questionable, and shewn
the limits of the probability of others, in notes; and if an
error has crept into the forms spoken of, and by me
believed to be correct, it will give me pleasure to be able
hereafter supplementarily to correct it. The form sequal
vachanhi was, however, only in a measure a theoretic forma-
tion; and I should not have ventured to ([G. Ed. p.318.]
exhibit it if I had not observed, in other words of the same
declension, i.e. in other bases terminating with a consonant,
the locative, which has entirely escaped Rask.
“Remark 2.—One might consider the o of ézotv instead
of a conjunctive vowel, as has been stated above (see
§. 221.), as a property of the base, i.e. as an unorganic
extension of it; or, in other words, regard it as a trans-
ition from the third to the second declension; a decla-
ration which must then naturally extend itself to the dual
termination ow of the whole third declension (aocio-sv, Bo-
- Tpvo-1v, Satpdvo-tv like AdKo-tv), and to all cases in the forma-
tion of words and arrangement of the same, where we have
represented an o foreign to the proper base as conjunctive
vowel. According to this, forms like weArrders, peArtomaAns,
guowwAoyia, Botpuders, Botpvddwpos, would be, under the pre-
supposition of the bases MEAITO, ®YSIO, BOTPYO, to be
divided into peArré-ers, and would lead us to expect the
nominatives peA:to-v, &c., which are not to be found. The
statement here given has this in its favour, that similar
cases occur also in cognate dialects, since in general that
declension which is the most in vogue and most used, is
prone, in certain cases, to receive into itself the other
declensions, which annex to their original base the final -
letters of the bases of the declension more in use. The
origin of ézoiv from “OO, of gepdvrov from SEPONTO,
was as it were the first commencement of the disease,
which came to its full developement in the Pali; since in
this language, which otherwise closely resembles the
Sanskrit, the bases which end with consonants are declined
in the old way only in the singular, but in the plural are
so corrupted, that, with the exception of the nominative
and the vocative of similar sound, and the genitive, which
at the same time supplies the place of the dative, they
have extended the old base by an unorganic a (=Greek o),
and have thus partly brought it from the Greek third
declension into the second; and in the singular, also,
most of the cases may, together with the old form, assume
more recent forms, which have originated in the manner
stated. In this manner, for example, the root ¥z char, “to
go,” forms its participle present partly from the original base
‘ata charant, or its corruption aq charat (see §. 129.), partly
from the augmented theme 4m charanta, and in part also
[G. Ed. p.319.] arbitrarily from ‘awa charant or ata
charanta, as follows (see Clough’s Pali Grammar, Colombo
1824, p. 25, and compare Burnouf’s and Lassen’s Essay,
p- 112 et seq.):
SINGULAR.
Th. CHARANT, CHARANTA, CHARAT.

Nom. charan,* charanté, nina
Ace. charant-am,t wail 5 idea
Instr. sis ipa charanté-n-a, charat-d,
Dat. like the Genitive,
charanta-smé,
Abl. veers ve sharcniesa’ charat-4.§
* The final q n is, as in the Prakrit ({. 10.), transmuted into the
Anuswara, which I here express, as in the Sanskrit, by n.
+ It might also be divided thus, charanta-m, and deduced from
charanta.
t Transposed, and with A for s (comp. §. 166.). These forms are
derived from the medial pronoun sma mentioned in §.166., which, in
the Pali also, has forced its way into the usual declension, The ¢, which
was to have been expected, is, as generally happens at the end of a word,
suppressed.
§ Charaté is, according to appearance, identical with the instrumental,
but
SINGULAR.
Th. CHARANT, CHARANTA, CHARAT,

Gen 5 oy Jie charanta-ssa, charat-é,
charanté,
Loe. jercharanta-omin charat-i,
or charanta-mhi,
charan,
Voe. jorchara” oS cu » hat tg
¥ or chard,
PLURAL. [G. Ed. p. 320.]
Pes het STN charantd,+ See

Ace. a charanlé, oie EX
eee ee
charantébhi,
as } ert
Dat. like the Genitive.
Abl. like the Instrumental.
Gen. Aap gray 2a - charat-am.
Loc. Fane e tes charanté-su, Se Pog
Voc. charanté, charanté, Pee
_ “Tf the Greek in its bases ending with a consonant had fol-
lowed the declension-confusing example of the Pali, one would
have expected, for instance, from ¢épwv a genitive pépovTou,
dative épovrw; and in the plural indeed, ¢epdv7wy from
but is, in reality, corrupted from charat-at, analogous with Zend forms
like ap-at (in §.180.): the suppressed ¢ is replaced by the lengthen-
ing of the preceding vowel, as in achard, “the went,” from achardt
(Clough, p. 106.).
* If this form really belongs to a theme in nt, as I believe, it has
sprung from the original form chara, by suppression of the concluding
nasal (comp. Burnouf and Lassen, p. 89); and in chard this deficiency is
replaced by lengthening the vowel.
+ According to the usual declension ending with a consonant one
would expect with charanté also ren, A the original theme
charant; as, for example, gunavanté is used with gunavantd, “ the vir-
tuons”; the former from gunavant, the latter from gunavanta.
®EPONT, but pepovror, pepovtous, bepovras, from PBEPONTO.

In this manner the form gepévrov in the dual, which has
been lost in Pali, would be clearly explained as derived from
®EPONTO ; but even when standing isolated, pepdvrow may
be justly referred to a theme ®EPONTO, as the first com-
mencement of a corruption which was further pursued in the
Pali; and I prefer this view of the matter now to that laid
down at §. 221. Both views, however, concur so far; and
thus much of my opinion may be looked on as proved,
that in q@epdvror, and all other dative-genitive forms of the
third declension, the o belongs neither to the original theme,
which lies at the root of all the other cases, nor to the
true case-suffix. |
[G. Ed. p. 821.] SINGUIAR.
Sanskrit. Zend. Latin. Greek. Gothic.
N. bharan, baran-s, feren-s, —pépar, Jiyand-s.*
Ac. bharant-am, barent-ém, ferent-em, épovt-a(v), fiyand.
Ins. bharat-é, barént-a, +. ie epee o Jfiyand.
' -D. bharat-é, barént-é, see Locat, see Loc. see Dat.
Ab. see Gen. _—_barant-at, ferent-e(d), .... init
G. bharat-as, barent-6,t ferent-is, epovt-os, fiyand-is.}
L. bharat-i, barént-i, D. ferent-i, D. pépovt-t, oo
V. bharan, baran-s, feren-s, pépuv, Siyand.
* Feind, “foe,” as “ hater,” see §. 125, p. 188.

+ See p. 210. Note §; with cha, barentas-cha (“ ferentisque’’).
t I imagined, p. 210, that I must, in this case, which before was not
proved to exist in VD bases, set down fiyand-s as a mutilation of,Jiyand-is
from fiyand-as, according to the analogy of other bases terminating with a
consonant /‘ahmin-s, brothr-s, §.191.); Grimm has (I. 1017.) conjectured
friyéndis or friyénds from JSriyonds. Since this, owing to the very valuable
additions made by Massmann to our Gothic authorities, the genitive
nasyandis of Nasyand (“ preserver, “ preserving”) has come to light (see
his Glossary, p. 158), by analogy with which I form fiyand-is,
nvaL.
Sanskrit. Zend. Greek.
N. Ac. Voc. bharant-du, barant-do, or baranta, pépovt-e.
} Védic, bharant-é,* ae dwar
1. D. Abl. bharad-bhyém, baran-bya,t pepovto-iv.t
Gen. Loc, bharat-és, barat-d? (p. 276, R. 1.) ----
PLURAL. (G. Ed. p. 322.]
N. V. bharant-as, barént-é,$ ferent-és, pépovt-es, fiyand-s.

Ace. bharat-as, barént-0,§ —ferent-és, —pépovt-as, fiyand-s.||
Instr. bharad-bhis, baran-bis§ ....- eae ae .
==
D.Ab.bharad-bhyas, baran-byé,{ ferent-i-bus, ...- ee
Gen. bharat-dm, _ barént-anm,t+ ferenti-um, epdvt-wv, fiyand-étt
Loc. bharat-su, due core épou-cr. [G. Ed.p.323.]
* See p. 230, Note *

+ Or barénbya. See p. 241 Note *, and p. 210. Note §.
t See p. 299. Rem. 2.
§ Barentas-cha, “ferentesque.” See p.210 Note §.
|| This form, which, owing to an oversight, is omitted in p. 260, is found at
Matth. 5. 44., and agrees with friydnds, “amicos” (“ amantes”), Matth. 5.
47.as generally with the declension of a root terminating with a con-
sonant. Comp. Grimm (J. 1017.).
{ See p.241 Note #, and p.210 Note.
** The Gothic dative, which I would have used also as the instrumental
(§. 243.), does not occur in roots ending in nd.
tt Or barant-anm. ee p. 266 Note t.
Tt This case certainly cannot be proved in bases in mJ; but may, how-
ever, be correctly deduced from the other bases ending with a consonant,
and from the elder sister dialects. See §. 245.
§§ I conjecture a transition into the a declension (comp. p. 299 Rem. 2.),
by suppressing
the nt; thus, perhaps, baraéshva (or -shu, or -shit, §.250.),
as Vend. S. p.354; g.usroasmog?g drégvatstt (read p yxy shit) for drégvat-
sit, from drégvat, in the strong
cases (§. 129.) drégvant ; on the supposition
that the reading
is correct, except
the false s. See
§.52.
SINGULAR.
Sanskrit. Zend. Latin. Greek. Gothic

N. dtmd’, asma’, sermo’, Saiuuy, ahma’.
Acc, dtmdn-am,asman-ém, sermon-em, Saiyov-a(v), ahman.
Inst dtman-4, asman-a, phot --++ DI. ahmin. (8. 132.
Dat. dtman-é, asmain-é, see Loc. see Loc. see Dative. —
Abl. see Gen. asman-at, sermon-ed), .... « rads
Gen. dtman-as, aiman-6,* sermon-is, Satuov-os, ahmin-s (8. 132.
Loc. dtman-i, asmain-i.D.sermon-i, Saipov-t, S553 |
Voc. diman, asman, sermo’, datuor, ahma’.
DUAL.

N. Acc. Voc. dtmén-du, asman-do, or aiman-a, Saipov-e.
Véda, dtmdn-a,
Instr. D. Ab. dtma’-bhyam, asma’-bya, D. G. Sdatpdvo-tv.t
Gen. Loc. dtman-és, asman-6? (p. 276, R. 1.), ....
PLURAL.
Sanskrit. Zend. Latin. Greek. Gothic, —

N. V. dtmdn-as, asman-é," sermon-és, Satpuov-es, ahman-s.
Ac. dtman-as, asman-é, sermon-és, daipov-as, ahman-s.
Instr. dtma’-bhis, asma-bis, .... (Sacpovo-giv), D. I. ahma’-m
D.Ab dtma’-bhyas, asma'-byé, sermon-i-bus, «+++ Fed
Gen. dtman-dm, asman-dm, sermon-um, Satpdv-wv, ahman-é.
Loc. dfma’-su, asma'-hva, .... daipuo'-c1, ts
|G. Ed. p. 324 | SINGULAR.

Sanskrit. Zend. Latin. Greek. Gothic,
N.. bhrétd, brata, frater, MATH, bréthar.

Ac. bhrdtar-am, brdtar-em,§ fratr-ém, atép-a(v), _ brdthar.
* Asmanus-cha, “ celique.”” + See p. 299, Rem.2. +t See p.241, Note t. |

9 Also ¢ e7G.usJs brathrém might be expected, as Vend, Sade, p. 357; |
Gerosd patrem (pathrém?), contrary to the theory of the strong cases
(§.129.), for patarém.
SINGULAR.
Sanskrit. Zend. Latin. Greek. Gothic.

In. bhrd*r-4, —_brathr-a, re ...- D. Inst. bréthr (see §. 132
D. bhrdtr-é, brdthr-é, see Loc. see Loe. be
Ab. see Gen. brdthr-at, —_fratr-e(d), .... “aoe
G. bhrdtur, bréthr-6,* fratr-is, matp-<ds, bréthr-s (see §. 13
L. bhrdtar-i, brathr-i D.fratr-i, natp-i, xe 9
YV. bhrdtar, brdtarét —frater, —-nae, bréthar.
t DUAL.

WN. Ace. Voc. bhrdtar-du, Véd. bhrdtar-4, bratar-do or brdtar-a, marép-€.
Inst. D. Ab. bhrdtri-bhydm. bratar-é-bya, TaTépo-
Gen. Loc. bhrdtr-ds, brdthr-d(2)
PLURAL-§ .
Sanskrit. Zend. Latin. Greek.
Nom. Voc. bhrdiar-as, brdtar-é, || fratr-és, mTarép-er
Accus. bhratré-n, brdthr-eus?** —fratr-és, marép-a
Instr. bhrdtri-bhis, bratar-é-bis, ie ee [G. Ed. p. 324
Dat. Abl. bhrdiri-bhyas, _brdtar-é-byé, fratr-i-bus, y
Genitive, bhrdtri-n-dm, brdthr-anm,{j _fratr-um, TAaTEép-t
Locative, bhrdtri-shu, aes cies TAT pa-¢
* Vide §. 194. p. 211, 1. 1. Note.

+ See p.216. Note ||. t See §. 44,
§ ror the Gothic, which is here wanting, see p. 253, Note f.
Il asgrssas Aso); brdtaraé-cha, “fratresque.”
{ See §. 127. Note.
** Perhaps also brdthr-6, bréthras-cha (“fratresque”), according to the
analogy of dthr-0, “ignes,” from dtar. See §.239.
tt See p. 266, Note Tf.
x
SINGULAR.
Sanskrit. Zend. Greek, Latir

N. A. V. manas, mané.* pévos, genus."
Instr. . manas-d, mananh-a,t Sis pede con
Dat. manas-é, mananh-é, see Loc. see Le
Abl. see Gen. mananh-at, bee ge
Gen. manas-as, mananh-d (mananhas-cha), _réve(a)-os, ge
Loc. — manas-i, manah-i,(see p. 316,G.ed.) D. péve(o)-, gen
* Manas-cha, “mensque,’’ “mentemque.”

+ M. Burnonf remarks, in his review (in the separate impression. p. 11),
that in this class of words the instrumental ending is generally long, 3
I, in like manner, had remarked forms enough of this kind with a long 4, _
but in passages where also many a’s, originally short, appear to be length-
ened at the termination, and which, therefore, I was not willing to bring q
into account: moreover, the cases could not be included, where, through
the particle Ass cha, a preceding .s @ is preserved in its original length.
After deducting these two classes from forms in anhd, the computation —
might perhaps turn out in favour of the short a given above. I have,
however, as yet not applied any closer reckoning: it would, however,
surprise me if, on more exact calculation, but still in departure from the &
fate of other polysyllabic words ending with a shortened a, the advantage —
in this particular case should incline to the side of those words which .
retain the long vowel, which I would then gladly restore. Noone will
deny that the collation of MSS. is of great importance in deciding many _
grammatical and orthographical questions, although I believe I may assert ‘ti
that even a single lithographed MS. opens a rich field to inquiries and :
important grammatical observations: for although it is very full of errors,
it nevertheless shews no systematic opposition to what is correct; and
many expressions, passages, and turns recur so frequently, that, taken —
together, they canin a measure supply the place of a comparison of other
MSS. For the rest I had at my command the edition of Olshausen of
the three first chapters and part of the fourth of the Vendidéd, with the —
various readings attached to it, so that, through these means, I was not
heft entirely destitute of MSS,
wisi. [G. Ea. p. 326.]

N. Ac. V. manas-i, ain péve(o)-e.
I. D. Ab. mané-bhydim, mune-bya(p. 316 G.ed.), D.G. pevé(c)o-w.*
G.L. = manas-és, mananh-é(?) (p. 297 G. ed.),
PLURAL.
Sanskrit. Zend. Greek. Latin.

N. Ac. V. mandns-i, mananh-a,t éve(o)-a, gener-a.
Instr. mané-bhis, mane-bis, (uéveo-puv,) mE
Dat. Abl. mané-bhyas, mane-byé, see Loc. gener-i-bus.
Genitive,manas-dm, mananh-anm, pevé(c)-wv, —_—gener-um.
Locative, manas-su, mané-hva, éveo-ci, re
SINGULAR, MASCULINE AND FEMININE. |G. Ed. p.327-]
eatin
Nom. durmands, dushmando (§. 56°.). duopevns (§. 146.)
Accus. durmanas-am, dushmananh-ém, ducpevé(c)-a(v).
Voc. durmanas, ae ducpevés.
The rest like the simple word.
DUAL
N.Ac.V. durmanas-éu, ‘
Vis, dernanasdt dushmananh-a (?) ducpuevé{c)-e.

PLURAL.
N. Voc. durmanas-as, dushmananh-é (ai-cha), Svopevé(a)-es.

Accus. durmanas-as, dushmananh-o (ai-cha), 3vopevé(c)-as.
* See
p. 299, Rem. 2.
T See p. 245, Note}. It was, however, from an oversight that J,
as was observed at p.253, Note §. read in the Vendidad SAde, p. 127,
AWZEG EY REmenha - it should be SUG E/ némanha, and may also be
considered the instrumental singular; then we should have in this pas-
sage, which recurs three times, the instrumental in »sw3u anha in both
editions three times with a short a.
t See p. 230, Note*.
x 2
308 FORMATION OF CASES
SINGULAR, NEUTER.

Nom. Ac. V. durmanas, dushmané (a’-cha), Suopevés.
“ Remark.—It was remarked in §. 152. (comp. §. 146.), that
the = in forms like pévos, edyevés, belongs to the base, and
is not the nominative character ; and that the = in forms like
tetupos has come from 7, and in like manner belongs to the
theme. M. Reimnitz, who, in (p. 54, &c.) his pamphlet men-
tioned at p. 294, G. ed., agrees with this view, first given in
my treatise “On some Demonstrative Bases,” wishes to look
upon the = in the masculine terudas as belonging to the
base, and arising out of 7; in which I cannot agree with him,
as I, according to the view generally taken, consider the
final letters of rervpas as marks of the nominative, before
[G. Ed. p.328.] which the final letter of the base is suppressed
on account of the incompatible association of ro (comp. §. 99.),
and replaced by lengthening the preceding vowel; as, for
example, in “éAas for uéAavs. The Sanskrit has a few bases
in n which, differing from the ruling principle (see §. 139.),
run parallel in the nominative to the Greek néAds;_ thus,
panthds, “the way,” from panthan, accusative panthin-am.
Only in this panthds the lengthening of the a can be less re-
garded as a compensation for the rejected n than in the Greek,
because it extends also to the other full cases (§. 129.), with
the exception of the vocative; but perhaps the lengthening
of the a has originally taken place only in the nominative,
and has thence imparted itself, when the reason of this
prolongation was no longer perceived, to those cases which
otherwise stood upon an equal footing with the nomina-
tive. Thus one says A@Tq mahdn, “great” (from the theme
mahant, properly a participle present from wz mah, “ to
grow”), with the vowel of the concluding syllable length-
ened, according to the analogy of the Greek form, as
Aéywv. The Sanskrit word, however, retains the long vowel
also in the other strong cases (muhdntam “ magnum,” mahantas

“magni,” mahdntdu, “peyaAw”), with the exception of the vo-
cative; while the usual participles present leave the a short
in all the strong cases. In most exact accordance, however,
with the Greek participle present stand the Sanskrit pos-
sessive adjectives, which are formed by the suffix vant
(Greek evr for Fevr, in peArréers and others) and mant (in the
weak cases vat, mat). These lengthen, that is to say the a
only, in the nominative singular; so, for example, dhanavén,
* dives ”* (from dhana, “riches”), dhanavant-am, dhanavant-du,
dhanavant-as, as Aéywv, AéyovTa, AeyovTw, AeyédvTes.
OLD SCLAVONIC DECLENSION.t [G. Ed. p.329.]

255. Before we enter upon the province of Sclavonic
Grammar, we must endeavour to explain its system
of sounds; and although it is not requisite to specify
all the minutiz of the subject, we must, nevertheless,
bring into notice those parts which are indispensable to
the understanding of the Grammar. It is therefore our
principal object, in the following remarks, to exhibit the
connection of the Old Sclavonic sounds with those of the
elder languages, of which they are either the true trans-
* If,as has been remarked in another place,

the suffix aq vant has
maintained itself inthe Latin in the form lent (as opulents), it would not
be surprising ifthe weak form at vat, without the interchange of v with 1
but with the weakening ofthe ato i, had its representative in the Latin
divit, which stands in the same relation to dhanavat, by passing over the
middle syllable, as malo to marolo.
+ It is stated by Professor Bopp, in the preface to the second published
portion of this Grammar, commencing with the formation of cases in
general, that it had not occurred to him to direct his attention at an
earlier period to the Sclavonic tongues: having subsequently considered
the subject, he found sufficient reason to include them in the same
family of languages, and accordingly devotes to its principles of declension
the supplementary section which follows.— Editor.
31G. - FORMATION OF CASES.
missions, or corruptions more or less vitiated. We give

therefore, for the first time, a history of the Sclavonic
sounds, in which, however, as is natural, as far as their value
is concerned, we have nothing new to bring forward; and in
this respect follow only the teaching of native grammarians.
(a..\—The Old Sanskrit wa has so far experienced, in the
Sclavonic, an exactly similar fate to that which has befallen it
in the Greek, that it is most frequently supplied by e or o
(e, 0), which are always short: it very rarely remains a. In
the interior of the bases, also, e and o are interchanged as in
Greek; and as, for example, Adyos is related to Aéya,
so, in the Old Sclavonic, is brod, “ferry,” to bredé, “I wade
through ;” voz, “carriage,” to vezd, “I ride in a carriage.”
And as, in the Greek, the vocative Adye is related to the
theme AOTO, so is, in the Old Sclavonic, rabe, “O slave,” to
rabo, nominative rab, “a slave.” The o has more
weight than e, but a more than o; and hence a
corresponds most frequently to a Sanskrit 4, so that,
for instance, in the Old Sclavonic, forms in a answer to
the feminine bases in wt 4 (comp. vdova, “ widow,” with
fayat vidhavd), which, in the vocative, is in like manner
abbreviated to o (vdovo!), as above o to e. As final
vowel, also, of the first member of a compound, @ is
weakened to 0; for instance, vodo-pad, “waterfall,” vodo-
pot, “ water-drinker,” for voda-; just as in the Greek
Movoo-tpadis, Moveo-pidns, and similar compounds, which
[G. Ed. p.830.] have shortened the feminine a or 7 to o.
Even if, therefore, a is in the Old Sclavonic a short vowel,
I nevertheless regard it, in respect to grammar, as the long
0; so that in this the Old Sclavonic stands in a reversed
relation to the Gothic, in which a has shewn itself to us as
the short of 6, and,in case of abbreviation, 6 would become
a, exactly as in the Old Sclavonic a becomes o.
(b.)— = i and $@ both appear in the Old Sclavonic as i,
and the difference of the quantity is removed, at least I
OLD SCLAVONIC. 311
IN THE
ter ¢ is anywhere
do not find that a longer or shor
with
spoken of. Let schivd, “I live,” be compared
sila; and, on the
statin jivdmi ; sila, “virtue,” with yte
the root fag vid, “ to
other hand, vidyeti, “ to see,” with
3fa védmi, the Old
know,” to the Guna form of which,
vyed my, infin. vyes-t
Sclavonic vyemy (abbreviated from
f, so that vid and
for vyed-ti,) «J know,” assimilates itsel
erent roots. The
vyed in the Sclavonic appear as two diff
ly in the Old Scla-
short = i, however, appears frequent
as in the Greek
vonic also in the corruption to (€), e
that is to say, the
and the Old High German (§. 72.);
i, and the numeral
bases in i shew, in several cases, e for
osition in the
three (fa tri) appears frequently in comp
g. treptitye, “ trivium.” So, also, péte-shestvye,
form tre, e.
sovery frequently
édormopia from PUTI (§. 260.). Theiisal
al dadyat, “they
*suppressed, e.g. in the 3d person plur
give,” Sanskrit qefa dadati; sit, “they are,” Sanskrit
vowel
uf santi. Where i forms a diphthong with a
writ ing with a
preceding it, it is marked in the old
“ strife.”
short mark, which we retain, e. 9. boi,
in the forms
(c).—s u and & @ have, in the Old Sclavonic, In
me y+
which are retained most correctly, both beco
answers to 4
this manner, for instance, by (infin. by-ti)
to Dobrowsky’s incorrect
* The suppression here noticed of final i refers
er, the final i in Old Sclavonic has
orthography. In point of fact, howev Do-
e b y; ¢-9- that which
either been retained unaltered, or has becom
writes dadjat , “they give,” sit, “they are,” should be
browsky, l.c.,
Regarding the nasalized
corrected to AAAATD, dadanty, (ATb sunty.
vowels, see §. 783. Remark.
y, as, like the Greek
+ We express, as in Polish, the yery or dull é by
the old short or long u.
v, where it is original it supplies the place of
(by Gretsch II. p. 666.), as
It is pronounced in Russian, according to Reiff
ally ; according to
in the French oui, spoken very short and monosyllabic
i (Heym, p.5). This
Heym, nearly like i, in union with a very short
of this letter (Reiff,
does not, however, remain the same in all positions
labials like a dull thick ¢
1. c.), and it sounds after consonants other than
(i sourd et étouffé”).
bhi, “to be; svekry, “mother-in-law,"to wysvasré ; myshy,

“mouse,” to aH mdsha; syn, “son,” to Wa séinu;
chetyri, técoapes, with @ag chatur (in the theme), nomina-
tive masculine wratta chatwdras. The instances of y for :
Zu are, nevertheless, more rare than those where y
corresponds to the long & @; for the short u, as in |
the Old High German (§. 70.), has for the most part
[G. Ed. p. 831.] become o; and thus, for example, snocha,
“daughter-in-law,” answers to @at snushd; oba, “both,”
to gut ubhd (Védie form), Zend sss» ubd. Hence, also,
the old u declension has, in many cases, become similarto
the o declension, which, according to (a.), has arisen from
wa; and, on the other side, 0 may also, but only in
substantives, participate in those forms which belong
only to the genuine u declension: whence it is easily
perceived that the genius of the language could not
everywhere distinguish further the two kinds of 0, in
their history, indeed, far separated from one another,
but phonetically identical.
(d).-Unorganic y, i.e. y as representative of original
vowels other than ¥ u or & w%, is not uncommon in the
grammar; that is to say, the personal termination my
(Ist person plural), like the Latin mus, has arisen
from the more ancient mas; and if the bases in a (for
‘at d) have y in the nominative plural (vdovy, “ vidue ™),
still the y here is so much the less to be looked upon
as a case termination, as no account could be given
of y in this sense; and with bases in ya the a of the
base is also really retained (volya, “voluntates”). But
as the y exerts the force of an Umlaut on an o suc-
ceeding it, by which that vowel is changed to an e, so
I think that to an z following the 0, without the interven-
tion of another letter, the force of a reactive Umlaut must
be ascribed, even if this force is not everywhere exerted,
and that some y’s must be declared to be the Umi/auts
ofo; that is to say, as soon as so much has been re-
IN THE OLD SCLAVONIC. 313
cognised in the Old Sclavonic adjectives, that their

bases all end either in o or yo (changed by the Umlaut
to ye), and are thus sister forms to the Greek, like ATA@O,
‘ATIO; andof the Sanskrit, asWa Swéta, “ white,” feadivya,
“heavenly ”;—so soon, I say, as the abbreviation of the
base in the masculine nominative has been recognised
(nov, novus, for novo), then will it be no longer said wit.
Dobrowsky (p. 318) that the definite adjectives are derived
from the primitives (indefinite) by annexing, according
to the measure of the final letter of the primitive, either
yi or i.* If, however, I may trust that I have obtained
an accurate knowledge of the organization of the Old
Sclavonic grammar on any point, it is on this, that the affix
in the nominative singular of definite adjectives consists
not in yi or i, but in 7 as a mutilation of yo from ya
(aya), and in the feminine of ya from yd (G. Ed. p. 332}.
(at yd). This also appears to me subject to no manner
of doubt, that if, for example, the compound word svyatyé
comes from the word svyato, “ holy,” its acknowledged
theme, the y is a euphonic product from o, through the in-
fluence of the i which is added to it. This i has, in some
cases, in which it has been dropped, still in a degree, in its
euphonie operation, left its reflection, and thereby the
proof of its former existence. Thus, for instance,
svyaty-m, “per sanctum,” from the older svyatyim,
svyaty-ch, “ sanctorum,” and “in sanctis,” from svyatyi-ch,
corresponds to the indefinite forms svyato-m, svyatye-ch
(for svyato-ch).| At times, through the said pronominal
syllable i, the preceding o may be changed at will into y
* Dobr. also himself, p. 493, considers simplei or ii as the definitive

adjunct; but in considering, as he there does, blagyi as the confluence of
blag and i, he appears to look upon the y as having arisen from the i of
the suffix, and not to acknowledge in it the final vowel of the simple
adjective root.
+ In the oldest MSS., according to Dobr. p. 502, the more full forms
yich, yim, yimé occur in the plural, for ym, ych, ymi.
or not: thus the interrogative exhibits the forms ky,

“quis?” (Dobr. 500 and 343.), kyim, “per quem?” kyiich,
“in quibus, quorum?” kyim, “quibus?” kyimi, “per
quos? with kot, kotm, kotch, kotmi. The possessive
pronouns allow no euphonic reaction at all to the de-
monstrative i, which forms the last member of them,
and they always retain their radical 0; e. g. mot, “ meus,”
motm, “per meum,” not myt, myim. As to the definite
form of the adjective bases in yo, which Dobrowsky forms
through the addition of i, I have not the slightest
doubt that here, also, a simple i is the defining element,
for the first i is clearly the vocalization of the y of the
primitive base; so that therefore, for example, sinii
“the blue,” is to be divided, not into sin-iz, but into
sini-t. The primitive adjective is sounded in the nomina-
tive which is deprived ofall inflection and of the last vowel
of the base—siny, the y of which appears as 7 in the nomi-
native plural masculine, just as in the definite pronoun,
sini, “ cewrulei,” sinii, oi “ ceerulet.” In order, however, here
fully to explain the nature and origin of the. definite
declension, and not hereafter to be compelled to repeat
what is already settled, it may be stated that its pro-
nominal defining addition is identical with the Sanskrit
relative base q ya, which is most correctly preserved
in the Lithuanian, in which language *ya signifies “he”
(ya-m, “to him,” ya-mé, “ in him”). The nominative
yis, “he” (for yas), has given the y an assimilating
influence, as is the case with all bases in ya (§. 135.).
The feminine, also, is pronounced in the nominative,
through assimilation, yi for ya; but the genitive e
e
ee
yos, and all the other cases, are easily perceived through
the declension- of ranka, “hand,” and giesme, “song,”
[G. Ed. p.333.] from GIESMYA (p. 169, Note). The
* Written ja in the text. This passage furnishes a good reason for

writing the Germanic j by y, as has been done throughout this translation.
Old Sclavonic has, in all the masculine bases ending with

a vowel, suppressed this vowel in the nominative and
accusative; and since the vowel has dropped from the
Sanskrit-Lithuanian base 4 ya, ya—which, according to(a.),
-‘ makes one expect yo in the Old Sclavonic, from which,
according to (n.), must be formed ye*—the y must be
pe
een
changed into a vowel; hence, i, “he,” “ him,” which
must, therefore, on no account be placed together with
the Latin-Gothic is, from the base i, In the nomina-
tive singular masculine, however, this Sclavonic pro-
noun occurs in all the three genders, not isolated, but in
union with the particle sche, which has preserved to it
the old relative meaning: i-sche means as well “qui”
as “quem”; ya-sche, “que”; yii-sche, “quam”; and ye-sche
ae
es
hemieeiheee quod.” Now as i means “he,” ya, “she,” and ye, seh
I could not imagine how one could create the definitive
. adjective forms svyaty-i, svyata-ya, svyato-e (for svyatoye),
accusative svyaty-i, svyati-yi, svyato-e, in their opposition
to the indefinites svyat(o), svyata, svyato, differently from
Dobrowsky (p. 493), and perhaps other grammarians
before him, have done, namely, by the addition of the
pronoun here under discussion;{ for this pronominal
suffix supplies the place of the article of other languages ;
and the Lithuanian language uses the same pronoun
- * Hence in the genitive ye-go, dative ye-mi, loc. ye-m, the e of which
Dobrowsky wrongly ascribes to flexion, because he everywhere seeks the
_ base in the nominative. However, the base ye has not fully maintained
itself before all terminations beginning with a consonant, but become, in
like manner, shortened to i: in i-m, “‘ per eum,” and iis, i-mi “ per eos,”
i-ch, “ eorum,” “‘in iis,” for ye-m, &c.
+ What Grimm (by Wuk, p. xl.) remarks against this declaration has
not convinced me; least of all can I, for the above reasons, concede to
him that the i ofsvyatyi has any thing to do with the a of blinda, “the
blind” (from blindan, §. 140.) ; so that syyatyi would belong to the indefi-
nite declension ;and, on the other hand, svyat, contrary to the Sclavonic
Grammarians, would be to be removed from the indefinite into the defi-
nite forms.
for the same object, ie. equally in the emphatic, or,

as it is also termed, definite declension of the adjective;
and certainly so, that, through all cases, both the adjec-
tive which precedes and the pronoun which concludes are
declined, while, in the Sclavonic, in most cases the pronoun
only is provided with the inflexions of case, but in some
[G. Ed. p.334.] it has utterly disappeared, and in others
is still to be recognised in the y for o mentioned above. —
(e.)}—The Sanskrit diphthong z é I have found always ren-
dered, in the Old Sclavonic, by ye, in similar forms; so
that after weakening the zw é, to compensate for this,
the semi-vowel y has made its appearance, to which, in 4
this union, a particular legitimacy would be, according —
to (c.), to be ascribed. Let pyena, “ foam,” be compan
with ta phéna; svyet “light,” with Wt svéta; vyemy, “ [
know,” with fa védmi. The most important’ cases in
the grammar wth ye corresponding to z@é are the dual 7
case forms of the feminine and neuter, and those of the
imperative, in accordance with the Sanskrit potential of —
the first conjugation. .
(f.)—The Sanskrit diphthong wt 6 (from a+u) is repre-
sented in the Old Sclavonic by # (#);* so that the first
* Although this vowel may at times be pronounced short, still this much,
at least, is certain, that, according to its origin and its definition, it is long.
In Bohemian it appears in two forms, as au and uw: the former is pro-
nounced ou, but the writing points to an older and different pronunciation, —
in which the a was accurately preserved in its place: the u is pronounced
short, whence, however, it cannot be deduced that this short u perhaps —
corresponds to the Sanskrit g and Greek v, and that au is its intensitive
or Guna; but, on the contrary, only the w retained in the au corre- —
sponds to the Sanskrit $ u, and the w which stands alone in Bohemian ~
is a weakening of the au; so that, from this, the concluding element v —
alone is left: etymologically, that is to say, the Bohemian au, as also u, q
answers to the Sanskrit ft 6, and also to the Sclavonic é (g), only that—
the former is phonetically more exact, and without the loss brought about —
by time. Hence, also, usta (written vsta) “ora” corresponds to the San-
skrit BYE dsh, ha, “the lip”: more complete, however, is austne, “ by word
of -
~ element of the Indian diphthong has assimilated itself to

the second, and, in conjunction with it, presents a simi-
lar long vowel, as, in the Greek # (ov), two hetero-
geneous vowels, according to pronunciation, have united
themselves in a similar measure. As, according to (a),
the Indian short a has, in the Sclavonic, mostly become
short 0, we must consider the first element in the diph-
thong é# also (so we write the #) to be 0; and it be-
comes visible, too, in this form, when @ is resolved before
vowels into ov, (compare So(F)és from B8, [G. Ed. p. 335.]
§. 123.), while the Indian #t 6 becomes av before a vowel
(fx gavi=foFi, from zitgé). Now as, in the Sanskrit,
gu,
% aw, rise to 6 through Guna (§. 26.), and sté-shydmi
appears as the future of stu, so in the Old Sclavonic,
in like manner, y (cy) is interchanged with @; so that bd
in bi-du, “I shall be,” must pass as the Guna form of by
(in byti, “to be”): but ifa class of nouns, which in the
nominative-accusative terminate in a consonant or in
yerr (see k.), exhibit, in many oblique cases, the syllable
ov before vowel-endings, this ov must neither be consi-
dered, with Dobrowsky, for an augment added to the
base, nor can it be deduced from forms like synovi, “ from
a son” (Sanskrit qq sinav-e, from siinu), zynov-é, “sons”
(aaa siinav-as), that syn, in the nominative-accusative,
is an abbreviation of synd; and that therefore the yerr,
when it is added to the form syn, is a representative
or weak remainder of &: but it is clear, from (c.),
that syn, “/filius,” “filium,” if its final vowel, in its
most genuine form, had remained to it, would sound
syny, from which synov is the Guna intensitive, the
ov of which has arisen from ¢ through the influence
A
of mouth”; and even for vsta is to be found austa (Dobr. Bohm. Lehrg.
p.4.): ruka corres to the ponds
Lithuanian ranka, “hand” ; and hus to the
Sanskrit $# haisa, “ goose” ; for which, according to p. 319. rauka, hausa
was to have been expected. A distinction must here. according to §. 783.
Remark q. y., be made between Oy a, and & un.
of the vowel following it, but has remained in the

genitive plural also, after the ending has been dropp ed.
Let synov, “filiorum,” be compared with the Gothic
suniv-é (§. 247.) As, in the Sanskrit, the substantiv
bases in wu adopt the Guna form of the wu before the
vowels of the derivative suffix, so it is very remarkable
that, in the Old Sclavonic bases in y, also, this vowel
appears before certain derivative suffixes in its Guna
form ; e. g. domov-it from dom (DOMY), “ house”; binov-at,
“debtor,” from byn (BYNY ).* Derivative substantives
and adjectives in ov, ev (theme ovo, evo, the latter for
yovo, see n.), correspond to the Sanskrit in Wa ava; as
unea paindav-a (nominative as), “descendant of Pandu”;
"THT drtava, “seasonable,” from We radu, “season”: so,
in Old Sclavonic, Adamov, “Adamite,” from Adam —
(ADAMY ); zarev for zaryev, “ kingly,” from xar (theme~
ZARYY). For these formations, therefore, we must not, —
with Dobrowsky (322, 323), assume a suffix ov or ev,
but we must look upon the o alone, which, in the nomi-
native, is suppressed, as the derivative suffix (4DAMOV-O,
ZAREV-O). Through the Vriddhi increase (§. 29.) the
Old Sclavonic y becomes av, because a, according to (a.),
usually corresponds to #1 4: hence, from the root by,
“to be,” comes the causal baviti (infinitive), as in the
[G. Ed. p.336.] Sanskrit wrafaqa bhdvayitum. But
though sfaviti occurs as the causal of sta, this form may
have arisen in the perverted feeling of the language as an
irregularly analogous word to baviti. In order, then, still
more to establish, by a few other examples, the representa-
tion of the Indian #4 or #4 av by the Sclavonic #, we
find dst, “ mouth,” correspond to wre éshtha, “lip”; shdr
“sinister” (theme SHUYO), to wat savya; biditi, “to
awake”—a causal, whose primitive bdyeti has entirely
* Dobrowsky supports himself in these cases by calling ov a prefix (p. 329).

lost the vowel of the root—to @wfaaa bédhayitum, also

“to awake,” from yy budh, “to know.” ~ Thus giibiti is
the causal of gyb-né (1. P.), and stdditi of styd-né (Dobr.
360, 361.); while vyesiti is the causal of visyeti (see e.), as,
in the Sanskrit, tyfamA vésayitum, “to cause to enter,”
from fay vis, “to go in.”
(g.)—As the nasals* easily resolve themselves into u, so
the second element of the diphthong ¢@ sometimes also
supplies the place of a nasal in the cognate languages;
e.g. réka, “a hand,” Lithuanian ranka; pity, “a way,”
Sanskrit ware panthds, id. Latin pons; geluby, “a dove,”
columba; gisy, “a goose,” ¢@ hansa. The Polish has
preserved the old nasal in golamb, “a dove,” gansie, “a
gosling,” gansior, “a gander,” and in many similar
cases. Hereby the @ in the accusative of bases in a
(from #14), which are for the most part feminine, is
remarkably explained; compare vdové from vdova, “a
widow,” with fawatq vidhavdm, “ viduam.” Therefore
vdovit is to be derived from vdovo~m for vdova-m (see a.) ;
so that the a which is weakened to an o is contracted
with the nasal mark of the case tod. This view is further
supported by the consideration, that in Polish, also, the
corresponding feminine declension marks the final vowel
of the base with the same sign which, in the middle of
a word, expresses a nasal, which is governed according
to the organ of the following letter, but at the end,
probably through a corruption of sound, is said to have
an equal value with a ringing h. This nasalizing mark
recurs also in the Polish verb, and, indeed, exactly in
such a place where one had to expect a nasal, i.¢. in
the lst person singular and 3d person plural; and thus,
in Bandtke’s second and third conjugation, the so
marked ¢, e.g. in pieke, “I bake,” supplies the place of
the am of the first conjugation, as czyfam, “I read.”
* Cf. §.783. Remark.

320 FORMATION OF GASES
The Old Sclavonic has, however, excepting some ano-

malous remains of an older formation, @ in all the con-
jugations; and, according to what has been said, it
admits of no doubt, that in the second part of this diph- a’
thong (0+%) the personal character m, and in the first

part of the diphthong the conjunctive vowel, is retained.
When therefore, in the Ist person, an 0 corresponds to the
e (e) of nes e-shi, “thou carriest,” nes-e-t, “he carries "—
[G. Ed. p. 837.] for nest is for nes-o-u for nes-o-m from
nes-e-m—it must be assumed that the conjunctive vowel e,
before its confluence with the i, which has arisen out of m,
has passed into 0; as in Greek ov arises by the contraction —
of e and o, through the transition of e into o and o into v.

The same relation is to be found in the Old Sclavonic in
the 3d person plural, where, correspording to nes-e-m,
“ wecarry, nes-e-te, “ ye carry ” (comp. Aéy-e-re), the form
nesent is expected, but in place of it occurs nesdt in sur-
prising accord with the Greek Aéyouor for Aéyovo: from
Aéyovtt. The Polish has, like the Bohemian, relinquished
the character of the 3d person in the plural, as well as for
the most part in the singular, but everywhere retains, in
the first, the old and more powerful a (=), and marks this S
e
with the diacritical sign mentioned above, which, in the

middle of a word, supplies the place of a nasal function;
thus, sa, “they are,” corresponds to the Sanskrit wf
santi, Sclavonic sit. The Bohemian has also, in many ‘
conjugations, retained the old conjunctive vowel a in the
3d person plural, but, like the Sclavonic, permitted the n
to dissolve into a u; therefore, in wezau, “vehunt”
t
L
(wez-e-me, “vehimus,” wez-e-te, “ vehitis”), the u answers to

the n of azfat vahanti, “vehunt,” and the u which, in Bohe-
mian, is united with an a, is essentially different from
that which stands alone; for the latter answers to the
Old Sclavonic diphthong @ (s), but the former only to
the latter portion of the d, which, in the Old Sclavonic,
never stands alone, at least never occurs as u, but as y (c).
If, then, through what has been said, the vocalization of

the m or n, which is of such frequent occurrence in the Scla-
vonic, has been shewn with sufficient clearness, it is remark-
able that conversely, also, the latter portion of the @ (#) has
- occasionally been hardened into a nasal; and thus bidd, “I
_ will be,” is in Polish bende (written bede).
(h.}—In certain cases an old 4 (=) unorganically supplies
the place of the Sclavonic d, i.e. in the instrumental of
pronouns without gender, and all feminines; thus,
vdovoy-i, “through the widow,” answers to fawrat vidha-
vay-d; and toboy-i, “through thee,” to wat éway-d. Deno-
minatives also, in éyd (Ist per. pres.), in the Old Sclavo-
nic, correspond to the Sanskrit in wrarfa dydmi, as yerafa
Sabdéydmi, “1 sound,” from ye sabda, “a sound,”; facta
chirayami, “I hesitate,” from fat chira, “long”: thus,
in the Sclavonic, zielityd, “I greet,” “I kiss,” from ziel,
(ZIELO), “healthy”: vdoviyd from vdova, “ widow” (Dobr.
p-372.). Finally, words in dn (UNO) answer, as it appears,
to the Sanskrit pera of the middle voice, in dna, as
yunjana, “uniting,” from Fy yuj; so in the Old
Sclavonic, perin, (PER UNO), “ “Deus [G. Ed. p. 338.]
tonans,” from the root per, “to shake”; byegén, “ runner”
(BYEGUNO), from BYEG “to run” (Dobr. p. 289.).
(i.)—There are in the Sclavonic alphabet two marks, which
by some are called litfere aphone, but by Gretsch semi-
vowels; I mean the so-called soft yer,* and the hard yerr.
The former is represented by Gretsch as half i, and by
his translator, Reiff (47), as answering to the tones
‘mouillés’ of French (compare Kopitar, p. 5); and thus
schal®, “sympathy,” and ogon”, “fire,” are, in respec-
to the soft yer compared with the pronunciation of
travail and cicogne. This yer, therefore, denotes a tone
* In the original jer, pronounced, however, yer ; and hence y has been
substituted for 7 in all that follows.— Editor.
which is rather to be called a y than an i*; and it may

be said that in schal” and ogon” one hears quite as much
of a yas can be heard of this semi-vowel after a con-
sonant preceding it. Hence we mark it witha y, and
write the above words schaly, ogony, Old Sclavonie ogny.
In the words, too, which end with it in the uninflected
nominative and accusative singular, it occurs in several
oblique cases as a distinct proper y, e.g. in zarya, “ regis,”
zaryu, “regi,” from zary, “rex,” “regem.” On the consonant
which precedes it this yer has an influence which ren-
ders its pronunciation more mild, because its sound is —
somewhat broken by the y, which throws back its sound.
Etymologically the yer corresponds either to a final i of
the cognate languages, as in yesty, “he is” (fer asti,
éori, Lithuanian esti), kosty, “bones” (wfeq asthi), or
in the nominative and accusative singular of masculine
substantives and adjectives, to a y (q y), from which a
vowel has dropped; for the theme of siny, “cwruleus,”
concludes neither with ¢ nor with y, but with yo (euphoni-
cally ye, see n.); whose final vowel, suppressed in the
nominative and accusative masculine, appears, however,
in the feminine sinya, in its extension to a, while the
neuter sine for sinye has rejected the y.
(k.)—The hard yerr is represented by Gretsch as a semi 0,
——
but by Reiff, more correctly in my opinion, it is com-
pared to the French silent e and the Hebrew schva: itis —
therefore, to use the expression, equivalent to “nothing”; —
and one cannot perceive of what vowel the small, still
perhaps remaining vowel part of it is the residue. Conso-
nants preceding it have a stronger and free pronunciation;
(G. Ed. p. 839.] and Kopitar (p. 5) tells us that they are
pronounced before it sharp, and without echo, and that it
is for this reason called the hard yerr, and not on account
of its own pronunciation. We require, therefore, in the
* In the Carniolan dialect this sound has mostly disappeared; but —
where it has remained it is also written by a y; as, kony, “ horse.”
IN THE OLD SCLAVONIC. © 323
Roman character, no substitute for this mark, and

Dobrowsky also on its it at the end of words. Etymo-
logically, however, this yerr always represents a sup-
pressed mute vowel, only not always an 0, nor, as
Grimm conjectures (in his valuable Preface to Wuk’s
Servian Gramm. p. xxxtv) a u. Rather, each of the
three short fundamental vowels—a (as represented also
by o, e), i, u, (for which may stand y, 0),—is very fre-
quently dropped at the end of words; and although the
i is seldom entirely suppressed, more generally throwing
back its sound as y, nevertheless the vowel suppressed
after the m of rabo-m, “per servum,” and in Russian
replaced by yerr, is clearly, as we gather from the
Lithuanian, an i.
(/.)—I* believe I may assert, that in the whole extent of
the structure of the Sclavonic language, at least in
all the conditions of its noun and verb, not a single
final consonant occurs after which some termination,
which, through the cognate languages can be pointed
out as beginning with a vowel, has not been dropped.
Thus, the base NEBES, “celum,” forms, in the genitive
plural, likewise nebes, but the vanished termination
is, in Sanskrit, =m dm (ara nabhasdm, “ celo-
rum”), Greek wy (vedé(r)wv), Latin um, Gothic ¢. The
real final consonants, however, which, in the truly-pre-
served elder dialects of the Indo-European family, stand
as the foundation of the word, have utterly disappeared
in Selavonic polysyllables; e. g. from WA_as, €s is formed,
in the nominative plural, e (e); and synov-e answers to
forms like #fq@ siinav-as, Pédrpv-es.
(m.)—As far as regards the writing of those consonants
which, in the Sclavonie alphabet, properly correspond to
the Roman, we express the sound of the French j (zivycte,
in the Carniolan sh), as in Zend (§. 65.), by sch, that
of our German sch (=) by sh as in Sanskrit
* Cf£. §. 783. Remark.
¥2
324 * FORMATION OF CASES
and also as, in Sanskrit, the fsch by ch: for the

sound of the Greek ¢ (=ds) we retain @, and use z for
the sound of our German z (=t¢s): for y we write ch. In
regard to etymology, it is important to call attention
to the relation of this letter to sibilants, by means
of which snoche, “daughter-in-law,” corresponds to
the Sanskrit @qut snushd. Ch also, in - declension
and conjugation before certain vowels, passes into gs
[G. Ed. p.340.] (Dobr. pp. 39, 41), and in some cases
into sh (Dobr. 41.). Finally, in preterites like dach, “I
gave,” dachom, “ we gave,” the ch returns to the s (a s, 3)
whence it has proceeded, in the cases where a personal
ending beginning with a ¢ follows it; hence, daste, “ye
gave,” dasta, “ye two” and “they two gave.”"* As the
vowels exercise a multifarious influence in the trans-
formation of gutturals preceding them, we will further re-
mark that the ch under discussion maintains itself in the
3d person plural before @, but before @ appears as sh;
hence, dasha or daché, “ they gave.”
(n.)—}For the semi-vowel y (q y) the Cyrillian alphabet
gives the Greek 1, excepting in the cases for which the
inventor of the character has provided by particular |
letters set together according to their value, which, at
the same time, express the y with the following vowel;
that is to say, ya is never written by two letters. It
would, however, for this reason, be wrong to assume a
vowel ya, as this syllable, however it may be written,
still always unites in itself two sounds. For ye, also,
* Dobrowsky has, however, as it appears to me, not perceived the

irrefragable connection between the ch of dach and the s of daste, for he
considers the ch and ste, &c. as personal terminations (pp. 264, 383. 397);
and hence he nowhere informs us that ch before ¢ passes into s. More on
this subject when we come to the verb.
+ The vowels mentioned here, preceded by g, are, with the exception of
te ye, and yé, nasalised vowels (see §, 783. Remark); and hence pyaty,
“five,” must be pronounced panty (in the original character MA'Tb).
Cyril has provided by a simple sign, and yd is expressed

by an o in conjunction with ans But y often appears in
Sclavonic as a dialectic addition before vowels foreign
to the cognate languages. Compare yesmy, “I am,”
yam (for yadmy), “T eat,” pyaty, “five,” desyaty, “ten,”
yedin, “ one,” with the corresponding Sanskrit forms, asmi,
admi, panchan, dasan, adi (primus). An o which follows
is, in accordance with similar forms which we have
observed in the Zend and Lithuanian (§. 137. and p. 174,
Note*), changed into e through the influence of a y
preceding it. In like manner, in accordance with the
Zend and Lithuanian, the y, after it has assimilated a
vowel following it, has often itself disappeared, and has
left behind only its effect, and thereby the proof 1s its
former existence.*
* Dobrowsky does not express himself with sufficient clearness re-

garding this form, when he says (cap. IT. §. iii.) that 0 after y and liquid |
consonants is changed into e. According to this, one would believe that,
besides y, certain other consonants had the power of changing an o follow-
ing them into e. Dobrowsky understands—which, however, as far as
I know, he nowhere expressly says—under “consone liquide,” those
which, inconsequence of a following yer (y), have retained a more flowing
and softer pronunciation ; while he calls the consonants without yer “ con-
sone solide”’ (comp. | c. p. 267); so that no consonant is by nature and
of itself alone liquid, but receives this quality through a following yer
(a y without a vowel). Thus, in Dobrowsky’s second masculine declen-
sion, the consonants 7, ch, and ¢, in zary, “king,” vrachy, “ physician,”
and knya¢y, “prince,” are liquid. But as these words in the instru-
mental form zarem, brachem, knyafem, Dobrowsky ascribes the e for o
to the influence of a liquid consonant ;while, according to my opinion, the
consonants in these forms have no concern whatever in transforming 0 into
e, but for zarem, &c. zaryem must originally have stood. And as in this
form the y is the full semi-vowel, not entirely without a vowel sound, and
therefore not the expression of the yer without a vowel which softens the
consonant preceding it—as in the abbreviated nominative zary—so the r
also, in zaryem, was not liquid, and has not, according to my opinion, be-
come liquid after the dropping of the semi-vowel ; at least, ] find it nowhere
stated
[G. Ed. p. 341.) 256. We must now, in order to be able

to compare the true case-suffixes of the Old Sclavonic with
those of the cognate languages, first of all endeavour to ascer-
tain the final letter of the kinds of base which occur, as they
have for the most part been rubbed off in the singular
nominative, whence it has appeared as if these letters,
where they again present themselves in the oblique cases,
either belonged to the case termination, or were an addition
equally foreign to the base and to the termination, which has
been termed “augment” by Dobrowsky. After becoming
[G, Ed. p. 842.] acquainted with the true base, the case ter-
minations assume, in many points, an entirely different shape
from what Dobrowsky has represented (p. 460), with whom
we cannot concede to the neuter a nominative termination
o or e, but perhaps the advantage of having preserved, in pre-
ference to the masculine, the final vowel of the theme in this
case. For the practical use of the language, and to keep
simply within the limits of the Sclavonic language, all might,
notwithstanding, be assumed as inflexion which is usually
represented as such. It is not, however, here our object
to consider those syllables as supplying the place of gram-
matical relations which present themselves to the feeling
of the speaker as such, but only those which may be so
traced through the history of the language, and which, for
thousands of years, have subsisted as Grammatical forms.
257. To the masculine and neuter bases in ¥ a corre-
spond, in the Old Sclavonic as well as in Greek, bases in
o,* which vowel has disappeared in the nominative and
stated that the r and other consonants, in forms like zarem, knyafem,
golibem, lebedem, are differently pronounced from what they are in pirom,
votom, lobom, adom, of Dobrowsky’s first masc. declension. The difference
in the two classes of words is only this, that the former have a y for the last
letter but one of their theme, which, by the power of assimilation, has
changed the following 0 into e, which e, after the y has been dropped, does
not again become a.
* Dialectically the older a has, in certain cases, maintained itself, as in
accusative singular: so the corresponding a has disap-

peared in Gothic, except in the neuter (as Gothic dlinda-ta,
“eoecum,” in contrast with blind’-s, “cacus”): it has also
maintained itself frequently in the beginning of compounds
in the Gothic and Old Greek, where, according to the oldest
principle, the naked theme is required; as, nov, “novus,”
appears in many compounds as novo (novo-grad, “ new-
town”), but is then not to be considered as the neuter
novo, “novum,” but as the commo» theme [G. Ed. p. 343,]
of the masculine and neuter, in which as yet no difference of
sex is pointed out. The clearest proof that the class of nouns
under discussion corresponds to the Indian, Lithuanian, and
Gothic nouns in a, is afforded by their feminine bases in a
(for st 4); so that to the form rab (for rabo), “ servant,”
corresponds a feminine raba, “a maid”: that is to say, all Old
Sclavonic primitive adjectives, i.e. those with an indefinite
declension, correspond to the Sanskrit in a-s, d, a-m, Greek
o-s, (a), o-v, Latin u-s, a, u-m; much as one might be led
astray by outward appearance to seek in the adjectives, which
in the nominative masculine end in y (yer), and in the neuter
in e, as siny, “ ceruleus,” syne, “caeruleum,” an analogy to
Latin adjectives like miti-s, mite.
258. But I recognise in adjectives like that just men-
tioned, and in similarly-constituted substantives, as knya¢y,
“prince,” more, “the sea,” bases of such a nature as, with-
out the euphonic form mentioned at §. 255. (n.), must have
terminated in yo, whence ye; and hence, in the nominative
masculine—according to the suppression of the final vowel
of the base, y in this case—and in the neuter e retaining
the vowel and dropping the y. These. bases, therefore,
correspond to the Indian in q ya, the Greek and Latin in
the Carniolan, before all inflections beginning with m in the three num-
bers, as posla-m, “ through the domestic,” posla-ma “the two domestics.”
This word appears to be identical with q¥ puéra, “son,” Persian pisar
“son,” “boy,” “ young man,” and to owe its meaning to familiar address.
10, iu (ryto-¢, ayto-v, sociu-s, proeliu-m); that is to say, serdze

(nominative and accusative neuter), “heart,” corresponds to
the Sanskrit ¢qaq hridaya-m, which is likewise neuter.
The feminines, again, afford a practical proof of the jus-
tice of this theory, for the Sclavonic bases in ya correspond
to the Sanskrit feminine bases in gt yd Greek sa, Latin
ia); and this form, in the uninflected nominative, stands
opposed to the masculine termination y and neuter e, as
sinya, “ cerulea,” to siny, “ ceeruleus,” and sine, “ceruleum.”
[G. Ed. p. 844.) When an ior other vowel precedes the last
y but one of the base, the y in the nominative, and accusative
masculine is changed into the vowel 7; as, nyetit, “ nepos ex
. sorore” (Dobrowsky, p. 282). The corresponding feminine
form is iya, and the neuter ye, the y of which has arisen from
i of the form tye, which is to be supposed the original, after
dropping the last y but one. To the Sanskrit warq savya-s,
Wat savyd, Waya savya-m (sinister, a, um), correspond thus
shti, shtya, shte (compare Dobrowsky, p. 285).
259. The Old Sclavonic masculine and neuter bases in yo,*
with their feminines in ya, are, according to their origin,
of four kinds :—1l. Those in which, as in SHUYO=aaqy
savya, both the semi-vowel and the vowel following, from
the earliest period of the language, belong to the base of
the word; and this case is perhaps the most rare.
2. Such as originally end in i, to which an unorganic o
has been added; as, in the Lithuanian, the bases in i, in
many cases, change into the declension in ia (ie) (§. 193.
and p. 171, Note *). To this class belongs MORYO, nom.
more, “ the sea,” the e of which therefore differs widely from
* Where I fix the theme, I leave the euphonic law contained in
§. 255, (n.) unregarded, and I give SERDZYO as the theme of serdze
(“heart,” nom. acc.), although the latter is no other than the theme
modified according to that euphonic law, i.e. without inflection, as in
the Sanskrit vdch is laid down as the theme, although ch cannot stand at
the end of a word, but passes into #, as in the nominative vék, which is
properly identical with the theme.
the mare in Latin, corrupted from mari; so that the

Sclavonic y, which again makes its appearance in the geni-
tive morya, dative mory#, corresponds to the Latin e spoken
of. The Latin word must, however, in order to be
classed with the Sclavonic, be pronounced in the nominative
mariu-m. Neuter bases in i, without an unorganic augment,
are entirely wanting in the Sclavonic. [G. Ed. p.345.]
Among the masculines of this class of words chervy, “a worm”
(theme CHERVYO), answers to the Sanskrit @f& krimi
and the Latin VERMI, Old High German, WURMI ; and
gyaty (ZYATYO), “gener,” to the Sanskrit wife jati
feminine, “familia,” “genus,” from #4 jan, “to be born.”*
The third kind of bases in yo is that where the unorganic y
precedes a final 0, according to the euphonic disposition
mentioned in §. 255. (n.). So giisy (GUSYO) corresponds to
the Indian ¢a hasisa, “goose” (§. 255. g.). In the fourth
place there exist among bases in yo the words in which the y
as well as the following vowel is an unorganic addition.
Thus fnouns of agency in TARYO correspond to the
Sanskrit in az tar (¥ tri, in the strong cases aTt tdr,) to the
Latin in tér, and to the Greek in typ, twp; hence the nomi-
natives my-tary, schi-tary, and ¢latary (Dobrowsky, p. 295), and,
with y for a, pas-tyry, “shepherd.” Of this kind, also, are
the nouns of agency in TELYO, the / of which is clearly
an interchange with r (§. 20.), so that this suffix also con-
forms itself to the Sanskrit we far ; hence the nominatives
blago-dyetely, “beneficus,” pye-tely, “a cock,” from the root
pye, “to sing,” schately, messor,” spas-i-tely, * salvator.”t
* ¢ frequently answers to the Sanskrit = Jj, and for example §1 jnd,

“to know,” is in the Sclavonic {na (infinitive (nati).
+ But see p. 879. Note §.647.
} As these words stand in analogy with the infinitive in ¢i,in so far
that their suffix begins with a like consonant, Dobrowsky (pp. 292, 293)
derives them from the infinitive, and allows them simply ely as suffix (as
also simple ary for ary), as it has been the custom to derive also, in
the Latin, tor and turus from the supine. However, it is certain
the
260. To the Sanskrit feminine bases in a1 4 correspond

as has been already remarked, Old Sclavonie in a, To
[G. Ed. p.346.] this class of words, however, belong also
some masculines, particularly proper names, which are then
declined entirely as feminines, as in Latin nauta, ceelicola ,&c.
(§. 116.), on which we will not here dwell further. Among
the bases in i there are, in Old Sclavonic, no neuters, and only
a very small number of masculines—as in Lithuanian—
which -Dobrowsky, p. 469, represents as anomalous, as
though they were only irregulars of his second declension
masculine: they are, however, in reality, foreign to it, for
this very reason, that they end their theme with i, but
the former with yo, and in part with yy, (§.263.). It is only
in the nominative and accusative singular that these three
classes of words, from various reasons, agree; and, gosty,
“ouest,” from GOSTI* (Gothic GASTI, Latin HOSTT)
agrees with knyacy, “prince,” from KNY AGYO, and vrachy,
“medicus,” from VRACHYY. The masculine bases origi-
nally ending with n—there are but a few of them—form ~
most of their cases from a base augmented by i; KAMEN,
“stone” (Sanskrit wya#4_asman), is extended to KAMENT
and then follows GOSTJ.
261. To the Sanskrit feminine bases in ¥ i correspond
numerous Old Sclavonic bases of a similar termination
(Dobrowsky, decl. fem. 1v.); that is to say, the Sclavonic
agrees with the Sanskrit in the formation of feminine ab-
the suffixes JOR, TURU and the Sclavonic TARYO, TELYO, used to
borrow their ¢ not at first from another syllable of formation so com-
mencing. They form primitive words from the roots themselves, and not
derivatives from other words.
* Thus, also, PUTT, “a way” (Sanskrit afaa pathin), and LYUDI, pl.
num, nom. lytidy-e, “people,” Gothic LAUDI, nom. lauths, “a person,’ the
au of which, according to §.255. (f:), is represented bya (#), and, according
to §.255. (m.),has gainedaprefixed y. GOSPODI, “a master” (comp. fa
pati, Lithuan PATT and Gothic FADIJ) is in fact irregular, as it passes
into several kinds of theme in its declension.
stracts in TJ, as PA-MYA-TI, “memory,” nom. pamyaty,

from the root MAN, as in Sanskrit af. mati (for manti),
“ spirit,” “ meaning,” from #4 man, “to think “* (compare
memini). These words weaken, indeed, in [G. Ed. p. 347.]
the nominative and accusative, their i to yer, but in no case
overstep their original base by an unorganic addition; and
hence they must not, on any account, be looked upon as of the
same base with the majority of masculines terminating simi-
larly in the nominative and accusative singular. But
Dobrowsky’s third feminine declension is of a mixed nature
(zerkovy, “a church”): in this we recognise some words
which have, by Guna, changed a Sanskrit final & @ to ov;
and from this form several cases, as from a base ending with
a consonant—e.g. zerkv-e, genitive singular and nominative
plural—but so that the o is suppressed béfore vowel termina-
tions. In some cases the theme extends itself by an un-
-organie i, in others by a; and also before these exten-
sions of the base the o of the syllable ov is suppressedt;
e.g. zerkviy-d, “per ecclesiam,” zerkvi, “ecclesia,” zerkvii,
“ ecclesiarum,” zerkva-m, “ ecclesiis,” zerkva-ch, “in ecclesiis,”
zerkva-mi, “per ecclesias.” The dative locative zerkvi is
doubtful, as this case could have no other sound than
zerkvi, whether it come from ZERKOFP or from ZERKYTV I.
* Dobrowsky (p.355) imputes, in my opinion wrongly, the n of po-
myant, “I remember,” and some similar bases, to derivation, instead of
supposing that the radical is suppressed before ¢, in analogy with the
Sanskrit, and as, in Greek, rdois from TAN, Sanskrit afre tati-s, “a line”
(as extended), for afera tanti-s,
+ The example given by Dobrowsky, zerkory, “a church,” nevertheless
does not apply to monosyllables, as krovy, “ blood” (Sanskrit yay kravya,
neuter, “fiesh’”’), nor to those polysyllables in which two consonants
precede thesyllabie ov ; for yatrrach and krvach would be equally imprac-
ticable (comp. Gretsch by Reiff, p. 163). Brovy, “eyebrow,” also appears
to form all its cases from a theme BROVI, an extension of the Sanskrit
¥ bhri, feminine, by the addition of i,with a Guna of the q%. The
nominative plural is hence brovi (Dobrowsky, p. 115), not brov-e.
Some words of this class have, in the nominative, y, and

[G. Ed. p. 348.] thus svekry agrees with Waa swasri-s,
“ sucrus” (§. 255. ¢.); others have, at will, ovy or vi, with
o suppressed ;hence zerkovy or zerkvi.
262. Among bases in wu (Greek v) of the cognate lan-
guages only masculines have maintained themselves in the
Old Sclavonic. They, like the bases in 0, suppress their
final vowel in the nominative and accusative, but in the
remaining cases this letter shews itself either with Guna
changed to ov or @ (§. 255. f.), or without Guna, as o
(§. 255. ¢.); and in the latter form it appears also in the
beginning of compound words as a naked theme. Hence
it is more probable, that anciently for syn, “filius,” “filium,”
stood syno rather than syny (§. 255. c.).* With this simi-
lar conformation of theme of the old bases in a and uy, it
is not surprising that two kinds of bases, which in their
origin are widely different, run very much into one another
in the Sclavonic declension; and that, in the more modern
dialects, these two declensions, which were originally so
strictly separate, have fallen almost entirely into one.
263. As in the o bases which have arisen from Wa, a y
preceding introduces a difference of declension, which we,
in §. 258., have represented as purely euphonic, the same phe-
nomenon makes its appearance also in the y bases, by means
of which their Guna form is articulated ev (for yev) instead
* We term this class of words, nevertheless, bases in y; for although

their final letter never occurs as y, still, according to }. 225. (¢.), y is the
most legitimate, even if it be the most rare, representative of the Sanskrit
Zu. But should it be wished to call them bases in 0, they would not be
distinguished from the order of words, which, according to §. 257., bear
this name with more right. The term u bases would be appropriate only
so far as here, under the u, might be understood, not the Old Sclavonic x
(etymologically
= 6), but the Sanskrit g uw or the Latin u of the
fourth declension, which, in the Old Sclavonic, has no real existence.
of ov.* If, however, with Dobrowsky, we di- (G. Ed. p. 349.]

vide the Old Sclavonic masculines—with the exeeption of the
bases in i, §. 260.—into two declensions, and in doing this de-
sire, as is natural, to ground the division on the final letters of
the bases, we must place knyaéy, “ prince” (nominative) of
Dobrowsky’s second declension in the first, and by the side
of rab, “a servant”: on the other band, the words syn,
“son,” and dom, “a house,” of Dobrowsky’s first masculine
declension must be transferred to the second declension
as mutilated y forms. Of the paradigma here given by
Dobrowsky, vrachy, “medicus,” adheres most strictly to the
true y declension, and, according to §. 255. (n.), opposes
ev to the ov of SYNY. On the other hand, words inflected
like zary, “a king” (nominative), clearly form the nomi-
native and genitive plural from bases ini; hence zary-e,
“kings,” zarii, “of kings,” from ZARI; as gosty-e, “ho-
spites,” and gostii, “ hospitum,” from GOSTI. In the dative
plural and instrumental singular the form zare-m is doubt-
ful: in this and other words, also, of obscure origin, it re-
mains uncertain whether the more contracted theme in 3,
or the more extended in YY; is the older; but it is certain
that several old i bases have migrated into this declension
by an unorganic addition; for instance, ogny, “ fire” (nom.),
dative ognev-i, from OGNYY, agrees with the Sanskrit sftq
agni, Latin TGNJ, Lithuanian UGNI+ It (G. Ed. p. 350.]
* Without Guna, the final of the base is pronounced e for ye from yo
(§. 255. n.); and hence, in the cases without Guna the yy bases are just
as little to be distinguished in their inflection from the yo bases, as, in
the instrumental singular, syno-m (from the theme SYNY) from rabo-m
(theme RABO). In the beginning of compound words, also, the yy bases
end like those in yo, with e for ye.
+ As regards words inflected like mravii, the only proof which could
bring them under the head of the y bases is the vocative sing. mraviyt :
that they, however, although they have borrowed this case from the y
declension, originally belong to the o declension, is proved by their
feminine in iya and neuters in iye or ye (Dobrowsky, p. 282).
a.
deserves here to be further remarked, that in the more modern

dialects of the Sclavonic stock, the two masculine declensions
here spoken of have been transfused almost entirely into one,
which has taken several cases regularly from the old wu
declension, in which, however, from the point of view of
the more recent dialects, e.g. in the genitive plural of
the Polish and Carniolan, ov, ow, form an exception as a
case termination. In the Old Sclavonic, also, rab (theme
RABO), “a servant,” may optionally form several cases from
a theme RABY (for rabii); and for rab, “ servorum,” we
may also have ,rabov: and in the nominative plural of
this class of words we find also ov-e, according to the
analogy of synov-e. On the other hand, the adjective
masculine o bases (the indefinites) of the y declension have
admitted no irregular trespassings any more than the
pronouns.
264. Bases ending in a consonant are, under the limi-
tation of §. 260., entirely foreign to the masculine: on the
other hand, there are neuter bases in en, es, and at (
which are important for the system of declension, because
the case suffix, commencing with a vowel, divides itself so
much the more distinctly from the base ending with a
consonant. The bases in en correspond to the Sanskrit
in Wa an, and have preserved, too, in the uninflected
nominative, accusative, and vocative, the old and more power-
ful a, but with the euphonic prefix of a y (see §. 255, n.),
and with the suppression of n of the base (see §. 139.).
All of them have an m before the termination en; so that
men is to be considered as the full formative suffix of the
word, which answers to the Sanskrit HA man—e.g. in @Aq
karman neut., “deed”—and to the Latin men; that is to say,
SYEMEN (nominative syemya, “seed,” from the base sye)
answers to the Latin se-men; and imen, “a name,’ is a
mutilation of ata ndman, “nomen.” The bases in es
answer to the Sanskrit neuter bases in as, as_nebes,
“heaven,” Sanskrit =pta nabhas. In the (G. Ed. p.351.]

nominative, accusative, and vocative, they relinquish the con-
cluding s (according to §. 255. l.), and afterwards strengthen
the e to o (§. 255. a.). We cannot, therefore, any longer com-
pare the o of nebo with the Sanskrit-Zendian o, which has
arisen out of a+u. As in this abbreviation of es to o the
neuter es bases in the cases mentioned become similar to the
o bases, it is then—on account of the influence of these cases,
and because the nominative principally gives the tone in
the declension, and shews in the oblique cases as inflec-
tion that which is in itself deficient,—it is then, we say,
not surprising, if the original o bases at times admit an es
in the oblique cases, particularly when we consider the ori-
ginal great extension of these neuter bases terminating in s
(compare §. 241.), which induces the conjecture, that many
words, now declined as o bases, were originally domiciled in
the bases in es. On the other hand, Dobrowsky proves that
there is no admixture of es in the thoroughly legitimate
adjective o bases. [t is also clear, from §. 255. (2), that
the bases in yut* in the uninflected cases must lay aside
the #, and follow cya, not aea mahat (“ magnum™) and
caput.
265. Of the class of words in r mentioned in §. 144. two
feminine words have remained in the Old Sclavonic which
derive most of their cases from the genuine r bases, but
in others increase the original base by an unorganic i, or
also by ya (compare the Lithuanian in §. 144.): in the nomi-
native singular, however, in accordance with the Sanskrit and
Lithuanian, they suppress the r. These are, mati, “mother,”
and dshchi, “daughter”; in the latter only occurs the increase
of the base by ya (in the nominative accusative and dative
plural); the declension ofthe former springs [G. Ed. p. 352.]
* They are all derivatives from names of animals, and denote the
young of the animal mentioned.
partly from MATER, e.g. mater-e, “ matris,” and matres

(uarép-es), partly from MATERL, e.g. matery, “ matrem.”
266. *In order now to pass over to the formation of
cases, the nominative and accusative have lost the case-
signs s and m, with the exception of the bases in a, which
present in the diphthong @ (g), a contraction of the vocalized
nasal with the final vowel. of the base shortened to a, (see
§. 255. g.); hence vodd, “ aquam,” from vodo-i. The instru-
mental has, in the feminine, and the pronouns which have
no gender preserved the genuine Sanskrit inflection; but
it is to be remarked of the feminine bases in é that they
change this vowel before the termination @, (for 4, see
§. 255. h.), not into simple y, but into iy; so that in this
respect the Old Sclavonic agrees more closely with the
Pali, which, in the corresponding class of words, changes
the final i before all the vowel endings into iy, than with
the Sanskrit. Hence, let kostiy-d, from KOSTI, “bones,”
be compared with the Pali uifrat pitiy-a (from piti, “ joy”),
for the Sanskrit what prify-d. Masculines and neuters have
mt for their instrumental ending; and this is, I have no
doubt, an abbreviation of the Lithuanian mi, and comes there-
ore from 6i (§. 215.).
267. The dative has, in the singular, a common ending with
the locative, and, in fact, the Old Sanskrit i (§. 195.); hence,
imen-t, “in nomine,” and “ nomini”; synov-i, “ filio,” brachev-i,
“medico,” fron SYNY and BRACHYY (§. 263.), with
Guna.f If the case-sign is suppressed, the preceding ov
[G. Ed. p. 353.] becomes é, and ev (from you) becomes yd;
hence, also, synd, “ filio,” with synov-i, and zaryil, “ vegi,” with
* Cf. §. 783!.
+ For m, according to Dobrowsky, we should read Mb my.
} Hence I am now disposed, contrary to §.177., to assume for the
Lithuanian a common origin for the two cases, although in their received
condition they are externally separated from one another, as is the
case in Old Sclavonie, also, in several classes of words.
the y bases, but prefer, however, the abbreviated form d,
hence rabi, from RABO, more rarely rabov-i. The o bases
of the adjectives, and of these there are, in the mascu-
line and neuter, only o bases, and those of neuter substan-
_ tives have alone the uninflected form in @; hence, e.g.
blagi, “bono,” masc. neut. ; sinyd, “ ceruleo,’ masc, neut.:
slovil, “verbo,” moryt, “mari”: not blagov-i, sinev-i, slovor-i,
morev-i. In masculine names of inanimate things this
uninflected form in @ extends itself also to the genitive.
and locative; hence domé, “of the house,” “to” and “in
the house”: but in the dative is also found domov-i, and in
the locative domye.* The pronouns of the 3d person mas-
culine and neuter—with exception of the reflexive—have
in the dative, in like manner, the uninflected @; for the
form mi in to-mé, “to this,” is clearly from the Sanskrit
appended pronoun @ sma (§. 165. &c.), which has extended
itself in the cognate European languages so much, and
under such different forms, and this, in the Old Sclavonic,
would necessarily give the base SMO, from which, after
dropping the s, would come the dative md, as rabé from.
RABO.
268. While the o bases, as has been shewn above, have
borrowed their dative from the y declension, the y bases
appear, in the locative, to have intruded on the o class;
for synye answers to rabye, from RABO from RABA
(§. 255. a.); but the ye of rabye is, according to §. 255. (e),
clearly from the Sanskrit & é of ¥& vriké from ym vrika,
and answers to the Lithuanian wilké from ([G. Ed. p. 354.)
WILKA (§.197.). As, however, in Lithuanian, from SUNU
comes sunu-ye, so may also the Old Sclavonic synye require
- * Masculine names of inanimate things all follow the declension of dom

(theme DOM Y), although very few among them, according to their origin,
fall into the class of the old ¥ u, i.e. of the Latin fourth declension, but
fer the most part correspond to Sanskrit bases in W a.
Z
to be divided into syn’-ye ;and this is rendered the more pro-

bable, as the feminine a bases, also, have in the locative ye
for a-ye ; hence vod'-ye, “in aqua,” from VODA, answers to
the Lithuanian ranko-ye (for ranka-ye) from ranka.* In bases
in i, masculine and feminine, it might appear doubtful
whether i, with which they end in the dative and locative—e. g.
. pati, “in the way,” kosti, “in the bone”—is to be ascribed
to the theme or to the inflection: as, however, in the
genitive, (to which belongs an i, though not« through any
inflection), they have just the same sound, and otherwise
never entirely give up the 7 of the base, except in the in-
strumental plural, it is more natural to consider the forms
puti, kosti, uninflected, just like domd, “in the house.” We
may also look upon the i in the dative and locative of those
bases, which have y as the last letter but one, as nothing
else than the vocalization of this y; the ¢ therefore, of
knyati, mori, brachi, voli, represents nothing else than the y
of the masculine bases KNYACYO, VRACHYY, and of
the neuter MORY0O, and feminine VOLYO.
269. In the genitive the terminations as, os, is, which
in the cognate languages, are joined to bases ending with a
consonant, must, according to §. 255. (/.), drop the s, but the
[G. Ed. p. 355.) vowel appears as e¢in all the bases ending
with a consonant (§§. 260. 264.): hence imen-e, “of the name,”
* It must be allowed that here occurs the very weighty objection, that
the f minine form rankoye in the Lithuanian, and vodye in the Sclavonie,
might stand in connection with the Sanskrit §TqT= dydm in
Jjihwdy-dm (5. 202.) ; so that, after dropping the m, as in the Zend (§. 202.),
the preceding vowel, which in the Zend is already short, would, through
the enphonic influence of the y, become e. As the bases iné in the
Lithuanian, down to a few exceptions, are feminine, so might also awiye
from awi-s, “a sheep,” be divided into awiy-e, and compared with aaa
maty-dam, from mati or fray bhiy-dm from bhi (comp. in §.266. kostiy-é,
for kosty-i, from KOSTT).
answers to alata ndimn-as, nomin-is ; nebes-e, “ of the

heaven,” to #71aa nabhas-as, vépe(s)-os; muter-e to matr-is,
fytpés. The pronominal forms also follow this analogy:
‘men-e, “mei,” teb-e, “tui,” seb-e, “sui,” because, in the
oblique singular cases, MEN, TEB, SEB are their themes.
We recognise the fuller Sanskrit genitive ending @& sya in
-the pronominal genitive termination go, as to-go = wet ta-sya
(§. 188.). This comparison might alone be sufficient in place
ofall proof; but, over and above, is to be remarked the easily
adopted hardening of the semi-vowel y to g (comp. p. 121
G. ed.), and in the Prakrit to ¥ 7 (8. 19.); finally, let the
high degree of improbability be considered, that the Sclavonie
should have formed an entirely new genitive termination,
foreign to all the cognate languages. Now, if the g of the
termination go is taken for a hardening from y (q y), then
the Old Sclavonic has preserved exactly as much as the
Greek of the termination sya; and go answers to the
Greek 10, and to-go, “hujus,” to the Greek ro-fo. As,
however, in Sclavonic, the sibilants are easily interchanged
with gutturals (see §. 255. m.), one might also conjecture
the g of go to be a corruption of the Sanskrit s and the
semi-vowel of @& sya, which had been lost. This conjec-
ture cannot entirely be put aside; but in any case, even in
this supposition, the termination go remains connected with
w@ sya and so. As, however, in the Old Sclavonic, g is else-
where exchanged only. with € and sch (Dobr. p. 41), but not
with s, in my opinion the derivation of g from y (q y) is
to be preferred to that from s.
270. The substantive and adjective (indefinite) o bases,
in disadvantageous comparison with the pronouns which
hold fast the old form, have lost the genitive termination go;
but for it, in compensation for the lost termi- [G. Ed. p. 356.1
nation, they have retained the old a of the base, instead of,
according to §. 255. (a.), weakening it to 0; hence raba, “servi.”
nova (=Sanskrit nava-sya) “novi.” Now, although the y bases
z2
in the genitive end in a, the comparison of the form syna, “filii,”

with the Lithuanian and Gothic sunadé-s, sunau-s, and the
Sanskrit séind-s (from sénau-s), teaches that the a here is only
a Guna element, but foreign to the proper base, as well as to
the case-suffix, which, according to §. 255. (b.), must disappear.
271. The feminine bases in a, with the exception of
those which have a penultimate y, change that a in
the genitive into y; hence vody, “aque,” from VODA,
but volya, “voluntatis,” with unaltered base, from VOLYA.
I ascribe that y, as well as that in the nominative plural, to
the euphonic influence of the s, which originally ends the
form (see §. 255. d.): this, however, does not obtain if a y
precedes the a; hence volya, “voluntatis,” is identical with
the theme. On the other hand, the feminine pronominal
bases in @ have preserved a remarkable agreement with
the Sanskrit pronominal declension; for if ta, “this” (at
the same time the theme), forms to-ya in the genitive, I do
not doubt of the identity of the ending ya with the San-
skrit syds (§. 172.), as in the word weqra tasyds, of the same
import, for the final s must, according to §. 255. (/.), give
way; but the a of the Sclavonic ya directs us, according
to §. 255. (a.), to an Indian “t 4, just as the preceding o
points to a short wa. The irregularity, therefore, in the
shortening of the Sclavonic termination lies only in the drop-
ping of the sibilant before y, as, in the Greek, roto, from
7 ta-sya, and in the to-go, for to-(s)yo, mentioned in §. 269.
272. In the vocative, which in the cognate languages
is without any case suffix (§. 204.), o is weakened to e (e) and
a to o (§. 255 a.), hence nove (from NOVO, “new”), for
[G. Ed. p.357.] Sanskrit #4 nava, is identical with the Latin
nove, and answers to the Greek vé(F)e: from VODA, “water,”
comes vodo ; but from VOLYA, according to §. 255. (n.), vole
for volyo: and so from KNYACYO, “prince,” knyashe* for
* ¢ before e becomes sh.

Bases in yy change their y by Guna to a (§. 255. f.),

in analogy with §. 205.; hence vrachyé—more commonly,
with y suppressed, vrachii—* medice!” from VRACHYY On
the other hand, y bases without y for their penultimate letter
commonly omit the Guna, and weaken their final vowel,
like the o bases, to e; hence syne, “oh son!” more rarely
syn (Dobr. p. 470), =Gothic sunau, Lithuanian sunail, San-
skrit stinéd from sunau.
DUAL.
273. By preserving a dual, the Old Sclavonic surpasses

the Gothic, in which this number is lost in the noun:
it exceeds, in the same, the Lithuanian in the more true
retention of the terminations, and it is richer than the
Greek by one case. The agreement with the Sanskrit
and Zend is not to be mistaken: let the comparison be
made.
SANSERIT. ZEND. OLD SCLAVONIC.
N. Acc. V. m. ubhd (ambo Védic), ubd, oba.

f. n. ubhé, ubé, obye (§. 255. n.).
I. D. Ab. m.f.n. ubhé-bhydm. ubéi-bya, 1. D. obye-ma(§. 215.)
G. L. m. f. n. ubhay-és, ubdy-6, oboy-t.t
* The ye, which precedes the termination ma, may be compared with
the Sanskrit é in plural forms, as FaTe vrikébhyas: ye-ma, however,
occursin the Old Sclavonie only in dvye-ma, “ duobus,” “per duos,” and
some pronouns. The usual form of substantive o-bases before this ending
is that with an unchanged 0, as sto-ma, from sto, “a hundred”; and the
final a of feminine substantives also remains unchanged, as dyeva-ma, from
DYEVA, “agirl.”
T The form %, for the Sanskrit ending 6s, is, according to §. 255. (/-)
and (/), necessary: the Zend certainly approaches the Old Sclavonic in
casting away the s voluntarily. The oy, which precedes the termina-
tion #, clearly corresponds to the Sanskrit Wy ay (see §. 225.) and the
Zend
[G. Ed. p.358.] The Sanskrit ubhé, as neuter, comes, ac-

cording to §. 212., from the theme ubha, in union with the
case-suffix 7; and the feminine ubhé is an abbreviation of
ubhay-du, and is therefore without a case termination (§. 212.).
The Old Sclavonic, which runs parallel to the Sanskrit in
both genders, and, according to §. 255. (/.), opposes ye to the
Indian @é, no longer recognises the origin of this ye, and
regards it entirely as a case-suffix before which the final
vowel of the theme appears to be suppressed. Therefore,
also, neuter bases ending in a consonant make ye their
termination, if the imenye, “two names,” given by Do-
browsky, p. 513, actually occurs, and is not a theoretic for-
mation. In feminines, however, the termination ye extends,
exactly as in Sanskrit, only to bases in a (for Sanskrit 4,
§. 255. a.); but in such a manner, that those with’ y as the
last letter but one in the theme reject the termination ye,
and vocalize the y of the theme; hence dyevye, “two girls,”
from dyeva, but steCi, “two steps,” from STECYA. The
feminine bases in i, in the dual case under discussion,
answer to the Sanskrit and Lithuanian forms mentioned
at §§. 210. 211., as pati, “two sirs,” from fa pati;
(G. Ed. p.359.] auiz, “two sheep,” from AWI; only
that, according to §. 255. .), the i in the Sclavonic is not
lengthened; as dlani from DZANI (nominative singular
Zend éy or ay (see p.277); but that occurs only in dvoy-é=Sanskrit

dway-és, “of two,” “in two” m.f.n., and in toy-d=Sanskrit tay-ds,
“ of these two,” m.f.n. The genitives and locatives of the two first persons
also rest on this principle, only retaining the older a—nayt, viyit.
For the rest, however, the final vowel of the theme is rejected before
the termination %, as st’-% (Sanskrit shatay-6s) from STO, ‘a hundred,”
dyev-4 fram DYEVA, “a girl”; and thus occurs, also, together with
dvoyi, the syncopated form'dvi. Although the Lithuanian generally
does not drop the final s, still the @ mentioned in §, 225. may be identical
with the Sclavonic 4; as in the Zend, also, in this termination the ¢ is
often dropped.
dlany), “vola manus.” On the other hand, the masculine y

bases do not follow this principle, but suppress the final
vowel before the case-suffix a; hence syn-a, “two sons,”
from SYNY.
PLURAL.
274. In the plural, the masculine nominative termina-
tion e (e) for the most part answers to the Greek eg, and,
according to a universal rule of sounds, omits the s
(§. 255. L); hence synov-e, “the sons,” @q7@ siénav-as:
compare fézpv-es, kamen-e, “the stones,’ for wWyATTe
asmén-as (§.21.); compare daiuov-es, gosty-e, “guests” (theme
GOST 1), for the Gothic gastei-s,and Greek forms like récv-es.
The bases in o take, as in Lithuanian do the corresponding
bases in a, i as their termination (see §. 223.), but before
this reject the o of the base; hence rab’-i, “servants,” for
rabo-i (comp. AvKo-:), as in Latin /up-i for lupo-i. Neuters
have a for their ending, like the cognate dialects, with the
exception of the Sanskrit with i for a; nevertheless, slova,
“verba,” from SLOVO—as ddoa from AQPO—answers to
Védic forms like vand, “ woods,” from vana; and the same
thing obtains which, §. 231. p. 267 G. ed., has been said of
Gothic, Greek, and Latin, regarding the relation of the a of
the termination to the o of the theme. As regards the bases
ending in a consonant, let imen-a, “names,” be compared
with the Latin nomin-a and Gothic namén-a; nebes-a, “ the
heavens,” with vege(c)-a; and telyat-a, “ calves,” with Greek
forms like cdpat-a. Feminines, with the exception of the
class of words in ov mentioned at §. 261., have lost the no-
_ minative ending; hence volya, “voluntates,” is the same as
the theme and the nominative singular; and [G. Ed. p. 360.]
from KOSTI, “bones” (Sanskrit asthi, neuter) comes the
nominative singular kosty, and the plural like the theme.
275. The accusative plural is, in feminine and neuter
nouns, the same as the nominative, and therefore in the former
mostly without inflection, exactly as in the few masculine

bases in 27; hence gosti for the Gothic gasti-ns. Bases in 0,
without y preceding, like RABO, change this o into y, as
raby, “servos”; at least I cannot believe that this y is to
be looked upon as the case-suffix; and I pronounce it to be
the euphonic alteration of the o of the base, through the
influence of the consonant of the inflection which has
been dropped (comp. §. 271.): as in Lithuanian, also, the
corresponding class of words often changes the final vowel
(a) of the base into u; hence wilki-s, “lupos,” answering
to the Gothic vulfa-ns and Sanskrit vrikd-n. But if the
Old Sclavonic bases in y, of animate creatures, form
owy in the accusative plural, and thus synovy, “ filios,”
answers to the Lithuanian suni-s (from SUNU), this
very Lithuanian form, as well as the Gothic and Sanskrit
sunu-ns, FAA sini-n, prove that the Sclavonic form is
unorganic, and formed from an augmented theme SY NOVO,
according to the analogy of raby. Bases in yy in this case
follow bases in yo (from ya, §. 255. a.), which, preserving the
old a sound, give ya, as in the genitive singular (see §. 270.);
hence vrachya, “medicos,” like knyatya, “principes”: but —
forms, also, like doschdevy, analogous with synovy, occur, fol-
lowing the euphonic rule, §. 255. (n.).
276. The view here given is the more incontrovertible,
as in the dative, also, synovo-m, “ filiis” (compare rabo-m),
is clearly formed from a theme SY NOVO, increased by 0,
corresponding to the Lithuanian sunu-ms. This dative
suffix m, for the Lithuanian ms (from mus, §. 215.), according
[G. Ed. p.361.] to §. 255. (/.), extends itself over all classes
of words, and appears to be attached by a conjunctive vowel
e to bases terminating with a consonant; but, in fact, it is
to be considered that these, in the cases mentioned as also in
the locative (see §.279.), pass over into the é declension, as
a final i, before the signs of case m and ch, becomes e: and a
similar metaplasm occur's in the Lithuanian, and indeed, to a
much greater extent (§. 125. sub finem, comp. §. 126.); hence
imene-m, imene-ch, from IMENIJ from IMEN, “names,” as
koste-m, koste-ch, from KOSTT, “bones.”
277. Less general is the instrumental ending mi, an-
swering, subject to the loss required by §. 255. (/.), to the
Lithuanian mis, Sanskrit bhis, and Zend bis. This ter-
mination mi is, however, in masculine and neuter nouns
for the most part lost (comp. Dobr. pp. 473 and 477);
and is preserved principally, and indeed without exception,
in feminines, as well as in a few masculine 7 bases: a final
i of the base is, however, suppressed before the termina-
tion mi. Let kost’-mi be compared with afaira asthi-bhis,
from «feq asthi, “bone”; vdova-mi with fawarfva vidhavé
bhis, from fawat vidhavé, “a widow.” The instrumentals
raby, synovy, are, like the accusatives of similar sound,
uninflected (§. 275.); the i of knyat, vrachi, is the vocali-
zation of the y of the bases KYNACYO, VRACHYY,
after the loss of the final vowel; and the y of neuters
terminating in a consonant, like imeny “‘ per nomina,” is to be
explained by a transition into the o declension, and is there-
fore analogous to raby, slavy, similarly to the o of the Greek
dual forms like da:pudvor (p. 318 G. ed. Rem. 2.).
278. Dobrowsky (p. 461) represents ov, y, ii, ev, en, yat,
and es, as plural genitive terminations; but in reality the
_ suffix of this ease has entirely disappeared, and in bases in 0,
a, and y, has also carried away those final vowels with it, while
bases in i double that vowel; hence rab, [G. Ed. p. 362.]
“servorum,” from RABO; vod, “aquarum,” from TODA; syn.
“filiorum,” from SY NY; kostit, “ossium,” from KOSTT; imen.
“nominum,’ from IMEN ; nebes, “ celorum,” from NEBES.
The n and s of imen, nebes, would, without the former protec-
tion of a following termination have been dropped, as in
Sclavonic we have only a second generation of final conso-
nants; while the former, with the exception of a few mono-
syllabic forms, has, according to §. 255. (/.), disappeared.
279. The termination of the locative plural is ch

throughout all classes of words, and has been already, at
§. 255.(m.) recognised as identical with the Indian ¥ su,
and therefore, also, with the Greek o:: compare, also, the
Zend ayo kha, for the Sanskrit swa, in §. 35. Before
this kh, o passes into ye, exactly as the corresponding
Sanskrit W a into vé (see §. 255. ¢.); hence rabye-ch,
servis,” answers to yaq vriké-shu, “in lupis.” Bases in yo—
and those in yy follow their analogy—suppress, however,
before this ye, their preceding y, as in similar cases;
hence knyatye-ch, “in principibus,” not knyatyy-ch from
KNYACYO. A final a remains unchanged ; hence vdova-ch,
“in viduis,” answers to the Sanskrit vidhavd-su. For bases
in i, and consonants, see §. 276.
280. For an easier survey of the results obtained for
the Old Sclavonic case-formation, we give here, in order
to bring under one point of view all the kinds of theme
existing in Old Sclavonic, and to render their comparison
with one another easy, the complete declension of the
bases: RABO, m.“a servant,” KNYACYO, m. “a prince,”
SLOVO, n. “a word,” MORYO, n. “a sea” (Dobr. p. 476,
§. 11.), VODA, f. “water,” VOLYA, f. “will,” GOST, mi.
“a guest,” KOSTI, f. “a bone,” SYNY, m. “a son,” DOMY,
m. “a house,” VRACHYY, m. “a physician,” KAMEN, m.
[G. Ed. p. 863.] “astone,” JMEN, n. “aname,” MATER, f.
“a mother,” NEBES, n. “heaven,” TELY AT, n.“acalf.”* In
* The above examples are arranged according to their final letters,

with the observation, however, that o represents an original short a, and
hence precedes the a for Sanskrit @ (§.255.a.). All bases in ¢ have a y
before the preceding a; this semi-vowel is, however, readily suppressed
after sibilants; hence ovcha for ovchya, Dobr. p. 475; and hence, also,
from lizyo come (nom. lize) the genitive, dative, and nominative accusative
plural liza, lizi, for lizya, lizyt. Ifin bases in yo, m.n., and in femi-
nines in ya, an 7 precedes the semi-vowel, this involves some apparent
variations
INTHE OLDSCLAVONIC. £347
those forms of the following table in which a part of the word
is not separated from the rest, thereby shewing itself to
be the inflection, we recognise no inflection at all, ie. no
case-suffix; but we see therein only the bare base of the
word, either complete or abbreviated; or also a modifica-
tion of the base, through the alteration of the final letter,
occasioned by the termination which has been dropped
(compare §. 271.). In some cases which we present in the
notes, base and termination have, however, been contracted
into one letter, by which a division is rendered impossible.
With respect to the dual, which cannot be proved to
belong to all the words here given as specimens, we
refer to §. 273.
variations in the declension, which require no particular explanation here

(see, in Dobr. mravit, m. p. 468 ;ladiga, f.p.478; and tichenye,n. p. 474,
With regard to zary, “a king,” see §.263).
348 FORMATION OF CASES. -
[G. Ed, p. 364.) ' SINGULAR.

THEME, NOM. Accus. INSTR, DATIVE. GEN. Loc.
RABO, m.' rab’, rab’, rabo-my, rabi,'8 raba,”' rabye,”
KNYACYO. m.” knyaty’, knyaty’, knyate-my, knyalyi, knyatya,' knyagi,
SLOVO, n2 _ slovo, slovo, slovo-my, _ slovii, slova,?! slovye,”>
MORYO,n2 more, more, _more-my, moryi, morya,"| —_mori,
VODA, f.4 voda, vodi,"® —_vodoy-ii,'® —vod'-ye,'® —_vody,™ vod’-ye,®
VFOLYA, f. — volya, volyi,” —voley-ii,'® voli, velya, voli,
GOSTI, m.> _gosty, gosty, goste-my,'" gosti,” gosti, gosti,
KOSTI, £2 kosty, kesty, kostiy~i,'® _kosti,® kosti, kosti,® —
SYNY,mé syn’, syn’, syno-my,!? synov-i, — syna,”" synye,> —syx
DOMY, mm.’ — dom’, dom’, domo-my, domov-i, domi, domi,
VRACHYY.m°S vrachy’, vrachy’, vrache-my, vrachev-i, vrachya,”® vrachi, ora
KAMEN,m.? kamy’* .... kamene-my, kamen-i, kamen-e,* kamen-i,
IMEN, n."° imya, imya, imene-my, imen-i, imen-e,2* imen-i, “a
MATER, f£.'' mati, aig Rol By eee es mater-i, _mater-e,* materi,
NEBES, n. nebo, nebo, nebese-my, nebes-i, nebes-e,** —nebes-i, =
TELYAT,n.* telya, telya. telyate-my, telyat-i, _telyat-c,** telyat-i, -
' Comp. p. 278, &e. ? Sce §§. 258.259 * Comp. pp. 275, 276. * Comp.p
5 Comp. p. 286. ® Comp. p. 288. 7 See p.337, Note. 8 See §. 26
* Comp. p.304. The cases wanting come from KAMENT (sce §. 260.); w
also, kamene-m, kamene-ch (§.266.); and whence, also, might be derived the¢
and locative kamen-i, which I prefer, however, deriving from the original theme,
as in MATER. 4
Comp. §.139. | See § 265. and comp. p.805. '? Comp. p. 806. and §. 147.
18 See §. 264. '4 Dobr. p. 287. '® See §. 266.
16 Comp. Sanskrit jihway-d, ke. See §. 266. ” Comp. Lith. pati-mi,
18 Or rabovi, §. 267. 19 See *. 268.
20 The i may also be ascribed to the mark of case, and the dropping of the final
of the base may be assumed ; but in the genitive of the same sound, the i clearly bell
to the theme.
21 See §. 270. : 2 See §. 271.
23 More commonly vracha, and in the vocative, vracht. See p. 347, Note.
% See }. 269. 5 See §. 268. % Or syne.
- IN THE OLD SCLAVONIC. 3419
PLURAL. [G. Ed. p. 365.]

nom. voc.! accus.* InsTR.° DATIVE. GEN. - LocaTive.®
rab’-i, raby, raby, rabo-m, rab’, rabye-ch.
knyaGi, knyalya, knyati, knyafe-m, knyaly’, knyate-ch
slova, slova, slovy, slovo-m, slov’, slovye-ch,
morya, morya, mori, more-m, mory’, morye-ch.
vody,* vody, voda-mi, voda-m, _vod’, voda-ch.
volya, volya, volya-mi, volya-m, voly’, ~ —volya-ch.
gosty-e, _gosti, gost’-mi, _ goste~m, _—gostit, goste-ch.
kosti, kosti, kost’-mi, —_ koste-m, kostit, koste-ch.
synov-e, synovy,' _—synovy,* —synovo-m,* synov, synovye-ch.*
domov-e, domy, domy, domo-m, domov, dome-ch,
vrachev-e, vrachya, vrachi, vrache-m,_ vrachev, vrache-ch.
rates a2 et owe ¢ kamene-m, .... kamene-ch.
imen-a, imen-a, imeny, imene-m, imen, imene-ch.
mater-eé, > ta yt mater-mi, matere-m, Shir eh ° é
nebes-a, nebes-a, nebesy, nebese-m, nebes, nebesye-ch.”
telyat-a, telyat-a, telyaty, telyate-m, telyat, tetyate-ctte
? See§. 274. 2 See §.271. 3 See§. 275.

* From SYNOVO, see §.275. In the locative occur also synovo-ch
and synove-ch.
5 See§. 277. 6 See§. 276. 7 See §. 278. ® See §.279.
9 One would expect nebese-ch ;but in this case ech and yech are fre-
quently interchanged with one another, and the form yech appears to
agree better with the preceding s (comp. Dobrowsky, p. 477).
350 ADJECTIVES.
ADJECTIVES.
LG. Ed. p. 366.] 281. The declension of the adjective is not

distinct from that of the substantive; and if some inflected
forms, which in the Sanskrit and Zend belong only to the
pronouns, have, in the cognate languages, emerged from the
circle of the pronouns, and extended themselves further, they
have not remained with the adjectives alone, but have
extended themselves to the substantives also. As regards
the Greek, Latin, and Sclavonic, we have already ex-
plained at §§. 228. 248. and 274. what has been introduced
from pronominal declension in those languages into
general declension: we will here only further remark that
the appended syllable sma, in §. 165. &c., which, in Sanskrit,
characterises only the pronominal declension, may in the
Pali be combined also, in several cases, with masculine
and neuter substantive and adjective bases, and indeed
with all bases in a, i, and u, including those which, origi-
nally terminating in a consonant, pass by augment or
apocope into the vowel declension; thus the ablative and
locative singular of késa, “hair,” is either simply késd
(from késdt, see p. 300), késé, or combined with sma or its
variation mha, késa-smd, késa-mhd, késa-smin, késa-mhi. In
the Lithuanian, this syllable, after dropping the s, has, in
the dative and locative singular, passed over to the adjec-
tive declension, without imparting itself to that of the sub-
stantive, and without giving to the adjective the licence of
renouncing this appended syllable; as, géram, “bono,” —
geramé, “in bono.” According to this principle it would
be possible, and such indeed was lately my intention,
ADJECTIVES. 351
to explain the agreement of the Gothic full adjective dative,

as blindamma (from blindasma, §. 170.), with [G. Ed. p. 367.]
pronominal datives like tha-mma, “to this,” i-mma, “to him”;
but the examination of the Old Sclavonic declension, in which
the indefinite adjectives remove themselves from all admix-
ture of the pronominal declension, and run entirely parallel
to the German strong substantive, not to the weak, has
led me to the, to me, very important discovery, that
Grimm’s strong and Fulda’s abstract-declension-form of
adjectives diverges in not less than nine points from the
strong substantives (i.e. those which terminate in the
theme in a vowel), and approaches to the pronominal de-
clension for no other reason than because, like the definite
adjectives in the Sclavonic and Lithuanian, they are com-
pounded with a pronoun, which naturally follows its own
declension. As, then, the definite (so I now name ‘the
strong) adjectives are defined or personified by a pronoun
incorporated with them, it is natura! that this form of de-
elension should be avoided, where the function of the in-
herent pronoun is discharged by a word which simply pre-
cedes it; thus we say guter, or der gute, not der guter, which
would be opposed to the genius of our language; for it
still lies in our perception that in guter a pronoun is con-
tained, as we perceive pronouns in im, am, beim, al-
though the pronoun is here no longer present in its original
form, but has only left behind its case-termination. In
comprehending, however, the definite adjective declension,
the science of Grammar, which in many other points had
raised itself far above the empirical perception of the lan-
guage, was here still left far behind it; and we felt, in
forms like guter, gutem, gute, more than we recognised, namely,
a pronoun which still operated in spirit, although it was no
longer bodily present. How acute, in this respect, our percep-
tion is, is proved by the fact that we place the definite form of
the adjective beside the ein when deprived [G. Ed. p. 368.]
352 ADJECTIVES.
of its definitive pronominal element; but, in the oblique

cases, beside the definite eines, einem, einen, the indefinite:
ein grosses, eines grossen (not grosses), einem grossen (not
grossem). In the accusative. grossen is at the same time
definite and indefinite; but in the former case it is a bare
theme, and therefore identical with the indefinite genitive
and dative, which is likewise devoid of inflection; but in
the latter case the n evidently belongs to the inflection.
282. The pronominal base, which in Lithuanian and
Old Sclavonic forms the definite declension, is, in its origi-
nal form, ya (=Sanskrit a ya, “ which”); and has, in the
Lithuanian, maintained itself in this form in several cases
(see below). In the Old Sclavonic, according to §. 255. (a.),
yo must be formed from ya; and from yo again, ac-
cording to §. 255. (n.), ye or e: but the monosyllabic na-
ture of the form has preserved it from the suppression of
the y, which usually takes place in polysyllabic words. In
some cases, however, the y has vocalized itself to 7 after
the vowel has been dropped. It signifies in both lan-
guages “he”; but in Old Sclavonic has preserved, in union
with sche, the old relative meaning (i-sche, “ which”), The
complete declension of this pronoun is as follows :—
SINGULAR.
LITHUANIAN. OLD SCLAVONIG.
Nominative, m. yis f. yi, m. i,* f, ya," n. ye.™

Accusative, m. yin, f. yen, ee 4 f. yt, n. ye.
Instrumental, m. yu, f. ye, m. n. im. f. yeyt,
Dative, m. yam, f. yet, m.n. yemtt, f. yet,
Genitive, m. Y0, f. yds, m. n. yego, 3 yeya,
Locative,. _m. yame, f. yoye, m.n. yem, f. yet,
* Occurs only as the relative in union with sche.

ADJECTIVES. 353
PLURAL.
LITHUANIAN. OLD SCLAVONIC.
Nominative, m. sie (yi), f. yos, m. i,” f. n. yo.*

Accusative, m. yis, f. yes, m.f.n. ya.
Instrumental, m. yes, f. yomis, m. f. n. = imi.
Dative, m. yiems, f. yoms, m. f. n. = im.
Genitive, m. f. yi, m.f.n. ich
Locative, m. yisd, f. yosd, m. f. n. ich.
DUAL. [G. Ea. p. 369.]
LITHUANIAN. OLD SCLAVONIC.
Nominative, m. yu (y#), f. ye alia

Accusative, m. yun, f. yin, a
Dative, m. yiém, f. yom, Instr. Dat. m. f. n. yima
Genitive, m. f. yi, Gen. Loc, m. f. n. yeyil.
283. The Lithuanian unites, in its definite declension,
the pronoun cited—which, according to Ruhig (Mielcke,
p- 52.), signifies the same as the Greek article—with the
adjective to be rendered definite ; so that both the latter, and
the pronoun, preserve their full terminations through all the
cases; only the pronoun in some cases loses its y, and the
terminations of the adjective are in some cases somewhat
shortened. Géras, “good,” will serve as an example.
MASCULINE.
SINGULAR. DUAL. PLURAL.
Nominative, gérasis,t geriyu, gerieyi.

Accusative, geranyan, = geruyun, geriisus,
Instrumental, geriyu, Pee geraiseis.
Dative, geramyam, girtemsiom,{ —_geriemsiems.
Genitive, geroyo, aes gertiyil.
Locative, geramyame, ..-- geritsiise.
Vocative, gerasis, geriyu gerieyi.
* See Note on preceding page.

+ Or gerassis,by assimilation from gerasyis, as,
in the Prakrit y fre-
quently assimilates
itself to apreceding
s, as tassa, “‘ hujus,’’ for WA tasya.
t The s of the adjectiveis here not in its place, and appearsto be
borrowed
from the plural.
AA
354 ADJECTIVES,
FEMININE.
SINGULAR. DUAL. PLURAL.
Nominative, —geroyi, gerieyi, gerosos,
Accusative, geranyen, — geriyin, gerases.
Instrumental, geraye, wok geromsomis.
Dative, geraiyei, gerémsom,* —_geromsoms.
Genitive, gerosiés, gertyt, gertyd.
Locative, geroyoye, 20 gerososa.
Vocative, geroyi, geriyt, gerosos.
LG. Ed. p.370.] 284. The Old Sclavonic, differing from
the Lithuanian, declines only in some cases the adjective
together with the appended pronoun, but in most cases the
latter alone. While, however, in the Lithuanian the appended
pronoun has lost its y only in some cases, in the Old Sclavonic —
that pronoun has lost, in many more, not only the y but also
its vowel, and therefore the whole base. Thus the termi-
nation alone is left, For more convenient comparison we
insert here, over against one another, the indefinite and
definite declension: svyat (theme SVYATO), “holy,” may
serve for example:
SINGULAR.
MASCULINE. FEMININE.
Indef. Def. Indef, Def.
Nominative, .svyat, svyaty-t," svyata. —svyata-yu.
Accusative, svyat, svyaty-i,' svyatd, —_svyatii-yd.
Instrumental, svyatom, svyaty-m, _ svyaloytl, svyato-yi®
Dative, svyatil, svyato-mil, svyatye, svyato-i.*
Genitive, svyata, svyata-go, svyaty, —svyaty-yd.
Locative, svyatye, svycto-m,? _svyalye, svyato-i.t
* See Note { on preceding page. ;

' See §. 255. d. 2 Or svatye-m, in which, as in the Lithuanian, the
adjective is inflected at the same time.
3 The indefinite and definite forms are here the same, for this reason,
that svyato-yeyt, as the latter must originally have been written, hasdropped
the syllable ye. The adjective base svyata has weakened its 0 to a
before the pronominal addition (§. 255. a.), just as in the dative and loca-
tive svyato-i, where an external identity with the indefinite form is not
perceptible. * Or svyatye-i. Comp. Note 2.
Indef. De. Indef. Def.
Nom:native, svyati, svyati-i svyaty svyaty-ya.
Accusative, svyaty, svyaty-ya, svyaty, — svyafy-ya,
Instrumental, svyafy, _ svyaty-imi,’ svyata-mi, svyaty-imi’
Dative, svyatom, svyaty-imi,’ svyata-m, svyaty-im.'
Genitive, svyat, svyaty-ich, svyat, svyaty-ich.
Locative, svyatyech, svyaty-ich,* svyata-ch, sryaty-ich.”
SINGULAR. PLURAL.
NEUTER.
Indef. Def. Indef. Def.
Nom. Accus. svyato, svyato-e, svyata, svyata-ya.
* I give those forms which, according to Dobrowsky (p. 302.), occur in

tle oldest MSS., in place of the more ordinary forms, which have lost
the i of the pronominal base: svyaty-mi, svyaty-m, svyaty-ch.
® Although in the pronominal declension the genitive plural 1s exter-
nally identical with the locative, we must nevertheless, in my opinion,
separate the two cases, in respect to their origin. I find, however, the
reason of their agreement in this, that the Sanskrit, which in this case is
most exactly followed by the German and Sclavonic, in pronouns of the
third person begins the plural genitive termination witha sibilant, Sanskrit
sam, Gothic zé (for sé,§.248.). This s, then, has, in Old Sclavonic, become
ch, just like that of the locative characteristic F su (9.279.). The nasal of
BTA sam must, according to rule, be lost (§. 255. 1.): the vowel, however,
has, contrary to rule, followed it, as also in the ordinary declension the
termination &m has entirely disappeared (§.278.); and the same relation
which imen, “nominum,” has to the Gothic naman-é, tye-ch, “ horum,”
has to thi-ze. This tye-ch, however, answers as genitive to the Sanskrit
Wate ¢é-shdm, and aslocative toWy té-shu; ye being used inboth cases
for @ é, according to §. 255. (e.)
* See NotesSand6. The identity with the masculine and neuter forms
arises from this, that the grave a of the feminine adjective base is changed
into the lighter 0; and this again,as in the masculine neuter,
is con-
~ verted, according to §. 255. (d.}, into y.
4a2
356 ADJECTIVES.
[G. Ed.p.371.] 285. As in the Sanskrit the preponderating

majority of adjective bases end in the masculine and neuter
in a, and in the feminine in 4d; and as this class is, in the
Old Sclavonic, only represented by bases in 0, yo in the mas-
culine and neuter (see §. 257.), and a, ya in the feminine;
it is not surprising that in German also, with the excep-
tion of a few in u (of the comparative and participle
present), all other adjective bases, in their original con-
dition, end in a, feminine o for 4 (§. 69.). It is, however,
remarkable, and peculiar to the German, that its adjectives,
in their indefinite condition, have all lengthened their theme
[G. Ed. p. 372.]_ by an unorganic n, and that in substantives
the class of words in n appears to be the most generally made
use of, inasmuch as a large number of words, whose bases
in Gothic terminate in a vowel, have, in the more modern
dialects, permitted this to be increased by n. The reason,
however, why the indefinite adjectives—not simply in part,
and for the first time in the more modern dialects, but
universally, and so early as in Gothic—have passed into
the n declension, is to be sought for in the obtuseness of
the inflection of this class of words, which, according to
§§. 139. 140., in common: with the Sanskrit, Latin, and
Greek, omits the nominative sign, and then, in variance
from the older languages, dispenses also with the dative
character, upon the loss of which, in Old High German,
has followed, also, that of the genitive character. This ab-
sence of the animating and personifying mark of case
might belong to the indefinite adjective, because it feels
itself more exactly defined through the article which pre-
cedes it, or through another pronoun, than the definite
adjective, the pronoun of which incorporated with it, has
for the most part left behind only its case terminations.
In the Lithuanian and Sclavonic, in which the article is
wanting, and thereby an inducement further to weaken the
declension of the indefinite adjectives, the Jatter stand on an
ADJECTIVES. 357
equal footing with Grimm's strong declension of substautives,

i.e. they maintain themselves, without an unorganic conso-
nantal augment, in the genuine, original limits of their base.
286. As the feminine, where it is not identical, as in
adjective bases in i in the Sanskrit, Greek, and Latin,
with the theme of the masculine and neuter, is always,
in the Indo-European family of languages, made to diverge
through an extension or an addition to the end, it is
important for German Grammar to remark—and I have
already called attention to this point in another place—
that the feminine of the German indefinite adjective, in
variance from the principle which has been ([G. Ed. p. 373.]
just given, has not arisen from its masculine, but from an
older form of the feminine; e.g. the primitive feminine
BLINDA m. n. “blind,” has extended itself in the indefinite to
BLINDAN, and the primitive feminine BLINDO to BLIN-
DON: one must not, therefore, derive the latter, although it
is the feminine of BLINDAN m., from this, as it is entirely
foreign to the Indo-European family of languages to derive
a feminine base through the lengthening of the last letter
but one of the masculine and neuter. As far as regards the
_ declension of BLINDAN m., it follows precisely that of
AHMAN (p. 322G.ed.),and BLINDAN n., that of NAMAN
(p. 176 G. ed. &c.); the fem. BLINDON differs from the mas-
culine only by a more regular inflection, since its 6 remains
everywhere unchanged, while a, in the genitive and dative
singular, is, according to §. 132., weakened to i; therefore—
MASCULINE. NEUTER. FEMININE.
Theme, BLINDAN. BLINDAN. BLINDON.
SINGULAR. PLURAL. SINGULAR. PLURAL. SINGULAR. PLURAL.
N. V.blinda’,’ blindan-s, blindé,? blindén-a? blindé@, blindén-s.
ice. blindan, blindan-s, blindé,? blindén-a,?_ blindén, blindén-s.
Jat. blindin,! blinda’-m, blindin,' blinda’-m, blindén, blindé-m,
xen. blindin-s, blindan-é, blindin-s,'blindén-é? —blindén-s, blindén-63
? See §.140, 2 See §. 141. 3 See §. 245.
358 ADJECTIVES.
287. In order, then, to examine the definite declension of

adjectives in Gothic, we will, in the first place, for the pur-
_ pose of bringing into view their agreement and discrepancy
with substantives and simple pronouns, place by the side
of each other the declension of the definite BLINDA m. n.
and BLINDO f., and that of VULFA m., “ wolf,” DAURA
n., “a. gate,” GIB O f, a gift,” and the interrogative
[G. Ed. p.874.] HVAm.n., “ who? “what?” HVO f.; further,
that of MIDYA m. n. (medius), MIDYOf., by that of HARYA
m., “an army,” BADYA n., “a bed,” KUNTHYO f., “news,”
and HVARYA m. n., “ who 2” “ what 2” HV ARYO £
MASCULINE.
SINGULAR. PLURAL.
N. vulf’s, blinds, hva-s, vulfés, blindai, hvai,*

A. vulf’, blindana, _hva-na, vulfa-ns, blindans, hva-ns.
D. vulfa,t blindamma, hva-mma,° _ vulfa-m, blindaim, hvai-m.
G. vulfi-s, blindis, hvi 8, vulf”-é, blindaizé, hvi-zé. —
V. vulf’, _blind’s, spends vulfés, blindai, ....
N. haryi-s,° midyis,! hvaryi-s, hary4s,” midyai, hvaryat
A. hari,’ — midyana, —_hvarya-na, harya-ns, midyans, hvarya-ns. :
D. harya. midyamma, hvarya-mma, harya-m, midyaim, hvaryai-m.
G. haryi-s, midyis, hvary-is, hary-é, midyaizé, hvaryaizé.
.
V. hari, midyis, ‘eee haryés, midyai, Paen
|
1 See §. 135. ® See §. 228. 5 See §.171. |
{
2 See §.297. + See §. 160,
6 From harya-s, see §. 135.
7 The nominative in adjective bases in ya does not oceur, unless perhaps :
||
in the fragments which have last appeared; and I have here formed it by
analogy with haryis and hvaryis. Grimm gives midis (I.170.). If, l.¢., the
form yis is considered as unorganic, and, in regard to midis, if its analogy |
with Aardus is remembered, then Grimm is wrong in taking MJDI for the |
theme, as in reality HARDU is the theme of hardus. The true theme ~
MIDYA occurs, however, in the comp. midya-sveipains, “ deluge,” and
answers
ADJECTIVES. 359
NEUTER.
SINGULAR, PLURAL.
N. A.V. daur’, dlindata,? hva® daura, blinda, hvé.

N. A.V. badi, midyata,’ hvarya-ta, badya, midya, hvarya.
FEMININE. [G. Ed. p. 375 ]
SINGULAR, PLURAL.
N. giba, blinda, hvé. ibés? —blindds,” hvéds?

A. giba, blinda, hed. gibé-s, —_blindés, —_hvé-s.
D. gibai,* ~—blindai,® —hvizai.® gibé-m, _blindaim, hvai-m.
G. gibé-s, _blindaizés,* hvizd-s.8 —gib’-6, ~—_blindaizo, hvi-zd.
V. giba, blinda? Rone gibés, blindés, ....
N. kunthi," midya, hvarya. kunthyés,” midyés,” hvaryos.”
A. kunthya, midya, hvarya. kunthyé-s, midyds, hvaryd-s.
D. kunthyai,” midyai,” hvaryai.” kunthyé-m, midyém, hvaryé-m.
G. kunthyé-s, midyaizés, hvaryaizds.” kunthy-6, midy’d, hvary’6.
V. kunthi, midya. iA kunthyds, midyés, hvaryés.
answers to the Sanskrit quy madhya. Formed from midya as theme,

midyis would be clearly more organic than midis. Adjective i bascs,
which could be referred to hardu-s as u base, do not exist, but only sub-
stantive, as GASTT, nom. gasts.
® Compare Zend forms like G3259—0 titirim, ‘‘ quartum,” from asyy2s pe
titirya (§. 42 ).
® Hva, with suppressed termination, for hvata, Olid High German huaz,
see §§. 155. 156.; for blindata also blind ; and so for midyata also midi.
10 The form hvé, which, like some others of this pronoun, cannot be
shewn to occur, is, by Grimm, rightly formed by analogy from thé,
“hee.” Grimm here finds, as also in the accusative singular, the 6 in
opposition to the a of blinda surprising: the reason of the deviation,
however, is fixed by §§. 69. 137. 231.
N See p. 173. Note +. 12 See §. 161. 8 §_172,
For kunthya, from kunthyé, by suppression of the final vowel of the
base, which again appears in the accusative, but shortened to a (see
§.69.); but here, also, the final vowel can be dropped; hence kunthi as
accusative. Luc. 1.77.
360 ADJECTIVES.
If, then, it is asked which pronoun is contained in the German

definite adjective, { answer, the same which, in Sclavonice
[G. Ed. p.376.] and Lithuanian, renders the adjective defi-
nite, namely, the Indian relative ya (aya). This pronoun
in German, indeed, in disadvantageous comparison with the
Lithuanian and Sclavonic, does not occur isolated in its
inflected state; but it is not uncommon in the history of lan-
guages, that a word has been lost in regard to its isolated
use, and has been preserved only in composition with other
words. It should be observed, too, that a demonstrative
i base must be acknowledged to belong to the Sanskrit,
which, in Latin, is completely declined; in Gothic almost
completely; but in Sanskrit, except the neuter nomina-
tive accusative idam, “this,” has maintained itself only in
derivative forms, as ¥f i-ti, 3raq it-tham, “so,” gam iy-at,
“so much,” $¢yé-drisa, “such.” The case is the same
in Gothic, with the pronominal base ya: from this comes,
in my opinion the affirmative particle ya, as in other
languages. also, affirmation is expressed by pronominal
forms (i-tu, Wat ta-thd, “so,” ovrws), and further yabai, “if,”
analogous with ibai, “whether,” ibaini, “lest”; as also,
in Sanskrit, af yudi, “ if,’ comes from the same base, and
to this, as I now believe, the Greek ei—the semi-vowel
being laid aside—has the same relation as in Prakrit, in the ~
3d person singular present, ai, e.g. Wakbhamai, “he wanders”
(Urvasi by Lenz, p. 63), has to the more usual wfe adi,
for the Sanskrit wfa afi. In Prakrit, too, wz jai (I. ¢.
p- 63 on j for y, see §. 19.), really occurs for yadi; so
that in this conjunction, as in the 3d person of the present
Aéyer from Aéyer:), the Greek runs parallel to the cor-
ruption of the Prakrit. If, however, in ef the Sanskrit
a y has disappeared, as in the Molic vypyes=Sanskrit
yushmé, it appears as h in 6s, which has nothing to do with
the article 6, 7, where h falls only to the nominative mascu-
line and feminine, while in 6s it runs through all the cases, as
ADJECTIVES. 361
in Sanskrit the q y of qq ya-s. To this [G. Ed. p.377.]

WA_ yas, 6c, in regard to the rough breathing, bears the same
relation as duets to war yushmé, alw, a&ytos to qq yaj, “to
worship,” “to sacrifice,” aq yajya, “to be worshiped ;” topiv
to qy yudh, “to strive,” a yudhma, “strife” (comp. Pott,
pp- 236. 252.). But to return to the Gothic YA, let us further
observe yah,* “and,” “also,” with A enclitic, of which hereafter,
and yu, “now,” i.e. “at this time,” “already” (comp. Latin jam).
It also clearly forms the last portion of hvar-yis (for yas), as,
in the Sclavonic, this pronoun often unites itself with almost
all others, and, for example, is contained in ky-i, “who?”
although the interrogative base also occurs without this
combination.
288. In Gothic definite adjectives the pronominal base
YA shews itself most plainly in bases in u. Of these,
indeed, there are but a few, which we annex below,f but
a ya shews itself in all the cases, and these in blinds differ
from the substantive declension, to such an extent that
before the y the u of the adjective is suppressed, as in
Sanskrit before the comparative and superlative suffixes
iyas, ishtha; e.g. laghiyas, “more light,” laghishtha, “ most
light,” for laghv-tyas, laghv-ishtha from laghu; and as,
even in Gothic, hard-izé, “more hard” (according to
* The / may assimilate

itself to the initial consonant of the following
word, and thus may arise yag, yan, and yas, and in conjunction with thé:
yatihé, “or” (see Massmann’s Gloss.).
T Aggvus, “narrow,” aglus, “heavy,” glaggvus, “ industrious,” hardus,
“hard,” manvus, “ready,” thaursus, “dry,” thlaqvus, “tender,” seithus,
“late,” filus, “much,” and, probably, Anasqvus, “tender.”” Some occur
only as adverbs, as glaggvu-ba, “industriously.” In addition to the adverb
Jilu, “much,” since Grimm treated this subject the genitive filaus has been
found (
filaus mais, “for much more,” see Massmann’s Gloss.), which is
the more gratifying, as the adjective u bases had not yet been adduced in
this case
362 ADJECTIVES,
[G. Ed. p. 378.] Massmann, p. 48), for hardv-izé from —

HARDU. Hitherto, however, only the accusative singular
masculine thaurs-yana, “siccum,” manv'-yana, “ paratum”; the
accusative singular neuter manv'-yata; the dative plural
hnasqu'-yaim are adduceable, if Grimm, as I doubt not, is
right in ascribing to this word, which is not to be met
with in any other case, a nominative hnasqvus.* Finally.
also, the accusative plural masculine unmanv'-yans, amapa-
oxevactous (2 C. 9. 4.), although, in this case, blindans is not
different from vulfans. These examples, then, although
few, furnish powerful proof; because, in the cases to be
met with, they represent an entire class of words—viz.
the definite adjective in u—in such a manner, that not a
single variety of form occurs. It may be proper to annex
here the complete definite declension of MANVU, as it is
either to be met with, or, according to the difference of
cases, is, with more or less confidence, to be expected :—
MASCULINE. FEMININE.
SINGULAR. PLURAL+ SINGULAR. PLURAL.
N. manvu-s, (manv'-yai), manvu-s, (manv'-yds),
Ac. manv'-ya-na, ee crmendi (mane eee
D. (manv’~ya-mma),manv Feces epee (manv -yaim).
G. manvrau-s, (manv-yaizé), (manv'~yaizé6s), (manv'-yaizd,
|G. Ed p.379.] NEUTER.
SINGULAR PLURAL.
Nom. Accus. manv’-ya-ta,} (manv’-ya).
* I am the more inclined to agree with him, as a few other adjective _

bases in vu occur. Perhaps a euphonic influence of the v on the vowel __
which follows it is also at work ; as at times one finds in the Prakrit a final )
a changed through the influence of a preceding T 2, t r, or @1,tos u,
So Urvasi, p. 72,-dlu, télu, dvaranu, for kdla, tdla, dvarana; p.71,
mandharu for manéhara, ..
+t Without inflection and pronom. manvu, as wag swddu, 700, Lithue

anian darku.
ADJECTIVES. 363
“Remark 1.—Grimm finds (I. 721.) the identity of the fe-

minine with the masculine remarkable, since he, as it appears,
looks upon s as an originally mere masculine termination
(comp. I. c.824, 825.2 *). That, however, the feminine has
equal claim to s as the nominative character, and that it is
entirely without inflection where this is wanting, I think
I have shewn in §§. 134.137. Adjective bases in i, which
in the Gothic, as in the Lithuanian and Sclavonic, are
wanting, end, in the Sanskrit, Greek, and Latin, in the
nominative of both genders, in is; and only the neuter is
devoid of inflection: compare yfta Suchi-s m. f., “clean,”
suchi n., with ipi-s, iSpi, facili-s, facile. Adjectives in u, in
Sanskrit, frequently leave, in like manner, the feminine base
‘undistinguished from the masculine and neuter, and then
end, according to §. 234., in the nominative in u-s; so pandu-s
m.f., agrees with manvu-s above, and the neuter pdndu
with manvu. If two consonants do not precede the final
Zu, as in pdndu, the feminine base may, except in com-
pound words, be lengthened by an ¢, which is particularly
characteristic of this gender; and thus qTgt swddwi, “the
sweet” (theme and nominative), answers to the Greek
word deta, which is lengthened by an unorganic a (§. 119.),
for 73F:a; and swédu-s answers both as feminine and mas-
culine nominative to the Gothic manvus. In the Sanskrit,
also, a short u in the feminine base may be lengthened, and
thus the feminine of wq tanu, “thin,” is either tanu or
tani, whence the nominative fani-s; and tanwi, as substan-
tive, means the “slender woman.” The Lithuanian has
adjective bases in u, as szwiesu-s, m. “light,” “clear,”
(compare Wa Swéta, “ white,”) which nevertheless, in seve-
ral eases, replace the u bya; as szwiesém dangui, “to the
bright heaven”: in some, too, they prefix an i to the a,
the assimilating power of which changes the a into e
(comp. p. 169 Note); as, szwiesiems dangums, “to the bright
heavens.” The feminine is, in the nominative, szwiesi, the
364 ADJECTIVES.
[G. Ed. p. 880.] final i of which is evidently identical with

the Sanskrit $7 in swddwi. In the oblique cases, how-
ever, an unorganic a also is added to the Lithuanian i, as it
has been in deta: this ia, however, becomes either by eu-—
phony, e (comp. p. 174, Note *), e.g. accus. szwiesen, accus.
plural szwiesés ;or it happens, and that, indeed,in the majority —
of cases that the z is entirely suppressed, so that SZWI1ESA
passes as the theme; as szwiesds rankbs, “of the bright hand” —
(gen. szwiesai rankai (dat.). The i of ia, however, appears,
as with the participles, to have communicated itself from
the feminine to the masculine,
“Remark 2.—With the accusative manvyana which has
been cited, the conjectured dative manvyamma is least
doubtful. That Grimm should suggest forms like hardv-
amma, hardv-ana, arises from his regarding amma, ana, as
the dative and accusative terminations of the pronoun and
adjective; while, in fact, the terminations are simply mma
and na. When, therefore, HARDU, in the dative and
accusative, without annexing a pronoun, follows never-
theless the pronominal declension, the cases mentioned
must be written hardu-mma, hardu-na, analogous with
tha-mma, tha-na, i-mma, i-na. If, however, contrary to
all expectation, forms like hardvamma, hardvana, shew
themselves, they must be deduced from hardu-ya-mma,
hardu-ya-na ; so that after suppressing the y, the preceding
u, in the place in which it would be left, has passed into v.
With regard to blindamma, blindana, blindata, it is doubtful
whether they ought to be divided blind’-(y)amma, blind’-(y)ana,
blind’-(y)ata, as analogous with manv(u)-yamma, manv(u)-
~yana, manv(u)-yata, or blinda-(ya)mma, &e.: I have there-
fore left them, as also the corresponding forms from
MIDY/A, undivided. If the division blinda-mma, &ce. is
made, nothing is left of the pronoun, as in the Old Scla-
vonic dative svyato-ma, and as in our expressions like beim,
am, im, except the case-termination, and the adjective base
ADJECTIVES. 365
has preserved its a. If, however, the division biind’-amma,

&c. is made, to which I now give the preference, and
which is also adopted-by Grimm, though from a different
point of view, then the pronoun has only lost its y, as in
some cases of the Lithuanian definite, e.g. in geriis-us for
geris-yus (see p. 353); and with respect to the y which has
been dropped and the vowel which is left, blind’-amma
would have the same relation to blind’-yamma as midums,
“the middle man” (theme MZDUM4), to its Sanskrit cog-
nate form of the same import, awa madhyama, whose rela-
tion to MIDUMA I thus trace—the latter has softened
the first a to i, and has changed the middle a, through the
influence of the liquid, into u; and both, however, have, ac-
cording to §. 66., suppressed the semi-vowel.
“Remark 3.—Although, in the accusative plural mascu-
line, blindans is not different from vulfans, and the simple
word BLINDA could not form aught but (G. Ed. p. 381.]
blinda-ns; nevertheless the word manv-yans, mentioned above,
which is of the highest importance for the Grammar, as well
as the circumstance that where any inflections peculiar to
the pronoun admonish us of the existence of an inherent
pronoun in the definite adjective, this inheritance really
exists ;—these two reasons, I say, speak in favour of dividing
thus, blind’-ans, and of deducing it from blind-yans. Just in
the same manner the dative blindaim, both through the aim,
which occurs elsewhere only in pronouns, as through the
word hnasqv-yaim, mentioned above, declares itself to be
an abbreviation of blind-yaim; but blindai proves itself
only by its pronominal inflection (compare thai, hvai, San-
skrit % té, % ké) to be an abbreviation of blind’-ya.
“ Remark 4.—In the Sanskrit, in some cases an i blends
itself with the final a, which, with the a of the base, be-
comes é; hence the instrumental plural of the Véda dialect
and of the Prakrit, wafwa aswé-bhis from aswa, agate
kusumé-hin from kusuma. To this é@ answers the i in
366 ADJECTIVES.
Gothic pronominal datives like hvai-m, “ quibus,” tha-im
“his”; as the German dative, in accordance with its origin, —
is identical with the old instrumental. We were, however,
compelled, before we had a reason for seeking the pronoun
YA in the Gothic definite adjective, to give to the exten-—
sion of the base in German a wider expansion by an i
which means nothing, than it has in the Sanskrit; while we
have now every reason, where, in Gothic definites, an 7
unsubstantiated by the oldest grammar shews itself, to re-
cognise in the i a remnant of the pronominal base YZ,
either as a vocalization of the y which so often occurs in
the Sclavonic (see p. 354), or the i may be considered as
an alteration of the a of YA, as in the Lithuanian geras-is
for yeras-yis, (p. 353). The latter view pleases me the bet-
ter because it accords more closely with blind’-amma,
blind’-ana, &c., from blind’~yamma, blind’-yana. The vowel,
then, which in blind’-amme, &c., maintains itself in its
original form, appears, in this view, as 7 in the feminine
singular genitive blindaizés—which is to be divided blinda-
izés—from blinda-yizdés; and this yizés is analogous with
hvizés, thizds, from hvazés, thazés, = Sanskrit kasyds, tasyds
(§. 172.). We must not require blindé-izds — because
BLINDO is the feminine adjective base—for there is
a reason for the thinning of the 6, in the difficulty of
placing the syllables together, and a is tae short of 0
(§. 69.). For the rest, let it be considered, that in
the Sclavonic the graver feminine @ before its union
with the pronoun is weakened to the lighter masculine o
(p. 354, Note 3.); and that a diphthong oi in the Gothic
[G. Ed. p.382.] is never admissible; on which account
salbé, “I anoint,” in the subjunctive suppresses the 7, which
belongs to this mood (salbés, salbé, for salbdis, salbdi). In the
feminine dative one should expect blindaizai for blindai,
which is simple, and answers to gibai, while the remaining
German dialects are, in this case, compounded in the very
ADJECTIVES. 367
‘same manner: in Old High German the genitive is plentera,

and the dative plinteru.* In the genitive plural mascu-
line and neuter the ai in blindaize might be substantiated
through the Sanskrit zé of the pronominal genitive, as
- Ware téshdm, “ horum™; and therefore the division blindai-ze
or blind’-(y)aize should be made: as, however, the mono-
syllabic pronominal bases, in which one would rather ex-
pect a firm adherence to the old diphthong (comp. §. 137.), do
not retain it, and thi-zé, “ horum,” hvi-zé, “ quorum,” as weak-
ened forms of tha-zé, hva-zé, are used; and in the feminine
thi-z6, hvi-zé6, for thé-26, hvé-z6,=Sanskrit td-sdm, ki-sdm;
I therefore prefer to substantiate in a different way the ai
in blindaizé m.n., and blindaizé f., than by the Sanskrit é
of té-shdm m. n. (f. td-sdm), which, moreover, would not be
applicable to the feminine form blindaizé ; and [ do it, in
fact, by the pronominal base YA, so that blinda-izé blinda-
izd, is the division to be made according to the analogy of
blinda-izés.
* Remark 5.—The nominative masculine and feminine has
kept itself free, in Gothic, from union with the old relative
base, and has remained resting upon the original, as
received from the Sanskrit, Greek, and Latin. The mas-
culine blinds, also, through the very characteristic and
animated s (see §. 134.). has cause to feel itself personified
and defined determinately enough. Even if blinds could be
looked upon as an abbreviation of blindeis (comp. altheis,
“old,” from the base 4ZLTHYA, according to Massmann),
or of blindais, to which the Old High German piinter
would give authority, I should still believe that neither the
one nor the other has existed in Gothic, as even the u bases,
* The Gothic ai would lead us to expect é, and this, too, is given by

Grimm. As, however, with Kero, the doubling of the vowel, and, with
Notker, the circumflex is wanting, I adopt in preference a shortening of
the e, or leave the quantity undecided.
368 ADJECTIVES.
like manvu-s above, which, in the oblique cases, shew so

clearly the pronominal base Y4, have not received it in
the nominative singular of the personal genders. In Old
High German however, the pronoun spoken of has had
time, in the space of almost four centuries which intervene —
between its oldest memorials and Ulfilas, to raise itself up
from the oblique cases to the nominative; which was the
more desirable, as the Old High German substantive declen- —
[G. Ed. p. 888.] sion in the nominative masculine, in dis-
advantageous comparison with the Gothic, omits the mark of
case. Pliniér (the length of the é is here rendered certain)
is contracted from plinta-ir (for plinta-yir); for the Old High
German é corresponds, according to §.'78., to the Gothic ai.
In the feminine, therefore, the form plintyu, which occurs
in the chief number of strict Uld High German authori-
ties, and those which, as Grimm remarks, are the oldest
of all, has good substantiation, and corresponds very fitly
to the masculine plintér; and in the nominative and accusa-
tive plural and neuter the form plint-yu, with regard to the
retaining the y of the pronoun, is more genuine than the
Gothic blind-a for blind-ya. The form plintyu, moreover,
answers to feminine pronominal forms like dyw, “the” (f),
syu, “she,” désyu (dé-syu), “this”* (f.), and to the instru-
mental masculine and neuter dyu (in the interrogative huiu),
where all authorities concur in retaining the 7 or y; while
in the adjective, Otfrid, and, as Grimm remarks, here and
there Isidore and Tatian, have w for yu, For explanation,
* As in the Old High German é and j (y) are not distinguished in

writing, it remains uncertain in many, if not in all cases, in what placesof
the memorials which have come down to us the sound j, and in what that
of i is intended; as even where the Gothic has aj, it may become ¢ in
the Old High German. If, however, in the analogous adjective forms
like plintju one reads j, which is supported by the Gothic (p 362), we
must, i my opinion, leave it in the above forms also. Grimm writes diu,
siu, but désju ; and expresses, p. 791, his opinion regarding the #.
ADJECTIVES. 369
however, of the pronominal forms which have been men-

tioned, it is important to consider, that in the San-
skrit the pronominal base fa, or the sa which supplies
its place in the nominative masculine and feminine. unites
‘itself with the relative base q ya, by which the first pro-
noun loses its vowel. Compare, then—
SANSKRIT. OLD HIGH GERM. OLD SCLAVONIC.
em syd (=syd,) “hec,” syu, dyu, _—ta-ya.

wm tydm, “hanc,” dya, tii-yit.
B tyé “his” dy tii
mR fyds, “he,” “has,” dyd, ty-ya.
mita tydni, “hee,” dyu, ta-ya.
Here, then, in a manner as remarkable as convincing, the
relation is proved in which the Old High German forms
mentioned stand to the Gothic sé, thé, thai, [G. Ed. p. 384.]
thés, thé: one must first transpose these into syd, thyd, &c.,
before they can pass as original forms for the Old High Ger-
man. Our mother tongue, however, in the case before us,
obtains more explanation through the Sclavonic, where the
demonstrative base ZO may indeed be simply inflected
through all the cases: in several, however, which we have
partly given above, it occurs also in union with YO. It is
most probable, that in the Old High German the combina-
tion of the base of the article with the old relative pronoun
has extended itself over all the cases of the three genders;
for that it does not belong to the feminine alone is seen
from the masculine and neuter instrumental form dyu
(d’-yu), and from the dative plural, where together with
dém occurs also dyém (diém), and, in Notker, always
dien. According to this, I deduce the forms dér, dés, opis
&e., from dyer, dyes (for dyis), dyemu (from dyamu);s
that, after suppression of the vowel following the y, pid
letter has vocalized itself first to i and thence to @ Ac-
cording to this, therefore, dés, and the Gothic genitive
BB
370 ADJECTIVES.
thi-s, would be, in their origin, just as different as in the

accusative feminine dya and thé. In the neuter, on the
other hand, daz—for dyaz, as Gothic blind’-ata for blind-
yata—the vowel of the base DYA is left, and the semi-
vowel, which above had become é (from 7) has disappeared.
Further support of my views regarding the difference of
bases in the Gothic tha-na and the Old High German
dé-n (I give the accusative intentionally) is furnished by
the demonstrative désér, which I explain as compounded,
and as, in fact, a combination of the Sanskrit @ tya, men-
tioned at p. 383 G. ed., for taya, and = sya for sa-ya, the
latter of which has a full declension in the Old Sclavonie,
also, as a simple word. Déser stands, therefore, for dya-sdir
(¢=ai); and our Modern German dieser rests, in fact, upon —
a more perfect dialectic form than that which is preserved
to us in the above désér, namely, upon dya-sér or dia-sér;
referred to which the Isidorean dhéa-sa, mentioned by
Grimm (I. 795.), at least in respect of the first syllable, no
longer appears strange, for dhéa from dhia for dhya,*
answers admirably to the Sanskrit @tya, and the final —
syllable sa answers to the Sanskrit Gothic nominative |
form sa (Greek 6), which has not the sign of case.
“Remark 6.—The adjective bases which from their first
origin end in ya, as MIDYA=Sanskrit madhya, are less ;
favourable to the retention of the y of the definite pronoun;
for to the feminine or plural neuter plint’-yu for plinta-~yua
midy’-yu would be analogous, which, on account of the diffi-
[G. Ed. p. 885.] culty of pronouncing it, does not occur, but
may have originally existed in the form midya-yu, or mid-
ya-ya; for the masculine nominative midyér is from midya-r
for midya-yar, as, in Gothic, the feminine genitive-form
midyaizés from midya-yizdés. If, however, according to this —
even hvar-yaizés (from hvar-yayizés) be used, and analogous
* D, th, and dh are interchanged according to different authorities. —

ADJECTIVES. 371
forms in several other cases, so that the base YA is therein

doubled, we must recollect, that in the Lithuanian also the
base YA, besides its composition with adjectives, combines
itself, also, with itself, for stronger personification; and,
indeed, in such a manner, that it is then doubly declined,
as yis-sai (for yis-yai*), ‘he’; yo-yo, ‘of him,’ &e.”
289. The participle present has, in Gothic, preserved
only the nominative singular masculine of the definite
declension, e.g. gibands, “giving,” which may be deduced
as well from a theme GIBAND, according to the analogy
of fiyand-s (see p. 164), as from GIBANDA, according to
the analogy of vulf’-s (§.135.). The Pali (see p. 300) and
Old High German support the assumption of a theme
GIBANDA, as an extension of the original GIBAND;
whence, then, by a new addition, the indefinite theme
GIBANDAN has arisen, as, above, BLINDAN from
BLINDA; and it is very probable that all unorganic nr
tases have been preceded by an older with a vowel ter-
mination: for as al] bases which terminate in a consonant
(nd, r, and n, §. 125.) are in their declension, with the excep-
tion of the nominative nd-s, alike obtuse; . [G. Ed. p. 386.]
so it would not be necessary for GIBAND, in order to
belong, in the indefinite adjective, to a weak theme, or one
with a blunted declension, to extend itself to gibandan (com-
pare p. 302), unless for the sake of the nominative gibanda
(see §. 140.).
290. In the Pali, no feminine theme charanté has been formed
from the unorganic theme charanta, mentioned at p. 319 G. ed.
* Ruhig (by Mielke, p. 68) wrongly gives ai as the emphatic adjunct,

as the doubling of the s in tassai, szissui, yissai is clearly to be explained
through the assimilative power of the y (see p. 353, Note +). The termi-
nation ai answers to the neuter fai, mentioned at 3. 157., for tat, which
latter is contained in the compound tat-tai (comp. kok-tai, tok-tai). After
two consonants, however, the y is entirely dropped; hence e.g. kurs-ai,
not kurs-sai.
BE2
372 ADJECTIVES.
for the masculine and neuter form eharanta has arisen from
the necessity of passing from a class of declensions termi-
nating in a consonant into one more convenient, terminating
with a vowel in the theme. The Sanskrit, however, forms
from bases terminating in a consonant the feminine theme
by the addition of a vowel (é, see §. 119.); e.g. from charant m,
comes charantt, and there was therefore no reason in the
Pali to give also to the more recent form charanta a
feminine theme charantd. Here, again, the Gothic stands
in remarkable accordance with the Pali, for it has pro-
duced no feminine base GIBANDO from the presupposed
GIBANDA; and therefore, also, the indefinite G[BANDAN
has no feminine, GZBANDON, nom. gibandé, answering to
it (as BLINDON to BLINDAN); but the feminine form
gibandei_ (ei=7, §. 70.., which has arisen from the old
theme GIBAND, in analogy with the Sanskrit eharant?,
has become GIBANDEIN, by the later addition of an n.
Hence, according to §. 142., in the nominative gibande
must have arisen. It is not, however, right to regard this
nominative as a production of the more recent theme, but
as a transmission from the ancient period of the language,
for it answers to the feminine Sanskrit nominative cha-
ranté (§. 137.), and to Lithuanian forms like sukanti, * the
turning,” for which a theme sukantin is nowise admis-
sible. In Latin, bases in i or @, originally feminine, must
have arisen from adjective bases terminating with @
consonant; thus FERENTI from FERENT (compare
§. 119. genitré-c-s): and this feminine 7, as is the case in
Lithuanian, as well with the participles (see p. 174, Note) as
- [G. Ed. p. 887.] with the adjective bases in wu (p. 363), has
in some cases no longer remembered its original destination,
and been imparted to the other genders: hence the ablatives
in i (for i-d), genitive plural in i-um, neuter plural in ia
(ferenti(d), ferenti-um, ferenti-a); and hence is explained,
what must otherwise appear very surprising, that the
ADJECTIVES. 373
participles, when standing as substantives, freely take this

i, which is introduced into them from the feminine adjec-
tive (infante, sapiente).
~ “Remark.—In the yu of kepan'yu, the Old High German
feminine of képantér, I recognise the regular defining ele-
ment, as above in pliniyu, answeringto the masculine plintér.
On account of the participial feminines in yu, therefore,
_ it is not requisite to presuppose masculines in yér, accord-
ing to the analogy of midyér, midyu, midyaz, partly as
képentér and képantaz, incline, in none of their cases, to the
declension of midyér, midyaz, and also as the derivative
indefinite base in an has sprung from KEPANTA, and not
from KEPANT YA: therefore m. képanto (= Gothic gibunda),
f.n. képanta (=Gothice gibandé). This only is peculiar
to the Old High German participle present, in relation to
other adjectives, that in its uninflected adverbial state it
retains the defining pronominal base YA in its contrac-
tion to i; therefore képanti, “ giving,” not kepant, like plint,
It is, however, to be observed, that there is far more
frequent occasion to use this form divested of case termi-
nations in the participle present, than in all other adjec-
tives, as the definite form in nds in Gothic, in the
nominative singular masculine, corresponds to it; and as
it may be assumed, that here the i supplies the place of
the case termination, which has been laid aside; so that it
is very often arbitrary whether the definite form of the
participle, or the uninflected form in i, be given. So in
Grimm's hymns (II. 2.), sustollens is rendered by the unin-
flected ufpurrenti, and baptizans by taufantér, although the
reverse might just as well occur, or both participles might
stand in the same form, whether that of the nominative
or adverbial. As regards the Old Saxon forms men-
tioned by Grimm, namely, slipandyes or sldpandeas,
“ dormientis,” gnornondyé, “ merentez,” buandyam, “ habitan-
tibus,” they should, in my opinion, be rather adduced in
374 ADJECTIVES
proof of the proposition, that the participle present has,

in the dialect mentioned, preserved the defining element —
more truly than other adjectives; and that those forms
have maintained themselves in the degree of the Gothic —
[G. Ed. p.388.] forms like manvyana, mentioned at p. 362,
than that a theme in ya belonged to the Old High German
participle present before its conjunction with the pronominal
syllable.” a
DEGREES OF COMPARISON.
291. The comparative is expressed in Sanskrit by the
suffix tara, feminine tard, and the superlative by tama,
feminine tamd, which are added to the common mas-
culine and neuter theme of the positive; e.g. punya-
-tara, punya-tama, from punya, “pure”; guchi-tara, suchi-
-tama, from guchi, “clean”; balavat-tara, balavat-tama, from
balavat, “strong.” In the Zend, through a_ perver-
sion of the language sup tara and xs¢¢~ téma unite
themselves with (in place of the theme) the nominative
singular masculine; e.g. ashsobyur huskétara (Vend. 8.
p- 383) from huska, nominative masculine b >» huskd,
“dry” sEepbpwysedss spéntétema from spénta, “holy”;
SE epoca 7Ge7eb vérethrazanstema (Vend. S. p. 43) from
veréthrazant, nom. veréthrazans, “ victorious” (literally,
“ Vritra-slaying ”).* According to my opinion attara owes
* The participle present zant, the nominative of which I recognise in

S956 97 See véréthra-zans, rests on the analogy of the frequently-~
occurring roshgnd upa-zéit, “let him strike”; since, in fact, the root zan
(Sanskrit gan) suppresses its final vowel, and has treated the a which
remains according to the analogy of the conjugation vowel of the first and
sixth class (see p. 104). The Sanskrit radical gq han, “slaying,” which ap-
pears in Fart Vritra-han, “ Vritra slaying,” and similar compounds, has,
in Zend, taken the form jan, the nominative of which is gusy_ jdo (Vend. S,
p. 43),
DEGREES OF COMPARISON. 375
its originto the root @ éri (tar, §.1.), “to [G. Ed. p.389.)
step beyond ” “to place beyond” (e.g. “over a river”); hence,
also, the substantive tara, “a float.” In the Latin, as Lisch
has acutely remarked, with this root are connected the pre-
position frans, and also terminus, as that which is overstepped,
and probably also fra, in in-tra-re, penetra-re. The superla-
tive suffix I derive, with Grimm (IIL 583.), from that of the
comparative, although I assume no theoretic necessity that the
superlative must have been developed through the degree of
the comparative. But tama, asa primitive, presents no satis-
factory etymology: I formerly thought of the base w= tan,
“to extend,” whence, also, raros could be explained; but then
wa tama would be no regular formation, and I now prefer
recognising in it an abbreviation of farama, partly be-.
cause the superlative suffix =¥ ishtha may be satisfac-
torily considered as derived from its comparative éyas,
through the suffix tha, which, in the Greek, is contained in
the form of to, as well in so-to¢g as in tatos, for taptos or
tapotros. In this manner, therefore, is formed taro-s and
waa tama-s: they both contain the same primitive, abbre-
viated in a similar manner, but have taken a different de-
rivative suffix, as in 7éun-rTos contrasted with waa panchama,
“the fifth”: the vowel, however, is more truly retained
in the derivative taros than in its base repos. In Latin,
waa tama-s has become timu-s (optimus, intimus, extimus,
ultimus); and, by the exchange of the é with s, which
is more usual in Greek than in Latin, simus; hence,
p. 43), andis analogous to the Sanskrit panthds, from panthan, mentioned

at p. 308. More usually, however, do in Zend nominatives stands in the
place of the Sanskrit Gn of the suffix vant and vdis ; so that, in Zend, the
sign of the nominative has taken the place of the Indian n, the said sign
being 0 for s, according to §.56°. In gus» vdo, from ata vais, the Zend
o may also be looked upon as belonging to the base (comp. Burnouf’s
Yagna, Notes, p. cxxviii. &c.).
y\
376 ADJECTIVES.
maximus (mac-simus) for mag-simus. However, the simus

is generally preceded by the syllable is, which we wil
hereafter explain. .
292. As in comparatives a relation between two, and in
[G. Ed. p.390.] .superlatives a relation between many, lies
at the bottom, it is natural that their suffixes should also be
transferred to other words, whose chief notion is individual-
ized through that of duality or plurality: thus they appear in
pronouns, and wate katara-s is “which of two persons?”
and @aaQ katama-s, “ which of more than two persons?”
wHATA ékataras is “ one of two persons,” and ékatama-s, “one
of more than two.” It is hardly necessary to call attention
to similar forms in Greek, as mérepos (for Kérepos), ExATEpOS.
In €xaotos the superlative suffix (c7o¢ for :otos) presents a
different modification from that in ékaftama-s, and expresses
“the one of two persons,” instead of “the one of many
persons.” In Latin and German, indeed, the suffix tara
is not in use in genuine comparatives, but has maintained
itself in pronouns in Latin in the form of TERU (ter, teru-m),
and in Gothic in that of THARA; hence uter, neuter, alter;
Gothic, hva-thar,* “which of two persons?” Old High German,
[G. Ed. p.391.]- Auédar, which has remained to us in the
adverb weder, as an abbreviation of the Middle High Ger-
* The Gothic resembles the Latin in withdrawing the sign of the

nominative from its masculine bases in ra, as the latter does from
its corresponding bases in ru. Hence, above, hvathar for hvathar(a)s, as
alter for alterus; so also vair, *‘ man,” = Latin vir for viru-s. This sup-
pression has, however, not extended itself universally in both languages.
In the Gothic, as it appears, the s is protected by the two preceding con-
sonants; hence akrs, “a field” (comp. Grimm, p. 599); still the adjective
nominatives gaurs, “ mournful ” (theme Gaura, comp. Sanskrit 91< ghéra,
“terrible ”), and evérs, “ honoured,’”’ occur, where this cause is wanting,
where, however, the preceding long vowel and the diphthong aw may
have operated. In vair, indeed, a diphthong precedes ; but the @ is here
first introduced through the euphonic law 82. If, in Latin, in adjective
bases in ri, only the masculine has predominantly given up the s, with the
preceding
man, combined with a particle of negation newéder. Anthar,

also, our anderer, belongs here, and answers to the Sanskrit
Watq antara-s, whose initial syllable is the same which in
wa anya, “alius,” has united itself with the relative base
aya. From this wa anya comes anyatara, “alter.” If,
however, Wat antara means, in general, “the other,” the
comparative suffix is here intended to denote the person
following after, passing over this thing; so is, also, the
Latin ceferus to be considered, from ce as demonstrative
base (compare ci-s, ci-tra); and so, also, in Sanskrit, tara,
“the other,’ comes from the demonstrative base i, as, in
Latin, the adverb iterum from the same base.”* In our
German, also, wieder is the comparative suffix, and the
whole rests, perhaps, on a pre-existing Old High German
word huia-dar or hwyadar, with a change of the inter-
rogative meaning into the demonstrative, as in weder, ent-
-weder. The wie in wieder, therefore, should be regarded as,
p- 370, die in dieser; and herein we may refer to the Isidoric
dhéa-sa.
293. In prepositions, also, it cannot be surprising if one
finds them invested with a comparative or superlative suf-
fix, or if some of them occur merely with a comparative
termination. For at the bottom of all genuine prepositions,
preceding i, while e.g. the feminine ccris might have permitted its is to
have been removed, just as well as the masculine, I can find the reason of
firm adherence of the feminine to the termination is only in the circum
stance that the vowel 7 particularly agrees with that gender, as it is in
Sanskrit (although long), according to (. 119., the true vowel of formation
for the feminine base. In Gothic, the suppression of the nominative sign
# is universal in bases in sa and si, in order that, as the final vowel of the
base is suppressed, two s should not meet at the end of the word; hence
e.g. the nominative drus, “a fall,” from DRUSA ; garuns, “a market,”
from GARUNSI, f.
* I have traced back the comparative nature of this adverb, which
Voss derives from iter, “the journey,” for the first time in my Review of
Forster's Sanskrit Grammar in the Heidelb. Jahrb. 1818. i. p. 479.
378 ADJECTIVES.
at least in their original sense, there exists a relation between

[G. Ed. p. 392.] two opposite directions — thus, “over,”
“from,” “before,” “to,” have the relations “under,” “in,” “to-
wards,” “from,” as their counter-poles and points of com-
parison, as the right is opposed to the left; and is always
expressed in Latin, also, with the comparative suffix, dezter
(efeya dakshina), sinister. As, however, the comparative na-
ture of these formations is no longer recognised in the present
condition of the Latin, the suffix fer admits of the further
addition of the customary ior (dexterior, sinisterior, like
exterior, interior); while the superlative timus has affixed
itself to the core of the word (deztimus or -tumus, sinistimus).
The prepositions which, in Latin, contain a comparative
suffix, are inter, preter, propter, the adverbially-used subter,
and probably, also, obiter (compare audacter, pariter).* To
inter answers the Sanskrit Wat antar, “among,” “between”;
for which, however, a primitive an is wanting, as in Sanskrit
the relation “in” is always expressed by the locative. Notwith-
standing this, antar, in regard to its suffix, is an analogous
word to wIat pratar, “in the morning,” from the preposition
[G. Ed. p. 893.] pra, “before,”fwith a lengthened a,as in the
* I was of opinion, when I first treated this subject (Heidelb. Jahrb.

1818, p. 480), that ob-i-ter must be so divided, and i looked upon as the
vowel of conjunction. As, however, the preposition ob is connected with
the Sanskrit wy abhi, “to,” “towards,” the division obi-ter might also be
made, and the original form of the preposition recognised in obi : observe
the Sanskrit derivative wire abhi-tas, “near,” from abhi with the suffix
tas. The common idea, however, that obiter is compounded of ob and
iter cannot entirely be disproved, partly as then obiter would be a similar
compound to obviam.
+ Comp. ni, pari, prati, for ni, &c. in certain compounds, Formations
which do not quite follow the usual track, and are rendered intelligible by
numerous analogies, are nevertheless frequently misunderstood by the
Indian Grammarians. Thus Wilson, according to native authorities,
derives WATT antar from anta, “end,” with rd, “ to arrive at,” and the
analogous
Greek zpwi from apo. For the relation “under,” the San-
skrit has the preposition wwa adhas, which I have else-
where explained as coming from the demonstrative base
waa; from which, also, come Wat a-dhara and waa, a-dhama,
“the under one,” or “the most under,” to which inferus and
infimus are akin, as fumus to yaa dhiima-s, “ smoke,” and,
with a nasal prefixed, as in éudi in relation to wfz abhi,
and in aude, “ambo,” answering to wit ubhdu, Old Scla-
vonic oba. The suffixes wt dhara and wa dhama are, in my
opinion, only slightly-corrupted forms of the tara and tama
mentioned: in § 291.; as also in waa prathama, “the
first,” m. from pra, “before,” the 7 sound of the suffix is
somewhat differently transposed. The suffix dhas of adhas,
“beneath,” however, has exactly the same relation to fas,
in wae afas, “from here,” as dhara, dhama, have to tara,
tama; and therefore adhas, as a modification of atas, is, in
Tespect to its suffix, a cognate form of subtus, intus. The
usual intention of the suffix wa tas, like that of the Latin
tus, is to express distance from a place. In this, also, the
Greek @ev (from Ges, comp. §. 217.) corresponds with it,
which, in regard to its 7’ sound, rests on the form ya dhas
in wa _adhas (§. 16.), as the latter also serves as the pat-
tern of the Old Sclavonic suffix dd, which only occurs in
pronouns, and expresses the same relation as 7a tas, Oev,
tus: e.g. ovo-idd, “hence,"* ono-iidd, “thence.” The form
di, however, corresponds to the euphonic alteration, which
a final as in the Sanskrit must suffer before [G. Ed. p.394.]
sonant letters (§. 25.), viz. that into 6 (see §. 255. f.), which in
Zend has become fixed (§. 56°.).
analogous word prdétar from pra, with at, “to go.’’ A relation, never-
theless, between anta, “end,” and antar, “among,” cannot perhaps be
denied, as they agree in the idea of room. They are, however, if they
‘are related, sister forms, and the latter is not an offshoot of the former.
- * The demonstrative base OVO answers remarkablyto the Zend
asa) ava, witho for a,accordingto §. 255. (a.).
380 ADJECTIVES,
“Remark.—Dobrowsky p. 451 gives ddd as the full

form of the suffix, just as he also lays down a suffix ddye,
which forms adverbs of place, as kidye, “ where?” ondidye,
“there.” As, however, the definitive pronoun, which has
been treated of at p. 353, &c., exists in these two adverbs,
didd, ddye, and forms, with sche, diddsche, ddyesche, for yddd,
&c.; and as this pronoun is, in general, so frequently
compounded with other adverbs, there is every reason to
assume that it is also contained in ovo-idi, ono-idi,
on'-tidye, ¢-tdye, and others. But how is the @ itself in
u-dt, yti-dye, to be explained? I cannot speak with confi-
dence on this point; but as, according to §. 255. (g.), in the last
element of the diphthong @ a vocalised nasal is sometimes
recognised, yudi, yidye, might be regarded as corruptions
of yondi, yondye, and, in respect to their nasal, be compared
with the Latin inde, unde, from J, U. Ydadye, yddyf, might
also have proceeded from the feminine accusative yd, which
would again conduct us to a nasal (§. 266.): this accusative
would then stand as theme to the derivative adverb, as our
preposition hinter, Old High German hintar, has arisen
from hin, a petrified accusative, on which the Gothic
hina-dag, “this day,” “to day,” throws light. Before the
suffix dye, however, elder form de, occur also the pronouns
in a simple form, as gdye, “where?” (more anciently kde,
with the final yowel of the base KO suppressed); zdye (older
sde), “here”; idyesche, “ where” (relative). As e (¢), accord-
ing to §. 255. (b.), frequently stands as the corruption of an
older i, I recognise in the suffix de the Sanskrit fa dhi,
from wfy adhi, “over,” “upon” “towards,” (from the demon-
strative base a), which, in Greek, is far more widely diffused
in the form of 0 (7601, dAAo6:)”
294. In German, even more than in Latin, the preposi-
tions shew themselves inclined to combine with the com-
parative suffix, To the Sanskrit wag antar, Latin inter, men-
tioned above (at p. 392, G. ed.), corresponds our unter, Gothic
uncar, with u for the old a, according to §. 66.* If, how-

ever, the, in my opinion, incontrovertible original identity
of the latter with the two former is recognised, [G. Ed. p.395.]
one must not, with Grimm (III. 260.), derive undar from the
preposition und, “as far as,” &c., by a suffix ar, and so again
divide the dar; for undar,t as transmitted from an ancient
period of the language, was already formed, before the
existence of a German dialect, and the abovementioned
preposition has only to dispose itself according to the
relations of sound mentioned in §§. 66.91. The matter
-is different with the Old High German af-tar, “after,”
for the primitive language, or languages, transmit to
us only Wy apa, azo, “from”; to which, in the spirit of
Wat antar, inter, subter, &c., the old comparative suffix
has first united itself upon German ground. In Gothic,
aftra. means “again,” which I look upon as an abbrevia-
tion of aftara, as in Latin extra, intra, contra, and others,
as feminine adjectives, from ezfera, &c. In regard
to the termination however, ajftra, and similar forms
in tra, thra, appear to me as datives, i.e. original in-
strumentals (§. 160.), as also, in the Sanskrit, this case occurs
as an adverb, e.g. in "ata antaréna, “between.” Per-
haps, also, the Sanskrit pronominal adverbs in éra, although
they have a locative meaning, like qq yatra, “where,”
are to be regarded as instrumental forms, according to
the principle of the Zend language (§. 158.), and of the
gerund in q ya, (Gramm. Crit. §. 638. Rem.), so that their
tra would be to be derived from wtt tard: compare forms
like Aquat manushya-trd, “inter homines” (Gramm. Crit.
* Regarding dar and tar for thar, see §.91.

+ Grimm however, also, at II. 121. &ec., divides bréth-ar, vat-ar
(“brother,” “ father”), although the many enalogous words denoting rela-
tionship in the German and the coguate languages clearly prove the T sound
to belong to the derivative suffix (see Gramm. Crit. §. 178. Rem.).
* 882 ’ ADJECTIVES.
§. 252. suff. trd). As aftra is related to aftar, so is the Gothiz

vithra, “against,” to the Old High German widar, our wider,
the primitive of which is supplied by the Sanskrit through its
[G. Ed. p.396.] inseparable preposition fa vi, which ex-
presses separation, distraction, e.g. in visrip, “to go from one
another,” “to disperse.” Exactly similar is the Sanskrit
fa ni, to which I was the first to prove the meaning “below”
to belong,* and whence comes the adjective at# nicha, “low”
(Gramm. Crit. §, 111.), the base of our nieder, Old High Ger-
[G. Ed. p.897.] man. ni-dar.-+ From hin-dar, Old High
German hin-tar, comes our hin-ter which has already been
discussed (p. 394, G. ed. compare Grimm. III. 177. c.).
In the Old High German sun-dar, Gothic sun-dré,
“ seorsim,” afterwards a preposition, our sondern, dar is,
in like manner, clearly the comparative suffix, and the
base appears to me, in spite of the difference of signi-
* It is usual to attribute to it the meaning “in,” “‘into,”’ which cannot
in any way be supported.
+ Grimm assents to my opinion, which has been already expressed in
another place, regarding the relationship of fa ni and nidar (III. 258,
259): he wishes, however, to divide thus nid-ar, and to suppose a Gothic
verb nithan, nath, néthun, to which the Old High German gindda (our
Gnade) may belong. Does, however, gi-ndda really signify humilitas?
It appears that only the meaning gratia can be proved to belong to it;
and this is also given by Grimm, I. 617. and II. 235. gratia, humanitas,
where he divides ki-nd-da, which appears to me correct, and according to
which nd would be the root, and da the derivative suffix ; as in the etymo-
logically clear ki-wd-da,“ afflatus,’’ to which the Sanskrit gives qT wa,
to blow,” as root, the Gothic gives vé (§. 69.) (vaia, vaivé). To gi-nd-
-da, indeed, the Sanskrit supplies no root nd, but perhaps nam, “ to bend
oneself,” the m of which, according to the laws of euphony, is suppressed
before ¢, which does not produce Guna; as nata, ‘‘ bent,” nati, ‘* bending,’
with the preposition sam, san-nati, which Wilson explains by “‘ reverence,”
‘¢ obeisance,” “ reverential salutation.” As the Gothic inseparable prepo-
sition ga, Old High German gi or ki, is, as Grimm first acutely remarked,
identical with the Sanskrit sam, gi-nd-da has much the same formation
with san-na-ti: it would, however, still better agree with the feminine
_-~passive
fication, related tothe Sanskrit _ sam, “with” (compare

Gothic samath, “ together with,” Old High German samant),
and the u, therefore, is from a, according to §. 66. The
Latin con-tra, however, is nearly just as much opposed in
meaning to its primitive cum; and as cum (compare ctv)
belongs, in like manner, to 4 sam, so sundar, sundré, and
contra, would be, in a double respect, sister forms. Observe,
also, the Gothic samath, Old High German samant, “to-
gether with”: the latter answers surprisingly to the
Sanskrit waw samanta (from sam+anta, “an end”), the
ablative of which, samantdt, as also the adverb, samantatas,
mean “everywhere.” Perhaps, too, in all other Old High
German adverbs in nt (Grimm. III. 214.), the said wt anta
is contained, for the meaning “end,” cannot be unexpected
in adverbs of place and time, and, like Mitte, “mid,”
passive participle san-na-id. Be that as it may, so much is certain, that

there is no necessity for a hypothetic Gothic base nith or nath, either for
the substantive gi-nada or for the preposition nidar, as they can be fully
set at rest by the existence of a Sanskrit primitive f& ni, ‘‘ below,” and
the comparative suffix dar, which frequently occurs in prepositions, And
as the circumstance that genuine original prepositions never come from
verbs, but are connected with pronouns, I must, with regard to its etymo-
logy, keep back every verb from our nidar. Grimm wishes also to divide
the Gothic preposition vi-thrd, Old High German wi-dar, into vith-ra,
wid-ar, and to find their base in the Anglo-Saxon preposition widh,
English with, Old Sclavonic wid, Old Norman vidh, Swedish vid, Danish
ved, which mean “with,” and, according to appearance, are wanting in
_the Gothic and High German. If, however, one considers the easy and
frequent interchange of v, b, and m (atft vari, “ water,”"—=mare, Bporés—=
ana mritas, “mortuus”), one would rather recognise, in the above pre-
positions, dialectic variations of sound from the Gothic mith, which is of
the same import with them (=the Zend ROG mat), and which, in most
of the dialects mentioned, maintains itself equally with the other forms;
as it often occurs,in the historyof languages, that the true form
of a word
is equally preserved with a corruption of it.
384 ADJECTIVES.
(compare inmitten, “in the midst”) and Anfang, “begin-

ning,” it attaches itself first to the prepositional ideas:
therefore hinont, “this side,” enont, “that side,” would be
the same as “at this end,” “at that end.” With regard
to the comparative forms there is, further, the Old High
German for-dar, fur-dir (“porro,” “‘amplius”), our fir-der
to be mentioned, whence der vordere, vorderste.
[G. Ed. p.398.] “ Remark 1—As we have endeavoured
above to explain the Gothic af-tra and vithra as datives, I be-
lieve I can with still more confidence present the forms in
thré or taré as remarkable remains of ablatives. Their mean-
ing corresponds most exactly to that of the Sanskrit ablative,
which expresses the withdrawing from a place, and to that of
the Greek adverbs in ev ; thus hva-thrd, “whence?” tha-thré,
“thence,” yain-thré, “hence,” alya-thré, “from another
quarter,” inna-thré, “ from within,” uta-thré, “from with-
out,” af-taré, “ from behind,” dala-thré, “from under,” and.
some others, but only from pronouns, and, what is nearly the
same, prepositions. I might, therefore, derive dalathré,
not from dal, “a valley,” but suppose a connection with
the Sanskrit swt adhara, “the under person,” with aph-
zeresis of the a and the very common exchange of the r
with J (§. 20.). Perhaps, however, on the contrary, thal is
so named from the notion of the part below. As to the
ablative forms in tard, thré, the 6 corresponds to the San-
skrit dt (§. 179.), with 6, according to rule, for wr 4 (8. 69.),
and apocope of the ¢; so that 6 has the same relation to
the to-be-presupposed 6t that in Greek ov7w has to ovTws,
from oftwr (§. 183. Note * p. 201). Many other Gothic ad-
verbs in 4, as sinteind, “always,” sniumundé, “ hastily,” spranté,
“ suddenly,” thridys, “thirdly,” &c., might then, although
an ablative meaning does not appear more plainly in them
than in the Latin perpetuo, cito, subito, tertio, and others, be
rather considered as ablatives than as neuter accusatives of
indefinite (Grimm’s weak) forms; so that ¢hridyé would
answer to the Sanskrit ablative trit’ydt while the ecmmon

Gothic declension extends the ordinal bases in a by an
unorganic n; thus THRIDYAN, nom. thridya. It must
be further observed, that all unorganic adjective bases in
an are, in general, only used where the adjective is ren-
dered definite through a pronoun preceding it; that there-
fore the forms in 6, which pass for adverbial, are, for the
very reason that no pronoun precedes them, better as-
signed to the definite (strong) declension than to the inde-
finite; especially as most of them are only remains of
an old adjective, which is no longer preserved in other
cases, and, according to their formation, belong to a period
where the indefinite adjective declension had not yet re-
ceived the unorganic addition ofan n. As to the transla-
tion of todvavriov, 2 Cor. ii. 7., by thata undaneithé, here of
course andaneithé is the neuter accusative; but the in-
ducement for using the indefinite form is supplied by the
article, and tovvavtiov could not be otherwise literally ren-
dered. The case may be similar with 2 Cor. iv. 17., where
Castiglione takes thata andavairthé for the [G. Ed. p. 399.]
nominative, but Grimm for the adverbial accusative: as it
would else be an unsuitable imitation of the Greek text,
where 70 does not belong to aitika, but to €Aadpdv. In my
opinion, however, it can in no case be inferred from these
passages that the adverbs in 6, without an article preceding
them, belong to the same category. Moreover, also, anda-
neithé and andavairthé do not occur by themselves alone ad-
verbially. As, then, thré has shewn itself to us to be an
abbreviation of thrét, it is a question whether the suppres-
sion of the ¢ by a universal law of sound was requisite, as
in Greek, and in the Prakrit, all 7’ sounds are rejected
from the end of words, or changed into =. It is certain
that the T sounds (t, th, d)) which, in the actual condi-
tion of the Gothic, are finals, as far as we can follow their
etymology, had originally a vowel after them; so that
cc
386 ADJECTIVES.
they are final sounds of a second generation, comparable

in that respect to the Sclavonic final consonants (§. 255. 1).
This holds good, for example, with regard to th, d, in the
3d person singular and plural, and the 2d person plural
= Sanskrit fa ti, wfat anti, a tha or # ta; and I explain the
th or d, which, in pronominal bases, expresses direction to
a place, as coming from the Sanskrit suffix y dha (e ha);
which, in like manner, in pronouns expresses the locative
relation. .The passing over from the locative relation to
the accusative, expressing the direction whither, cannot be
surprising, as, even in Sanskrit, the common locative ad-
verbs in tra, and the ablatives in tas, occur also with accu-
sative meaning, i.e. expressing the direction to a place
(see tatra in my Glossary). The Sanskrit suffix 4 dha
appears, in common language, abbreviated to ha, and is
found indeed, only in i-ha, “here,” from the pronominal
base i and az sa-ha—in the Vedic dialect and Zend sa-dha—
which I derive from the pronominal base sa. It ought,
according to its origin, and consistently with the usual
destination of the suffix dha, to mean “here or there”: it
has, however, become a preposition, which expresses “ with.”
The adverb ¥@ tha, “here,” is, in Zend, s@s idha,* and fre-
[G. Ed. p.400.] quently occurs in combination with as jna,
“not”; so that serv naédhal means “ nor,” answering to
rosy néit, “‘neither” (literally “not it,” from na +izé, §. 33.).
From s»~s ava and wr. aéta, “ this” (mas.), comes s@ys»as
* Vend. Sade, p. 8368. several times: aspasteand Lesasl MOS HF

iman idha vaché framrava, “hee hie verba enuntia,” which Anquetil
translates by ‘‘en pronongant bien ces paroles.” In the same page also
occurs repeatedly »s@s adha, with the same meaning, from the demon-
strative base a, as in the Véda’s Wy adha (Rosen’s Sp. p. 10), without
perceptible meaning.
+ a+i makes é, according to §.2.; and from nédha is formed, by §.28.,
naédha.
avadha and sexprss aéta-dha (Vend. S. p. 164). To the

Zend-Védic suffix dha corresponds most exactly the Greek
6a, in évOa and évrad-6a, “here.” Perhaps évOa and OS
i-dha, $ tha, are, with regard to their base, identical ;
évOa, therefore, is for tv@a from {6a (comp. in, inde), as nasals
are easily prefixed to another consonant, and thus du¢@/ an-
swers to Wf abhi, dupw to WH ubhdu, Old Sclavonic oba;
but adda, in the triple compound év-1-ad6a, is completely
the Zend se» avadha, whose theme ava has been con-
tracted in the Greek to at (compare av-@: and ad-récs, the latter
being combined with the article), but in the Old Sclavonic it is
more correctly preserved in the form of OVO.* To the word
zm thatya, “ of this place,” which is derived from ¥¢ tha
through the suffix @ tya, corresponds the Greek év@dctos,
with o from +; compare, with regard to the suffix, the Latin
propitius from prope, and, in the Gothic, frama-thya, “a
foreigner,” through which the preposition fram shews itself
to be an abbreviation offrama. As in the Sanskrit the suffix
7 tya belongs only to local adverbs and prepositions, so might
also the Gothic ni-thyis, “cousin” (for ni-thyas, §. 135.), as
propinquus, or one who stands somewhat lower in relationship
than a brother, &c.,f be derived from the ([G. Ed. p. 401.]
* Before my acquaintance with the Zend, and deeper examination of

the Sclavonic, I believed I could make out the Greek base av to agree
with the Sanskrit amu, “ille,” by casting out the m (as xodpos with ku-
mara): now, however, 3q ava and OVO have clearly nearer claims to
take the Greek forms between them.
+ Terms of relationship often express the relation, of which they are
the representatives, very remotely, but ingeniously. Thus Ay naptri,
“a grandson,” is, I have no doubt, compounded of na, “not,” and pitri,
“father”; and “‘not-father” is regarded as a possessive compound, “ not
having as father,”’ in relation to the grandfather, who is not the father of
the grandson. In Latin it would be difficult to find the etymology of
nepos (nepot-)—and the same may be said of our word neffe—without the
aid of the word Vater, which is fully preserved from the Sanskrit. In the
cc2 meaning
388 ADJECTIVES.
ancient preposition ni, mentioned at p. 382, from which,

in Sanskrit, nitya actually comes, but differently related,
and with a signification answering less to the meaning
of the preposition, namely, sempiternus. In consideration
of the aspirates in Greek being easily interchanged, and,
e.g. in the Doric, "OPNIX is said for "OPNIO, one may also
recognise in the syllable yo, in forms like wavta-yd-Oev
mavTa-yo-ce, TOAAaYéce, and others, a cognate form of the
suffix 6a, dha, or of the corrupted € ha (comp. §. 23.). At
the bottom of these forms lies, in my opinion, as the theme,
the plural neuter, which need not be wondered at, as mavra
and woAA&é are also used as first members of compounds
(moAAG-onpos, mavTad-poppos). Tavrayo might, in the iden-
tity of its suffix with 6a, dha, or ha, mean “everywhere”;
whence may then be said mavrayé-ce, “from everywhere,”
&c., as we combine our locative adverbs wo and da with
her and hin (woher, wohin); and in Greek, also, éxet6i, éxeice,
éxeidev, which might literally mean in illic, versus illic, ab
illic, as éxet is a local adverb. Forms in yo, however, are in
a measure raised to themes capable of declension, though
only for adverbs, and develope, also, case-forms, as TavTaxov,
mavrayot (old locative and dative), avray7. The addition
of new suffixes or terminations to those already existing,
but which are obsolete, appears to me assuredly more natural
than, as Buttmann supposes, the introduction of an un-
meaning ay or eyen ayo, in which case we should have
to divide mavt-ayé-6ev, &c. But as the yo under discus-
sion has arisen from @a, dha, I think I recognise in the
xe of Hye a corruption of the suffix 6, from fy dhi; in
which respect might be compared d@yy:, as a sister form to
meaning of Weffe the negation of the relationship of father points to the

uncle. The Indian Grammarians, according to Wilson, see in naptri the
negation, but not the father, but the root yat, “to fall,” and a Un&di
suffix tri.
‘afy adhi, “to,” “towards,” with a nasal introduced. Asa

third form in which the Védic-Zend suffix dha appears in
Greek, I notice ce, with o for 6, y dh, as peros from mq
medhya, “midst,” the y of which has assimilated itself,
in the form péocos, to the c. The suffix ce, however, in
that it is altered from its original intention to denote
rest in a place, to the expression of motion to a place,
answers to the Gothic th or d, whence we set out in this
examination, in forms like hAva-th, mdé-ce, “whither?” also
hvad—John xiii. 3. hvad gaggis, mov tnaryers—yain-d, €xel-ce,
alya-th, @\Ao-ce. To the Zend idha, Greek év6a, corresponds
i-th; which, however, contrary to the original intention of
the form, does not mean “thither,” but is used as a con-
junction—“ but,” “ if,” “then” (1 Cor. vii. 7.). To this class,
also, belongs ath, which only occurs in combination with than
—ath-than, “but,” like ith-than; andit has [G, Ed. p. 402.]
the Védic-Zend a-dha as prototype (§. 399.). Thad, in com-
bination with the relative particle ei, which is probably con-
nected with q@ ya, has preserved the original locative
_ meaning together with the accusative, and thad-ei may be
cited as “ where” and “whither.” The d in these forms,
answering to the Greek 6, agrees with therule for the transmu-
tation of sounds (§. 87.); and it is to be observed that medials
at the end of a word freely pass into aspirates—compare bauth,
bu-dum (§.91.);—so that the Gothic 7 sound of the suffix
under discussion, after it has, in one direction, diverged from
the Greek, has, in another, again approached it.
“Remark 2.—As we have above recognised ablatives in
the formations in thré, taré, so we find in this comparative
suffix, also, a remnant of the Sanskrit locative; in which,
however, as in the adverbs in th, d, the expression of
repose in a place is changed into that of motion to a
place—in hidré,* “hither,” Mark xi. 3. Luke xiv. 21.; hva-dré,
“whither?” John vii. 35. On the other hand, yaindré ac-
*® Vide
§.991.
890 ADJECTIVES,
tually occurs with a locative meaning; tharei leik, yaindré

galisand sik arans, ‘Srov To cpa, éxet cvvayOjcovra of aérot.’
Compare these forms with the Sanskrit, as, adharé, “in
the lower,” and the Lithuanian wilké (§. 197.). That, how-
ever, the Gothic é, which in the genitive plural masculine
and neuter answers to the Sanskrit = 4 (§. 69.), moreover —
corresponds to @é, is proved by preterites like némum,
‘we took,’ answering to the singular nam; as, in Sanskrit,
afaa némima, ‘we bent ourselves, answers to WAR nanama
or Wata nandma, ‘I bent myself.’ ”
295. The superlative suffix #a éama occurs in the Gothic
also in the form of TUMAN, nominative tuma, or, with
d for t in prepositional derivations, either simply or in com-
bination with the common superlative suffix JS7T4; thus,
af-tuma, “posterus,” af-tumists, “postremus,” hin-dumists, “ ex-
tremus.” If one considers the Indian suffix wa tama, to
have suffered apocope of the a—as in Latin, also, timus ap-
pears abbreviated to tim in adverbs like viri-tim, caterva-tim,
which I have already, in another place (Heidelb. Jahrb. 1818.
p. 480), explained, together with forms like Jegi-timus, as
superlatives—one may look for that fam in the Gothic cor-
[G. Ed. p.403.] rupted to tana, after the analogy of the ac-
cusative masculine of pronouns, like tha-na =A tam, Tov, hva-
-na = %_ka-m, “whom?”; and accordingly regard the pre-
positional derivations in tana, dana,as superlative forms; thus,
Gothic af-tana, “ behind”; hindana, népav, Old High German
ni-dana, “under” (compare our hie-nieden, “here below.” As,
however, in Old High German there exist, also, formations
in ana without a preceding ¢ sound (Grimm III. 203, &c.),
it is a question whether innana “within,” dzana “abroad,”
forana shortened to forna “from the beginning,” férrana
“moppwbev,” rimana “from a distance,” héhana “ tyd6ev,”
heimina “oixoOev,” have lost a ¢ or a d preceding the a;
or if they are formed after those in tana, dana, in the
notion that the whole of the suffix consists merely of ana;
or, finally, whether they rest on some other principle.
The preposition obar, “over,” Gothic ufar, which answers

to the Sanskrit waft upari, Greek trép, has, in the
same manner, an adverb obana, “above,” corresponding
to it.
296. In the Sanskrit the appellations of the quarters
of the heavens come from prepositions in combination
with the root wz anch, “to go”; thus the east is denoted
as “that which isbefore,” by arg prdnch, from apra, “before”;
the west as “that which is over against it,” by wr _pratyanch,
from wfa prati, “opposite”; the south as “that below,”
by wary avdnch, from wa ava, “below”; and its opposite
pole, the north, as “that above,” is called seq udanch,
from 3a ut, “up.” Now it is remarkable that in German
the names of the quarters of the world shew themselves
through their terminations, Old High German tar and tana,
or as they so frequently occur in prepositions, dar, dana, to be
derivations from prepositions, though the nature of their
origin has become obscure. The custom of the language
disposes of the forms in r and na in such a manner,
that the former expresses the direction whither (Grimm.
IIf. 205.), the latter the direction whence, which, however,
was not, perhaps, the original intention of the terminations,
both which seem adapted to express the same direction;
the former comparatively, witha glance at [G. Ed. p. 404.]
that which is opposite, the latter superlatively, in relation
to all the quarters of the globe, as, p. 376, waac
ékatara, “one of two persons,” but wana ékatama, “one
of many persons.” The west may perhaps be most satisfac-
torily explained, and in fact, as being etymologically pointed
out to be that which lies over against the east, as in Sanskrit.
For this object we betake ourselves to the prepositional
base wi, mentioned at p.382, whence the comparative
wi-dar. We do not, however, require to deduce wés-tar,*
* By writing wé, Grimm marks the corruption of the e from i, in which
I readily agree with him.
392 ADJECTIVES.
“towards the west,” wés-tana,“. from the west,” from

the derivative widur; but we may keep to its base wi,
with the assumption of a euphonic 9; as in the Sanskrit,
also, some prepositions terminating in vowels in certain
combinations, and before consonants which are disposed
to have an s before them, assume this letter; e.g. pra-
tishkasa for pratikasa; and as in Latin abs, os (for obs),
from ab, ob (§.96.). But if it were preferred to deduce
wéstar, wéstana, from the derivative widar, it would
then be necessary to force the d of derivation into
the base, and, according to § 102., change it into s.
The east is more difficult of explanation than the west
—Old High German és-tar, “towards the east,” 6s-tana,
“from the east,’—for several prepositions start up toge-
ther that would gladly sustain this quarter of the heavens.
It is not necessary that the preposition after which the
east is named should elsewhere, also, be received as a
German preposition; for in this appellation a prepo-
sition might have incorporated itself, which, except in this
case, is foreign to the practice of the German language.
[G. Ed. p. 405.] It may therefore be allowable for us,
first of all, to turn to a preposition which, in the Indian
language, is prefixed to the south, and, in the German,
may have changed its position to the east; the more so,
as, with prepositions, the principal point is always where
one stands, and the direction to which one is turned;
and one may, with perfect justice, turn that which is at the
bottom to the uppermost, or to the front. In Zend, ava,
which in Sanskrit signifies “ below,” exists as a pronoun,
and means “this”; and as this pronoun is also proper to
the Sclavonic (OVO, nom. ov), and occurs in Greek as au,
(av-0, aitds, see p. 387), it need not surprise us to find an
obsolete remnant of this base in German, and that the
east is taken as the side opposed to the west. Here it
may be necessary to observe, that in Sanskrit the pre-
position ava, in like manner, annexes a euphonic s; from
avas, therefore, by suppressing the last a but one, would

arise (as in Greek av) aus (different from our aus, Old
High German dz, Gothic a, in Sanskrit 3@ ut, “ up ™),
and hence, according to §. 80., ds: the old northern form
is austr, austan. The Latin aus-ter might then—to which
Grimm has already alluded (Wiener Jahrb. B. 28. p. 32)—
be placed with more confidence beside the Old High
German as a sister form, and led back by the hand of our
comparative suffix to the preposition, which in Sanskrit
has given ‘its name to the south, bold as it at the first
glance might appear, if we declared aus-ter and waTy_ avdnch
(ava +anch), “southern,” to be related. The derivations
from haurio, or avw, certainly deserve less notice. As,
however, the juxta-position of austar with the Latin auster
and the Indian preposition ava, avas, is most suitable,
we refrain from giving other prepositional modes in
which one might arrive at the appellation of the east in
German. As the most natural point of departure, we
cannot place it in so subordinate a position to the west as to
mark it out as “not west” (a-ustar from a-_ [G. Ed. p. 406.]
-wéstar). We turn now to the south, in Old High Ger-
man sun-dar, “towards the south,” sundana, “from the
south,” the connection of which with the sundré, sundar,
mentioned at p. 383, is not to be mistaken. The south,
therefore, appeared to our ancestors as the remote dis-
tance, and the reason for the appellation of this quarter
of the heavens being clearly in allusion to space, is a new
guarantee for the prepositional derivation of the names for
east and west, as also for the fact that the designation of the
north, too, has subjected itself to a preposition, although it is
still more veiled in obscurity than that of the three sister
appellations. We cannot, however, omit calling atten-
tion to the Sanskrit preposition fwa nis, which signifies
“out, without,” and before sonant letters, to which d belongs
(§. 25.) according to a universal law of euphony, appears
394 ADJECTIVES.
in the form of nir, which it is also usual to represent as

the original form.
297. In the Old Sclavonic the Indo-Greek compara-
tive suffix occurs in vtoryi, “the second” (m.), in which
the definitive pronoun is contained (p. 352): vtory-i, then,
is formed from vtoro-i (§. 255. d.), in which the cardi-
nal number dwa is melted down to », corresponding in
this respect to the Zend 6b in b-yaré, “two years,” but
singular, with 6 as a hardened form from v. To the
Sanskrit qa katara, “which of two? m.” (Gothic hva-thar)
and qat ya-tara, “which of both,” corresponds etymolo-
gically, the Old Sclavonic ko-tory-i (as definitive), older
ko-tery-i and ye-ter, feminine ye-tera (ye-repa) neuter
ye-tero. The origin of these two pronouns is, however,
forgotten, Sacer with their sag eek veupiecd 8 ; for
kotoryi means “who?” and yeter, “some one’ cornet
p. 352). Dobrowsky (p. 343), however, in which he is
[G. Ed. p. 407.] clearly wrong, divides the suffix into
ot-or; for although the interrogative base KO may
lay aside its 0, and combine with the demonstrative base to
(kto, “quis?” Dobr. p. 342), still it is more in accordance
with the history of language to divide ko-toryi than kot-
oryi or koto-ryi, as the formation or would there stand
quite isolated; and besides this the pronoun i, “he,”
from yo, does not occur in combination with the demon-
strative base to, and yet ye-ter is said.
298, A small number of comparatives are formed in
Sanskrit by $7@ Zyas, and the corresponding superlative by
#8 ishtha, in which ishtha, as has been already remarked
(p. 389.), we recognise a derivation from fyas in its con-
traction to ish (compare ish-ta, “ offered,” from yaj), so
that the suffix of the highest degree is properly ¥ tha,
through which, also, the ordinal numbers aqua chatur-thas
(rérap-ro-s), and awa shash-thas (éx-ros), are formed, for
the notion of the superlative lies very close to the ordinal
numbers above two, as that of order does to the super-

latives, and hence the suffix wa tama occurs in ordinal
numbers; e.g. fayfrara vinsati-tama-s, “the twentieth,”
wherefore ma, in forms like Taya pancha-ma-s, “ the fifth,”
may be held to be an abbreviation of tama. To the form
ish, contracted from zyas—euphonic for is—in Greek and
Zend is, corresponds the Latin is, in the superlatives in
is-simus, which I deduce through assimilation from is-timus
(comp. §.101.); the simple is, however, which, viewed
from Latin, is a contraction of ids (§.22.), appears in the
simple form in the adverb mag-is, which may be compared
with yeys in péyic-tos. In the strong cases (§. 129.) the
Indian comparative shews a broader form than the fyas
above, namely, a long @ and a nasal preceding the s, thus
$uia fydns (see §.9.), This form, how- [(G. Ed. p. 408.]
ever, may originally have been current in all the cases,
as the strong form in general (§. 129.), as is probable
through the pervading Jong 6 in Latin, idris, idri, &c., if
one would not rather regard the length of the Latin o as
compensation for the rejected nasal: compare the old
accusative mel-idsem, mentioned in §. 22., with Sanskrit
forms like m<taiwa_gar-iydis-am (graviorem). The breadth
of the suffix, which is still remarkable in the more
contracted from yas, may be the cause why the form
of the positive is exposed to great reductions before
it; so that not only final vowels are rejected, as gene-
rally before Taddhita suffixes* beginning with a vowel, but
whole suffixes, together with the vowel preceding them,
are suppressed (Gramm. Crit. §. 252.); e.g. from afana
mati-mat, “intelligent,” from mati, “understanding,” comes
mat-iyds; from balévat, “strong” (“gifted with strength,”
*°The Taddhita suffixes are those which form derivative words not
primitives
direct from the root itself,
396 ADJECTIVES,
from bala+vat), bal-tyas; from kshipra, “quick” (from

the base kship, “to throw”), comes kshép-tyas; from
kshudra, “insignificant,” kshéd-iyas; from tripra, “ satis-
fied,” trap-iyas; since with vowels capable of Guna the
dropping of the suffix is compensated by strengthening
the radical syllable by Guna, as in the Zend vaédista;
which Burnouf (Vahista, p. 22) deduces, as it appears to me,
with equal correctness and acuteness from vidvas (vidvé,
§.56°., Sanskrit vidwas), “knowing.” With respect to
trapiyas, from tripra, let it be observed that ar, as Guna of
ri, is easily transposed to ra (Gramm. Crit. §. 34°.): compare
the Greek @dpaxov for éSapxov; marpéor for matapar (see
p- 290, G.ed.). Ina similar manner M. Ag. Benary explains
the connection of variyas with uru “great,” with which he
rightly compares the Greek evpis (Berl. Jahrb. 1834. I.
(G. Ed. p. 409.] pp. 230, 231). But variyas might also
come from vara, “excellent,” and uru might be an abbrevia-
tiou of varu, which easily runs into one. To the su-
perlative afzg varishtha, which does not only mean Jatissi-
mus but also optimus, the Greek dproros (therefore Fap:o7os)
is without doubt akin, the connection of which with evpis one
could scarcely have conjectured without the Sanskrit. Re-
markable, too, is the concurrence of the Greek with the
Sanskrit in this point, that the former, like the latter, be-
fore the gradation suffix under discussion, disburthens itself
of other more weighty suffixes (compare Burnouf's Vahista,
p. 28); thus, éy@:o70, ALOK LTTOS, OLKTIOTOS, KUDIGTOS, MIKLTTOS,
aAyioros, from €yOpos, &c., exactly as above kshépishthas and
others from kshipra; and I believe I can hence explain, ac-
cording to the same principle, the lengthening of the vowel in
uyKioTos, HGooov, from paKpds, on which principle also rests
the Guna in analogous Sanskrit forms—namely, as a com-
pensation for the suppression of the suffix. The case is
the same with the lengthened vowel in forms like @ac<cov,
aocov, where Buttmann (§. 67. Rem. 3. N.**) assumes that
- DEGREES OF COMPARISON. 397
the comparative « has fallen back and united itself with
thea (z); while, in my opinion, a different account is to be
given of what has become of the « in forms like @acswy,
Bpacowy (§. 300.). The formation of péy:cros from péyas,
from péyado-s, is similar to the origin, in Sanskrit, of
afeg banhishtha, from bahula, “ much”; from bahu, “ much”
comes bhtiyishtha; and péy-icTos, in relation to METAAO, has
lost as much as banh-ishtha, compared with bahula, only that
the Sanskrit positive base is compensated for the loss of ula by
the addition of a nasal; which therefore, as Ag. Benary
(l. c.) has very correctly remarked, rests on the same
principle with the Guna in kshépishtha, &c.*
“ Remark.—It will then, also,be necessary ([G. Ed. p. 410.]
—as Burnouf(Yacna, p. 131) first pointed out, but afterwards
(Vahista, p. 25), in my opinion, wrongly retracted—to explain
the z é of sréyas, “ better,” sréshtha, “the best,” as coming
from the i of srz, “ fortune,” by Guna, instead of the common
view, in which I formerly concurred, of substituting a useless
gra as positive, and hence, by contraction with éyas, ishtha,
forming sréyas, gréshtha. From sré comes the derivative sri-
mat, “fortunate,” from which I deduce sré-yas, sré-shtha, by
the prescribed removal of the suffix,t although one might
* The Guna, however,in the gradation forms under discussion, might

also be accounted for in a different way, namely, by bringing it into con-
nection with the Vriddhi, which occurs before many other Taddhita
suffixes, especially in patronymics, as Fq@@qa vaivaswata, from
‘viraswat. On account of the great weight of the gradation suffixes fyas,
ishtha, which has given rise to the suppression of the suffix of the positive
base, the initial vowel also of the same would accordingly be raised by
the weaker Guna, instead of by the Vriddhi, as usual (§.26.). Be that
how it may, one must in any case have ground to assume an historic con-
nection between the Grecian vowel-lengthening in pyxictos, Oaccov, and
others, and that of Sansk:it forms like kshép-yas, k»hépishiha.
+ If there existed, as in Zend, a srira, one might hence also derive the
above gradations.
398 ADJECTIVES.
expect in the superlative gray-ishtha, euphonic for $ré-ishtha ;

and on this ground it is that Burnouf takes his objection.
But as in Greek éxa-o7os, é2d-o705 (see p. 376), in spite of
the want of the « of so7o¢, are nevertheless nothing else than
superlative forms, I do not see why, in certain cases, in
Sanskrit, also, the suppression of an i may not hold good.
This happens, moreover, in sthé-shtha from sthi-ra, “ fast,”
sphé-shtha from sphi-ra, “ swollen,” and pré-shtha from priy-a,
“dear.” In the latter case, after removing the suffix a,
the preceding y, also, must retire, since priy is only a
euphonic alteration of pri (Gramm. Crit. §. 51.) As to the
derivation, however, of the meanings melior, optimus, from
a positive with the meaning “fortunate,” it may be further
remarked, that, in Sanskrit, ‘fortune ” and “splendour”
are generally the fundamental notions for that which is
good and excellent ; hence, bhagavat, “ the honourable,” “ the
[G. Ed. p. 411.] excellent,” properly, “the man gifted with
fortune ”; for our besserer, bester, also Gothic bat-iza, bat-ists,
are associated with a Sanskrit root denoting fortune (bhad,
whence bhadra, “‘ fortunate,” “ excellent”), which Pott was
acute enough first to remark (Etymol. Inquiries, p. 245), who
collates also bétyan, “to use.” The old d gives, according
to §. 87., in the Gothic ¢, and the Sanskrit bh becomes b.
It might appear too daring if we made an attempt to refer
melior also to this root ; but cognate words often assume the
most estranged form through doubled transitions of sound,
which, although doubled, are usual. It is very common for
d to become 1(§. 17.), and also between labial medials and the
nasal of this organ there prevails no unfrequent exchange
(comp. §. 63.). If, also, the Greek BeAtiwv, BéATicTos, should
belong to this class, and the t be an unorganic addition, which
is wanting in BéA-repos, BéA-TaTos, BeA would then give the
middle step between wg bhad and mel. The ideal positive
of Bertiwv, namely aya6ds, might be connected with wary
agidha “deep,” with which, also, the Gothic géths (theme
géda) is to be compared, with 6, according to rule, for st 4

(§. 69.), and medials for Greek aspirates, according to §. 87.
299. From the strong theme fafa ¢ydas, mentioned at
§. 298., comes the nominative ¢ydén, with the suppression
of the final letter rendered necessary through §.94. The
vocative has a short a, and sounds zyan. To tyan answers
the Greek jwv, and to the vocative ¢yan answers tov; to
the neuter fyas (N. A. V.), identical with the weak theme,
corresponds the Latin ius (§. 22.). The Greek, however,
cannot become repossessed of the s, which is abandoned
in Sanskrit in the nominative and vocative masculine for
legitimate reasons, since it declines its comparative as
though its theme terminated from the first with v; hence
accusative fov-a for the Sanskrit fataq dydzis-am, Latin
idr-em (ids-em, §. 22.), genitive tov-os for tyas-as, idr-is.
However, one might, as Pott has already, I believe, noticed
somewhere, reduce the contracted forms like feArTiw,
BeArious, to an original toca,ioces, tocas, corresponding to
tydnsam, tydnsi (neuter plural), zydns-as, ¢yas-as, the o of
which, as is so common between two vowels, would be
rejected.* On the other hand, v, except in ([G. Ed. p. 412.]
comparatives, on the presupposition that the contracted forms
have rejected an vy and not oa, is suppressed only in a few
isolated words (A7cAAw, Tloce:d&, etka, andovs, and a few
others), which, however, the theoretic derivation of the com-
parative = renders very embarrasing. We would therefore
prefer giving up this, and assuming, that while the Sanskrit
in the weak, i.e. in the majority of cases, has abandoned
the former consonant of ns, the Greek, which was still
less favourable to the vo-, has given up the latter, as
perhaps one may suppose in the oldest, as it were, pre-
Grecian period, forms like BeAtiovea. It is, however,
remarkable, that while all other European sister lan-
* Comp. p. 325 G. ed.

400. ADJECTIVES.
guages have only preserved the last element of the

comparative ns—the Latin in the form of r—and while ~
the Sanskrit also shews more indulgence for the s than
for the n, the Greek alone has preserved the nasal;
so that in the comparative it differs in this respect
from all the other languages. Without the intervention
of the Sanskrit and Zend it would be hardly possible to
adduce from the European sister languages a cognate
termination to the Greek twv, jov; or if idr and fev should
be compared, one would think rather of a permutation of
liquids,* than that after the Greek v the prototype of the
Latin r, namely o, has originally existed.
300. In Zend, the superlatives in seus ista are more
numerous than the corresponding ones in Sanskrit, and re-
quire no authentication. With regard to their theory,
Burnouf has rendered important service, by his excellent
[G. Ed, p. 418.] treatise on the Vahista; and his remarks are
also useful to us in Sanskrit Grammar. In form aseys
ista stands nearer to the Greek soro-s than the Indian ishtha,
and is completely identical with the Gothic ista, nom, ist’-s
(§.135.), as the Zend frequently exhibits ¢ for the Sanskrit
aspirates. The comparative form which belongs to ista is
much more rare, but perhaps only on account of the want of
occasion for its appearance in the authorities which have been
handed down to us, in which, also, the form in tara can
only scantily be cited. An example of the comparative
under discussion is the feminine sw )9.0G masyéhi, which
occurs repeatedly, and to which I have already elsewhere
drawn attention.f It springs from the positive base
* Comp. 4.20.
+ Berl. Jahrb. 1831. I. v. 372. I then conceived this form to be thns
arrived at, that the y of the Sanskrit éyast had disappeared, as in the geni-
tive termination /é, from ep sya: after which the 7 must have passed inte v,
Still the above view of the case, which is also the one chosen by Burnout:
SassusG masa’, “great” (maid, masah, masanh, S§. 56°. 56°.),

and confirms, like other Zend forms, the theory which holds
good for the Sanskrit, that other suffixes fall away before
the exponents of the comparative and superlative relation
under discussion. If yéhi is compared with the Sanskrit
feminine base zyas?, the loss of the ¢ shews itself, and then
the a has, through the power of assimilation of the y (§. 42.),
become é and s has, according to §. 53., become A. In
the loss of the 7 the Zend coincides with the Sanskrit forms
like: sré-yas, mentioned at p. 397, with which, also, bhé-yas,
“more,” and jyd-yas, “older,” agree. Greek comparatives
with a doubled ¢ before wy, as xpeiccwy, Bpacowy, EAdoowr,
are based on this; which, according to a law of euphony
‘very universally followed in Prakrit, have assimilated the y to
the preceding consonant, as elsewhere aAAos [G. Ed. p. 414.]
from aAyos, Gothic alya-, Latin aliu-s, Sanskrit anya, are
explained (Demonstrative Bases, p. 20). In Prakrit, in the
assimilations which are extremely common in this dialect,
the weaker consonant assimilates itself to the stronger,
whether this precedes or follows it; thus anna, “the
other,” from anya, corresponds to the Greek a@AAos; the San-
skrit tasya, “hujus,” becomes tassa; bhavishyali, “he will
be,” becomes bhavissadi,* divya, “heavenly,” divva; from
‘is simpler, and closer at hand, although the other cannot be shewn to be
impossible ; for it is certain that if the y of iyas had disappeared in Zend,
‘it would fall to the turn of the preceding i to become y.
* Comp. Zocopa:, from écyoua, with SrtA syami,incomposition with
attributive verbs. It may be allowed here preliminarily to mention
another interesting Prakrit form of the future, which consistsin this, that
the Sanskrit s passes into h, but the syllable q ya is contracted to i,
herein agreeing with the Latin i in eris, erit; amabis, amabit, &c.; as,
karihisi,
“ thou willst make,” from karishyasi ;sahihimi, “* I willendure,”
from sahishydmi, instead of the medial form sahishyé (Urvasi, by Lenz.
-p.59).
DD
402 3 ADJECTIVES, .
which. it is clear that v is stronger than y, a3 it also is

more powerful than r; hence savva from sarva, “every-
one.” It is remarkable that the i also of iti “thus” as-
similates itself to the following ¢; hence, ¢ti, which, in pro-
nunciation, naturally leans upon the word preceding.
Therefore one might thus also, without presupposition of
a form yw, establish the assimilation from iwv. As to the
transition of the consonant of the positive base into o (kpé:o-
-cwv, Bogo-cwv, Bao-cwv, pho-cwv, éddc-cwv, &e.), to which
the y has assimilated, the transition of 7, 0, 6, into o need
least of all surprise us (see §. 99.); but with regard to the
gutturals, the Old Sclavonic may be noticed, in which, be-
sides what has been remarked in §. 255. (m.), y, i, and e—
which latter comes very near the vowel combined with a
y, and is frequently the remainder of the syllable ye—
exert an influence on a guttural- preceding them, similar
[G. Ed. p.415.] to that which the comparative y or 1 produces
in Greek. Before the i, namely, of the nominative plural,
and before ye in the dative and locative singular, as before é
and ye of the imperative, ch becomes s; e.g. gryes-i from
gryech, as §ac-cwv from Gao-yuv, from tay-; g becomes &
e.g. prati from prag, as peiGwv, dAiGwv, from peCywy, dArTyav,
from jiey-, dAry-; & becomes ch, while in Greek « is modified
in the same way as y On account of the contracted nature
of the € (=<dc) no assimilation takes place after it, but the y
entirely disappears, or, in peiCwr, is preree into the interior
of the word (comp. §."119 .), aS in duelver, yeipwv, which lat-
ter may be akin to the Sanskrit wat adhara, “the under
‘(m),” consequently with apheresis of the a (comp. §. 401.),
With the superlative péy:oros compare the Zend aseuss§
mazista, where ¢ z, according to §. 57., answers to the San-
skrit h of Heaq_mahat, “great “; while in the above Jev709 39956
masyéhi, as in the positive masa (euphonica!ly masé), 8
‘stands irregularly for z, as if the Zend, by its permutation
of consonants in this word, would vie with the Greek; but
a
DEGREES OF COMPARISON, 403
owe find, Vend.S. p. 214, by3¢0¢ mazyé, with z, which I hold

to be a neuter comparative; thus, gus gs yycasg mazyd
-vidvdo, “the more (literally greater) wise.”
301. As in the Latin comparative a suffix has raised
itself to universal currency, which in Sanskrit and Greek
is only sparingly applied, but was, perhaps, originally,
similarly with the form in fara, repo-s, in universal use; so
the German, the Sclavonic, and Lithuanian, in their degrees
-of comparison everywhere attach themselves to the more
rare forms in Sanskrit and Greek; and indeed in the Gothic
the suffix of the comparative shews itself in the same short-
ened form in which it appears in the Sanskrit, Zend, Greek,
and Latin, in its combination with the superlative suffix
(see §. 298. p. 395 &c.), namely, as is; and this most plainly
in adverbs like mais, “more,” whose con- (4G, Ed. p. 416.]
nection with comparatives in the Sanskrit, &c., I first pointed
out.in the Berl. Jahrb. (May 1827, p. 742). We must divide,
therefore, thus, ma-is; and this word, as well in the base as
‘in the termination, is identical with the Latin mag-is (comp.
péryio-tos, p. 402); whence it is clear that the Gothie form
has lost a guttural (compare ma-jor and mag-ior), which, in
_mikils, “great”—which has weakened the old a to i—appears,
according to the rule for the removal of letters (§. 87.), as k.
Mais, therefore, far as it seems to be separated from
it, is, in base and formation, related to the Zend maz-yé
(from maz-yas), which we have become acquainted with
above (p. 415 G. ed.) in the sense of “ more.”
“Remark.—There are some other comparative adverbs
in is, of which, the first time I treated of this subject, I
was not in possession, and which Grimm has since
(IIL 589, &c.) represented as analogous to mais. He has
however, afterwards, |. c. p. 88, agreed, with Fulda, in viewing
hauhis, av@repov, as the genitive of the positive hauhs, “ high,”
Yet hauhis stands in exactly the same relation to hau-
hiza, “the higher,” that mais does to maiza, “major.”
DD2
“404 2° ‘ADJECTIVES, © (of
‘Compared with the Zend maz-yé and Greek péft-wv, one

‘might believe the z in. maiza belonged to the positive base,
particularly as the Old High German adds a second compara-
‘tive suffix to its adverb mér, answering to the Gothic mais
:(mériro, ‘ major’) because in mér no formal expression of thie
-comparative relation was any longer felt. Raithtis, which
Grimm wishes ‘to leave under the forms which, III. p. 88,
are considered as genitive, seems to me properly to signify
‘potius, or our rechter ; and 1 consider it, therefore, as a com-
‘parative, although the Old High German réhtes, examined
-from the point of view of the Old High German, can: only
.be a genitive, and the comparative adverb is réhtér. The
»comparative ga-raihtéza, ‘justior,, which may be cited in
“Gothic, does not prevent the assumption that there may
have been also in use a raihtiza, as in all adjectives
‘iza may just as well be expected as éza; for, together
with the comparative adverb frumézé, ‘at first’ (R. xi. 35),
eoccurs the superlative frumists. Perhaps, however, the
-genius of the Old High German language has allowed itself
‘te be deceived through the identity of the comparative
‘suffix is with the genitive termination i-s; and taking some
-obsolete comparatives, which have been transmitted to it
[G..Ed. p.417.] for genitives, left them the s, which, in
‘evident comparatives, must pass into 7; but is also still re- ,
‘tained as s in wirs, ‘pejus.’ I prefer to consider, also, allis,
‘omnino,’ as @ comparative, in order entirely to exclude the
Gothic apparent genitive adverbs from the class of adjectives,
In the Old High German, together with alles, ‘omnino,’ exists
alles, ‘aliter,’ which, according to its origin, is an essentially
different word—through assimilation from alyes, as above
(p. 414 G..ed.) dAAos—in which the comparative termination,
‘in the Latin ali-ter and similar adverbs, is to be observed. The
probability that these forms, which, to use the expression,
are clothed as genitives, are, by their origin, comparatives,
is still further increased thereby, that together with eines,
DEGREESOF COMPARISON. 405
“semel, and anderes, ‘aliter, there occur, also, forms in the

guise of superlatives, namely, einesf, ‘once’ (see Graff,
p- 329), and anderesf, ‘again. Some comparative adverbs
of this sort omit, in Gothic, the i of is; thus min-s,
‘less’ (compare minor, minus, for minior, minius), perhaps
vair-s, ‘worse, which is raised anew into. vairsiza, ‘pejor,,
and may be connected with the Sanskrit avara, ‘posterus,
as above yefowy was compared with wut adhara;_ seith-s,
“*amplius’ (from seithu, ‘late’); and probably, also, suns,
* statim,’ and anaks, «subito. ”
302. The comparative-suffix is required in Gothic, where
the consonant s is no longer capable of declension,* an un-
#rganic addition, or otherwise the sibilant would have been
necessarily suppressed. The language, however, preserved
this letter, as its meaning was still too powerfully per-
ceived, by the favourite addition an, which we have seen
‘above, though without the same urgent necessity, joined to
participial bases in nd in their adjective state (§. 289.). As,
then, s comes to be inserted between two [G. Ed. p. 418.] a
‘vowels, it must, by §.86.(5.), be changed into z: hence the
‘modern theme M4IZAN, from the original MAJS, which
has remained unaltered in the adverb. The nominative mas-
culine and neuter are, according
to §§. 140. 141., muiza, maizé:
On the other hand the feminine base does not develope itself
from the masculine and neuter base M@4AIZAN—as in general
from the unorganic bases in an of the indefinite adjectives
* A base in s, as the abovementioned

mais, would not be distinguished
from the theme in all the cases of the singular, as also in the nominative and
accusative plural, as, of final double s, the latter must be rejected (comp.
drus, “fall,” for drus-s from drusa-s, §. 292. 1st Note). In the nominative
and genitive singular, therefore, the form mais-s must have become mais;
just as, in the nominative and accusative plural, where ahman-s comes
from the theme ahman. The dative singular
is, in bases ending in a con-
sonant, without exception devoid of inflection ; and so is the accusative,
insubstantives of ay kind.
406 - ADJECTIVES,
no feminines arise—but to the original feminine base in ¢,
which éxists in the Sanskrit and Zend, an f is added, as in
the participle present; thus MAIZEIN (ei=@, §. 70.), from
mais+ ein, answers to the Zend feminine base of the same
import, sw'739995¢ masyéhi, and Sanskrit forms like m<tzat
gartyas-i, from gartyas. The nominative mdizei may then,
according to §. 142., be deduced from MAIZEIN, or may
be viewed as a continuation of the form in Zend and San-
skrit which, in the nominative, is identical with the theme
(§. 137.); in which respect again the participle present
(§. 290.) is to be compared. These two kinds of feminines,
namely, of the said participle and the comparative, stard
in Gothic very isolated ; but the ground of their peculiarity,
which Jacob Grimm, III. 566, calls still undiscovered (com-
pare I. 756), appears to me, through what has been said, to be
completely disclosed ; and I have already declared my opinion
|G. Ed. p.419.] in this sense before.* The Old High German
* Berl. Jahrb. May 1827, p. 748, &c. Perhaps Grimm had not yet,
in the passage quoted above, become acquainted with my review of the
two first parts of his Grammar; since he afterwards (II. 650.) agrees with
my view of the matter. I find, however, the comparison of the transition
of the Gothic s into z with that of the Indian as into a sh inadmis-
sible, as the two transitions rest upon euphonic laws which are entirely
distinct ;of which the one, which obtains in the Gothic (§. 86. 5.), is just
as foreign to the Sanskrit, as the Sanskrit (§. 21. and Gramm. Crit. 101°.)
is to the Gothic. It is further to be observed, that, on account of the
difference of these laws, the Sanskrit q sh remains also in the superlative,
where the Gothic ha’ always st; not zt. In respect to Greek, it may
here be further remarked, that Grimm, hk c. p.651, in that language, also,
#dmits an original s in the comparative; which he, however, does not
look for after the v of ww», as appears from §. 299., but before it; so that
he wishes to divide thus pei-¢wv, as an abbreviation of peyi{ey ;and regards
the ¢ not as a corruption of the y, as Buttmann also assumes, but as
& comparative character, as in the kindred Gothic ma-iza, The Greek
wy, ov, would, according to this, appear identical with the unorganic Gothic
an in MAIZAN; while we have assigned it, in §, 299., a legitimate
foundation, by tracing it back to the Sanskrit dns,
DEGREESOF COMPARISON. 407
has brought its feminine comparatives into the more usual

path, and gives, as corresponding to the Gothic. minnizei,
“the lesser” (fem.), not minnirt, but minnira. The Gothic
sibilant, however, was,in the High German comparatives, in
the earliest period transmuted into r, whence, in this respect,
minniro, minnira, has more resemblance to the Latin minor
than to the Gothic minniza, minnizei.
303. The comparative suffix in the Gothic, besides is,
iz-an, exhibits also the form ds, éz-an: it is, however
more rare; but in the Old High German has become so
current, that there are more comparatives in it in éro
(nominative masculine), 6éra (nominative feminine and
neuter), than in iro, ira, or éro, ra. The few forms in
OZAN which can be adduced in Gothic are, svinthéza,
“fortior” (nominative masculine), frédéza, “prudentior,”
fruméza, “prior,” hlaséza, “hilarior,” garaihtéza, “justior,”
framaldréza, “provectior etate,” usdaudoza, “ sollicitior,”
unsvikunthéza, “ inclarior” (Massmann, p. 47), and the ad-
verbs sniumundés, “ crovdaotépwe,” and alyaleikés, “ étépws.”
How, then, is the 6 in these forms to be explained,
contrasted with the i of JS, IZAN? 1 believe only
as coming from the long a of the Sanskrit strong themes
fydns or yans (§§. 299. 300.), with 6, according to rule, for
wt a (8. 69.). If one starts from the latter [G. Ed. p. 420.]
form, which, in the Zend, is the only one that can be
adduced, then, beside the nasal, which is lost also in the Latin
and in the weak cases in the Sanskrit, yéns has lost in
the Gothic either the 4 or the y (=j), which, when the
4 is suppressed, must be changed into a vowel. The
Gothic és, dz, and still more the Old High German ér,
correspond, therefore, exactly to the Latin dr in minor,
minér-is, for minior. There is reason to assume that, in
the Gothic, originally y and 6 existed in juxta-position to
one another; and that for minniza, “the lesser,” was used
minnyéza, and for frédéza, “the more intelligent,” frédydza,
408 “9°. ADJECTIVES. ~~ ~ 7
The forms which have lost the y are represented in Latin’

by minor, minus, and plus, and those with 6 suppressed by
mag-is. One cannot, however, in Gothic, properly require
any superlatives in OSTA, nom. ést'-s, corresponding to the
comparatives in 6s, 6z; because this degree in the San-
skrit, Zend, Greek, and Latin always springs from the
form of the comparative, contracted to is, ish. It is, how-
ever, quite regular, that, to the fruméza, “ prior,” corresponds
a frumists, “primus,” not frumésts. To the remaining
comparatives in éza the superlative is not yet adduced;
but in the more recent dialects the comparatives have
formed superlatives with 6, after their fashion; and thus,
in the Old High German, ésé usually stands in the super-
lative, where the comparative has ér: the Gothic furnishes
two examples of this confusion of the use of language, in
lasivésts, “infirmissimus” (1 Cor, xii, 22.), and armésts, “ miser-
vimus” (1 Cor. xv. 19.).
304. In the rejection of the final vowel of the positive base
before the suffixes of intensity the German agrees with the
cognate languages; hence sut’-iza, from SUT'U*, “sweet”;
[G. Ed. p.421.]. hard’-iza, from HARDU, “hard”; seith-s
(thana-seiths, “ amplius”), from SEITHU, “late”; as in the
Greek 7diwv from ‘HAY, and in the Sanskrit daghtyas from
laghu, “light.” Ya is also rejected; hence spéd’-iza, from
SPEDYA, “late” (see p. 358, Note 7.); reik’-iza, from
REILK YA,“ rich.” One could not therefore regard the 6, in
forms like frddéza, as merely a lengthening of the ain FRODA
(§. 69.), as it would be completely contrary to the principle
of these formations, not only not to suppress the final vowel
of the positive base, but even to lengthen it... The -expla-
nation of the comparative 6 given at §. 303, remains therotenp
the only one that can be relied upon.
“ The positive does not occur, but the Sanskrit swdédu-s and Greek 1
Jead us to expect a final wv.
* 305. In the Old Sclavonic, according to Dobrowsky, p. 332;

&c., the comparative is formed in three ways, namely,
(1) By masculine iz, feminine shi, neuter yee; as, dini2,
“the better (m.)”; dnshi, “the better (f.)”; dnyee, “the best
(n.),” from a positive which has been lost, as batiza, melior,
and dueivwy; and it is perhaps connected in its base with
the latter, so that a may have become o (§. 255. a.), but p, %
as frequently occurs with n; and this d, with the preceding
o, has become d(#).* Mnit, “the lesser, (m.) ” fem. menshi,
neuter mnyee, spring, in like manner, from a positive which
has been lost. Boliz, “the greater,” fem. bolshi, neuter bolyee,
may be compared with the Sanskrit baliydn, “the stronger”
(p. 396), fem. baliyast, neuter baliyas.t| For [G. Ed. p.422.] _
bolit is also used bolyet; and all the remaining comparatives
which belong to this class have yet for iz, and thus answer
better to the neuter form yee. If, as appears to be the case, the
form yei is the genuine one, then ye answers to the Sanskrit
yas Of jyd-yas, bhii-yas, sré-yas, &c. (§. 300.), and the loss of the
s is explained by §. 255.(/.): the final i of ye-é, however, is the
definitive pronoun (§..284.), for comparatives always follow,
in the masculine and neuter, the definite declension. In the
feminine in shi it is easy to recognise the Sanskrit s? of Zyas-é,
oryas-i, and herewith also the Gothic
zei (oblique theme ZEIN,
* The ain dycivey appears to me to be privative; so that peivey would

seem to be a sister form to the Latin minor, Gothic minniza, Sclavénic
‘mnit; and dyeivey would properly signify “the not lesser,” “the not
more trifling.” Perhaps this word is also inherent in omnis; so that 0
for a would be the negation, which, in Latin, appears as in; where it
may be observed, that, in Sanskrit, a-sakrit, literally “not once,” has taken
“the representation of the meaning “‘ several times.”
+ The positive velit, withv for 5 and e for 0, occurs only in this de-
finite form (Dobr. p. 320) ; the primitive and indefinite form must be vel.
With respect to the stronger o corresponding to the weaker letter e
(§. 255. a.), boli, in the positive, answers to the manner in which vowels
are strengthened in Sanskrit, as mentioned at ). 298.
410 ADJECTIVES.
p.418 G. ed.); that is to say, bol-shi, “the greater (fem:),”

corresponds to the Sanskrit q@tvat baliyast, “the stronger
(f.),” and menshi, “ the lesser,” to the Gothie minn-izei. While,
therefore, the Sclavonic masculine and neuter have lost the 3
of the Sanskrit yas, the feminine has lost the ya of yas-é.* This
feminine shi, also, in departure from (2) and (3), keeps free
from the definite pronoun. There are some comparative
adverbs in e, as the abbreviation of ye (§. 255. n.), which in
like manner dispense with the definite pronoun; thus, dné,
“better”; bole, “greater”—in Servian MSS. dnye, bolye 3
[G. Ed. p. 423.] pache, “more,” probably related to mayts,
nGoowv; so that (which is very obscure) the final vowel of
pache for pach-ye, for reasons which have been given before,
is, in fact, identical with the Greek co of xdc-cov, for MAT-YOVe
The ch of pache may, according to p. 415 G. ed., be regarded
as a modification of k, as the first ¢ of waocov has developed
itself from y. Thus the ¢ of dol¢-yee, “longer” (neuter and
adverbial ), as euphonic representative of the g of dolg, dolga,
dolgo (longus, a, um), answers remarkably to the Greek ¢
in petCwv, 6AiCwv, for pelywv, dAtywv.. That, however, the
positive dolg is connected with the Greek doArydés needs
scarce to be mentioned. Somewhat more distant is the
Sanskrit eta dirgha-s, of the same meaning, in which
the frequently-occurring interchange between r and J is
- ® It may be proper here to call remembrance to the past gerund,

properly a participle, which in the strong cases vdns, nom. mase. van for
vans, fem. ughi, neuter vat (for vas), corresponds to the Sanskrit of the
reduplicated preterite in vas. The Old Sclavonic has here, in the nomi-
native masculine, where the s should stand at the end, lost this letter,
according to §.255.(/.), as by-v, “qui fuit,” but dy-vshi, “que fuit” ;”
and in the masculine also, in preference to the comparative, the s again
appears in the oblique cases, because there, in the Sanskrit, after the ¢
follow terminations beginning witha vowel; so in rek-sh, ‘‘ eum qui diwit,”
the sh corresponds to the Sanskrit vdns-am, as rurud-vdns-am, “ eum qui
ploracit,” PStR
DEGREES OF COMPARISON. 4ib
to be noticed (§.20.), The ¢ of doAryos, however, shews

itself, by the evidence of the Sclavonic and Sanskrit, to be
an organic addition. Let garyee, “ pejus,” be compared
with the Sanskrit gariyas, “ gravius,” from guru, “heavy "—
according to Burnouf’s correct remark from garu, as this
adjective is prononnced in Pali—through the assimilating
influence of the final u, to which the kindred Greek Bapus has
permitted no euphonic reaction.
- (2) The second, by far the most prevalent form of the
Old Sclavonic comparative, is nominative masculine shit,
feminine shaya, neuter shee. The i of shii is the definitive
pronoun, which, in the feminine, is ya, and in the neuter e
for ye (§§. 282. 284.). After the loss, then, of this pronoun,
there remains shi, sha, she; and these are abbreviations of
shyo, shyc, shye, as we have seen, p. 332, G. ed., the adjective
base SJNYO (nominative siny), before its union with the
defining i, contracted to sini (sini-7, neuter sine-e for sinye-ye.
The definite feminine of SIN
YO is sinya-ya}; and as to the
feminine comparatives not being shya-ya but sha-ya, this
rests on the special ground that sibilants gladly free
themselves from a following y, especially [G, Ed. p. 424.]
before a (Dobrowsky, p.12); so in the feminine nomi-
natives désha, sisha, chasha, for sisya, &c. (Dobr. p. 279).
The relation of the comparative form under discussion
to the Sanskrit qa yas and Zend wayy yai (p. 401)
is therefore to be taken thus, that the ya which precedes
the sibilant is suppressed, as in the above feminines in
shi; but for it, at the end, is added an unorganic YO,
which corresponds to the Gothic-Lithuanian YA im the
themes NIUYA, NAUYZA, “ new,” answering to 4% nava,
NOVU, NEO, Sclavonic NOVO. This adjunct YO has
preserved the comparative sibilant in the masculine
and neuter, which, in the first formation, must yield to
the euphonic law, §.255. (2) Examples of this se-
cond formation are, dn-shiz, “the better (m.),” feminine
412 LOO TT ADIECTIVESS 1077
dn-shaya, neuter dn-shee ;pist-shit from pst, theme PUSTO?

“desert.” Hence it is clear that the final vowel of the
positive base is. rejected, as in all the cognate languages,
however difficult the combination of the ¢ with sh. Even
whole suffixes are rejected, in accordance with §. 298.; as,
glib-shit from gliébok, “deep” (definite, gliboky-t), sladshié
from sladok, “sweet.”*
(3) Masculine yeishii, feminine yeishaya, neuter yelehows
but after sch, sh, and ch, ai stands for yey: and this av evidently
stands only euphonically for yat, since the said sibilants, as
[G. Ed. p. 425.] has been already remarked, gladly divest
themselves of a following y: hence blasch-aishii, “the
better” (masculine), from blag (theme BLAGO), “ good,”t
since g, through the influence of the y following, gives
way to a sibilant, which has subsequently. absorbed
the y; compare éAiC-wy, for 6Ary-twv, dAry-yov (p. 402):
so tish-aishit, from tich (theme ZTICHO), “still{ as in
the Greek 6éc-cwv from tayis. As example of the form
* I hold ko, whence in the nom. masc. k, for the suffix of the positive
base, but’ the preceding o for the final vowel of the lost primitive; and
this o corresponds either to a Sanskrit a, according to §. 255. (a.), or to an
Z u, according to §. 255. (¢.);. for example, ¢ano-k, “thin,” theme
TANOKO, corresponds to the Sanskrit tanu-s, ‘‘thin,” Greek ravv; and
slado-k to the Sanskrit swadu-s, “ sweet,” with exchange of the v for J,
according to §.20. Thus the above slad-shi¥ shews itself to be originally
identical, as well in the suffix of the’ positive as of the other degrees with
the Greek #3-ioy and Gothic sut-iza (§. 804.), far as the external diffe-
rence may separate them; and to the Sclavonic is due, as to the truer
preservation of the fundamental word, the preference above the Greek
and Gothic, although, on account of the unexpected transition of the
v into J, the origin of the Sclavonic wordis more difficult to recognise.
t Dobrowsky says (p. 334) from blagyi (this is the definite, see §.284.\:
it is, however, evident that the comparative has not arisen from the adjec-
tive compounded with a pronoun, but from the simple indefinite one.
t Compare the Sanskrit adverb ¢éishnim, “ still, silent,” and refer to
}.255. (m.).
DEGREES OF COMPARISON. ‘413
with yei, yin-yetshit, * junior,” from yin, ‘may ~serve.

Whence comes, then, the yet or at (for yat), which distin-
guishes this formation from the second? It might be sup-
posed that to the first formation in yet, where, for example,
also yin-yet, “the younger((m.),” oceurs, that of the
second has also been added, as in Old High German
‘mérero, “ the greater ” (masculine), and in Gothic, probably,
vairsiza, “ the worse” (p. 405), are raised twice to the com-
parative degree; andas, in Persian, the superlatives in terén,
in my opinion, contain,as their last element, the compara-
tive sare iydis, which forms, in the nominative masculine,
éyan, and from this could be easily contracted to én. In
Persian the comparative is formed through ter ;.as, behter,
“the better,” whence behterin, “ the best.” _Now it deserves
remark, that in Old Sclavonic the formation before us fre-
quently occurs with a superlative meaning, while in the
more modern dialects the age relation is expressed
through the comparative with nai, “ more,” prefixed (pro-
bably from mai = Gothic mais, according to §. 225.1). The
only objection to this mode of explanation [G. Ed. p. 426.]
is this, that the element of the first formation ye-z has not
once laid aside the definitive pronoun i, which is foreign
to the comparative; so that therefore in yiin-yet-shit the
said pronoun would be contained twice. There is, how-
ever, another way of explaining this yetshitor (y)atshiz,
namely, as an exact transmission of the Sanskrit fyas or yas,
from which the second formation has only preserved the
sibilant; but the third, together with this letter, may have
retained also that which preceded. Still, even in this method,
the i of yer, (y)a’, is embarrassing, if it be not assumed that
it owes its origin to a transposition of the 7 of ya.
306. As to the remark made at p. 400, that among the
European languages the Greek only has preserved the
nasal, which the Sanskrit shews in the strong cases of the
comparative suffix 7ydas, I must here admit a limitation in
414 ; ADJECTIVES,
favour of the Lithuanian, which, exceeding in this point

the Greek, continues not only the nasal,* but also the com-
parative sibilant through all the cases. For an example,
gerésnis, “the better” (m.), may serve, with which we would
compare the Sanskrit gariydasam, “graviorem™” (nominative
gartydn). It may be, but it is not of much consequence
to us, that gerésnis and garfydns (strong theme) are also
connected in the positive base; so that, as according to
p. 398, in Greek and Gothic goodness is measured by depth,
in Lithuanianit is measured by weight. The Sanskrit com-
parative under discussion means, also, not only “heavier,”
or “very heavy,” but also, according to Wilson, “highly
venerable.” In order, however, to analyze the Lithuanian
gerésnis, we must observe that gerésnis stands for gerésniae,
and the theme is clearly GERESNIA; hence genitive
gerésnio, dative gerésniam; as géro, gerdm, from géra-s,
[G. Ed. p. 427.] The termination ia, therefore—for which
ya might be expected, the y of which, as it appears for
the avoiding of a great accumulation of consonants, has
been resolved into i—corresponds to the unorganie addi-
tion which we, p. 411, have observed in Sclavonie compara-
tives. We have now geresn remaining, which I regard
as a metathesis from gerens,t through which we come
very near the Sanskrit gariydus. But we come still
nearer to it through the observation, that, in Lithuanian, e
is often produced by the euphonic influence of a preceding
y or i (§. 193.).. We believe, therefore, that here also we
may explain gerésn as from geryasn (geryans), and further
recall attention to the Zend ywwyys0G masyéhi (§. 300.).
* In the Lith. comparative adverbs like daugiaus, “ more,” mazaus,

“less,” I regard the u as the vocalization of the n; thus daugiaus from
dauyians, where ians=Skr. iydns of the strong cases.
+ This has been already alluded to by Grimm (III. 635, Note *), who
has. «owever, given the preference to another explanation,by which esnis
is similarly arrived at with the Latin issimus,
The emphasis upon the e of géresnis may be attibutable
to the original length in the Sanskrit strong theme gariydzs.
Hence the astonishing accuraey may justly be celebrated
with which the Lithuanian, even to the present day, con-
tinues to use the Sanskrit comparative suffix yds, or
- gather its more rare form preferred in Zend ydns.
307. The Lithuanian superlative suffix is only another
modification of the comparative. The nasal, that is to
say, which in the latter is transposed, is, in the superlative,
left in its original place: it is, however, as often happens,
resolved into u,* and to the s which ends the theme in
the Sanskrit, which, in Lithuanian, is not declinable (§. 128.),
is added ia: hence GERAUSIA, the nominative of which,
however, in departure from gerésnis, has dropped, not. the
a, but the i; thus gerausa-s, gen. gerausio, and, in ‘the femi-
nine, gerausa, gerausiés; in which forms, ([G. Ed. p. 428.)
contrary to the principle which is very generally followed
in the comparative and elsewhere, the i has exercised no
euphonic influence.
“Remark.—With respect to the Sanskrit gradation-
suffixes tara, taina, I have further to add, that they also oc-
cur in combination with the inseparable preposition 34 ut ;
hence ué-tara, ‘the higher,’ ut-tama, ‘the highest,’ as above
(§. 295.) af-tuma, and in Latin ex-timus, in-timus. I think,
however, I recognise the base of ut-tara, ut-tama, in the
Greek %¢ of to-repos, to-ratos, with the unorganic spir. asp.,
as in €xarepos, corresponding to the Sanskrit ékafura-s, and
with o from 7 (compare §. 99.), in which it is to be remarked
that also in the Zend for ut-tara, ut-tama, according to
§. 102., us-tara, us-téma, might be expected.
* Comp. §. 255. «g.); in addition to which it may be here further

remarked, that in all probability the w also in Gothic conjunctives like
haitau, haihaityau, is of nasal origin.
nosiakneles ae
NUMERALS.
CARDINAL NUMBERS.
. 308. I. In the designation of the number one great dif-
ference . prevails among the Indo-European languages,
which springs from this, that this number is expressed by
pronouns of the 3d person, whose original abundance
affords satisfactory explanation regarding the multiplicity
of expressions for one. The Sanskrit éka, whose com-
parative we have recognised in the Greek éxdrepos, is, in
my opinion, the combination of the demonstrative base 4
of which hereafter, with the interrogative base ka, which
also, in combination with api, “also” (nom. mase. ké’pi),
signifies “ whoever”; and even without this api, if an in-
terrogative expression precedes, as Bhagavad-Gita, IL. 21,
at a yen me ae wate ef aR kathan sa purushali
Partha kan ghdtayati hanti kam, “ How can this person, O
Partha, cause one to be slain, (or)slay one?” The Zend ss
. [G, Ed. p. 429.] aéva, is connected with the Sanskrit pro-
nominal adverbs éva,“ also,” “only,” &c., and évam, “so,” of
which the latter is an accusative, and the former, perhaps,
an instrumental, according to the principle of the Zend lan-
guage (§. 158.). The Gothic ain’-s, theme 4/NA, our einer,
is based on the Sanskrit defective pronoun éna (§. 72.) whence,
among others, comes the accusative masculine éna-m, “ this.”
To this pronominal base belongs, perhaps, also the Old Latin
oinos, which occurs in the Scipionian epitaphs, from which
the more modern dnus may be deduced, through the usual
transition of the old 4 into u, which latter is, lengthened
to make up for the i suppressed. Still -dnus shews, also, a
surprising resemblance to the Sanskrit dna-s, which pro-
perly means “ less,” and is prefixed to the higher numerals
in order to express diminution by one; as, énavinshati,
“undeviginti,” dnatrinshat, “undetriginta.” This énas could
CARDINAL NUMBERS. 417
not have appeared in Latin, more accurately retained than

under the form of dnu-s, or, more anciently, éno-s. The
Greek ‘EN is founded, it is highly probable, in like manner,
on the demonstrative base eq éna, and has lost its final
vowel, as the Gothic 47N4A, in the masculine nominative
ains ; with respect to the é for é compare éxatepos. On the
other hand, oios, “ unicus,” if it has arisen from oivos compare
oinos), as weit from peifova, has retained the Indian diph-
thong more truly, and has also preserved the final vowel
of eq éna. If ovos, the number one in dice, really has
its name from the idea of unity, one might refer
this word to the demonstrative base wa ana, Sclavonic
ONO (nominative on, “that”), which also plays a part
in the formation of words, where ovy corresponds to
the Sanskrit suffix and (feminine of the masculine and
neuter ana), if it is not to be referred to the medial
participle in dna, as povy to ména. The Old Sclavonie, yedin,
“one,” is clearly connected with the Sanskrit arfe ddi, “the
first,” with y which has been prefixed according to §. 255. (n.):
on the other hand, in the Lithuanian wiena-s, [G. Ed. p. 430.]
if it is connected with the Gothic 47NA and Sanskrit wa
éna, an unorganic w has been prefixed. In regard to
to the ie for z é compare, also, wies-te, “ knowledge,” with
afa védmi, “I know.”
*Remark.—The German has some remarkable expres-
sions, in which the number one lies very much concealed
as to its form, and partly, too, as to its idea: they are, in
Gothic, haihs, “one-eyed,” hanfs, “one-handed,” hails,
“Jame,” and‘ halbs, “half.” In all these words the num-
ber one is expressed by ha; and in this syllable I recog-
nise a corruptionof the abovementioned Sanskrit @ ku for
@a éka, “one,” which is founded on the universal rule
for the mutation of consonants (§. $7.). It would he
erroneous to refer here to the Zend »sw ha of nog7egasy
ha-keret, “ once” (Sanskrit waa sakrit), as the Zend w h
EE
418 NUMERALS.
stands, without exception, for the Sanskrit @ s, to which

the h in Gothic never corresponds.* J. Grimm compares
haihs with cecus (II. 316), not with the purpose of following
out the origin of these cognate words, but in order to
prove the transition of the tenuis into the aspirate; for the
simple aspiration stands in Gothic instead of kh, which
is wanting. These words are, however, so far connected,
that, in both, the word eye is contained. It is only the
question whether the one-eyed in Latin has also Jost the
other eye, and if the blind (cécus), in regard to etymology,
has not preserved one eye left. This appears to me
more probable than that the blind in Gothic should reco-
ver his sight, though but with one eye. The theme of
haihs is HAIHA: one may, then, divide HATHA into
HA-IHA or into H-AITHA; thus the latter portion of this
compound word is assuredly connected with the word wy
aksha, “eye,” in Sanskrit, which only occurs at the end of
compounds ; so that of the compounded q ksh only the first
portion is left, while the Zend sygyas ashi, “eye ”—\which, in
like manner, I have found only at the end of compound words,
aS Fras csvas-ashim, “the six-eyed”—has pre-
served the last element: the Latin ocus, however (the primi-
tive base of oculus), preserves only the first like the Gothic. If
in HAIHA the diphthong ai is left entirely to the share of the
eye, we must assume that the a is introduced through the
euphonic influence of the h (§. 82.), and that 47H stands for
[G. Ed. p.431.] JHA, and this for 4HA; as fimf from
wa pancha; fidvér from arate chatwar. But ifthe aof AIHA
is allotted to the numeral, which appears to me more correct,
then the A in this word has not introduced any euphonic a,
because, with the aid of the first member of the compound, the
* Connected, however, with this designation of “one,” which is taken

from the pronominal base sa (Greek 6), may be the Greek 4 in d-aAovs,
disposition of the h to ai was already satisfied. We must

further recall attention to the Latin cuocles, in which,
however, the notion of unity is evidently represented only
by the ec, for the c must be left to the ocles as a derivative
from oculus: c@cus, however, if @ is the correct way of
writing, and if the number one is contained therein, would
spring from ca-icus; and the Indian a, therefore, is weak-
ened, as in Gothic, to i, which, in Latin compounds, is the
usual representative of an a of the base (§.6.). Let us
now examine” the one-handed. Its theme is, in Gothic,
HAUFA, nominative abbreviated haufs; so that here, as ina
skein, two bases and a pronominal remnant, as mark of case,
lie together. The numeral is here the most palpable ele-
ment: it is more difficult to search out the hand. In the
isolated state no theme nfa could be expected; but in com-
pounds, and also in prefixed syllables of reduplication,
a radical vowel is often rejected; as, in the Sanskrit
witaa jagmima, “ we went,” of the root 7 gam, only gm is
left; and in the Greek, aixtw for aimétw, ET, which corre-
sponds to the Sanskrit wq pat, “to fall,” is abbreviated to
mt. We shall, therefore, be compelled to assume that a vowel
has fallen out between the n and fof HA-NFA. Ifit was ani
which was displaced, then NJF'A might pass as a transposi-
tion of the Sanskrit wf pdni, “hand,” with f for p, accord-
ing to §.87. In H4A-LTA, “lame ”—nominative halts—must
ha again pass for a numeral, and ha-/ta may originally signify
“ one-footed,” for it is (Mark ix. 45.) opposed to the Gothic
tvans féluns habandin, “having two feet,” where it is said
‘it is better for thee to enter into life with one foot, than
having two feet to be cast into hell.’ It is at least certain,
that a language which had a word for one-footed would
very fitly have applied it in this passage. If the last element,
however, in HA-LTA means the foot, we must remember
that, in Sanskrit, several appellations of this member are
derived from roots which mean “to go.” Now, there is, in
EE2
420 NUMERALS.
Gothic, a root LITH, “to go,” with an aspirated f, indeed;

but in compounds the consonants do not always remain
on the same grade which they adopt in the simple word;
[G. Ed. p. 432.] e.g. the t of quatuor appears as d in many
derivatives and compounds, without this dthereby dissembling
its original identity with the ¢ of quatuor and “aq chatur.
So, then, HA-L7'A may stand for HA-LITHA; and it may
be remarked, that from the root LIT comes, also, lithus, “the
limb,” as that which is moveable. Before I pass on to
the explanation of halb, I must mention that J. Grimm
divides the pronoun selber, as it appears to me very pro-
perly, into two parts; so that the syllable st of the
Gothic silba devolves on the reciprocal (sci-na, si-s, si-k).
With respect to the last portion, he betakes himself to
a verb leiban, “to remain,” and believes that silba may,
perhaps, have the meaning of “that which remains in
itself, enduring.” Be this as it may, it is clear that halbs
—the theme is HAL BA—might be, with equal right, divided
into two parts; and it appears to me, that, according to its
origin, this word can have no better meaning than, per-
haps, “containing a part”; so that the ideas one and a
part, remnant, or something similar, may be therein ex-
pressed, and, according to the principle of the Sanskrit
possessive compounds, the notion of the possessor must be
supplied, as in the already explained haihs, “having one
eye.” In the Gothic, also, laiba means “remnant.” It
scarcely needs remark, that halb is no original and simple
idea, for which a peculiar simple word might be ex-
pected, framed to express it. The half is one part of the
whole, and, in fact, equal to the absent part. The Latin
dimidius is named after the middle through which the division
went. The Zend has the expression »s§ 4 naéma, for halb,
according to a euphonic Jaw for néma, which in Sanskrit,
among other meanings, signifies “part”: this is probably
the secondary meaning, and the half, as part of the whole,
the original. If it is so, 44 néma appears to me a very

ingenious designation for a half, for it is a regular contrac-
tion of 4 na, “not,” and ¥€ ima, “this or that”; and the
demonstrative therefore points at the “this or that” portion
of the whole excluded by the negative na. In Sanskrit,
halb is termed, among other appellations, arf@ sdmi, in
which one recognises both the Latin semi and the Greek yz,
and the three languages agree in this also, that they use
this word only without inflection at the beginning of com-
pounds. As to its origin, atfa sdmi may be viewed as a
regular derivative from #a sama, “ equal,” “ similar,” by a
suffix i, by which the suppression of the final vowel, and
widening of the initial vowel of the primitive, become neces-
sary. If this explanation is well founded, ([G. Ed. p. 433.]
then in this designation of halb only one part of the whole,
and, indeed, one equal to the deficient part, would be ex-
pressed, and the atfa sami would be placed as érepov over
against the deficient ETEpov ; and the Sanskrit and German
supply each other's deficiencies, so that the former expresses
the equality, the latter the unity, of the part; i.e. each of
the two languages only semi-expresses the half. As to
the relation, however, of the Greek 7uscvs to Hus, it follows
from what has been already said—that the latter is not an
abbreviation of the former, but the former is a derivation
from the latter; and indeed I recognise in ov the Sanskrit
possessive swa, “suus,” which, remarkably enough, in Zend
enters into combinations with numerals with the meaning
“part”; e.g. ass thri-shva, “a third part,” ara 7Gasgs
chathru-shva, “a fourth part.” In the accusative these
words, according to §. 42., are written ¢9xy576 thri-shd-m,
¢prBr7Gasys chathru-shim, of which the last member comes
_very near to the Greek cuv of jyiouv. “Hyui-ovs means
therefore, “having one equal part,” and the simple yy
means only the equal. The Sanskrit designation of “the
whole” deserves further to be mentioned, #a@qa_sa-kala-s,
422 NUMERALS.
which, as signifying that which joins the paris and unites them,
is opposed to the German halb as applying to one part, and
in a measure furnishes a commentary and guarantee for the
correctness of my view of the latter. The word wae sakala
consists, though this is scarcely perceptible, of @ sa, “ with,”
and a@at kald, “part,” so that, if the latter is regarded
in the dual relation—and the last member of a compound
may express each of the three numbers—aa@ sakala ex-
presses that in which the two parts are together. Thus the
word @Aayq sam-agra, “ full,” is used especially in regard to
the moon, as a body with points, i.e. that in which the two
points touch one another. Transposed into Greek relations of
sound sakala-s would give, perhaps, 6xaAos, or oxeAos, or
6xoAos; but from this the present dAog has rejected the middle
syllallable, as is the case in xépos, Kovpos, compared with
wate kumdra-s, “a boy.”
309. IL. The theme of the declension is, in Sanskrit, dwa,
which is naturally inflected with dual terminations: the
Gothic gives for it tva, according to §. 87., and inflects it, in
the want ofa dual, as plural, but after the manner of pronouns:
[G. Ed. p.484.] nominative tvai, tvds, tva; dative tuaim; ac-
cusative tvans, thvés, tva.* The Sanskrit displays in the dual
* One would expect ¢vé, on account of the form being monosyllabic

(§.281.). In the genitive masculine and neuter I should look for tvi-zé,
after the analogy of thi-zé, “horum,” from THA,or tvaizé, according to the
analogy of the definite adjectives (§.287. p. 3874 G. ed.), and according to the
common declension tv’-é (p. 276). However, the form tvaddyé occurs three
times in the sense of duorum; whence it is clear that the genitive of the
base T'VA was no longer in use in the time of Ulfila. The form tvaddy’-é
belongs to a theme TVADDYA (as hary’-é from HARYA), and appears,
from the ordinal number, which in Sanskrit is dwi-tiya for dwa-tiya, to
have introduced itself into the cardinal number. From ¢vaddyé, by
rejecting both the d—of which one is, besides, superfluous—and by
changing the gy into a vowel, we arrive at the Old High German zueié,
according to Isid. zweiyé, as fior from fidvor ;also definite, zweiéré, which,
in Gothic, would be tvaddyaizé. Grimm appears, on the other hand, to
have
no difference between the pronominal declension and the

ordinary one, and dwéu is declined like vrikdu (p. 274),
dwé feminine like dhdré (p. 285), and dwe neuter like dané
(p. 276). As, however, the notions of number are much
akin to those of the pronouns; and as wet alpa, “a little,”
forms, in the nominative plural masculine, way alpé (§. 228.);
so from the masculine theme dwa, if it had a plural,
might be expected dwé, to which, according to §. 78., the
Gothic tvai would correspond, which it is not requisite
to regard like adjectives terminating similarly, as if com-
pounded with a definite pronoun, espe- [G. Ed. p. 435.]
cially as a genitive tvaizé, which would make the latter
view necessary, does not occur. To tvai corresponds, also,
bai, “both,” from the theme BA, neuter ba, dative baim, accu-
sative masculine bans, which is to be deduced through
apheresis from the Sanskrit base ubha, Old Sclavonic oba
(nominative and accusative dual), from the base OBO. In
Zend the masculine of the number two is 25»4 dva (for dvd,
§. 208.), with which the Old Sclavonic dva is identical, while the
feminine neuter dvye answers to the Sanskrit dwé (§. 255. e.).
The Zend neuter is duyé, with euphonic y (§. 43.), and the v
resolved into u. In the Greek and Latin dvw, dvo, duo, the
have taken occasion, from the Old High German forms, to suppose a
Gothic tvaiyé and tvaiaizé, in which I cannot agree with him. The Old
Northern, by exchanging the dental medials with gutturals, gives tvaggya
for the Gothic ¢vaddyé. In the accusative plural feminine is found, in
Gothic, together with ivés also tveihnés, which presupposes a masculine and
neuter base TVETHNA. fem. TVEIHNO; and in which the an-
nexed HNA reminds us of the appended pronoun ¥q sma, discussed
at §. 165. &c., which, by metathesis, and with the alteration of the s into
h, has in Prakrit and Pali taken the form mha (comp. §. 169.). On this
Gothic TVEIHNA is based the Old High German nominative and
accusative masculine zuéné with léss ofthe A. The feminine, however,
appears in Old High German free from this addition, and is in the nomi-
native and accusative =ud, also abbreviated zua (comp. §.69.).
424 NUMERALS.
old v is, in the same way, resolved into the u, but the final
vowel of the base is not abandoned: d%w answers to the
Védie masculine dwd (§. 208.); but in distinguishing the
genders the Greek is surpassed by the Latin and the
other European sister languages. The Lithuanian has du
in the nominative masculire, and dwi in the nominative
feminine; with the closer explanation of which, and
their dual declension, we will not here occupy ourselves
further. It is, however, to be remarked of the Sanskrit nu-
meral, that the a of dwa is, in the beginning of compounds,
weakened to i (compare §. 6.): hence dwi, which is repre-
sented by the native grammarians as the proper theme
(comp. p. 102). The Greek, in which $F: is inadmissible,
gives in its stead 3:; hence, d:uj7wo = fgata dwimdtri (theme),
“having two mothers.” The Zend and Latin agree in
the corruption of this dwi very remarkably, in this point,
that they have both dropped the d and have both hardened
the v to b; hence ayasceusessdss bipaitistana, “with two
nipples,” like biceps, bidens, and others. From this abbre-
viated bi, comes, in both languages, also the adverb bis,
“twice,” in contrast to the Sanskrit dwis and Greek
dis: the Greek 3, however, in compounds, cannot be re-
garded as an abbreviation of d/s, as is wont to be done.
The German dialects, with exception of the Old High Ger-
[G. Ed. p. 436.] man, require, according to §. 87., tvi for dvi,
as the initial member of compounds; this is furnished by the
Anglo-Saxon in compound words like (vi-féte, “ bipes,” tvi-finger,
“duos digitos longus,” tvi-hive, “bicolor.” The Old High
German gives zui (=zwi) or qui; e.g. zui-beine, “ bipes,”
qui-falt, “duplex” (Grimm III. 956.). The adverb zuiro,
more fully zuiror, also quiro, “twice,” belongs, according to
its formation, but not without the intervention of another
word, to the above dwis, dis, bis ; but it is clear, from the
Old Northern tvis-var, that ro has arisen from sva by
apocope of the a and vocalization of the v, perhaps more
anciently to u, and thence to o (§.77.) as in déo (also diu),

“a servant,” genitive diwe-s, from the base DIWA.
Whence comes, however, the Old Northern svar, which
occurs also in thrisvar, “thrice,” and with which the En-
glish ce in twice, thrice, is connected. I believe that
the s, which precedes the var, is certainly identical with -
the s of fzq dwis, dic, and faa tris, tpic, but the an-
nexed var corresponds to the Sanskrit substantive vdra,
which signifies period and time; hence ékavdra, “once”
(see Haughton), and vdramvdram, “repeatedly.” Hence .
comes the Persian bdr, e.g. bdr-i, “once”; and as the
original meaning of this word is “time,” and we have
already seen, in Persian, the transition of the v into 5, we
may hence very satisfactorily explain the Latin ber in
the names of months; and Septem-ber, therefore, is literally
the seven-time, i.e. the seventh time-segment of the year.
But to return to the Old Northern svar, in trisvar, thrisvar,
which we must now divide into fris-var, thris-var, accord-
ing to the explanation which has been given, the idea of
time, is expressed therein twice, which is not surprising,
as in the Old High German mériro, also mentioned above,
the comparative suffix is twice contained, because it is no
' longer felt the first time, by the genius of the language,
with sufficient clearness. As then, in Old High German,
first the r, and more lately also the o (from v), of s-var has
been dropped, we see, in the Middle High ([G. Ed. p. 437.]
German drir, from dris, the form again returued into the
original limits of the Sanskrit-Greek tris.
310. Ill. The theme is, in the Sanskrit, Greek, Latin,
Lithuanian, and Old Sclavonic, ZRJ, whence in the
Gothic, according to §.87., THAI, and exactly the same in
Zend, according to another law of sound (§. 47.). The
declension of this base is, in most of the languages
mentioned, perfectly regular: it is only to be remarked
of the Gothic, in which, however, all the cases cannot be
426 — NUMERALS.
adduced, that on account of the word being monosyllabic,

the i is not suppressed before vowel terminations, but
becomes ty (compare the Pali, §. 226.): hence the genitive
thriy-é, and nominative neuter thriy-a (§. 233.). Besides
these, the dative thri-m and the accusative thri-ns may be
cited. The Sanskrit forms the genitive from an extended
theme traya, hence trayd-n-dm; while the Zend thry-anm ,
or thray-anm comes from the original base. Both lan-
guages, however, agree in this, that f# éri, 8 thri, is
only a theme of the masculine and neuter; and although,
according to its termination, it might quite as well be
assigned to the feminine, nevertheless the feminine num-
ber has an appellation peculiar to it, which is rather
different from fri, thri, of which the theme is fisar (fay
tisri, §. 1.), the a of which, in the Sanskrit nominative,
accusative, and vocative, is irregularly suppressed; hence
faaa tisras} for tisaras, Zend Urrsuasseo tisaré.
[G. Ed. p. 488.] 311. IV. The Sanskrit feminine theme
‘amaq chatasar (chatasri) follows the analogy of the tisar
just mentioned; and the similarity between the two forms is
so great that it appears, which is perhaps the fact, that the
number three is contained in the fourth numeral; so
that tisr-as would be a weakened form of tasr-as, and
the cha prefixed to the number four would be identical
with the particle, which means “and,” and which, in other
places, is attached to the end of the word. If one wished
to press still farther into the deep mystery of the appel-
lations of numbers, one might moot the question whether
* With this extended theme one may compare the Old High German
nominative masculine drié in Isidor, which belongs to a theme DRIA,
with pronominal declension. The feminine drié, from the base DRI1O0,
of the same sound, presupposes in like manner a masculine and neuter
theme DRIA.
+ In the accusative, tisras is more organic than free tisris, as it must
stand according to the common rule (comp. §. 242.),
the syllables tasa in the theme cha-ta-sar, might not be

considered as identical with the demonstrative bases of the
same sound. I do not think, at least, that any language
whatever has produced special original words for the
particular designation of such compound and peculiar
ideas as three, four, five, &c.; and as the appellations of
numbers resist all comparison with the verbal roots,* the
pronominal roots remain the only means by which to
explain them. Without attempting to resolve the diffi-
culties in the individual numbers, we will express the
conjecture, that the operation of speech with regard to
the numbers might originally be expounded nearly in
this manner—that one might perhaps say, “it, this, that,
and it, and this,” &c.: thus the pronouns might actually
suffice better than they appear to do in the forms of
numerals which lie before us. But an obscuration of
the original clearness of this method, which would occur
in the course of time, would be owing also [G. Ed. p. 439.]
to this, that a simple or compound word might undertake im-
mediately to designate this or that number, and no other
one, though equally adapted to denote it.
312. The masculine and neuter of the number four have,
in Sanskrit, 4ratt chatwér as the strong theme, and 4qq chatur
as the weakt; hence, nom. masc. chatwér-as, accus. chatur-as,
nom. accus. voc. neut. chatwér-i: the gen. masc. and neut.
is irregularly chatur-n-dm for chatur-dm, since, according
to the analogy of bases terminating with a vowel, a nasal
#* Only in three might one perhaps think of the Sanskrit root q ¢77,
“ trans-gredi,” and consider three, therefore, as the more (than two).
This verbal notion of passing over, adding, is,however, also the only pos-
sible one which could be blended with the names of numbers.
t To §. 129. is further to be added, that from the strong theme springs
also the form of the nom., ace., and voc. plural of the neuter; while this
kind forms the whole singular and dual from the weak theme.
428 NUMERALS.
is introduced (§. 246.). In the Zend the strong theme is

Jus Guys chathwér, according to §. 47.; hence, nom. mase.
Vases rsys chathwdré ; and the weak theme is, by trans-
position, Gass chathru; as, chathru-mdhim, “ four months”
(accus. sing.), Vend. 8. p. 248. For the Sanskrit genitive
‘ara chaturndm, we find Gyrjavr7Gasys chathrusnanm (1. ¢.
pp. 204 and 206, with @ inserted, G wepsasr7 Gass chathrusa-
nanm); but in the beginning of compound words it is
more frequently found chsh arses chathwaré; so that the
weakening consists merely in the shortening of the 4, and,
according to §.44., an @ is added to the r; as chalwaré-
poitistanydo, “of her with four teats” (gen. fem., Vend. S.
p- 83). As to the European sister languages, one must
expect, according to §. 14., for ch, gutturals and labials,
hence, in Gothic fidvér, and aspirates for smooth letters,
according to §.87. .This fidvér is based on the strong theme
“ait chatwdr, but in the state of declension extends the
theme by an unorganic i, hence dative fidvéri-m, the only
adduceable case. In Old Northern the nom. masc. is fiéri-r.
[G. Ed. p.440.] The original theme jfidvér appears in the
compound fidvér-tiguns, “ forty” (accus.): on the other hand,
jidur in fidur-dégs, “four days,” is referable to the Indian
weak theme chatur; whence, however, it should not be
said that the weak theme of the German, Lithuanian, and
Sclavonic has been brought from an Asiatic original site,
for it was as easy for the Gothic, by suppressing the last
vowel but one, to contract its fidvdr to fidur—like thiu-s
“servant,” from thiva-s, gen. thivi-s—as for the Sanskrit to
abbreviate chatwdr to chatur. The Lithuanian theme fol-
lows the example of abbreviation in its interior, but
extends the theme at the end; the masc. nom. is kefuri,
and the feminine keturios: KETURIA serves the latter as
theme: the masculine keturi is analogous with ger, “ the
good” (see p. 251, Note {), and therefore has KETUR/E,
euphonic for KETURIA, as its base. The genitive and
CARDINAL NUMBERS, 429
accusative masculine keturi-d, keturi-s, proceed from the base

KETURi. The Old Sclavonic gives CHETYR/ as the mas-
culine and feminine theme, and inflects the masculine like
GOSTTI, and the feminine like KOSTT (p. 349); hence nom.
chetyry-e, chetyri, just as in the third numeral triy-e, “tri”; and
the feminine form may, in both, represent also the masculine,
and always supplies the neuter. But the collective chet-
vero, and the ordinal number chetverty-z, stand in closer
agreement with the Indian strong theme “rat chatwér:
the Latin guatuor, also, which, in disadvantageous comparison
with the cognate languages, has lost the capability of declen-
sion, and the Greek téccap-es, tétTap-ec, rest on the strong
waite chatrdras ; so that tétrapes, just like the Pali form
sara chattérd, has gained its last £ by assimilatson of the
semi-vowel. The Prakrit form, also, which I am not able
to quote, will scarcely be other than chaftéré (comp. §. 300
p- 414 G. ed.). With regard to the initial 7 let reference be
made to §. 14., by which thistis accommo- ([G. Ed. p. 441]
dated with the Zolic misvpes, which refers itself to the weak
theme Wag chatur. With the Zend transposition of the weak
theme to chuthru(p. 439 G. ed.), at the beginning of compounds,
agrees surprisingly the Latin quadru, in quadrupes and other
words. The adverbial s, by which fra dwis, “ twice,” and
fra tris, Zend thris, “thrice,” are formed, is, in the San-
skrit chalur, suppressed by the rule of sound mentioned
in §. 94.; hence chatur, “four times,” for chaturs. That
the latter has originally existed one learns from the Zend
transposed form 41997 Gags chathrus. The Latin has already,
in the number three, without being forced by a compulsory
law, dropped the s, and hence ter and quater appear only
as internal modifications, of the cardinal numbers.
313. V. Sanskrit q panchan, Zend Jpyro panchan.
Lithuanian penki,* Greek évte, /Eolic meunxe, Gothic
* This is the nominative masculine ; the feminino is penkios, and holds

the
430 NUMERALS -
Jimf,*® Latin quingue, Old Sclavonic pycty.t The Sanskrit

Zend panchan is the theme, and the genders are not dis-
|G. Ed. p. 442.] tinguished in this and the following num-
bers; hence the nominative, accusative, and vocative have
always singular neuter forms (therefore pancha, according to
§. 139.): the other cases shew plural terminations; as, geni-
tive waram =panchdndm, Zend GH 9A panchananm
(Vend. S. p. 52). By this irregularity in the declension the
Sanskrit and Zend prepare us in a measure for complete want
of inflection in Greek and Latin. Moreover, it is remark-
able that not one of the European languages will at all recog-
nise the final nasal, while, nevertheless, that of saptan.
navan, and dagan is found also in Gothic and Lithuanian;
and in Lithuanian, also, that of weq ashtan, “eight”
(aszlini). The Greek has frequently preserved an old a
the same relation to it that keturios does to keturi (p. 428). The‘same
obtains with the appellations of the numbers 6, 7, 8, 9, of which we give
only the masculine.
* Occurs only uninflected: in the declined theme, the unorganic addi-
tion of an i must be expected, as in FTDVORT; and as is also actually
the case in Old High German in this number, and the appellations for the
six to ten inclusive. In Gothic, however, occur also saihs, “six,” sibun,
“seven,” ahtau, “eight,” and taihun, ‘ ten,” only uninflected, and there-
fore without the unorganic i; but from niun, “nine,” comes the genitive
niun-é, which indeed might also have proceeded from a theme WJUN or
NIUNA, but which I doubt not comes from WIJUNT.
+ The theme is PYAT'J, and is inflected like KOSTT (p. 348), and
with singular terminations; so that one has to look upon this nume-
ral as a feminine collective, beside which the object numbered stands
in apposition in like cases. The same obtains with the appellations for
the numbers 6 to 10 inclusive. As to the formal relation of PYATJ
to panchan, we must observe, that of the latter, in Sclavonic, only the
syllable pa is represented by pya (§. 225. n.); but 7 is a derivational
suffix, as in SHESHTT, “six,” DEVYATT, “nine,” and DESYATT,
*‘ten,” and corresponds to the Sanskrit suffix ¢i in the multiplied numbers
vinsati, “ twenty,” shashti, “sixty,” &c.
before a nasal originally there, while it has preferred

weakening the same to e before other consonants; hence
éruva(y. v), Erufav, but éruvpe(r) ; TéTUpa(ut) but rérude(re ;
and so éxra, évvéa, déxa: not révta, however, but xévre. It
might therefore well be assumed, that the nasal in Indo-
Zend numerals is a later addition, but that cha is the par-
ticle signifying “and,” which, in the number four, we have
taken for the prefix (§.311.). In Latin, also, guinque is, in
regard to its termination, similar to words connected with
the particle que, as in wévre the enclitie te, which is akin
to que and cha (see §. 14.) appears to be contained. This
Leimg the ease, I would prefer regarding pan in 73 pancha
as euphonic for pam, and the m as a neuter case-sign ; but
the pa which remains over as a pronoun, and indeed as
identical with the ka which occurs in the number one (§. 308),
in regard to which one might advert to the [G. Ed. p. 443.]
old Latin pidpid for quidquid, motos for xotos, &c. Five would,
therefore, literally mean “and one,” and in fact that one
which is to be added to four.*
314. VI. Sanskrit wy shash, Zend ..3».0% csvas, Lithu-
anian szeszi, Old Sclavonic shesty (theme SHESHTT, p. 430,
Note +), Gothic saihs (see §. 82.), Latin sex, Greek €&. One
may justly suppose that the guttural which begins the
Zend word has also existed in Sanskrit, for instance, ay
* Ag. Benary, who likewise recognises in pancha the particle “ and,”’

seeks to compare the preceding syllable with pdéni, “hand” (Berl. Jahrb-
1833. II. p.49). If, however, a connection exists between the appellations
of the hand and five, the former word might be named from the number
of the fingers; as one might also venture an attempt to explain digitus
and éaxrvAos with the number “ten,” and our “finger,” Gothic figgrs
( =fingrs), theme FIGGRA, with fiinf (fimf) ; so that in this word no
transition of the guttural organ into the labial has taken place. I do not
think it probable that finger in named from fangen, “to seize’; also. as
far as regards the Greek and Latin, the appellation of each single finger
is more likely to be derived from the total number than from pointing
(deixvups).
432 NUMERALS.
kshash, for sh is otherwise not an initial syllable in Sanskrit,

and also no original sound, but that sibilant which is only
admissible with a preceding k (§.21.). In Latin, Greek,
and German the guttural appears to be transposed, for
sex is the transposition of zes.
315. VII. Sanskrit awa saptan, Zend jasodasw haptan, no-
minative and accusative WH sapta, »pdasw hapta (see §. 313.),
Greek ér7d, Latin septem, Lithuanian septyni, Old Sclavonic
sedmy (theme SEDMI). The m of septem and sedmy seems to
me to have been introduced from the ordinal number, which
is, in Sanskrit, saptama, nom. masc. saptama-s, and in Scla-
vonic sedmyi. The same holds good of the termination of
osmy, “ eight,” and the Latin novem, decem, Sanskrit navama-s,
[G. Ed. p.444.] “the ninth,” dasama-s, “the tenth”; for it
is not probable that the n of the Sanskrit cardinal number
has become m in the abovementioned languages, as m is
very frequently corrupted ton, especially at the «nd of words,
where, in Greek, this transition is necessary; while the re-
verse method of the n to m scarcely occurs anywhere.
316. VIII. Sanskrit rq ashtan or wet asi:tdu; from the
former the nominative and accusative ashta, from the latter
again aghidu; Zend yaseoawas astan, nominative aspwas asta,
Lithuanian asztin?, Gothic ahtau, Greek oxrw, Latin octo,
Old Sclavonic osmy (theme OSM/). The Sanskrit ashidu
and the analogous oxtw appear, as it were, in a dual dress
(see §. 206.); nevertheless, ashtdu is, in my opinion, just as
much as ashtan,a bare theme, and has perhaps proceeded
from the latter form, which occurs only in Zend, by the
resolution of the n to u, which is so common (comp. p. 415,
Note ), and the lengthening of the a; if it is not preferred
to develope it from ashtas, according to the analogy of
§. 206. From wet ash tdu comes, by suppression of the last
element of the diphthong, ashid-bhis, as/td-bhyas, ashid-su,
as rd-bhis, &c., from réi, “thing,” “riches,” while ashidn,
in the cases mentioned, forms regularly ashtabhis, ashto-
bhyas, ashidsu (comp. p. 304). The genitive has only one

form, namely, weTaty ashidndm. The strength of the du
of ashidu is preserved, also, in the cognate languages, and
indeed in the Latin octav-us, Greek oydoos for oydoF-oc, and
in German forms as ahtowe-n, dative, according to Notker
the cardinal number from ahtowi-m, from the theme
AHTOWI. But if ashidu were connected in its base with
“tat chutur, “ four,” there would be strong reason for con-
sidering the former form as the dual, expressing four twice,
and for assuming that an unorganic corruption of a dual
termination, which made its appearance in the earliest
antiquity, has grown up with the theme.
317. TX. Sanskrit waq navan, Zend [G. Ed. p. 445.]
Asay navan (nominative and accusative nava), Gothic niun
—by contracting the va to u and weakening the a to i, as is
so common, §. 66.—Latin novem (see §. 315.), Greek évvéa,
Lithuanian dewyni, Old Sclavonic devyaty (theme DEV YATI)
The last two appellations appear foreign to the system of
the other sister languages: they are based, however, as I
have already remarked in another place,* on the facile
interchange of a nasal with the organically corresponding
medial on which, among others, rests the relation between
Bporés and 474 mritas, “ mortuus.” As regards the origin
of this numeral term, there exists a close connection in re-
spect of form with the expression for “new” (Sanskrit nava).
That, however, a relation of ideas actually exists between
the two designations, as Ag. Benary first acutely conjec-
tured (Berl. Jahrb. 1832. ii. p. 50), appears to me likewise
probable; for without recognis‘ng a dual in ashidu, and
without excluding the thumbs in reckoning by the fingers,
the number. nine can still only be thought of with refe-
rence to the earlier numbers, and as next to eight, and
* Historical and Philological Transactions of the Academy of Letters for

‘the year 1833, p. 168,
FF
434 NUMERALS, ">
nine, in contrast with eight or all the preceding’ numbers,
is just as much a new number, as that which is new itself
is always a something later and successive, a this corre-
sponding to the old that. As a case in point, observe
the Latin secundus from sequor. One must also admit that
it would not be surprising if any former number what-
ever, excluding one, were named after the idea of that
which is new, and that this origin is most intimately con-
nected with the pronominal origin of other numerals,
|G. Ed. p.446.] 318. X. Sanskrit ey dasan, Zend
‘pss928 dasan (nominative and accusative dasa), Greek dé«a
Latin decem, Lithuanian deszimt, deszimt’-s and deszimtis (the
two first indeclinable), Old Sclavonic desyaty (theme DESYATI
see §. 313. Note f), Gothic taihun. Concerning the ai and u of
tathun, see §§. 66. and 82.: the consonants have obeyed the law
‘of removal (§. 87.). The Greek, rather than the Sanskrit,
therefore serves as prototype to the Gothic in regard
to the second consonant; and we hive laid down in
§. 2L. the Sanskrit Se as a proportionably modern sound.
‘Tf, then, in this corruption, the Lithuanian and Sclavonie
agree with the Sanskrit, this may be so explained, that
these languages, guided independently by the Sanskrit and
Zend, but with the same euphonic feeling, have transformed
‘an old guttural to a sibilant ;* in which change of sound, how-
‘ever, the Sclavonic, in other cases, goes farther than the
Sanskrit (comp. p. 415 G. ed.). If, however, we desire to base
on historical tradition the peculiar coincidence with the San-
skrit and Zend in the case before us, and some others, we
must arrive at this through the assumption that the Li-
thuanian and Sclavonic races at some period wandered
from their original settlement in Asia, when corruption
* But not universally, where, in Sanskrit, yt s is found; for asman

“a stone,” nom. asmé, is, in Lithuanian, AK MEN, nom, akmii (§. 189.)
and in Old Sclavonic KAMEN, nom. kamy (§. 264.).
shad already entered into the language, which did not exist
_at the time when the Greeks and Romans transplanted the
Asiatic original language to Europe.
319. X[—XX. The smaller numbers are combined with
the expression for ten: Sanskrit watgqa ékddasan, stew.
-dwidasan, THEW trayddasan, WAeWA chaturdagan, &e. ;
Zend yasssasgyyasy
7028 aévandasan (2), yassvasgsg dvadasan ;*
Greek évdexa, dadexa, tpicxaidexa, tescaper- [G. Ed. p. 447.]
kaidexa; Latin undecim, duodecim, tredecim, quatuordecim;
Lithuanian wienolika, dwylika, trylika, keturolika; Gothic
-ainlif (1 C. xv. 5.), tvalif,t fimftaihun, “ fifteen”; Old Sclavonic
chetyrinadesyaty, “ fourteen,” pyatynadesyaty, “ fifteen,” &e,
“Remark.—Before the simple dasan (from dakan) had
-been changed in the Gothic into {aihun, according to the
# These maybe deduced from the ordinals aévandasa, dvadaga (Vend.

S. p. 120). So also chathrudasan, “fourteen,” panchadagan, “fifteen,”
fromechathrudasa, “the fourteenth,” panchadasa, “the fifteenth.” The nasal
-in aévandasa appears to have proceeded from m, and to be an accusative
sign, for the whole stands 1. c. in the accusative (aévandasém). Bythis
doubt is thrown on the aévandasan given above, and perhaps aévédaégan,
‘or, according to the original principle of the compound, aévadasun might
be expected. In one other passage, indeed, occurs the nominative of the
‘ordinal aévandaié (1. c. p. 280): it is, however, clearly a false readi
and the sense requires the accusative, as governed by spshasy iy
frasnaéiti, which Anquetil renders by a atteint ; thus, FEWASG Was P.a5
sostasyadd aévandasém frasnaéiti, “decimum attingit” ; and in
the following analogous constructions the ordinal number also stands
alwaysin the accusative. The form aévandasém, from aévamdasém, is
remarkable, also, in a phonetic respect, because elsewhere in Zend a final
m is not governed by the organ of the following letter.
+ I do not take the tva here, with Grimm (I1.947.), for the neuter, but,
according to the principle of genuine compounds, for the theme (compare
§.112.), whence the nom. masc. tvai. Tva may also—and this appears
to me more correct—be regarded, without the Gothic being conscious of
the formation, precisely as the abbreviation of the Sanskrit dwd, which is
a lengthening of the theme dwa, as ék4 from éka.
FF2
436 NUMERALS, —
comparatively recent law for the alteration of sounds

(compare §. 82.), it may have happened that, through the
very widely-diffused disposition for exchanging the d with
/, and through the not less common permutation between
gutturals and labials—through which, among others, the
relation of fidvér to the Lithuanian keturi and Latin quatuor
becomes explicable—the dasan contained in ekd-dasan
“eleven,” and dwd-dasan, “twelve” (from dakan), may have
passed, in Gothic, into LIBI. Through the dative tva-libi-m,
genitive tva-lib’-é, LIBI is preserved, in fact, as the true
theme; so that each a of dasan is weakened toi. The f of
[G. Ed. p. 448.] the uninflected tval/if is, therefore, not to be
explained according to §. 87., but according to §. 93%; and if
the theme /ibi has not obeyed the law for the mutation of
sounds, the objection, which has been raised by Graff
(Old High German Thesaurus, p. 317) against my ex-
planation, is removed by what has been remarked in
§. 89., for we refer to fidvdér, not fithvér, The Latin
quadraginta, also, for quatraginta, and the Greek d'ydoog for
Sxroos, ERdouos for Exropos, and several others, may be
noticed, in support of the proposition that the nume-
ra] formations in the choice of the degree of the organ of
the consonants have not always remained in the custo-
mary path; and in cumbrous compounds the medials are
more admissible than the smooth letters and aspi-
rates.* To remove the objection which may be taken
on the ground that LIBI is so very different from
the form of taihun, we may remark, that, in French
* The Anglo-Saxon endleofan, endlufan, compared with, tvelf, and

the Old Friesian andlova with twilif, should not make us doubt, since
the Anglo-Saxon ¢o corresponds to the Sanskrit a of dasan and Gothic i
of lif, as in the relation of seafon (Old Friesian siugon) to the Sanskrit
saptan, Gothic sibun. Let, then, the Old Friesian o of lova be regarded
like that of siugon. To the Sanskrit chatwdr, Gothic fidvér, correspond
the Anglo-Saxon feover, Old Friesian fiuwer.
also, the number ten, in compounds like on-ze, dou-ze,

trei-xe, is so remote from the expression of the simple
ten, that one would hardly venture to pronounce the syl-
lable ze to be akin, or originally identical with dix, if it
were not historically certain that onze, douze, &c., have
arisen from undecim, duodecim, and that therefore ze is a
corruption of decim, as dix is a less vitiated form of decem.
If, then, onze, douze, &c., have assumed the appearance of un-
compounded words through the great alteration of the expres-
sion for the number ten contained in them, the same holds good
with regard to our ei/f and zwé/f, in which, perhaps, as in
onze and douze, a connection with ein and zwei may be
recognised, but none with zehn ; and in the English eleven,
also, the relation to one is entirely obliterated. But with
regard to our using for thirteen, fourteen, &c., not dreilf,
vierif, or similar forms in /f, but dreizehn, vierzehn, &c.,
in which zedn is just as unaltered as the drei and vier,
this arises from the Germans having forgotten the old Indo-
European compounds for these numbers, and then having
compacted the necessary expressions anew from the elements
as they exist uncompounded. Nay, even [G. Ed. p. 449.]
the Greek has reconstructed afresh, as well as it could, its
numerals from thirteen upwards, after that the old more
genuine compounds had fallen into disuse; but this has been
done, I must say, in a clumsy, awkward fashion, by which the
addition of a particle signifying and was found requisite in
an attempt at extreme perspicuity, while édexa, dddexa,
move more freely, and are suited to the spirit of the ancient
compounds. The literal meaning, too, of tp:cxaidexa (for
tpidexa) is “thrice and ten,” and the numeral adverb 7zpic,
instead of the bare theme 7pz, is here just as much a mistake
as the masculine plural’ nominative serves as a reproach to
the rercapecxaidexa, and is inferior in purity to the Sanskrit
chatur-dasan, not chatvdras-daéan (chatvdré-dasan). On the
ether hand, the Sanskrit, in the designation of the number
438 NUMERALS.
thirteen, commits a similar error, and awkwardly gives in~

stead of tri-dasan, trayé-dasan—euphonic for trayas-dasan—
where the masculine plural nominative instead of the theme,
which is adapted for all genders, is not well selected. The
Latin tre-decim is therefore a more pure formation, as it
dispenses with a case-sign in the first member of the
compound: just so the Lithuanian try-lika, not trys-lika.
This lika, which concludes the form, in all Lithuanian
adding numerals (eleven to nineteen), exchanges the old d.
for J, as in German, and is therefore as far estranged.
from the simple deszimi’s as the Gothic libt from taihun ;
partly, as the second consonant in lika has maintained
itself in its oldest form received from the Greek, and has
not become a sibilant; so that lika and déxa resemble each
other very closely. The Lithuanian lika, therefore, is de-
rived, like the Gothic Uibi and the French ze in onze, douzey
&e., from the old compound which has been handed down,
and cannot, therefore, be censured for its want of agree-
ment with the simple number ten: it is no longer con-
scious of its meaning, and, like an inanimate corpse, is car=
ried by the living inferior number. As, however, the smaller.
number in these compounds is still living, so that in the
feeling of the speaker the numbers wieno-lika, dwy-lika, &e.
do not appear as independent simple designations of num-
bers—as, perhaps, septyni is felt to be independent of each of
the earlier numbers—so, naturally, in these compounds the
first member has kept tolerably equal pace with the form which
it shews in its isolated state; on which account wieno-/ika, if it
is regarded as an ancient compound from the time of the unity
of language, or perhaps as derived from wateyry ¢kd-dasan,
[G. Ed. p. 450.]| has nevertheless undergone, in its initial
member, arenovation; as also in Gothic ainlif, in Greek évdexa,
in Latin undecim, have regulated their first member according
to the form which is in force for the isolated number one. Ou
the other hand, daddexa is almost entirely the Sauskrit dwd-dasa
(w for4,according to §. 4.), and is as similar to it as possible, as

v (F) in Greek cannot be pronounced after consonants, and in
the first syllable, also, could not assimilate itself to the prece-
ding consonant (compare térrapes from rétFapes), for ddadexa
could not be uttered. In Latin, duodecim has formed its first
member exactly after the simple form: on the other hand,
the French has paid no regard to the form in which the prece-
ding number appears in its isolated state, but has left the
composition entirely in the old form, only with the abbre-
viations which time has by degrees introduced. With refe-
rence to the isolated state of the smaller number, it would
have been, perhaps, necessary in French to have said unze,
deuze, troize, &c. After what has been stated, I think no
one can any longer doubt, that in our eilf (elf) and zwilf,
strange as it at the first glance may appear, a word is cou-
tained expressing the number ten, and identical in its origin
with dasan, déxa, and zehn.. If, however, the older LIBIJ,
lif, and Lithuanian lika, be regarded without the suspicion
arising, that in them corrupt though very common permu-
tations of sounds may have preceded, then one would propose
in Lithuanian a root lik, and in Gothic lif or lib (Gothic
af-lifnan, “relinqui, superesse,” laibés, “reliquie™), which both
signify “to remain,” and are also connected with each other
and with the Greek Aci‘zw (AID). Grimm, who has recog-
nised (II. 946) the original identity of our lif and the Lithu-
anian lika, has perhaps allowed himself to be led astray by
Ruhig in the meaning of these expressions, and deduces the
latter from Jikti, “lingui, remanere,” the former from leiban,
“manere.” Ruhig, according to Mielcke, p. 58, holds lika for
the 3d person plural, since he says, “Composition in the car-
dinal numbers from ten to twenty takes place by adding
the 3d person plural number present indicative lika (from
liki: s. liekmi); scil., the tenth remains undisturbed with the
simple number, e.g. one, two, &c.; which addition, how-
ever, in composition degenerates into a declinable noun of
the feminine gender, according to which, also, the preceding
440 NUMERALS.
[G. Ed. p. 451.] simple number must be regulated.”* The

languages, however, do not proceed so pedantically; and if
they hold any thing understood, as very commonly happens,
they do not expressly state that any thing remains over to
be expressed. It is certain, however, that the Sclavonic lan-
guages, in their expressions for eleven to twenty, do not keep
back any thing to be understood, but form those expressious,
after the loss of the old, no longer intelligible compounds,
anew, with the annexed preposition na, “over”; e.g. in Old
Sclavonic, where the numbers eleven, twelve, thirteen, no
longer occur, chetyri-na-desyaty, “ four over ten.” The ordi-
nal numbers for eleven and twelve are yedinyt-na-desyaty,
“ the first over ten,” vtoryi-na-desyuty, “ the second over ten.”
In the same manner proceeds the twin sister of the Lithuanian
—accompanying it, but corrupted—the Lettish, in which
weenpazmit signifies “eleven,” as it appears to me, with con-
traction of the d(e)s of desmit, “ten,” to z, and overleaping the e.
This procedure in Lettish has no doubt originated from the
older lika being no longer intelligible. If it was to be so
understood, as Ruhig has taken it, its form would be palpable,
and the Lettians might have been satisfied with it. With re-
ference to the composition of the numerals under discussion,
there remains to be noticed a most remarkable coincidence
of the Lithuanian and German with a Prakrit dialect,
which coincidence, when I formerly touched upon this
* Grimm’s view iscertainly much more natural, “ten and one over,
two over.” Only it would be to be expected, if the language wished to
designate the numbers eleven and twelve as that which they contain more
than ten, that they would have selected for combination with one and
two a word which signifies ‘and over, or more,” and not an exponent of
the idea “to leave,” *‘to remain,” It would, moreover, be more adapted
to the genius and custom of the later periods of the language, not to
forget the number ten in the newly-formed compounds, like the Lettish
and Sclavonic. J. Grimm, in his “ History of the German Language,”
p. 246, agrees with ny explanation of ei/f, zwélf, and analogous forms in
Lith. and Sclavonie,
subject,* was not yet known to me, and which has been
since then observed by Lenz in his edition of Urvasi (p. 219).
In this dialect, then, the number ten is pronounced simply
@¢ daha—approaching closely to the Gothic tai/un—but
at the end of the compounds under notice raha: r and J,
however, are, according to §. 17., most intimately connected.
Hitherto only, ae véraha, “twelve,” from gteq dwddasa,
and Weert atthdraha, “eighteen,” from were as! tddasa,
can be cited, but still from them it is probable that the other
numerals too, which fall under this cate- [G. Ed. p. 452.]
gory, have an r for d, apparently to lighten the word loaded
by the prefixing of lesser numbers, by exchanging the d for
a weak semi-vowel. Now it is a remarkable coincidence
that if we were desirous of not seeing a mutation of
letters in this raha we should be led to the root rah, “to
leave,” which is probably identical with the verb, to which
recourse has been had for the explanation of the corre-
sponding Lithuanian and German numeral forms.t I
thought I had exhausted this subject, when I -vas led by
other reasons to the Hindisténi grammar, where I was
agreeably surprised by perceiving that here, also, the
number ten, in the designation of eleven, twelve, &c., has
taken another lighter form than in its simple state, in
which it is pronounced das.{ But in the compounds under
discussion this becomes rah,{ and, for example, bdrah,
* Influence of the Pronoun on the formation of Words, p.27; and
Histor. Philol. Trans. of the Academy for the year 1833, p. 178, &c.
+ The a of rah has been weakened in the cognate languages to i:
hence linquo, Lithuanian liki, Greek Xeirw (fAurov), Gothic af-lif-na.
In respect to the consonants, we refer the reader to §) 20.23.: remark,
also, the connection of the Lithuanian Jaka, “I lick,” with the Sanskrit
root Jib, “to lick.” Since writing this note, I have come to the conclu-
sion that it is better to concur with Benfey, in assigning the Latin Linguo,
Greek Aeixa. Gothic af-lif-na, to the Skr. root rich, from rik, “to leave.”
t The text has des and reh but as these sounds are incorrect, I have
altered them, as well as some other inaccuracies in the Hinddstani nume-
rals which follow.—Translater.
442 NUMERALS,
“twelve,” answers to the abovementioned Praikrit 91tz

béraha, and, like this, has proceeded directly from the
Sanskrit original form gteq dwddasa, without heeding
the form of the simple do, “two,” and das, “ten.” It
may be proper here to quote all the Hindistani compounds
which belong to this subject, together with the corre-
sponding Sanskrit words of which they are the corrup-
tions. We annex, also, the. number twenty, and nine-
teen which is related to it as being twenty less one, as
also the simple lower numbers in Hindistani.
[G. Ed. p. 453.]
. HINDUSTANI. SANSKRIT, NOMINATIVE.
ék 1, ig4-rah, 11, ékddasa 11.
do 2, ba-rah 12, dwidasa 12.
in 3, . térah 13, trayédasa 13.
char 4, chau-dah 14,* chaturdasa 14,
panch 5, pand-rah 15, panchddasa 15.
chhah 6, sé-lah 16, shédasa 16,
sat 7, sat-rah 11, saptadasa 17.
ath 8, athd-rah 18, a-hiddasa_ 18.
nau 9, wunnis 19, dnavinsati (“undeviginti’) 19,
das 10, bis 20, vinsati 20.
320. XX—C. The idea of ten is expressed in Sanskrit
by wifa Sati, ra sat or fa ti; in Zend by s055 saiti, soass
sata, or so ti; and the words therewith compounded are
substantives. with singular terminations, with which, in
Sanskrit, the thing numbered agrees in case, as in ap-
position, or is put, as in the Zend, in the genitive, as
* The retention of the d is here clearly to be ascribed to the cireum-
stance that the lesser number ends with r, although in the Hindistani
eorruption this is no longer present. The Bengali has assimilated the r
to the following d, hence chduddo; but, as a general rule, the Bengali in
these compounds changes the d into r, and in all cases suppresses the
Hindastani h ; as égdro, “eleven,” bdro, “ twelve,” téro, “thirteen.”
+ This form merits particular notice, as, through its 7 for the r found
elséwhere, it comes so near to the Lithuanian and German lika, lif. The
Bengali is shdlo. i
dependent upon it. Oceasionally, too, one finds these

numerals in Sanskrit used adjectively, with plural endings.
Compare, [G. Ed. p. 454]
SANSKRIT. ZEND. GREEK. LATIN.
20, fasta viisati, sesassogh visaiti, eixare, viginti,
30, fiqa frinSat, 20992539376 thrisata, tpraxovra, triginta.
40, chatwérinsat, chathwarésata, tescapaxovta, quadraginta.
50, panchdsat, panchdsata, nevtyjKovta, quinquaginta.
60, shashii, esvasti, éfnkovta, sexaginta.
70, saptati, haptditi, éRdounkovta,t septuaginta.
"80, asiti, asia éydonxovra, — octoginta.
90, navati, navaiti, évevnKovTa, __nonayinta.
100, sata-m, sate-m, €-KaTO-V, centu-m.
“Remark.—I hold sati, gat, sata, ti, to be abbreviations
of dasati, dasat, dasata, and therefore derivations from
dasan, “ten,” by a suffix ti, ta, or ¢: the former is
# The numerals in sata, answering to the Sanskrit forms in éaé, aré

nenters, and occur, like the forms in #2, very frequently in the 6th and
12th Fargard of the Vendidad, but only in the accusative singular, in
which gatém might also belong to a theme sat. That, however, sata is
the theme and the neuter form is clear from Vend. S. p. 230. (in the
7th Fargard), where puncha satém (panchdsatém), “ fifty,”’ stands as nomi-
native. Fromesvasti, “sixty,” haptéiti, “seventy,” and navaiti, “ninety,”
we find the accusative csvastim, haptditim, navaitim: on the other hand,
in the 12th Fargard, occurs several times vésaiti (also written visati and
visati) as accusative of visaiti, which perhaps is a dual neuter form (two
decades), and according to this would stand for visaiti (§.210.). But if
the final vowel is retained in its original form it is a singular neuter. It
is, however, remarkable, that only this final i, and no other, is again found
in the cognate Latin and Greek forms.
+ This and the following number are renovated forms, in whieh th
first member proceeds unorganically from the ordinal number. We might
have expected énrjxovra, Gxréxovra, for the latter Ion. éySexovra. In
évevnxovra the two v are separated from each other: the -epic form ¢y7-
xovra is more genuine.
444 NUMERALS.
in Lithuanian and Sclavonic, already contained in the

simple deszimt's, deszimtis, Old Sclavonic desyaty. With
regard, however, to the ten being expressed without
abbreviation in the languages mentioned, in compounds,
also—as in Lithuanian dwideszimli (or tis), “twenty,”
trysdészimli (or tis), “thirty,” and in Old Sclavonic che-
tyridesyaty, “ forty.”* pyatydesyaty, “ fifty” —I do not consider
[G. Ed. p. 465.] this as a more true retention of the original
form, but as a new formation. The Lithuanian, too, from
forty upwards, separates the two numbers, and puts the
former in the feminine plural, e.g. keturios deszimtis, “ forty,”
penkios deszimtis, “fifty”; in which it is surprising that
deszimtis, also, does not stand in the plural. The Gothic
method in this numeral category is of comparatively
recent date: it has lost, as in thirteen, &c., the ancient
compound, and gives, in the numbers under seventy
(sixty docs not occur), tigus, masculine, as the expression
for ten, and declines this, and in twenty, thirty, the lesser
number also, with regular plural terminations: hence the
accusatives tvanstiguns, thrinstiguns, fidvértiguns, fimftiguns,
genitive thriyétigvé. The substantive igus, however, is
the etymological quaver to taihun, and LIBI: it is related
to the former essentially, the aspirate having become a
medial (see §. 89.), thus rendering the a, which, in taihun,
is brought in by the rule of sound mentioned in §. 82.,
superfluous. Advert, also, to the Latin medials in ginii,
ginta, contrasted with the Greek xati, kovra, which answer
better to déxa. Tigu-s may be identical with the San-
skrit ordinal .daga, nominative masculine dasa-s, which
occurs only in compounds, as duddasa-s, “the twelfth.”
To this dasu-s, therefore, is related tigu-s in regard to
its u, as fétu-s to pdda-s, “a foot.” In the numbers
seventy, eighty, and ninety, ten is denoted by the neuter
* Twenty and thirty do not occur.

substantive téhund (theme TEHUNDA, genitive téhundi-s)

hence sibun-téhund, “seventy,” ahtau-téhund, “eighty,”
niun-téhund, “ninety.” The é of this TEHUNDA stands
as the representative of the ai of tathun, and I hold DA to
be the ordinal suffix, which has introduced into the com-
mon ordinals another unorganic WJ, or, according to Grimm,
follows the weak declension; hence 747HUNDAN, nomi-
native taihunda, “decimus.” | Hereby, then, it becomes still
more probable that the abovementioned tigus also is
originally an ordinal number. In our New German this
word has transformed itself to zig or ssig (dreissig), and
is found also in siebenzig, achtzig, neunzig, Old High
German sibunzog, ahtozog, niunzog, or -zoc, and zéhanzog
(zoe), Gothic taihuntéhund, “a hundred.” The Sanskrit-
Zend sata, “a hundred,” which is a neuter substantive—
nominative qaa satam, ¢¢ex4s satém—in my opinion owes
its designation to the number ten (dagan), whence it is
formed by the suffix ¢a—the suppression of the final nasal
is regular ;—so that it is to be regarded as an abbreviation
of dasata, as above, yifa sati, wa sat, and the Zend sexs
sata for dasati, &c. This abbreviation, however, which
has given to the word the stamp ofa primi- [G. Ed. p. 456.]
tive expression specially created for the idea “a hundred,”is
proved to be of the highest antiquity by the consentaneous
testimony of all the cognate languages, Greek xatov (éxarév
is, verbatim, “one hundred”), Latin centum, Lithuanian
szimta-s (masculine), Old Sclavonic sto (at once theme and
nominative and accusative neuter).* The Gothic hund and
Old High German hunt (theme HUNDA, HUNT4A) occur
only in compounds, as tva-hiinda, thria-hunda, zuei-hunt,
driu-hunt, where the lesser number is likewise inflected.
That also yfa sati, qa sat, and the corresponding words
* In Zend sta occurs frequently for gata, and just so in the numbers
compounded therewith.
446 NUMERALS.
in the cognate languages, have in the earliest periods lost the
initial syllable of the number ten, and with it the lingual
remembrance of the same; and that in fagfa visisati, $0.32599.4)
visaiti, eikatt, eixoot, viginti, the single elements have lain
together undisturbed for thousands of years, affords a fresh
proof of the agreement of the languages which have most
faithfully preserved their ancient construction. I would
not, however, wish to maintain that the loss of the d of
the number two in the above forms falls under the period
of the unity of languages; and that it may not have hap-
pened that each of the four individual languages, having
become weary of the initial double consonant in a word
already encumbered by composition, may have disbur-
thened itself of the initial sound, as we have above seen
the Latin and Zend, independently of each other, produce
bis from dwis, and bi from dwi, and as, in agreement with
the abbreviation of fasrft vinsati, the Prakrit dialect men-
tioned at p.451G.ed. has laid aside the din the number
twelve also (vdraha for dwdraha). It is remarkable that the
four oldest and most perfect languages of the Indo-European
family in the category of numerals before us, have lost ~
exactly as much of the number ten as the French in the
forms for eleven, twelve, &c.; and the ze of douze is
therefore identical with the Sanskrit sa of fagfa vinsati.
The Sanskrit and Zend, however, in a later corruption
which is unsupported by the Greek and Latin, have
caused the word dasati to be melted down to the deri-
vation suffix ti, and this ¢i corresponds to the French ¢e
of trente, quarante, &c. The numbers which have been
thus far abbreviated begin, in Sanskrit and Zend, with
sixty, ufe shashti (ti euphonic for ti), seya8»20 csvasti. To
the ati of fagfa vinsati sass visati, regularly corresponds
the Doric kati of eikat:, while in the Latin ginti the smooth
[G. Ed. p.457.] letter has sunk toa medial, as in ginta=Kovra
of the higher numbers. In Sanskrit. the # of vinsati,
triasat, chatwdérinsat, is surprising, and one might imagine

a transposition of the nasal, so that in the Latin ginti,
ginta, centum, and in the Gothic HUN DA, “one hundred,”
it would stand in its proper place. For the rest, chatwd-
rinsat shews its relation to the neuter chatwéri (see §. 312.);
as also tpra, Tecoapa, in tprGKovta, TeccapaxovTa, are, in my
opinion, plural neuter forms, with the termination length-
ened in tp:&, and originally, also, in teccapa, as the Ionic
reccapyxovta, Doric tetpaxovra,* Latin quadraginta, prove.
These forms excite the conjecture, that, in Sanskrit, the
introduction of the nasa] may, contrary to the explanation
attempted above, have the same object that, in Greek, the
Jengthening of the termination has, namely, an emphatic
Yepetition of the prefixed number, which is also percep-
tible in the long i of the Zend visaiti, as in the long a of
wary, aegege: FERIA
WA panchdsatém from panchan
(§. 318.), and to which again the length of mevrijxovra,
quingquaginta, runs parallel. The Zend chathwaré, in
“aspassrghsesCays chathwarésata, “ forty ”(Vend. S. p.-380), is
likewise stronger than cha-thru-sata, which might have
‘been expected from §.312. As «2:39 Sata is a neuter, to
‘which, in Greek, xatov or kovtov would correspond, Kxovre
‘therefore, and the Latin ginfa, are best explained as neuters
‘in the plural, by which the neuter nature of tp:& and recocapa
is still more authenticated. An auxiliary vowel, which
‘merely facilitated the combination, and which might be
‘assumed in é&jxovra, would at least be very superfluous in
‘the theme TPI; and it is much more probable that é&y, too
is a lengthened plural neuter. Compare é&d-xis, eEawhods,
and the remarks on zavra and 7oAAd, p. 401, G. ed.
* The for ais explained by §.4. As to the suppression of the vowel

before the o, rerpw answers to rerpa in rerpdxis, terpazAovs, which in like
manner are based on plural neuter forms instead of the theme.
448 ; NUMERALS.
ORDINAL NUMBERS.
321. While, in designating the number one, the greatest

variety obtains amongst the Indo-European languages, they are
(G. Ed. p. 458.] almost unanimous in their designation of the
jirst, which idea none of the languages here treated of derives
from the corresponding cardinal number: Sanskrit wen
prathama-s (nom.), Zend Leesadd frathémé (8. 56..), Latin
primu-s, Lithuanian prima-s, Gothic frum’-s (for fruma-s,
§. 135.), or indefinite fruma (theme FRUMAN, §. 140.), or,
with newly-added superlative suffix, frumist’-s, Old High Ger-
man érisiér, usually indefinite éristo (from the adverb ér, * be-
fore”), Greek mp@ros, Old Sclavonic pervyt. war prathama,
from the preposition pra, has been already discussed (p. 393
G. ed.); so the Greek 2pG@ro¢ is derived from the correspond-
ing preposition mpd, the lengthening of which to zpw accords
with the Sanskrit prd in pratar, “in the morning” (see p. 392
G.ed.). The suffix TO is an abbreviation of the Sanskrit
tama or thama, which occurs even in Sanskrit in [
chatur-tha-s, “ the fourth,” and qa shash-tha-s, “ the sixth,”
as also in Latin in the form of TU in quartus, quintus,
sexlus, while in Greek this abbreviation extends to all the
ordinal numbers, exclusive of devrepos, EBdouos, and dydoos.
In Lithuanian the corresponding 7'A of four runs through
all, but in such wise, that together with septintas, asztuntas,
occur also sékmas, dszmas, which correspond to the Sanskrit
waa saptama-s, WEAR ashtama-s, in which the last portion
of the superlative suffix tama or thama has remained; of
which kind of division, also, WaHH panchama-s,
navama-s, and eyAq dasama-s, partake, which therefore com-
plete, by their suffix, the tha of chaturtha, so that both united
present the perfect word. The Zend agrees herein with the
Sanskrit, only that its USasodasy haptathé agrees more
with septintas than with aWAq saptama-s and septimu-s; and
ORDINAL NUMBERS. 449
that also be wd pug-dhé, “the fifth,” belongs more to

the European cognate languages, in which it comes nearest
to the Lithuanian penk-ta-s. The Lithuanian, however, is
more true to the original form, as its sister, the Zend, has
softened two original smooth letters, as (G. Ed. p. 459.]
in Greek, dy3o0¢ for oxtoos ; and, besides this, has aspirated
the last, rejected the nasal (comp. p. 94, basta from bandh),
and irregularly changed the a to u, as in “ONYX, corre-
sponding to the Sanskrit a@ nakha, “a nail.’ In the
numbers from eleven to twenty the superlative suffix, in
Sanskrit and Zend, is abbreviated still more than in the
simple eqH dasama, »¢¢3453 daséma, and of all the deri-
vational suffix only the a is left, before which the a of the
primitive word must fall away, according to a universal
principle for the derivation of words; as, ZIeq dwd-
dasa, 23339395s»4 dvadasa, “the twelfth”; “watq chaturdasa,
283923997659 chathrudasa, “the fourteenth.” The Latin
appears to prove that this abbreviation is comparatively of
recent date, and it goes beyond both the Asiatic sisters by
its undecimus, duodecimus, not undecus, duodecus ; but has, as it
were, exhausted itself in the effort which the continuance of
these heavier forms has cost.it; and has given up the ana-
logous formations in the very place in which the German
cardinal numbers have lost the old compound in lif: hence,
tertius decimus for the lost tredecimus, &c. An imitation, how-
ever, of the abbreviation which we have just remarked im the
Sanskrit-Zend dasa is supplied by the Greek and Latin in the
forms octav-us, oy30(F)-os, where, of the ordinal suffix, in like
manner, only the final vowel is left: we might have expected
Gydoyos, octomus. In the very remarkable coincidence which
here exists between the said languages, it must seem strange
that, in the remaining designations of the ordinal numbers,
the Latin is a much truer colleague to its Asiatic sisters
than to the Greek; and it preserves this character, also, in
annexing, from twenty upwards, the full superlative suffix
simu-s (from timu-s=aaq tama-s); thus vicesimus or vige-
GG
450 . NUMERALS.
[G, Ed, p. 460.] stmus, trigesimus, as in Sanskrit vinsalitama-s,

trinsattama-s.* In Latin, however, the termination nti or nta
of the primitives is rejected, and in compensation the pre-
ceding vowel is lengthened in the form of é Compare, in
this respect, the comparative formations discussed in §. 298.
The Greek shews its more rare superlative suffix, correspond-
ing to the Sanskrit ¥¥ ishtha, in the ordinal numbers like
eiKooTos, TpLaKooTOs, With the loss of the: of soTo¢, as in ExacTos,
nootos. Here also, therefore, as in Latin, the 71, o1, and vta
of the cardinal number are rejected. The German languages
employ in like manner the superlative suffix in numbers from
twenty upwards: hence, Old High German dri-zugésto, “the
thirtieth,” fior-zugésto, “the fortieth”: but in the numbers from
four to nineteen the 7AN or DAN, in Gothie, corresponds,
according to the measure of the preceding letter (§. 91.), to
the suffix of the cognate languages, as in =qaa chaturtha-s,
rétapto-s, quartu-s, ketwir-ta-s. ‘The N, however, is an unor-
ganic addition, after the principle of the indefinite adjective
declension (§. 285.), which is followed by the ordinal numbers,
with the exception of 1 and 2 in the older dialects; while
the New German has also introduced the definite—vierter,
“fourth,” fiinfter, “fifth,” &e.; hence, Gothic FIMFTAN,
nom. mase. fimfta.t
[G. Ed. p. 461.] |322. From the weakened base fgdwi “two”
(p. 424), and from the fa tri, “three,” contracted to q tri, the
Sanskrit forms the ordinal numbers bya suffix f7ya; hence dwi-
tiya-s, trittya-s.. This suffix is easily recognised in the Latin ter-
* However, this and the higher numbers may follow the analogy of
ékddasa-s, “the eleventh”; hence, also, vinsa, trins-a, &e. In Zend
I am unable to quote the ordinal numbers from twenty upwards.
+ Incompounds like fimftataihunda, “ the fifteenth,” the lesser number
has either preserved the original theme while still free from the n, which
was added more lately,—for the lesser number in these compounds does
not partake of declension,—or fimfta is here the regular abbreviation of
the theme FIMFTAN, since, as I have already elsewhere remarked
(Borl. Ann. May 1827. p. 759), bases in n, in strict accordance with the
Sanskrit, drop the n in the beginning of compounds.
ORDINAL NUMBERS, 451
tius, as also in the Old Sclavonic fretit, fem. tretiya, which, like
all the ordinal numbers, hasonly a definite declension, in which,
however, the particular case occurs, that the defining element
is brought with it direct from the East, while the ty? of
chetwertyi and others, in which, in like manner, a connection
with ata tfya might be easily conjectured, is, in fact, con-
nected with the a tha, TO, TU of 4y@ chaturtha, térapros,
quartus, and has arisen from the indefinite theme in TO
(comp. the collective chetvero, §. 312.), according to §. 255. (d.),
although the simple word in most of the formations falling
under this category no longer exists. The same relation,
then, that chetvertyt, shestyt, have to chaturtha-s, shashtha-s,
“sedmyi, osmyt, have to awa saptama, WEA ashiama; and
pertyi, “the first,” to Y% pirva, “the former ;” which ex-
pressions, in Sclavonic, remain only in combination with
the pronominal base YO (§. 282.). The Zend has rejected
the ¢ of the suffix fzyu, and abbreviated dwi to bi; hence
2339s bitya, asysps76 thritya, in which it is to be remarked
that the y, which is thus by syncope united with the ¢ at a
comparatively later period, has gained no aspirating influence
(§:47.). To this Zend tya corresponds, by similar suppression
of the middle# the Gothic DYAN (from dya, §. 285.) in
THRIDYAN, nom. masc. thridya, the y of which in the Old
High German dritto, has assimilated itself to the preceding f,
in analogy with the Prakrit forms and Greek comparatives,
like Odccwv, xpeiccwy, kpeirtwv, mentioned at p. 402. Still
closer, however, lies the comparison with d:rrés, tprrrds
(8:06, tprccds), which are evidently, in [G. Fd. p. 462.]
their origin, one with the corresponding Sanskrit-Zend ordinal
numbers; and, in respect of their reduplicated consonant, have
the same relation thereto that the Old High German dritto has
to the Gothic thriyda. Regarding tvaddyé, “duorum,
see p. 422, Note *: the place of the ordinal numberis supplied
by the pronoun anthar (see p. 377), Old High German andar,
Middle High German ander. Our zweiter, however, is a new
unorganic formation. The Old Sclavonic ofory? (see §. 297.5
452 NUMERALS.
answers, in respect to its derivation, to the Greek Setrepos,

and, in abbreviation of the base, to the Zend bitya, only that
it has lost also the é of the Sanskrit dwi-ttya, in regard
to which we have, in §. 297., adverted to the Zend gZsdas
b-ydré*, “two years.”
- $23. We give here a general view of the ordinal numbers
in the feminine nominative singular, since in this case the
agreement of all the languages strikes the eye more than
in the nominative masculine. The Gothic forms which do
not occur we give in parentheses, formed theoretically, and
according to the Old High German.
[G. Ed. p. 463.] NOMINATIVE FEMININE.
SANSKRIT. ZEND. GR. DOR. LATIN. GOTHIC. LITHUANIANe) OLD SCLAVONIC.
prathamaé, frathema,' mpora prima, fruma, pirma, perva-ya.
dwitiyd, bitya, Seurépa, altera, anthara, antra, viora-ya.
tritiyd, thritya, tpira, tertia, thridy@, _ tréchid, treti-ya.
chaturthé? —_tilirya, rerapta, quarta, (fidvordd’), ketwirta, chetverta-ya.
panchamé, pugdha, wépnra, quinta, fimfté’, — penkta, pyata-ya.g
shashthd, estva,* extra, seala, saihst@’", szésatay shesta-ya.
saptamé, haptatha, €Bddua, septima, (sibundd’), sékma, sedma-ya.
ashtamd, astema, Oydea, octava, ahtud’, Gszma, osma-ya.
navamaly nduma, evvdtad, nona, niundé’, dewinta,® devyata-ya.®
dasam, ° daséma, Sexdra, decima, taihundé’, deszimta, desyata-yae
ékidasé, — aévandasa,> Evdexdra, undecima, (ainlifts’), wiendlikta, yedina-ya-na-desya:
vinsati tama, visaititéema? elkootd, vicesima, «+++ dwideszimta, vioraya-na-desyaty
# We should read thus §. 297. for byare, as accusative singular (see

Olshausen, Vend. 8. 43).
1 More usually paoirya, masc. paoiryé, by which the Sclavonie pervy?,
pervaya, is, as it were, prepared,
? Also turtyd, masc. turtya-s, on which is based the Zend téirya,
masc. ¢tiiryd. The suppression of the syllable cha might announce the
looser connection of the same with the remaining portion of the word,
and thereby support the conjecture expressed at §. 311.
* The ¢ of pyataya, masc. pyatyi, has nothing in common with the ¢ of
the cardinal number pyaty ; the proper primitive is pya (see p. 430 Note t),
whence PYATT by the suffix TJ, and PYATO, fem. PYATA, by the
suffix T'O, fem. 7'A (see §.322.). The same holds good with regard to
shestaya in relation to shesty, &e.
* By transposition and syncope from csvasta, as must be expected from
the cardinal number x25 »05Gd esvas.
5 Regarding the d for n, see §. 317. 6 See §.319, Note *, p. 435.
ORDINAL NUMBERS, 453
~ “Remark.—As the old a@ of the preposition @ pra has

been weakened to i—as in guingue, answering to panchan
—the Latin prima appears distinct from the preposition
pro, and is decidedly not derived from a Roman soil, but
is, as it were, the continuance of the Indian prathamd, the
middle syllable being east out. A similar weakening of
the vowel is exhibited in the Greek adverb zpiv, which is
hereby, in like manner, brought into connection with the
preposition zpé. In the comparative prior only the pr of
the preposition, which forms the base, is left, as the i be-
longs to the comparative suffix. In Lithuanian the m of
the superlative formation has introduced itself also into
the preposition pirm, ‘before’; but the unaltered pra stands
as prefix. To the same base, however, belongs also pri, ‘by,
before,’ as well isolated as prefixed. The Gothic fruma shews
the same relation to prathamé that the Latin ([G. Ed. p. 464.]
and Lithuanian do: the u of fru has arisen from a through
the influence of the liquid (§. 66.). In the cognate preposition
fram, ‘before, by,’ &c., the original vowel has remained,
and in this form, as in the Lithuanian pirm, the superla-
tive mis contained. On ¥ pra is based, also, faur, ‘ before,”
with transposition of the u of fru-ma, and with a prefixed,
according to §. 82.
NUMERAL ADVERBS.
324. The adverbs which express the ideas “twice,’
“thrice,” “four times,” have been already discussed
(p. 435 G. ed.). Let the following serve for a general
view of them :—
SANSKRIT. ZEND. § GREEK. LATIN. OLD NORTHERN.
dwis, __ bis, dig, sts, tvis-var (p. 436 G. ed.).
_ tris, thris, Tois, ter, thris-var.
chatur,* chathrus, .... Quater, «+6
* According
to §. 94. for chaturs.
454 NUMERAL ADVERBS.
The Greek forms in xc like rerpdkis, mevraxis, &c., in re-

gard to their suffix, do not belong to this class, but xg answers
to the Sanskrit gas (§.21.), the a being weakened to i;
this sas, however, forms adverbs from words which ex-
press a great number, multitude or number, as gatasas,
“by hundreds,” sahasrasas, “by thousands,” bahusas, “of
many kinds,” ganagas, “in swarms.” The original idea of
the suffix in both languages is that of repetition, but e.g.
Satusas is an indefinite repetition of a hundred, while in
éxatovraxis the repetition is strictly defined by the numeral,
How stands it, then, with the Latin forms like quinquies,
sexies, &c.? I believe that in respect to their suffix they are
connected neither with the forms in s like dwis, dis, nor with
[G. Ed. p. 465.] those in xig (as), by suppression of the
guttural; but as ¢oties, quoties, evidently belong to this class,
which are also pronounced quofiens, totiens, this probably
being the more genuine form, as in Greek, in a similar case,
7i0évg is more genuine than tifeio (§. 138.), I therefore
prefer bringing these forms in ens, es, into conjunction
with the Sanskrit suffix vant (in the weak cases vat),
which signifies, in pronominal bases, “ much,” but else-
where, “gifted with,” and the nominative of which is, in
Zend, vans, e.g. chvans, “how much,” for chivans. This
suffix has, in Sanskrit, in combination with the interroga-
tive base ki, and the demonstrative base i, laid aside the
v; hence kiy-ant, iy-ant—weak form kiyat, iyat—nomina-
tive masculine kiydn, iy4n; this ant for vant answers there-
fore to the Greek ENT (nominative masculine ers), e.g. in
pedrtéers, and also to the Latin ens, in totiens, quotiens, which
indeed are, in form, masculine nominatives, but must also be
considered as neuters, as in the participles, too, in nf, the
masculine nominative has forced its way into the neuter.
Now comes the-question whether we ought to divide toti-ens
quoti-ens, or tot-iens, quot-iens? In the former case fof,
quot, would have preserved, in this combination, the @
NUMERAL ADVERBS. 455
which belongs to them, for they are based on the San-

skrit afa tati, “so much,” @fa kati, “ how much “;" and the
ens in toti-ens would, according to that, express the “time,”
and fofi, ‘so much.” In the division tof-iens, however,
we should have to assume that in iens, the abovementioned
demonstrative yr iyant, “so much,” is contained, but in
such wise, that only the meaning of the suffix is still per-
ceived. Under this supposition guingu-ies [G. Ed. p. 466.]
would, accordingly, express “ five-somuch” (times); in the
former case, however, the i, as quinqui-es, octi-es, would have
to pass as representative of the e and o of guinque, octo,
and that of sezies as a conjunctive vowel, or as an accom-
modation to the prevailing analogy. In any case, how-
ever, the identity of the suffix ens, es, with the Sanskrit
ant, from vant, is highly probable. The Sanskrit expresses
_ the idea “ times” from five upwards by kritwas ; as, Taga
panchakritwas, “five times.” This kritwas comes from rit,
“making,” which in sakrit, “once,” is sufficient of itself:
the annexed vas, however, might, by exchange of the ¢
for s (compare §. 156. Note *), have arisen from vat, which
should be given above as the weak theme for vant; as, idvat,
“so much,” ydvat, “how much” (rel.. With srit from
kart (§. 1.) is clearly connected the Lithuanian karta-s,
“time,” a masculine substantive, which, like the defining
number, is put in the accusative, in order to make up for
the adverbs under discussion; e.g. wienan kartan, “ once,”
di karti, “twice” (accusative du), tris kartis, “three
times.” In Old Sclavonic the corresponding krat or kraty
is not declined, and the former appears to be an abbrevia-
* These are neuters, which, in common with the numerals

panchan, “five,” &c. (§.313.), have, in the nominative, accusative, and
vocative, a singular form; in the other cases, plural terminations; while
in Latin quot, tot, like quinque, &c., have become completely inde-
clinable.
456 NUMERAL ADVERBS.
tion of the latter, for it cannot be brought into direct

comparison with the Sanskrit ga krit on account of §. 255. (1.):
kraty, however, is to be deduced from #r@e kritwas, by sup-
pression of the v. With regard to the y for as compare
§. 271.
325. Through the suffix wtdha the Sanskrit forms ad-
verbs in sense and in form, corresponding to the Greek |
in ya, which, therefore; have altered the JT sound of the
suffix into a corresponding guttural, by the usual exchange
of organ in aspirates, as in OPNIX for OPNIO, and in the
forms mentioned at p.401 G.ed. Compare,
[G. Ed. p. 467.]
fut dwi-dhé,* dinya,
faut tri-dhd, Tpi-YXOe
smut chatur-dha, rétTpa-Ya.
weut pancha-dhd, évta-xa.
© “Divided into two parts,” Sav. V. 108.
END OF VOL. I.
14, Henrietta Street, Covent Garpen, Lonpon;
20, SouTH FREDERICK STREET, EDINBURGH.
SATALOGUE OF SOME WORKS

PUBLISHED BY
WILLIAMS AND NORGATE.
gnostic’s Progress, An, from the Known to the Unknown. 268 pp.
Crown 8vo. cloth 58
tlantis, the New; or, Ideals Old and New. Emanating from the
Masters of Man belonging to the Great Races, who conducted the past Course
of the Civilization, Culture and Education of Humanity. By a Disciple of
Buckle. “Crown 8vo. cloth 4s 6d
aur (F. C.) Church History of the First Three Centuries, Trans-
lated from the third German Edition. Edited by the Rev. Attan Mewnzizs.
2 vols. 8vo. 2Is
— Vide Theological Translation Fund Library.
ur (F.C.) Paul the Apostle of Jesus Christ, his Life and Work,
his Epistles and his Doctrine. A Contribution to the Critical History of
Primitive Christianity. Edited by E. Zerzer. Translated by Rev. ALLan
Menzies. 2 vols. 8vo. cloth 2I8
eard (Rev. Chas.) Lectures on the Reformation of the Sixteenth
Century in its Relation to Modern Thought and Knowledge. Hibbert
Lectures, 1883. 8vo. cloth. (Cheap Edition, 4s 6d) tos 6d
card (Rey. Chas.) Port Royal, a Contribution to the History of
Religion and Literature in France. Cheaper Edition, 2 vols. Crown 8vo. 12s
card (Rev. Dr. J. R.) The Autobiography of Satan. Crown 8vo.
cloth . 78 6d
ble for Young People. A Critical, Historical, and Moral Hand-book
to the Old and New Testaments. By Dr. H. Oorr and Dr. J. Hooyxaas,
with the assistance of Dr. Kuznen. Translated from the Dutch by the Rey,.
P. H. Wicxstgep. Vols. I. to IV., Old Testament, 19s; V. VI, New
Testament, 12s. Maps. 6 vols. Crown 8vo. cloth 3Is
leek (F.) Lectures on the Apocalypse. Edited by T. Hosspacn.

Edited by the Rev. Dr. S. Davipson. 8vo. cloth, 10s 6d
2 Catalogue of some Works
Cobbe (Miss F. P.) The Hopes of the Human Race, Hereafter an

Here. Essays on the Life after Death. With a Preface having special}
reference to Mr. Mills’ Essay on Religion. Second Edition. Crown 8yo.}
cloth : 58
Cobbe (Miss F, P.) Darwinism in Morals, and (13) other Essays.
(Religion in Childhood, Unconscious Cerebration, Dreams, the Devil,}
Auricular Confession, etc. etc.) 400 pp. 8vo. cloth (pub. at ros) §s
Cobbe (Miss F. P.) The Duties of Women. A Course of Lectures
delivered in London and Clifton. Second Edition. Crown 8yo. cloth $s
Cobbe (Miss F. P.) The Peak in Darien, and other Riddles of Life
and Death. Crown 8vo. cloth 78 6d
Cobbe (Miss F. P.) Broken Lights. An Inquiry into the Present
Condition and Future Prospects of Religious Faith. ‘Third Edition. Crown
8vo. cloth 58
Cobbe (Miss F. P.) Dawning Lights. An Inquiry concerning the
Secular Results of the New Reformation. 8vo. cloth 58
Cobbe (Miss F. P.) Alone to the Alone. Prayers for Theists, by
several Contributors. ‘Lhird Edition. Crown 8vo, cloth, gilt edges 58
Corvichen (R.) The Philosophy of all Possible Revelation, &c. Post
8vo. cloth 73 Od
Davids (T. W. Rhys) Lectures on the Origin and Growth of Religion,
as illustrated by some Points in the History of Indian Buddhism. Hibbert
Lectures, 1881. 8vo. cloth 10s 6d
Echoes of Holy Thoughts: arranged as Private Meditations before a
First Communion. Second Edition, with a Preface by the Rev. J. Hamitron
Tuom, of Liverpool. Printed with redlines. Crown 8yo. cloth as 6d
Evolution of Christianity, The. By Cuartes GILL. Second Edition,
with Dissertations in answer to Criticism. 8vo. cloth 128
Ewald (Professor H.) Commentary on the Prophets of the Old Tes-
tament. Translated by the Rev. J. Frep. Smirn. Vol. I. Yoel, Amos,
Hozea, and Zakharya, 9-11. Vol. II. Yesayah, Obadya, Micha. Vol. III.
Nahum, Sephanya, Habaqquq, Zakharya, (12-14) Yeremiah, Vol. IV.
Hezekiel, Yesaya (40-46) with Translation. Vol. V. Haggai, Zakharya,
Malaki, Jona, Baruch, Appendix and Index. Complete in 5 vols. 8vo.
cloth each ros 6d
—— Vide Theological Translation Fund Library.
Ewald (Professor H.) Commentary on the Psalms. (Poetical Books
of the Old Testament. Part 1.) ‘Translated by the Rey. E. Jounson, M.A.
2 vols. 8vo. cloth each 10s 6d
——Vide Theological Translation Fund Library.
Published by Williams and Norgate. 3
: Ewald (Professor H.) Commentary on the Book of Job. (Poetical

' Books, Part 2.) Translated by the Rev. J. Freperick SmirH. 8vo.
cloth tos 6d
j —— Vide Theological Translation Fund Library.
Gould (S. Baring) Lost and Hostile Gospels. An Account of the
Toledoth Jesher, two Hebrew Gospels circulating in the Middle Ages, and
extant fragments of the Gospels of the first three Centuries of Petrine and
Pauline origin. By the Rev. S. Barine Goutp. Crown 8vo. cloth. 7s 6d
Hanson (Sir Richard) The Apostle Paul and the Preaching of
Christianity in the Primitive Church. By Sir Ricuarp Davis Hanson,
Chief Justice of South Australia, Author of “‘The Jesus of History,’ “ Let-
ters to and from Rome,” etc. 8vo. cloth (pub. at 12s) 7s 6d
Hausrath. History of the New Testament Times. The Time of
Jesus, by Dr. A. Hausratn, Professor of Theology, Heidelberg. Translated,
with the Author's sanction, from the second German edition by the Revs.
C. T. Porntine and P. QuenzerR. 2 vols. 8vo. cloth 21s
Hibbert Lectures, vide Beard, Davids, Kuenen, Miiller, Pfleiderer,
Renan, Renouf, Reville.
Horne (Rey. W.) Religious Life and Thought. By Witt1am Horng,
M.A., Dundee, Examiner in Philosophy in the University of St. Andrews;
Author of “ Reason and Reyvelation.”” Crown 8vo. cloth 35 6d
Jones (Rev. R. Crompton) Psalms and Canticles, selected and pointed

for Chanting. 18mo. cloth ts 6d
—— Anthems, with Indexes and References to the Music. 18mo. cloth Is 3d
—— The Chants and Anthems, together in one vol. 2s 6d
—— A Book of Prayer in 30 orders of Worship, for public or private Devotions.
12mo. cloth ‘2s 6d
—— The same with the Chants. 18mo. cloth 38
Keim’s History of Jesus of Nazara, considered in its connection with
the National Life of Israel, and related in detail. Translated from the German
by A. Ransom and the Rev. E. M. Getparr. In 6 vols. 8yo. cloth
! each tos 6d
Kuenen (Dr. A ) The Religion of Israel to the Fall of the Jewish
State. By Dr. A. Kuenen, Professor of Theology at the University, Leyden.
Translated from the Dutch by A. H. May. 3 vols. 8vo. cloth 31s 6d
Kuenen (Professor A.) Lectures on National Religions and Universal
Religions. Delivered in Oxford and London. By A. Kuenen, LL.D., D.D., 6d
Professor of Theology at Leyden. Hibbert Lectures, 1882 Ios 6~
Macan (Reg. W.) The Resurrection of Jesus Christ. An Essay —

in Three Chapters. Published for the Hibbert Trustees. 8vo. cloth 5s
Mackay (R. W.) Sketch of the Rise and Progress of Christianity.
8yo. cloth (pub. at 10s 6d) 6s
Martineau (Rey. Dr. James) Religion as Affected by Modern Material-
ism ; and Modern Materialism : its Attitude towards Theology. A Critique
and Defence. 8vo. 2s 6d
The Relation between Ethics and Religion. —8vo.
—— Ideal Substitutes for God considered. 8vo.
Mind: a Quarterly Review of Psychology and Philosophy. Contri
butions by Mr. Herbert Spencer, Professor Bain, Mr. Henry Sidgwick, Mr.
Shadworth H. Hodgson, Professor Flint, Mr. James Sully, the Rev. John
Venn, the Editor (Professor Croom Robertson), and others. Vols. I. to X.,
1876-8 5, each 12s. Cloth, 13s 6d 12s per annum, post free
Miller (Professor Max) Lectures on the Origin and Growth of Religion,
as illustrated by the Religions of India. Hibbert Lectures, 1878. 8vo.
cloth Ios 6d
Oldenberg (Prof. H.) Buddha: his Life, his Doctrine, his Order.
Translated by Witt1am Hoey, M.A., D.Lit., Member of the Royal Asiatic
Society, Asiatic Society of Bengal, &c., of H. M. Bengal Civil Service. Cloth,
gilt 18s
Peill (Rev. G.) The Three-fold Basis of Universal Restitution.
Second Edition. Crown 8vo. cloth 3s
Perrin (R. S.) The Religion of Philosophy; or, The Unification of
Knowledge: A Comparison of the Chief Philosophical and Religious Systems
of the World, made with a view to reducing the Categories of Thought, o
the most general terms of existence, to a single principle, thereby establishing
a true conception of God. 8vo. cloth 16s
Pfleiderer (O.) Paulinism. An Essay towards the History of the
Theology of Primitive Christianity. Translated by E. Peters, Esq. 2 vols.
8vo. cloth
Paul on the Davcicouieat of Christianity. Translated by the Rev. J. FRepERICK

Smitu. Hibbert Lectures, 1885. 8vo. cloth 10s 6d
Poole (Reg. Lane) Illustrations of the History of Medieval Though
in the Departments of Theology and Ecclesiastical Politics. 8vo. cloth 1os 6d
Protestant Commentary, A Short, on the New Testament, with
General and Special Introductions. From the German of Hilgenfeld, Holtz.
mann, Lang, Pfleiderer, Lipsius, and others. Translated by the Rev. F. Hy
Jonegs, of Oldham. 3 vols. 8yo. cloth each Ios 6d
Renan (E.) On the Influence ot the Institutions, Thought and Culture

of Rome on Christianity, and the Development of the Catholic Church.
Translated by the Rev. C. Bearp, of Liverpool. Hibbert Lectures, 1880.
_ (Cheap Edition, 2s 6d) 8vo. cloth tos 6d
Renouf (P. Le Page) Lectures on the Origin and Growth of Religion,

as illustrated by the Religion of Ancient Egypt. Hibbert Lectures, 1879.
Second Edition. 8vo. cloth los 6d
Reville (Prof. Albert) Prolegomena of the History of Religions.
: By Arsert Revitte, D.D., Professor in the Collége de France, and
Hibbert Lecturer, 1884. Translated from the French. With an Introduc-
tion by Professor F. Max Mutter. 8vo. cloth tos 6d
Reville (Prof. Albert) Lectures on the Origin and Growth of Religion,
as illustrated by the Native Religions of Mexico and Peru. Translated by the
Rev. P. H. Wicxsteep, M.A. Hibbert Lectures, 1884. 8vo. cloth 1os 6d
Samuelson (Jas.) Views of the Deity, Traditional and Scientific; a
Contribution to the Study of Theological Science. By James SamMvuELson,
Esq., of the Middle Temple, Barrister-at-Law, Founder and Former Editor of
the Quarterly Journal of Science. Crown 8yo. cloth 4s 6d
Savage (Rev. M. J.) Beliefs about the Bible. By the Rev. M. J.
Savace, of the Unity Church, Boston, Mass., Author of “ Belief in God,”
“ Beliefs about Man,” &c., &c. 8vo. cloth 7s 6d
Schrader (Prof. E.) The Cuneiform Inscriptions and the Old Testa-
ment. ‘Translated with Additions by the Author, by the Rev. O. C. Wu1rTE-
House. Vol. I. witha Map. 8vo. cloth los 6d
Schurman (J. G.) Kantian Ethics and the Ethics of Evolution. A
Critical Study, by J. Goutp Scuurman, M.A., D.Sc., Professor of Logic
and Metaphysics in Acadia College, Nova Scotia. Published by the Hibbert
Trustees. 8vo. cloth 5s
Seth (A.) The Development from Kant to Hegel, with Chapters on
the Philosophy of Religion. By Anprew Sera, Assistant to the Professor
of Logic and Metaphysics, Edinburgh University. Published by the Hibbert
Trustees. 8vo. cloth 55
Sharpe (S.) History of the Hebrew Nation and its Literature, with
an Appendix on the Hebrew Chronology. Fourth Edition. 487 pp. 8vo.
cloth 7s 6d
Sharpe (S.) Bible. The Holy Bible, translated by Samus. SHARPE,
being a Revision of the Authorized English Version. Fourth Edition of the
Old Testament ; Eighth Edition of the New Testament. 8vo.roan 45 6d
Sharpe (S.) The New Testament. Translated from Griesbach’s
Text. 14th Thousand, fcap. 8vo. cloth 1s 6d
Smith (Rev. J. Fred.) Studies in Religion under German Masters.

Essays on Herder, Goethe, Lessing, Franck, and Lang. By the Rev. J.
Freperick SmiTH, of Mansfield. Crown 8vo. cloth 58
Spencer (Herbert) Works. The Doctrine of Evolution. 8vo. cloth
First Principles. Sixth Thousand 16s
Principles of Biology. 2 vols. 348
Principles of Psychology. Fourth Thousand. 2 vols. 36s
Principles of Sociology. Third Edition. Vol. I, 21s
Ceremonial Institutions. Principles of Sociology. Vol. II. Part r 7s
Political Institutions. Principles of Sociology. Vol. II. Part 2 12s
Ecclesiastical Institutions. Principles of Sociology. Vol. II. Part3 58
The Data of Ethics. Principles of Morality. Fourth Thousand. Part 1 8s
Spencer (Herbert) The Study of Sociology. Library Edition (being
the Ninth), with a Postscript. 8vo. cloth Ios 6d
— Education (Cheap Edition, Seventh Thousand, 2s 6d) 6s
—— Essays. 2 vols. Third Edition 16s
Essays (Third Series). Third Edition 8s
Spencer (Herbert) The Man versus the State. 15; or on better
paper, in cloth as 6d
Spencer (Herbert) The Philosophy of M. Comte—Reasons for
Dissenting from it. (Republished from ‘ The Classification of the
Sciences,’ &c., 1864) 6d
Spinoza. Four Essays, by Professors J. Lanp, Kuno FiscHer,
Van Vioren, and Ernest Renan. Edited, with an Introduction, by
Professor W. Kniceut, of St. Andrews. 8vo. cloth 58
Stokes (G. J.) The Objectivity of Truth. By Gesorce J. Sroxkes,
B.A., Senior Moderator and Gold Medallist, Trinity College, Dublin; late
Hibbert Travelling Scholar. Published by the Hibbert Trustees. 8yvo. cloth 55
Strauss (Dr. D. F.) New Life of Jesus, for the People. The
Authorized English Edition. 2-vols. 8vo. cloth 245
Taine (H.) English Positivism. A Study of John Stuart Mill.
Translated by T. D. Haye.- Second Edition. Crown 8vo. cloth 38
Talmud of Jerusalem, ‘Translated for the first time into English by
Dr. Moses Scuwas, of the Bibliothéque Nationale, Paris. Vol. I. The
Treatise of Berakhoth (Blessings). Foolscap 4to. Qs
Tayler (Rey. J. J.) An Attempt to Ascertain the Character of the
Fourth Gospel, especially in its relation to the first three. New Edition.
8vo. cloth 58
Ten Services of Public Prayer, taken in Substance from the “ Common
Prayer for Christian Worship,” with a few additional Prayers for particular
Days. Crown 8vo. cloth, 2s 6d; with Special Collects 38
—— 32mo. Is; with Special Collects 1s 6d
Psalms and Canticles. (To accompany the same.) Crown 8vo. 1s 6d
With Anthems ad
Thoughts for every day in the Year. Selected from the Writings of
Spiritually-minded Persons. By the author of “Visiting my Relations.”
Printed within red lines. Crown 8vo. cloth as 6d
Theological Translation Fund. A Series of Translations, by which
the best results of recent Theological investigations on the Continent, con-
ducted without reference to doctrinal considerations, and with the sole purpose
of arriving at truth, will be placed within reach of English readers. A
literature which is represented by such works as those of Ewald, F, C. /
Baur, Zeller, Roth, Keim, Noldeke, &c. in Germany, and by those of Kuenen,
Scholten, and others in Holland.
Three Volumes annually, for a Guinea Subscription. The Prospectus,
bearing the signatures of Principal Tulloch, Dean Stanley, Professors Jowett,
H. J. Smith, Henry Sidgwick, the Rev. Dr. Martineau, Mr. W. G.
Clark, the Rev. T. K. Cheyne, Principal Caird, and others, may be had.
33 Volumes published (1873 to 1884) for £11. 11s.
Protestant Commentary, a Short, on the New ‘Testament. 3 vols.
Keim’s History of Jesus of Nazara. 6 vols.
Baur’s Paul, his Life and Work. 2 vols.
Baur’s Church History of the First Three Centuries. 2 vols.
Kuenen. The Religion of Israel. 3 vols.
Ewald. Prophets of the Old Testament. 5 vols.
Ewald’s Commentary on the Psalms. 2 vols.
Ewald. Book of Job.
Bleek, on the Apocalypse.
Zeller, on the Acts of the Apostles. 2 vols.
Hausrath’s History of the New Testament Times. 2 vols.
Pfleiderer’s Paulinism. 2 vols.
Reville’s Prolegomena of the History of Religions.
Schrader’s The Cuneiform Inscriptions and the Old Testament: 2 vols.
In the Press.
Pfleiderer’s Philosophy of Religion. 2 vols.
All new Subscribers may purchase any of the previous volumes at 7s instead of
tos 6d per volume. A selection of six or more volumes from the list may also be
had at the Subscribers’ price, or 7s per volume.
Vickers (J.) The History of Herod; or, Another Look at a Man
Emerging from Twenty Centuries of Calumny. Crown 8vo. cloth 6s
Williams (Dr. Rowland) The Hebrew Prophets. Translated afresh
and illustrated for English Readers. Two vols. 8vo. cloth 22s 6d
Wright (Rev. J.) Grounds and Principles of Religion. Crown 8vo.
cloth 3s
Zeller (Dr. E.) The Contents and Origin of the Acts of the Apostles,
critically investigated. Preceded by Dr. Fr. Overseck’s Introduction to the
Acts of the Apostles from De Wette’s Handbook, Translated by Joseru
Dare. 2 vols. 8vo. cloth " ais
8 Catalogue of some Works published by Williams and Norgate.
PAMPHLETS.
Athanasian Creed. ‘Two Prize Essays. By C. Peabody and C. S. Kenny Is
Bastard (T. H.) Scepticism and Social Justice : Is
Beaumont (Rev. G.) Science and Faith. A Discourse Is
Beard (C.) William Ellery Channing. In Memoriam. A Sermon. i12mo. 6d
Beard (C.) The Kingdom of God. A Sermon 6d
Beard (C.) The House of God, and two Sermons by Rev. R. A. Armstrong Is
Bennett (W.) Popular Contributions towards a Rational Theology 1s 6d
Butler's Analogy: A Lay Argument. By a Lancashire Manufacturer Is
Carpenter (Prof. J. Estlin) The Church of England during the Middle Ages _ Is
Drummond (Prof. J.) Philo and the Principles of the Jewish-Alexandrine
Philosophy Is]
Drummond (Prof. J.) Religion and Liberty Is
Gordon (Rev. A.) Gospel Freedom. A Sermon 6d
Hopgood (Jas.) Disestablishment and Disendowment of the Church of
England 6d
Hopgood (Jas.) An Attempt to Define Unitarian Christianity 6d
Howe (Rev. C.) The Athanasian Creed. Two Discourses Is
Infinite Love. A Meditation OK) 18
Jesus of Nazareth and his Contemporaries Is
Journey to Emmaus. By a Modern Traveller 2s
Lisle (L.) The Two Tests : the Supernatural Claims of Christianity tried by two
of its own Rules. Cloth 1s 6d
Marriage of Cana, as read by a Layman 6d
Martineau (Rev. Dr. James) New Affinities of Faith; a Plea for free Christian
Union. 12mo. Is
Must God Annihilate the Wicked? A Reply to Dr. Jos. Parker Is
Odgers (J. Edwin) Our Church Life: its Significance and Value 6d
Reasonable Faith, A, the Want of our Age Is
Sharpe (S.) Journeys and Epistles of the Apostle Paul Is 6d
Sidgwick (H.) The-Ethics of Conformity and Subscription Is
Tayler (Rev. J. J.) Christianity: What is it? and What has it done? Is
The Recent Prosecutions for Blasphemy Is
Who was Jesus Christ? 8vo. sewed 6d
Wicksteed (Rev. P. H.) The Ecclesiastical Institutions of Holland. 8vo, Is
WILLIAMS AND NORGATE,

14, HENRIETTA STREET, COVENT GARDEN, LONDON;
anv 20, SOUTH FREDERICK STREET, EDINBURGH.
ISN
~
AN AN
SX SWAN
TO AN WRK RSV WWW.
SS : WX \ \SSSN\
SS IX \
NK XK ASS
RR QVwyg
SS A \ S
WN A
\ NN
AX
WN
SY AX
SNS
Se SS ~
SSRSV
~ WSS
SS
AX S A
SEN AS
SAA
.S SS
AQ \ \
AN \\ SX SS XX A
WAV SY
SSS NS SS
CX
RR V@7
TARA
RQ
WQQVQ
WS
ASS
: IGG
\ \\ S
\\
Ss AS
Sy SAY x

Comparativegrammar Sanskrit Greek Latin Latvian German

Uploaded by

Copyright:

Available Formats

Comparativegrammar Sanskrit Greek Latin Latvian German

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Comparativegrammar Sanskrit Greek Latin Latvian German

Uploaded by

Copyright:

Available Formats

HONn

Digitizedby the Internet /

TRANSLATED FROM THE GERMAN

EDWARD B. EASTWICK, F-.R.S., F.S.A.

WILLIAMS AND NORGATE:

20, SOUTH FREDERICK STREET, EDINBURGH.

the adoption of the Work as a Lecture Book at Oxford, and

Te study of Comparative Philology has of late years bean

for, is conveyed, are necessary to bring within their reach

the most acceptable of its class to English students: it

tion of the Vergleichende Grammatik. He has accord-

I conremPate in this work a description of the compara-

* Sanskrita signifies “adorned, completed, perfect ”; in respect to lan-

grammatical constitution, it stands in the most intimate relation

endowments of exceeding richness, and, with the capability

* Frankfort.a. M. 1816. A translation of my English revision of

of itself, and, with it, of the entire Germanic stock, to the

not fear that that practical and profound research in utrdgque

of cases produced by the Pehlvi translator as verbal persons,

* J give the Zend expressions according to the system of representation

remark on the age of the Vocabulary to which I advert; while

et fais néaesch.” I do not insist on translating the adjective asguQ valma

one in question cannot therefore be ascribed to any very late

opened field. My observations, derived from the original texts

monitory voices of the Asiatic as well as the European sisterhood.

narily difficult in languages not now thoroughly understood to

tions, while with Grimm it has a dynamic signification. A comparison

this class of words,* and which will, therefore, be treated in

CHARACTERS AND SOUNDS.

_ a short a united with a following ¢ becomes z é (equivalent

as if the diphthong z é united its first element @ with the

presented by 7 or w than by a long alpha: and though in the

* The original has devr, but, as observed in p. 1, in European texts it

various but sure appearances, which I shall further illustrate

ample, pénas, “lord,” at the end of conapounds, is weakened

which admits of more freedom than the genitive. In the

endeavoured, in my Grammar, to remedy this confusion in the

convertible to Anuswiara before any consonant except a semi-vowel or a

the variable nasal, which, under certain conditions, passes into

nating sounds are almost always governed by the following -

terminal, and in pronunciation retains their respective sounds, according

a remarkable accordance with the Sanskrit rule of euphony

* No native scholar would read these as bhaavah or kudhan,as the

respective non-aspirates, with a clearly audible 4; thus, for

* The original here adds—‘‘ We designate the aspirate by a comma,

dig.” As regards the sonant aspirates, the ¥ gh of gharma,

* The original has g and g;

mas, “we cleave,” faafa_chhinadmi, “I cleave,’ answers

* Here, also, it may be doubted if similar modifications of the dental

vowel, or letters of its own class: it otherwise governs itself

from #t dyé. Nominal bases in y do not exist. <r at the end

* It is scarcely correct to say “often,” as the instances are rare: nor

waita anfara-s, “the other,” alter; az vad, “to speak,”

* More usually ¢; the sh is reserved for the eerebral sibilant.— Editor.

belongs to the lingual class. It often steps, according to

* Cf, Hartung, p. 106.

r before ¢. The Latin protects the s usually at the end of

We give here a general view of the Sanserit characters,

Semi-Vowels....... ay, tr, GL av.

* Berlin Journal, Feb. 1827, p. 254.