V.

SASANIAN ART
CHAPTER 28

SAS.ANIAN ARCHITECTURE
A. HISTORY
By OSCAR REUTHER·
ASANIAN architecture must have continued the Parthian tradition . H ence early

S Sasanian architecture was presumably more like late Parthian than late Parthian was
like early P arthian, built four centuries prior; and it would be equally reasonable
to suppose that early Sasanian architecture was closer to late Parthian than it was to late
Sasan ian, constructed four centuries later. Yet this inference, though apparently obvious,
is difficult to sustain, for the relative dates of most of the few existing ruins from both the
P arthian and Sasanian periods are uncertain, the latter being especially subject to di spute.
The palace of Firuzabad, for example, was at one time supposed to be not Sasanian at
all, but built under the last of the Achaemenids, and hence the palace of Sarvistan,' which
is clearly later in style, had, for the moment, to be classed as Seleucid; but the Firuzlibad
palace is now definitely identified as that built by Ardashir, according to Tabari,
prior to his insurrection against the Arsacid Artabanus V in A.D. 224. Hence it is the
earliest Sasanian monument.' Sarvistan, on the other hand, is variously dated between
the fourth and sixth centuries. The divergence of opinion and insecurity in dating
Sasanian buildings are shown by the fact that one scholar formerly classed as Arsacid
the Taq-i-Girra, which is important for the history of architecture because of the use
of the horseshoe arch, and believed that a relatively early dating for it must be conceded,
but now this same scholar considers it late Sasanian.' The Taq-i-Kisra, the vast ruin
of the royal palace of Ctesiphon, is attributed by some to the reign of Shapur I
(A .D. 24I-72) on the testimony ofIbn al- Muqaffa' and for stylistic reasons,' while others
• Translated by PHYLLIS ACKERMAN. Z F . SARRE- E . H ERZFELD, Iranische Fe1sreliefs,
I The ruins of Fln1zabiid and Sarvis tan were first Berlin, 1910, p. 128. There is an error in TabarI's
thoroughly studied and measured by C. FLANDIN and account (see 'T. NfiLDEKE, Geschichte der Perser und
P . COSTE, who published the results in : Voya~e en Araber zur Zeit der Sasaniden, aus der arabischen
Perse pendant les annees 1840 et 184 t, Perse anClenne, Chronik des Tabari, Leyden, 1879, p. 1 I), for he
Paris, I8SI-4J Text pp. 23- 7, 36- 45; Illustrations, I, speaks of a palace that Ardashlr had built in Gur which
PIs. 28 , 29, 34-42. FLANDIN (ibid., p.p. 27, 40) was call ed a!-!irbai, and a fire temple. The /irbtzi was
realized that the buildings were to be attributed to the really the fire temple (see p. 553).
Sasanian period. M. DIEULAFOY, L'art antique de la 3 Herzfeld, in SARRE-HERZF ELD, op. cit" pp. 234-5,
Perse, Paris, 188 I) IV, pp. 59- 76, tri ed to show that advances the opinion that the Taq-i-Girra shou ld be
the domed buildings of Fars, Flruzabad, and Farrash- considered pre-Sasanian, but recently he has expressed
band (which he called Ferachabad) date from the Achae- the view that it is late Sasanian : Archaeoiogische Mit-
menid period. His thesis was contradicted by Perrot: teilungen aus Iran, II (1930), p. 80.
G. PERROT-C. CHIPJEZ, Histoire de l'art dans l'Anti- 4 The opinion that the Taq-i-Kisra was built by
quite, v, Paris, 1890, pp. 561-78. R. P. SPIERS, Shapur I was first advanced by Herzfeld: SARRE-
Architecture east and west, London, 1905, p. 73, also HERZFELD, Archll.ologische Reise im Euphrat- und
opposed this view, and the Sasanian attribution has Tigrisgebiet, H, Berlin, t 920, p. 76. M. STRECK,
never since been questioned. Se1eucla und Ktesiphop, in Der alte Orient, XVI (t 9 I 7),

493
 OSCAR REUTHER
accept as historically correct the tradition embodied in its name, identifying it as the
ivan built by Khu sraw I Aniishirviin (A.D. 531-79).' The dating of the palace excavated
at Damghan is similarly contested. The excavators class it as late Parthian, although
the rich stucco ornament is stylistically very similar to that found at Ctesiphon and
Kish and called by some late Sasiinian (but see pp. 642-5 and pp. 601-42).' How
difficult it is, under these circumstances, to discuss Sasiinian architecture can best be
realized by imagining how the history of Greek architecture would be written if the
dating of the Parthenon varied by two, three, or even four hundxed years.
All this uncertainty is attributable chiefly to the fact that the buildings classed as
Sasiinian are in many respects extraordinarily different from each other. Moreover, some
are unique, so that they can scarcely be dated by analogy. Again, though considerably
more Sasiinian than Parthian monuments are known by name and location, we have
clear information on the essential construction of relatively few. Many are shapeless
heaps of ruins which could be discussed only after excavation, yet where excavation has
been undertaken, the reconstruction has not been determined as fully or conclusively
as that of the Parthian buildings excavated at Ashur, while the architectonic and decora-
tive completion of the standing Sasanian monuments is not by any means as sure as
that of the ruins of Hatra. The representations of buildings on two Sasiinian metal
salvers give valuable supplementary information, and the architectural elements cut
in the rock at Taq-i-Biistan provide evidence as to details, but both demonstrate the
variety in the essential architectural elements of this period.
This wide range is doubtless due not merely to differences of date, but also to divergent
local traditions, and if anyone group of Sasanian buildings cannot be readily related to
Parthian buildings, this is probably because the Partllian architecture of the region in
question is unknown.
We are most fully informed concerning the SOUtll and the southwest.' In Fars,
the ancient birtllplace of tl,e Persian state, lie Igakhr, the ruins of which have not
yet been sufficiently explored, Firiizabad, Qal'a-i -dukhtar,' Girra,' Shiipiir (PI. 260 D),'

p. 37; E . DIEZ, Die Kunst der islamischen Volker, Tepe I:litllir must, because of its stylistic similarity to
Berlin-Neubabelsberg, 19 I 5, p. xiii, and others fol- that found at Kish and Ctesiphon, be comparatively late.
lowed this view. 3 HERZFELD, Khur1f.san, Der Islam, XI (192 I),
I F. WACHTSMUTH, D er Raum, Marburg, 1920, pp. 147-52, gives a survey of the known Sasanian
pp. '50-5· monuments. HERZFELD, Reisebericht, Z eilSchrjft der
2 A short report of the 193 I excavations of the Joint Deumhen Mo~~enl4ndischen Gesellschaft, N .F. v (1926),
Expedition of the University Museum and the Pennsyl- pp. 225-84, lists Sasanian buildings found in Fars,
vania Museum of Art under the direction of Dr. Erich Sistan, and Khurlisan.
Schmidt, at Tepe l:Ii ~ar near Damghan, was made by 4 HERZFELD, op. ci t., pp . 252-3; idem, Archaeo-
A. U. POPE, JlluSlraud L ondon News, March 26,1932, logical history of Iran, London, 1935, pp. 95-6;
PP. 4 82-4 . Dr. SC HMI DT published a summary ad- R. BYRON, Note on the Qal'a-i-Dukhtar at Flruzabad,
vance report, Tepe Hissar excavations, 193 I , Museum Bullelin of the American Instilule for Persian Art and
1ournal, XXIII ( 1933), pp. 455-9' Schmidt believes, on Archaeology, 7 (1934), PP' 3-6 .
the basis of the numismatic evidence, that the build- s DIEULAFOY, op . cit ., pp. 77- 8; HERZFELD, Reise-
ing which he excavated must be dated in the third bericht, p. 256; idem, Archaeologi cal history of Iran,
century A.D . E. KtiHNEL, Die Ausgrabungen der pp. 9 -3.
zweiten Ktesiphon-Expedition 193J /2 , Berlin, 1933, '
6 FI.ANDIN-COSTE, op . cit., Text pp. 46-50,
pp. 22-5, concluded that the stucco ornament found at Illustrations I, Pis. 45-7.

494
 SASANIAN ARCH ITECTU RE. A. HISTOR Y
and Sarvistan, to mention only those remains that are most important architecturally.
In Khuzistan, the old Susiana, stand the Ivan -i-Karkha,' and the Dizful and
Shushtar bridges (PI. 26IBA).' In ' Iraq-i-'arab, in addition to the ruins of Dastajird'
and Ctesiphon,' which have been known for many centuries, Sasanian buildings have
been excavated within the area of the old Babylonian city of Kish (see Chapter 28 c)
and the Lakhmid capital I:Iira.' Qa!r-i-Shirin, Hawsh-Kuri,' the Taq-i-Girra,' and
I H. C. RAWLINSON, March from ZoMb to tectural description with careful drawings is given by
Khuzistan, Journal of the Royal Geographical Society, IX Herzfeld in SARRE-H ERZFELD, Archaologische Reise,
(1839), p. 71, first described the Ivan-i-Karkha. A de- 11, pp. 46-76, 1II, PIs. XXXVlII-XLlV, and IV, PIs. CXXIII-
tailed description, measured drawings, and photographs CXXVII. Herzfeld's interpretation has been verified and
were published by DIEULAFOY, op. cit., pp. 79- 98, completed at important points by the recent excava-
Figs. 55- 62, P is. VII- IX. See also SPIERS, op. cit., tions: see REUTHER, Die Ausgrabungen der deutschen
pp. 82 - 3; and SARRE- HERZFELD, lranische Fels- Ktesiphon-Expedition im Winter 1928/9, Berlin,
reliefs, pp. 130-1. 1930; idem, The German excavations at Ctesiphon,
% H . LAYARD, Description of the province of Antiquity, III (1929), pp. 434-51; E. KUHNEL, Die
Khuzistan, Journal of Ihe Royal Geographical Society, XVI Aus~rabungen der zweiten Ktesiphon-Expedition,
(1846), pp. 27-8, 3 I ; DlEULAFOY, op. cit., pp. 105- 12. Berlm, 1933. The results have not yet been defini-
3 HON. C. KEPPEL, Travels in Mesopotamia, I, tively published. .
London, 1827, p. 274; C. J. RICH, Narrative of a S The excavations at Kish under the direction of the
residence in Koordistan, II, London, 1836, pp. 251-6; late L. C. Watelin, including the additional excavation
SARRE-HERZFELD, Iranische Felsreliefs, p. 237; idem, undertaken for the American In stitute for Persian Art
Archaologische Reise im Euphrat- und Tigrisgebiet, and Archaeology, were first reported by S. LANGDON,
II, pp. 89-93, IV, PI. CXXVlIl. New light on early Persian art, The Illuurated London
.. The first modern European to report on the ruins News, February 14, 1931, p. 261; idem, Persian art
of Ctesiphon was PIETRO DELLA VALLE, Viaggi, Rome, discoveries at Kish, ibid., March 7, 1931, p. 369; idem,
1662, I, pp. 4 I 0-11; and many European travellers of A Christian nave at Kish? The discovery of a second
the eighteenth and nineteenth centuries described the Neo-Persian palace, ibid., April 25, 1931, p. 697;
ruins of the Sasanian capital. The most important idem, Palace Three, the bath of the Sasanian kings at
accounts are listed by SARRE-HERZFELD, Archaolo- Kish, ibid., February 20, 1932, p. 273; and The Tim es
gische Reise im Euphrat- und Tigrisgebiet, II, p. 49, n.T, (London), February 2, March 7, 1932. Owing to the
as follows; E. IVES, Voyage to India and Persia, London, death of M. Watelin , the final report on the American
1775, tt, pp. 283-93; KARSTEN NIEBUHR, Reisebe- Institute expedition was not rendered, but the Institute
schreibung nach Arabien, Copenhagen, 1778, II, pp. is now endeavouring to obtain the necessary data and
305-6; G.A. OUVIER, Voyage dans l'Empire Ottoman, hopes to publish it in the near future in its Bulletin. An
Paris, 1804, II, pp. 433-6; SIR J. M ACDONALD KINNEIR, account of the l:IIra excavations is given by D. T. RICE,
Geographical memoir on the Persian Empire, London, The Oxford expedition to Hira, Antiquity, VI ( 193 2),
1813, p. 253; RICH, Narrative of a residence in Koord- pp. 276- 91 i idem, Hira, Journal of the Royal Central
istan, H, pp. 395, 404; J. S. BUCKINGHAM, Travels in Asian Society, XIX (1932), pp. 254- 68 ; idem, Excava-
Mesopotamia, London, 1827, II, Chapter XIII; KEPPEL, tin~ the city of Hira, Discovery, XlII (1932), pp. 161-
Narrative of a journey from India to England, I, pp. 3; Idem, The Oxford excavations at l:IIra, Ars Is/amico,
124-36; Capt. R. MIGNAN, Travels in Chaldaea, Lon- 1(1934), pp. 51- 73.
don, 1829, Chapter IV, pp. 68-86; J. BAILLIE FRASER, 6 The first European traveller to mention the ruins of
Travels in Koordistan and Mesopotamia, London, 1840, Qa~r-i-Shlrln was PIETRO DELLA VALLE, Viaggi, de-
Il, pp. 1-8; idem, Mesopotamia and Assyria, Edin- scritti da lui medesimo in lettere familiari all' erudito
burgh, 1842, p. J 55. Lieut.-Col. CHESNEY, The expedi- arnica Mario Schipano, Venice, 1659, Parte seconda,
tion for the survey of the rivers Euphrates an~ Tigris, pp. 4- 5. RICH, op. cit., II, JP. 264- 7 and 269- 72,
London, 1850, I, p. 35; FLANDIN-COSTE, op. Cit., Text, gives striking descriptions of f,Ja~r-i - ShIrfn and Hawsh-
pp. 174-7, Illustrations, IV, Pis. 216-18; LWARD, KurI, with a sketch of the Chahar Qapu. Plans, sec-
Discoveries in the ruins of Niniveh and Babylon, Lon- tions, and some photographs were first published by
don, 1853, p. 570 ; Frh. M. v. THIELMANN, Streifzuge J. DE MORGAN, Mission scientifique en Perse, IV (1896),
im Kaukasus, etc., Leipzig, 1875, p. 380; DIEULAFOY, pp. 341 - 60, PIs. XL- LIV. Photographs of the Chahar
L'art antique de la Perse, v, pp. 61-78, Pis. lIt-VI; Frh. Qapu were published in SARR~-H ERZFELD, Iranische
M. v. OPPENHEIM, Vom Mittelmeer zum Persischen Felsreliefs, pp. 236- 8, PI. II. Qa~r-i- Shlrln was a~ain
Golf, Berlin, 1900,11, p. 284; SARRE-HERZFELD, Iran- studied by BELL, Palace and mosque at UkhalQir,
ische Felsreliefs, pp. 129-30; G. L. BELL, Amurath to pp. 44-50, Pis. 51-73, with a decidedly more accurate
Amurath, London, 191 I, pp. 179-83; idem, Palace and plan than that given by de Morgan, numerous photo-
mosque at Ukhai4ir, London, 1914, pp. 129-32, graphs, and a very complete description .
134-8. A detailed historical, topographical, and archi- 7 RAWLINSON first described the Taq-i-Girra 111

495
 OSCAR REUTHER
Taq-i-Bustan' lie on the old road which leads from 'Iraq, through the Zagros passes,
past Kirmanshah to H amadan. Towards Sulaymani is the monument of p aikuli.' In
the interior of Persia Sasanian architectural remains are rare. In I~fahan only some
capitals have been found,' and in Na\allz are ruins of a small building.' Architectural
remains have been found in Rayy and near Varamin, and parts of a Sasanian palace
have Iieen excavated near Damghan (see Chapter 28 B). At Yazd, which is today a
Zoroastrian centre, no Sasanian traces have been found. In east Iran near Raba\-i-Safid
in Khurasan between Turbat-i-I;Iaydari and Mashhad, and also on the site of the old
city of Shahristan' ruins presumably of Sa san ian fire temples have been discovered, and the
structures at Kuh-i-Khwaja suggest that, did we but know them, the Sasanian monuments
of this region would probably be interesting.' But Khurasan and Sistan are almost, as
far as Sasanian architecture is concerned, terra incognita, which is especially unfortunate,
for this is the district where certain important connecting links should be sought. The
transmission of the Sasanian architectural style into Central Asia must have been effected
here. Moreover, the Buddhist ruins in Turfan, since they are in part definitely Sasanian
in character (see Chapter 48 B) , show that there was a connexion between Iran and Eastern
Asia, and if any elements entered, it is here that they must have penetrated.
The earliest Sasanian monument, the palace of Firuziibad (PIs. 146-7), originally called
Gur, was really built in the Parthian period, for the district was still nominally controlled
by Artabanus V. Yet while it is, like the Parthian palaces in Mesopotamia, a monumental
ivan building, and like the Hatra buildings is vaulted throughout, it differs markedly
Journal of tht Royal Geographical Society, IX, p. 34. The DIEULAFOY, op . cit., V, pp. 95-108, Figs. 69- 77 ;
first dependable drawings of the monument were pub- SARRE-H ERZFELO, Iranische Felsreiiefs, pp. 199-2 t 2,
lished by FLANDIN- C OSTE, op. cit., Text, pp. 72- 3, PIs. XXXVI- XXXIX. A complete series of photographs
IllustratIOns, IV, PIs . 2 14-1 5- See also DE. MORGAN, of Taq-i-Bustan, with many details, is included in the
op. cit., pp. 335-9, Fig. 204, PI. XXXIX. Herzfeld in files of the Architectural Survey of the American Insti-
SARRE-HERZFELD, op. cit., pp. 233-5, gives an tute for Persian Art and A rchaeology, and M. U.
accurate description, drawings of details which Monneret de Villard is systematically re-studying the
differ considerably from Coste's, and a photograph monument for the Institute .
(PI. XLVU). :t The Pilikull monument was first investigated by
I The most thorough study of Taq-i-Bustan, with Rawlinson: see E . THOMAS, Early Sasanian inscrip-
numerous drawings and photographs, is given by tions, London, 1868, pp. 56- 60. Accurate drawings
HERZFELD, Am T or von Asien, Berlin, 19 20, pp . 57- and a convincing reconstruction were first published by
ISS, P Is. XXVJl-LXV. The publication also includes an HERZFELD, P aikuli, Monument and inscription of early
extensive bibliography, in which the following are the history of the Sasanian Empire, Berhn, 1924, I,
most important titles: J.-B. TAVERNIER, Les six voyages pp. 1-10, Figs. 1-2 and 6-9, II, Pis. 1-9. Additional
en Tur'luie, en P erse et aux Indes, Paris, 1671-81, details are given by HERZFELD in Zeifuhriftder Deut.schen
p. 3 I 6 i J. OTTER, Vorage en Turquie et en Perse, avec Morgenltindischen Gesellscha/t, N.r. V, p. 227.
une relation des expeditions de Thamas Kouli Kan, Pari s, 3 FLANDIN-COSTE, op . cit., Text, p. 22, Ill ustrations,
1748, pp. 184- 8 ; A.J . SI LVESTRE DE SACY, M emoires Pis. 27, 27bis; HERZFELD, Am Tor von Asien, p. I It.
su rdiverses antiquites de laPerse,Paris, 1793, pp . 211- -4 A. GODARD, Le monument Sasanide, Ailliir-e lriin,
70; MACDONALD KINNEIR, Geographical memoir of the , (1936), pp. 79-82 .
Persian, Empire, London, 1813, pp . 132-6; R. KIR S HERZFELD in Zeit.schrift der Deufuhen Morgen-
PORTER, Travels in Georgia, Persia, Armenia, Ancient ltindischen Gesellscha/t, N.r. V, pp. 269, 275 .
Babylonia, etc., II, London, 1822, pp . I7 1-8; KEPPEL, 6 On the basis of his excavations made in 1929,
Narrative of a journey from India to England, H, HERZFELD concludes that the only Sasanian building
pp. 35- 47; RAWLINSON, in 'Journal of the Royal Geo- at Kuh-i-Khwaja represents work of a second building
graphical Society, IX, p. 1 16; FLANDIN- COSTE, op. cit., period at the Qal'a-i-Rustam, and even this dates on ly
Texte, pp . 1-6, Illustration s, PI s. 2- 12; DE MOReAN, from the third century : Archaeological history of Iran,
op . cit ., pp. 307-35, Figs. 180-2 01, PIs. XXXIV-XXXVIII; p. 66 . See p. 418, n. I. .
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
from any known Parthian architecture. The novel features must be accounted typically
Sasanian, for Fars, where this architectural style developed and whence it spread later
to the rest of the kingdom, was the fatherland of the new dynasty which, in reaction
from Arsacid Hellenism, sought a nationalist revival.
One architectural detail which appears in the palace of Firuzabad, and likewise in the
mountain castle of Qal'a-i-dukhtar near Firuzabiid, which was probably also built by
Ardashir, was clearly a deliberate revival from the Achaemenid style. This is a concave
cornice decorated with three rows of overlapping upright petals forming an imbricate, a
design characteristic of the palaces of Persepolis
(Fig. 127 and PI. 146 B; cf. PI. 86 B). It is because
of tllis moulding, on tile lintel of the door con-
necting the inner rooms and also above the niches
which break up tile walls, that the palace of Firuzabad
was at one time tllOught to be the work of a Satrap
of Xerxes,' for it was considered improbable that
a Sasanian architect would have been intentionally
archaistic and used an architectural element from a
building then six centuries old. But actually this
exact repetition of an Achaemenid form points to
Ardashir as the builder of the palace, for thereby he -'::-0-, -!4-~
doubtless thought to mark witll a seal oflegitimacy '~ . ''' ....
the palace that was to house his hard-won throne. FIG. 12 7. Flruz.b.d,Palace of Ardashlr,
door frame.
The ruins of Persepolis, where the lintels of the
doors and niches are conspicuous, must certainly have seemed to him and his con-
temporaries, as it did to later descendants, a monument to national greatness. Ardashir
and his architect could not seize on tile columns of Persepolis, more of which must
certainly have been standing then, as an element to copy, because columns had no
structural architectonic place in the kind of building to which they were accu stomed,
Indeed, the palace of Firuzabad is typically a columnless building, The old Persian
cavetto cornice, to which a symbolic significance was probably attached, could be taken
over without any adjustment. This cornice is the only recognizable Achaemenid element
in Firuzabad, but in Shapur, the city founded by Shapur I southwest of Persepolis
on the plains of Kazarun,' the small tower-like masonry structure has bull -head pro-
tomes taken from the Persepolis capitals, and the moulding on the edge of the round
fountain basin is overlaid with a petal imbricate like that on the Firuzabiid door .lintels.
But while there was this conscious return to a few Achaemenid details, the Hellenistic
inheritance was not allowed to lapse. For the very cyma moulding' on which at
Shapur the leaf imbricate is applied represents a form that does not appear in Achaemenid

'D IEULAFOY, op. CI. t., p. 59. outline of which consists of a concave and a convex
:: FLANDIN- COSTE, op. cit., I, PI. 46. line.-Ed.
3 A eyma moulding is a moulding of the cornice, the

3s 497
 OSCAR REUTHER
architecture ; and in Firiizabad the door and niche frames on which the lintels rest are
of a type that is wholly unknown in early Persian architectural design, but is clearly
Hellenistic in origin. They are formed of rectangular mouldings, and are defined by
pilasters carrying a semicircular archivolt, also with mouldings.
The palace of Firiizabad may seem archaistic in other respects when compared to
Parthian buildings, especially those at Hatra, but this cannot be considered the result of
a deliberate revival of early Persian architectural form s. More probably the accession to
power of the aristocracy of Fars provided an opportunity of using for the first time on a
monumental scale local building methods which represented a traditional style rooted in
the soil and in the life of the people.
STRUCTURAL FORMS
The peculiarities of the style are evident not only in the plan and form of the rooms
of the palace at Firiizabad, but especially in the construction. The walls and vaults
are made of unfinished broken stone or of coarse round cobbles set in gypsum mortar,
which takes the place of the clay that was employed as mortar in the Ancient East,
for example, in Hittite buildings, and is still commonly used in the Orient for native
buildings constructed of field stone (Pis. 260 A, 281, and 335). The stones are not really
built up like masonry, but are simply piled on top of each other, without any bond, the
walls being modelled, as it were, in the rapidly setting mortar, and the stones packed in
only as a filling. Were it not for the mortar the walls wopld immediately tumble into
a shapeless heap, and this is exactly what has happened where rain water has seeped in,
broken down tile mortar, and thus loosened tile stones. The same technique is found
also in later Sasanian buildings in Fars, and the method was obviously indigenou s and
spread from there to the frontier districts between ' Iraq-i- 'arab and Ardilan, where similar
conditions prevail. Thus it was used in the Alvand Valley in all the buildings built by
the later Siisanian rulers, especially Khusraw II, of which Qa~r-i-Shirin is the centre.
On ti,e plains of Khiizistan and 'Iraq-i- 'arab tile ancient local brick construction was,
as might be expected, continued in the Sasanian period. Walls were constructed of
unfired brick wlt11 clay mortar, laid in the traditional Babylonian manner, with reeds
or rushes between the courses.' Fired bricks were laid, at least for the foundation s, '
in tile peculiar upright courses found in tile Parthian constructions at Ashur (see
p. 422, Fig. 99 a, tI) , and were set with gypsum. The foundation s of the Taq-i-Kisra,
the royal palace in Ctesiphon, were built in this fashion, as were the round piers of the
older of the two churches excavated there, and the same lay occurs in the piers of the
Sasanian palace in Damghan. Moreover, thi s technique was continued in Damghan in
tile early Islamic period, for the columns of tile Tiirik-khana (see Chapter 39 C and
PI. 258), a mosque built in the eighth century, are also constructed with upright and
horizontal courses alternating,
I R. KOLDEWEY, Die Tempel von Babylon und R EUT HER, Di e Innenstadt von Babylon, Leipzig,
Borsippa, Leipzig, 191 I, p. 5; idem, D as wieder- 1926, pp. 8'2 , 129- 30, Figs . 8S, 86.
erstehende Babylon, Leipzig , 1925, pp. 10, 62;
 SASANIAN ARCHITECTU RE. A. HISTOR Y
The Sasanian architects, like the sculptors, when they sought permanence used
stone. Thus the But-khana at PaikUli, which was erected by Narse (293- 303) as a
victory monument, was faced with masonry. In Shapur are the remains of masonry
buildings which probably date from the reign of Shapur I. The Taq-i-Girra is built
of large, carefully fini shed, masonry blocks, and Taq-i-Bustan was partly constructed in
masonry, partly cut, with consistent architectural elements, directly in the living rock.
Thus masonry construction was used from the beginning to the end of the Sasanian
period, but it was specifically limited to commemorative monuments or to buildings
which had to be especially solid, for example, the piers of ti,e bridge over the Karun
River near Dizful (PI. 261 B), and other bridges
(see Chapter 39 M).
All the known Sasanian buildings are vaulted,
whetller they are built in the rubble technique
indigenou s to Fars, which ought perhaps to be
considered typically Sasanian, or in brick, as
in Khuzistan, 'Iraq-i- 'arab, or Sistan; and the
form of the vaults remains the same, regardless
of the difference of material and the consequent
essential difference in vaulting nlethod, save FIG. I28. Construction of an arch with centering
set on the supporting members.
that vaults of a more developed type were
impossible, even later, in the coarse rubble. That the Sasanian architects could also,
when it seemed necessary, construct a masonry vault is shown by the barrel vault of
the Taq-i-Girra.
Two kinds of barrel vaults are found at Firuzabad. One was built with centering. A
relatively tllin layer of stone and gypsum mortar was first laid on ti,e scaffolding, and
then when the mortar was set, tI,is sheet served as centring for the bulk of the stone and
mortar which constituted the actual vault. This method required only a rather light
centering which was erected on the sustaining walls after they had been built up as far as
ti,e imposts. This type of barrel vault or arch has, as a result, a characteristic form con-
sidered typically Sasanian, the diameter being greater than the breadth of the room or ti,e
door, as ti,e case may be, so that ti,e vault or arch above the imposts sets back a certain
distance in relation to ti,e walls or the jambs (Fig. 128). The second type is tim already
found in Ashur in the Parthian period, built without any centring. When large spaces
had to be vaulted, for example ivans, ti,e builders, in order to reduce the span as much as
possible, gradually shifted inwards the horizontal courses of bricks above the imposts, so
that the walls approached each otl,er in a slight curve, and the opening above was roofed
with vertical courses describing a large segment of a circle. Thus the half-elliptical or
parabolic section of vaults or arches, which is so typical of Sasanian architecture, did
not result from considerations of stability, but from the actual process of construc-
tion (Fig. 129), and doubtless the method was an ancient tradition. It is noteworthy
that the Egyptian New Kingdom parabolic arches resulted from precisely the same
499
 OSCAR REUTHER
process.' The gigantic parabolic barrel vault over the ivan of the Taq-i-Kisra in Ctesi-
phon is the largest example of this construction without centring (PI. 151). Here the
span is 8+ feet (25'60 metres), while the span of the ivan vault at Firlizabad, built
with flat stones in place of bricks, is, despite this crude material, +3! feet (13'30
metres). Occasionally the lowest courses of a barrel vault or arch built in this way
are set forward a little in relation to the supporting walls or columns, but this is not
the case in the big Taq-i-Kisra vault. There the surface of the vault is in the same

FIG. 129. Diagram of a Sasanian barrel vault with parabolic cross-section,

built without centering, with vertical semicircular courses.

vertical plane with the walls, divided from them by a narrow brick cornice. The vaults
of the side rooms, on the other hand, are set forward from the walls, but irregularly,
and at Sarvistan and Qa§r-i-Shirin this system of advancing the vault or arch in relation
to the supporting construction seems to have been the rule. Undoubtedly it developed
from the Sasanian masons' tendency to build up a fal se vault with overhanging lays,
and later, in Islamic building, it became a stylistic postulate (cf. e.g. Pis. 277 B, 306).
In contrast to the parabolic outline of uncentred vaults, the vaults built with centring
have a semicircular cross-section, because the centring could most easily be con-
structed in a semicircular form. Thus both at Firuzabad and in later Sasanian buildings
parabolic and semicircular barrel vaults are found side by side. Considerable emphasis
was clearly laid on rounding evenly the crown of the arch or vault. At Firlizabad some
of the vaults and door arches are built Witll fal se vaulting, made by stepping forward
I Parabolic barrel vaults are found in the granaries between the Egyptian and Sasanian vaulting method has
ofth~ Ramesseum: RLEPs,us,DenkmalerausAgypten been noted by DrEULAFOY, op. cit., p. 2 I; and SPIERS,
und Athiopien, I, Berlin, 1849, PI. 89. The similarity op. cit.} p. 62.

500
 SASANIAN ARCHITECTU RE. A. HISTOR Y
successive courses, up to the crown. This gives a pointed-arch cross -section, but the
mason sought to modify this and round the crown by laying each course, as he got nearer
the top, with an increasing overhang on the inner side, and in the end the rounded effect
was achieved by compensating with plaster. In Qa~r- i -Shirin there are pointed vaults,'
but these are only on subordinate rooms, such as corridors in the lower story.
All the ivans and the oblong rooms of Firiizabad are barrel vaulted, while the three
square rooms that run across the building behind the front ivan are domed. These are
not, however, the earliest domes known in the Orient. The domed tombs of Ur show
that the dome principle already existed, at least in germ, in the third millennium B.C.
in Mesopotamia,' and domed buildings represented on an Assyrian relief from
Kuyunjik have repeatedly been
cited. 3 But the particular form of
dome used at Firiizabad has no
known antecedents to the west of
Iran. It is quite misleading to
connect this type with Roman-
Hellenistic dome construction. The
Firiizahad type is based, rather, on
what is doubtless a very ancient
solution of the problem of roofing
a square space without using wood, FIG. 130. Diagram of method of constructing an
Iranian vault on squinches.
eitl,er permanently or temporarily,
when no large monoliths are available. In constructing a brick barrel vault without
staging, the mason starts at the back wall with a course describing a segment of a
circle and advances with a succession of parallel courses. In this otl,er type of
vaulting he begins by laying a small arched course diagonally in one corner and
advancing with increasingly larger courses until he has the segment of a cone, which
he repeats in each of the other corners, these half-cones meeting at tl,e centre of each
side of the room. If the work is continued in tl,is way, the remaining opening, which
is square in horizontal projection, is closed, and there results the form of vault used to
this day on houses in all the towns and villages of Khurasan (Fig. 130)' This squinch
vault may be considered the antecedent of tl,e squinch dome as it appears at Firiizabad:
I BELL, op. cit., p. S0, PI. 51, Fig. I. single room, some with high domes approximately
:z C. L. WOOLLEY, The Sumerians, Oxford, 192 8, parabolic in contour, which can be interpreted as cor-
pp. 37, 191. belled, and others with domes in the shape of half-
• • • • • • • globes, which must have been built in radial courses
Domes were used on circular structures in Elam at and thus represent a true vaulting technique. These
least as early as the beginning of the third millennium buildings are represented as the dwellings, not of
B.C., judging from a representation on a stamp seal (see Assyrians, but of foreigners. The modern descendants
p. 290, and Fig. 70 d).-Ed. of this type of dome building are the qubiib still
3 LAYARD, A second series of the monuments of common in Syria and the Saruj: see E. BANSE, Die
Nineveh, London, 18S I,PI. 17. The relief, found by Gubab-Hutten Nordsyriens und Nordmesopotamiens,
Layard in the Palace of Sennacherib in Kuyunjik and Orienfa/ische.s Archiv, II (191 I), pp. 173-9.
dating from the beginning of the seventh century B.C., .. DIEz, Die Kunst der islamischen VOlker, p. 78,
shows clearly a group of buildings, each consisting of a defined the principle correctly and introduced the term

501
 OSCAR REUTHER
The funnels of the squinches are not carried, in this version, to the middle of each
side, but a space is left between them; these spaces are filled with level courses on a
graduall y increasing horizontal curve until a horizontal circle is made at the crown
of the squinch arches, and this supports the dome. The dome, which has a half-elliptical
cross -section, is built up with horizontal circles and it can be completed without any
scaffolding (Fig. 131).
The difference between this Eastern squinch dome and the Roman-Hellenistic dome,
such as appears in large scale on the
Pantheon, must be kept clearly in
mind. In the Roman -Hellenistic
West, dome construction developed
from the problem of roofing with
stone, in the simplest possible man-
ner, the round rooms of the thermre,
as the Stabian Baths in P ompeii
show.' A roof more or less like
a dome was made by building up
successive circular courses of stone,
each projecting inwards beyond the
one below until the circle was
closed. If the round unit was com-
bined with rectangular rooms, as
FIG. 13 I. Diagram of an Iranian dome on squinches.
occurs in the same thermre, the
corners of the square were rounded out with niches. This device is found in most
of the later round and octagonal thermre, out of which the Christian baptisteries
developed.' The square with niched corners or the octagon could also be used as
a transition zone between a rectangular room and a dome, thereby solving the problem
of setting a dome on a square (Fig. 132 a, b), but this kind of dome on diagonal
niches had originally nothing whatever to do with the dome on squinches. The
latter is without doubt purely Oriental in origin, and was most probably first used
in some specific district in Persia, where it was developed out of the simple squinch
'squin ch vault' (Trompengtwolbe) . If Herzfeld's dating Stabian a, p. 8, Pis. I , II; A. M ICHAELIS, Die neuen
of the Kuh-i-Khwaja buildings is correct, the oldest Bltder in Pompeji, Archtlologische Zeitung, XVII (1 859),
known standing examples of such vaults, dating from pp . 4'2-3, PI. CXXIV. Professor H. Sulze, Dresden,
the first century A.D . , are to be found there: HERZFELD, who has restudied the Stabian baths and is preparing
Archaeological history of Iran, p. 66 . DIEULAFOY'S a publication on them, has kindly informed the writer
attempted explanation of the o ri ~ in of the squinch- that the frigidarium was originally the laconicum, and
dome (op . cit., pp. 5, 8, 10) is enttrely erroneous. certainly when serving this purpose had no niches.
I The frigidarium of the Stabian Baths in Pompeii These were first set into the walls when the room was
is the oldest known example of this kind of Roman adapted to its new use as a cold bath.
rotunda with four niches: see G. MINERVINI, Bulltllino 1 L. VON SYBEL, Christl iche Antike, II, M arburg,
Archaeologico Napolilano, N.S. IV, 3 (77) ( 185 5), p. 20; 1909, pp, 309-10; O. WULFF, Altchri stliche und
ibid., V ( 1857), supplementary plate, VI, I (12 5) ( 1857), byzantimsche Kunst, Berlin-Neubabelsberg, 19 t 3,
p. I, PI. I, Fig. t; NJ CCOLlNI, Le case ed i monumenti pp. 244-5; W AC HTSMUT H, D er Raum, II, Marburg,
di Pompei, I, Naples, 1854, Therme presso la porta 193 5, p. 12 3.

502
 SASANIAN ARCH IT ECTU R E. A . HISTO R Y
vault. Since the latter appears in the Kuh-i-Khwaja ruins of the first century A.D., the
dome on squinches may very well have originated in east Iran. It is owing to this
basic difference in principle between Roman domes and Sasiinian, that R oman Imperial
architecture, in so far as it was vaulted, employed primarily round rooms or rooms with
concave curved walls in palaces and thermre, and to extend the limits of the rooms semi-
circular apses were used, either singly or in the form of the triconchos.' This spatial
form was, on the other hand, entirely alien to Sasanian architecture, for in Siisanian plans
only rectangular or square rooms appear, and extensions are effected with niches
which likewise are rectangular,
when, as at Sarvistiin, they
are covered with half-domes.
Where a semicircular apse does
occur, as in the sole and un-
precedented instance in Kish
(see p. 538), it must be attri-
buted to influence ultimately
a
emanating from R ome.
Until the advent of the
Byzantine style the true squinch
dome was confined to Persia
and the adjacent countries,
passing thence into Central
Asia) together with the barrel FIG . J 32 0, D. Plan and diagram of a Roman domed rotunda with
vault constructed of circular niches on the diagonal axes masked by exterior right angles .
courses. The Buddhist and Manichaean religious buildings built of unfired brick in the
ruined city of Khocho on the northern edge of the T arim Basin have only squinch
domes and uncentred semicircular vaults (see Chapter 48 B) .
The clumsy type of squinch dome used at Firuzabad, where the dome rests its weight
on all the walls, developed into a more elegant form. Thus on the little building
with a single dome at Farrashband, in the Girra Valley, which perhaps stands closest in
date to the palace of Firuzabad, tl,e dome is carried on four strong corner piers which are
connected by arches. Each arch is closed with a niche so tl,at the room is widened out
on the four axes. The great square main room of the palace of Sarvistan is built in the same
way, but the dome, which is ovoid and set on a squinch zone defined below and above by
dentated cornices, is built, not of rubble, but of fired brick. The architect of Sarvistan
clearly attempted still another step forward in the small room in the southeast corner

I Attempts to roof a square space with a dome-like architects used for large buildings a round plan ex-
vault appear very early in Egypt and also in Etruria. tended with niches. On the origin of the pendentive
A satisfactory solution of the problem with the use of see K. A. C. CRESWELL, Early Muslim architecture,
spherical pendentives is found only relatively late in Umayyads, Early 'Abbasids, and Tu!unids, I, Oxford,
Roman Syria and Asia Mi nor, and is at fi rst limited to 1932, pp. 3°4-23.
small buildings. Even in late Imperial Rome the
 OSCAR REUTHER
of the palace, where he placed a round pier at each corner, connected by arches, but
set the dome, not on this square of arches, but on walls outside the arches (Fig. 133).
The pillars carry only the four arches, on top of which are horizontal courses of
masonry, making a gallery accessible from the roof terrace.
In comparison with this attempt to enlarge the room and distinguish the walls that
carry the load from those that merely define the
interior, the interesting domed square building
which stands isolated on the terrace of the
building in Qa!r-i-Shirin, called today the
Chahar QapG, represents a definite regression. '
The dome is destroyed, but its form can be
deduced from the rubble squinches which are still
' .. standing (see Figs. 158 and 159). It had a span
of 53 feet (16' 15 metres) , the largest Sasanian stone
dome known, and it rested on all the walls.
The four great doors, with round -arched vaults
built of brick, are only openings in the extra-
ordinarily thick walls and do not play an active
structural part as the arched niches at Farrash-
band do.
The Qa!r-i-Shirin buildings seem on the
whole technically primitive and inept, especially
in the vaulting, though they date from the end
of the Sasanian period.' This is probably due

. to the fact that the architect, to satisfy his patron,

had to work very fast with the available material,
FIG. 133 . Sarvistan, Palace, interior of domed coarse, round, field stones which he could u se
room in the southeast corner.
at all only becau se of the rapidly setting gypsum
mortar. Regular lays and smooth, even surfaces were impossible with such material, and
thick, clumsy walls and vaults were unavoidable. The bricks which are used for the
door arches of the domed square building of the Chahar QapG, and which probably
constituted the shell of the dome, must have been brought from elsewhere, hence as
few as possible were employed.
The vaulting of the palace of Sarvistan, though perhaps a little earlier than the Qa!r-i-
Shirin buildings, is far more evolved, presumably because the architect had readily
workable stone. This structural superiority is even more evident in the two oblong
rooms than in the dome. These were roofed with a combination of barrel vaults, now
I DE MORGAN, op. cit., p. 350; BELL, op. cit., p. 53. p. 164. Yaqilt and MustawfI were also familiar with
Z Ibn Rusta, who wrote about 903 (c . '290 H.), de- this tradition, which has persisted down to the present
scribes the ruins of Qa ~r-i-S h ttln as the remains of the day in the name of the rums, and there is no reason for
palace built by Khusraw II Parvtz : M. ]. DE GOEJE, doubting its accuracy.
(Ed .), IB N RU STA, AI-a'iaq an naftsa, Leyden, 1892,
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
destroyed, and half-domes. In front of the long walls stand pairs of short, cylindrical,
coupled pillars, connected by a horizontal plinth, and joined to the wall with small
arches. Rectangular piers develop from the round pillars, and between these are
thrown half squinch domes (Figs. 134 and 152). Thus there is a series of niches vaulted
with half-domes, and it is the frontal arches of these and the piers between which carry
the vault, instead of simple sus-
taining walls, as at Firiiziibiid.
The additional width provided
by this niche plan was of scarcely
any practical value. A man can
only just pass between the pairs .. ,
of pillars and the wall. It was the
effect of greater width that was
sought, and especially a relief from
the sense that the walls were too
restricting. This is certainly the
problem that confronted the archi-
tect, but he had quite a good
practical knowledge of statics
and understood quite well both
the approximate continuation of
the line of pressure of his vaults
and the possibility of distributing
the thrust and counterthrust by .: . ,.:
..'-< .-',::,'
means of a system of interior .. .. ;. '

buttresses connected by arches.

This is a first tentative step to-
wards the constructional system
of which Gothic architecture is FIG. 134. Sarvistan, Palace, interior of barrel-vaulted room on the
south side, with half-domed niches.
the supreme realization.
Other Sasiinian and early Islamic rooms are vaulted by this same method, for example,
the Siisiinian church which has been excavated at Ctesiphon. Rectangular piers, instead
of pairs of round shafts, are used here, while in the church of Mar Tahmazjird
in Kirkiik, which can be dated in the eighth or ninth century, the paired round shafts
are used again,' but both here and in the approximately contemporary great hall
of the palace of Ukhayqir the passage between the .columns and the wall is missing, the
columns being half engaged.' This, however, is of course a regression. The rooms
I BELL, Churches and monasteries of the Tur pp. 330-6, Fig. 308.
'Abdin and neighbouring districts, Zeilschrift fur z REUTHER, Ocheidir, Berlin, 191 2, pp . 20-2 ,
Geschichte der Archi1ektur, Beiheft IX (1913), pp. 100, Pis. VII, XIV; BELL, Palace and mosque at Ukhaic;lir,
102, Pis. XXVI, XXVII; SARRE-HERZ FELD, Arch:to- pp. 12-14, Pis. 4, If, IS, 16.
logische Reise im Euphrat- und Tigrisgebiet, II,

3T 505
 OSCAR REUTHER
excavated at Damghan and Kish, which open out as ivans, have narrow side aisles
defined by rows of thick, round piers and doubtless they were vaulted in tllis same
manner (see pp. 581-2), and the projecting porches of the 'Imarat-i-Khusraw in Qa§r-i-
Shirin and the Hawsh-Kuri palace ought also to be reconstructed as triple ivans. Rows
of round piers are shown at the sides in the original survey plans,' but beyond these
there must have been solid walls, for the piers would not have sufficed to take up
the thrust of the necessary vault, and a trabeated roof would be contrary to all that is
known about Sasanian architecture. Even a superficial excavation, which could easily
be carried out, would determine this question.
Conditions were more favourable for the development of vaulting and so for varying
the shapes of the rooms where brick was used, as on the alluvial plains of 'Iraq and
Khuzistan. The ruins known today as the Iviin-i-Karkha show a notable advance in
this respect.' The building consisted of a single room which was symmetrically sub-
divided into a central square about 30 feet (9 metres) to a side, outlined by four great
arches, and two wings of this same widtll and each about 65 feet (20 metres) long.
The central square was roofed with a dome on squinches, remains of which were still
to be seen in the nineteenth century. The wings were spanned by five wide arches,
tllrown from wall to wall, and the spaces between these were covered with transverse
barrel vaults. The facing walls, which are, in consequence, only screens, are pierced with
windows (Fig. 135). This plan solved the problem of making high windows in tlle
long walls of a vaulted room without weakening tlle members that carried the vault.
The problem was solved in Rome and in medieval Europe by using successions of
groined vaults, but these were not known to the Siisanian builders. The series of trans-
verse vaults also appears, however, in medieval Europe, notably in the Burgundian
church of Saint Philibert in Tournus,' and the system was used there to solve the same
problem. In Islamic Persia this type of vaulting continued to be very common
(see Chapter 39 H and PI. 446). The preliminary stages of the development were
already apparent in Parthian architecture, where stone beams were laid above a grating
of arches (see p. 425 and Fig. 102), a roofing method which may have been taken over
from the same source by the Parthian architects and the builders of the J:Iawran.
If barrel vaults were substituted for the stone beams, the ivan-i-Karkha vaulting system
would result, and in the Parthian palace at Ashur this step had already been taken, for
a room was divided by arcades and the three aisles thus defined were barrel-vaulted,
giving essentially the Ivan-i-Karkha scheme (see p. 423 and Fig. roo). Because of this
similarity the ivan-i-Karkha has been considered Parthian.' The ground plan, however,
of three units symmetrically arranged, is so far in advance of any Parthian or early
I DE MORGAN, op. cit., Pis. XLII, L1; BELL, op. cit., 3 C. Duuo--G. V. BEZOLD, Die kirchli che Bau-
p. 45, when she investigated the 'Imarat-i-Khusraw in kunst des Abendlandes, Stuttgart, 1892, PI. 137;
Qa~r-i-ShlrIn) found no trace of the piers or columns DIEULAFOY, op. cit., p. 1631 assumes that there was
reported by de Morgan. some cannexion between Persian construction like that
2 DIEU LAFOY, op. cit., pp. 79- 88, Figs. 55-62, of the Ivan-i-Karkha and St. Philibert in Tournus.
Pis. V I-IX; SARRE-HERZFELD, Iran ische Felsreliefs, 4 H. GLOCK-E. DIEz, Die Kunst des Islams,
p. 130; SPIERS, op. cit.} p. 82, Fig. 37. Berlin, 1925, pp. 14, 127.

506
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
Siisanian vaulted structures known that the building can have been built only at a
comparatively late date.
All in all, Sasanian vaulting is extraordinarily clumsy. Rubble construction tended

:; :. . -, .. '

.... . ..' " ,

' . ' '"

.. ".
..
.' .
. '" ..
' " '.

', " ,

FIG. 135. Ivan-i-Karkha: cross-vaulted room.

to make walls and the shells of vaults very thick, and though bricks permit more
economical work, the masons evident! y used them in t!,e same proportions. As the
dimensions of the rooms increased, the builders obviously thought it necessary to make
the vaults spanning them correspondingly stronger. The barrel vault of tile Taq-i-
Kisra is at the crown more than 3 feet (I metre) thick, and above the springing becomes
50 7
 OSCAR REUTHER
nearly 6 feet (1'80 metres) thick. The builder believed that he had to make the walls
that take up the thrust of this unprecedentedly heavy barrel vault 13 feet (4 metres)
thick at the impost and over 23 feet (7 metres) thick below. The gigantic dimensions
obviously discouraged him from breaking up the sustaining walls with niches as the
architect at Sarvistan had done. The simplest form seemed to him the safe st. But this
simplicity of construction should not be taken as evidence of an early dating for the
Taq-i-Kisra. It was due to uncertainty in the face of great dimensions for which there
were no precedents that would provide practical knowledge. Anxiety lest the thrust of
the vault should damage the articulation of the brickwork led the architect to reinforce
the frontal arch of the great ivan with wooden beams at the level of the imposts and the
beginning of the actual vault. He also inserted reinforcing beams into the back wall of
the barrel vault, obviously with the intention of preventing cracks. Such reinforcing
already appears in Babylonian architecture,' and the ancient tradition still persists in
Islamic building today (see Chapter 39 c). The series of rooms on either side of the
great ivan must have had transverse vaults like those in the Ivan-i-Karkha. Here, where
he was relieved of the onus of vast scale, the architect clearly felt that he was freer.
In a vaulted architecture like that of the Sasanian period the arch necessarily played a
conspicuous role. Doors and niches were the principal motifs of exterior and interior
design, and both were arched. When there was a barrel-vaulted ivan its frontal arch, as in
the Parthian period, dominated the whole fas:ade. The contour of the arch of course was
determined by that of the barrel vault, which was developed from the vaulting technique.
The half-elliptical or parabolic arch, which is regarded as typically Sasanian, was
evolved from un centred vaulting. The type is well exemplified in the ivan fas:ades of
the palaces of Firuzabad and Sarvistan, while the parabolic arch of the Taq-i-Kisra
(84 feet (25.60 metres)) was, until it broke down in 1888, the largest arch visible on a
fas:ade in the whole world (PI. 149).
Yet the parabolic arch was apparently not regarded by the Sasanian architects as
aesthetically desirable. They permitted the form to be revealed only when compelled
by the customary method of vaulting without centring, especially as frontal arch of
their barrel-vaulted ivans. All the doors and niches at Firuzabad were, as far as can be
determined, framed in somewhat stilted semicircular arches; this is true also of the
other Sasanian buildings, and the arches of the blind arcades on the fas:ade of the
Taq-i-Kisra are semicircular. Door arches, owing to the habit of erecting the centring
of the vault on the walls, are often set back at the impost, in contrast to the parabolic
arches developed from uncentred vaulting. This form of door arch, slightly wider at
the impost than the space between the jambs, had been used in late Babylonian con-
struction, notably in the doors in the walls, still standing, of the excavated temple of
the Goddess Nana in Warqa, which building dates from the third century H.C.' The

I KOLDEWEY, Das wiedererstehende Babylon, pp. 62, deutschen Ausgrabungen in Warka), Deutsche Allge-
70 ,82, 114. meine Zeilung, No. 12 2, Beiblatt 14 (March 19 34).
2 ANDRAE, Auf den Spuren :tltester Kultur. (Die

508
 SASANIAN ARCHITECTU RE. A. HISTOR Y
motif was continued into early Islamic building, appearing, for example, in the palace
of UkhaY9ir and the Tarik-Khana in Damghan (PI. 258). Though the form did
result from a method of vaulting, it evidently ,
was not considered aesthetically unsatisfactory,
for some of the niches cut in the rock at Taq-
i-Bustan have this shape, and no technical
necessity constrained the designer there.
It has been suggested that the recession of
the arch from the line of the jambs was com-
pensated in the plaster surfacing, and that this
gave rise to the horseshoe form.' But since
the Sasanians did not object to the outline we
cannot assume that the discrepancy was con-
cealed with plaster. However, since this arch,
set back at the imposts, resembles the horseshoe
arch in that both have a diameter greater tllan
the width between the supports, the one may
have opened the way for the other.
The only horseshoe arch known for a long
time in Sasanian architecture was that in the
Taq-i-Girra, which stands isolated near the top
of the Zagros Pass at the western entrance to
Persia (PI. 153 c).' The date of this little
structure, a barrel-vaulted ivan without any side
rooms, is still disputed. The monumental con-
struction out of large- scale, carefully finished
ashlar, set without mortar, the Attic moulding
of the sode, and the mouldings of the impost r-------··
and of the upper cornice, as well as the almost
Classically drawn guilloche which is the only ' - - - - - - - - -- -- - _ .
. d r FIG. 13 6. Taq-i-Girra, details of socle, impost,
ornament, Ilave a11 been cIte as arguments lor archivolt profile, and upper cornice.
an early dating,' and have even been used to
support a Parthian attribution (Fig. 136).' But contrary evidence is provided by three
features tllat occur together in Christian buildings of upper Mesopotamia, but not earlier
I DIEULAFOY, op. cit., p. 37, gives this explanation to assign a date to it-certain features point to a
of the origin of the horseshoe arch. Herzfeld (SARRE- period between the second and fifth centuries, which, of
~ERZFELDJ Iranische Felsreliefs, p. 129) opposes this course, takes us into the Sasanian period, yet no certain
View, example of a Sasanian building of cut stone is known'
2. See p. 493. Recently the question of the date has (p. 139).
been discussed by CRESWELL, Early Muslim Archi- 3 SARRE-HERZFELD, Iranische Felsreliefs, p. 234.
tecture, pp. J 37-9. He concludes that: 'This monu- 4 Fig. 136 has been drawn after Herzfeld in SARRE
ment, which lies on the border between Mesopotamia -H ERZFELD, Iranische Felsreliefs, Figs. I 12-13, and
and Persia, in style belongs to the Christian architec- Flandin in FL\NDIN-COSTE, op. cit., IV, PI. 215.
ture of the former country'; yet says: 'It is difficult
 OSCAR REUTHER
than the fourth century: the mouldings of the arc hi volt ; the horizontal projection of
the arc hi volt over the imposts ; and the horseshoe arch. These three unusual features
appear in the portal of the church of Mar Ya'qiib in Nisibis dated 359,' and this shares
still another peculiarity with the Taq-i-Girra: the large- scale, carefully fini shed masonry
construction, set without mortar. The horseshoe arch does appear earlier in Syria, if
the canopy tomb in Brad in the Jibal Sim 'an is correctly dated in the third or even
second century;' but there is after all more basis for dating the Taq-i-Girra in the
Sasanian than in the Parthian period, though not at the end of the former.'
Moreover, this Taq-i-Girra horseshoe arch is now no longer the sole known example
of the form in Sasanian architecture, for a double door has been found in a dwelling

, ,
FIG. 137. Ctesiphon, al-Ma'aric,l: excavated dwelling FIG. 138. Detail from si lver plate with
house, 6th century, detail of double door with horse- fortress, Hermitage Museum (PI. 233 B,
shoe arches. and Fig. 149), showing arcade of horse-
shoe arches.

excavated at Ctesiphon, which must date from the middle of the sixth century, with
horseshoe arches formed of stucco carried on pilasters and pairs of half-columns, set
forward, witilOut capitals (Fig. 137).' The fortress represented on the Hermitage silver
salver also has horseshoe arches in the arcades (Fig. 138), and the archivolt mouldings,
which are drawn as concentric semicircular arches, are cut back on the diagonal above
the imposts in the same way as the door arches at Ctesiphon. Finally, horseshoe arches
occur in the radial columnar arcade framing the central rounde! on the bronze salver in
the Staatliche Museen, Berlin (Fig. 139 and PI. 237)'
The stilted semicircular arch continued to be the standard form throughout the history
of Sasanian architecture, as the many hundreds of blind arcade niches and door arches
at Firiizabad and Ctesiphon show. The horseshoe was, as far as we can at present judge,
only occasionally substituted as a variant, and it was not introduced until late in the
period. This interpretation of the historical place of the horseshoe arch is also supported
by ti,e far more conspicuous role that it played in early Islamic architecture. At
I SARRE-HERZFELD, Archaologische Reise im Eu- 1904-5 and 1909, Division II, Section B, Leyden,
phrat- und Tigrisgebiet, li, pp. 336-43, Figs. 3 I 6-q; 1920, p. 299, Fig. 329, and PI. xxv.
BELL, in ZeitJChrij' fiir Geschichle der ArchittclUr, IX, 3 See p. 493, n. 3 .
pp. 96-9. Fig. 34. .. KUHNEL, Die Ausgrabungen der zweiten Ktesi-
2 H. C. BUTLER, Syria, Publications of the Prince- phon-Expedition, p. 8 and Fig. 4.
ton University Archaeological Expedition to Syria in

510
 SASANIAN ARCHITECTU RE. A. HISTOR Y
Ukhay<;lir it is used in the great blind arcade of the Court of Honour, in the small
arcade which, as a miniature gallery, form s the top course of the main building, and
also in window arches,' and the blind arcades of the court fas:ade of the gate in 'Amman
provide a second example, farther to the west.' The arches of the transepts and the lower

FIG. '39. Detail from bronze salver, Staatliche Museen, Berlin (see PI. 23 7),
showing radial arcade with pointed horseshoe arches .

arcade arches of the U mayyad Mosque in Damascus are also clearly horseshoe shaped, as
is likewise the relieving arch over the central western doorway.'
One may assume that al- Valid' s architect derived this motif from Syrian Christian
architecture, but the Ukhay<;lir and 'Amman examples must have been taken over from
Sasanian prototypes. The form did not last long in the East, disappearing from 'Iraq
and Syria during the first centuries after the Hijra, but in western Islam it won so firm
a foothold that it is still in use there today. The only horseshoe arches so far found in
Persian Islamic architecture are far to the east. There are a number in the mosque
of Peshavaran, in east Sistan, one at DClst Mul,1ammad, likewise in Sistan, and one
I REUTHER, Ocheidir, Leipzig, 1912, p. 14, are published by J. STRZYGOWSKI, Mschatta, II , 'Jahr-
Fig. 14, pp. J I, 50, and Fig. SI; BELL, Palace and buch der A:. oniglichtn preuftiJChen KunstJammlunge1t, xxv
mosque at Ukhaic.iir, pp. 8, 24, 166. (1904), pp. 350-3, Fig. I I 6, and PI. XII. See the part
2 DI EULAFOY, op. cit., V, p. 102, Figs. 92, 93, gave of n. I on p. 531 concerning 'Amman .
the first illustrations, after drawings by Mauss . SPIERS, 3 SPIERS, op. cit., Fig. I I I, shows the horseshoe
op. cit., Fig. 39 , reproduces a drawing after a photo- shape of the tran sept arch. CRESWELL, Early Muslim
graph. An accurate drawing of the monument was architecture, pp . 11 I and 118, and Figs. 63, 64.
made by B. Schulz. His drawings and his description

5II
 OSCAR REUTHER
remains standing in the ruins of the guard-house at Qal'a-i-Bist (Chapter 39 F) , which
is today within the boundaries of Afghanistan.'
The Syro-Mesopotamian origin of the horseshoe arch has not been conclusively
established.' It cannot have been directly derived from a timber construction form, for
the Near East, where wood was scarce, had always built in stone and brick. There is more
to be said for the theory that it was derived from the outline of tile conch-shell, which was
used inside the arches of niches.' Shell niches were a favourite decorative element in
Syro- Roman architecture; but they appear with the horseshoe frame only after this
form had been established for the discharging arches of Syrian church portals and in
arcades. Finally, it has been suggested that the horseshoe arch came into Western
Asia from the Indo- Buddhist culture, where it had developed out of the vault- shaped
roofing of houses fashioned of bamboo,' and had come to be tile characteristic gable of
the Chaitya hall, so tllat a symbolic meaning accrued to it as a feature of the dwelling of
a deity or holy person and it became an essential element in religious architecture. That
Buddhist precedents exerted an influence in east lriin well after the advent of Islam is
shown by the horseshoe arches ofPeshavariin, Diist Mul~ammad, and Qal'a-i -Bist, which
clearly were derived from this source. But such a definitely Indo-Buddhist form must
have penetrated Iriin in the pre- Islamic period, for this type of arch had not been used
for centuries in India itself or in the neighbouring Buddhist regions to the northwest.
The source of the pointed arch in Sasanian architecture is connected with this problem.
The fact that there are pointed -arched barrel vaults, built in corbelled construction, at Qa!r-
i-Shirin might suggest that the pointed arch was a common feature of Sasanian architec-
ture, especially since it had already occurred much earlier as the result of similar construction
farther to the west. When the towers of the city wall of Dura-Europos on the Euphrates
were excavated, it was found that the doors in them had, in place of horizontal lintels,
corbelled courses of masonry.' The corners of most of the masonry blocks are cut off in
a straight line, making a triangle tllat corresponds to the Mycenaean discharging triangles. 6
In some cases the surface is fini shed by cutting the blocks in a broken instead of a straight
line, which results automatically, as it were, in a pointed arch. Corbelling fini shed in this
way is found over a door in tile tower near tile temple of the Palmyrean gods in Dura.
The broken line is rounded out on each side, making two bowed jambs, which intersect
in a point at the crown. The city wall of Dura was rebuilt in the Partilian period, about
30 B.C., but tllese towers with the door in them survived from the Seleucid construction.'

I POPE, The mosque at Qal'a-i-Bist, Bulletin of the Cullura Espanola, HI (1906), pp. 785-8 I I.
American bUli/Ule for Persian Art and ArchaeQlogy, IV S P. V. C. BAOR-M. I. RO STOVTZEFF) The excava-
(1935), pp. 7- 1 I. tions at Dura-Europos) Preliminary report of second
1 The much discussed question has been cri tically season of work, October 1928-April 1929) New
summarized again in CRESWELL, Early Muslim archi- Haven, 1931, Pis. XXII-XXIV.
tecture, pp. J 37-9. 6 W. J. ANDERSON-R. P. SPIERS) Architecture of
J SARJI,E-HE RZFELD, Iranische Fe1sreliefs, p. 23 5. Greece and Rome) London) 1902) Figs . I) 6 .
4 The Indian derivation was advanced by G. T. 7 The writer is indebted for this information to
RIVO IRA, Architettura musulmana, sue origini e suo Dr. Armin von Gerkan) who studied the ruins of Dura-
sviiuppo, Milan, 19141 pp. 114- 23, and G6MEZ Europos in the spring of 1934.
MORENO, Excursi6n a traves del area de herradura,
•

SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y

The Sasanian builders who arrived at pointed-arched barrel vaults in the same way
seem, on the contrary, to have determinedly avoided the visible effect of a pointed arch,
for no Sasanian building has yet been found in which the form is revealed, whether in
corbelled construction or in true vaulting. The niches of the gallery on the back of the
Taq-i-Kisra fa~ade (PI. 152 B), which have been cited as evidence that the pointed arch
was current in Sasanian architecture, are not really pointed.' For while the bricks which
constitute the soffit form an angle at the peak, this is purely accidental and represents no

FIG. 14.0. Ctesiphon, Taq-i-Kisra, detail of FIG. 14 I. Detail from silver plate with fortress,
niche on rear of fal):acle (see PI. 152 B). Hermitage Museum (PI. 233 B and Fig. 149),
showing portal arch.

structural function (Fig. 14.0). Hence it is reasonable to assume that these little arches
were rounded out with a concealing surface finish.
Pointed arches do, however, appear on the bronze salver in the Staatliche Museen,
Berlin, showing a building with three domes.' The arcade in the upper part of tlle flat
fa~ade consists of round arches carried on columns, but the arch of the portal, the
only aperture, is drawn with a point. Again a pointed arch, but with haunches in the
form of a horseshoe arch, appears twice in the radial arcade on the salver, the rest of it
consisting of normal horseshoe arches. The portal arch of the castle depicted on the
silver plate in the Hermitage (PI. 233 B and Fig. 141) likewise shows an unmistakable
point. The artists who drew these arches must have known this form in actual buildings
of tlleir time, that is, of the sixth century A.D. 3 But it would be an error to conclude from
this that the pointed arch was characteristic of Sasanian architecture. The buildings of
I HERZFELD, in SARRE-HERZFELD, Archaologische (1933), p. 7. This publication throws new light on a
Reise im Euphrat- und Tigrisgebiet, II,p. 66, corrects whole series of problems in Sasanian architecture.
Dieulafoy's error on this point (op. cit., v, p. 69), and 3 The L:nin~rad sil~er plate has been. attributed to
in Fig. 169 gives an accurate drawing of the niches: the early Sasaman perIOd by SARRE, Die Kunst des
idem, Die Genesis cler islamischen Kunst und das alten Persien, Berlin, 1923, p. 53. The writer is in-
Mshatta-Problem, Der Islam, I (1910), p. I 12, n. I. debted to Professor KUhnel for the opinion that there is
The drawing in Fig. t 40 was made by the writer after no reason against dating in the late Sasanian period, the
a sketch done on the site in 1929. sixth century, and the use of the horseshoe and ogee
2 POPE, A Sasanian garden palace, The Art Bulletin arches would tend to support this dating.

3U
 OSCAR REUTHER
the period that have survived testify too decisively against this. Moreover, the arches
represented on the two metal plates are not composed of two segments of a circle
meeting at a point, the shape that is usually produced by corbelling and also as a rule
by vaulting. They are keel or ogee arches, a form which again indicates a connexion
with Indian cultures, for it is as characteristic of the outer line of the Chaitya gable as the
horseshoe arch is of the inner line, and it constantly recurs as a symbolic ornament
in Indian art, both in the original combination with the horseshoe arch and also
separately. It is especially common as a frame for the aedicules in which a figure of a
god or of the Buddha is set.'
That this ogee arch may have come into Iranian Sasanian architecture from Indian
Buddhist art, perhaps even associated Witll its religious significance, is, to say the least,
not improbable. Through the door of both buildings depicted on the metal plates is
seen an altar, such as is commonly represented on Sasanian coins. Thus the fire altar
seems to stand in an aedicule framed by an ogee arch. Most mi~rab s are in the form of
an ogee or pointed arch witll a lamp shown within the niche, which is surrounded by
an inscription quoting the Light Sura (Sura 24, v. 35).' The fire altar seen through the
pointed-arched portal might conceivably have had some connexion with this type of
mil.nab, tllOugh such an interpretation involves two rather hazardous assumptions: first,
that designs like those on the metal plates were numerous; and second, that the subsequent
designers of the mi~rab s misinterpreted them and saw in the portal and altar a unit.'
In any case, the pointed arch was not a feature of Sasanian secular architecture, and
apparently it was not granted general aesthetic status until towards the end of the sixth
century, when it was used in Christian Syria, notably in the domed basilica of Qa~r
ibn Wardan, and there it developed out of the vaulting system.' Early I slamic builders
in Syria and 'Iraq used the form commonly.' It is, however, possible to contend that
the ogee arch, which came from quite another source, and was first incorporated as an
architectural element in late Sasanian building, opened the way for the introduction of
the pointed arch into Islamic architecture.

EXTERIOR DESIGN
The form of the arches, whether as apertures, notably as ivan openings, or as members
of a blind arcade, or as niches, was an important factor in Sasanian exterior elevations.
What are the stylistic relations of Sasanian exterior design? Sasanian structural methods,
I Forthe penetration into Sasaniandesign of a related for fire temples were numerous and all offered this
motif likewise apparently·Indian see pp. 67 S-6.-Ed. image of the altar seen through the door. Nor need the
2 K. WULZINGER, Die Piruzmoschee zu Milas, in artist have made the mistaken assumption that the altar
Festschrift ZUi Hundertjahrfeier def Technischen was in an aedicule. The door with the altar framed in
Hochschule, Karlsruhe, 1925, p. 20 ; SARRE-HERZ- it might well have been an immediate transcription of
FELD, Archaologische Reise im Euphrat- und Tigris- the familiar visual impression (see pp. 87O-I).-Ed.
gebiet, I V, pp. 286-7. 4 BUTLER) op. cit.) p. 32) Pis. I) II) III.
3 On the other hand, the artist might have made the S REUTHER, Ocheidir, p. 3 j BELL) Palace and
transition not from the representation of a fi re altar mosque at UkhaiQir) p. 165 j CRESWELL, op. cit.)
seen through a door on a metal plate (or any other PP·27 8- 80 .
object), but from his own repeated direct perception,
 SASANIAN ARCHIT E CTU RE. A, H ISTO R Y
especially the types of vaulting and hence the basic spatial form s, were indubitably
derived from Oriental, primarily Iranian, antecedents, and had nothing to do with R oman
vaulting. The typical exterior designs of Sasanian buildings have also been considered
fundamentally Oriental, the assumption being that they represented a deliberate return
to Ancient Eastern principles ; or to avoid such an hypothesis, buildings like Firuzabad
and Sarvistan have been classed as Achaemenid, which puts them more than five centuries
earlier than they are usually dated.' A more careful study, however, shows clearly that
in exterior design also the Sasanian architects followed Parthian precedents, utilizing the
Hellenistic heritage transmitted by the Parthian builders. They worked with H ellenistic
elements, but they used the H ellenistic elements in an Oriental manner, composing their
exteriors pictorially instead of structurally, so that they were merely a screen without
any mechanical relation to the vaulted structure behind them. As a result, these exteriors
have a distinctive character in which the dual inspiration is quite evident.
This is most clearly exemplified by the Taq-i-Kisra (Pis. 1+9, 152 A). That this gigantic
fa~ade has Roman-Hellenistic affiliations has never been denied. The blind architecture
composed of engaged columns, entablatures, and niches has been credited to Palmyrean
influence ;' but since the fa~ades of the Ashur Parthian buildings have become known,
the connexion of the Taq-i-Kisra with these has been beyond argument. On either side
of the parabolic ivan arch the huge brick wall is divided into six stories, and the same
principle is found in the fa~ade design of the Parthian palace at Ashur (see pp. +15-6) .
Moreover, the plan of the palace in Ctesiphon, as the excavations have shown, corresponded
with that of the Parthian palace court,' inasmuch as there was a second ivan fa~ade con-
fronting the one that is still standing, with an enclosed court between them. Thus
the H ellenistic-Roman features of the architecture of Ctesiphon do not come directly
from a Western source, but are a Parthian inheritance. Indeed, a local 'Iraqi tradition
may have been followed, and the Partllian royal palace at Ctesiphon may have had an
ivan fa~ade of this type, which served as model for the Sasanian palace.
The Taq-i-Kisdi is so called because according to tradition it is the remains of tile
Ivan-i-Khusraw, the palace of Khusraw I Anushirvan (A.D. 531-79); but this has been
denied on the testimony of Ibn al-Muqaffa' R6zbih, whom f,Iamza al-I!fahani quotes
as stating that the ruins were the remains of a palace built by Shapur I, an assertion tllat
J;Iamza himself denies. A second argument that has been advanced against tile late dating
rests on tile tlleory that tile columns in an architectural design of tile sixth century would
not carry the entablatures directly, but would be connected by arches.' This objection,

I See p. 4 93 , n. t. tance to the passage of the Sasanian Chronicle, the

2 This view is expressed by Herzfeld, in SAR.RE- Khudhay-Noma, translated by Ibn al-Muqaffa' Rozbih
HERZFELD, op . ci t., p. 75 . and reported by l:iamz.a al-I!}fahanI, stating that the
J R EUTHER, Die Ausgrabungen der deutschen tvan then existing in Mada'in (Ctesiphon) was built by
Ktesiphon-Expedition im Winter 1928- 9, p. 21. Shapiir I, despite the fact that I:Iamza himself im-
4 H ERZFELD, op . ci t., 11, p. 75, finds an argument mediately thereafter cites the opposing view of a Zoro-
for an early dating in the fact that the engaged columns astrian priest livin g in his own day, who told him that
on the fa~ade carry the entablature directly and are not this was an error and that the Ivan still standing in Ctesi-
connected by arches, and attributes decisive impor- phon was built by Khusraw II ParvIz . See p. 493, n. 4.
 OSCAR REUTHER
however, is not sound. The double, superimposed columns which appear on the exterior
of the apse of the basilica in Qalb Lawza carry the cornice directly.'
The similarity of the fa~ade, in essential principles, to Parthian buildings might be a
stronger argument for dating the Taq-i-Kisra early, but parallel designs also appear in
early Islamic architecture, and especially the north front of the central court at Ukhay4ir,
among others, bears witness to the long life of the tradition.' Thus these Parthian prin-
cip
es could have been used at Ctesiphon just as well under Khusraw I as under Shapur I
(A.D. 24I-72).
There are, moreover, three specific indications that the Taq-i-Kisra dates from
relatively late in the Sasanian period, notably two features in common with sixth
century Byzantine architecture, and a detail found in contemporary Syrian buildings
and early Islamic architecture. The use of a 'colossal order' of engaged columns to
join two stories of a fapde was already common in Roman architecture, as numerou s
Roman triumphal arches and the fa~ades of Roman city gates show.' But the combina-
tion of very large and very small elements of the same type, for example, of two stories
of arcades markedly different in scale such as appear in conjunction with the 'colossal
order' on the Taq -i -Kisra fa~ade, is not found in the architecture of the Mediterranean
countries before the sixth century. Again, the principle of coincidence of the vertical
axes of the various stories, a postulate to which Classical architecture adhered rigidly,
was not rejected until the sixth century.' Thus on the sides of the nave of Hagia Sophia
there are five units in the lowest arcade and seven in each of the galleries above, so that
here, as on the Taq-i-Kisra fa~ade, there is no continuity in the main vertical members.
Finally, the detail that recurs in Syrian buildings of the sixth century and in early
Islamic architecture is the arcade moulding on the ivan archivolt. This is found, for

I M. DE VOGUE, Syrie centrale, Architecture civile 3 The Gate of Augustus in Nimes is a compara-
et religieuse du I au VII siecie, Paris, I86S,Pls. uS, tively early instance of the 'colossal order', which was
134; BUTLER, Publications of an American archaeo- employed again in late Renaissance architecture,
logical expedition to Syria in 1899-1904, Part H, especially by Michelangelo and Palladio. The Roman
Architecture and other arts, New York, J 904, triumphal arch which shows the principle most clearly
p.2 23 · is that of Trajan in Timgad: see J. DURM, Baukunst
.z The first careful study of the Palace of Ukhay4ir, der Etrusker und Romer, Handbuch der Architektur,
which lies in the desert south of the Shethatha oasis, II, 2, Stuttgart, 1905, Figs. 502, 799.
was undertaken by L. MASS IGNON, who published his -4- An outstanding example of the disregard of
findings in: Note sur Ie ohateau d'al Okhaider, Comptes vertical continuity in Byzantine elevation design is to
Rendus de l'Academic des inscriptions et Belles-LetlrtJ, be found in the mosaic of San Apollinare Nuovo,
J 909, pp. 202-12; and idem, Les chateaux des princes Ravenna, representing the Palace of Theodoric ;
de Hirah, Gazelle de; Beaux-Arts, 4eme periode, 1 (u) O. W ULFF, Altchristliche und byzantinische Kunst,
(1909), pp. 297-306. Massig non considered Ukhayc;lir Berlin-Neubabelsberg, 19 I 4, Fig. 373. The only
either Sasanian or Lakhmid. That it was built in the Roman Imperial example known to the writer is the
early Islamic period was recognized by BELL, Palace and 'Storied tomb' in Petra, to which the Taq-i-Kisra
mosque at Ukhaic;lir, pp. 160-8, and REUTHER, Ocheic;lir, fa~ade is compared by Bell (op. cit., p. 129), yet this is
p. I. These two publications give full illustrations of the not a fa~ade with two stories, but an attempt to repre-
buildings, which are oneof the most revealing sources of sent on the flat rock wall two fa~ades) one behind the
information on the persistence of the Sasanian architec- other: see H. KOHL, Kasr Fira'un in Petra, Leipzig,
tural tradition into Early Islamic building. BELL, op. 19IO, p. 39; and K. WULZINGER in W. BACHMANN-
cit., pp. 129-33, sees the parallel between the Taq-i- C. WATZINGER-T. WIEGAND, Petra, Berlin-Leipzig,
Kisra fa~ade and the court fa~ade of the Palace of 1921, pp. 16-17·
Ukhayc;lir. Cf. also REUTHER, op. cit., p. 3 I.

5I6
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
example, on the apse arch of the church at Qalb Lawza, in Ukhay~ir,' and at Samarra;'
and it does not appear in any building prior to the sixth century. Thus these three
features all tend to support the identification of the Taq-i-Kisra as the palace of
Khusraw I.
Compared with the Baroque character of the Taq-i-Kisra fa\,ade, in which H ellenistic-
Roman architectural elements are employed with a complete disregard of the inherent
principles of balanced proportion and of vertical coincidence, the architecture of the
palace of Firiizabad seems rather archaic. The fa\,ade had, like that at Ctesiphon, a
great ivan arch in the middle, but this faced not on an enclosed court, but, as at
Hatra, towards the outside. The rest of the Firiizabad fa\,ade is shown, in the survey
drawings and reconstructions, as a flat wall with a row of rather shallow niches at the
top, and possibly another just above the base course.' This reconstruction, which has
also been followed in the model (PI. I47C, D), might, however, be questioned. The
stucco facing that originally covered the coarse rubble construction, in which all the
architectural details were executed, has entirely disappeared, and the fa\,ade, which even
in Flandin's time was in very bad condition, has now collapsed. Blind architecture
with a series of stories, such as we see on the Taq-i-Kisra and the Parthian buildings of
Ashur, may possibly have been used at Firiizabad, also, with a scaffolding of engaged
columns and mouldings worked out in stucco in rather low relief. This would, of course,
be in opposition to the very effective designs in high relief still to be seen on the side
walls and the back, which are in better condition. These have been compared to the
half-round vertical mouldings and rectangular channels used by the Babylonians, and
cited as a deliberate revival of Ancient Eastern elements, another example of the archaism
which undoubtedly did pr,?mpt the Firiizabad builder to use the Achaemenid cavetto
cornice. But just as the door and niche frame s which carry this cornice are in essence
Hellenistic, so, too, the Firiizabiid exterior designs, which are composed of round-
arched niches with half-columns in front of them, can be explained only as derivatives
of Hellenistic architectonic principles.'
On the other hand, the change in proportions from the Hellenistic norm, whereby
niches and column shafts became tall and slender, obviously was dictated by innate
Oriental taste. To the master builder of this palace the half-columns were not structural
elements, constituting a tectonic scaffolding, but only a means of lightening the rigid,
massive, exterior walls by creating vertical furrows, and thus they did fill the same role
as the flute and channel treatment of Babylonian buildings. In the final analysis these
exterior designs are really projections on a flat surface of the Helleni stic columnar hall,
and when they are so regarded, the palace of Firiizabad corresponds to a significant
I DE VOOUE, op. cit., Pis.
126, 129; BUTLER, op. cit., PI. 40. DIEULAFOY, op. cit.} PJ. XVJI, omits this lower
pp. op. cit., pp. 2S, 3'2, Figs. 24, 26. row of niches in his reconstruction. His photograph,
222-3; REUTHER,
1 SARRE-HERZFELD, Archaologische Reise im PI. H, is so much retouched that one cannot judge from
Euphrat- und Tigrisgebiet, I, p. 81, Figs. 31, 32. it. Recently the whole front building at Flriizabad has
J FLANDIN-COSTE, Voyage en Perse, I, show in a collapsed.
drawing of the ruins, PI. 38, a lower row of niches, .. This is also the opinion of Herzfeld, in SARRE-
which also appears in the reconstruction of the fac;ade, HER.'lFELD, Iranische Felsreliefs, p. 129.
 OSCAR REUTHER
degree with the 'Peripteros' in Ashur (see p. 436). In each case the motif derived from
this columnar hall is used on the side and rear fa9ades, and not on the ivan fa9ade. In
the court of the palace of Firuzabad the niches, which are worked out in low relief
without the half-columns, frame the small ivans which are symmetrically placed on
either side.

. ...
•
• \
, .... '.
.4. •.
....
. -'"
"
. . . l. _

".... ' " "''- ;': ·i.~. ··/·:,,:,.',:..: >,:.

", :
';:' .'; '."," '. ".
. .. . ..... ; ' .:.. . ' ' .
.
.: ...•..... :"."..., ' ':
. . '
.. '
",'

:." . . "
' '.,: ~ :,•.:.

. :,. ; .. ' . ......,. . ",": ... :.:. ;,"

",

FIG. 14 2 . Sarvistan, Palace: proposed reconstruction of exterior.

The fapde design of the third Sasanian palace of which a great part is still standing,
that at Sarvistan, also follows a Parthian prototype. Across the front are three ivan
arches, the arrangement that had been used on the fa9ade of the Parthian Ashur temple
in Ashur and on some of the buildings at Hatra (see p. 436 and Fig. III).' At
the ends are engaged columns, and a group of three engaged columns is on either side
of the central ivan, the whole series being connected by a cornice, if the reconstruction
is correct (Fig. 142). Here again is a Hellenistic fa9ade reinterpreted according to the
Iranian conception, which deprived the columns of structural function and employed
them as vertical strips applied to the wall, preferably arranged in close ranks in order
to intensify the effect. Similarly, many of the engaged columns on the Taq-i-Kisra
fapde are doubled, not only the small ones framing the niches of the half- stories, but
likewise the large ones of the lowest course.
I ANDRAE-LENZEN, op. cit., Fig. 42.

518
 SASANIAN ARCH IT ECTU RE. A. HISTOR Y
At Taq-i-Biistan, also, a triple ivan composition was used or projected, for the plan
may not have been completely executed. The large central ivan, which is cut in
the living rock, was flanked by two smaller masonry ivans extending in front of the
central ivan.
Another type of three-arched fas:ade must have been used at Kish and at Damghan
on the three-aisled halls which were open in the ivan form. Here the arch of the central
ivan was evidently much higher and wider, decisively dominating (see Fig. 167 a). If
the reconstructions of the projecting porches of the 'Imarat-i-Khusraw at Qa!r-i-Shirin
and the Hawsh-Kuri palace as triple ivans are correct, essentially the same fas:ade must
have appeared here (see Fig. 152).
The round piers built of brick or stone which carried the vaults at Damghan
and Kish and also at Qa!r-i-Shirin, and appeared on the fas:ade carrying the
frontal arch, are not columns in any exact sense. They originated in the Iranian
vaulted construction, and it was clearly irrelevant whether they were round or
rectangular. In the Parthian hall in Ashur, built with four piers, they are rectangular,
as they are, also, in the Sasanian church at Ctesiphon. At Damghan they are round,
6 feet (2 metres) in diameter, and were probably very low, of gross cylindrical shape,
quite unlike a column, but in essential idea like the early Sumerian thick, round, brick
piers used at Uruk.' They could not, of course, have been derived from a Sumerian
prototype, but in every instance they resulted from the same conditions, from dispropor-
tionately heavy brick construction. In the Parthian peripteros they are equated with the
Greek column, and the round form may have been derived from this relation; but they
are not in any sense its legitimate successor.
The only true derivative of the Greek column in Sasanian architecture is the half-
column used in fas:ade designs, for example in arcades ; but this had so degenerated that
it had often lost the essential differentiation of the column into base, shaft, and capital.
Engaged columns without either base or capital have long been known on early Islamic
fas:ades, notably at Samarra, R aqqa, Ukhay<;iir, and 'Amman, but their use in fas:ades
of Hellenistic origin was clearly typical of Sasanian architecture. The horseshoe arch
of the double door excavated at Ctesiphon (see p. 510) is carried by half-cylindrical
shafts, above which, as at 'Amman, there is a plinth instead of a capital, and the columns
that carry the radial arcade on the Berlin bronze salver are represented in exactly the
same way (PI. 237)' The columns of tl,e top blind gallery of the castle on the
Hermitage silver plate are also drawn without base or capital.' Here, in true Roman
fashion, they stand in front of tl,e arcade and carry an entablature (Fig. 149). Hence
it seems most probable that the engaged columns on the Taq-i-Kisra fapde should also
be reconstructed without bases and capitals.'
I E. HEI NRIC H, Vierter vorlaufiger Bericht uber 2 Compare also the coupled columns ca rryi ng the
die von dec Notgemeinschaft dec deutschen Wissen- arcade, each pair with a single impost block, on the Pope
schaft in Uruk-Warka unternommenen Ausgrabungen, green-glazed bottle (PI. 192 B).-Ed.
Berlin, 193'2, p. 12, PI. 8; idem, Schilfund Lehm, 3 SPIERS, op. cit., p. 80, also assumes that the
Berlin, 1930, pp. 31-5, Fig. 13, and PI. 6. columns on the T~ q-i-Ki s ra fa ~ade lacked capitals. On
 OSCAR REUTHER
The Sasanian architect was certainly not aware that these half-columns had originally
functioned as structural elements. The logical equivalent of a column in the Sasanian
system is the thick, round pier built of brick or stone, remnants, as it were, of what had
originally been a solid wall, and these, also, probably carried the arch without the
interposition of any capital, as they do in the mosque of Nayin and the Tarik-Khana
in Damghan (PI. 258). Thus Sasanian architecture, as exemplified at Firiiziibad, Sar-
vi stan, Ctesiphon, Qa~r-i -Shirin, and also, in so far as we can judge, at Kish and
Damghan, was essentially astylar.
There is, however, a Sasanian columnar architecture. The capitals of Taq-i-Biistan,
Bistiin, I ~fahan, and Qal'a-i- Kuhna (PI. I53 A, B),' while they are purely Sasanian in
ornament, are derived from Byzantine impost capitals, hence they must have carried
Byzantine arcades.' Thu s they represent a foreign intrusion, which, moreover, probably
appeared only late in the period, for in the ornament of three of them are relief portraits
of Khusraw II, which dates them between 590 and 628. Arches on columns of this type
appear also on Sasanian metal vessels, for example, on one of the plates in the H ermitage
Mu seum (PI. 2 07B) . Here the columns which carry the arch that frames the aedicule
have an impost capital covered with foliate ornament while the base is decorated with a
rosette, and on the H ermitage vase with women bearing offerings there is a columnar
arcade with rosette capitals.' But these columns supporting arcades are in no sense
peculiar to Iranian design. Columns of another kind, however, have long been known at
Taq-i-Biistan. These are only carved in low relief on the back wall of the grotto con-
taining the equestrian statue, but since they are connected with a horizontal concave
moulding they must have been truly architectural in origin (PI. I68 A). H ence they prove
that Sasanian Persia had a columnar architecture.' That no remains of this architecture
have yet been found is due to the fact that the columns and beams were wood and there-
fore unable to survive the centuries. Excavation of Sa san ian sites would undoubtedly show
that in certain parts of Persia wooden columnar structures were at least as common as
they are today, for the exact parallel of the Taq-i-Biistan type is still found in northern
and western P ersia and especially the Kurdish districts (Fig. I43 a, 6) . In early Islamic
times the style was evidently considered specifically Persian, for Ziyad, MU'aviya' s
governor, when he built the hypostyle mosque at Kiifa, employed a Persian architect,
who is said to have worked previously for Khusraw.'
This columnar architecture was of early Persian, or, more accurately, Median deriva-
tion, and even the specific design of tl,e Taq-i-Biistan capital must approach very closely

the other hand, Herzfeld (in SARRE-H£R.ZFELD, Orfevreriesasanide, Moscow-Leningrad, 1935,PI.45.

Archaologische Reise, 11, p. 69) assumes that there For other foliate capital s see PI. 246 B. A lost ewer (see
were capi tals. Fig. 252) shows an arcade carried by coupled column s,
1 The Qal'a-i-Kuhna capital was removed to T:tq-i- each pair carrying a single expand ing Impost block,
Bustan in I 936.-Ed. and interposed between this and the springing of the
1 FLANDIN-COSTE, op . cit., I, Pi s. 17 , 18,27,28 ; arch is a figure, in full round or very hig h relief, of
H ERZFELD, Am T or von A sien, Berlin, 1920, p. lIZ. a kn eeling eros in caryatid pase.-Ed.
l See V. SM IR NOFF , L'argenteri e orientale, St. 4 See Fig. 30S.-Ed.
Petersburg , 1909, PI. XLVI; or J. OUELI-C. TREVER, S CRESWELL, Early Muslim architecture) p. 36.

520
 SASANI AN ARCRIT ECTU R E. A. RISTO R Y
that of the ancient prototypes. A slender fluted shaft carries a rectangular plinth on
which is set the capital. This broadens out towards the top in a trapezoid, which has
scalloped sides. The original capital, like the Ionic capital to which it bore an essential
relation, was evolved by developing in a bilaterally symmetrical form the decoratively
rounded and curved ends of the saddle beam.' The Persepolis and Susa columns
(Pis. 87, 101) with the bull and unicorn protomes, the volutes and leaf calices, already
represent a remote derivative.
In Persia today and in 'Iraq, which follows Persian architectural styles, the wooden
columnar structure with horizontal
beams and a flat timber roof is com-
monly combined with vaulted con-
struction. The Imamzadas in Najaf,
Karbala,' Ka?;imayn, Samarra, and
Qumm, as well as such mosques as the
Masjid-i-Naw in Shiraz, the mosque
at Qiimisha, the shrine of Shah 'Abd al-
'A?;im, and various buildings in Mazan-
damn, are typical modern examples of a b
porticoes with slender wooden columns
in conjunction with vaulted ivans and
dome chambers, and buildings of FIG. 143 a, h. Wooden capitals from modern Iranian style
buildings in 'Iraq: a. in Kirkuk; h. in KarbaJa.
I!fal1an dating from the reigns of Shah
'Abbas and his successors (Pis. 4-73, 4-77 B) show how effectively these two hetero -
geneous architectural principles could be combined. No Sasanian ruins indicate
whether this practice was current in that period, but the fact has now been proven,
an essential addition to our knowledge of the history of architecture obtained with-
out costly journeys or excavations. 3 The building represented on the Berlin bronze
salver, though domed, has on each side very slender columns on a vase-shaped base,
with calyx capitals, carrying a rather curious entablature. The artist, like the early
Egyptian draftsmen, employed no perspective and so has given a conceptual (vorstellig)
rather than a perceptual (anschaulich) representation, showing that the vaulted and domed
structure had porches with slender wooden columns and a flat timber roof. In view
of this evidence the reconstruction already proposed of the Amarat-i-Khusraw in Qa!r-i-
Shirin, with a 'tarma' with wooden columns, must be accepted as at least possible.'
The usual cornice on Sasanian monumental buildings must have consisted of stepped
crenellations. Thu s the Taq-i-Biistan rock-cut fa~ade is crowned with these crenellations
(PI. 159B), as are the monument at Paikiili, the building represented on the Berlin
I H ERZFELD, op. cit.) pp . 104-8, Figs. 23, 24, 26; 3 Seep. 513, n. 2 .
REUTHER, Das Wohnhaus in Bagdad und anderen .. F. OUMANN, Zur Deutung des romischen Kerns
St~d ten des Irak, Berlin, 191 0 , p. 63, Figs. ISZ, 153. im Trierer Dom, Bonner 'Jahrbllcher, Heft 12.7 (1922),
2 A. NOLDEKE, Das Heiligtum al-Husains zu Ker- Figs. 13, 14.
bela, Berlin, 1909, PI. v.
3X 521
 OSCAR REUTHER
bronze salver, and the fortress engraved on the Hermitage silver plate, and it has always
been assumed that the palaces of FirGzabad and Sarvistan had this kind of roof coping.
The Ancient Eastern form of battlements,' which was transmitted to the Sasiinian architects
by the Parthians (Fig. 14+ a),' had rectangular steps such as appear at Taq-i-BGstan, but
in the two metal plate designs the battlements are stepped in acute angles (Fig. 144 b,c),
a variant of the Ancient Eastern form which must be considered specifically Sasanian.
The type appears later in Buddhist Central Asia and on Fii\imid and AyyGbid buildings
in Cairo.'

a
a. From the commemorative monument of Narsc
(293-30 3), Paiku!1 (see Fig. 164).

\1\1\1\1\1
b ,
h. From bronze salver with building, Staat~ c. From silver plate with fortress, Hermi-
liche Museen, Berlin (PI. 23 7) . tage Museum (PI. 233 B and Fig . 149.)
FIG. 144 a-c. Stepped battlements.
Below the crenellations on the PaikGli monument is a row of circular loopholes, and
a series of ovals is drawn below the battlements of the building on tile Berlin bronze
salver. This combination of stepped battlements witll circular embrasures is an inheri-
tance from the Ancient East, for it is represented as early as the Assyrian period on
fortresses shown in reliefs.' The round loopholes were metamorphosed into rosettes
when the battlements had ceased to be a means of defence and become mere decoration,
as at Dur Sarrukin, and later on Parthian buildings in Mesopotamia. At Ctesiphon were
found stucco roundels about 3 feet (c. I metre) in diameter, with a circular hole in the
centre, framed by a rosette composed of anthemion segments (Pi. 173 A) . These were
combined with small, square, brick pillars to make a kind of balu strade, which
I V. PLACE, Ninive et l'Assyric, Paris, 1857, III, J HERZFELD, op. cit., Fig. 3, represents a number of
Pis. 34, 35; W . JORDAN, Konstruktionselemente as- step.ped crenellations from Parthian and early Islami c
syrischer Monumentalbauten, Berlin, 1910, pp. 16-17 , bUildings and also from Buddhist buildings in Turfan,
Figs. 9, 10; W . ANDRAE, Die jungercn Ischtartempd and discusses the Ancient Eastern origin of this form.
in Assur, Leipzig, 1935, Figs. 2, ); idem, Das Gottes- 4 Fortresses with battlements and circul ar em-
haus und die Urformen des Bauens im Alten Orient, brasures are often shown on A ssyrian reliefs : see
Berlin, 1930, p. 81, Fig. 82. L AYARD, A second series of the monuments of Nine-
Z Figure after HERZFELD, Paikuli, pp . 2-4, Fig . 2 . veh, London, I8sr, PIs . 19, 29, 36,39 B.

522
 SASANIAN ARCHITECTU RE. A. HISTOR Y
certainly was the equivalent of the row of loopholes. Reconstructions of the Taq-
i-Kisrii fa~ade have shown a H ellenistic entablature such as undoubtedly crowned
the buildings at Hatra, and this would be a correct assumption if the architectural
details were directly of Western derivation.' But the loophole balustrade shows
that here, as on the other Sasanian palaces, there was a battlemented cornice, hence
in this respect, also, the design was consistent with the Parthian tradition. The recon-
struction (Fig. 145) shows these Taq-i-Kisra battlements stepped in right angles, but
more probably tl,e angles should be acute. All the buildings of the palace precincts

. ,: -.
FIG. 14-5. Ctesiphon) 'Taq-i-Kisra: detail of stepped battlement, completed with
stucco rosettes (PI. I7 3 A).

in Ctesiphon had, judging from the conditions of the find, tl,is pierced and crenellated
coping and there are many Islamic examples of the rather strange combination of a
pierced balustrade crowned with crenellations, which is inexplicable until tl,e pierced
balustrade is understood as a degeneration of the row of loopholes.'
The battlement, as an element of fortress design, may have seemed to the Sasanian
architect a symbol of strength and permanence,' for he used it only on solid walls, not
on the light wooden columnar halls. Thus crenellations appear only above the fapde
of the domed building on the Berlin bronze salver, not on the porches, which are
fini shed with an entablature. On tl,e otl,er hand, the motif doubtless served aestlletically
to lighten the top of the building in order to mitigate the effect of heaviness which was
an unavoidable result of Sasiinian building methods.
Bow windows are a characteristic feature of the domestic architecture of Western
I H crzJeld in his attempted reconstruction,SARRE FLANDIN- COSTE, op. cit., Pis. 2 1 7, 2 1 8, place over
- H ERZFELD, Archiiologische Reise, PI. XLI , assumes a the Ivan arch extrados a dentated mouldi ng of diagonal
low plinth with a dentated moulding below it, which is bricks.
cut by the extrados of the Ivan arch. SPIERS, op. cit., :z H ERZFELD, op. cit., p. 80.
Fig. 36, shows a comparatively high moulding and an 1 For the symbolic significance see pp. 833-4 , 869-
architrave stepped up with several projectin g courses 70 ,87 8.- Ed•
of bricks, with a wide frieze and a simple cornice plinth.
 OSCAR REUTHER
Asia, and the building represented on the Hermitage silver plate shows that such con-
structions, usually called in Persia today shiih-nishfn,' were already used in Sasanian
architecture. On either side on both stories there are projections of this type, partly
supported by brackets and with the roof carried by little columns. It is clearly indicated
tI,at these were built of wood.
DECORATION
Except for the masonry buildings of Taq-i-Girra and Shapur, and the architectural
motifs cut in the living rock at Taq-i-Bustan, the Sasanian ruins, in so far as tlley are
standing at all, are now only gigantic skeletons, stripped to the bone. The rubble
buildings in Fars and in the border districts of western Persia must obviously be clothed,
in imagination, with stucco, and the brick buildings in 'Iraq also were completely faced
with stucco, in which both the finer architectural motifs and ti,e ornament were
executed, a method that had been used in the Mesopotamian river valleys as early as the
Seleucid period (see p. 412). The only decorative form which developed out of brick
building as such was the dentated frieze, which consisted of bricks set diagonally
between projecting courses. This moulding outlines ti,e squinch zone of the dome
chambers at Firuzabad and at Sarvistan, and emphasizes there also ti,e impost line of the
vaulted rooms (Fig. 134 and PIs. 147B, 148c). Later, at 'Amman, it is imitated in
masonry, and it was the antecedent of the zigzag archivolts that appear both at
'Amman and at Ukhay<;lir.' The rest of the smaller architectural elements and the
decoration of Siisiinian buildings were independent of the material and form of the
structure, wrought in the stucco that covered the actual building like a garment.
Only very scanty remains of such stucco facings are to be found in situ. In Firuzabad
ti,e jambs and lintels of the doors and ti,e wall niches in the three dome rooms are still
in place. In the 'Imarat-i-Khusraw in Qa!r-i-Shirin applied architectural elements
forming two stories were still in existence a short time ago: below, a round arched door
framed by half-columns and capitals, more or less of the Doric order, carrying a flat
archivolt; and above that, a series of engaged columns, the space between them varying
in a rhythmical sequence, set on a string course with a simple profile. All this has now
disappeared.' A stucco facing, which also included columns, still adhered to the Qal'a-
i-hazar-dar in Darra-i-shahr, a ruin which unfortunately has not been tllOroughly
studied.'
In recent years a number of fragments of stucco ornament have been recovered in
'Iraq, at Ctesiphon, Kish, and Barghuthiyat, and in North Persia, at T epe J:Ii!ar near
Damghan, and near Varamin.' Most of them are remnants of wall revetments consisting,
I REUTHER, Das Wohnhaus in Bagdad, pp. 6, See also REUTHER, Ocheidir, p. 35 and PI. xx.
100-2. 3 DE MORGAN, op. cit" p. 346, Fig. 208; B ELL,
:: DIEULAFOY, op. cit., p. 102, first noted the zigzag Palace and mosque at Ukhai<;lir, p. 44.
mouldings and archivolts in •Amman and their relation ... DE MORGAN, op. cit., p. 372 and PI. LXII.
to the similar forms in Flriizabad. A photograph by S On the conditions of these finds see REUTHER, Die
Mr. Robert Byron seems to show that the niches on the Ausgrabungen der deutschen Ktesiphon-Expedition
court wall at Ftruzabild also had zigzag archivolts. im Winter 1928-9, pp. 12, 27, '32; KOHNEL, Die
 SASANIAN ARCHITECTU RE. A. HISTOR Y
as a rule, of motifs combined to form continuous designs, the same motif being
repeated, or several being used together recurring in a regular sequence, both vertically
and horizontally. The technique favoured this arrangement, for the revetments were
made up of plaques approximately a foot (c. 30 em.) square, so that the units could be
multiplied and simply added mechanically one to another, just as was later the case with
the tiles used in Islamic architectural decoration (see Chapter 40). Indeed, these Sasanian
stucco plaques must be considered the antecedents of the glazed tiles. In addition to
these stucco tiles which covered walls and the surfaces of vaults, and the borders and
mouldings that framed them, remains of stucco figural reliefs have also been found.
Some are practically life-size and they must have played an important role in architec-
tural decoration.
Very few purely architectural elements in stucco have been found. A decorated
circular cushion found at Tepe I:Ii~ar probably served as capital of one of the round
piers that supported the arches of the triple ivan (see Fig. 168), but there are no true
capitals such as appear in the reliefs of Taq-i-Biistan or on the metal vessels showing
architectural motifs, where there are usually capitals in the form of leaf calices; and
column bases are likewise missing. This, however, is not surprising, since the Sasanian
architect often eliminated these members in designing his engaged columns. On the other
hand, it seems curious that almost no mouldings have come to light,' such as were used
at the Taq-i-Girra on socies, imposts, and archivolts, and such as are seen also, rendered
in stucco, on the door and niche frames at Firiizabad. Evidently the Sasanian stucco
modeller gave up the architectural details of Hellenistic derivation, more or less modified,
which his Parthian predecessor had employed, and this inference is confirmed by the
fact that these details are likewise conspicuously absent from the rock-cut fa"ade at
Taq-i-Biistan, where the only architectural elements in the exact sense of the word are
the columns and entablature cut in relief on the back of the grotto.
There is not even, strictly speaking, an archivolt at Taq-i-Biistan as there is on the
door and niche arches at Firiizabad and on the frontal arch of the Taq-i-Girra. For a
true archivolt represents a curved beam and is derived from a Greek architectural form,
whereas at Taq-i-Biisran the inner line of tile main arch is defined by a torus decorated
with an acanthus leaf garland, crowned by a concave moulding that is ornamented with
Rowers alternating with acanthus leaves (Fig. 146a).' Instead of resting on the impost of

Aus~rabungen der zweiten Ktesiphon-Expedition, p. 484; E. F. SCHMlDT, Tepe Hissar excavations,

Berhn, 1933, pp. 16-25; J. H. SCHMIDT, L Expedi- 1931, 'Ihe Mu seum Journal, XXIII (1933), pp. 455-9.
tion de Ctesiphon en 1931-2 (Syria, xv (1934»), pp. I A few fragments of mouldings were found at

8-18; S. LANGDON, New light on early Persian art, Kish (Fig. 147 a, b), but unfortunately there is no in-
The Illustrated Londotl News, February 14, 193 I, p. 26 I; dication of when or how they were onginally used. A
idem, Persian art discoveries at Kish, ibid., March 7, frag ment of a half-round stucco moulding, bearing a
1931, p. 369; idem, A Christian nave at Kish? The dis- simple incised chevron, was found in the hig h place
covery of a second Neo-Persian palace, ibid., April 25, above the temple at Raba~-i-Safrd by Mr. D. M. Wilber
1931, p. 69 7 ; idem, 'Palace Three', the bath of the of the Architectural Survey of the American Institute
Sasanian kings at Kish, ibid" February 20, 1932, for Persian Art and Archaeology.-Ed.
p. :2.73; POPE, New relics of a great period in Persian % Fig ure after a photograph by Dr. W. Bachmann.
art, 'Ihe Illustrated L ondon N ews, March 26, 1932,
,-
/.

OSCAR REUTHER

.. '. .

. .; ' ...- '.

.: . ..... .''. ,.
. . : ': :

.. ' ' ..
. .. .
FiG. 146 b. Taq-i-Biistan, crown of archivolt.
..
' ..

the arch, this torus terminates on each side in a wholly

unarchitectonic pattern of a fluttering scarf, the ends
of which flutter like the ribbons with which the royal
diadem is tied at the back of the head in portraits of
Sasanian kings (see PI. 162 A- C). Clearly the motif
represents the diadem,' and had, not an architectonic,
but a symbolic significance. That this interpretation
is correct is confirmed by the crescent surmounting the
crown of the arch, for this symbol usually appears on
the royal diadem above the forehead (see Fig. 146 b).'
Sections of stucco arch mouldings of this same kind
have been found at Ctesiphon, Kish, and Damghan,
but in place of the leaf garland the torus is ornamented
with scale imbricates, zigzags, or rows of quatrefoils,
while the extrados carries border patterns composed of
different types of palmettes, flowers, and pomegranates,
alternating in evenly spaced rows (PI. 171 A-D). Those
found at Ctesiphon and Damghan also end above the
impost in the floating scarf ends. The best preserved
examples are those found at Kish (Fig. 148, and see
Fig. 176 a- d).
This recurrence of the diadem motif as an arch
moulding seems to be of primary importance in dating
the stucco, for inasmuch as it occurs on the grotto

I DIE VLAFOY, op. cit., p. 96, explains the Taq-i-Bustan archivolt as

a diadem.
2 Figure after a photograph by Dr. W. Bachmann, with details
completed from a photograph by Mr. Pope.

FIG. 146 a. Taq-i-Bustan, panel and archivolt on right (see PIs. r67,
168 cJ.
....

,
 SA SAN I A N ARC HIT. E C T U R E. A. HIS TOR Y
arch of Taq-i-Bl,stan, which was cut for Khusraw II about A.D. 600,' the stucco in
the style to which the diadem motif belongs would probably likewise be late Sasanian.
But on the other hand one must take into account Oriental conservatism, and such a
form might, because of its symbolic significance, have been retained for a long time
even in its architectural adaptation.
The problem of dating the Sasanian stucco ornament is still unsolved and it is im-
portant for the history of Sasanian architecture, for in the first place the date of the

FIG. 14 7 fl , D. Frag ments of stucco mouldings, Kish. Scal e J! 3.

ornament is usually the same as that of the building where it is found, and in the second
place in order to reconstruct a dated ruin it is necessary to know what decorative style was
current at the time when it was built. All the stucco so far found, whether at Damghan,
Kish, or Ctesiphon, represents the same style. Palmettes, flowers, and leaves, for example,
are all treated in essentially the same plastic manner throughout, and the stylistic varia-
tions that do occur in some of the motifs are very slight. Thus the style was current
over a wide area. Yet the obvious conclusion that these buildings must be approximately
contemporaneous is contradicted by certain archaeological evidence. Circumstances at
Damghan point, in the opinion of the excavator, to the second half of the third century.
There is reason to assign the Kish buildings to the time of Shiipur II,' that is to say,
to the middle of the fourth century, while the stucco at Ctesiphon is attributable to the

I This datin g has now been questioned: see K. ( 193 7), pp. 79-9 7. For a summary of the thesis see
E RDM ANN, Zur Chronologie cler sasanidischen Kunst, p. 819. n. 3.-Ed.
For.schungtn und ForlSchrilU , XIII (1937), pp. 169-70 ; 2. Originally thoug ht to be Bahram Gur (see p. 642,
idem, Das Datum des Ta~-i Bustan, Ar.s Is/amica, IV n. I).
 OSCAR REUTHER
sixth century.' Thus one must assume that this decorative style underwent very few
changes, remaining practically stationary for three centuries, and some of the elements
continue almost unmodified in early Islamic ornament.
The difference is very marked between this stucco ornament and that found on the
early Sasanian buildings, notably the petal imbricate on the cavetto cornices over the
doors at Firiizabad, but this does not constitute
decisive proof that the stucco is late, for these
early Sasanian motifs were deliberate Achae-
menid revivals, adopted for symbolic or
propaganda purposes, and hence they might
well have been soon succeeded by motifs
of quite another style, or even coexistent
with these. The alternative style might have
been taken from contemporary industrial
arts. In the late Sasanian period there are
certainly two distinct styles. The ornament
at Taq-i-Biistan has, aside from the diadem
archivolt, little in common Witll the stucco
ornament, and one decorative motif at
Taq-i-Biistan has no parallel in the stucco
ornament so far recovered. This is tile
handsome acanthus tree on the fa~ade beside
tile arch (Fig. 146 and PI. 168 c). The decisive
technical divergence between stone carving
and stucco modelling might explain tllis one
point of difference, for tile sculptor, work-
ing on a continuous surface, was free to
choose any scale, while the stucco modeller
FIG. 148. Stamp for stucco from Kish, floating made his patterns in discrete, self-contained,
end of 'archivolt'. Scale 1: 3.
square units. Hence the 'cosmological tree'
was rendered in stucco as a small repeating pattern (PI. 172 D). The difference is,
however, deeper than tllis. The Taq-i-Biistan tree was developed from a form of the
acantilUs motif which has no connexion whatever Witll either the Roman or Byzantine
version, but is purely Hellenistic.' The stucco foliation, on the contrary, is never the
true acanthus, but is derived from the anthemion or half-anthemion, and hence from a
plant motif anterior in Classical ornament to the acanthus.' The same delicately drawn

I The opinion that this stucco ornament is all late

Sasan ian is advanced by KOHNEL, Die Ausgrabungen On the basis, however, of Dr. Erdmann's recent re-
der zweiten Ktesiphon-Expedition, pp. 23-4' On the search (see p. 527, n. I), Prof. KUhne! now tends to
• other hand, SCHMIDT, op. ci t., pp. 455-9, believes that modify his earlier view.-Ed.
he can show that the Tepe l:Ii~ar stucco is early z HERZFE LD, Am Tor von Asien, p. 78.
Slsanian (see p. 579, n. 2). 1 The development of the acanthus out of the

528
 SASANIAN ARCHITECTU RE. A. HISTOR Y
fan of acanthus leaves combined with floral elements constitutes also the tree or bouquet
motifs that decorate the Taq-i -Biistan capitals (Pis. 153 B, 168 A). This acanthus ornament
seems not to have been destined to any further development, at least not in architectural
decoration, while the repertoire of the stucco modellers, especially the patterns of a
marked conventional character, took a conspicuous place in early Islamic decoration.
The stucco fragments are all remnants of the surface covering of the various buildings,
but these buildings can themselves be only tentatively reconstructed according to the
excavated ground plans; the bits of stucco can be only partly fitted together; and their
architectural use can be conclusively shown by the condition of the finds in only a few
cases. Hence though reconstructions may come quite close to the fact, the question of
just how the stucco was applied cannot be finally answered.
A more or less definite idea of its original distribution has, however, been obtained
in one villa excavated at Ctesiphon. The main part of the building consisted of three
ivans around a court, but evidently only the two facing ivans were decorated.' On these
two sides the walls, including the interior walls of the ivan, were covered up to the
springing of the ivan arch with square stucco plaques divided into rectangular panels
with borders. Along the top of this dado, a continuation of the Ancient Eastern
dado, was a torus which was repeated on the archivolt. Above this moulding the
wall was fini shed Witll smooth plaster, the only plastic ornament being the torus
around the arch, and perhaps figural reliefs in the spandrels. This combination of a
smooth surface and areas of relief ornament avoided excessive decoration, and the contrast
enhanced the effect of the designs. The back of the ivan niche seems to have been
decorated up to the impost line, the pattern being the royal symbol of the wings. This
part of the ivan, which was the principal room, was thereby probably designated as the
place of honour, and presumably the field of the arch above had a corresponding
decoration. But that this was covered, as has been suggested, with the usual square
stucco plaques seems doubtful, for the contour of the arch would have cut into the
repeating unit of the design awkwardly. The ivan barrel vault was certainly covered
with the plaques showing boars or bears running through reeds or bushes (PI. 177 A, c),
an arrangement reminiscent of the Taq-i-Biistan decoration, for the side walls of tlle
ivan there are covered with the famous hunting relief (Pis. 163-6). In another villa
the ivan arch was clearly decorated with the coffered pattern composed of octagons and
squares (PI. 172 B) .
Taq-i-Biistan, indeed, is the only Sasanian monument that still has all its
ornament in place. It stands to actual structural architecture in much the same
relation as the Achaemenid rock tombs stand to the palaces of Persepolis, for it is
only a miniature representation of a palace created as a setting for commemorative
sculpture; but its scheme should probably be followed in reconstructing tlle palaces
where stucco ornament has been found. In the spandrels of each fa~ade ivan arch
anthemion is demonstrated by A. RIEGL, Stilfragen, I KtiHNEL, op. cit., pp. 16-1 8, Fig. 8; J. H.
Berlin, 1893, pp. 208-33. SCHMIDT, op. cit., pp. 9-10, Fig. 8.

3Y 52 9
 OSCAR REUTHER
there would then be relief figures, corresponding to the flying Victories, and figural
reliefs would also be applied to the back wall of the ivan.
A point of difference between Taq-i-Biistan and the stucco facades, as they must
be reconstructed, is the acanthus trees which make the rectangular panels on either
side of the ivan arch resemble pilasters supporting it. Yet the trees could hardly have
played this role in the designer's mind, for he was evidently interested only in
decoration and symbolism, and it was these that led him to select this place for the
'cosmological tree'.
The Sasanian decorative style is wholly unstructural. The stucco patterns covered the
surfaces solely to enrich them, like a textile pattern, a tendency also evident in early
Islamic ornament. Even the shafts of columns were sheathed with decoration and so
robbed of their structural character. There were decorated shafts at Qal'a-i-hazar-dar,'
and the round columns of the triple ivan at Damghan, completely covered with lattices
and imbricates, are evidence of the Sasanian architect's desire to clothe his supports
with applied ornament, a perverse preference from the architectonic point of view. The
shafts of the columns carrying the arcade on a silver dish in tlle Hermitage Museum
(PI. 221 c) are covered with a foliate design. Similarly, the columns of the western
maq!iira of the Ulu ]ami' in Diyar-Bakr 'Amida, which come from a sixth century
building, are sheathed in ornament,' and the patterns, for example a swastika meander
set with rosettes, belong to the same repertoire as the late Sasanian ornament. Again,
the column shafts in Coptic Egyptian buildings are similarly decorated, one of several
indications of Sasanian influence in the Christian architecture of the Nile Valley. But
while ornamented shafts seem to have been characteristic of Sasanian architecture, regular
flutings were sometimes worked in the stucco coatings (see pp. 586-7 and Fig. 170) ,
or in other cases they were left quite smooth.'
The only other sure indications of the distribution of ornament in Sasanian architec-
ture are found on the two metal plates with buildings engraved on them. The castle
on the Hermitage silver plate (Fig. 149) has, below, a blind arcade with narrow arches,
which resembles the Mesopotamian moulding and channel decoration even more strikingly
than do the sides of Firiizabad, and above this is a succession of horizontal zones of
ornament, some of which carry patterns that appear in stucco borders, for example, a
row of half-palmettes laid flat. The top zone might be interpreted as the diagonal brick
dental, represented as if seen from above, but a frieze of rectangles laid diagonally beside
one another also appears on early Islamic buildings in both Persia and Syria,' and the
motif was doubtless generally current. The decoration of the building on the Berlin
bronze salver at one point is comparable to that at Taq-i-Biistan, for flanking
t DE M ORGAN, op. cit" p. 372, PI. LXII. Fig. 208.
:2 J. STRZYGOWSKI-M. VAN B ERC H EM, Amida, . . . . . . . .
Heidelberg, 1910, pp. 157-62, Fig. 78. The columns represented on the Cairo bronze ewer
1 Smoothly finished shafts were found in the Ctesi- (PI. 2+6 B) have diagonal flutes.-Ed.
phon excavattons, and the shafts of the corner columns .. B ELL, Palace and mosque at UkhaiQir, PI. 79,
on the PaikiilI monument are also smooth: see HERZ- Fig. 2.
FELD, Paikuli, Figs. 2, 6 i and also DE MORGAN, op. cit.,

53 0

/ .
\
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
the ogee arched door are tree motifs. But the radial arcade that frames the roundel
is more important, for it shows the decoration of a blind arcade.
None of the stucco fragments so far
found seem to have belonged to a
fa~ade like that of the Taq-i-Kisrii,
with courses of blind arcades. The
only other comparable design hitherto
known has been that of the citadel gate-
way at ~mman, which was formerly
called Sasanian and related to the Taq -i-
Kisra. Now it is generally accepted as
U mayyad, but many features are purely
Sasanian.' The four ivan fa~ades
sUrtounding the small court are treated
with horizontal courses of blind arcades
composed of horseshoe arches carried
on half-columns, without either bases
or capitals. The ornament, carved in
very low relief on the stone, looks as
if it had been copied from Sasanian
stucco decoration, many of the patterns
being directly parallel to motifs found
in the sasanian stucco. A tree motif,
a vine, Of an all-over pattern covers the
back of each niche. Most of the spand-
FIG. 149· Detail from silver salver, Hermitage Museum
reIs are fill ed wi th h al f -palmettes treate d (see PI. '33 B and Fig. (41 ). showing elevation offortress.
like wings such as appear on the Berlin
salver flanking the portal, supporting the entire building, and in the radial arcade
t SPJERS, op. cit., p. 86; DIEULAFOY, op. cit., the Baukunst dec Armenier und Europa, Vienna,
pp. 101-4, had already compared the decoration of the 19 I 8, pp. 46 7-9, Strzygowski connected 'Amman
gate of 'Amman with the fa~ade of the Taq-i-Kisra with Armenian churches like the Church of St. John
and recognized the Perso-Sasanian characteristics, in Mastara (ibid., Fig. 59), and suggested that the
notably of the dentated moulding, the zi~ag archi- central square was domed The discovery of the
volts, the engaged columns without capitals In the blind Parthian palace at Ashur shows that the four-Ivan plan
arcade, and the half-domes on squinches on the Ivans. did not originate in the madrasa, but was a very early
Because of the pointed and horseshoe arches, however, elaboration of the ordinary IVan building, and the Ivans
he assumed that it must date from the end of the at 'Amman have nothing to do with the Armenian
Sl1sanian period, or more probably from after the Hijra, apse ; nor would there be any connexion between the
though he did not attribute it to the Umayyad period 'Ammltn gate and Armenian churches even if the
because it had nothing in common with known central square were domed, which seems improbable,
Umayyad buildings in Syria. His views about the date judging from the condition of the ruins. The identifica-
were not generally accepted. STRZYGOWSKI, Mshatta, tion of 'Ammlin as Umayyad is now almost universally
II, ]ahr/Juc}: der Koniglich Preujischtn Kunstsammlun$tn, taken for granted, for example by DIEZ, Die Kunst der
xxv (190+), p. 352, considered that the building might islamischen Volker, p. 28; while Creswell (in a letter)
be as late as the twelfth or thirteenth century, on the hoJds that the building, with another still more damaged
ground that the madrasa with the four-Ivan plan first construction that belongs with it, is Lakhmid.
came into Syria from the East at this time. Later, in

53 1
 OSCAR REUTHER
spandrels. Moreover, the arcade design on the salver finds its counterpart on the
'Amman fa<;ade.
Similar ornament was probably used in the niches and spandrels of the Taq-i-Kisra,
and the large engaged columns at least, which presumably lacked both base and capital,
may well have been sheathed in ornament. How they terminated at the top, however,
is difficult to say. The archivolts of the doors and of both the large and small niches
should be reconstructed as flat mouldings, such as appear on the horseshoe arches found
at Ctesiphon, and also, in various forms, at Qa!1'-i-Shirin.' The horizontal bands corre-
sponding to entablatures had, in all probability, no mouldings, or at the most very flat
ones. The brickwork here is absolutely smooth, with no projecting courses that could
have carried heavy stucco mouldings. Certainly a dripstone or cyma is out of the
question. The fact that the fa<;ade is built up in receding steps also constitutes a decisive
argument against the theory that the mouldings followed a Hellenistic scheme, for the
horizontal bands were recessed in relation to the columns. Most probably they were
flat strips of ornament divided into zones, perhaps remotely related to a Classical entabla-
ture. The frontal arch of the great ivan also clearly lacked a projecting brick course
that could have carried archivolt mouldings with a profile in proportion to the huge
scale. If there was a diadem moulding, it must have been in very low relief and been
connected with the round-arched arcade moulding. But the flat spaces above the impost
zone on either side of the arches would not be large enough for the flutterin g ribbons.
Owing to the intense light which brought out the relief, stucco ornament was most
effective on fa<;ades and in the open ivans, and since the strong chiaroscuro was in itself
colouristic, polychromy was evidently eschewed.' Most of the stucco fragments found
were covered only with a thin ivory white wash or a coating of finer plaster. In interiors
stucco seems to have been used only to a limited extent, but here it was, at least sometimes,
coloured. Fragments showing traces of paint have been found at Damghan and in tl,e
church excavated at Ctesiphon,' where they had decorated the sanctuary. In addition to
the white ground there are indications of blue, red, and green, and certain parts were picked
out with gold. There is reason to believe that the stucco fragments in tl,e Staatliche
Museen, Berlin, which came from Ni?;amabad, were originally all richly polychromed.
Mural paintings were evidently used in especially important rooms. Ammianus Mar-
cellinus describes a summer palace near Ctesiphon as having hunting scenes painted on
the walls.' Mural painting was certainly practised in the Parthian period, even if the
wall paintings in Dura-Europos be not accepted as Parthian (see Chapter 41), for
remains of murals were found in Ashur, and in the ruins of Kuh-i-Khwaja extensive
paintings are preserved which must be attributed to the first century A.D. S Fragments
of frescoes were recovered on the site of the baths in Ctesiphon, and other bits of
I BELL, op. cit., p. 51, PI. 66, Figs. I, ~. 19 17, p. 28.
% But see pp. 590, 673.-Ed. S HERZFELD, Archaeological history of Iran,
1 R EUTHER, Die Ausgrabungen der Deutschen pp. 67-74. For a discussion of the iconography of
Ktesiphon-Expedition im Winter 1928-9, p. 12. these see pp. 875-8.
4 M. STRECK, Se1eucia und Ctesiphon, Leipzig,

53 2
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
paintings were found in a villa in the neighbouring ruins at al-Ma'ari<;l, where a quantity
of stucco was also excavated.' A whole room, including the surface of the vault, seems
to have been painted here, with figural scenes involving groups of richly clad personages .
The bright colours, red, rust, rose, ochre, yellow, yellow-green, and blue, were laid on
a thin engobe of plaster that covered the unfired brick walls. The outiines were drawn
in black, and in places the design was enriched with gold.
Glass mosaic was used instead of painting in the Taq-i-Kisra at Ctesiphon. Small
fragments of this mosaic have been found, but they give no indication of the nature of
the designs.' We know, however, from reports of Arab historians, that Khusraw himself
was depicted on the wall of the tilrone room, mounted on a yellow horse, as he appeared
at the siege of Antioch. 3 The mosaics must have covered the upper part of the walls and
the vault, as was usual in Byzantine decoration, the lower part of tile walls being faced
with plaques of coloured marble.
Quite a number of fragments of polished marble plaques came to light in the palace
precincts. Both the material and the technique for tilis wall treatment as well as the glass
mosaic, were in all probability brought from Byzantine Syria,' which was so readily
accessible, and plaques of coloured glass and motiler of pearl were also used. Mas' udi
states that Khusraw Anushirvan took as booty from Syria into 'Iraq marble, various
kinds of fusayjisa, and coloured stones, and had the buildings in the city of Rumiya
that he founded decorated with mosaics such as he had seen in Antioch and otiler Syrian
cities. Barhebraeus recounts that Khusraw II, before tile invasion of Heraclius, had all
the marble taken out of the Christian churches in the territory which was in danger
and had it removed to Ctesiphon.' Both citations give a just measure of the high value
placed on such materials. When the Arabs systematically plundered the palace they
stripped off the glass mosaics and the marble revetments, to use the material again in the
houses of their new lords, so that only a few bits of refuse have been found. The
floors in the rooms of state were made of thick marble plaques, but these also had
been entirely removed.
PALACE PLANS
That the principal Sasanian ruins still standing are palaces, and that Zoroastrian reli-
gious buildings that can be identified are relatively few and unimportant, is doubtiess due
to the fact that the Islamic conquerors had less motive for destroying the former. The
palaces had, as it were, a neutral status, and the Arabs even regarded them with a certain
sentiment, which did not, to say the least, detract from their chance of survival. The
royal palace at Ctesiphon, which they looked at with awe as one of the wonders of
the world, served as a measure of the greatness of Islam' s victory, and their poets
I KtlHNEL, op. cit., p. 2 S; J. H. SCHMIDT, op. cit., tion; fazwiq means 'mercury',
p. 18. 4 KOHNEL, op. cit., p. 26.
, R EUTHER, op. Cit.,
. p. 22 s STRECK, op. cit., pp. 3 I, 33, citing C. B ARB IER DE
3 SARRE-HERZFELD, Archaologische Reise im Eu- M EYNAR D-PAVET DE COURTEILLE (Trans.), Ma ~oud i,
phrat- und Tigrisgebiet, II, p. 70, cite Qazvinl, who Les Prairies d'or, II, Paris, 186 5, p. 199; and Chronicon
speaks of laz.wj~ in order to indicate the type of decora- syriacum des Barhebraeus, Beirut, 1890, p. 156, 2.

533
 OSCAR REUTHER
used the majestic building as a symbol for the transitoriness of earthly power. It
was only considerably later that the 'Abbasids decided to raze it, and even then there
were dissenting voices in their own entourage.' On the other hand, most of the Sasa-
nian buildings that have been excavated have also been identified as palaces, which
would seem to indicate that palaces were by far the most important commissions
of the architects of this period, and absorbed almost all the architectural skill. In
this respect Sasanian architecture is akin to both Neo-Babylonian and Achaemenid
architecture, for in Babylon, and Persepolis too, technical and artistic capacity were
almost entirely concentrated on making the ruler's official residence the symbol of his
authority.
In Ardashir's palace in Firlizabad, the oldest of the Sasanian buildings, this point of
view is apparent not only in the scale- the ground plan (Fig. 150) covers a rectangle of
180 feet (55 metres) by 341 feet (104 metres)-and in the massive construction, with
walls more than 13 feet (4 metres) thick, but also in the grandiose character of the
whole design.' The great barrel-vaulted ivan, together with the rooms in correspond-
ing pairs on either side, forms a symmetrical spatial unit, the axis of which controlled
not only the whole building, but also the immediate setting, which must be conceived
of as some sort of garden. This is indicated by the great circular pool in front of the
ivan, which lies on the axis of the building. The marked symmetry of the whole
layout expresses the conception that everything should be subordinated to one dominant
principle.
In ground plan and in external design a dual division of the building is very marked.
The great ivan, with the side rooms flanking it, and the series of three square domed
rooms lying behind it, together constitute a front edifice, which clearly served for court
functions, the chief features being the ivan, which formed the sole entrance, and the
middle domed room on the same axis.
This plan of an ivan and a domed square room in communication, which appears for
the first time at Firlizabiid, became just as important in eastern Islamic architecture as the
I SARRE- HERZFELD, op. cit., H, p. 63 , n. I, have :z FLANDIN-COSTE, op. cit., PIs. 38-42. The
collated the references to the destruction of the Palace ground plan (Fig . 150) is based on Coste's plan) which
of C'esiphon by ai-Mansur (754-75 (13 7-60 H.)). has the dimensions on it and seems more dependable
There are four varying versions; ( I) that it was done by than Dieulafoy's (op . cit.) PI. XIII). Mr. Robert
M a n ~u r and an anonymous P ersian Secretary of State Byron's photographs also were used. These show)
(G. SALMON (Trans.), AL-KHATlB AL-BAG H DADf, L'in- among other things) that the niches flanking the court
traduction topographique a l'histoire de Baghdadh, ivans are not of equal height) as Coste's elevation draw-
Paris, 1904, p. 92 (translation, p. 180»; (2) that it was ing represents them. The model (PI. 147 c) D)) which
done by Man~ur and his Vaztr, Kh1i.lid al-BarmakI was made by H. Schulz) is based on the ground plan
(F. W OSTENFELD (Ed.),Jacut's Geographisches W~rter­ and Mr. Byron's photographs. Herzfeld (Reise-
buch (i.e. YAQUT'S Mu'jam al-Buldan), Leipzig, 1866, bericht) Zeitschrift der Deulschen Morgenldndischen Ge-
1, p. 425); (3) that it was Man~ur and Sulayman ibn se/lschaft) N.r. v (1926), p. 252) remeasured the ruins
Khalid (G. LESTRANGE (Ed.)) 1;IAMO-ALLAH MusTAwrl and so found that Dieulafoy's ground plan was taken
of QazwIn, The geographical part of the Nuzhat al- from Coste's and repeats the same slight inaccuracies.
Quliib) Leyden) 19 19) p. 45 (translation)) ; (4) that it Herzfeld's results have not yet been published. Ac-
was Hariin ar-RashId and Yaf)ya ibn Khalid (C. BARBIER cording to Mr. Byron's photographs the front part of
DE M EYNARO-PAVET DE COURTEILLE) Ma~oudi) Prairies the palace is largely destroyed today) only a part of the
d'or) II) p. 187 ; and C. J. TORNBERG (Ed .)) IB N AL- three dome chambers remaining.
ATHiR) Kitab al-KamIl) Leyden) 1851-76) v) P.43 8).

534
 SASANIAN ARCHITECTU RE. A. HISTOR Y
ivan cross plan of the Parthian palace of Ashur, but in the Sasanian palace the role of
the ivan was modified. In Ashur and Hatra the ivan was an independent room,
the main room, indeed, of the building,
as it is still today in the ivan houses of
Syria and Persia. In Ardashir' s palace it
functioned as an entrance hall or ante-
room, which in the court ceremonial
served probably for the public receptions
usual in the Orient, and thus it corre-
sponded to the Divan-i-' amm of modern
Persian and Indian palaces. The domed
room behind was obviously the room in
which the ruler held his private audiences,
the Divan-i-khass, in later Persian idiom.
Persian Islamic religious architecture
retained this plan, the domed room
becoming the J:!aram of the mosque
(see Chapter 39 c).
This front section at Firiiziibiid has
been equated with the Apadana or
Tachara of the Achaemenid palace plan,
which had its roots in the so-called
Khi!ani.' According to this theory the
wide columnar vestibule flanked by
minor rooms became, in the vaulted con-
struction of Firiizabiid, the deep, barrel-
vaulted ivan, while the rectangular throne
room behind, with many columns,
changed into the square domed room.
But actually the Firiiziibiid plan repre-
sents, not a Sasanian variation in terms of
vault and dome building of the Achae- o,-,,,~,,~.IL.'~"~";::.O___''.O'_ _....l:lO,-_-,,,',o, -_-,"""
FIG. ISO. Firuz1ibad, Palace, g round plan.
menid palace, but rather the combination
of two types of buildings, each independently evolved: the iviin building, and the single-
roomed domed building. The chief palace at Hatra (see p. +31, Fig. 103) shows very
clearly how they came to be coupled together, for tl,ere the square room enclosed in a
corridor has been added subsequently to the south ivan. The Mesopotamian Parthian
architect roofed the square room, which, as a temple, had a comparable use, witll a barrel
I SARRE-HnZFELD, lranische Felsreliefs, pp. front part of the vaulted Sasanian palace was derived
127-8; BELL, Palace and mosque at UkhaiQir, pp. 66- from the Achaemenid palace plan, as does also, though
8. OELMANN, Hilani und Llwanhaus, Bonner Jahr- on other grounds, WACHTSMUTH, Der Raum, p. 145.
cucher, cxxvu (1922), p. 2 I 7, opposes the view that the

535
 OSCAR REUTHER
vault, because he did not know how to set a dome on a square. The symmetrical arrange-
ment on an axis of the ivan plus the square domed room and the necessary minor
rooms resembles the Achaemenid palace plan, not because the one deliberately imitated
the other, but because Persian court ceremonial remained the same.
The second unit at Firuzabad which lay behind the first was much lower, but the
two were connected and were on the
same axis. This was an ivan building
with an inner court, with two ivans
directly opposite each otl,er, each
flanked by enclosed rooms, symmetri-
cally balanced on either side. A stair
led up to the roof terrace, which was
doubtless used as part of the dwelling,
in the usual Oriental fashion.
A second palace of Ardashir' s, not
far from Firuzabad, shows the ivan
and the domed room in a separate
building. This is the castle of Qal'a-
i-dukhtar, which is set on a high
rocky peak rising above a gorge.'
The building consists of three parts
on three terraces, one above the other.
On the lowest is an outer court, and
from this a winding ramp, inside one
0,......_. _.,",
'.L. t_._._,,,,,l,O"-_"',,-_-,,l,O,-_:.2.,,~,,, of the two massive towers that con-
FIG. lSI. Sarvistan, Palace, ground plan. stitute the fa~ade of the castle itself,
leads to a second court surrounded by
ivans and vaulted rooms which corresponds to the rear building at Flriizabad. Above
this is an open terrace, witll a stone tl,rone dais. A great barrel-vaulted ivan opens off
this, with a square domed room behind it, both enclosed in a colossal round tower
which stands on the very top of the mountain, dominating the whole landscape.
In Sarvistan, also, an ivan with a square domed room behind it is tl,e decisive feature
of tl,e palace plan, and behind tl,at again is the inner court surrounded by rooms as at
Firiizabad, but in tl,is case the rigid symmetry of arrangement is lacking, the rooms
following one anotl,er in a random succession and varying in shape, showing much
more freedom of design than Ardashir' s palace (Fig. 151). Moreover, the Sarvistan
palace differs from that at Firuzabad in having a number of secondary ivans and entrances
from the outside.' The main ivan is shallower and the domed room is larger, but these
I See p. 494, note 4. closes the east side of the court, which corresponds with
Z The plans that have been published differ from each the usual Sasanian palace plan. DnULAFOY, op. cit.,
other at certain points. FLAN.DIN-COS~E) op . cit.) PI. III, has the opening of the Ivan closed with a wall .
PI. 7. 8) show an Ivan on the axiS of the wmg that en- The ground plan (Fig. 151 ) is drawn after Coste's plan,
 SASANIAN ARCHIT ECTU RE. A. HISTOR Y
variations must have been due to special requirements in the use of the building or to the
personal preference of the owner. This palace was formerly thought to have been built
for Shapiir II (310-79), but more recently it has been attributed to the fifth, or even

.' ." .
. '......

", .
" " ," ,"

. ... . . .
:. ' .", .

FIG. 152 . Diagram showing vaulting system of the Palace at Sarvistan.

sixth century.' The highly evolved technique of the vaulting is evidence of a rather
late date, as is the use of rooms of varying shapes (Fig. 152). Moreover, whereas at
Firiiz1ib1id the form of the vaults is hidden by the walls, here it is visible above the
cubic mass of the construction, and this and the fact that the building quite lacks the
archaic power and compactness of Firiiz1ib1id likewise argue for a later dating.

which gives the impression of being more accurate than showing the vaulting system (Fig. 152) this is taken into
Dieulafoy's, since the dimensions are given. In this it account.
is dear that the court is oblong, and not square, as I Herzfeld attributes the palace of Sarvistan to the
Dieulafoy shows it. On the other hand, Dieulafoy's reign of Bahram Gur (420-38), whose premier (buzurg
photograph (PI. 148 A) shows that the room on the framadhiir), Mihr Narse, built it: SARRE-HERZFELD,
left of the front Ivan was roofed with a dome and not a Iranische Felsreliefs, p. 13 I, n. 3 j idem, Archltologische
barrel vault, as Coste represents it. In the drawing Reise im Euphrat- und Tigrisgebiet, II) pp. 332-3.

3Z 537
 OSCAR REUTHER
The building excavated at Damghan also seems to represent a well advanced stage in
the evolution of Sasanian architecture, though an early dating has been claimed for it
(see pp. 582, n. 2, 642-5). The section of the ground plan that has been uncovered
shows that here, again, an ivan communicating with a domed room constituted the focu s
of the layout, and hence the identification as a palace seems correct; but the connexion
between the ivan and the domed room was much closer here than at Firiizabad or Sarvistan,
for they communicated through a wide arched opening. The ivan there was a three-aisled
hall, and the vaulting should be reconstructed like that of the two barrel-vaulted rooms at
Sarvistan. Narrow doors in the side aisles opened into a passage way, running on either
side of the domed room to the back, and side arches opened from the room into these
corridors. An arch in the back wall of the room provided a perspective and entrance
into another room, which formed the terminus of the whole series, but the excavation
did not reveal the exact character of this further room. The plan is very similar to
that of a three-nave church with a domed transept intersection.
The room excavated in the building designated as Palace II at Kish (see pp. 586-7 and
Fig. r696) has also been compared to early Syrian churches, and an apse, semicircular in
plan, projecting out of a small rectangular room on the axis of the central nave does,
indeed, show real similarity to the choir of a Christian church. If one reconstructs the
transept with a vault, one gets a central dome with two barrel-vaulted side arms and
the apse roofed with a half-dome. The apse is an un-Sasanian feature and does point
to Syria, but there it is an element, not only of churches, but also of palaces. The
closest parallel is the three-aisled hall with the triconchos at Mshatta, which was to have
been. the audience hall or throne room of the Caliph.' It opened on the court with a
three-arched fa~ade, and this was doubtless the plan at Kish also. Moreover, at Kish
there was a forecourt with engaged columns on the side walls, and this, togetller with the
three-aisled hall and the transept with the apse, formed, as in Mshatta, a central unit
which was surrounded by the subordinate rooms and corridors, the whole layout form-
ing a rectangle.
Palace I at Kish is a standard ivan building of the same type as the rear building at
Firiizabad, with two ivans, one larger than the other, confronted (Fig. r69 a). The arrange-
ment of the subordinate rooms is so similar in the two Kish buildings that their function
must have been identical, and both likewise have two small round pools symmetrically
placed in front of the ivan. Moreover, botll show the same curious type of entrance,
I The three-aisled hall with the triconchos at pp. 104-46) that Mshatta dates from the Umayyad
Mshatta was once thought to bea Christian church, and period and was built by ValId II. Creswell goes into
the castle a monastery built by a Byzantine emperor the problem of the triconchos (pp. 382-6) and shows
(5 . MERILL, Eastef the Jordan, 188 I, pr.:2. 56-6 3). The 'that the trefoil plan was first used In the hot chambers of
dating of Mshatta was long in dispute. CRESWELL, provincial Roman baths, and at Rome for mausolea;
Early Muslim architecture, pp. 390-405, summarizes that the adoption of this feature for a throne room or
the complicated arguments. He agrees with the thesis audience hall first took place in Syria; and that the
defended by Herzfeld (Genesis der islamischen Kunst further development seen at M shatta, in which the
und das Mshatta-Problem, Der Islam, I (1910), triple-apsed throne room is preceded by a basilical hall,
pp. 122-43; idem, Mshatta, I:Ilra und Badiya, Jahr- was due to Egyptian influence' (p. 386).
buch der PreujJischen KunJlSammlungen, XLII (192 I),
 SASANIAN ARCHITECTU RE. A. HISTOR Y
indirect passages instead of a portal on the main axis. Similar narrow corridors are found
in the Parthian palace at Ashur (see Fig. 106), and circuitous entrances which prevented
passers-by from seeing in had been used before this in ancient Babylonia.' Interrupted
narrow passages of this same type commonly lead to the courtyard in modern Islamic
domestic architecture (see Chapter 30 L). Obviously, then, both the Kish palaces served the
same end, and since Palace I was clearly the dwelling of a person of rank, that must likewise
have been the status of Palace II. But the arrangement of the entrance and the lack of the
ante-chamber unit found, for instance, at Firiizabad show that this was a purely private
residence. The three-aisled main room of Palace II is the equivalent of the simple
ivan of Palace I, and therefore it must have been vaulted as the corresponding room at
Damghan doubtless was, although the columns or round piers built up of bricks and
covered with stucco in which flutings were modelled are more slender than at Damghan
and proportionately farther apart.
The Sasanian builders may have been encouraged to use the three-aisled hall, in place
of the ivan, by the example of the Syrian basilica, but the Sasanian plan had had quite a
different history. The basilica was divided into three units by setting two rows of
columns in a large hall. The Sasanian piers had been by degrees moved inwards from the
wall that carried the vault. Thus the resemblance between the Syrian and the Sasanian
plan is merely fortuitou s.
The 'Imarat-i-Khusraw or I;Iajji Qal'asy is the only one of the Qa!r-i-Shirin ruins
that can be confidently identified as a palace, that is, a building planned for the residence
and official ceremonies of a prince, and its modern name suggests that it was built
for Khusraw II Parviz (590-628).' In general design it is quite similar to another
palace, the Hawsh-Kuri, which must be of about the same date, and both are distinctly
different in layout from the type of palace represented by the buildings at Firiizabad
and Sarvistan.' The main buildings of the 'Imarat-i-Khusraw and the Hawsh-Kuri
both stand on rectangular terraces, reached by double ramps. Terraces of this type
existed at Ctesiphon also, for tl,e excavations uncovered in the mound of ruins south of
the town, known as Tell Dhallab, a square terrace with double ramps ; and a second
terrace, which very probably carried part of the royal palace, lies directly south of
the Taq-i-Kisra. This is today called the I;Iarim-i-Kisra, a name which may
perpetuate the memory of its original identity. These terraces approached by ramps
recall PersepoIis, and evidently the principle of building on a terrace was continued
from Achaemenid times, even though no examples of tl,e Parthian or early Sasanian
periods are known.'
J REUTHER, Die Innenstadt von Babylon (Merkes), those points where it is evident that the ruins were in
pp. 87-8. ' See p. 495, n. 6. better condition than when Miss Bell saw them. The
3 The ground plan and reconstruction of the three-Ivan fa~ad e is adopted because this form appeared
Palace of Khusraw in Qa~r-i-Shlrln (Figs. 153, 154) are also in the Kish and Damghan excavations .
based primarily on the plans made by Miss Bell (op. • The temple of Kangavar, since it is a cult build-
cit., Pis. 53, 54), which make a more convincing Im- ing, is not relevant.
pression than those of de M organ (op. cit., Pis. XLII, . . .
XLVI); but the latter have been taken into account at At Kangavar there was 'a great arched building

539
 OSCAR REUTHER
The whole layout of the 'Imarat-i-Khusraw is gigantic, the building, which stands in
walled precincts, the King's Paradise Park, being 1,220 feet (372 metres) long and 623
feet (190 metres) wide (Fig. 153). The palace is accurately oriented, the west-east axis
being marked by a narrow water channel 1,805 feet (550 metres) long, while the entrance
gate of the park is about 354 feet (c. ro8 metres) farther to the east on the same axis.
The terrace, 935 feet (285 metres) long, 321 feet (98 metres) wide, and about 25 feet
(c. 8 metres) high, was in part carried at the edges on vaults, and was approached by a
double ramp on the main axis, and two other ramps symmetrically on the cross axis.
Above this rose the palace itself, with its symmetrical ground plan, including the state
rooms and the king's private apartments (Fig. 154).
The two plans that have been published are ambiguous at a number of points. It is
certain that a deep porch or anteroom on the axis lay in front of the main building,
projecting from it into the great plaza at the east end of the terrace. In the older plan
this anteroom is outlined Witll a single row of columns on either side. The subsequent
investigator who made the other plan saw no trace of these columns, but found piles of
rubbish, leading to the supposition that there had been heavy side walls, carrying a
barrel vault. Evidently both the columns and the walls had once existed, and the ante-
room was a triple ivan hall like those at Damghan and Kish (see Figs. 167a, 170).
If this antechamber corresponds to the ivans of the Fars palaces, it must have com-
municated with a square domed throne hall. Both the plans show a rectangular room
with its main axis at right angles to the axis of the anteroom, but here again there is a
discrepancy. The more recent plan justifies the assumption that it was barrel vaulted,
tlle other, made when the ruin was in better condition, would indicate a central dome
carried on transverse arches with a barrel-vaulted wing on each side. This would corre-
spond to tlle Ivan-i-Karkha, and similar rooms appear in Qa~r-i-Shirin and Hawsh-
Kuri as gate buildings.
From this room one passed, through a door on the main axis, into a large square
unit, 88t feet (27 metres) on a side, surrounded by a columned corridor. Only the
corner pillars of this have been distinguished in the rubbish heaps, and these are shown
as heart- shaped in section. It has been assumed that this space was roofed, and tlle
suggestion has even been advanced that it was domed.' But the slender masonry columns
of the corridors and the relatively weak enclosing walls of the area, both built in the
crude cobble and gypsum rubble, could never have carried a dome with a diameter of
61 feet (18·6 metres). The single domed building of the neighbouring Qal'a-i-Chahar

standing on a platform, and dating from the days of Yiaqiit, III, 50, 169, IV, 120,381). Again Yaqut tells
Khusraw Parvlz., being constructed with columns and of a great arched building surmounting a platform at
of mortared brickwork . . . . Yakut asserts that the Mldharastan on the Khuri\si\n road, originally part of a
platform, where the Sassanian building stood, was 20 palace belonging to Bahram Gur, and set in the middle
ells above the ground level, and Mustawfi adds that the of a Paradise Park (ibid., p. 19 I ).-Ed.
great stones for its construction had been brought from I DE MORGAN, op. cit., p. 344; BELL, op. cit.,
the mountain of Bisutt1n', (LE STRANGE, Lands of the p. 45, designates this space as the 'audience hall', and
Eastern Caliphate, pp. 188-9, citing I ~~akhrt, 196; therefore also assumed that it was roofed.
Ibn l:Iawqal, '2 S6; Ibn Rusta, 167. Muqaddast, 393;

540
 v, v,

u. u.

c
"
,i·

F ,

i, ,
", .'

+
, .' " ,' ,
C '~(

"I ·1

FlO. 153. Qa~r-i-ShtrIn, 'Imllrat-i-Khusraw: ground plan.

 OSCAR REUTHER
Qapii shows how domes were built at Qa~r-i-Shirin, and the great amount of counter-
weight deemed necessary. The large square area of the 'Imiirat-i-Khusraw must, there-
fore, have been open, that is to say, it was a court with a peristyle of columnar arcades
and an ivan opening on the main axis.

..
. . .:: :
, '

. ..' ..' , .
"
', ,'

",
.
'.'
."
.' .. :. ,
. ' . .. '
... .
. :. , . '

• •, n'

. .;
. '. . , ', .

", :

" " .

.. '.

. ' ... , .. . : .. : ',:

'

. .. .. '.. . . .
" . : ..'..: ..' .. ..
... . . . .' .
' ."
', ':: ': , ... .' : :: . .
. . . . . ' ..' . .. . .
'

. .. ..;
" "
'. "
. . .. '

.... . "
" .
'
"'.: : ', -
.. .
. ' .. ' . .
'

. ..' ., . . :' .,: .

'. .
. .. . ', '
' .. . .
. . . . ... . .
. ' ... .
...
. ' .. ,'

FIG. 154. Qa ~r-i- Shlrin) 'Imarat-i-Khusraw: proposed reconstruction of exterior.

Thus the whole plan would coincide with the usual layout as seen at Firiizabad and
Sarvistiin: (I) a public reception room in the form of an ivan opening outwards ;
(2) an enclosed vaulted room as a private audience hall or Diviin-i-khiiss; (3) a court
with an ivan on the main axis.
On either side of the small ivan on the court were groups of rooms symmetrically
542
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
arranged, and out of these opened two ivans, facing on the opposite side, on a court
that lay beyond to the west. This court is divided, in the more recent plan, by a wall
cutting across it, with a portal on the axis, and on each side is a passage connecting a
series of separate buildings which could be entered only from the passages. Each of these
has its own interior court, ivans, and rooms, and some have also a kitchen and small
service court. To the west, this row of houses was continued on a lower level, surround-
ing the end of the terrace, and on the south the terrace widened out asymmetrically
into two arcaded courts connected by a vaulted gate- house, with buildings on the sides,
while on the north side were several building units combined in a group.
The houses on the terrace at the side of the main building may have been the tarim,
while those farther back would have supplied accommodation for attendants. The arcaded
courts might have served for the animal baiting, so popular with Sasanian monarchs.
The narrow court, more than 325 feet (100 metres) long, into which one could look
from the palace terraces, would have been especially well adapted to this sport.
The layout of the Hawsh-Kurl' is, in principle, very similar to that of the ' Imarat-
i-Khusraw. Only the gross feature s of the ground plan of this have, however,
been recorded, so that nothing but the major sequence can be traced: a three-aisled
front ivan; a rectangular room, the main axis at right angles to the axis of the ivan,
possibly with a central dome; and a court with smallivans on the axis. The dwelling
houses, likewise ivan buildings with inner courts, were side by side on the west end
of the oblong rectangular plan. Here again the layout, though on a much smaller scale
than that of the 'Imarat-i-Khusraw, is symmetrical and unified,' and all the parts are
arranged on an axis, as were also the palace precincts. The main axis, on either side of
which lay large courts, likewise symmetrically arranged, was continued to the east in an
avenue 575 feet (175 metres) long, with a small building like a kiosk at the beginning.
The palace fa~ade defines the transverse axis, and this was continued towards the north
by a straight avenue that led to the gate of the huge, almost square park. In the south-
west corner of this park was a building with three courts which may have contained the
stables and accommodations for hunting animals.
The layout of the Taq -i -Kisra, the largest in scale of all the Sasanian royal residences,
followed a scheme, revealed only by excavations (Fig. 155), quite unlike the usual formula
exemplified by the palaces at Firiizabad, Sarvist1in, and Damghan, and equally distinct
from the terrace palace type represented at Qa~r- i -Shirin and the Hawsh - KurL' From
the ivan, which opens towards the east, one went through a door on the axis in the back
wall into a series of smaller connecting rooms, and from these into a very large enclosed
room, equal in width to the ivan, which also must have been barrel vaulted. On either
side of the ivan and divided from it by a corridor apparently vaulted like the Ivan-i-
Karkha was a room, large, but smaller than that on the main axis. If the part of the
I DE MOR.GAN, op. cit., PI. LI. phon, also, ought to be reconstructed with a domed
Z REUTHER, Die Ausgrabungen cler Deutschen room behind the great Ivan: SARRE-HERZFELD,
Ktesiphon-Expedition, p. IS and Fig. 10. Surface Archaologische Reise, p. 62 and Fig. 16 7.
evidence seemed to indicate that the Palace of Ctesi-

543
qt'u,~,~,L~'u,~,~,'~P _______'2P~____~l~~____-"~p~ I ' - , - - --.N
£iti6tif1,9 ~CDvated
• Walls _Wafts
FIG. ISS. Ctesiphon, Taq-i-Kisra (Ivan-i-Khusraw): ground plan based on results of the
excavations of 1928-9.

544
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
plan uncovered in the excavations be symmetrically completed, a rectangular building
results, which includes two successive series of rooms on a central axis, each consisting
of three large barrel-vaulted rooms. The middle rooms dominated the whole plan
because of their greater breadth and height, that in the front series being open as an
ivan, and these were divided from the side rooms by corridors. A zone of smaller
apartments ran the full width of the building on the transverse axis, serving both to
separate and to connect the front and back series of large rooms. At the ends of this
row of smaller rooms were square chambers, which may have been domed, though just
because the rooms were square we cannot conclude with certainty that that was the case.
If, however, domes were used at Ctesiphon, they did not play the dominant part that
they did at Firuzabad, Sarvistan, or, judging from all the indications, Damghan, at
least as far as we can determine from the half of the palace which is all that is as yet .
known. Of the other half which stood opposite we can only say with confidence that
it, too, had an ivan opening on to the court that lay between the two fa~ades .
The use of the palace and of its various parts can be only tentatively suggested, save
where the buildings themselves provide indications or there is documentary evidence of
probable relevance. The great ceremonial audiences of the Sasanian royal court at
Ctesiphon, of which Arab historians speak, were almost certainly held in the big ivan
and the court in front of it. The throne of the Khusraws must have stood here, with
the crown suspended from the ceiling, as Tabari recounts, and the costly curtain, which,
according to al-Kha\ib aI-Baghdadi, the soldiers of Sa'd ibn Abi'l- Waqqa! burnt
when the city was sacked.' The long wall-enclosed rectangle south of the Taq buildings
may have been the arena for the animal baiting or royal exhibition hunts, such as
we see on the reliefs of Taq-i-Biistan, but this is only surmise. The dwelling
houses of the palace precincts at Ctesiphon were probably on the terraces which later
took the place of this arena,' though this, again, is only a presumption. The private
apartments could not, however, have been in the Taq itself, for its gigantic scale made
it unsuited to domestic use.
BATHS
A building excavated to the west of the Taq building was certainly the baths.J
Masonry water channels and conduits made of clay pipe, the remains of a warming
oven, a well, and a masonry platform for resting, all point to the purpose of the build-
ing, and since indications were found that it was decorated with murals, mosaics, and
marble revetments, it is evident that the baths played an important part in Sasanian court
life, though until these ruins were uncovered we had not known of the existence of
baths of this character prior to the U mayyad period. Thus the Islamic ~ummam was
not developed solely from the Roman thermae.
Palace III at Kish, an ivan building, has been tentatively identified as a swimming
I T. N OLDEKE, Geschichte der Perser und Araber cit., p. 60.
zur Zeit der Sasaniden, Aus der arabischen Chronik z R EUT HER, op. cit., p. 24.
des TabarT, Leyden, J879, pp. 221-2; S TREC K, op. 3 K OU NEL, op. cit., p. 12.

4A 545
 OSCAR REUTHER
bath,' first because a rectangular water tank takes the place of the interior court; second,
because this is surrounded by rooms, also in the form of ivans, a very unusual feature;
and third, because it has a very narrow entrance. The arrangement of the entrance,
however, is the same as in Palaces I and II, both certainly residences, while the water
tank, the ~aw'!l, is an unfailing feature of the court of a Persian house even today, just
as the ba~ra is of a Syrian house, and the tanks of some Islamic palaces in India fill
the entire inner court.' These, however, are neither baths nor swimming tanks. The
expanse of water in hot summer weather serves to soften the air, and the sight of the
water gives a pleasant illusion of coolness and freshness. The Siisanians preferred to
build their palaces beside springs or water courses, or else they made careful provision,
as at Qa!r-i-Shirin, for an ample water supply. In Ardashir' s palace in Firuziibiid
there was a spring which fed the round pool in front of the ivan,' and the description of
a palace built by Kaviidh near Asak in Khuzistan mentions a similar pool,' while
according to Mustawfi a palace at Shiz had in the court a bottomless pool or lake, fed
by springs with water that petrified objects immersed in it.' Perhaps since there were
no springs available in 'Iraq, tanks were built as substitutes, as Palace III in Kish
shows.'
FENESTRATION AND VENTILATION
Protection against summer heat is almost as important a consideration in domestic
architecture in the East as protection against winter cold is in the North. The inhabi-
tants of 'lriiq and South Persia today are metamorphosed during the daytime in summer
into temporary troglodytes, taking refuge in the sard-iib, vaulted living cellars under-
neath their lightly built houses (see Chapter 39 L). In old Babylonia the house itself was
built like a cave, closed against the burning rays of the sun with thick, windowless
walls. 7 This practice had not yet altered by the Siisanian period, and tlle thickness of the
walls of Ardashir's palace at Firuziibiid is certainly due not only to the requirements of
tlle boulder rubble construction and to the need of a counterweight for the great thrust of
the clumsy, thick-shelled vaults, but also undoubtedly to the idea that heavy walls would
provide protection from the heat. No windows admitted daylight directly into the
rooms, the doors apparently affording light. The three domed rooms must have been
I LANGDON, in Th e Illustrated L ondon News, S Ibid., p. :2.:2. 4.
February 20,1932, p. 273. 6 Notable among the S;Is!nian palaces that have
. . . . now disappeared but are mentioned by the Arab
But corrected by LANGDON, Excavations at Kish geographers is one that stood three leagues from Hama-
and Barghuthiat, I , Sassanian and Parthian remains in dan in the village of Jiihasta. It had been built by
central Mesopotamia, 'Iraq, I (1 934),(, 115, where he Bahram Gur and 'was a huge structure, with halls,
suggests that the building was a 'roya summer house', passages, and chambers, in part cut out of the live
with the pool for a cooling effect. -Ed. rock. At the four corners were sculptural female
:1 R EU THER, Indische PaHlste und Wohnhlluser, figures, and along one face of the buildtng ran an in-
Berlin, 1925, pp. 25, 82, PI. 160. scription in Old Persian (Far;;yah.) commemorating
J HERZFELD, Zei/schriJt der Deutsche1t Morgenliind- the conquests of the Chosroes'. (LE STRANGE, op. cit.,
ischen Geselbchaft,N.F" v, p. 254; idem, Archaeological p. '95. citing Ibn al-Faqlh).-Ed.
history of Iran, p. 96. 7 REUTHER, Das W ohnhaus in Bagdad, pp. 15-
4 LE STRANGE, The lands of the Eastern Caliphate, :2.0 ; idem, Die Innenstadt von Babylon, p. 90.
p. '44·
546
 SASANIAN ARCRITECTU RE. A. RISTO R Y
completely dark, unless one assumes that there was fenestration in the crown of the
dome, which would be consistent with the condition of the ruins, and is very probable
judging from the present practice in Persia, where a small opening serves to illuminate a
very large area, for example, in the bazaars (see Chapter 39 L). Ventilation was provided
by a system of masonry air channels piercing the thick walls which enclose the domed
rooms, with shaft openings between the domes, and by means of these air was driven
down into the rooms by the wind.'
At Sarvistan no trace has been found of this ventilating device, which corresponds
exactly to the 6ad-g;rs or wind-towers (see PI. 462 and Chapter 42), customary today in
domestic architecture in Persia and 'Iraq (see Chapter 39 L). Another system of
ventilation seems to have been used there. Clay pipes enclosed in the shell of the
dome are shown in one plan, while the other assumes that there were narrow slits
in the dome.' In either event, there were small openings in the vault which admitted
both light and air. A kind of restricted fenestration was apparently intended. A parallel
device, small openings in the vault filled with coloured glass, is found in the Oriental
~ummam today.' The shell of the ivan vault of the Taq-i-Kisra has clay pipes running
through it at regular intervals, but since this is an open ivan tlley could have had no
connexion with either light or ventilation. Moreover, similar pipes are found in the
lower part of the vault where it is several yards thick, and the outer ends of these are
stopped up with mortar. It has been suggested that they may have served for the sus-
pension of lamps, but this does not seem probable, for the lamps would have hung very
close together at tl,e sides of the room almost touching the walls and increasingly far
apart towards the middle. It is possible that ornaments may have been set into the holes.'
PRIVATE HOUSES
All the dwellings of the ' Imarat-i-Khusraw represent the simplest version of the ivan
plan, with either a single ivan opening into a court, or two, opposite each otller, one
serving for the summer, the other for the winter, and only one chamber on either side
of the ivan. The dwelling houses excavated at Ctesiphon, which for archaeological
reasons are attributable to the late Sasanian period,' also had the court and ivan as
central features, but some of these were large buildings with many rooms, adequate to
a more pretentious scale of living than is evident in the dwellings at 'Imarat-i-Khusraw.
Thus there were varying developments of the ivan plan, but the basic principle always
remained the same (Fig. 156a). The open ivan was at the middle of one, two, tllree, or
probably, if a typical plan (Fig. 1566) is completed, four sides of the court, making,
J REUTHER, Das Wohnhaus in Bagdad, p. 95; lighting system, which is typical of 1}.ammam domes,
DIEULAFOY, op . cit., p. 33. in examples in Cairo.
2 See FLANDIN-COSTE, op. cit., I, PIs. 28-9; , D IEULAFOY, op. Cit.,
. pp. 72-3; SPI ERS, op. Ctt.,
.
DIEULAFOY, op. cit., IV, PI. VlII. pp. 77-8; SARRE-H ERZFELD, Arc h~olog i sche Reise
3 For example the Yeni Kaplija in Brusa: WILDE, 1m Euphrat- und Tigrisgebiet, II , p. 65'
Brussa, Berlin, 1909, p. 100, Fig. 137. COSTE, Monu- 5 The ground plans (Fig. 156 a, b) are taken from
ments modernes de la Perse, Paris, 1865, gives the those given by Wachtsmuth, in Ktl'HNEL, op. cit.,
section of a bath in Kashan . FRANZ PASCHA, Baukunst Figs. 2, 3, 5, and 6.
des Islam, Darmstadt, 1896, Figs. :2 Q-23, shows this

547
 OSCAR REUTHER
in the last instance, the ivan cross plan that had already appeared in the Parthian palace
at Ashur (see p. 433), and the rooms on either side were more or less symmetrically
arranged. Though the plans of private houses were far less symmetrical than the palace
plans, they were unified by having the ivan and the rooms behind it all on one axis,
and if a square room adjoins the ivan, we have the same scheme as at Firuzabad,
Sarvistan, and other palaces.

\
N

/
-
Courlyard

a b
FIG. 156o, h. Ctesiphon, dwelling houses, 6th century, ground plans: n. Umm as-Sa'atir; b. al-Ma'ariQ.

The ivans were not all of equal importance. The largest was evidently the principal
room, while the others were sometiines only shallow niches. In some cases they were
wider in front than at the back and stepped back. Occasionally across the front was a
columnar vestibule called in 'Ir:iq today the tarma,' forming with the ivan an inverted
T (1.) in ground plan; or thetarma might run along the court, but it never developed into
a continuous peristyle. A similar vestibule opening on the court in the same relation to
the ivan is found built with masonry columns in the palace of Ukhay4ir, which dates
from the early Islamic period.'
The enclosed rooms which surrounded the rest of the court between the ivans usually
assumed one of three forms. Some are square with four supports defining a square.
This type of room is already seen in the Parthian palace in Ashur (see Fig. 106) and
before that in the subordinate rooms of the Hadish and other residential palaces at
Persepolis (Fig. 74). It recurs in the palace of Ukhay4ir,' vaulted as at Ashur. Others
I REUTHER, op. cit" pp. 2. and 8 -13. 3 REUTHER, Ocheldir, p. 37, Figs. 36 and 37.
2 REUTHER, Ocheldir, p. 41, Fig. 4- and PI. Ill.
 SASAN IAN A RCH IT E CTU RE. A. H ISTO R Y
are a long rectangle, often communicating with a square room on the axis; or, finally,
a rectangle or square, extended with one or more rectangular niches.
These niche rooms are of special interest in the history of Eastern architecture, for
the niche is a derivative of the ivan, which means that towards the end of the Sasanian
period the ivan, originally only a room opening outwards, had come to be utilized in an
enclosed room which, replacing the court, should be considered a court roofed over.
Thus in each niche room the principle of the ivan house was repeated in a compressed
form. This room with an ivan alcove became one of the most important motifs in
Islamic architectural design. In private houses in Syria and Egypt today the so-called
qii'a is this type of room.' The part equivalent to the court, called in Cairo the durqii'a,
off which open one or more alcoves, is higher than the alcoves, and thus it can be
lighted by windows in the upper part of the walls, and presumably the Sasanian niche
rooms also had this 'basilica' form of fenestration. The qii'a in modern Syrian houses
very often has three ivan arms, so that the room assumes the inverted T (.1) ground
plan, and in one of the Sasanian houses at Ctesiphon (Fig. 156 b) a three- branched room
of this type was clearly the show room of the house, communicating on the axis with
the ivan which opened on to the court. Precisely the same arrangement is found in
modern houses in Aleppo.
The fanna, which appears as an incidental element in some of the ivan houses, is
the distinguishing feature of quite another type of house, known only from a re-
construction based on traces found in the ruined city of Darra -i -Shahr.' This porch
or vestibule, with round columns built of stone, open on to the street or garden,
formed the front of a detached house with no court. In plan there is no point in common
with the houses found at Ctesiphon, and also there is no relation to the usual houses in
Persia and the adjacent countries today, save for the farina, which is a characteristic
feature of modern Persian and 'Iraqi domestic architecture. Thus the houses of Karbala
have a farina of this type opening on to the street, but more often the farina surrounds
the interior court, and slender wooden columns are used instead of the stone or
brick piers.' These slender wooden columns were derived from the early Persian style
(see p. 5), and so we must conclude that the farina house was indigenous during the
Sasanian period to western and northern Persia, as well as the central valleys, all districts
where it is today the predominant dwelling, and also that it was combined as a second
story with the ivan house, which had developed from wholly different origins.
The ruins of private houses at Darra-i-Shahr indicated a two- story plan, and the
reconstruction follows this interpretation. Probably most of the Sasanian city buildings
had several stories, and the engraving on the Leningrad silver plate indicates that the
upper stories projected and were corbelled out, as is usually the case in the Islamic
cities of the Near East today. Whether, or to what extent, the houses excavated at
Ctesiphon consisted of more than one story it is impossible to say. No traces of
I REUTHER, Die Qa'a, Jahrbuch der asialischen Z DE MORGAN, op. cit., p. 371 and Fig. 222.
KUnJt, II (1925), pp. 208-16. 3 See n. I on p. 548.

549
 OSCAR REUTHER
stairways were found, but there could have been stairs in the narrow passageways run-
ning through the ground plans like conduits to connect the more remote groups of
rooms, and if they were constructed of wood they could have completely disappeared.
Whether the palaces had upper stories is also impossible to say conclusively. At Hatra
the living rooms were in two stories on either side of the ivan (see p. 430). The smaller
rooms at Ctesiphon that lay between the great ivan and the large rooms must presumably
have had a story above them. In the north wing of the desert castle of Ukhay<;lir, which
includes the great entrance hall, three vaulted stories are still standing.' This dates
presumably from about the middle of the eighth century, but it continues both in general
conception and in detail the traditions of Sasanian palace architecture, hence we can
conclude that parts of the Sasanian palaces had more than one story, while other parts
were all on the ground level. It has been suggested that the prototype of Ukhay<;lir
was the famou s castle of Khavarnaq near I:Iira,' and this must also, like Ukhay<;lir,
have had one part that was several stories high, for in the poem of 'Abd al- ' Uzzii it is
said that Sinimmiir the architect 'built it high like a mountain witl, high peaks' , and in
another place we are told that from the top Nu'man ibn Imr' al-Qays could look out
over his flourishing land to the lake of Najaf.3 A city house might be built up in several
stories for lack of space, but that could certainly not have been the motive in the case
of a palace standing in a park or, as at Ukhay<;lir, in the open desert. The aim must
rather have been to get purer air above the sultriness ,and dust of the ground level, and
to have a fine view over the surrounding country ..

CULT BUILDINGS
The problem of the Zoroastrian cult buildings in Siisanian Persia has been until
recently a subject of lively dispute. The names of the seven great fire sanctuaries are
known, and there are records of many other fire temples having existed, but very few
of these have been located, and neither the Islamic nor the Christian accounts enable
us to form any definite idea of their appearance. Nor does an acquaintance with the
cult buildings and rituals of the Parsees in India or of the Zoroastrians still living in
Persia, the Gabri, contribute anything to a solution. Recently, however, some light
has been shed on the whole question, and it would seem that certain views, advanced
long since, are indeed correct. 5
I REUTHER, Ocheidir, pp. 15-20 and 27-9, 3 NOLDEKE, op. cit., pp. 83-5'
Pis.III, lV, and Vll; BELL, Palace and mosque at 4 This is somewhat confirmed by TabarI's statement:
Ukhaiq.ir, p. S, Pis. z, 3, 4. 'On the terrace [probably terrace roof] of this Khavarnaq
2 N~LDEKEJ Geschichte der Perser und Araber, one had the desert on one side; the air (which blew
pp. 79-80; G. R OTHSTEIN, Die D ynastie der Lahmiden from this side) is the best air in the world'. H . ZOTI N-
m al-1;lira, Berlin, 1899, p. lSi B. M EISSNER, Von BERG (Trans.), Chronique de Tabari, II, LXIX, Paris,
Babylon nach den Ruinen von I:Iira und Ijuarnak, 1869, p. I 09·- Ed.
Senduhriften der deutschen OrimfgesellschaJt, 190 I, 5 The whole problem of Persian temple building
No.2; L. MASS IGNON, Mission en Mesopotamie, was discussed by OUMANN, in Archiiologischer AlIzeiger,
Memoires pub/ib par Ie; membres de I'InJlilut Fran;ai; III/IV (1921), pp. 274-87 . Recent discoveries of Sa-
d'Archiologie Orientale au Caire, XXVIU (I 9 10), pp. 36-7, sanian temple buildings prove that his statements,
PI. 37; idem, Khawarna~, in Houtsma, Encyclopaedia of based on rather scanty facts, were on the whole rig ht.
bJom, JJ (1927), p. 1001. These discoveries were made chiefly by Herzfeld: see
 SASANIAN ARCHITECTU RE. A. HISTOR Y
The ruins of Takht-i-Sulaymiin, that crown a mountain 8,000 feet (2,500 metres)
high about a hundred miles (c. 160 km .) southwest of Lake Urmiya and a hundred
west of Zinjan were thought nearly a century ago to be the remains of the city of
Canzaca described by Cedrenus and Theophanes, and known to the Arabs as Shiz,'
and the remains of a great fire temple with a dome 29! feet (9 metres) and originally
very high, were thought to have been the temple destroyed by Heraclius, where Adhar-
gushasp, the sacred fire of Adharbayjiin, had been worshipped since mythical times.
The general ruins, which are very extensive, including nearly twenty Parthian, Sasiinian,
early and middle Islamic structures, are surrounded by a powerful masonry wall more
than 45 feet (IS metres) high, and three-quarters of a mile in circumference, strengthened
by twenty- seven bastions. Sherds prove a continuous occupation of the site since from
about the third millennium B.C. to at least the fifteenth century A.D. Although collapsed,
enough remains of the fire temple to show that it is true to plan; a dome setting on
four arches opening into a vaulted corridor, clearly visible in the aerial photographs.'
In the Girra Valley, between Firuzabad and Kazarun, stand the ruins of four domed
buildings of this type which are called, by the inhabitants of the district, 'Chahiir Taq',
or 'Four Arches'. One of these, near the village of Farrashband, has long been known
and was considered, by its discoverer, Achaemenid, like the palace of Firuzabad. 3
All four were recognized as fire temples by the scholar who found the other three, and
identified them as the temples built, according to Tabari, by Mihr Narse, Vazir of
Yazdijird I, and of Bahram Gur, in four villages near Gur that belonged to him.
Tabari gives the names of these temples, and this scholar believes that one of these names,
'Friiz-marii-iivar-khudhiiyii' (that is, 'Come unto me, Lord'), is perpetuated in the
modern geographical name Farrashband.' All four structures, two of whkh are large,
Reisebericht, Zei/Schrijt dey Deuuchen Morgen/dnd;;chen 1906, p. 13 I, fully accepts Rawlin son's identification
GesellschaJt, N.F., v, pp. :169-76; and idem, Archaeo- of Takht-i-Sulayman with the Shlz of the Arab writers
logical history of Iran, pp. 66 and 88-93. Recently the and probably also with the Ganjak and Janza of the
problem has been discussed by U. MONNERET DE PerSians, the Gazaka, or Canzoch, of the classical
VILLARD, The fire temples of Iran, Bulle/in of/he Ameri- writers. Jackson visited the ruins and believes that the
can Institute f or Iranian Art and Archatology, IV (1936), domed brick building described by Rawlinson really
pp. 175-84. is a Sasanian fire temple. The site has now been sur-
I H. C. RAWLIN SON, Notes on a journey from veyed, photog raphed, and the extant monumen ts
Tabriz. through Persian Kurdistan to the ruins of planned, by the Architectural Survey staff of the Ameri-
Takhti-Soleiman, Journal of the Royal Geographical can Institute for Iranian Art and Archaeology. The
Society, x (1844), pp. 77-8, cites J. GOAR (Ed.), Thea- studies confirm the views of Rawlinson and Jackson ,
phanes, Chronographia, Paris, 1655, p. :2.58; and besides adding much new information concerning
G. XVLANDER (Ed.), Cedrenus, Historia, Basle, 1566, the history of the site. For a preliminary announce-
p. 338. He identifies Canzaca with ShIz, which ment with plans and photographs see ARTHUR UPHAM
QAzvINi, Athar al-bilad, pp. 66-9, describes briefly. POPE and DONALD WILBER, Bulletin of the American
LE STRANGE, Lands of the Eastern Caliphate, p. 224, Institute for Iranian Art and Archaeology, v, 1937.
summarizes the various descriptions of ShIz as given Z Taken by Dr. Erich Schmidt.
by the Arab~an geographers other than QazvInI: by Ibn 1 DIEULAFOV, op. cit., pp. 77-8, Figs. 56, 57.
Khurdadhblh, aI-Ibn Faqlh, MustawfI, and Yaqo.t. " The first report on the other three temples in
Rawlinson's theory that the ruins of Takht-i-Sulayman the Cirra valley was made by HERZFELD, in Zeitschrifl
are the remains of Canzaca-Shlz, and that the brick der D eutschen Morgenlandischen Geselluhaft, N.F., v,
building still standing was the famous fire temple, has p. :2.56. He gives a detailed description with plans in
been questioned by A. HOUTUM-SCHINDLER, Reisen Archaeological history of Iran, pp. 91-3, Figs.
im nordwestlichen Persien, 1880-:2., p. 3:2.7. A. V. W. 12- 1 3.
JACKSON, Persia past and present, New York-London,

55 1
 OSCAR REUTHER
two small, are built on the same principle and correspond to the building at Takht-i-
Sulayman. The square cella is roofed with a dome on squinches which is carried by four
pillars connected by arches, and this is surrounded by a narrow passage, roofed on the sides
with barrel vaults and at the corners with small domes on squinches. This passage-
way had an exit on each of the four sides, as one of the best preserved ruins shows, in
the form of a round -arched portal. The exterior of the domed structure shows above
the portal a series of small niches, and the thick- shelled dome carried on a square base
rose above this like a flat cap. In the proposed reconstruction a round opening is shown
on the crown.' In front of the building was a wide, but shallow, entrance court with a
small barrel-vaulted building set symmetrically at either end. The structural method
employed is in itself sufficient to establish the Sasanian attribution of these buildings.
The thick walls are formed of broken stones set in gypsum mortar and the vaulting,
also, is typically Sasanian. That these were fire temples is
beyond question because of the complete correspondence
with Tabari' s account.
, Other ruins of buildings of this type are known near
"•
,o Shahristan in Sistan, the old Ram- shahr-i-Kaivishtasp-
shah, though here only the ground plan can be determined,
and still another in exceptionally good condition has
been found in Khurasan on the road from Turbat-i-
o~.'~' ~.~. ~~~ .~==~ "~~~m M
0"." by Do"," WHb.,. I;Iaydari to Mashhad at a place now called Raba~-i-Safid,
FIG. 157· Rab'\-i-Safld, ruined fire and this building, which is known locally as the Qal'a-i-
temple, ground plan. D ukhtar, h as now been measured and p hotograph ed
(Fig. 157).' Like the buildings in Fars, it is constructed of coarse fieldstone with
gypsum mortar and it resembles in every essential the ruins in the Gird Valley. Here,
however, the dome over the square cella is carried, not on squinches, but on primitive
projections in the corners, reinforced with wooden ties. On each side, in the base of the
dome, above the parabolic arch, is a small round-arched window. The crown of the
dome has now broken down so that it is impossible to say whether there was originally
an opening there. The corridor in this case did not surround the whole structure but
was confined to two sides and part of the third.
The second great structure in Qa!r-i-Shirin, the Chahar Qapii, has likewise been
identified as a palace like the 'Imarat-i-Khusraw. The large, square, domed structure
standing apart at the western end of the court, which was surrounded on three sides by
buildings like the dwellings of the 'Imarat-i-Khusraw (Fig. 158), was, according to this
theory, the throne room.' But this hardly seems probable in view of the marked
divergence in layout from the other known Sasanian palaces, and the domed building,
I By HERZFELD, Archaeological history of Iran, fire temple, Bulletin of the American Institute for Iranian
Fig. 12. Art and Archaeology, v (1937).
2 By D. M. WILDER and A. U. POPE, for the Archi- 3 DE MORGAN, op. cit., pp. 349-50. The interpre-
tectural Survey of the American Institute for Iranian tation of the Qal'a-i-Chahar Qapii is also questioned in
Art and Archaeology. See WILBER-POPE, A Sasanian BELL, Palace and mosque at Ukhaic,lir, p. 44.

55 2

•
- - - - - -- -- - -- - ----_.

SASANIAN ARCHITECTU RE. A. HISTOR Y

the Chahar Qapii, which has given the ruins their modern name, may, as has been
suggested, have been a fire temple.' True, there is today no surrounding passage which
is a distinctive feature of all the other presumed fire temples, and if the Parsee ritual
precepts specifying that the sun's rays shall never fall on the sacred flame prevailed
during the Sasauian period, the
enclosing passage would have been
essential; but originally there may
well have been one. For above
each arch is a round-topped win-
dow, as in the Qal'a-i-Dukhtar, and
if the arches opened originally
directly out of doors there would
have been no reason for fenestra-
,
tion. Moreover, there is a recess
in the wall between the top of the
arch and the bottom of the win-
")1_ _ _ , N
dow, distinguishable in spite of
the crumbling surface. This could
have served as impost, supporting
the spring of the barrel vault roofing ,, .I
the passage (Fig. 159).' Inasmuch
as these passages are in every known
instance narrow, the vault would
have had only a small span; hence
the outer wall need not have been ~ \~ ,30
heavy, so that the traces of it could
easily be concealed by the surround-
ing rubbish heaps. The Chahar
Qiipii is the largest building of
the type known, being 82 feet (25
metres) to a side with a dome
59 feet (18 metres) across, and it 0u.. ~.. ,".I~",,,!P'-_.:l',o,-_3o:,O_ _4l:,o_-,\CO_--,,',O=-_7",,o'--"i8,om
must, in its massive simplicity, have FIG. 158. Qa!}r-i-Shlrln, the Chahar Qapu, ground plan.
been quite impressive.
The modern name of the ruins, while it cannot be taken as proof of its original use,
may have some reference. ChaMr Qapii means Four Portals and is equivalent to the
name of the presumed temple ruins in the Gina Valley. Moreover, Tabari states that
when Ardashir built Khurra, which was later renamed Firiizabad, on tl,e site of the ancient

I SARRE-.HERZFELD., Iranische. Fe1sreliefs, p. 239; and photographs of G. ~. BELL, op. cit., Pis. ~41 67,
OUMANN, 10 Archiiologzscher Anzetger, 192 I, p. 275. 69-72; and photographs In possession of the wrIter.
Z Figures 158 and I S9 were drawn after the survey

553
 OSCAR REUTHER
Giir, he erected a building called 'a\-Tirbal' and a fire temple.' 'A\-Tirbal' was, accord-
ing to I~\akhri, the Arabic name of a building called by the Persians the Ivan, which was
still standing in Firiizabad in hi s day and was ascribed to Ardashir.' The remains of
this building, which lie right beside the ruins of the well known Sasanian tower in what

o 0 0

o 0 0 I,)

o 0 0 0

9,!,.i",,1P I,!> 2f>m

FIG. I 59. Qa~r-i-ShIrln) the Chahar QapG.: cross-section of a proposed reconstruction of
the main domed building.
is now Firiizabad qadim, show a built-up terrace, square in plan, faced with masonry.'
Four flights of stairs lead up to the platform, one on each side, and on this stood, accord-
ing to the old descriptions, the Tirbal. It has been suggested that the name Tirbal was de-
rived from the Greek tetrapylon (four-way arch), and indicated the form of the building,
and according to this same thesis the modern name of ChaMr Qapii or Chahar Taq per-
petuates an ancient designation of the Sasanian temples, representing likewise a synonym
of the Greek tetrapylon for which the Arabic form of at-fir6alwas retained into the tenth
century.' The name resulted from the four openings into the cella, one on each side.
The building represented on the bronze salver in the Staatliche Museen, Berlin
1 NOLDEK.E, Geschichte der Perser und Araber, fire temple in Flruzabad, ibid., v, p. 198.
etc., p. I I. 3 FLANDIN-COSTE, op. cit., Pis. 36, 37.
Z A. D. MORDTMANN, El Isztachri, Das Buch des 4 HERZFELD, in Zeil$chrif/ de,. Deutschen Morgen-
Landes, Hamburg, 1846, p. 67; M. J. DE GOEJE (Ed.), ldndischen Ge;el/5chajt, N,F., v, p. 2SSi idem, Archaeo-
I ~ takhrl, Bibliolheca geographorum arabicorum, II, Ley- logical history of Iran) p. 90.
den, 18 70, p. 123; Ibn al-FaqIh likewise mentions the

554
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
(Fig. 160), gives considerable support to the hypothesis that the Siis1inian fire temple was
a square domed building surrounded by a vaulted passage, with four corner domes. That
it was probably meant to be a temple is shown by the fact that through the ogee-arched
door is seen a fire altar such as appears on so many Siisiinian coins (see PI. 251 c, E).

FIG. 160. D etail from bronze salver, Staatliche Museen, Berlin (PI. 237), showing elevation of building.

On the other hand, it is not impossible that the artist had in mind another famou s building,
either one for which he used the familiar fire-temple scheme, or one which did actually
follow this plan. When the drawing on the salver was first published, and its impor-
tance for the history of Siisiinian architecture was pointed out, it was suggested that the
building was a small garden palace, which, however, as the presence of the altar indicated,
served more than profane purposes, and it has now been argued that it represents the
Takht-i-Tiiqdis, the famou s throne of Khusraw (see pp. 775-8, 878).' That altars might
I POPE, A Sasanian garden palace, The Art Bulle/ill, was intended to represent the Takht-i-Taqdis, the
xv (1933), p. 10. Dr. Ackerman is of the opinion throne of Khusraw II, but specifies that in the design
that the building depicted on the Berlin bronze salver 'the standard fire temple plan was followed' (see p. 777).

555
 OSCAR REUTHER
be found in buildings that were not specifically temples is indicated by the fact that the
Byzantines in the Battle of Malalya (A.n. 576), when they sacked Khusraw's tent,
found in it a fire altar.'
The interpretation of the drawing as representing a building Witll a large central dome
and a small dome at each corner is undoubtedly correct.' The draftsman shows the
front elevation, with the pair of corner domes on the near side concealing the correspond-
ing domes behind them, just as he has drawn
only the nearest column of tlle series support-
ing the roof of the columnar porch on either
side, indicating the deptll of the porch by
drawing the base and cornice turned upright.
The actual effect of the building becomes quite
clear in a perspective rendition (Fig. 161 a), and
it is evident that the plan (Fig. 16xb) must have
consisted of a square of piers connected by
arches, carrying the dome, and a surrounding
vaulted passage roofed at each corner with a
a dome. This corresponds exactly with the type
of building actually standing in the Gird
Valley. The altar visible through the door
must have stood in the centre of the dome
chamber. The only unsolved question concerns
b the actual arrangement of the stylar porches.
FIG. 16 14) b. Proposed reconstruction and plan of They can be thought of as only on the two
building represented on bronze salver) Staatliche sides of the main building, as the draftsman has
Museen, Berlin (PI. 237).
shown them; but, on tlle otller hand, it would
be more consistent with the whole conception of the building if they were on all four
sides, tllat is, peripteral, each separate from the others. This interpretation involves the
assumption that the artist omitted the columnar vestibule on the nearest side in order
to give a clearer view of tlle faorade, with the door and the blind arcade.
The similarity of this structure to garden palaces and kiosks of the Islamic period in
Persia has been quite rightly emphasized,' for these, also, have as essential elements a
lightly constructed, flat roofed columnar porch or vestibule, combined with a domed
central unit. The Chinili Kiosk in Istanbul and the $afavid buildings in I~fahan, the
Chihil Sutun, tlle Ayina -khana, and the Haft Dast, were all characteristic examples of
the type (see Chap. 39 L). There is, however, an even more fundamental similarity to the
great Imamzadas, the sanctuaries to which the Shi'a devout make pilgrimage. The shrine
of 'Abbas in Karbala provides, perhaps, the clearest illustration of tllis relation, for it
consists of a square domed cella with an opening on each side and a vaulted passage
1 F. JUSTI, Geschichte des alten Persien s, Berlin, , p OPE, op. Cit.,
. p. 4.
1879, p. 208 . 3 Loc, cit.

55 6
 SASANIAN ARCHITECTU RE. A. HISTOR Y
surrounding it, though in this instance the plan is complicated by an outer series of small
ivans and chapels. To this is added a farma across the main fa~ade, with tall, slender
columns.' It is very probable that these Imamzadas, which perpetuate a number of
traditional features of pre ~ Islamic cult buildings, represent an architectural continua-
tion of the Sasanian fire temple.
East Christian buildings have also been related to the fire temples. The mortuary
church of al-Mun4ir in Ru!afa-Sergiopolis, built towards the end of the sixth century,
is undeniably very similar to the supposed Sasanian temples, aside from the apse and
the paSiophoria, which were essential for Christian ritual,' but the Praetorium of
Musmiya, which dates from the second century A.D., represents the same type of con-
struction in a Roman -Hellenistic form, with columns in place of the piers, 3 and if
there was a connexion between Christian buildings of the type of the sepulchral church
of al-Mun4ir and Iranian temples, this may have been not a direct relation but a parallel
development from a common root, the one following its course within the Hellenistic,
the other within the Iranian tradition.'
The question has also been raised whether the Zoroastrian cult buildings were not the
forebears of the Armenian churches which, instead of following the Syrian basilica style,
have a central domed unit. It has even been claimed that the church of Bagaran, built
about 630, represents a pure Iranian fire temple turned into a Christian house of
worship.' A fire temple did exist at Bagaran, on which Ardashir bestowed certain
benefits. Moreover, there might have been a relation between fire temples and church
design in Armenia. Christianity and Zoroastrianism contested bitterly there for cen-
turies. Time and again the Armenian Christians rose to protect their native customs and
their faith, and destroyed the Zoroastrian temples ; and when the Persians put down these
revolts they, in their turn, demoli shed the churches, and once more advanced the creed
of Zoroaster. Finally, the Armenian grandees, after an insurrection, to forfend the
wrath of the victorious Yazdijird II, ostensibly forswore their Christianity in a fire
temple. Hence the churches, in order to be less conspicuous, might have assumed the
guise of Zoroastrian cult buildings. If this be so, however, the fire temples in Armenia
must have differed from those in Sasanian Iran, for while the form is in general the same,
the semicircular apse of the Armenian churches is quite alien to Sasanian architecture.
This feature may, however, have been added when the type was adapted to Christian
uses. In any case, the Armenian domed square with four apses, which is most clearly
exemplified at Bagaran, is proved by its semicircular apses not to be a Sasanian architec-

I A. NOLDEKE, Das Heiligtum al-I;fusains 2U been had a dome been destroyed. The similarity of the
Kerbeli, PI. III. church to a Persian fire temple would, if this were the
:l S. GUYER, in SARRE-HERZFELD, Arch~ologische case, be decidedly limited.
Reise im Euphrat- und Tigrisgebiet, II, pp. 39-43, l M. DE VocUi, L'architecture civile et religieuse
Figs. 151-6; SPANNER-GUYER, Ru ~Ma, Berlin, 1926, dans la Syrie centrale, Paris, 1865-77, p. 4 6, PI. 7.
pp. 42--1-,66-9, PIs. 31, 32, Guyer and Herzfeld be- .. SPANNER-GUYER, Ru~ifa, pp. 68-9.
lieve that the great central square of the church was 5 STRZYCOWSKI, Bagaran, Wasmuths L trikon dtr
originally covered with a pyramidal wooden roof, for Baukunst, I, Berlin, 1929, p. 285.
there was no debris in the Interior, as there would have

557
 OSCAR REUTHER
tural form (see p. 538), and it is only when these apses are cut off that such a building
assumes the form of the temples in the Girrii Valley.
The early Arabic authors do not give a clear idea of the appearance of the Sasanian
fire temples, but their accounts can to a certain extent be connected with some of the
existing ruins. The temples of the Magi are repeatedly spoken of as domed buildings.
Qazvini describes the great fire temple of Karkiiya in Sistan as having two domes under
which the sacred fire burned, each carrying a horn.' This might refer to a half-moon
finial, such as the designer of the Berlin bronze salver shows on top of the central dome
of his building, though it is possible that the Karkiiya symbols really represented horns,
a persistence of an Ancient Oriental symbol of divine might. The peak of the dome
on the temple of Shiz, which has been equated with the Takht-i-Sulayman, bore a half-
moon, according to Qazvini, and this was a talisman for the preservation of the city.'
The main building of the great fire temple at Nawbahar in Balkh, in Khuriisan,' was
crowned, Yaqiit says, with a great dome called al- U stun, and around this central unit
were 360 rooms in which dwelt the priests, one for each day of the year. Evidently
living quarters were nsually connected with temples, for according to Mas' iidi Ardashir
retired into a fire temple to spend his declining years in contemplation.' These
residential buildings probably formed an enclosed group of structures like Buddhist or
Christian monasteries. This corresponds Witll tile layout of the Chahar Qapii in Qa~r-i­
Shirin, where the isolated domed building is surrounded by subordinate structures that
may well have served as priests' houses. The walls of the temple at Nawbahar, according
to Yaqiit, were decorated with precious stones and covered Witll patterned silk hangings,
and there were numerous figures of the gods, one of which was especially venerated by
pilgrims who came there from Kabul, India, and China. From the top of the dome
floated a great silk banner, and from this and the figures of the gods worshipped by
foreigners it has been deduced tllat the temple of Nawbahar was originally a Buddhist
sanctuary, and the name has been explained as a perversion of Naw Vihara,' Vihara
being, of course, tile Sanskrit term for a monastery. Mas'iidi also speaks of this temple
which, according to him, was dedicated to the moon.' He mentions green silk banners
which floated from tall masts (see Chapter 67), and speaks of the height and solidity of
the structure and the great extent of tile temple precincts which were enclosed in a wall
(see Chapter 39 E), but his account contributes nothing to a more precise conception
of tile architecture.
While there is no epigraphic proof that any of the ruins so identified are remains of

1 LE STRANGE', op. cit., pp. 341-2., has brought to- 3 LE STRANGE, op. cit., pp. 421-2; BARBIER DE
~ether the information on the temple in Karkiiya, which MEY NA RD, op. cit., p. 569 .
IS described by Ya'qG.bl, IgakhrI, Ibn I;Iawqal, • B ARBIER DE MEYliARD--PAVET DE C OURTEILLE
MuqaddasT, Mas'udt, Yaqut, and QazvlnI. (Trans .), Ma~oudi , Les Prairies d'or, II, p. 160.
z RAWLINSON, op . cit., pp. 68- 9, g ives QazvlnI's 5 RAWLINSON, Monograp h on the Oxus, Journal
description of the fire temple of ShIz. The silver of/he Royal Geographical Society, XLII (1872), pp. Slo-
crescents on the dome are mentioned by Yaqut: I!,
B ARB IER DE MEYNARD, Dictionnaire geographique de 6 BARBIER DE MEYNARD (Trans.), Ma~oudi, Les
la Peese, Paris, 1861, pp. 367-8 . Prairies d'or, IV, Paris, 186 5, pp. 47-9.

, 5S 8
 SASANIAN ARCHITECTU RE. A. HISTOR Y
fire temples, the internal evidence for this is conclusive. Thus the identifications of
the four buildings in the Gird Valley with the temples of Mihr Narse mentioned
by Tabari can scarcely be doubted.' Yet caution is advisable in attempting any deduc-
tions from these small, unimportant buildings, as to the character of the great Sasanian
fire sanctuaries. It seems safe to assume that the central unit, the temple itself with
the altar that bore the perpetual fire, was, like these, a square domed cella with a sur-
rounding vaulted passage. Moreover, since the building on the topmost terrace of
the Qal'a-i-Rustam at Kiih-i-Khwaja was certainly a fire temple (and the discovery
of the stone altar puts this beyond question) the chain of development is complete, back
through the Parthian to the Achaemenid period. The essential features are the square cella
and the surrounding passage. The Sasanian fire temple is differentiated from its Parthian
antecedents by the four great arched openings which connect the cella with the passage.
These openings in the cella walls are lacking in the Kiih-i-Khwaja temple and in Hatra,
where only a comparatively small door gives access to the cella, and if the reconstruction
is correct, this was also the case in the Susa 'iiyadana'. Furthermore, the Sasanians,
like the Parthians (see pp. 435-6) probably had also other types of cult buildings.
Tabari speaks of a temple of Anahit in Igakhr, in which the forebears of the Sasanian
family functioned as priests and to which Ardashir sent the head of his vanquished enemy .
as a gruesome offering.' Since there are representations of Anahit and Hormuzd in some
of the Sasanian rock reliefs (Pis. 157 B, 160 A, B, and see Fig. 252), statues of these
gods were probably worshipped in temples, especially as the fi.gures of the gods in the
great temple at Nawbahar show that in at least one instance there was idolatry in the
Zoroastrian cult. Temples in which statues of the gods were displayed may well have
been quite different from the fire temples. Perhaps they were like Babylonian, Greek, or
Indian temples in plan, consisting of a series of rooms on an axis, with an anteroom and
adytum or holy of holies, in which the figure of the god stood.
Palace II excavated at Kish (see p. 538) shows a plan which, though it corresponds with
the central unit of every Sasanian palace, would be appropriate for such a Zoroastrian
temple. A forecourt, the importance of which was emphasized by the richness of the
stucco ornament, ~pened into a triple columnar hall, which could have served as pronaos ;
and a small domed room with an apse could have been the adytum, where the statue of
the god stood. Thus if the identifi.cation as a palace is not accepted, the most reasonable
alternative is that it was a Zoroastrian temple. Stucco busts of Shapiir II were found
in the court, and neither Christians nor Manichaeans would have had any motive for
decorating a religious building with portraits of a king who, as head of the Zoroastrian
state church, was opposed to their organizations and persecuted them with merciless
cruelty. The Christians did sometimes enjoy some toleration for reasons of internal
politics (see p. 78), and even occasionally exercised a certain influence (see p. 79),
but not under Shiipiir II (see p. 77), and the Manichaeans were never personae gratae.
I HERZFELD, Archaeological history of Iran, 2 N6LDEKE, Geschichte dec Perser und Araber,
pp. 66-7· p. 17.

559
 OSCAR REUTHER
MANICHAEAN CULT BUILDINGS
One Manichaean sanctuary is known, albeit not on Iranian soil, ruin K in the temple
city of Khocho, which otherwise is wholly Buddhist (see Chapter 48 B). The site is
far to the east of the utmost extension of the Sasanian empire, at the northern end of the
Tarim Basin. From the eighth century it was the residence of Uighur princes who
adhered to the creed of Mani, until the Qirghiz conquered and destroyed Khocho in
843 (229 H.). In one part of the extensive plan, called the library, numerous Mani-
chaean manuscripts were found, proof of the original status of the building. The
architecture is in every respect typical of the locality. The construction was carried out
entirely with unfired bricks, and the rooms were vaulted, in the characteristic Sasanian
fashion, with barrel vaults built up in circular courses, and domes on squinches. The
principal building, which has been identified as the place for fasting required by the
Manichaean cult, formed an enclosed rectangle divided into three sections by walls that
cut across it. The rest of the ground plan has not been deciphered. To the north stands
a building with a barrel-vaulted room in the middle and a wing covered with two domes
on either side. This may have been the residential quarters, for there are indications of
cells and some desiccated bodies were found in the building. A square domed building
with four doors, which stands apart to the west of the main building and constitutes
an independent unit, corresponds almost exactly to the Chahar Qapu at Qajr-i-Shirin.

SASANIAN CHRISTIAN CHURCHES

Far fewer Christian churches built during ti,e Sasanian period have so far been found
in the territory that was then Sasanian than in the neighbouring countries. To be sure,
most of the Armenian churches are post-Sasanian in date, but Syria plays an outstanding
role in the history of early Christian architecture. Only three Siisanian Christian churches
are known, all excavated in 'Iraq, at Ctesiphon and I;!ira.' Christian documents refer
to a good many churches in the cities of the Siisanian empire, but tlley also say that
they were repeatedly destroyed during periods of anti-Christian activity, especially
under Shapur II, Yazdijird I, and Bahram GUr.' The later kings were comparatively
tolerant towards the Christians within their dominions, and we know that Khusraw II,
under the influence of his Christian wife, had two churches built in Ctesiphon. This
change in attitude was principally due to the circumstance that after the Council of
Ephesus the Persian Christians formed a separate church, loyal to the doctrines of
Nestorius and antipathetic to Western Christendom, so that the Persian monarchs no
longer feared them as sympathizers of the arch-enemy, Byzantium (see pp. 77-8). Thus
the churches in Ctesiphon and in the cities that perpetuated Seleucia, which were in
existence during the late Sasanian period and in part survived the fall of the dynasty,
I REUTH BR, Die Ausgrabungen del' Deutschen pp. '279-80; Ars Is/amica, I (1934), pp. 54-5, Figs. S, 6.
Ktesiphon-Expedition im Winter 192 8-9, pp. I I-IS, Z E. SACHAU, Chronik von Arbela, Abhandlungen
Fig. 5; RICE, in Journal of the Royal Central Asian der Berliner Akademie, Philosophiuh-hislorische KlasJe,
Society, XIX (1932), pp. 264-6; idem, in Discovery, XIII No.6 (1915), p. 75.
(193'2), pp. 161-3; idem, in Antiquity, VI (1932),

560
 '1

SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y

were used for N estorian rites, an important fact which must not be overlooked. In
Khokhe or New Seleucia stood the cathedral which was the official church of the Katho-
likos, the 'Great church of Seleucia'. This was demolished by Shapur II and again by
Bahram Gur, but was each time rebuilt. We also hear of a Church of the Virgin, a
Church of Saint Sergius, a convent of Saint Pethion, and a Church of Saint Narkos, all
in Ctesiphon.' Moreover, in addition to Seleucia and Ctesiphon, Takrit, Kirkuk,
Arbela, and other cities were Christian centres, and must all have had important churches,
while Persia as well as 'Iraq certainly had Christian churches at this time. The Bishops
of Persis, who challenged the privileged position of the Katholikos of Seleucia, must
have had their official churches.'
The church excavated at Ctesiphon, which, like all the monumental buildings there,
was built of fired brick set with gypsum mortar, is purely Sasanian in design, with no
reminiscences whatever of Syrian church architecture (Fig. 162 a). The reconstruction of
the oblong building, 89 feet (27' 18 metres) by 49 feet 5 inches (15'06 metres) wide, should
follow the indications provided by the barrel-vaulted rooms at Sarvistan. In place of the
pairs of round piers there were simple rectangular shafts close to the long walls, with which
they were connected by small parabolic arches, still standing. The niches thus formed
must have been roofed either with small barrel vaults, or, more probably, as at Sarvistan,
with half-domes, while the nave was roofed with a barrel vault set above the niches.
The church supplants an older one in which heavy round piers, instead of the rect-
angular ones, stood in front of the long wall (Fig. 1626). The nave of this older church
can be reconstructed like the ivan of the palace at Damghiin, that is, like the rooms at Sar-
vistan, and the same type of interior design is evident in the ruins of the two churches at
I;Iira, which have been described as three-naved columnar halls of the basilica type, J one
of them also with round columns (Fig. 162 c).' These three-naved Sasanian churches,
however, were not derived from the Western basilica, but rather the triple nave was
developed, like the triple ivan of the palaces (see pp. 504-6), by gradually separating the
pillars from the side walls to which they were originally attached.
The eastern part of the building that stands in the citadel of Diyar Bakr provides a
valuable demonstration of this process. 5 The local tradition that this was an ancient
Nestorian church has been disputed, and it has been contended that the vaulted hall,
to the west side of which a larger domed building was clearly added considerably later,
was originally built for Islamic religious purposes; but however that may be, the
essential interior design is unmistakably the same as that at Sarvistan.' The rectangular
I STRECK, op. cit., p. 44, has assembled all the ( 19 I 5), p. 352, explains the eastern part of the double
information on churches in Seleucia and Ctesiphon. building in the capital of Diyar Bakr as a Christian
2 Loc. cit. 3 See p. 560, n. 1. church. Bell, who studied and photographed the monu-
4 The plan is drawn from RI CE, in Ar; blamica, 1 ment, doubted that the eastern part was Christian in
(1934), Fig. 6. origin, especially because of the masonry, which she
S BELL, in Zeilschrift [fir Geschichle der Architeklur, describes as a 'patchwork of re-used material; the
Beihcft 9, pp. 9'2-5, gives a description, and in Figs. masonry of the nave arcades betrays its Moslem origin
30-2 ground-plan and sections. by its character, the small siz.e of the stones and the
6 STRZYGOWSKI, Die sasanid ische Kirche und ihre manner of the arch building is quite unlike early
Ausstattung, Monatshifie fur Kuns/wi;ulIscha/t, VIII Christian work' (op. cit., p. 95). The writer's own

56r
-~--

OSCAR REUTHER
piers which divide the side naves from the central nave, and in front of which stand
engaged columns that carry the frontal arches, are connected with the walls by round
arches, and above this spring the stilted round arched barrel vaults which end at the
walls with half-domes. The central nave, which is almost 30 feet (9 metres) wide and
39 feet (12 metres) long, is spanned by an oval dome which is carried on squinches at

.1
a h ,
a, b. Ctesiphon, Qa!}r Bint al-Qac;lI:
a . Lower level. b. Upper level. c. I;IIra.
FIG. 162 a-c. Excavated churches, ground plans.

the corners. This is certainly not the original roof, but in form it in all probability
retains reminiscences of an older type of dome, and this reflection of earlier practices
appears also in the small vaults of the side naves, which end in half-domes on squinches.
This combination of vault and dome on squinches, which was first noted at Sarvistan,
appears in the roofing of the mosque at Ukhay4ir' and of the narthex of the 'Adhra
Church in I;Ia.kh in the Tur 'Abdin,' and one must take into consideration the possibility
that the central nave of the church in Ctesiphon may have been treated in the same way.
The characteristic features of theseSasanian naves recur in the Chaldaean, Nestorian, and
Jacobite churches in northern Mesopotamia, especially in the Tur 'Abdin, both in the
inspection in 19 I 7 convinced him that the lower part of Euphrat- und Tigrisgebiet, II, p. 3.h believes the whole
the nave arcade and the vaults of the side aisles may be building to be IslamIC.
pre-Islamic. The dome is undoubtedly late, thoug h it J REUTHER, Ocheidir, p. 24 and PI. XXVI.
may have been substituted for an earlier dome. Herz- 2 BELL, op. cit., pp. 82-3, Figs. 23, 24.
feld, jn SARRE-HERZFELD, Archaologische Reise im

562
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
convent churches with transverse naves, like Mar Ibrahim and Mar Ubil in Midiyat and
Mar Gabriel in Qartmin, and in the parish churches with longitudinal naves like Mar
Cyriacus in Arnas and various others.' These churches supplement our knowledge of
Sasanian church architecture, for while few date from the Sasanian period, in design they
agree at essential points with the churches excavated in 'Iraq, though they represent simpli-
fied provincial examples of the Sasanian type. Their evidence must, moreover, be used
only with reservations. The heavy rubble vault construction and the thick walls enclosing
rooms roofed with barrel vaults are unmistakably Oriental in character, but the masonry
columns, capitals, lintels, and mouldings
are a witness to the exceptional tenacity in
this region of Hellenistic traditions.
The pillars which form the niches are not
separated from the walls, but are connected
with them, a feature which we may assume
represents a retrogression in the style, and
the niches terminate in arches or short barrel
vaults in place of the half-domes (Fig. 163).
The main nave is spanned with a heavy
barrel vault, closed on the narrow sides with .,
a curtain wall that has no structural function. • •
The Sasanian tradition is still more evident Fw. 16 3. Midiyat, churches of M ar IbrahIm and
Mar Ubil, ground plan s.
in details of the tomb chamber of the con-
vent of Mar Tahmazjird in Kirkiik,' though here again the pillars are attached directly
to the wall. But between them are small half-domes on corner squinches as at Sarvistan,
and the pairs of half-cylinders which form the front of the piers correspond to the
coupled circular shafts there.
From tllese church interiors, which perpetuated an older, less evolved form or which
represent a reversion to a prior stage in the evolution of the style, it is evident that the
side passageways of the churches at Ctesiphon and I;Iira were really not true naves,
such as result from setting a row of supporting members in a room of a prior determined
shape, or are constituted by the addition of a unit on either side, but rather they were
originally only rows of niches, in which the dividing walls had been opened up, making
a continuous corridor.
That a true three-nave plan was sometimes used for Sasanian churches is suggested by
some of the churches in the Tiir'Abdin,at Mossul,and in this vicinity. Thus the Church
I BELL, op. cit., pp. 61-109, gives an inclusive mische Baudenkmlller, Leipzig, 191 I, pp. 3-43;
account of these buildings. A preliminary report of her SARRI-HERZFELD, ArchllologischeReise im Euphrat-
observations had been published in J. STRZYGOWSK I - und Tigrisgebiet, II, pp. 289-303, 329-4 8.
M. VAN BERCH EM, Amida, Heidelberg, 1910, pp. 224- 1 BELL, op. cit., p. 100, Figs. 35, 36, Pis. XXVI,
62. One single church is described by GUYER, Surp xxvu; SARRE-HERZFELD, op. cit., II, pp. 330-3,
Hagap (Dj inn Deirmene), eine Klosterruine der Kom- Fig. 308; W. BACHMANN, Kirchen und Moscheen in
magene, Reper/onum for KunHwissenschajt, xxxv (1912), Armenien und Kurdistan, Leipzig, 19 t 3, pp. 18- 20,
pp. 484-508; see also C. PREUSSER, Nordmesopota- Pis. 17, 18.
 OSCAR REUTHER
of the Virgin, which stands at the southwest corner of the Mar Gabriel Convent in
Qartmin, is a long building divided into three naves by rectangular pillars joined by
arches and roofed with three parallel barrel vaults.' The same type of triple-nave,
barrel-vaulted, pillared hall appears in the Mar Ahudeme and in other Mossul churches,
as well as in the Ulu Jami' in Kirkiik, which was originally a Christian church.' But
here again no genetic relation to the Western basilica can be assumed, though familiarity
with the Syrian type of church, represented in monumental form by the basilica in
Mayafarqin, built during the reign of Khusraw II,' may have reinforced the tendency
to this form of building. The Nestorian and Jacobite churches with three naves must
be regarded as three oblong halls joined to make a single unit, with the dividing walls
reduced to rows of pillars.
The west wall of the church uncovered at Ctesiphon had no openings in it, entrance
into the nave being effected by three narrow doors in the long walls on north and south.
This same peculiarity is found also in the Tiir 'Abdin churches, where the entrances
are likewise, as a rule, in the long walls of the rectangular structure, though here there
is a vaulted narthex, and no trace of this feature was found either at Ctesiphon or in the
. In the convent churches with a transverse nave the narthex is on
two churches at Hira.
the west side, but in the parish churches with a longitudinal nave it is on the south.
Thus in entering it is always necessary to cross the main axis and, as is architecturally
foreordained, the entrance doors open into the wall niches.
On the eastern sides of the church excavated at Ctesiphon and also of the two at J:Iira
there is a group of three rectangular rooms which can together be considered the
'choir'. At Ctesiphon the middle one is the widest, and judging from the arrangement of
the plan must have held the altar. The two narrower ones on the sides can be designated
the prothesis and the diaconicon. They were entered from the nave through narrow
doors and were also connected by small entrances with the altar room, which opened
into the nave with a wider door. There was, however, a door here, not the open chancel
arch usual in Western churches. The choir rooms of the church excavated in Mound XI
at J:Iira are the same, but the second church there shows a variation in plan. The entire
width of the altar room was open into the nave, there were double doors from the side
naves into the side rooms, and no doors between these and the altar room.
Here again, in this arrangement of the choir the Sasanian ruins correspond with the
North Mesopotamian churches, where the altar room has a flat back wall instead of a
round apse. Choirs of this type are found in Mar Ibrahim in Midiyat and Mar Maika
in the Tiir 'Abdin, and the Mossul churches and the Ulu Jami ' Kirkiik have likewise a
straight outer wall at the choir end.' The rounded apse of Mar Gabriel in Qartmin,
Mar Ya'qiib al-J:Iabis in Salah, and the oblong churches in Kefr Zeh, Arnas, and
Midiyat must have been due to Syrian or Armenian influence. Thus the rectangular,

I BELL, op. cit., Fig. 5. • of Ibid" Figs. 9, 10 ; S A RRE-HERZFELD, op. cit.) II,
1 SARRE-HERZFELD, op. cit.) II, p. 329, Fig. 307 . Fig. 307; Ill, PIs. CHI, e Vil, CVIIl, CIX.
3 BELL, op. cit.} pp. 86-8 .

..
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
as opposed to the round choir must be considered N estorian, that is, Sasanian, and this
rectangular choir, called today the madhbah, is connected with the nave, known as the
haykal, in both Nestorian and Jacobite churches, only by a door varying in width.
This is the result of certain features of the ritual, which would indicate that these
aspects of the ceremony were already established in the Sasanian period.'
At Ctesiphon the altar room was widened out in the middle on either side by niches
that extended from the long walls, and thi s suggests that it was roofed in the middle with
a dome on squinches, with a barrel vault running out on either side. This form of
vaulting may also have been used on the altar room of the church in Mound XI at
I;Iira. Support for this interpretation is, again, found in the Mesopotamian churches,
where the madhbah is often emphasized with a dome, for example, in Mar Petros and
Mar Ahudeme in Mossul, and also in the Ulu Jami' in Kirkiik, though a simple barrel
vault might also have been used, as is usually the case on altar rooms of the Tiir 'Abdin
churches.
The ruins disclosed at Ctesiphon and I;Iira give some idea of the equipment and
decoration of the Sasanian churches. Both the I;Iira churches had, in the middle of the
central nave, a raised platform. In the church in Mound XI, which was more completely
uncovered, this platform was flanked by curved seats, and it has been suggested that it
may have served the same purpose as the Schola Cantorum in Western churches.' At
Ctesiphon four holes forming a square in the gypsum floor indicate that there were
supports for a mensa or ciborium, and on the back wall beyond this was a high, stepped
projection. Fragments of mural paintings were found there, and also the headless,
painted stucco torso of a saint, or perhaps of Christ Himself. Numerous fragments of
stucco ornaments were likewise polychromed and gilded in parts.' The treatment of
the drapery on the headless torso recalls that of the flying 'angels' on the spandrels at
Taq-i-Biistan, and there is an even closer connexion between the stucco ornament and
the palmettes and round archivolt mouldings found in the various Sasanian palaces, an
indication of the relatively late date of the decoration, though the building itself may
have been earlier. The walls of the altar room excavated in Mound V at I;Iira were also
decorated with two layers of paintings, the lower in the Sasanian style and obviously
dating from the seventh century, the subsequent over-painting similar to the early
Islamic murals at Samarr:i, but clearly Christian, for the cross is a conspicuous feature.
Crosses appear also in a number of plaster fragments, both here and in the other
church at I;Iira,' and it has been suggested that they played a specific role in the cult,
corresponding to that of the icon in the Orthodox church.
The exteriors of the Sasanian churches were undoubtedly characterized by the same
massive concentration and effect of weight that are so typical of the churches of the
I On the peculiarities of the choirs of Chaldaean, Z RI CE, in Ars b/amiea, I, p. 56.
Nestorian, and Syrian Jacobite c,hurches see A. BAUM- 3 REUTHER, Die Ausgrabungen der Deutschen
STARK, Ein Alterskriterium im nordmesopotamischen Kte s i~h o n-Expeditionl pp. 12-13 .
Kirchenbau, Oriens ChriS/ian us, N . S., v (19 I5), pp. I I 1- • ICE, op . .
CIt., 6
pp. 5 S,S.
12; SAlUtE-HERZFELD, op. cit., II, pp. 29 7- 8.
 OSCAR REUTHER
Tiir ~bdin, where the cubic mass is scarcely relieved by windows or any architectonic
elements. The gable and span-roof form s, concealing the barrel vault, found in the
Tiir ~bdin, can hardly be postulated for 'Iraq, where a level terrace roof must have
been used, if the vault or especially the domes were concealed.
The features typical of these comparatively small churches were probably equally
characteristic of the more important ones, so that we can assume that the 'great church
of Seleucia', the official church of the Katholikos, was an oblong, barrel-vaulted build-
ing with a choir that had no apse, though it may have been richer and more elaborate in
form, perhaps with a dome and, contiguous to this, rectangular arms roofed with half-
domes. The 'Adhra church in I:Iakh gives a hint of what it may have looked like,
though the round apse design with shell niches and columns is wholly un-Sasanian.'
If the mortuary church extra muros at Ru!afa' were, as has been suggested, closely
related to Sasanian fire temples, we might infer that in Persia also there was probably
still another type of church derived from the fire-temple plan. But inasmuch as the
Ru!afa church is not a derivative of the fire temple, but a parallel development pre-
sumably from a common origin, such an inference would hardly be justified.

TOWERS
The church tower and the square minaret originated in Syria, deriving immediately
from the ancient watch towers. The Sasanian churches had no towers and the Eastern
round minarets have nothing to do with Christian church towers, having developed
from an entirely different source (see p. 929). The mosque towers, however, are in
general associated with a hypothetical Iranian religious tower, which was presumably
connected with the Zoroastrian cult, a theory based chiefly on philological evidence.'
There are two Arabic names for the mosque tower: mi'dhana, meaning the place where
the mu'adhdhin calls out his prayer, the usual term in the southern and western
Mediterranean countries ; and mantira, whence is derived the European term, minaret.
'Maniira' means philologically a place where fire burns or shines, and though the origin
of the name would thus seem to indicate that the conception was derived from the
lighthouse, the thesis has been advanced that in the Islamic East both the word and
the structure go back to tower-like buildings on which burned the Zoroastrian fire.
The Persian synonym of the Arabic maniira is iitash giih, meaning 'place where there is
fire', and this has for the most part been connected with the idea of some kind of tower
intended for the fire cult.
The ruins of a tower now 82 feet (25 metres) high, in Firiizabad, beside the remains
of the Tirbal, are thought to have been an iitash giih. That the structure dates from the
Sasiinian period has never been questioned. The reconstruction formerly proposed'
I BELL, op. cit., pp. 28- 9, Figs. 23-4. Baudenkm~lerJ I, Berlin, 19 t 8, pp. I 13-16 i CRESWELL,
1 See p. 55 7, n. 2 . Evolution of the minaret, with special reference to
3 On the problem of the origin of the minaret see Egypt, Th e Burling/on Magazine)XLVI!J (1926), pp. 134-
H. THIERSC H, Pharos, Antike, Islam und Orient, 40 , 2 52-8,290-6.
Leipzig, 1909, pp. 97-172; E. DIIZ, Churasanische of DIEULAFOY, op. cit., pp. 79- 80, Fig. 58 , PIs. XIX, xx.

566
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
shows a massive tower, square in plan, the top of which was approached by an exterior
steep spiral ramp. It was suggested that such a tower with a ramp had been derived from
the Babylonian temple tower, the ziggurat, and according to thi s theory this Sasanian
tower constituted the connecting link between the ziggurat and the unu sual spiral minarets
of the 'Abbasid mosques in Samarr;;, which are round in plan but essentially the same in
principle.' More recent investigation, however, has revealed that the Firiizabad ruins
had been misinterpreted, and that the section still standing was only the core of the
tower, around which wound a ramp that was vaulted, the whole having originally
been enclosed in another outside wall. Thus the Firiizabad tower was, in exterior
design, rectangular.' This new interpretation of the form, which must henceforth be
accepted, is no argument, however, against the identification of this building as an iitash
giih, and the immediate proximity of the temple ruins does tend to show that the tower
was connected with the Zoroastrian cult. Indeed, the tower may well have been a kind of
gigantic fire altar on which, on special occasions, burned under the open sky the sacred
fire which was preserved in the temple;' or else it may have been a 'high place' for
certain Zoroastrian ceremonies connected with the invocation of the stars.
It has been suggested that the Takht-i-Rustam i.n the Shahriiyaz district near Tehran
was such a 'high place'. It is a terrace building on a steep isolated basalt peak 650 feet
(200 metres) high. One platform, built of flat stones without mortar, crowns the
summit of the hill, a second has been constructed two-thirds of the way down. Near
the Qal'a-i-Dukhtar at Raba\-i-Safid is anotherSasanian building standing on a high cliff
near the brink of a gorge, the Qal'a-i-Pisar, and this also is thought to have been a
'high place' for Zoroastrian worship, which seems sometimes to have been connected
with temples.'
Ammian and Zosimos mention towers that Julian's army saw in 360 on their
march across the Babylonian plains during the war against Shiipiir II,' but it is a
question whether these were ruin s of Babylonian ziggurats, or towers of city walls, or
'high places' of the Zoroastrian cult built like towers, such as the Firiiz1ibiid tower very
probably was. Ammian describes a tower, which stood where the 'royal canal' branched
off from the Euphrates, as being built like the Pharos, which means that it must have
consisted, like the famou s lighthouse of Alexandria, of a number of stepped-back
stories. This statement has led to the conclusion that Ammian saw a ziggurat, which
at that time might still have been in sufficiently good condition to show its original form.'
But on the other hand, it is quite possible that Zoroastrian towers built for the same
purpose as that in Firiizabad were also constructed in a succession of diminishing stories.
The Leningrad silver plate contributes some information bearing on this discussion.

I HERZFELD, Samarra, Berlin, 1907, pp. 26-30. ElIcyclopaedia of Islam, tJJ (193 4), p. Soc.-Ed.
1 HER ZFELD, in Zeitschrift der D eutschen M orgen- 'H ERZFELD, op. CIt.,
. pp. 233, 275.
liindischen GeseIIschaft, N ,F,) V, p. '2'27. s AMMIAN US MARCELLI NU5, XXIV, 2, 6; ZOSIMOS,
J A pre-Islamic cult practice involving a fire on top XVIl, 8 (Historia, Book III).
of a tower lasted long into Islamic times, as a special 6 HERZFELD, Samarra, p. 29.
festival, at aI-Muzdalifa: see F. BUHL, al-Muzdalifa,
 OSCAR REUTHER
It has been assumed that the designer was not representing here an actual episode, but
was depicting an allegorical 'Castle of Faith', defended against inimical powers, a
symbol of the Zoroastrian church militant.' The fire altar, resembling a column, which
is visible through the door, shows that the illustration must have had some religiou s
significance. Moreover, it seems improbable that a fortress would have had an exterior
arcade, and the Sasiinian forts, so far as one can judge from the existing ruins, were
quite different in appearance. The building on the plate must have been intended as a
two-storied rectangular tower. The crenellated crown was in all probability used also
in the Firiizabad tower, and this element taken over from fortress architecture, which
constituted the usual cornice on Sasiinian monumental buildings, may have been used
on towers of this type as a symbol of enduring power and permanence (but see
pp. 833-4, 86 9-7 0 ).
MEMORIAL MONUMENTS
Rectangular towers were also certainly used commonly in Sasanian architecture for
memorial buildings. The But-Khana of Paikiili was not, as was formerly thought, a
fire temple,' but a memorial structure, essentially the same in function as the rock reliefs,
for it was built solely to carry the royal busts in relief and the inscription commemo-
rating the war of the builder, Narse (A.D. 293-303), against Bahram III Shahanshah.
It was a tower 29t feet (9 metres) square in plan, and though there are now only
scattered remains, it has been shown that originally it was a massive construction of
rubble and mortar faced with masonry (Fig. 164). The corners were treated as engaged
columns with bell capitals.' The cornice consisted of stepped crenellations Witll a row
of circular embrasures below. On each side were bu sts in full relief above the inscrip-
tion. It is impossible to determine whether tl,ere was an interior stair leading to the
terrace roof, but if there was, there must have been an entrance, which would have had
some kind of architectural treatment.
The choice of a tower for this purpose was undoubtedly influenced by memories of
the older tomb towers. The Parthian mausolea in Hatra with their corner columns or
pilasters were the specific prototypes, with Hellenistic details (Fig. II3). The monumental
sepulchre was entirely alien to Sasanian Persia owing to the Zoroastrian custom of
exposing the dead, but the tower was taken over as a purely memorial monument,
essentially the same in idea as the victory towers of eastern Islam (Pis. 355-6) and India.
Moreover, the monument ofPaikiili was not the only one of ti,e kind. In the Mamaseni
district there is a ruin now called Mil-i·Azdahar which likewise was a Sasiinian masonry
tower, obviously very similar to the But-Khana, though it bears no inscription.'
Another structure may also, judging from the plan and the monumental construction,
have been a memorial. On the road from Ram-Hormuz to Khiizistan is the Rahdar
darvaza-i-gach, a massive masonry gateway with a great central round arch and a smaller
1 SARRE, Die Kunst des alten Persien, p. 69. iandischen Gesellschaft, N,F., v, p. 227.
2 See p. 4, note I. • Ibid., p. 258.
3 HERZFELD, in Zei/Schrift der Deuuchm Morgcn-

568
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
arch on either side (Fig. 165). The modern name means 'customs gate'.' The striking
similarity of this building to a triple triumphal arch suggests that it was a Sasanian
version of the Roman type of memorial.
The fas:ade of Taq-i-Biistan, the most important Sasanian memorial monument, was

FIG . 164' Paikult, commemorative monument of Narse (2 93-30 1) (reconstruction) .

also designed to have a three-arched fas:ade. ' To be sure, as a memorial it was primarily
the work of sculptors, but architecture, too, played an important role. Nor was it
limited to providing the setting for the reliefs, for the ivan grottoes and the additional
rooms built beside them and now destroyed, or perhaps planned and never built, were
intended as a temporary residence for the king when he went to the Paradise Park that
had been created around the spring. Thus the modern name Taq-i-Biistan, meaning
the 'Arch in the Garden', is correct. The three -arched fas:ade, which was at least planned
if not actually constructed, corresponded to the triple ivan plan. Taq-i-Biistan may,
I C. A. DE B ODE, Travels in Luristan and Arabi- see H ERZFELD, Am Tor von Asien, p. J 56, n. 114.
stan, London, 1845, I, p. 390. Herzfeld assumes that the Partho-Sasanian triple-
:t HERZFELD, op. cit., p. 229. On the relations of the arched fa~ade and the Roman triple triumphal arch had
triple-arc he~ f~r;ades of H a~ra , of the D a~az.a-i-gac h, a Hellenistic prototype, as does OELMANN, in Bonner
and at Taq-I-Bustan to the tnple Roman tnumphal arch 1"hrbu(heY, Heft 12 7, p. t 5 t.

569
 OSCAR REUTHER
therefore, be considered a sculptural rendition of a summary representation of a palace,
converted into a memorial monument by its sculptural treatment.
Finally, the Taq-i-Girra (PI. 153 c) may also have been a memorial, for it is very
difficult to see what practical purpose could have been served by this little masonry
structure isolated against the vast background of the Zagros Pass. In conception it is
only the central architectural motif, the ivan, separated from the total house plan and
converted into a monument.
Judging by the large- scale ashlar set without mortar of which tlle Taq-i-Girra is
constructed, and also by certain stylistic features, notably the horseshoe arch of the

FIG. 16 S. Rahdar darvaza-i-gach.

archivolt, and the mouldings, the building certainly had close relations with the architec-
ture of the neighbouring countries immediately to the west of the Sasanian empire, and
indeed in all probability it was actually the work of North Mesopotamian stone masons. '
Numerous statements of the Arab historians and geographers show that foreign work-
men were employed on Sasanian buildings, and while many of these assertions may
merely be a reflection of the superior prestige of Western architecture, as, for example,
where Sinnimar the builder of the castle of Khavarnaq in J:Iira is spoken of as a 'Rrtmi',
a Roman or Byzantine,' still ~tller reports of this kind doubtless had some historical
basis. They all refer to engineering constructions.
BRIDGES
Mas'iidi and Tabari' recount that Shapiir I, after his victory over the Romans in 260,
employed the Roman prisoners that he had captured with the Emperor Valerian to
I There is a marked similarity both stylistically and :t ROTHSTEIN, Die Dynastie der Labmiden in al-
techn ically between the Taq-i-Girra and Mar Ya'qub I:::IIra, p. IS, has brought together the information on
in Nisibis, the older parts of which afe built of large- Sinnimar given by the Arabic writers. According to
scale ash lar with little or perhaps no mortar at all J:famza al-I~fahanl (lOS), the RiimI who worked for
(SARRE-HERZHLD, Archllologische Reise im Euphrat- Nu'mln built not the Kh ovornoq, but the Sinnln castle,
und Tigrisgebiet, II) r. 338). It is, however, true that while Tabarl (852; 14) quotes a poem of 'Abd al-
the later Christian bUildings of Mesopotamia were en- 'Uzzl which says merely that the builder was an 'ilj,
tirely constructed with mortar, while the tendency to that is a foreigner in general. See also J. HALivy,
build with cut stone laid dry certainly persisted longer Khawarnaq et Sinimmar, R evue Simi/ique, xv ( 1907),
in Syria: see B UTLER, Publication of an American p. 101; and HERZFELD} Am Tor von Asien, p. 102.
archaeological expedition to Syria in 1899-1900, Part 3 NOLDEKE, Geschichte der Perser und Araber,
11, Architecture and other arts, New York-London, pp. 32-3; BARBIER DE MEYNARD (Trans.), Ma~oud i,
1904, p. 13. On the Taq-i-Girra see p. 493, n. 3; Les prairies d'or, IV} p. 186.
P. 495, n. 7; p. 509 , n. 2.

570
 SASANIAN ARCHITECTU RE. A. HISTOR Y
build the bridge over the Kariin and the dam at Shiishtar. These structures, long known
and repeatedly described, are still partly in use today, though they have been many times
repaired and improved. Tabari says the bridge was 1,000 ells long, which corresponds
with the length of the actual bridge, 1,693 feet (516 metres), and the forty-one piers
that support it are very similar to Roman bridge piers. For example, the piers of the
Roman bridge across the Rhine near Mainz were pentagonal and were set on a pyramid of
the same ground plan, a rectangle with a triangle applied so that it projects against the
current, and the Shiishtar piers show the same type of pentagonal plan.' This is only
one of a variety of forms commonly used in Roman bridge building. Again, the
bridge at ShiishtaI (PI. 261 A) and also the smaller bridge over the Ab- i - Diz near Dizfiil
(PI. 26IB) are thoroughly Roman in structural method. The core of the piers is composed
of a coarse cobble and monar concrete, and this is faced with masonry, the alternate
courses of which penetrate the core, thus integrating the mass, a far more solid form
of construction than that used by the Sasanian builders of the Paikiili monument. On
the other hand, the Shiishtar bridge piers are very close together, and the bays, which are
now vaulted with Islamic arches, are very narrow, and in these respects the design is
quite un-Roman; but these features may be explained by the fact that the structure
was intended to serve not only as a bridge, but also as a dam. The width of the piers
is 29t feet (9 metres) , while the bays are only about 23 feet (c. 7 metres) wide. The
cross arches of the Roman bridge over the Moselle at Trier are about 65t feet (c. 20
metres) across. Moreover, the span of the Shiishtar bridge cross arches varies, for the
builder set the piers where he found islands in the conglomerate through which the
river has cut its bed (see Chapter 39 M).
That Sasanian bridges show Roman relations cenainly does not mean, however, that
the Sasanians were dependent on Roman skill for their engineering. The oldest
known substantially constructed bridge was that built by Nebuchadrezar over the
Euphrates which is described by Herodotus and Diodorus. The brick piers of this
bridge were uncovered in the course of the excavations there. ' The masonry bridge
connecting Ctesiphon and Seleucia in the Parthian period was most probably a deriva-
tive, technically, from the Babylonian tradition.' If Shapiir made use of Roman
prisoners in carrying out his great building projects on the Kariin and the Ab-i-Diz,
it must have been simply to take advantage of available labour and skill, not because
he thought that they alone would be capable of carrying out the work, or even believed
1 DJ EULAFOY, op. cit.) V, pp. I05-u, gives a long 3 STRECK, Seleucia und Ktesiphon, p. 37 ; idem,
description of the bridges of Shushtar and Dizfui, and Die alte Landschaft Babylonien nach den arabischen
in Figs. 97, 98 a plan and sketch of one bay of the latter. Geographen, pp. 2.69- 70; l:iamza al-Il?fahant, J. M. E.
A. NEUBURGER, Die Technik des Altertums, Leipzig, GOTTWALDT (Ed.), I, Leipz.ig, 1844, p. 31, states that
1919, p. 478, Fig. 638 , illustrates a cut-water of the there were still remains of the bridge in the bed of the
Roman bridge over the Rhine from a model. Tigris, so that boatmen had to beware at low water.
2. R. KOLDEWEY, Das wiedererstehende Babylon, IHAKHRY, Masalik al-mamalik, 87, I, writes that in his
pp. 193-5; F. WETZEL, Die Stadtmauern von Baby- day (c. 950 (c. 340 H.)) there was no longer anything to
lon, Leipzig, 1930, pp. 54-7, PIs. 51, 77, 7Si C. be seen. Both designate the bridge as 'fJanfara' which
MERCKEL, Die Ingemeurtechnik im Altertum, Berlin, signifies a vaulted arched bridge, and Il?!akhrt says that
1899, pp. 60-9· it was built of fired bricks.

57I
 OSCAR REUTHER I

them superior to his own master builders. A bridge at Idhaj in Khuzistan might,
judging from the descriptions of Yaqut and Qazvini,' have had connexions with Roman
building methods. The structure, known as the Qan\ara Khurra-zadh becau se it had been
built by Ardashir's mother, spanned the river with a single gigantic arch (it is said to
have risen ISO ells above the water level), and this was built of masonry blocks held with
iron clamps set with lead solder. This is a typical Greek technique, but it was also used
later in Roman bridge building.' Thus it is impossible to say whether the Qan\ara
Khurra-zadh followed the Hellenistic-Seleucid tradition of masonry construction, or
whether it had been built under the direction of Roman engineers. One detail cited by
Yaqut and Qazvini is very curious and technically hardly explicable. Both write that
the foundation s of the buttresses were made of a combination of lead and slag.
The Pul-i-Zal near Khurramabad,' which spans a deep gorge with a parabolic arch,
is typically Sasanian, as is also the centre arch of the Zab bridge at Altiinkiiprii.' While
these may not have been built until the early Islamic period, the parabolic arch form
is characteristic of Sasanian construction, as are the style of vaulting with upright
circular courses of bricks above projecting courses, and the reinforcing with wooden
beams. The piers are made of concrete faced with masonry, a method of construction
usual in Sasanian bridge building and obviously chosen because it would resist the
wear of the water so much better than bricks. The arches, on the other hand, were
commonly built of bricks in the current vaulting technique. Thus if the bridges of
Shushtar and Dizful as well as other bridges at more northerly sites (see Chapter 39 M),
do reveal the Roman collaboration mentioned by the Arab descriptions, other bridges
still in existence show that there was also a native tradition in bridge building.
FORTRESSES
Moreover, that Roman and, later, Byzantine technicians did not exercise a dominant
influence even on Sasanian engineering is clear from the Sasanian fortresses, which,
I YXQUT (BARB I ER DE MEYNARD, Dictionnaire gee- iron and lead. See also RAWLINSON, Notes on a march
graphique, pp. 461-2) writes that the bridge had a from Zohab to Khuzistan, Journal of the Roy/Ii Geo·
single arch with a span of more than 1,000 ells and graphical Society, IX (1839), p. 82 .
500 ells high. This fantastic exaggeration of dimen- . . . . .
sions is matched by the peculiarity of the technique For the full description given by Yaqut see
that he describes, for he asserts that the bed for the Chapter 39 M).
foundations was made of lead and iron and also that in 2 The Pons Cestius and the Pons Fabricius in
building the arch a mixture of lead and slag was used. Rome were built in this technique: see C. HOLSEN,
Qazvlni's description of the bridge is more detailed, but Jahresbericht fiber die Topographie der Stadt Rom,
repeats the extraordinary statements of Yaqut concern- Milleilungen de; Kaiser/icll Deunchen Archaologi;chen In-
ing the use of lead and iron . LAYARD, in Journal of the ;liIUIJ, Romische AbteilulIg, 1899, pp. 282-3; MERCKEL,
Royal Geographical Society, XVI (1846), p. 95, believes op. cit., pp . 281-3, Fig. 88 . The masonry blocks of the
that the MaJamlr of today has taken the place of the old Roman bridge at Mainz were also bound together with
Idhaj (Eididj), and he undertook to find the traces of the iron clamps: see J. GRIMM , Der rOmische BrUckenkopf
famous Qantara Khurra-zadh in the remains of brick- in Kastel bei Mainz und die dortige Romerbrucke,
built piers of a bridge that spanned the Karlin River Mainz, r882, p. 26.
below the valley-plain of Malamlr, where it flows 3 HERZFELD, Bericht uber architologische Beo·
through a deep gorge. But inasmuch as these piers are bachtungen im sUdlichen Kurdistan und in Luristan,
built of fired brick, they do not correspond with the Archaeologisclle Milleilungm au; Iran, I (1929), p. 73 .
descri~tions of Yi\qlit and QazvlnI, for they state that .. Idem, in SARRE-HERZFELD, Architologische Reise
the bridge was a masonry structure held together with im Euphrat- und Tigrisgebiet, II, pp. 322-6, Fig. 306.

57 2
 SA SAN I A N ARC HIT E C T U R E. A . HIS TOR Y
judging from the limited evidence at our disposal, were quite distinctive in character
and markedly different from late R oman or Byzantine military architecture.
The 8asfulian military structures concerning which we have the most adequate infor-
mation are, besides the square citadel of Qa!r-i-Shirin called the Qal'a-i- Khusraw, the
city walls of Ctesiphon and of Dastajird, of which there are plans and descriptions.'
Both had massive rounded towers or bastions, proj ecting from the face of the wall to
a depth equal to their breadth. At Ctesiphon the wall was built in the Babylonian
fashion of unfired brick and clay mortar and was nearly 35 feet (10 metres) thick. It
was protected against outside attack with battering rams by a steeply piled-up scarp.
Every 125 feet 6 inches (38'30 metres) there was a tower 30 feet (9'30 metres) wide.'
The city wall of Dastajird, called today by the Arabs Zindan, is built of fired brick
and gypsum mortar, and is 54 feet 6 inches (16·60 metres) thick. The towers here are
about 38 feet (11·60 metres) wide, and are closer together, only 58 feet (17'7 metres)
apart. Imbedded deep in the wall is a vaulted passage for the guard, with narrow
embrasures that run to the outer surface on the diagonal. It was the strongest city defence
that had ever been built in the Near East, yet Khusraw II, who built it, seems to have
deemed it insufficient protection against H eraclius,' for he fell back on Ctesiphon.
Rounded bastions were exclusively used, so far as we know, only in the Sasanian
fortresses of this period. The military architects of the Ancient East were familiar with
the rounded tower, as the examples of the city wall of Ur and later on of the citadel of
Zenjirli, the ancient Sham'al, in Syria show.' But in Babylonian and Assyrian military
architecture in the first millenniu m B.C. only rectangular towers appear, and this is the
only form found also in the Parthian city wall at H atra.' The Greeks of the Seleucid
period also used round towers, especially in exposed parts of their fortifications. Y et
rectangular towers were the rule on all the H ellenistic fortification walls of the N ear
East, although the Classical writers on fortifications recognized the advantages of semi-
circular or polygonal towers, since they offer greater resistance to battering -rams and
to bombardment with. catapults. First the Greeks and then the Romans, Vitruvius
and later Vegetius and Philo of Byzantium, recommend the use of such towers.'
Roman fortifications show a combination of round and rectangular towers, or some-
I For accounts of notable Sasanian mi litary strong- u ber die von der Deutschen Forschungsgemeinschaft in
holds see p. 74, n. 2- Ed. U ruk-Warka un ternommenen Ausgrabungen, Abhand-
:l R EUTHER, Die Ausgrabungen der deutschen Ktesi- lungen der PreujJischen Akademie der WisJenscha/ten,
phon-Expedition im Winter 1928- 9, pp. 6-9; idem, Philosophisch-hislorische Xlasse, NO. 4 (1936), pp.41-
In Antiquity, III (1929), p. 44 9. H erzfeld's supposition 3, Fig. I. I n general, however, rounded towers on
(op. cit., pp. 53-5) that the city wall of Ctesiphon was a fortifications are very u nusual in the Ancient East. O ne
double wall and con nected by transverse walls without exception is the sem icircular bastion in Susa : see
any projecting towers has been shown by the excavations D IEULAFOY, L 'acropole de Suse, Paris, 1890, P I. II .
to be erroneous. Babylonian and Assyrian fort ress wall s and also the
J SA RRE- HERZFELD, op. cit. H, pp. 89-93, Figs. H ittite walls at Boghaz-KOy have, without exception,
175-7· rectangular projectin g towers.
of KOLDEWEY, D ie Architektur von Sendschirli, 5 A NDRAE, Hatra, II, Leipzig, 1912, pp. 25-59,
Berlin, 1898, pp. u 8, 131- 2, P is. 28-30. Semi- Figs. 24-63.
circular towers were already in use in the city wall of 6 VITRUVIUS, Book I , Chapter 5. 5 ; P hilo, see
Uruk, which dates from t he third millennium B.C. : A . DE R OCHAS D'AIOLUN, Principes de la fortification
see A. VON H ALLER, in Sieben ter vorlaufiger Beri cht antique, P ari s, 188 1, p. 33.

573
 OSCAR REUTHER
times round towers alone, especially in Gaul' and Syria, where the fortified frontier
camps like Odhruh and al-Lejjiin have only round towers.' Yet the walls of large
cities of the fourth and fifth centuries A.D.- Rome, Constantinople, Salonika, and
Antioch-had rectangular towers, and only a subordinate place was granted to demi-
octagonal, pentagonal, or triangular towers, the last appearing at Salonika.' All
the Sasanian city walls that we know, on the other hand, are characterized by huge
rounded bastions, and they are found also, though proportionately smaller in size,
on the Sasanian citadels and fortresses, like the Qal'a-i-Khusraw in Qa!r-i-Shirin.'
These rounded bastions are continued into the early Islamic period. Nothing remains
of the inner circular 'Abbasid city at Baghdad, but it undoubtedly had round towers
such as appear at Raqqa, which also was founded by Man!iir, and likewise in other
early Islamic fortifications, for example, Qa!r-al-J:Iayr and Ukhay<:lir.' There and in
Samarra, however, the round towers had already become merely architectural elements,
features in the design of the enclosing walls of palaces and mosques.' Thus an Ancient
Eastern architectural principle was perpetuated.
CITY PLANNING
The Sasanians were evidently conscious of the distinctive character of their architec-
tural style as compared Witll foreign styles, for when Khusraw Aniishirvan destroyed
Antioch in 540 and founded a new city, where the inhabitants could be installed, near
Ctesiphon, called New Antioch-i-Khusraw, he had it laid out on exactly the same
plan as the original Antioch, according to Tabari, and had tile houses built in the Syrian
style.' Moreover, tllere were a Hippodrome and other buildings tllat would hardly
have been included among the public works of a Persian city of the period. The
remains of this city, which tile Arabs called Riimiya, tile 'Roman', are still to be seen
in the ruins called Biistan-i-Kisra, a little over a mile (c. 2 km.) southeast of the
Taq-i-Kisra, and the excavations there have uncovered a powerful city wall, built of
sun-dried bricks, as was usual in 'Iraq.' It is noteworthy that tile towers are rectangular.
Square towers were also used in the city wall of the original Antioch. Thus the fortifica-
tions, also, of the new city followed tile late Roman or Byzantine method.
I A . BLANC HET, Les enceintes romaines de la R. FORSTER, Antiochia am Orontes, Jahrhllch des Kaiur-
Gaule, Paris, 1907, p. 262; H . THIERSCH, Pharos, Neh Delluehen Archiiologi;chen instilllU, XII (1897), pp.
Leipzig-Berlin, 1909, pp. 142, 25 I. 130-45, Figs. 2, 5-8.
2 R. E. BR UNNOw-A. V. DOMASZE WSKI, Die Pro- .. DE MORGAN, op. cit., IV, pp. 354-7, PI. XLIX,
vinzia Arabia, Strassbu rg, 1904-9, II, pp. 5, 8, 24, 9 S, Fig. 214 .
Pis. XLI-XLII. S SARRE-HERZFELD, Archaologische Reise im
3 Concerning the wall of Aurelian at Rome see Euphrat- und Tigrisgebiet, II, pp. 121-2; idem.,
J. A. RICHMOND, The city wall of Imperial Rome, p. 356; REUTHER, Ocheidir, pp. 9-10, Pis. 111, IV, VI,
Oxford, 1930. The wall of Con stantinople on the land XXII; BELL, Palace and mosque at Ukhai~ir, pp. 4, 6,
side has recently been measured by H. Lietzmann, Pis. 1,2,3,6-9.
F. Krischen, et al. : see LIETZM ANN, Die Landmauervon 6 HERZFELD, Samarra, Berlin, 19°7, p.. 20, 39.
Konstantinope1, Berlin, 1929. The city fortifications of 7 NOLDEKE, Geschichte der Perser und Araber,
Salonika are described by O. TAFRALl, Topographie de pp. 165,239. .
Tessalonique, Paris, 1913, pp. 65-81 . The city walls 8 RE UTHER, Die Ausgrabungen der Deutschen
of Antioch, the remains of which, still standing, date Ktesiphon-Expedition im Winter 192 8-9, pp. 35-7;
from after the earthquake of 528, are described by SARRE-HERZFELD, op. cit., pp. 59-60.

574
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
Only two sections of the Biistan-i-Kisra wall remain, forming a right angle. This
would indicate that the enclosed area was rectangular, perhaps square. In this respect,
too, the Roman model was followed. But we have little definite information on the
Sasanian principles of city planning. Although the sites of a number of Sasanian cities
are known, almost no research on this problem has yet been undertaken. The oval,
almost circular layout of Ctesiphon dates probably from ti,e Parthian period, and that
this form of plan, which seems to have been typical of Eastern cities, was current in
the Parthian period is shown by the layout of Hatra and Darabjird (see p. 44I). The
round scheme, however, recurs in the 'Abbasid period in Man§iir's al-Madina al-
Mudawvara, the round city of Baghdad, hence in all probability it persisted tllrough the
Sasanian period (see Chapter 42). But square Sasanian cities are also mentioned. Thus
according to Mustawfi, Shapiir I laid out Nishapiir, which was named after him, in the
form of a checker-board with sixteen squares.' The author adds that the Sasanians were
accustomed to laying out their towns in the shape of an animal or an object. In
another passage he tells how Ardashir had Shiishtar laid out in ti,e form of a horse
(see Chapter 42).' The idea that a certain layout, usually astrologically determined,
would be propitious for a newly founded city was widespread in the East. Regular
rules for this are found, for example, in ti,e discussions of city planning in ti,e Manasara
and ti,e otl,er Silpa-Sastras.'
Since we know so little about Sasanian city plans, it is difficult to say how far
aesthetic considerations were taken into account. But it hardly seems accidental that the
church excavated in Ctesiphon lies directly on the east-west axis running through the
ivan of the royal palace. Apparently the arrangement on an axis, which determined
the placing of buildings in the palace precincts, purely for aesthetic reasons, also
prevailed over a wider area, and the same principle is found, rigidly maintained, at the
'Imarat-i-Khusraw in Qa§r-i-Shirin. That this principle persisted centuries after the
Arabic conquest is shown by Samarra, on the one hand, and ti,e I§fahan of Shah 'Abbas
(see Chapter 42), on the other, to cite only an early and a late example. Nor is it
necessary to assume that Hellenistic or Roman traditions influenced this practice. In
any society organized in a hierarchy of classes or under an autocracy, symmetry and
axial plans predominate, not only in the design of individual buildings, but also in the
arrangement of groups of buildings, and, finally, in the layout of the inclusive group,
the city.

In the last analysis, Sasanian architecture cannot compare in excellence to either

late Roman or Byzantine architecture. Nor is this conclusion vitiated by the limited
range of known monuments. None of the vanished palaces mentioned by the Arab
writers is said to have surpassed the ivan -i - Khusraw, and Ibn I;Iawqal is quite convincing
I LE STRANGE (Trans .), MUSTAWFl, Nuzhat al- 3 RAM RAZ, Essay on the architecture of the
Quhib, Leyden, 1919, p. 147. Hindus, London, 1834, pr. 41-4; K. PFEIL, Die
2 Ibid" p. 107. The author is indebted for this indische Stadt, Leipzig, 1935, pp. 29-30.
reference to the Editors.

575
 OSCAR REUTHER
when he says that the Persians had nothing else like it.' It was more important even than
the other palace in Ctesiphon, the White Castle, which the Caliph al-Muktafi (902-8
(290-6 H.)) had razed so that he could use the materials to complete the at-Tiij palace
in Baghdad, begun by his predecessor al-Mu'ta9id.' Yet the Royal Palace of Ctesiphon,
which seemed to the amazed Arabs one of the wonders of tile world, could not be
compared as an architectural creation (if tile whole can be judged by the remaining part)
to any of the thermae of the late Roman Empire, nor is it comparable, if one accepts
the opinion that it was built in the sixtll century and wishes to measure it by its con-
temporaries, to any of the great Byzantine churches of this period, to say nothing of
Hagia Sophia. Despite its imposing colossal scale, the weakness of this, the outstanding
Sasanian monument, is unmistakable.
The design of the Taq-i-Kisra fa~ade has often been very severely criticized, chiefly
because of the lack of coincidence of the vertical axis, and the combination of Classical
motifs markedly different in scale; and the architect has been accused of a barbarous
misunderstanding of the basic architectonic requirements of me style whence he had
borrowed his fa~ade elements.' This could be attributed to me fact mat they were taken
from foreign sources, and in an alien environment had lost meir original significance.
But if these practices be faults, it is not only the architect of tile Taq-i-Kisra who was
guilty of them, for me same features appear also in me work of Sixtll century Byzantine
architects, and it is more probable mat me architects of mis time deliberately disregarded
the Classical principles of balanced proportions and of vertical coincidence in favour of
a complex rhythm. A millennium later the same architectural tendencies again prevailed,
for Michelangelo, Palladio, and the Baroque architects all used the 'colossal order'
to join two stories, while Pietro Lombardo, for example, chose to abandon vertical
continuity on me courtyard fapde of me Palazzo Ducale at Venice, and the German
Renaissance architects very often decisively rejected me axioms of the Classical orders,
preferring a varied rhythm and a picturesque distribution of accents.' Nor is it a major
defect tlmt me architect of Khusraw's palace concealed the actual structure of his
building behind a decorative screen, much of which had notlling behind it. He was
minking in terms of a court, and his primary aim was to enclose this. The front of his
building was not a fa~ade in tile Western sense, but a part of the court wall, which had
to be completed on me other sides.
A far more serious fault is the clumsiness and heaviness of the actual construction of
the walls, arches, and vaults, the incredible excess in sheer bulk of material which the
I M.J. DE GOEJE(Ed .), InN HAUQAL, Bib/io/h eea Geo- POPE, review of Archaeological history of Iran, Bulletin
graphrm~mArabico,.um, Il, Leyden, 1870-94, p. 167, 10. of the Americ4n Ins/illite for Persian Art 4nd Archaeology,
Z F. WOSTENFELD (Ed.), YAQUT, Leipzig, 1866, I, IV (1936), p. 164.
p. 109, IS; H, p. 304. See also STRECK, op. ci t., pp. .. J. BURCKHARDT, Der Cicerone (H. WOLFFLIN
Z 59-60 i LE STRANGE, Baghdad during the Abbasld (Ed .)), Berlin-Leipzig, 193$, I, pp. 198-9; H . WILLICH,
Caliphate, Oxford, 1900, pp. 39, 2.$2-4· Die Baukunst der Renaissance in Italien, Berlin-
1 HERZFELD, Samarra, p. 44; Idem, in SARRE- Neubabelsberg, 1914, pp. 135-6; M. WACKERNAGEL,
HERZFE LD, Archll.ologische Reise im Euphrat- und Die Baukunst des 17. und 1 8. Jahrhunderts in den
Tigrisgebict, II, pp. 68-70; idem , Archaeological his- germanischen Landern, Berlin-Neubabelsberg, t 9 I 5,
tory of Iran, pp. 94-5; for a criticism of this view see pp .• 8, 33.
 SA SAN I A N ARC HIT E C T U R E. A. HIS TOR Y
architect, because he had had no experience with such a gigantic structure, felt was
necessary, and also other technical defects, shown, for example, by the inadequate
wooden reinforcements at the impost of the great ivan arch.
On the other hand, the noble clarity of the ground plan, in so far as it can be deciphered
in the excavated area, is at the opposite pole from the inept construction, and this contrast
is seen again in the latest Sasanian monuments. There is no relation whatever between
the grandiose plan of the ' Imarat-i-Khusraw and the defective technical execution. It
must be conceded that the Sasanian architects did often have to carry out splendid
conceptions with materials that were most unfitting, both structurally and architec-
turally. This disparity between intention and means might, on the one hand, have been
due to the consciousness of a great past, of which the monarchs and their architects
wished to be worthy, a determination not to fall short of the magnificent scale of the
Achaemenid monuments ; yet, on the other hand, it might be an indication that, even
to the end of the period, Sasanian architecture had not matured.
But just because Sasanian architecture did not fully mature, it had within it the germs
of further development, and this was carried out by the Persians after the House of
Sasan fell before the Arab onslaught. Such continued evolution was possible because
Sasanian architecture had freed itself far more definitely than Parthian architecture from
the Hellenistic ideal and had far deeper roots in the native Iranian tradition.
That Sasanian architecture was the successor of Parthian architecture seems historic-
ally unmistakable. Yet seen as a whole, it does not represent a direct continuation of the
Parthian tradition as exemplified by the Parthian buildings at Ashur and Hatra. From
the very beginning the development takes an oblique course, deflected by the strong
stream of influence that flowed out of Persis, native land of the Dynasty. Many
Hellenistic elements which were still evident at the end of the Parthian period dis-
integrated, or even wholly di sappeared; new tendencies were added and released the
capacity for further growth of old prin<:iples. The desire to perpetuate Achaemenid
traditions was of little real significance and did not survive the early period. At the
most it resulted only in the adoption of a few superficial features. The central impetus
came from the native structural practices, up till then limited in scope, now given the
opportunity for a great development. Thus tl,e line of progress led away from forei gn
contributions to the basic Iranian heritage, as the palace of Ardashir at Firllzabad bears
witness, with its strong, archaic, satisfying monumentality.
Eastern Islamic architecture, as the direct heir of the Sasanian tradition, transmitted
its Persian components to the architecture of the I slamic world. This is not apparent in
the earliest West Islamic monuments. The Umayyad buildings in Syria follow, in tl,e
main, the Syrian style of tl,e sixth and seventh centuries. But at Mshatta and 'Amman
it becomes evident that the roots of Islamic architecture, even west of the Euphrates
and in the Syrian desert, are closely intertwined with the Persian Sasanian development.
The interdependence becomes even more marked when the 'Abbasids make Baghdad
their capital, and so move the fulcrum of the Islamic empire towards tl,e east. From
577
 OSCAR REUTHER
this time on it is obvious that the fall of the House of Sasan did not entail the death of
the Sasanian architectural style, but rather opened the way for its direct evolution in a
wider sphere. The early Islamic structures of Ukhayc:lir, Samarrii, Damghan, Nayin,
the tomb of Isma'il the Samanid, and a number of other places continue the principles
of Sasanian architecture unmodified. Structure, layout, design, motifs, and ornament
all remain essentially the same, and the changes that are introduced in the first centuries
after the Hijra are in the nature of evolution, and only in a small degree the result of out-
side influences. The full maturity, which the style had not reached during the Sasanian
period, is now attained, and thereafter its influence extends both west and east in the
Islamic world.
A SURVEY OF
PERSIAN ART
FROM PREHISTORIC TIMES TO
THE PRESENT

ARTHUR UPHAM POPE

Editor
PHYLLIS ACKERMAN
A nil/anI Editor

VOLUME I
TEXT

PRE-ACHAEMEN ID, ACHAEMENID

PARTHIAN AND SASANIAN PERIODS

PUBLISHED UNDER THE AUSPICES OF THE

AMERICAN INSTITUTE FOR IRANIAN ART
AND ARCHAEOLOGY

OXFORD UNIVERSITY PRESS

LOND ON AND NEW YORK
19 3 8