Geochimica et Cosmochimica Acta Vol. 42. pp.

II99 lo 1212
0 Pergamon Press Ltd. 197X. PrintedinGreatBritain

Rare-earth abundances in chondritic meteorites

N. M. EVENSEN.P. J. HAMILTONand R. K. O’NIONS
Lamont-Doherty Geological Observatory of Columbia
University, Palisades, NY 10964, U.S.A.

(Received 29 August 1977; accepted in revised form 4 April 1978)

Abstract-Fifteen chondrites, including eight carbonaceous chondrites have been analyzed for rare
earth element (REE) abundances by isotope dilution. These analyses complement and extend earlier
isotope dilution REE determinations in chondrites, performed in other laboratories, so that coverage
of major chondrite classes is now complete. An examination of this body of precise and comparable
REE data from individual chondrites reveals that only a small proportion of the analyses have flat,
unfractionated REE patterns within experimental error. A statistical procedure is used to derive revised
chondritic abundances of REE by selection of unfractionated patterns. A number of the remaining
analyses show Eu anomalies and fractionated patterns consistent with magmatic fractionation as
encountered in the products of planetary differentiation. However, many patterns exhibit features not
readily explicable by known magmatic processes; in particular, positive Ce anomalies are often encoun-
tered. Abundance anomalies can be quantitatively determined by the use of a least-squares curve
fitting procedure. The wide variety of anomalous patterns and the uncertainties in model parameters
preclude detailed modeling of the origin of anomalies, but it is probable that at least some arise
from fractional condensation in the solar nebula, as has been demonstrated for Allende inclusions.
Elemental abundance anomalies are found in all major chondrite classes. If these anomalies are
ignored, the range and nature of variation within chondrite classes are consistent with a parent body
model, in which solid-liquid or solid-solid equilibria induce variations from an unfractionated bulk
composition. Absolute abundances in the H, L and LL parent bodies are almost twice those of the
E parent body.
The persistence of anomalies in chondritic materials relatively removed from direct condensational
processes implies that anomalous components are resistant to equilibration or were introduced at
a late stage of chondrite formation. Large scale segregation of gas and condensate is also implied,
and raises the possibility of bulk variations in REE abundances between planetary bodies.

INTRODUCTION cumulative effects of all fractionations undergone by

the material since its origin as primordial condensate,
THE RAREEARTHELEMENTS (REE) comprise fourteen the regularity of the normalized REE patterns in even
naturally occurring elements with extremely similar the most highly differentiated of terrestrial, lunar and
chemical properties. This chemical similarity renders meteoritic samples provides a striking demonstration
them less susceptible than other groups of elements of the chemical coherence of these elements. The
to mutual fractionation in geochemical processes, and smooth patterns obtained imply a similar smooth
thus simplifies the interpretation of those fraction- variation with atomic number in those chemical
ations that are observed in natural systems. properties relevant to the geochemical fractionations
In particular, it may be expected that abundances experienced by the materials. The effects of most geo-
of REE in primitive meteorites should closely ap- logic processes affecting REE abundances can, in fact,
proximate the relative abundances present in the solar be understood in the context of simple crystal-chemi-
nebula from which the Sun, planets and meteorites cal properties of the REE, principally ionic radius and
are ultimately derived. The measurement of REE oxidation state. The interpretation of REE patterns
abundances in chondrites thus provides fundamental of evolved materials is discussed in detail in reviews
information on the ‘cosmic abundances’ of the ele- by HASKIN et al. (1966, 1968, 1970).
ments (SUESSand UREY, 1956; CAMERON,1959, 1968). Fundamental to these applications of chondritic
These not only constrain models of nucleosynthesis REE abundance data is the notion of the uniformity
(e.g. BURBIDGEet al., 1957; TAYLOR, 1960), but also of the relative abundances of REE in individual chon-
provide the initial abundances from which more drites, and the implication that the variety of REE
evolved materials have formed. paterns observed today are the result of a variety of
The most widespread use of chondritic REE abun- fractionation processes acting upon relatively homo-
dances arises from the observation that when REE geneous starting materials with chondritic relative
abundances in other natural materials are divided by abundances of REE. The first survey of REE in some
the corresponding chondritic abundances, the result- twenty individual chondrites, performed by neutron
ing values typically exhibit a smooth and regular vari- activation analysis (NAA), demonstrated a reasonable
ation with atomic number (HASKINand GEHL, 1962; degree of consistency in relative abundances within
MASUDA, 1962; C~RYELL et nL, 1963). Since these the accuracy (- 10%) permitted by the method at that’
chondrite-normalized values presumably reflect the time (SCHMITTand SMITH,1963; SCHMIIYet al., 1963,

1199
1200 N. M. EVENSEN, and R. K. O’NIONS
P. J. HAMILTON

1964). Subsequent to this pioneering work, techniques residue in 6M HCI was occasionally incomplete and a resi-
allowing more precise determination of REE abun- due amounting to < I?; of the sample persisted. Initially
fusion of the residue in LiEOX was attempted, but this
dances by mass spectrometric isotope dilution (ID)
procedure was found to contribute high and extremely
were introduced (SCHNETZLER et al., 1967; GAST et variable blanks in the light REE. A large portion of Abee
al.. 1970; SHIMIZU.1974; H~OKERet a/., 1975). How- was therefore dissolved to allow REE in the soluble frac-
ever, almost a decade elapsed after the studies of tion to be measured directly. The residue was found to
SCHMITTet al. (1963. 1964) before the ID technique have no detectable REE content in excess of the LiB02
blank, and so in view of this observation and the small
was applied to a similar survey of REE in chondrites amount of residue remaining, it has been centrifuged off
(MASUDAet al., 1973; NAKAMURAand MASUDA. 1973; and discarded when present. The chemical separation of
NAKAMURA.1974). These studies were considerably the REE was performed using anion-exchange techniques
more precise than the early NAA work. and demon- essentially identical to those described previously by
strated more clearly than the previous data that chon- HAWKER et al. (1975). This procedure involves the scpar-
ation of the REE as a whole from a first anion-exchange
drites as a class are far from homogeneous in REE column, employing a CH&OOH-HN03 mixture as elu-
abundances. They further demonstrated that some in- tant, and separation of the REE into La. Ce-Gd and
dividual chondrites display strikingly anomalous REE Dy-Yb fractions on a second column by elution with a
patterns of a form not readily explicable in terms of CH,OH-HNO, mixture. Chemical separation blanks are
in most cases negligible (IO-” g or less for each rare
familiar crystal-liquid chemical fractionation.
earth), and are given in Table 1.
Recently B~YNTON (1975) has attempted to explain Isotopic analyses were performed on a VG-Micromass
the extremely anomalous REE pattern observed by 30 mass spectrometer equipped with a Daly scintillation
TANAKAand MASUDA(1973) in a Ca-Al-rich inclusion detector with fast ion counting logic as well as a conven-
from the C3 chondrite Allende. He pointed out that tional Faraday collector. Following HOOKERer nl. (1975).
the REE were analyzed in three fractions (La, CeCrd and
available thermodynamic data imply that condensa- Dy-Yb). The mass spectrometric technique only differs in
tion of REE from a gaseous medium (solar nebula) the use of a triple filament assembly for the analyses of
can potentially result in fractionations of markedly Dy-Yb fraction. which results in improved sensitivity for
different character than those imposed by crystal- the metal species and minimal interference. The field values
corresponding to the masses of interest are selected by an
liquid partitioning, and can at least qualitatively on-line computer and all data reduction. including interfer-
account for the anomalous Allende pattern of ence correciion for ld2Nd0+ on ‘?eO’ and *%jrn+
TANAKAand MASUDA(1973). and 13’CeO+ on 13sLaO+. are made on-line. The maani-
In view of the varied and often anomalous REE tude of the ion-currents measured varies between lOy14
and IO-“’ A depending upon the particular element and
patterns observed in chondrites. and the possibility
analysis. In general preliminary data are acquired on small
of explaining some anomalous patterns as resulting ion beams using the Daly scintillation detector. and then
from fractional condensation in the solar nebula, a progressively larger ion beams are measured using either
program of REE abundance determinations in chon- the Daly detector or the Faraday cup. Analyses were fre-
dritic meteorites has been undertaken in this labora- quently checked in separate mass spectrometer runs.
The precision of REE determinations is considered to
tory. The primary objectives of this program are: (1) be 1-2:~ for all the REE.
to augment the data base of high precision REE
determinations in chondrites by extending the survey
of MASUDA et al. (1973). NAKAMURAand MASUDA
(1974) and NAKAMURA(1974); (2) to attempt to verify RESULTS
the existence of anomalous REE patterns as noted The results obtained in 18 analyses of 15 chondrites are
by these workers; and (3) to investigate individual displayed in Table 2. The classification of meteorites used
meteorites in detail in an effort to gain further under- in the Table and throughout the text is that of VAN
SCHMUSand WOOD (1967). as modified by VAN SCHMUS
standing of the nature and origin of the variations
and HAYES (1974)and WASSON(1974) for the carbonaceous
and possible anomalies. In this paper we discuss chondrites. Representatives of every chondrite group have
results obtained to date with respect to the first two been analyzed. except for CV chondrites. exemplified by
of these objectives. Allende for which many other analyses exist (e.g. TANAKA
and MASUDA.1973).Particular emphasis has been placed
ANALYTICAL PROCEDURES
The third of the above objectives. the detailed analysis
of individual objects. requires a capability for the precise
determination of the REE in samples of much less than Table I. Procedural blanks (ng)
100 mg at chondritic abundance levels: i.e. on samples an
order of magnitude smaller than those previously reported
in the literature. The analytical techniques were. therefore,
La
designed with this goal in mind. and niost of the analyses Ce LO.12
reported here were performed on 50-300 mg samples. This Nd 0.05
sample size may accentuate the effects of small scale in- sm 0.03
homogeneity in the samples, in comparison with previously ELI 0.004
Gd 10.02
published chondrite analyses. 0.007
DY
Samples. with an aliquot of a mixed REE spike solution ET 0.004
added, were digested in a HF-HNO, mixture in a PTFE Yb 0.007
decomposition vessel at 140°C. Subsequent solution of the
 Rare-earth abundances in chondritic meteorites 1201

Table 2. Rare-earth element analyses

Meteorite TYPO Yeight La Ce Nd Sm Eu Cd 3Y Er Yb

Ivuna CIl 193 0.231 0.610 0.459 0.146 0.0550 0.193 0.239 0.157 0.156
orguei1 1 CII 163 0.264 - 0.512 0.168 0.0708 0.217 3.277 0.189 0.184
orguei1 2 CIl 36.9 0.226 0.569 0.422 0.142 0.0524 0.183 0.233 - -
Cold Bokkex
mid c>12 105 0.352 0.916 0.638 0.217 0.0839 0.283 0.337 0.228 0.225
i?ixhel CM2 216 0.336 0.936 - 0.217 0.0779 0.279 0.333 0.220 0.215
CN2 20.5 0.249 0.655 0.476 0.148 0.0619 0.201 0.245 0.174 0.205
012 178 0.330 0.859 0.665 0.212 0.0785 il.273 0.340 0.222 0.221
co3 285 0.433 1.190 0.838 0.256 0.0939 0.343 0.419 0.329 0.273
co4 351 0.472 1.270 0.935 0.304 0.112 n.404 0.479 0.348 0.321
Abee 1 F.4 231 0.330 0.777 0.461. 0.152 0.0546 0.207 0.243 0.167 0.159
Abee 2 E4 630 0.197 0.526 0.378 0.122 0.0487 0.171 0.203 0.143 O.ILE
Abee 3 E4 48.7 - 0.692 0.521 0.168 0.0715 0.278 0.281 0.186 !l.>80
Indarch E4 41 0.318 0.776 0.567 0.183 0.0720 0.227 0.275 0.188 0.195
KWW H4 53.2 0.207 0.64i 0.452 0.145 0.0549 0.194 0.232 0.161 0.148
Richardton H; 93.2 0.334 0.840 0.600 0.194 0.0691 0.255 0.313 0.216 0.19C
Bluff 1.5 301 0.259 0.617 0.491 0.160 0.0762 0.223 0.279 0.208 0.291
Pamallea Li.3 377 0.366 0.938 0.707 0.233 0.0868 0.313 0.387 0.269 0.258
OllWWX4 LL5 293 0.291 0.739 0.577 0.100 0.0847 0.255 0.31s 0.220 c.213

on the carbonaceous chondrites, since few of these have MAWDA et al. (1973; 3977), NAKAMURAand MASUDA
been measured by isotope dilution. The new data also pro- (1973) and NAKAMURA (1974). hereafter collectively referred
vide some overlap with previously analyzed meteorites. to as NMT. To these may be added three analyses by
The rare earth patterns, normalized to the average abun- GAS et al. (1970) and a whole-rock analysis from
dances in CI chondrites (see discussion below) are dis- SCHNETZLER and BOTTINO (1971).
played in Figs. 1 and 2. The primary source of neutron activation analyses of
Individual REE vary in abundance by at least a factor chondrites is the pioneering work of SCHM)~~et al. (1963;
of two among chondrites, and in some cases much more. 1964) and SCHMITT and Smrrr (1963). which was the first
The patterns themselves have a varied assortment of systematic study of REE in meteorites. These early data,
shapes, and in order to discuss the general features of REE together with an analysis by HASKINand GEHL(1962), are
variation in chondrites and examine possible correlations of lower precision than the ID analyses and do not permit
with various classes, it is useful to consider our data in the resolution of fine detail in REE patterns. More recent
the context of previous REE analyses. activation analyses using improved techniques are given
for two chondrites in HASKIN et al. (1968).
Comparison with previous data Many analyses of REE in Allende have been obtained
Almost all of the published isotope dilution analyses of by ID (GAST et al., 1970; TANAKAand MASUDA,1973;
REE in individual chondrites are found in the work of MASUDA et al., 1973; NAKAMLJRA, 1974; TANAKAet of.,

i “-I
0.0

L(LLL 0 BLUFF (LS)

0 PARNALLEE lLL3)
x OUMNZA ILL51

I.6 -

1.4ro-Q

I ’ 1 1 I 1 1 I I
I’ fi 1
.a Ce Nd Sm ‘5 6d Oy Er Yb La Ce Nd Sm Eu 6d Dy Er Yb

Fig 1. REE abundances in carbonaceous chondrites from Fig 2. REE abundances in E, H, L and LL group chon-
this study, normalized to average CI chondritic drites from this study, normalized to average CI chondritic’
abundances. abundances.
1202 N. M. EVENSEN,P. J. HAMILTONand R. K. O’NIONS

197.5).NAA (WAKITAand SCHMI~. 1970; WANKE er al.. circularity in the selection of ‘typical’ chondrites to be used
1974) and spark-source mass spectrometry (MARTINand in deriving a chondritic average, but this effect can be miti-
MASON. 1974). Because of the extreme heterogeneity and gated to a large extent by a suitable statistical treatment
complexity of Allende and the fact that most of the ana- of the data.
lyses have been performed on separated phases. the major- On the assumption that chondritic abundances as
ity of these analyses are not relevant to a discussion of observed today represent varying degrees and modes of
chondritic REE abundances. Similar comments apply to departure from a uniform primordial abundance pattern
data on separated phases of other chondrites (e.g. HONDA reflective of cosmic abundances, there should be a tendency
and SH~MA,1967; SCHMITTet al., 1968; SCHNETZLER and for relative abundances to cluster at the values character-
BOTTINO,1971; ALLENand MA.SON,1973). istic of the original undifferentiated material. An objective
These ID and NAA analyses together with those pre- assessment of the cosmic abundances can then be provided
sented here thus comprise the available data base of whole- by finding the relative abundance values that match the
rock REE analyses of chondrites, and will be used exten- largest possibte subset of analyses within some given error.
sively in the discussion that follows.* The 56 ID analyses Although in principle statistical techniques such as clus-
of 39 meteorites include examples of all of the major ter analysis could be applied to the data, the large number
classes of chondrites and many of the petrologic types. of free parameters (REE abundances) to be determined and
The data are of comparable precision, and since almost the limited data base have led us to adopt an iterative
all analyses were made in two laboratories, they should approach that is simpler but remains both objective and
be intercomparable with a minimum of uncertainty. Data statistically reasonable. The data are first normalized to :
obtained on BCR-1 in this laboratory agree well with the approximate ‘average chondrite’ values. For each sample
analyses of NAKAMURAand MASIJDA(1973) and HOOKER the standard deviation of the normalized abundances of
et al. (1975). There is general agreement between data the individual REE in the sample is then computed, and
reported here and those of NMT where overlap of meteor- the sample is either accepted as flat or rejected, according
ites exists, though the internal inhomogeneity of chondrites to whether the y0 standard deviation falls within a given
precludes close comparison. The 22 activation analyses of ‘window’ or not. The relative abundances of the flat pat-
21 chondrites include I2 meteorites not analyzed by ID. terns are then averaged and used as normalizing values
and were primarily obtained in one laboratory. Compari- for the next iteration. This process is repeated until a self-
son of ID and NAA results is difficult; because of the consistent set of flat patterns is obtained, i.e. one whose
larger errors and additional elements analyzed by NAA. average generates the same set of flat patterns. At this
the data obtained by the two methods is generally dealt point the window is narrowed and the above process
with separately in the statistical analyses that follow. repeated. For each value of the window the mean and
The analyses of the data base exhibit a range of variation standard error are computed for the relative abundance
far in excess of their experimental precision. This raises of each REE, and the average standard error for n/l REE
the very real question of choosing ‘normal’ or ‘average’ analyzed is taken as a measure of consistency. This average
chondritic abundances. The practice of nor~lizing error may be expected to decrease with narrowing of the
observed REE abundances for ‘chondritic’ values is so use- window as aberrant patterns are successively rejected, until
ful (e.g. HASKINrr al.. 1966) that it would be highly desir- the level of random analytical error is reached. At this
able to introduce suitable values at the onset of this dis- point further rejections (narrower window) should not de-
cussion. Other workers have tended to use either averages crease the spread of the resulting data set by an amount
of individual analyses (e.g. SCHMITTet al., 1963. 1965; NAK- proportional to the decreasing number of analyses; this
AMURA.1974) or analyses of chondrite mixtures (e.g. HAS- would lead to a rise in the standard error. This is precisely
KIN et al.. 1968, 1970: SCHNETZLER and PHILPOTTS.1969; the behavior observed in practice. The set of anafyses yield-
HUBBARDand GAST. 1971). MASUDAet al. (1973) chose ing the minimum error then comprises the most internally
to normafize their values to the L6 chondrite Leedey. consistent subset of the original data base. and the average
These choices are mutually incompatible, but Leedey and abundances derived from it have the lowest standard error
the Nakamura average have similar relative REE abun- obtainable.
dances, while differing significantly in absolute abundances. The above treatment was applied separately to the ID
Such distinctions may have little relevance in the interpre- and NAA analyses of our data base. For the ID data.
tation of REE patterns of highly evolved materials, but starting with a window of IO?4 deviation, the process con-
can potentially prove important in the underst~ding of verges to the same values regardless of the abundances
small but ~alytjcally resolvable differences between or employed for the initial normalization. within wide limits.
within chondrites. Individual analyses of Leedey and Indarch, the averages
Whether averaging of chondritic abundances is done of SCHMITTet al. (1964) and NAKAMURA(1974) and the
arithmetically or mechanically (by mixing of chondrites chondrite composite of HASKIN er al. (1968. 1970) were
into composites), the results of such averaging are evidently all employed as starting values with identical final results.
dependent upon the choice of chondrites or even particular The minimum standard error is obtained for a 2’?/;,window.
portions of chondrites used. and in the case of composites, which yields a set of nine analyses, though use of a slightly
no assessment of the individual samples can be made prior larger window has negligible effect on the abundance
to their inclusion. The data base of precise analyses is now values or their errors, as given in Table 3. Since the
large enough that a more systematic selection of analyses nominal precision of the ID analyses is l-Z”/;, these nine
to be averaged may be made. It should be possible to analyses may be considered identical within analytical
arrive at chondritic abundance values which are relatively errors.
free of the effects of including aberrant or evolved materials When the treatment is applied to the NAA data, a win-
in the fina! average. There is, of course, a fundamental dow of 30% must be set initially to produce convergence
to a unique set of tinal values inde~ndently of the starting
values. The minimum error is found for a 207; window,
which. accepts 19 of the 22 analyses as flat. The resulting
relative abundances (Table 3) are of lower precision than
* Supplementary data tables comprising a compilation those from the ID data but include those elements not
of whole-rock chondrite analyses, the same data norma- determinable by isotope dilution. To obtain the final rela-
lized to the average CI abundances derived in this paper, tive abundances of Table 3, the NAA results have been
and results of the anomaly detection procedure described adjusted to give a least-squares best fit to the ID results.
below, may be obtained from the authors on request. then combined in a weighted average to yield best estimate
 Rare-earth abundances in chondritic meteorites 1203

Table 3. Best estimates of unfractionated cbondritic REE abundances

CI **
NM* ID+ Co&inede AVerapt!q C.A.U.
La 1.000 t 0.021 1.000 t 0.007 1.000 * 0.007 0.2446 0.480
ce 2.973 t cl.160 2.598 * 0.008 2.608 t 0.008 0.6379 1.241
Pr 0.4lmt 0.0155 0.0149 0.09637
0.394fJt 0.187
w 3.928 t 0.040 1,936 t 0,012 1.937 t 0.012 0.4738 0.896
Sm 0.67042 0.0132 0.6266* 0.0026 0.62%+ 0.0026 0.1540 0.279
EU 0.2442+ 0.0067 0.2364t 0.0012 0.2372* 0.0012 0.05802 0.104
Cd 1.003 + 0.034 0.8312t 0.0023 0.8351? 0.0023 0.2043 0.354
:b 0.1594t 0.0045 0.1531k 0.0043 0.3745 0.0643
DY 1.013 10.029 1.036 + 0.005 1.039 2 0.005 0.2541 0.426
HO 0.2414+ 0.0044 0.2318? 0.0042 0.05670 0.093?
Er 0.70991 0.0162 0.6761f 0.0031 0.6787f 0.0031 0.1660 0.271
Tilt 0.109oi 0.0033 0.1047* 0.0032 0.02561 0.0413
Yb 0.60811 0.0167 0.67582 0.0036 0.67482 0.0036 0.1651 0.260
LU 0.0024 0.1034t 0.0003 0.1038t 0.0003 0.02539
0.1070?: 0.0396

*La-normalized abundances from NAA data alone. Errors are one

standard error
t ~-normalize abundances from ID data alone. Errors are one stan-
dard error.
g La-normalized average of ID and NAA values, weighted by respect-
ive variances. after multiplying ID values by 1.0412, the average ratio
of NAA to ID values.
C Combined relative abundance from previous column, multiplied by
0.2446 to produce a best fit to absolute abundances measured on CI
chondrites.
**Cosmic abundance units (atoms/fO~ atoms Si). derived from CI
Average by assuming 10.3% Si in Cf chondrites (MOORE, 1971).

relative abunda~~s for all 14 REE. Because of the large i~~vidual REE and irregular patterns. These present a dif-
data base used and the objective nature of the treatment ficult problem in interpretation, both because of their var-
applied, these relative abundances are believed to be iety and because many of them bear little resemblance to
superor to previous values in the literature, though for the familiar types of patterns encountered in analyses of
the pofyisotopic REE they are very close to the values terrestrial and Iunar materiafs. It would seem that pro-
of NAKAMURA(1974) cesses different from those encountered in planetary differ-
The assignment of absolute REE abundance values is entiation must have produced the anomalous patterns in
more problematical. In the absence of Si determinations these meteorites. Before attempting to understand these
on afiquots of samples analyzed for REE, it is impossible more unusual patterns, it is wortbwh~le examining possible
to determine to what extent variations in absolute abun- mechanisms of REE fractionation.
dances among flat patterns represent true variations of Si-
normalized abundances and to what extent they result
simply from variable amounts of diluent phases in the REE ~ffE~~RY AND ~RA~I~NAT~ON
material analyzed. Using the average Si values compiled The most regular fractionations of REE result from
by MOORE(1971) for the various chemical groups of chon-
drites to normalize absolute abundances still leaves a fac- crystal-liquid partitioning of REE in the trivalent
tor of two variation in Si-normalized abundance among state. The progressive decrease of ionic radius with
Rat patterns, as noted by NAKAMURA (1974). In the case increasing 2 in trivalent REE (the knthanide contracw
of the Cl chondrites, genera@ considered most representa- tion) typically produces smooth variation in partition
tive of cosmic abundances, the limited range of REE abun-
dance allows fairty precise absolute and Si-normalized
abundances to be derived; these are also presented in
Table 3, and are the values most comparable with previous
‘cosmic’ abundance estimates. It is noteworthy that four
of the nine ID patterns identified as flat by the iterative
procedure are from CI chondrites.
Figure 3 illustrates five of the nine patterns setected as
flat within 2”/, normalized to the average CI values of
Table 3. whiie Fig 4 compares the CI average with two
composites and the NAKAMURA (1974) average. The abso-
lute abundances in CI chondrites are lower than the aver-
age of the meteorites in the composites and average, but 1 ’ I , i
relative abundances show a reasonable match within ex- Lace Nd SmEu Gd Dy Er Yb Lu
perimental errors of the techniques employed. The Naka-
mura average is evidently ctosest in relative abundances Fig. 3. Examples of REE patterns showing no significant
to that derived here. The majority of the ID anafyses in fractionation relative to the average chondritic abundances
the data base, however, are not flat within experimental derived in the text. Data sources: Leedey (MASUDAet al,,
error. They form a widely varying assortment of smooth 1973); Murray, Ivuna (this study): @een’s Mercy, Orgueil‘
but not flat patterns patterns that contain anomalies of @IAKAMUFCA, 1974).

G.c.I.42/8-z
t20Q N. M. EVENSEN.P. J. HAMILTON and R. K. O’NIONS

La Ce Pr Nd San Eu Gd Tb Dy Ho Et Tm Yb Lu
Fig. 4. REE abundances in two chondrite composites (SCHNEITZLER and PHILP~. 1949; HASKIN et
al., 1970) and one chondritic average (NAKAMIJRA, 1974). normalized to the average CI chondritic abun-
dances derived in the text. Abundances of Sm and Nd in the Schnetzler and Philpotts composite
have been lowered by 11% to conform to their revised standard values (SCHNETZLERand PHtt.WTTs
1971).

coe&ients. A smooth appearance of chondrite- conden~tion sequences. BGYHTON(1975) states that

normalized REE patterns of evolved materials is com- the ionization potentials of the REE form a regular
monly taken as evidence of their derivation from sequence with atomic number and hence cannot in-
chondrite-like precursors through fractionation as tri- duce irregular fractionations, but the available data
valent ions (see reviews by HAXIN et al., 1966. 1968, are somewhat mutually inconsistent.
1970). While the variety of geochemical fractionation pro-
Deviations from this regularity of behavior occur cesses is enormous, they tend to be reducible (at least
when altervalent states are present. The divalent state as they are modelled) to a single basic concept; that
of Eu is most readily produced, and is common in of a multiphase assemblage at equilibrium, in which
moderately reducing environments. The reduced the minor elements are partitioned between the major
charge and larger ionic radius render it similar to phases. The distribution coefficients are regarded as
Ca’+ and cause it to partition into plagioclase a th~rmod~ami~lly determinate function of the
strongly relative to trivalent REE (DEBAR and phase commotions and the pressure and tem~rature
PHILPOTIS,f968). The Ce4+ ion is encountered under of the assemblage; more complex models are derived
oxidizing conditions, particularly in aqueous solution. by compounding this simple one. Wowever, manipula-
Although other altervalent ions, particularly Yb’+, tion of phases (as in fractional crystallization, for
can be formed, they require extreme conditions and example) can only amplify or combine the fraction-
there is no direct evidence for their existence in ations inherent in the relevant phase equilibria. Thus
natural systems. The electronic structures of the neu- smoothly varying distribution coefIlcients will lead to
tral lanthanides are less regular than in the trivalent smoothly varying fractionations, while anomalies,
ions. Again Eu and Yb tend to differ from other lan- once induced, will persist through subsequent more
thanides, e.g. lower heat of vaporization and higher regular fractionation processes.
atomic volume (e.g. COTTONand WILKINSON,1972). The differential incorporation of REE from various
The REE are comparatively refractory, and most nucleosynthetic regimes can produce absolute abun-
models of nebular condensation have proposed their dance anomalies, as well as apparent anomalies in
condensation as a group at high temperature, in solid both ID and NAA analyses by inva~dating the
solution within the early Ca-Al-rich condensate (e.g. assumption of uniform natural isotopic abundances.
perovskite, GRO~WAN and LARIMER,1974). BO~NT~N Nowever, extensive searches for isotopic anomalies of
(1975) examined condensation equilibria of the REE REE in meteorites (MURTHY and SOIMIT~, 1963;
in more detail, and found that their condensation EU~STERet al., 1970; LUGMAIR et al., 1975; MCCUL-
occurs over a wide range of temperature and in a LOCH et al.. 1976, 1977) have revealed only small
sequence that is not a smooth function of atomic ( < 1%) efkcts in one Allende inclusion (LUGMAIRef
number. The major points to be observed about gas- al., 19783,insuffcient to significantly affect the results
solid partitioning are: (1) the irregular form of the discussed here.
partitioning, as compared with solid-liquid or solid-
solid equilibria; and (2) the tendency for parallel be- ANOMALOUS RARE-EARTH PATTERNS
havior of the La-Cd and GdLu sub series. The gen-
eral form of the irregular ~st~bution will be charac- Identl~cation of~~~rnal~es
teristic of all fractionations induced by this type of The REE abundances in a sample as measured
equilibrium. today reflect the combined effects of all fractionations
If condensation took place from a plasma rather undergone by the sample. The unusual nature of
than neutral gas (ARRHENIUSand ALFVEN, 1971; many chondritic REE patterns demonstrates that they
ALFVENand ARRHENIUS,1976), the ionization poten- have encountered processes unlike the familiar crys-
tials of the elements become important in deter~n~ng tal-chemical fra~tionations that have operated during
 Rare-earth abundances in chondritic meteorites 1205

planetary differentiation. It would be useful to remove percent deviations from the values predicted from the
the effects of any crystal-chemical fractionation that curve are computed in the same manner as the resi-
has contributed to the observed REE patterns, and duals.
thus display more clearly the nature of any additional The fitting program has been run using as abscissa
processes that may have operated. A possible both atomic number and the ionic radii of WHIT-
approach is suggested by the fact that crystal-chemi- TAKER and MUNTUS(1970). and with both linear and
cal processes acting on trivalent REE result in frac- logarithmic ordinates. The results obtained by all
tionation patterns that are smooth functions of methods are reasonably comparable, and are in gen-
atomic number or ionic radius. The following pro- eral accord with those suggested by a more subjective
cedure was therefore applied to each analysis of the appraisal of the data. The average residuals tend to
data base: be minimized by use of atomic number and logarith-
(1) The REE abundances are normalized to the Cl mic abundances, and so these results will be used
average of Table 3. here.
(2) A third order polynomial is fitted to the normal- The ID and NAA data have been dealt with separ-
ized abundances by a least-squares procedure. This ately. The limit L was set at 3% for the ID analyses,
order of polynomial possesses sufficient degrees of a value slightly greater than their nominal precision.
freedom to fit any smooth curve likely to result from This allows the fitting routine to detect small Eu
crystal-chemical fractionation, but will not fit irregu- anomalies that are readily apparent in the plotted
lar patterns. Residuals, defined as (observed - pre- patterns. For the NAA data, values of L less than
dicted)/predicted, are computed for each element. The % 15% produce an excessive number of anomalies;
root-mean-square (RMS) residual is also calculated values from iS-25% yield a reasonable proportion of
by taking the square root of the sum of the squared non-anomalous patterns. A value of 200/, was chosen
residuals divided by F, the degrees of freedom. In- for the treatment reported here.
itially F is equal to the number of REE analyzed The results of this procedure are tabulated in Table
minus the four degrees of freedom used in the poly- 4, along with the RMS residual remaining after fitting,
nomial fitting. If ail residuals are less than some limit as a measure of the effectiveness of the smoothing
L,, the procedure is terminated at this point. procedure. Figure 5 summarizes the data for ail
(3) If any residuals greater than L are present, that detected anomalies (those exceeding 3% in ID and
element contributing most to the RMS residual (nor- 20% in NAA data).
mally the one showing the greatest deviation) is From Fig. 5 it can be seen that Eu anomalies are
flagged as anomalous, F is reduced by one, and step by far the most common, as might be predicted.
(2) is repeated with the remaining elements. This cycle Negative anomalies are about as numerous as posi-
continues until no residuals greater than L are found tive, and the total range in either direction is compar-
or F reaches a value of one. able. A large proportion of the Eu anomalies lie
(4) When the final best fit curve has been computed, within It lo”/, but most of these are probably real;

La Ce R Nd Sm Eu Gd Tb Dy Ho Er Tm Yb L

Fig. 5. REE elemental abundance anomalies computed by the least-squares curve fitting procedure
described in the text. Anomalies greater than 3% for ID data and greater than 20% for NAA data
are plotted. Letters next to the anomaly points indicate chondrite groups where known; those for
NAA data are enclosed in parentheses.
1206 N. M. EVENSEN,P. J. HAMZLTON
and R. K. O?-&QNS

Table 4. E~menial abundance anomalies of REE in &ondrites, expressed as % deviations from best-fit
po$nozniai
I-* ._--. . _- __.

TYW * Source+ FUG’ La Ce Pr Nd Sm Eu Gd Tb Dy Er Tm Yb I."

Resid.

13.4
0.7
9.6
1.4
0.7 +4
2.1 413
1.9
Boriskfno 10.6 +41 +33
Cold Bokkeveid 1.7 -%I +4
Mighei T.6 +40
a.4
I.1
1.4
1.7 +I1
7.2
1.2
9.7
0.7 -5 -42
1.5 +10
8.1
2.0 .c7
1.6 -7 -6
9.6 +40
13.9 +32 +43 t27
1.1 +15 +10 -17 -17 +203
2.1 -?
2.5 16 +R
2.2 -7
1.0 +12
Indnreh 11.3 +36 +27
1.5
2. h *9
1.7 -17 -SO
2.h -h -37 -15 -1:
11.5 +45
9.3
Jajh deh kot Lalo E6 1.R
H 1?.1
N 1.3
H3 0.9
H3 1.0 +fJ
Ii4 0.4 +14 -in -23
0.9 -22
1.6
11.1 -36
9.5 +3i
1.0 -7 +6 +1x
0.4 +8 -9
5.9 -42
0.7 -4
1.8
1.4 -?7
0.3 -I
1.0
3.9 +41 t32
1.2 +221 Sll
0.7 +311 -5 Cl3
L5 1.6 -8 +24 +R
Lh 1.7 -h -11 -11
1.h -11
Lb 6.7 +91
2.h -7
2.1
Kyushu L6 8.9 +41
Leedey L6 0.8
Modof i1905) L& -35
*17
Peace River Lh
Parnallee LL3
LL5 ilR
LL6
LLh +4
LLb
, -10 t4 "tl3
-16
? -14
0 +4

* As given in WATSON(19743.
t (1) HASKINand GEHL, 1962; (2) SCHMITTet al.. 196;; (3) SCNM~TIand SMITH, 1963; (4) SCHMllr
ef al.. 1964; (5) HASKIN et al., 1968; (6) GAST et al., 1970; (7) SCXNETZLERand BDTT~NO,1971; (8)
MASUDAet al.. 1973; (9) NAKAMURA and MASUDA, 1973; (10) NAKAMURA,1974; (11) MASUDAet a/*,
1977; (12) this work.
 Rare-earth abundances in chondritic meteorites 1207

Eu anomalies are particularly easy to detect because

of the close proximity of Sm and Gd. Eu is the only
element to commonly form isolated anomalies (Table
4).
The numerous Ce anomalies are more unexpected,
and include the two largest anomalies found, in two
fragments of Khohar (NAKAMURA and MASUDA,
1973). Positive anomalies predominate, both in size
and number; the comparative scarcity of complemen-
tary negative anomalies unless fortuitous, poses
major problems in interpretation.
Gd anomalies are also numerous, but the largest
of these appear in the NAA data and are positive.
It can be seen from Table 3 that Gd shows the great-
I ’ I I I , I t I
est discrepancy between the ID and NAA averages, LOCO Nd Sm Eu Gd Dy Er Yb Lu
being almost 20% higher for the NAA data; the posi-
tive Gd anomalies disappear from the NAA analyses Fig. 6. Contoured percentiles of the distributions of indi-
if the NAA average alone is used for the fitting pro- vidual REE abundances in chondrites, incorporating all
published ID data. Zero and 100th percentiles correspond
cedure. This is in accord with the conclusions of HAS-
to minimum and maximum observed abundances respect-
KIN et al.(1968), who found similar systematic differ- ively. Heavy dashed line is the arithmetic mean; 50th per-
ence betwen their Gd values and the analyses of centile is the median. Note that percentile contours do
Schmitt and coworkers. not represent actual patterns but merely connect points
Anomalies of other elements are fewer and more computed for each element individually.
scattered. They seldom appear in isolation, but rather
occur in combination within a given analysis, and are initial composition. A similar but, less marked effect
symptomatic of an overall irregular form of the pat- is seen in the heavy REE, which preferentially parti-
tern (Table 4). Most striking of these is the -200% tion into clinopyroxene.
Yb anomaly in Abee (NAKAMURA and MKWDA, 1973), NA~~RA f1974) has presented plots of Sm abun-
a~ompanied by pronounced irregularity of the dance vs the normalized Eu/Sm ratio for individual
lighter rare earths. However many other patterns meteorite samples, and suggests that the E, H and
show irregularity far in excess of the nominal analyti- L-LL groups define distinct correlation lines on this
cal error. diagram. If all of the analyses of our data base are
plotted in this manner, the correlations are less con-
Urigin of anomalous patterns
vincing, particularly the distinction between H and
The most common anomalies, those of ELI, may L group chondrites. However the approximate corre-
arise from fractionation of plagioclase-bearing assem- lations are at least explicable in terms of the above
blages, their presumed mode of origin in planetary comments regarding Eu fractionation if reference is
environments (SCHNETLLERand PHILPOIIX, 1968, made to Fig. 7. Correlation lines of the type de-
1970). In the case of chondrites, however, we must scribed will arise whenever the range of variation of
also consider other possible cosmochemical fraction- Eu abundance is smaller than that of Sm (or by exten-
ations. In particular, gas-solid equilibria can produce sion, the other REE). The greater the contrast in
Eu anomalies (BOYN’ION, 1975). These should be variability, the more closely defined the regression
accompained, though, by anomalous behavior of lines will appear. Furthermore, if different chondrite
other elements, particularly Yb. Table 4 demonstrates groups are characterized by different absolute abun-
that Eu anomalies frequently appear in isolation, or dances of ELI, approximately parallel lines will be
at least predominate in those samples where they are defined for each group. Figure 7 demonstrates that
present. this is exactly the case. The significance of this obser-
Further evidence bearing on the origin of these Eu vation in terms of parent body processes will be dis-
anomalies is provided by the correlation between the cussed shortly.
abundances of the other REE and the sign and mag- The other frequently anomalous element is Ce.
nitude of Eu andmalies. This results in a much Actually Ce anomalies are seldom present in isola-
smaller variation of Eu abundance relative to other tion, but can sometimes dominate other anomalies
REE (Fig. 6). In soli~liquid fractionation, the REE present (e.g. Khohar). It is tempting to attribute Ce
in general are excluded from the major mineral anomalies to the fractionation of quadrivalent Ce, but
phases found in chondrites. The bulk ~Ii~liquid dis- possible m~hanisms remain obscure. NAKAMURAand
tribution coefficients of pla~~la~-wring assem- MASUDA(1973) suggest that the high vapor pressure
blages are therefore * 1 for trivalent REE but closer of ceric oxide (CeO,), as compared with other REE
to unity in the case Qf Eu 2+. This leads to precisely oxides of the M203 form, may be significant, but do
the type of effect seen in Fig. 6, when variable not offer a possible mechanism. Sc~wrr et al. (1964)
amounts of fractionation occur in systems of similar noted positive Ce anomalies in Orgueil and the two
1208 N. M. EVENSEN,P. J. HAMIL~Nand R. K. O’NIONS

material would have to be very abundant to balance

the overall positive anomaly of these meteorites. It
is interesting to note that positive Ce anomalies have
been found in many lunar samples (MASUDAet al..
1972, 1974; LAUL et al., 1972). This raises the possibi-
lity that small differences may exist in the relative
abundances of REE in different planetary bodies.
However the anomalies are small and found in
enriched, highly fractionated materials.
None of the remaining elements displays any clear
tendency to form isolated or dominant anomalies,
with the exception of Yb in one Abee analysis (NAKA-
MURA and MASUDA,1973). These authors suggest that

the divalent state of Yb may be involved in the pro-

duction of Yb anomalies. BOYWoN (1975) has con-
cluded that divalent Yb is not significant in the en-
vironment of the solar nebular, and points out that
his model is capable of explaining the very large Yb
anomaly seen in an Allende inclusion (TANAKAand
MASUDA,1973) without appeal to the divalent state.
However, the origin of the Abee Yb anomaly remains
I.0 unclear.
08 The remaining anomalies tend to occur in com-
ia ce Nd Sr&uGd IJy Er YbLu binations within indi~dual chondrites, often includ-
ing Eu or Ce anomalies, but with no clear pattern
Fig. 7. Contoured percentiles of the distributions of indi- to their associations. if condensation equilibria of the
vidual REE abundances in published ID data for E, H type described by BOYNTON(1975) were significant in
and L group chondrites. Lightly dashed lines indicate NAA
the formation of the chondrites, gas-solid fraction-
data where it falls outside the extreme limits of the ID
data: heavy dashed lines are arithmetic means; 50th per-
ation would leave a characteristic imprint which
centile is the median. Note the greater sensitivity of the would tend to persist through subsequent reprocess-
mean to aberrant data (e.g. Yb in E. Ce in L chondritesl ing, though perhaps in a variety of forms. The uncer-
Percentile figures are omitted on the diagram for the L tainties of the thermodynamic data and the variety of
group because of the close spacing of the lines.
patterns observed in presumably fractionally con-
densed inclusions from Allende (TANAKA and
H chondrites Holbrook and Modoc. In an extensive MASUDA,1973; MARTIN and MASON, 1974; TANAKA
discussion they consider the possibility that the et al., 1975; CONARD et al., 1975) preclude a unique
Orgueil anomaly could result from differential disso- definition of a condensationally induced imprint.
lution of REE by percolation of water. The solubility Nevertheless, solid-gas equilibria remain the most
product of ceric hydroxide, Ce(OH),, is much smaller plausible mechanism for the production of irregutar
than those of trivalent hydroxides, so differential solu- patterns as opposed to the smooth fractjonations in-
tion would leave a Ce-rich residue. There are many duced by solid-solid and solid-liquid equilibria.
objections to this model, which led SCHMITTet ai. Some interesting correlations do appear between
(1964) ultimately to reject it. The most serious diffi- samples having anomalous REE patterns. Abee ana-
culty is the necessity for a highly oxidizing environ- lyses, for example, yield a variety of irregular patterns
ment to produce the Ce4+ ion. A high oxidation state (Fig. 2; SCHMITT et al., 1963; NAKAMURA and
is uncharacteristic of meteorite materials. though the MASUDA,1973). More well defined parallelisms may
volatile content of carbonaceous chondrites would be observed in the CO chondrites (Fig. 1; a similar
make them the most likely sites of such processes; downturn of La and Yb appears in the analysis of
however, they also contain highly reduced phases. Felix by SCHMITTer ai., 1963) and the L-LL group
Differential solution would require abnormally high (Bluff, and to a lesser degree Parnallee and Olivenza
starting abundances of REE, so a precipitation mech- Fig. 2). Understanding of such regularities must await
anism would seem to be required. An appropriate set- more careful examination of these objects and greater
ting for such a process is difficult to imagine. knowledge of possible fractionation mechanisms in
A further difficulty in the interpretation of the extraterrestrial environments.
observed Ce anomalies is the comparative absence
of negative anomalies. The mean of the analyses of
IMPLICA~ONS FOR SOLAR
our data base shows an _ iO% positive Ce anomaly
SYSTEM EVOLUTION
(Fig. 6). Some anaiyses of Ailende components show
small negative Ce anomalies (e.g. TANAKA and It is clear from the analyses discussed here that
MASUDA.1973; MARTINand MASON,1974). but such the concept of uniform and homogeneous abundances
 Rare-earth abundances in chondritic meteorites 1209

of REE in chondrites is without observable founda- The heterogeneity of REE within chondrites is an
tion (HASKINet al., 1970; MASUDA et al., 1973). This important observation bearing on the fractionation
may appear paradoxical in view of the widespread of solid matter following condensation. If the anoma-
and successful use of ‘chondritic abundances’ in the lous patterns are produced during condensation, then
interpretation of REE patterns observed in other at least some physical segregation of gas and solid
materials. These viewpoints can, however, be recon- must have occurred. This could resuh either from
ciled if we examine the relation between these two accretion during the conden~tjon process or from
sets of measurements. motion of the condensed grains into a region having
The interpretation of REE abundances in differen- a gas composition different from that in which the
tiated materials normally consists of inferring the grains originally formed. Either process would have
nature of crystal-chemical fractionations capable of to occur during the interval in which the REE are
producing the observed pattern from an assumed pre- condensing. Their condensation is believed to have
cursor. The smoothness of chondrite-normalized REE been largely completed at an early stage of the total
patterns (perhaps excepting Eu) implies that this pre- condensation sequence (e.g. GROSSMAN and LARIMER,
cursor did not contain si~ifi~nt elemental abun- 1974; BOYNTON,1975); if so, fractionation of REE
dance anomalies relative to normal chondritic abun- may be indicative of physical processes occurring in
dances, and that no fractionation processes capable the earliest stages of the formation of sohd matter
of producing elemental anomalies (again, excepting in the solar system.
Eu) have operated on the precursor. Furthermore, Fractionation by simultaneous condensation and
there is a tendency for the less differentiated products accretion would result in segregation of early and late
of magmatic processes on the Earth and Moon to REE condensates over distances of the order of the
exhibit relatively flat REE patterns, leading to the radius of the largest bodies accreted by the end of
conclusion that un~ff~entiated terrestrial and lunar REE conden~tion. For these fractionations to have
material possesses approximately chondritic refuter persisted in the ordinary chondrites, these initially
abundances of REE. Since the above observations accreted bodies would have to have been broken
apply to a wide variety of samples collected from apart and their inner and outer portions separated,
many regions of the Earth and Moon, and since mag- mixed individually with later bulk condensate, and
matic processes presumably sample large regions of formed into chondrites, while maintaining the segre-
their interiors, it is reasonable to conclude that the gation of early and late REE condensates. The unlike-
bulk REE abundances of these objects are approxi- lihood of this process makes it more reasonable to
mately chondritic (say within + 10% in relative abun- suppose that fractionation of REE condensates
dance; the absolute abundances cannot be inferred resulted from di~erential motion of grains and gas
to much better than a factor of 2). (CAMERON,1973). This raises the possibility that
Further information on the validity of average bodies accreting in different portions of the solar
chondrite values as representative of the bulk compo- nebula may exhibit bulk fractionation of REE relative
sition of the earth is provided in the case of Sm and to each other or to chondrites.
Nd by studies of the isotope systematics of Nd in The REE-containing early condensate represents a
terrestrial materials. The decay of 14’Sm to ‘43Nd small portion of the total solid material condensed
allows the Sm/Nd ratio in the mantle source area and would therefore contain high abundances of REE
of a terrestrial sample of known age to be deterred (e.g. TANAKAand MASUDA,1973; WXNKEet at, 1974;
from its present-day Nd isotopic composition. Recent GROSSMAN et af., 1977). Incorporation of a small car-
studies (DEPAOLOand WASSERBURG, 1976; HAMILTON rier grain bearing fractionated REE could produce
et al., 1977, 1978) have demonstrated a remarkably detectable anomalies in a much larger mass of chon-
consistent Sm/Nd ratio for the Earth’s mantle, during drite. Equilibration of the carrier phase with bulk
the Archaean, at a value very close to that derived condensate, either before or after aggregation of the
here. chondrite, would distribute REE throughout the
Earth, Moon and chondrites were formed from the major phases in accordance with the appropriate dis-
solar nebula; beyond this their genetic re~tionship tribution coefficients. The scale of hetcro~nejty in the
is uncertain. The Earth is unexcelled among accessible resulting assemblage would be a function of the fre-
objects as a representative sample of the solar nebula, quency of carrier grains and the scale of equilibration.
but its bulk composition must be inferred indirectly. Duplicate analyses of Khohar and Abee by NAKA-
The chondrites are believed to be the most represen- MURAand MA~UDA(1974) strongly imply gradation
tative samples that we can directly observe, but are from anomalous to normal REE over a scale some-
some 25 orders of magnitude smaller in mass. The what larger than the approximately gram-sized frag-
marked variation within and between chondrites ments analyzed. The anomaly-bearing phases must be
makes it difficult to determine a meaningful average. at least as widely dispersed as this.
We must therefore conclude that while infaences There is increasing evidence that chondrite forma-
from terrestrial, lunar and meteoritic samples are in tion is associated with processes occurring near the
general concordance, the relative cosmic abundances surface of meteorrite parent bodies (e.g. DODD, 1969;
of the REE remain uncertain in detail, FREDRIKSSON,1969; BUNCH and ST~FFLER, 1974;
 1210 N. M. EVENSEN,P. J. HAMILTONand R. K. O’NIONS

RAJAN, 1974; FODCJRand KEIL, 1976). However apart between about 1.3 and 1.5. The E parent body is
from the possible role of regolith dynamics in produc- clearly much lower, in accord with the conclusion of
ing the textural properties of chondrites, it is far from NAKAMURA(1974), in the range 0.8-1.2. The H parent
clear what parent body processes may have influenced body is poorly defined, but is probably in the range
their chemical properties. 1.1-1.8. The comparatively narrow range of E and
The classification of chondrites has been carried L group Eu abundances and the appreciable differ-
out on the basis of major element chemistry and ence between them suggests that (1) the material
petrology. Observed groupings of these properties are forming these parent bodies was fairly homogeneous
commonly considered indicative of derivation from to begin with (NAKAMURA, 1974); (2) effective homo-
distinct parent bodies (e.g. WASSON, 1972). and thus genization occurred following parent body accretion ;
inter-group variations may be considered due to or (3) the samples originated from small regions of
accretionally induced variations of parent body bulk the parent bodies. It is also significant in this regard
chemistry. However inn-a-group variations could well to note the narrow range encompassed by the median
represent the influence of parent body processes act- half of the abundances lying between the 25th and
ing upon a relatively uniform bulk composition. 75th percentiles in Fig. 7.
Figure 7 displays the distribution of absolute REE It is somewhat surprising to find a relatively nar-
abundances within the E, H and L chondrite groups. row range of Eu abundances accompanied by enor-
The range of variability of non-anomalous REE in- mous anomalies of Ce and Yb, since Eu is normally
creases in the sequence L. H. E, but it is not clear the element that most readily exhibits anomalous be-
whether this variation is indicative of parent body havior. The implication is that material with a high
processes or is a sampling effect. However. the range proportion of anomalous to normal REE (particularly
of Eu variability is more significant. Clearly any hypo- Eu) was introduced into some of the chondrites, either
thesis seeking to account for the observed intra- a material distinct from the parent body contribution,
group variation must explain the limited range of Eu or in a form which escaped homogenization during
abundances as compared with the other REE. parent body differentiation. Identification of anomaly-
In the earlier discussion of Eu anomalies it was bearing components, if present. is crucial to an under-
pointed out that plagioclase-bearing assemblages standing of these questions.
have distribution coefficients much closer to one for The carbonaceous chondrite groups show a narrow
divalent Eu than for the trivalent REE. If such frac- range of internal variation compared to the E. H and
tionation acted upon an unfractionated precursor (e.g. L groups (Fig. 1) but collectively span a large total
parent body) to produce the observed range of vari- range. The variation in median absolute REE abun-
ation. then the normalized REE abunances of the pre- dances (Fig. 8) parallels that in average Si (MOORE,
cursor must be bracketed by the observed range of 1971). but exceeds the Si variation in magnitude so
Eu abundances. Furthermore. the Eu abundances de- that Si-normalization reduces but does not eliminate
limit the extreme range of heterogeneity present in the variation. The comparatively narrow range in
the unfractionated starting materials. Again, similar absolute abundances within carbonaceous chondrite
arguments apply to the heavy REE in clinopyroxene groups offers additional evidence that parent bodies
bearing assemblages. The reduced variability of heavy may have been relatively homogeneous upon accre-
REE can be seen in the L and to some extent in tion and that subsequent processes produced the
the H group, and provides additional evidence that range of variation observed in the more evolved
the intra-group variation arises from crystal-chemical meteorite types.
processes. It is not possible at present to assess the
scale on which these variations are present, i.e.
whether they are characteristic of entire meteorites
or of individual samples taken for analysis. However.
most of the analyses of E, H and L group chondrites
were performed on gram-sized or larger samples. so
if internal heterogeneity is present, it must be on a
relatively coarse scale. Of course the relevant equili-
bria may be solid-solid as well as solid-liquid: the r
Fe/Si ratio in enstatite chondrites is also much more
variable than in H and L chondrites, and MASON
(1966) attributed this to metamorphic effects rather
than a primary origin. On the other hand, NAKAMURA
and MASUDA (1973), NAKAMURA (1974) and MASUDA Lo Ce Nd SmEuGd Dy Er Yb Lu
et al. (I 977) offer at least suggestive evidence for mag-
matic processes being involved in the genesis of par- Fig. 8. Comparison of the median abundances for indivi-
dual groups of chondrites, relative to the average CI abun-
ticular meteorites. dances derived in the text. Data from published ID ana-
In either case, the normalized abundances in a lyses and this work. Numbers in parentheses refer to the
hypothesized L parent body are closely delimited number of analyses represented in each group.
 Rare-earth abundances in chondritic meteorites 1211

SUMMARY BUNCHT. E. and STIFLER P. (1974) The Kelly chondrite:

a parent body surface metabreccia. Conrrih. Mineral.
1. Both relative and absolute REE abundances Petrol. 44, 157-171.
show clear and sometimes marked variation between BURBIDGEE. M., BURBIDGEG. R., FOWLER W. A. and
and within individual chondrites (HASKIN et al., 1970; HOYLEF. (1957) Synthesis of the elements in stars. Rev.
Mod. Phys. 29. 547-656.
MASUDAet a/., 1973; NAKAMURAand MASUDA, 1973 ; CAMERONA. G. W. (1959) A revised table of abundance
NAKAMLJRA,1974). The concordant relative abun- of the elements. Astrophys. .I. 129. 676-699.
dances of a number of samples indicate that ‘average CAMERONA. G. W. (1968) A new table of abundances of
chondrite’ values may reflect solar nebula abundances the elements in the solar system. In Origin and Distribu-,
but are not representative of the majority of indivi- tion of the Elements (editor L. H. Ahrens), pp. 125-143.
Pergamon Press.
dual chondrites. CAMERONA. G. W. (1973) Accumulation processes in the
2. The variations from nebular abundances primitive solar nebula. Icarus 18, 407-540.
encountered in individual chondrites cannot be attri- CONARDR. L., SCHMITTR. A. and BOYNTONW. V. (1975)
buted solely to crystal-liquid fractionation (MASUDA Rare-earth and other elemental abundances in Allende
inclusions. Meteoritics 10, 384-387.
et al., 1973; NAMAMURA and MASUDA, 1973; NAKA-
CORYELLC. D.. CHASEJ. W. and WINCHESTER J. W. (1963)
MURA, 1974). Details of processes are obscured both A procedure for geochemical interpretation of terrestrial
by the wide variation in observed patterns and by rare-earth abundance patterns. J. Geophys. Res. 68, 559.
uncertainties in the parameters of fractionation models COTTONF. A. and WILKINSONG. (1972) Advanced Inor-
ganic Chemistry. Interscience.
(BOYNTON,1975), but it appears likely that solid-gas
DEPAOL~ D. J. and WA~~ERBURG G. J. (1976) Inferences
(e.g. nebular condensation) equilibria were involved. about magma sources and mantle structure from vari-
3. The persistence of condensational anomalies in ations of ‘43Nd/‘44Nd. Geoohvs. Res. Lett. 3. 743-746.
non-carbonaceous chondrites together with their in- DODD R. T. (1969) Metamorphism of the ordinary chon-
ternal heterogeneity implies that fractional condensa- drites: a review. Geochim. Cosmochim. Acta 33, 161-203.
EUGSTERO., TERA F., BURNETTD. S. and WASSERBURG
tion was accomplished by large scale segregation of G. J. (1970) Neutron capture effects in Gd from the Nor-
grains and gas. The anomalous components escaped ton County meteorite. Earth Planet. Sci. Lett. 7, 436-440.
homogenization either by their refractory nature or FOD~R R. V. and KEIL K. (1976) Carbonaceous and non-
by being introduced into the chondrites at a relatively carbonaceous hthic fragments in the Plainview, Texas
chondrite: origin and history. Geochim. Cosmochim. Acta
late stage of their evolution. The possibility exists that
40, 177-189.
major solar system bodies have slight differences in FREDERIK~~ON K. (1969) The Sharps chondrite-new evi-
their relative bulk REE abundances. dence on the origin of chondrules and chondrites with
4. With the exception of individual abundance analyses by E. Jarosewich and J. Nelen. In Meteorite
Research (editor P. M. Millman), pp. 155165. Reidel.
anomalies, the range in REE abundances within each
GAST P. W., HUBBARDN. J. and WEISMANNH. (1970)
of the E, H, L and LL groups is consistent with the Chemical composition and petrogenesis of basalts of
expected nature of magmatic processes operating on Tranquility Base. In Proc. Apollo I I Lunar Sci. Co&
homogeneous starting materials. The inferred abso- pp. 1143-I 163. Pergamon Press.
lute REE abundances of the starting materials (e.g. GROSSMANL. and LARIMERJ. W. (1974) Early chemical
parent bodies) are distinctly lower in the E group history of the solar system. Rev. Geophys. Space Phys.
12. 71-101.
chondrites than in the ordinary chondrites. The car- GROSWAN L., DAVISA. M., OLSEN E. and SANTOLIQUID~
bonaceous chondrite groups display appreciably less P. M. (1977) Chemical studies of condensates in the
internal variation. Murchison type 2 carbonaceous chondrite. In Lunar
Science VIII, pp. 377-379. Lunar Science Institute.
Acknowledgements-We are particularly indebted to M. Houston.
PRINZ of the American Museum of Natural History for HAMILTONP. J., O’NIONS R. K. and EVENSENN. M. (1977)
providing the majority of the meteorite specimens analyzed Sm-Nd dating of Archaean basic and ultrabasic vol-
in this study. We also wish to thank W. S. BROECKER. canics. Earth Planet. Sci. Lett. 36. 263-268.
P. HAMLYNand S. R. CARTERfor helpful comments and HAMILTONP. J., O’NIONSR. K., EVENSENN. M.. BRIDGE-
discussion, and W. V. BOYNTON,L. A. HASKIN,N. NAKA- WATERD. and ALLAARTJ. H. (1978) Sm-Nd investiga-
MURA,J. A. F%ILPO~~S and R. A. SCHMITTfor their careful tions of the Isua supracrustals, West Greenland: implica-
reviews of the manuscript. R. COHEN and G. WE~XLMANN tions for mantle evolution. Nature 272. 41-42.
provided valuable technical assistance. This work was sup HASKINL. A.. FREYF. A., SCHMITTR. A. and SMITHR. H.
ported by NASA grant NSG-7200. L-DGO Contribution (1966) Meteoritic, solar and terrestrial rare-earth distri-
No. 2693. butions. Phys. Chem. Earth 7. 167-320.
HASKINL. A. and GEHL M. A. (1962) The rare-earth distri-
REFERENCES bution in sediments. J. Geophys. Res. 67. 2537-2541.
HASKINL. A.. HASKINM. A., FREY F. A. and WILIJEMAN
ALFVENH. and ARRHENIUS G. (1976) Evolution of the solar T. R. (1968) Relative and absolute terrestrial abundances
system. NASA pub. 345. of the rare earths. In Origin and Distribution of the Elr-
ALLEN R. O., JR. and MA.~~NB. (1973) Minor and trace mews (editor L. H. Ahrens). pp. 889-912. Pergamon
elements in some meteoritic minerals. Gcochim. Cosmo- Press.
chim. Acta 37, 1435-1456. HASKIN L. A., HELMKEP. A., PASTERT. P. and ALLEN
ARRHENIUSG. and ALFVENH. (1971) Fractionation and R. 0. (1970) Rare earths in meteoritic, terrestrial. and
condensation in space. Earth Planer. Sci. Letr. 10. lunar matter. In Activation Analysis in Geochemistry und
253-267. Cosmochemistry (editors A. 0. Brunfelt and E. Steinnes).
BOYNTONW. V. (1975) Fractionation in the solar nebula pp. 201-219. Universitetsforlaget. Oslo.
condensation of yttrium and the rare-earth elements, HONDAM. and SHIMAM. (1967) Distribution of rare earths
Geochim. Cosmochim. Acta 39. 569-584. among component minerals of Bruderheim chondritc.
I212 N. M. EVENSEN,P. J. HA~~ILT~Nand R. K. O’NIOMS

Earth Plawt. Sci. Lrtt. 2. 344-348. SCHMITTR. A. and S~trtr R. H. (1963) Implications of
HOOKER P.. O’NIONS R. K. and PANKHURSTR. J. (1975) similarity in rare-earth fractionation of nakhiitic meteor-
Determination of rare-earth elements in USGS. stan- ites and terrestrial basalts. Nature 199. 5s.551.
dard rocks by mixed-solvent ion exchange and mass SCHMITTR. A.. SMITHR. H.. LATCHJ. E., MOSENA. W,,
spectrometric isotope dilution. Chetn. Grol. 16. 189-196. OLEHY P. A. and VASILEVSKIS J. (1963) Abundances of
HUBHARDN. J. and GAST P. W. (1971) Chemical composi- the fourteen rare-earth elements, scandium and yttrium
tion and origin of nonmare lunar basalts. Proc. Second in meteoritic and terrestrial matter. Geochim. Cosmochim.
Lunar Sci. Co@. Ge~hjrn. Cosrn~hjt?~. Acta Suppl. 2, pp. Acra 27. 577-622.
999-1020. MIT. SCHMI~ R. A.. SMITHR. H. and OLEH~ D. A. (1964) Rare-
LAULJ. C.. WAKITAH.. SHOWALTER D. L.. BOY~JT~N W. V. earth. yttrium and scandium abundances in meteoritic
and SCHMITTR. A. (1972) Bulk rare earth and other and terrestrial matter-II. Geochim. Cosmochim. Acfu 28.
trace elements in Apollo I4 and IS and Luna I6 samples. 67-86.
Proc. Third Lunar Sci. COI$. Geochim. Cosmochim. Acta SCHMITTR. A.. SMITHR. H. and QLEHY D. A. (1968) Rare
Suppl. 3. pp. I I8 I -I 200. MIT. earth abundances in meteoritic chondruies. In Origin and
LUGMAIRG. W.. SCHEININ N. B. and MARTI K. (19753 Disrrihution of rhe Elements (editor L. H. Ahrens), pp.
Search for extinct ‘46Sm--I. The isotopic abundance of 273-282. Pergamon Press.
r4’Nd in the Juvinas Meteorite. Ear& Pkmef. Sci. Luff. S~HNETZLER C. C. and B~TTINO M. L. (1971) Some alkali.
27, 79-84. alkaline-earth and rare-earth element ~ncentratjons and
LUGMAIRG. W.. MARTI K. and SCHEININN. B. (1978) In- the Rb-Sr age of the Lost City mereorite and separated
complete mixing of products from r-. p-. and s-process phases. J. Geophys. Res. 76, 4061-4066.
nucleosynthesis: Sm-Nd systematics in Aliende inclusion SCHNE+ZLERC. C. and PHILPOTTSJ. A. (1968) Partition
EK l-04- I. In Lunur and Platwtary Science IX. pp. coefficients of rare earth elements and barium between
672-674. Lunar and Planetary Institute. igneous matrix and rock-forming mineral pheno-
MARTINP. M. and MACONB. (1974) Major and trace ele- trysts--I. In Origin and Distribution of the Elements
ments in the Aliende meteorite. Narurc 249. 333-334. (editor, L. H. Ahrens), pp. 929-939. Pergamon Press.
MACONB. (I 966) The enstatite chondrites. Geochim. Costno- SCHNETZLER C. C. and PHILPO~TSJ. A. (1969) Genesis of
chin,. Acru 30. 2339. the calcium-rich achondrites in light of rare-earth and
MASUVA A. (1962) Regularities in variation of relative barium ~n~ntrations. In meteorite Research (editor
abundances of lanthanide elements and an attempt to P. M. Millman). pp. 206216. Springer.
anaiyse separation-index patterns of some minerals. J. SC~INETZLER C. C. and PHILPOTT~J. A. (1970) Partition
Earrh Sci.. Nagoya Unir. IO. i73- 187. coefficients of rare earths between igneous matrix mater-
MASUVAA.. NAKAMURAN.. KURASAWAH. and TANAKA ial and rock-forming mineral phenocrysts-II. Geochim.
T. (1972) Precise determination of rare-earth elements Cosmochiru. Acra 34, 331-340.
in the Apollo I4 and I5 samples. Proc. Third Lunar S~HNETZLERC. C., THOMASH. H. and PHILPOTTSJ. A.
Sci. ConJ. Geochim. Cosmochim. Acta Suppl. 3. pp. (1967) Determination of rare-earth elements in rocks and
1307-1313. MIT. minerals by mass spectrometric, stable isotope dilution
MASUVAA.. NAKAMURAN. and TANAKAT. (1973) Fine technique. Anal. Chem. 39. 1888-1890.
structures of mutually normalized rare-earth patterns of SHIMIZUN. (1974) An isotope dilution technique for anaiy-
chondrites. Ge~hj~)l. Cosmochitn. Acfu 37, 239248. sis of the rare-earth elements. Carnegie Inst. Wash. Yrbk
MASUVAA., TANAKAT.. ASAKURAJ. and SWMIZUH. (1977) 73, 941-943.
REE. Rb, Sr and Ba abundances in Yamato 0). (k) and SUESS H. E. and UREY H. C. (1956) Abundances of the
(m) meteorites. Antarcfic Record 5&, 197-203. elements. Rev. Mod. Phys. 28, 53-74.
MASUDAA.. TANAKAT., NAKAMURAN. and KURASAWA TANAKAT. and MASUDAM. (1973) Rare-earth elements
H. (1974) Possible REE anomalies of Apollo I7 REE in matrix, inclusions and chondruies of the Aliende
patterns. Proc. Fifth Lunar Sci. Con&. Geochim. Cosmo- meteorite. Icarus 19. 523-530.
chim. Acta Suppl. 5. pp. 1247-1253. Pergamon Press. TANAKAT., NAKAMURAN., MASUDAA. and ONUMA N.
MCCULL~CHM. T., DE LAETERJ. R. and R~%AN K. J. R. (1975) Giant oiivine chondruie as a possible later stage
(1976) The isotopic com~sit~on and elemental abun- product in the nebula. Nature 256, 27-38.
dance of lutetium in meteorites and terrestrial samples TAYLORS. R. (1960) The abundance of the rare-earth eie-
and the “6Lu cosmochronometer. Earrh Planer. Sci. ments in relation to their origin. Geochim. Cosmochim.
Left. 28, 308-322. Acfa 19, IO&112.
MCCULL~CHM. T., ROSMANK. J. R. and DE LAETERJ. R. VAN SCHMIJSW. R. and HAYESJ. M. (1974) Chemical and
(1977) The isotopic and elemental abundance of ytter- petrographic correlations among carbonaceous chon-
bium in meteorites and terrestrial samples. Geochim. drites. Geochim. Cosmochim. Acta 38, 47-64.
Cosmochim. Acta 41, 1703-l 708. VAN SCHMUS W. R. and WOOD J. A. (1967) A chemicai-
MOORE C. B. (1971) Silicon. In Handbook of Elemental petrologic classification for the chondritic meteorites.
Abundances in Meteorites (editor B. Mason), pp. &ochi& Cosmochim. Acta 31, 747-765.
125-129. Gordon and Breach. WAKITA H. and SCHMITT R. A. (1970)
. Rare-earth and other
MURTHY V. R. and !SCHMITT R. A. (1963) Isotope abun- elemental abundances in the Allende meteorite. Nature
dances of rare-earth elements in meteontes-I.- Imphca- 227. 478479.
tions of samarium, europium, and gadolinium to the WANKEH., BADDENHAUSEN II., PALMEH. and SPELL B.
early history of the solar system. J. Geophys. Res. 68, (1974) On the chemistry of the Ailende inclusions and
911-917. their origin as high temperature condensates. Earth
NAKAMURA N. (1974) Determination of REE, Ba, Fe, Mg, Planet. Sci. Lert. 23. i-7.
Na and K in carbonaceous and ordinary chondrites. WA~.S~NJ. T. (1972) Formation of ordinary chondrites.
Geochim. Cosmochim. Acta 38, 757-775. Rev, Geophys. Space Phys. 10, 71 i-759.
NAKAMURAN. and MA~UDA A. (1973) Chondrites with WASSONJ. T. (1974) Meteorites-Clussificarion and Proper-
peculiar rare-earth element patterns. Earth Planet. Sci. ties. Springer.
Lert. 9, 359-364. WHITTAKER E. J. W. and MUNTUS R. (1970) Ionic radii
RAJANR. S. (1974) On the irradiation history and origin for use in geochemistry. Geochim. Co~rn~~im. Acra 34,
of gas-rich metedrites. Geochim. Cosmocbim. Acta 38. 945-956.
777-788.