SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES ? NUMBER 25
Inclusions
in the Allende Meteorite
Brian Mason
and S.R. Taylor
ISSUED
OCT 51982
SMITHSONIAN PUBLICATIONS
SMITHSONIAN INSTITUTION PRESS
City of Washington
1982
ABSTRACT
Mason, Brian, and S.R. Taylor. Inclusions in the Allende Meteorite. Smith-
sonian Contributions to the Earth Sciences, number 25, 30 pages, 25 figures, 5 tables,
1982.?Six discrete groups of inclusions have been distinguished in the Allende
meteorite. Groups I, V, and VI are mostly melilite-rich chondrules, although
some have been extensively altered to fine-grained aggregates; Groups II and
III are mostly fine-grained aggregates made up largely of spinel and fassaite;
Group IV are olivine-rich aggregates and chondrules. Each group has a
distinctive trace-element pattern, most clearly shown by the rare-earth (RE)
distribution pattern. Group I has an unfractionated pattern (except for a
small positive Eu anomaly) at about 10-15 times chondrites; Group II has a
highly fractionated pattern with depletion of the heavier lanthanides (Gd-Er)
and negative Eu and positive Tm and Yb anomalies; Group III has an
unfractionated pattern at about 20 times chondrites, except for negative Eu
and Yb anomalies; Group IV has a relatively unfractionated pattern at 2-4
times chondrites; Group V has an unfractionated pattern at 10-20 times
chondrites; Group VI has an unfractionated pattern at 10-20 times chondrites,
except for positive Eu and Yb anomalies (i.e., complementary to Group III).
The complex patterns of trace element distribution in these Allende inclusions
indicate a complex history of formation of this meteorite from the solar nebula.
OFFICIAL PUBLICATION DATE is handstamped in a limited number of initial copies and is recorded
in the Institution's annual report, Smithsonian Year. SERIES COVER DESIGN: Aerial view of Ulawun
Volcano, New Britain.
Library of Congress Cataloging in Publication Data
Mason, Brian Harold, 1917-
Inclusions in the Allende meteorite.
(Smithsonian contributions to the earth sciences ; no. 25)
Bibliography: p.
1. Allende meteorite. I. Taylor, Stuart Ross, 1925-. II. Title. III. Series.
QE1.S227 no. 25 [QB756.A44] 550s [523.5'1] 82-600091 AACR2
Contents
Page
Introduction 1
Acknowledgments 2
Experimental Methods 2
Results 3
Group I 7
Group V 12
Group III 12
Group VI 16
Group II 17
Group IV 20
Platinum-Group Elements 20
Th/U Ratios 20
La/Th Ratios 20
Zr-Hf-Y-Er 20
Unclassified CAIs 22
Discussion 24
Literature Cited 28
in
Inclusions
in the Allende Meteorite
Brian Mason
and S.R. Taylor
Introduction
The fall of the Allende meteorite in northern
Mexico on 8 February 1969 was a unique event
in the history of meteoritics. One of us (BM) was
in the field immediately after the event, and even
a cursory examination of the first specimens re-
covered showed that the meteorite was a Type 3
carbonaceous chondrite but with components
never before seen?notably giant chondrules (Fig-
ure 1) which proved to have a remarkable mineral
association (melilite-titanian fassaite-spinel-anor-
thite) and white to pink irregular aggregates. The
Smithsonian Institution field party collected
about 150 kg of meteorites and distributed ma-
terial to scientists in 37 laboratories in 13 coun-
tries within a few weeks of the date of fall. Clarke
et al. (1970) published an account of the field
occurrences and gave a comprehensive account of
the morphology and composition of the meteor-
ite. They distinguished the following major com-
ponents: (a) matrix, (b) chondrules, (c) irregular
aggregates, (d) dark inclusions. The chondrules
were divided into two groups on the basis of
chemical and mineralogical composition: (I) Mg-
rich olivine, sometimes with clinoenstatite and
interstitial glass (similar to the common chon-
Brian Mason, Department of Mineral Sciences, National Museum of
Natural History, Smithsonian Institution, Washington, D. C. 20560.
S.R. Taylor, Research School of Earth Sciences, Australian National
University, Canberra, A.C.T., Australia 2600.
drules in Type 3 meteorites), and (II) Ca- and Al-
rich chondrules (first recognized as a significant
component in Allende, although a similar chon-
drule had previously been described (Christophe
Michel-Levy, 1968) from the Vigarano C3 chon-
drite) . Chemical analyses were given of the bulk
meteorite, matrix, Mg-rich chondrules, two
Ca,Al-rich chondrules, a Ca,Al-rich aggregate,
and a dark inclusion.
The availability of large amounts of Allende
and the concurrent distribution of lunar samples
resulted in extensive and intensive research on
this meteorite, so that the literature on Allende
now probably outweighs that on any other me-
teorite. Researchers have concentrated on the
Ca,Al-rich inclusions (both chondrules and ag-
gregates), partly because their compositions sug-
gest that they might be early condensates from
solar nebula and partly because of the isotopic
anomalies recognized therein (first oxygen (Clay-
ton, Grossman, and Mayeda, 1973) and later
other elements). Research on other components,
however, has not been neglected; see, for example,
publications on dark inclusions (Fruland et al.,
1978) and on Mg-rich chondrules (Simon and
Haggerty, 1980).
The useful acronym CAI has been widely
adopted to refer collectively to Ca,Al-rich chon-
drules and aggregates in Allende (and other C3
meteorites). Its use obviates the sometimes subjec-
1
SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
FIGURE 1.?Allende 3529-G, a melilite-fassaite-spinel-anorthite chrondrule, 25 mm in diameter;
note the size contrast with the small (~1 mm) olivine-rich chondrules in the surrounding
matrix.
tive judgment whether a specific inclusion crys-
tallized as a molten droplet (possibly plastically
deformed and/or fragmented) or originated as an
aggregate of solid particles.
ACKNOWLEDGMENTS.?We thank Ms. P. Os-
wald-Sealy, Mr. J.M.G. Shelley, and Mr. N. Ware
for assistance in the analytical work. One of us
(BM) enjoyed the hospitality of the Max-Planck-
Institut fur Kernphysik and the advice and help
of Dr. A. El Goresy during part of this research.
The Smithsonian Research Foundation contri-
buted part of the costs of the research, and this
contribution is gratefully recorded.
Experimental Methods
A large number of Allende specimens in the
collections of the National Museum of Natural
History (NMNH) were inspected, and those
showing prominent and readily extractable chon-
drules and aggregates were selected for detailed
examination. From these, over 40 chondrules and
NUMBER 25
aggregates have been extracted and subjected to
complete chemical and mineralogical analysis,
and many others investigated to a lesser degree.
We point out, however, that our selection was
biased towards the larger and more easily ex-
tracted inclusions and cannot be considered as
statistically representative of the total variety of
Allende inclusions. This has been clearly demon-
strated by the work of Kornacki (1981). He iden-
tified 189 fine- to medium-grained inclusions in
17 Allende thin sections but found none of the
melilite-rich chondrules on which so much re-
search has been published. Most of our inclusions
were extracted from NMNH 3529, a 32-kilogram
stone; in this stone, as in most Allende specimens,
the inclusions are irregularly distributed, as can
be seen in Figure 2. This figure shows that some
pieces from a single stone may contain one or
more of these large melilite-rich chondrules,
whereas others have none. Each piece of Allende
that we examined had many CAIs that were too
small for our planned investigation.
As far as possible the CAIs were extracted from
the matrix by simple mechanical tools and hand-
picking to avoid introducing contaminants; how-
ever, in some instances methylene iodide was used
to separate the denser matrix from the less dense
CAIs. Mineral identification and analysis were
carried out using microscope, X-ray diffraction,
and electron microprobe techniques. Composi-
tional data for major elements and some minor
elements were obtained by microprobe analyses
of glasses prepared by fusion. The trace element
data were obtained by spark source mass spec-
trometry (SSMS), using the procedures described
by Taylor (1965, 1971) and Taylor and Gorton
(1977). The precision is about ?5% and the ac-
curacy about ?5%, but they vary somewhat,
being poorer for elements below mass number
105 than for those of higher mass number. Recent
data show ?2% agreement with isotope dilution
mass spectrometric data for Sm and Nd. Some
trace elements could not be determined because
their concentrations were below the sensitivity of
the technique. Among the lanthanides, Lu was
not determined because of its use as an internal
standard. The single mass number (169) for Tm
is subject to interference from a multiple carbon
ion, and this element cannot be accurately deter-
mined at low concentrations (<0.3 ppm). The
169 line, however, was unexpectedly strong in
Group II samples, indicating the presence of a
positive Tm anomaly. The rare earth (RE) data
were normalized to chondritic abundances given
by Taylor and Gorton (1977).
The platinum-group elements, normally below
detection limit in the SSMS technique, were ac-
cessible in these samples. After extensive testing
and calibration, the following mass numbers were
found to provide reliable analytical data: 101Ru,
103Rh, 106Pd, 185Re, 190Os, 191Ir, and 195Pt. All other
mass numbers had various interferences.
Results
The results of our chemical analyses are sum-
marized in Table 1. In the discussions which
follow, these results are combined with the data
on 20 Allende CAIs published by Mason and
Martin (1977). They determined major and mi-
nor elements in 17 CAIs from the Allende mete-
orite and distinguished three groups on the basis
of RE distribution patterns: Group I (10 melilite-
fassaite-spinel-anorthite chondrules) had unfrac-
tionated chondrite-normalized RE distribution
patterns at 10-15 times chondritic abundances
with a positive Eu anomaly; Group II (5 aggre-
gates) with La-Sm abundances similar to Group
I but with rapidly diminishing abundances of the
heavier RE, except for a positive Tm anomaly;
Group III (2 aggregates) with unfractionated RE
abundances about 20 times those of chondrites,
except for large negative Eu and Yb anomalies.
(Group IV comprised olivine-rich chondrules and
aggregates with relatively unfractionated RE
abundances at about three times chondrites; CaO
and AI2O3 are relatively low (<10%), and hence
Group IV samples are not discussed in this paper.)
Taylor and Mason (1978) analyzed 18 CAIs
(Table 1) and extended this classification with
Group V (unfractionated RE distributions with
no significant anomalies) and Group VI (RE
SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
* .*
FIGURE 2.?Four pieces of Allende 3529, illustrating the irregular distribution of large CAI
chondrules, CAI aggregates, and dark inclusions.
NUMBER 25
SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
TABLE 1.?Analytical data on Allende CAIs by group and sample (4-digit sample numbers are
NMNH catalog numbers; letters and 2-digit sample numbers are individual CAIs from NMNH
3529)
Constituent
SiO2
TiO2
AI2O,
FeO
MgO
CaO
Na2O
Ba
Pb
Th
U
Th/U
Zr
Hf
Zr/Hf
Nb
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
Tm
Yb
2 RE
Y
Eu/Eu*
La/Yb
Eu/Yb
Y/Er
Zr/Y
Pt
Ir
Os
Re
Pd
Rh
Ru
I
Z
30.86
1.21
28.85
0.30
9.25
29.35
0.18
75
7.2
0.58
0.09
6.4
97
2.35
41
3.75
5.36
12.8
2.03
9.74
2.77
L37
3.62
0.67
4.62
1.17
3.55
0.45
3.38
52
34.8
1.34
1.6
0.41
9.8
2.8
7.8
7.4
7.0
0.57
1.00
1.09
8.9
40
20.46
0.57
52.35
5.80
11.39
6.78
2.67
8.6
0.34
0.88
-
2.0
-
-
0.13
14.6
38.2
5.74
29.4
7.62
0.29
2.87
0.50
2.41
0.13
0.14
1.04
0.014
103
0.99
0.20
1040
21
7.1
2.0
-
-
-
-
-
-
5284
29.70
0.58
39.40
4.47
11.29
12.54
1.96
12.5
0.84
0.72
0.030
24.3
3.6
0.10
36
0.94
8.46
23.0
3.20
16.8
5.17
0.35
2.77
0.49
2.62
0.15
0.12
0.80
0.58
64.5
1.51
0.29
14.6
0.60
12.5
2.4
0.98
0.87
0.37
_
0.42
-
1.27
II
3655B
32.76
0.98
35.95
2.63
9.83
17.20
0.57
52
1.52
0.57
0.025
22.1
7.3
0.18
41
1.82
9.21
22.7
3.44
18.7
5.38
0.52
3.57
0.58
3.12
0.25
0.28
0.74
1.36
69.9
3.43
0.37
6.8
0.38
12.3
2.1
0.62
0.35
0.27
-
0.21
0.15
0.73
43
23.58
0.58
42.0
7.9(1
17.29
7.67
1.46
19
1.35
0.65
0.033
19.5
3.43
_
_
0.64
8.72
23.1
3.98
18.3
5.41
0.40
3.07
0.55
2.81
0.16
0.18
0.88
0.36
68
1.85
0.30
24
1.1
10.2
1.85
0.72
0.37
0.35
-
-
0.34
1.23
4691
19.75
1.64
41.86
0.73
7.54
28.41
0.18
50
-
1.57
0.17
9.2
10.3
0.40
26
4.37
20.3
_
7.7
37.8
11.0
1.04
5.26
0.96
4.95
0.28
0.37
1.92
3.25
_
3.80
0.42
6.21
0.32
10.3
2.7
-
-
1.00
0.86
1.86
5242
35.05
0.66
23.90
13.18
12.76
14.01
0.44
O 42
PERCENT
26.46
2.32
36.25
3.52
8.86
22.94
0.21
17.95
0.79
51.14
2.15
11.53
16.26
0.28
PARTS PER MILLION
11
1.32
0.21
0.057
3.7
5.3
0.15
33
1.59
3.3
12.8
1.53
6.81
2.30
0.35
0.48
0.09
0.53
0.07
0.18
0.38
1.10
29.9
1.77
_
3.0
0.32
9.8
3.0
42
_
1.10
0.10
11
95
3.20
30
3.64
7.51
19.4
2.68
14.1
4.12
1.09
6.15
1.13
7.80
1.95
5.94
0.85
4.52
77
41
23
0.82
0.35
0.041
8.4
-
_
_
1.43
3.66
9.50
1.46
7.52
2.19
0.50
2.89
0.54
3.61
0.85
2.60
0.37
1.67
37
24
RATIOS
0.69
1.7
0.24
6.9
2.3
0.64
2.2
0.30
9.2
-
III
41
26.67
1.63
37.12
0.98
14.24
19.19
0.21
51
1.06
0.66
0.10
6.7
81
2.40
34
4.03
5.20
14.4
2.01
10.0
3.35
0.77
4.02
0.81
5.38
1.29
4.00
0.60
3.45
55
31
0.66
1.5
0.22
7.8
2.6
PLATINUM-GROUP ELEMENTS (PPM)
0.76
0.66
0.63
_
0.82
0.21
1.56
10.7
8.0
7.5
0.52
1.55
0.91
8.86
24
19.8
19
1.50
0.55
2.01
28.4
12.9
11.4
10.4
0.84
0.84
1.37
15.1
46
26.44
2.33
36.03
3.13
7.46
24.10
0.46
58
1.28
1.66
0.13
13
108
3.30
33
5.64
11.6
_
4.30
22.0
6.37
1.33
8.10
1.58
10.6
2.68
7.74
1.11
5.29
_
50
0.57
2.2
0.25
6.5
2.2
10.9
8.8
7.7
0.90
1.75
0.82
13.3
21
27.35
1.27
33.15
2.43
11.62
23.93
0.25
36
3.3
0.55
0.12
4.7
69
2.05
34
5.83
5.25
13.2
1.88
9.88
2.85
1.00
3.58
0.68
4.67
1.14
3.30
0.48
3.02
51
28.7
1.0
1.74
0.33
8.7
2.4
18.7
9.33
8.89
0.75
1.27
2.54
16.2
V
3655A
29.51
1.78
29.68
1.00
11.36
26.49
0.19
37
1.24
0.62
0.092
6.8
112
2.74
41
5.41
4.89
13.3
1.99
10.4
3.00
1.01
4.19
0.78
5.36
1.30
4.12
0.62
3.85
55
42
0.88
1.3
0.26
10.2
2.7
11.8
10.7
9.85
0.80
1.17
1.15
13.3
3682
29.58
1.42
27.66
4.10
10.71
26.28
0.22
46.5
0.52
0.093
5.7
102
2.92
35
3.56
4.08
11.2
1.60
8.31
2.65
0.89
3.56
0.68
4.63
1.18
3.97
0.47
2.80
-
33
0.90
1.5
0.31
8.3
3.1
11.3
4.48
4.41
0.37
1.42
1.85
8.6
45
26.34
1.38
34.26
1.82
4.05
32.04
0.07
46
1.41
0.64
0.11
5.6
88
2.13
41
7.02
6.36
-
2.59
13.9
4.20
1.53
5.63
1.03
7.39
1.64
4.30
0.73
4.2
-
40
1.0
-
0.36
9.3
2.2
10.7
12.6
10.5
0.94
2.29
0.91
20
44
27.57
1.34
35.56
1.12
6.05
27.95
0.37
46
2.1
0.39
0.095
4.2
91
2.07
44
5.31
4.24
11.9
1.54
8.06
2.30
1.31
3.00
0.59
3.96
0.98
3.00
0.39
3.42
45
37
1.54
1.24
0.38
12.3
2.5
20
8.4
7.2
0.57
1.10
2.84
15.3
VI
Y
30.31
1.50
28.72
0.87
10.79
27.59
0:23
37
2.6
0.48
0.20
2.4
90
1.97
46
7.80
4.43
13.6
1.65
8.53
2.65
1.14
3.24
0.62
4.51
1.06
3.45
0.46
4.44
50
38
1.22
0.99
0.26
11.0
2.4
15.1
9.1
9.6
0.97
1.85
1.32
10.7
47
32.78
1.13
27.10
7.94
9.77
20.54
0.62
27
2.0
0.71
0.43
1.7
67
1.91
35
7.47
4.75
12 A
1.67
8.21
2.50
0.95
3.20
0.63
4.18
0.99
2.94
0.45
4.59
48
26
1.0
1.03
0.21
8.8
2.6
14.8
7.9
7.1
0.62
1.34
1.51
12.3
patterns the reciprocal of Group III, about 15
times chondrites with positive Eu and Yb anom-
alies). They noted, however, that major and trace-
element distributions for Groups I, III, V, and VI
show many similiarities. The application of this
RE classification is illustrated in Figure 3. The
CAIs clearly cluster in discrete groupings; the
only discrepancy is 4698, a Group III inclusion
which plots among the Group II inclusions be-
cause of an anomalously low Yb content.
Since the CAIs are characterized by their high
Ca and Al contents, it is of interest to examine
this relationship in detail, especially since Ahrens
and von Michaelis (1969) have shown that the
Ca/Al ratio (by weight) is almost constant at 1.09
in chondrites and many achondrites. This rela-
tionship for the Allende CAIs is shown in Figure
4; the point for Allende (bulk) falls on the 1.09
line, but the points for the individual CAIs scatter
widely. The points for Groups I, V, and VI cluster
fairly closely above or just below the 1.09 line
(except 36, which has a unique mineralogy con-
sisting largely of andradite). Group II and III
CAIs plot below and generally well away from
NUMBER 25
60h
40
30
20[-
La/Yb
10
8
6
4
3
2h
1 -
-
?40
B4692
B43
-
B5284
n4698 ?37
B3598
-4691
3655BB"
B3803
D3593
46 D D42
D?
D41
i I ?
O21
3682Q Q45
O3655A
? 47
I 1
GROUP
?
?
o
?
?
3658* i
1
II
III
V
VI
33
Z
171
0.2 0.4 0.6 0.8 1.0
Eu/Eii*
1.2 1.4 1.6
FIGURE 3.?Plot (semi-log) of La/Yb ratio vs. Eu/Eu* (europium anomaly) for Allende CAI
groups (omitting Group IV) (4-digit sample numbers are NMNH catalog numbers; G, O, Y, Z,
and 2-digit sample numbers are individual CAIs from NMNH 3529).
the 1.09 line and show a much wider range of Ca
and Al contents than the other groups. The in-
dividual groups will now be discussed in detail.
GROUP I
Group I CAIs appear to be relatively restricted
in chemical and mineralogical composition (Fig-
ures 3, 4) and are readily identified since they are
all coarse-grained (grain size 0.5-3 mm) chon-
drules consisting largely of melilite and titanian
fassaite, with minor amounts of spinel and anor-
thite (Figure 5). Ten have been chemically ana-
lyzed, of which the data on nine (3529-26, 29, 30,
31, 32, 33, G; 3658; 3898) were reported by
Mason and Martin (1977), and for the tenth
(3529-Z) the analytical data are given in Table 1
(5171, which was reported as a Group I CAI by
Mason and Martin (1977) has been reclassified
as Group VI). The RE data are summarized in
Figure 6 for minimum (3529-31), average of 10,
and maximum (3529-Z) values. RE data on CAIs
by Gast, Hubbard, and Wiesmann (1970) and by
Wanke et al. (1974) fall within the range of these
Group I inclusions.
The mineralogical composition of the Group I
inclusions is also relatively uniform. The data in
Table 2 have been derived from microscopic ex-
amination of thin sections, microprobe analyses
of the minerals, and the bulk compositions in
Table 1. The principal variable in the mineral-
ogical composition is the amount and composi-
SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
35
30
25
c
a>o
a>20a.
4-?
O)
|15
o
(0
o 10
5
-
_
GROUP
? 1
? II
a III
o v
? VI
-
_
*36
-
/
/
0Allende
/^ (BULK)
S 1 I
,3898 /
O45 /
,31 ,29 /
? 5171 #Z -3 XT B4??1
v* * / *
* 3655Ae >^
3682 3658 x^
/^O Q21 D46
? 32/ DO
/ D41
. O/ ,3655B
\^X B5242
? ?^\/ B5284
/ B4692 D3593 Q 4698 D42
.37
B3803 43
,40
1 1 1 1 1 1 1 1 1
10 15 20 25 30 35
AI2O3, weight percent
40 45 50 55
FIGURE 4.?The relationship between CaO and AI2O3 in individual samples (sample
identifications as in Figure 3).
tion of the melilite. The pyroxene is highly alu-
minous (AI2O3, 16%-22%) and titanian (TiO2,
3.9%?19.1%); a detailed account of these pyrox-
enes was published by Mason (1974). Spinel and
anorthite have compositions close to stoichiomet-
ric MgAl2O4 and CaAl2Si2O8. Melilite shows al-
teration along grain boundaries to a fine-grained
material consisting largely of grossular, sometimes
with minor nepheline; these minerals probably
contain the minor amounts of FeO and Na2O
shown by the bulk analyses.
Two of these Group I inclusions (3529-29, 3898)
contain minor amounts (<5%) of rhonite, whose
unit-cell formula is Ca4Xi2(Si, Al)i2O40, where X
includes Mg, Fe, Ti, and Al. Rhonite and fassaite
in 3529-29 are illustrated in Figure 7, and micro-
probe analyses of these minerals are given in
Table 3. Rhonite in Allende CAIs is distinctly
different from terrestrial rhonite in its high con-
tent of Al, Ti, and V and its low Si and Fe
content. The high vanadium content is especially
remarkable, since the content of this element in
the bulk meteorite (90 ppm) is not exceptional.
In meteoritic rhonite and fassaite the titanium is
evidently present as both Ti3 and Ti4 (Dowty and
Clark, 1973; Fuchs, 1978). In the formulas cal-
culated from the analyses in Table 3, the titanium
is arbitarily calculated as equal amounts of Ti3
and Ti4; this procedure results in excellent agree-
ment with the structural formulas.
Figure 7 shows that the rhonite is rimmed and
veined by black opaque material. This material
has been described as a fassaite-spinel-perovskite
symplectite by Haggerty (1977) and El Goresy,
Nagel, and Ramdohr (1977). Haggerty interprets
this symplectite as a decomposition product of
NUMBER 25
FIGURE 5.?Microphotograph (transmitted light) of 3529-G, a Group I melilite-rich chondrule
25 mm in diameter; melilite is present as large prismatic crystals, fassaite and anorthite occur
as equant grains interstitial to the melilite, and spinel is disseminated as small grains throughout
the other minerals.
rhonite, whereas El Goresy and his coworkers
infer simultaneous precipitation of fassaite + per-
ovskite from a silicate liquid around pre-existing
rhonite and spinel. The Ca and Ti are present in
rhonite in the correct proportions to produce
perovskite in considerable amounts; some CAIs
show patchy concentrations of perovskite grains
suggestive of pre-existing rhonite.
Haggerty (1977:389) stated: "These cores (of
rhonite) are unquestionably the earliest constitu-
10 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb
FIGURE 6.?RE distribution patterns for Group I CAIs (31 = minimum RE concentrations,
Av = average of 10 Group I CAIs, Z = maximum RE concentrations).
FIGURE 7.?Allende 3529-29: rhonite (darker gray, black borders) and fassaite (lighter
gray) enclosed in melilite; idiomorphic spinel grains, average 0.05 mm, are present in
melilite and fassaite.
NUMBER 25 11
TABLE 2.?Mineralogical composition of Group I inclusions
(estimated weight percentages of minerals; range and mean
Ak content of melilite)
Inclusion
3529-G
3529-Z
3529-26
3529-29
3529-30
3529-31
3529-32
3529-33
3658
3898
Melilite
45
50
50
60
50
70
35
60
50
75
Fassaite
25
25
25
15*
20
15
35
20
15
10*
Spinel
20
15
5
15
25
10
25
15
20
10
Anor-thite
10
10
20
5
5
5
5
5
15
5
Ak range
and
mean
28-40, 34
18-43, 29
11-65, 42
18-35, 26
21-40,28
18-33, 24
44-70, 65
25-48, 31
22-32, 28
13-25, 19
* Includes rhonite.
ent in the CAT (Ca-Al-Ti-rich inclusion)." Our
observations suggest the reverse, that rhonite was
the last mineral to crystallize. The sequence of
crystallization appears to be spinel-melilite-fas-
saite-rhonite. As Clarke et al. (1970:39) pointed
out, the sequence spinel-melilite-fassaite is that
predicted from studies of crystallization of melts
in the CaO-MgO-Al2O3-SiO2 systems and is sup-
ported by textural relations in the melilite-rich
chondrules, in which small euhedral crystals of
spinel are included within all the other minerals;
melilite is present as large zoned crystals (indicat-
ing crystallization over a considerable tempera-
ture interval), whereas fassaite occurs as smaller
xenomorphic grains in the interstices of the mel-
ilite crystals. These relations are present in 3529-
29 and 3898, with the rhonite enclosed in the
fassaite. The extreme enrichment of Ti and V in
the rhonite probably represents a concentration
of these elements in the residual liquid after
practically all the material of the chondrule had
crystallized as spinel and melilite, both essentially
free of these elements. Boivin (1980) has demon-
strated that in terrestrial rocks rhonite is a late-
crystallized phase, after spinel and pyroxene.
Data on some of these Group I inclusions have
been reported in the literature. The 39Ar-40Ar
systematics of 3529-Z have been studied by Her-
zog et al. (1980), who find it contains excess 40Ar
over that produced by in situ decay of 40K. Nie-
meyer and Lugmair (1980) report the Ti isotopic
composition in this inclusion. Jessberger et al.
(1980) have published K-Ar data on 3529-26. El
Goresy et al. (1979) have reported a number of
accessory phases in "Fremdlinge" in 3529-29, in-
cluding a calcium phosphate containing Mo and
Ru, and Fe- and V-rich oxides; in this inclusion
Jessberger et al. (1980) have reported K-Ar data,
El Goresy and Ramdohr (1980) have reported Os
depletion in "Fremdlinge," and Jordan et al.
(1980) have found an unusual Xe isotopic com-
position. Mason and Martin (1974) separated
melilite and pyroxene from 3529-32 and analyzed
these minerals for major, minor, and trace ele-
ments; they recorded high contents of Mo and
the Pt elements (except Pd) in both minerals and
suggested that these elements were present as
minute grains of native metal, a suggestion that
TABLE 3.?Rhonite and fassaite compositions in Allende
inclusions (nd = not determined)
Constituent
SiO2
TiO2
AI2O3
Cr2O3
v2o3
FeO
MnO
MgO
CaO
Na2O
Total
1
15.58
23.15
28.38
0.20
3.09
0.18
0.02
15.18
13.86
nd
99.64
2
16.12
19.80
28.57
0.26
3.32
2.80
0.03
14.97
14.07
0.03
99.97
3
30.50
18.64
21.19
0.06
0.83
0.03
0.04
5.68
23.80
nd
100.77
4
30.23
19.05
20.27
0.00
0.85
0.05
0.02
5.85
23.74
0.02
100.08
Column 1: Rhonite, 3529-29; calculated formula
Ca3.88 (Mgr,.91 Ti'$2.27Ti42.27 Alo.79 Vo.6f>-
Column 2: Rhonite, 3898; calculated formula
Ca3.95 (Mg5.85 Ti3i.95Ti4i.95 AI1.03 V0.70 Feo.6i)-
(Si4.23Al7.97)O40
Column 3: Fassaite, 3529-29; calculated formula
Cao.97(Mgo.32Ti 0.27Ti 0.27AI0.11V0.03)-
(Sii.i6Al0.84)O6
Column 4: Fassaite, 3898; calculated formula
Cao.98 (Mgo.MTi' 0.27 Ti 0.27 Alo.07 V0.03)-
(Sii.i6Al0.84)O6
12 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
has since been confirmed. Wark and Lovering
(1980) have provided a detailed petrographic
description of 3529-33, including the minute
metal grains rich in Pt-group elements; this inclu-
sion (identified as Dl) was analyzed by Gray,
Papanastassiou, and Wasserburg (1973), who
showed that it had a very low 87Sr/86Sr ratio;
Jessberger et al. (1980) have published K-Ar data.
Material from 3898 (their D7) was analyzed by
Gray, Papanastassiou, and Wasserburg (1973),
who reported compositions of melilite and pyrox-
ene, contents of Rb and Sr, and 87Sr/86Sr ratios;
Rb is extremely low (0.01-0.02 ppm) and 87Sr/
Sr is the lowest yet measured, identifying this
material as a very early condensate from the solar
nebula. Niederer, Papanastassiou, and Wasser-
burg (1980) found Ti isotope anomalies, and
Becker and Epstein (1981) consider it a FUN
inclusion (inclusions showing isotopic fractiona-
tion (F) and unidentified nuclear (UN) effects),
48, 17/anomalous in Ca, O, and O.
GROUP V
Group V inclusions were distinguished by Tay-
lor and Mason (1978) on the basis of their RE
distribution pattern, which is essentially unfrac-
tionated with no significant anomalies (Figure 8).
Four such inclusions have been recognized, and
their mineralogical compositions are given in
Table 4. In major-element chemistry and in min-
eralogical composition and texture they are prac-
tically indistinguishable from Group I inclusions.
TABLE 4.?Mineralogical composition of Group V inclusions
(estimated weight percentages of minerals; range and mean
Ak content of melilite)
Inclusion
3529-21
3529-45
3655A
3682
Melilite
30
70
30
40
Fassaite
40
20
50
40
Spinel
20
10
10
10
Anor-thite
10
0
10
10
Ak range
and
mean
17-47, 33
4-15, 9
12-56, 24
20-43, 26
Inclusion 3529-45 does have some unique fea-
tures; it consists largely of Al-rich melilite (note
the low Ak content given in Table 4), and the
fassaite is associated with small areas of inter-
grown perovskite, spinel, and hibonite (Figure
9)?this mineral association is highly suggestive
of decomposed rhb'nite. Minor and trace element
concentrations (Table 1) are similar to those in
Group I inclusions.
Steele and Hutcheon (1979) have published an
Al-Mg isochron determined by ion microprobe
on coexisting spinel, melilite, and hibonite in
3529-45. Herzog et al. (1980) have reported min-
eral compositions and 39Ar-40Ar systematics on
3655A.
GROUP III
Group III inclusions are characterized by an
unfractionated RE distribution pattern at 10-
40 times chondrites, except for negative Eu and
50
- (A
- 0)
ch o
- O
? 3655-A
o 3529-21
? 3682
? 3529-45
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb
FIGURE 8.?RE distribution patterns for Group V Allende CAIs.
NUMBER 25 13
FIGURE 9.?Spinel (cubes)-hibonite (laths)-perovskite (white) intergrowth in Allende
3529-45 (reflected light); this association suggests an origin by the decomposition of
rhonite.
Yb anomalies of equal magnitude (Figure 10).
The first pattern of this kind was described by
Tanaka and Masuda (1973); this was determined
on a portion of 3529-O. Mason and Martin (1977)
described two Group III inclusions (3593 and
4698). We have studied four Group III inclusions,
and their mineralogical compositions are given in
Table 5. In major-element composition they are
similar to Group I inclusions, except for higher
TABLE 5.?Mineralogical composition of Group III inclu-
sions (estimated weight percentages of minerals; range and
mean Ak content of melilite)
Inclusion
3529-O
3529-41
3529-42
3529-46
Melilite
40
30
40
55
Fassaite
40
30
15
20
Spinel
20
20
35
20
Anor-thite
0
20
10
5
Ak range
and
mean
25-40, 35
22-39, 30
1-4, 3
3-12, 6
AI2O3 and lower CaO contents. Group III inclu-
sions, however, show a variety of textural rela-
tionships, from coarse-grained chondrule-like ob-
jects to fine-grained irregular aggregates; some
show complex internal rimming associated with
the alteration of melilite to grossular (Figure 11).
Inclusion 3529-42 has areas extremely rich in
hibonite (Figure 12); inclusion 3529-46 has areas
rich in perovskite, suggestive of the decomposition
of rhonite (Figure 13).
Minor and trace element concentrations in
Group III inclusions (Table 1) are generally sim-
ilar to those in Group I inclusions (except for the
Eu and Yb anomalies).
Lorin and Christophe Michel-Levy (1978) have
described metal grains, vanadium oxide inclu-
sions, and Mg isotopic anomalies in 3529-42.
Wark (1981a) has published analyses of nephe-
line, sodalite, anorthite, spinel, grossular, andrad-
ite, and pyroxene from 3593 and estimated its
primary and altered composition.
14 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
50
20
10
? 3529-0
? 3529-42
D 3529-41
? 3529-46
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb
FIGURE 10.?RE distribution patterns for Group III Allende CAIs.
FIGURE 11.?Photomicrograph (transmitted light) of Allende 3529-46, a Group III CAI,
consisting largely of melilite (white) and spinel, the melilite much altered to grossular
(gray, fine-grained). Note extensive internal rims.
NUMBER 25 15
FIGURE 12.?Photomicrograph (reflected light) of a portion of Allende 3529-42, a
Group II CAI, showing hibonite (laths) and spinel (cubes) in melilite.
FIGURE 13.?Photomicrograph (reflected light) of a portion of Allende 3529-46,
showing a concentration of perovskite (white, grains ~ 2-4 micrometers).
16 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
GROUP VI
Group VI inclusions were characterized by
Taylor and Mason (1978) as having flat RE
distribution patterns except for positive Eu and
Yb anomalies (Figure 14). These patterns are
complementary to those of Group III, although
the Eu and Yb anomalies are not so pronounced
in the Group VI inclusions. Inclusions 5171 and
3529-36, described in Mason and Martin (1977)
as Group I and unclassified, should also be in-
cluded in Group VI. Group VI inclusions show a
wide range in chemical and mineralogical com-
positions and in textures. Inclusion 3529-Y is a
melilite-rich chondrule, quite similar to the nu-
merous Group I chondrules, as is inclusion 5171.
Inclusion 3529-44 is an irregular medium-grained
white inclusion (possibly a deformed chondrule)
50
- CO
ch o
- O
o 3529-Y
? 3529-44
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb
FIGURE 14.?RE distribution patterns for Group VI Allende CAIs.
FIGURE 15.?Photomicrograph (transmitted light) of Allende 3529-47, a Group VI CAI,
showing melilite (white) largely altered to grossular-andradite (gray, fine-grained) and
rimmed by diopside.
NUMBER 25 17
?^w. x, 3^ " * Plte
FIGURE 16.?Photomicrograph (transmitted light) of Allende 3529-36, a Group VI CAI,
showing andradite (dark gray to black) with lighter areas of nepheline, sodalite, and
melilite.
consisting largely of melilite (mean composition
AIC27) which is extensively altered to grossular.
Inclusion 3529-47 is an unzoned aggregate, 15
mm in maximum dimension, consisting largely of
gehlenitic melilite (mean composition Ak3); the
melilite is extensively altered to grossular or an-
dradite and is rimmed by diopside (Figure 15).
Inclusion 3529-36 is a large rounded aggregate,
possibly a deformed and altered chondrule; it is
unique in Allende inclusions so far examined in
consisting largely of andradite (Figure 16), possi-
bly pseudomorphous after melilite or fassaite.
Except for the Eu and Yb anomalies, the trace-
element concentrations in Group VI inclusions
are similar to those in Group I inclusions.
Becker and Epstein (1981) have reported a 29Si
deficiency in 3529-Y.
GROUP II
Group II inclusions show a highly fractionated
RE distribution pattern, characterized by the
light RE (La-Sm) at 10-50 X chondrites, a neg-
ative Eu anomaly, a very rapid decline in relative
abundances from Gd to Er, a strong positive Tm
anomaly (Tm ? Sm), and Yb abundance com-
parable with that for Eu (Figure 17). This pattern
was first recognized by Tanaka and Masuda
(1973) in their Allende inclusion G (inclusion
3598 in Mason and Martin (1977)). Mason and
Martin described five Group II inclusions, and
we have found five more (Table 1).
All the Group II inclusions except one are fine-
grained (average grain size 0.01-0.04 mm) irreg-
ular aggregates (Figure 18), white to pink (the
pink color usually indicates a richness in spinel).
The exception is 4691, a large irregular inclusion
(maximum dimension 15 mm) consisting largely
of melilite (Ak5_i2, average Ak7) with abundant
spinel inclusions, and a little fassaite and anor-
thite (Figure 19). Group II inclusions show a wide
range in major-element composition (Table 1 and
Figure 4); most of them show relatively high
alkali and FeO contents (Na2O up to 4.7%, FeO
up to 13%). The wide range in major-element
composition is reflected in a wide range in min-
18 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
50 -
20 -
10 -
5 -
2 -
1 -
.5 -
I a
?*? i
ites
- o
- .c
~ a>
- a.
E" (0
CO
-
-
O D-___
^ 3529-40 1
o 3655-B
? 3529-43
a 4691
? 5242
* 5284
H
t \
/H I\f \ I X\ \\ /# \\
1 1 1 1 1 1 1 I
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb
FIGURE 17(upper).?RE distribution patterns for Group II
Allende CAIs.
FIGURE 18(right).?Photomicrograph of Allende 3529-37, a
fine-grained Group II CAI, consisting of fine-grained melil-
ite, fassaite, and spinel, with some nepheline and sodalite;
black material is fine-grained matrix.
eralogical composition. Modal composition can-
not be precisely determined because of the fine-
grained nature of these inclusions; microscopic
examination, X-ray diffraction, and microprobe
data show that the minerals include major mel-
ilite, fassaite, and spinel, minor nepheline and
sodalite and possibly anorthite, and accessory
perovskite and hibonite; grossular and grossular-
andradite are usually present, possibly as altera-
tion products of melilite. Figure 20 illustrates the
texture of a typical fine-grained Group II inclu-
sion, as revealed in the scanning electron micro-
scope. It is markedly porous and consists of an
aggregate of jUm-sized anhedral mineral grains,
NUMBER 25 19
FIGURE 19.?Photomicrograph (transmitted light) of Allende 4691, a coarse-
grained Group II CAI, consisting of melilite (white) enclosing innumerable
micrometer-sized spinel grains; dark material is surrounding matrix.
FIGURE 20.?Scanning electron micrograph of Allende 4692, a fine-grained
Group II CAI; white is largely spinel and pyroxene, with some hibonite; gray
is nepheline and sodalite, black is pore space.
20 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
sometimes aggregated into chains; the whole ap-
pearance is suggestive of a flocculated mass of
dust particles.
As well as their characteristic RE patterns,
Group II inclusions are distinctly different from
other GAIs in other trace elements. They are
relatively depleted in Y, Zr, Hf, Nb, and the
platinum-group elements.
Niederer, Papanastassiou, and Wasserburg
(1981) have published Ti isotopic data, and Chen
and Wasserburg (1981) report on U-Pb isotopic
compositions in 3529-40. For 4691 Wark (1981b)
has estimated primary and altered compositions;
Jessberger et al. (1980) have published K-Ar data;
and Niemeyer and Lagmair (1980) have reported
Ti isotopic compositions.
GROUP IV
We have no data on Group IV inclusions be-
yond those published by Mason and Martin
(1977). As mentioned previously, these olivine-
rich inclusions are relatively low in Ca and Al
and on that account should perhaps be excluded
from the CAIs.
PLATINUM-GROUP ELEMENTS
(Data from Table 1 only)
Although the RE have proven most useful in
distinguishing among the various classes of Al-
lende inclusions, other trace elements show some
characteristic differences. The Group II inclusions
contain abundances of Ru, Rh, Pd, Os, Ir, and
Pt at about chondritic levels (our analytical tech-
nique was not sensitive enough to detect Re at
this level).
In contrast, Groups I, III, V, and VI are en-
riched in the refractory trace elements Ru, Re,
Os, Ir, and Pt at 10-20 times chrondritic levels.
They show a relative depletion in Rh (5 X Cl
abundances) and particularly in Pd (2-3 X Cl
abundances) (Figure 21). There is a good corre-
lation between the relative abundances and the
condensation temperatures for these elements in
Groups I, III, V, and VI (Figure 22). Thus the
element fractionations are clearly linked to rela-
tive volatility. It is curious that the Group II
inclusions show about chondritic levels of the Pt-
group elements, despite the evidence of extreme
fractionation evident in the RE patterns.
TH/U RATIOS
(Data from Table 1 only)
The analytical data for these elements are both
precise and accurate by the SSMS technique.
Wide variations among the ratios are found
among the inclusions (Table 1). Only in Group
VI is the ratio chondritic (3.8) or below. Groups
I, III, and V have ratios ranging from 4.7-13 with
little overlap between the groups. One Group II
inclusion (5242) has a chondritic ratio, but the
others show values up to 24.3. This variation is
attributed mainly to changes in the abundance
levels of uranium relative to the more refractory
element thorium.
LA/TH RATIOS
(Data from Table 1 only)
Thorium appears to behave in a very similar
manner to the light RE, La, in Groups I, III, V,
and VI. The La/Th ratio is 8.50 ? 1.45, which is
essentially equivalent to the chondritic ratio of
8.51. The correlation coefficient (r) for 12 samples
is 0.963, illustrating that there is no fractionation
between these elements. In contrast, the Group II
inclusions (6 samples) have La/Th = 13.50 ? 1.5,
with a correlation coefficient of 0.969. In these
inclusions La is thus enriched uniformly over Th
by a factor of 1.6.
ZR-HF-Y-ER
(Data from Table 1 only)
The extreme behavior of Er in the Group II
inclusions makes it of interest to identify other
FIGURE 21 (upper).?Platinum-group element distribution
patterns in Allende CAIs.
FIGURE 22(lower).?Correlation between relative abun-
dances and condensation temperatures for platinum-group
elements in Allende CAIs.
NUMBER 25 21
0.1
Ru Rh Pd Re Os Ir Pt
30 _
'?? 10 I
o
O
Eo.
a
?a
a
1.0
1
-
IIIII
iii i
i
i i
Pt
4
Pd
i
i i
i
Ru
<
I
i
-
> <
Ir
i
i
Os
I I I I 1
I ill
i
1000 1500 2000
Condensation Temperature K
22 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
elements showing similar characteristics. The par-
allel behavior of Y to the heavy RE in many
geochemical processes also makes it useful to com-
pare its distribution in the Allende inclusions. In
Groups I, III, V, and VI, Y correlates with the
heavy RE, since they have patterns parallel to
those of chondrites. In Group II, Y correlates with
Er but not with neighboring Ho or Tm. Accord-
ingly, Y and Er must have similar volatilities.
The Y/Er ratio is constant at 9.4 ? 1.8 for all
inclusions. The correlation coefficient r=0.945 for
18 samples. The chondritic ratio for Y/Er is 9.39,
and, accordingly, yttrium and erbium show a
closely coherent relationship during cosmochem-
ical processes.
Other elements which show a close correlation
with Y and Er are Zr and Hf. Zr/Y is constant
for all inclusions at 2.34 ? 0.30. Our Zr data by
SSMS are less precise than those for Hf, but the
Zr/Hf ratio is 36 ? 4.9 for all inclusions. This
value is not distinguishable from the chondritic
ratio of 33, the difference being due to the less
precise nature of our Zr values. Accordingly, these
four elements must have similar volatility char-
acteristics and remain locked together in cosmo-
chemical processes.
UNCLASSIFIED CAIS
We have found some CAIs, and others have
been described in the literature, that appear to
be different from any of those described above.
Unfortunately, for most of them no RE data are
available.
Inclusion 3509 is a chalky-white chondrule, 8
mm in diameter (Figure 23). It was found in one
of the first Allende specimens collected and was
described by Clarke et al. (1970:39), as follows:
A single chondrule of type c has been found. It was observed
on a broken fragment of the meteorite, being unusually large
(8 mm in diameter) and a uniform chalky-white color. It is
so fine-grained that the individual minerals could not be
identified in grain mounts in immersion oils. X-ray powder
photographs provided positive identification for sodalite,
nepheline, clinopyroxene (probably fassaite), and olivine,
and this mineralogy is consistent with the bulk chemical
composition of the chondrule (Table 3, column 8). By the
X-ray procedure of Yoder and Sahama (1957) the compo-
sition of the olivine was found to be Fa26, in good agreement
with the FeO/(FeO+MgO) ratio in the chemical analysis.
This occurrence of sodalite (NasAleSieC^C^, with 7.3 per-
cent Cl), the first recorded from a meteorite, is particularly
intriguing, since the chlorine content of stony meteorites is
usually low (of the order of 100 ppm) and combined as
chlorapatite. Microprobe analyses have confirmed the pres-
ence of nepheline and sodalite, and have provided some
significant compositional data; in particular, potassium is
notably fractionated between the two minerals, the nephe-
line containing 1.5 percent K and the sodalite less than 0.1
percent. The extremely fine-grained texture of this chondrule
suggests that it solidified as a glass and subsequently devi-
trified.
Grossman and Steele (1976:150) commented that
"Clarke et al. (1970) gave a very accurate descrip-
tion of an amoeboid aggregate but they classified
it as a type c Ca-Al-rich chondrule" [their italics].
Since they never saw the specimen, their reclas-
sification of it as an amoeboid aggregate is evi-
dently based on their interpretation of the above
description; however, this inclusion is certainly
not amoeboid.
The phase composition, estimated from the
chemical analysis and the minerals present, is
(weight percent): olivine (Fa26) 35, nepheline 30,
sodalite 20, pyroxene 15. Mason and Martin
(1977) determined minor and trace elements; the
RE concentrations are low (at about average
chondritic abundances, but with a positive Eu
anomaly (Eu/Eu* = 2.9). This inclusion does not
fall in any of the previously described RE groups.
Wasserburg and Huneke (1979) have shown
that this inclusion is extremely enriched in 129Xe
evidently produced from 129I, presumably incor-
porated in the halogen-rich mineral sodalite.
Unfortunately, despite many searches through
large collections of Allende material, no further
specimens of this unique chondrule type have
been found.
Inclusion 3510 has been observed only in thin
section, so no chemical data are available except
for microprobe analyses of the individual min-
erals. It is a chondrule, 2.4 mm across (Figure 24),
FIGURE 23.?Allende 3509: a, chondrule, 8 mm in diameter,
in matrix; b, scanning electron microphotograph showing
olivine (light gray, radiating), sodalite (black, interstitial),
and diopside (light gray, blocky, at lower left).
NUMBER 25 23
E N T E R
24 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
FIGURE 24.?Photomicrograph (transmitted light) of anor-
thite-fosterite-spinel chondrule (2.4 mm across) in Allende
3510.
consisting of anorthite, Ang6 (80%), forsterite, Fai
(15%), and spinel (5%). From the composition of
the minerals and the mode, the bulk composition
(weight percent) is SiO2, 40.3; A12O3, 32.7; FeO,
0.2; MgO, 10.4; CaO, 16.1; Na2O, 0.3, which
clearly puts it in the CAI classification.
Fuchs (1969) described an irregular inclusion,
1.0 X 0.7 mm, consisting of cordierite with alu-
minous enstatite, anorthite, spinel, olivine, and
sodalite. This is the first, and so far the only,
record of cordierite in a meteorite. Repeated
searches through a large collection of Allende
specimens have failed to find another specimen.
Cordierite, anorthite, and spinel are liquidus
phases in the 10% MgO plane which approxi-
mates the composition of many Allende CAIs,
and this inclusion may represent a slightly more
SiO2-rich composition than other CAIs.
Several CAIs with apparently unique miner-
alogical compositions have been mentioned in the
literature, and investigations of their RE distri-
bution patterns may reveal additional groupings.
Discussion
Grossman (1975:443) proposed a classification
of coarse-grained Allende CAIs into two types, as
follows:
Type A inclusions contain 80-85 per cent melilite, 15-20 per
cent spinel, 1-2 per cent perovskite and rare plagioclase,
hibonite, wollastonite and grossularite. Clinopyroxene, if
present, is restricted to thin rims around inclusions or cavities
in their interiors. Type B inclusions contain 35-60 per cent
pyroxene, 15-30 per cent spinel, 5-25 per cent plagioclase
and 5-20 per cent melilite. The coarse pyroxene crystals in
Type B's contain 15 per cent AI2O3 and 1.8% Ti, some of
which is trivalent.
Grossman's classification has been widely, if
somewhat uncritically, accepted by researchers
on Allende CAIs; however, it is extremely restric-
tive in its compositional limits, as is shown in
Figure 25. The requirement that Type A inclu-
sions contain at least 80% melilite means that
CaO in the bulk composition must be 32% or
greater (since melilite contains 40%-41% CaO).
We have plotted on Figure 25 the estimated
mineral composition of our coarse-grained CAIs,
none of which falls within the limits set by Gross-
man for his Type A and Type B inclusions. Either
his classification is too restrictive, or it is not
applicable to the coarse-grained inclusions we
have studied. Unfortunately his RE data (Gross-
man and Ganapathy, 1976) are inadequate to
correlate his Type A-Type B classification with
the RE groups defined in this paper.
Since the first descriptions of CAIs in the Al-
lende meteorite there have been lively discussions
as to their genesis. Marvin, Wood, and Dickey
(1970) described several CAIs and suggested that
they represented early, high-temperature conden-
sates from the solar nebula; they noted the simi-
larity between the chemistry and mineralogy of
the CAIs and the sequence of compounds calcu-
lated by Lord (1965) to be among the highest
temperature condensates to form in a solar ne-
bula. This suggestion has been adopted and elab-
orated by several researchers, most notably by
Grossman (1973, 1980), who concluded (1980:
584) that "the trace element results are thus very
powerful evidence that coarse-grained inclusions
formed in a gas-condensed phase fractionation
process." Other experts, e.g., Kurat (1970), have
concluded that CAIs are residues of an evapora-
tion process, whereby more volatile components
have been lost, resulting in the observed enrich-
NUMBER 25 25
Melilite
Pyroxene
FIGURE 25.?Mineralogical composition of coarse-grained Allende CAIs and the com-
positional limits of Type A and B Allende inclusions as defined by Grossman (1975).
Rest
ment in the refractory elements. This process has
been elaborated by Kurat, Hoinkes, and Fred-
ricksson (1976) and by Chou, Baedecker, and
Wasson (1976).
The variety and complexity of the Allende
CAIs, and particularly the existence of distinct
groups characterized by unique RE distribution
patterns, indicate a complex origin. It is likely
that their genesis was equally complex, and that
neither fractional condensation nor fractional
evaporation is the complete answer.
The Group I inclusions comprise a highly co-
herent group, very uniform in chemical and min-
eralogical composition. They are all coarse-
grained melilite-rich chondrules, and the texture
and crystallization sequence of the primary min-
erals strongly supports an origin as liquid silicate
droplets. The major elements?Si, Al, Mg, Ca,
Ti?are those predicted to condense at high tem-
peratures in a cooling solar nebula (or to be
residual in a fractional evaporation sequence).
The small amounts of Fe and Na shown in bulk
analyses of Group I inclusions were probably
introduced in a gas phase after the inclusions had
solidified; in all of them the melilite is partly
altered along grain boundaries to a fine-grained
material consisting largely of grossular, often with
some nepheline. Group I inclusions show a uni-
form enrichment in RE and other refractory trace
elements at 10-20 times chondritic abundances
26 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
and a small positive Eu anomaly, probably linked
to the affinity of the element for the melilite
structure. The minimum 87Sr/86Sr ratios recorded
in some of them is additional evidence for their
early formation, yet it is difficult to equate them
with direct condensates from a cooling solar ne-
bula. Most experts (e.g., Grossman, 1972) believe
that such condensation must be gas-solid, not gas-
liquid (although Blander and Katz (1967) have
argued for nonequilibrium condensation in the
form of supercooled liquid droplets). Gas-solid
condensation would produce a dust of refractory
solids, which must then be segregated from the
cooling nebula and melted to give the chondrules.
In this connection the giant size (up to 25 mm
diameter) of the melilite-rich chondrules, in com-
parison to the common olivine-rich chondrules
(average 1 mm diameter), may be significant. It
indicates that some minimum amount of refrac-
tory-rich condensate had to accumulate before
melting took place and suggests that some special
factor may have been involved?an internal heat
source such as postulated by Clayton (1980).
Group V inclusions are very similar to Group
I inclusions in major and trace elements, and in
mineralogy, but lack the positive Eu anomaly.
They are melilite-rich chondrules, indistinguish-
able from Group I chondrules, except for 3529-
45, which is a nonspherical melilite-rich inclusion,
possibly a deformed chondrule.
Group III inclusions are more diverse in texture
and composition, but all show characteristic neg-
ative Eu and Yb anomalies superimposed on a
flat RE distribution pattern at 10-40 times chon-
dritic abundances. This pattern is understandable
in terms of the relatively volatility of the RE
elements; Eu and Yb are the most volatile of
these elements, and the observed pattern could
be the result either of fractional condensation or
fractional evaporation. The Group III inclusions
comprise both coarse-grained (3529-O, 3529-41,
3529-42, 3529-46) and fine-grained (3593, 4698)
types; the fine-grained inclusions are notably
richer in alkalies (Na2O, 1.3%-1.8%) than the
coarse-grained inclusions (Na2O, 0.21%-0.46%).
Possibly the coarse-grained inclusions were
formed from fine-grained precursors by heating
and the loss of alkalies, without affecting the
distribution of the more refractory elements.
Group VI inclusions show RE distribution
patterns essentially complementary to those for
Group III inclusions, i.e., undifferentiated except
for small positive Eu and Yb anomalies (the Eu
and Yb anomalies are not so pronounced as those
in the Group III inclusions). Group VI inclusions
are more varied in texture and mineralogy than
other groups; inclusions 3529-Y, 5171, and prob-
ably 3529-44 are coarse-grained melilite-rich
chondrules, whereas 3529-36 and 3529-47 are
fine-grained aggregates. These inclusions range
widely in major-element compositions; the mel-
ilite-rich chondrules are low in FeO, whereas the
fine-grained aggregates are high in this compo-
nent. The texture of 3529-36 suggests that it is a
metasomatically altered melilite-rich chondrule,
whereby the melilite and fassaite have been al-
tered to an aggregate of nepheline, sodalite, and
andradite; this alteration has not affected the
concentration of minor and trace elements (ex-
cept Ba, which is anomalously low).
Group II inclusions present the most intriguing
problems of all the Allende CAIs. They are all
fine-grained aggregates (except 4691) and show
a wide range in major-element compositions.
Nevertheless, the RE distribution patterns are
unique and remarkably uniform, except for 5242,
which has considerable lower RE concentrations
than the others (Figure 17). Inclusion 4691 is
unique among Group II inclusions in several
respects. It is the only coarse-grained such inclu-
sion; its major-element composition is distinctive
in having the highest CaO and TiC>2 and the
lowest S1O2, FeO, MgO, and Na2O contents; and
it has the highest RE concentrations. All these
features strongly suggest that 4691 may be a
Group II inclusion that has experienced a high
degree of fractional evaporation, whereby more
volatile elements have been lost with the produc-
tion of a refractory residue, consisting largely of
melilite and spinel (note that the melilite is almost
pure Ca2Al2SiO7, the refractory end of the melil-
ite series).
Boynton (1975, 1977) has provided an exten-
sive discussion of the significance of the Group II
NUMBER 25 27
RE distribution. He concluded (1977:184) that
The strong depletion of most of the heavier RE and the
peculiar Eu, Yb, and Tm anomalies are all related to the
volatility of the RE and are readily understood in terms of
an equilibrium gas-solid fractionation. Because the Group II
RE pattern requires a gas/solid fractionation and because it
is depleted in the most refractory RE, these aggregates must
indeed be condensates since refractory residues would be
expected to be enriched in the most refractory RE.
He has since found the complementary RE pat-
tern in a CAI from the Murchison meteorite
(Boynton, Frazier, and Macdougall, 1980).
Wood (1981) has recently presented a scenario
for the origin and early evolution of the solar
system which is based to a large extent on his
interpretation of the significance of the CAIs in
the Allende meteorite. He considers that the CAIs
are unlikely to be condensates but are distillation
residues, and that therefore the depletion of
super-refractory elements in Group II CAIs can-
not have been accomplished by fractional con-
densation in the solar nebula. He concludes that
this depletion, and presumably a number of other
properties of the components of primitive meteo-
ritic material, must be relics of pre-solar-system
fractionations among different populations of in-
terstellar dust grains.
Literature Cited
Ahrens, L.H., and H. von Michaelis
1969. The Composition of Stony Meteorites, III: Some
Inter-Element Relationships. Earth and Planetary
Science Letters, 5:395-400.
Becker, R.H., and S. Epstein
1981. Silicon and Oxygen Isotopes in Some Selected
Allende Inclusions. In Lunar and Planetary Science,
XII, pages 56-58. Houston: Lunar and Planetary
Institute.
Blander, M., and J.L. Katz
1967. Condensation of Primordial Dust. Geochimica et
Cosmochimica Acta, 31:1025-1034.
Boivin, P.
1980. Donnees experimentales preliminaires sur la sta-
bilite de la rhonite a l'atmosphere. Bulletin de
Mineralogie, 103:491-502.
Boynton, W.V.
1975. Fractionation in the Solar Nebula: Condensation
of Yttrium and the Rare Earth Elements. Geochim-
ica et Cosmochimica Acta, 39:569-584.
1977. A Brief Review of the Chemistry of Ca,Al-rich
Inclusions in the Allende Meteorite. Meteoritics,
12:183, 184.
Boynton, W.V., R.M. Frazier, and J.D. Macdougall
1980. Trace Element Abundances in Ultra-refractory
Condensates from the Murchison Meteorite. Me-
teoritics, 15:269.
Chen, J.H., and G.J. Wasserburg
1981. The Isotopic Composition of Uranium and Lead
in Allende Inclusions and Meteoritic Phosphates.
Earth and Planetary Science Letters, 52:1-15.
Chou, C-L., P.A. Baedecker, and J.T. Wasson.
1976. Allende Inclusions: Volatile-Element Distribution
and Evidence for Incomplete Volatilization of
Presolar Solids. Geochimica et Cosmochimica Acta,
40:85-94.
Christophe Michel-Levy, M.
1968. Un chondre exceptionel dans la meteorite de Vi-
garano. Bulletin de la Societe Francaise de Mineralogie
et de Cristallographie, 91:212-214.
Clarke, R.S., Jr., E. Jarosewich, B. Mason, J. Nelen, M.
Gomez, and J.R. Hyde
1970. The Allende, Mexico, Meteorite Shower. Smithson-
ian Contributions to the Earth Sciences, 5: 53 pages.
Clayton, D.D.
1980. Chemical Energy in Cold-Cloud Aggregates: The
Origin of Meteoritic Chondrules. Astrophysical
Journal, 239:L37-L41.
Clayton, R.N., L. Grossman, and T.K. Mayeda
1973. A Component of Primitive Nuclear Composition
in Carbonaceous Meteorites. Science, 182:485-488.
Dowty, E., and J.R. Clark
1973. Crystal Structure Refinement and Optical Prop-
erties of a Ti3+ Fassaite from the Allende Mete-
orite. American Mineralogist, 58:230-242.
El Goresy, A., K. Nagel, and P. Ramdohr
1977. Type A Ca,Al-rich Inclusions in Allende Meteor-
ite: Origin of the Perovskite-Fassaite Symplectite
around Rhonite and Chemistry and Assemblages
of the Refractory Metals (Mo, W) and Platinum
Metals (Ru, Os, Ir, Re, Rh, Pt). Meteoritics, 12:216.
El Goresy, A., K. Nagel, B. Dominik, P. Ramdohr, and B.
Mason
1979. Ru-bearing Phosphate-Molybdates and Other
Oxidized Phases in Fremdlinge in Allende Inclu-
sions. Meteoritics, 14:390, 391.
El Goresy, A., and P. Ramdohr
1980. Osmium Depletion in Fremdlinge in an Allende
Inclusion: Possible Supernova Material. In Lunar
and Planetary Science, XI, pages 257, 258. Houston:
Lunar and Planetary Institute.
Fruland, R.M., E.A. King, and D.S. McKay
1978. Allende Dark Inclusions. In Proceedings of the Ninth
Lunar and Planetary Science Conference, pages 1305-
1329. New York: Pergamon Press.
Fuchs, L.H.
1969. Occurrence of Cordierite and Aluminous Orthoen-
statite in the Allende Meteorite. American Mineral-
ogist, 54:1645-1653.
1978. The Mineralogy of a Rhonite-bearing Calcium
Aluminum Rich Inclusion in the Allende Mete-
orite. Meteoritics, 13:73-78.
Gast, P.W., NJ. Hubbard, and H. Wiesmann
1970. Chemical Composition and Petrogenesis of Basalts
from Tranquility Base. In Proceedings of the Apollo
II Lunar Science Conference, pages 1143-1163. New
York: Pergamon Press.
Gray, CM., D. A. Papanastassiou, and G.J. Wasserburg
1973. The Identification of Early Condensates from the
Solar Nebula. Icarus, 20:213-239.
Grossman, L.
1972. Condensation in the Primitive Solar Nebula. Geo-
chimica et Cosmochimica Acta, 36:597-619.
28
NUMBER 25 29
1973. Refractory Trace Elements in Ca-Al-rich Inclu-
sions in the Allende Meteorite. Geochimica et Cos-
mochimica Ada, 37:1119-1140.
1975. Petrography and Mineral Chemistry of Ca-rich
Inclusions in the Allende Meteorite. Geochimica et
Cosmochimica Ada, 39:433-454.
1980. Refractory Inclusions in the Allende Meteorite.
Annual Review of Earth and Planetary Sciences, 8:559-
608.
Grossman, L., and R. Ganapathy
1976. Trace Elements in the Allende Meteorite, I:
Coarse-grained, Ca-rich Inclusions. Geochimica et
Cosmochimica Ada, 40:331-344.
Grossman, L., and I.M. Steele
1976. Amoeboid Olivine Aggregates in the Allende Me-
teorite. Geochimica et Cosmochimica Ada, 40:149-155.
Haggerty, S.E.
1977. Refinement of the Ti-Cosmothermometer in the
Allende Meteorite and the Significance of a New
Mineral, R2+Ti3O7, in Association with Armal-
colite. In Lunar Science, VIII, pages 389-391. Hous-
ton: Lunar Science Institute.
Herzog, G.F., A.E. Bence, J.F. Bender, G. Eichhorn, H.
Maluski, and O.A. Schaeffer
1980. 39Ar-40Ar Systematics of Allende Inclusions. In
Proceedings of the Eleventh Lunar and Planetary Science
Conference, pages 959-976. New York: Pergamon
Press.
Jessberger, E.K., B. Dominik, T. Staudacher, and G.F.
Herzog
1980. 40Ar-39Ar Ages of Allende. Icarus, 42:380-405.
Jordan, J., E.K. Jessberger, T. Kirsten, and A. El Goresy
1980. Alien Xenon in Allende Inclusions. Meteoritics,
15:309, 310.
Kornacki, A.S.
1981. Are CAI's Condensates of Distillation Residues?
Evidence from a Comprehensive Study of Fine-
to Medium-grained Inclusions in the Allende
C3(V) Carbonaceous Chondrite. In Lunar and Pla-
netary Science, XII, pages 562-564. Houston: Lunar
and Planetary Institute.
Kurat, G.
1970. Zur Genese der Ca-Al-reichen Einschliisse im
Chondriten von Lance. Earth and Planetary Science
Letters, 9:225-231.
Kurat, G., G. Hoinkes, and K. Fredriksson
1975. Zoned Ca-Al-rich Chondrule in Bali: New Evi-
dence against the Primordial Condensation
Model. Earth and Planetary Science Letters, 26:140-
144.
Lord, H.C. Ill
1965. Molecular Equilibria and Condensation in a Solar
Nebula and Cool Stellar Atmospheres. Icarus,
4:279-288.
Lorin, J.C., and M. Christophe Michel-Levy
1978. Heavy Rare Metals and Magnesium Isotopic
Anomalies Studies in Allende and Leoville Cal-
cium-Aluminum Rich Inclusions. In Lunar Science,
IX, pages 660-662. Houston: Lunar and Plane-
tary Institute.
Marvin, U.B., J.A. Wood, and J.S. Dickey
1970. Ca-Al-rich Phases in the Allende Meteorite. Earth
and Planetary Science Letters, 7:346-350.
Mason, B.
1974. Aluminum-Titanium-rich Pyroxenes, with Special
Reference to the Allende Meteorite. American Min-
eralogist, 59:1198-1202.
Mason, B., and P.M. Martin
1974. Minor and Trace Element Distribution in Melilite
and Pyroxene from the Allende Meteorite. Earth
and Planetary Science Letters, 22:141-144.
1977. Geochemical Differences among Components of
the Allende Meteorite. Smithsonian Contributions to
the Earth Sciences, 19:84-95.
Niederer, F.R., D.A. Papanastassiou, and GJ. Wasserburg
1980. Titanium Isotope Anomalies in Allende Inclu-
sions. Meteoritics, 15:339, 340.
1981. The Isotopic Composition of Titanium in the Al-
lende and Leoville Meteorites. Geochimica et Cos-
mochimica Ada, 45:1017-1031.
Niemeyer, S., and G.W. Lugmair
1980. Ti Isotope Anomalies in "Un-FUN" Allende In-
clusions. Meteoritics, 15:341.
Simon, S.B., and S.E. Haggerty
1980. Bulk Compositions of Chondrules in the Allende
Meteorite. In Proceedings of the Eleventh Lunar and
Planetary Science Conference, pages 901- 927. New
York: Pergamon Press.
Steele, I.M., and I.D. Hutcheon
1979. Anatomy of Allende Inclusions: Mineralogy and
Mg Isotopes in Two Ca-Al-rich Inclusions. In
Lunar and Planetary Science, X, pages 1166-1168.
Houston: Lunar and Planetary Institute.
Tanaka, T., and A. Masuda
1973. Rare-Earth Elements in Matrix, Inclusions, and
Chondrules of the Allende Meteorite. Icarus,
19:523-530.
Taylor, S.R.
1965. Geochemical Analysis by Spark Source Mass Spec-
trography. Geochimica et Cosmochimica Ada, 29:
1243-1261.
1971. Geochemical Applications of Spark Source Mass
Spectrography, II: Photoplate Data Processing.
Geochimica et Cosmochimica Ada, 35:1187-1196.
Taylor, S.R., and M.P. Gorton
1977. Geochemical Application of Spark Source Mass
Spectrography, III: Element Sensitivity, Precision
30 SMITHSONIAN CONTRIBUTIONS TO THE EARTH SCIENCES
and Accuracy. Geochimica et Cosmochimica Ada,
41:1375-1380.
Taylor, S.R., and B. Mason
1978. Chemical Characteristics of Ca-Al Inclusions in
the Allende Meteorite. In Lunar and Planetary Sci-
ence, IX, pages 1158-1160. Houston: Lunar and
Planetary Institute.
Wanke, H., H. Baddenhausen, H. Palme, and B. Spettel
1974. On the Chemistry of the Allende Inclusions and
Their Origin As High Temperature Condensates.
Earth and Planetary Science Letters, 23:1-7.
Wark, D.A.
1981a. Alteration and Metasomatism of Allende Ca-Al-
rich Materials. In Lunar and Planetary Science, XII,
pages 1145-1147. Houston: Lunar and Planetary
Institute.
1981b. The Pre-Alteration Compositions of Allende Ca-
Al-rich Condensates. In Lunar and Planetary Science,
XII, pages 1148-1150. Houston: Lunar and Pla-
netary Institute.
Wark, D.A., and J.F. Lovering
1980. More Early Solar System Stratigraphy: Coarse-
grained CAIs. In Lunar and Planetary Science, XI,
pages 1208-1210. Houston: Lunar and Planetary
Institute.
Wasserburg, G.J., and J.C. Huenke
1979. I-Xe Dating of I-bearing Phases in Allende. In
Lunar and Planetary Science, X, pages 1307-1309.
Houston: Lunar and Planetary Institute.
Wood, J.A.
1981. The Interstellar Dust As a Precursor of Ca,Al-rich
Inclusions in Carbonaceous Chondrites. Earth and
Planetary Science Letters, 56:32-44.
Yoder, H.S., and T. Sahama
1957. Olivine X-ray Determinative Curve. American Min-
eralogist, 42:475-491.