petrogenesis of olivine-phyric shergottites sayh al ... · petrogenesis of olivine-phyric...

doi:10.1016/S0016-7037(03)00171-6

Petrogenesis of olivine-phyric shergottites Sayh al Uhaymir 005 and Elephant MoraineA79001 lithology A

CYRENA ANNE GOODRICH1,2,*

1Hawaii Institute of Geophysics and Planetology, School of Ocean and Earth Science and Technology, University of Hawaii at Manoa,1680 East-West Road, Honolulu, HI 96822, USA

2Max-Planck-Institut fu¨r Chemie, P.O. Box 3060, D-55020 Mainz, Germany

(Received July 18, 2002;accepted in revised form February 21, 2003)

Abstract—Martian meteorites Sayh al Uhaymir (SaU) 005 and lithology A of EETA79001 (EET-A) belongto a newly emerging group of olivine-phyric shergottites. Previous models for the origin of such shergottiteshave focused on mixing between basaltic shergottite-like magmas and lherzolitic shergottite-like material.Results of this work, however, suggest that SaU 005 and EET-A formed from olivine-saturated magmas thatmay have been parental to basaltic shergottites.

SaU 005 and EET-A have porphyritic textures of large (up to�3 mm) olivine crystals (�25% in SaU 005;�13% in EET-A) in finer-grained groundmasses consisting principally of pigeonite (�50% in SaU 005;�60% in EET-A), plagioclase (maskelynite) and� 7% augite. Low-Ti chromite occurs as inclusions in themore magnesian olivine, and with chromian ulvo¨spinel rims in the more ferroan olivine and the groundmass.Crystallization histories for both rocks were determined from petrographic features (textures, crystal shapesand size distributions, phase associations, and modal abundances), mineral compositions, and melt compo-sitions reconstructed from magmatic inclusions in olivine and chromite. The following observations indicatethat the chromite and most magnesian olivine (Fo 74–70 in SaU 005; Fo 81–77 in EET-A) and pyroxenes(low-Ca pyroxene [Wo 4–6] ofmg 77–74 and augite ofmg 78 in SaU 005; orthopyroxene [Wo 3–5] ofmg84–80 in EET-A) in these rocks are xenocrystic. (1) Olivine crystal size distribution (CSD) functions showexcesses of the largest crystals (whose cores comprise the most magnesian compositions), indicating additionof phenocrysts or xenocrysts. (2) The most magnesian low-Ca pyroxenes show near-vertical trends ofmg vs.Al2O3 and Cr2O3, which suggest reaction with a magma. (3) In SaU 005, there is a gap in augite compositionbetweenmg 78 and 73. (4) Chromite cores of composite spinel grains are riddled with cracks, indicating thatthey experienced some physical stress before being overgrown with ulvo¨spinel. (5) Magmatic inclusions areabsent in the most magnesian olivine, but abundant in the more ferroan, indicating slower growth rates for theformer. (6) The predicted early crystallization sequence of the melt trapped in chromite (the earliest phase) ineach rock produces its most magnesian olivine-pyroxene assemblage. However, in neither case is the totalcrystallization sequence of this melt consistent with the overall crystallization history of the rock or its bulkmodal mineralogy.

Further, the following observations indicate that in both SaU 005 and EET-A the fraction of solidxenocrystic or xenolithic material is small (in contrast to previous models for EET-A), and most of thematerial in the rock formed by continuous crystallization of a single magma (possibly mixed). (1) CSDfunctions and correlations of crystal size with composition show that most of the olivine (Fo 69–62 in SaU005; Fo 76–53 in EET-A) formed by continuous nucleation and growth. (2) Groundmass pigeonites are inequilibrium with this olivine, and show continuous compositional trends that are typical for basalts. (3) TheCSD function for groundmass pigeonite in EET-A indicates continuous nucleation and growth (Lentz andMcSween, 2000). (4) The melt trapped in olivine of Fo 76 to 67 in EET-A has a predicted crystallizationsequence similar to that inferred for most of the rock and produces an assemblage similar to its modalmineralogy. (5) Melt trapped in late olivine (Fo� 64) in SaU 005 has a composition consistent with theinferred late crystallization history of the rock.

The conclusion that only a small fraction of either SaU 005 or EET-A is xenocrystic or xenolithic impliesthat both rocks lost fractionated liquids in the late stages of crystallization. This is supported by: (1) highpigeonite/plagioclase ratios; (2) low augite contents; and (3) olivine CSD functions, which show a drop innucleation rate at high degrees of crystallization, consistent with loss of liquid. For EET-A, this fractionatedliquid may be represented by EET-B.Copyright © 2003 Elsevier Ltd

1. INTRODUCTION

Eighteen of the 26 meteorites believed to be Martian rocksare classified as shergottites (Table 1). Shergottites are com-monly divided into basaltic and lherzolitic types. The basaltic

shergottites are pyroxene-plagioclase basalts, and the lher-zolitic shergottites are olivine-pyroxene cumulates derivedfrom basaltic magmas (McSween and Treiman, 1998). How-ever, recently discovered Martian meteorites include olivine-phyric basalts (Table 1) that appear to represent a third type ofshergottite (Goodrich, 2002). The relationship of these rocks tothe basaltic and lherzolitic shergottites is not clear. Are theyproducts of mixing between basaltic shergottite-like magmas

* Author to whom correspondence should be addressed ([email protected]).

Pergamon

Geochimica et Cosmochimica Acta, Vol. 67, No. 19, pp. 3735–3771, 2003Copyright © 2003 Elsevier Ltd

Printed in the USA. All rights reserved0016-7037/03 $30.00� .00

3735

and lherzolitic material? Do they represent magmas that couldhave been parental to both basaltic and lherzolitic shergottites?Or are they products of entirely different magma types and/orprocesses? This paper examines the petrogenesis of olivine-phyric shergottites Sayh al Uhaymir (SaU) 005 and lithology Aof Elephant Moraine A79001 (EET-A).

1.1. Basaltic and Lherzolitic Shergottites

The basaltic shergottites Shergotty, Zagami, QUE 94201,Los Angeles, NWA 480, NWA 856 and Dhofar 378 (Table 1),consist predominantly of clinopyroxene (pigeonite and augite)and plagioclase (now shock-produced glass or maskelynite),and have basaltic or diabasic textures. Shergotty and Zagamicontain cumulus pyroxene (Stolper and McSween, 1979; Mc-Coy et al., 1992), whereas QUE 94201 and Los Angeles (whichhave higher plagioclase/pyroxene ratios) may represent magmacompositions (Kring et al., 1996; McSween et al., 1996; Rubinet al., 2000; Mikouchi et al., 2001a; McKay et al., 2002).Nevertheless, the absence of olivine in all basaltic shergottites,and their low bulk mg (�100 � molar Mg/[Mg � Fe]) valuesof �23 to 52 (Table 1), indicate that they represent fraction-ated, rather than primary, magmas (Stolper and McSween,1979).

The lherzolitic shergottites ALHA77005, LEW 88516, Y793605, YA 1075 and GRV 99027 (Table 1), consist predom-inantly of coarse-grained olivine and poikilitic pigeonite (Mc-Sween et al., 1979a, 1979b; Harvey et al., 1993; Ikeda, 1994a,1997; Treiman et al., 1994a). In contrast to the basaltic sher-gottites, they have much lower plagioclase contents and higherbulk mg values (�70), and they contain chromite. Their min-eralogy is consistent with early accumulation from magmashaving the crystallization sequence of inferred primary basalticshergottitic magmas (McSween et al., 1979a, 1979b).

1.2. Olivine-Phyric Shergottites

In the context of the division of shergottites into basaltic andlherzolitic types, EETA79001 was always an anomaly.EETA79001 consists of two lithologies separated by an obvi-ous contact (Steele and Smith, 1982; McSween and Jarosewich,1983). Lithology B (EET-B) is a clinopyroxene-plagioclaserock resembling the basaltic shergottites. Lithology A (EET-A), however, is distinct from either the basaltic or the lher-zolitic shergottites, and in the light of recent discoveries can beconsidered to be the first known olivine-phyric shergottite. Itcontains megacrysts of olivine, orthopyroxene, and chromite ina finer-grained groundmass of pigeonite and plagioclase. Pet-

Table 1. Shergottitesa

ol (%)b Fo opx (%) pig (%) aug (%) plag (%) Oxidesc mgd Ref.

Basaltic shergottitesShergotty — — — 36.3 33.5 23.3 ti, il 46 1, 2Zagami — — — 36.5 36.5 21.7 ti, il 52 1, 2EET-Be — — — 39.5 20.0 29.1 ti, il 43 1, 2QUE 94201 — — — 33.7 10.1 46.0 ti, il 43 3, 4Los Angeles — — — 38–44 43–45 ti, il 23 5, 6NWA 480 — — — 41 31 25 ti, il 34 7NWA 856 — — — 45 23 23 ti, il 49 8Dhofar 378 — — — 49 47 ti, il 9

Olivine-phyric shergottitesEET-Ae 10–13 81–52 3.4–7.2 55–63 3.2–6.5 16–18 chr, ti, il 61 2, 10, 11DaG 476f 14–24 79–62 1.5–3.5 50–53 tr–2.9 12–17 chr, ti, il 68 12, 13, 14SaU 005g 21–29 74–62 1 48 �7 15 chr, ti, il 68 15, 16Dhofar 019 7–12 73–25 — 57–63 26–27 chr, ti, il 58 17NWA 1068h 21 72–42 — 52 22 chr, ti, il 59 18NWA 1195 81–60 19

Lherzolitic shergottitesALHA77005 60.2 73–69 — 9.5 3.7 9.5 chr-ti, il 71 2, 20LEW 88516 45.9 70–64 — 25.3 12.0 7.0 chr-ti, il 70 20, 21Y 793605 40.4 75–64 — 51 7.4 chr-ti, il 70 22, 23YA 1075 75–68 24GRV 99027 73 25

a Abbreviations: ol � olivine; Fo � forsterite; opx � orthopyroxene; pig � pigeonite; aug � augite; plag � plagioclase (maskelynite); ref. �reference; EET � Elephant Moraine; QUE � Queen Alexandra Range; NWA � North West Africa; DaG � Dar al Gani; SaU � Sayh al Uhaymir,ALH � Allan Hills; LEW � Lewis Cliffs; Y � Yamato.

b Modal abundances.c chr � chromite; ti � titanomagnetite (chromian ulvospinel); il � ilmenite; chr-ti � chromite-titanomagnetite solution.d 100 � molar Mg/(Mg � Fe) in bulk composition.e EETA79001 lithology A and lithology B.f And possibly paired meteorites DaG 489/734/670/876. Range of Fo contents varies among specimens.g And possibly paired meteorites SaU 008/051/094/060/090. Range of Fo contents varies among specimens.h Possibly paired with NWA 1110 (26).References: (1) McSween (1985) and references therein. (2) Banin et al. (1992) and references therein. (3) McSween et al. (1996). (4) Dreibus et

al. (1996). (5) Rubin et al. (2000). (6) Mikouchi (2000). (7) Barrat et al. (2002a). (8) Jambon et al. (2002). (9) Ikeda et al. (2002). (10) Steele andSmith (1982). (11) McSween and Jarosewich (1983). (12) Zipfel et al. (2000). (13) Folco et al. (2000). (14) Mikouchi et al. (2001b). (15) Zipfel (2000)and this work. (16) Dreibus et al. (2000). (17) Taylor et al. (2002). (18) Barrat et al. (2002b). (19) Irving et al. (2002). (20) Treiman et al. (1994a).(21) Dreibus et al. (1992). (22) Ikeda (1997). (23) Warren and Kallemeyn (1997). (24) Yanai (2002). (25) Lin et al. (2002). (26) Goodrich et al. (2003).

3736 C. A. Goodrich

rogenetic studies of EET-A have largely focussed on modellingit as a mixture of basaltic and lherzolitic shergottite types.Textural and compositional characteristics of the megacrystssuggested disequilibrium with the groundmass, which led to theidea that they are xenolithic remnants of assimilated ultramaficmaterial. McSween and Jarosewich (1983) calculated that thegroundmass of EET-A could be produced by mixing �10%olivine, 26% orthopyroxene and 0.5% chromite with a magmasimilar to EET-B. However, Wadhwa et al. (1994) showed thatthe energy required to assimilate this material was more thancould plausibly be provided by latent heat of crystallization. Analternative model discussed (McSween and Jarosewich, 1983;McSween, 1985; Wadhwa et al., 1994) was magma mixing,with the megacrysts originating as phenocrysts in one of themagmas. Mittlefehldt et al. (1999), however, showed fromtrace element data that the lherzolitic endmember in any mixingmodel for EET-A should have contained little melt, and sug-gested that the energy constraints of assimilation could besatisfied by impact melting. Although no one model has beengenerally accepted for the origin of EET-A, the idea that itsmegacrysts are in some sense xenolithic has, and some authors(e.g., Treiman, 1995) have even referred to them as lithology X.

DaG 476 (discovered in Libya in 1998/1999) is a porphyriticolivine basalt consisting of large olivine crystals, lesser or-thopyroxene, and chromite grains in a finer-grained ground-mass of pigeonite and maskelynite (Zipfel et al., 2000), and wasthus the first example of a lithology like EET-A to be found asa whole meteorite. Its general similarity to EET-A led to theidea that its megacrysts might be xenolithic, and to discussionof mixing models for its petrogenesis (Folco et al., 2000; Zipfelet al., 2000; Mikouchi et al., 2001b; Wadhwa et al., 2001).However, textural and chemical characteristics of DaG 476make mixing models for its origin less compelling than in thecase of EET-A, and some authors (Zipfel et al., 2000) havesuggested that DaG 476 represents a previously unrecognizedtype of shergottitic magma.

SaU 005 (found in Oman �1 yr after the discovery of DaG476) is similar to DaG 476 in texture, mineralogy and mineralcompositions, bulk chemical composition, and exposure age(Zipfel, 2000; Dreibus et al., 2000; Patsch et al., 2000). How-ever, detailed mineralogical differences between the two me-teorites and the large distance between the two sites at whichthey were found, indicate that they are not paired (Zipfel,2000). SaU 094 (Gnos et al., 2002) has been possibly pairedwith SaU 005 (Grossman and Zipfel, 2001).

Three more shergottites (Dhofar 019, NWA 1068 and NWA1195) with mineralogical and textural similarities to EET-A,DaG 476, and SaU 005 have recently been discovered. It isclear that these six rocks share petrographic features that dis-tinguish them from either of the two established shergottitetypes (Table 1), and I have proposed (Goodrich, 2002) that theybe designated by the term “olivine-phyric shergottite.” Theirrelative abundance suggests that they may not simply be mix-tures of basaltic shergottite-like and lherzolitic shergottite-likematerials. They may instead represent distinct martian magmatypes (e.g., Zipfel et al., 2000; Taylor et al., 2002), possiblymore primitive than any previously recognized (Irving et al.,2002). In light of these new developments, it seems worthwhileto reconsider the petrogenesis of EET-A, through a comparisonwith other olivine-phyric shergottites.

This paper addresses the petrogenesis of SaU 005 andEET-A. For each of these rocks, I examine petrographic fea-tures (textures, crystal shapes and size distributions, phaseassociations and modal abundances) and mineral (olivine, py-roxenes and spinels) compositions to reconstruct its crystalli-zation history. I then use magmatic inclusions in olivine andchromite to determine the compositions of magmas that werepresent at various stages of its history, and compare the crys-tallization sequences predicted for these magmas to its modalmineralogy, mineral compostions, and inferred crystallizationsequence to test various petrogenetic models. In addition, I usethe compositions of these magmas to examine possible rela-tionships between these olivine-phyric shergottites and the ba-saltic and lherzolitic shergottites.

2. SAMPLES AND ANALYTICAL METHODS

Three thin sections (#s 1, 3 and 4) and one thick section of SaU 005,three thin sections (. . .,76; . . .,68; and . . .,94) of EETA79001, and onethin section (. . .,29) of ALHA77005 were studied. Back-scatteredelectron images, X-ray maps, and quantitative analyses were obtainedusing the JEOL JXA 8900RL electron microprobe at Johannes Guten-berg Universitat in Mainz, and the JEOL JSM-LV5900 scanning elec-tron microscope and Cameca SX-50 electron microprobe at the Uni-versity of Hawaii. Operating conditions for quantitative analyses were15-keV accelerating potential and 12 to 30 nA beam current foranalyses of silicate phases, and 20-keV accelerating potential and �20to 30 nA beam current for analyses of chromite and Fe-Ti oxides.Natural and synthetic oxides and silicates were used as standards.Counting times ranged from 10 to 40 s. PAP �–�-z corrections wereapplied to the analyses. For analyses of glasses, a defocussed beam(�2-�m diameter) was used wherever the size of the area permitted. Totest whether loss of alkali elements occurred in these analyses, someglasses were analyzed using 10-keV accelerating potential, 5-nA beamcurrent, and a 5-�m beam diameter. These tests did not show signifi-cantly higher values for Na2O or K2O compared with 15-keV analyses.Analyses of mixed phases (designated broad-beam analyses) wereperformed with a defocussed beam (�2–5-�m diameter).

3. GENERAL PETROGRAPHY AND MINERALCOMPOSITIONS

3.1. SaU 005

SaU 005 has a porphyritic texture of large olivine crystals(commonly in clusters) in a finer-grained groundmass consist-ing principally of low-Ca pyroxene and maskelynite (Figs. 1aand 1c). Chromite occurs as rare inclusions in magnesianolivine, and more abundantly with chromian ulvospinel rims asinclusions in more ferroan olivine and discrete grains in thegroundmass. Occurrences and compositions of spinels are de-scribed in detail in section 4.1.

The abundance of olivine was determined by automatedpoint-counting of X-ray maps to be 21 to 29% (by area), similarto that (25%) reported by Zipfel (2000). Grain shapes arecommonly subhedral to euhedral (Figs. 1a and 1c). Sizes (max-imum length and average width) were measured for all olivinecrystals (�50 �m) in an area (excluding shock-melted veinsand pockets) of �325 mm2 (276 crystals, yielding a density of�85 crystals/cm2). Individual crystals in clusters were distin-guished optically, and then measurements were made on com-bined elemental X-ray maps (e.g., Fig. 1a). Seventy-four per-cent of all crystals have lengths � 0.5 mm, with a maximum inthe distribution at �0.25 mm (Fig. 2a). For the remaining 25%,abundance decreases regularly with increasing length up to �2

3737Petrogenesis of SaU 005 and EETA79001 lithology A

mm. Only a few larger crystals (up to �3 mm) were observed.A crystal size distribution (CSD) plot of ln(n) vs. length, wheren is the slope of the cumulative crystals per volume vs. lengthfunction (Marsh, 1988), is linear with negative slope (�1.74)for crystals 0.1 to 1.5 mm in length, shows a dropoff for smallercrystals, and is horizontal for larger crystals (Fig. 2c). Mostcrystals have aspect ratios of 1 to 2; however, crystals � 2 mmin length all have aspect ratios of 2 to 3.5.

Compositions were obtained for 113 olivine crystals, com-prising all those (�50 �m in size and not obviously shockmelted) in an area of �100 mm2. For every crystal, maximumand minimum forsterite (Fo) contents were determined througha combination of line profiles and point analyses, using Mg andFe X-ray maps as a guide in selecting locations for analysis. Forzoned crystals, maximum Fo contents are always located nearcenters and minimum Fo contents are always located at edges.In most cases, zonation is concentrically regular or subregular.Forsterite contents range from 74 to 62 (Fig. 2e; Table 2).Crystals � 0.5 mm in length (74% of all crystals) are predom-

inantly Fo 65 to 63 in composition, and only slightly zoned. Allcrystals � 1 mm in length have minimum (edge) Fo contents ��65. Most have maximum Fo contents extending only to �Fo71. Only the few crystals � 2 mm in length have moremagnesian (up to Fo 74) core compositions.

Magnesian low-Ca pyroxenes (mg 75–77) having Wo con-tents consistent with orthopyroxene composition (Wo 4–6)occur as rounded, �50 to 200 �m sized inclusions in the highlymagnesian (Fo 73–74) cores of the largest olivine crystals (Fig.3a; Table 3). They have �0.5 to 1.0% Al2O3, 0.4 to 0.5%Cr2O3 and 0.05 to 0.1% TiO2 (Fig. 4a). Magnesian augite (mg78, Wo 34–35), with 1.8 to 2% Al2O3 (Table 3), also occurs inseveral of the inclusions. Low-Ti chromite grains are com-monly associated with these pyroxenes (Fig. 3a).

The groundmass contains �48% low-Ca pyroxene and 15%maskelynite, and has an average grain size of �130 �m(Zipfel, 2000). Augite occurs as small (�100 �m), irregularly-shaped grains, and was determined by automated point-count-ing of combined elemental X-ray maps to comprise at most 7%

Fig. 1. SaU 005 (a, c) and EETA79001 (b, d). Combined elemental X-ray maps (Red � Ca K�, Green � Fe K�, Blue� Al K�) in (a, b). Olivine is light to medium green, pigeonite is dark green to brown, maskelynite is blue. Augite,phosphates, and (in SaU 005 only) veins of terrestrial Ca-carbonate are red to orange. Image of EETA79001 (b) showscontact between lithology A (top) and lithology B (bottom). Lithology B has a higher abundance of maskelynite and augiteand is coarser-grained than the groundmass of lithology A. Olivine abundance is �25% in SaU 005 and �12% in lithologyA of EETA79001 (EET-A). Larger areas of olivine in SaU 005 are clusters of two to five crystals. Note that olivines inEET-A are more strongly zoned than those in SaU 005. Collages of back-scattered electron images (BEI) in (c, d). Small,bright grains are chromite and Fe-Ti oxides.

3738 C. A. Goodrich

of the rock (this number includes whitlockite and Ca-carbon-ates, which were not distinguished from augite). Minor phasesin the groundmass are whitlockite, pyrrhotite, pentlandite, chro-mite, chromian ulvospinel (or titanomagnetite) and ilmenite.Maskelynite compositions as reported by Zipfel (2000) areAn51-65Or0.3-0.9 and were not investigated in this work.

Low-Ca pyroxenes in the groundmass are pigeonite, withWo � 6 and mg � 75 to 67 (Table 3). They have abundant,shock-produced twin lamellae typical of pigeonite. No opticallyor compositionally distinct cores of orthopyroxene were ob-served, consistent with the report of Zipfel (2000). Composi-tional trends (Fig. 4a) show generally increasing Al2O3 (start-

Fig. 2. Olivine in SaU 005 and EET-A. (a, b) Histograms of number of crystals versus maximum length. SaU 005: 276crystals in �325 mm2 (85 crystals/cm2). EET-A: 46 crystals in �195 mm2 (24 crystals/cm2). (c, d) Crystal size distribution(CSD) plots, based on data in (a, b), of ln(n) vs. length, where n is dNV*/dL and NV* is cumulative number of crystals pervolume (NV � �NA1.5). Lines are regressions through the ranges of lengths shown. (e) Maximum (central) and minimum(edge) forsterite (Fo) contents for all crystals in an area of 100 mm2 in SaU 005. Crystals � 0.5 mm in length (74% of allcrystals) are mostly Fo 65 to 63 in composition, and only slightly zoned. All crystals � 1 mm in length have minimum Focontents � � 65. Most have maximum Fo contents � Fo 71. Only the few crystals � 2 mm in length have more magnesiancore compositions, up to Fo 74.


ing from values comparable to those found in theorthopyroxene) and Wo as mg decreases from 75 to �71, anddecreasing Al2O3 and Wo as mg decreases from �71 to 68.Cr2O3 contents decrease and TiO2 contents increase (again,starting from values comparable to those found in the orthopy-roxene) over the entire compositional range, but show increasesin slope from mg � 71 to 68. In addition, both Wo and Al2O3

(and possibly also Cr2O3) show notable spikes (nearly verticaltrends) at the most magnesian compositions (mg � 74–75),which are similar to those reported for the most magnesianpigeonite in DaG 476 (Zipfel et al., 2000). Augites in thegroundmass (Wo � 30–34) have compositions of mg � 73 to69, with Al2O3 contents of �1.8 to 2.5% (Table 3).

3.2. EET-A

EET-A has a porphyritic texture (Figs. 1b and 1d) with �15vol.% megacrysts in a finer-grained groundmass (Steele andSmith, 1982; McSween and Jarosewish, 1983). The megacrystsconsist of olivine (�10–13%) and low-Ca pyroxene (2–4%)that is referred to in the literature as orthopyroxene. Olivine andlow-Ca pyroxene megacrysts are generally isolated from oneanother, but some composite grains have been observed (Mc-Sween and Jarosewich, 1983). Low-Ti chromite grains occur asinclusions in the megacrysts and with chromian ulvospinel rimsin the groundmass, and are commonly considered part of themegacryst assemblage. Spinels are described in section 4.2.The groundmass consists of pigeonite (55–63%), augite (3–6%) and maskelynite (16–18%), with minor ulvospinel, ilmen-ite, phosphate, pyrrhotite and mesostasis (Steele and Smith,1982; McSween and Jarosewich, 1983). The CSD function forgroundmass pyroxenes is linear with negative slope in therange 0.1 to 0.4 mm (Lentz and McSween, 2000).

Steele and Smith (1982) and McSween and Jarosewich(1983) described olivine grains in EET-A as having highlyirregular external forms. However, observations made in thisstudy show that many grains are subhedral (Figs. 1b and 1d),and many of those with highly irregular shapes appear to havebeen sheared and/or disrupted by veins of late shock melt.Olivine sizes (maximum length) were measured for all crystalsin an area of �195 mm2 (46 crystals, yielding a density of �24crystals/cm2). The distribution peaks at �1 mm, with �57% ofcrystals having lengths between �0.9 and 1.4 mm (Fig. 2b).Sizes extend up to �2.7 mm (McSween and Jarosewich [1983]report megacryst sizes as large as 5 mm, but it is not clear if

these are single crystals). The crystal size distribution (CSD)function (Fig. 2d) is linear with negative slope (�1.35) forcrystals �0.9 to 1.9 mm in length, shows an extreme dropofffor smaller crystals, and is horizontal for larger crystals. Oli-vine compositions range from Fo 81 to 53 (Steele and Smith,1982; McSween and Jarosweich, 1983). In this study, it wasobserved that the most magnesian compositions (Fo � 76) arerare and occur only in the core regions of the largest (�1.5 mm)crystals. Most crystals are zoned from �Fo 76 to 63 (similar tocrystals described by Steele and Smith, 1982). McSween andJarosewich (1983) and Steele and Smith (1982) emphasized theirregularity of zonation contours, but many crystals appearconcentrically zoned. Smaller crystals tend to consist entirelyof more FeO-rich compositions.

Low-Ca pyroxene megacrysts consist of irregularly-shaped,magnesian cores with more ferroan coronas (Figs. 3b and 3c).Steele and Smith (1982) and McSween and Jarosewich (1983)noted that although it is difficult to determine the structuralstate of EET-A low-Ca pyroxenes from optical properties, dueto shock effects, the more Mg-rich compositions appear to beorthorhombic. In this work it was found that the magnesiancores of low-Ca pyroxene megacrysts are commonly free oftwin lamellae, while their coronas, as well as all groundmasslow-Ca pyroxenes, show shock-produced twinning typical ofpigeonite (Fig. 3b). This distinction is correlated with differ-ences in minor element trends. Although the cores have alimited range of mg (�80–82), they show large variations inAl2O3 (�0.4–2.1 wt.%) and Cr2O3 (�0.5–1.1%) contents,even within single grains, resulting in nearly vertical mg-Al2O3

and mg-Cr2O3 trends (Fig. 4b). Al2O3 and Cr2O3 contents arepositively correlated, and the highest concentrations of theseelements occur near the outer edges of the grains. Wo contentsof cores are mostly 2 to 3 (consistent with orthopyroxenecompositions), though the data hint at a nearly vertical mg-Wotrend as well. Small inclusions (�50 mm) of low-Ca pyroxeneobserved in olivine of Fo 73 are similar in composition (mg 81,Wo 2), and also show significant variation in Al2O3 and Cr2O3

(Fig. 4b). In contrast, coronas around the cores and all ground-mass low-Ca pyroxenes have less magnesian (mg � 78) com-positions and show much shallower compositional trends (Fig.4b). Al2O3 and Wo increase as mg decreases to �60, and thendecrease as mg decreases to � 60. Cr2O3 decreases continu-ously as mg decreases, but shows a slight spike at mg � 60.TiO2 contents are very low (�0.05%) in the magnesian cores,and increase continuously with decreasing mg in coronas andgroundmass low-Ca pyroxene (Fig. 4b). The most ferroanlow-Ca pyroxene analyzed in this study was mg � 57, butMcSween and Jarosewich (1983) report compositions extend-ing to mg � 50. Augites in the groundmass range in compo-sition from mg � 65 to 50. The general compositional trendsobserved here for low-Ca pyroxenes are consistent with thosereported by McSween and Jarosewich (1983). However, thedistinct vertical trends shown by the megacryst cores (Fig. 4b)have not previously been described.

Section . . .,68 has an exceptionally large (5 mm) low-Capyroxene megacryst, called X-14 (Berkley et al., 1999; Berkleyand Treiman, 2000), with several isolated, irregularly-shapedpatches (or cores) that are more magnesian (mg � 84–86) thancores of the low-Ca pyroxene megacrysts found in most thinsections. Data from Treiman and Berkley (private communica-

Table 2. Olivine in SaU 005.a

1b 2 3 4

SiO2 38.2 37.3 36.7 36.2Cr2O3 0.1 0.06 0.05 bd1c

FeO 23.7 26.2 30.4 32.5MgO 37.6 35.1 31.7 29.8MnO 0.50 0.53 0.58 0.62CaO 0.19 0.39 0.26 0.26Total 100.3 99.6 99.7 99.4Fo 73.9 70.5 65.3 62.0

a Selected analyses showing full range of forsterite (Fo) contents.b Center of large crystal in Fig. 3a.c Below detection limit.

3740 C. A. Goodrich

Fig. 3. Orthopyroxenes in SaU 005 and EET-A. (a) Orthopyroxene (mg 77–75) in SaU 005 occurs only as rounded, 50to 200 �m-sized inclusions (dark grey) in the most magnesian (Fo 74–71) olivine, in some cases associated with magnesian(mg 78) augite (not distinguishable at the contrast level shown). White grains are low-Ti chromite. This olivine crystal isone of the largest (see Fig. 2), and is zoned from Fo 74 in the center to Fo 65 at the edges (“wings” on either side are separatecrystals). Collage of BEI. (b) Megacryst consisting of magnesian (mg 82–80) orthopyroxene core surrounded by pigeonitein EET-A. Crossed-polarized transmitted light. Core is free of twin lamellae, while rim shows polysynthetic twinningcharacteristic of pigeonite. (c) BEI of same grain as in (b). White dots correspond to analysis points labelled “opx cores”in Figure 4b.


tion) and obtained in this work show that these cores (Wo 3–4)have near-vertical trends in Al2O3 and Cr2O3, similar to thoseof cores in the common megacrysts but offset to higher mg(Fig. 4b).

McSween and Jarosewich (1983) designated low-Ca py-roxenes with Wo 3 to 5 (corresponding to mg � 73) asorthopyroxene and those with Wo � 5 (mg � 73) as pigeonite.In this paper, I will refer only to low-Ca pyroxene identified bythe near-vertical minor element trends shown in Figure 4b (andcommonly also by the absence of twin lamellae) as orthopy-roxene (regardless of its present structural state). Other low-Capyroxene (identified by shallower minor element trends and thepresence of abundant twin lamellae) will be referred to aspigeonite (despite the fact that the most magnesian membershave low Wo contents consistent with orthopyroxene compo-sitions).

4. SPINELS IN SaU 005 AND EET-A

4.1. Spinels in SaU 005

The main occurrences of spinels in SaU 005 are summarizedin Figure 5. Low-Ti chromite occurs as individual grains (type1) included in olivine of Fo 74 to 70 composition, and incomposite grains as cores rimmed by chromian ulvospinel(type 2) included in olivine of Fo � 69 to 62 and in thegroundmass. Chromian ulvospinel also occurs as individualgrains included in olivine of Fo 69 to 62 and in the groundmass,and as daughter crystals in melt inclusions that occur in Fo 69to 62 olivine. In addition, a magnetite-rich spinel occurs inmicron to submicron-sized intergrowths with pyroxene in oli-vine of all compositions.

4.1.1. Type 1 chromites

Chromites included in olivine, which are rare, are subhedralto anhedral grains �25 to 35 �m in size (e.g., Fig. 3a). Theyhave Cr-rich, low-Ti compositions (Fig. 6a; Table 4, analyses 1and 2) with �1.7 to 2.0% ulvospinel (100 � molar 2Ti/[2Ti �Cr � Al]) component, and show a small variation in Cr# (molarCr/[Cr � Al]) from �0.77 to 0.81. Their magnetite (100 �molar Fe3�/[Fe3� � Cr � Al � 2Ti]) components, calculatedfrom electron microprobe analyses following the method ofCarmichael (1967), are �2 to 3% (Fig. 7a). They show anormal zonation trend, with fe# (molar Fe2�/[Fe2� � Mg])increasing and Cr# dereasing from center to edge (Fig. 8a). Theobserved variation in fe# (�0.77–0.78), however, is small,which indicates that subsolidus Fe/Mg reequilibration has oc-curred. Olivine-spinel Fe/Mg equilibration temperatures deter-mined from the calibration of Fabries (1979) are �820°C forthe edges of the grains and �900°C for their centers.

4.1.2. Type 2 composite Chromite-Ulvospinel grains

Eight composite spinel grains included in Fo 69 to 62 olivineand eighteen occurring in the groundmass were examined indetail. They have identical properties in the two occurrences.Chromite cores and chromian ulvospinel rims are distinguishedfrom one another texturally (Figs. 9a and 9b). Cores are per-vaded by short thin cracks (�1–20 �m long, submicron inwidth), which commonly end abruptly at the core-rim bound-ary, while rims are almost completely crack-free (though insome cases systems of larger cracks extend through both coreand rim). This textural distinction is revealed most clearly in

Table 3. Pyroxenes in SaU 005 and EET-A.

SaU 005 EET-A

opxa augb pigc pigd pige pigf pigg augh opxi opxJ opxk pigl pigm pign

SiO2 54.3 50.9 53.8 53.7 52.3 52.5 52.5 51.3 54.9 54.3 55.2 54.7 50.8 51.3TiO2 0.07 0.18 0.10 0.11 0.18 0.28 0.52 0.58 0.04 0.08 0.05 0.05 0.42 0.51Al2O3 0.55 2.0 0.73 0.88 1.39 1.22 0.83 1.8 0.42 2.1 0.50 0.46 1.2 0.73Cr2O3 0.43 0.91 0.47 0.43 0.54 0.56 0.34 0.83 0.52 1.0 0.64 0.54 0.49 0.28FeO 14.1 8.9 15.3 15.7 15.0 16.7 18.4 12.6 13.1 12.5 12.0 13.9 21.3 23.7MgO 27.2 17.7 25.5 25.7 24.3 21.9 22.0 16.8 30.1 27.1 30.4 27.6 17.9 18.2MnO 0.48 0.33 0.55 0.53 0.55 0.58 0.61 0.45 0.46 0.43 0.42 0.51 0.68 0.72CaO 2.0 17.1 3.7 3.0 5.7 5.8 4.5 14.1 0.93 2.8 1.0 2.0 6.5 4.6Na2O 0.05 0.29 0.05 0.07 0.08 0.09 0.12 0.24 0.07 0.11 0.03 0.03 0.11 0.07Total 99.2 98.3 100.2 100.1 100.0 99.6 99.8 98.7 100.5 100.4 100.2 99.8 99.4 100.1Wo 3.9 35.2 7.3 6.0 11.1 11.8 9.1 29.7 1.8 5.5 1.9 3.8 13.5 9.5mg 77.5 78.0 74.8 74.5 74.3 70.0 68.1 70.4 80.4 79.4 81.9 78.0 60.0 57.8

a High-mg, low-Wo inclusion in Fo 74 olivine (Fig. 3a).b High-mg augite inclusion in Fo 74 olivine (Fig. 3a).c High-mg, low-Al groundmass pigeonite.d High-mg, low-Wo groundmass pigeonite.e High-mg, high-Al and high-Wo groundmass pigeonite.f Low-mg, highest-Wo, high-Al groundmass pigeonite.g Lowest-mg groundmass pigeonite.h Groundmass augite.i Lowest-Al opx, from core in Figs. 3b and 3c.J Highest-Al opx, from core in Figs. 3b and 3c.k Highest-mg opx, from core in Figs. 3b and 3c.l Low-Wo, high-mg pigeonite from corona in Figs. 3b and 3c.m Highest-Wo and -Al groundmass pigeonite.n Lowest-mg groundmass pigeonite.

3742 C. A. Goodrich

back-scattered electron images (Figs. 9a and 9b) and, except inrare cases, is difficult to see in reflected light. Cores are anhe-dral to subhedral, sometimes with corroded and/or embayedshapes. Overall grain shapes are anhedral to euhedral and arenot necessarily controlled by core shapes. Rims are not alwayspresent on all sides of the cores; i.e., cores extend to the edgesof the grains in some places (e.g., Fig. 9a lower left and rightcorners) but not in others. Core sizes range from �10 to 300�m, and overall grain sizes range from �25 to 370 �m.

Cores have low-Ti compositions similar to those of type 1chromites. Three distinct patterns of core-rim zonation wereobserved. Individual grains commonly show different of thesepatterns (including all 3) in different profiles. In pattern 1 (Fig.6a; Fig. 10a, profile 1; Table 4, analyses 3 and 4) cores havenearly constant ulvospinel (2.4 0.4%), and a small range ofCr-Al variation similar to that of type 1 chromites (Cr# de-creasing from center to core/rim boundary). Their magnetitecontents are �1 to 3% (Fig. 7b). Fe#s are uniform within mostcores, and vary from �0.74 to 0.81 among cores (Fig. 8b). Oneexceptional core shows normal fe#-Cr# zonation similar to thatof type 1 chromites.

Chromian ulvospinel rims follow a trend of ulvospinel vari-ation (�10–42%) at nearly constant Al content (Fig. 6a; Fig.10a, profile 1; Table 4, analyses 5 and 6). Magnetite contents(�2–9%) and fe#s (�0.77–0.85) are generally higher thanthose of cores. They are zoned (from core/rim boundary to edgeof grain) with Cr# decreasing (�0.81–0.75) as magnetite andfe# increase (Figs. 7b and 8b). There is a distinct gap inulvospinel component (from �3.5–10%) between cores andrims (Fig. 6a), and a discontinuous change in zonation trend.

Pattern 2 (Fig. 6b; Fig. 10a, profile 2; Table 4, analyses 7–9)differs from pattern 1 only in that cores deviate from a trend ofstrict Cr-Al variation, showing slight enrichment (from centerto core/rim boundary) in ulvospinel (up to �7%) and magnetitecomponents. There remains a small gap in ulvospinel contentbetween cores and rims. Pattern 3 (Fig. 6c; Fig. 10a, profile 3;Table 4, analyses 10–12) occurs where rims are absent andcores extend to the edge of the grain. For grains included in Fo69 to 62 olivine, this occurs only where the grains protrudefrom their olivine hosts into the groundmass. This pattern ischaracterized by smoothly increasing (from center to edge)ulvospinel and magnetite contents, with decreasing Cr# and fe#(Figs. 6c, 7c, and 8c). It is similar to that of pattern 2 cores, butat slightly higher Al contents and extending to higher ul-vospinel contents (�13%) that bridge the gap seen in patterns1 and 2.

Superimposed on these general compositional patterns areseveral other effects. Near melt inclusions, cores show strongdeviations from the patterns described above. They are depletedin chromite and magnetite and enriched in spinel and ul-vospinel components, and have lower fe#s (Fig. 10a, profile 4;Table 4, analysis 13). They show a zonation pattern (Figs. 6c,7c, 8c) that is similar to the pattern 3 trend but with thesignificant exception that magnetite decreases instead of in-creases (Fig. 11). This pattern was also observed around somelarge cracks. In addition, cores show an unusual pattern ofback-scattered electron contrast, which in some grains (Fig. 9a)has a fine lamellar structure. Quantitative analyses failed toresolve chemical variations that might be responsible for thispattern, but X-ray imaging suggests variations in Ti content,possibly indicating exsolution of ulvospinel or ilmenite. Rimshave a faintly-revealed fine lamellar structure, which is prob-ably a result of late oxidation-exsolution of ilmenite.

4.1.3. Pyroxene-Spinel intergrowths

Micron to submicron-sized pyroxene-spinel inclusions (Fig.9e), similar to those described by Zipfel et al. (2000) in DaG476 and by Ikeda (2001) in DaG 735 (paired with DaG 476),

Fig. 4. Plots of mg vs. Al2O3, Cr2O3, Wo and TiO2 for low-Capyroxenes in (a) SaU 005 and (b) EET-A. In EET-A both the commonorthopyroxene cores and the unusual X-14 (data from this work andfrom Berkley and Treiman, private communication) show nearly ver-tical trends in Al2O3, Cr2O3 and possibly also Wo, which are distinctfrom the shallower trends shown by pigeonite (coronas on cores and ingroundmass). In SaU 005, similar spikes in Al2O3, Cr2O3 and Wo areshown by the most magnesian groundmass pigeonite.


occur in olivine of all compositions. They tend to be elongatein shape, showing parallel alignment along crystallographicdirections in their hosts. In some cases the pyroxene andchromite occur in a symplectic intergrowth. They appear to bemore abundant in the small, ferroan olivines than in the moremagnesian olivine, and as reported by Ikeda (2001) for DaG735, also slightly larger. Analyses were obtained from only afew of the larger inclusions, showing the pyroxene to be pi-geonite of Wo �11 to 16 and the spinel to be Ti-poor andmagnetite-rich.

4.2. Spinels in EET-A

The main occurrences of spinels in EET-A are analogous tothose in SaU 005 (Fig. 5). Type 1 low-Ti chromites occur asinclusions in a magnesian range of olivine compositions (in thiscase Fo 81–60), while type 2 composite grains of chromitecores with chromian ulvospinel rims occur as inclusions in themore ferroan range of olivine compositions (Fo 59–53) and inthe groundmass. Low-Ti chromite also occurs as trapped crys-tals in melt inclusions that occur in Fo � 76 to 60 olivine. Inaddition, tiny pyroxene-spinel intergrowths were observed inolivine.

4.2.1. Type 1 chromites

Chromites included in olivine are significantly more abun-dant than in SaU 005. They are �15 to 40 �m in size, with

euhedral to subhedral shapes. They have low-Ti compositionssimilar to type 1 chromites in SaU 005, with nearly constantulvospinel content of �2% (Fig. 6d) and magnetite contents of�3% (Fig. 7d). Likewise, they show limited normal zonation,but with higher Cr#s (�0.87–0.84) and fe#s (�0.78–0.82),and a slightly larger range of fe# variation (Fig. 8d). Olivine-spinel equilibration temperatures are �950 to 1000°C (Fabries,1979).

4.2.2. Type 2 composite Chromite-Ulvospinel grains

As in SaU 005, composite spinels that occur as inclusions inferroan olivine and those that occur in the groundmass havevirtually identical properties. They consist of cores and rimsthat are distinguished from one another texturally (Figs. 9c and9d). Cores are pervaded by cracks that end abruptly at thecore/rim boundary, while rims are largely crack-free (as in SaU005, systems of larger cracks can extend through both core and

Fig. 5. Occurrences of spinels and melt inclusions in SaU 005 andEET-A. Low-Ti chromite occurs as inclusions in olivine of Fo 74 to 70in SaU 005 and Fo 81 to 60 in EET-A. Composite grains consisting oflow-Ti chromite cores with chromian ulvospinel rims, as well asindividual grains of chromian ulvospinel, occur as inclusions in themore ferroan olivine (Fo 69–62 for SaU 005 and Fo 59–53 for EET-A)and in the groundmass in both rocks. Melt inclusions occur in low-Tichromites in all settings in both rocks. Melt inclusions occur in olivineonly of Fo 69 to 62 in SaU 005; these inclusions commonly containdaughter crystals of chromian ulvospinel, but do not contain chromite.Melt inclusions occur in olivine only of Fo 76 to 60 in EET-A; theseinclusions commonly contain trapped crystals of low-Ti chromite,without ulvospinel. The most magnesian olivine cores in both rocks aredevoid of melt inclusions. In both rocks, olivine contains tiny exsolu-tions of pyroxene plus a magnetite-rich spinel (not illustrated).

Fig. 6. Compositions of spinels in SaU 005 and EET-A in the systemchromite (Cr) – spinel (Al) – ulvospinel (2Ti). (a) SaU 005. Type 1chromites (inclusions in olivine of Fo 74–70) in black. Type 2 com-posite grains of chromite cores with chromian ulvospinel rims (whichoccur as inclusions in olivine of Fo 69–62 and in groundmass),zonation pattern 1, in green. (b) SaU 005. Type 2, zonation pattern 2.(c) SaU 005. Type 2, zonation pattern 3, in black. Reaction rims aroundmelt inclusions in type 2 chromite cores in red. (d) EET-A. Type 1chromites (inclusions in olivine of Fo 81–60) in black. Type 2 com-posite spinels (which occur as inclusions in olivine of Fo 59–52 and inthe groundmass) in green. In contrast to SaU 005, type 2 spinels inEET-A show zonation pattern 1 only. Spinels in ALHA77005 shown inmagenta for comparison.

3744 C. A. Goodrich

rim). Cracks in the cores are more extensively developed intobranching systems, and more easily seen in reflected light thanin SaU 005. Both cores and overall grain shapes range fromanhedral to euhedral. Core sizes range from �80 to 400 �mand overall grain sizes range from �100 to 500 �m.

Core-to-rim compositional zonation is also similar to that ofSaU 005, except that only pattern 1 is observed (Fig. 6d; Fig.10b, profiles 5–7). Cores are similar to type 1 chromites, withnearly constant ulvospinel of �2 to 3% (Fig. 6d), magnetite of�3% (Fig. 7d; Table 4, analyses 14 and 15), and a limitedrange in Cr# variation. However, Cr#s of type 2 cores aredistinctly lower than those of type 1 chromites, whereas in SaU005 they are similar (Figs. 7d and 8d). Also in contrast to SaU005, most cores show significant variation in fe# (Fig. 8d) from�0.77 to 0.88, and are normally zoned (e.g., Fig. 10b, profile5).

Chromian ulvospinel rims (Table 4, analyses 16 and 17)follow a trend of ulvospinel variation (�18–67%) at nearlyconstant Al content (Fig. 6d; Fig. 10b, profiles 5–7). Magnetitecontents (�4–9%) and fe#s (�0.88–0.92) are higher thanthose of rims in SaU 005 (Figs. 7d and 8d), and increase fromcore/rim boundaries to edges of grains with decreasing Cr#(Fig. 10b, profiles 5–7). The gap in ulvospinel content betweencores and rims is larger than for type 2 chromites in SaU 005(Fig. 6); in addition, there appears to be a significant gap inmagnetite content (Fig. 7d; Fig. 10b, profiles 5–7).

Although cores do extend to edges of grains in places (e.g.,Fig. 9c, bottom), no deviations in their composition such aszonation pattern 3 in SaU 005 were observed. Furthermore,reaction rims were not observed around melt inclusions. Somecores contain fine lamellae of a Ti-rich phase, which (as dis-cussed above for SaU 005) may be exsolved ulvospinel orilmenite. Rims commonly have a fine lamellar structure (Fig.9d), which is probably a result of late oxidation-exsolution ofilmenite.

4.2.3. Pyroxene-Spinel intergrowths

Micron to submicron-sized intergrowths of pyroxene andspinel, similar to those in SaU 005, are abundant in olivine ofall compositions in EET-A. No analyses of either the spinel orthe pyroxene were obtained.

4.3. Spinels in ALHA77005

To compare spinels in SaU 005 and EET-A with spinelsknown to be a cumulus phase in another shergottite, spinels inlherzolitic shergottite ALHA77005 were examined. Thosestudied were larger grains (�50–200 �m in size) occurring inassociation with pyroxene and maskelynite (spinel grains in-cluded in oikocrystic olivine were specifically avoided) toprovide the best analogy to type 2 spinels in SaU 005 and

Table 4. Spinels in SaU 005 and EET-A.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

SiO2 0.25 0.31 0.19 0.18 0.17 0.18 0.15 0.19 0.17 0.49 0.18 0.17 0.18 0.24 0.25 0.17 0.11TiO2 0.63 0.71 0.95 1.21 3.91 13.31 1.57 2.34 6.49 0.96 1.41 4.49 1.50 0.66 0.92 6.31 19.23Al2O3 8.68 10.81 8.21 8.97 7.95 7.03 10.00 11.02 9.54 8.29 9.08 11.20 14.25 8.14 7.85 6.16 4.00Cr2O3 56.4 53.0 57.3 56.1 52.5 33.3 53.4 50.9 44.8 57.1 55.4 45.6 50.4 57.4 56.1 45.6 22.7FeO 28.3 29.1 27.6 28.0 30.2 38.8 29.5 30.4 34.1 27.5 28.2 31.1 28.1 28.5 30.7 37.7 49.5MgO 4.49 4.22 4.92 5.03 5.06 5.48 4.48 4.72 4.80 4.83 5.12 5.69 5.33 4.63 3.00 2.92 3.00MnO 0.33 0.40 0.37 0.41 0.37 0.51 0.41 0.42 0.47 0.41 0.36 0.40 0.38 0.40 0.40 0.39 0.55CaO 0.03 0.02 0.01 0.05 0.05 0.18 0.03 0.01 0.03 0.03 0.08 0.20 0.06 0.00 0.00 0.01 0.03ZnO 0.11 0.13 0.05 0.05 0.12 0.03 0.08 0.10 0.07 0.04 0.08 0.06 0.05 0.03 0.07 0.08 0.04V2O3 0.71 0.78 0.79 0.85 0.85 1.96 0.94 1.09 1.30 0.79 0.78 1.11 0.98 0.56 0.66 1.14 2.46Total 99.9 99.5 100.4 100.9 101.2 100.8 100.6 101.2 101.8 100.4 100.7 100.0 101.2 100.6 100.0 100.5 101.6

Cations on the basis of 4 oxygen atoms

Si 0.009 0.011 0.006 0.006 0.006 0.006 0.005 0.006 0.006 0.017 0.006 0.006 0.006 0.008 0.009 0.006 0.004Al 0.353 0.438 0.333 0.360 0.320 0.284 0.402 0.438 0.379 0.335 0.365 0.446 0.556 0.330 0.324 0.255 0.165Cr 1.540 1.441 1.556 1.511 1.416 0.901 1.442 1.356 1.196 1.550 1.493 1.219 1.320 1.561 1.555 1.266 0.629Fe3� 0.045 0.052 0.034 0.038 0.034 0.070 0.045 0.051 0.055 0.027 0.043 0.071 0.018 0.052 0.045 0.107 0.119Ti 0.016 0.018 0.024 0.031 0.100 0.343 0.040 0.059 0.165 0.025 0.036 0.114 0.037 0.017 0.024 0.167 0.507V 0.020 0.022 0.022 0.023 0.023 0.054 0.026 0.030 0.035 0.022 0.021 0.030 0.026 0.015 0.019 0.032 0.069Mg 0.231 0.216 0.252 0.256 0.257 0.280 0.228 0.237 0.241 0.247 0.260 0.286 0.263 0.237 0.157 0.153 0.157Fe2� 0.771 0.786 0.760 0.760 0.827 1.041 0.797 0.807 0.907 0.763 0.761 0.807 0.760 0.767 0.854 1.000 1.331Zn 0.003 0.003 0.001 0.001 0.003 0.001 0.002 0.003 0.002 0.001 0.002 0.001 0.001 0.001 0.002 0.002 0.001Mn 0.010 0.012 0.011 0.012 0.011 0.015 0.012 0.012 0.013 0.012 0.010 0.011 0.011 0.012 0.012 0.012 0.016Ca 0.001 0.001 0.000 0.002 0.002 0.007 0.001 0.001 0.001 0.001 0.003 0.007 0.002 0.000 0.000 0.000 0.001Ulvo 1.7 1.9 2.5 3.1 10.2 35.3 4.1 6.0 16.8 2.5 3.7 11.6 3.8 1.7 2.4 17.0 52.6Sp 17.9 22.3 16.9 18.3 16.2 14.6 20.4 22.3 19.4 17.1 18.5 22.7 28.2 16.7 16.4 13.0 8.6Chr 78.1 73.2 78.9 76.7 71.9 46.5 73.2 69.1 61.0 79.0 75.7 62.0 67.1 79.0 78.8 64.5 32.7Mag 2.3 2.6 1.7 1.9 1.7 3.6 2.3 2.6 2.8 1.4 2.2 3.6 0.9 2.6 2.3 5.5 6.2fe# 0.769 0.784 0.751 0.748 0.763 0.788 0.778 0.773 0.790 0.755 0.745 0.738 0.743 0.764 0.845 0.867 0.895Cr# 0.813 0.767 0.824 0.807 0.816 0.761 0.782 0.756 0.759 0.822 0.804 0.732 0.704 0.826 0.828 0.832 0.792

(1–13) � SaU 005. (14–17) � EET-A. (1) Type 1, center of grain. (2) Type 1, edge of grain. (3–6) Type 2, pattern 1. Points 1, 9, 10 and 14 ofprofile 1 in Fig. 10a. (7–9) Type 2, pattern 2. Points 1, 6 and 8 of profile 2 in Fig. 10a. (10–12) Type 2, pattern 3. Points 1, 5, 7 of profile 3 in Fig.10a. (13) Type 2, reaction rim around melt inclusion in core. Point 0 of profile 4 in Fig. 10a. (14–17) Type 2. Points 1, 12, 13 and 18 of profile 5in Fig. 10b.


EET-A. None of the grains examined show extensive crackssuch as those seen in type 2 chromites in SaU 005 and EET-A,or a textural distinction between cores and rims (Fig. 9f). Ninegrains were analyzed quantitatively. All have similar center-to-edge zonation profiles, with no sharp core/rim distinction (Fig.6d; Fig. 10b, profile 8). Grains have centers with high-Cr,low-Ti compositions similar to type 1 and cores of type 2chromites in EET-A, and zone smoothly with increasing ul-vospinel and magnetite components, decreasing Cr#, andslightly increasing fe#, generally following the trend shown byrims of type 2 chromites in SaU 005 and EET-A. There is nogap in ulvospinel content such as that seen in SaU 005 andEET-A spinels. These compositional data are in agreement withthose reported previously by McSween et al. (1979b) and Ikeda(1994a) for chromites occurring as inclusions in late crystalliz-

ing phases (ferroan pyroxenes and maskelynite) inALHA77005.

5. MELT INCLUSIONS IN OLIVINE AND CHROMITE INSaU 005 AND EET-A

5.1. General

Inclusions were identified optically and then examined bySEM and EMPA. A few objects originally thought to be meltinclusions were subsequently recognized, on the basis of min-eralogy and mineral compositions, to be patches of ground-mass. A few melt inclusions in SaU 005 that show signs ofterrestrial alteration (presence of carbonate veins and/or verylow analytical totals for glasses suggesting the presence ofH2O) were excluded from further study.

Fig. 7. Magnetite content vs. Cr# for spinels in SaU 005 and EET-A. (a) SaU 005. Type 1 chromites (inclusions inolivine). (b) SaU 005. Type 2 composite spinels (chromite cores with ulvospinel rims), zonation patterns 1 and 2. Cores ofdifferent grains shown in different black symbols. Rims of same grains shown by same symbols in blue. (c) SaU 005. Type2 composite spinels, zonation pattern 3 (black) and reaction rims around melt inclusions (red). (d) EET-A. Type 1 chromitesin green. Type 2 composite spinels shown as in (b): cores of different grains � different black symbols, rims of same grains� same symbols in blue.

3746 C. A. Goodrich

Melt inclusions in olivine occur only in a limited range ofhost compositions in both SaU 005 and EET-A (Fig. 5). In SaU005 they occur in the more ferroan olivine (Fo 69–62; average64 2), located in the outer zones of large (�500 �m) crys-tals, or near the centers of smaller crystals; the more magnesianolivine cores (Fo 74–70) of large crystals are free of meltinclusions. In EET-A, they occur in olivine of Fo 76 to 60; themost magnesian (Fo 81–77) cores and the more ferroan (Fo59–53) outer zones of crystals are free of melt inclusions. Inboth rocks, multiple inclusions per crystal are common.

Melt inclusions occur in both type 1 chromites and thechromite cores of type 2 composite spinel grains in both SaU005 and EET-A. Multiple inclusions per grain are common, and

in some cases (observed in SaU 005 only) they are concentratedin zones outlining the core near the core-rim boundary.

General properties of the inclusions are summarized in Table5. In terms of most of these properties, inclusions in olivine inSaU 005 are distinguished from the other three groups. Allinclusions are generally rounded (Figs. 12–15). Inclusions inolivine in SaU 005 range from �10 to 130 (average �60) �min size. Inclusions of the other three groups are smaller: thosein chromite in both SaU 005 and EET-A average �12 �m, andthose in olivine in EET-A average �30 �m.

All inclusions consist principally of pyroxene and “glass.” Insome cases the “glass” is homogeneous; in others it containsblebs and/or dendrites of a nearly pure silica phase. Inclusions

Fig. 8. Fe# vs. Cr# for spinels in SaU 005 and EET-A. (a) SaU 005. Type 1 chromites (inclusions in olivine). (b) SaU005. Type 2 composite spinels (chromite cores with ulvospinel rims), zonation patterns 1 and 2. Cores of different grainsshown in different black symbols. Rims of same grains shown by same symbols in blue. (c) SaU 005. Type 2 compositespinels, zonation pattern 3 (black) and reaction rims around melt inclusions (red). (d) EET-A. Type 1 chromites in green.Type 2 composite spinels shown as in (b): cores of different grains � different black symbols, rims of same grains � samesymbols in blue.


in olivine in SaU 005 include both cases (Fig. 12), which aredistinguished as Type I (pyroxene � type 1 glass) and Type II(pyroxene � type 2 glass � silica phase). In most inclusions ofthe other three groups, interpyroxene areas are so small that itis not possible to determine from BEIs whether the “glass” ishomogeneous (Figs. 13–15). However, the presence of thesilica phase in these areas can be inferred from large hetero-geneitites in SiO2 content, and in a few inclusions (e.g., Figs.

13b and 14a) silica blebs or dendrites can be distinguished inBEI but not resolved by EMPA.

Iron sulfide is a minor phase in inclusions of all groups.Minor phosphate and Fe-Ti oxides (ulvospinel or ilmenite)occur in Type II inclusions in olivine in SaU 005. Inclusions inolivine in EET-A contain grains of low-Ti chromite (Figs. 14band 14d) whose large size precludes crystallization as a daugh-ter mineral (that is, if integrated into the trapped melt compo-

Fig. 9. BEI of spinels in SaU 005, EET-A and ALHA77005. (a, b) SaU 005. Type 2 composite grains of low-Ti chromitecores with chromian ulvospinel rims, which occur as inclusions in olivine of Fo 69 to 62 and in groundmass. Melt inclusions(round, black) in cores. Numbered lines correspond to compositional profiles shown in Figure 10a. (c, d) EET-A. Type 2composite grains of low-Ti chromite cores with chromian ulvospinel rims, which occur as inclusions in olivine of Fo 59–52and in groundmass. Melt inclusion (round, black) in core in (d). Numbered lines correspond to compositional profiles shownin Figure 10b. (e) SaU 005. Micron to submicron-sized exsolutions of magnetite-rich spinel and pyroxene in olivine. (f)ALHA77005. Numbered line corresponds to compositional profile shown in Figure 10b.

3748 C. A. Goodrich

Fig. 10. Representative center-to-edge compositional profiles for spinels in SaU 005, EET-A, and ALHA77005. (a) SaU005, type 2 composite spinels (chromite cores with ulvospinel rims). Profiles 1 and 4 show zonation pattern 1. Profiles 2and 3 show zonation patterns 2 and 3 respectively. Profile 4 begins at a melt inclusion, others do not. (b) type 2 compositespinels (chromite cores with ulvospinel rims) in EET-A, and spinel in ALHA77005. Positions of some profiles marked onBEI in Figure 9. Ulvospinel � 100 � molar 2Ti/(2Ti � Cr � Al); magnetite � 100 � molar Fe3�/(Fe3� � 2Ti � Cr �Al); Cr# � molar Cr/(Cr � Al); fe# � molar Fe2�/(Fe2� � Mg).


sition they would result in a melt with an unrealistically highCr2O3 content) and so were probably trapped as solid grainsalong with melt. One inclusion in chromite in SaU 005 containsa grain of olivine, which can be identified from its composition(see below) as a daughter mineral.

Point-counting showed that inclusions in olivine in SaU 005contain 30 to 60 vol.% pyroxene, which occurs as thin rims andskeletal/dendritic crystals (Fig. 12). Inclusions in the otherthree groups contain 50 to 90 (on average �70) vol.% pyrox-ene, which occurs as thick rims and/or massive/blocky crystals(Figs. 13–15) and only rarely as skeletal/dendritic crystals (e.g.,Figs. 13c and 15a).

5.2. Compositions

5.2.1. Inclusions in Olivine in SaU 005

Pyroxenes in inclusions in olivine in SaU 005 have high Wo(44–56), Al2O3 (�7–15%) and TiO2 (�1.1–3.6%), and lowNa2O (�0.3 0.07%) and Cr2O3 (0.15 0.1%) contents (Fig.16). They contain significant amounts of phosphorus (up to 2%P2O5), which from structural formula calculations appears to besubstituting for Si (as in other cases of pyroxenes crystallized insmall closed systems; e.g., Goodrich, 1984). Their Fe/Mg ratiosaverage 0.6 to 0.7 (it is evident from BEIs that they arenormally zoned, but the crystals are so small that the full rangeof zonation was not recorded in the analyses). Pyroxenes inType I inclusions show slightly less compositional variationthan those in Type II inclusions (Fig. 16). This apparent dif-ference may be only a result of the smaller number of analysesfor Type I inclusions (25 vs. 83). Average compositions forType I and Type II pyroxenes are given in Table 6. The TypeII average is based on only the 52 analyses that have excellentcation totals. The Type I average includes analyses of slightlylesser quality, all of which show slight excesses of SiO2 that areprobably due to overlap with surrounding glass. Aside from theeffect of this, and a slight difference in Fe/Mg ratio, the twoaverages show only small differences that do not appear to besignificant.

Glasses in Type I inclusions (type 1 glass) in olivine in SaU005 show a relatively homogeneous composition (Fig. 16,Table 6), both within and among inclusions. It contains �68%SiO2, 17.5% Al2O3, and 8% CaO, and has an extremely lowCr2O3 (�0.02%) content. FeO and MgO contents are also verylow (�2.0 and 0.2%, respectively), and Fe/Mg ratios rangefrom �2 to 24 (with large errors). The only significant variationit shows is in Na2O content, which ranges from 1.5 to 3.2%(inclusion averages). This variation is not an analytical artifact(see section 2), and most likely results from variable degrees oflate volatile loss (shock-induced?). The highest observed value(3.2%) is taken to be the best estimate of Na2O content beforethis loss (Table 6).

Analyses of glass (type 2 glass) and silca-rich blebs in TypeII inclusions together form extensive compositional trends thatpass through the composition of type 1 glass (Fig. 16). Broad-beam analyses intended to sample mixes of type 2 glass and thesilica-rich phase also fall on these trends and cover nearly thesame range of compositions. It is inferred from these trends thattype 2 glass and silica-rich blebs have a complementary rela-tionship, which can be described by the equation: type 1 glass� 70 to 80% type 2 glass � 20 to 30% silica phase.

Most analyses of the silica-rich blebs probably have someoverlap with glass, as the blebs are generally small, and thepure silica phase is probably represented only by the mostsilica-rich analyses (�95% SiO2, 5% Al2O3). Jagoutz (1989)

Fig. 11. Comparison between zonation pattern 3 and reaction rimsaround melt inclusions for cores of type 2 composite spinels in SaU005. The two patterns are similar in showing enrichment of Al and Ti,and depletion of Cr. In addition, both show an increase in fe#. Theydiffer in that reaction rims around melt inclusions show a depletion inmagnetite component, while pattern 3 shows an enrichment. Bothpatterns appear to result from reaction of low-Ti chromite cores withevolved liquid. Ulvospinel � 100 � molar 2Ti/(2Ti � Cr � Al �Fe3�); magnetite � 100 � molar Fe3�/(2Ti � Cr � Al � Fe3�);chromite � 100 � molar Cr/(2Ti � Cr � Al � Fe3�); spinel � 100� molar Al/(2Ti � Cr � Al � Fe3�).

3750 C. A. Goodrich

and Harvey et al. (1993) found evidence that a similar phase inmelt inclusions in shergottites ALHA77005 and LEW 88516 iscristobalite. Variation in the composition of type 2 glass prob-ably results from differences in the amount of silica phaseformed (rather than overlap) because analyses of glass withineach inclusion are fairly uniform (with the exception of twoinclusions [e.g., Fig. 12d] in which the “glass” itself appears tobe a very fine-grained mixture of glass � silica phase).

The only element for which the above relationship does notalways hold is Ca, which is too low in glasses of some Type IIinclusions (Fig. 16). This appears to be due to the presence inthese inclusions of small patchy areas containing feathery crys-tallites and having high CaO contents, which suggests thatincipient crystallization of further pyroxene in these areas de-pleted the remaining glass in Ca.

Based on these observations it is concluded that, to first-order, all inclusions in olivine in SaU 005 represent a singletrapped liquid composition and crystallized approximately thesame amount of pyroxene, leaving a residual liquid representedby type 1 glass. In Type II inclusions this liquid further sepa-rated out a silica-rich phase, and sometimes crystallized a smallamount of additional pyroxene.

5.2.2. Inclusions in Chromite in SaU 005

Both pyroxenes and glasses in inclusions in chromite in SaU005 (Fig. 17) are compositionally distinct from those in inclu-sions in olivine. Pyroxenes have a broader range of lower Wocontents (�8–43, avg. � 34), significantly lower Al2O3 (1.7–6.6%) and TiO2 (0.2–1.9%), higher Cr2O3 (1.0–2.7%), andrelatively uniform lower Fe/Mg ratios (0.47 0.1). Theycontain similar amounts of phosphorus, which again appears tobe substituting for Si.

Most “glasses” are heterogeneous and analyses show com-positional trends (Fig. 17) indicating that they are mixes of trueglass and a silica-rich phase similar to that in inclusions inolivine (in one inclusion silica blebs can be seen in BEI, butareas of silica and glass cannot be resolved by EMPA: Fig.13b). All have SiO2 contents greater than or equal to that (68%)of type 1 glass in inclusions in olivine in SaU 005 and extend-ing to nearly pure analyses of the silica-rich phase (95% SiO2),

suggesting that this phase is more abundant than in inclusionsin olivine and that the bulk glass has higher SiO2. This issupported by averages of “glass” analyses for inclusions fromwhich several analyses were obtained, which show 74 to 75%SiO2 (Table 6, column 6), and also by a high modal abundanceof silica blebs in the one inclusion in which they can be seen.In one exceptionally large inclusion (Fig. 13a) the glass ishomogeneous, with 74% SiO2, (Table 2, column 7), and maytherefore be a good representative of the bulk glass composi-tion for all inclusions. In addition to having higher SiO2 thantype 1 glass in inclusions in olivine, it has lower Al2O3

(�16.5%), CaO (2.4%) and Na2O (2.6%), higher Cr2O3 (0.5–0.6%), and much lower Fe/Mg (1.4). Broad-beam analyses ofthe inclusions are consistent with mixes of the observed py-roxenes and “glasses” (Fig. 17).

The one grain of olivine that occurs in an inclusion inchromite can be identified as a daughter mineral (rather than agrain of primary olivine trapped along with melt) because itcontains significant phosphorus (�0.75% P2O5) that appears tobe substituting for Si (see Goodrich, 1984).

5.2.3. Inclusions in Olivine in EET-A

Although melt inclusions were observed in olivine of Fo 76to 60 in EET-A, most of the data were obtained from inclusionsin Fo 76 to 67. Of the inclusions in Fo 67 to 60 host compo-sitions, only two (both of which are in Fo 60) were largeenough to yield usable data.

Compositions of the majority of pyroxenes in these inclu-sions are similar to those of pyroxenes in inclusions in chromitein SaU 005 in Al2O3, TiO2, Na2O and P2O5 contents (Figs. 17and 18), and in this regard are distinct from pyroxenes ininclusions in olivine in SaU 005 (the only analyses which fallin the compositional range of pyroxenes in olivine in SaU 005are those from the two inclusions in Fo 60 olivine). However,they show a bimodal distribution of low-Ca (Wo 3–13) andhigh-Ca (Wo 40–48) compositions (one analysis with interme-diate Wo may be a result of overlap), and have low Cr2O3

contents similar to pyroxenes in inclusions in olivine in SaU.Their Fe/Mg ratios are relatively homogeneous (0.40 0.10).

“Glasses” in most inclusions are also similar to “glasses” in

Table 5. General properties of melt inclusions in SaU 005 and EET-A.

SaU 005 in olivine (Fo 69–62)a

SaU 005 in low-Ti chromite(Fo 74–70)b

EET-A in olivine(Fo 76–60)a

EET-A in low-Ti chromite(Fo 81–60)bType I Type II

Number 8 17 23 13 12Size (avg.) 15–90 (50) �m 10–130 (60) �m �3–70 (12) �m 20–50 (30) �m 5–30 (12) �mDaughter phases High-Ca pyx.

Type 1 glass.High-Ca pyx.Type 2 glass.Silica phase.

Low to high-Ca pyx.Glass.

Silica phase.

Low & high-Ca pyx.Glass.

Silica phase.

Intermediate-Ca pyx.Glass.

Silica phase.Minor phases Fe-sulfide. Fe-sulfide.

Phosphate.Ti-Fe oxide.

Fe-sulfide. Fe-sulfide.Low-Ti chromite.

Fe-sulfide.

Pyx morphology Thin rims.Skeletal.

Dendritic.

Thin rims.Skeletal.

Dendritic.

Massive.Blocky.

Rarely skeletal.

Thick rims.Massive.Blocky.

Massive.Blocky.

Rarely skeletal.Vol.% pyx (avg.) �30–60 (�40) �30–60 (�40) �50–80 (�70) �50–80 (�70) �50–90 (�70)

a Range of olivine compositions in which melt inclusions occur.b Range of compositions of olivine containing low-Ti chromites (without ulvospinel-rich rims).


inclusions in chromite in SaU 005 (Figs. 17 and 18), and appearto be mixes of a silica-rich phase and silica-depleted glass.Their SiO2 contents are, again, greater than or equal to that oftype 1 glass in inclusions in olivine in SaU 005, suggesting thatthe bulk glass has higher SiO2 than that glass. Unfortunately,no inclusions appear to contain homogeneous glass. The bestestimate of the bulk glass composition, given by the average ofanalyses from a typical inclusion, is �77% SiO2, 15% Al2O3

and 1.5% CaO (Table 7, column 2). The two inclusions thatoccur in Fo 60 olivine are exceptional in that they containsilica-rich blebs and areas of silica-depleted glass (SiO2 �68%) large enough to resolve by EMPA (Fig. 14d). Both showa low abundance of the silica phase, indicating bulk glasscompositions with lower SiO2 than other inclusions.

5.2.4. Inclusions in Chromite in EET-A

Only a few analyses of discrete pyroxenes and “glasses”were obtained for inclusions in chromite in EET-A. Both aresimilar to those in inclusions in chromite in SaU 005 (Fig. 18),though pyroxenes show only intermediate Wo contents (17–36). “Glass” analyses show a large range in Si/Al ratio, andclearly reflect extremely unrepresentative sampling of silica-depleted glass and the silica-rich phase. Unfortunately, no

single inclusion provided a plausible sample of bulk glass. Thecompositions of these inclusions are also represented by broad-beam analyses, which are consistent with mixes of the observedpyroxenes and glasses (Fig. 18).

5.3. Present Bulk Compositions (pbcs)

The present bulk composition (pbc) of an inclusion is definedto be the bulk silicate composition of its presently visibleportion. Pbcs were constructed from the petrographic observa-tions described above (ignoring minor phases such as phos-phates) and examined for consistency between their predictedearly crystallization sequences as determined by MAGPOX(Longhi, 1991) and the observed mineralogy (principally py-roxene types) of the inclusions.


As discussed above, it appears that Type I and Type IIinclusions in olivine in SaU 005 have the same pbc, which canbe calculated simply as 40 vol.% average type 1 glass � 60vol.% average pyroxene, using the highest observed Na2Ocontent for type 1 glass and the average pyroxene from Type IIinclusions (Table 6). Glass and pyroxene volume proportions

Fig. 12. BEI of inclusions in olivine (Fo 69–62) in SaU 005. (a-b) Type I inclusions consisting of pyroxene (thin rimsand skeletal crystals) and homogeneous glass. Bright spherule in (b) is iron sulfide. (c-d) Type II inclusions consisting ofpyroxene (thin rims and skeletal crystals), a silica-rich phase (black blebs and dendrites), and silica-depleted glass. In (d)the “glass” appears to contain fine quench needles of the silica-rich phase.

3752 C. A. Goodrich

were weighted by reasonable estimates of density (�pyx � 3.3;�glass � 2.3) to give weight proportions. The resulting compo-sition is given in Table 6 (column 4) and shown in the Olivine–Quartz-Plagioclase (Ol-Qtz-Plag) phase system in Figure 19a.It is saturated only with augite (not visible in Fig. 19a), con-sistent with the presence of high-Ca pyroxene as the onlydaughter phase in the inclusions.


The pbc of inclusions in chromite in SaU 005 can be calcu-lated as 70 vol. % average pyroxene � 30 vol. % bulk glass.Two possible compositions for the bulk glass were used, yield-ing two possible pbcs. “Bulk glass” 1 (Table 6, column 6) is anaverage of mixed analyses of glass � silica phase from atypical inclusion. “Bulk glass” 2 (Table 6, column 7) is thehomogeneous glass in the large inclusion shown in Figure 13a.“Bulk glass” 1 has slightly higher SiO2 and lower Al2O3 than“bulk glass” 2; otherwise they are similar. Volume proportionswere weighted by densities, using the values given above, toyield weight proportions. Pbc1 and Pbc2 are given in Table 6(columns 8 and 9) and shown in the Ol-Qtz-Plag system in

Figure 19b. A third pbc estimate (pbc3) is provided by theaverage of all broad-beam analyses (Table 6, column 10). Itappears from this average that broad-beam analyses sampledless pyroxene than the average abundance indicated by point-counting, and in fact this composition can be satisfactorilymodelled as a mixture of �52 vol.% average pyroxene � 48vol.% average glass. Pbc1 and Pbc2 are saturated only withpigeonite and evolve to pigeonite-augite cosaturation, consis-tent with the presence of only intermediate- and high-Ca py-roxenes in the inclusions. In contrast, pbc3 is orthopyroxene-saturated. Therefore pbc1 and pbc2 appear to be more accurateestimates of the bulk composition of these inclusions and willbe used for the reconstruction of the primary trapped liquid(PTL) composition (with pbc3 providing an estimate of theuncertainty in CaO content). These compositions differ signif-icantly from the pbc of inclusions in olivine, in having lowerAl2O3, CaO, Na2O and TiO2 contents, and higher Si/Al ratios(Table 6).


Because the inclusions in olivine in EET-A occur in a broadrange of host compositions (Fo 76–60), it seems likely that

Fig. 13. BEI of inclusions in chromite in SaU 005. All inclusions consist of pyroxene and “glass.” Pyroxene occurspredominantly as massive or blocky crystals, and only rarely (c) as skeletal/dendritic crystals. The “glass” is heterogeneousin SiO2 content and in most cases appears to be a mixture of a silica-rich phase and silica-depleted glass. In rare cases (b),the silica-rich phase can be distinguished in BEI (contrast of this image has been enhanced to show this). The inclusion in(a) is exceptionally large, and unusual in that the glass is homogeneous. This chromite grain is the same one as shown inFigure 9a. Pit in inclusion in (c) is SIMS analysis spot.


they sample an evolving melt and should have a range of pbcs.However, in general the data obtained here are not adequate todistinguish differences in bulk composition among the inclu-sions because no single inclusion is completely sampled.Therefore, only an average pbc can be obtained. There is,however, direct evidence for distinct bulk compositions is thatthe two inclusions that occur in Fo 60 have significantly morealuminous pyroxenes and bulk glasses than those in Fo 76 to 67(consistent with their representing a more evolved melt). There-fore, because the interest here is primarily in the earliestmagma, which would have been preserved most closely in themore magnesian olivine, these two inclusions have been elim-inated from the data used to calculate the pbc.

The pbc is calculated as 70 vol.% average pyroxene � 30vol.% bulk glass, weighted by the densities used above (Table7). The resulting composition (Fig. 19c) is saturated only withorthopyroxene and evolves to orthopyroxene-augite cosatura-tion, consistent with the bimodal distribution of low- andhigh-Ca pyroxene compositions in the inclusions. It is similarto the pbc of inclusions in chromite in SaU 005 in having lowAl2O3 and high Si/Al compared to inclusions in olivine in SaU005.


An estimate of the pbc (pbc1) of these inclusions is made as70 vol. % average pyroxene � 30 vol. % average glass (Table7; Fig. 19c). The average of all broad-beam analyses (Table 7;Fig. 19c) provides another estimate (pbc2). Pbc2 has higherAl2O3 (and lower Si/Al) than pbc1, and also (as was the casefor inclusions in chromite in SaU 005) appears from Figure 19cto have a smaller pyroxene (mafic) component. However, inthis case, increasing its pyroxene content (by addition of py-roxene from inclusions in chromite) cannot result in a compo-sition like pbc1, because its CaO content is already too high(Table 7). Therefore, the PTL will be calculated from pbc1,taking the upper limit on Al2O3 content from pbc2, and thelower limit on CaO content from the lowest pyroxene/glassratio (�60 vol.% pyroxene) that would still result in a pigeo-nite-saturated composition (these limits are shown as an ellipsein Fig. 19c). The pbc of inclusions in chromite is clearly similarto that of inclusions in olivine in EET-A and inclusions inchromite in SaU 005 in having low Al2O3 and high Si/Alcompared to the pbc of inclusions in olivine in SaU 005 (Tables6 and 7; Fig. 19).

Fig. 14. BEI of inclusions in olivine (Fo 76–60) in EET-A. Inclusions consist of pyroxene and “glass.” The “glass” isheterogeneous in SiO2 content and appears to be a mixture of a silica-rich phase (contrast in [a] has been enhanced to showthis) and silica-depleted glass. Pyroxene occurs predominantly as thick rinds or massive/blocky crystals, and only rarely asskeletal/dendritic crystals. Bright grains in (b) and (d) are low-Ti chromite. The inclusion in (d) is one of two exceptionalinclusions (both in Fo 60 olivine) in which blebs of the silica-rich phase are large enough to analyze. These two inclusionshave notably less pyroxene and more aluminous compositions than those in more magnesian olivine.

3754 C. A. Goodrich

5.4. Primary Trapped Liquid (PTL) Compositions

5.4.1. General

Primary trapped liquid compositions are reconstructed frompbcs by addition of that portion of the surrounding host mineralthat crystallized from the trapped melt onto the original walls ofthe inclusions, plus corrections for any chemical exchangereactions that occurred between inclusions and their hosts.Crystallization of the host mineral as the sole wall phase occursnaturally, because a primary trapped melt must be saturatedwith the host mineral. The host offers a ready nucleation site,and nucleation of other phases is commonly suppressed. Meltstrapped in olivine, for example, tend to crystallize large vol-umes of wall olivine in excess of the equilibrium amount. In allbut the most rapidly quenched cases, this olivine quickly re-equilibrates Fe and Mg with its host (or even with the largerbody of magma surrounding the host crystal) and concommi-tant reequilibration of the residual melt in the inclusion se-verely depletes it in FeO (Danyushevsky et al., 2000; Gaetaniand Watson, 2000). No other important exchange reactionsoccur, as olivine does not accommodate significant quantititesof any other cations, and hence serves as a relatively impervi-ous container. Thus, reconstruction of the PTL for an inclusionin olivine involves only addition of olivine and exchange of Mg

for Fe. The final Fe/Mg ratio is defined by the requirement ofequilibrium with olivine of host composition. The amount ofolivine to be added, however, can only be determined if (1) theFeO content of the PTL is known by independent means(Danyushevsky et al., 2000) or (2) an additional constraint,such as co-saturation of the PTL with a second phase, isavailable.

In contrast, for melt inclusions in chromite crystallization ofthe host phase onto inclusion walls is negligible (Kamenetsky,1996), being limited by the low solubility of Cr2O3 in spinel-saturated basaltic melts (Roeder and Reynolds, 1991). Hence,for quenched inclusions the pbc essentially preserves the com-position of the PTL. Commonly, however, sufficiently rapidcooling does not occur, and there can be Fe/Mg exchangebetween inclusions and their hosts. In addition, if water ispresent in the melt (even in minute quantities) a closed-systemreaction will cause oxidation of FeO in the melt (2FeO � H2O3 Fe2O3 � H2) and exchange of this Fe for Cr and/or Al in thespinel (Fe2O37 Cr2O3, Al2O3), hence depleting the inclusionin FeO and enriching it in Cr2O3 and/or Al2O3 (Zlobin et al.,1990). This reaction generally leaves no record (e.g., zonation)in the chromite, which is effectively an infinite reservoir.These effects must be reversed to derive PTL compositions.The proper Fe/Mg ratio can be determined by the require-

Fig. 15. BEI of inclusions in chromite in EET-A. Inclusions consist of pyroxene and “glass.” The “glass” is heterogeneousin SiO2 content and appears to be a mixture of a silica-rich phase and silica-depleted glass. Pyroxene occurs as thick rindsand massive/blocky crystals, and only rarely as skeletal/dendritic crystals (a). Contact between chromite core and ulvospinelrim can be seen in (b).


ment of equilibrium with olivine inferred to have cocrystal-lized with the host chromites. The amount of FeO that hasbeen lost can be determined from the constraint of olivinesaturation and/or estimated from the Al2O3 and Cr2O3 con-tents of the inclusions (it can be assumed that latter shouldbe zero, and the former may be known by independentmeans).


The Fe/Mg ratio of these inclusions is low (0.73, implyingequilibrium with olivine of Fo � 81) indicating that they have

reequilibrated with their hosts, and so must be corrected in thereconstruction of the PTL. The choice of final Fe/Mg ratio isbased on the average host olivine composition (Fo 64 2;Fe/Mg � 0.56), and the PTL is defined to have Fe/Mg � 1.6(KDol/liq � 0.36). The amount of olivine to add is determinedby the requirement that the PTL be cosaturated with olivine andat least one other phase, which is reasonable because the hostolivines are relatively Fe-rich and the groundmass containspyroxenes of compositions appropriate for cocrystallizationwith such olivine. It can be seen from Figure 19a that the PTLmust be close to cosaturation with olivine, low-Ca pyroxene

Fig. 16. Compositions of pyroxenes, glasses, and silica-rich phase in inclusions in olivine in SaU 005. All oxides inwt.%.

3756 C. A. Goodrich

and plagioclase, and inspection of the system orthopyroxene-plagioclase-wollastonite (Opx-Plag-Wo) shows that it must beclose to cosaturation with augite, low-Ca pyroxene and plagio-clase. Addition of �20 mol.% olivine results in a compositionwhich (after correction of Fe/Mg) is co-saturated with olivineand pigeonite (Wo 9.6, mg 69) and within a few percent ofcrystallization of being saturated with augite. This PTL is givenin Table 8, and is shown in the Ol-Plag-Qtz system in Figure 20and the Opx-Plag-Wo system in Figure 21.


The Fe/Mg ratio of these inclusions is also low (0.38, im-plying equilibrium with olivine of Fo 88–89), indicating thatthey have reequilibrated with their hosts, and so must becorrected. In this case, the PTL is assigned an Fe/Mg ratio(0.83) based on equilibrium with the most magnesian olivine(Fo 76) in which the inclusions occur, to yield an approxima-tion to the earliest magma they represent. The amount ofolivine to add is determined by the requirement that the PTL becosaturated with olivine and low-Ca pyroxene, which is appro-priate because of the presence in EET-A of low-Ca pyroxenethat appears to be in equilibrium with Fo 76 olivine. Additionof �20 mol.% olivine results in a composition cosaturated witholivine of Fo � 76 and low-Ca pyroxene (Wo 4) of mg � 78.This PTL is given in Table 8 and shown in Figures 20 and 21.

A previous estimate of the composition of melt trapped inolivine in EET-A, made by Treiman et al. (1994b), has muchhigher Al2O3 and lower Si/Al than the PTL determined here(Fig. 20). This discrepency can be explained by the observationthat most of the data of these authors were obtained from thin

section . . .,68 (Treiman, private communication), in which Ifound that all melt inclusions in olivine occur in Fo 64 to 60,the most ferroan range of host compositions. These inclusionsare more aluminous than those in Fo 77 to 67 and, as discussedabove, were not included in my PTL reconstruction. Thisconfirms the supposition that melt inclusions in olivine inEET-A record an evolving magma composition, and shows thatthe composition of Treiman et al. (1994b) represents (except inmg, which was incorrectly chosen for equilibrium with the mostmagnesian olivine in EET-A) a later stage of this evolution thanthe PTL reconstructed here.


Because of evidence (presented above and discussed in detailbelow) that the low-Ti chromites crystallized before the earliestolivine in SaU 005, it can be assumed that the melt they trappedwas in equilibrium with olivine of Fo 74 or higher. However,the present Fe/Mg ratio of the inclusions is dominated by thatof their pyroxenes (�0.47), which would be in equilibriumwith olivine of Fo 69 or lower, and the one grain of olivine thatoccurs in an inclusion is Fo 60. Most of the low-Ti chromitecores have relatively homogenous Fe/Mg ratios of �3.5 to 3.7,similar to those of the earliest ulvospinel-rich rims. Because theulvospinel-rich phase occurs in olivine (as rims around low-Ticores) only of Fo 69 or lower, it is inferred that it did not beginto crystallize until olivine compositions had evolved to thispoint. It therefore appears that both chromite cores and theirmelt inclusions reequilibrated Fe/Mg with the ulvospinel phaseand/or olivine of Fo 69 or lower (at temperatures of �850°C orlower, according to the olivine-spinel geothermometer of Fab-

Table 6. Melt inclusions in SaU 005.

Inclusions in olivine Inclusions in chromite

Type I, avg.pyx (25)a

Type II, avg.pyx (52)b

Type I,avg. glassc pbcd Avg. pyx (33) “Bulk glass” 1e “Bulk glass” 2f pbc1g pbc2h pbc3 (avg. bb)i

SiO2 46.1 45.5 67.9 56.8 50.0 75.2 73.9 55.8 55.1 59.0TiO2 2.3 2.3 0.3 1.3 0.72 0.33 0.47 0.63 0.66 0.60Al2O3 9.8 10.3 17.5 13.9 4.31 13.2 16.5 6.4 7.1 9.1Cr2O3 0.15 0.15 0.02 0.09 1.75 0.51 0.58 1.5 1.5 1.7FeO 10.4 10.4 2.0 6.1 12.5 0.75 0.89 9.8 9.8 7.2MgO 8.4 9.5 0.20 4.7 14.8 0.33 0.35 11.5 11.4 8.6MnO 0.25 0.29 0.06 0.17 0.46 0.02 0.04 0.36 0.36 0.31CaO 22.1 21.7 7.9 14.6 15.9 2.0 2.4 12.7 12.7 11.7K2O 0.01 0.01 0.08 0.05 0.01 0.20 0.20 0.05 0.06 0.12Na2O 0.29 0.24 3.2 1.7 0.23 2.42 2.6 0.73 0.79 0.73P2O5 0.71 0.48 0.84 0.67 0.54 0.67 0.90 0.57 0.62 0.57Total 100.5 100.9 100.0 100 101.2 95.6 98.8 100 100 99.6Wo 52.6 50.5 34.3Fe/Mgj 0.69 0.62 5.7 0.73 0.47 1.3 1.4 0.48 0.48 0.47Si/Alj 6.9 14.8 13.2 11.0

a Average of all analyses with cation totals (� Fe � Mg � Ca � Mn � Na � Cr � Ti � viA1, on the basis of 6 oxygen atoms) � 1.95–2.01.b Average of analyses with cation totals � 1.98–2.01.c Average of glass averages for 8 Type I inclusions, with Na2O adjusted to 3.2%.d 49% Type II avg. pyx � 51% Type I avg. glass (by wt).e Average analysis for glass-silica phase mix in typical inclusion.f Average of 6 analyses of homogeneous glass in large inclusion with no silica phase (Fig. 13a).g 77% avg. pyx � 23% “bulk glass” 1 (by wt).h 77% avg. pyx � 23% “bulk glass” 2 (by wt).i bb � broad-beam analysis.j Molar ratios.


ries, 1979). To correct for this reequilibration, the Fe/Mg ratioof the PTL is assigned a value (�0.95) appropriate for equi-librium with Fo � 74, to provide a minimum estimate of its mg.

Furthermore, the pbc is not saturated with olivine (ratheronly with low-Ca pyroxene) and olivine is present as a daughtermineral in the inclusions only in one rare case. This can beaccounted for by loss of FeO from the inclusions via thereaction discussed, above in which Fe2� is oxidized to Fe3�

and exchanged for Cr3� and/or Al3� from the host chromite(Zlobin et al., 1990). The high Cr2O3 content of the inclusions(1.5%, when it should be �0%), provides clear evidence thatthis reaction has occurred. In addition, the observation that latereaction rims around the inclusions show a deficit in magnetite

component, while analogous reaction rims adjacent to ground-mass show an increase (Fig. 11), can be explained by earlydepletion of the trapped melts in Fe3� (and in H2O necessaryfor the build-up of Fe3� in late liquids).

Estimating the amount of FeO-loss by constraining the PTLto be minimally saturated with olivine (that is, cosaturated witholivine plus low-Ca pyroxene), results in (after correction ofFe/Mg ratio) the PTL composition given in Table 8 and shownin Figures 20 and 21. It is in equilibrium with olivine of Fo �74 and low-Ca pyroxene (Wo 4) of mg � 78. However, theamount of FeO-loss estimated in this calculation is greater thancan be accounted for by the Cr2O3 in the inclusions via theabove reaction. If some Al2O3 was also acquired by reaction

Fig. 17. Compositions of pyroxenes and glasses, and broad-beam analyses for inclusions in chromite in SaU 005. Alloxides in wt.%. Scales same as in Figure 16.

3758 C. A. Goodrich

with host chromite it would account for additional FeO-loss.Unfortunately, there is no way to constrain the amount ofAl2O3-gain, but the Al2O3 content of the inclusions is alreadyfairly low and the amount gained is unlikely to be very high(nevertheless, the Al2O3 content of the reconstructed PTL istaken as an upper limit, as shown by the uncertainties inplagioclase component in Figs. 20 and 21). One possibility toaccount for this FeO discrepency is that, in addition to Fe3�-Cr3� exchange between inclusions and hosts, a rim of puremagnetite was precipitated (if all the FeO addition necessary torestore the pbc to olivine saturation was in the form of mag-

netite, this magnetite would nevertheless be volumetricallyminor –constituting, for example, a rim 0.1 �m wide around a10-�m diameter inclusion). This would result in significant lossof FeO from the inclusions, with no effect on other elements.Such a rim is unlikely to be preserved, as it would be volu-metrically insignificant relative to the chromite host and wouldeasily reequilibrate with the adjacent chromite. Thus, althoughthe record of FeO-loss from the inclusions cannot be docu-mented in detail, the proposed mechanism is plausible, andsince FeO-loss is the only known effect that could account forthe failure of the pbc to be olivine-saturated, reconstruction of

Fig. 18. Compositions of pyroxenes and glasses in inclusions in olivine in EET-A. Compositions of pyroxenes andglasses, and broad-beam analyses for inclusions in chromite in EET-A. Oxides in wt.%. Scales same as in Figures 16 and17.


the PTL simply by addition of sufficient FeO to achieve olivinesaturation is considered appropriate. The resulting composition,of course, represents only the minimum olivine component ofthe PTL (Fig. 20).


Because of evidence (presented above and discussed in detailbelow) that the low-Ti chromites in EET-A crystallized beforeor with the most magnesian olivine, we can assume that themelt they trapped was in equilibrium with olivine of Fo 81 orhigher. However, the inclusions have experienced Fe/Mg re-equilibration. Their pyroxenes have Fe/Mg ratios (�0.64) in-dicating equilibrium with olivine of �Fo 59; the chromite coresare zoned with Fe/Mg ratios ranging (center to core/rim bound-ary) from �3.5 to 7.3, with the outer zones being similar to theulvospinel-rich rims; and the ulvospinel-rich rims occur inolivines only of Fo 59 or lower. Therefore, it appears that bothchromite cores and their melt inclusions equilibrated Fe/Mgwith the ulvospinel-rich phase and/or olivine of Fo 59 andlower (at temperatures of �1000–800°C, according to theolivine-spinel geothermometer of Fabries, 1979). To correct forthis reequilibration, the PTL is assigned an Fe/Mg ratio (0.65)appropriate for equilibrium with Fo 81 olivine.

In addition, the pbc of these inclusions is not olivine-satu-rated and has high Cr2O3 (1.6%), indicating that they have lostFeO to and gained Cr2O3 (and possibly Al2O3) from their hosts.This can account for the higher Cr2O3 and possibly also thehigher Al2O3 content of their pbc compared to that of theinclusions in olivine. The PTL is calculated from the pbc byaddition of FeO/MgO to achieve minimum olivine saturation(that is, cosaturation with olivine plus low-Ca pyroxene), fol-lowed by adjustment of Fe/Mg ratio. The resulting composi-tion, which is in equilibrium with olivine of Fo � 81 andorthopyroxene (Wo 3) of mg � 83, is given in Table 8 andshown in Figures 20 and 21. Again, this composition represents

only the minimum mg value and olivine component of the PTL(Fig. 20).

6. DISCUSSION

In sections 6.1 to 6.3, I discuss observations concerningolivine, pyroxenes and spinels in SaU 005 and EET-A thatallow reconstruction of the crystallization histories of theserocks, and in section 6.4 I summarize those histories. Insection 6.5 I examine the crystallization sequences predictedfor the trapped melt compositions determined in this work,and compare them to the crystallization histories of therocks. In sections 6.6 and 6.7, I discuss models for thepetrogenesis of SaU 005 and EET-A and consider possiblerelationships between these rocks and the basaltic and lher-zolitic shergottites.

6.1. Olivine

The CSD function for olivine crystals in SaU 005 (Fig. 2c)shows three regimes, which can be interpreted in terms of CSDtheory (Marsh, 1988). The linear form and negative slope of thecurve in the size range 0.25 to 1.5 mm (encompassing �95% ofall crystals) indicates a period of continuous nucleation andgrowth. The sharp dropoff of the curve for crystals � 0.25 mmindicates either a decrease in nucleation rate due to loss of meltat high degrees of crystallization, or a cessation of nucleationdue to phase equilibria but with continued growth (annealing ofsmaller crystals). The horizontal form of the curve for crystals� 1.5 mm in size (�5% of all crystals) indicates an overabun-dance of large crystals that may result from entrainment ofcumulates (previously grown crystals from the same magma),or more complex processes such as addition of xenoliths oradmixture of a phenocryst-bearing magma. Combining theseinterpretations with the correlation of crystal size and compo-sition shown in Figure 2e, leads to the conclusion that crystals

Table 7. Melt inclusions in EET-A.

Inclusions in olivine Inclusions in chromite

Avg. pyx (27) “Bulk glass” a pbcb Avg. pyx (6) Avg. “glass” (6) pbc 1b pbc 2 (avg. bbc)

SiO2 51.5 76.9 57.7 49.5 79.8 56.2 58.6TiO2 0.33 0.07 0.27 0.43 0.11 0.35 0.30Al2O3 3.91 14.6 6.5 4.6 10.1 5.8 9.0Cr2O3 0.51 0.06 0.40 1.77 0.92 1.57 1.53FeO 13.6 1.3 10.7 16.9 1.8 13.4 9.8MgO 20.7 0.19 15.9 14.7 0.9 11.5 10.1MnO 0.47 0.07 0.38 0.59 0.07 0.47 0.37CaO 9.0 1.5 7.3 11.9 2.6 9.7 9.4K2O 0.01 0.21d 0.05 0.01 0.15 0.04 0.06Na2O 0.13 1.13 0.36 0.19 1.44 0.48 0.56P2O5 0.45 0.57 0.48 0.40 0.80 0.49 0.70Total 100.6 96.6 100 101.0 98.7 100 100.4Wo 18.6 26.0Fe/Mge 0.37 3.8 0.38 0.64 1.1 0.65 0.55Si/Ale 15.1 13.5 16.3 11.0

a Average of mixed glass-silica analyses in typical inclusion.b 77% average pyroxene � 23% “bulk” glass (by wt).c bb � broad-beam analysis.d Average K2O for all “glass” analyses.e Molar ratios.

3760 C. A. Goodrich

with central Fo values � 70 may be accumulated phenocrystsor added xenocrysts, whereas those with central Fo � 70formed by continuous nucleation and growth in a magma nowrepresented by most of the material in the rock. This indicatesthat there was a change in crystallization conditions whenolivine compositional evolution reached Fo � 70. The obser-vation that melt inclusions occur only in olivine of Fo � 70(indicating a higher growth rate than for olivine of Fo � 70) isfurther evidence of this change of conditions.

The shape of the CSD function for olivine crystals in EET-A(Fig. 2d) is similar to that for SaU 005. The linear form andnegative slope of the curve in the size range 1 to 2 mm indicatesa period of continuous nucleation and growth. The dropoff ofthe curve for crystals � 1 mm in size indicates a sharp drop innucleation rate at high degrees of crystalliztion. The horizontalform of the curve for crystals � 2 mm in size suggests, as inSaU 005, addition of phenocrysts or xenocrysts. Although therelationship between composition and size has not been quan-

Fig. 19. Compositions of inclusions in olivine and chromite in SaU 005 and EET-A in the phase system Olivine-Plagioclase-Quartz (projected from Wollastonite). For each set of inclusions, phase boundaries shown are those appropriatefor the reconstructed present bulk composition (pbc). Diagrams are in oxygen units (see Longhi [1991], for computation ofend-members).


tified as rigorously for EET-A as for SaU 005, the fact that themost magnesian compositions (Fo � 76) were observed only as

rare cores of the largest crystals suggests that, analogous to theargument made above for SaU 005, olivine of Fo � 76 accu-

Table 8. Primary trapped liquid (PTL) compositions.

SaU 005 EETA

Inclusions in olivine Inclusions in chromite Inclusions in olivinea Inclusions in chromiteb

SiO2 51.7 50.4 51.6 51.7TiO2 1.0 0.58 0.21 0.32Al2O3 11.1 �6.1 5.0 �5.4Cr2O3 0.07 �1.3 0.31 �1.4FeO 16.5 17.9 21.7 16.6MgO 5.8 10.6 14.7 14.4MnO 0.14 0.33 0.29 0.43CaO 11.7 11.5 5.6 8.9K2O 0.04 0.05 0.04 0.04Na2O 1.4 0.7 0.28 0.44P2O5 0.53 0.54 0.37 0.45Total 100 100 100 100Fe/Mgc 1.6 0.95 0.83 0.65Si/Alc 7.8 �14 17.7 �16.2Ca/Alc 0.95 �1.7 1.0 �1.5

a Average of PTLs determined from pbc1 and pbc2 for inclusions in olivine.b Based on pbc1 for inclusions in chromite.c Molar ratios.

Fig. 20. Primary trapped liquids (PTL) represented by inclusions in olivine and chromite in SaU 005 and EET-A in thephase system Olivine-Plagioclase-Quartz (projected from Wollastonite). For inclusions in chromite in SaU 005, symbolsrepresent compositions calculated from pbc1 and pbc2 and the ellipse shows estimated uncertainties (the Al2O3 contents ofpbc1 and pbc are only upper limits, and it is not known whether the PTL was saturated with olivine � low-Ca pyroxeneor with olivine only). For inclusions in chromite in EET-A the symbol represents the composition calculated from pbc1 andthe ellipse shows estimated uncertaintites (the Al2O3 content of pbc2 is higher than that of pbc1, but either could be onlyan upper limit; and it is not known whether the PTL was saturated with olivine � low-Ca pyroxene or with olivine only).Ex is the composition of the parent magma of the megacryst assemblage in EET-A as estimated by Longhi and Pan (1989).The recalculated Ex (Longhi, private communication) is based on new data from this work. The composition of Treimanet al. (1994b) for the PTL in olivine in EET-A represents a later stage of evolution than that determined here. See text fordiscussion. SU � bulk composition of SaU 005 (Dreibus et al., 2000). EA � bulk composition of EET-A (McSween andJarosewich, 1983). AH � estimated composition of parent magma for lherzolitic shergottite ALHA77005 (McSween et al.,1988). Diagram is in oxygen units (see Longhi [1991], for computation of end-members).

3762 C. A. Goodrich

mulated in the form of phenocrysts or was added as xenocrysts,and olivine of Fo � 76 formed by continuous nucleation andgrowth in a magma now represented by most of the material inthe rock (the observation that the smallest crystals tend to havethe most FeO-rich compositions further supports continuouscrystallization in the bulk of the rock). In addition, as in SaU005, the fact that melt inclusions were not observed in olivineof Fo � 76 indicates a change in crystallization conditions atthis point in olivine evolution.

6.2. Pyroxenes

In SaU 005, the most magnesian pyroxenes are orthopyrox-ene (Wo 4–6, mg 77–75) and augite (Wo 35, mg 78) that occuras inclusions in the most magnesian (Fo 74) olivine (Fig. 3a).This assemblage appears to be in equilibrium (Fig. 22a). TheKDFe/Mg implied are KDol/opx � � 1.2 and KDol/aug

� 1.3 to 1.4, which are similar to those determined experi-mentally by Longhi and Pan (1989) for SNC-like magmacompositions (KDol/lpyx � 1.1, KDol/aug � 1.1–1.5) in the range1215 to 1240°C. In EET-A, the most magnesian pyroxene isorthopyroxene (Wo 2–5, mg 80–84) that occurs as opticallydistinguishable (twin-free) cores of megacrysts, and appears tobe in equilibrium (Fig. 22b) with the most magnesian (Fo81–77) olivine (KDol/opx � 1.2).

Magnesian low-Ca pyroxenes in both rocks show distinctspikes (nearly vertical trends) in Al2O3, Cr2O3 and Wo (Fig. 4).In SaU 005, they occur in the most magnesian groundmasspigeonites, whereas in EET-A they occur in the orthopyroxenecores of megacrysts (including X-14). Such trends are unusual,and several possibilities are considered to explain them. Onepossibility is that the most magnesian low-Ca pyroxenes rep-

resent populations of exotic crystals (e.g., xenocrysts, or phe-nocrysts in an introduced magma) with inherited near-verticaltrends in Al2O3 and Cr2O3. However, this seems unlikelybecause vertical trends would not be expected from any crys-tallization or melting process, and the highest Al2O3 contentsobserved would imply crystallization from magmas with unre-alistically high Al2O3 contents. Furthermore, the observationsthat in SaU 005 the orthopyroxene inclusions in olivine do notshow these trends (but have low Al2O3 and Cr2O3 contents),and that in EET-A the orthopyroxene cores have corrodedshapes and the highest Al2O3 and Cr2O3 contents occur neartheir edges, suggest reaction of early formed low-Ca pyroxeneswith liquid (with the orthopyroxene inclusions in olivine inSaU 005 having been shielded from this reaction). For exam-ple, the compositional spikes may result from the reactionorthopyroxene � liquidf pigeonite, which is common in earlyevolution of low-Ca magmas (Longhi, 1991). However, Al, Crand Ca crystal/liquid distribution coefficients for magnesianorthopyroxene and pigeonite are similar (e.g., Longhi and Pan,1989), and so this reaction should not produce large changes incontents of these elements. Alternatively, they might resultfrom resorption of chromite (this reaction will be discussedbelow) and the consequent increase of Cr2O3 in the liquid.However, this reaction would not result in significant increasesin Al2O3 or CaO. Thus, there appears to be no explanation forthese trends by closed-system evolution. The most likely pos-sibility seems to be that these pyroxenes are exotic crystals thatoriginally had low Al2O3, Cr2O3 and Wo contents (similar tothe lowest observed), and acquired their higher ones by reaction(disequilibrium) with the magma they were introduced into.

In both rocks, the groundmass pigeonite (except the mostmagnesian pigeonite in SaU 005) follows compositional trends(Fig. 4) that are typical for evolution of basaltic magmas:Al2O3, Wo and TiO2 increasing and Cr2O3 decreasing as mgdecreases, with reversals in the Al2O3 and Wo trends at mg �71 in SaU 005 and mg � 60 in EET-A that record the onset ofplagioclase crystallization. These trends indicate that crystalli-zation proceeded without interruption beginning with pigeoniteof mg � 73 in SaU 005 and mg � 78 in EET-A, consistent withthe evidence that olivine crystallization was continuous startingfrom Fo � 69 (equilibrium pigeonite mg � 73) and Fo � 76 inEET-A (equilibrium pigeonite mg � 79). The linear CSDfunction for groundmass pyroxenes in EET-A (Lentz and Mc-Sween, 2000) further supports this conclusion.

In SaU 005, augite shows a compositional hiatus betweenthat (mg 78) which occurs in the most magnesian (Fo 74)olivine, and that which occurs in the groundmass (mg 73–69).The most magnesian composition in the groundmass impliesequilibrium with olivine of Fo � 67 (Fig. 22a). This hiatus isfurther evidence for a change in conditions after crystallizationof the most magnesian phases. In EET-A, all augite, whichoccurs in the groundmass, is relatively ferroan (mg 65–50). Themost magnesian composition implies equilibrium with olivineof Fo � 59 (Fig. 22b).

6.3. Spinels

6.3.1. Crystallization Sequence

The occurrences of low-Ti chromite and chromian ul-vospinel in SaU 005 and EET-A (Fig. 5) indicate that: (1)

Fig. 21. Primary trapped liquids (PTL) represented by inclusions inolivine and chromite in SaU 005 and EET-A in the phase systemOrthopyroxene-Plagioclase-Wollastonite (projected from Olivine). Forinclusions in chromite in SaU 005, symbol represents compositioncalculated from the pbc1/pbc2 average and ellipse represents estimatesof uncertainty (the Al2O3 content of pbc1/pbc2 is only an upper limit;pbc3 provides the lower limit on CaO content). For inclusions inchromite in EET-A, the symbol represents composition calculated frompbc1 and the ellipse represents estimates of uncertainty (the Al2O3

content of pbc2 is higher than that of pbc1 but both are upper limits; thelower limit on CaO content is obtained by calculating pbc1 with alower pyroxene/glass ratio). SU � bulk composition of SaU 005(Dreibus et al., 2000). EA � bulk composition of EET-A (McSweenand Jarosewich, 1983). AH � estimated composition of parent magmafor lherzolitic shergottite ALHA77005 (McSween et al., 1988). Dia-gram is in oxygen units (see Longhi [1991] for computation of end-members).


low-Ti chromite crystallized as the sole spinel before and/orwith the more magnesian olivine (Fo 74–70 in SaU 005 and Fo81–60 in EET-A) in both rocks; (2) low-Ti chromite ceased tocrystallize at some point before olivine compositional evolutionreached Fo 69 in SaU 005 and Fo 59 in EET-A; and (3)chromian ulvospinel began to crystallize when olivine compo-sitional evolution reached Fo 69 in SaU 005 and Fo 59 inEET-A.

Furthermore, the observation that, in each rock, chromitecores of composite spinels (type 2) have ulvospinel contents,magnetite contents, and Cr-Al zonation patterns (Figs. 6–8)that are nearly identical to those of chromite inclusions inolivine (type 1), indicates that they formed at the same time asthe type 1 chromites and were only later overgrown withchromian ulvospinel crystallizing from a more evolved magma.Compositional differences between type 2 and type 1 chromitesin SaU 005 can be attributed to interaction of type 2 chromiteswith the more evolved magma from which the chromian ul-vospinel crystallized (while type 1 chromites were shieldedfrom the magma by their olivine hosts). For example, nearlyuniform fe#s within most type 2 chromites indicate a higherdegree of reequilibration with their surroundings than in type 1chromites. Furthermore, zonation patterns 2 and 3, both ofwhich show deviations from the zonation trend of type 1chromites via increasing Ti and Fe3�, can be attributed toreaction with the liquid from which chromian ulvospinel crys-tallized (reaction rims around melt inclusions, which are dis-cussed below, as well as zonation around large cracks, indicatea similar reaction). The observation that all three zonationpatterns sometimes occur in a single grain indicates that reac-tion of chromites with liquid was non-uniform. This could beexplained if some sides of the grains were temporarily shielded

by adjacent crystals (probably of pyroxene or olivine) beforeovergrowth of rims began (pattern 1) or were shielded laterafter reaction with liquid (pattern 3). Such complexities areconsistent with the presence of the chromite grains and othercrystals in a liquid in which the liquid/crystal ratio was stillhigh. The reverse Cr#-fe# zonation (decreasing fe# with de-creasing Cr#) seen in pattern 3 can be explained as a result ofFe/Mg reequilibration at � 900°C (Fabries, 1979), indicatingan even higher degree of reequilibration than that seen in type1 chromites and in type 2 chromites showing zonation pattern2.

The absence of zonation patterns 2 and 3 in type 2 chromitesin EET-A indicates that they did not react significantly with theevolved liquid from which chromian ulvospinel crystallized.This suggests more rapid cooling in EET-A than in SaU 005.Furthermore, the slightly lower Cr#s and greater degree of fe#zonation in cores of type 2 compared to type 1 chromites inEET-A suggest that the former represent a slightly later popu-lation. This implies a high degree of fractionation, furtherindicating more rapid cooling for EET-A than for SaU 005. Thelarger gap in TiO2 and magnetite contents between cores andrims of type 2 grains in EET-A (compared to SaU 005) indi-cates either a higher degree of fractionation and/or a larger gapin the spinel stability field.

6.3.2. Origin of the Spinel Compositional Gap

Spinel grains exhibiting a compositional gap similar to thatseen in type 2 spinels in SaU 005 and EET-A (Figs. 6 and 10)occur in both terrestrial (Haggerty, 1976, and referencestherein) and lunar (El Goresy et al., 1971, 1976; Haggerty andMeyer, 1971; Taylor et al., 1971) basalts. In addition, in many

Fig. 22. Crystallization sequences for the minerals in SaU 005 (a) and EET-A (b), inferred from the observationspresented in this work. The CSD functions for olivine, combined with compositional information, indicate that the mostmagnesian olivine and pyroxenes in each rocks represent accumulated phenocrysts or added xenocrysts. Other argumentssuggest that they are xenocrysts.

3764 C. A. Goodrich

magma bodies early chromite (crystallizing alone or with oli-vine) disappears from the crystallization sequence before theappearance of titanomagnetite, leaving a gap in which no spinelphase appears to be stable (Irvine, 1967; Evans and Moore,1968). Based on the observation that in layered sequences ofperidotite–orthopyroxenite–clinopyroxenite of the Muskox in-trusion chromite is present in the peridotite but absent in thepyroxenites, Irvine (1967) suggested that chromite and or-thopyroxene may be involved in a peritectic reaction similar tothat responsible for the change from olivine to orthopyroxenecrystallization. Irvine’s suggestion was confirmed by the ex-perimental data of Hill and Roeder (1974) for spinels crystal-lized over a range of temperatures and oxygen fugacities. Thisstudy demonstrated that at basaltic liquidus temperatures andlog10 f O2 � �9 (�QFM), the crystallization of early, chro-mite-rich spinel is terminated by the appearance of pyroxene,with which it is in reaction relationship; after an interval of�25 to 50°C, spinel reappears and crystallizes as titanomag-netite. In contrast, at log10 f O2 � �9, chromite crystallizationis continuous and there is complete solid solution betweenchromite and titanomagnetite.

These experimental studies indicate that the low-Ti chro-mites in SaU 005 and EET-A must have crystallized beforepyroxene. Since in both rocks the most magnesian low-Capyroxenes appear to be in equilibrium with the most magnesianolivine, this implies that the chromite also crystallized beforemost of the olivine (and is therefore the earliest phase) and ispresent as inclusions in olivine with Fo contents as low as70/69 in SaU 005 and 60/59 in EET-A only because of dis-equilibrium persistence. They also show that the spinel com-positional gap evidenced by low-Ti chromite cores havingcompositionally sharp contacts with chromian ulvospinel rimscan be a result of closed-system magmatic evolution. It doesnot, therefore, require complex processes such as magma mix-ing or a xenocrystic origin for chromite cores, though it is notinconsistent with such processes. Furthermore, the presence ofchromian ulvospinel as inclusions in late olivine (Fo 69–63 inSaU 005; Fo 59–53 in EET-A) indicates that in the crystalli-zation of these rocks olivine did not have a peritectic relation-ship with low-Ca pyroxene, since it was still crystallizing (notjust persisting) at the relatively evolved stage at which chro-mian ulvospinel began to crystallize. Finally, the results of Hilland Roeder (1974) suggest that ALHA77005 formed at higherf O2 than SaU 005 and EET-A, such that chromite-titanomag-netite solid solution was continuous (though the exact f O2 limitobserved by these authors may not be strictly applicable be-cause of compositional differences between the shergottitemagmas and the terrestrial basalts used in the experimentalstudy).

6.3.3. Origin of Cracks in Chromite Cores of Type 2 Spinels

The remarkable textural distinction between low-Ti chromitecores (riddled with cracks) and chromian ulvospinel rims(largely crack-free) of type 2 spinels in SaU 005 and EET-A(Fig. 9) strongly suggests that the chromites experienced somephysical stress before they were overgrown with ulvospinel.Possible explanations include: (1) expansion of glasses in meltinclusions upon solidification; (2) stress due to their chemicaldisequilibrium with an evolving magma (as discussed above);

or (3) stress associated with disruption and/or assimilation ofxenocrystic chromites by a magma (thermal shock?). The firstseems unlikely because the SiO2-rich compositions of theglasses imply low solidification temperatures, while ulvospinelmay have begun to crystallize at temperatures as high as1160°C (Hill and Roeder, 1974). The other two possibilities aresupported by the absence of these cracks in type 1 chromites,which suggests that they were protected by their olivine hostsfrom interactions with the surrounding magma. However, thesecond seems unlikely, because the chromite-pyroxene reactionrelationship has been observed in a variety of rocks, and crackssuch as these have not been reported in the chromite cores ofcomposite chromite-ulvospinel/titanomagnetite grains in otheroccurrences (e.g., El Goresy et al., 1971, 1976; Taylor et al.,1971). The third possibility seems the most likely, although theexact mechanism of crack formation is not known. Further-more, the absence of similar cracks in cores of cumulate spinelgrains in ALHA77005 suggests that they do not form underconditions of continous evolution of a magma.

6.3.4. Pyroxene-Spinel Exsolutions in Olivine

Inclusions similar to the tiny pyroxene-spinel intergrowthsthat occur in olivine in SaU 005 and EET-A have been ob-served in olivine from a variety of rocks including terrestrialdunites (Irving et al., 1992), Nakhla (Yamada et al., 1997;Mikouchi et al., 2000), lunar troctolite 76535 (Gooley et al.,1974; Bell et al., 1975), ungrouped achondrite QUE 93148(Goodrich and Righter, 2000) and polymict ureilite DaG 165(Goodrich and Keil, 2002). They are thought to have formed bysimultaneous two-phase exsolution resulting from oxidation ofFe2� and Cr2� in olivine (Mosley, 1984; Goodrich andRighter, 2000). The presence of these inclusions in olivines inSaU 005 and EET-A indicates late oxidation in both of theserocks.

6.4. Crystallization Histories of SaU 005 and EET-A

6.4.1. SaU 005

Low-Ti chromite is the earliest phase in SaU 005 (Fig. 22a).It probably co-crystallized with olivine of Fo � 74 which is notpresent (due to continous reequilibration or crystal settling).During growth of chromite crystals, samples of the magmafrom which they were crystallizing were trapped as melt inclu-sions. By the time olivine compositional evolution had reachedFo � 74, orthopyroxene (mg � 77–75) and augite (mg � 78)had begun to crystallize. Crystallization of chromite ceasedwhen pyroxenes appeared. Chromite grains were no longer inequilibrium with the magma, but many persisted and weretrapped as inclusions in later growing phases. As olivineevolved to Fo � 70, orthopyroxene was replaced by pigeonite(mg 75–74) in the crystallization sequence. All olivine, chro-mite and pyroxene formed up to this point represent eitheraccumulated phenocrysts or added xenocrysts in the presentrock. The vertical compositional trends shown by the low-Capyroxenes (except those included in olivine), and the pervasivecracks in chromites (except those included in olivine) in thisassemblage suggest that they are xenocrysts.

Most of the material in SaU 005 then crystallized from amagma, with olivine (Fo 69–62) forming by continuous nu-


cleation and growth and some continued nucleation on theearlier (accumulated or added) crystals. Pigeonite (mg 74–68)may also have nucleated to some extent on the earlier low-Capyroxenes. The growth rate of olivine was now higher than inthe previous stage of evolution, leading to trapping of melt asinclusions. Ulvospinel began to crystallize when olivine com-positional evolution reached Fo � 69, and formed rims onpersisting low-Ti chromite grains as well as individual grains.It also grew as daughter crystals in trapped melts. Augiteappeared in the crystallization sequence shortly after ul-vospinel. Plagioclase began to crystallize when pigeonite com-position reached mg � 71 (and olivine reached Fo � 66).

6.4.2. EET-A

Low Ti-chromite is the earliest phase in EET-A (Fig. 22b). Itprobably co-crystallized with olivine of Fo � 81 which is notpresent (due to continous reequilibration or crystal settling).During growth of chromite crystals, samples of the magmafrom which they were crystallizing were trapped as melt inclu-sions. By the time olivine evolution reached Fo � 81, orthopy-roxene (mg 84–81) had begun to crystallize. Crystallization ofchromite ceased when pyroxene appeared. Chromite grainswere no longer in equilibrium with the magma, but neverthelesspersisted and were trapped as inclusions in later growingphases (and in melt inclusions). All olivine, chromite, andorthopyroxene formed up to this point represent either accu-mulated phenocrysts or added xenocrysts in the present rock.The vertical compositional trends shown by the orthopyroxeneand the pervasive cracks in chromites (except those included inolivine) in this assemblage strongly suggest that they are xe-nocrysts.

Most of the material in EET-A then crystallized from amagma, with olivine crystals (Fo 76–53) forming by continu-ous nucleation and growth (though at high degrees of crystal-lization, nucleation rate decreased dramatically) and some con-tinued nucleation on the earlier (accumulated or added) olivinecrystals. The growth rate of olivine was higher than in theprevious stage of evolution, leading to trapping of melt asinclusions. Pigeonite (beginning at mg � 78) nucleated onearlier orthopyroxene to form rare low-Ca pyroxenemegacrysts, but mostly formed by continous nucleation andgrowth to form the finer-grained groundmass. Both ulvospineland augite (mg 65) began to crystallize when olivine compo-sitional evolution reached Fo � 60. Plagioclase began to crys-tallize when pigeonite reached mg � 60 (and olivine reachedFo � 55). The absence of trapped melts in olivine of Fo � 60may indicate a return to slower growth and/or be associatedwith the decline in nucleation rate.

6.5. Trapped Melt Compositions and Their PredictedCrystallization Sequences

6.5.1. SaU 005

From the previous analysis, it is clear that the inclusions inlow-Ti chromite in SaU 005 represent the melt that was presentwhen the earliest phases were crystallizing. This melt has ahigh Si/Al ratio and low Al2O3 and Na2O contents (Table 8;Fig. 20), which indicate its primitive nature, and a notably highCaO content (�11.5 wt.%). As discussed above, its olivine

component must have been higher than the minimum calcu-lated (the composition given in Table 8 is cosaturated witholivine and low-Ca pyroxene) because chromite must havecrystallized before pyroxenes. The predicted equilibrium crys-tallization sequence for this melt, with an additional 10%olivine component (an arbitrary amount, simply added to en-sure that some olivine is already present when pyroxene ap-pears), is shown in Figure 23a. After 10% crystallization ofolivine, it reaches cosaturation with olivine and low-Ca pyrox-ene (the addition of olivine has no effect on the the crystalli-zation sequence after pyroxene appears). Augite appears early,after �11% crystallization of low-Ca pyroxene. Plagioclaseappears after �47% crystallization of pyroxenes. The finalphase assemblage (including the estimated 10% olivine) con-tains �43% augite, 17% pigeonite and 23% plagioclase. Thisassemblage is very different from the bulk modal mineralogy ofSaU 005 (shown in Fig. 23a for comparison), which has � 7%augite and �50% pigeonite. However, its early crystallizationsequence (Fig. 23a, bottom), in the interval �1275 to 1230°C,produces an assemblage very similar to the magnesian assem-blage in SaU 005 that has been shown to represent eitheraccumulated phenocrysts or added xenocrysts–olivine of Fo 74to 70, low-Ca pyroxene (Wo 4–6) of mg 77–75, and augite(Wo 35) of mg 78. These comparisons lead to the conclusionthat the melt trapped in chromite does, indeed, represent theparent magma of the earliest phases in SaU 005, but it cannotbe the parent magma of the whole rock. Therefore, the earlyphase assemblage must be xenolithic.

In contrast to the melt trapped in chromite, the melt trappedin olivine in SaU 005 (Table 8; Fig. 20) has high Al2O3 and lowSi/Al, which indicates that it is a relatively evolved liquid andis consistent with its occurrence in the more ferroan olivine(avg. Fo 64). Its predicted crystallization sequence furtherdemonstrates this. Its liquidus temperature is low (�1140°C),and although it is cosaturated with olivine (Fo 64) and pigeo-nite (mg 9.6, mg 69), augite appears after only a few percentcrystallization and essentially replaces pigeonite (consistentwith the observation that pigeonites more ferroan than mg � 68were not observed in SaU 005), and plagioclase appears after�9% crystallization. The final phase assemblage contains�52% plagioclase and 37% augite, which shows clearly thatthis melt is not representative of the magma from which thebulk of SaU 005 formed, but rather represents a late stage ofcrystallization.

6.5.2. EET-A

As for SaU 005, it is clear that the inclusions in low-Tichromite in EET-A represent the melt that was present whenthe earliest phases were crystallizing. This melt is similar to theearly melt in SaU 005 in having a high Si/Al ratio and lowAl2O3 content, (Table 8; Fig. 20), but it has a significantlylower CaO content (8.9%). The predicted equilibrium crystal-lization sequence for this melt (again, with �10% additionalolivine) is shown in Figure 23b. After 10% crystallization ofolivine, it reaches cosaturation with olivine and low-Ca pyrox-ene. Augite appears after �34% crystallization of low-Ca py-roxene, and plagioclase appears after �56% crystallization ofpyroxenes. The final phase assemblage (including the estimated10% additional olivine) contains �40% pigeonite, 24% augite,

3766 C. A. Goodrich

and 17% plagioclase. This assemblage differs significantlyfrom the modal mineralogy of EET-A (shown in Fig. 23b forcomparison), which has �60% pigeonite and 6% augite. How-ever, its early crystallization sequence (Fig. 23b, bottom), in theinterval �1350 to 1300°C, produces an assemblage very sim-

ilar to the early assemblage in EET-A that has been shown torepresent either accumulated phenocrysts or added xenoliths–olivine of Fo 81 to 77 and low-Ca pyroxene (Wo 2–3) of mg 84to 80. This leads to the conclusion that the melt trapped inchromite does, indeed, represent the parent magma of the

Fig. 23. Low-pressure, equilibrium crystallization sequences predicted by MAGPOX (Longhi, 1991) for trapped liquidcompositions reconstructed from melt inclusions. (a) Trapped liquid in inclusions in chromite in SaU 005, with an estimated10% additional olivine. Modal abundances in SaU 005 shown for comparison. (b) Trapped liquid in inclusions in chromitein EET-A, with an estimated 10% additional olivine. (c) Trapped liquid in inclusions in olivine (Fo 76–67) in EET-A.Modal abundances in EET-A shown for comparison in (b) and (c).


magnesian megacryst assemblage in EET-A, but it cannot bethe parent magma of the whole rock. Therefore, as in SaU 005,the early phase assemblage must be xenocrystic or xenolithic.

The composition of the parent magma of the magnesianmegacryst assemblage in EET-A was previously estimated byLonghi and Pan (1989), based on the mineral compositionaldata of Steele and Smith (1982) and experimentally determinedpartition coefficients. This composition (Ex) has significantlyhigher Al2O3 and lower Si/Al than the composition determinedhere from melt inclusions in chromite (Fig. 20). However, theAl2O3 content of Ex was determined principally by the lowestAl2O3 content (0.8%) that Steele and Smith (1982) reported forEET-A orthopyroxene. The observations presented here (Fig.4b) show that Al2O3 contents of the orthopyroxene actuallyextend to much lower values (0.4%). Recalculation of Ex (JohnLonghi, private communication) using this new lowest Al2O3

value results in a parent magma composition that has muchlower Al2O3 and higher Si/Al, and, in terms of all majorelements is very similar to that determined here (Fig. 20).

The PTL composition reconstructed here for melt trapped inolivine of Fo 76 to 67 in EET-A (Table 8) is similar to the melttrapped in chromite in having high Si/Al and low Al2O3 (Fig.20), but has a lower CaO content. The predicted equilibriumcrystallization sequence for this melt is shown in Figure 23c. Atits liquidus, it is in equilibrium with olivine of Fo 76 andlow-Ca pyroxene (Wo 4) of mg � 80. Plagioclase appears after�67% crystallization (when olivine evolution has reached Fo57) and augite after �78% crystallization (when olivine evo-lution has reached Fo 55). The final phase assemblage contains�58% low-Ca pyroxene, 15% plagioclase, and 5% augite.Completely fractional crystallization (more germane here thanin SaU 005 because of the stronger zonation of minerals inEET-A) produces more augite (�10%) and less low-Ca pyrox-ene (�38%). In both cases, the final assemblage is generallysimilar to the bulk mineralogy of EET-A (shown in Fig. 23c forcomparison), and the points at which plagioclase and augiteappear (relative to olivine evolution) are similar to those in-ferred for the rock (Fig. 22b). These comparisons suggest thatthe melt composition determined from inclusions in olivine of�Fo 77 to 67 in EET-A may be a reasonable approximation tothe composition of the melt from which most of the material inEET-A crystallized.

6.6. Models for the Petrogenesis of SaU 005 and EET-A

6.6.1. Closed-System crystallization of a magma of bulk rockcomposition

The possibility that either SaU 005 or EET-A crystallizedfrom a magma of its bulk rock composition can be ruled out. Amelt having the composition of bulk EET-A (mg 61) is inequilibrium at its liquidus with olivine of Fo 83, which issimilar to the most magnesian olivine (Fo 81) in the rock, butdoes not become cosaturated with orthopyroxene until �26%olivine has crystallized (Fig. 20). A melt having the composi-tion of bulk SaU 005 (mg � 68) is in equilibrium at its liquiduswith olivine of Fo 86, which is considerably more magnesianthan the most magnesian olivine (Fo 74) in SaU 005, and doesnot become cosaturated with low-Ca pyroxene until �34%olivine has crystallized (Fig. 20). Thus, for either SaU 005 or

EET-A, the magnesian olivine-orthopyroxene assemblagecould not have been in equilibrium with a melt having thecomposition of the bulk rock.

6.6.2. Closed-System crystallization with accumulation ofolivine

The melts trapped in low-Ti chromite in SaU 005 and EET-Aappear to have been in equilibrium with the earliest, mostmagnesian phases in these rocks. However, neither of thesemagmas could be the parent magma of its respective wholerock, even with accumulation of olivine phenocrysts, becauseboth produce significantly higher augite/pigeonite ratios thanthose observed (Figs. 23a and 23b).

6.6.3. Incorporation of xenocrysts

The model suggested by the results of this study is that thechromite and most magnesian olivine and pyroxenes in bothSaU 005 and EET-A are xenocrystic, but the fraction of solidxenocrystic (or xenolithic) material is small and most of thematerial in each rock formed by continuous crystallization of asingle magma. This model differs from previous models forEET-A (e.g., Steele and Smith, 1982; McSween andJarosewich, 1983; Wadhwa et al., 1994), in which all of theolivine was believed to be xenocrystic. One of the foundationsfor this belief was that olivine grains have irregular shapes andzonation contours indicating disequilibrium with the ground-mass. However, I believe that this observation has been over-emphasized. Many grains, in fact, are subhedral and showconcentric zonation (Fig. 1b). Furthermore, irregular shapesand zonation contours could be explained by the rapid growthand high degree of fractionation evidenced in olivine compo-sitions (while the more regular shapes and zonation contours ofolivine in SaU 005 are consistent with its lower degree offractionation), and irregular grain shapes may also be partly dueto effects of late shock. Furthermore, the CSD function forolivine in EET-A (Fig. 2d) does not indicate any disruption ingrowth history (such as would result from resorption) in thebulk of the rock.

The model suggested here does, however, for either EET-Aor SaU 005, present the apparent problem that subtraction ofthe material inferred to be xenolithic from the bulk compositionhas a small effect, and does not leave a composition that cancrystallize most of the rock (similar to the bulk compositionitself, it is not cosaturated with olivine and low-Ca pyroxenebut has an excess of olivine). This leads to the conclusion thatSaU 005 and EET-A must have lost fractionated melts late intheir crystallization. Several observations support this conclu-sion. First, their olivine CSD functions (Figs. 2c and 2d) showdecreased nucleation rates at high degrees of crystallization,which may indicate loss of melt. Second, they have highpyroxene/plagioclase ratios and low augite contents (andEET-A has an excess of pigeonite and deficit of augite relativeto that expected from fractional crystallization of its PTL inolivine). And third, in the case of EET-A, there is directevidence for segregation of late melt in the contact with EET-B.McSween and Jarosewich (1983) discussed the possibility thatEET-B represents a fractionated liquid derived from the samemagma as EET-A, but dismissed this idea because they did not

3768 C. A. Goodrich

believe that the olivine in EET-A crystallized in-situ. However,if the model presented here is correct, this is no longer anobjection. Indeed, the gradational nature of the contact betweenlithology A and lithology B (Fig. 1b) is more suggestive ofmagmatic layering than of a clast/host relationship. Further-more, melts present at high degrees of crystallization (�75%,slightly less for fractional crystallization) of the PTL in olivinein EET-A are similar to EET-B in having high Fe/Mg and CaOcontent, and crystallize a pigeonite-augite-plagioclase assem-blage similar to the mineralogy of EET-B.

6.6.4. Magma mixing

Although in the model suggested here the fraction of solidxenocrystic material in either SaU 005 or EET-A is small, it ispossible that the xenocrysts were phenocrysts in an admixedmagma. The obvious candidate for that magma is the melt thatwas in equilibrium with the xenocrystic material. Assumingthat the PTL in olivine in EET-A represents the magma fromwhich most of the material in the rock crystallized, and that itis a mixture of the PTL in chromite plus an assimilatingmagma, we can place some contraints on the composition of thelatter. Mixing of the PTL in chromite in EET-A with a basalticshergottite-like magma such as QUE 94201 (Dreibus et al.,1996; Warren and Kallemeyn, 1996, 1997) could not producethe PTL in olivine, because both endmembers have higher CaOcontent than this melt. Furthermore, the mafic component of thePTL in olivine is nearly as high as that of the melt trapped inchromite, which means that it could not have any significantcomponent of a non olivine-saturated magma such as a basalticshergottite magma. This leads to the conclusion that if magmamixing did occur, then all melts involved were olivine-satu-rated. It also seems inescapable that if magma mixing occurredwhen the xenocrysts were added to SaU 005, the assimilatingmagma was olivine-saturated. Unfortunately, we lack the con-straint that we have for EET-A because the melt trapped inolivine is a late one and does not represent the melt from whichmost of the rock crystallized. However, the high CaO contentof the PTL in chromite in SaU 005 indicates that its contribu-tion to a mixed magma would have to be small (since the CaOcontent and augite abundance of SaU 005 are similar to thoseof EET-A), and therefore the assimilating magma must havebeen olivine-saturated. Furthermore, SaU 005 has an evenhigher abundance of olivine and higher bulk mg than EET-A,which indicates that non olivine-saturated basaltic shergottite-like magmas are not likely to have been involved in its petro-genesis.

6.7. Relationship of SaU 005 and EET-A to Basaltic andLherzolitic Shergottites

The model presented here, that only the most magnesianphases in SaU 005 and EET-A are xenocrystic, leads to theconclusion that all magmas involved in their petrogenesis wereolivine-saturated (whether magma mixing occurred or not).Therefore, SaU 005 and EET-A did not form by mixing ofbasaltic shergottite-like magmas and lherzolitic shergottite-likematerials. The magmas from which most of the material ineither of these rocks crystallized may, however, have beenparental to basaltic shergottites.

Were the parent magmas of the xenocrysts, or the magmasfrom which most of the material in each of these rocks crys-tallized, similar to the parent magmas of lherzolitic shergot-tites? Unfortunately, there are no completely reliable estimatesof the compositions of lherzolitic shergottite parent magmas.McSween et al. (1988) calculated a parent magma composition(an intercumulus liquid composition, assumed to represent co-existing equilibrium melt) for ALHA77005 by subtracting cu-mulus phases from the bulk rock composition. However,Longhi and Pan (1989) argued that this cannot be a true meltcomposition, because it is not cosaturated with olivine andlow-Ca pyroxene (Fig. 20). Attempts to estimate parent magmacompositions from melt inclusions in olivine have been madeby Ikeda (1994b) and Zipfel and Goodrich (2001) forALHA77005, and by Harvey et al. (1993) for LEW 88516.However, Goodrich and Harvey (2002) showed that these in-clusions probably record an evolving magma composition andthe average melt compositions determined are not parent mag-mas. Preliminary data for melt inclusions in chromite inALHA77005 (Goodrich and Harvey, 2002) show a more prim-itive composition (lower Al2O3 and higher Si/Al), and mayyield a true parent magma. At this point, however, we cannotdirectly compare the SaU 005 and EET-A magmas with lher-zolitic shergottite parent magmas. Nevertheless, the parentmagma of the xenoliths in SaU 005 (in contrast to that inEET-A) is an unlikely candidate for a lherzolitic shergottiteparent magma, because its high CaO content produces an earlyassemblage that includes augite. Furthermore, the spinel stabil-ity gap seen in SaU 005 and EET-A suggests that all magmasinvolved in their petrogenesis were more reduced than thatfrom which ALHA77005 formed.

Finally, it should be emphasized that although SaU 005 andEET-A appear to have had similar petrogenetic histories, theymay not be closely related rocks. SaU 005 is strongly LREE-depleted (Dreibus et al., 2000) and belongs to a group ofmartian meteorites (which includes olivine-phyric shergottitesDaG 476 and Dhofar 019, as well as basaltic shergottite QUE94201) identified by their high initial �Nd143 values and thepresence of 142Nd anomalies as originating from a reservoir orreservoirs formed early in martian differentiation history (Ja-goutz and Jotter, 2000; Jagoutz et al., 2000, 2001; Borg et al.,2001), while EET-A has a flatter REE pattern (Banin et al.,1992) and less pronounced 142Nd anomaly (Harper et al., 1995)and cannot be derived (at least not wholly) from the samereservoir. Thus, similar petrogenetic processes may have oper-ated at different times in martian differentiation history.

7. CONCLUSIONS

1. Petrographic and mineral compositional observations indi-cate that the chromite and most magnesian olivine andpyroxenes in SaU 005 and EET-A are xenocrystic. In bothcases, however, the fraction of solid xenocrystic (or xeno-lithic) material is small (only a few percent), and most of thematerial in each rock formed by continuous crystallizationof a single magma (which may or may not have been amixed magma).

2. Melts trapped in chromite in SaU 005 and EET-A representthe parent magmas of their respective xenocrysts. Neither ofthese melts could be the parent magma of its host rock.


3. Melt trapped in olivine in EET-A represents a reasonableapproximation to the magma from which most of the mate-rial in EET-A crystallized. Melt trapped in olivine in SaU005 represents a late stage of magmatic evolution.

4. If the model discussed here is correct and only a smallfraction of the material in either SaU 005 or EET-A isxenocrystic or xenolithic, then these rocks must have lostfractionated liquids late in their crystallization. For EET-A,this fractionated liquid may be represented by EET-B.

5. All magmas involved in the petrogenesis of these rocks wereolivine-saturated.

6. SaU 005 and EET-A did not form by mixing of basalticshergottite-like magmas and lherzolitic shergottite-like ma-terial. The magmas from which most of the material in theserocks crystallized may, however, have been parental tobasaltic shergottites.

7. Olivine-phyric shergottites SaU 005 and EET-A provideevidence for a greater diversity of martian magma types thanis apparent in the basaltic and lherzolitic shergottites.

Acknowledgments—I thank J. L. Berkley, A. M. Fioretti, A. Greshake,C. D. K. Herd, K. Keil, A. N. Krot, G. Libourel, J. Longhi, H. Y.McSween, D. K. Ross, E. R. D. Scott, S. Singletary, E. Stolper, G. J.Taylor, L. A. Taylor, A. H. Treiman and J. Zipfel for helpful discus-sions; B. Schulz-Dobrick, E. Macsenaere-Riester and D. K. Ross forassistance with electron microprobe analyses; and J. Zipfel, E. Jagoutzand the Antarctic Meteorite Working Group (particularly D. W. Mit-tlefehldt and C. Satterwhite) for providing samples. I especially thankJutta Zipfel for suggesting the study of melt inclusions in SaU 005 thatled to this paper. Careful reviews by L. Folco, A. J. Irving, an anon-ymous reviewer, and the associate editor H. Palme led to significantimprovements in this manuscript and are greatly appreciated. This workwas supported by NASA grant NAG-11591 (K. Keil) and the Max-Planck Society. This is Hawaii Institute of Geophysics and Planetologypublication #1282 and School of Ocean and Earth Science and Tech-nology publication #6161.

Associate editor: H. Palme

REFERENCES

Banin A., Clark B. C., and Wanke H. (1992) Surface chemistry andmineralogy. In Mars (eds. H. H. Kieffer, B. M. Jakosky, C. W.Snyder, and M. S. Matthews), pp. 594–625. University of ArizonaPress, Tucson.

Barrat J. A., Gillet P., Sautter V., Jambon A., Javoy M., Gopel C.,Lesourd M., Keller F., and Petit E. (2002a) Petrology and chemistryof the basaltic shergottite North West Africa 480. Meteoritics Planet.Sci. 37, 487–499.

Barrat J. A., Jambon A., Bohn M., Gillet P., Sautter V., Gopel C.,Lesourd M., and Keller F. (2002b) Petrology and chemistry of thepicritic shergottite North West Africa 1068 (NWA 1068). Meteorit-ics Planet. Sci. 37, 3505–3518.

Bell P. M., Mao H. K., Roedder E., and Weiblen P. W. (1975) Theproblem of the origin of symplectites in olivine-bearing lunar rocks.In Proc. 6th Lunar Sci. Conf., pp. 231–248.

Berkley J. L. and Treiman A. H. (2000) Petrologic history of a complexpyroxenite xenolith in the EET-A79001 Martian meteorite [abstract].Lunar Planet. Sci. 31, #1729.

Berkley J. L., Treiman J. H., Jones J. H., and Mittlefehldt D. W. (1999)Highly magnesian orthopyroxenite xenoliths in EET-A79001: Im-plications for Martian magmas and differentiation [abstract]. LunarPlanet. Sci. 30, #1588.

Borg L. E., Nyquist L. E., Reese Y., Wiesmann C.-Y., Ivanova M.,Nazarov M. A., and Taylor L. A. (2001) The age of Dhofar 019 andits relationship to the other martian meteorites [abstract]. LunarPlanet. Sci. 32, #1144.

Carmichael I. S. E. (1967) The iron-titanium oxides of salic volcanicrocks and their associated ferromagnesian silicates. Contrib. Min-eral. Petrol. 14, 36–64.

Danuyshevsky L. V., Della-Pasqua F. N., and Sokolev S. (2000)Re-equilibration of melt inclusions trapped by magnesian olivinephenocrysts from subduction-related magmas: Petrological implica-tions. Contrib. Mineral. Petrol. 138, 68–83.

Dreibus G., Jochum K. H., Palme H., Spettel B., Wlotzka F., andWanke H. (1992) LEW88516: A meteorite close to the “Martianmantle [abstract].” Meteoritics 27, 216–217.

Dreibus G., Spettel B., Wlotzka F., Schultz L., Weber H. W., JochumK. P., and Wanke H. (1996) QUE 94201: An unusual Martian basalt[abstract]. Meteoritics Planet. Sci. 31(Suppl.), A39–A40.

Dreibus G., Spettel B., Haubold R., Jochum K. P., Palme H., Wolf D.,and Zipfel J. (2000) Chemistry of a new shergottite: Sayh al Uhaymir005 [abstract]. Meteoritics Planet. Sci. 35(Suppl.), A49.

El Goresy A., Ramdohr P., and Taylor L. A. (1971) The opaqueminerals in the lunar rocks from oceanus procellarum. In Proc. 2ndLunar Sci. Conf., pp. 219–235.

El Goresy A., Prinz M., and Ramdohr P. (1976) Zoning in spinels as anindicator of the crystallization histories of mare basalts. In Proc. 7thLunar Sci. Conf., pp. 1261–1279.

Evans B. W. and Moore J. G. (1968) Mineralogy as a function of depthin the prehistoric Makaopuhi tholeiitic lava lake, Hawaii. Contrib.Mineral. Petrol. 17, 85–115.

Fabries J. (1979) Spinel-olivine geothermometry in peridotites fromultramafic complexes. Contrib. Mineral. Petrol. 69, 329–336.

Folco L., Franchi I. A., D’Orazio M., Rocchi S., and Schultz L. (2000)A new martian meteorite from the Sahara: The shergottite Dar alGani 489. Meteoritics Planet. Sci. 35, 827–839.

Gaetani G. and Watson B. (2000) Open system behavior of olivine-hosted melt inclusions. Earth Planet. Sci. Lett. 183, 27–41.

Gnos E., Hofmann B., Franchi I. A., Al-Kathiri A., Hauser M., andMoser L. (2002) Sayh al Uhaymir 094—A new martian meteoritefrom the Oman desert. Meteoritics Planet. Sci. 37, 835–854.

Goodrich C. A. (1984) Phosphoran pyroxene and olivine in silicateinclusions in natural iron-carbon alloy, Disko Island, Greenland.Geochim. Cosmochim. Acta 48, 1115–1126.

Goodrich C. A. (2002) Olivine-phyric martian basalts: A new type ofshergottite. Meteoritics Planet. Sci. 37, B31–B34.

Goodrich C. A. and Righter K. (2000) Petrology of unique achondriteQueen Alexandra Range 93148: A piece of the pallasite (howardite-eucrite-diogenite) parent body? Meteoritics Planet. Sci. 35, 521–535.

Goodrich C. A. and Harvey R. P. (2002) The parent magmas oflherzolitic shergottites ALH 77005 and LEW 88516: A reevaluationfrom magmatic inclusions in olivine and chromite [abstract]. Mete-oritics Planet. Sci. 37(Suppl.), A54.

Goodrich C. A. and Keil K. (2002) Feldspathic and other unusual clastsin polymict ureilite DaG 165 [abstract]. Lunar Planet. Sci. 33,#1777.

Goodrich C. A., van Niekerk D., and Morgan M. L. (2003) NorthwestAfrica 1110: A new olivine-phyric shergottite possibly paired withNorthwest Africa 1068 [abstract]. Lunar Planet. Sci. 34, #1266.

Gooley R., Brett R., Warner J., and Smyth J. R. (1974) A lunar rock ofdeep crustal origin: Sample 76535. Geochim. Cosmochim. Acta 38,1329–1339.

Grossman J. N. and Zipfel J. (2001) The Meteoritical Bulletin 85.Meteoritics Planet. Sci. 36, A293–A322.

Haggerty S. E (1976) Opaque mineral oxides in terrestrial rocks. InOxide Minerals. Min. Soc. Am. Short Course Notes. Vol. 3, Hg101–Hg300.

Haggerty S. E. and Meyer H. O. A. (1971) Apollo 12: Opaque oxides.Earth Planet. Sci. Lett. 9, 379–387.

Harper C. L., Jr., Nyquist L. E., Bansal B., Wiesmann H., and ShihC.-Y. (1995) Rapid accretion and early differentiation of Marsindicated by 142Nd/144Nd in SNC meteorites. Science 267, 213–217.

Harvey R. P., Wadhwa M., McSween H. Y., Jr., and Crozaz G. (1993)Petrography, mineral chemistry, and petrogenesis of Antarctic Sher-gottite LEW88516. Geochim. Cosmochim. Acta 57, 4769–4783.

Hill R. and Roeder P. (1974) The crystallization of spinel from basalticliquid as a function of oxygen fugacity. J. Geol. 82, 709–729.

Ikeda Y. (1994a) Petrography and petrology of the ALH-77005 sher-gottite. Proc. NIPR Symp. Antarc. Meteorites 7, 9–29.

3770 C. A. Goodrich

Ikeda Y. (1994b) Petrology of magmatic silicate inclusions in the AllanHills 77005 lherzolitic shergottite. Meteoritics Planet. Sci. 33, 803–812.

Ikeda Y. (1997) Petrology of the Y-793605 martian meteorite. Antarct.Meteorite Res. Nat. Inst. Polar Res., Tokyo 10, 13–40.

Ikeda Y. (2001) Magmatic and unusual inclusions in olivine grains inthe Dar al Gani 735 shergottite [abstract]. Meteoritics Planet. Sci.36(Suppl.), A86.

Ikeda Y., Takeda H., Kimura M., and Suzuki A. (2002) A newshergottite from Oman, Dhofar 378 [abstract]. Lunar Planet. Sci. 33,#1434.

Irvine T. N. (1967) Chromian spinel as a petrogenetic indicator. Part 2.Petrologic applications. Can. J. Earth Sci. 4, 71–103.

Irving A. J., Hirschmann M. M., and Kuehner S. M. (1992) Exsolutionof chromite and diopside from mantle olivine: Montana dunitexenoliths and the Twin Sisters dunite [abstract]. Eos: Trans. Am.Geophys. Union 73, 336.

Irving A. J., Kuehner S. M., Hupe A. C., and Hupe G. M. (2002)Olivine-phyric basaltic shergottite NWA 1195: A very primitivemartian lava [abstract]. Meteoritics Planet. Sci. 37(Suppl.), A69.

Jagoutz E. (1989) Sr and Nd isotopic systematics in ALHA 77005: Ageof shock metamorphism in shergottites and magmatic differentiationon Mars. Geochim. Cosmochim. Acta 53, 2429–2441.

Jagoutz E. and Jotter R. (2000) New Sm-Nd data on Nakhla minerals[abstract]. Lunar Planet. Sci. 31, #1609.

Jagoutz E., Jotter R., and Dreibus G. (2000) Evolution of six SNCmeteorites with anomalous neodymium-142 [abstract]. MeteoriticsPlanet. Sci. 35, A83–A84.

Jagoutz E., Jotter R., Dreibus G., and Zartmann R. (2001) New U-Pbisotope data on SNC meteorites [abstract]. Lunar Planet. Sci. 32,#1307.

Jambon A. J., Barrat J. A., Sautter V., Gillet P., Gopel C., Javoy M.,Joron J. L., and Lesourd M. (2002) The basaltic shergottite NorthWestern Africa 856: Petrology and chemistry. Meteoritics Planet.Sci. 37, 1147–1164.

Kamenetsky V. (1996) Methodology for the study of melt inclusions inCr-spinel, and implications for parental melts of MORB from FA-MOUS area. Earth Planet. Sci. Lett. 142, 479–486.

Kring D., Gleason J. D., Hill D. H., Jull A. J. T., and Boynton W. V.(1996) QUE94201, a new martian meteorite that may represent abulk melt rather than a cumulate fraction [abstract]. Lunar Planet.Sci. 27, 705–706.

Lentz R. C. F. and McSween H. Y., Jr. (2000) Crystallization of thebasaltic shergottites: Insights from crystal size distribution (CSD)analysis of pyroxenes. Meteoritics Planet. Sci. 35, 919–927.

Lin C.-Y., Zhang F.-S., Wang H.-N., and Zhang W.-L. (2002) AntarcticGRV9927: A new member of SNC meteorites [abstract]. LunarPlanet. Sci. 33, #1562.

Longhi J. (1991) Comparative liquidus equilibria of hypersthene-nor-mative basalts at low pressure. Am. Mineral. 76, 785–800.

Longhi J, and Pan V. (1989) The parent magmas of the SNC meteor-ites. In Proc. 19th Lunar Planet. Sci. Conf., pp. 451–464.

Marsh B. D. (1988) On the interpretation of crystal size distributions inmagmatic systems. J. Petrol. 39, 553–599.

McCoy T. J., Taylor G. J., and Keil K. (1992) Zagami: Product of atwo-stage magmatic history. Geochim. Cosmochim. Acta 56, 3571–3582.

McKay G., Koizumi E., Mikouchi T., Le L., and Schwandt C. (2002)Crystallization of shergottite QUE 94201: An experimental study[abstract]. Lunar Planet. Sci. 33, #2051.

McSween H. Y., Jr. (1985) SNC meteorites: Clues to Martian petro-logic evolution? Rev. Geophys. 23, 391–416.

McSween H. Y., Jr., and Jarosewich E. (1983) Petrogenesis of theElephant Moraine A79001 meteorite: Multiple magma pulses on theshergottite parent body. Geochim. Cosmochim. Acta 47, 1501–1513.

McSween H. Y., Jr., and Treiman A. H. (1998) Martian meteorites. InPlanetary Materials (ed. J. J. Papike) Rev. Mineral. 36, 6-1–6-53.

McSween H. Y., Jr., Taylor L. A., and Stolper E. M. (1979a) AllanHills 77005: A new meteorite type found in Antarctica. Science 204,1201–1203.

McSween H. Y. Jr., Stolper E. M., Taylor L. A., Muntean R. A.,O’Kelley G. D., Eldridge J. S., Biswas S., Ngo H. T., and Lipschutz

M. E. (1979b) Petrogenetic relationship between Allan Hills 77005and other achondrites. Earth Planet. Sci. Lett. 45, 275–284.

McSween H. Y. Jr., Lundberg L., and Crozaz G. (1988) Crystallizationof the ALHA77005 shergottite. How closed is a closed system[abstract]? Lunar Planet. Sci. 19, 766–767.

McSween H. Y. Jr., Eisenhour D. D., Taylor L. A., Wadhwa M., andCrozaz G. (1996) QUE94201 shergottite: Crystallization of a Mar-tian basaltic magma. Geochim. Cosmochim. Acta 60, 4563–4569.

Mikouchi T. (2000) Pyroxene and plagioclase in the Los AngelesMartian meteorite: Comparison with the Queen Alexandra Range94201 Martian meteorite and the Asuka 881757 lunar meteorite[abstract]. Meteoritics Planet. Sci. 35, A110.

Mikouchi T., Yamada I., and Miyamoto M. (2000) Symplectic exso-lution in olivine from the Nakhla martian meteorite. MeteoriticsPlanet. Sci. 35, 937–942.

Mikouchi T., Koizumi E., McKay G., Le L., and Schwandt C. (2001a)Experimental crystallization of the QUE94201 basaltic shergottite[abstract]. Lunar Planet. Sci. 32, #2100.

Mikouchi T., Miyamoto M., and McKay G. A. (2001b) Mineralogy andpetrology of the Dar al Gani 476 martian meteorite: Implications forits cooling history and relationship to other shergottites. MeteoriticsPlanet. Sci. 36, 531–548.

Mittlefehldt D. W., Lindstrom D. J., Lindstrom M. M., and MartinezR. R. (1999) An impact-melt origin for lithology A of martianmeteorite Elephant Moraine A79001. Meteoritics Planet. Sci. 34,357–367.

Mosley D. (1984) Symplectic exsolution in olivine. Am. Min. 69,139–153.

Patsch M., Altmaier M., Herpers U., Kosuch H., Michel R., and SchultzL. (2000) Exposure age of the new SNC meteorite Sayh al Uhaymir005 [abstract]. Meteoritics Planet. Sci. 35(Suppl.), A124–A125.

Roeder P. L. and Reynolds I. (1991) Crystallization of chromite andchromium solubility in basaltic melts. J. Petrol. 32, 909–934.

Rubin A. E., Warren P. H., Greenwood J. P., Verish R. S., Leshin L. A.,Hervig R. L., Clayton R. N., and Mayeda T. K. (2000) Los Angeles.The most differentiated basaltic martian meteorite. Geology 28,1011–1014.

Steele I. M. and Smith J. V. (1982) Petrography and mineralogy of twobasalts and olivine-pyroxene-spinel fragments in achondrite EET-A79001. J. Geophys. Res. 87, A375–A384.

Stolper E. and McSween H. Y., Jr. (1979) Petrology and origin of theshergottite meteorites. Geochim. Cosmochim. Acta 43, 1475–1498.

Taylor L. A., Kullerud G., and Bryan W. B. (1971) Opaque mineralogyand textural features of Apollo 12 samples and a comparison withApollo 11 rocks. In Proc. 2nd Lunar Sci. Conf., pp. 855–871.

Taylor L. A., Nazarov M. A., Shearer C. K., McSween H. Y., Jr., CahillJ., Neal C. R., Ivanova M. A., Barsukova L. D., Lentz R. C., ClaytonR. N., and Mayeda T. K. (2002) Martian meteorite Dhofar 019: Anew shergottite. Meteoritics Planet. Sci. 37, 1107–1128.

Treiman A. H. (1995) S � NC: Multiple source areas for Martianmeteorites. J. Geophys. Res. 100, 5329–5340.

Treiman A. H., McKay G. A., Bogard D. D., Mittlefehldt D. W., WangM.-S., Keller L., Lipschutz M. E., Lindstrom M. M., and Garrison D.(1994a) Comparison of the LEW88516 and ALH77005 martianmeteorites: Similar but distinct. Meteoritics 29, 581–592.

Treiman A. H., Lindstrom D. J., and Martinez R. R. (1994b) The parentmagma of xenoliths in shergottite EET-A79001: Bulk and traceelement composition inferred from magmatic inclusions [abstract].Lunar Planet. Sci. 25, 1417–1418.

Wadhwa M., McSween H. Y., Jr., and Crozaz G. (1994) Petrogenesisof shergottite meteorites inferred from minor and trace elementmicrodistributions. Geochim. Cosmochim. Acta 58, 4213–4229.

Wadhwa M., Lentz R. C. F., McSween H. Y. Jr., and Crozaz G. (2001)A petrologic and trace element study of Dar al Gani 476 and Dar alGani 489: Twin meteorites with affinities to basaltic and lherzoliticshergottites. Meteoritics Planet. Sci. 36, 195–208.

Warren P. H. and Kallemeyn G. W. (1996) Compositional-petrologicinvestigations of eucrites and the QUE94201 shergottite [abstract].In Proc. 21st NIPR Symp. Antarctic Meteorites, pp. 195–197.

Warren P. H. and Kallemeyn G. W. (1997) Yamato-Y793605,EET79001, and other presumed Martian meteorites: Compositionalclues to their origins. Antarct. Meteorite Res. 10, 61–81.


Yamada I., Mikouchi T., Miyamoto M., and Murakami T. (1997)Lamellar inclusions in olivine from Nakhla (SNC) meteorite [ab-stract]. Lunar Planet. Sci. 28, #1393.

Yanai K. (2002) New Martian meteorite identified as a lherzoliticshergottite similar to ALH-77005 meteorite [abstract]. Lunar Planet.Sci. 33, #1248.

Zipfel J. (2000) Sayh al Uhaymir 005/008 and its relationship to Dar alGani 476/489 [abstract]. Meteoritics Planet. Sci. 35(Suppl.), A178.

Zipfel J. and Goodrich C. A. (2001) REE in melt inclusions in olivine

of ALHA77005 [abstract]. Meteoritics Planet. Sci. 36(Suppl.),A232.

Zipfel J., Scherer P., Spettel B., Dreibus G., and Schultz L. (2000)Petrology and chemistry of the new shergottite Dar al Gani 476.Meteoritics Planet. Sci. 35, 95–106.

Zlobin S. K., Kamenetskiy V. S., Sobolev A. V., and Kononkova N. N.(1990) Chrome spinel inclusion data on the parent melt for theparallel-dike complex in the Maynits-Zone ophiolites in the Koryakupland. Geokhimiya 11, 1595–1604.

3772 C. A. Goodrich

petrogenesis of olivine-phyric shergottites sayh al ... · petrogenesis of olivine-phyric...

Documents