Lunar Mineralogy:A Heavenlydetective Story

American Mineralogist, Volume 61, pages l0S9-l I16, 1976

Lunar mineralogy:a heavenlydetective story. Part II1

JosrpHV. SnttrueNo Intt M. SrnslE Department of the Geophysical Sciences, Uniuersity of Chicago Chicago, I II inois 6063 7

Contents

Abstract | 060 Introduction t06r Generaldiscussion of rock tvoes:relation to modelof Moon t06r Lunar specimenswith high MglFe l 066 (a) selectedspecimens of special interest r 066 (b) others I 070 Lunar specimenswith lower MglFe r07| (a) high-KREEPmaterial, mostly noritic t07I (b) low-KREEPnoritic material 107I (c) anorthositic to12 (d) graniticand rhyolitic 1074 (e) marebasalts t074 (f) ANr I 075 Origin of rock types I 075 Discussionof minerals lo'76 Feldspars r076 Olivines 107'7 Pyroxenes 1079 (a) relationof majorelements to rock type and crystallizationconditions 1079 (b) minor and traceelements 1080 (c) exsolution,inversion, and sitepopulation l08l (d) deformation 1082 Othersilicates.... 1082 Spinels 1084 (a) generalchemistry 1084 (b) spinelsfrom non-mare rocks I 084 (c) spinelsfrom mare rocks . I086 (d) miscellaneous 1086 Ilmenite I 087 Armalcolite, zirkelite, and possiblerelated phases . I089 Baddeleyite,corundum, and rutile l09l Apatite and whitlockite ... 1092 Sulfides,carbide, phosphide,and metals I 093 (a) sulfides I093 (b) coheniteand schreibersite I 095 (c) metallic Fe,Ni,Co minerals 109'7 Conclusion I 106 Appendix:additional lunar minerals..... r r07 Acknowledements I r08 References I 108 r059 1060 J. V. SMITH AND I. M. STEELE

Abstract

This secondpart is a revisedand expandedversion ofthe 1973Presidential Address, and incorporatesnew data publishedup to November 1975. The earlierlist of lunar mineralsis expandedto includecordierite, molybdenite, farringtonite, akagan6ite,unidentified Cl,Fe,Zn-bearing phases, bornite(?), aluminum oxycarbide(?), monazite,thorite(?), pyrochlore(?), titanite, and graphite. A general discussionof rock types is set in the context of the phase relations in the silica-olivine-anorthiteand Fe-FeS systems,and a model for early primary differentiation followed by partial melting of cumulates.Most of the returned rocks are metamorphosed brecciascomposed of severalmaterials. Lunar specimenswith Mg-rich mineralsare reviewed in detail becausethey may representearly crystalline differentiates.Emphasis is placed on the chemicalcomposition of olivine and pyroxeneas a meansof sorting out the host rocks, while texturesand modesare given lesssignificance because of erratic mechanicaland metamorphic processes.Rocks with Mg-rich minerals are classifiedas dunitic [with olivine For, ,n, Cr,Al,Fe,Mg-spinel, and sometimesa titanate (armalcolite?)f,toctolitic (Mg-subgroup with olivine Foru-", and Fe-subgroupwith variable olivine mostly Foro-ru),spinel-troclolitic (di- verseolivine Fo", ,0,rare Mg-rich pyroxene,and Mg,Al-rich spinel).Rocks with mineralsof lower Mg-content are classified as high-KREEP material, mostly norilic (pyroxenes mostly Enuo-roFsru-ruWor-u,rare augite, relatively Na-rich plagioclase Anro rr), low-KREEP noritic (pyroxenesmostly Enrr-rrwo-3 but some Enrr-o.Wonu-.r,plagioclase Anrr-rr), anorthositic, granitic and rhyolitic, mare basalts, and ,4NI group. Important aspectsof the following minerals are discussed. Feldspars: trace elementanalysis by ion microprobe; cation vacancyin anorthite; experimental reproduction of texturesin basalts;twin distribution; domain texturesfrom electron microscopy:unusual ternary compositions. Oliuines: correlation of Al,Ca,Ti,Cr,Mn, and Ni with type of host rock; slight Mg,Fe ordering; magnetichyperfine peaks; reaction of olivine clastsin metamorphosedbreccias. Pyroxenes: correlationof Ca,Mg,Fe contentwith host rock and interpretationin termsof crystal-liquid differentiation;low content of Na and K; content of minor elementsincluding Ti,Al, and Cr; estimation of crystallizationand cooling conditions from exsolutionphenom- ena and site populations;interpretation of disorientationand cracking. Other silicates; possiblelimit on pressurefrom occurrenceof silicapolymorphs; limit on availability of water from rarity and compositionsof amphibole;possibility of garnet occurring in deep-seatedrocks; uncertainidentification of melilite. Spinels: apparent absenceof trivalent iron; new schemefor plotting compositionson a tetragonalpyramid with ulvdspinelat the apex;composition trend from Mg,Al-spinel toward chromite-rich spinel and then to ulvdspinelas the host rock trends from troctolitic to mare basalt;correlation of zoning from chromite to ulvcispinelin mare basaltwith relativestage of crystallizationof silicates;reduction of Ti-rich spinel to iron plus rutile or ilmenite plus Ti- poor spinel;partition of Zr betweenilmenite and ulvdspinelas a thermometer. Ilmenile: Mg/Fe ratio which probably depends on temperature and Mg content of coexisting(Mg,Fe)-silicates; deformation occurring at low shock pressure;interpretation of intergrowthswith rutile, spineland pyroxenein terms of exsolution,reduction or shock. Armalcolite:contains about 5 percentTi3+Tia+Ou; breaks down to ilmeniteand rutile below about 800-l000oC; has an Mg/Fe ratio which correlateswith that of coexistingsilicates; has an upper limit for pressurestability at severalkbar. Zr-rich and Cr,Zr-Ca-rich grains in brecciaswere commonly describedas armacolite,but may not be isostructural;their Mg/Fe ratios correlate with coexistingsilicates; because they are found with Mg-rich silicatesthey may occur deep in the Moon. Baddeleyite and rutile: Wide compositional ranges. Phosphate and phosphide: relative occurrence depends on the redox conditions. Whit- lockite carriessubstantial REE. Apatite is rich in F and Cl, and OH is probably absentor very low. Schreibersiteapparently derived from meteoritic debris and directly from lunar rocks, and the partition of Ni and P betweenschreibersite and iron mineral givesa usefulthermome- ter for metamorphosedbreccias. Troilite: ubiquitous in lunar rocks; anticorrelationof Fe and S in mare basaltsindicates LUNAR MINERALOGY

loss of S; positive correlation of Fe and S in non-mare rocks indicates meteoritic contamination; Ti partitioning with ilmenite is temperaturedependent. Cohenite: results from meteoriticdebris, and probably from solar wind. Iron metals'. ubiquitous; indicate a low redox state; result from both meteoritic contamination and indigenousprocesses for which Co and Ni contentsmay provide a fairly reliableclue; occur in veins and vugs indicating presenceof vapor. The sum total of data provides important controls on geochemicaland petrologicmodels. Extensivecrystal-liquid fractionation, impact differentiation,shock and thermal metamorphism, vapor transferand changeof redox conditions,and late meteoriticaddition are needed to explain the mineralogy.Although considerableuncertainties remain, primary differentiation of an olivine-richdry Moon is reaffirmed as a valuable model upon which many subsidiary processesincluding partial melting of cumulates,formation of an Fe,S-richcore, and impact metamorphismcan be addedas subsidiaryparameters. The cluesoffered to the lunar detective are heavily biasedby samplingand confusedby multiple processes,and much further research is needed.

Introduction The appearanceof the review of lunar minerals by Part I (Smith, 1974) set lunar mineralogy in the Frondel (1975) makes it unnecessaryto elaborate context of a model Moon, crudely classifiedthe rock hereon the strictly mineralogicfeatures, and the pres- types, enumerated the observed minerals, and re- ent review concentrateson crystal-chemical,petro- viewed the propertiesof feldsparand olivine. Taylor logic, and geochemicalaspects of lunar mineralogy. (1975) provided a well-balancedview of the general Minerals not discussedin detail, and not listedin Part propertiesof the Moon, but his detailedgeochemical I, are given in the appendix. model has somecontroversial features especially concerning the origin of mare basaltsand KREEP-rich rocks.The developmentof the early Earth and Moon General discussionof rock types: was reviewed by Smith (1975), and a model was relation to model of Moon developed for origin of the Moon by catastrophic The confusion in lunar rock nomenclature has capture followed by accretionof the Earth-orbiting been partly reducedas petrologistsand geochemists debris. Simultaneousaccretion of solar-orbiting ma- realizedthat most lunar rocks are hybrids which have terial by a proto-Earth and an Earth-orbiting proto- undergonesequential changes including erratic shock Moon was also consideredlikely, while origin of the metamorphism. Most lunar petrologistsbelieve fhat from or volatilization the Earth Moon by fission of the outer part of the Moon (perhapsl0 km thick) is was thought to be unlikely. Uncertaintiesin models dominated by debris produced by repeatedimpact for the bulk composition of the Moon were empha- from -4.5 to -4.0 Gy into a complex assemblageof particular, sized: in existing estimatesfor Ca differ igneousand metamorphic rocks. Impacts by bodies four-fold. some tens of kilometersacross yielded superimposed Moon-wide blankets ranging from a few kilometers 'Note by J. V. Smith Revised version and expanded of Presi- thick at the edge of an impact basin to some meters dentialAddress, Mineralogical Society of America, deliveredat the 54th Annual Meeting of the Society,l3 November 1973.The first thick at great distances.The debris consistedof com- part was published in Vol. 59, p. 231-243, 1974, but various plex mixtures of material excavateddown to depths circumstancesmade it impossibleto preparethe secondpart imme- of some tens of kilometers mixed in with dispersed diately thereafter. This was actually fortunate becausethis second material from the projectile.Secondary impacts and part, completed in November 1975,could incorporate the exten- differentiationduring transport causedfurther com- sive data and ideaswhich marked the end of the Apollo program. plication, and could account for the bulk present [Final revisionof the manuscriptin May 1976allowed addition of of referencesto several papers appearing after November 1975.] The regolith. Impact basinswere partially filled with hot first part is brought up to date, the rock types are evaluated, and debriswhich becamemodified by metamorphismand the remainingminerals are reviewed.Although a PresidentialAd- volcanism.Mantle materialuplifted the basinfill, and dressis personal,few recentPresidents could maintain that their late mare basaltsprovided a cap. The ejectablankets work was carried out totally by themselves.Ian Steelehas been a remarkablyproductive colleague since the secondyear of my lunar are dominated by metamorphosedbreccias ranging program, and it is appropriatethat we preparedthis articlejointly from poorly consolidated material to partly-melted even thouqh it modifiesa tradition samples. These complex assemblageswere further 1062 J V. SMITH AND I, M. STEELE

modified by smaller impacts to produce a veneerof were invented and terrestrial terms were used in a regolith. ' loose or extendedway. A glossaryis given in pages Most non-mare rocks are inhomogeneous,and i-xi of the Proceedings of the Fifth Lunar Science some controversy among lunar petrologists arose Conference.Many namesare confusing,and we shall merely because they had examined different sub- try to useclear terms such as "polymict brecciawith samples.Other controversiesarose from differentin- anorthositebulk composition." terpretationsof the samecomplex texture. Thus some Apart from the mare basalts, which can contain thin sectionsfrom rock 14310show sporadicclots of large amounts of ilmenite and significantamounts of plagioclasewhose irregular grain boundariesare con- diopsidecomponent, most lunar rock compositions sistent with solid-stateannealing, whereas the great can be representedclosely on the ternary plot of bulk of the area is dominatedby an undoubtedig- silica-olivine-plagioclase.The liquidus relations for neous texture. After considerabledispute on frag- the pure SiOr-MgrSiO.-CaAlzSizOssystem at I at- mentary evidence, most investigators now believe mosphere dry conditions (Fig. I ) show crystalliza- that 143l0 was producedby nearlycomplete melting tion fields for forstQrite, Mg,Al-spinel, anorthite, of regolith in which only the clots and trace elements Mg-pyroxene, and tridymite or cristobalite(Ander- provide evidenceof the earlier hybrid state (seepa- sen, l9l5). These crystallization fields change little pers by Ridley et al., 1972,James, 1973,Gancarz et with substitution of Fe for Mg and NaSi for CaAl al.,1972,and many papersinthe Proc.3rdLunar Sci. (e.g. Walker et al., 1975), and a single generalized Conf. for conflicting ideas: note that Crawford and diagram is sufficientfor considerationof melting of Holfister, 1974, from the presenceof Na-rich cli- lunar highlandsmaterial at low pressure.Key features nopyroxene and low-Al hypersthene, interpreted of the diagram are (a) incongruent melting of bulk 14310as a product of melting 200 km deep in the compositionsin the region pqt to Mg,Al-spinel and Moon). We acceptthe generalopinion in 14310,and Mg,Al-depletedliquid; equilibriumcooling results in suspectthat many lunar rocks with igneoustextures reaction of spinel with the liquid to produce olivine are the products of impact melting of complex and plagioclasebelow the solidus(Fig. 2) for pressure breccias. lessthan Skbar;(b) incongruentmelting of pyroxene Most Apollo and Luna specimensare metamor- to olivine and Si-rich liquid; with increasingpressure phosedbreccias in which detailedmineralogical study the incongruent melting is reduced, and is finally has revealedextensive chemical and physicalchanges eliminated at severalkilobars, (c) decreasingtemper- betweenmineral grains accidentallythrown together. atures along the cotectic curves qr and rs to the A good idea of the complexity can be obtained from the.following papers: Albeeet al., 1973;Anderson e/ al., 1972: Bence et al., 1973, 1974; Cameron and - Fisher, 1975;Chao, 1973;Grieve et al., 1975:Kurat mg 0.7 1.0 et al., 1974;Simonds et al., 1973, 1974:'Steele et al., - - Ti-rrchmore 1972;Warner, 1972,Warner et ql., l9'74.All grada- bosolt tions occur from (a) monomict brecciaswith angular mineral claststo (b) low-gradepolymict brecciaswith open pores and residualglass to (c) medium-grade polymict brecciaswith partial recrystallizationof the mineral clastsalong with completeelimination of the PL46IOCLASE glass to (d) high-grade breccias ranging from ones with hornfelsictextures to oneswith igneoustextures. As the grade increases,the mineral compositionsbe- come homogenized,and information on the primary Mg-rich Co-rich constituentsbecomes lost. Partial melting and vapor oliv ine feld spor transport may affect the texturesand bulk composi- Ftc l. Crystallization fields of SiOr-Mg-rich olivine-Ca-rich tlons. plagioclase system for dry conditions at I atmosphere (Fe-free, Becauseso few even-texturedrocks occur on the Andersen, l9l5; mg 0.7, Walker et al , 1973b) Temperatures for Fe-freesystem See Lipin (1975) for abstracton data for mg 06 Moon, petrologistsand geochemistswere forced to Labelsp-w indicatereference points discussedin thetext H and M userock terms in a chemicalrather than a textural or representpossrble bulk compositions of the lunar crust and the mineralogical sense; furthermore many acronyms whole Moon LUNAR MINERALOGY 1063

km depth Fe,Ni,S-rich liquid to form a core, (b) gravitational sinking of Mg-rich olivine and minor amounts of heavy complex oxides to form a mantle, and (c) development of a late liquid rich in plagioclaseand pyroxene components which would ultimately form the complex crustal rocks. The Fe-rich core might become a magneticdynamo. The Fe-FeS eutecticis close to 1000' C for all lunar depths, whereas the melting curve for iron risesfrom -1520 to -1650" (Brett, 1973;Usselman,1975). A S-richliquid would be strongly superheatedif the supposedselenotherm is correct, thereby reducing the viscosityand easing fluid-dynamicalproblems of an early lunar dynamo. During progressivecrystal-liquid differentiation the Fe contentof the liquid would increase,as would the kb "incompatible" elements(e.9. K, GPo amount of the grovitolionolpressure REE, Ba, P, Zr, and U) which enter the octahedral and tetrahedralsites of common silicatesonly in tri- FIc. 2 Some relevantphase equilibria for a Moon with olivine, pyroxene,and plagioclaseas the dominant minerals See text for vial amounts. Late partial melting and volcanism origin of curves. would confuse the situation greatly, as would the effectsof major and minor impacts. Figure 3 shows eutectic s at 1222"C for the Fe-free system; (d) how the mean density of the Moon could be ex- decreasingtemperatures over the whole diagram as plained by this kind of petrological model. Even if Fe substitutesfor Mg and NaSi for CaAl. only the outer part of the Moon melted,as proposed Figure 2 showssome relevantcontrols on the min- by many workers (e.g. Taylor, 1975),many of these eralogy of the Moon. The present-dayselenotherm results from interpretation of the electrical con- (km) ductivity profile deduced from perturbation of the deplh SolidFe,Ni solar wind (e.9.Sonett, 1975; Shankland, 1975). As- densily 7.5 sumption that the mantle consistslargely of Mg-rich '\very rore g/cn3 4.5- - olivine essentiallyfree of Fe3+leads to a model-de- t ilmbniterich cumulote FeSmell neor5.0 pendent estimateof the presenttemperature profile (Dyal et al., 1974) using olivine data of Duba et al. Densilyof 4.0 , olivineond Q97a); assumptionof Mg-rich pyroxenegives a sim- 1 pyroxene m0re Forr*Co,Al-Enuu ilar profile(Heard et al.,1975).Although the details t Fo z / roremullrple 0xtdes *______[-l are controversial, the general trend of the seleno- t/ 70 therm is with 5:-:::eo 51 consistent seismicevidence for melting ---;;/'F-:gobi below - l000km depth: if the molten liquid is basal- =-too'" | tic, the required temperature for 1000 km depth is Ty^^;i;1,.,zo"c tsoo"c -1500' /-oia,nlnsiie Foru okb 38kb C which is consistentwith the proposed gronit-=--='ii*E - 1000km curve. The selenothermis estimated for the entire 1500 t250 t000500 0 depth (r3 early Moon reaching the temperaturefor extraction rodiuskm plot) of basalticmelt from an ultrabasicbulk composition Flc 3. Mineralogicalmodel of Moon expressedas densitypro- when account is taken of near-surfacecooling over file The small diagram shows the variation of density of olivine and low-Ca pyroxenefrom the past 4 Gy. Complete melting of Mg-rich olivine 20oC and 0kbar to I 500"C and 38kbar. the latter of which are possibleconditions at -l00km depth. A would require temperatures up to 1800-2000' C, possible lunar core might consist of either Fe,Ni or Fe,S or a which would exacerbateproblems of heat sources. mixture of both A mantle might consistdominantly of olivine and Figure 3 is a suggestedmineralogical model of the low-Ca pyroxene.The crust is probably dominatedby rocks rich in Moon expressedas a density-depth profile. Details pyroxene and plagioclase(point l/, Fig. I). Mare basalts may result from will be given elsewherein a paper on the chemical melting of cumulatesincluding ones rich in ilmenite. Note that many small variants would satisfythe mean densityof composition of the Moon. Complete differentiation 3, 3.343 gcm including a mantle whose olivine was uniform at of the Moon would result in (a) extraction of -Fo,, From Smith (1976). J V. SMITH AND T. M STEELE featuresof crystal-liquid fractionation would occur task to find out whether lunar rocks and minerals (e.g. Murthy et al., l97l). give clues on the applicability or otherwise of this Ignoring the possibleFe-rich core and the minor model. [Elsewherewe shall consider the validity of constituents,the bulk composition of a completely models proposed by other workers based on bulk differentiatedMoon might approximate to point M compositionsrich in pyroxene(Ringwood) or Ca,Al- on Figure I, which liesin the field outlined by Walker rich refractory condensate(D. L. Anderson).] et al. (1973a). Olivine would be the liquidus phase A simpleapproach is to examineall the bulk analy- whatever the depth in the Moon. Sinking of olivine ses of lunar sampleswithout taking account of the would result in the bulk composition of the remain- mineralogicalfeatures. Wood (1975) separatedmare der moving away from the olivine corner. Depending from highlandsmaterial by a plot of Ca/Al us.TiO,. on the exact position of M and the extent ofcrystal- Those specimens with CalAl (atomic) ) [0.786 liquid reequilibration,the subsequentdifferentiation -O.229log TiO, (wt.7o)lwere classifiedas mare ba- could take differentpaths. Figure 2 showsphase rela- salts. The remaining samples were split into high- tions for a bulk composition with equal amounts of KREEP and low-KREEP varietiesby the line KzO * forsterite and anorthite (based largely on Kushiro P2Ou(wt.Vo : 0.945 -0.00893 (7onormative plagio- and Yoder, 1966). At low pressure,anorthite and clase). Figure 4 shows the resulting modal plots of forsterite are the subsolidusphases. Anorthite dis- olivine-plagioclase-quartzfor the two groups of pre- appearsat the solidus(-1320') giving spinel * for- sumed highlandssamples. Many of the bulk analyses steritet liquid, then olivinedisappears at -1380' C by broad-beam electron microprobe analysisare for leaving spinel * liquid, with complete melting at small lithic fragments, and both the sampling and 1500-1600'C. In the subsolidusregion, anorthite analyticalerrors are significant.Nevertheless the data and forsterite are replacedby anorthite, spinel, and show interestingtrends. The KREEP-rich specimens two pyroxenesabove about Skbar, and then by gar- tend to lie near the cotecticqr of Figure l, consistent net and pyroxeneabove about 15kbar.The detailsare with proposals for origin from cotectic primary or controversial,but the generalrelations are sufficient secondarymelts, but the spreadis so wide that other for the presentpurpose. Depending on the bulk com- origins are indicated as well (e.9. hybridism). The position M and the temperatureprofile of the Moon, KREEP-poor specimenstend to lie in the lower right, garnetmight form in the lunar mantle.Assuming that and this might be explained by derivation mainly the temperature profile of the early Moon is con- from primary cotectic liquids in the anor- trolled by the "basalt" solidus, garnet should not thite-plagioclase-silicasystem, with a bias causedby form at depths lessthan -300km. Only half of the the tendency for plagioclaseto rise to the lunar sur- lunar volume is insidethis depth, and olivine may be face becauseof its lower density.A few samplesare the sole crystallizing silicate there, thus eliminating very rich in olivine, and we shall focus attention on garnet except perhaps for the products of trapped them becausethey may derivefrom the deep crust or liquid. On the other hand, the liquid may have upper mantle. reacheda composition sufficientlyrich in An mole- The remainder of this section will concentrateon cule to yield some primary garnet. Further complica- evaluation of the rock types in the lunar samples, tions can arise in the depth range 100-300 km be- using the chemistry of olivine, pyroxene and spinel cause of the possible assemblageanorthite * two (Figs. 5-12) as the principal mineralogic clues. Be- pyroxenes* spinel. cause all lunar petrologists assume that the lunar For depthsless than 100km, crystallizationwould highlands are dominated by Ca-rich plagioclaseand be controlled principally by the relationsin Figure 1. low-Ca pyroxeneas expressedby rock terms such as Point fl on the olivine-plagioclasecotectic would highland basalt and noritic anorthosite,we shall de- mark the final disappearanceof olivine when account liberately focus attention on the mineralogy of con- is taken of elimination of the incongruentmelting of troversial materialswhich may representdeep-seated pyroxene at several kilobars pressure (or of sub- rocks. Study of rocks from large igneouscomplexes solidus reaction). The lunar crust would be domi- and granulite terrains on Earth, together with in- nated by Mg-rich pyroxene and Ca-rich plagioclase vestigation of xenoliths transported by kimberlites with a bulk composition near -F1.Subsequent differ- and alkali basaltsfrom the upper mantle, providesan entiation would terminateat the eutectics. Of course, excellent basis for establishing mineralogical and this ideal theoretical situation could not occur, and chemical criteria. Under plutonic conditions, the manv idiosvncratic variations must occur. It is our grain sizeis often coarse,and cumulatetexture occurs LUNAR MINERALOGY 1065

OURRTZ

HIGHKREEP PLOTOO?5 HIGfILRNDSSRHPLES

LOhKREEP I,.IHOLEROCK HIGHLRNDSSBI1PLES RNRLYSI S

LITHIC FRHC,IIENT

FNORTHI TE

OLIVINE ,o,rt. FNORTHI TE

FIc. 4. Principal components of highlands samples selectedfrom computerstore. Wood (1975).

in some igneous complexes; however, granulation Becausethere are so few rocks with simpletexture and recrystallizationcan lead to a variety of textures and chemistry, the lunar detectiveis forced to work from porphyroclastic to equant (e.g. Mercier and with complex rocks. As an example Figure 5 shows Nicolas, 1975; many papers in Phys. Chem. Earth, textural features of 67075,a polymict breccia with Vol. 9). Chemical featuresshould provide easiercri- anorthosite bulk composition (Brown et al., 1973; teria especiallyfor lunar specimensin which impact Smith and Steele,1974; McCallum et al.,197Sb).The metamorphismis so common. The chemicalfeatures overall view (a) shows the angular clasts of plagio- indicative of deep-seatedorigin include lack of zon- clase,pyroxene, and olivine, while the magnifiedpor- ing, presenceof high-pressureminerals, chemical par- tions (b-f) show special features described in the titioning indicative of high-pressure, coarse ex- figure legend. Particularly interesting is the coarse solution and strong cation ordering indicative of exsolution of pyroxene clasts (d-f), which indicates prolonged annealing, low content of certain minor prolonged subsolidusannealing (this is the coarsest efements(e.g. Ca in olivine), and chemical features recorded in lunar pyroxene). Although individual inherited from a high-pressureancestor. In addition, pyroxenegrains have uniform compositions(except, certain chemical featuresindicate early crystallization of course,for the exsolution)the pyroxenesof differ- from primitive liquids of which the simplestisthe mg ent grains in the samethin-section range in composi- ratio (: atomic MglMg+Fe) of the ferromagnesian tion from EnnoFsuuWonto En.uFs.rWosand from minerals.Rocks composedof early crystalsshould be EnroFsrrWon,to EnorFsrnWonnwhile the olivines low in the "incompatible elements" which tend to range from Fonoto Fouu.The simplestexplanation is end up in granitic residua,and should haveaccessory that 6'7075 is composed of the compacted debris from minerals rich in most transition metals of the first different parts of a layered igneous complex which series(Ti-Ni). underwent prolonged annealing before disruption 1066 J. V. SMITH AND I, M STEELE

(Brown et al.,1973).lf 67075had beenstrongly meta- morphosedor melted,the aboveevidence would have been lost.

Lunar specimens with high Mg/Fe

All the lunar missions brought back soils and brecciaswith rare grainsof olivine whosehighly magnesian composition (-Fono) is like that of olivines from the Earth's upper mantle (e.g. Heiken, 1975). Steeleand Smith (1972)located four Apollo 14 fragments with ultrabasicaffinities, and many more have since been discoveredincluding some larger specimenswith indicativetextures. All thesespecimens are more magnesian(mg -0.9) than the common material (bulk mg mostly < -0.75). We first describesome specimenswith particularly interestingtextures and mineralogy (Table I and Figures 6, 7, and l2), and then summarize pertinent mineralogical data for other Mg-rich materials. (a) Selected specimens of special interest. The clearesttextural evidencefor deep-seatedprocesses is FIc. 5. Photomicrographsof 670'75,a polymict breccia with from 76535(Gooley et al., 1974),a 155g rake sample anorthositebulk composition composed mainly of -5mm grains of plagioclase (a) Dark portion at left consistsof olivine (-Fouo, fractured) and (587o,Anr6), olivine (37Va,Fo'6) and orthopyroxene plagioclase (unfractured) clasts set in plagioclase-rich matrix. Light portion shows various clasts of plagioclasegrains plus a (4Vo,Eng.). Bence et al. (1974) found similar 2-4mm larger clast of polycrystallinemicro-anorthosite with small equant fragments in Apollo l7 soils. The grain boundaries pyroxenes;minor olivine clasts(-Fotu) also occur are curved and tend to meet at 1200,and the texture (b) Polycrystallinemicro-anorthosite clast, mostly with triplejunc- is that of a thoroughly annealedrock which has not tions and polygonal texture Small pyroxene grains are hard to been brecciated or shock-metamorphosed. The recognize Same texture as clast in (a). (c) Polycrystallineclast dominated by shocked 2 mm grain of pyroxene hae the spacegroup P2,ca (Smyth, 1974a) anorthite Matrix showsat upper left, while other anorthite grains which resultsfrom unusuallyeffective segregation of with interstitial pyroxene grains occur at the right of the clast. Mg and Fe. Although the depth of crystallization (d) Pyroxene clast showing coarse and fine exsolution lamellae inferred from thermodynamic estimates is con- The matrix containsangular plagioclaseclasts. troversial (Albee et al., 1975),all authors agreethat (e) Clast composedof 2 mm pyroxenegrain with coarseand fine intergrowth (central region) and portion of large plagioclase(dark 76535was annealedin the solid statefor a long time regionat lower left). The compositeclast is surroundedby matrix. at considerabledepth: indeed the textural features Matte areas are cracks filled with epoxy The pyroxene grain is can be matched in terrestrial granulite rocks. Inter- traversed by a band of plagioclasegrains with curved boundaries esting details are: (a) the plagioclasecontains orien- (gray and white grains running NNW through the center). The (5-15 wt.VoNi and up to centralpart is enlargedin (f). ted incfusionsof iron needles (f) Central part of (e). Spineloccurs as black inclusions(E-shaped 5VoCo), chromite, and an unidentifiedphase; (b) the at lower left and rod-shaped at upper and lower right). Pyroxene olivine contains discrete chromite inclusions and exsolution lamellae occupy the left and right parts of the field of symplectic intergrowths of chromite and pyroxene; view. The centralone-third consists of plagioclasegrains, the outer (c) symplecticintergrowths of orthopyroxene,diop- ones of which tend to be scallopedwhile the inner onestend to be side and Al-Mg-chromite occur betweenolivine and ellipsoidal.Similar plagioclasegrains occur as dark-grayentities at the margin and just insidethe lower edgeof the pyroxenegrain in plagioclasegrains; (d) mosaicclusters of orthopyrox- (e). The origin of this plagioclaseis unknown but injection of ene, diopside, chromite, Fe-Co-Ni metal, troilite, plagioclase into fractures followed by prolonged annealing is one whitlockite, chlorapatite, baddeleyite, plagioclase, possibility.Faint horizontal lines are blemisheson film. and K-feldspar occur sporadically,and are most eas- (a) (b) (c) and (e) 2 mm across,(d) I mm across,and (f) 0.5 mm ily interpreted as the products of trapped primary across. (a) and (c) plane-polarized light; others crossedor partly- crossedpolars. (a) and (c) from 67075,2;others 67075,51 liquid augmented by exsolved material; (e) some See McCaflum et al (1975b) for other photographs. metaf particles consist of a-7 assemblageswhose LUNAR MINERALOGY r067

Trslr I Analysesof mineralsin selectedlunar peridotites

o tn l1 T2 t?

sio2 39'9 55.9 55.5 52.5 44.9 - 40.1 5!.3 52.7 44.8 0.19 +2.4 40.8 Tio2 0.03 0.4 0.3 o.T o.o2 a.g-r.2 <0.01 0.31 0.54

roral 99.2 100.1 100.2 99.9 99.5 - 100.4 101.0 99.9 roa.z 99.8 LaL.5 99.8

14 L5 16 17 18 rg 2a 2L 22 2? 2)J 2\

s i0^ 55.a J6.5 +5.2 o.o - -t.tla ^ 53.2 ]lLq 0.00 0.00 Ti0^ 0.78 0.15 - 0.03 o.or 98.o 76.9 - o.9 o.04 t7.o 4.0^o- 5.r 2.4+ 34.3 tT.3 5B.o o.oo 5.9 5".+ ATTDQE o.41 14.0 12.7 0.3 O.5 O.04 o.4 8.6 0.43 Fe0 5.8 5.7a o.02 9.3 9.4 0.1 9.6 9.8 5.L 0.05 8.9 28.9 Mn0 0.0 J.09 - 0.43 Mgo 33.5 35.7 o.02 2a.\ 20.4 <0.1 13,1 49.3 33.1 0.ol 2L.4 11.5 ca0 0.0 <0.01 18.4 0.0 <0.1

Total 100.7 10o,5 99.7 1Or.O lOO.5 IOO.1 1O0.11OO,5 100.1 loc.O 100.0 101.2

L-6 76535 troctolitic granulite (Gooley et al.r 1974), f olivine Fog7.5, 2,3 bronzite as single grains Erg6.2F"lO.9tdo2.9 rrd in mosaic inlergrowths and symplectites EngT,2Fs1f.5Wof.3, 4 djopsjde in sJ-mplectiles and nosaics Ln'l;I.,"' . ^fs .Wo, o. anorrhi Le Ab" /An-tr ,0r^ -. K2oo.09 r AZ-rvs-chromiLeranse ro. alr jor"i'.\-o.", ur"o tr^."iil :Jt'.e-E'.8 .t t,2-6,L, taenite co 1.8-4.2 I'ti 18,0-33,0; 7-l] 72415 brecciated dunite (Albee eL a1.,1974b), 7 olivine FoBBr llio 0.04, 8 low-Ca pyroxene EngLlslll,iogr t hieh-Ca pyroxene En54Fs5 Wo4I, lo anorthite AbrAnr2orl others 2, K2O o.09 BaO 0,04, 11 Al-l'{g-chronite, VrO, A.35, ZrO2 o.AB, also Fe-metal Co f.3-2.2 Ni 24.5-3I.8 t'Ig 1,3 Cr o.35 Si 0.04; 12-20 brecciated peridotite clasts in 15445 breccia (Anderson, 1973; Ridrey eL aL., L973) L2 olivine (A) afso NiO 0,03, 13 olivine (R), f4 orthopyroxene (A), 15 orthopyroxene (R) 16 anorthite (A) also K20 0.12, 17 Cr-spinel (A), f8 cr-spinel (il) middle of four analysesj 19 rutile (e) afso NbZa5 I.6, 20 arnalcoLile (A);2--25 brecciared pe"ido'i'e'ragmen's in 3243 soil (Bence eL al., o 19.1\,21 olivine (mean ol- ,22 py-o*"ne (mean of ljrsr I anallses).23 anorLliie (mean of 2, KAA O.AT), 24 spinel (mean of 2), 25 ilnenite.

composition indicatesa cooling rate consistentwith silicatesin terrestrialperidotite xenoliths (Basu and some tens of kilometers below the lunar surface. MacGregor, 1975;Dawson and Smith, 1975),and The origin of the chromite-pyroxenesymplectites exsolution,especially from pyroxene,combined with is controversial.Gooley et al. (1974) suggestedthat solid-staterecrystallization provides a plausible but they formed by a reaction Ol + An * chromite - not definitive explanation of the terrestrial occur- Opx f Cpx * Al-Mg-chromite, whereas Albee rences.Perhaps chromite tendsto intergrow with sili- et al. (1975) argued for co-precipitation during cate under deep-seated metamorphic conditions magmaticcrystallization. Bell and Mao (1975)pro- whateverthe sourceof the silicate;if so, the origin of posedthat similarsymplectites in12415 brecciated the lunar symplectiteswill be hard to discover. dunite and in severallunar olivinesresulted from Alf largegrains in 7 6535have an inclusion-freerim solid-statebreakdown of primary garnet trapped which results from diffusion of exsolvedmaterial to in olivine. The garnet would be very Cr-rich the grain boundaries. Whatever the details of the (-Ca' rMgorFeo rCr, rAlo.rSi.O,,for 72415 dunite). annealing mechanisms,we endorse the mineralogic Chromite forms vermicular intersrowthswith several implications that 76535 is the product of prolonged 1068 J. V. SMITH AND I, M. STEELE

claseclast and minor chromite with TiOu ( l7o.Unpublished. I l. Polycrystalline inclusion, lmm across,in mare basalt 74275(Meyer and Wilshire, 1974). 12. Dunitic fragment in polymict breccia 14318 Olivine Fouz with average For". Minor plagioclase ", An". with cluster at Anr, (Kurat et al., 1974). ". Oliuine-plagioclase-rich (" troctolilic" ) I Clast in rake sample 15335."lgneous" texture (Steeleet al , ST TzO 1972\.2. Troctolitic granulite 76535and 2-4 mm fines76503. See Table I and text. 3 Clast in breccia14321. Granular assemblageof anorthite Anrr, and accessoryK-feldspar (Compston et al 7 olivine, , oq Spinel 1972). 4. Several 2-3 mm fragments in breccia 14320,4. Oli- .t- TroctoIi tic vine-plagioclase An"u intergrowth with texture similar to cres- -l t62295 cumulate of layered ultrabasic rocks of Rhum (Brown et al., 1972). 5 Fragment, I mm across,in l-2 mm fines 14166,6 70Voplagio- clase An,u,30% olivine,2% diopside,8 tiny grains Spinel-A north itt Cr-Zr-armalcolite (Steele and Smith, 1972). 6. Fine-grained an- -0livine ( 3 Opx nealed clast in recrystallized noritic breccia fragment 76503,6,21. I Cotoclos ite PlagioclaseAnr., olivine, minor orthopyroxene,and augite(Bence et al., 1974).7. Troctolitic breccia77017. Olivine Fo"' and plagio- claseAn"u, patchesin feldspathic breccia Interpreted as xenolith in ,, ta tl mare basalt(Brown el al., 1974).8. Clastsin stratifiedBoulder l, t0 Station 2, Apollo l7 Most have fracturedolivines in granulated 9 groundmass of olivine and plagioclase,but some have basaltic texture Plagioclase is An.. ,uOro, r n with An usually over 8870 9-?e o- (Stoeseret al.,1974) 9. Clasts in stratifiedBoulder l, Station 2, .+_ Apollo 17.Granulated ANT (Stoeseret al., 1974).10.Luna 20 soil fragmentswhose bulk compositionis noritic or troctoliticanortho- - +76535 T'oc' I to lit ic '1 site (Prinz et al, 1973a). ll. ditto for anorthositic norite and troctolite compositions.For specimens8-l l, the analysesare dis- played as histograms. 12. "Basalt" fragment 14162,41-1407-ll 30% skeletal olivine For. microphenocrysts,l07o Na,Al,Cr-rich pyroxenes En.oFsuWo.. and EnurFsrrWo2T5,5070 plagioclase and l0% glassymesostasis (Powell and Weiblen, 1972). Dun itic _ *72415 Sp inel-ano r t hi t e-o rt hop y ro xe ne-ol iu i ne cat aclasit e ll L 15445,10clast in breccia(Anderson, 1973).2. 15445Type B t00 90 80 70 60 clasts in breccia(Ridley et al , 1973) 3.73263,1,112-4 mm soil fragment (Bence el al , 1914). See Table I and text for details.

nol.% Fo Spinel troctolitic FIc. 6. Compositions of olivine in selectedsamples, many of 1.62295 rock. lVo xenocrysts of olivine Forr-ro, anorthite mg-rich composition. Ao"" pink spinel (mg O.75,3 5 mol.7oCrzO.) and Fe-Ni-Co "u, meLal.99Vo fine-scale intergrowth of olivine Fo"r ,", plagioclase Oliuine-rich (" dunitic" ) Arr"u spinel(zg 0.9-0.76,2 mol9o CrzO"),and glassyresiduum ",, l. Soil fragmenl73242,4,Al4:olivine brecciawith minor plagio- (Walker et al., 1973b).2. 62295 rock Olivine phenocrysts Forr-". claseand spinel.One clast of pink spinel and plagioclase(Steele with overgrowths For. ,, set in groundmass of anorthite and Smith, 1975b). 2. Dunite clast in 14068,10breccia: polyg- An"oAb"rOro.skeletal olivine For"-rr, spinel,ilmenite, elc (Agrell ranular mosaic of olivine For" with single chromite grain Helz et al., 19'73).3. 62295 rock. Olivine (xenocrysts Fo".-.", inter- (1972). 3. Dunite clasts 72415-18:see text (Albee et al., 1914b). 4. growth Fo., .r), plagioclase (xenocrysts An".-"r, intergrowths and Soil fragment 78422,2,A18: olivine breccia, I mm across, with quench grains Anro spinel (xenocrysts9-16 mol 7ochromite, "r), minor plagioclase, chromite and armalcolite (Steele and Smith, intergrowth2-4 mol % chromite)(Hodges and Kushiro, 1973).See 1975b). 5. Soil fragment 14002,7,E-l-8:shocked single olivine also Brown et al (1973)and Weiblen and Roedder(1973) 4.4 mm grain with minor chromite (Steeleand Smith, 1972).6. Soil frag- clast 67435-14in breccia.Cumulus of olivine Fo"', ,, o and spinel ment 78502,17,A34: olivine brecciawith minor chromite, armalco- poikilitically enclosedin plagioclaseAD6 u ,, o Prinz et al., 197lb). lite, and plagioclase(Steele and Smith, 1975b).7. Soil fragment 5 Luna 20 plagioclaseAnrr,rr-olivine Fo"-spinel fragment (Tara- 78442,2,A15:olivine breccia,I mm across,with minor chromite, sov et al., 19'73).6 I mm fragment 14321,76of olivine -Fo"u, armalcolite,diopside, and plagioclase(Steele and Smith, 1975b).8 anorthite An"n, and spinel in breccia(Steele, 1972).7. Rake sample Soil fragment 78442,2,A16: olivine breccia, I mm across, with 65785 with coarse-grained cenler (65Vo An"r,30Vo Fo"o-"", spinel minor plagioclaseand armalcolite (Steeleand Smith, 1975b) 9. zonedoutwards from 2.6to l2.6VoCrzOs and FeO l0 to l2%). Rare Breccia 14063,14.Fine-grained clast of 7070 olivine, 30Voplagio- pyroxenes, ilmenite, Ni-iron, troilite, Zr-rutile, whitlockite, claseAn.", and minor bronzite and chromite (Steeleand Smith, Cr-Zr-REE armalcolite, K-feldspar, chromite, and farringtonite 1972). 10. Breccia clast 78502,17,815:olivine breccia with plagio- (Dowty et al., 1974b).8. Thirty-four fragments of spinel troctolite LUNAR MINERALOGY 1069 metamorphismat considerabledepth (tensof kilometers?), probably of an igneous cumulate, in which -.a-N ond TA subtle effects resulted from exsolution, diffusion and m0l reduction.Haskin et al. (1974)concluded from analy- o/ lo o dunitic sesof REE, K, Rb, Sr, and Ba that 76535was origi- o troctoliiic nally an olivine-plagioclase cumulate containing tr sp-0n-opx-01 8-16 percent parent liquid with no appreciableEu cotoclosite x anomaly and about 20 timeschondritic abundanceof spinel-troctolitic + 60618,l- | REE. The age data are inconclusive(reviewed by Haskin et al., 1974,p. l2l4), but original crystalliza- En 80 OU 40 tion at -4.3 Gy is indicated. Frc 7. Compositions of pyroxene in selectedsamples, many of Five olivine-rich fragments72415-72418 were col- Mg-rich compositions. Numbered specimens can be identified lected from a l0 X 20cm clast of metaclastic breccia from the appropriate sub-group of specimenslisted in Fig. 6. 72435. Olivine Foru-r, accounts for 9l percent of Specimen60618,t-l was describedas a spinel-olivine-anorthosite, 72415 as crystals up to one cm set in a granular and is listed as No. l4 in the spinel-troctoliticsub-group. Other data for the spinel-troctolitic group are lumped together without matrix of olivine with the samecomposition (Albee er numbered identification N and TA and AN and T are items l0 al., 1974b).Some large olivinescontain strain bands and I I of the troctolitic group marked by small recrystallizedolivines 50pm across. Plagioclase Anrr-rr, Cr-spinel, high- and low-Ca This evidenceplus the relativelyhigh CaO content of pyroxeneand metal occur as clastsand as inclusions the olivine (0.l3 wt.7o)suggests a higher temperature in the olivine, sometimesas symplecticintergrowths. of equilibration for 72415 than 76535. The 72415 The compositions of the two pyroxenes are closer brecciateddunite is particularly important becauseof than for those in 76535(Table l, Fig. 7) and the low- the tentatiueconclusion from the Rb-Sr isotopicdata Ca pyroxenemay be a clinopyroxene(EnrrFsrrWor). that crystallizationoccurred very early (- - 4.6 Gy). Polymict breccia 15445contains white clasts un- in Luna 20 sample with plagioclaseAnr,-r, mean An"", olivine usually rich in olivine, spinel,and pyroxene(Ander- Forr,ro mean Fo"n, low-Ca pyroxene Enrr-rrFsrr-r,Wor,r, and son, 1973; Ridley et al., 1973) which were called spinel rzg 0. I -0 5 cr 0-0.2 (Prinz et al., 1973a).9. Fragment 100I 9- peridotite. Benceel al (1974) found similar minerals 22. PlagioclaseAn"u., olivine For,, spinel and glass(Keil et al., in 2-4mm fragments in 73263 soil and incorrectly 1970). 10. I mm clast l43l9,lla-3, PlagioclaseAn"u, olivine Fo,. a new rock type, spinel and spinel mg 0.54 cr 0.027 TiO,03-1.0 wt.7o (Roedder and stated that they represented Weiblen, 1972a). ll. 0.5 mm breccia clast 14303,508.Olivine cataclasite,not found in other missions.Although the Fo""u, plagioclase An,o and spinelzg 0 55 cr 0.068TiO, l.l wt.% textureswere badly altered by shock, some coarse- (Roedder and Weiblen, 1972a).12. Type A spinel-brearinglithic grained relics (- I mm) of unbrecciatedolivine and fragmentsin Apollo 16 aod 17soils. Olivine euhedralto subhedral Cr-spinel indicate an original coarse texture. The and plagioclase as later laths. Trace metal, troilite (?), mesostasis, chemical analysesof the minerals differ considerably and rutile(?). No pyroxene (Weiblen et al , 1974). 13. Type B spinel-bearinglithic fragmentsin Apollo l6 and l7 soils.Olivine in detail (Table l), but the generalfeatures are quite interstitial to plagioclase. Minor spinel, trace metal, troilite (?), similar including the very low CaO and CrzOain the ilmenite and l-8Vo mesostasis (Weiblen el a/., 1974). 14. olivines,low CaO and high but variable AlrO, in the "Spinel-olivine anorthosite" 60618, l-1. Rake sample. Photo- low-Ca pyroxenes,and the low TiO, and moderate micrograph shows 4 mm shocked plagioclase grain Anr" with at- WhereasAnderson reported mi- tachedirregular grains of plagioclase,minor olivine, Mg-Alspinel, CrrO, in the spinels. and pyroxene. Pyroxene also as veins in the large feldspar grain. croprobe analysesof rutile (?) and low-Zr "armalco- Assigned to the Mg-troctolite group becauseof mineral composi- lite" (?), Benceet al. reported high-Mg ilmenite. Two tions (Dowty et al., 1974d). 15. One Apollo l5 and two Luna 20 Fe-rich pyroxenes in73263 were ignored becausethey fragments. Data for spinels and one olivine (Reid, 1972) 16 cannot be in equilibrium with the other minerals,and Histograms of olivine compositions for spinel troctolite fragments presumably represent erratic grains incorporated in Luna 20 and Apollo 16 (Taylor et al , 1973). l'7. 2-4 mm fragments in Al6 soils, denoted Type A Feldspathic Intersertal during brecciation. The mg ratios of all the other Igneous Rocks. Quench, diabasic, and poikilitic textures, with rare olivines and pyroxenesare mostly 0.90-0.92.Ander- evidence of thermal metamorphism. An*-*, Fo 2 74, pyroxene mg son (1973) concluded that the 15445clasts were ig- - 0.2, Mg-Al spinel. Only other mineral data is histogram of neous cumulates formed at moderate pressure(e.9. olivine compositions(Delano et al., 1973). 18. Luna 20 fragment followed by partial recrystallization during with very fine-grained igneous texture. 6570ADr"-,00, l5Vo Fo"o,2Vo 2kbar) Mg-Al spinel and -20Vo interstitial phase. Interpreted as impact granulation at 950 + 50'C and P > 1.3 + 0.5 kbar. meft which lost volatiles (Dowty et al., 1973a). Bence et al . argued that the assemblageof 1070 J V. SMITH AND I. M. STEELE

Al-enstatite, forsteritic olivine, pleonaste,and anor- variety of textures (describedas fine-grained basaltic, thite is stablefrom 3 to 5kbar. Whatever the details, annealed cataclastic, metamorphosed, brecciated, the overall evidence suggests that these cataclastic,and altered) and a wide range of olivine spinel-plagioclase peridotitic clasts originated at and pyroxene compositions.One aberrant specimen sometens of kilometersdepth. Anderson made inter- with Fo., (No. 7) could be interpretedas a cognate esting speculationsabout the opaque minerals and xenolith in a mare basalt.The large troctolitic gran- the minor elementcontent of the silicates,and these ulite 76535 has the same olivine composition as the ideasshould be testedby thorough study of all speci- first type, and indeed might serve as a guide for mens in this group: in particular the relations be- interpretationof all the first group. The secondgroup tween armalcolite and ilmenite need further ex- of fragments can be interpreted as the products of planation in view of the stability data obtained by impact mixing of diverse materials near the lunar Lindsleyet al. (1974a).The pattern of rare earthsin a surface.Whereas the CaO content of olivines of the 15445clast was found to be inconsistentwith coexis- first group is mostly below 0.I weight percentsugges- tence of olivine, orthopyroxene, and Cr-pleonaste tive of plutonic crystallization,that of the second with any known basalticlunar liquid, and Ridley el group is mostly above 0.I weight percent(e.g. Prinz al. (1973) suggestedthat garnet had been presentby et ql., 1973a,Fig. l2) suggestiveof near-surfacecon- analogy with the pattern found for pyropic garnet in ditions. The first group was found only in Apollo 14 terrestrial peridotites. Further work is needed, but and Apollo l7 samples,whereas the second group one should ponder the possible presenceof garnet was strongly representedin Luna 20 samplesas well. peridotite and eclogite on the Moon. For Mg-rich Tentatively we define a group of Mg-troctolites with bulk compositions,pressures over l5kbar and depths unzonedolivines in the range85-91 Fo and a group over 300km would be needed,assuming a temper- of Fe-troctolites with diverse textures and olivines ature over 1000'C (Fig. 2). less magnesianthan Forr. The Apollo l7 fragments (b) Others. From the viewpoint of bulk composi- describedby Stoeseret al. (1974) show a bimodal tion, other Mg-rich specimens could be classified distributionon the histogramof line 8 in Figure 6, mostly as dunitic, troctolitic, and spinel-troctolitic. and may representa mixture of the proposed two Figures6, 7, and l2 show major elementsof olivine, types of troctolites. Further study, especiallyof mi- pyroxene,and spinel (s./.) from thesetypes. nor elements,is desirable.Lines 9, 10, and 1l of Dunitic: Polycrystallineolivine-rich fragmentsoc- Figure 6 are histograms for specimensdescribed as cur in brecciasand soils, and the legendto Figure 6 "anorthosite," and "ANT." Without further study of lists all the specimensfor which chemical composi- textures, mineralogy, and minor elements,it is not tions and sometextural data were given.All are from possibleto interpret thesespecimens in detail, but we Apollo 14 and 17. Specimens1-8 have olivrne com- suspect that most of them are polymict breccias positions in the range Fo.uto Forr, suggestingorigin which have undergone various degrees of meta- from a near-uniform source. Chromite occurs in morphism. Preferring to rely on chemistry rather nearly all these dunitic fragments,and the composi- than on textureand mode,we shallassume that speci- tionsfall in a singleregion on a plot of Mg,Al,Cr, and mens 8, 9, 10,and ll are mixtures of severaltypes of Fe (Fig. l2); furthermore the TiO, contentsare below material. 4 weight percent,which distinguishesthem from most Spinel-troctolitic: The olivine compositions in chromitesof mare basalts.Armalcolite occursin sev- Figure 6 are widely scatteredfrom Fo' to Foro,with eral fragments, and its mg value ranges from 0.6 to the majority from Foro to Foru. Pyroxenesare vol- 0.7 and correlateswith mg of the coexistingolivine umetricallyunimportant and tend to scatterwidely in along the trend of Steele(1974). Specimens 10, ll, composition(crosses in Fig. 7). Spinelsare rich in Mg and 12, and perhaps 9, have olivine compositions and Al and low in Fe and Cr; their pink color is which range to lower Fo values. diagnostic.Rock texturesare often igneous, with a Troctolitic: Polycrystallinefragments rich in pla- distinction between early megacrysts and late gioclaseand olivine occur in brecciasand soils, and groundmass.Most, if not all, of this group can be the legend to Figure 6 lists all specimenswith pub- explained by melting of mixtures of olivine and lished chemical compositions and textural data. plagioclasewhose bulk composition lies in the area There are two principal types, one with non-brec- pqt of Fig. 1, followed by cooling sufficientlyrapid to ciated texturesand olivine compositionsrestricted to allow preservationof the spinel. Impact melting of Foru-., (specimensl-6, 12) and the remainderwith a regolith and partial melting of olivine-plagioclase LUNAR MINERALOGY l07t cumulatesnear the lunar surfaceare the two favored mechanisms(e.g. Agrell et al., 1973; Hodges and Kushiro, 1973;Walker et al., 1973b).The extensive --..*--i literature on specimen 62295 provides an excellent \ '.i guide to the evidenceand ideas.In this specimenare . !:. i two generationsof olivine: an early generationvari- ! ously described as xenocrysts or phenocrystswith _"______.j composition Fo.n_ruand a later generationof micro- crystalsoften skeletaland with composition Foro_rr. At leastsome of the spreadof olivine compositionsin Fo Figure 6 resultsfrom a contrast betweenearly Mg- 200 rich and late Fe-rich crystals,but the total spreadis Frc. 8 Composition ranges of olivines and pyroxenes tn so large that diverse source materials must be in- KREEP-rich rocks. The shading and stippling are qualitative analysesof pyroxenes;by far the volved. Until a thorough study has been made of indicationsof the frequencyof majority occur in the dark region. Data sources:Apollo l2 gray- textures and mineral compositionsof all membersof mottled breccias(Anderson and Smith, l97l); Apollo l2 feldspa- this group, we feel that interpretation is difficult. On thic norites(Brown el al.,l97l);12032,45 lithic fragments(Quaide the whole, we suspectthat some are impact melts, et al., 1971);noritic clastsin 14313breccia (Floran et al., 19'12); while some are the products of rapidly-crystallized 14053and 14073basalts (Gancarz et al., l97l); 14276basalt (Gan- (Steeleand Srnith, 1973);4-10 magmasfrom the lunar interior. Perhapsthe Fe-rich carz et al , 1972\:67'749basalt clast mm Apollo l5 fines with basaltic texture (Powell et al., 1973); ones tend to fall into the first group while the Mg-rich basaltclasts in 15465breccia (Cameron and Delano, 1973);Apollo ones mostly or entirely fall into the secondgroup. l2 coarsefines, Type A crystallinenorite-anorthosites (Wood el al, l9'7l\:Apollo l2 KREEP fragments(Meyer el a/., l97l). Note Lunar specimenswith lower MglFe that many other data are available. In part I, a crude distinction was made between ANT, KREEP, and mare-type specimens. After is obtained it is not certain what is the Surface 1973, lunar petrologists became increasingly dis- distribution of KREEP-rich materials.Texturally, satisfiedwith thesesimple distinctions,though it was KREEP-rich materialsrange from glassesto breccias difficult to establishprecise criteria for a betterclassi- to rare basalts. Most crystalline specimenshave a fication. The following attempt to sort out the miner- mineralogy dominated by plagioclase and low-Ca alogical evidenceis rather subjective. pyroxene (commonly orthopyroxene, but pigeonite (a) High-KREEP material, mostly noritic. Essen- occurs), and can be called noritic. Figure 8 summa- tially all lunar rocks have detectableKREEP com- rizes the composition range of pyroxenesobtained ponents, and the key feature is whether the KREEP from the specimenslisted in the legend.Compilation content is high or low. Unfortunately the KREEP of erratic data from diverse sourcescannot give a components(and related ones including Zr and Ba) statisticallymeaningful frequency distribution. How- occur mainly in the interstitial and late phaseswhich ever, when account is taken of the tendency to list are very fine-grainedand difficult to characterize.ln unusual compositions, most pyroxenes from small and coarse-grainedspecimens, non-random KREEP-rich noritic material lie in the range samplingis a problem. Somebulk analyseswere made En.o-roFsr.-ruWor-u.A fairly strong concentration by broad-beam electron microprobe techniquesfor band extendsto augites- EnnrFs'uWono,while a few which the accuracyof K and P was marginal.Bearing scatteredanalyses extend to Fe-rich compositions. all thesefactors in mind, it is not certain whether the Olivines are uncommon and mostly lie in the range concentrationsof KREEP components show a bi- Fo* to Forr. Plagioclasestend to be sodic compared modal or a continuous distribution. The plot of to other lunar specimens,with a range mostly from weight percent (KrO + P2Ou)us. weight percent AnnoAbroto AnruAbru(Part I, Fig. 2). plagioclaseby Wood (1975,Fig. 3) showsa continu- (b) Low-KREEP noritic material. The legend to ous distribution, though there is a strong tendency Figure 9 summarizes selected specimens. Three for the population density to drop abruptly as the Apollo l7 specimens(Civet Cat clastin72255, rock KREEP content increasespast the line (KrO + PrOd) 78235,and orthopyroxene megacrystsin soils) have : 0.945-0.00893(norm. 7oplagioclase). Orbitaldata minerals with similar compositions: plagioclase show a marked concentration of radioactivity near Atnr-ro, orthopyroxene Enrr-rrWo-r, and augite the Imbrium region, but until greater areal coverage Eno, nrWo..-n.. All are coarse-grained,and norite 1072 J. V. SMITH AND I M. STEELE

D Hd and as lamellaein the Civet Cat clast, as interstitial grainsin 78235,and as a componentof rare KREEP- rich veins in the orthopyroxene megacrysts from soils.Although the bulk KREEP content is low, there are KREEP-rich spots or veins in the specimens. Perhapsthese result from permeation of the norites by minor amounts of KREEP-rich fluids, or erratic concentrationof residualliquids. Specimens5 and 6 in the legend to Figure 9 are En to Fs clastsin breccias,and the wide range of plagioclase BO 60 40 compositions a diverse 5J to Fo suggests origin. (c) Anorthositic. Petrographically, anorthosites should havemore than 90 percentplagioclase, but the terms anorthosite and anorthositic have been used An loosely.Few photomicrographshave been published toAAt I of anorthositic specimens,but we suspectthat most Ftc 9. Compositions of pyroxene,olivine, and plagioclasein are polymict breccias.The 19 specimenslisted in the low-KREEP norite and 67075anorthositic breccia Detailsof spec- legendto Figure l0 probably show the entire rangeof imens:l. CivetCat clast,2.5 cm 1ong,72255,Boulder l, Station2, texturesand compositions.Many specimenscrudely South Massif, Apollo 17. 35Volight streaks of pure plagioclase lumped into the ANT category are rich in feldspar, Anr, ,nOror-, opartly shockedto maskelynite.6590 dark streaksof orthopyroxeneEnr, ,oFsrr_rrWor_owith abundant ilmenite plates in cleavages,rare augite Ena2_nuFs",roWoq6_a6?.Ssmall grains and lamellaein opx, accessorycristobalite, baddeleyite, ilmenite, chromite, metallic iron, troilite, and Nb-rutile (Stoeseret al., 1974) 2. mol. Norite rock 78235with l-5 mm plagioclaseand orthopyroxene, chipped from boulder, Station 8, Apollo l7 Cumulate texture. Heavily shocked 60Vo anorthite Ah* u_"nu, 30Vo Enr" ,oFs,r-roWo,,, l}Vo glassy veins, interstitial augite EnnrFs"Wonu,silica, whitlockite (MgO 3.6 CezOg2.0 wt %), fluo- rapatite, rutile (Cr,O3 3.0 and Nb?), iron (Co 3 Ni 2), chromite, baddeleyite(TiO,4), and troilite (McCallum et al.,l975a;lrvinget al., 1974). 3. Six l-2 mm orthopyroxene crystals Enr"-.,Fs,.-2oWor(AlrO.l.l CrrO, 0.6 wt.Vo)with minor attached plagioclaseAnrr rn, (Fe 0.1 K 0.1 wt.7o),mesostasis, and veins of 4 ot An65 silica mineral, K-Ba feldspar, phosphate, chromite, diopside Enn"Fsn-uWo."-c,troilite, and iron (Ni 3.9 Co 2.8). Probably refatedto 78235(lrving et al., 1974).4. Micronorite fragmentsin | 432I polymict breccia.Granoblastic, hornfelsic and vestigialsub- ophitic textures Unzoned pyroxeneswith total compositionrange Frc. 10. Compositions of pyroxene,olivine and plagioclasein En.rFs,,Wo,to Enu.FsrrWon(TiOr 0 5 Alros 0.6 CrrO.0 3), plagio- anorthositic specimens l. 2 mm fragments of microanorthosite clase inclusions Anru-ru, rare high-Ca pyroxene, ilmenite, iron, 10085-4-10a.Microphenocrysts of anorthite An", u in granulitic zircon and whitlockite (Grieve et al., 1975) 5. Nine recrystallized mosaic of anorthite, olivine For,,rr, clinopyroxene,ilmenite (9 6 noritic fragments with granular pyroxene occurring as clasts in wt.7o MgO), rutile, and trace of troilite and iron (Agrell et al., microbreccias of Apollo l5 soils. Interpreted as monomict 1970) 2 Anorthositic basaltfragment 10059-27.Anorthite Ane64, breccias PlagioclaseAhru pyroxene Enuo-"oFsru,"Wor-rr, oli- minor olivine Forr,, augite Enou rWor" r, pigeonite "u, "Fsru vine Fouu-ss(Cameron et al., 1973'). Enu,nFsrorWo.r, and ilmenite (6 17" MgO) (Keil el al., 1970).3. Anorthosite fragment 10085-07,6.Anorthite Anrr * and or- 78235 has a definite cumulus texture. An obvious thopyroxene En""oFsru ,Won . with lamellae of clinopyroxene EnoroFs,.uWoruo(Reid et al., 1970). 4. Anorthosite fragment sourcewould be a disrupted plutonic complex. The 14258,28,1419-7 Bytownite Anr" oOr, ,, I7o diopside Enn,u range of mg of the pyroxenes is close to that for FsuuWonu o as inclusions and interstitial grains, and llVo pyroxenesin spinel troctolites,but the compositions ilmenite(8.2 wL.VoMgO) (Powell and Weiblen, 1972) 5 Fragment of the two pyroxenesare more highly separatedas D of brecciated anorthosite in 14257,2 coarse fines Anorthite befits a lower temperatureof equilibration. The or- AD"or-rur, minor olivine Fo"o,pigeoniteEnuo uuFsrr,roWo, ,, and trace ilmenite (Klein and Drake, 1972) 6. 23 anorthositic frag- thopyroxenes fall to the extreme Mg-rich end of the ments of 67712soils. Anorthite An**r, olivine FoTa*r,and pyrox- compositionalmaximum for KREEP pyroxenes(Fig. enes with narrow composition ranges at Eno.FsroWon,and 8). The augite is rare, and occurs as discretegrains EnrrFs,rWo,(Simkin et al.,1973).7. Cataclasticanorthosite 61016. LUNAR MINERALOGY 1073

Plagioclase An"u containing blebs of separate pyroxenes and show a similar range of mineral compositions. ", Enn,Fs,uWon,and Enu"FsroWo,Trace silica mineral (Steeleand Probably most ANT specimenshave bulk composi- Smith, 1973) 8. Cataclasticanorthosite 60215 97Voanorthosite tions corresponding to noritic anorthosite and Ahru-"r, orthopyroxeneEn.r-rrFsro_rrWor_r, interstitial to granu- norite. Historically, the GenesisRock, lated feldspar, very rare iron. Clasts of Anru, En..FsrrWo, and anorthositic Cr-spinel (mg O.27cd.65) (Meyer and McCallister,1973). 9. Cata- 15415,is the most interesting,and indeed its texture clastic anorthosite60025 3 mm tabletsof plagioclaseAn"o ,u in has beeninterpreted in differentways. Even on Earth, fine-grained recrystallized matrix. 2Vo Enro-.rFsno-nrWo,,o and the origin of anorthosites is very controversial, En.rWonnFs,,occurring partly as crushed grains in matrix, and though most petrologists probably favor origin by partly as inclusions and recrystallizedgrains. Trace silica and cumulation from a liquid followed by solid-statean- opaques(Walker et al , 1973b).10.Cataclastic anorthosite 60025. 95VoAn""-"n, Enuo ,.Fsn, .rWor-. and Enr, ,rFs,u-r,Wonr_n.mostly nealing which producesmany textural variants. Fig- as discretegrains, trace ilmenite and Cr-spinel (Hodgesand Kush- ure 5 showsseveral textures in the polymict anortho- iro, 1973). I l. Granoblasticanorthosite 606t9 Rake sample In- sitic breccia67075, which was interpreted by Brown homogeneousdistribution of mineralsbut compositionsuniform et al. (1973) as the product of a disrupted layered Ang u-", ,, Foro ,r, Enuu-rrFsrr-r.Wo3 E[aa il- 11, o"Fsro_,uWo.,nu, by plagioclaseinter- menite (MgO 5-8 wt.%), chromire (TiOz 3-5VoMgO 5.67oAl,O3 igneous complex dominated 12-15%) and rutile (Dowty et al., 1974d). l2 Granoblastic spersedwith minor augite(now exsolvedto two pyrox- anorthosite60677 2 mm clast in rake sample breccia.Anr, , ,, enes).In spiteof the problems,we assumethat all ", FoTa7. (Dowty et al., 1974d).13. Nine Apollo l6 ferroan anortho- lunar anorthositeshad an igneousorigin which has site rake with samples cataclastictextures and veins of melt, usu- been obscured to different degreesby brecciation, ally recrystallized.62275 is mostly maskelynite.For brevity, only shock metamorphism,mixing and annealing. composition rangesgiven here.An"" 2 s?o, exceptfor An., u ,r, ln 62275 Rare olivine mostly Fo* u",but Fono_uufor 65789.High- Pyroxene, olivine, ilmenite, Cr-rich spinel, silica and low-Ca pyroxenesas separategrains or intergrowths, with minerals. rutile. iron, and troilite occur erratically, latter most common: most are Enro-roWo^,,and En."-nrWo-4s. and samplingproblems precludeestimates of modes Rare chromite in 3 specimens(TiO" l-3%o MgO l-4% Al,O, and mineral associations.Although incorporation of ll-2l%o) and ilmenite in 3 specimens(MgO 2-5Vo)(Dowty et al., grainscannot be ruled out, we believethat 1974d\. 14. Genesis rock t5415 Polymetamorphic texrure with extraneous evidence of cataclasis and recrystallization. 97-99Vo anorthite all these minerals occur in the anorthositic assem- Afl ,u ,", 2% EnnoFsr"Wor4,trace EnrrFsooWo3,trace silica mineral, blage, though not necessarilyin equilibrium. The trace ilmenite (MgO | .0 MnO 0 6 wt Vo)(James, 1972; Hargraves pyroxenesoccur either as separategrains or inter- and Hollister, 1972;Stewart et al ,1972;-Steele and Smith, l97l; grown lamellae, and the wide composition gap in- Smith and Steele,1974) 15. Cataclasticanorthosite rake sample dicatessubstantial solid-state annealing. Augite pre- 15362. 97Vo Anr", ,r rOro,-o 2Vo Ens"FsrrWonr, 0 57o ", Enu.Fso,Wo,,0.370 ilmenite (Al,Oa 1.6 MnO 0.7 MgO 1.9 wt %), dominates several-fold over low-calcium pyroxene, two grains Al-chromite (AlrO, 9.1 Cr,O,48.8 FeO 35.1 MgO I 2) and there is a wide spread of mg from 0.8 to 0.4. (Dowty et al ,1972a',Nehru et al., 1974) [6 Cataclasticanortho- Olivines occur sporadically,and the range of mg is site 15264,19fragment )907o Ango ee!average 96, orthopyroxene similar to that for the low-Ca pyroxenes. Most mean EnrrFsrrWo3,trace iron and chromite (Mason, 1972). 17. plagioclase are betweenAnrn and Anee, Cataclasticanorthosites from Apollo 16.Abstract reports various compositions textures,pyroxene and plagioclasecompositions, and absenceof but some analysesgo into the bytownite range.The silicamineral. Specimens 64819, 60215, 61016, 65315 and 60015 spinels(Fig. l2) are rich in Fe and Cr. have Enu,,-uu oFs' o s.5Woq5_r o and En' n nnrFsr", ,, n,, the pyroxenesand the significanceof "Woo* The origin of (Dixon and Papike, 1975). l8 Four anorthositic fragments in their mg ratio is controversial. Smith and Steele Apollo l5 soils, one from Luna 20 and three from Apollo 16. (1974: see also Part l, p. 237) suggestedthat the PlagioclaseAttrr-"r, olivine scatteredvalues between Fonu and Forr, high-Ca pyroxene Enoo-n"Fs,o-roWo."-r,and low-Ca pyroxene plagioclaseincorporated substantial Ca(Fe,Mg)SLOa mostlyEnro suFs12-2rWo,u. Notplotted in Fig. l0(Cameronet al., which exsolvedat lower temperatureto give pyroxene 1973).19. Anorthositic polymict breccia67075. Contains clasts of Ca(Fe,Mg)Si2O6and a silicamineral. Compilation of microanorthositewith different amounts of olivine and pyroxene available data (e.g. legend to Fig. l0) shows that (Fig. 5 ) Plagioclaseis An"u irrespectiveof textural state.Olivine "" silica mineralsoccur in only some anorthositicspeci- as isolatedunzoned clasts with bimodal compositionrange Fono-., less abundant than the and Fouo-uuPyroxene as augite, pigeonite, orthopyroxene,and mens and are always much partially invertedpigeonites either as individual grainsor lamellar pyroxenes.Of course,one can suggestthat primary intergrowths up to 40pm across. Composition ranges for clasts olivine combined with exsolvedsilica to give pyrox- and matrix may show clusters.Blue-gray chromites occur as rn- ene. The mg rutio of plagioclaseis lower than that for clusions in olivine or as crushed grains in groundmass, while pyroxene crystallizing from the same liquid, and chromite-ilmenite-iron intergrowthsoccur as small inclusionsin source liquid of pyroxenes. Data combined from Brown et al. (1973), El Goresy Smith and Steelededuced that the et al (1973a), McCallum et al. (1975b), Ghose e/ al (1975) and lunar anorthosites would be more Fe-rich if the Smith and Steete(1974). pyroxeneswere primary rather than secondary.The t074 J V. SMITH AND I M. STEELE modal and textural data on anorthositic speclmens able but low amounts of NarO. The sparsedata on are too poor to give a test, but we now believethat olivine and pyroxenecompositions (Fig. I 1) are scat- the major part of the pyroxeneis primary: the reasons tered and tend to be Fe-rich. The presenceof K-rich are (a) the olivinesin Figure l0 have a similar range feldspar, phosphates,and zircon obviously gives a of mg to the low-Ca pyroxenes in spite of erratic KREEP-like affinity to the granitic and rhyolitic modal variations and (b) the spinelsare very Fe-rich specimens.All workers agree on an origin from (Fig. l2). If this conclusionis valid,the parentliquids highly-differentiatedliquids but the relative roles of of the anorthositescould have been very Fe-rich. progressivefractional crystallization (Ryder et al., Predicted mg values are 0.6-0.2 for extreme crys- 1975),partial remelting,and liquid immiscibility are tal-liquid fractionation (Part I, Fig. 4), and higher controversial. values0.8-0.4 for complete crystal-liquid equilibra- (e) Mare bqsalts. We shall not attempt to review tlon. in detail the mineral data for mare basalts.Because Dowty et al. (1974d) usedthe termferroan anortho- mare basalts are undoubted igneous rocks crystal- site for a "widespread and distinctive lunar rock lized from magmas,the bulk compositionsare mean- type," but we are not convinced that a clear dis- ingful. Readerscould consult Taylor (1975,Chapter tinction has been made from other anorthosites.We Hd deliberately placed their spinel-olivine anorthosite Di 60618,1-lwith our spineltroctolite group becauseof chemical similarities and inconclusivetexture (their Fig I b). They distinguished two granoblastic anorthosites whose ferromagnesian minerals are more magnesianthan those in the ferroan anorthosites. Hubbard et al. (1974) used bulk analysesof minor and trace elements to distinguish between En Fs anorthositic rocks, but the relation to the detailed Fo Fo mineralogyand mineral chemistryis not clear.At this 20 time we prefer not to attempt a subdivision of the r00 80 60 anorthositic specimens,but agree that the present FIc I l. Compositions of pyroxene and olivine in granitic and and rhyolitic fragments in evidencesuggests more than one origin. Becauseof r[yolitic specimens. I Granitic 14306,14321 and 14270 breccias. Several contain cores of the mineralogical evidence for a hybrid origin of K-feldspar and quartz,and one (14306,53)has 0.4 mm grainswith many anorthositicspecimens, we arguethat chemical planar faces Most are glassyor havegranophyric textures. Analy- distinctionsmade from bulk compositionsshould be sis of Fe-augite microphenocrystand K-feldspar (NarO 0.5 CaO less powerful than those made from the individual 0 8 BaO 2.6) in 14306,53granite clast (Anderson et al., 1972).2 (0 from Boulder I, Station 2, minerals. Whatever the future of research on Rhyolitic clasts 5-2 mm) in breccias Apollo l7 Mafic, non-mafic,holocrystalline, and glassytypes Fig. anorthositicspecimens, the tendencyfor mg to be low ' 3 showstotal rangeof pyroxenecompositions in all clasts,but indi- in the pyroxenes,olivines and spinels must be ex- vidual rangesare smaller. K,Ba feldspars(BaO l.l-3.6), plagio- plained. clase,and unique ternary feldsparsnear AnroAbloOrno.Cristobal- (d) Granitic and rhyolitic. Although no largerocks ite, olivine For,-n., ilmenite, troilite, iron, REE-whitlockite, and in 12070with or fragmentsare known, numerous tiny clasts(up to zircon (Ryder et al., 1975) 3 Rhyolitic fragment granophyrictexture, olivine Forr, ferroaugiteand ilmenite(Marvin granitic 2mm) of material occur on the Moon. Most et al., 1971) 4. Polymict KREEP-rich breccia 12013permeated by specimensare quenched melts or fine-scaleinter- "granitic" component. Distinction betweenprimary and second- growths of K-rich feldspar and a silica mineral, but ary mineralswas difficult. Drake et al. (1970,Fig. 5g) showedtwo occasionalclasts (e.g. Anderson et al., 1972,Fig. 3) ranges of pyroxenes associated with granitic areas- K-feldspars show coarsegrains in a equigranulartexture. Because Enoo-uoFsnrrrWoa-16 and Enrr-r.Fs, noWoro-.0, Oruu-roAn, quartz and tridymite, ilmenite,Cr-ulvcispinel, troil- of the 15, small sample size, the number of accessory ite, iron, apatite,whitlockite, and zircon. Seealso Lunatic Asylum minerals is hard to determine,but minor pyroxene, (1970). 5. Clasts of granitic composition in t4303, 14319,14320, olivine, phosphates,zircon, troilite, and iron have and 14321 breccias. Glassy to crystalline K-feldspars (Na,O beenrecorded. Ryder et al. (1975)reported numerous 0.6-1.4 BaO l-3 CaO 0 2-1.0) and plagioclase(Ab 8-27VoOr of granitic clastsin Boulder l, Station 2, Apollo 17, and l-6%) (Roedder and Weiblen, 1972b).6. Granoblastic clast "microgranite" in 14321breccia I X 1.5 mm. Mosaic of equant various described textural variants and unusual ter- grains of K-feldspar (NarO 0.3, CaO 0.3 BaO 1.9),silica mineral, nary feldspars.K-feldspars in granitic specimensusu- minor zircon, apatite,and whitlockite.Also rhyolitic glasses,some ally carry about 1-2 weight percent. BaO and vari- devitrified(Grieve et al ,1975). LUNARMINERALOGY IO75

4) for a useful introduction. Idiosyncrasiesin the core, an olivine-rich mantle, and a plagioclase- cooling history of fragments torn out of different pyroxene-richcrust; levelsof the same lava flow produced extremecom- (b) intensebrecciation of surfacerocks, and complexity in the zoning of the pyroxenes.Many of the plex volcanic,plutonic, and metamorphic activity in details are irrelevant from the viewpoint of overall the upper 200km during the next 0.5Gy; minor lunar petrogenesis.Mare basaltsdefinitely have di- amounts of incoming projectileswere incorporated; verse bulk compositions and mineralogy requiring (c) KREEP materialtended to be dispersed;on the differentsource regions, and partial melting of hybrid largestscale, Fe-rich ANT rocks would tend to over- rocks is a popular but not fully acceptedmodel. lie Mg-rich peridotitic rocks with late oxide-rich Although there are subtle differencesin the chemical cumulatesin the middle, however,on a smallerscale, zoning which will be mentioned in later sections, layered plutons would develop sporadically in the pyroxene compositions tend to lie at the left of the near-surfaceregion; region labeled FETI in Figure 3 of Part I: indeed all I (d) after -4.OGy, consolidation of a lunar crust; mare basaltshave pyroxeneswith mg < -0.7. Oli- igneous activity mostly confined to partial melting vinesoccur only in some mare basaltsand also obey and migration of interstitialliquid; extrusionof mare the boundary at -0.7. Strong zoning occurs to very basalts produced by remelting mixtures of early Fe-rich compositions,sometimes with formation of cumulatesand trapped liquids; the early melts would pyroxferroite. Plagioclasestend to lie in the range tend to becomemixed up in the surfacedebris, and Anru-ru. Spinels are very Fe-rich and zoned from only the later melts would be easily recognizableas chromite to ulvdspinel. KREEP minerals occur at the mare basalts. low concentrationsin the late residua. Some of the above ideas have become widely ac- (J) ANf. To conclude this section, a few words cepted, but the nature of the lunar interior and the are neededon the catch-allacronym for anorthositic, origin of mare basaltsare highly controversial.Many noritic, and troctolitic specimens.Materials with of the mineralogical and petrological data can be these bulk compositionsdominate the non-mare lu- explainedsatisfactorily by early melting of only the nar samples,but most are metamorphosedpolymict outer part of the Moon, followed by migration of a breccias.We have found no simpleway of sorting out molten zone into the interior. All simple models for the mineral propertiesand in despairrefer readersto the origin of mare basaltshave apparently been aban- the summary diagrams in Part I. Most ANT speci- doned, and multi-stagemodels are beingexplored. At mens have ferromagnesianminerals withmg 0.5-0.8. one extreme are models involving remelting of late Pyroxenesdominate over olivines,and low-Ca pyrox- cumulates produced during primordial differenti- enesdominate over high-Ca pyroxenes.Most plagio- ation (e.g. the use of phase-equilibrium data by claseshaveAn)90percent.Afewspecimenscanbe Walkeretat., 1975,toexplainhigh-Ti lunarbasalts classifiedinto the groupsgiven earlier.The remainder from remelting of ilmenite-rich cumulates), and at can be interpreted as a heterogeneousrag-bag of the other extremeare modelsinvolving late melting in material dominated by plagioclaseand low-Ca pyro- the interior of the Moon (e.g.Kesson and Ringwood, xeneprobably producedby mixing up the productsof 1976).We refer readersto the Proceedingsof the 1975 multiple igneous and metamorphic processes.De- Mare Basalt Conference(obtainable from the Lunar tailed study of minor and trace elementsof mineral ScienceInstitute) and of the SeuenthLunar Science and rock clasts should enable progressto be made, Conference.Whatever the problems concerning the but the task is daunting becauseof the loss of evi- inside of the Moon, we believe that many denceby partial melting and solid-statediffusion. petrographicand mineralogicdata can be fitted into any schemeinvolving early differentiation of all or most of the Moon (e.9. 76535 and 15445 from a Origin of rock types plagioclaseperidotite zone, 67075from a disrupted Although very complicated models can be devel- layered pfuton, 62295 from melting of an Mg-rich oped, and indeed must be necessaryfor individual pfagioclase-peridotite zone, 14310 from incomplete breccias,we suggestthat all the data on the rock melting of plagioclase-richdebris). types and mineral compositions can be accom- We shall considerthe petrologic model in detail in modated in the following simple framework: a discussionof the controls on the bulk chemical (a) crystal-liquid differentiation of the entire compositionof the Moon (Smith and Steele,in prep- -4.5Gy Moon about with formation of Fe,S-rich aration). The remainder of this review is devotedto t076 J. V SMITH AND I, M. STEELE

9 . ochondrites B o noritic tr spinel-ol-plog cotoclosite I o/o 1" o Murchison Wl. + X onorihositicspineltroctolitic tn TPXK o o Allende Ti02 . dunitic troctolitic I . al + 4;.! ! .lYc- I 2 a a '-hd; o 32 FeAl20a

TICBC o2 (-' t 0.4 mg 0.6

0.8

Mqlzoo lllgCr20a otomiccr Frc. 12. Major-elementcomposition of spinelsfrom non-mare rocks mg and cr are atomic Mgl(Mg*Fe) and Crl(Cr*Al) respectively. The numbers refer to the specimen sequencein the appropriate figure legend for the rock type (see Fig. 6 for dunitic, troctolitic, spinel-anorthite-orthopyroxene-olivinecataclasite and spineltroctolitic; Fig.9 for noritic; Fig. l0 for anorthositic) Composition ranges are given for the following: M all meteorites(Bunch and Olsen, 1975), EC equilibrated chondrites (Snetsingeret sl., 1967)' UC unequilibratedchondrites (Bunch e/ at.,1967), achondrites(Bunch and Keil, l97l; Lovering, 1975),Murchison C2 meteorite(Fuchs el (Smith at , 1973),Type A and B inclusions,Allende meteorite(Grossman, 1975),TPXK terrestrialperidotite xenoliths in kimberlite and Dawson, 1975),TXAB terrestrialxenoliths in alkali basalts(various sourcesespecially Basu and MacGregor, 1975),TPAC terrestrial podiform depositsin alpine complexes(various sourcesincluding those tistedby Smith and Dawson, 1975),TICBC terrestrialigneous complexeswith basalticcompositions (many sourcesincluding those for tholeiitic compositions,e.g. Bushveld,and thosefor metamor- phosedArcheancomplexes,egFiskenaesset: seeSmithandDawson, lgT5,forreferences),Rhumultrabasiccomplex(Henderson, 1975) In the auxiliary plots for TiOr, trendsfor most terrestrialspecimens are omitted to avoid overlap.Most such specimenshave lessthan I wtVo TiO,.

discussionof the mineral groups, especiallyon the masses,especially those from troublesomemolecular cluesthey provide to petrologicand geochemicaldif- ions. From plots of Li us.Ba and Mg us.Ba, Meyer et ferentiation. ql. recognized distinct compositional ranges for plagioclasesfrom anorthositicand gabbroicrocks (Li Discussionof minerals l-2ppm, Ba 5-25ppm, Mg 150-350ppm),KREEP basalts (Li l0-30, Ba 60-220, Mg 750-1500),mare Feldspars basafts(Li2-14, Ba l5-350, Mg > 1700)and a high- The sectionson feldspar and olivine bring up to Ba group of plagioclase clasts from Apollo 14 date the correspondingsections in Part I. breccias(Li 14-50,Ba 200-750,Mg 200-1100).They Meyer et al. (1974) investigatedthe potential and concluded that these Apollo 14 brecciascould not limitations of a low-resolution ion microprobe for have derived from a pre-Imbrium regolith because analysisof trace elementsin lunar plagioclase.Only they should then have containedmore plagioclaseof the elementsLi, Mg, K, Ti, Sr, and Ba occur in the first type believed to represent the original lunar sufficient concentrations to give statistically valid sig- highlands.[Data for 14063,a fine-grainedclastic rock nals that can be corrected accurately for overlapping with severalcomponents, do not fall with data for the LUNAR MINERALOGY t07'l

other Apollo 14 breccias.lApollo 16 sampleswith literature: e.g. Wenk et al. (1972) found that for metaclastictexture contain many plagioclaseclasts plagioclasein rocks 12051,14053, and 14310the twin with low contentof trace elementseven though the frequencieswere generallyAlbite ) Carlsbad-Albite bulk contentof traceelements is high. Microprobe - Carlsbad - Pericline (or Acline) - Baveno ) analysisof minor and traceelements is a powerful Albite-Ala ) Manebach, but there were minor dif- tool for decipheringthe origin of plagioclaseespe- ferencesfrom rock to rock. ciallyin anorthositicrocks and breccias. Many papersdescribed the domain texturesof lu- Detailed crystal structurerefinement by Smyth nar plagioclases.Wenk et al. (1973) compiled data ( I 975) of anorthite Nao e36Ks.ee5Cae.errFeo.oorAlr.nrrSir.o,obtained by electron microscopy on out-of-stepdo- from troctoliticgranulite 76535 demonstrated l0 per_ mains, and proposed a phasediagram to explain the cent vacancyin the M siteof type(000). This resultis occurrenceof b and c domains and of the Huttenlo- surprisingfor a strongly-annealedspecimen, though cher intergrowth. Subtle features of diffuse X-ray perhapsit is inheritedfrom primarycrystallization of diffractions of both lunar and terrestrial plagioclase a defectanorthite from an earlierigneous melt. Smith were interpreted by Jagodzinski and Korekawa suggestedthat the vacancymight result from ex- (1973,1975)in terms of steppeddomain boundaries. pulsionof Fe upon reductionto metal. Shock and recrystallizationproduced many com- Lofgrenet al. (1974)reproduced experimentally plex texturesin lunar plagioclase(e.g. Heuer et al., the texturesand mineral chemistryof Apollo 15 1974), and high-speed particles left tracks. All these quartz-normativebasalts. Megacryst-groundmasseffects reported in numerous papers testify to the texturecould be produced by steadycooling and does complex history of the lunar regolith with regard to not requiretwo discontinuousperiods of crystalliza- impact of both meteoritesand nuclear particles. tion.All the syntheticplagioclases were rich in minor Ryder et al. (1975)reported feldspars with unusual elementsand werefound from electronmicroprobe ternary compositions from Apollo 17 granitic clasts analysesto containI l-23 percentCao.uAlSirO, for- (seeearlier). Those in holocrystallineclasts are com- mula-unit.Crawford's (1973) description of the zon- positionally uniform in any clast and lie near ing and chemicalvariation of plagioclasein mare AnuoOrnoAbrowhereas those in a rim betweenplagio- basaltsis closelymatched by thatof Bryan( I 974)for clase crystals and granitic groundmass range from quenchedoceanic basalts. Delayed growth of plagio- plagioclasecomposition to AnurOrroAbru.Some clasts clasefrom basalticmelts is a seriousproblem for contain intergrown lamellae of K-feldspar and determinationof stable phase-equilibria(Gibb, plagioclase,probably the result of unmixing. The 1974\. ternary compositionslie well within the stability field Dowty et al. (1974a)studied the occurrenceof of two coexistingfeldspars, and some kind of non- twins in Ca-rich plagioclaseof feldspar-richrocks equilibrium crystallizationor metasomatismseems to attributed to the lunar highlands.Carlsbad and be necessary. Carlsbad-Albitetwins as well asAlbite and pericline Niebuhr et al. (1973)interpreted the ESR spectrum typeswere found in rockswith igneoustexture which of I 5415plagioclase in terms of lines from Mn,+ and .-had not beenseverely brecciated. In cataclasticrocks, from Fe3+in two structural sites.Only I percentof only Albite and Periclinetwins were found, and the iron was trivalent. Dowty et al. suggestedthat original Carlsbad and Carlsbad-Albitetwins had disappearedby breakage Oliuines along twin boundaries.Albite and Periclinetwins Following up the ideason minor elementsin Part I, commonlyshowed features attributed to origin by Steeleand Smith (1975b)measured Al, Ca, Ti, Cr, deformation.Anorthites can undergosuch twinning Mn, and Ni down to the 50ppm level in Mg-rich very easilyand reversiblybecause the Si,Al order olivineswhich may derive from the early lunar crust doesnot blockthe twin process.Carlsbad and Carls- or deeperenvironments. Low Ca contents (0 to 0.1 bad-Albitetwins are not infrequentin granoblastic wt.Va CaO) consistently occur only in olivines from rocks, and probably arise from high-temperaturedunitic and troctolitic rocks, especially breccias, solid-staterecrystallization. In terrestrialplagioclase, whereasolivines from spineltroctolites, mare basalts, suchtwins tend to occurin igneousspecimens and are and ANT materialsfrom Luna 20 haveCaO over 0.1 absentin low and mediumgrades of metamorphism:weight percent(Fig. l3). Although the effectof bulk however,some do occurin thegranulite facies. Many composition is not known, simple analogy with ter- other reportsof plagioclasetwinning occur in the restrial specimensindicates that high-Ca olivinesde- l07E J. V. SMITH AND I. M. STEELE rive from volcanicenvironments and low-Ca ones siveearlier crystallization of Ti-poor olivine. System- from annealedones. The distinctlylowest Ca content atic study of other transition metals(V,Co) and other (0.01wt.7o) is for olivinein 15445peridotitic catacla- minor and trace elementsin olivine should place fur- site(see earlier). ther controls on models for the crystallization of The highestcontents of AlzOrfound in newanaly- olivine. sesof lunarolivines are 0.08 weight percent, and most Burns et al. (1973) concluded that there is no valuesare below 0.04 weight percent. The highvalues unequivocal optical-absorption evidence for sub- in the literature,collected in PartI (Fig.7), shouldbe stitution of Cr2+ in lunar olivines and pyroxenesbe- discardedunless they are confirmed by newanalyses. causeof possibleoverlap of absorption bands with Currentlyit appearsthat seriousrevision of tech- those for Fe'+. Furthermore they concluded that niques(probably with respectto determinationof the there is no reliable evidencefor assigningbands to background)is neededin someelectron microprobe Cr'+. If the chromium in lunar olivinesis indeed all laboratories. trivalent, it will be necessaryto ascribe the high Cr High Cr analyses(up to 0.15 wt.7o)and low Ni content of some lunar olivinesto high crystallization analyses(up to 0.05wt.7o) in Mg-rich olivinescon- temperaturerather than to low oxidation state.The firm earlier analyses.A weak anti-correlationbe- occurrenceof high Cr content in olivine from terres- tweenCr andNi mightbe explainable in termsof the trial komatiite lavas could be explainedin the same oxidationstate being so low that Ni2+ is removed way (e.g. Green, 1975). Detailed study of olivines upon reductionto the metallicstate while Cr'+ is synthesizedfrom Cr-bearing systemsat controlled reducedto the divalentstate which shouldfavor en- temperature,pressure, and oxygen fugacity is badly trance into olivine. However,the data scatterso needed. much, and there are so many indicationsof com- Hewins and Goldstein (1974) studied the Ni and plexitiesin lunarrocks, that suchan ideais presented Co contentsof Ni-Fe metal grains and their olivine with great hesitation.Furthermore a weakpositive host from Apollo 12 mare basalts.Cobalt concentra- correlationindicates coupled substitution between Cr tions in olivine ranged from 300ppm at the core to andAl. Titaniumranges up to 0.1weight percent and 70ppm at the rim. The ColNi distribution between appearsto be higherin olivinesfrom plagioclase-richolivine and metal indicated non-equilibrium, and volcanicrocks; this may resultfrom plagioclasetend- showed that the metal did not crystallizefirst. Per- ing to crystallizefrom laterdifferentiates of the bulk haps late reduction by sulfur loss (seelater), affects Moon, whoseTi contenthas been enriched by exten- the distribution of transition elements.Hewins and

Ave.=O.O75 Sp.fio.t.Auno iOt.-Prinz etol.(19730) sp.Troci. (a)__---zweibren etol.(1974)

tic",?octolific-

o.r o.2 0.3 o.4 CoO in Olivinc(wt.%)

Frc. 13.Calcium in lunar olivines.Steele and Smith (1975b). LUNAR MINERALOGY 1079

Goldstein (1975a) concluded that the olivine and Mg,Fe,Ca on the well-known quadrilateral metal of 76535 troctolitic granulite equilibrated at En-Fs-Di-Hd areuseful, as illustrated by Reidel a/. 900 + l00oc, a range not inconsistent with that (1970), Prinz et al. (1973a),and Steeleand Smith deducedfrom the coexistingpyroxenes. (1973).The useof pyroxene(and olivine)composi- Ghose et al. (1975)observed slight Mg,Fe segrega- tionsto distinguishrock typeswas illustrated earlier. tion in olivine from cataclasticanorthosite 67075,a In general,the mg value decreaseswith progressive rock which containsorthopyroxene whose Mg,Fe or- crystal-liquiddifferentiation, thereby allowing test- der indicatesequilibrium at 640.C. In the olivine, ing of petrologicmodels. The pyroxenegrains from 0.525 + 0.002 Fe atoms enter the Ml site and 0.4g0 coarse-grainedrocks are individuallyhomogeneous the M2 site. exceptfor exsolution,and the bulk composition Huffman et al. (1975) interpretedmagnetic hyper- probablygives a good clue to the original parent fine peaks of Mdssbauerspectra of olivine-rich lunar liquid. Highly metamorphosedbreccias have rela- samplestaken at 5oK in terms of superparamagnetic tively uniformpyroxenes, whereas poorly-metamor- clusters of Fe2+ spins in mirror sites.Annealing at phosedones have diversepyroxenes (e.g. Warner, 1000"Cfor I day reducedthe hyperfinepeak intensity 1972,Fig. l0). Rapidly-crystallizedpyroxenes show by 30 percentin an olivine from a lunar basalt.Ter- complex zoningtrends which are not evenuniform restrial olivines of unspecifiedorigin showed no hy- over the samethin sectionor evenin differenttra- perfine peaks. When detailed time and temperature versesof the samecrystal. The detailsare very com- studies have been made, the hyperfine spectra may plex, and only a brief accountcan be given here. provide anotherthermometer for high temperatures. However,the data are consistentwith the general Olivine clasts in high-grade metamorphosed conclusionthat pyroxenenucleates easily and that brecciasreact with the matrix. Cameron and Fisher the mg value of the first pyroxenedepends fairly (1975)described coronas of pyroxene,ilmenite, and closelyon the bulk mg. Subsequentzoning depends plagioclasearound olivines in Apollo l4 breccias, on many factorsincluding competition with other and attributed them to diffusion of Ti, Mg, and Fe. mineralsfor the componentsin the decliningreser- The distribution of mg betweenolivine and basaltic voir of liquid, melt temperature,viscosity, and cool- liquid was usedby many workers in models for crys- ing rate. tal-liquid fractionation in the Moon. Roeder and The main emphasishas beenon pyroxenesfrom Emslie(1970) found that rng(olivine)/rng(tiquid) was marebasalts which show many textures ranging from independentof temperaturefrom I 150to 1300.C and skeletalcrystals in vitrophyresto megacryst-ground- of oxygen fugacity from l0-0.? to l0-r2 atm. Their massassemblages. Initially it was believedthat the ratio of 0.30 was found to apply approximately to latter textureresulted from early slow crystallization syntheticdata for lunar basaltsexcept that high ilme- (perhapsin a deep-seatedmagma chamber) followed nite content causedsome displacement.For hydrous by rapid crystallization(perhaps upon extrusionat systemsat l5kbar, the ratio rangesfrom 0.52 at the the lunarsurface as a lavaflow). However, Lofgren et iron-wiistite buffer to 0.04 at the magnetite-hematite al. (1974)were able to reproduceall the texturesof buffer (Mysen, 1975).Because the oxidation stateand Apollo l5 quartz-normativebasalts merely by vary- water content of the Moon may havechanged greatly ing the rateof monotoniccooling. Indeed many spec- since its inception, the fractionation betweenolivine imensof lunarbasalts probably came from the same and early liquid may have changedwith time. individuallava flow and differmerely in their close- nessto the surface;of course,initial quenching Pyroxene combined with multiple extrusioncould give complex Pyroxeneis the major mafic phasein the returned coolinghistories. The observationsby Dowty et al. (1973b,1974c) on Apollo l5 basaltsillustrate the complexities. Sector zoning, often associatedwith skeletal growth,produces quite differentchemical trends in the samecrystals (e.g. Hollister et al., l97l; Benceet ql., 197l; Boydand Smith, I 97I ). Delayedcrystalliza- (a) Relation of major elements to rock rype and tion of plagioclase(see earlier) results in a higher crystallization conditions. For classificationor char- contentof CaAlrSizO,in the residualmagma, while acterization of rocks, soils, and breccias,plots of onsetof plagioclasecrystallization results in a sharp J, V. SMITH AND I. M STEELE drop of Ca and Al in the zoning trend of the pyroxene (e.g.Hollister et al.,l97l;Bence et al.,l97l; Dowty et al., 1974c).After about two-thirds of the basaltic liquid has crystallized, the remainder tends to be trapped in pockets,and the composition of the crystallizing pyroxenesis affectedby the idiosyncrasiesof the local environment.Most pyroxenecrystallization occursjust abovethe vault-shapedsolvus, but discon- tailed analysesof Na and K are needed. major, mi- tinuities in some zoning trends from pigeonite to Broad overall correlationsbetween the to those augite compositionssuggest straddling of the solvus. nor and trace elements are rather similar Cr tends to Finally, very fine oscillatory zoning probably results found for terrestrial pyroxenes; e.g. tend to from diffusion effects at the crystal-liquid surface. correlate with Mg, while Al, Ti, and Cr After the initial surge of studiesof pyroxenezoning, most investigatorshave settledfor random analyses which provide a statistical averaging of the above effects.Bence and Papike (1972)gave detailsofzon- ing trends for Apollo ll, 12, 14, 15, and Luna 16 speclmens. Random Ca, Mg, Fe plots of pyroxenesfrom mare of basaltsmostly fall in the region labeledFETI in Part near-absenceof Na and K, the substitutions of the I, Figure 3. There are many dozensof such plots, and Ti, Al, and Cr can be interpreted in terms 1972): those of Dowty et al. (1973b) are good examples. following components (Bence and Papike, and .&f+AlSiAlO6. Although there are slight differencesfrom one text- W+Tto* Alrou, _rll+Cr3+SiAlO., Cr2+SiOt ural type to another,all plots show mg < - 0'7 with The components il+Ti3+AlSiO6 and of pyrox- the major population between0.7 and 0.6. Extreme have also been invoked. Zoning trends by many in- zoning trends ultimately produced the pyroxenoid enes from mare basalts were plotted such pyroxferroite.Relative concentrations of Ca-rich and vestigatorsand interpreted in terms of. factors involving Ca-poor pyroxenesdefinitely vary from one type of as: beginning of plagioclasecrystallization approach mare basalt to another: thus Ti-rich basalts 70215 decreaseof ,41*AlSiAlO. component and and 75035 are strongly dominated by Ca-rich pyro- to the l:2 ratio of Ti to Al for the rll+Tia+AlzO. greater xenes(Longhi et al., 1974)whereas most basaltshave component; presenceof Ti3+ by a TilAl ratio both Ca-poor and Ca-rich pyroxenes. than 0.5; early removal of Cr'+ leaving Cr-depleted of ilmenite or To summarize'.the peridotite suite is characterized liquid; correlation with crystallization in crystallization rate. Ea4Y by high-Mg, low-Ca pyroxenesaccompanied by only spinel; and changes (seeearlier for references)are sporadic rare diopside, the KREEP suite is also pyroxenesfrom 14310 probably becauseof the high dominated by low-Ca pyroxenes,but theseare some- inriched in M'?+AlrSiO. crystallization at high what lessrich in Mg and are accompaniedby rather AlrO, of the bulk rock, but invoked. Most ANT rocks more Ca-rich pyroxenes,the mare basaltsuite is even pressurehas also been with diverseorigins and poorer in Mg and has a greaterproportion of Ca-rich are metamorphosedbreccias of minor elementshas pyroxene, and the ANT conglomeration occupiesa textures,and the significance evaluation of the minor middle region. This generaltrend from Mg-rich, Ca- not been evaluated.Careful pyroxenes from unusual poor to Fe-rich, Ca-medium compositions can be and trace elements in the perhapsthe most noticeable explained by cotectic crystallizationmoving off the Mg-rich rocks is needed: high ALO' content of py- simple olivine-plagioclase-silicaternary towards a feature in Table I is the cataclasites. Ca-rich pyroxene end-member. Of course, partial roxenes in 15445 and'73263 with an ion microprobe melting is involved as well as progressivecrystal- Qualitative areal scans Mn, Li, Na, K, liquid fractionation, metamorphism,and hybridism. showedtrends for Mg, Ca, Ti, Al, (Bence (b) Minor and trace elements. In conformity with and Ba in some pyroxenesfrom mare basalts of some the low content of volatile elementsin the Moon, Na and Autier, 1972). Higher concentrations edgesor at pigeon- and K are much lower than for terrestrialpyroxenes' elementswere found at crystal from liquid trapped in Potassiumis below detectionin routine electron mi- ite-augite interfaces,probably croprobe analyses(- < 0.05 wt.7o)and probably is defects. (c) and site population. Prior below 0.01 weight percentin most samples.Sodium is Exsolution, inuersion LUNAR MINERALOGY l08l

to the Apollo program, the phaserelations and poly- developedonly in pyroxenesannealed at sometens of morphism of the pyroxeneswere only partly under- kilometers depth in rocks ascribableto a peridotitic stood, and major new discoverieswere made on lunar facies. samples. The state of knowledge in 1969 is ex- Coarse exsolution of lunar pyroxenes is uncom- emplified by Special Paper Number Two of the Miner- mon, and most specimenshave exsolution which is so alogical Societyof America, J. J. Papike,Ed. papike fine that X-ray diffraction and electron microscopic et al. (1973)interpreted pyroxene structures from the techniquesare needed.In general for mare basalts, viewpoint of topology and structural linkages. the pyroxenescrystallized mostly as augiteor pigeon- Ross e/ al.(1973) delineatedthe augite-pigeonite ite (often as zoned crystals) which exsolved small solvus at I atm, using pyroxenesfrom mare basalt amounts of pigeoniteand augite,respectively, during l2l2l. Lindsley et al. (1974b)determined the compo- cooling. Pigeonitescrystallize with C2/c symmetry sitions of augiteand hypersthenecoexisting at 810"C and invert to P2r/c symmetry upon cooling below a and l5kbar under hydrothermal conditions, and temperatureranging from 1000'C for Mg-rich speci- found that the composition gap was similar to those mens to 500' for Fe-rich ones (Prewitt et al., l97l). observedfor coexistingaugite and either pigeoniteor Passsagethrough the inversionresults in an antiphase hypersthene in various lunar specimens (aug- domain texture which is revealedby dark-field elec- ite-pigeonite-orthopyroxeneintergrowths in crystals tron micrography (e.g. Champness and Lorimer, from 67075anorthositic breccia,Brown et al., 1973, l97l; Champnesse, al., l97l; Christieet al., l97l McCallum et al., 1975b; inverted pigeonite clasts Lally et al., 1972:Ghose et al.,1972). Very complex from 14082and 14083breccias, Papike and Bence, lamellar structureswere observedwhich require sev- 1972; inverted pigeonite in fractured clast of eraf stagesof exsolution (e.g. Ghose et al., 1973).ln "anorthositic gabbro" in breccia 15459, Takeda, slowly cooled rocks, inversion from pigeoniteto or- 1973).All thesespecimens with coarseintergrowths thopyroxene may occur. Unfortunately it is difficult of augite and low-Ca pyroxene(s)are the result of to determine from textures whether exsolution pro- high-temperature crystallization followed by pro- ceededby spinodal decompositionor heterogeneous longed annealing in the solid state. Inversion from nucleation. Lally et al. (1975)suggested that the fol- pigeonite to orthopyroxene is sporadic. Although lowing experimentaldata yield the best information preciseconditions cannot be determined,the cooling on the cooling rate: differenceofB anglefor exsolved history and other evidence require burial at some pyroxenesin quickly-cooledrocks, textural evolution kilometers depth probably of gravitationally-layered and domain sizein moderately-cooledrocks, and the plutonic bodies.Analogies were drawn with the well- width of diffusionzones and the domain morphology known Skaergaard and Bushveld complexes on in slowly-cooledrocks. Earth. Finally, the distribution of cations between the Low-Ca pyroxenes with Mg-rich compositions M(l) and M(2) sites dependson temperature, and have orthorhombic symmetry (e.g. those in the Mg- can be examined by either Mrissbauer resonance rich specimensdescribed earlier, and in fragments study of the Fe distribution or by X-ray crystalstruc- ascribedto KREEP-rich rocks, Fuchs, l9?l). These ture analysisofthe electrondensity. Interpretation of pyroxenes show no evidenceof inversion from pi- both types of data is difficult for various technical geonite,and presumablycrystallized directly with or- reasons(e.9. Dowty et al., 1972b;Schiirmann and thorhombic symmetry. This is consistent with the Hafner, 1972),but the evidencefor equilibration of occurrenceof low-Ca, Mg-rich pyroxenesin terres- cations between the M(l) and M(2) sites down to trial rocks. However, bronzitesfrom troctolitic gran- temperaturesof around 600'C even for pyroxenes ulite 76535 and norite 78235 (Smyth, 1974a;Steele, from mare basalts seemsconvincing (e.g. Hafner et 1975)have the spacegroup P2rcarather than Pbcain al., l97l; Ghose et al., 1972;Brown and Wechsler, terrestrial bronzites. Crystal-structure analysis by 1973;Yajima and Hafner, 1974).Such low temper- Smyth (oral communication, 1975)showed segrega- atures have appeared surprising in view of the per- tion of Mg and Fe into four sitesinstead of the usual sistenceof pigeonite and the expectedrapid cooling M(l) and M(2) sites.The new spacegroup was also of thin basalticflows. However, the Mg, Fe distribu- found in pyroxenefrom the Steinbachmeteorite. pro- tion cdn be modified by atomic diffusion on a unit- longed subsolidusannealing must be neededto pro- cell scale, whereas polymorphic inversion requires mote the strong ordering of Mg and Fe, aswas infer- major structural changesover much larger distances. red from the textural features of the lunar and Suggestionsof a late episodeof metamorphicheating meteoritic specimens.Perhaps the new spacegroup of basalticspecimens may require further thought. 1082 J, V. SMITH AND I. M. STEELE

might be Some non-pyroxene phases were found as ln- tridymite on the Moon (and in meteorites) effectof water, clusions in pyroxenesand probably result fiom ex- explainedby absenceof the catalytic Rare quartz solution: oriented ilmenite and green spinel in or- but other effectscannot be ruled out. from coarse- thopyroxene clasts from 14321breccia (Gay et al., grains in lunar soils may be relics (1971) suggestedthat 1972); oriented ilmenite lamellae in orthopyroxenes grained granites.Charles et al. at from Apollo KREEP-rich soils(Fuchs, l97l);plate- quartzin granitic specimensimplies crystallization long as the quartz lets of chromite in pyroxene En..FsrrWo, grains of over 400 bars for a dry system,so did not invert from 14258 soils (Haggerty, 1972a); chrome-spinel la- crystallized from a melt and pressurecorresponds mellaeparallelto (001)and perpendicularto (001)in tridymite or cristobalite.This pigeonite from Luna 20 soil (Ghose et al., 1973); to a depth of over 10km. with a seven-repeat oriented lamellaeof ilmenitein a hedenbergitecrystal Pyroxferroite, a pyroxenoid insteadof pyrox- from Luna 20 soil (Haggerty, 1973a).A cluster of silicatechain, commonly crystallized basalts.Its compo- iron needlesin pyroxenefrom 12021basalt was inter- ene from the late residuaof mare becauseof preted by Walter et al. (1971) as the result of late- sition range was not delineatedprecisely pyroxene,but stageintroduction of Fe rather than from exsolution. complex textural relationswith Fe-rich Table 2' (d) Deformation. Disorientation and cracking of is probably Mgo-o.ruFeo.r-o.rCao.r-o.rrSiOe' lunar pyroxenes were explained by several factors: No. 1, gives a representativeanalysis' Crystal-struc- (Burnham, differentialthermal contraction of pigeoniteand aug- ture refinement showed cation disorder the chain ite (Carter et al., l97l), inversionfrom C to P lattice 1971),and the number of tetrahedrain stoichiome- symmetry of pigeonite (Brown et al., l97l), rapid repeat was not controlled by the cation (Lindsley growth (e.g. Carter et a\.,1970),and shockdeforma- try. Pyroxferroite crystallized metastably tion (e.g. Denceet al., 1970;Engelhardt et al., 1970). and Burnham, 1970). also occurs in Careful study is neededto separateall thesefactors as Tranquillityite, Fer(Zr,Y)rTiasisor4, important for emphasizedby Carter et al. (1971). Although some the late residua of mare basaltsand is 2)' The lamellae in pyroxenes are reasonably ascribed to its high content of REE and U (Table 2' No. geome- shock deformation, most arisefrom exsolution' Even crystal structure is unknown, but hexagonal work (Lover- the most casualstudy of brecciasreveals that pyrox- try was indicated in preliminary X-ray to bad- enes are much less affected by shock than plagio- ing et al.,l97l). Tranquillityitebroke down 75055 basalt (El clases:indeed some soil fragments contain plagio- deleyite,ilmenite, and pyroxene in clasewhich was convertedcompletely to maskelynite Goresy et al., 1974). ori- glass whereas the pyroxene was merely fractured. Zircon promisesto provide usefulclues to the pointed out by G' M' Although the possibility should be consideredthat a gin of soils and breccias.As material, plagioclase-richmagma might be produced by filter- Brown, Zr is always high in KREEP-rich acronym pressingthe liquid producedby shock deformationof and indeed he tried to popularize the is Hf plagioclase-pyroxenerocks, no evidence has been KREPZ. The most abundant minor element percent in adduced. which rangesfrom 0.n to about 3 weight some dozen analyses(e.g. Table 2, Analysis 4)' Ter- Other silicates restrial zircons carry similar amounts of Hf, which granitic rocks' The silica mineralscristobalite and tridymite occur tends to increasefrom gabbroic to including Al' P, Ti, Fe, as late crystalsin mare basalts.Several papers in the Various other minor elements the P-rich zircon Proceedingsof the Apollo I I and l2 conferencesde- Mg, and Ca were reported, and is noteworthy (Table 2' scribestacking faults and complex inversiontwinning from 14321polymictbreccia Pink, colorless,and brown grains were of cristobalite.Quartz is very rare in basalts,but is a Analysis 3). types in breccias'A characteristicmineral in granitic fragmentswhere it is found in lunar soils,and several all rock types is ur- usually intergrown with a barian sanidine.Electron detailed study of zircon from it can be used to dis- microprobe analysesreport 0.0n-0.n weight percent gently neededto test whether polymict breccias' Unlike of AlrOs, TiO2, FeO, CaO, NarO, and KrO in cristo- lntangle components of varietiesare unzoned,and balite and tridymite, but the accuracy is unknown terrestrialzircon, the lunar more than one stage of because of problems from secondary fluorescence there is no evidence for and small grain size.On Earth, cristobaliteand tridy- growth. in lunar samplesin mite have not beenfound in ancientrocks becauseof Amphiboles are extremelyrare near-absenceof hy- inversion to quartz. Persistenceof cristobalite and conformity with the absenceor LUNAR MINERALOGY 1083

TlsI-s 2 Somerepresentative analyses of lunarminerals

II t? I5 17

P^O- o,60 4L.2 39.8 sio2 46.0 14.0 32.7 54.5 42.7 39.9 10.4 o.25 0,1 4.2 4.24 0.18 o.6r Tio2 o .4 19,5 0.16 0.1 0,1 27.5 7r.2 7r.3 62.r 1.82 85.3 - ZIOZ L7.A 66.9 oJ. t 32.8 3a,2 94.7 0.70 - Hf02 a.2 J.a 0 .89 a.ot At za3 0,3 1.1 a.2a a.75 16.7 12.3 0.43 )-.9 0 ,15 0,54 0,82 nl a .2+ r.3 r.7 10.3 o,29 o,13 2.65 Feo 46.0 42.0 o.09 0.4 12,2 14.2 7.4+ 16.2 13.4 8.1 22.A 0.45 0.6r. 3.BB o.70 Mr,o 0.8 0.3 o.16 o.z 0.4 r.2 o,08 o.a2 0.04 0.09 0. 30 0.17 A.a2 0.07 <0.01 Meo a.7 o.0 0,02 16.3 9.9 o,o3 8.8 9.2 2.8 4.33 o.14 0.03 r.95 0.10 cao 6.0 1.3 0. 15 2,13 12.0 8.7 3.55 0.4 3.2 4.32 o.16 a.j2 42.2 51.3 Y^0- - 2.8 -nf nf 10.5 r.95 a.57 Nb^o- - 0.3 0.49 7.L Na20 8.69 r.3 2.7 nf 0.09 <0.01 Kzo r.49 a.5 1.8 nf

Total IOo.2 99.2 raa"6 99"9 97.r 97.7 100.5 99.+ 99.7 99.It 98.8 9D.a 10o.3 100. 5 98. E 98.3 99.I

p 1 Pyroxferroite, Table 4, Chao et a1. (t97A). Tranquitlityite, labte 2, lovering et al, (1971), U 7oppn. 3 ZiTca^, micronorite clast in 14321 polynict breccia, Grieve et aI. (1915). 4 Zircon, 12013 poltmicr KlFIpy breccia, Haines et al. (1971), Lr- 100ppn. 5 Amphibole, fOO58 basatt, Gay et al. (1970), t L2 + 0,3. a Anphibote, grains fron 12021 basatt, Dence et aI. (1971), 4 0.4 c1 0.2. 7 Amphibole, breccia fragment ln 14319 fines, l,lason el al. (1972). I Garnei, foose grains fron f2o2l basalt, Trall] et a1. (1970). 9 zlrkefite, basalt 15538, ttrark e-! al. (t973), La2a3 0.29 cezo3 l-.63 !r2o3 0.43 Nd2ca 2.13 sgo3 t.08 Gd203 1.45 Tbao3a.42 ry2o32.a9 Ho2o3 o.38 Fr2o3 1.Lb Tn2o3 o.1r yb2o3 r.24 Lu2ca c.44 uo2 0.14. 10 Arnalcolite, basalt :19 :'09 10022, Table 2, Anderson et ar. (19701, tl "zi-arma1co1iie", ureccia 78{+:, greete (19/4), t2 "ct-zt-ailMrcofitel, breccia 78442, steele (1974). 13 "cr-zr-awrcoliter', breccla'18442, steele (1974). 13 z,r,15102 so11 rragment, liaggerry (r9-i2c\. .V2A3 14 Baddeleyite' Luna 20 grain, Brett et a]. (1973), a.06. f5 Rutile, lnclusions in ilmenite of KRFIlpy Il+162 nicrobreccia, Hlava et 4L (1972), x2a3 O.22. 16 uhitlockite, 14310 basalt, Griffin et aL. (\912), F

drous fluids which have such strong effectson Earth. nite, except for their very low FeO and very high Gay et al. (1970)chipped a small grain from a vug rn MgO (Olsenet al., 1973b).A fascinatingspeculation mare basalt 10058and reported the electron mlcro- is whether amphibole occurred in an early hydrous probe analysisNo. 5 of Table 2. This magnesioarf- Moon which ultimately lost its water, or whether vedsonitecomposition is surprisinglyhigh in Na and amphibolewas alwaysvery sparsein a Moon which low in Ca with respectto other lunar minerals.An was alwaysdry. ion microprobe analysisfor H and an electronmlcro- Existingreports of mica and layer silicatesin lunar probe analysisfor Cl would be interesting.Charles et samplesare too sketchy to have any geneticsignifi- al. (1971) synthesizedthe hydrous equivalent, and cance.The possibilityof meteoriticinput into lunar discussedthe stability relationsfor lunar amphibole. finesmust be borne in mind. Dence et al. (1971) reported aluminotschermakite Loose grains of almandine-richgarnet (Table 2, grains from a bag containing 12021 mare basalt, No. 8) from mare basalt 12021 were interpreted as and suggestedthat the amphibole together with rare constituentsof late differentiates(Traill et al., garnet (see later) crystallizedmetastably. The elec- 1970).Of moregeneral interest is the speculationthat tron microprobe analysis(Table 2, No. 6) shows a garnet precursorwas involved in the generationof low but significantalkalies, low F and Cl, and near- ol-opz-plag cataclasite15445 (Ridley et al., 1973). equal CaO, MgO, and FeO. Mason et al. (1972) For all modelsof the Moon which involvecomposi- observed loose grains in 14163fines with chemical tions dominated by olivine, pyroxene, and plagio- and optical propertiesindicative of Fe-rich richterite clase,garnet must be consideredas a possiblecon- (Table 2, No. 7) and arguedagainst it being a terres- stituent at depth (seeearlier). As the outer part of a trial contaminant. It is surprisingthat further amphi- hot Moon cooled to the present estimated seleno- boles have not turned up in the extensivescanning- therm, the boundary for garnet stability could have electron micrographs of vugs and in mineralogical moved up by 50 to 100km. Many interestingspecula- studiesof lunar fines. Further work is needed.Am- tions can be made on possiblechanges of lunar den- phibolesfrom enstatitechondrites are richteriteswith sity and radius with implications for surface mor- a general similarity to the lunar magnesioarfvedso- phology (e.g. lineaments) and heat content. If the J. V SMITH AND I, M, STEELE lunar interior is dominated by olivine, these effects AB,O4 compositions without invoking substantial could be rather small, but if plagioclasecoexisted Fd+. This is consistentwith the generally reduced with olivineat substantialdepth, the reactionof ol t state of lunar rocks, and indeed Haggerty (1912a) an * gt -| px could be important. The possiblerole of suggestedthat some Cr is divalent in lunar spinels the basalt-eclogitetransition was discussedexten- though firm evidenceis lacking. The major elements sively,e.g. by Ringwood and Essene(1970) and An- are Mg, Al, Ti, Cr3+,and Fd+. Manganeseincreases derson(1975). from 0.0r to 0.r weight percentas Fe increases'Van- The/speculationthat the Moon was derivedfrom a adium (expressedas VrOr) has a similar range but high-temperaturecondensate of the solar nebulawas there are insufficientdata upon which to basecorrela- forcefully advocated by D. L. Anderson (e.5. 1975), tions. Calcium and silicon were reported at the 0.0tt but runs into severe problems from phase-equilib- to 0.r weight percent level (e.g. Nehru et al.' 1974)' rium constraints.Melilite is one of the mineralstypi- and presumably enter the crystal structure, though cal of high-temperaturecondensates, as exemplified analytical error must'be consideredfor small spinel by dkermanite-richones from the Allende meteorite crystals. Most published analysesof lunar spinels (e.g. Grossman, 1975).Eleven uniaxial grains, five omit one or mote minor elements,and a systematic and six respectivelywith optical propertiessimilar to study is badly needed.There is no valid evidencefor those of ikermanite and gehlenite,were reported by deviationof any lunar spinelfrom the ABrO4 compo- Massonet al. (1972)in Apollo 14 fines,but no meli- sition. lites have beenreported in lunar rocks. Further study The following schemefor calculatingend members is needed to verify whether these grains are indeed is basedon that of Buscheet al. (1972)and is unam- melilite, and whether they have compositionslike biguous:(a) calculatethe number of metal atoms for those of the Allende melilites.Whether meliliteoc- 4 oxygens;(b) add Si to Ti, V to Cr' and Ca * Mn to curs in the lunar interior is purelyspeculative. Rele- Fe givrng TI, CR, and FE; (c) use up the TI as vant phaserelations were given by Yoder (1973).Of FerTIOn (all lunar spinelscontain enough FE); (d) course, the chance of hypothetical melilite from the calculate FECRTO. and assigneither excessCR to deep interior of the Moon being found in surface MgCRrOo or excessFE to FEAlzOo; (e) calculate rocks is very small. However, a detailed study of remaining Mg as MgAlrOo; (f) note whether residual lunar fines is desirable to identify the crystals re- Mg or Al occurs;(g) recalculateend membersto 100 ported by Masson el a/. percent. Both Buscheet al. (1972)and Haggerty(1971a) Spinels plotted analyseson a Johnstontype oftrigonal prism rectangular base betweenspinel (sensustricto) (a) Generalchemistry. There are no direct chem- with MgAlrOa (Sp), picrochromite MgCrrOo (Pc), chro- ical analysesfor Fd+ and Fe'+ in lunar spinels,but FeCrrOn (Cm), and hercynite FeAlrO. (Hc), electron microprobe analysescan be calculated to mite and a roof-line betweenFerTiO, (Up) and MgrTiOo' Unfortunately this geometry is unnecessarilycom- uI vijspine I not needed Fe2Tr0o {Fe,Mg) plex becausethe MgrTiOn end-memberis Fel in lunar spinels (except for skeletal crystals in a shocked basalticglass, Sclar et al., 1973),but is im- plied by the method of plotting' The simplestgeome- apex concenlr0ll0nS try is a tetragonal pyramid with FerTiOnat the for morebosolls (Fig. 15). To plot an analysis(a) calculatethe metals 4 oxygens; (b) calculate Up from onorthosiiic to a total of Ti(+Si); (c) add Mn and Ca to Fe and subtract2 (Ti hercynite I = FeAl20a\, cnr0mlTe + Si) to give Fe', then calculatemg' Mg/(Mg + FeCr20a = Fe'); (d) add V to Cr to give Cr', then calculatecr' Cr'/(Al * Cr'); (e) plot point p in either of the two sprnel oiompercent MgA l20a MqCr20a diagramsat the upper left of Figure 14 usingmg' and u', (f) plot point 4 betweenp and the end member oflunar spinelsand composi- Frc. l4 Graphical representations Up using the Up component. Becausedata points tion ranges for spinelsfrom severallunar rock types. ST spinel on a perspectivedrawing of the tetragonal-pyramid troctolitic, SOPC spinel-olivine-plagioclasecataclasite, DNT du- unambiguously nitic, noritic, and troctolitic. SeeFigs. 12, 16, and l7 for details (middle left diagram) are not located LUNAR MINERALOCY

unlessthe line pqUp is drawn, the triple projection els from lunar anorthositesand mare basalts:other (upper left diagram) is usuallybest for detailedstudy. aspectsof the mineralogy of achondrites,of course, When Ti is low, Figure 12 showsan easyapproxima_ resemblethose of lunar minerals.Some composi- : tion. Calculatemg atomic Mg/(Mg * Fe) and cr : tional overlap occurs with lunar spinelsand Cr-rich atomic Cr/(Cr + Al), and plot on a rectanglerather spinels from other meteorites whose compositions than a square in order to fit lantern slides better. were reviewedby Bunch and Olsen (1975).The Al- Then plot TiO, on adjacentrectangles as weight per- lende and Murchison meteoritescarry spinelswhich cent. For specimensrich in FeCrrOn,I weight percent are mostly very rich in Mg and Al, but a few have TiO, correspondsto about 3 mole percent Up. For substantialFe and Cr (e.g.Fuchs el al., 1973:-Gross- spinelsfrom mare basalts,a simple approximation is man, 1975):further study may extend the composi- the line between(Fe, Mg, Mn) (Cr, Al, V)rOn and tion range. (Fe, Mg, Mn)r(Ti, Si)Onin the upper right diagram of Terrestrial spinelscarry substantialFe8+ and tend Figure 14.The positionsof lunar ilmenite(Fe, Mg) to zone to magnetite rather than ulvdspinel. Speci- TiO, and armalcolite(Fe, Mg) TirOuare also shown. mens with low Fe and Ti are summarizedin Figure The large diagram in Figure l4 summarizesthe 12, and most have only a few weight percentof FerO. principalcomposition trends of lunar spinels.Those and TiOr. Spinelsfrom deep-seatedrocks, terrestrial from non-mare rocks lie in two clusters, one near peridotite xenoliths in kimberlite (TPXK) and terres- spinel(s.s.) and one trendingtowards chromite.As trial xenolithsin alkali basalt(TXAB), lie in bandsat Fe and Cr increase,so does the ulvdspinel com- lower mg than for lunar spinels except for Al-rich ponent. Spinelsfrom mare basaltstend to lie in a tube compositionswhere overlap occurs.Podiform depos- betweenmagnesian-aluminian chromite and ulv6spi- its in alpine complexesgive spinels (TPAC) which nel with relatively low population in between. range to lower mg valuesat higher Cr-content. Spi- Crystal-structurerefinements of chromite, ulv6spi- nels from terrestrial igneous complexes of basaltic nel, and a pink Mg, Al-rich spinel showedthat Crs+ composition cover a very wide range,and the factors and Tia+ enter the octahedralB site,that Mg2+ enters which govern their composition are not fully under- the tetrahedralI site,and that Fe2+enters both sites stood. Probably the oxygen fugacity, water content, with a strong preferencefor the I site (Takeda et al., and extent of later metamorphism have important 1974).There are no X-ray data indicating lower than effects.Spinels from layeredcomplexes produced un- isometric symmetry for spinels,but severalauthors der reducing conditions from tholeiitic basalt or ul- (e.g. Haggerty, 1972a) reported anisotropic optical tramafic composition tend to have cr -0.6-0.8 and reflectivityfor Ti-rich spinels. mg -0.2-0.7 (e.g. Irvine, 1974), while those from (b) Spinels from non-mare rocks. Figure 12 metamorphosed Archean complexes tend to have coordinates the data which have already been dis- spinelswith lower cr.The closestfit to low-Ti spinels cussedin relation to petrologicdifferentiation of non- mare rocks. The analysestend to lie in a band from Mg, Al-rich spinelsto Fe, Cr-rich ones with higher concentrationsat the two ends. Titanium tends to 0'o increasewith increasingFe and Cr content but there is a wide scatterwhich should be much greater than experimentalerror. Detailed data for an anorthosite 12 cB and an anorthositic 0.7 xta norite from Apollo l5 are shown I xJ ol2 in Figure 16. mg l-l.'ls 's ,,Qfi1 Rangesfor various types of meteoritic and terres- 0.8o.'r:'- zt:l) trial spinels are shown for comparison.. Chromites .12,18 ,r.'6.i$---- r:_-..-15 from equilibrated .3 chondrites (EC) and unequili- o-cl? brated ones (UC) are richer in FeCrrOa than any lunar spinel, thereby providing a possible criterion for characterizingsuch types MgA l20a 0.1 0.20 0.4 0.8 t.2 of meteoritic spinelsin o/" lunar regolith; howevermetamorphism might change cr wl Ii02 FIc. 15. mg vs. cr and wtVo TiO, for the composition in brecciasor grains broken out of spinels from spinel troctolitic samples(dots) and from spinel-anorthite-orthopyroxene-oli- breccias. Spinels from achondrites have lower vine cataclasites(squares). The numbersare keyedto the legendof FeCrrOa,and their range partly overlapsthat of spin- Fig. 6. 1086 J. V SMITH AND I. M STEELE

low mg for low- from lunar dunitic, noritic, and troctolitic rocks is I 1 from brecciaclasts have unusually should be made of with spinelsfrom the Rhum layeredcomplex whose cr spinels.Detailed investigation from all speci- dominant rock is allivalite (i.e. mostly bytownite and minor and major elementsof spinels in the legend to olivine): however, facile comparison is to be es- mens listed under spinel troctolitic chewedespecially in view of the textural and compo- Figure6. from the liquid in the sitional complexitiesreported by Henderson(1975) Mg,Al-rich spinelcrystallizes (Fig. 1), but should and Cameron(1975): several other papersin Volume system SiOr-anorthite-olivine at equilibrium to 39, Number 6/7, Geochim.Cosmochim' Acta areper- rlact completely with the liquid temperature tinent. produce anorthite * forsterite at lower of spinel moves the bulk With this background one can hazard a model for ipig. Z) unless cumulation system'Walker et al' (1973b) the origin of the lunar spinels.Those from anortho- composition out of the Kushiro (1973) interpreted rock sitic rocks are remarkably high in FeCrzOoand lie at and Hodges and quenched partial melt from a the oppositeend from the spinelsfrom spineltrocto- 62295as the rapidly rock, in which melt lites and the spinel-olivine-plagioclasecataclasites, parental olivine-spinel-anorthite of spinel'olivine, and while thosefrom dunitic, noritic, and troctolitic rocks iome phenocrystsor xenocrysts Metamorphism of this and lie in between.A possibleexplanation is that during plagioclasewere carried' should result in partial or primary differentiationof the Moon, the early liquids rirnitut spinel troctolites the Mg,Al-spinel' Because were low in Fe. Cr, and Ti. Prolonged precipitation complete elimination of than silicates,it should sink of low-Cr, Mg-rich olivinejoined later by plagioclase spinel is much denser thereby causingdepletion of resultedin ultimate build-up of liquids relatively rich rapidly in lunar liquids in Fe, Cr, and Ti. The oxidation statewould be high AlrO' and MgO. from Apollo l4 and other poly- enough to preventformation of substantialCr2+ ' The Mg,Al-rich spinels show reaction rims against sp-ol-pl cataclasitesrepresent the early mineralogy, mict brecciascommonly richer in Ti, Cr, and Mn while the anorthositesrepresent cumulates crystalliz- the matrix; theserims are (Roedderand Weiblen, 1972a)' ing from the later liquids. Spinel troctolites with than the parent spinel rocks- Numerous composi- Mg,nt-rich spinels might result from remelting of (c) Spinelsfrom mare spinels from mare rocks (e'g' early cumulates,while the noritic' dunitic, and troc- tion plots exist for Nehru et al', 1974),but the tolitic spinels would representintermediate stages' Haggirty, 19726,1973a, becauseof incompletecorrela- Mare basalts would result from remelting of late situation is confusing zoning trends of the spinel and co- cumulatesrich in ilmenite and Fe,Ti-rich spinel,and tion betweenthe In general, early. spinels are would precipitate the chromite-ulvcispinel series crystallizing silicates. Mg,Al-chromites which are euhedraland enclosedin olivine when that mineral is present'Late spinel has ulvcispinel composition and occurs either in the groundtutt as separatecrystals or more commonly as rims on earlier spinels.The intermediatehistory varies from one rock to another and from one place rock. The morphology of the seeearlier for comments on garnet). to another in the same whether crystallizationis progres- The composition of spinelsmay provide a clue to zoning dependson or reaction occurs' The the origin of spineltroctolites. Although the available sive or whether resorption on competition with other min- data are scatteredand probably of variableaccuracy' compositiondepends erals. In broad outline, the composition range is shown

from the cataclasites,and differ only in the absenceof pyroxene. On the mg,TiO2 plot of Figure 15, the pyroxene. basaltscontain spinelswhich zone dis- ipinels from all these materialshave low TiOr' The Many mare (e.g' those from Apollo l5 basalts)' oiher spinelstend to lie in broad bands in which cr continuously (1974)discussed the existingideas for the and TiOz increaseasmg decreases'Specimens l0 and Nehru et at. LUNAR MINERALOGY l0E7

Apollo o t56t0 bosoIts ol microgobbro t205t,59 o 15105,5 t2039,3 ol-phyricbosoli () \l

,.- Apolloi5 feldspothic px-phyr ic microgo bbro bosolts . t5388,t0 high-AlKREEPy o l5ll8,1-1 fe ldspoth ic boso lts l5ll6,t-4 pendottte /.15382,7 ' a -r5366,r-r 15385,1-1 :15359,2-2 |5308 onorthositic L\\ n0rite *".'r* t5362.1 nblinosite Sp Up

Flc 16.Composition rangesof spinelsin selectedlunar rocks. Apollo l2 and Apollo l5 data were calculatedfrom atomic fractions given in Buscheet al. (197l\ and Nehru et al. (19'73)using the proceduredescribed in the text. discontinuity and favored early crystallization of Figure l6 shows detailsfor severalunusual rocks. chromite followed by cessation of spinel growth, Basalt 12036,9(upper left) contains three types of while pyroxene continued to crystallizefollowed by spinel, one of which is unusually rich in hercynite resumption of spinelgrowth producing an ulvdspinel (Buscheet al., 1972):the unusual spinelsoccur as 3 rim. Laboratory synthesesshow that the solvusin the grains in early Al-rich pyroxenemelt inclusions.Two FerTiOn-FeCrrOn-FeAlrO4system (Muan et al., of the high-Al, KREEPy basaltsfrom Apollo 15have 1972)crests below l000oC for compositionswith cr spinelsfalling in the generalrange from chromite-rich > 0.6, thereby ruling out exsolutionas a possible to ulv6spinel-richcompositions, but the third one has explanation for spinelsgrowing above I 100"C from Mg,Al-rich spinelslike thosefrom spineltroctolites. lunar basaltic melts. Nehru et ql. (1974\ correlated (d) Miscellaneous. Ulvcispinel in mare basalts is the presenceor absenceof a composition gap with often reduced to ilmenite + iron (e.g. Haggerty, texture and composition of Apollo l5 rake samples, l97lb) and very rarelyto rutile * iron (ElGoresy e/ and for pyroxene-phyric basalts suggestedthat the al.,1972). Ti-rich spinelsfrequently were reducedto gap is smaller for rocks which cooled more slowly. Ti-poor spinel * ilmenite * iron, and Haggerty Dalton and Hollister(1974) found that in 15555ba- (1972a) suggesteda reduction indicator TAC : salt the M'?+CrSiAlO,content of pyroxenedecreased TiOr/(TiO, + AlrO3 f CrrO3). Experimentalstudies and the M'+Al2SiO. content increasedwhen Cr-rich by Taylor et al. (1972) showed that the former reac- spinel crystallized,and that M'?+TiAlrOucontent in- tion can proceed under slightly lessreducing condi- creased sharply when ulvdspinel crystallized. El tions than the latter. The stability curvesrange from Goresy et al. (1975) used spinel zoning as an in- log oxygen fugacity --16 at 1000'C to --18 at dicator of the crystallization relations with olivine, 850'C. McCallister and Taylor (1973) concluded pyroxene,and plagioclasein Apollo l2 and l5 mare from laboratory studiesthat the reductionof ulvcispi- basalts. nel to iron and ilmenite usually involves nucleation 1088 J. V SMITH AND I M. STEELE and growth, and that ilmenite lamellae develop in rich spinel, while the partition of elementsbetween ulvcispinelonly under conditionsslightly more reduc- ilmenite and silicatemay provide useful indicatorsof ing than for equilibrium of all three phases. physical conditions when laboratory calibrations The partition of Zr betweenilmenite and ulv6spi- have been made. nel in the presenceof iron and ZrOz depends on The crystal structuresof lunar and terresterialil- temperature,and favors ilmenite 3-fold near I 100'C menites are essentiallythe same in being based on and 1.5-fold near 850oC (Taylor and McCallister, highly ordered Fd+ and Tio+ (Raymol{ and Wenk, 1972).The highestreported concentrationsfor lunar 197l). Whereas terrestrial ilmenites commonly con- basaltsare 0.4 weight percentZr in ilmenite and 0.2 tain FerOain solid solution, lunar ilmenitesshow no weight percent in ulvdspinel. Observed ratios for definitive evidencefor such substitution, and indeed Apollo l5 rocks indicate that some contain ilmenite there is evidence for minor substitution of TirOr. and ulvcispinelwhich equilibrated down to - 850"C Pavicevicet al. (1972) interpreted Ihe L X-tay spectra whereas others were quenched from near-liquidus of ilmenitesfrom 10047and 12063basalts in terms of temperatures. Fe2+ and Tia+, but small intensity differenceswith El Goresy et al. (1972)suggested that the partition spectral lines for ilmenite from 14053basalt were of Mg between ilmenite and ulvdspinel approaches interpreted in terms of Tis+substitution. Bayet et al. equilibrium more closelyduring subsolidusreduction (1972) interpreted electron microprobe, X-ray and than during initial crystallization, and interpreted Mossbauer data of terrestrial and lunar ilmenites' chemical analysesin terms of phaseequilibrium data and concluded that Fe8+ was present in the former ofJohnson et al. (1971). and Ti3+ in the latter. These and other data on the Spinel lamellaeoccur in ilmenite (seelater). Gay et occurrenceof Ti'+ in lunar ilmenitesshould be aug- al. (1972) reported trails of beadsabout lpm across mented by systematic study to check for possible in olivine from Apollo l4 breccia,and Roedder and correlation with paragenesisand oxidation state. Weiblen (1971) reported rosettesup to lOpm across Electron microprobe analysesof ilmenitesfrom non- in olivine in 12018basalt. Both types of inclusion chondritic meteoritesshowed no excessor deficiency may be spinel, but definitive identification awaits of Ti (Bunch and Keil, 1971).Most lunar ilmenites electron-opticalstudy. appear to be stable towards reduction, but Mao et al. The occurrenceof magnetitein lunar specimensis (1974) reported breakdown to ferropseudobrookite highly controversial.Trace elementdata for bulk lu- and iron, and to rutile and iron in Apollo 17 soils. nar soils indicateincorporation of - I percentcarbo- These reactions indicate very reducing conditions -16 naceouschondrite which should bring in some mag- (e.g. log oxygen fugacity below if the tempera- netite: indeed Wood et al. (1971) found magnetite ture were 1000"C), perhapsresulting from incoming spherulesin a fragment of carbonaceouschondrite in solar protons. 12037fines. However, Jedwab and Herbosch(1970) The main compositional substitution is Mg for did not find any crystalsin Apollo 1l soilswith the Fe2+. In ilmenites from mare basalt, MgO ranges peculiar morphology of magnetitefrom Type I car- from about 2 to 0.2 weight percent,the amount ap- bonaceous meteorites. Numerous studies by mag- parently decreasingfrom early arborescentcrystals to netic, M6ssbauer,and electron spin resonancetech- tiny ones in the late residuum (e.g. Cameron, 1970; niques reported in Volume 3 of the Proceedingsof Smith el al., 1970).llmeniterims on armalcolitecon- both the Fourth and Fifth conferencesare subjectto tain up to 5 weight percentMgO, and the presenceof various interpretations.Magnetite may have beenin- oriented rutile lamellae indicates decomposition of corporated into lunar soils but destroyedby reduc- armalcolite. Steele (1974) found that the ilmenites tion involving solar protons or lunar-derivedreduc- from metamorphosedApollo l7 brecciaswere con- ing materials. sistentlyhigher in MgO (4.6-7.6, average6.5 wt. Vo) than those from ilmenite basalts (0.8-3.5, average 1.8); this confirms the interpretation of Brett et al' Ilmenite (1973) for Luna 20 specimens.The former ilmenites Ilmenite can be found in essentiallyall types of were chemically homogeneousand had apparently lunar specimensranging from scattered grains in equilibrated with the silicatesduring metamorphic anorthositeand ultrabasicmaterials to about 20 per- recrystallization.Ilmenites from unmetamorphosed cent of ilmenite-rich mare basalts. Particularly im- brecciasshowed considerablecompositional scatter, portant are its relationsto armalcolite,rutile, and Ti- and presumably have diverse origins. Nehru et a/. LUNAR MINERALOGY l0E9

(1974)plotted histogramsof MgO for ilmenitesfrom to test this. Becauseof overlap betweenthe TiK6 and Apollo l5 rake samples.The wide scatterof data for V K" lines,V is difficult to detectin Ti-rich minerals, most rocks indicatesdisequilibrium but nevertheless but reported valuesrange up to 0.4 weight percent. MgO of the ilmenitetends to increasewith bulk rzg of Shock deformation of ilmenite producesmultiple the host rock. Ilmenite from brecciated peridotite twinning, which can be seenin ilmenite from many fragments in 73263 soil (Table 1, analysis25) con- lunar breccias.X-ray study of natural and experimen- tains 11.6weight percentMgO. Under equilibrium, tally deformed ilmenite by Minkin and Chao (1971) the principal factor which determinesthe partition of showedthat distortion is greatestin reciprocallattice Mg between ilmenite and coexisting silicates (e.g. directionsperpendicular to a. Sclar (1971) produced pyroxeneand olivine) is temperature.Anderson et c/. sigmoidal bending and microfaulting in experimen- (1972) proposed a thermometer for the distribution tally-shockedilmenites; these features are common in of Mg/Fe betweenilmenite and pyroxene,and gavea ilmenite from lunar breccias.Sclar er al. (1973) ob- tentativecalibration basedon thermochemicalcalcu- tained further evidenceon the twinning producedby lation and some observationson natural minerals. shock deformation of terrestrialilmenite, and found F. C. Bishop (personal communication) confirmed that optically visible twins were produced at 35-75 the generalrelations from laboratory syntheses,but kbar peak pressure, and that initial twinning on found considerableshifts in the proposed relations. {0001}was followedby twinningon ll0il}. This peak The MglFe ratio of the ilmenite increaseswith bulk pressureis much lower than for production of twin- Mg/Fe ratio of the rock and with increasingtemper- ning in pyroxenes and feldspars. Sclar et al. also ature as Mg is transferred from the silicate to the obtained evidenceof shock-induced reduction and ilmenite. The lunar data can be explained satisfac- cation disordering. torily by the new calibrations,and values for equili- Ilmenite frequentlyoccurs with orientedrutile, and bration temperatures in metamorphic lunar rocks textural studiessuggest that most intergrowthsresult will ensue.Interpretation of ilmenite from mare ba- from breakdown of armalcoliteupon falling temper- salts will require careful examination of textures to ature (seenext section) rather than reduction. How- coordinate the crystallization times of the ilmenite ever, Scfar et al. (1973) attributed some occurrences and silicates. to shock reduction. Exsolution of spinel from ilme- The reported rangesof minor elementsin ilmenite nite was reported by several investigators e.g. Dence are MnO - 0.2-0.6 weight percant,Cr2O, - 0-1.0, et al. (1971). Haggerty (1973a) noted ilmenite ex- CaO - 0-0.4, Al,Oa - 0-2.8, ZrO, - 0-0.4, but some solution in pleonastespinel. Fuchs (1971) and Gay et values may be too high becauseof fluorescenceor al. (1972) interpreted oriented ilmenite plates in or- overlap onto other minerals during electron micro- thopyroxene as exsolution products. Ilmenite in- probe analysis.The high values of AlrO3 especially clusions in plagioclasefrom anorthosites were as- should be investigatedin view of the errors found in cribed to exsolution of Ti from plagioclaseby Smith analysesof olivine. No reliablepatterns emerge from and Steele(1974). correlation of the minor elementconcentration with Ilmenite of euhedral shapeoccurs in vugs (e.g. in rock type, but systematicstudy may well prove fruit- Apollo 14 breccias:McKay et al. 1972). Scanning ful when erratic errors are eliminated. electron microscopyof specimensfrom 10072basalt The partitioning of Zr between ilmenite and ul- showed growth steps and globular over-growths of vcispinelwas mentionedearlier, and interpretationof silicate and sulfide (Jedwab, l97l). These observa- observations for several well-known basalts (e.g. tions, and those on other mineral types testify to the 12052 and 14310) indicated cooling rates near the presenceof a widespreadvapor phase in the outer solidus in the region of 100'C per day. Arrhenius el part of the Moon. al. (1971) noted that lunar ilmenites contain 6 to 8 times more Zr than terrestrial ones, but the several Armalcolite, zirkelite and possible related phases proposedexplanations remain untestedby laboratory The identificationof somefine-grained Ti- and Zr- syntheses.El Goresy et al. (1974) found that CrrO, rich minerals in the lunar rocks is uncertainbecause was higher (0.93-1.05wt. 7o) in secondaryilmenite of lack of diffraction data. Peckett et al. (1972) made rims around armalcolite than in primary ilmenite considerableprogress by plotting compositionson a (0.39-0.65)of ilmenite basalt 74243. The CaO con- ZrO"-FeO-TiO, ternary diagram, and, Brown et ul. tent of ilmenite may correlatepositively with that of (1975) extended the data to eight contrastedcom- the bulk rock, but careful measurementsare needed positions. There is no problem with ilmenite 1090 J.V SMITHANDI M STEELE

(Fe,Mg)TiO, describedin the precedingsection. The the type of armalcoliteand the bulk compositionand mineralszircon (ZrSiO.; HfO, 0-3 wt. 7r), baddeley- crystallization path of the host basalt. Perhapsthe ite (ZrOz; FeO 0-7, TiO" 2-3, HfO, 1-3, YrOs 0.2), morphology depends on the nucleation conditions. rutile (TiOr, NbrO, 0.n-7, Fe"O"0-8, CrzO' 0-3, ZrO, Armalcolite shows Mg/Fe zoning which may be cor- 0-2) and tranquillityite Fe2"+(ZrI Y)zTirSi'Ozoare relatablewith the composition of the host liquid and well established(see next section)by electronmicro- the crystallizationtemperature. Taylor and Williams probe and optical methods though systematicstudy (1974\ correlatedthe ColNi ratio of coexistingiron with diffraction techniquesis needed.Zirkelite and with the presenceof armalcolite, and suggestedthat zirconolite are two names assigned to a mineral armalcolite occurs only in rocks depletedin Ni. which occursin the late residuumof mare basaltsand The low-Zr armalcolites (e.g. Table 2, Analy' in the groundmassof KREEP-rich breccias.The ter- sis 10) range from about Mgo..rFeor.Tirou to restrial equivalentshave not been fully established, Mgo ,uFeo.rTirou,and carry minor CrrOr (l-2 wt. Vo) and there may be only one actual mineral for which and AlrOe(l-3 wt. %). Tracesoccur of MnO, VrO., (0'2 the name zirkelitemay have priority. Chemicalanaly- CaO, and ZrOr, probably always weight per- ses of the lunar mineral show ZrOz 30-45, TiO, cent. The trivalent elementsCr and Al should sub- 25-35,CaO 3-ll, FeO 4-ll, and YrO33-ll weight stitute as Ml+TiOu, therebyreducing the atomic frac- percentwith substantialamounts of REE, Th, and U tion of Ti. High contentsof Ti in electronmicroprobe (e.g.Table 2, Analysis 9). Wark et al. (1973)gave the analyses suggest substitution of Ti'+ as the generalformula A2+B24+Ozbased on data for natural Ti;*Ti'+Ou component (Lindsley et al., 1974a)as a and synthetic specimens.X-ray study showed mon- consequenceof the highly reducedstate of mare ba- oclinic symmetry. The above Zr-rich minerals occur salts,perhaps to the extent of 5 to l0 mole percent. fate in the differentiation when Zr and other "in- The alternativeof structuralvacancies is inconsistent compatible" elements have been concentrated by with crystal structureanalyses of lunar and synthetic prior crystallizationof silicates.For details,see Fron- armalcolites (Wechsler et al., (1975). Preliminary del (1975). analysessuggest considerable disorder of Fe, Ti, and The name armalcolitewas usedfor severalkinds of Mg between the two cation sites, but similarities in lunar mineralswith about 70 percentTiO, (Haggerty, ionic radii and X-ray scatteringfactors causedsevere 19'736).The kind occurring as a primary phasein Ti- problemsin estimationof site population. Terrestrial rich mare basalts from Apollo 11 and l7 definitely armalcolitesfrom kimberlitesmay contain some fer- has the pseudobrookite type of structure and con- ric iron (Cameron and Cameron, 1973; Haggerty, tains little Zr. Becauseof the reduced state, it con- 1973c,1975) in responseto a more oxidizing environ- tains no significant Fe'*, and its composition ment. (Fe,Mg)TirOuis obtained from the terrestrialequiva- The stability relations of low-Zr armalcolite were lent FeB+TiOuby substituting Ti + (Fe,Mg) for included in the study by Lindsley et al. (1974a)of the 2Fe+ . This armalcoliteexists as two textural variants join ferropseudobrookite FeTirOu-karrooite (a) in both Apollo 1l and l7 ilmenitebasalts as blue- MgTirOu.Armalcolite is stableat higher temperature, grey discretegrains with or without a mantle of Mg- whereasilmenite solid solution and rutile are stableat rich ilmenite; (b) only in Apollo 17 medium- to lower temperature.Armalcolite beginsto break down coarse-grainedilmenite basalts as tan-colored dis- at ll30'C for mg 1.0 and 1000'C for mg 0'5, and crete grains, either in mutual and sealedbut cuspate breakdown is complete at 800'C for mg 0.5. Lipin grain-boundary contact with ilmenite or in cores and Muan (1974),Lindsley et al. (1974a),and Kesson mantled by ilmenite. Type (b) is frequently twinned. and Lindsley (1975) showed that trF+ ions (Al, Cr Haggerty (1973d) invented the terms "ortho" and and Ti) stabilizearmalcolite to lower temperaturesat "para" which were not acceptedby other workers low pressure(approx. by 100'C for lunar armalco- because (a) single-crystal X-ray data were in- lites). Pressureraises the stability limit for armalco- distinguishable(Smyth, l9'74b),and (b) the rangesof lite with an increaseof 300"C for pure MgTirOu at chemical composition overlap (Papike et al., 1974:. 7.5kbarand about 100-150"Cfor armalcolitescon- Taylor and Williams, 1974).However, the grey vari- taining substantial Al, Cr, and Ti3+ (Kesson and ety apparently tends to have higher MgO and CrrO, Lindsley, 1975).Phase relations for Ti-rich mare ba- and lower FeO than the tan variety (Taylor and Wil- saltswere highly controversial,but the original prob- liams, 1974;El Goresy et al., 1974).The terms grey lems appear to have been largely resolvedby Walker and tan low-Zr armalcolite appear suitable. et al. (1975)and O'Hara and Humphries(1975). Ar- El Goresy et al. (1974)noted a correlation between malcolite is a primary phase on the liquidus only at LUNAR MINERALOGY l09l low pressurefor certain bulk compositions rich in (Steele,1974). Haggerty (1973b) classified these Ti- TiO, component.Papike et al (1974)observed zoning rich mineralsas Type 2 andType3 armalcolitesand in armalcolite from Apollo 17 basalts in which mg distinguished'themchemically by relativelyhigh con- decreasedoutwards in responseto decreasingmg of tentsof Cr, Zr, and Ca for Type 2 andZr for Type 3. the crystallizingbasalt. Because the low-temperature The principal rangesof analysesare as follows limit of armalcolite solid-solution decreasesfaster (Cr,Zr,Ca-richfirst; Zr-rich second):SiO, 01.-0.6, than the temperatureof the basalticliquid, armalco- 0.1-0.3;TiO, 65-71,68-72; ZrO" 4-8,0.5-4; ALO, lite ceasedto crystallizeat the "suicidal" value of ng l-2, 0.5-1.3;Cr,O' 4-l l, l-2; FeO6-13, 8-18; MnO : 0.66. 0.1-1.3,0-0.1; MgO 2-4,7-12, CaO 3.1-3.7, 0.3-0.7; Anderson (1973) reported microprobe analyses REE 0.5-1.5,not detected;Nbrou 0-0.4,0-0.6. Rep- consistentwith low-Zr armalcolite and rutile occur- resentativeanalyses are given in Table2, Nos. I I and ring in ol-sp-opx-plag cataclasite15445, while Bence 12.These ranges can be calculatedapproximately to et al. (1974) reported ilmenite in similar cataclasites M+ Mt+Oi formulaewhen Zra+ is addedto Tin+and from 73263. These occurrencesneed further study accountis takenof substitutionof the trivalentele- becausethe oxidesmay provide limits on crystalliza- ments as Mt+M1+Oi. However,the cation totals tion conditions. For example,the Mg-rich composi- tend to lie about0-0.15 below the idealvalue of 3, tion of the 15445armalcolite(?) would be unstable perhapsbecause some Ti is trivalent.Steele (1974) below I 100'C at Okbarand about 1400.C for 7kbar. obtainedlow oxidetotals especially for theCr,Zr,Ca- Low -Zr armalcolite underwent several post-crys- richspecimens, and Peckett et al.(1972)had a similar tallization reactionsin ilmenite basalts(many papers: problemwhich led themto suggestpresence of light e.g. El Goresy et al., 1974;Haggerty, 1973b).Some elementsparticularly Li; however,other analysts ob- armalcolite reacted with liquid to form ilmenite. tainedgood totals.Perhaps choice of standardsand Some underwent solid-statebreakdown to Mg-rich correctionformulae is responsible.The simplestpro- ifmenite * rutile (e.g.Haggerty, t973b, p.781) while posalis that thesetwo compositionclusters represent some reactedwith ulvdspinelto form ilmenite. Reac- variantsof the pseudobrookitestructure, perhaps tion of ulvcispinelwith iron to form ilmenite and with a strongtendency to formationof an ordered TirOu has also been observed.Haggerty (1973b) de- superstructure.However, armalcolite and other scribedthe reaction of niobian rutile and ilmenite to pseudobrookitesare stable only at high temperature, form a vein of armalcolite in grain 12070,35. whereasthe Zr- and Cr,Zr,Ca-richspecimens appear Usselmanet al. (1975),on the basisof cooling rate to be stableunder metamorphic conditions, perhaps experiments,suggested that armalcolite rimmed by occurringin the range500-900"C. Levy et al. (1972) an oriented intergrowth of equimolar ilmenite f ru- reportedisometric symmetry for reflectionoptics of tile formed at depth, but with a maximum of 100km. the Ca-Cr-Zr variety,instead of the biaxialoptics for Ringwood and Essene(1970) found that low-Zr ar- a pseudobrookitestructure, but this may resultfrom malcolite was not stable in compositions similar to metamictization(unpublished X-ray data of I. M. those of ilmenite basaltsabove pressuresof -4kbar. Steele).Analogies have been drawn betweenthe It seemsfairly safe to conclude thatlow-Zr, Fe-rich Cr,Zr,Ca-richspecimens and zirkelite,but the latter armalcolite is not presentin the lunar mantle though containsmuch moreCa and lessTi. Steele(1974) further experiments are desirable. Walker et al. used invertedcommas for the terms "Zr-" and (1975) concluded that ilmenite rather than armalco- "Cr-Zr-armalcolite",but therules of the IMA Com- lite was in the sourceregion of mare basalts.See later missionon New MineralNames indicate the need for for comments on Mg-rich armalcolites. neutralterms. Peckettet al. (1972)used the term Sclar e, al. (1973) reported armalcolite rich in PhaseX for the Ca-Cr-Zr-rich mineral,while Hag- Fe(mg 0.26) and Alros (7 wt. Vo)as minute skeletal gerty (1973b)used the term Zr-armalcolite for the crystalsin a fragment of shocked basalt. Zr-rich one. Only structuraldata can resolvethe The last group of Ti-rich minerals occurs as tiny problem,but in the meantime,we shall use"Zr-" grains in brecciasfrom Apollo 14, 16, and 17. The and "Ca-Cr-Zr-titanate" asneutral terms. Haggerty maximum size is only about 50pm, and no X-ray (1972c)reported phase Zl in 15102,and discussed its diffraction data have been obtained. Several reports possiblerelation to zirkelite(1973b); see Table 2, emphasizecomplex textural relations(e.g. Haggerty, Analysis13. 1973b;Albee er al., 1973)and the correlation between Both "Zr-" and "Ca-Cr-Zr-titanate" occur in mg of the silicatesand the complex oxides indicates complexintergrowths with otheroxides. Albee et al. that a metamorphic equilibration has occurred (1973)showed the mineralsoccurring within 50pmof J. V, SMITH AND I M. STEELE each other, but not in contact, in metaclasticfrag- ZrO" 0.7 but no REE in rutile from an Apollo 14 ment 61156; Mg-rich ilmenite,Nb-rich rutile and sili- KREEP fragment(Table 2, Analysisl5). Stoeserel cates formed part of the cluster. Haggerty (1973b) al. (1974) reported a mineral from the Civet Cat described complex assemblages including (a) norite clast with TiOz 73 percent, Nb2Ou 20, CrrO" 4, "Ca-Cr-Zr-titanate", Mg-rich ilmenite, chromite, ZrO" 2, and FeO I percent. The highest reported rutile, and iron and (b) "Zr-titanate", ilmenite,chro- content of ZrO" is for a rutile spinel troctolite 65785 mite, and Zl . Other assemblageswere reported which (ZrO, 3.8, NbzO, not found, CrrOs 0.6, FeO 0.7, involved zirkelite, but not the unknown titanates.All Alros 1.1;Dowty et al.,1974b).Probably all Nb-rich theseassemblages, and onesinvolving baddeleyiteare specimensoccur in KREEPy rocks or patches of explainableby a sequenceof solid-statereactions in- rocks, but the factors which govern the other ele- volving Ti-Cr-Zr-Fe-Mg oxides affected by changes ments are unknown. The rutile(?) in the 15445 of temperatureand oxidation state.Steele (1974) de- sp-ol-opx-plag cataclasitecontained 1.6weight per- liberatelyselected some Apollo l7 metabrecciaswith cent NbzOu, but only trivial amounts of other ele- narrow compositionranges of silicates.These showed ments.Jedwab (1973)reported severaltypes of rutile a regular partition of zg betweentitanate and olivine in lunar soils including a blue rutile with no detect- with Ko (oxide/olivine) -0.3 for "Zr-titanate" and able N b andZr. The blue rutile occurredas an 0.5mm -0.1 for "Ca-Cr-Zr-titanate." Ilmeniteshad similar aggregateof micrometer grains, and may have re- Ko to the latter. Thesepartition data may be relevant sultedf,rom condensation. to the possiblecoexistence of complex Ti-oxides and Corundum aggregatesfrom 10084fines were in- Mg-rich silicatesin the lower crust or upper mantle, trepreted as condensatesof impact-derived vapor but the effectof pressuremust be evaluated. (Kleinmann and Ramdohr, 1971).A 200pm crystal The titanates carry a substantial amount of rare was found in 14163fines by Christophe-Michel-Levy elementsincluding up to 0.6 weight percent Nbrou. et al. (19'12),but no corundum has been reported in The "Ca-Cr-Zr-titanates" are characterized by lunar rocks. about 1 weight percent total REE. Other minerals such as rutile and apatite, however, contain greater Apalite and whitlockite quantitiesof one or more of theserare elements. Whitlockite -Car(POn), is apparentlymore abundant than apatite -Cau(POn)'(F'Cl) in lunar rocks' Baddeleyite,corundum, and rutile Both phosphatesoccur in the high-KREEP rocks or Thesethree oxides are much lessabundant than the in KREEP-rich areas of low-KREEP rocks, some- multiple oxides. times as intergrowths readily characterizedby cath- Baddeleyite(ZrOr) is found in the KREEPy assem- odoluminescence. Particularly beautiful are the blages. Although there are no systematic studies, scanning electron micrographs of vugs containing scattered electron microprobe analyses reported hexagonalprisms of apatite or hexagonaltablets of HfO, l-3 weight percent,FeO 0-7, and TiO, l-8 whitlockite(e.g.McKay etal.,1972). Manybreccias weight percent (e.g.Table 2, Analysis l4). Brown et and lunar rocks contain veins and vugs lined with al. (1975) reported from an Apollo l7 basalt a phosphates,silicates, ilmenite, and iron;these testify Ti-Fe-baddeleyite with YrO, 1.8weight percentand to ubiquitous vapors penetratingthe surfacerocks. HfO, 0.8. Prewitt and Rothbard (1975)compared terrestrial, Rutile (TiOr) generallyoccurs in associationwith lunar, and meteoritic whitlockites by X-ray and ilmenite. Lamellae parallel to {0112} of ilmenite in- chemical methods. Whitlockite from the Estacado dicate exsolution, whereasirregular intergrowths of meteorite yielded a crystal structure with one phos- Mg,Fe-rich rutile and Mg-ilmenite indicate break- phate tetrahedroninverted with respectto the crystal down of armalcolite (Haggerty, 1973b).Primary ru- structure of a terrestrial whitlockite: the ideal for- tile tends to occur as twinned, randomly-oriented, mula is CarrMgzNazP'rOu..A terrestrialwhitlockite euhedral crystals associatedwith ilmenite. Some from the Palermo quarry had the formula analysesof primary rutile are rich in Nb, and Marvin Carrr(Mg,Fe)sH'.6P14O86, while synthetic ones could (1971)reported NbrO,6.4 weightpercent, Cr"Or3.2, be made either with or without hydrogen Ta"OuO.2,VrOs 0.4, LarOs0.4,and CerOs0.8 in rutile (CarrMgrHrP,rOu., CarsMgrPrrOuu).Unfortunately grains from microbreccia fragment 12070,35,while Prewitt and Rothbard could not locate a lunar Hlava et al. (1972) reported Nbrou 7.1, CrrOe 2.6, whitlockite large enough for single-crystal X-ray LUNAR MINERALOGY 1093 study.About 80electron microprobe analyses of lu- only one (farringtonite) has been observedin lunar narwhitlockites record up to 25elements with weight rocks so far. concentrationsup to two atomicpercent (e.9. Table 2, Analysis whitlockite l6). Lunar may haveinverted Suffides, carbide, phosphide and metals from a high-temperaturepolymorph (Griffin et al., 1972),which would tolerateexcess Ca and extensive The detailedreview by Frondel (1975)emphasized substitutionof REE. Prewittand Rothbard(1975) the crystallographicand mineralogic aspectsof these believedthat the old namemerrillite should be used minerals and allows us to concentrateon petrologic for the meteoriticmineral, and that a new name and geochemical features. Many papers discussed wouldbe neededfor the lunarmineral when detailed whether the minerals have a lunar, meteoritic, or X-ray work had given a proper characterization.In hybrid origin, but simpleanswers cannot be expected the meantime,however, the namewhitlockite is re- becauseof complex effectsproduced by shock and tainedfor the lunarmineral. Columns l6 and l7 in thermal metamorphism and by melting of polymict Table 2 containanalyses of whitlockiteand apatite breccias. Furthermore the bulk Moon was itself from 143l0 basalt,for whichspecial care was taken probably accretedfrom early bodies,and many types with standardization.Fluorine and chlorineare al- of such bodies were available, not necessarilythe wayslow in lunar whitlockites;SiO, is consistently same as those now representedby meteoritesin mu- reported at the 0.n percent level or greater; FeO seums. The extensiveevidence for precipitation of rangesfrom 0.r to 6 weightpercent, while MgO tends iron and troilite in vugs,for reduction of Fe'+ to Feo to vary inverselyfrom 4 to 0.n weightpercent; YrO, by sulfur volatilization and carbon oxidation, and for rangesfrom I to 3 percent;CezOs and NdrO, occur redox conditions allowing movement of phosphorus up to 3 percent,while other REE occurat theO.n to I between oxidized minerals (phosphateand silicate) percentlevel; NarO rangesfrom 0.0nto 0.6 weight and metallic phases(schreibersite and iron minerals) percent;SrO usuallywas not reported,but 1 percent adds to the problems. However, considerableprog- in 15475whitlockite (Brown et al., 1972).Probably ress has been made in establishingpossible criteria the compositionof lunar whitlockitedepends on lo- for distinguishinglunar from meteoritic phases,and cal idiosyncrasiesduring the crystallizationof late at leastsome specimenscan be assignedwith consid- residua,and themg ratioflor example may be a guide erable confidence.In addition, various texturesand to the degreeof fractionationduring such crystalliza- elementpartitionings are usefulas indicatorsof meta- tion. Vapor-depositedwhitlockite may havea differ- morphic equilibration. An understandingof the me- ent compositionthan whitlockitecrystallized from tallic minerals and alloys is important for inter- liquid: in particular,the REE may be absentin the pretation of magneticphenomena. former. (a) Sulfides. The principal sulfide is troilite, essen- Some 40 electronmicroprobe analyses of lunar tially FeS. In mare basalts,troilite can occur with apatitesrecord 0.0n-2 weight percent Cl and -2.5-4 blebs of iron mineral, and Skinner (1970)interpreted percentF, and when accountis taken of the high troilite-iron intergrowthsin 100'72basalt as products error for F thereis no needto invokethe presenceof of an immiscible liquid which separatedat I 140" C OH. However,ion microprobeanalyses for H would from the silicatemagma after substantialcrystalliza- be veryinteresting to checkfor the relativeavailabil- tion of ilmenite and pyroxene.In Apollo l2 basalts, ity of Cl,F, andOH. Apatitesfrom recrystallizedsoils troilite commonly occurs with ilmenite and spinel exposedto solarprotons might incorporatesignifi- rather than with metallic iron, but does occur inter- cant OH, and vapor-depositedapatite might also grown with iron as a late product of crystallization havea distinctivecomposition. The rangesof other (e.g. Taylor et al., 1971). Probably its occurrence elementsare: SiO, 0.n-2 weightpercent indicating depends on a combination of factors including the substantialreplacement of P by Si; FeO0.0n-2; MgO bulk S content and the state of oxidation. Troilite is 0.0n-0.n:Y zOe0.n-2i CerO, 0.0n-3; NdrO3 0.0r-1, U one of the minerals occurring in cracks in norite 20-1000ppm. Thehigh U contentsmight permit age 78235 (Irving et al., 1974) and in the mosaic assem- determinationby ion microprobetechniques. blagesinterpreted as the final products of crystalliza- The relativeoccurrence of phosphateand phos- tion of troctolitic granulite 76535 (Gooley et al., phide(e.g. schreibersite) depends on the redoxcondi- 1974), to mention just two occurrencesin coarse- tions (next section).Fuchs (1968) recorded several grained rocks. Indeed troilite seemsto be ubiquitous other types of phosphatesin meteorites,of which in lunar rocks (except perhaps for some anortho- J, V SM]TH AND I M. STEELE sites). This is not surprising when considerationis Moon model in Figure 3 and the possible seleno- given to the factors governing the migration of S on therm in Figure 2, S could existas an Fe, FeSeutectic the Moon. liquid at depths greater than 300km in the present During primary differentiationof the Moon, a liq- Moon and at evenshallower depths before the Moon uid rich in Fe, FeS,and the metal- and sulfur-seeking cooleddown. elementswould form wherever the temperature ex- Troilite is a significantconstituent of many typesof ceededabout 1000' C (Fig. 2), and its high density meteorites,but the contribution of impact debris to would cause it to sink. Coexisting silicate liquids the S contentof lunar surfacerocks, especially impact should contain sulfur at the level of at least 0.0n breccias,is hard to evaluate becauseof the easeof weight percent by analogy with data for terrestrial volatilization of S. About 5 percent(range I to 1Vo)of rocks(summarized by Bishopet al.,1975);indeed the coarse metal-rich particles from Apollo 14 and 16 S content of mare basaltsindicates 0.n percent(Brett, soilscontain sulfide (Goldstein and Axon, 1973).Al- personalcommunication). Crystallizationcould lead though the Co and Ni contentsof metallic iron may to rocks relativelyricher or poorer in S dependingon not distinguishrigorously betweenmeteoritic and lu- the efficacyof differentiation.Particularly important nar sources,the wide range of Co and Ni contentsof for interpretation of the sulfide and iron minerals is metallic iron mineral associatedwith sulfidein these the extent of post-crystallizationmigration of S and coarse particles indicates both lunar and meteoritic Fe, especiallythe escapeof S vapor and associated sources.Spherules, 50pm to 5mm across,showed S reduction of Fe2+ into Feo. Gibson and Moore contents from 0.05 to 26 weight percent, and had (1973) noted a negative correlation between S and complex textureswhich were reproducedexperimen- Fe" (not total iron) in Apollo l5 and l7 basalts,and tally by rapid cooling of drops of synthetic liquid Brett (1975)found similarcorrelations for Apollo l2 (Blau and Goldstein,1975). These spherules occur in basalts.The incrementalratio AS/AFe' was closeto most lunar soils,and wereinterpreted as the products but substantially less than the theoretical ratio of of shock melting of metallic and non-metallic mate- unity for atom-for-atom reductionof Fe2+in basaltic rial, probably of both lunar and meteoriticorigin. melt as follows: Synthetic compositions rich in both S and P appar- S'?-(melt)* Fe2+(melt): S' (vapor) f Feo (crys- ently yielded two immiscible liquids, one of which tal). Disproportion of troilite to iron mineral * S crystallized to troilite with small inclusions of would also require a unit ratio. Brett supported the nickel-iron while the other crystallizedto nickel-iron suggestionby Gibson and Moore that the observed grains surrounded by schreibersite(Fe,Ni)sP and a correlations resulted principally from outgassingof few blebs of troilite: rare lunar fragments showed S, but the ultimate fate of the S is unknown. Sato e/ evidenceof similar textures.Brett (1975) observeda al. (1973) made detailed measurementsof the redox strong positive correlation between S and Feo (zot conditions of lunar basalts and argued from ther- total iron) in non-mare rocks, and suggestedthat mochemical calculationsand the expectedpresence meteoritic material with approximately l0 times Feo of C in the Moon that reduction of lunar rocks by to S had been incorporated. conversionof carbon or carbideto carbon monoxide Troilite from a vug was found to have a super- would be a secondplausible mechanism at the lunar structure with a and c doubled over those for the surface. Probably more than one processoccurs in simpleNiAs structuretype (Evans,1970). the reduction of lunar rocks, but the extensiveevi- Electron microprobe analysesof minor elementsin denceof migration of S in lunar surfacerocks and the troilite are questionablebecause of problems from greater abundance of S than C suggestthat S vol- secondary fluorescenceand overlap into inclusions. atization plays the stronger role in determining the Publisheddata show Mg 0.02-0.13weight percent,Si presentredox state.Gibson and Moore (1973)found 0-0.8, P not detected,Ca 0-0.6, Ti 0-0.7, Cr 0-0.1, that 12 to 30 percentof the S and C in lunar breccias Mn 0-0.4 but mostly <0.05, Co 0-0.3, Ni 0-6.3, Cu could be volatilizedunder vacuum in I day at 750" C 0-0.4. High values of Co (0.3) and Ni(6.3) were and almostall at I100' C. The extensiveobservations found only for troilite occurring with iron in a glob- of troilite in vugs(e.g. in Apollo 14 breccias;McKay ule in 12001fines (Champnesset al., l97l), and all et al., 1972) testify to the migration of S-bearing other values are 0-0.1. High values of Cu were re- vapor. On an even larger scale,the migration of S in ported only for troilite coexistingwith copper metal the lunar mantle and especiallyfrom the hypothetical in 10045basalt (Simpson and Bowie, 1970). High lunar core must be considered seriously. For the valuesof Mg, Si, Ca, and Mn should be rechecked, LUNAR MINERALOGY 1095 but consistentlyhigh values of Si and Ca were ob- derived from iron meteorites.Electron microprobe tained for troilite from 12051basalt by Keil et al. analyses showed Ni 0.8-1.7 weight percent, Co (1971). <0.02-0.25, P 0.03-0.08 (Goldstein et al., 1972;El The Ti content of troilite is greater(up to 0.5 wt.%) Goresy et al., l973a,b; Goldstein and Axon, 19731' when it coexists with ilmenite or ulvcispinelthan Taylor et al., 1973b). Carbide from a melted frag- when it coexists with metallic iron (0.0n) in lunar ment of iron meteoritehad 6 percentNi and up to 8 basalts,and although fluorescenceerror cannot be percent P (Goldstein et al., l97O). Carbon from the ruled out, exchangereactions may be responsible(El solar wind should tend to reach a steady level on Goresy et al., l97l; El Goresy et al., 1972;Taylor et mineral surfacesas proton stripping balancesthe in- al., l97l). Troilite annealed with iron, ulvdspinel, coming carbon. Meteorite impact should result in and vapor at 700 to 1000oC showed systematicre- transfer of some carbon from surfaceto interior of duction of Ti content from 0.8 weight percent at grains, while some could be converted to gaseous 1000" C to 0.3 at 700o C, while the Ti content of speciesespecially carbon monoxide. Sputtering and metallic iron dropped from 0.8 to 0.4 (Taylor et al., erosion by micrometeoritesshould changethe grain 1973a).Observed Ti contentsof lunar troilite coexist- size. Many papers (e.g. Moore et al., 1974) have ing with ilmenite could be interpretedas the result of reported the C and S contents of lunar specimens. metamorphism to temperaturesfrom about 900 to Pilfinger et al. (1974) found that the content of iron substantiallybelow 700" C. carbide in lunar fines (as measuredby evolution of Mackinawite was tentatively identified from its CDo upon dissolutionby deuteratedacid) is propor- strong bireflectanceas tiny spots associatedwith tional to the amount of fine-grained((lpm) iron troilite from basalts 12018and 12063(El Goresy et metal measuredby ferromagneticresonance spectros- al., l97l; Taylor et al., l97l). Subsolidus reaction copy. Both the carbide and iron metal when normal- below 150' C is indicated. Taylor and Williams ized to the solar wind exposureare proportional to (1973)reported opticaland chemicalidentification of the bulk iron content of the fines,which indicatesthat chalcopyrite and cubanite along cracks and grain carbide synthesisoccurred only with Fd+ reducedto boundaries of troilite in 12021 basalt. Both Feo by the solarwind. In general,lunar carbidetends Cu-Fe-sulfides may have exsolved at low temper- to be higher in the fines, and only a small part is of ature from Cu-bearingtroilite, and both contain 0.9 meteontlcongln. weight percentCo. Clanton et al. (1975)reported that Schreibersite,(Fe,Ni)rP, is important becauseit in vugs of 76015breccia the order of crystallization forms only under fairly reduced conditions, and its from vapor was silicate, ilmenite, Ni-Fe mineral, relationshipto phosphateshould placea limit on the troilite, chalcopyrite (Cr 0.5 wt.Vo),pentlandite (Fe oxygen fugacity-temperaturerelation. Furthermore 35.5Ni 29.1Cu 0.3 Cr 0.2),and chromite(?). Sphal- it is a common constituent of meteorites,especially erite in "rusty" rock 66095contained 28 mole percent irons, and its occurrencein lunar rocks might provide FeS and occurred as reaction rims in and around a measure of meteoritic contamination. Unfortu- troilite grains (El Goresy et al., l973a,b) or as indi- nately solid-statereactions cause problems. The dis- vidual grainsattached to or enclosedby troilite grains tribution of Ni between schreibersiteand metallic (Taylor et al., 1973b).The latter authors argued that iron provides a thermometer. the uniform composition and the latter texture were The Type I enstatite meteorite fragment in an inconsistentwith the idea of alteration by aZn-bear- Apollo l5 soil (Haggerty, 1972d)contains I percent ing solution proposed by the former authors. The schreibersitetogether with kamacite,troilite, and ni- identification of all other speciesof lunar sulfidesis ningerite (Mg,Fe,Mn,Cr)S. A fragment from Apollo tentative. I I finescontaining iron mineral with Neumann lines, (b) Cohenite and schreibersite. The carbon chem- schreibersite,cohenite, and troilite is probably mete- istry of lunar samplesis extremelycomplex because oritic, and Widmanstiitten lines were probably re- of the input of carbon from the solar wind and be- moved by shock (Frondel et al., 1970).A fragment cause reaction of carbon with iron in silicatesin- from breccia 10046 contained symplectic inter- volves oxidation-reduction processes. Grains of growths indicative of crystallization from a P-rich cohenitelarge enoughto be visibleby optical micros- eutectic (Goldstein et al., 1970). A particle from copy were reported only in a few fragments in fines Apollo l2 finescontained coarse kamacite grains plus and 66095 "rusty" breccia. The coexistingminerals an intergrowth of small kamacitegrains set in schrei- and textures indicate that the host frasments were bersite matrix, while another particle showed den- r0% J. V, SMITH AND I. M. STEELE drites of kamacite and taenite surrounded by schrei- Figs.l7a and b) showlow P in the schreibersiteand bersiteand troilite (Goldstein and Yakowitz, l97l). high P in metalliciron (exceptfor ltem 8, Fig. l7b); For the latter particle, the Ni distribution between thesecompositions are close to thosefor the 1000"C the taenite,kamacite, and schreibersite(Fig. l7) in- isothermin the syntheticFe-Ni-P system(Doan and dicated equilibration to 500-600" C basedon experi- Goldstein,1970). The insertto Figurel7b showsthe mental data of Doan and Goldstein (1970) for the Fe-Ni-P system.Metamorphism in an ejectablanket followed by mechanicalremoval would be a possible explanationof the equilibration to relativelylow temperature.These and other data (e.g. Goldstein et al., 1972) indicate that iron meteorites were impacting the lunar surface and that their debris was under- going complex reactionsincluding melting and metamorphism. The stateof oxidation of some iron meteoriteswas estimated by Olsen and Fuchs (1967) using calcu- lations for the assemblage chlorapatite-metallic wl. o/ iron-pyroxene. These showed that schreibersiteis /o stable for conditions more reducing than the iron P wi.istitebuffer, while phosphate is stable for more oxidizing conditions. McKay et. al. (1973) heated synthetic silicate glass doped with Car(PO4)2,NiO, and CorO, in a controlled stream of H, and CO2. Charges quenched from 1250' at oxygen fugacities o devitrifiedgloss ra lower than l0-13-10 atm containedspherical inter- . mesosiosis- rich growths of schreibersiteand metallic iron interpreted as the product of immiscible Fe-P liquid. The oxidation conditions in lunar rocks probably range over severalorders of magnitude around the iron-wiistite uto20304050 buffer, and straddle the phosphide-phosphateequi- librium. Because of the possibility of growth of wt.% Ni schreibersite from calcium phosphates, P.idley et al. Ftc. 17.Partition of P andNi betweencoexisting metallic iron (1972)argued that presenceof schreibersitewas not a and schreibersite.(a) Apollo l6 rakesamples, Gooley et al. (1973). sufficientcriterion for meteoritic contamination. Lu- (b) Other data:l-4 Browne, al. (1973),I 62295spinel troctolite, 2 60335feldspar-phyric troctolite, 3 15082soil fragment,4 68815 nar schreibersite result from could one or more of troctolite clast in remobilizedbreccia (phosphide interpreted as three processes:(a) simple incorporation of mete- productof liquid quenchedfrom 1080"C) 5 14310basalt, El oritic schreibersite,(b) shock heating of meteoritic Goresye/ al (1972).6 64455highland basalt, Grieve and Plant phosphate(Goldstein et al., 1972),and (c) conversion (1973).766095 metamorphosed breccia, mean of severalanalyses, (1973a).8 of lunar phosphate.Conversely meteoritic schreiber- Ef Goresy et al. 66065breccia, typical analysesfor sphericalintergrowths, McKay e! al. (1973).9 66095metamor- site could be convertedto lunar phosphate. phosedbreccia, metal-schreibersite-cohenite association, Taylor Gooley et al. (1973) studied the occurrence of et al. (1973b).l0 14310basalt, Axon andGoldstein (1972). ll-14 schreibersiteand iron in Apollo l6 rake samples,and phosphide-metalparticles in 14003and 14163fines, Axon and their data were supplementedand endorsedby those Goldstein(1972). 15 Type I enstatitemeteorite in Apollo l5 fines, (1972d). particle of Brown et al. (1973), El Goresy et al. (1973a), Haggerty l6 Apollo l2 soil F2, kamacite-schreibersiteeutectic, Goldstein and Yakowitz (1971').17 Grieve and Plant (1973), (1973), McKay et al. Misra Apollo l2 soilparticle F3-2, kamacite-taenite dendrites in schrei- and Taylor (1975), and Reed and Taylor (1974) for bersite-troilitematrix, Goldstein and Yakowitz(1971).See Misra other Apollo l6 rocks. Figure l7a. showsthe data on andTaylor (1975, Fig. 5) and Reedand Taylor (1974, Fig.9) for Ni and P distribution obtained by Gooley et al. graphicaldata on Apollo l6 samples.Data for 7 particlesfrom soil (1973), and Figure l7b shows tabulated data for 78501in whichiron-phosphide assemblages coexist with anorthosite,basalt, or agglutinatewere given by Goldsteinet al. (1974): many types given of specimens by various authors. equifibration temperatures of 375-475'Cwere indicated. Inset dia- Coexistingschreibersite and metallic iron irr samples gram shows 3-phasetriangles betweentaenite, kamacite,and quenched from high temperature (open circles in schreibersiteat 550,650, and 750oC,Doan andGoldstein (1970). LUNAR MINERALOGY t09'1 experimental data for the three-phasetriangle for phosphatewhile a secondtype occurs as tiny grains. taenite, kamacite,and schreibersiteat 750, 650, and The assemblageplotted as Number l0 in Figure l7b 550" C. For bulk Ni contents to the right of the apparently annealed at 700o C while assemblage kamacite-schreibersiteside of the triangle, the tie Number 5 may not be in equilibrium.When account lines betweenkamacite and schreibersitestay almost is taken of metamorphism and oxidation-reduction parallel, whereasfor bulk compositionsto the left of reactions,the complex data are not inconsistentwith the taenite-schreibersitejoin, the tie lines between derivation of 14310by melting of a regolith (but see taeniteand schreibersiteare also almost parallel.The Crawford and Hollister, 1974,p.aft). data of Gooley et al. (1973)indicate that schreibersite In summary, the coexistenceof schreibersiteand and iron exchangedNi to -600" C in mesostasis-rich iron provides a useful parameteron thermal anneal- rocks,to -550' C in diabases,and to various temper- ing, while the relative amount of schreibersiteand atures from 600 to below 550' C for rocks with phosphateprovides information on the redox condi- poikilitic texture.Reed and Taylor (1974)reported a tions. Meteoritic schreibersitecan apparently be ei- similar temperaturerange for 100 metallic particles ther augmentedby lunar-derivedschreibersite or dis- from Apollo 16 soils and rocks. Both Gooley et al. sipated by conversion to phosphate; unless it is (1973)and Misra and Taylor (1975)emphasized that accompaniedby other iron-bearing minerals in rec- schreibersitewas rare or absentin the poikilitic rocks, ognizabletextural or chemicalassemblages, it is not a and Misra and Taylor noted that whitlockite was reliable indicator of meteoritic contamination. unusually abundant in them. Gooley et c/. suggested (c) Metallic Fe,Ni,Co minerals. Introduction. that this resulted from conversion of phosphide to Most metallic minerals in lunar specimens are phosphateas a result of slower cooling, while Misra Fe,Ni,Co alloyswith up to 100percent Fe, 60 percent and Taylor pointed out that the distinctly high P Ni, 8 percentCo, and small amounts of other sider- contents of the metallic iron in poikilitic rocks in- ophile elements.Trivial amounts of metallic copper dicated a different thermal history which included occurring in some specimens(listed by Dwornik el equilibration at a higher temperature. No phos- al., 1974)appear to be indigenous,while indium and phide-phosphate intergrowths hav€ been reported. aluminum metal probably are contaminants.Most of Reed and Taylor (1974) statedthat the P contentsof the Fe,Ni,Co alloys were characterizedonly by op- metallic iron minerals in the Apollo 16 rocks were tical and chemical methods,and djstinction between higher than those for metallic iron minerals from kamacite (a type with low Ni) and taenite (r type meteorites,and suggestedthat P was incorporated with high Ni) varieties was not always made. Al- into metallic iron minerals from meteoritic debris though details of interpretation were controversial, upon reduction of lunar phosphate. magnetic techniquesyielded much additional chem- Spinel troctolite 62295 has a quench texture, but ical information, especiallyfor submicroscopiciron the Ni distribution betweenschreibersite and taenite produced by reduction. The literature is confusing (Brown et al., 1973;Misra and Taylor, 1975) in- becauseearly data wereinterpreted in terms of simple dicatesequilibration below 550' C (Fig. l7b, No. I ). ideas now seen to require further evaluation, and The coarsenessof schreibersitegrains (called rhab- becauseit is difficult to cross-relatethe data obtained dite when occurring as needlesand elongatecrystals) by the different techniques.By 1975, it was certain provides an estimate of time-temperaturerelations. that iron minerals derived from at least four partly Axon and Goldstein (1972) concluded that several interrelatedsources: (a) rocks and magmasfrom the metal-phosphidefragments from Apollo l4 soilshad lunar interior in which the composition of the iron been annealed to 500-600' C for periods up to I mineralscorrelates with the silicates.sulfide. and va- month to 50 years. Gooley et al. (1973) concluded por phase, (b) simple meteoritic contamination in that rhabdite lamellae,4pm wide, in poikilitic rock which the iron mineralssurvived the impact, (c) com- 64657 would have required growth for about 2 plex meteoriticcontamination in which the iron min- months at 600' C. erals were transmuted during the impact and mixed Basalt 143l0 was controversialbecause of the high with material already on the Moon, (d) reduction of content of trace elementscharacteristic of meteoritic Fe2+in silicates,sulfide, and magma by various proc- contamination. Axon and Goldstein (1972) and El essesinvolving shock, escapeof S vapor, addition of Goresy et al. (1972)presented evidence for more than H and C from the solar wind, and conversionof C one generationof schreibersite,one type occurring as and H to escapingCO and HrO. Furthermore iron coarse grains in contact with iron, troilite, and Ca was transported in the vapor phase, was mixed up 1098 J. V. SMITH AND I M. STEELE and recrystallizedin breccias,and was generallyre- of the maria by internally-derivedbasalts would pro- mobilized during all the complex processesin the vide fresh surfaceswhich should receiveonly minor lunar crust. There is no needto belabor the problems amounts of earlier debris from late impacts in the resulting from oxidation-reductionreactions in view highlands.The regolith on the mare surfacesshould of earlier comments.However, it is necessaryto em- be especiallysuitable for recording meteoritic con- -3Gy. phasizethe problems of determining the amount of tamination sinceabout In the context of this metallic Fe,Ni,Co resulting from meteoritic con- complex situation, (a) Ganapathy et al. (1970) and tamination. others noted the very strong depletion of certain ele- Recent evidence from oxygen isotopes (Clayton ments including Ir and Au in Apollo I I mare basalts, and Mayeda, 1975)indicates that the precursorsof and calculated that the higher abundance of these the Earth, Moon, and "differentiated meteorites" (i.e. elementsin Apollo I I fines could be produced by eucrite, howardite, hyperstheneachondrite, enstatite about 2 percent undifferentiatedmaterial from the chondrite and achondrite. and nakhlite) could have solar nebula such as carbonaceouschondrite, (b) been geneticallyrelated in the solar nebula, whereas Morgan et al. (1974) interpretedthe Au, Ir, Ge, Re, the ordinary chondrites(types H, L and LL), carbo- and Sb contents of rock fragments from highland naceouschondrites (C2 and C3), and ureilites must soils in terms of debrisbelieved to have arisenfrom 6 have formed separately.The simplestexplanation is basin-formingprojectiles of distinctivechemical com- that the Earth, Moon, and differentiatedmeteorites position, and (c) Goldstein and co-workers (latest formed largely from material which was originally papers:Goldstein et a|.,1974; Hewins and Goldstein, interacting in the same general region of the solar 1975b) interpreted the Ni and Co contents of lunar nebula, whereas the other meteorites formed else- metal in terms of five components(meteoritic, high- where, but other possibilitiesmust be consideredin- Fe and high-Co basaltic, supermeteoritic,and sub- cluding a time variation. There is considerableargu- meteoritic-to be defined later). The (c) data were ment and some observationsin favor of present-day obtained by electronmicroprobe analysisof individ- meteorites arising from either the asteroid belt or ual metal grains, whereasthe former data were ob- cometary debris or both. Becausethe capture cross- tained by neutron activation analysis of bulk frag- section dependsstrongly on the relative orbital elements whose very minor component of metal ments, the relative population of meteoritesimpact- undoubtedly carries the trace siderophile metals ing the Earth and the Moon should have changed (W2inke et al., l97l; Wlotzka et al., 1972). Partic- considerablywith time. A searchfor meteoriticdebris ularly important is the conclusion by Morgan et al. on the lunar surface must bear in mind how the (1974)from plots of Au,Ge,Ir and Au,Sb,Re that the surfacedeveloped with time. In terms of our favored basin-formingprojectiles do not resembleiron mete- model, most of the Moon would have accretedabout orites and show little resemblanceto chondrites -4.5Gy, and shortly afterwardssevere differentiation which have recently impacted the Earth. Wlotzka et would have resulted in sinking of most of the Feo, al. (1972, 1973)made neutron activation analysesof Coo, Nio, and other siderophile elementsinto the Ir,Au,As,W and other trace elementsof bulk metal lunar interior. The nature of the remaining metal separated from Apollo 14, 15, and 16 fines and phase in the upper mantle may be revealedby the breccias,and also made electronmicroprobe analyses ultrabasic fragments emphasizedearlier, or by the of Co and Ni for individual particles.After reviewing mare basalts if they really derive from late remelting several possibilities,they concluded that, although of early deep-seatedcumulates. Down to -3.9Gy the the meteoritic component added during the past 3 to lunar surfacewould be changeddrastically by a flux 4 Gy had a siderophilecontent like that of undiffer- of incoming projectiles of which the large basins entiatedsolar material,earlier meteoriticcomponent would provide the most dramatic evidence. Each had a different composition. Wasson et al. (1975) ejecta blanket should contain some of the projectile concluded that earlier meteoritic contamination was greatly diluted by excavated crustal materials, the richer in siderophileelements than material arriving ratio probably varying considerably from place to after -3.7Gy, and went on to question the identi- place. Scouring of the lunar surface by the debris fication of debris from basin-forming projectiles by cloud, metamorphism,reduction, and vapor transfer Morgan el al. Although several fragments of stony should causefurther complications.Smaller impacts meteorites were discovered in lunar fines (e.g. a car- would rework the ejectablankets, and melting would bonaceouschondrite, Wood et al., l97l1'an enstatite produce plagioclase-richand other rocks. Flooding chondrite, Haggerty, 1972d), and many iron-rich LUNAR MINERALOGY 1099

fragments appear to be heated relics of iron mete- also concluded that the population frequencyvaries orites (e.g. Goldstein and Axon, 1973;Goldstein et approximately as the inversesquare of the grain vol- al., 1975), most meteoritic debris may have been ume, and that there are two separatepopulations, transformedduring incorporation into lunar regolith. one nearly equidimensional with coercive forces It is obvious that great caution is neededin attempt- >2000 Oe and one very elongated.The former can be ing to interpret the chemistry of lunar iron, and in- identified with spheresenclosed in glass,and the lat- deed in using the term "meteoritic." ter perhapswith needlesor whiskersobserved in sev- The magnetic properties of lunar specimensare eral optical and electron microscopicstudies. largely determined by the metallic iron minerals. Electron microprobe analysesof submicroscopic Much iron in unmetamorphosedbreccias, glasses, particlesare difficult, but some analysesindicate low and fines occurs as tiny particles pepperinga glassy Ni and Co for iron particles in glasses.The Curie matrix, as displayedin transmissionelectron micro- points for pure Fe, Co, and Ni are 770, 1131,and graphs (e.g. Agrell et al., l97O; Lally et al., 1972 358' C, respectively.The latestthermomagnetic data Housley et al., 1973).Widely-separated iron particles (Nagataet al.,1974,1975)show Curie pointsof lunar about 0.0lpm across in a diamagnetic matrix are samples straddling the 770" C transition for pure superparamagneticand show viscousremanent mag- iron. Becauseof the opposite temperatureeffects of netization (e.g. Gose et al., 1972). For increasing the subsitution of Co and Ni, all interpretationsare grain size, the magnetic phenomena are governed completely ambiguous: however, the histogramsare first by a single domain (-0.03rrm) and then by consistentwith all lunar rocks containing some pure multi-domain assemblages()lpm) typical of ferro- iron, and with the breccias,fines, and some but not magnets.Gose e/ al. (1972), Cisowski et al. (1974), all igneousrocks containing Ni-bearing iron as well and otherscorrelated the increasinggrain sizeof iron (see later). Mdssbauer resonancemeasurements of in metamorphosed breccias with the metamorphic the hyperfine field allowed Housley et at. (1972) to grade deduced from the silicateand oxide minerals, concludethat the iron metal in 10084fines contained and Pearceet al. (1972),Tsay and Live (1974), and lessthan 1.5percent Ni on auerageand perhapsnone othersmimicked the observedphenomena with meas- at all; becauseelectron microprobe analysesof coarse urementson glassyfines or syntheticglasses annealed particlesfrom 10084fines showed particles with up to at 600 to 1045" C. Wasilewski(1974) pointed out the l6 percentNi (seelater), the fine iron particlesshould possible magnetic effects from antiferromagnetic be very low in Ni. 7-Fe,Ni, and referred to the demonstrated occur- Although there are some technical subtleties,the rence of 7-iron in lunar breccias(Lally et al., 1972). amount of Feo was estimated from the saturation Early interpretationsof Mcissbauerand electron fer- magnetization,the ferromagneticresonance, and the romagnetic resonancespectra in terms of magnetite Mcissbauerspectrum. The principal features of the or other Fe3+-bearingphases appear to have been data given by Housley et al. (1974), Huffman et al. replaced satisfactorily by new interpretations (e.g. (1974), Pearce et al. (1974), Cisowski et al. (1974), Tsay et al., 1973b) based on tiny grains of super- and Pearceet al. (1975)are: (a) low values(<0.05) of paramagneticiron; however, the controversyhinges Feo,/(Feo* Fe'2+)in most mare basalts,(b) diverse on subtle argumentsinvolving computer simulation values for highland rocks (0 to 0.15), (c) relatively of resonance profiles, chemical composition, and small range for lunar soils (0.02 to 0.08) no matter shape,and on observedtemperature variation. Some whetherthey derivefrom highland or mare locations, relevantrecent papers are: Forester(1973), Griscom and (d) increasingvalues as the grain sizeof the soil et al. (1973), Tsay et al. (1973b), Weeks (1973), decreasesand the grain size of the iron increases,with Weeks and Prestel(1974), Friebeleet al. (1974), and a specifictrend for each of the Apollo 14, 16,and l7 Tsay and Live (1974).Many mineralogicalproperties suites.The latter trendswere explainedby increasing of lunar fines and breccias require a high state of reduction of the iron as a soil becamemore mature reduction,and any ferric-oxide phasesshould tend to during meteoritebombardment. be reducedto either ferrous phasesor iron metal. The formation of iron minerals in brecciasand Dunlop et al. (1973)concluded that single-domain soils must involve many complex problems (e.5. Ag- particlesof iron, l0-20nm across,are the major car- rell et al., 1970).Reduction of ferrous iron from the riers of natural magnetic remanencein soil breccia silicates,oxides, and sulfidesaugments the iron al- 14313.From detailed analysisof viscousremanence, ready available from pre-existing lunar rocks and thermoremanenceand magnetic granulometry, they meteoritic debris. Such reduction can occur as the re- l 100 J V. SMITH AND I. M STEELE

sult of loss of CO, S, HrO, and O, either upon ^ Juvinoseucrite meteoriticimpact or from thermal metamorphism.In- creasinggrade of metamorphismand recyclingof lunar cosmicrolio brecciasby fresh impacts should result in a tendency towards hybridization of all the different types of iron. Reduction of silicate.oxide. and sulfide could

_o yield iron low in Ni and Co, and this could dilute the bulk Ni and Co concentrationsof iron from most mete- lende da AI oritic debris and most lunar rocks. The solar wind +' 0.lpm with protons and carbon = implants the outer atoms, and the present-dayflux of protons could reduceabout lffglcm2 Fe during the past 3 Gy if no protons were lost during the impact. However, the escapevelocity for the Moon is only 2.4 km/sec, and some impacting bodies,especially present-day micro- meteorites,should hit at such a high velocity (10-25 km/sec) that a substantialfraction of material is lost. The situation before -4Gy is especiallyuncertain, Frc. 18. Co and Ni content of nickel-iron: summary of mete- but processesassociated with impacts must have oritic material.Note changeof scalefor Ni at 2Owl.Voand for Co caused substantial reduction of iron in rocks, aI2 wtVa breccias,and finesperhaps for depthsin the range of lrons'. dala for bulk compositions taken from Scott (1972, Fig l6) with principal regionsfor types I, IIa, b, tld, IIIa, Illc, IVa, IVb, kilometers. and others outlined.The unusual Butler iron with Co-rich kama- The next sections describe some of the miner- cire (Ni 6-6 5 co I 7), taenite (Ni l6-45 Co 0.6), and plessiteis alogicalproperties of iron metal in lunar rocks in the shown separately (Goldstein, 1966).Euuites and howardiles'.dala context of the above introduction. Emphasiswill be howardite and for homogeneous kamacite grains from the Binda placedon the Co and Ni content.Data for meteoritic Macibini eucrite from Lovering (1964) Five grains from the Ju- are shown in Fig- vinas eucrite showed diversebulk compositions (filled triangles; specimensrecently fallen on Earth WZinkee! al ,l97l).lron in micrometerintergrowth with troilite in ure 18 to provide a possible referencepoint. For the Moama eucrite has Ni 0.04 Co 0.16 wt.7o(Lovering, 1975) convenience,the scalesfor Ni and Co change at 20 Duke (1965) reported Ni for Ni-Fe metal in l0 basalticachon- and 2 weight percentrespectively, causing two kinks drites.but Co was not measured:6 have Ni {l wL%o,3have2-49o in the line for the cosmic ratio Ni/Co - 20. A de- Ni, and one contains taenite (3890)with kamacite (47o) We ob- is given in the figure legend.The tained new data for iron in the Ibitira eucrite describedby Wilken- tailed explanation ing and Anders (1975):iron associatedwith troilite has Ni 0.9 Co bulk compositions of metal from irons and non-car- 08-1.1, and that with ilmeniteand spinel has Ni 0.1-03 Co bonaceouschondrites have more than 5 percentNi, 0.4-0.6. and the Ni/Co trend cuts across the line for the Carbonaceouschondrites'. large filled circles are for metal grains (high Cr 016-1.0 and high P 0-3.2 wt%) usuallyenclosed in forsteritecrystals in black matrix of Type II (C2) carbonaceous enstatitechondrites (Wasson and Wai, 1970) An abstractby Goo- chondrites (Olsen el al., 1973a). Inclined crossesshow composi- ley and Moore (1973) reported generally low Ni in diogenites tions of kamacite and taenite (some as aggregates)in 7 Type III (13Vo;O.lVoin Garland and Shalka;up to 527oin lbbenbiiren)and carbonaceouschondrites (Fuchs and Olsen, t973). Kamacite high Co ()l%o; up to 870in lbbenbtiren and 27Voin Roda). The dominatesin 5 specimens,but in Allende and Ornans the metal is metalphases ofthe ordinary chondritesshow complextextures and mostly taenitegiving bulk compositionsnear 67 and 6l wt.7oNi zoning profilesof Ni and Fe, but detaileddata are not availablefor Note that one cross for kamacite lies outside the ringed area at Ni Co (eg. Wood, 1967;Taylor and Heymann, l97l; Dodd, l97l; 5 2 Co 6.9. Goldstein and Doan. 1972\.The Ni contentsof bulk metal from Other meleoriter: the hexagons show compositions of bulk Ni-Fe mesosideritesrange from 7.0 to 9.0 wt.7o(Powell, 1969;Wasson el metal in enstatite chondrites (ec), pallasites, bronzite-enstatite al., 1974); although no specific Co analyses were made, general chondrites (bec) and hypersthene-olivine chondrites (hoc) as re- chemicalrelations indicate similar levelsto thosein the iron mete- ported by Lovering (1964)from a Ph.D. thesisby Creenland.The ofltes. dotted line labeledenc showsthe compositionrange for Ni-Fe in The star at Ni 5 Co 0.3 is the predictedvalue for conversionof enstatitechondrites (Keil, 1968)which lies at higher Co than the all Fe to the metallic statein normal or carbonaceouschondrites mean reported by Lovering (1964);the horizontal cross at Ni 3 3 (Wlotzka et al., 1972).Not plotted are the following data for bulk Co 0.61 is for the Kota-Kota enstatitechondrite. The only Ni-Co metal analyzed by neutron activation (W?inke et al ,1910): Juvinas pair for an enst^titeachondri te is f or the Busteespecimen (Wai and eucrite Ni 29 Co 0.62,Norton County aubrite 10.3,027 and9.l, Knowles, 1972),but Ni analysesand generalchemical properties 0.29, Pultusk H-chondrite: 9.1, 0.46, Ramsdorf L-chondrite 13.9 suggestthat enstatite achondrites have a similar range of Co-Ni to 0 63. LUNAR MINERALOGY il01 cosmicratio. Reductionof all the Fe to the metallic statewould bringall thechondritic metal close to the star at Ni 5 Co 0.3 weightpercent (Wlotzka et al., 1972),but only partial reductionis to be expectedfor 2l chondritesincorporated into lunarsurface rocks, be- t202 I causesome Fe shouldremain in the silicates;indeed I 4 the enstatitechondrites should become oxidized with fr lossof the Si from the metalto form silicate.Kama- -.- cite and taenitein Type III carbonaceouschondrites arerespectively poorer and richerin Ni/Co than the ) 1i....lz cosmicratio, and the bulk compositionof the metal b.Q i2/ from the Allende meteoriteis very rich in Ni. The ty'; +- / metal from most eucritesand howarditeshas rela- = tively low Ni and high Co, but metal from the Moamaeucrite has no detectableNi and only a little Co. In generalthe metalsfrom the eucritesand how- ardites except for the Macibini eucrite can be matchedwith thosefrom the marebasalts. Mare basalts. Two typesof metal occur in mare basalts(Fig. l9). That in the Ti-rich mare basalts found at Apollo ll and l7 sitesoccurs only asblebs 20 40 in troilite(see earlier). The final liquid should be very r0 low in Ni and fairly low in Co as a resultof prior wr.% Ni crystallizationof olivineand/or pyroxene. Although Frc. 19. Co and Ni content of nickel-iron: mare basalts.Note few analyseswere made, Co apparentlyranges from change of scale for Ni at 20 wt.7o and for Co at 2 wL.Vo.Apollo I l: 1.4 to 0 weight percentwhile Ni decreasesslightly numerous generalstatements that Ni is very low ((0. l) and Co variable between 0 and 0.7; I datum from Simpson and Bowie from 0.4 to 0.00 weight percent.The metal in the (1970). Apollo l2: rangesgiven by different authors for same rock relativelylow-Ti basaltsfrom Apollo 12 and l5 sites do not agree completely; ranges for various rocks are outlined crystallizedthroughout the entire crystallization from data of Brett et al. (1971), Busche e/ al. (1971), Cameron rangeof the silicates,and its Ni contentdrops from (1971), El Goresy et al (1971), Reid et al. (1970),Taylor et al. as high as 30 weightpercent for inclusionsin early (1971), Wlotzk a et al. (1972),and Wood et al. (1971);inclusions in 12009,and 12022 are shown by special sym- percent olivines from 12004, olivineto aslow as0 weight for crystalsin the bol; ranges for metal enclosed by olivine from 12004 and 12022 final residuumwhere a little troilitemay occur.Dif- (Hewins and Goldstein, 1974) are stippled.Luna 20: one datum, ferentrocks show different composition ranges, and Tarasov et al. (1973) Apollo 15: rake samples, Dowty et al. differentinvestigators reported inconsistent ranges (1973b); Fig 8 shows plots for individual rocks; olivine-phyric for the same rock (e.9. for 12004).However, the samples tend to have Ni-rich metal compared to pyroxene-phyric gabbros similarcomposition ranges but extending general explained ones.Four show trendis clear,and the datacan be to lower Co contentswhen Ni is zero (Taylor et al., 1975;Taylor by the expectedtendency for the early ferromagne- and Misra, 1915).Apollo l7: Taylor and Williams (1974),Meyer sianminerals to extractMg andNi out of the liquid, and Boctor (1974) allowingFe to increase,while Co tendedto be inter- mediatein behavior(Hewins and Goldstein,1974). weight percentwere for metal coexistingwith chro- Idiosyncrasiesmay ultimatelybe correlatedwith the mite, but fluorescenceerror must be considered. complexbehavior of the other minerals,especially Selected non-mare rocks. Figure 20 shows the pyroxeneand oxideminerals (see earlier). Whether Co/Ni contentsof metal from some of the key rocks the negativecorrelations between S and Feoin mare mentioned earlier.The metal in mosaic assemblages basaltsresult from lossof S duringcrystallization or of the troctolitic granulite 76535 consists of both whetherthey resultfrom inheritanceof propertiesof high-Ni and low-Ni varieties.Most particlesconsist the sourceregion of mare basalts(perhaps early of one or the other, but some form a 2-phaseassem- cumulates)is undecided(Gibson et al., 1975).Taylor blagesimilar to the "clear taenite" found in ordinary et al. (1973a)correlated the Cr contentof Fe,Ni chondrites(Taylor and Heymann,1971) and in lunar metalin 15536basalt with that of coexistingoxides soils(Axon and Goldstein, 1973).A strong diffusion and other minerals;the highestvalues of 0.2 to 0.7 profile was found at the 76535 kamacite-taenite l 102 J. V. SMITH AND I. M, STEELE

high-Co part of the range for mare basalts,and one L20 t20t3 n might concludegenerally that the metal in the upper mantle and lower crust has equilibrated with the Fe,Mg silicatesand tends to be relatively rich in Ni troctolite has very Ni-rich ,n and Co. The 67435spinel metal, while the 62295spinel troctolite has metal with 7-17 weightpercent Ni and 0.7-l.l percentCo whose b-e 1.5 range overlaps that for the 14310 plagioclase-rich +' basalt. Unfortunately the data for this basalt are = somewhat inconsistent.James (1973) showed two rather wide ranges,one for P-free metal fairly closeto the cosmic line and centeredon Ni l4 Co 0.9, and a particularly wide range for P-bearing metal which 14310P-free tendsto straddlethe cosmicline and which extendsto 14310P- beorinq pure Fe. Hewins and Goldstein (1975b) explaineda narrow trend obtained by El Goresy et al. (1971) in wr. % Ni terms of fractional crystallizationinvolving simulta- neouscrystallization of metal and silicate.The mean Ftc 20. Co and Ni content of nickel-iron: selectednon-mare value 14310metal given by Ridley et al. (1972) rocks. Note change of scalefor Ni at 20 wt.Vaand for Co al2wt.Vo. of ?6535troctolitic granulite, singlegrains and kamacite-taeniteas- appears relatively high in Co. Data for metal in semblages,Cooley el al. (1974), oriented particles in the interior of plagioclase-richbasalts 14276 and 68415mostly lie in plagioclasegrains contain Ni 5- I 5 Co up to 5Vo:72415dunite clast, the range for 14310metal. Perhapsall three are de- Albeeet al. (l97ab); 78235norite, interstitial grains associated with rived from melted regolith, and are somewhat in- whitlockite, apatite, rutile, baddeleyite,chromite, and troilite, McCallum et al. (1975a); "noritic" fragmentsin Apollo l7 soil homogeneous. appear to be similar to 78235,metal in vein and as inclusion in Plagioclasecrystals in Luna 20 fines also contain orthopyroxene, Irving el al. (1974);67435 spinel troctolite frag- rods of Fe about 0.5pm wide and severalto l00rrm ment with coarse texture, Prtnz et al. (1973):62295 spinel troctolite long (Brett et al., 1973 Bell and Mao, 1973).Up to with quench texture, metal associatedwith schreibersite,Brown e, four orientations of the rods occur. Rounded blebs al. (1973),Misra and Taylor (1975),horizontal crossesshow indi- formed when the rods reached the surface. Electron vidual data; 14310plagioclase-rich basalt, many data, heavy lines outline ranges for P-bearing and P-free metal measured by James microprobe analysesrevealed about I percent Ni (1973), stippled area shows range obtained by El Goresy et ai. (Brett er al.) and 2.1 percent Ni (Bell and Mao). (1971)as reportedby Hewins and Goldstein(1975a, Fig. l), cross Exsolution of Fe from plagioclase is the favored shows mean value of range Ni 5.5+5.4 Co 0.8+0 4 obtained by mechanism for growth of theSe rods because they Ridfey er al. (1972),seealso Gancarzet al. (1971);68415 plagio- in the interior but not at marginsof plagioclase clase-richbasalt, large filled circlesshow data obtained by Taylor occur et al. (1973a) and Gancarz er al. (1972): 14276plagioclase-rich grainsin the 76535granulite. Presumably Fe exsolved basalt,filled squares,Gancarz et al. (1972);granitic part of 12013 from grain margins migrated into the mosaic assem- KREEP-rich polymict brecciacontained metal with 21.5-2370Ni blages.Exsolution of Fe requiresa reduction mecha- but no data for Co, Drake et al. (1970); metal in fine-grained nism. Many lunar rocks, fragments,and clastscon- aggregatesof plagioclase in 60015 anorthosite believed to result veins or intergranular clots of complex from shock melting, Ni 0.ll-0.13 Co 0.20-0.32,Sclar and Baur tain fig',l4\. assemblagescontaining several of the following phases: iron, troilite, baddeleyite, apatite, whit- boundaries.Metal also occurs as oriented needlesin lockite, several silicatesincluding diopside. Perhaps the interior but not at the grain marginsof the plagio- the whole crust is pervadedby vapors which react in clase.If analogywith the pure Fe-Ni systemis valid, subtle ways with the main minerals. Metal grains in the kamacite-taeniteassemblages reacted down to the granitic part of 12013 KREEP-rich polymict 400oC, and the inferred cooling rate is a few tens of breccia contain 22 percent Ni (Drake et al., 1970), degreesper million yearsfor the range 800 to 400'C. and the Co content should be measured. The724l5 dunite clastcontains metal with Co and Ni Anorthosites (sensu stricto) contain little or no ranges which overlap those for the troctolite. The metal. Most workers did not report iron metal in the metal from the 78235norite and Apollo 17 "noritic" 15415anorthosite (Genesis Rock), but Hewins and fragments has only 2 percent Ni but relatively high Goldstein (1975b) located very tiny grains devoid of Co (2.5-3.0).All thesecomposition rangesfall in the Ni and Co, possibly attributableto shock reduction. LUNAR MINERALOGY I 103

IndeedSclar and Baur (1974)reported only 0.1 Ni and 0.3 weight percentCo in metal enclosedin fine- grained aggregatesof plagioclasebelieved to have resultedfrom shock melting of 60015anorthosite. Figure2l showsranges of Co and Ni for metal in "anorthositic"rocks studiedby Hewinsand Gold- stein (1975b).Those for coarse-grainedspecimens t5455 mostlyfall in the rangefor metalin marebasalts (Fig. 19), whereasthose for recrystallizedspecimens tend to be richer in Ni and lower in Co. Considerable overlap occurs for the latter and the metal in the plagioclase-richbasalts and 62295spinel troctolite a (Fig. 20). (-) Fines. The iron in the fines rangesfrom coarse b.S fragments up to - lmm across down to sub- microscopicparticles enclosed in glass.As described + = 6443.5;78t55 in the introduction,the latter appearto be essentially l, pure iron formedby reductionof silicate.Many mor- i' phologieswere reported (e.9. Frondel et al., 1970). li i Trtr't't..rr,r, tt ! I This sectionis concernedprincipally with the coarse | / ! L----'. ' 1,"' particlesfor which extensivemetallographic and elec- ,t/ .// tron microprobestudies were made mostly by J. L Goldstein,F. Wlotzka,and their respectiveco-workers.Figures 22,24,25, and 26 summarize the analyses of the bulk Co and Ni for severalhundred particles from soilsat the Apollo sites,while Figure23 shows (oo9-" the assignmentscheme proposed by Goldsteinet a/. ,,,Irt'z'5 (1974).Because of spacerestrictions, we have combined togetherdata for soils of similar nature, but there are subtle differencesdiscussed in detail by wt.% Ni Goldstein et al. Although we agreewith the general Frc. 21. Co and Ni contentof nickel-iron:feld- conclusions,we adopt a rather more scepticalatti- spar-rich(anorthositic) specimens. Note changeof tude to assignmentof metal in lunar fines.Based on scalefor Co at 2 wt.Vo.Described in abstract(Hewins the data plottedin Figures18, 19, 20, and 21, we and Goldstein, 1975b).Full lines show rangesfor concludethat (a) coarse anorthositesand dashedones for recrvs- the high-Feand high-Cobasaltic tallizedanorthosites. rangesare continuous,and the lower limit of the high-Co basalticfield should be placedhorizontally microtextureswhich were studied systematicallyby at Co - I weight percent,(b) the meteoritefield Blauand Goldstein (1975) using optical and scanning appliesonly to a restrictedgroup of relatively"un- electronmicroscopy plus electronmicroprobe analy- differentiated"meteorites with ColNi near the cos- sis. Cooling rates were deducedby comparingtex- mic ratio, and doesnot encompassthe verydifferenti- tures with those of quencheddroplets of synthetic ated eucritesand howarditeswhich tend to overlap moltenalloys. About 30 percentof the lunar spher- the lunarbasalts, (c) non-basalticlunar rocks such as ules had dendritic or cellular texturesindicative of anorthosite,norite, and granulitecontain iron in the rapid quenching,and Blau and Goldsteinconcluded basaltichigh-Co range,(d) recrystallizedrocks such that 6 out of 65 spherulespartially or completely as 14310and the recrystallized"anorthositic" rocks solidifiedby radiation lossduring flight in the lunar of Figure2l occupywide rangesof uncertainmean- vacuum,that 3 solidifiedupon conductionloss to a ing. Furthermorewe note that shock-reducedlunar silicatesubstrate, and that about 9 cooled by both iron has low Ni from magneticdata, and probably radiation and conduction.The remaining70 percent fallsin the high-Febasaltic range of Gol dsteinet al. had globular metal areas,and somehad solid-state Metallic spherules, 50pm-5mm across, from precipitates;a variety of processesincluding rela- Apollo ll, 12, 14, 15, and 16 finesshow complex tively slow cooling or even reheatingin an ejecta J. V, SMITH AND I M. STEELE

at2.7 and 3.5 weight x The S and P contentsaverage t.0 2.35 percentwith most valuesbelow 8 and l0 weight percent. All six radiation-cooled spherulesare S-rich. o lunorspherules --{' c) Phosphoruscould havederived from either reduction ).i, x X phosphide. x.ttxx x of phosphateor from original Experimen- bs , x ...x.tI" x x tal data indicatedthat Fe,Ni,Co,S,Palloy forms two 0.5 xtxxx x x xx # )40(x F( xx x = n$( xx immiscibleliquids upon melting.This explainedboth *)ffi x x -xx the sulfidehusks found around many lunar spherules Xz' and the enrichment of Co and Ni in the phosphide- u5t0n+ rich areas. That the spherulesoccur in all Apollo 1520 soils, irrespectiveof whether they are of mare or wt oloNi highland type, confirms assignmentto a meteoritic origin. Becausemost spheruleshave Ni > 5, hybrid- FrG.22. Co and Ni content of nickel-iron: lunar spherules.Bulk compositions. Many textural and chemical details in Blau and ism with reduced metal from lunar silicate sources Goldstein (1975).One datum at Ni 0.2 Co 2.35 omitted shcfuldbe small. Furthermore the extremepaucity of high-Co metal arguesagainst formation of spherules blanketcould be possible.The Co and Ni contents from lunar rocks exceptin rare instances.This study mostlylie in a band(Fig. 22) fromNi 6 Co 0.35to Ni of large lunar spherulesgives confidencein attempt- 20 Co 0.8,which closely matches the band for irons ing to interpretthe more complex data for other types and non-Qarbonaceouschondrites (Fig. l8), includ- of iron in lunar soils. (1975) ing the tendencyfor low Ni specimensto havemore Blau and Goldstein listed the many refer- glasses Co than thecosmic ratio while the high Ni oneshave encesto small metal spherulesin and on of less Co than the cosmic ratio. The simplestex- which the mounds revealedby scanningelectron mi- planationis that the spherulesmostly result from croscopy are particularly spectacular.Although re- the smaller bulk iron in a rangeof meteorites,and that con- quiring further study, it is likely that poorer in Ni and Co, and to have taminationwith lunar iron is insignificant.Three spherulestend to be in Carter and spheruleswith Ni ( 5, one with high Co(2.3),and formed by reduction of Fe2+ silicates. (1972) two with low Co(0.2)may resultfrom lunaror mete- McKay simulatedthe moundsexperimentally. we togetherthe oritic sourcesor a mixtureof both; for examplethe As a crude approximation, brought irregular particles in fines one at Ni 3 Co 0.25 falls near the range for the numerous data for large (Apollo Macibinieucrite. from highland sites 14,Apollo 16,and Luna 20; Fig 24), siteswith high-Ti mare basalt(Apollo I I and l7; Fig. 25), and siteswith low-Ti mare basalt (Apollo 12 and l5; Fig. 26).lt must be emphasized that in the later Apollo missions,the fines from the differentsubregions should and do vary considerably. The crystallization agesof the basalts at the sites are o (-) Apollo 14 (Al-rich) -4.OGy, Apollo I I (Ti-rich) -3.8 -3.6, -3.8 -3.7, be to Apollo l7 (Ti-rich) to Apollo 12 (quartz and olivine) -3.4 to -3.2, and +' "":s lfra' = -aglgoritic Apollo l5 (quartzand olivine)-3.4 to -3.3 (Taylor, - - 1975).Naively one might test whethermeteoritic con- DOS; tamination had changedwith time by comparing the high-Fe GHAclossificoiion metal from the various fines,but the uncertaintiesare immense. All the plotted data are for large metal particlesmostly near 100pmacross, but the sizepop- wt%Ni ulationsare differentfor eachsoil (e.g.Apollo 14and l5 soils,Wlotzka et al.,1972, between20 and l00pm Frc. 23. Metal classification of Goldstein et al. (1974) based on after grinding the soils; Apollo 15, Goldstein and Ni and Co content. Abbreviated as M (meteoritic), SM (super- 1972, 5}-250pm; Apollo 16, Goldstein and meteoritic),Fe (basaltichigh-Fe), Co (basaltichigh-Co), and l-4 Axon, (submeteoritic with Ni l-49o and Co below basaltic high-Co). See Axon, 1973,over 200pm). their Figures6 and 7 for histogramsof metal in lunar soils. For these relatively large, irregular particles, the LUNAR MINERALOGY I r05 simple distribution for spherulesis replacedby complex distributions. Immediately obvious is a strong highlondsoils tendencyfor all soilsto havea concentrationof metal . 2.0 /\ 14003,14163 particleswith bulk compositionsnear Ni 6 Co 0.45, U x 63501,65701,68501 and indeed the tendency is so high that individual o L20 points could not be plotted therein without overlap. Actually the cluster for each Apollo site occupiesa t6 o r.w slightly differentregion, and the clustersfor the older () x x sites tend to center about Ni 6, whereasthose for the d-e younger Apollo 12 and 15 sites center about Ni 5. The reason for the clustering is unclear. Certainly E r.0 there is no simple match with the spherules,and one . .xt F..^ 2tx might speculatethat an early meteoritic component x {(}. a /x tendsto be dominated by relativelyNi-poor material, whereasa later meteoritic component shows a wide range of Ni like that for meteoritesnow impacting the Earth. Wlotzka et al. (1972) pointed out that "normal completereduction of all Fd+ to Feo in a or u carbonaceouschondrite" would give iron with 5 per- 5 l0 15 20 30 cent Ni and 0.3 percent Co (star in severalof Figs. u/t % Ni l8-26.) Neutron activation analysesof Co, Cu, Ga, Frc. 24. Co and Ni content of nickel-iron: soils from highland Ge, As, Pd, Ir, and Au for bulk metal from 14163 sites.Note changeof scalefor Ni at 20 wtVo and for Co at2wt.Vo. fines matched quite well with calculated values for Apollo l4 (Goldsteinet al., 1972;Wlotzka et al., (1972).Apollo l6 either a reduced carbonaceouschondrite or for a (Goldsteinand Axon, 1973).Luna 20 (Goldsteinand Blau, 1973). hexahedrite.The detailsare complex, but in general The ring outlinesa very heavyconcentration in all threeApollo l6 soils. the agreementresults becauseall these types of meteorite are relatively undifferentiated. The highland soils contain very little iron with Co ) l, indicating little contribution from high-Co basaltic,anorthositic, noritic, and granulitic rocks. The fairly strong concentration of pure iron may result 2.0 more from reduction of Fe'+ than from contribution of high-Fe basalticcomponent. Most of the data fall in a parallelogramfrom Ni 0 Co 0 weight percentto 10,0.3;10,0.9; and 0,0.6.Much of this rangeoverlaps o 1.5 those for recrystallized"anorthosites" in Figure 2l and feldspar-richbasalts in Figure 20. The simplest bQ explanation is that the iron of these highland soils consistsmainly of a complex mixture of metalsfrom # = an ancient meteoritic component and a variety of feldspar-rich rocks. Perhaps thorough study of the chemical composition of silicatesand glassesadher- ing to someof the metal particleswould allow further progressto be made. The soils from Apollo 12 and 15 mare regions contain much Co-rich metal, and there is a high concentration for 15601soil centeredon Co 1.3 Ni 5r015 0.25. A high concentrationof pure Fe occurs in all wt.% Ni these soils. Soil 15601 obtained from the edge of FIc. 25. Co and Ni content of nickel-iron: soilsfrom Apollo I I Hadley Rille is shown separately from the other and 17. Note change of scale for Ni at 20 wt.Voand for Co aI 2 Apollo l5 soils becauseit has no concentration of wt.7o.10084 (Wlotzka et al., 1972).78501and 78510(Goldstein el metal grains near Ni 5 Co 0.5 and is presumably al., 1974). Strong concentrations are ringed. I t06 J V. SMITH AND I, M STEELE

percent, 4 contentsare below 0.2 weight therebyruling . O t2oot out simple addition of Si-rich metal from enstatite 3 x l56Ol chondrites.Sulfur and P are also very low exceptfor 2n metal particles quenched from high temperature. Tungsten contentsare much higher than for metal in "undifferentiated" meteorites(Wlotzka et al., 1972), and addition consequent upon reduction of W in ol5 c) ,." lunar rocks is indicated. Chromium is usually low, but one particle from Apollo l4 soils contained l7 b.e weight percent Cr (Goldstein et al., 1972). #,^ Becausemost data were obtained on coarse par- = t.u ticles or on bulk samples,the role of the smaller particles is rather undefined. Electron microprobe analysescould be extended to particles only a few J n5 micrometersor lessin diameter.Systematic study of the composition and texture of adhering or containing silicateor glassis mostly lacking, though Gold- steinet al. have provided somequalitative indications uU 5 t0 15 2030 (e.9. Goldstein and Axon, 1972). Because silicate should be more abundant than metal in the average wt.% Ni product of the solar nebula,several percent of mete- FIc 26 Co and Ni content of nickel-iron: soilsfrom Apollo l2 oritic silicateshould occur in the lunar regolith. Per- and 15. Note changeof scalefor Ni at 20 wI.Voand for Co at 2 haps criteria can be developedfor recognizingsuch wI.Vo. (Wlotzka (Goldstein 12O01 et al., 1972). Apollo l5 and material when attachedto iron of meteoritic origin. Axon, 1972) Strong concentrationsare ringed. The one labeled 15071occurs only for that soil. The concentrationat the origin Breccia. Although there are many data on the occurs for all soils. SeeAxon and Goldstein (1973) and text for composition of iron in breccias,we have not found data on 2-phaseCo-rich particles any obvious patternsdifferent from thosein fines.As mentioned repeatedly,metamorphism of mixed de- dominated by mare basalt.The soils from Apollo I I bris has yieldedvery complex materialsranging from and l7 mare regions contain a wide range of metal weakly metamorphosed breccias up to partially- compositionseven though the local rocks are high Ti melted rocks. Many of the finesare merely disrupted basaltswhose metal has Ni -0 and Co ( - l. Sub- breccias.Recrystallization and vapor transport have stantial contribution from either lunar non-mare modified the iron metal in breccias,and the grain size rocks or "differentiated" meteoritesis indicated; of and chemical properties give information on the course, fragments of feldspar-rich rocks including metamorphic grade. Detailed studiesof iron and as- anorthositesand norites have beenreported by many sociatedsilicate and other minerals in relativelyun- workers. metamorphosedbreccias may allow deciphermentof Goldstein and co-workers reported many textures early processesat the lunar surface. and chemicalvariations too complex to discusshere Conclusion in detail: among those reported are structureless metal (usually kamacite), kamacite * isothermal This review of the mineralogy of the Moon has taenite,ragged d2, acicularmartensite, zoned taenite, been set deliberately in a petrological-geochemical d + I t interstitial phosphide, metal * massive context with auxiliary useof geophysicalparameters. phosphide,metal * phosphideeutectic, and metal * Although the mineralogycan be interpretedplausibly sulfide.In general,some grains have equilibrated to in the context of an early differentiatedMoon whose low-temperature assemblagessuch as coexisting surfacewas extensivelyreworked by various igneous kamacite and taenite,whereas others show different and metamorphic processes,the provenanceof most levels of adjustment to lower temperature. Partic- samplesis too uncertainfor definitiveinterpretations. ularly important was the conclusion (Axon and The cluesprovided by the ultrabasicand basic rocks Goldstein, 1973) that some high-Co particles in and fragments are very tantalizing, and combined Apollo l5 soils had annealedto 350'C at conditions with the equally fragmentaryand tantalizinggeophy- probably found only at a depth of l0-20km. Silicon sicaldata, allow speculationabout the lunar interior. LUNAR MINERALOGY I 107

In recent years increasingemphasis on igneousand ence. Certainly in the specificfield of mineralogy,the metamorphic processes in meteorites has been implications of lunar mineralogy for the origin and coupled with theoreticalarguments for catastrophic developmentof meteoritesand planetswill be mined dynamics to favor high-temperature processesin for a very long time indeed. early bodies in the solar system.Such high-temper- Perhaps some readerswill feel that we have at- ature processes unfortunately tend to reduce the tempted to make too much of the slender clues in number of cluesto the mineralogy of the early solar lunar mineralsto which we reply with a phrasefrom system and to the origin of the planets. In a good Robert Browning's famous soliloquy by the Floren- detectivestory, the clueslead inexorably to a unique tine painter Andrea del Sarto: "Ah but a man's conclusion,but the caseof the Origin of the Moon is reach should exceedhis grasp/Or what's a heaven still open. The lunar detectiveshave hauled in ex- for." cellentclues, but many remain undiscovered.Further exploration of the lunar surfacewith detailedsample Appendix: Additional lunar minerals collection is needed when reconnaissancehas been The following minerals extend the list in Part I: completed of other targetsin the Solar System;per- cordierite (Albee et al., 1974a),inclusion in spinel haps mankind will learn the futility of expenditures grain of a plagioclase-richlithic clast of sample on dreadful war machinesand will direct its aggres- 72435,no chemical or optical data listed; molybdenite sions to colonization of parts of the Solar System. (Jedwab, 1973), micrometer-sizeparticles in fines, Scientificstudy of the Moon might then enter a new probably contaminant from lubricant in samplecabi- phase in which field and laboratory studies would nets;farringtonite (Dowty et al., 1974b), (4pm grain consolidate and extend present knowledge. In the associatedwith nickel-iron. troilite. and a Ca-silicate meantimesubstantial advances should come from the (probably plagioclase)all enclosedby spinelin spinel following studiesof lunar mineralogy:(a) carefulsur- troctolite clast 65785, chemical analysis showed 5l face study of all lunar samplesnow in storage for percent P2OE,4l percent MgO, 4 percent FeO;'aka- further coarse-grainedrocks or clasts, (b) detailed ganbite (Taylor et al., l974a,b), identified by X-ray study of all the Mg-rich specimens,especially the and optical spectrain the "rusty" materialpreviously "periodotitic" ones, with particular emphasis on ascribedto goethite(?)in referenceslisted in Part I, those minor and trace elementswhich undergo sub- attributed to terrestrial oxyhydration of probably stantial fractionation, (c) coordinated study of un- precursorsFe metal and lawrencitein Apollo l6 spec- metamorphosedpolymict brecciasto searchfor min- imens; unidentifiedCl,Fe,Zn-bearing phases, probably eralogicand geochemicalclues leading to a hierarchy a sulfateand a phosphateassociated with akagan6ite; of processes,(d) study of redox conditions as a func- still uncharacterizedmineralogically (El Goresy et al., tion of crystallizationage to test for an over-all ten- 1973b seealso Taylor et al., 1973b);bornite(2) (Car dency to progressivereduction, (e) detailedstudy of ter and Padovani, 1973),chemical identification dur- the chemistryof coexistingiron and silicateminerals ing SEM study of metallic spherulein 68841 fines, in fines and breccias to obtain further clues to the occurs with FeS between interlocking crystals of amount and type of meteoritic contamination as a schreibersiteand metallic iron; aluminum oxycarbide function of time, (f) laboratory studiesof potential (?) (Tarasov et al., 1973),20pm grain in troctolitic geochemicalindicators such as the partition of transi- fragment from Luna 20,electron microprobe analysis tion metals between silicate,metal, and liquid as a gave 74 percent Al, 20 percent C, and 4 percent 0; function of temperatureand redox state.These and monazite(Lovering et al., 1974), l0pm grain identi- many other studiesto be invented by ingenious in- fied chemically by electron microprobe from mesos- vestigatorsshould reduce the uncertaintiesso often tasis of 10047 mare basalt; thorite(?) (Haines et al., mentioned in this review. 1972), abstract reporting chemical identification of Few scientific endeavors can have been so exciting one U-rich grain in 14259,97 soil; pyrochlore (2) and so full of good fellowship as the study of the (Hinthorne et al., 1975),abstract describing chemical mineralogyof the Moon. We are greatly indebtedto identificationof 5pm grain in 76535granulite, 20 per- our scientific colleaguesfor their brilliant studies cent UO, + ThOr. Titanite, alias sphene,reported in which made this review possible.Scientific historians Part I as a doubtful mineral, was chemicallyidentified will probably evaluatethe scientificexploration of the by Grieve et al. (1975),and occursas anhedral,(5pm Moon as one of themost productive advancesin the grains associatedwith ilmenite and troilite in basaltic geological sciencesand indeed in the whole of sci- mesostasisof polymict breccia 14321.Zr-rich phases I 108 J V. SMITH AND I M STEELE

show even further complexities, and Brown el al. -. T. F. Bneztures, J JACoBy eNo J. V. SurrH (1972) (1975) now record 8 contrasted compositions for Thermal and mechanical history of breccias 14306, 14063, 14270,and 14321.PLC J, 819-835. zirkelite-type phases. Graphite was found with FeS - AND J. V Sunu (1971) Nature, occurrence,and exotic as a nodule in an a-iron fragmentof soil 14003(Gold- origin of "gray mottled" (Luny Rock) basaltsin Apollo l2 soils stein e/ al., 1972). and ttreccias.PLC 2,431-438. - AND l0 Orsens (1970) Armalcolite: a new mineral from the Apollo ll samples.PLC I, 55-63. ANDERSoN,D. L. (1975)On the compositionof the lunar interior. Acknowlegments J Geophys.Res 80, 1555-1557. AnnHrNtus, G., J. E EvrnsoN, R W. FrrzcrnALD ANDH. FuJrrA We thank all our colleagues at the University of Chicago for (1971)Zirconium fractionationin Apollo ll and l2 rocks PLC help and advice,particularly E Anders A. T. Anderson, R. N. 2 , 169-t7 6 R and L. Grossman.Technical help from l. Clayton, Ganapathy Axou, H. J. nNo J. l. Got-osruN (1972)Temperature-timerela- Baltuska, R. Banovich, O. Draughn and T. Solberg was much tionshipsfrom lunar two phasemetallic particles (14310, 14163, appreciatedWe thank B. Mason and R Brett for helpful criticism 140O3). Planet Sci Lett. 16, 439-447. and the Publicationsand ExecutiveCommittees of MSA for their Qarth - AND - (1973) Metallic particles of high cobalt difficult decisionson this unusually long manuscript. Generous content in Apollo l5 soil samples.Earth Planet. Sci.Lett. 18, financialaid camefrom NASA grant NGL l4-001-l7l Res,but the I 73- I 80. researchcould not have been carried out without electron micro- Blsu, A R. nNl I. D. MACGREcon(1975) Chromite spinelsfrom probe and X-ray facilities built up from NSF grants including the ultramafic xenoliths Geochim. Cosmochim Acta, 39, 937-945. grant to the Materials ResearchLaboratory (NSF DMR72-03028 Blvsn, C., J Fsr-scHr,H. ScHur-znNo P. RUrcsrccen (1972)X- A04). The cost of publishing this paper has been met by a special ray study and M6ssbauer spectroscopy on lunar ilmenites grant from the National Aeronauticsand SpaceAdministration. (Apollo ll) EarthJlanet. Sci Leq 16,273-274. Berr, P M rNo H. K Mno (1973) Optical and chemicalanaly- References sis of iron in Luna 20 plagioclase. Geochim Cosmochim Acta, The followrng reference abbreviations are used 31,755-759. throughout: - AND - (1975) Cataclasticplutonites: possible keys to the evolutionary history of the early Moon. L,S VI,34-35. Apollo l5 The Apollo I5 Lunar Samples (Lunar Sciencelnstitute, Seealso PLC 6.231-248. Houston, Texas) Btrlcs, A. E. er.rp B. Aurrrn (1972) Secondaryion analysisof V, VI III, IV, V, Z1 (Lunar Science LS III, IV, Lunar Science pyroxenes from two porphyritic lunar basalts. Apotlo t5, Institute, Houston, Texas) t9r-194. PLC I ProceedingsApollo Il Lunar ScienceConference, Geochim. _, J W DELAN., J. J. peprrs AND K L. CerunnoN (1974) Cosmochim' Acta Suppl - petrology of the highlands massif at Tauris-Littrow: an analysis PLC 2,3,4,5,6 Proceedings2nd etc. Lunar Science Conkrence, of th" i_a.. soil fraction. pLC 5,.:g5-g21.. Geochim. Cosmochim. Acta Suppl. 2, etc. _ AND J. J. pAplKE (1972) pyroxenes as recorders of lunar basalt petroqenesis:chemical trends due to crystal-liquid inter- AcRELL,S. O., J. E. Acnrll, A. R Anuolo eNo J. V. P. LoNc action.pLC J,431-469. (1973) Some observationson rock 62295.LS IV, 15-17. eNn D H. LrNosLey (1971) Crystallization -, J. H. ScooN, I. D. Muln, J. V. P LoNc, J. D C. histories of clinopyroxenes in two porphyritic rocks from McCoNNrr-rnxnA.Prcrerr(1970)Observationsonthechem- OceanusProcellarum.PLC2,559-574. istry, mineralogy and petrology of some Apollo ll lunar sam- S. SusNo eNo J. W. DeuNo (1973) Pyroxene ples PLC 1,93-128. poikiloblasticrocksfromthelunarhighlands.PLC4,597-611. Alnss, A. L., A. A. CHooos, R. F. Dvrrapr,A. J. G,cNc,o'nzeNo Brssop, F C., J. V., SrrrrrunNo J. B. DnwsoN (1975) Pentlan- D S GoloulN (1974a) Preliminary investigation of boulders 2 dite-magnetite intergrowth in De Beers spinel lherzolite: review and 3, Apollo 17, Station 2: Petrology and Rb-Sr Model Ages. of sulphides in nodules phys Chem. Earth,9, 323-337. LS V' 6-8. Bleu, P. J lNo J. I. Golt>srrrN (1975)Investigation and simula- -' D. A. PepeN,cstrsstou tionofmetallicspherulesfromlunarsoils.Geochim.cosmochim, nNo G. J. Wlssensunc (1974b) Dunite from the lunar high- Acta,39,3O9-3i4 lands:Petrography, deformational history, Rb-Sr age.lS Z BoyD,F.R.ANDD.SnarrH(1971)Compositionalzoninginpyrox- 3-5' enes from lunar rock 12021.Oceanus Procellarum.J Petrol. -, Dnpror.r (1975) sym- R. F. Dvvsr AND D. J. Spinel lZ.439_464. plectites: high-pressure solid-state reaction or late-stage mag- Bnerr, R (1973) A lunar core of Fe-Ni-S Geochim cosmochim. matic crystallization? LS vl, l-3. Acta' 37' 165-170' -, A. J. GANcARZnNo A. A. cuooos (1973)Metamorphism - ofApollo 16 and 17 and Luna 20 metaclasticrocks at about 3.95 (1975) Reduction of mare basalts by sulfur loss LS AE: samples 61156, 64423,14-2, 65015, 6'1483,15-2,76055, VI'89-91' 22N6.22007. PLC 4.569-595. -, P. Burrrn, Jn, C. Mryen, Jn., A. M. Rsto, H. Tersoe ANDERSEN,O, (1915)The systemanorthite-forsterite-srlica. Am nNo R. WtLt-l.lvs(1971) Apollo l2 igneousrocks 12004,12008, J. Sci., 4th ser ,39, 407-454. t2009, and 12022:a mineralogical and petrological study. PIC ANoensoN, A. T. (1973) The texture and mineralogy of lunar 2,301-317. periodotite, 15445,10.J. Geol.8l,219-226. -, R. C. Goolev, E. Dowrv, M. PntNz ANDK. KEIL (1973) LUNAR MINERALOGY t t09

Oxideminerals in lithic fragmentsfrom Luna 20fines. Geochim. and implications on the originof metallicmounds on lunar '16l-773. CosmochimActa, 31, glasses.PLC 3, 953-970. BnowN,G. E. nNo B. A. WpcHsr-en(1973) Crystallography of - ANDE. PeoovnNr(1973) Genetic implications of some pigeonitesfrom basalticvitrophyre 15597. PLC 4, 887-900. unusualparticles in Apollo l6lessthan lmm fines668841,1I and BnowNG. M., C. H. ErvrBLrus,J. G. Holr-rNo,A. Prcxrrr lNo 6994t.t3.PLC 4. 323-332. R. PHrr-lrps(1971) Picrite basalts, ferrobasalts, feldspathic no- Cnnrrn,N. L., L. A. FBnNllorz, H. G. Ave'l,tl-LEMANTlr-ro I. S. rites,and rhyolitesin a stronglyfractionated lunar crust. PIC 2, LruNc (1971)Pyroxenes and olivinesin crystallinerocks from 583-600. the Oceanof Storms.PLC 2,775-795. AND_ (1972)Mineral_chemical -, I. S.Lsur.rc, H. G. AvE'hLLEMANTeNo L A. FERNANDEZ variationsin Apollo l4 and Apollo l5 basaltsand granitic (1970)Growth and deformationalstructures in silicatesfrom fractions.PLC 3.141-157 Mare Tranquillitatis.P LC I, 26'l-285. -, A. Prcrerr, R. PsrllrpsrNo C. H. EruEr-Eus(1973) Min- CHnurNrss,P. 8., A. C. DuNHnv,F. G. F. Grss,H. N. Gtles, eral-chemicalvariations in The Apollo l6 magnesio-feldspathic W. S. MncKrrzre, E. F. Srunpnl euo J. ZussulN (1971) highlandrocks. PLC 4, 505-518. Mineralogyand petrologyof someApollo l2 lunar samples. C. H. Er*.ru-eusAND R. PHrr-r-rps(1974) Min- PLC 2.359-376. eral-chemicalproperties of Apollo l7 marebasalts and terra - ANDG. W. LonrnrEn( 197 I ) An electronmicroscopic study fragments.LS V,89-91. ofa lunarpyroxene. Contrib. Mineral. Petrol.33, l7l-183. R. Psrr-lrpsAND C. H. Eusleus(1975) Mineralogy CHeo, E. C. T. (1973)The petrologyof 76055,10,a thermally and petrologyof Apollo l7 basalts.LS VI,95-97 and PLC 6, metamorphosedfragment-laden olivine micronoritehornfels. r-13. PLC 4. U5-664. Bnynr.r,W. B. (1974)Fe-Mg relationshipsin sector-zonedsub- - ANDl4 orsBns (1970)Pyroxferroite, a newcalcium-bearing marinebasalt plagioclase. Earth Planet.Sci. Lett.24, l5'l-165. iron silicatefrom TranquillityBase. PLC I, 65-79. BuNcs,T. E. nNo K. Kerl (1971)Chromite and ilmenite in non- CHrnr-ss,R W., D. A. HEwtrr ANDD. R. WoNrs(1971) HrO in chondriticmeteorites. Am. Mineral.56, 146-157. lunarprocesses: The stability of hydrousphases in lunarsamples lNo K. G. SNrrrsrNcen(1967) Chromite composi- 10058and 12013PLC 2,645-664. tion in relationto chemistryand texture of ordinarychondrites. CHnrsrre,J. M., J. S. Lnllv, A. H. Hpurn,R. M. FIsusn,D. T Geochim.Cosmochim. Acta, 31, 1569-1582. Gntccs lNo S.V. RADcLIFFE(1971) Comparative electron pet- - ANDE. Or-srN(1975) Distribution and significance of chro- rographyof Apollo I l, Apollo 12,and terrestrial rocks. PLC 2, mium in meteorites.Geochim. Cosmochim. Acta, 39, 9ll-927. 69-89. BunNHlu, C. W. (1971)The crystalstructure of pyroxferroite CsnrsropHe-MrcHrr--Levy,M., C. Lnvv, R. Clvp ANDR. PIER- from MareTranquillitatis. PLC 2,47-5'l. nor (1972)The magnesianspinel-bearing rocks from the Fra BunNs,R G., D. J. VeucnlN, R. M. Anu-ErolNo M. WrrNrn Mauroformation. PLC 3, 887-894.Given incorrectly as (1973) (1973)Spectral evidence for Cr3+,Ti8+ and Fd+ ratherthan in PartI Cr2+ and Fe8+ in lunar ferromagnesiansilicates. PLC 4, Crsowsxr,C. S.,J. R. DuNr.r,M. Fullrr., M. F. Rosr ANDP. J. 983-994. Wnstlrwsxt (1974)Impact processesand lunar magnetism. Buscse,F. D., G. H. CoNRAD,K. Krrl, M. PnrNz,T. E. BuNcH, PLC 5,2841-2858. J. EnlrcuueNnNo W. L. Queror (1971)Electron microprobe Cr-eNroN,U. S.,J. L Cnnrrn nNnD. S. McKev (1975)Vapor- analysesof mineralsfrom Apollo l2 lunarsamples. Units. New phasecrystallization of sulfides?LS VI, 152-154.See also PIC Mexico Insl Meteorit..Soec. Pub.3. 6,7 t9-728 -, M. PnlNz, K. KelL eNo T. E. BUNcH (1972) Spinelsand CuvroN, R. N. luo T. K. Mnvso,c (1975) Genetic relations thepetrogenesisofsomeApollo12igneousrocks.Am.Mineral. betweentheMoonandmeteorites.LSVI, l55-l57andPLC6, 51, t729-174'7. t76t-t't69. CeupnoN,E. N. (1970)Opaque minerals in certainlunar rocks ConrsroN,W., M. J. VERNoN,H. Benny,R. Ruoowsxr,C. M. from Apollo ll. PLC 1,221-245. Guv nNo N Wnnr (1972)Age and petrogenesisof Apollo l4 - (1971)Opaque minerals in certainlunar rocks from Apollo basalts.LS III, l5l-153. 12.PLC 2,193-206. Cnnwrono, M L. (1973)Crystallization of plagioclasein mare - (1975) Posrcumulusand subsolidusequilibration of chro- basalrs.PIC 4, 705-717. mite and coexistingsilicates in the EasternBushveld Complex. - ANDL. S. Horr_rsren(1974) KREEP basalt:A possible Geochim.Cosmochim Acta. 39. l02l-1033. partialmeft from the lunarinterior. PLC 5,399-419 CnnenoN,K. L. ,qNoM. CeurnoN (1973)Mineralogy of ultra- Hot-ltsren(1974) Spinel-silicate co-crystal- mafic nodulesfrom Knippa Quarry,near Uvalde,Texas. Ceol Der-toN,J. lNo L. S. Soc.Am. Abslr.Programs,5,566. lizationrelations in sample15555. PLC 5,421429. - ANDJ. W. DnLeNo(1973) Petrology of Apollo 15con- Dnwsott,J. B. nNo J. V. SMIrH(1975) Chromite-silicate inter- sortiumbreccia 15465 PLC 4,461466. growths in upper-mantleperidotites. Phys Chem.Earth, 9, A. E. BsNcr rNo J. J. Prexr (1973)Petrology of 339-3 50. the2-4mm soil fractionfrom theHadley-Apennine region of the DELANo,J. W., A E. BErcr,J. J. PnptxsrNo K. L. CeusnoN Moon. EarthPlanet. Sci. Lett. 19,9-21. (1973)Petrology of the 2-4mm soil fractionfrom the Descartes - ANDG. W. FISHER(1975) Olivine-matrix reactions in regionof the Moon and stratigraphicimplications. PLC 4, thermafly-metamorphosed Apollo l4 breccias.Earth Planet. Sci. 537-55r. Lett. 25, 197-20'1. DeNce,M R., J. A. V. Douclls, A. G. PLANrnno R. J. Tnllll Clnren, J. L. lrqo D. S. McKlv (1972)Metallic mounds pro- (1970)Petrology, mineralogy and deformationof Apollo ll ducedby reductionof materialof simulatedlunar composition samDles.PLC I, 315-340. ll t0 J. V. SMITH AND I M, STEELE

_ AND_ (1971)Mineralogy and petrol- W GeNrNrn (1973b)Zinc, lead, chlorine and FeOOH-bearing ogy of someApollo l2 samples.PLC 2,285-299. assemblagesin the Apollo l6 sample66095: origin by impact of DrxoN,J. R. euo J. J. P,rprxr(1975) Petrology of cataclastic a comet or a carbonaceouschondrite. Earth Planet. Sci Lett. lE, anorthositesfrom the Descartesregion of the Moon: Apollo 16. 4lt4t9. LS VI. t90-192. -, P. R nvoonn eNo L. A. TAYLoR ( 197I ) The opaqueminer- DonN,A. S. nNoJ. I. Gor-osrstN(1970) The ternaryphase dia- als in the lunar rocks from Oceanus Procellarum. PLC 2' gram,Fe-Ni-P. Metal. Trans. l, l'159-1767. 219-235. Dooo, R. T. (1971)The petrologyof chondrulesin the Sharps -, L. A. Tnvlon nuo P RnvooHY.(1972) Fra Mauro crystal- meteorite.C ont rib. M ineral. Pet rol. 31, 201-227 . line rocks: mineralogy,geochemistry and subsolidusreduction Dowrv, E., K. Kerl ANDM. PnrNz(1972a) Anorthosite in the of the opaque minerals.PLC 3,333-349. Apollo l5 rakesample from SpurCrater. Apollo 15,62-66. ENcrLHnnor, W. VoN, J. AnNor, W. F. Milllrn AND D. SrdFF- AND- (1973a)Major-element vapor fraction- lrn (1970) Shock metamorphism of lunar rocks and origin of ation on the lunar surface:an unusuallithic fragmentfrom the the regolith at the Apollo I I landing site PLC I, 363-384. Luna 20 fines.Earth Planet.Sci Lett 21,91-96. EvrNs, H. T. Jn. (1970)The crystallographyoflunar troilile. PLC AND- (1974a)Plagioclase twin lawsin lunar /. 399-408. highfandrocks; possiblepetrogenetic significance. Meteoritics, Fr-oneN.R. J., K. L. C,tuenou, A E. BrNcn AND J. J. PAPIKE 9,183-197. (1972) Apollo l4 breccia 14313:A mineralogicand petrologic AND- (1974b)Igneous rocks from Apollo l6 report. PLC 3,661-6'71 rakesamples. PLC 5,431-445. FoRESTER,D. W. (1973) Mtissbauersearch for ferric oxide phases AND- ( | 974c)Lunar pyroxene-phyricbasalts: in lunar materials and simulated lunar materials. PLC 4, crystalfizationunder supercooledconditions. J. Petrol. 15, 269't-270r. 4t9453. FnresrLe.E. J., D. L. Gnlscou,C. L. MnnQu,qnor,R.A Wrrxs -, M. PnrNznNo K. Kerl (1973b)Composition, mineralogy, nNo D Pnrsret (1974) Temperature dependenceof the ferro- and petrologyof 28 marebasalts from Apollo l5 rakesamples. magneticresonance linewidth of lunar soils,iron and magnetite PLC 4.4234M. precipitatesin simulatedlunar glasses,and nonsphericalmetallic AND- (1974d\Ferroan anorthosite: A wide- iron particles.PLC 5,2729-2736. spreadand distinctivelunar rock type.Earth Planet Sci.Letr. FnoNorr-. C , C. Klerrv Jn., J. Iro ANDJ. C. Dnnrr (1970)Miner- 24, t5-25 alogical and chemicalstudies of Apollo I I lunar fines and se- -, M. Ross nNo F. Currrrrl (1972b\Fe'z*-Mg site distribu- lectedrocks. PLC I,445-474. tion in Apollo 12021clinopyroxenes: evidence for biasin Mdss- FnoNosr-, J. W. (1975) Lunar Mineralogy. Wiley-lnterscience. bauermeasurements, and relationof orderingto exsolution. New York PLC3,48t-492. FucHs, L. H. (1968)The phosphatemineralogy of meteorites.In, Dure. M. J., L S. McCellurrr.G. A. McK,rv ANDD. F. WEILL P. M Millman, Ed., Meteorite Research, Reidel, Dordrecht. (1970)Mineralogy and petrologyof Apollo 12 sampleNo. - (1971) Orthopyroxene and orthopyroxene-bearingrock 120f3:A progressreport. Earth Planet. Sci Lett.9,103-123. fragments rich in K, REE, and P in Apollo 14 soil samples Dusn, A., H. C. Hreno eNn R. N. ScHocK(1974) Electrical 14163.Earth Planet Sci.Lett.12,170-174. conductivityof olivineat high pressureand undercontrolled -AND E OlsrN (1973) Composition of metal in Type III oxygenfugacity. J. Geophys.Res. 79, 1667-1673. carbonaceouschondrites and its relevanceto the source-assign- DurB, M. B. (1965)Metallic iron in basalticachondrites. ,/. ment of lunar metal. Earth Planet Sci.Lett 18, 379-384 Geophys.Res 10, 1523-152'l. lNn K. J. JeussN (1973) Mineralogy, mineral- DUNI-oR,D. J., W. A. Gose,G. W. PEARCEnNn D. W. StneNc- chemistry, and composition of the Murchison (C2) meteorite. wev (1973)Magnetic properties and granulometryof metallic Smithsonian Contrib Ealth Scl., No. 10, l-39 iron in lunarbreccia 14313. PLC 4,2977-2990. GeNlpnrHv, R . R. R. Krnvs, J. C. Llur- nNo E. ANorns (1970) Dwonrrr, E. J., C. S. AnNrr-1,R. P. CHnrsrrnN,F. CurrrrrA, Trace elementsin Apollo I I lunar rocks:implications for mete- R. B. FrNxrlueu, D. T. LrcoN,Jn. nNoH. J. Rose,Jn. (1974) orite influx and origin of Moon. PLC I, lllT-1142. Chemicaland mineralogicalcomposition of Surveyor3 scoop -, J. W. MoRcAN,H HrcucHr,E ANDERSlNn A. T. ANorn- sample12029,9. PLC 5, 1009-1014. soN (1974) Meteoritic and volatile elementsin Apollo l6 rocks Dvnl, P., C. W PnnrrNAND W. D. Duly (1974)Temperature and in separatedphases from 14306.PLC 5,1659-1683' and electricalconductivity of the lunar interior from magnetic GnNcnnz,A. J..A. L ALsrErNo A. A. CHoDos(1971) Petrologic transientmeasurements in the geomagnetictail. PLC 5, and mineralogicinvestigation of somecrystalline rocks returned 3059-307t. by the Apollo l4 mission.Earth Planet Sci. Lell. 12' l-18. El Gonrsv, A., M. Pnrxz eNo P. Rnploosn(1975) Zoning in AND- ( | 972) Comparative petrology of Apollo spinelsas an indicatorof thecrystallization histories of Apollo l6 sample 68415 and Apollo l4 samples 14276and 14310.Earth f 2and l5marebasalts.Geol. Soc. Am.,Abslractswith Programs, Planet. Sci.Lett 16,307-330. 7, to67. Gey. P.. G. M. BlrrrcnoFr AND M. G. BowN (1970) Diffraction -, P. Revoosn rNo O. MroeNBAcH(1973a) Lunar samples and Mdssbauerstudies of mineralsfrom lunar soils and rocks. from Descartessites: opaque mineralogyand geochemistry. PLC t , 481497. PLC 4,733-750. -, M. G. BowN AND I. D. Murn (1972) Mineralogical and - ANDH. J. BenNuenor(1974) Taurus-Lit- petrographic featuresoftwo Apollo l4 rocks. PLC 3,351-362. trow TiOr-rich basalts:opaque mineralogy and geochemistry -, C M. BlNcnopr AND P G. L. Wlrltrtr,ts PLC 5.627-652. (1971) Mineralogical and petrographic investigationof some M. Plvrdrvlt. O. MeoeNsrcu.O. Milllsn nNo Apolfo l2 samples.PLC 2,3'17-392. LUNAR MINERALOGY llll

GHosr, S, l. S. McCnllur*l eNo E. Trny (1973) Luna 20 pyrox- surface,as observedin glasscoated Apollo l6 samples.PLC 4, enes: exsolution and phase transformation as indicators of 667-679. petrologichistory. Ceochim Cosmochim Acta, 37,831-839. GnrrnrN,W. L., R. Aulr nxo K. S. Htrrn (1972)Whitlockite and -, G. Nc eNo L. S. Wnlrsn (1972) Clinopyroxenesfrom apatitefrom lunar rock 14310and from Odegirden,Norway. Apollo l2 and l4: exsolution,domain structure and cation or- EarthPlanet. Sci. Lett.15, 53-58. der PLC -i, 507-531. Gnrscou,D. L., E. J. Fnreser-snNo C. L. Mlnounnor (1973) -, C. WeN ero l. S. McCnrruu (1975)Late thermal history Evidencefor a ubiquitous,sub-microscopic "magnetite-like" of lunar anorthosite67075: Evidencefrom cation order in oli- constituentin the lunarsoils. PLC 4,2'109-2727. vine and orthopyroxene. LS V(,282-283. GnossMe:r,L. (1975)Petrography and mineralchemistry of Ca- Ctss, F. G. F. (1974) Supercoolingand the crystallization of rich inclusionsin the Allendemeteorite. Geochim. Cosmochim. plagioclase from a basaltic magma. Mineral. Mag.39,641-653. Acta,39,433-454. Grssol.r,E. K. Jn., S. CHnnc, K. LsrroN, C. W. Moonr eNo HerNen,S. S., D. Vrnco nNo D. WARBURror.r(1971) Cation G. W. PEARCE(1975) Carbon, sulfur, hydrogen and metallic distributionsand cooling history of clinopyroxenesfrom iron abundancesin Apollo l5 and l7 basalts.LS Vl,290-292. OceanusProcellarum. PLC 2. 9l-108. See also PLC 6, 1287-1301. Hnccsnrv, S. E. (l97la) Compositionalvariations in lunarspin- - ANDG. W. Moone (1973)Carbon and sulfur distributions els.Nature Phys. Sci.233, 156-160. and abundancesin lunar fines.PLC 4, 1577-1586. - (l97lb) Subsolidusreduction oflunar spinels.Nature Phys. - AND - (1974\ Sulfur abundancesand distributionsin Sci.234,I l3-l 17. the valley of Taurus-Littrow. PLC J, 1823-1837. - 11972a)Apollo l4: subsolidusreduction and compositional Golosretn, J. L (1966) Butler, Missouri: an unusual iron mete- variationsof spinels.PLC 3,305-332. orite. Science 153, 975. - (1972b)Luna I 6: anopaque mineral study and a systematic - ANDH. J. AxoN (1972)Metallic particlesfrom 3 Apollo l5 examihhtionof compositionalvariations of spinelsfrom Mare soils. Apollo 15, 78-8 l. Fecunditatis.Earth Planet.Sci. Lett. 13,32E-352. - AND - (1973) Composition, structure, and thermal - (1972c\The mineralchemistry of somedecomposition and history of metallicparticles from 3 Apollo l6 soils,65701, 68501 , reaction assemblagesassociated with Cr-Zr, Ca-Zr and and 63501.PLC 4,751-775. Fe-Mg-Zr titanates.Apollo l5, 88-91. ANDS. O. AcRELL(1975) The structureand thermal - (1972d)An enstatitechondrite from Hadley Rille. Apollo history of five large metal particles from the lunar regolith. LS /J, 85-87. vt. 303-305. - (1973a)Luna 20: mineral chemistryof spinel,pleonaste, nNo C. F. YeN (1972) Metallic particles in the chromite,ulvdspinel, ilmenite and rutile. Geochim.Cosmochim. Apollo l4 lunar soil PLC 3, 1037-1064. Acta,37, 857-867. - AND P. J. Blnu (1973) Chemistry and thermal history of - ( 1973b)Armalcolite and genetically associated opaque min- metal particlesin Luna 20 soils.Geochim. Cosmochim. Acta,31, eralsin the funarsamples. PLC 4,777-797. 847-855. - (1973c)The chemistryand genesisof opaqueminerals in -AND A S Donr, Jn (1972)Theeffectofphosphorusonthe kimberlites.Intl. Conf.Kimberlites, Extended Abstr. 147-149. formation of the Widm?instatten pattern in iron meteorites.Geo- - (1973d)Ortho- and para-armalccllite samples in Apollo 17. chim Cosmochim Acta, 36, 51-69. NaturePhys. Sci 242,123-125. --, E P. HrNornsoN ANDH YAKowrrz(1970) Investigation - (1975)The chemistryand genesisof opaqueminerals in of lunar metal particles.PLC 1,499-512. kimberlites.Phys. Chem. Earth, 9, 245-307. -, R. H. HEwrNs nNo H. J. AxoN (1974) Metal silicate HlrNss.E. L.. A L. ALBEE.A. A. CHoDosAND C. J. Wnssensunc relationshipsin Apollo l7 soils.PLC 5, 653-6'll. (1971)Uranium-bearing minerals of lunar rock 12013.Earth - AND H Ylrowrrz (1971) Metallic inclusionsand metal Planet Sci.Lett. 12, 145-154. particlesin the Apollo l2 lunar soil. PLC 2, 177-191. -. A J. G,qNcenz.A. L. ALBEEAND G. J. Wrssensunc GooLEy, R. C., R Bnrrr nNo J. L. WARNER(1973) Crystallization (1972\Theuranium distribution in lunarsoils and rocks 12013 history of metal particles in Apollo l6 rake samples.PLC 4, and 14310.ZS 1//,350-351. 't99-810 Hencnlves,R. B. nNo L. S. Hollrsrrn (1972)Mineralogic and - AND J. R Srrlvru (1974) A lunar rock of petrologicstudy of lunar anorthositeslide I5415,18.Science, deep crustal origin: sample 76535. Geochim. Cosmochim Acta, t75,430432. 3E, 1329-1339. Hesxrr, L. A., C.-Y.Ssrs, B. M. BeNslr-,J. M. Rsoors, H. - AND C. B. Moone (1973) The metal in diogenites. Mete- Wrrsrr,rnuNexo L. E. Nyeutsr (1974)Chemical evidence for the oritics,8,4l (abstr). originof 76535as a cumulate.PLC 5, 1213-1225. Hrnno, H C.,A. Dusl, A. J. PrwrrsrrrAND R. N. ScHocr(1975) Cose, W A , G. W. Penncs, D. W. SrnnNcwAy AND E. E. Electricalconductivity studies: refinement of the selenotherm. Lnnsol (1972) Magnetic propertiesof Apollo l4 brecciasand LS vr, 355-357. their correlation with metamorphism.PLC 3,2387-2395. Httrsr, G. (1975)Petrology of lunar soils.Reu. Geophys. Space GnteN, D. H. (1975)Genesis of Archean peridotitic magmasand Phys.13, 567-587. constraintson Archean geothermalgradients and tectonics Ce- Herz, R T. (1972)Rock 14068:an unusuallunar breccia. PLCJ, ology,2, 15-18. 865-886. Gnrevs,R. A. F., G A. McKny, H. D. SvrrrHAND D. F WEILL HeNosnsoN,P. (1975)Reaction trends shown by chrome-spinels (1975) Lunar polymict breccia 14321:a petrographic study. Geo- of the Rhum layeredintrusion. Geochim. Cosmochim, Acta,39, chim Cosmochim Acta, 39, 229-245 I 035- I 044. - AND A. G. PlnNr (1973) Partial melting on the lunar Hruen,A. H.,J. M. Cuusrre,J. S. Lnlly ANDC. L. Nonp,Jn. I l 12 J. V. SMITH AND I M. STEELE

(1974) Electronpetrographic study ofsome Apollo l7 breccias. Kerl, K. (1968)Mineralogical and chemical relationships among PLC 5,275-286. enstatitechondrites. J. Geophy s. Res. 73, 6945-6976. HrwtNs, R. H lNo J I. ColosrnrN (19774)Metal-olivineassocia- -, T. E. BuncHrNo M. PnrNz(1970) Mineralogy and compo- tions and Ni-Co contents in two Apollo l2 marebasalts. Earth sitionof ApolloI I lunarsamples. PLC 1,561-598. Planet. Sci Lett. 24, 59-70. -, M. PRrNzAND T. E. BuNcs(1971) Mineralogy, petrology, - AND - (t975a) Comparison of silicateand metal geo- andchemistry of someApollo l2 samples.PLC 2,319-341. thermometers for lunar rocks. lS VI, 356-357. KEssoN,S. E. nNo D. H. LtNosr-sv(1975) The effectsof Al+e, - AND- ( I 975b)The provenanceof metalin anorthositic Cr+3,and Ti+s on the stabilityof armalcolitein vacuumand at rocks. LS Vl,358-360 and PLC 6,343-362. 7.5kbar. LS VI, 472-474See also PLC 6,9ll-920. HrNrHonNr, J. 8., R. CoNuo euo C. A. ANoenseN (1975) - ANDA. E. Rrxcwooo (1976)Mare basalt petrogenesis in a Lead-lead age and trace element abundancesin lunar troctolite, dynamicMoon. Earth Planet. Sci Lelt.30, 155-163. 76535.LS VI,373-375. Krrrr, C. Jn.nNo J. C. Dnnxr (1972)Mineralogy, petrology, and Huvl, P. F., M. PnrNz nNo K. Kul (1972)Niobian rutile in an surfacefeatures of some fragmentalmaterial from the Fra Apollo l4 KREEP fragment. Meteoritics, 7, 479-485. Maurosite. PLC 3, 1095-lll3. Hooces, F. N nuu I Kusurno (1973) Petrology of Apollo 16 KLrrNueuu,B. nNo P. RAMDoHR(1971) Alpha'corundum from lunar highlandsrocks. PLC4, 1033-1048. the lunar dust Earth Planet.Sci. Letl. 13' 19-22' HolLrsrsn, L S., W. E. TnzcrrNsrr, Jn., R. B. Hlnonnvss nuo KuRAr.G,. K. Kerr nxo M. Pntxz(1974) Rock 14318: a polymict C. G Kulrcx (1971) Petrogeneticsignificance ofpyroxenes tn lunar breccia with chondritic textsre. Geochim.Cosmochtm. two Apollo l2 samples.PLC 2, 529-557. Acta,3E,| 133-l146. Houslrv, R. M., E. H. Crnlrr, N. E. PeroN lNo I. B. Golosenc Kusurno,I. nNoH. S.Yoorn, Jn.(1966) Anorthite-forsterite and (1974) Solar wind and micrometeoritealteration of the lunar anorthite-enstatitereactions and their bearingon the basalt regolith PLC 5, 2623-2642. eclogitetransformation . J. Petrol. 7' 33'l-362. -! R. W. Gnnrr AND M. Asorl-Gewno (1972) Study of Lnllv, J. S.,A. H. Hrurn, G- L. Nono,Jn. lNo J. M. Csnlsrte excessFe metal in the lunar fines by magnetic separation, Mdss- (1975)Subsolidus reactions in lunarpyroxenes: an electronpet- bauer spectroscopy, and microscopic examination. PLC 3, rographicstudy. Contrib. Mineral. Petrol. 51,263-281. I 065- | 076. -. R. M FrsHER,J. M. Cnnrsrtr, D. T. GRIccs,A. H. nNo N. E. PlroN (1973)Origin and characteristics Heurn, G. L. Nono, Jn. ,tNoS V. Rnocrrrrt (1972)Electron of excess Fe metal in lunar glass welded aggregates.PLC 4, petrographyof Apollo l4 and l5 rocks.PLC 3,401-422. 2737-2749. Levv. C.. M. CsnrsropHp-Mtcutt--Levv,P. PIcor ANDR. CAYE Hunnnno, N. J, J. M. Ruooes, H. WrrsueNN,C.-Y. SHrH,orNo 1972\ A new titanium andzirconium oxide from the Apollo l4 B. M. BeNsnr-(1974) The chemical definition and interpretation samples.PLC 3, I I l5-l 120. of rock types returned from the non-mare regions of the Moon. LrNosrEv,D. H.lNo C. W. BunrHet'.t(1970) Pyroxferroite: stabil- PLC 5,1227-t246. ity and X-ray crystallographyof synthetic CaoluFeo ruSiOg Hunrurlr.l,G. P., G. R. Dulrvynr AND R. M. Frsuen (1975) pyroxenoid.Science, 168, 364-367. Superparamagneticclusters of Fd+ spins in lunar olivines.l,S -. S. E. KsssoN.M. J. HlnrzuAN ANDM. K. CusHtvt+tt VI, 414-416. See also PLC 6, 757-772. (1974a\The stabilityof armalcolite:experimental studies in the -, F. C ScHwenrn, R M. FrsHrn nNo T. Nnc,crl (1974) systemMgO-Fe-Ti-O. PLC 5, 521-534. Iron distributionSand metallic-ferrousratios for Aoollo lunar -, H. E. KrNcAND A. C. TuRNocK(19'14b) Compositions of sam ples: M dssbaueraiidhagnetic analyses.P LC 5, 277 9 -2794. syntheticaugite and hypersthenecoexisting at 810'C: appli- InvrNe, T. N. (1974) Petrology of the Duke Island ultramafic cationsto pyroxenesfrom lunar highlandsrocks. Geophys. Res. complex, Southeastern Alaska Geol. Soc. Am Mem. 138. Leu l,134-136. lRvlNc, A. J., I. M. SrsrI-r nNo J. V. SlurH (1974)Lunar noritic LrptN,B. R. (1975)The systemanorthite-olivine-silica and its fragments and associated diopside veins. Am Mineral. 59, bearingon the petrogenesisof lunar crustalrocks. Ceol.Soc. l 062- l 068. Am. Abstr.with Programs,T, ll'12. JlconzrNsxr, H. rNo M. Konxnwn (1973)Diffuse X-ray scatter- - ANDA. MuAN(1974) Equilibria bearing on thebehavior of ing by lunar minerals PLC 4,933-951. titanatephases during crystallization ofiron silicatemelts under - (1975) Diffuse scatteringby domains in lunar stronglyreducing conditions. PLC 5,535-548. and terrestrialplagioclase. LS VI,429-431. LorcnsN.G.. C. H. DoNALDsoN,R. J. Wllulus, O. Mult-tNs Jeues, O. (1972)Lunar anorthosite15415: texture, mineralogy and nNo T. M. UssrlneN (1974)Experimentally reproduced tex- metamorphic history. Science, l7S, 432436. turesand mineralchemistry of Apollo 15 quartz normative - (1973) Crystallization history of lunar feldspathicbasalt basalts.PLC 5, 549-567. 14310. U. S. Geol. Suru. ProJ. Pap.841. LoNcHr.J . D. WlI-rsn, T. L. Gnove,E. M. Srolprn ANDJ. F. Jrowln, J. (1971) Surfacemorphology of free-growingilmenites H,rvs(1974) The petrology of theApollo l7 marebasalts' PLC and chromites from vuggy rocks 10072,31and 12036,2.PLC 2, 5.447469. 923-935. LovERrNc,J. F. (1964)Electron microprobe analyses ofthe metal- - (1973)Rare-micron-size minerals in lunar fines.PLC 4, lic phasein basalticachondrites. Nature,203'70. 86I -874. - (1975) The Moama eucrite-a pyroxene-plagioclase- - ANDA HERBoscH(1970) Tentative estimation of thecon- plagioclase-adcumulate.Meteoritics, 10, l0l-l 14. tribution of Type I carbonaceousmeteorites to the lunar soil. - ANDl4 orHsns(1971) Tranquillityite: A newsilicate min- PLC t. 55t-559. eralfrom Apollo I I andApollo l2 basalticrocks. PLC 2,3945. JouNsoN,R. E.,E. WornueuN rNo A. MurN (1971)Equilibrium -. D. A. Wnnr. A. J. W. Crr.rnow ANDR. BRrrrEN(1974) studies in the system MgO-"FeO"-TiOz. Am. J. Sci. 271, Lunar monazite:a late-stage(mesostasis) phase in marebasalt. 278-292. EarthPlanet. Sci. Lett.2l. 164-168. LUNAR MINERALOGY I I 13

LuNntrc AsyLulr (1970) Mineralogic and isotopic investigations of projectiles, from meteoritic elements in Apollo l7 boulders. on funar rock 12013.Earth Planet Sci.Lett 9,137-163. PLC 5, t703-1736. Mno, H K., A. El Gonnsy nNo P. M. BBrt (1974\ Evidenceof MurN, A, J. Hrucr ANDT. LdFALL(1972) Equilibrium studies extensivechemical reduction in lunar regolith samples from the with a bearingon lunar rocks. PlC.3, 185-196 Apollo l7 sire PLC J, 673-683. MunrHv, V. R., N. M. EveNsoNAND H. T. Hlr-l (1971)Model of M,rnvrN, U B. (1971) Lunar niobian rulrle Earth Planet Sci.Lett early lunar differentiation. Nature,234,267 and 290 I t. 7-9 MysEN, B. (1975) Partitioning of iron and magnesium between -,J A. Wooo,G. J Tnvlon, J B. REID,Jn.,B.N powpll, crystals and partial melts in peridotite upper mantle. Contrib. J. S. Drcrrv, Jn. lNo J F Bowtn (1971)Relative proporrions Mineral. Petrol 52. 69-76. and probable sources of rock fragments in the Apollo 12 soil Nncnrl, T., R. M. FrsHER,F. C. ScuwenrR, M. D. FulLsn eNn sampfes.PLC 2, 679-699. J. R DUNN (1975) Effects of meteorite impacts on magnetic MrsoN, B (1972) Mineralogy and petrology of lunar samples properties of Apollo lunar materials. LS VI, 587-589 and PLC 15264,19,15274,12,and 15314,59.Apollo 15,t35-136. 6,3llL-3t22. -, W. G. MrlsoN eNn J. Nrr-pN (1972) Spinel and horn- -, N. Sucrur.e, R. M. Frssrn, F. C. ScswBRER,M D. blendein Apollo l4 fines L,S III, 512-514. Fullrn ANDJ. R. DuNr (1974) Magnetic propertiesof Apollo MessoN, C. R , L B. Surru, W. D Jllrrssol, J L. Mcl-lucsleN ll-17 lunar materials with special reference to effects of mete- eNo A. VoLsonrx (1972) Chromatographicand mineralogical orite impact PLC 5,2827-2839. study of Apollo l4 fines.PLC 3,. 1029-1036. NrHnu, C. E, M. PnrNz,E. Dowrv ANDK. KEIL (1973)Electron McCnllrsren, R. H. nNo L. A. Tlylon (1973) The kinetics of microprobe analysesof spinel group minerals and ilmenite in ulvcispinelreduction: synthetic study and applicationsto lunar Apollo l5 rake samples of igneous origin. Irst. Meteorit., UniD. rocks Earth Planet Sci. Lett 11,357-364. New Mexico, Spec. Publ. 10. McCnlluu, l. S., E. A. MATHEZ,F. P Orenrune eNo S. GHosr _ AND _ (19j4) Spinel_group minerals ( | 974) Petrologyand crystalchemistry of poikilitic anorthositic and ilmenite in Apollo l5 rake samples.Am. Mineral. 59, gabbro77017. PLC 5,287-3O2. t220-t235. _ AND _ (1975a) petrology of noritic NreruHn, H. H., S. Zrrnl' eno S. S. Hnrrrn (1973)Ferric iron in cumulates:samples 78235 and 78238.l.S V(,534-536. Seealso plagioclasecrystals from anorthosite 15415.PLC 4,971-982. PLC 6,395-414. O'Hnne, M J. eNo D. J. HuvpHntrs (1975)Armalcolite crystalli- -, F. P. Ornuuna AND S. GHosE (1975b) Mineralogy and zation, phenocryst assemblages,eruption conditions and origin petrology of sample 67075 and the origin of lunar anorthosites of eleven high titanium basalts from Taurus Littrow. LS VI, Earth Planet Sci. Lett 26,36-53 6t9-621. McK,rv, D. S., U. S. Cr-nNroN, D A. MonnrsoN AND G. H. Or-srN,E. ANDL. H. FucHs ( 1967)The stateof oxidation of some Ltolt (1972) Vapor phasecrystallizarion in Apollo l4 breccia. iron meteorites.Icarus, 6, 242-253. PLC 3,739-752. AND W C. Fonnrs (1973a)Chromium and phos- McKey, G A., S. J KnrosLs,qucHAND D F. Wsrr-l (1973)The phorus enrichment in the metal of Type II (C2) carbonaceous occurrence and origin of schreibersite-kamaciteintergrowths in chondrites. Geo chim C osmo c h im. A cta, 31, 2037-2042. microbreccia66055. PLC 4,811-818 -, J. S. HurnNrn, J A. V. Doucles AND A. G. P1aNr Mrncten, J-C C. eNo A. Nrcolns (1975)Textures and fabrics of (1973b)Meteoritic amphiboles.Am Mineral.58, 869-872. upper-mantleperidotites as illustratedby xenolithsfrom basalts. PAprKE,J. J. ANDA. E. BsNcE (1972) Apollo l4 inverted pigeon- J Petrol 16,454-487 ites: possible samples of lunar plutonic rocks. Earth Planet. Sci Mnvrn, C., JR, D. H. ANosnsoNnNo J. G. Bnetlsy fi974) Ion Lett 14,176-182. microprobemass analysis of plagioclasefrom "non-mare" lunar ANDD. H. LrNosLev(1974) Mare basaltsfrom the samples.PLC 5. 685-706. Taurus-Littrow region of the Moon. PLC 5,4'll-504. -, R. Bnrrr, N. J. Huselno, D A MonnrsoN. D S C. T. Pnrwrrr, S. SunNo ANDM. C,cvrnoN (1973)Pyrox- McKly, F K. Arrrrx, H. Tlrcroe ANDE. ScHoNDFELD(1971) enes: comparison of real and ideal structural topologies. Z. Mineralogy,chemistry, and origin of the KREEP componenrin K risrallogr. 138, 254-273. soil samplesfrom the Oceanof Storms.PLC 2, 393-411. Pnvr6rvr6, M., P. RltvtoouRAND A Et- Gonesy (1972) Electron Mrvrn, C. E. nNn H. G. Wrlssrne (1974) "Dunite" inclusion in microprobeinvestigations of the oxidation statesof Fe and Ti in lunar basalt 742'75 LS I/, 503-505 ilmenite in Apollo ll, Apollo 12, and Apollo 14 crystalline MeyEn, H O. A nNo N. Z. BocroR (1974)Opaque mineralogy: rocks PLC 3,295-303 Apollo 17,rock 75035 PLC 5.707-7t6. Prnncs, G. W., W. A. Gosr ANDJ. LINDSAv(1975) Reduction of - AND R. H. McCnllrsrnn (1973)Mineralogy and petrology iron and regolith soil maturity. LS V(,634-636 of Apollo l6: Rock 60215,13.PLC 4, 66t-665 -, D. W. SrnlNcway AND W. A. Gosr (1974) Magnetic MrNxrN,J A. eNo E C. T. Cnro (1971)Single crystal X-ray properties of Apollo samples and implications for regolith for- investigationof deformation in terrestrialand lunar ilmenite mation. PLC 5, 2815-2826. PLC 2,237-246. -, R. J. Wrllrnns AND D. S. McKAy (1972) The magnetic Mrsu, K C. nNo L. A. Teylon (1975)Correlation betweennative properties and morphology of metallic iron produced by sub- metal compositionsand the petrology of Apollo l6 rocks. lS solidus reduction of synthetic Apollo I I composition glasses. VI, 566-568. See also PLC 6,615-639 Earth Planet. Sci. Lett. 11,95-104. Moone, C B , C. F. Lewrs nNo J. D. Cnrpr (1974)Total carbon Prcrrrr, A., R. PHtr-ltesAND G. M. BnowN (i972) New zirco- and sulfur contents of Apollo l7 lunar samples. PLC 5, nium-rich minerals from Apollo l4 and l5 lunar rocks. Nature, l 897- I 906 236,2t5-217 MonceN, J. W , R. Gnunp,rrHy, H. HrcucHr, U. KnAsrNnijHr_ Prlr-rNcrn,C. T., P. R. Dlvrs, G. Ecr-rNror,t,A. P. Gowln, A. J. nNo E. ANoens (1974) Lunar basins:tenative characterization T. Jur-r-,J. R. Mexwrlt-, R. M. Housr-By AND E. H. CTRLIN ll 14 J, V. SMITH AND I M. STEELE

(1974)The associationbetween carbide and finelydivided metal- Snro, M., N. L Hlcrl-lNc AND J. E. McLnNo (1973) Oxygen lic iron in lunar fines.PLC 5,1949-1961. fugacity values of Apollo 12, 14, and 15 lunar samples and PowsLL, B. N. (1969) Petrologyand chemistry of the mesoside- reducedstate of lunar magmas.PLC 4, 106l-1079. rites-l Textures and composition of nickel-iron Geochim. ScstjnltnNN, K. AND S. S. HnnNsn (1972) Distinct subsolidus Cosmochim Acta,33,789-810. cooling historiesof Apollo 14 basalts.PLC 3, 493-506. -, F. K ATTKENeNo P. W. WrrslrNl (1973) Classification, Sclen, C B. (1971) Shock-inducedfeatures of Apollo l2 micro- distribution, and origin of lithic fragments from the Hadley- breccias.PLC 2,817-832. Apennine region.PLC 4,445460 - ANDJ. F. Bnurn (1974)Shock-induced melting in anortho- -, AND P W Wrrsr-EN (1972) Petrology and origin of lithic sitic rock 60015 and a fragment of anorthositic breccia from the fragmentsin the Apollo 14 regolith. PLC 3,837-852. "picking pot" (70052).PLC 5,319-336. Pnrwrtr, C T., G. E BnowN eNo J. J. Prrrrr (1971)Apollo 12 S. J. Plcrrnr AND H. A. ALPERIN(1973) Shock clinopyroxenes:high temperature X-ray diffraction studies.PIC effectsin experimentally shocked terrestrial ilmenite, lunar ilme- 2, 59-68. nite of rock fragments in l-l0mm fines (10085,19),and lunar - AND D. R. RorHseno (1975) Crystal structuresof mete- rock 60015.127.PLC 4,841-859. oritic and lunar whitlockites.LS Vl,646-648. Scorr. E. R. D. (1972) Chemical fractionation in iron meteorites PnrNz,M., E. Dowry, K. Krrr ANDT. E. Buncu (1973a)Miner- and its interpretation. Geochim Cosmochim. Acta, 36, alogy, petrology and chemistry of lithic fragments from Luna 20 t205-1236. fines:origin of the cumulate ANT suite and its relationshipsto SH,+NrllNo, T. J. (1975)Electrical conduction in rocks and miner- high-aluminaand mare basalts Geochim.Cosmochim. Acta,37, als: parameters for interpretation. Phys Earth Planel- Interiors, 979-r006 t0,209-2t9. _ AND _ (1973b) Spinel troctolite and SruxrN.T., A. F. NooNAN, G. S Swtrzen, B. M,orsoN,J. A. NEt-sN anorthositein Apollo l6 samples.Science, 119,74-'76. nNo W. G. MELSoN (1973) Composition of Apollo 16 fines Queror, W., V OBERBEcx,T BuNcn euo G. Polrowsrr (1971) 60051, 60052, 64811, 648t2, 67711,67712, 68821, and 68822. Investigations of the natural history of the regolith at the Apollo PLC 4,279-289. 12site.PLC 2,701-7t8. SrruoNns,C. H., W. C PHtNrrv nNo J. L. WanNrn (1974)Petro- RAyMoND,K. N. eNo H. R. Wsrr (1971)Lunar ilmenite (refine- graphy and classificationof Apollo 17 non-mare rocks with ment of the crystal structure) Contrib. Mineral. Petrol 3O, emphasis on samples from the Station 6 boulder. PLC 5, I 35- I 40. 337-353. Rrro, S J B lNo S R. Tevlon (1974)Meteoritic metalin Apollo -, J. L. WnnNen AND W. C. PHrNNrv (1973) Petrology of [6 samples Meteoritics, 9, 23-34. Apollo l6 poikilitic rocks PLC 4,613-632. Rsro, A. M , J. Z. Fnlzrn, H. FuJrrA rr.ro J. E. Evnnsor (1970) SrvpsoN, P. R. AND S H. U. Bowlr (1970) Quantitative optical Apolfo I I samples:major mineral chemistry.PLC 1,749-761. and electron-probestudies of opaque phasesin Apollo I I sam- Rrrl, J. B. Jn. (1972)Olivine-rich, true spinel-bearinganorthosites ples PLC 1, 873-890 from Apollo l5 and Luna 20 soils-possible fragments of the SrrNNsn, B. J. (1970) High crystallizationtemperatures indicated earliestformed lunar crust. Apollo i,5, 154-157. for igneousrocks from Tranquillity Base.PLC 1, 891-895. Rrolrv, W. I., R. Bnrrr, R. J. Wrllu.vs, H. TAKEDAnNo R. W. Surru, J. V (1974) Lunar mineralogy: A heavenly detective story. BnowN (1972) Petrology of Fra Mauro basalt 14310 PLC 3, Presidential Address, Part L Am. Mineral. 59,231-243. I 59- I 70. - (1976) Development of the Earth-Moon system with impli- -, N J Hunreno, J. M. RHoDEs, H. WrrsvrNN eNo B. cationsfor the geologyof the early Earth. In, B. F. Windley, Ed., BeNsnr-(1973) The petrology of lunar breccia 15445and pet- Proc NATO Adu. Study 1nsl. Leicester, 1975. John Wiley and rogenetic implications. J Geol 81, 621-631. Sons. London. RlNcwooD, A E. eNo E EssENE(1970) Petrogenesis ofApollo I I -, A. T. ANDERSoN,R C NswroN, E. J. OlsrN, P. J. basalts,internal constitution and origin of the Moon. PIC 1, WyLLrE,A. V. CREWE,M. S. IsllcsoN nlu D. JoHNsoN(1970) 769-799. Petrologic history of the Moon inferred from petrography, min- Rorpnsn, E. lNo P W. Wprsr-rN(1971) Petrology of silicatemelt erafogy,and petrogenesisof Apollo I I rocks PLC I,897-925. inclusions,Apollo I I and Apollo l2 and terrestrialequivalents. - AND J. B. DnwsoN (1975) Chemistry of Ti-poor spinels, PLC 2,507-528 ilmenites and rutiles from peridotite and eclogite xenoliths. - AND - (1972a') Occurrence of chromian, hercynitic Phys. Chem Earrh, 9, 309-322. spinel ("pleonaste") in Apollo l4 samples and its petrologic - AND I. M. Srnrls (1974) Intergrowthsin lunar and terres- implications. Earth Planet Sci. Lett. 15,376-402. trial anorthosites with implications for lunar differentiation. Am. - AND - (1972b) Petrographic features and petrologic Mineral 59, 673-680. significanceof melt inclusionsin Apollo l4 and l5 rocks.PLC 3, Svvru, J. R. (1974a) Low orthopyroxene from a lunar deep crustal 251-279 rock: a new pyroxene polymorph of spacegroup P2rca. Geophy. - AND - (1914) Petrology of clasts in lunar breccia Res Lett. 1,27-29. Seealso PLC 6,821-832. 6T9tsPLC 5, 303-3r8. - (1974b) The crystal chemistry of armalcolites from Apollo Rosoen,P. L. nNo R F Etrsup (1970)Olivine-liquid equilib- 17. Earth Planet Sci Letr.24,262-270. r\um. Contrib.M ineral Petrol.29, 2'75-289. - (1975)A crystalstructure refinement ofan anomalouslunar Ross,M., J. S.HursNrrn ero E. Dowrv (1973)Delineation of the anorthite.LS vl,756-758. Seealso PLC 6,821-832. one atmosphereaugite-pigeonite miscibility gap for pyroxenes SNnrsrNcrn, K. G., K. Krrl nro T. E. BuNcH (1967) Chromite from lunarbasalt 12021 Am. Mineral 5E,619-635. from "equilibrated" chondrites. Am. Mineral. 52' 1322-1331. Rvorn,G., D B Srossrn,U. B. M,cnvlNlNn J F Bowen(1975) SoNrrr, C. P. (1975)Solar-wind induction and lunar conductivity. Lunargranites with uniqueternary feldspars. PLC 6,435-449. Phys Earth Planet Interiors, 10, 313-322 LUNAR MINERALOGY ut5

SrsEI-p,I. M. (1972) Chromian spinels from Apollo 14 rocks. - AND K C. Mrsne (1975) Mineralogy and petrology of Earth Planet. Sci Leu 14,190-194. Apollo l5rock 15075LS 21,801-803.SeealsoPLC6,l65-179 - (1974) Ilmenite and armalcolite in Apollo 17 breccias.Am. -, D. R. Unln,+NN,R. W. Hopprn ero K. C. Mrsnl (1975) Mineral 59, 681-689. Absolute cooling ratesof lunar rocks basedon the kineticsofZr - (1975) Mineralogy of lunar norite 78235: second lunar diffusionin opaqueoxides: applications to Apollo l5 rocks from occurrence o( P2,ca pyroxene from Apollo 17 soils. Am. Min- Elbow Crater. LS V|,798-800 Seealso PLC 6, l8l-191. eral 60,1086-1091. - ANDK. L. Wrllrll.rs (1973)Cu-Fe-S phasesin lunar rocks. - ANDJ. V. Surrs (1971)Mineralogy of Apollo 15415"Gen- Am Mineral. 58, 952-954. esis rock": source of anorthosite on Moon. Nature, 234, - AND - (1974) Formational history of lunar rocks: I 38- 140. applications of experimental geochemistry of the opaque miner- - - AND (1972) Ultrabasic lunar samples. Nature Phys. als.PLC J, 585-596. Sci 240,5-6. AND R. H. McCnlrrsrrn (1972)Stability relations - - AND (1973) Mineralogy and petrology of some of ilmenite and ulvdspinel in the Fe-Ti-O system and appli- Apollo l6 rocks and fines:general petrologic model of Moon. cation of thesedata to Iunar mineral assemblages.Earth Planet. PLC 4, 5t9-536 Sci Lett 16,282-288. - - AND (1975a)Lunar touchstones:I Minerals,II Rock TAYLoR, S. R. (1975) Lunar Science A Post-Apollo View Per- Types. LS V|,765-770. gamon, New York. - - AND (1975b) Minor elementsin lunar olivine as a Tnetlt-, R J., A G. Pr,rNr ANDJ A. V. Doucrns (1970)Garnet: petrologicindicator. PLC 6, 451-467. first occurrencein the lunar rocks.Science,169, 981-982. -, - ANDL. GRoSsMAN(1972) Mineralogy and petrology Tsly, F AND D. H Ltvr (1974) Ferromagneticresonance studies of Apollo l5 rake samples Apollo 15, 16l-164. of thermal effects on lunar metallic Fe phases. PLC 5, Srswnnt, D B., M Ross, B. A. MoncnN, D E. Appr-Evnl, J. S. 2737-2746. Hue sNsn ANDR. F. Covlrrlu (1972)Mineralogy and perrology -, S L. MANATTrrrp S I. CHnx (1973a)Magnetic phasesin of lunar anorrhosite15415. LS III.726-728. lunar fines: metallic Fe or ferric oxidesl Geochim. Cosmochim. Srorssn, D. 8., U B. MARvrN,J A Wooo, R. W. Wolpr nNo J. Acta,37,l20l-l2l t. F Bowrn (1974) Petrologyofa stratifiedboulder from South D. H. Lrvs AND S. I. CHrN (1973b)Metallic Fe Massif,Taurus-Littrow PLC 5, 355-377. phasesin Apollo l6 fines: Their origin and characteristicsas Surc, C., R. M. Anu-Ero AND R. G. BunNs (19'14)TI3+/Ti'+ revealedby electron spin resonancestudies. PLC 4, 2751-2761 ratiosin lunar pyroxenes:implications to depth of origin of mare UssrlueN, T N (1975)Experimental approach to the stateofthe basaltmagma. PLC 5,717-726. core: Part l. The liquidus relationsof the Fe-richportion of the TAKEDA,H (1973) Invertedpigeonites from a clast of rock 15459 Fe-Ni-S sysremfrom 30 to l00kb. Am. J. Sci.275,278-290 and basalticachondrites. PLC 4,875-885. -, G. E. LorcnrN, C. DoN,qlosoNAND R. J Wrr_lrnvs(1975) -, M. Mry,cN,roroAND A. M. Rrro (1974) Crystal chemical Experimentallyreproduced textures and mineral chemistriesof control of element partitioning for coexisting chro- high titanium mare basalts LS VI, 835-837.See also PLC 6, mite-ulvdsprnel and pigeonite-augitein lunar rocks. PLC 5, '127-741 997-1020 . Wn, C. M rNo C. R. Kxowr-rs (1972)The metal phaseof the Tnnnsov, L S., M A. Nesnnov, I. D SutvelsEvsKy, E. S. Bustee enstatite achondrite. Mineral Mag 38, 627-629 Mnr,cnov ANDV. I. IvANov (t973) Mineralogy of anorthositic Wnlrrn, D., T L. Gnove, J. LorcHr, E. M Srolpsn nNo J F. rocks from the region of the crater Apollonius C (Luna-20). Hr,vs (1973a)Origin of lunar feldspathicrocks. Earth Planet PLC 4,333-349 Sci.Letr 20, ll25-1316. Tnvlon, G. J., M. J Durr, M. E. Helleu, U. B. MnnvrNluo J. -, J. LorlcHI, T. L. Gnovr, E. Srolpen AND J. F HAys A. WooD (1973) Apollo 16 stratigraphy: the ANT hills, the (1973b) Experimental petrology and origin of rocks from the Cayley Plainsand a pre-lmbrian regolith. PLC 4,553-568. Descartes Highlands PLC 4, 1013-1032. - AND D. HrynlNrl (1971)The formation of cleartaenite in E M. Srolern, T. L. Gnovr ANDJ F. Hlys (1975) ordinary chondrites.Geochim Cosmochim.Acta, 35, 175-188. Origin of titaniferous lunar basalts.Geochim. Cosmochim Acta, TAyLoR, L. A. nuo R. H. McCellrsrrn (1972) An experimental 39, t2t9-t235 investigation of the significance of zirconium partitioning in Wllrrn, L S.,B M. FRENcH,K F. J. HErNRrcH,P. D. LowMAN, lunar ilmenite and ulv6spinel. Earth Planet. Sci. Lett. 17, Jn., A. S DolN aN I. Aolsn (1971) Mineralogical studiesof 105-109 Apollo l2 samples.PLC 2,343-358. AND O. SARDr(1973a) Cooling historiesof lunar WANre, H, F Wlorzrn, E. Jecourz ANDF BEGEMANN(1970) rocks based on opaque mineral geothermometers. PLC 4, Composition and structure of metallic iron particles in lunar 8 t 9-828 ,.Rust" "fines."PLC l,931-935. -, H. K Mro rNn P. M. Belr (1973b) in rhe Apollo - AND9 orHsns (1971)Apollo l2 samples:Chemical compo- 16 rocks. PLC 4,829-839. sition and its relation to sample locations and exposure ages,the AND _ (1974a) Identificationof the hydrated two componentorigin of the varioussoil samplesand studieson iron oxide mineralakagan6ite in Apollo l6lunar rocks.Geology, lunar metallic particles PLC 2, | 187-1208. l,429-432. Wlnr. D. A A. F. Rrrn. J. F. Lovrnrrc ANDA. EL Gonssy AND_ (1974b)B-FeOOH, akagan6ite,in lunar . (1973) Zirconolite (versus zirkelite) in lunar rocks. Z.S 12, rocks. PLC 5,743-748. '164-766' -, G KuI-lsnuo ltto W. B. Bnvrr (1971)Opaque mineralogyand texturalfeatures of Apollo l2 samplesand a com- WARNER,J.L.(1972) Metamorphism of Apollo 14breccias.PLC parisonwith Apollo ll rocks.PIC 2, 855-871. 3,623-643. ll t6 J V SMITH AND I M. STEELE

-, C. H- SrvoNos lrqu W. C. PHrNNey(1974) Impact-induced (1972)Twin laws, optic orientation,and compositionof plagio- fractionation in the lunar highlands.PLC 5,379-39'7 clasesfrom rocks 12051,14053 and 14310'PLC -t, 581-589 Wrsn-Ewsrt, P. (1974) Possible magnetic effects due to fine par- Wrur. H.-R., W F. MUrlln ANDG THoMAs (1973) Antiphase ticle metal and intergrown phasesin lunar samples. The Moon, domains in lunar plagioclase.PLC 4,909-923. ll, 30t-3tt. WrlrrNrNc, L L. nro E. ANDERS(1975) Some studies of an WessoN, J. T., C. Cuou nNo W. V. BoyNroN (1975) Temporal unusual eucrite: lbitira. Geochim Cosmochim. Acta, 39, decreaseof siderophile/iridiumratios in mature lunar soils.LS 1205-1210 vr,857-859. Wrllrlns, K. L. nxo L. A. Tnvlon (1974)Optical propertiesand -, R. Scunuov, R. W. Brlo ANDC. CHou (1974) Mesoside- chemical compositionsof Apollo 17 armalcolites.Geology, I' rites I. Compositions of their metallic portions and possible 5-8. relationship to other metal-rich meteorite groups. Geochim.Cos- Wlorzre, F., E. J,ccourz, B. SrrrrEI-, H. BeloeulusrN, A. mochim Acta. 3E. 135-149. BALACEscueNo H WAurn (1972) On lunar metallic particles - ANDC. M . Wnr ( 1970)Composition of the metal,schreiber- and their contribution to the trace elementcontent ofApollo l4 site and perryiteof enstatiteachondrites and the origin of ensta- and l5 soils.PLC 3,1077-1084. tite chondritesand achondrites Geochim.Cosmochim. Acta,34. -, B. Spetrsl nro H. WANrr (1973)On the composition of I 69- t 84. metal from Apollo l6 finesand the meteoriticcomponent. PIC WscHsr-En,B. A., C. T. Pnpwrrr lNro J. J. Plerrs (1975)Structure 4, 1483-1491. and chemistry of lunar and synthetic armalcolite LS VI, Wooo, J. A (1967) Chondrites:their metallic minerals,thermal 860-862 histories,and parent planets.Icarus,6, l-49. Wrrrs, R A (1973) Ferromagneticphases of lunar fines and - (1975) Lunar petrogenesisin a well-stirredmagma ocean. breccias:electron magnetic resonance spectra of Apollo l6 sam- PLC 6,1087-1102 ples PIC 4,2763-2781. -. U B. Mnnvlu, J. B. Rslo, Jn.,G. J. TAYLoR,J. F. Bowrn, -AND D. PnrsrEl (1974)Ferromagnetic resonance properties B. N PowElt- r.No J. S. DIcxev, Jn (1971) Mineralogy and of lunar finesand comparisonwith the propertiesof lunar ana- petrology of the Apollo l2 lunar samples.Smithsonian Astrophy- logues PLC 5,2709-2728 sical Obseruatory Spec. ReP 333. Wsrsr-sN,P W, B. N. PowELLAND F K. ArrrnN (1974)Spinel- Yurun, T nun S. S. Hnrrsn (1974) Cation distribution and bearing feldspathic-lithicfragments in Apollo l6 and l7 soil equilibrium temperature of pigeonite from basalt 15065.PLC 5' '169-784. samples:clues to processesof early lunar crustalevolution. PLC 5,749-767. Yoosn, H. S. Jn (1973) Melilite stability and paragenesis. - AND E. Roroosn (1973) Petrology of melt inclusionsin Fortschr Mineral. 50' 140-173. Apollo samples 15598 and 62295,and of clasts in 67915 and severallunar soils.PLC 4,681-703. Manuscript receiued,January 12, 1976; accepted Wrur. E . A. Glnuser.. H ScHwnllen AND V. TRoMMSDoRFF for publication,May 11,1976.