American Mineralogist, Volume 78, pages360-376, 1993 A concise compilation of petrologic information on possibly pristine nonmare Moon rocks Paur, H. W.LnnBN Institute of Geophysics and Planetary Physics,University of California, Los Angeles,California 90024, U.S.A. Ansrn.q.cr To facilitate systematic study of the surviving compositionally pristine (endogenously igneous)rocks of the ancient lunar crust, a compilation has been generatedof all likely samples,along with key information on the petrologic characteristicsand chemistry of each sample. The compilation includes 260 samples.Besidesinformation related to the likelihood of each sample being truly pristine (i.e., mainly its texture and siderophile element abundances), information is compiled on mineral content, listing major phases presentas well as basic information on mineral compositions, on size (expressed as mass), and on whether a reasonably comprehensivechemical analysis has been published. The compilation also classifies the samplesinto seven categories of confidencein the pristine composition of the samples,reflecting an estimation of the relative likelihood that each arguably pristine sample is in fact pristine. For many purposes, it is crucial to avoid inclusion of polymict rocks in a data base. On petrologic diagramssuch as a plot of average Mg' [Mg' : 100 x molar Mg/(Mg + Fe)] in a low-Ca mafic silicate vs. average An content in plagioclase,rocks in the top three categoriesof the confidence in pristine character appear distinctly bimodal in composition, with roughly half belonging to a ferroan suite characterized by high An despite relatively low Mg'. When samplesof low to moderate pristine characterare included, the bimodality appears lessdistinct. Sample masscan also be important. With a data base restricted to samplesmore massive than I g, there is a clear distinction in feldsparcontent and bulk density betweenferroan and nonferroan (Mg- suite) rocks, such that only the ferroan-suite rocks are likely to have formed as flotation cumulates.With a data baseincluding smaller samples, the samebasic pattern is seen, but only in a blurred form, as the two rock types show considerableoverlap in their modal feldspar contents. INrnooucrroN The roughly 70-km thick nonmare or highland portion of the Moon's crust constitutes at least 99o/o of the total volume ofthe lunar crust and represents the only essen- tially primordial crust available for geologicalstudy (no primordial Earth rocks have been found). This crust has endured countless large meteoritic impacts. As a result, nearly all available rock samples from it have been al- tered by brecciation and melting. Most of the available rocksare polymict: i.e.,lithic masses of finely mixed rub- ble from unrelated sources, usually including minor com- ponents of meteoritic derivation that are clearly detect- able from siderophile-element enrichments. An important distinction can be drawn, at least in principle, between the majority of rock samplesthat are polymict breccias (including impact melt breccias),and the minority that are compositionally pristine, meaning that they survived the meteoritic bombardment with sufrciently limited brecciation and melting suchthat their bulk compositions represent individual, unmixed, endogenously igneous rocks. Polymict breccias can constrain the aggregate char- acteristicsof mixtures of precursor rocks, but, except in 0003-004x/93l0304-0360$02.00 the caseof regolith breccias(where statistical effectsbe- come important), the provenance and the number of components represented by the mixture (e.g., 4, or l0o?) are generally not well constrained. Only pristine rocks can be appropriately interpreted as products ofpurely endog- enous igneous processes, and pristine rocks are clearly essentialfor assaying the original petrologic diversity of the crust. Since the last compilation of known and suspected pristinerocks (Ryder and Norman, 1979), the number of such samples has grown by roughly a factor of three. Also during this period, the ion microprobe has matured into an almost standard analytical tool, and other microanal- ysis techniques such as PIXE have been under steady development, with tremendous potential for application to small samples such as lunar rocks. Clearly, the time is ripe for publication of a new compilation. This paper is the outgrowth of gentle prodding from NASA's Lunar and PlanetarySampleTeam, whoseChairman at the time was John W. Delano. The data compilation is intended to be as complete as possible,in terms of inclusion of all suspectecl pristine nonmare rocks. However, the compi- 360
17
Embed
A concise compilation of petrologic information on ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
American Mineralogist, Volume 78, pages 360-376, 1993
A concise compilation of petrologic information on possibly pristine nonmareMoon rocks
Paur, H. W.LnnBNInstitute of Geophysics and Planetary Physics, University of California, Los Angeles, California 90024, U.S.A.
Ansrn.q.cr
To facilitate systematic study of the surviving compositionally pristine (endogenouslyigneous) rocks of the ancient lunar crust, a compilation has been generated of all likelysamples, along with key information on the petrologic characteristics and chemistry ofeach sample. The compilation includes 260 samples. Besides information related to thelikelihood of each sample being truly pristine (i.e., mainly its texture and siderophileelement abundances), information is compiled on mineral content, listing major phasespresent as well as basic information on mineral compositions, on size (expressed as mass),and on whether a reasonably comprehensive chemical analysis has been published. Thecompilation also classifies the samples into seven categories of confidence in the pristinecomposition of the samples, reflecting an estimation of the relative likelihood that eacharguably pristine sample is in fact pristine. For many purposes, it is crucial to avoidinclusion of polymict rocks in a data base. On petrologic diagrams such as a plot of averageMg' [Mg' : 100 x molar Mg/(Mg + Fe)] in a low-Ca mafic silicate vs. average An contentin plagioclase, rocks in the top three categories of the confidence in pristine characterappear distinctly bimodal in composition, with roughly half belonging to a ferroan suitecharacterized by high An despite relatively low Mg'. When samples of low to moderatepristine character are included, the bimodality appears less distinct. Sample mass can alsobe important. With a data base restricted to samples more massive than I g, there is aclear distinction in feldspar content and bulk density between ferroan and nonferroan (Mg-suite) rocks, such that only the ferroan-suite rocks are likely to have formed as flotationcumulates. With a data base including smaller samples, the same basic pattern is seen, butonly in a blurred form, as the two rock types show considerable overlap in their modalfeldspar contents.
INrnooucrroN
The roughly 70-km thick nonmare or highland portionof the Moon's crust constitutes at least 99o/o of the totalvolume ofthe lunar crust and represents the only essen-tially primordial crust available for geological study (noprimordial Earth rocks have been found). This crust hasendured countless large meteoritic impacts. As a result,nearly all available rock samples from it have been al-tered by brecciation and melting. Most of the availablerocks are polymict: i.e., lithic masses of finely mixed rub-ble from unrelated sources, usually including minor com-ponents of meteoritic derivation that are clearly detect-able from siderophile-element enrichments. An importantdistinction can be drawn, at least in principle, betweenthe majority of rock samples that are polymict breccias(including impact melt breccias), and the minority thatare compositionally pristine, meaning that they survivedthe meteoritic bombardment with sufrciently limitedbrecciation and melting such that their bulk compositionsrepresent individual, unmixed, endogenously igneousrocks. Polymict breccias can constrain the aggregate char-acteristics of mixtures of precursor rocks, but, except in
0003-004x/93l0304-0360$02.00
the case of regolith breccias (where statistical effects be-come important), the provenance and the number ofcomponents represented by the mixture (e.g., 4, or l0o?)are generally not well constrained. Only pristine rocks canbe appropriately interpreted as products ofpurely endog-enous igneous processes, and pristine rocks are clearlyessential for assaying the original petrologic diversity ofthe crust.
Since the last compilation of known and suspectedpristine rocks (Ryder and Norman, 1979), the number ofsuch samples has grown by roughly a factor of three. Alsoduring this period, the ion microprobe has matured intoan almost standard analytical tool, and other microanal-ysis techniques such as PIXE have been under steadydevelopment, with tremendous potential for applicationto small samples such as lunar rocks. Clearly, the time isripe for publication of a new compilation. This paper isthe outgrowth of gentle prodding from NASA's Lunarand Planetary Sample Team, whose Chairman at the timewas John W. Delano. The data compilation is intendedto be as complete as possible, in terms of inclusion of allsuspectecl pristine nonmare rocks. However, the compi-
360
WARREN: NONMARE MOON ROCKS 361
lation is also designed to be compact and accessible. Mygoal has been to produce a compilation that is succinctenough to be scanned, or even thoroughly scrutinized,without a major investment of time. This goal could onlybe met by adopting some rather draconian measures, e.g.,reducing petrographic descriptions of diverse, complexlyidiosyncratic textures to a handful of abbreviations (Ta-ble l). The need remains for a compilation of pristinerock information that is both deep and wide. Such a com-pendium is planned, but it will have to be a bulky doc-ument, and its preparation and publication may take years.
INronulrroN coMPTLED rN TABLE I
Basic descriptive columns
Sample identification. The samples are of three basictypes: (l) A few pristine samples constitute all or nearlyall of the mass of a large solid chunk of Apollo lunarmaterial. Such samples are identified in Table I simplyby the five-digit generic NASA sample number; e.g.,16535. (2) Another type of sample constitutes a smallerfragment of Apollo lunar material, such that the samplehas been identified by NASA (and thus in Table l) as aspecific subsample of a generic regolith sample; e.g.,14141,7069 is a 23-mg particle from regolith samplel4l4l. (3) The third type of sample is a pristine clastwithin an Apollo or lunar-meteoritic polymict breccia. InTable 1, such clasts are identified by the official numberfor one of the samples studied from the clast, modifiedto have a "c" in place of a comma between the genericand specific components of the identif ier; e.g.,MAC88104c97 is a clast that has been studied (in part)using lunar meteorite sample MAC88104,97.
Rock type. The column labeled "Rock type" classifiesthe samples into eight petrologic groups, based on ascheme that has been loosely established by past re-searchers in this field (Fig. l). The single most abundantpristine rock type, the ferroan anorthositic suite (FAS)rocks, are a geochemically distinctive type, readily distin-guished by their anomalous combinations of high-Anplagioclase plus relatively low-Mg' mafic silicates [Mg' :100 x molar Mg/(Mg + Fe)1. In general, FAS rocks alsohave high feldspar contents, and they are widely assumedto be products of plagioclase cumulate flotation over aprimordial magmasphere (e.g., Warren, 1990).
Another distinctive and common rock type is KREEP,characterized by major-element and modal-mineralogicdiversity (from olivine-basaltic to granitic), but by in-compatible trace-element concentrations that are high andin a distinctive pattern of element/element ratios; e.9.,La/Yb = 2.1 x chondrites (e.g., Warren, 199 1). KREEProcks tend to have relatively high contents of silica min-erals and potassium feldspar, but in most cases thesephases are subordinate to pyroxene (which typically isonly moderately Fe-rich) and plagioclase. In some cases,rocks are classified as KREEP even without data on in-compatible elements, because the samples are petro-graphically similar to known KREEPy rocks from the
l Rock Classification Scheme Employed in Table 1\
"Fer roan"? (p lag ioph i le ra t iosvs . MS' ) . . . . . .YF$. . . . . . FAS
Fig. 1 Summary of the simple rock classification schemeadopted for Table I (see text).
same site. The uniform KREEP incompatible elementpattem may reflect a common derivation from the finalresidual melt of the global magmasphere.
Most of the relative few granitic rocks (i.e., rocks richin potassium feldspar and a silica mineral) have an in-compatible element pattem that is rich in heavy rare-earth elements and Th and is markedly different fromKREEP. In Table l, these rocks (but not KREEPy gran-ites) are listed as granitic. This category includes a fewrelatively fine-grained rocks described as felsites in theliterature.
Alkalic suite rocks are apparent intrusives distin-guished by plagioclase with relatively low An (or bulk-rock CalNa ratios), and (where data exist) high concen-trations of incompatible trace elements. Most alkali-suiterocks are highly anorthositic. They may be a subclass (oran extensioh) of the Mg suite (see below), or KREEP-related, or both.
The remaining pristine nonmare rocks constitute a di-verse m6lange of apparent cumulates that are broadlygrouped as the Mg suite. They are subdivided into ultra-mafic, troctolitic, noritic, and gabbronoritic subclassesusing cohventional (mode-based) igneous nomenclature.The Mg-gabbronorites are in many respects the most dis-tinctive of these subclasses (James and Flohr, 1983). TheMg suite may postdate the magmasphere as it appears tobe fundamentally unrelated to the FAS rocks.
Table I also includes a few rocks classif,ed as (?) oreven simply as mare. These are rocks that arguably mightbe fine-grained gabbronorites, or else are unambiguouslymare clasts from highland breccias that were originallydescribed without a clear indication of their mare affinity.
Confidence class. This column classifies the samples onan arbitrary scale reflecting the estimated confidence intheir individual pristine character. This classification isdescribed in detail below.
Mass. For lithic clasts within polymict breccias, massescannot be directly measured. In a few cases masses havebeen estimated by previous workers. More commonly,the description states only the longest dimension of the
362 WARREN: NONMARE MOON ROCKS
Tasl-e 1, Compilation ol information on possibly pristine nonmare Moon rocks
Sample'
Con- Chem- TS Sid. vs.fid- ical pho- cutoff ;
Rock ence Mass analy- to no.type-' class (S) sis? pub? meas-f
7 0 0 3 Y N 0 7 ; 44 0 0 1 1 Y N < 6 4 ; 1 ( N i )8 0 1 5 Y N 0 1 9 ; 4 n o t f o u n d6 0 0 2 1 i m p u r e N . 1 3 ; 2 -
5 0 0 2 1 Y N < 4 9 ; l ( N i )6 0 0 1 5 Y N . 1 4 ; 4
5 0 2 1 Y N5 0 1 N Y6 0.022 Y N <3;1 (Ni)6 0U4 Y , l . ] <3; 1(Ni)6 0 0 2 1 Y Y
1 8 (laths) YY0 4 (laths) N
1? ( laths) Y(<0 5) ( laths) (YY)
relict > 0 31 (laths) YY0 51 1I
1 6
N N Y
N N ?
N N Y
NN (Y)
NN (Y)
I "plutonic" - (Y)? "plutonic - (Y)? ,ptutonic _ (y)
3 0 - ( Y )1 8 - Y1? ( laths) Y NN Y
NN moderateglassy mostly glY Y rightY (Y) mod {ightY - ?Y - ?N N ?Y - l @ $
mod -tight
moderale?Y NN rightNN l@sYY mod {ightYY (N) mod -tight
Y - ext l@seY NN trghtY - tightY - (tight)(Y) Y moderate
YY moderatemod -tqht
(Y) - '?
NN - l@seYY moderateNN l@se(Y) (Y) nod -tight
Y NN tightYY tightY Y . Y t i g h iY N mod {ightNN l@seNN l@s(N) - mod J@se(N) mod J@se
l@seN - moderateNN mod i@seNN l@se
moderate?YY (Y) moderate?Y - moderateNN l@*Y Y tightY (YY) mod -tight
moderate?YY moderateYY - mod l@sYY - mod -l@*
NN l@seNN - l@seNN l@seNN l@seNN l@seNN l@seNN l@seNN - l@seNN l@seNN l@seYY Y tighrYY - tighi
not lound
6 0 2 8 Y Y . 0 9 , 2
5 0 1 N Y7 0 7 Y N 0 1 5 , 35 0 0 3 Y YI 5 3 Y Y 0 0 6 ; 5 n o t f o u n d
"plutonic" - (Y)(Y)(Y)
Y Y Y YY6 0 0 0 4 Y Y
6 0 0 1 5 Y Y6 O I N Y5 0 . 1 Y Y 9 . 1 ( N | )5 , . t N N6 ' . t N N5 0 3 1 Y Y < 6 ; 1 ( N i )4 0 3 4 Y Y 4 4 ; 3 C o / N i > o 45 0 1 1 Y Y 3 6 ; 1 ( N i )6 0 . 1 Y Y < 3 ; 1 ( N i )I 0 6 7 Y Y 0 3 : 5 n o t f o u n d8 0 9 Y Y 0 0 6 ; 5 n o t f o u n d 65 4 2 Y Y 3; I (Ni) not lound 37 0 5 Y Y 1 4 ; 5 < 0 2 % N i 57 3 2 Y Y 0 0 7 ; 5 C o / N a > o 8 67 7 5 Y Y 0 2 3 j 5 6
1 - ( Y )? plutonic" Y (?) (Y)? "plutonic (Y)
? 'plutonic (Y)
I 0043 Y Y7 000001 Y YI 0005 N Y7 00001 N Y7 00007 N Y 1 ? Y Y
1 5 (laths) YYYY
2 Y Y
{<0 5) ( laths) YY
3 > 1 Y3 - Y3 1 5 ( l a t h s ) Y Y3 1 5 ( r a t h s ) Y Y3 -0 5 (laths) YY3 1 (laths) YY3 1 (laths) YY3 1 (laths) YY3 o3(laths) YY3 2 (laths) YY3 1 (laths) YY3 2 (laths?) YY5 2 ( b ' n @ )
3 r e l i c t > 2 -
6 relict 26 - 5 Y Y3 ?1 1 4 "relict
4 1 53 relict 23 relict 23 ?5 "coarse3 1 (laths) YY6 2 35 4 Y Y5 1 2 ( b i n @ )
6 l0 (bintr ) Y5 relict 2?
3 0 62 5 relict6 4
"olutonic (Y)
0 3 (laths) YY NN (Y)0 3 Y ? 10 5 (laths) YY NN (Y)? "plutonic ' (Y)2 9 Y Y Y Y
5 0 1 7 Y Y 2 1 ; 1 ( N i )7 0 U Y Y < 1 5 ; 3 f o u n d , n m 35 0 1 Y Y < 7 a 2 n o t f o u n d 3B 2 5 Y Y 0 0 4 . 4 < 0 0 2 % N r 65 0 0 8 D S A Y - 39 2 6 9 Y Y 0 0 8 ; 3 < 0 5 % N i 64 l t 4 1 Y Y 1 4 ; 4 16 0 1 Y Y < 1 4 ; 3 l o u n d , n m 37 0 12 Y Y <12;2 found, nm 37 0 6 2 \ Y < 1 8 , 2 -
7 0 1 5 Y Y5 0 4 4 Y Y 7 : 1 ( N i )5 0 2 Y Y 9 ; 1 ( N i )5 O 7 4 Y Y 7 : 1 ( N i )5 0 4 1 Y Y < 6 ; l ( N i )5 0 1 0 Y Y 8 : 1 ( N i )5 0 0 5 Y Y <8; 1 {N i )
NN (Y)
NN (Y)
NN (Y)
NN (Y)
NN {Y)
NN {Y)
NN (Y)
NN (Y)
NN (Y)
NN (Y)
(Y)
5 O 2 2 Y Y < 1 2 0 ; 1 ( N i ) -
5 0 1 4 Y Y 5 : 1 ( N i )5 0 3 Y Y 1 0 ; 1 ( N i )5 0 3 2 Y Y <6; I (N i )
7 1 t Y Y 0 2 6 ; 48 1 0 Y Y 0 1 2 : 5 rel ict>l poiki l i t ic? - Y
T r @ t / S 5 1 5 Y YT r o c t 7 3 Y Y 0 1 7 ; 4Nori t ic 9 200 Y Y 008;4 co-r ichTr@t 5 0.1 Y Y <14; 3F A S 4 0 . t Y Y 1 2 ; 3F A S 6 0 . 1 Y Y < 2 ; 3N o r i t i c 5 0 1 Y Y < 1 4 ; 3 n o t t o u n dNori t ic 5 0.1 Y Y <14;3 lound, nmGranit ic 5 0.1 Y Y <12;3 noi loundF A S 6 0 . 7 Y N 0 6 ; 3K R E E P 5 0 1 6 Y YN o r i t i c A 0 3 4 Y N < 0 2 2 ; 4 C o / N i > lNoritic A 006 N Y - CorichF A S 7 4 6 0 0 Y Y - < 1 0 % N iF A S I 1 8 3 6 Y Y O O 4 ; 4Tr@t 6 0.7 N Y >35%NiF A S 8 3 5 5 Y Y 0 0 6 ; 4FAS 5 16 impure N <6;4F A S 7 1 2 0 Y Y 3 3 ; 4F A S 8 3 0 0 Y Y 0 5 ; 1 ( N i )
- YY tight??(Y) YY Y tight
Y Y Y - t i g h t(Y) noderate
- ( Y ) Y Y m o d e r a t e ?- ( Y ) Y - m o d e r a t e ?- ( Y ) Y - m o d e r a t e ?- ( Y ) Y Y m c d e r a t e ?- (Y) YY Y moderate?
NN (Y) NN - l@*- Y Y - t r g h t
(Y) (Y) mask (NN) mod -tight
YY tEht- genomict Y N mod light- Y ? Y Y t i g h t- (Y) YY N right- Y2 YY mod -tight
r e l i c t Y N N t i g h t- Y Y N
WARREN: NONMARE MOON ROCKS 365
TABLE 1. Continued
Sample
ComparisonITE patternvs KREEP
Low-Ca px
Age Plag Mg'(Ga)$ An ratio ratio
Olivine Pigeonite Modal Modal Modal ModalMg' Mg' feldspar olvn or high-Ca low-Caratio ratio volo/" silica? pyrox? pyrox? Ref.ll
Y WARSY WAB2Y JAM3Y JAM3Y RYOIY MAR2Y NOR1Y E B I 1Y RYO1Y WARSY JAMSY JAM5Y E B I 1Y WAR2Y DOW1Y WAR2Y OOW1Y DOWIN DOW2Y OOW1Y WAR3- MAR3Y MAR3Y MAR3Y MAR3Y MAR3Y NOR2Y NOF2
LINlY MCG'IY JAM4Y W A R I 2Y R Y D IY SAL2Y SAL1Y RYO1Y LAU3Y LAU3N WAR3Y WAR3Y JAM1Y ECKIY ECK1_ WAR6Y WAR3- WAR3Y JAM2Y JAM2Y WAB12Y WAR1OY WAR12Y RYD1Y RYO1Y WARg
-. Rock types: FAS : ferroan-anorthositic suite; Alk S. : alkalic suite;Ultram. : ultramafic; Troct. : Mg-suite troctolitic; Troct./S: Mg-suitetroctolitic with Mg-rich spinel; GN : Mg-gabbronoritic. Other abbrevia-tions: Y : yes; N : no; mode r. : modal recombination of mineral anal-yses; DBA: defocused-beam electron probe analysis; TE: trace ele-ments; TS: thin section; comp: composition; nm : not measured;binoc.: based on binocular-micioscopic observation; text. : texture;subo : subophitic; symplec. : symplectites; cataclas. : cataclastic;
mask: maskelynitized plagioclase; mod. : moderately; ext. : extremely;gl. : glass; ITE : incompatible trace element; v : very; FA : ferroan an-orthosite; px or pyrox : pyroxene; Olvn or Ol -- olivine; Si : silica mineral;pn: pnmary.
f Siderophile elements vs. cutoff; number of siderophile elements mea-sureo
+ Siderophile and FeNi-based meteoritic contamination class.$ In the age column, N : Nd; S : Sr, P : Pb (in zircon).ll References: COMI, Compston et al. (1984); DOW1, Dowty et al.
(1974a); DOW2, Dowty et al. (1974b) P5; EBll, Ebihara et al.(1992)P22;ECKI, Eckert et al. (1991); GOOl, Goodrich et al (1984) P15; GOO2,Goodrich et al. (1985) P15; GOO3, Goodrich et al (1986) P16; HAS1,Haskin et al. (1 973) P4; HERI , Hertogen et al. (1 977) P8; HUBl , Hubbardet al (1971); HUB2, Hubbard et al (1974) P5; HUN1, Hunter and Taylor
clast as observed outcropping on the breccia. For thesecases, Table 1 lists an estimated mass, derived by assum-ing that the clast's density is 3.0 g,/cm3, and modeling theclast's volume as orthorhombic, with the longest dimen-sion three times the length of the shortest dimension, andthe intermediate dimension twice the length of the short-est dimension. Masses estimated in this fashion are shownin italics. Of course, some clasts are far from orthorhom-bic, and the smallest dimensions are occasionally muchlarger or smaller than the scale (one-third of the maxi-mum dimension) assumed by this formula. However,based on years of experience with chipping clasts apartfrom lunar breccias, this formula gives a realistic predic-tion of the true mass for most breccia clasts. It should beborne in mind that the clasts chosen for chipping are notan entirely random sampling of breccia clasts: one of thecriteria that motivates the chipping is an apparently largesize, so a bias is introduced in favor of clasts that happento have their minimum dimensions hidden inside thebreccia. Perhaps for this reason, clasts often turn out tobe disappointingly shallow. In any case, by adopting thisuniform formula, the actual reported maximum dimen-sions of the clast can easily by recovered from Table Iby the formula m: (1.5 x M"', where m is the maxi-mum dimension in centimeters, and M is the mass ingrams. For a few clasts where no description of eithermass or maximum dimension is available, a default massof 0.1 g (italicized) is entered in Table l.
Chemical analysis? The column labeled "Chemicalanalysis?" refers to bulk-rock analyses. Unless otherwisenoted, these analyses are complete for all but one or twomajor elements and generally include at least a few traceelements.
Published photo of thin section? This column is in-cluded because some important textural characteristicsare essentially qualitative, and thus verbal descriptionscan be biased by the perceptions (or prejudices) of thepetrographer. In marginal cases, a reader can better for-mulate his or her own opinion if at least one photomi-crograph has been published.
Constraints on possible pristine composition
The various lines ofevidence that can be useful in as-sessing the likelihood of a given sample being pristinewere reviewed by Warren and Wasson (1977). This meth-odology has not changed much, although Ryder et al.(1980) supplied a more comprehensive review and jus-tification of the use of compositional data on FeNi metalto infer whether or not meteoritic contamination ispresent.
The emphasis that Warren and Wasson (1977) placedon siderophile elements has occasionally been ques-tioned, most forcefully and often by Ringwood (e.g.,Ringwood and Wiinke, 1989). Certainly siderophile ele-ments should not be considered proof for or against pris-tinity, in isolation from all other evidence. Indeed, sev-
7653677035c13077O35c22al1As77035c22912007707517721577075t77215771 15c1977539c15782U;178235t7825578235t7825578424;878504;217A527ALHAS I 005c32(ap)ALHAsl00sc3qhFA)ALHA81005c4( F")MAC88104c7("wx1 )MAC88105c86( 'wx2")
MAC88105c97("w2 )
inconclusive
Ytpoor
low Hl,Yb
in@ndusave
low, Yb{ich
low La,Th,Ht
inconclusive
inconclusive
inconclusive
anconclusive
low, Yb{ich
4 38S4 37N
4 34S4 4 N
9 2 693
-'88
9 0 8
9 5 29 5 89 3 8
949 5 89289 5 4
9 4 8
9 6 89 6 9
867 9 078
' 8 1
7 1 2
4 9 2
7 9 28 1
7 9 7
5 2 25257 8 25 4 06 2 5
7060
(-50)(-80)
55
83
8 8 572 7 1
7 0 5olololololN
Y
Y
Y
Y
YYNY
YY
N
Y
ol
ol?
olol
ololol
WAR3ECKlECKlRYDI
63
775 2 2
5 5 07 6 74 0 455
-70
9940
a26750
35
8570
YY, pd
Y, pri
Y, pri
YY, pri
WIN.1W A R 1 2W A R l OWAR3
LAU3LAU3WAR6GOO2GOOl
W A R 1 3WARl3JOL2
(1983) P13; JAM1, James and Hammarstrom (1977) P8; JAM2, Jamesand Mccee (1979) P10; JAM3, James et al. (1984) P15; JAM4, James etal . (1987) P17; JAMS, James et a l . (1989) P19; JAM6, James et a l . (1991)P21; JOL1, Jolliff (1991) P21; JOL2, Jolliff et al. (1991); JOL3, Jolliff eta l . (1991) P21; LAU1, Laul et a l . (1983) P14; LAU2, Laul (1986) P16;LAU3, Laul et al. (1989) P1 9; LlN1, Lindstrom (1 984) P15; LlN2, Lindstromet al. (1984) P15; LlN3, Lindstrom et al. (1988) P18; LlN4, Lindstrom etal . (1989) P19; MA1 , Ma et a l . (1981); MAR1, Mart i et ar . (1983) P14;MAR2, Marvin and Warren (1980) P11; MAR3, Marvin et al. (1987) P17;MAR4, Marvin et al (1 991 ) P21 ; MCG1, Mccee (1 987) P17 ; MORl, Morriset al. (1990) P20; NORl, Nord and Wandless (1983) P13; NOR2, Normanand Taylor (1 992); ROS1, Roseet al. (1975) P6; RYD1, Ryder and Norman(1979); RYD2, Ryder and Norman (1980); RYD3, Ryder (1985); RYDa,Ryderet al. (1988) Pl8; RYDS, Ryder and Sherman (1989); RYD6, Ryder
and Mart inez (1991) P21; SAL1, Salpas et a l . (1987) P17; SAL2, Salpaset al. (1988) P18; SHE1, Shervais et al. (1983) P14; SHE2, Shervais eta l . (1984) P15; SHl l , Shih et a l . (1985); SHl2, Shih et a l . (19. .90); SlMl,Simon et a l . (1983) P14; SlM2, Simon et a l . (1988) P18; STO1, Stof f leret al. (1985) Pl5; TAY1, Taylor and Mosie (1979); TAY2, Taylor et al.(1983); WANl, Wanke et a l . (1975) P6; WARI, Warner et a l . (1980);wAR2, warren and Wasson (1978) P9; wAR3, Warren and Wasson (1979)Pl 0; WAR4, Warren and Wasson (1 980b) P1 1 ; WARs, Warren et al (1 981 )Pl2; WAR6, Warren et al- (1983) P13; WAR7, Warren et al. (1983); WAR8,Warren et al (1983) P14; WAR9, Warren et al (1986) P16; WAR10,Warren et a l . (1987) P17; WARl1, Warren et a l . (1990) P20; WAR12,Warren et a l . (1991a) P21 ; WAR13, Warren and Kal lemeyn (1991); WlNl ,Winzer et al. (19741
eral obviously pristine rocks with siderophileconcentrations well above the cut-offlevel recommendedby Warren and Wasson (1977) have subsequently beenfound (e.g., Warren et al., 1990). However, the sidero-phile cut-off was never meant to be an upper limit sinequa non. For example, Warren and Wasson (1977) clas-sified the 78235 cumulate norite and the 72415 cumulatedunite as pristine, despite slightly elevated siderophileconcentrations, based on textural, mineralogical, and in-compatible trace-element characteristics. Although notself-sufficient or infallible, the siderophile element ap-proach is undeniably a powerful tool for assessing thelikelihood that a given sample is contaminated with ma-terial derived, directly or indirectly, from metal-rich me-teorites (and -92o/o of meteorite falls are metal rich). Datafor unbrecciated mare basalts as well as the few obviouslymonomict nonmare rocks indicate that truly pristine rocksconsistently have far lower levels of highly siderophileelements than typical highland polymict breccias (Haskinand Warren, 1991). Conceivably a lunar breccia mightbe contaminated with meteoritic matter and not by otherlunar materials. However, the lunar surface is almost en-tirely covered to a depth of several meters by powderyregolith. Unless a rock is at the very surface as the brec-ciation process begins, that process can hardly inject me-teoritic matter without also injecting material from theintervening regolith (and its coarser equivalent megarego-lith, which is 2-3 km thick). Thus, cases in which only
meteoritic matter is added during brecciation must beexceedingly rare. The opposite process, formation of apolymict breccia without introduction of a detectable sid-erophile enrichment, is probably more common; yet veryfew extraordinarily siderophile-poor samples do not ap-pear at least possibly monomict.
Textural evidence is harder to summarize concisely,and also hard to assess with complete objectivity, becauseof the complex mix of characteristics that constitute atexture. Recent studies of the Sudbury impact structure(Grieve et al., l99l) and of an Apollo l4 metal-rich rockwith medium-grained silicates of probable impact-meltorigin (Warren et a1., 1991b), demonstrate that only themost coarse-grained lunar rocks (and arguably not eventhese) may be safely distinguished from impact meltproducts on the basis oftexture alone.
At any rate, I will not attempt here to provide a com-plete justification of the relative weighting I attach to sid-erophile elements, various aspects of texture, and otherrelevant criteria. The format of this compilation shouldmake it relatively easy for a reader who is so inclined toadopt his or her own formula for assessment of the like-lihood of pristinity.
Siderophile elements and FeNi-metal compositions.Table I includes a column that records bulk-rock sider-ophile data in an abbreviated form. For the purpose ofconstraining the likelihood that the sample is contami-nated with meteoritic matter, the most relevant datum is
370 WARREN: NONMARE MOON ROCKS
the lowest chondrite-normalized (and reliably measured)bulk-rock siderophile concentration. Although highervalues for other siderophile elements in the same samplemight reflect a meteoritic component with a differentiatedsiderophile pattern, they more likely reflect an indigenouspattern, or even in a few cases laboratory contamination.The table shows the lowest siderophile ratio for eachsample, using an average of all published data for eachsiderophile element in each sample, and using I x l0 o
times CI chondrites as the normalization factor (i.e., 3.3pg/g for Ni, I I pglg for Re, 150 pe/e for Os, 140 pe/e forIr, and 44 pilg for Au). The same column also recordsthe number of these elements determined, because find-ing one out of six elements below the cut-off is slightlyless impressive than finding one out of one. One otherhighly siderophile element that has been determined inmany of these samples, Ge, was not included for thiscompilation, because its concentration might be influ-enced by its moderate volatility.
The next column of Table I records FeNi metal com-positions, which are considered to favor the pristine char-acter of the samples if they are far from the range of mostmetal in the lunar megaregolith, which is primarily de-rived from meteorites, 4-8 wo/o Ni, 0.3-0.6 wt0/o Co; andespecially if the ColNi ratio is much greater than the ratio(0.05) of chondritic meteorites. Table I records metalcompositions as regolithic if they are close to this range.If they are far from it, either the Ni content or the Co/Niratio is given.
The next column gives the siderophile and FeNi class,a summary evaluation of the likelihood that the sampleis meteorite free, based on the combined evidence frombulk-rock siderophile measurements and FeNi-metalcompositions (i.e., the two previous columns). On an ar-bitrary scale, the classes range from a value of 6 for sam-ples with the strongest indications that meteoritic con-tamination is absent, down to a value of I for sampleswith strong indications that meteoritic components arepresent.
Textural characteristics. The column listing maximumgrain size should be self-explanatory. Note, however, thata tiny clast 5 mm across can hardly be expected to havegrains >5 mm. The igneous character column is used toindicate samples that based on textural evidence havebeen interpreted as relatively coarse-grained igneous rocks(clast-poor lunar impact melts might be considered ig-neous too, but their textures are generally fine-grained).The cumulate character column is used to indicate caseswhere the texture reportedly shows features likely to re-flect origin as an igneous cumulate, i.e., a rock formed bygross segregation (fractional crystallization) of crystalsapart from their parental melt. The criteria by which suchtextures are distinguished are essentially qualilative, sopetrologic intuition, and even subjectivity, may be in-volved in classifying some rocks as cumulates. In prac-tice, lunar rocks are sometimes alleged to have relictcumulate textures where the only evidence is coarse granu-larity. The only truly suggestive textural signs of cumulate
origin are coarsely poikilitic, or quasi-poikilitic cumulusframework features (Wadsworth, 1985; Irvine, 1982). Ina typically small and brecciated pristine lunar cumulatesample, such a texture is only marginally discernible, asa few grains of one mineral (of intercumulus or heterad-cumulate origin) that are exceedingly anhedral, next to,and partly enclosing, grains of another mineral (of cu-mulus origin) that are blocky and subhedral to euhedral(perhaps the best example is shown in Fig. I of Warren,1990). However, not all cumulates are markedly poikil-itic (Wager and Brown, 1967), and, not all coarse grainedmafic igneous rocks are cumulates. Also, many fine-grained poikilitic lunar rocks are impact melt products.
The monomict character and cataclastic character col-umns should be almost self-explanatory. Textural indi-cations that a rock is monomict (i.e., clear absence offoreign lithic or mineral clasts) enhance the likelihoodthat the rock is pristine. In a few cases, a sample appearsto be a mixture free of meteoritic matter and limited toa single basic type of lunar rock but nevertheless a mix-ture of significantly different materials. Such samples arelisted in Table I as genomict. The most impressively doc-umented case of a genomict lunar rock is 60025 (Jameset al., l99l). Cataclasis can blunt one of the most pow-erful methods (evaluation of texture) for assessing thelikelihood of pristinity. However, a cataclastic rock mightlack overt textural indications of monomict origin andyet still retain a fully pristine composition, perhaps man-ifested by other traits (e.9., low siderophile concentra-tions).
The granulitic character column is used to register caseswhere the texture shows signs of recrystallization, i.e., anabundance of polygonal, equidimensional grains meetingaI l20 triple junctions. In principle, extensive recrystal-lization might result from a purely closed-system (i.e.,pristine) metamorphic process. However, a thoroughlygranulitic texture raises suspicion that at least for someof the more labile elements, concentrations may have beenaltered by chemical communication with the distant sur-roundings, which in general must include some polymict(nonpristine) materials. Several Apollo 17 granulitic an-orthosites appear to be quasi-pristine (Warren et al.,l99la). In any case, by obscuring the prior texture, ex-tensive recrystallization inhibits textural assessment ofpristinity.
Phase homogeneity. Most of the lunar crust apparentlyformed as igneous cumulates. On Earth, most cumulatesare adcumulates, with highly uniform plagioclase andmafic silicate compositions on a scale of centimeters todecimeters (Wager and Brown, 1967). Most (although notall) of the obvious lunar cumulates are similar. In con-trast, polymict breccias in general have nonuniform min-eral compositions. Hence, one column of Table I is usedto indicate the approximate degree to which plagioclaseand mafic silicates display compositional homogeneity.
Cornparison: Incornpatible element pattern vs. KREEP.This column indicates the degree to which the incom-patible element concentrations, and particularly the pat-
WARREN: NONMARE MOON ROCKS 3't I
tern of ratios among the incompatible elements, are con-sistent with contamination by KREEP. In the area of thecentral lunar near side, where all of the Apollo samplingwas conducted, KREEP appears to be an ubiquitous com-ponent of all highland regoliths, and KREEP is probablyalso dispersed, although not quite so evenly, throughoutthe megaregolith. As a result, incorporation of a minorKREEP component is almost inevitable for any polymictbreccia from the central lunar near side formed by large-scale or near-surface impact mixing. Addition of even aminor component of a material with exceptionally highconcentrations of incompatible trace elements can radi-cally alter the composition for those elements. Thus, theincompatible element pattern can demonstrate that littleor no KREEP has been added, and thus can add somesupport to the likelihood of the pristine character of someprimitive rocks. Figure 2 shows examples of the diverseincompatible element patterns of pristine nonmare rocks(a caveat: among Mg-suite rocks incompatible elementsdo not correlate well with mode-based rock classifica-tions; e.g., Warren et al., l98l). Of course, in many cases,e.g., pristine KREEP (!), the incompatible element pat-tern of the pristine rock is inherently KREEPlike.
A column listing the ITE and KREEP class (not shownin Table l) converts the relatively complex informationin the preceding column into a summary evaluation ofthe likelihood that the sample is KREEP-free. On an ar-bitrary scale (based on a semiquantitative but partly ar-bitrary calculation, too complex to describe here), theclasses range from a value of 6 for samples with thestrongest indications that KREEP contamination is insig-nificant, down to a value of I for samples with strongindications that KREEP is present.
Implausible as mixture? Another less crucial evalua-tion (not shown in Table l) concerns whether, aside fromincompatible and siderophile element constraints, thegeneral composition of a rock may suggest that it is prob-ably at least nearly monomict. For example, ultramaficrocks such as dunite or harzburgite are rare in the lunarcrust and so extremely different from most other crustalmaterials, it seems unlikely that two unrelated ultramaficmaterials would be mixed, without incorporating addi-tional components of more normal (-70o/o plagioclase)composition. The same can be said for extremely graniticrocks, devoid of normal (moderate to highly magnesian)crustal mafic silicates.
Age. Ancient ages provide circumstantial evidence thatthe rock has been involved to a relatively minor extentin impact mixing. Table I records only ages from the Sm-Nd, Rb-Sr, and U-Pb (in zircon) isotopic systems, whichappear to be relatively resistant to resetting by annealingand shock.
Mineralogic columns
Mineral compositional averages. The next three col-umns cite average compositions of the plagioclase, low-Ca pyroxene (orthopyroxene or unspecified low-Ca py-roxene), olivine, and pigeonite within the sample. Note:
E(!
= 100q)
=ILUJl,rJEYY 1 0
oE
o
E6U)
1
Incompatibre =,.'rn,$oliJf't=rtii"T"?"#rlin-^ *^==,Fig.2. Examples of incompatible trace element patterns for
pristine nonmare rocks, showing various degrees of dissimilarityto the normalization composition, average high-K KREEP(Warren, 1991). Samples affected by mixing with typical KREEPymegaregolith materials tend to have flat patterns at levels notgreatly below 103 on this scale. The examples shown are12033,425 (alkalic suite anorthosite), 14321 c1028 (granite; notethat all data for this sample are scaled down by a factor of 10),15295c41 (FAS anorthosite), 15382 (KREEP basalt), 61224,6(Mg-gabbronorite), 7 6 5 3 5 (Mg-suite troctolite), and 7 7 21 5 (Mg-suite norite). Data are from sources listed in Table l, plus thereview of Haskin and Warren ( 1991).
for several samples, hitherto-unpublished electron probedata by the author are included. The most notable casesare 7'1-ll5cl9, an alkalic-suite anorthositic troctolite forwhich no mafic silicate analyses were previously avail-able, and 65785c4, a spinel troctolite that is clearly pris-tine (Dowty et al., 1974b) but has relatively heteroge-neous silicates, particularly plagioclase. The compositionsshown for 65785c4 are averages that include the data ofDowty et al. (197 4b), along with the new data.
Modal mineral content. Table I lists the modal feldsparcontent, averaged from all available descriptions. The ad-jacent columns specify whether olivine, a silica phase, orboth are present in the mode, whether high-Ca pyroxeneis present, and whether low-Ca pyroxene is present, andif so whether it appears to be of primary igneous (as op-posed to subsolidus exsolution) origin. In most cases wherea silica phase is found in a reasonably large rock, a varietyof accessory phases (such as ilmenite, potassium feldspar,apatite, whitlockite, or zircon) are also found; and exceptfor ilmenite, these phases are seldom found in large rocksthat lack a silica phase. The table also lists average Anratios for plagioclase, and Mg' ratios for olivine, low-Capyroxene, and pigeonite. Among the typically small andbrecciated pristine rocks, pigeonite is not always distin-guishable from orthopyroxene. Some of the compositionslisted as low-Ca pyroxene are probably Ca-poor pigeon-ite; the pigeonite column is used for instances where thepetrographic descriptions specify pigeonite, or where theCa contents are too high to be consistent with orthopy-roxene.
372 WARREN: NONMARE MOON ROCKS
References
For the sake ofbrevity, only one reference is cited foreach sample. In general, the most recent work on thesample is listed (a few exceptions were made in caseswhere a later work only barely adds to the informationavailable from earlier descriptions or analyses). Ofcourse,this method omits some important papers and, in a fewcases, the outstanding work on the sample. However, acomprehensive, critical review of the literature is not thegoal here (mainly because it would require a vastly longertreatment). Virtually all of the previous studies should betraceable from reference citations in the works that aredirectly cited here. The majority of the sources cited arepapers in Proceedings ofLunar and Planetary Science, orits predecessors, the Proceedings ofthe Lunar [and Plan-etaryl Science Conference. To facilitate utilization ofTable l, these papers are listed with "P,X-' at the end ofthe citation, where X is the number of the Proceedingsvolume.
TrrB pnosr,nM oF coNFTDENcE rN THEPRISTINE COMPOSITION
A compilation such as this must face the problem ofassigning, to the degree practical, a relative confidencescale for the possibly pristine rocks. When dealing withthe many possibly pristine rocks collectively, we can onlyhope to weed out unlikely pristine rocks by applicationof some type of rating system, quantitative or otherwise.(In this context, "quantitative" simply means involvingmore than two classes: pristine vs. not pristine. Providedthat more than two classes are invoked, then whetherthey are designated by numbers, or letters, or words, isimmaterial.) Some lunar samples are more probably pris-tine than others. If they were not, it would be impossibleto study pristine nonmare rocks because the vast majorityof nonmare lunar rocks are not pristine. The number ofclasses that can practically be resolved is debatable be-cause any such classification system involves great un-certainty (which stems partly from inherent uncertaintyin the methodology for distinguishing pristine rocks fromnonpristine ones and partly frorn ambiguities associatedwith individual samples). I have opted for seven classes,which range on an arbitrary scale from a value of 9 forthe most likely pristine samples, down to a value of 3 forthe most unlikely pristine samples deemed relevant forthis compilation. The four samples of class 3 are onesthat once seemed possibly pristine, but from the presentperspective are probably not pristine. The 70 samples ofclasses 4-5 arc marginal cases, which I recommend beignored in any interpretation sensitive to pristine char-acter. Class 6 comprises 65 samples that I recommend beused for some purposes, but with caution. Classes 7-9can safely be assumed pristine, although the degree ofconfidence increases slightly from 7 to 9. The total num-ber of samples compiled is 260, but six of these are prob-ably mare.
These confidence classifications were derived based on
the information in the middle portion of the table (fromthe Siderophiles column to the Age column, roughly inorder of decreasing importance), by means of a formulatoo complex to describe completely here. A detailed de-scription can be obtained from the author upon request.The single most important factor that determines theconfidence classification is the siderophile and FeNi class,but the various petrographic parameters (including phasehomogeneity) are collectively 1.7 times as important asthe siderophile and FeNi class. The comparison of ITEvs. KREEP plays only a minor role in determining theconfi dence classifi cation.
The reader may want to devise his or her own schemefor translating the information in the middle portion ofTable 1 into a classification for level of confidence inpristine composition. In the final analysis, when evaluatingthe hypothesis that a given sample is pristine, the casemust be judged individually, taking into account the in-finite complexity of the texture, the scope and reliabilityof the available siderophile data, the possible influenceof the bulk composition and petrologic affinity of thesample on its indigenous siderophile concentrations, etc.Nonetheless, I claim that the confidence class assign-ments in Table 1 give a worthwhile, albeit imperfect, in-dication of the strength, and especially the relativestrength, of the pristine composition vis-d-vis individualsamples.
DrscussroN
Importance of confidence in pristine composition:The An vs. Mg' diagram
Figure 3 is a diagram plotting average plagioclase Anratio vs. average low-Ca mafic-silicate Mg' ratio. This isthe classic diagram used to illustrate the anomaly posedby the low Mg' ratios of the FAS rocks relative to oth-erwise comparable Mg-suite rocks, which show a moregeochemically normal pattern of decreasing Mg' accom-panied by decreasing An. The FAS rocks not only deviatefrom that normal trend, they at least arguably do not evenoverlap it. Assessing the degree to which the FAS is geo-chemically distinct from all other components of the lu-nar crust is of crucial importance in terms of distinguish-ing between models that form the FAS as a distinct varietyof flotation cumulates from a primordial magma ocean(e.g., Warren and Wasson, 1980a; James, 1980; Warren,1990) and models that form the entire lunar crust bypiecemeal, serial magmatism (e.g., Walker, 1983; Longhiand Ashwal, 1985). In making such an assessment, thedistinctiveness of the FAS might be obscured if the set ofsamples examined includes a significant proportion thatare not pristine. Such samples have by definition ac-quired their bulk compositions (and mineral composi-tion) by impact mixing, a process that tends to homoge-nize and smear over differences among pristine materials.
Unless impact mixing of the lunar crust has been highlysystematic (impacts are of course random events, butstructure within the target crust should make the mixing
o o Pigeonite
E " u p x;,
. Olivine
o(5.9at,o(UEoo,(Uo
Confidence ) 8 E a
t'tJ Ip
o 8
o - oo - - _ - e
. FE o
cfo
48100 |
WARREN: NONMARE MOON ROCKS
Confidence 2 6
o
52 56 60 64 68 72 76 80 84 88 92 96 100
Fig. 3. Average plagioclase An ratio vs. average low-Ca mafic-silicate Mg' ratio, for data bases using four different thresholdsfor confidence in pristine character (Table 1). Note: A few ofthe pyroxene data plotted as opx might actually be analyses ofespeciallylow-Ca pigeonire (see text).
J / J
100+100
1 0 1
o l48
process somewhat systematic), a diagram such as Figure3 can be expected to show the FAS less distinctly if thedata base includes samples of dubious pristinity than ifthe data base is restricted to samples that are most as-suredly pristine. Figure 3 includes four versions of thesame diagram, using a range of cutoffs for the confidencein pristine composition. In the three versions with con-fidence >6, the FAS is clearly a distinct population ofsamples, and the distribution of points at the high-Anrange of the diagram is obviously bimodal. Only onesample (i.e., one plagioclase-olivine data point and oneplagioclase-opx point) is seen to have approximatelyintermediate mineralogical geochemistry. This is a cu-mulate-textured clast from lunar meteorite MAC88l04(Warren and Kallemeyn, l99l). Including the MAC88l04clast may create an apples to oranges comparison becausewhereas nearly all of the other samples included are froma relatively small region of the central near side high-lands, the MAC88l04 lunar meteorite is from some dis-tant portion of the lunar crust, where all rocks, FAS aswell as Mg suite, might be relatively low-Mg' (consistent
with such a model, a large proportion of the FAS rocksfound as clasts within lunar meteorites are hyperferroan).
The version of Figure 3 with the confidence cutoff re-laxed to > 5 shows a significantly different distribution.The distinctiveness of the FAS is blurred by such plau-sibly but uncertainly pristine samples as 10085,1169(Simon et al., 1983), 15459c279 (Lindstrom et al., 1988),and 76504,18 (Warren et al., 1986). Another factor isprobably also at work, however. These three samples areall uncommonly small. The original mass of 10085,1 169was merely 0.001 g, plus a presumably comparable massconsumed for a thin section. The mass of 15459c279 ]sunspecified, but probably << I g. The original mass of16504,18 was 0.098 g. Unless phase homogeneity is es-pecially tight, samples this small could be grossly unre-presentative in relation to a diagram such as Figure 3.Phase homogeneity is undocumented in the case of10085,1 169, moderate (?) for 15459c279, and moderatefor 76504,18 (in fact, ifa sample is as small as 10085, I 169,the rock is sampled so poorly that its phase homogeneitycan never be well constrained). Data for such samples
Confidence 2 7
g
o
tso
o . d a^ o q E "b " t
a o - E
$
:
Average plagioclase An ratio
52 56 60 64 68 72 76 80 84 88 92 96 100
374 WARREN: NONMARE MOON ROCKS
o.2l00r
Equivalent Diameter (cm)04 0 .9 2 4 I
diagram (e.g., Warren and Wasson, 1979) were simplyhistograms of modal feldspar content. Figure 4 is updatedto include more samples and, by using mass as an addeddimension, show more clearly just how significant themodal variations are. Figure 4 clearly indicates that themodal feldspar contents of FAS rocks tend to be signifi-cantly higher than those of other pristine rocks and thatthe average FAS feldspar content is as high or higher thanpredicted (-85 volo/o) for a series of flotation cumulatesover an appropriately ferroan (i.e., FeO-enriched, andtherefore dense) magma ocean (Warren, 1990). ScanningFigure 4, one can see that if samples smaller than I g areignored, the overlap between FAS modes and other modesalmost disappears. If only samples larger than 3 g (- Icm) are considered, there is no overlap.
Another interesting implication of Figure 4 is that modalfeldspar content tends (admittedly there are many excep-tions) to be higher among troctolitic Mg-suite rocks thanamong noritic and gabbroic Mg-suite rocks. This roughcorrelation might be expected, assuming derivation of allMg-suite rocks from fundamentally similar parent mag-mas, because the troctolitic varieties of Mg-suite rockstend to be more magnesian than the noritic and gabbroicvarieties (note the correlation between An or Mg' and theratio of filled to unfilled symbols among the non-FASsamples on Fig. 3). The proportion of feldspar generated
by cotectic plagioclase and mafic-silicate crystallization isdirectly proportional to the Mg' ratio of the parent meltand also generally higher for olivine + plagioclase crys-tallization than for pyroxene * plagioclase crystallization(e.g., Fig. 3 of Longhi and Pan, 1988). Thus, troctolitic,high-Mg' members of the Mg suite are expected to havehigher modal feldspar, on average, than noritic, low-Mg'members of the suite, as observed (Fig. a).
The larger samples of Mg-suite cumulates virtually allhave feldspar contents lower than predicted for cumu-lates floated over a dense, FeO-enriched magma (Warren,1990). Yet the high Mg' ratios of the Mg-suite rocks,especially the relatively feldspathic troctolitic types, im-ply that the parent magmas had relatively high Mg', andthus their flotation cumulates should be even more feld-spathic, on average, than the FAS flotation cumulates.The conclusion seems inescapable that ar least the ma-jority of the Mg-suite rocks are not flotation cumulates.
ACKNOWLEDGMENTS
I thank (?) my comrades on LAPST for encouraging me to produce this
compilation, and I thank O.B. James and G. Ryder for providing extraor-
dinarily helpful reviews and B. Jollifl U. Marvin, C. Neal, M. Norman,
R Schmitt, and S. Simon for their useful comments. This research was
supported by NASA grant NAG 9-87.
Rnrr:nnNcns crtpn
Compston, W., Williams, I S., and Meyer, C. (1984) Age and chemistry
ofzircon from late-stage lunar differentiates (abs.). Lunar and Planetary
Sc ience .15 .182 -183 .Dowty, E., Prinz, M , and Keil, K (1974a) Ferroan anorthosite: A wide-
spread and distinctive lunar rock type. Earth and Planetary Science
Letters. 24. l5-25.
i{ 80o
rUo 6 0ooo
IL
o ,r0T'o
E6 mF
! htr-o u o o
f ; E r y o t r r c t r . . t rU N U -E t r d D O
U NOO I tr
- t r
I I O l l D a I f ii r . - Y: o" a l t r l
d r r l r l r aItl
.n o _ o a- { t r .
o o r aa o
I o ; o ' t o o a A a r a
D l a
" lrAs oo lMqStocbfitic I
' lllelnq'iuc :lMssgabb0c .
o I MgS ulbarndc .I All€f,c sile oKNEEPy A
I Granruf o
0.1 I 10 100
Mass (grams)1m0 10000
Fig. 4. Modal feldspar plotted as a function of sample mass,restricted to samples with a confidence level of '6 and mass'0.01 g.
should not be entirely ignored, but they should not sim-ply be equated with large, well-sampled rocks known tohave tight phase homogeneity.
Importance of sample size for hypotheses linked tornodal mineral content
Table I includes masses, or estimated masses, for allsamples. Sample volume, or mass, is an important pa-rameter for any application of the pristine rocks that issensitive to their modal mineral content. Obviously, es-pecially for coarse-grained rocks, random scatter can bereduced by working with larger samples or increased byworking with smaller samples. Most of the rocks in ques-tion are cumulates, and cumulates are notoriously het-erogeneous in mode. Terrestrial cumulates commonlyfeature pronounced modal banding, alternating from maf-ic to anorthositic on a scale of meters or even centimeters(Wager and Brown, 1967). Nonetheless, modal variationsamong cumulates can be significant, if a large number ofsamples collectively (statistically) show modes that cor-relate with stratigraphic position in the cumulate pile, orwith cryptic layering. The stratigraphic type of correlationis not a realistic possibility in studies of the availablecollection of lunar rocks. However, the same uncertaintyconcerning the detailed provenance ofthe available lunarsamples makes correlations between modes and solid so-lution variations especially worthy ofstudy because suchcorrelations may help in assessing the likelihood of der-ivation from a common parent magma, e.g., the primor-dial magma ocean or, less ambitiously, in assessing thelikelihood of derivation from a common magma type.
Probably the most important correlation of this type isbetween modal feldspar content and geochemical classi-fication ofthe pristine rocks (Fig. a). Past versions ofthis
0r -0.01
Dowty, E , Keil, K., and Prinz, M. (1974b) Igneous rocks from Apollo 16rake samples Proceedings ofLunar Science Conferences, 5, 431-445.
Ebihara, M., Wolf, R., Warren, P. H., and Anders, E. (1992) Trace ele-ments in 59 mostly highlands Moon rocks. Proceedings of Lunar andPlanetary Science, 22, 417-426.
Eckert, JO., Taylor, L.A., Neal, C.R., Schmitt, RA, Liu, Y.-G., andPatchen, A.D. (1991) Cumulate lithologies and melt rocks from Apollo17 breccias: Correlation ofwhole-rock and mineral chemistry (abs).Lunar and Planetary Science, 22, 333-334
Goodrich, C A., Taylor, G.J., Keil, K., Boynton, W V , and Hill, D H.(1984) Petrology and chemistry ofhyperferroan anorthosites and otherclasts from lunar meteorite ALHA8I005. Proceedings of Lunar andPlanetary Science Conferences, 15, 87-94.
Goodrich, C.A., Taylor, G.J., and Keli, K (1985) An apatite-rich, ferroan,mafic lithology ftom lunar meteorite ALHA81005. Proceedrngs of Lu-nar and Planetary Science Conferences, 16,282-283
Goodrich, C A , Taylor, G.J , Keil, K, Kellemeyn, G.W , and Wanen,P H (1986) Alkali norite, troctolites, and VHK mare basalts from brec-cia 14304. Proceedings of Lunar and Planetary Science Conferences,1 6 , 3 0 s - 3 1 8 .
Grieve, R.A.F., Sttiffier, D., and Deutsch, A. (1991) The Sudbury struc-ture: Controversial or misunderstood? Journal of Geophysical Re-search, 96, 227 53-227 64
Haskin, L.A., and Warren, P.H. (l 99 l) Cbemistry. In G Heiken, D Vam-man, and B.M French, Eds., Lunar sourcebook: A user's guide to theMoon, p. 357-474- Cambridge University Press, Cambridge, U.K.
Haskin, L.A., Helmke, P.A., Blanchard, D.P , Jacobs, J W , and Telander,K (1973) Major and trace element abundances in samples from thelunar highlands. Proceedings of Lunar Science Conferences, 4, 1275-1296.
Hertogen, J , Janssens, M.J , Takahashi, H , Palme, H., and Anders, E(1977) Lunar basins and craters: Evidence for compositional changesof bombarding population. Proceedings of Lunar Science Conferences,8 , l 7 -45 .
Hubbard, N.J , Gast, P W , Meyer, C, Nyquist, L E., Shih, C.-Y , andWeismann, H. (1971) Chemical composition of lunar anorthosites andtheir parent liquids Earth and Planetary Science Letters, 13, 7 l-7 5.
Hubbard, N.J., Rhodes, J M., Wiesmann, H., Shih, C -Y , and Bansal,B M. (1974) The chemical definition and interpretation of rock typesreturned from the non-mare regions ofthe Moon. Proceedings of LunarScience Conferences, 5, 122'l -1246
Hunter, R.H., and Taylor, L.A. (1983) The magma ocean as viewed fromthe Fra Mauro shoreline: An overview of the Apollo 14 crust. Pro-ceedings ofLunar and Planelary Science Conferences, I 3, 591-602
Irvine, T.N (1982) Terminology for layered intrusions Journal of Pe-trology,23, 127-162.
James, O.B (1980) Rocks ofthe early lunar crust Proceedings ofLunarand Planetary Science Conferences, I l, 365-393.
James, O.B., and Rohr, M.K. (1983) Subdivision of the Mg-suite noriticrocks into Mg-gabbronorites and Mg-norites Proceedings of Lunar andPlanetary Science Conferences, 13, 603-614.
James, O.B., and Hammarstrom, J G ( 1977) Petrology of four clasts fromconsortium breccta 73215- Proceedings of Lunar Science Conferences,8,2459-2494.
James, O.B., Flohr, M.K., and Lindstrom, M.M (1984) Petrology andgeochemistry oflunar dimict breccia 6 I 0 I 5. Proceedings of Lunar andPlanetary Science Conferences, 15, 63-86
James, O.B., Lindstrom, M.M., and Flohr, M K. (1987) Petrology andgeochemistry of alkali gabbronorites from lunar breccia 67975. Pro-ceedings ofLunar and Planetary Science Conferences, 17, 314-330.
- (l 989) Ferroan anorthosite from lunar breccia 64435: Implicationsfor the origin and history oflunar ferroan anorthosites Proceedings ofLunar and Planetary Science Conferences, 19,219-243
Jarnes, O.B, Lindstrom, M M., and McGee, J.J (1991) Lunar ferroananorthosite 60025: Petrology and chemislry ofmafic lithologies. Pro-ceedings ofLunar and Planetary Science, 21, 63-87.
Jolliff, B.L ( I 99 l) Fragments of quartz monzodiorite and felsite in Apollo
Jollitr, B.L., Haskin, L.A., and Korotev, R.L. (1991) A ferroan regron ofthe lunar h ighlands as recorded in meteor i tes MAC88104 andMAC88l05. Geochimica et Cosmochimica Acta, 55, 3051-3071
taul, J C (1986) Chemistry ofthe Apollo 12 highland component. Pro-ceedings ofLunar and Planetary Science Conferences, 16, 251-261.
Laul, J.C, Papike, J.J., Simon, S.B., and Shearer, C.K. (1983) Chemistryof the Apollo I I highland component. Proceedings of Lunar and Plan-etary Science Conferences, 14, 139-149.
Laul, J C., Gosselin, D.C., Galbreath, K C., Simon, S.B., and Papike, J.J(1989) Chemistry and petrology ofApollo (17) highland coarse fines:Plutonic and melt rocks. Proceedings of Lunar and Planetary ScienceConferences. 19, 85-97
Lindstrom, M.M. (1984) Alkali gabbronorite, ultra-KREEPy melt rockand the diverse suite of clasts in Nonh Ray Crater feldspathic frag-mental breccia 67975. Proceedings of Lunar and Planetary ScienceConferences, 15, 50-62
Lindstrom, M M , Knapp, S A, Shervais, J W., and Taylor, L.A. (1984)
Magnesian anorthosites and associated troctolites and dunite in Apollo14 breccias. Proceedings of Lunar and Planetary Science Conferences,15 .41 -49 .
Apennine Front revisted: Diversity of Apollo I 5 highland rock types.Proceedings of Lunar and Planetary Science Conferences, I 8, I 69- I 8 5.
Lindslrom, M.M., Marvin, U B, and Mittlefehldt, D.W. (1989) Apollol5 Mg- and Fe-norites: A redefinition ofthe Mg-suite differentiationtrend. Proceedings of Lunar and Planetary Science Conferences, 19,245-254.
Longhi, J, and Ashwal, L.D (1985) Two-stage models for lunar andtenestrial anorthosites: Petrogenesis without a magma ocean Proceed-ings of Lunar and Planetary Science Conferences, I 5, 57 l-584.
Longhi, J, and Pan, V. (1988) A reconnaissance study ofphase bound-aries in low-alkali basaltic liquids. Journal of Petrology, 29, 115-147 .
Ma, M.-S., Schmitt, R.A., Taylor, G J., Warner, R.D , and Keil, K. (1981)
Chemical and petrographic study of spinel troctolite in 67435: Impli-cations for the origin ofMg-nch plutonic rocks (abs.). Lunar and Plan-etary Science, 12, 640-642.
Marti, K., Aeschlimann, U., Eberhardt, P, Geiss, J., Griigler, N., Jost,DT, t -aul , JC, Ma, M.S., Schmit t , R.A, and Taylor , G.J. (1983)Pieces of the ancient lunar crust: Ages and composition of clasts inconsortium breccia 61915. Proceedings ofLunar and Planetary ScienceConferences, 14, 165-17 5.
Marvin, U B, and Warren, P.H. (1980) A pristine eucrite-like gabbro
from Descarles and its exotic kindred Proceedings of Lunar ScienceConferences, I l , 507-52 1.
Marvin, U B , Lindstrom, M M, Bematowicz,T.J , Podosek, F.A., andSugiura, N. (1987) The composition and history ofbreccia 67015 fromNorth Ray Crater. Proceedings of Lunar and Planetary Science Con-ferences. 17 . 47 l-490.
Maruin, U B, Lindstrom, M.M., Holmberg, B.B., and Martinez, R.R.(1991) New observations on the quartz monzodiorite-granite suiteProceedings of Lunar and Planetary Science, 2 l, i 1 9- I 35
McGee, J.J. (1987) Petrology and precursors oflithic clasts from feld-spathic fragmental breccia 67975. Proceedings ofLunar and PlanetaryScience Conferences, | 7, 5 I 3-5 15.
Morris, R.W., Taylor, G.J., Newsom, H.E., Keil, K., and Garcia, S R(1990) Highly evolved and ultramafic lithologies from Apollo 14 soils.Proceedings ofLunar and Planetary Science Conferences, 20, 6l-75.
Nord, G.L., Jr., and Wandless, M.V. (1983) Petrology and comparativethermal and mechanical histories ofclasts in breccia 62236 Proceed-ings ofLunar and Planetary Science Confererices, 13, 645-657
Norman, M.D., and Taylor, S.R. (1992) Geochemistry of lunar crustalrocks from breccia 67016 and the compositioh of the Moon. Geochimi-ca et Cosmochimica Acta, 56, 1013-1024
Ringwood, A.E., and Wiinke, H. (1989) Comments on "Lunar meteorites:Siderophile element contents, and implications for the compositionand origin of the Moon" by P.H. Warren, E.A. Jerde, and G.W Ka1-lemeyn. Earth and Planetary Science Letters, 94, 165-166
Rose, H.J., Jr., Cuttitta, F., Berman, S., Brown, FW, Carron, M.K.,Christian, R P., Dwornik, E.J., and Greenland, L.P (1975) Chemical
376 WARREN: NONMARE MOON ROCKS
composition ofrocks and soils at Taurus-Littrow. Proceedings ofLunarScience Conferences. 5. I I 19-l 133.
Ryder, G (1985) Catalog ofApollo 15 Rocks (Curatonal Publication207 87), 1296 p NASA Johnson Space Center, Houston, Texas.
Ryder, G., and Martinez, R.R. (1991) Evolved hypabyssal rocks fromStation 7, Apennine Front, Apollo 1 5. Proceedings of Lunar and Plan-etary Science, 21, 137-150.
Ryder, G , and Norman, M. (1979) Catalog of pristine non-mare mate-rials (rewised edition), (Curatorial Publication 14565/603),234 p. NASAJohnson Space Center, Houston, Texas.
-(1980) Catalog of Apollo 16 rocks (Curatorial Publication 52-JSCl6904), I 144 p. NASA Johnson Space Center, Houston, Texas
Ryder, G., and Sherman, S.B. (1989) The Apollo l5 coarse fines (4-10mm) (JSC #24035),207 p. NASA Johnson Space Center, Houston,Texas.
Ryder, G, Norman, M.D, and Score, R.A. (1980) The distinction ofpristine from meteorite-contaminated highlands rocks using metalcompositions Proceedings of Lunar and Planetary Science Confer-ences, 1 l, 47 1-479.
Ryder, G., Lindstrom, M M., and Willis, K. (1988) The reliability ofmacroscopic identification of lunar coarse fines particles and the pet-rogenesis of 2-4 mm particles in Apennine Front sample I 5243. Pro-ceedings of Lunar and Planetary Science Conferences, 18, 219-232.
Salpas, P.A , Taylor, L.A., and Lindstrom, M M. (1987) Apollo l7 KREEPybasalts: Evidence for nonuniformity of KREEP. Proceedings of Lunarand Planetary Science Conferences, 17, 340-348.
Salpas, P.A., Lindstrom, M.M., and Taylor, LA (1988) Highland ma-terials at Apollo l7: Contnbutions from 72275. Proceedings of Lunarand Planetary Science Conferences, 18, I 1-19.
Shervais, J.W, Taylor, L.A., and Laul, J C. (1983) Ancient crustal com-ponents in the Fra Mauro breccias Proceedings of Lunar and PlanetaryScience Conferences, 1 4, 17 7 - 192.
Shervais, J.W., Taylor, L A , Laul, J.C., and Smith, M.R (1984) Pristinehighland clasts in consortium breccia 14305: Petrology and geochem-istry Proceedings of Lunar and Planetary Science Conferences, I 5, 25-40.
Shih, C.-Y, Nyquist, L.E., Bogard, D D., Wooden, J L., Bansal, B.M.,and Wiesmann, H. (l 985) Chronology and petrogenesis ofa I 8 g lunarganitic clasr: 14321,1062. Geochimica et Cosmochimica Acta,49,4l1-4t6
Shih, C.-Y., Bansal, B.M , Wiesmann, H., Dasch, E J., Nyquist, L.E., andBogard, D D (1990) Sm-Nd age ofa pristine norite clast from breccia15445 (abs.). Lunar and Planetary Science, 21, 1146-1141
Simon, S.B , Papike, J J., Shearer, C K., and Laul, J.C (1983) Petrologyofthe Apollo I I highland component Proceedings ofLunar and Plan-etary Science Conferences, 14, 103-138.
Simon, S.B., Papike, J.J , and Laul, J.C. (1988) Chemistry and petrologyofthe Apennine Front, Apollo 15 I KREEP basalts and plutonic rocks.Proceedings of Lunar and Planetary Science Conferences, I 8, I 87-20 1.
Stdffier, D., Bischofi A., Borchardt, R., Burghele, A , Deutsch, A., Jess-berger, E K., Ostertag, R , Palme, H., Spettel, 8., Reimold, W.U.,Wacker, K., and Wiinke, H. (1985) Composition and evolution of thelunar crust in the Descartes highlands, Apollo 16. Proceedings ofLunarand Planetary Science Conferences, 15, 449-506.
Taylor, G.J, and Mosie, A.B (1979) Breccia Guidebook No. 3: 67915(JSC #16242), 43 p. NASA Johnson Space Center, Houston, Texas
Taylor, L A , Shervais, J.W., Hunter, R.H., Shih, C.Y., Nyquist, L., Ban-sal, B., Wooden, J, and Laul, J.C. (1983) Pre-4.2 AE mare basalt vol-canism in the lunar highlands. Earth and Planetary Science Letters, 66,J J - 4 t .
Wadsworth, W.J (1985) Terminology of postcumulus processes andproducts in the Rhum layered intrusion Geological Magazine, 122,549-554.
Wager, L.R., and Brown, G.M. (1967) Layered igneous rocks, 588 p.Freeman, San Francisco.
Walker, D. (1983) Lunar and terrestrial crust formation Proceedings ofLunar and Planetary Science Conferences, 14, 17-25-
Winke, H., Palme, H., Baddenhaussen, H., Dreibus, G., Jagoutz, E., Kruse,H, Palme, C., Spettel, 8., Teschke, F., and Thacker, R. (1975) Newdata on the chemistry of lunar sarnples: Primary matter in the lunarhighlands and the bulk composition ofthe Moon. Proceedings ofLunarScience Conferences, 6, 1 3 I 3-l 340.
Warner, R D., Taylor, G.J , and Keil, K. (1980) Petrology of 60035: Evo-lution of a polymict ANT breccia. In J.J. Papike and R.B Merrill, Eds ,Proceedings of the conferences on the lunar highlands crust, p. 377-394. Pergamon, New York
Warren, P.H. (1990) Lunar anorthosites and the magma ocean hypothe-sis: Importance of FeO enrichment in the parent magma. AmericanMineralogist, 7 5, 46-58.
-(1991) KREEP: Major-element diversity, trace-element unifor-mity (almost) (abs.). In G.J Taylor and P H Warren, Eds., Workshopon Moon in transition: Apollo 14, KREEP, and evolved lunar rocks(LPI technical report 89-03), p 149-153- Lunar and Planetary Insti-tute. Houston. Texas
Warren, P.H., and Kallemeyn, G.W. (199i) The MacAlpine Hills lunarmeteorite, and implications of the lunar meteorites collectively for thecomposition and origin of the Moon. Geochimica et CosmochimicaA c t a , 5 5 , 3 l 2 3 - 3 1 3 8
Warren, P.H., and Wasson, J.T. (1977) Pristine nonmare rocks and thenature ofthe lunar crust Proceedings ofLunar Science Conferences, 8,22t 5-2235.
- (l 978) Compositional-petrographic investigation of pristine non-mare rocks. Proceedings ofLunar and Planetary Science Conferences,9, 185-217 .
-(1979) The compositional-petrographic search for pristine non-mare rocks-third foray Proceedings of Lunar and Planetary ScienceConferences, 10, 583-610
- (l 980a) Early lunar petrogenesis, oceanic and extraoceanic. In J.J.Papike and R B Menill, Eds , Proceedings of the conferences on thelunar highlands crust, p 81-99. Pergamon, New York.
- ( I 980b) Further foraging for pristine nonmare rocks: Correlationsbetween geochemistry and longitude Proceedings of Lunar ScienceConferences, | 1, 431-470.
Warren, P H , Taylor, G.J., Keil, K., Marshall, C., and Wasson, J T. (1981)Foraging westward for pnstine nonmare rocks: Complications for pet-rogenetic models. Proceedings of Lunar and Planetary Science Confer-ences, 12, 2 l -40
Warren, P.H., Taylor, G.J., and Keil, K (1983) Regolith breccia AllanHills A81005: Evidence oflunar origin, and petrography of.pristineand nonpristine clasts. Geophysical Research Letters, 10, 7'79-782.
Warren, PH, Shirley, D.N., and Kallemeyn, GW (1986) A potpourriof pristine Moon rocks, including a VHK mare basalt and a unique,augite-rich Apollo I 7 anorthosite. Proceedings of Lunar and PlanetaryScience Conferences, 16. 319-330.
Warren, P H., Jerde, E.A, and Kallemeyn, G.W. (1987) Pnstine Moonrocks: A "large" felsite and a metal-rich ferroan anorthosite. Proceed-ings ofLunar and Planetary Science Conferences, 17, 303-313.
- (1 990) Pristine Moon rocks: An alkali anorthosite with coarse au-gite exsolution from plagioclase, a magnesian harzburgite, and otheroddities. Proceedings of Lunar and Planetary Science Conferences, 20,3 l - 59 .
-(l99la) Pnstine Moon rocks: Apollo l7 anorthosites. ProceedingsofLunar and Planetary Science, 21, 5l-61
-(l99lb) A 4 4-gram, mostly metal rock from Apollo l4 with at-tached, cogenetic mafic silicates (abs.) Proceedings oflunar and Plan-etary Science, 22, 1469-1470.
Winzer, S.R., Nava, D.F., Schuhmann, S., Kouns, C.W, Lum, R.K.L,and Philpotts, l.A. (197 4) Major, minor and trace element abundancesin samples from the Apollo 17 Station 7 boulder: Implications for theorigin of early lunar crustal rocks. Earth and Planetary Science Letters,23,439-444.
MeNuscrrpr REcETvED DecsMspn 31, l99lMeNuscrrpr AccEprED DpceMsen l. 1992