A neric att M in er atro oistVol. 57, pp. 1729-1i47 (t972)

SPINEI,S AND THE PETROGENESIS OF SOME APOLLO12 IGNEOUS ROCKS1

F. D. BuscrrE, MARTTN Pnrmz, AND KLAUS Kn:rr, Department ofGeologg and Institute of Meteoritics,(Jruiuersity of New Merico,

Albuqrcrque, New Merico 87106

AND

T. E. Buwcn, Space Sci.ences Diuision,l'r,4S,4 Ames ResearchCenter, Mofiett Fi"eld, California 9[036

AssrnAcr

Spinels in four Apollo 12 igneous rocks have been analyzed with the electronmicroprobe. Spinel composition is complex and reflects solid solution betweenfive end members that can best be represented in a modified spinel prism in whichTia replaces Fea. Some rocks (12036, 12038) contain two main spinel groups withdivergent compositional trends. In addiiion, 12086 contains rare hercynite-richspinels in pyroxene melt inclusions; these have not previously been reported inIunar samples. Rock 12039 contains only one ulvtispinel-rich late spinel group.Rock 12051 has spinels which are often continuously zoned from chromite- toulviispinel-rich end members with no compositional discontinuities. Conclusions:(1) A three-dimensional representation of the complex spinel solid solutions inApollo 12 rocks and a knowlefue of textural and paragenetic relationships ofspinels and coexisting phases are necessary for interpreting their petrogeneticsignificance. (2) The presence of two groups of spinels with divergent trends,commonly found in Apollo 12 igneous rocks, is the result of crystallization ofearly chromite-rich spinel and olivine, terminated by a peritectie reaction ofthe type described by frvine (1967), and later crystallization of ulviispinel-richspinel, either simultaneously with, or later than, pyroxene. The ,,gap" in com-position between the two spinel groups is the result of a change in composition ofthe melt during the course of crystallization. fu rock 12039, complete resorptionof early spinel (and olivine) may have resulted in the formation of only one,late spinel group. In 12051, rapid crystallizaiion of early chromite-rich spinelsimultaneously with pyroxene, rather than olivine, results in continuously aonedcrystals, trending towards the FegTiOn end member. No divergence in composi-tional trends is noted, apparently because there is no intervening peritecticreaction. (3) No evidence is found in support of several other hypotheses sug-gested in explanation of the divergent Apollo 12 spinel trends, zuch as thepresence of an immiscibility field, or reaction of early chromite-rich spinel andmelt to yield late ulviispinel-rich spinel. (4) Melt of the composition of rock120i18 could not be the parent of an olivine-pyroxene cumuJate rock suchas 12036 (or 1X)40) because early 12038 spinels differ markedly from early12036 spinels. flowever, a melt intermediate in composition between 12036and 12038 could be parental to both rocks. The residual melt of this parent,

l This paper was first presented as a talk before the Annual Meeting of theGeological Society of America, November l-3, l9fl, Washington, D. C.

1729

1730 BUSCHE, PRINZ, KEIL, AND BUNCH

after olivine and pyroxene crystal settling, would be enriched in feldsparcomponents and this could explain the unusual major and trace element com-position of rock 12038.

INrnonucrrox

Irvine (1965, 1967) has shown that the compositions of chromianspinels may be used as indicators of the physieochemical conditions offormation for terrestrial ultramafic rocks, especially if the mineralcontent and the composition of coexisting phases is known. Jackson(1969) studied the compositional variations in coexisting chromiteand olivine in chromitite zones of the Stillwater complex, using a simi-Iar approaeh, and was able to achieve much insight into conditions ofcrystallization.

In Apollo 12 igneous rocks, spinels ranging in composition fromFeCrzOe to FezTiOe @.9., Haggerty and Meyer, 1970) are commonlyfound as accessory minerals. This is in contrast to Apollo 11 igneousrocks where these phases are sparse (Agrell et a1.,1970). Others whohave analyzed and/or discussed the implications of Apollo 12 spinelsinclude BreLt et dI. (1971), Brown et al. (1971), Cameron (1971),Champness et aI. (1971\, Drake et aI. (1970), El Goresy et aI. (1971) ,Gibb ef al. (1970), Haggerty (1971a,1971b, 7972),Keil et aI. (1977),Kushiro et aL. (1970), Kushiro and Haramura (1971) , Muan et aI.(1971, lg72\, Reid (1971), Simpson and Bowie (1971) ,Taylor et aI.(1971), Walter et aI. (1971), and Weill et al. (197I) .

In surveying this literature it becomes apparent that several prob-lems with regard to Apollo 12 spinels require further clarification. Oneproblem is the graphical portrayal of the complex compositions ofspinels which may be represented by five components: FeCrzO+ (Cm),MgCrzOe (Pc), FeAl2O+ (Hc), MgAl2Oa (Sp), and FezTiO+ (Uv) (theabbreviations will be used throughout this paper). Any graphicalmethod of porbrayal which stresses only two of these end members(e.g., Cm and Uv) , as was commonly done in the cited literature, losesbhe subtle distinctions that may exist in spinel compositions within agiven rock and/or between rocks. Haggerty (1971a) also points outthat spinels are more eomplex than initially realized and their repre-sentation by a binary solid solution series between Cm and Uv isinvalid.

Another unresolved problem is that of the presence or absence of agap or discontinuity (Haggerty and Meyer, 1970; Haggerty, 1971a)in the so-called chromite-ulviispinel series of Apollo 12 spinels. Hag-gerty (1971a) suggests that a distinct discontinuity exists betweenUvro and Uvos in the chromite-ulviispinel series of Apollo 12 spinels,

and that rew points

^::':':':::';":.:::r"

n""".n"*';::Apollo 12 rocks contain spinels whi.h do not have a compositional gap

are Brown et aI. (1971), El Goresy et at- (L97I), Keil et al' (1971) 'Reid (1971), and Walter et al. (1971). Haggerty (1972) has shown

that no compositional gap is present in Apollo 15 sample 15555'

The significance of th'e gap or discontinuity in the composition of

Apollo 12 spinels has been interpreted in several ways: (1) as a

miscibility gap in the chromite-ulviispinel series (Haggerty and Meyer,

1970; Walter et a1.,1971); (2) a peritectic reaction of the early Cm-

rich spinels with the melt resulting in the formation of later uv-richspinels (Brett et a1.,7971; El Goresy et at.,1971; Kushiro et al',1970;Kushiro and Haramura, 1971) ; (3) a combination of hypotheses (1)

and (2) (Taylor et u1.,1971); (4) termination of early crystallizationof cm-rich spinel and olivine by a peritectic relation, as outlined by

Irvine (1967), when a new phase (pyroxene) crystallizes'(Cameron,1971; Champness ef al., 7971; Gibb et at., 1970; Weill ef ol., l97L);(5) a combination of hypiitheses (2) and (4) (Haggertv, 1972); (6)

termination of crystallization of cm-rich'spinel as a result of rapid

change of physicochemical conditions accompanying eruption of the

magma to the surface (Weill et aI.,l97l); and (7) as the result of a

very steep ,,liquidus slope" of equilibrium crystallization during the

interval in which the liquid is saturated with spinels of these com-positions (Weill et at.,I97t).

The purposes of this paper are' therefore, as follows: (1) to find a

suitable method of plotting the complex compositions of lunar spinels

such that relationships within and between rocks can be observed and

described; (2) to present spinel data for four Apollo 12 igneous rocks(12036, 12038, 12039, 12051) which have different spinel populations,

and to relate the spinel compositions to the petrology of each sample;

and (3) to discuss the significance of the compositional gap or discon-

tinuity for each sample and the insight it may provide for the petro-

genesis of each rock.

Axer,'vrrcer, Motnops

spinels were analyzed with an ARI-EMX electron microprobe using chemi-

callyanalyzed minerals and synthetic glass for standards: CrnO", ALO", and V"O"

were analyzed using chromite; FeO, TiO", MgO, and MnO were determined

using ilmenite; and cao was analyzed using a synthetic basaltic glass. All analyses

were corrected for drift, deadtime, background, mass absorption, secondary

fluorescence by characteristic radiation, and atomic number using procedures

described by Keil (1967). Because of the inhomogeneity exhibited in some lunar

spinel grains (especially in rock 12051), great care was taken to return to the

same anillytical points for all elements on successive analyses of the same grain.

since no wet-chemical bulk analyses of rocks 12036 and 12039 afe available, thin

1732 BUSCHE, PRINZ, KEIL, AND BUNCH

sections of those samples were analyzed for bulk composition using the broadelectron bearn technique described by Prinz et aI. (lgIL).

PEnssNrarroN oF DATA

The complex compositions of lunar spinel solid solutions can bestbe represented in a triangular prism, modified after Irvine (1965)(Fig. 1). The major modification is that Fes* is replaced by Tia. alongthe upper edge of the prism, giving rise to the components MgeTiOrand FezTiOa. This modification appears to be justified because Ti is

Frc. 1. The spinel prism of trrvine (1965), modified to 6t the compositions oflunar spinels. Major modifications are the substitution of M9TiO. and FerTiO.for FegFeoi and Mg,Jeor which corresponds to substitution of ria* for Feu*,Projections to sides A, B, and C are shown.

i,{roo

\

o.5 t.or /Fe+Mo+Mg

APOLLO 12 IGNEOAS NOCKS

a major constituent of lunar spinels and no evidence for the preseneeof tr'e3' has been found in our data (in contrast with Haggeriy andMeyer, 1970).

The length axis of the prism represents the relative proportions ofFe and Mg, with the front of the prism being Mg-rich and the rearbeing Fe-rich. A triangular cross section gives the relative proportionsof Al, Cr, and Ti for a given Fe/Fe * Mg ratio. The small arnountsof Mn and Ca were added to Fe, and V was added to Cr. The sixcorners correspond to each of the components; MgrTiO+ is not neededin our calculations and, therefore, spinel analyses are recalculatedinto five components (Table 1).

In order to completely describe a spinel analysis, a point P (rep-resenting t'hat analysis) is projected to three sides of the prism, asshown in Figure 1. These projections are shown as sides A, B, and Cand all analyses are plotted as mole ratios since end members arecalculated. A projection to side A is made by plotting 2Ti/(Cr +Al + 2Ti) against Fe/(Fel+ Mg) which projects the point P to'sideA horizontally, parallel to the base. A projection to side B is madeby plotting Crl(Cr + Al) against Fel(Fe '* Mg) and the point Pprojects vertically downward to the base. A projection to side C ismade by plotting mole percentages of 2T| Cr, and Al (recalculated to100 percent) in a triangular plot and this projects the point P hori-zontally, parallel to the base. Since no spinels were found with aFe/ (Fe l* Mg) ratio less than 0.5, the lefthand sides of the projections

t .b l" 1, t l . "at ." - t

Oxides ( i le ight Percenr)

1733

Gloup No,

T 1 O 2A].2o 3C t 2 0 3v 2 0 3IeoMn0MgoCao

T o E a l

I T I I I I I I In le r - I I

I mediateRock 12036 I Rock 12038

I5 , 7 2 6 . 4 1 . 1 8 | s . o 2 r , 8 3 q . L

1 0 . 1 4 . 1 s s . o I r 7 . 1 6 : 7 2 . 9 04 s . 6 1 5 . 7 4 . 6 I 4 r . s 1 7 . 1 5 . 8l . o 1 o . 6 ' J 0 . 0 6 I 0 . 9 8 0 . 5 8 0 . 4 1

z g , t 4 8 , 3 1 3 . 4 | 2 1 . 6 4 8 ' 6 6 2 . \0 . 2 8 0 . 3 3 0 . 1 5 i 0 . 2 6 0 . 3 4 0 . 3 27 . 6 4 . 0 5 . 6 I 1 . 4 3 . 3 o , o 40 . 4 2 0 , 2 2 0 . 4 6 I 0 . 3 6 o . 4 4 o . t 3

I98 .81 99 .68 1o0.46 | roo '20 98 '85 101.80

Splnel ComPonents*(oole percent)

Rock 12051

6 . 5 2 7 . 7 3 2 . 81 0 . 8 5 . 2 2 . 4 84 2 . 8 U . 3 0 . 0 30 . 7 4 0 . 4 A O . 2 2

36.8 53 .2 62 .20 . 2 3 o . 2 4 0 . 2 51 . 3 0 0 . 4 6 0 . 2 10 . 3 1 0 . 3 9 0 . 4 7

9 9 . 4 8 1 0 0 . 9 7 9 8 . 7 6

6 0 . 3 2 5 , 2 1 . 0

1 6 . 6 8 . 5 5 . 75 . 9 2 , 4

r7 .2 63 .9 93 '3

Rockr2039

3 1 . 5

2 , 2 6o . 4 7

5 3 . 50 . 2 8o . 2 Io , 2 0

100.92

3 . 7

8 8 . 6

CnP cH CS pUv

5 6 . 1 5 , 4 5 , 38 . 0 1 7 . 8

t ) . 22 0 , 9 4 . L 1 8 . 21 5 . 0 7 2 . 7 1 . 3

54.4 25 ,9 8 .80 . 0 1

- 5 . 5 0 ! {

3 3 . 1 8 . 9r2 .4 59 .7 84 .8

*See tex ! fo r method o f ca lcu la t ion

Fo i 72 ind iv idua l ana lyses see Bus9he e t 41 . (1971) .

1734 BUSCHE, PBINZ, KEIL, AND BUNCH

A and B are shown as dashed lines, and are omitted in subsequentdiagrams (Figs. 2-4). The orientations of sides A, B, and C are main-tained in Figures 24for ease of viewing.

Haggerty (1971a) has made a similar modification of the spinel prismof Irvine (1965) for terrestrial spinels by substituting Tia* for Fe'*.However, his MSALOo-FeAlrOn and MgCrrOn-FeCrrOn edges are thereverse of oursl he has projected to the same three sides but has usedweight percentage oxide ratios rather than mole ratios; and his projectionto the_triangular rear of the prism is presented as a plot of TiOr/(AlrO, *Cr,O,-'* TiO,) against Cr,O,/(AlrO, * Cr"Oe) (weight percentages).Thus, his plots cannot be directly compared with ours, although pointsin sides A and B have the same Eeneral orientation.

F Some selected spinel analyses"are presented in Table 1 and the endmembers are given below each analysis. These end members werecalculated as follows: (1) all Mn and Ca are added to Fe2*, and V isadded to Cr, as is the case for the data plotted in Figures 2-4. (2)FerTiOa is calculated using up all available Ti; excess Ti never occurredand therefore MgrTiOn was not calculated; (3) FeCrrOa is calculated;any excess Cr is calculated as MgCrrOn; (4) FeAlrOn is calculated; anyexcess Al is calculated as MgAlrOa. A small excess sometimes remainsbut it is generally less than 0.5 mole percent and is ignored.

Sprmrr, CoupoSrrroxs AND Pprnor,ocrcel RplerroNsurpsOne hundred fifty spinels have been analyzed in four Apollo 12

rocks (12036, 12038, 12039, 12051). Seventy-two eomplete analyses,including structural formulae, have previously been published (Buscheet al., 1971). The remaining 68 spinels have be?n analyzed foi Fii;Ti, Cr, Al, and Mg only and these data are included in Figures24. It should be noted that, becaiise of the small concentrationsof Ca, Mn, and V, a complete analysis [ (Fe * Mn * Ca) /(Fe * Mn * Ca * Mg), 2Til(Cr +- V + At + 2Ti), and (Cr + y)/C" + V + A!)] versus an incomplete analysis [Fe/(Fe * Mg),2Ti,/(Or.,* Al + 2Ti), and Cr/(Cr * Al)1, on the scales of Figures 2-4,plot very elose to each other. Hence, the use of complete and ineom-plete analyses in the'same' diagram does not alter the spinel com-positional patterns.

rn the following section spinel compositions are related to texture,mineral content and composition, and buli< eomposition of each rock.

Rock 120s6 !i I !:'

Rock 12036 is a feldspathic peridotite and has been studied in detailby Prinz et al. (in prep.).This rockis a complex cumulate with early

APOLLO 12 IGNEOUS ROCKS

o8

re/re+us

r2036.9t2036,tor2036,11

Frc.2. Compositions of spinels from rock 12086 plotted as projections to sidesA, R, and C of.the spinel prism shown in Figure l. (in mole,ratios). G:roup I,II, and III spinels are outlined by dashed lines. .

t7ffi

+

L\

o o.4

Fdt* o.e

+. a t

i : ' o 4AI

AICr

,iir,

uu

ri::.;

1736 BASCTIE, PRINZ, KEIL, AND BANCH

+

o o.4

tr1o. .?

+

i:'o'4AI

t r (

..]----;. -),,

' r t- t--------"

,c--r., /r- ; i

ca.:;-----)/ |I l

oiI

0.6 0.8

Fe,/Fe+Mg0.6 0.8

Fe./Fe *'M.g

r2038.63t2039,3

Frc. 3. Composition of spinels from rocks 12038 and 12039 plotted as projec-tions to sides A, B, and C of the spinel prism shown in Figure 1 (in mole ratios).Group I and II spinels in rock 12038 are oritlined by dashed lines.

- 0.6

+

s\o o.r

trNr- 0.6

+

i: o.r(\t

APOLLO 12 IGNEOUS ROCKS r7e7

Fe/Fe +Ms

t205t,59

Frc,4. Compositions o{ spinels from rock 12051 plotted as projections to sidesA, B, and C of the spinel prism shown in Figure I (in mole ratios). Ranges incomposition of individual spinel grains are indicated by solid lines but, for rea-sons of clarity, are only shown in the projection to side A (Fig.4A).

AICr

af

1738 BUSCHE. PRINZ. KEIL. AND BUNCH

crystallized Cm-rich spinel and olivine (Foun-oo) poikilitically enclosedin chemicallSr homogeneous pigeonite-augite megacrysts up to 5 rnmacross. These megacrysts are incorporated into a second stage as-semblage; this assemblage includes cumulus olivine (Forr*u), followedby plagioclase (Ane6-77), spinels intermediate between Cm and Uv,chemically inhomogeneous pyroxene which rims the early pyroxenemegacrysts and, finally, accessory minerals. A bulk analysis of threethin sections of this rock by broad beam electron microprobe tech-niques is listed in Table 2 and in general is similar but somewhatmore mafic than the analysis of rock 12040'(Kushiro and Hararnura,1971). Modally, the rock contains 24 percenL olivine, 58 percentpyroxene, 12 percent plagioclase, 5 percent opaques, and 0.7 percentmelt inclusions (in volume percentages) (Prinz et aI., in prep.).

The composition of spinels is shown in Figure 2, representing analy-ses of grains in three thin sections. Spinels of this rock may be dividedinto three groups (Fig. 2). Group I spinels are Cm-rich but alsocontain significant proportions of Sp, Uv, and Pc (Table 1) . They areeuhedral, often intergrown octahedra that are chemically homogeneousand represent cumulus phases. This group presumably represents a

Table 2: BuIk composlt ions of four APo1lo 12 igtreous locks( ln weight percentages)

112036

2 3 412038,24 12039,3 12057,34

47.05 47 .3 45 .543 . 2 2 3 . 0 4 4 . 1 4

7 2 . 1 2 l O . 7 9 , 9 50 , 3 4 0 . 2 3 0 . 3 6

L7.9r 2 r . l 20 .L90 . 2 4 0 . 1 9 0 . 2 87 . 0 9 5 . 3 6 , 8 2

1L.46 L2 .L 11 . 330 , 6 4 0 . 3 6 ) ' 3 10 . 0 7 0 . 0 9 0 , 0 40 . 0 2 0 . 0 9 0 , o 7n . d . n . d . 0 . I 0n . d . n . d . 0 . 0 3

Sanple No.

S 1 0 2TiO2A120 3Cr?0 'Fe0MnoMgcCa0Na20K20P 2o5S1120-

Sub total

L e s s O = S

TOEAA

4 2 , 22 . 5 1

0 . 5 92 2 . 70 . 2 7

r 7 , o7 , 80 . r 70 . 0 20 . 0 6n . d .n . d .

9 9 . 5 8

9 9 , 5 8

9 9 . 7 61 0 0 . 1 6 1 0 0 . 5 0

0 ' 0 5

1 0 0 , 1 6 1 0 0 . 5 0 9 9 . 7 r

n.d. = not deter0ined

1. Broad beam electton D. lcroprobe analysis; avelage of thlee

t h i n s e c t i d d s ( 1 2 0 3 6 , 9 i L 2 O 3 6 , L 1 t 1 2 0 3 6 , 1 1 ) ( P r l n z e t a f . '

i n p r e p . )

2, Kushito and Baranura (1971)

3. Broad bean electron nicroptobe analysls of one thin sect ion(this paper)

4. Mase11 and l l t lk (1971)

APOLLO 12 IGNEOUS ROCKS 1739

solid solution series, with the earliest spinels having the lowest Fe/(tr'e * Mg) ratios. These spinels crystallized essentially simultane-ously with early olivine (Foun--uo) and vary mainly in Fe/ (Fe + Mg)and relatively little in CrlCr * Al) and 2Ti/ (h + Al + 2Ti) ratios.The compositional variations appear to be the result of partitioningof these elements between spinel and olivine.

Group II spinels formed together with other minerals in a later,second stage assemblage of minerals and are Uv-rich, but also con-tain significant proportions of Fc, Cm, and Sp (Table 1). These spinelsare usually subhedral to anhedral and are sometimes zoned with theUv component increasing toward the rim. There is a small hiatus incompositions between group I and II spinels, especially as seen inFigure 2A. More significant than this compositional hiatus, however,is the major change in the compositional trend from group I to groupII spinels: whereas group I spinels vary mainly in Fe/(Fe * Ms) butlittle in Cr/(Cr + Al) and 2Ti/(Cr * Al + 2Ti), group II spinelsvary little in Fe/(Fe + Me) but mainly in Crl (Cr + Al) and 2Til(Cr+ Al + 2Ti). Group II appears to represent a solid solution series,with the earliest spinels presumably being those with lowest zTi/(Cr+ Al + 2Ti) ratios. Group II spinels formed in a second stagemineral assemblage and essentially simultaneously with later pyroxen€that rims the pigeonite-augite megacrysts. Apparently, the eomposi-tional variations in the spinels are the result of element partitioningbetween spinel and pyroxene.

Group III spinels are raxe; only three grains were found. They arerieh in the Hc component and are the first spinels of this compositionto be reported in lunar samples. The three analyses are presented inBusche et al. (1971; Table 38) and one analysis is given in Table 1.All three grains were found in early pyroxene melt inclusions which aredescribed in detail by Prinz et aL. (in prep.). One of the melt inclu-sions in which group III spinels were found is relatively AlzOs-rieh(approximately 17 weight percent) and this may be partly responsiblefor the aluminous nature of the spinel.

A number of noteworthy facts become apparent from a comparisonof spinel compositions in rock 12036 and other Apollo 12 samplesstudied. The earliest group I spinels have the lowest Fe/(Fe + Mg)ratios of Cm-rich spinels, and group II spinels have the lowest Fe/(Fe * Mg) ratios of intermediate to Uv-rich spinels in Apollo 12rocks. This appeaxs to be a reflection of the low Fe/(Fe * Mg) raiioof the rock as indicated by the bulk analysis and of the melt fromwhich 12036 crystallized. Since 12036 is a cumulate rock, the bulkanalysis of the rock cannot be taken as the composition of its parent

I74O BUSCHE, PNINZ, KEIL, AND BUNCH

u.relt and, in fact, the parent melt must have had an even lower Fe/(Fe * Mg) ratio. Furthermore, the Cr/(Cr + Al) ratio of group Ispinels are among the highest of Apollo 12 Cm-rich spinels, indicatinga high Cr/(Cr + Al) ratio of the parent melt.

Roclc 12038

Rock 12038 is a fine-grained basalt that is unique among Apollo 12wet chemicaJly analyzed igneous rocks because of its relatively highAl2O3 and Na2O and low FeO contents (Table 2). This rock is alsorelatively rich in rare earth elements as compared to other Apollo 12igneous rocks (Schnetzler and Philpotts, 1971). Biggar et al. (1971)interpret the melt from which rock 12038 formed as being a low-pressure cotectic liquid that is probably parental to such magnesianrocks as 12018, 12020, and 12040; i,.e.,the latter three rocks are essen-tially 12038 melt with the addition of cumulus olivine, some pyroxene,and chromian spinel. Since rock 12036 is very similar to rock 12040,as shown by Prinz et a^1. (in prep.), this hypothesis presumably ap-plies to rock 12036 as well. Schnetzler and Philpotts (1971) point outthaf a melt of 12038 composition cannot be the direct parent of themagnesian Apollo 12 igneous rocks cited by Biggar et al". (7971) butthat rock 12038 may have formed from a later liquid differentiatedfrom a melt parental to those rocks which has been enriched in plagio-clase, and perhaps, clinopyroxene.

Compositionally, spinels in rock 12038 fall into two widely sepa-rated groups that are shown as groups I and II in Figure 3 (notethat groups I and II of rock 12038 are similar but not identical togroups I and II of rock 12036). Group I spinels are Cm-rich, butcontain considerable Sp and Uv. They are compositionally homo-geneous euhedral octahedra and are very sparse. One grain of a groupI spinel was found associated with a grain of olivine (Fo6e) (olivinein very rare in this rock). Since the rock also contains late fayaliticolivine, the presenee of Cm-rich spinel and olivine of Fo56 indicatesthat the latter phases are either the result of crystal settling or earlycrystallization from melt of 120,38 composition and incomplete resorb-tion. However, group I spinels are markedly aluminous (approxi-mately 17 percent Al2O3) [Table 1; Busche et aL. (797I)]. In fact,Chese are the most aluminous Cm-rich spinels yet reported fromApollo 12 rocks. Since rock 12038 is the most aluminous Apollo 12igneous rock it is suggested that these spinels crystallized from thismelt and, hence, are not the product of extensive crystal settling andaccumulation. Furthermore, the CrzOs content of spinels in rock12038 is low, and the Cr/(Cr + Al) ratios are among the lowest of

APOLLO 12 IGNEOUS ROCKS L74I

Apollo 12 spinels. Rock 12038 also has a low bulk CrzOa content(Kushiro and Haramura, 1971), giving additional supporb to theproposition that Cm-rich spinels (group I spinels) crystallized fromthis liquid.

Group I spinels have variable Fe/(Fe + Me) ratios but vary littlein2Ti/(h + Al + 2Ti) or Cr/(Cr + AI) ratios. Apparently, group

I spinels represent a solid solution series similar to that of group Ispinels in rock 12036. However, a major difference between the twogroups is that group I spinels of rock 12038 are much richer in AlzOsand poorer in Cr2Os than group I spinels of rock 12036. Since boththese groups crystallized simultaneously with olivine it is presumedthat this type of compositional trend is again the result of elementpartitioning between spinel and olivine.

Group II spinels are spaxse; they are generally anhedral and in-homogeneous, with enrichment of Uv towards the rim. They are highlyenriched in FeO and very poor in MgO; in fact, the Uv-content' maybe as high as 85 percent. In addition to the Uv component, they alsocontain some Cm and Hc (Table 1); apparently, group II spinels aremembers of a solid solution series. They are among the latest phasesto crystallize and probably form as late as the fayalitic olivine.

A single spinel analysis is intermediate between group I and IIspinels (Fig.3; Table 1);the sigrrificance of this spinel is not clear.

Roclc 12039

Rock 12039 is a medium-grained. basalt or microgabbro and one ofthe most Fe-rich and Mg-poor Apollo 12 igneous rocks (Table 2).It is most similar in composition to rock 12064 analyzed by Kushiroand Haramura (1971). Early olivine was not observed, and the rockcontains tridymite and cristobalite (Keil 'et aiL.,1971; Bunch et aL.,lg72) . Plagioclase and pyroxene are the predominant minerals; andpyroxferroite often mantles clinopyroxene.

Spinels in rock 12039 are subhedral and interstitial, and appear tohave formed late in the crystallization history of the rock. They arerelatively homogeneous and grain to grain compositional variationsare small. The compositions of three grains are shown in Figure 3and individual, complete analyses are given in Busche et al". (1971);

one representative analysis is given in Table 1. Reconnaissance studyof other spinel grains in this thin section indicates near-uniformity incomposition with the grains reported here and, hence, no additionalquantitative analyses were made. The spinels in rock 12039 are veryUv-rich and contain some Cm and Sp. The similarity between rocks12039 and 12064 is further expressed in the composition of spinels;

1742 BUSCHE, PRINZ, KEIL, AND BT]NCTT

both contain Uv-rich spinel only. The relatively uniform compositionof these spinels, therefore, indicates that immiscibility was not a fac-tor in their formation. Comparison of the mineralogical and chemicalcompositions of rocks 1203g and 12064 indicate close similarity.

Biggar et aI. (1971) suggest that rock 12064 is a low-pressurecotectic liquid and rooks such as l212l,1206b, and 120b2 are samplesof this lava to which pigeonite and, to a lesser extent, olivine andspinel have been added. They also show that a liquid of rock 120Mcomposition crystallized early olivine and Cm-rich spinel and thattheir disappearance is due to reaction with the liquid. They also notethat there is no evidence of a discontinuous ieaction of Cm-richspinel and liquid to form Uv-rich spinel in rock 12064. Thus, becauseof early resorption, rock 12064 contains no Cm-rich spinel, and crystal-lization of late Uv-rich spinel is the result of the composition of themelt and the physicochemical conditions under which it crystallized.Similar arguments appear to aceount for the Uv-rich spinels in rock12039. The possibility that early Cm-rich spinels settled out of thismelt rather than were resorbed back into it, seems less likely sinceBiggar et al. (1971) show that the composifion of rock 12039 is veryclose to a low-pressure cotectic liquid. If significant crystal settlingof Cm-rich spinel and olivine occumed then the melt would not beclose in composition to that cotectic liquid.

Roch 12051

Rock 12051 is a fine-grained basalt relatively rich in iron in com-parison to most other wet chemically analyzed Apollo 12 igneousrocks (Table 2). Brown et aI. (1977) refer to this rock as a ferrobasalt.No early olivine was found in our sections, but Brown et aL (lg7t)note its rare occurrence in their material. Fayalitic olivine and silicaare commonly present (Keil et al.,lgTI). Brown et al. (lg7l) note thatthe texture indicates cotectic crystallization of plagioclase and pyrox-ene, and we are in agreement with this interpretation.

The spinels in rock 12051 differ from those of most Apollo 12 igneousrocks. Many grains are inhomogeneous and some are zoned from amore Cm-rich core to a Uv-rich rim: However, the Cm-rich spinel hasa relatively high Fe/(Fe + Mg) ratio and zoning is continuous acrossthe so-called Apollo 12 spinel gap (as was also shown by Brown et aL.,1971). The zoning is diffuse and sharp boundaries or mantled grainsare not observed. Some grains appear to have more complex composi-tional inhomogeneities (Keil ef al., 1971, Fig. 8), and these may bedue either to unusual growth patterns as the result of rapid crystalliza-tion, or to geometrical effects as a result of the particular cut in thinsection through a zoned grain.

APOLLO 12 IGNEOAS ROCKS 1743

other two projections for the purpose of clarity. The patterns inFigure 4A indicate that early spinels with lowest Fe/(Fe * Mg) ratiosgrade continuously into Uv-rich compositions. Apparently, a. solidsolution relationship exists between early, Fe/(Fe * Mg)-poor andlate, Uv-rich spinels, and no immiscibility gap is observed. Themajor variations arcin2Ti/(Cr * Al + tli), although there is somevariation in Fe/(Fe + MS) and Crl(Cr + Al). The variation inFe/(Fe + Mg) is nearly entirely a variation in Fe, as Mg is neareonstant in all zoned grains that 'u'ere analyzed (see Keil et aI., lg7l,Fig. 8). As crystallization proeeeded, Cm-rich .spinel increased inFe/(Fe + Mg) ratio and the slope of the line through points inindividual grains (as shown in Fig. 4A) increases nearing verticai onthe right side of the diagram. Later spinels have increasingly Uv-richcores and rims and plot in the upper right eorner of the plot; thesemay be a later solid solution series.

Early spinels of 12051 crystallized essentially simultaneously withpyroxene and plagioclase and in this respect, are analogous to groupII spinels in rocks 12036 and 12088. The major difference is that thesegrains are sometimes cored b5' Cm-rich spinels, with relatively highFe/(Fe * MS) ratios, and are continuouslv zoned.

Coucr,usroNs

The following conclusions have been drawn from the study of spinercompositions for the four Apollo 12 igneous rocks studied:(1) There appears to be no evidenee for a miscibility gap in the spinelsystem in any of the four rocks studied. This finding is in agreementwith the experimental work of Muan et aI. (1972): they have shownthat no two-phase region is present at temperatures down to 1000.Cin that poriion of the spinel prism where Apollo 12 spinels plot (notethat there is no evidence for subsolidus reactions in the spinels siudiedhere). These rocks have spinels that may be divided into either one(12039), two (12038), or three (1203G) groups. Rock 120b1 has Cm-rich spinels which are zoned continuously to Uv-rich spinels. Spinelsform continuous solid solution series in rocks with one.spinel group.Separate spinel groups in rocks 12036 and 12088 appear to be the re-sult of crystallization at different stages in the crystallization historyof the rocks.

1744 BUSCHE, PNINZ, KEIL, AND BANCH

(2) There is no textural evidence in the rocks studied here for aperitectic relation involving reaction of early Cm-rich spinel withliquid to produce Uv-rich spinel as suggested by Kushiro et al. (1970).

A peritectic relation of this type should be manifested in texfuralevidence, such as resorption of Cm-rich spinel, which is not observed.Cameron (1971) and Weill et at. (197D have reached similar con-clusions.(3) Mineralogical and chemical data for spinels and coexisting phases,and the textures and bulk compositions of the rocks, point to a peri-tectic relationship for the origin of spinels of the type desmibed byIrvine (1967). Early Cm-rich spinel forms simultaneously with earlyolivine; forma;tion of these phases is terminated and is followed bypyroxene crystallization, often simultaneously with plagioclase' If newspinel crystallizes shortly thereafter, the compositional hiatus be-

tween old and new spinel is small, as in rock 120E6. If new spinelcrystallizes much later, then the compositional gap is large, as in rock12038. If early Cm-rich spinel is totally resorbed or settled out of themelt, only one Uv-rich spinel series is present, as is the case in rock

12039. There is no clear evidence that suggests that crystallization ofCm-rich spinel is terminated by eruption of the lava to the surface,or that a change in the "liquidus slope" of equilibrium crystallizationhas changed, as suggested by Weill et al. (1971) .(4) Rock 12051 contains spinels with trends sornewhat analogous togroup II spinels of rocks 12036 and 12038. Early Cm-rich spinel, withlow Fe/ (Fe + Ms) ratio, is absent but the raxe occurrence of earlyolivine was noted by Brown et al. (1971). It appears that the rapidcrystallization of this fine-grained rock resulted in the inhibition ofcrystallization of early spinel and olivine before the peritectic re-action took place that results in the onset of pyroxene and plagioclasecrystallization. Therefore, Cm-rich spinel with relatively high Fe/(Fe * Mg) ratios crystallized simultaneously with pyroxene andplagioclase and, since there was no interference from a peritectic re-

lationship, resulted in a continuous crystallization series between cm-and Uv-rich compositions. Again, this observation points to the lackof a miscibility gap in the complex spinel systems observed in Apollo12 roeks.(5) Haggeriy (1971a) has surveyed the literature on Apollo 12 spinelsand plotted their compositions in a modified spinel prism similar to

the one used in this paper. Most of the data points fall into two groups

similar to groups I and II of rocks 12036 and f2m8. We interpret thisfinding as an indication that many Cm-rich spinels formed early, be-fore the crystallization of pyroxene, and that many Uv-rich spinels

APOLLO 12 IGNEOUS ROCKS 7745

formed after the peritectic rela,tionship which produced pyroxene, oreven later in the crystallization history of the rock.

'We suggest that

the extent of a compositional gap between early and later spinels isa function of the time of crystallization of the spinel in the melt. Themajor change in the compositional trends between the two spinelgroups is interpreted as the result of the differences in element parbi-tioning between spinel and co-existing phases, especially spinel andolivine for early Cm-rich spinels, and spinel and pyroxene for laterUv-rich spinels. However, we would caution any interpretation ofspinel data without the detailed knowledge of the composition andzoning of spinel grains in relation to composition, texture and time ofcrystallization within the specific rock.(6) Biggar et al. (1971) suggest that melts of rock 12038 composi-tion may be the parent of olivine-rich rocks such as 12018, 72020,and 12040. Pnnz et a,t. (in prep.) have shown that rock 12CI36 is verysimilar to rock l2O4A. Hence, by analogy, a melt of rock 12038 com-position may be the parent of rock 12036. However, Schnetzler and

Fhilpotts (1971) argue that rock 12&38 could not be parental to theserocks and must have formed from a later liquid to which plagioclase,and perhaps, clinopyroxene were added. However, their propositionwas not supported by textural evidence.

Data on the composition of spinels for rocks 12036 and 12038 arerelevant to these hypotheses: The early Cm-rich spinels in rock 12038are more Al-rich and Cr-poor than those in rock 12036 and, there-fore, the latter could not have formed in a liquid of 12G38 composition.However, a liquid intermediate in composition between rocks 12036

and 12038 could be parental to both, as well as to the other magnesianApollo 12 igneous rocks. From this parent, 12036-type Cm-rich spinels

as well as early olivine and pigeonite-augite megacrysts could have

crystallized. This differentiated melt would then be greatly enriched in

the feldspar component and resemble the composition of rock 12038.

This model is then in essential agreement with the proposition of

schnetzler and Philpotts (1971) and would avoid the problem of not

obsewing textural evidence for plagioclase enrichment.

AcrNowmocrvrnxrg

We gratefully acknowledge the assistance of Mr. George Conrad in the

electron microprobe work, Mrs. Julie lfultzen, Misses D. M. Boyer and M. T'

Busch. and Messrs. Paul Elava and Norman wilderman in the electron micro-

probe and other data reduction. This work is zupported in part by NASA Con-

tract NAS 9-9365 and NASA grant NGL 32{0t063 (Klaus Keil, Principal In-

vestigator).

1746 BUSCHE, PRINZ, KEIL, AND BUNCH

RnnnnnNcns

Acnnr,r,, S. A., 4.. Pocnorr, F. R. Bovn, S. E. Haccnnrr, T. E. BuNcu, E. N.CeunnoN, M. R. DnNcn, J. A. V. Doucr,.rs, A. G. Pleur, R. J. Tnarrr,, O. B.Jeuns, K. Knrr,, aNn M. Pnr*z (1970) Titanian chromite, aluminian chromiteand chromian ulviispinel from Apollo 11 rocks. Proc. Apollo 11 Lunar Sci.ConJ., Geochim. Cosmochim. Acta, Suppl..1, 1, 81-86.

Brccan, G. M., M. J. O'Hena, A. Pncr<urr, eru D. J. Huupnnrns (1971) Lunarlavas and the achondrites: petrogenesis of protohypersthene basalts in themaria lava lakes. Proc. Second l-runar Sci. Conf., Geochim. Cosmochim.Acta, Su,ppl. 2, L, 617443.

Bnntr, R., P. Burr,on, C. Moynn, A. M. Rorn, H. Ternoa, eNo R. 'Wrlr,rervrs

(1971) Apollo 12 igneous rocks 12004, 12008, 12009, andl2W2: A mineralogicaland petrological study. Proc. Second Lunar Sci. ConJ., Geochim. Cosmochim.Acta, 9up,pl. 2, r,30I-317.

Bnoww, G. M., C. H. Eunlnus, J. G. Holr,eNn, A. Pncrorr, eNl R. Pnrr,r,rps(1971) Picrite basalts, ferrobasalts, feldspathic norites, and rhyolites in astrongl5r, fractionated lunar crust. Proc. SeconrJ Lunar Sci. Conf ., Geochi.m.Cosmochirn. Acta, Suppl.g, 1, b8B-600.

BuNcu, T. E., K. Knrr, aNr M. PnrNz (19?2) Mineralogy, petrology, and chemistryof lunar rock 12039. Meteoritics, 7,245-255.

Buscnn, F. D., G. H. CoNneo, K. Knrl, M. Pnrwz, T. E. Buxcn, J. E. Enr,rcnlrer.r,eNr W. L. Quarno (1g21) Electron microprobe analyses of minerals fromApollo 12 lunar samples. (Jni,u, New Mexico, Inst. Meteoritics, Spec. Pub. No.3, 61 p.

CeuonoN, E. N. (1921) Opaque minerals in certain lunar rocks from Apollo 12.Proc. Second Lunar 9ci. Conf ., Geochim. Cosm,ochim. Acta, Suppl. 2, L,193-206.

CuertrNnss, P. E., A. C. DuNrreu, F. G. F. Grsn, H. N. Grr,ps, W. S. Mac-KoNzm, E. F. Sruuprl, aNo J. ZusslreN (1971) Mineralogy and petrologyof some Apollo 12 lunar samples. Proc. Second, Lunar Sci. Conf., Geochim.Cosmochim. Acta, Su'ppl. p, t, tsEg_:B76.

Dnern, M. J., I. S. McClr,r,uiu, G. A. McKer, eNn D. F. Worr,r, (1970) Min-eralogy and petrologJr of Apollo 12 sample No. 12018: A progress report.EarLh Planet. Sci. Lett.9, 103-123.

Er, Gonesv, A., P. R-lrrnoHn, AND L. A. Teyr,on (1971) The opaque minerals inthe lunar rocks from Oceanus Procellarum. Proc. Second, Lunar Sci,, Conl.,G eo chim. C o srno chim. A ct a, Suppl. 2, r, 219-235.

Gnn, F. G. F., E. F. Sruntp'n'r, arn J. Zussrrew (1970) Opaque minerals in anApollo 12 rock. Earth Planet. Sci,. Lett. 9,217-224.

Ilaccnnrv, S. E. (1971a) Compositional variations in lunar spinels. Natwe Phgs.$ci,. zsz,156-160.

- (1971b) Subsolidus reduction of lunar spinels. Nature Phgs. 9ci.234, ll3-tt7.

- (1572) Luna 16: An opaque mineral study and a systematic examinationof compositional variations of spinels from Mare Fecunditatis. Earth Planet.Sci,. Lett. 73, 328-352.

- AND H. O. A. Mo'r-on (1970) Apollo 12: Opaque oxides. Earth, Planet. Sci.Lett.9,379-387.

APOLLO 12 IGNEOUS ROCKS t747

Invrxn, T. N. (1965) Chromian spinel as a petrogenetic indicator. Part I, Theory.C an. J . Earth Sci. 2, 648472.

- (1967) Chromian spinel as a petrogenetic indicator. Parl 2, Petrologicapplications. Con. J. Earth Sci.4,7I-103.

Jecrsox, E. D. (1969) Chemical variation in coexisting chromite and olivine inchromitite zones of the Stillwater complex. Econ. Geol'., Momo. No. 4, 4l-71'

Krrr,, K. (1967) The electron microprobe x-ray analyzer and its application inmineralogy. F orts chr. M ineral. 44, 4-fr6.

M. PnrNz, eNn T. E. Bur.rcn (1971) Mineralogy, petrology, and chemistryof some Apollo 12 samples. Pr'oc. Second, Lurwr Sci,. Cont., Geoch,im. Cosmo-chim. Acta, Suppl. z,1, 319-341.

Kusnrno, I., Y. N,meuune, elru S. Arrnroro (1970) Crystallization of Cr-Ti spinelsolid solutions in an Apollo 12 rock, and source rock of magmas of Apollo12 rocks [abst.]. Trans. Amer. GeophEs.Uni,on,5L,585.

axr I[. Henerrunr (1971) Major element variation of Apollo 12 crystallinerocks. Science, 17 l, 1235-1237 .

Mexwnr,r,, J. A., eNo H. B. Wrrr (1971) Chemical composition of Apollo 12 lunarsamples 12004, 12033, lnil, n052, and 12065. Earth Pl'anet. Sci. Lett' 10,285-288.

MueN, A., J. Heucr, E. F. OsronN, eNo J. F. Scrrarnun (1971) Equilibrium rela-tions among phases occurring in lunar rocks. Proc. Second Lutttar Sci.Conf ., Geocfuim, Cosmochi,m. Acta, Su'ppl. 2, t, 497405.

_, _, ewr T. Lijnelr, (1922) Equilibrium studies with a bearing on lunarrocks. latrstr.] Third, Lunar Sci,. Conl. (Reu. Abstr.)

Pntxz, M., T. E. Buxcn, eNn K. Knr, (1971) Composition and origin of lithicfragments and glasses in Apollo 11 samples. Contri,b. Mi.neral. PetrologE, 32,2tt-230.

Rnrn, J. B. (1971) Apollo 12 spinels as petrogenetic indicators. Earth, Planet. Sci'.Lett. 7o,351-356.

ScrrNorzr,un, C. C., er.ro J. A. Pnrr,porrs (1971) Alkali, alkaline earth, and rare-earth element concentrations in some Apollo 12 soils, rocks, and separatedphases. Proc. Second Lunar Sci. Conf ., Geochim. Cosmochim. Acta, SWpL2,2,ll0l-ll22.

Srursow, P. R.,.eNo S. IL U. Bowrn (1971) Opaque phases in Apollo 12 samples.Proc. Second Luntrr Sci. Conl., Ganclvim. Cosmoch,im. Acta, Suppl 2, 7,n7-2r8.

Terr,on, L. A., G. Kur,r,nnuo, ewo W. B. BnveN (1971) Opaque mineralogy andtextural features of Apollo 12 samples and a comparison with ApoIIo 11rocks. Proc. Second" Lunar Sci. Conf ., Geochim. Cosmochim. Acta, Suppl. 2,1, 855-871.

Wer,rnn, L. S., B. M. Fnnxcn, K. F. .L llnrNmcn, P. D. Lowrraw, A. S. Doer.r,er'ro I. Anr,nn (1971) Mineralogical studies of Apollo 12 samples Proc. Second,Innar Sci,. Cont., Geochim. Cosrnocfui,m. Acta, Suppl. 2, 7,343-358.

Worr,r,, D. F., R. A. Gnrnvn, I. S. McCer,r,ulr, eNo Y. Borrrwce (1971) Mineralogy-petrology of lunar samples. Microprobe studies of samples 12A21 artd 9,AT2;viscosity of melts of selected luna,r compositions. Pro,c. Secomd Iatnar Sci,.Conf ., Geochim. Cosmoch,im. Acta, Suppl.2,1,413-430.

Manuscri,pt receiued, Februarg 11,1912; accepted, for Wbli,cation, June 1, 1n2.