Petrogenesis of pyroxene-oxide intergrowths from kimberlite andcumulate rocks: co-precipitation or exsolution?

American Mineralogist, Volume 66, pages 723-740, IggI

Jeues R. GannrsoN, JR.' eNo LawRBNcB A. TAyLoR

Department of Geological SciencesThe University of TennesseeKnoxville, Tennessee 379 I 6

Abstract

Pyroxene-oxide intergrowths occur in rocks representing equilibration over a wide rangeof P-!-fo.conditionq these intergrowths have three modes ofoccurrence: (l) nodular pyroi-ene-ilmenite intergrowths from kimberlite, (2) "myrmekitic" pyroxene-oxide intergiowthsfrom cumulate rocks and mantle xenoliths, and (3) fne lamelJof spinel or ;lmenite in py-roxenes from cumulate rocks and mantle peridotites. Examination of three nodular cfi"opy-roxene-ilmenite intergrowths and an ilmenite-spinel websterite from the Elliott County kim_berlite, Kentucky and a cumulate norite from a small gabbroic pluton from Chester County,South Carolina has revealed intergrowths representing all three modes of occurrence. Thenodular clinopyroxene-ilmenite intergrowths are interpreted to have formed during co-pre-cipitation of Cr-poor clinopyroxene (73.0-76.8 mol9oDi) and picroilmenite (44.746.6 mil.EoMgTio, and 5.4-5.8 mol.vo Fero) at 1295.-l335oc from ihe .proto-nimberlitic,'liquidwhich produced the Elliott County kimberhte megacryst suite. The

-Elliott County websterite

xenolith contains fne (l-5 pm) lamellae of magnetite (12.6 mol.vo (Fe,Mg)Alroo and g.0mol.7o (Fe,Mg)rTiOa) and pleonaste (6.8 mol.Vo (Fe,Mg)FerOo and 0.1 mol .n lf e,Ugyrfioolwithin aluminous clinopyroxene (8.5 mol.zo caTs and 7.3 mol.vo Jd) and orthopyroxine fl.bmol.7o MgTs); these intergrowths are interpreted to have formed during '.exsolution" ofspinel from high+emperature (1105"-ll40oC), non-stoichiometric pyroxenes. Coexistingoxides suggest equilibration at 695oC and /o. : l0-r3.8. The Chester cumulate norite containsboth ilmslile lamellae (11.6 mol.vo MgTio3 and 5.7 mol.vo Fe2o) in orthopyroxene (4.4mol.7a MgTs) and clinopyroxene (5.6 mol.Vo CaTs artd 2J mol.Vo Id) and myrmekitic inter-growths of orthopyroxene and magnetite (0.g mol.vo (Fe,Mg)Alro o and 0.5 mol9o(Fe,Mg)rTioo) and minor ilmenite (14.3 mol.vo MgTio, and 3.g mol.vo Fe2o3). The lamellarintergrowths were produced during "exsolution,' from high+emperature (l200oC), non-stoichiometric pyroxenes. Myrmekitic intergrowths formed Ouring co-precipitation of ortho-pyroxene, magnetite, and ilmenite. Coexisting Fe-Ti oxides suggest equilibration temper-atures of 480"-5l5oc at f o. - l0-2r.o-10-23.?.

Introduction

Intergrowths of oxides and silicates have been re-ported from a variety of rock types representingequitbration over a wide range of temperature andpressure conditions. The most commonly reportedoccurrence is nodular pyroxene-ilmenite inter-growths found in kinberlite (Gurney et al., 1973;Boyd and Nixon, 1973; Dawson and Reid, 1970;Frick, 1973; Ilupin et al., 1973; McCallister el a/..

I Current address: School ofGeology and Geophysics, Univer-sity of Oklahoma, Norman, Oktahoma 73019.

w03-o0/.x/ 8 | /0708-0723$02.00

1975; Smith et al.,1976; Eggler et al.,1979; Haggertyet al., 1979). These intergrowths occur as large singlecrystals of clinopyroxene (e.g., Gurney et al.,1973\ ororthopyroxene (e.9., Frick, 1973) with lamellar orgraphic intergrowths of optically continuous pic-roilmenite (e.9., Figure la). Recently, Rawlinson andDawson (1979) reported a xenolith from the Weltvre-den Mine, South Africa, which consists of acicularorthopyroxene intergrown irregular$ with picroilne-nite, but 3ls9 ssataining large orthopyroxenes, gar-nets, and clinopyroxene-ilmenite intergrowths, de-scribed as having a quench texture. Schulze et al.

724 GARRISON AND TAYLOR: PYROXENE-OXIDE INTERGROWTHS

Fig l. (a) Coarse graphic clinopyroxene-ilmenite intergrowth (l-2C-0) from the Elliott County kimberlite; ilm : ilmenite; cpx :

clinopyroxene. Scale bar : I mm. (b) Cumulate ilmeniti-spinel websterite xenolith 1-29 from the Elliott County kimberlite' mt :

magnetite; ilm: ilnenite; sp: pleonaste; opx: orthopyroxene; cpx: clinopyroxene. Scale bar: I mm. (c) lamellae ofpleonaste (sp)

and magnetite (mt) in orthopyroxene from the cumulate ilmenite-spinel websterite l-29' Scale bar : 100 p,m' (d) lamellae of hematite

(hm) along intragrain dislocations in interstitial iknenite (ilm) in websterite l-29. Scale bar : 100 pm.

GARRISON AND TAYLOR: PYROXENE-OXIDE INTERGROWTHS

(1978) described lamellar pyroxene-ilmenite inter-growths in garnet pyroxenite xenoliths recoveredfrom kimberlite and from latite. Basu and MacGre-gor (1975) described wormy "symplectic,, inter-growths of Cr-spinel with both clinopyroxene and or-thopyroxene in peridotite xenoliths from alkali basaltand kimfsllite; they also described thin Cr-spinel la-mellae in the orthopyroxene. Dawson and Smith(1975) described similar "fingerprint" intergrowths ofCr-spinel with orthopyroxene and, more rarely,clinopyroxene from a variety of South African perid-otite xenoliths; these intergrowths were commonlyassociated with other silicates such as olivine andamphibole. Haselton and Nash (1975) observed a"eutectic-like" intergrowth of ilmenite and ortho-pyroxene in Lower Zone a (LZa) of the Skaergaardcomplex; they reported a similar intergrowth inApollo 16 breccia 60016 (re., similar to Figure 2a).Haggerty (1972) observed thin lamellae of Cr-spinelwithin clinopyroxene in Apollo 14 soil sample l4Z5B.Deer and Abbott (1965) described thin lamellae ofFe-Ti oxide (magnetite?) within clinopyroxene fromthe Kap Edvard Holm layered complex, and Morse

(1969) reported a sinilar occurrence of lamellar in-tergrowths of magnetite-ulvbspinel and ilmenitewithin clinopyroxenes from the Kiglapait layeredcomplex; Moore (1971) reported lamellar inter-growths of spinel and pyroxene from the Gosse Pilecomplex.

The coarse nodular pyroxene-ilmenite inter-growths from kimberlites have received the most at-tention because they have been considered to beunique to kimberlite and to be possible clues to thephysiochemical conditions surrounding kimberliteformation. Although many mechanisms have beenproposed to explain these coarse lamellar andgraphic intergrowths, their origin is still uncertain.The proposed mechanisms for intergrowth formationare: (l) eutectoid transformation of garnet (Ring-wood and Lovering, 1970), (2) exsolution from an"ilmenite-structured pyroxene" (Dawson and Reid,1970), (3) the replacement of pyroxene by ilmgnils(Frick, 1973), (4) eutectic crystallization (Gurney e/al., 1973; Boyd and Nixon, 1973; Wyatt, 1977), and(5) cotectic precipitation (Frick, 1973).

Most other occurrences of pyroxene-oxide inter-

Fig 2. (a) "myrmekitic" intergrowth of orthopyroxene with magnetite (mt) and ilmenite (ikn) from the Chester cumulate norite C-14.Scale bar : 100 pm. (b) lamellae of ilrnenite (ihn) parallel to the (010) plane of orthopyroxene in cumulate norite C-14. Scale bar : 100&m.

GARRISON AND TAYLOK PYROXENE-OXIDE INTERGROWTHS

growths are restricted to rocks representing deep-seated cumulates (1e., layered gabbroic rocks) orrocks presumably derived from such rocks (l'.e., web-sterites and fragments within lunar soils andbreccias), and, less commonly, harzburgites andlherzolites of mantle origin. These intergrowths di-fferfrom the nodular intergrowths found in kimfgilils intexture, scale, and the range of observed oxide com-positions. The pyroxene-oxide intergrowths fromcumulates and mantle peridotites have two modes ofoccurrence: (1) worm-like "myrmekitic" intergrowthsof oxide within pyroxene (e.g., Figure 2a), and (2)very fine rod- or plate-like lamellae (-l-3 pm thick)within pyroxene (e.9., Figures lc and 2b).

The "myrmekitic" intergrowths are texturallyquite similar to some of the pyroxene-ilmenite inter-growths from kimberlite, although the width of theoxide intergrowths is much smaller (-5-20 pm thick)than the coarse graphic intergrowths from kimberlite(-100-400 pm thick). These "myrmekitic" inter-growths, like the nodular intergrowths from kimber-lite, generally consist of one optically continuousoxide phase; occasionally, the nodular intergrowthsshow minor stress recrystallization to polygonal mo-saics. The fine lamellar intergrowths could also beoptically continuous, but the extremely small sizemake verification difficult. However, these lamellaeare usually crystallographically oriented along the(010) andlor the (001) planes of the host pyroxenes(e.g., Deer and Abbott, 1965; Morse, 1969).

The mineralogies of these two different inter-growths are similar. The "myrmekitic" intergrowthsfrom cumulates and mantle peridotites are com-monly intergrowths of ilmenite, magnetite, orCr-spinel and orthopyroxene, and less commonlyclinopyroxene (Haselton and Nash, 1975; Basu andMacGregor, 1975; Dawson and Smith, 1975). Thelamellar intergrowths generally consist of ortho-pyroxene or clinopyroxene with lamellae of ilmenite(Schulze et al., 1978; Morse, 1969), magnetite-ul-vdspinel (Morse, 1969; Deer and Abbott, 1965), andCr-spinel (Haggerty, 1972; Basu and MacGregor,1975; Suwa et al.,1975).

The "myrmekitic" pyroxene-oxide intergrowthsfrom cumulates and peridotites have been consideredthe result of (l) eutectic crystallization (Haselton andNash, 1975), (2) exsolution (Basu and MacGregor,1975), (3) subsolidus recrystallization (Dawson andSmith, 1975), and (4) cotectic crystallization (Daw-son and Smith, 1975).

The very fine lamellar intergrowths are generally

considered to be the result of exsolution2 of ilmeniteand spinel from pyroxene (Haggerty, 1972; Schulzeet al.,1978 Basu and MacGregor, 1975) because: (1)their crystallographically controlled orientations areconsistent with exsolution, (2) considering the kinet-ics of oxide nucleation and growth, it is difficult to vi-sualize such fine-grained lamellae precipitating froma melt, and (3) they are conmonly concentratedalong kink bands and deformed areas within the hostpyroxenes.

It is clear from the discussions above that there isstill much uncertainty and controversy about the ori-gins of all three types of pyroxene-oxide inter-growths. The purpose of this study is to re-evaluatethe possible models for the origin of these inter-growths in light of new mineralogical and chemicaldata from xenoliths in kimberlite and a sample ofcumulate norite from a layered complex. These sam-ples represent examples of all three types of pyrox-ene-oxide intergrowths. Included in this study are:

(l) Three coarse nodular clinopyroxene-ilmeniteintergrowths from the Elliott County kimberlite,Kentucky;

(2) a cumulate ilmenite-spinel websterite xenolith,from the Elliott County kimberlite, which containsthin lamellar spinel and magnetite intergrown withboth orthopyroxene and clinopyroxen€;

(3) a cumulate norite from the Chester pluton,South Carolina which contains both a "myrmekitic"intergrowth of magnetite and ilmenite with ortho-pyroxene and intergrowths of thin lamellar itnenitein orthopyroxene and clinopyroxene.

' Mineralogy and PetrograPhY

N o dular c linopyroxene -ilmenite int ergrowths fromkimberlite

Three nodular clinopyroxene-ilmenite inter-growths were collected from the Elliott County kim-

berlite in eastern Kentucky. These nodules range insize from 0.5 cm to 2.0 cm in diameter and consist oflarge single crystals of pale-green clinopyroxene withgraphic intergrowths of ilmenite (Figure la). Thesegraphic ilmenites (10M00 pm thick) form one opti-cally continuous crystal (up to 30 modal percent ofnodule), although occasionally they exhibit very mi-nor recrystallization to polygonal mosaics or showundulose extinction indicating minor strain. The host

2 Tfuoughout this paper, the term exsolution is used to describe

all possible subsolidus reactions involving the conversion of asolid solution to a new composition and a unique phase or phases.

GARRI S ON AN D TAYLOR P YRO XEN E- O XI DE I NTERGRO T|/TH S

clinopyroxene commonly shows undulose extinctionbut rarely recrystallization. The ilmenites frequentlyextend beyond the host clinopyroxene into the kim-berlite matrix, but terminate at the nodule boundary.This is probably a consequence of fragmentation andabrasion during transport through the kinberliteconduit. The grain boundaries between the graphicilmenite and host clinopyroxene are generally free ofalteration, although in some cases traces of phlogo-pite can be observed.

These nodular intergrowths from Elliott Countyare texturally similar to other such graphic inter-growths from kimberlites in South Africa (e.g., Gur-ney et al.,1973) and the United States (e.g., Eggler etal., 1979), although the graphic texture of the ElliottCounty intergrowhs are not as regularly arranged asintergrowths such as those described by Dawson andReid (1970). The Elliott County nodular inter-growths are comparable in size to most other inter-growths from kimberlites. We consider the ElliottCounty clinopyroxene-ilmenite intergrowths to berepresentative in size and texture of the nodulargraphic intergrowths commonly recovered from kim-berlite.

I lmenite- spinel webst erit e xenolith

An ilnenite-spinel websterite xenolith (sample l-29\,2 cm in diameter, was recovered from the ElliottCounty kimberlite in eastern Kentucky. This xeno-lith exhibits a crude layering in which magnetite (upto 20Eo of rock), ilmenite (up to 5Vo of rock), andgreen pleonaste spinel are interstitial to cumulusclinopyroxene and orthopyroxene (up to 2 mm i1length). The interstitial pleonaste is ubiquitously sur-rounded by Fe-Ti oxides (Figure lb), although ilme-nite and magnetite commonly occur without spinel.

Both clinopyroxene and orthopyroxene containvery fine "plate-like" lamellae of green pleonaste(-l-5 pm thick) and magnetite (l-2 pm thick); thelamellae are oriented parallel to the pyroxene (010)plane (Figure lc). In the clinopyroxenes, pleonasteforrns the majority of the lamellae; magnetite fonnsthe majority of the lamellae in orthopyroxene. Thepyroxenes occasionally exhibit bands diagonal to thelong dimension of the crystals in which the lamellaeare concentrated; these concentrations are probablvalong structural dislocations in the pyroxenes (1e.,kink bands), although optically this is difficult toconfinn. Both orthopyroxene and interstitial ilmeniteshow evidence of subsolidus reequilibration; ortho-pyroxene appears to have exsolved diffuse lamellae

of clinopyroxene, and ilmenite has exsolved lamellaeof hematite along intragrain dislocations (Figure ld).

Cumulate norite from layered pluton

Cumulate norite sample C-14 was collected from asmall layered gabbroic pluton in Chester County,South Carolina. This extremely fts5fu sample consistsof cumulus plagioclase (-50Vo of rock) and olivine(<l0%o of rock) enclosed by intercumulus ortho-pyroxene (-25Vo of rock) and brown hornblende(-l1Vo of rock); accessory amounts of clinopyroxene(-lVo of rock), biotite, magnetite, ilmsnifs, glssapleonaste, and pyrrhotite also occur interstitially.The olivine is generally always surrounded by ortho-pyroxene, only rarely is it surrounded by hornblendeor clinopyroxene. The orthopyroxene appears tohave crystallized prior to hornblende, clinopyroxene,biotite, and intercumulus Fe-Ti oxides.

The orthopyroxene, even when it occurs as a thinrim around olivine, appears to be ubiquitously asso-ciated with magnetite and ilmenite. The ortho-pyroxene is intergrown with magnetite and ilmeniteproducing a wormy "myrmekitic" texture (Figure2a). Itmenite is only a minor constituent of these"myrmekitic" intergrowths. When orthopyroxene oc-curs as a thin rim around olivine, the "mynnekitic"intergrowths give one the impression of a "s5m-plectic" reaction texture associated with the break-down of olivine, but these intergrowths also occurwithin the cores of large (up to 3 mm) ortho-pyroxenes which are quite distant from olivine, thusarguing against their origin by a simple eutectoidalbreakdown of olivine. The "myrmekitic" inter-growths are similar in texture and scale to the inter-growths reported by Haselton and Nash (1975), Basuand MacGregor (1975), and Dawson and Smith(re7s).

The orthopyroxenes also contain very fine (l-3 pmthick) "plate-like" lamellae of ilmenite oriented par-allel to the orthopyroxene (010) plane (Figure 2b).The only clinopyroxene noted in this sample alsocontains fine lamellae of ilmenite which are orientedparallel to the (010) plane and less conmonly to the(001) plane of the clinopyroxene. These lamellae aresimilar in texture and scale to the lamellae in the py-roxenes in the ilmenite-spinel websterite xenolithand to numerous other reported intergrowths fromcumulate rocks (e.g., Deer and Abbott, 1965; Morse,1969). The orthopyroxene contains fne lamellae (-lpm thick) of clinopyroxene; the clinopyroxene con-lains fine lamellae (-l pm thick) of orthopyroxene.

GARRISON AND TAYLOR: PYROXENE-OXIDE INTERGROWTHS

Table l. Representative av€rage analyses of clinopyroxenes andilmenites from nodular graphic cl inopyroxene-i lmenite

intergrowths from kimberlite

tion to clinopyroxene megacrysts from the ElliottCounty kimberlite (Garrison and Taylor, 1980; Fig-ure 3), although somewhat more Cr-poor (0.06-O.38Vo CtzOr). The temperatures estimated for theseclinopyroxenes, using the Diopside-Enstatite solvusof Lindsley and Dixon (1976) with mol.7o Di: 100 x2Ca/Ca+Mg*Fe and assuming coexisting ortho-pyroxene, range from 1295"-1335"C (Garrison andTaylor, 1980); these temperature estimates are withinthe range of temperatures obtained for the ElliottCounty clinopyroxene megacrysts (1305"-1390"C).These clinopyroxenes are quite similar in composi-tion to clinopyroxenes from intergrowths reported byDawson and Reid (1970), Gurney et al. (1973} Gur-ney et al. (1979), and Eggler et al. (1979).

The ilmenites from the Elliott County intergrowthsare picroilmenites containing 44.7-46.6 moI'VoMgTiO, (geikielite) and 5.4-5.8 mol.Vo FerO, (hema-tite); they fall well within the range of compositionsobserved for the Elliott County ilmenite megacrystsuite (Figure 4). The ilmenites from the ElliottCounty intergrowths are similar in composition tothose reported by Eggler et al. (1979) for clinopyrox-ene-ilmenite intergrowths from the Colorado-Wyo-ming kimlsrlite district (38.9-44.4 mol.Vo MgTiO'and 5.5-9.2 mol.Vo FerOr). They also have similarhematite content to ilmenites from the quench pyrox-ene-ilmenite xenolith (3.0-7.5 mol.Vo FerOr) reportedby Rawlinson and Dawson (1979), although theWeltvreden xenolith ilmenites have a somewhatlower geikielite content (32.541.3 mol.Vo MgTiO,).In contrast to the homogeneous nature of the ilme-nites from the Elliott County intergrowths, Haggertyet al. (1979) found the ilmenites from many Monas-tery Mine intergrowths to be zoned, ranging from293-36J mol.Vo MgTiO, and 6.3-34.2 mol.Vo FerO,;these ilnenites are lower in FerO, and higher inMgTiO, and FeTiO, proximal to the host pyroxene.The compositional ranges of the intermediate areasof the Monastery ilmgniles are shown in Figure 4. Itis interesting to note that the Monastery ilmenitemegacrysts have compositions that overlap the rangefor the ilmenites from the Monastery intergrowths-a situation somewhat analogous to the compositionalrelationship between the ilmenites from the ElliottCounty intergrowths and the Elliott County ilmenitemegacrysts. This suggests some genetic relationshipbetween the pyroxene-ilmenite intergrowths and themegacrysts of the host kimberlite.

If the assumption is made that the clinopyroxene-ilmenite system has remained closed since formation,the Fe*2-Mg distribution geothermometer of Bishop

CPX I L}I CPXl - 2 c - 0 r - 2 c - 0 1 - 2 7 - 0

I Lll CPX I Li.r1 - 2 7 - 0 1 - 3 0 - 0 r - 3 0 - 0

s i 0 ^

T i 0 ^

A l ^ 0 -z t

C r ^ 0 ^

' - 2 - a

Fe0

i ln0

MsoN i 0

Ce0

Na^0

t

Cat ionB a s l s

. . l v

T i

1+Fe-

7 +t e

Ig

l'ln

N i

C a

N a

E

5 4 . 8 ( 2 )

o . 5 0 ( 1 ) 5 \ , 2 ( 3 \

2 .36(2 ' , t 0 .57 (4 )

o . 0 6 ( 3 ) o . 3 3 Q l

o . / t

t r .48(7) 25 .7 Q)

o . l 5 ( 2 ) 0 . 2 4 ( 3 )

1 8 . 2 ( t ) 1 3 . 0 ( r )

o . 0 9 ( 4 )

17.9 (2'l

1 . 4 9 ( 3 )

5 \ . 9 Q \

0 .55Q\ 53 .2 (2 )

2 . \ o ( 2 \ l . o 4 ( 6 )

0 . 3 8 ( 5 ) r . 4 0 ( 4 )

o . ) )

4 . 7 r ( 1 3 ) 2 4 . 5 ( 3 )

o . r 3 ( 2 ) o . 2 0 ( 3 )1 8 . 6 ( 1 ) 1 3 . 5 ( r )

o . l 7 ( 4 )17 .o ( z \

t . 5 0 ( 2 )

5 \ . 3 Q \

0 .57Q\ 53 . \ (2 )

2 . 3 2 ( t r . o 3 ( 9 )

0 . 3 8 ( 2 ) 1 . 3 9 ( 2 )

5 . 1 4

4.54(9) 24 .9 (5 )

o . t 5 ( 2 ) o . l 9 ( 2 )

1 8 . 3 ( 1 ) 1 3 . 2 ( r )

o . l o ( 2 )

t 5 . 8 ( 1 )

1 . 5 5 ( 3 )

9 9 . 9 4 1 0 0 . 8 4

6 3

1 . 9 7 9

0 0 2 1

0 . 0 7 9 0 . 0 1 5

0 . 0 1 3 0 . 9 2 7

0 . 0 0 1 0 . 0 0 5

u . I r >

0. 1 3 l { 0 . 490

0 . 9 7 7 0 . \ \ 2

0 00r { 0 .005

0 . 0 0 1

0 . 1 0 4

1 0 0 . 1 7 r 0 0 . 7 0 9 8 . 9 r 1 0 0 . 3 5

6 1

1 . 9 7 5

0 .025

0 . 0 7 6

0 . 0 r 5

0 . 0 1 0

0 . t 4 1

0 . 9 9 8

0 . 0 0 3

o.655

0 . t 0 ,

3

o . 0 2 8

0 . 9 0 9

0 . 0 2 5

0 . i l 3

0 .464

0 . 4 5 5

0 . 0 0 3

0 . 0 0 2

1 .977

0 . 0 2 3

0.076

0 . 0 1 5

0 . 0 1 0

0 . r 3 8

0 . 9 9 3

o . o o 4

0.653

0 . r 0 9

0 . 0 2 8

0 . 9 1 4

0 . 0 2 5

0 . r 0 5

o . \ 7 5

o . \ 4 7

0 . 0 0 4

0 . 0 0 2

4 . 0 o l l 2 . 0 0 0 4 .oo r 2.000 ? aoa 2 . 0 0 0

* A l l F e d e t e r m l n e d a s F e i o x i d e s b a s e d d c r y s t a l c h e m l s t r y .l o e r r o r s i n O .

Mineral chemistry

Coexisting minerals in the xenoliths from kinber-lite and the cumulate norite were analyzed with aMAC 400 S electron microprobe utilizing the correc-tion procedures of Bence and Albee (1968) and thedata of Albee and Ray (1970). All minerals werefound to be extremely homogeneous. When grainsize permitted, four to ten analyses were made oneach grain and an average analysis and lo standarddeviation computed. A bulk analysis was made onpyroxenes containing fine lamellae utilizing standarddefocused beam analysis (DBA) techniques (50 pmdiameter beam). Representative average analyses ofminerals are included in Tables I through 3.

N odular intergrowths from kimberlite

Representative average analyses of clinopyroxenesand ilmenites from the three nodular intergrowthsfrom the Elliott County kimberlite are presented inTable l. The clinopyroxenes contain 73.0-76.8 mol.VoDi in solid solution and are quite similar in composi-

GA RRI SO N AN D TAYLO R: PYRO XEN E-O XI DE I NTERGROIYTH S

Table 2. Representative average analys€s ofcoexisting minerals from the ilnenite-spinel websterite xenolith

729

Cl inopyroxene 0 r thopyroxene Exsol ved Phases Coex i s t i nq PhasesBu l k (DBA) Hos t B u l k ( D B A ) H o s t I ' l agne t i t e Sp ine l S p i n e l I I men i te l.lagnet i te

s i 0 2 4 9 . 6 ( 1 6 )

r i 0 2 0 , 6 6 ( 7 )

Af2o3 7 .72(96)C r r l , 0 . 0 5 ( 3 )

Fe203**

Feo 5 .29 (89)

Mno 0 .09 (4 )

t l g o 1 4 . 8 ( 1 1 )

N i 0

CaO 20.\ (23)

N a r o 1 . 2 0 ( t 6 )

5 3 . 8 ( 2 ) 5 4 . 3 ( r )0 . 1 4 ( 4 ) o . r 3 ( 2 )4 . 8 0 ( r 6 ) 3 . 9 1 ( 4 )

' | . . r - , r , . | . . ' | - , r ,o . 1 9 ( 3 ) o . 3 r ( 2 )

3 0 . 3 ( 3 ) 3 1 . 6 ( 2 )

0 . 8 9 ( 8 ) 0 . 2 1 ( 3 )

0 . 0 5 ( 2 ) n . d .

3 . r 5 ( 1 4 ) o . o 4 ( l )

6 . 2 7 { d 7 ) 6 0 . l ( l )

0 . 1 7 ( 1 ) o . o 9 ( l )

6 2 . 2 5 . 8 8

1 6 . r ( 2 ) 1 5 . 2 ( 2 )

0 . 2 9 ( 1 ) o . 0 9 ( 4 )

1 2 . 6 ( 5 ) 1 6 . 7 ( 1 )

o .20 (4 )

4 4 . 4 ( 3 ) l . 5 0 ( 2 )

0 . 2 2 ( 1 ) 3 . 1 7 Q )o.08( 1 ) o .2 \ (2 ' , t

r 8 . 5 6 3 . 1

3 0 . 9 ( 4 ) 3 0 . 3 ( 5 )

0 . 3 2 ( 3 , o . 1 9 ( 3 )

5 .52(7) 2 .09Q)

l_11"' l_ll"'

5 0 . 3 ( 3 )

0 .74 (2 )6.20(15)n . d . r t

4 . 9 1 ( 1 3 )

0 . 1 2 ( 3 )

13 .7 Q l

2 t . 8 ( 3 )

1 . 3 3 ( 8 )

, . r r , t ,6 1 . 8 ( 2 )

0 . 1 3 ( l )

5 . 7 212.6 (z )

o . o 8 ( 2 )

1 8 . 9 ( 3 )

o-.-,-:,,,

r 9 9 . 8 tCat i onB a s i s 6

s i r . 8 1 9

9 9 . 1 0

6

t .863

0 . 1 3 7

0 . 1 3 2

t:: : :

0 . 1 5 1

o .758

a: : : :

0 .864

0 . 0 9 5

100.77

6

r . 8 8 2

0 . 1 1 8

0 . 0 7 9':::1

0 . 3 0 8

1 . 5 7 9

a: : : :

0 . 0 3 3

0 . 0 0 3

1 0 0 . 5 5

6

1 ' 8 9 6

0 . 1 0 4

0.057

' :::]

0.292

1 . 6 4 4

0 . 0 0 8

':::1

1 0 0 . 7 8

4

0 . 2 \ 9

0 .080

0 . 0 0 5I < 7 q

0 . 4 5 2

0 . 6 3 1

0 .008

99.30

4

1 . 8 6 6

0 . 0 0 1

0 . 0 0 2

0 . 1 3 6

0 . 3 3 4

o.655

0.002

0 .004

99.58

_.11 . 8 7 8

0 . 0 0 1

0 . 0 0 2

0 . 1 1 1

0 . 2 7 t

o . 7 2 8

0 . 0 0 2':::1

1 0 0 . 0 1

. : _

0 . 0 0 6

0 . 8 1 6

0 . 0 0 1

0 . 3 4 0

0 . 6 3 1

0 . 1 9 8

0 . 0 0 7

' :::]

100 . 7tf

..1.0 . r 3 80 .0410 .0071 .75\0.9350 . r 1 50 .006a::::

A r l v o . r 8 r

A f v l o . B 2

T i 0 . 0 1 7

C r 0 . 0 0 1?+

Fe'

F .2+ o . t 6 t

Hs 0 .809

Hn 0 .002

N I

C a 0 . 8 0 2

N a 0 . 0 8 5

E \ .029 \ .023 4 . 0 1 0 4 . 0 1 1 3 .000 3 .000 3 .000 2 .000 3.000

(1980; equation 13) can be used to estimate the equil-ibration temperature of the clinopyroxene-itnenitesystem. Using Bishop's geothermometer, temperatureestimates for the Elliott County nodular intergrowthsrange from 1520o-1560'C (at P: 50 kbar), a rangesomewhat higher than obtained using the Di-Ensolvus (i.e., 1295"-1335"C). Bishop (1980) noted asimilar discrepancy for other "high temperature" in-tergrowths; he attributed this to the non-apptcabilityof the two pyroxene solvus to these systems due tothe absence of orthopyroxene. It seems more plau-sible that this discrepancy is the result of errors in-troduced in the clinopyroxene-ilmenite geother-mometer by the reduced activity of FeTiO, inilmenite due to the presence of substantial FerO, insolution and by the reduced activity of CaMgSirOu inthe pyroxene due to the subcalcic nature of the inter-

growth clinopyroxenes, although most likely all threefactors contribute to the discrepancy. When the cal-culations are repeated with X$"r,o, : F e*2 / F e'2 *Mg*'+ 0.5Fe*'and XBi* : (2Ca/Ca+Mg+Fe) (MglMg*Fe), the temperature estimates range from1210"-1245"C, values somewhat more consistentwith the two pyroxene solvus temperature estimates.Since the true efects of Fe*3 in ilmenite and sub-calcic clinopyroxene on the experimentally cali-brated Mg-Fe*' geothermometer of Bishop (1980)are not known, these revised calculations must beonly considered illustrative.

I lmenite- spinel webst erite

Representative average analyses of coexistingphases from the cumulate websterite are presented inTable 2. The pyroxenes are very aluminous and rela-

730 GARRI SO N AN D TAY LO K PY ROXEN E-O XI DE I NTERGROWTH S

Table 3. Representative average analyses of coexisting minerals from the Chester cumulate norite

cl I nopyroxene

h o s t b u l k ( D B A )

o rthopy roxene I ane I I ae

h o s t b u l k ( D B A ) i l m e n i t e

o r i m a r y p h a s e s

i l n e n i t e m a g n e t i t e i l r e n i t e m a g n e t i t e s p i n e l

s i o 2 5 l . t ( 3 )

r i 0 2 0 , 8 2 ( 1 2 )

A r2o3 3 .77 ( | 5 )

c r z 0 3 0 . 1 4 ( 2 )

FerOr*

Feo 5 .49 (15 )

Mno 0 . 19 (2 )

MgO 14 .9 ( 2 )

N i 0

Cao 22. \ (2)

Na20 0.75e)

\ 9 . 7 ( 5 \ s 4 . 5 Q )1 . 2 2 ( 1 7 ) 0 . l 5 ( 2 )\ .zs3z) z .2 t (5 )

0 . 1 4 ( 2 ) n . d .

7 . 3 r u 2 5 ) 1 2 . 5 ( l )

0 . 2 1 ( 2 ) 0 . 3 5 ( 1 )t 5 . 9 ( 6 ) 3 0 . 1 ( 1 )

20.5 r59l 0.38(4)o . 8 0 ( 1 1 ) o . 0 3 ( 1 )

53.6 (3)

0 . 3 0 ( 1 0 ) 4 9 . 6 ( 5 )

2 . 3 8 ( 7 ) o . t 2 ( 1 )n . d . 0 . 1 3 ( 1 )

5 . 7 91 3 . r ( 6 ) 3 8 . 0 ( 5 )

0 . 3 5 ( 2 ) 1 . 4 9 ( 5 )2e.3 Q) 2.99Q)

n . d .

0 . 8 7 ( 1 4 )

0 . 0 3 ( l )

0 . 1 6 ( 5 ) 5 0 . 5 ( 1 )

0 . 3 3 ( 3 ) o . 0 9 ( 1 )0 . 5 8 ( 1 2 ) o . 0 4 ( 2 )

6 7 . 7 6 . 6 8

3 0 . 0 ( 5 ) 3 8 . 6 ( 5 )

o . 0 6 ( 2 ) o . 6 5 ( 1 )

0 . 5 0 ( 2 ) 3 . 7 0 ( 2 )o_'_:0"'

l_Tt"

0 . 1 6 ( 9 ) 0 . t 7 ( 3 )

0 . 4 8 ( t 4 ) 6 4 . 4 ( 2 )

o . 8 0 ( 3 ) 0 . 8 8 ( 5 )

6 5 . 8 0 . 0 0

30 .4 ( 6 ) 15 .5 Q )

0 . 0 5 ( 2 ) 0 . 1 4 ( 2 )

0 . 3 6 ( 9 ) 1 7 . 2 ( 1 )

0 . 2 1 ( 1 ) o . 1 6 ( 1 )

5 1 . 7 0 )

0 . 1 I ( 3 )

0 . 0 5 ( 1 )? q q

38.6 (2)

1 . 4 8 ( 4 )

3 .79G)0 . 0 5 ( l )

t 99.s6C a t i o nB a s i s

o

s i 1 . 8 9 2

A l l v o . r o 8

A t v l 0 . 0 5 6

T l 0 . 0 2 2

C r 0 .003

F"3+

Fe2+ 0 .169

i ls 0.822

I'ln 0 .005

N i

Ca 0 .887

N e 0 . 0 5 3

r 0 0 . 0 7

I . 845

0 . t 5 5

0 . 0 3 0

0 . 0 3 4

0 . 0 0 3

0 .228

0 . 878

0 .005

0 . 8 1 3

0 .057

100.22

6

1 . 9 3 3

0 . 057

0 . 024

0 . 003

0 . 3 7 1

1 . 588

0 . 0 1 0

0 . 0 1 3

0 . 001

99 .93

6

1 . 917

0 . 083

0 . 0 r 7

0 . 007

0 . 390

1 . 5 6 2

0 . 0 1 0

0 .032

0 .002

98 . t 2

0 . 0 0 4n a ? q

0 . 0 0 3

0 . 1 1 0

0 .800

0 . 1 1 2

0 .032

99 .77

0 . 0 0 3

0 . 9 5 8

0 . 0 0 1

0 . 074

0 .793

0 . 1 3 9

0 . 0 3 1

0 . 001

o o E ?

4

0 . 0 r 50 .0050 . 0 1 8t . 9 5 0o.9650.o290 .0020.006

r 0 0 . 3 r

3

0 . 0 0 3n o 2 ,

0 . 0 0 1

0 . 1 2 6

0 . 788

0 . 1 3 5

0 . 0 1 4

0 . 0 0 1

99 .26 98 .45

4 4

0 .022 1 .972

0 . 0 0 5 0 . 0 0 3

0 . 0 2 4 0 . 0 1 7

1 . 9 3 8 0 . 0 0 0

0 .98 r 0 .336

0 . 02 1 0 .655

0 .002 0 .003

0 . 0 0 7 0 . 0 0 3

4 . 0 t 7 4 . 049 4 . o r o 4 . 020 2 .000 2 .000 3 . 000 2 .000 3 .000 3 .000

* A l l F e d e t e r m l n e d a s F e : o x l d e s b a s e d o n c r v s t a l c h e m i s t r v . l o e r r o r s i n O . n . d , = n o t d e t e c t e d ; < 0 . 0 3 2 .

tively Cr-poor (<0.05 wt.Vo CrrOr); the clinopyroxene(MglMg+Fe : 0.83) contains 8.5 mol.7o CaTs(CaAlrSiO.), 7.3 mol.Vo Jd (NaAlSizO.), and 75.6mol.Vo Di-Hd (Ca(Mg,Fe)SLO.); the orthopyroxene(Mg/Mg+Fe : 0.85) contains 7.8 mol.Vo MgTs(MgAlrSiO) and 0.7 mol.Vo Di-Hd. Defocused beamanalyses (DBA) of both clinopyroxene and ortho-pyroxene reveal more aluminous and more Fe-richcompositions. The bulk clinopyroxene (Mg/Mg+Fe: 0.83) contains 12.2 mol.Vo CaTs, 6.5 mol.Vo Jd, and65.9 mol.clo Di-Hd; the bulk orthopyroxene (Mg/Mg+Fe : 0.84) contains 9.6 mol.Vo MgTs and 3.3mol.Vo Di-Hd.

The pyroxenes from the ilmenite-spinel websteriteare generally more Fe-rich than pyroxenes associatedwith nodular intergrowths from kimberlite (Figure3); generally they are similar to lamellae-bearing py-roxenes from numerous cumulate rocks. The web-

sterite clinspylexsls is quite similar to the pyroxenesdescribed by Deer and Abbott (1965), Moore (1971),Schulze et al. (1978), and Morse (1969), except thewebsterite clinopyroxene is generally higher in TiOr,NarO, and AlrO3 and lower in Cr'Or. The websteriteclinopyroxene has a Mg/Mg+Fe ratio (i.e., mg)within the ranges of those found for lamellae-bearingclinopyroxenes from cumulate rocks, although theMg/Mg+Fe ratio varies depending on the locationwithin the layered complexes. Likewise, the webster-ite orthopyroxene is similar to lamellae-bearing or-thopyroxenes from other cumulate rocks (Moore,l97l and Schulze et al., 1978), although generallylower in CrrO, and NarO. When compared to ortho-pyroxenes intergrown with "myrmekitic" oxides, thewebsterite orthopyroxene is much higher in AlrO3and lower in TiO, and CaO (Haselton and Nash,r97s).

50 I*

GARRISON AND TAYLOK PYROXENE-OXIDE INTERGROWTHS 731

Websterite l-29

* bulk Cpx* hosl Cprr hrlk opro host opx

Coars€ Intsrgrowths

O CpxI{or i te C-14

r bulk Opxa host opxa bulk Cpro host Cpr

Co. opx m€gacrysts

of opx- llmenita in Kimborlits

Mg 50 FeFig 3. Atomic Ca-Mg-Fe of pyroxenes from the nodularclinopyroxene-ilmenite intergrowths from the Elliott Countykimberlite, the iknenite-spinel websterite l-29, and the Chestercumulate norite C-14. The range of compositions for the ElliottCounty pyroxene megacrysts and orthopyroxenes in pyroxene-ilrnenite nodules from kirnberlite (Rawlinson and Dawson, 1979)are shown for comFarison.

Utilizing the petrogenetic grid of Herzberg (1978),temperature estimates obtained for bulk and hostclinopyroxenes are ll30'C and 1020"C, respectively,at P : 15 kbar. In rocks where orthopyroxene coex-ists with olivine and spinel, the AlrO, content of theorthopyroxene is relatively insensitive to pressureand serves effectively as a geothermometer (Obata,1976; Fujii, 1976; Danckwerth and Newton, 1978).Assuming that olivine, although not observed, is astable phase of this assemblage, the AlrO, content ofthe host orthopyroxene (3.91 vtt.Vo Al2O3) suggestsequilibration at l060oC (Fujii's data) to l0l5.C(Danckwerth and Newton's data) and the bulk com-position (4.80 tvt.Vo AI'OJ suggests equilibration atll40'C (Fujii's data) to ll05.C (Danckwerth andNewton's data).

The ilmenite from websterite contains 19.8 mol.VoMgTiO, and 17.0 mol.Vo FerOr. The ilnenite containsmore Fe2O3 and FeTiO, than the ilmenites from thenodular clinopyroxene-ilmenite intergrowths (Figure4), actually quite sinilar to the ilmenite lamellaefrom the garnet pyroxenite studied by Schulze et al.(1e78).

The interstitial magnetite contains 88.5 mo.7o(Fe,Mg)FerOo, 7.0 ntol.Vo (Fe,Mg)AlO o, and 4.1nol.Vo (Fe,Mg)rTiOo; the interstitial pleonaste con-tains 5.5 mol.Vo (Fe,Mg)FerOo, 94.3 moI.Vo(Fe,Mg)AlrO4, arLd 0.1 mol.Vo (Fe,Mg)rTiOo. Both

contain only minor amounts of the (Fe,Mg)CrrOomolecule; the magnetite contains 0.4 mol.Vo, and thepleonaste contains 0.1 mol.Va The fine lamellaewithin the pyroxenes have compositions quite closeto the primary interstitial spinels; the pleonaste la-mellae contain 6.8 mol.Vo (Fe,Mg)FerOo,93.0 mol.Vo(Fe,Mg)AlrOo, 0.1 mol.Vo (Fe,Mg)CrrO4, and 0.1moI.Vo (Fe,Mg)rTiOo and the magnetite lamellae con-tain 79.1 mol.Vo (Fe,Mg)FerO., 12.6 noI.Vo(Fe,Mg)AlOo, 0.3 mol.Vo (Fe,Mg)CrrOo, and 8.0mol.Vo (Fe,Mg),TiOo. The spinel lamellae are dis-tinctly lower in CrrO, than previously reported spinellamellae in pyroxene (e.g., Basu and MacGregor,1975; Haggerty, 1972).

The coexisting interstitial magnetite and iLnenitesuggest equilibration at 695"C and an fo,: lg-tta(re., QFM + 4.0) (Buddington and Lindsley, 1964;Lindsley, 1978). This temperature is over 300'Clower than the temperatures estimated for the ex-solved pyroxenes, suggesting that the oxides contin-ued to reequilibrate to lower temperature conditions.The ilmenite contains extremely fine lamellae ofhematite (-l pm thick) which, if the ilmenite wasisochemical, suggests further reequilibration down toas low as 400oC. It is interesting to note that the in-terstitial spinel and magnetite lie on the magnetite-

Wehsterite l-29

* primary llncnite

Coarse Intorgrowihs

O graphic llmenite

Nor i te C-14

A primary l lmcnite

I myrmekitic

O lamellae

s-o oaA

FeTiO3 50 MeTio3Fig 4. A ternary plot of Fe2O3-FeTiO3-MgTiO3 showing thecompositions of ilmenites from the nodular clinopyroxene-ilrnenite intergrowths, the iknenite-spinel websterite xenolith 1-29, a'I.d the Chester cumulate norite C-14. The compositionalranges of ilmenite megacrysts from Elliott County and ilmenitesfrom Monastery intergrowths (Haggerty et al., 1979); S = ilrnenitefrom the Skaergaard intergros'th and 60016 : ilmenite from60016 intergrowth (Haselton and Nash, 1975). GW : ilmenitelamellae in clinopyroxene from garnet websterite studied bySchulze et al. (1978).

Co

732 GARRI SO N AN D TAYLO R: PY ROXEN E-O XI DE I N?:ERGROWTH S

(4 :6)

(3:4)

iligh P-T tp tt low p.f

Fig. 5. Schematic diagram ofthe solid solution ofpyroxene and spinels at high P-Tand low P-Z Stippled area are ranges of solidsolutions. Numbers in ( ) refer to cation/oxygen ratio of end-members. Px: pyroxene; Mt = magnetite; Sp : pleonaste. Note the rangeof nonstoichiometry implied by the solid solutions (see text for discussion of non-stoichiometry).

PI

tt 8p

spinel solvus (Turnock and Eugster, 1962) at aboutthe 500"C isotherm. This seems to indicate that theinterstitial spinel phases also continued to reequili-brate, although it is realized that in utilizing thissolvus we are extrapolating from a 2 component sys-tem into a multi-component system, and the effectsof MgO on this solvus are not completely known.

Cumulate norite

Representative average analyses of coexisting min-erals from the cumulate norite sample from the Ches-ter pluton are presented in Table 3. The norite clino-pyroxene is somewhat lower in AlrO, and NarO thanthe clinopyroxene from the cumulate ilmenite-spinelwebsterite; the host clinopyroxene (mg :0.83) con-tains 5.6 mol.Vo CaTs,2.7 mol.Vo Jd, and 81.3 mol.VoDi-Hd. The buft analysis (DBA) of the clinopyrox-ene (mg: 0.79) is richer in AlrO3 than the exsolvedhost; it contains 6.3 mol.Vo CaTs, 1.9 mol.Vo Jd, and7l.l mol.Vo Di-Hd. As expected from the presence ofthe ilmenite lamellae, the bulk clinopyroxene is moreFe-rich and has more NaTiAlSiOu molecule (3.2mol.Vo) than the host (2.2 mol.Vo). The cumulate no-rite orthopyroxene is lower in AlrO3 31d higher inFeO than the orthopyroxene from the ilmenite-spinel websterite (Table 2). The host orthopyroxene(mg : 0.81) contains 4.4 nol.Vo MgTs and 1.3 mol.VoDi-Hd; the bulk orthopyroxene (mg: 0.80) is morecalcic and contains 4.6 mol.Vo MgTs and 3.1 mol.VoDi-Hd. Temperature estimates for bulk and hostclinopyroxene, estimated from the petrogenetic grid

of Herzberg (1978), are l200oC and 1050'C, respec-tively, at P :4-5 kbar (also estimated from the pet-rogenetic grid).

The pyroxenes from the cumulate norite are gener-ally more Fe-rich than the pyroxenes from the nodu-lar pyroxene-ilmenite intergrowths from kimberlite(Figure 3); they are quite similar to lamellae-bearingpyroxenes from other cumulate rocks (Deer and Ab-bott, 1965; Morse, 1969; Moore, l97l). The hostclinopyroxene from the Chester norite is quite simi-lar to other lamellae-bearing clinopyroxenes fromcumulates, except that it is higher in TiOr, althoughclinopyroxene bulk compositions (1.e., with includedilmsnils) such as those reported by Deer and Abbott(1965) have similar ranges of TiO,. The clinopyrox-ene from the intergrowths in the garnet pyroxenitestudied by Schulze et al. (1978) is lower in Al,O. andricher in NarO. The orthopyroxene from the "myr-mekitic" intergrowth in Apollo 16 breccia 60016 (Ha-selton and Nash, 1975) is similar in composition tothe Chester norite orthopyroxene.

The primary intercumulus ilmsnils in the Chesternorite is virtually identical to the fine ilmenite la-mellae in the pyroxenes and the itnenites in the"myrmekitic" intergrowths (Figure 4); the inter-cumulus itmenite contains 6.4mol.Vo FerO. and 13.7mol.Vo MgTiO.. The ilmenite from the "myrmekitic"intergrowths has 3.8 nol.Vo FerO, and 14.3 mol.VoMgTiOr; the ilmenite lamellae contain 5.7 moI.VoFerO, and ll.6 mol.Vo MgTiOr. These ilmenites havesimilar FerO, to the ilmenites from the nodular inter-

GARRISON AND TAYLOR: PYROXENE-OXIDE INTERGROWTHS

growths from kimberlite, although the FeTiO, mole-cule is more abundant. The ilmenite from the "myr-mekitic" intergrowth is intermediate in compositionto the ilmenites from the "myrmekitic" intergrowthsfrom breccia 60016 and the Skaergaard (Figure 4),but quite distinct from ilnenites from the webste-rites.

The intercumulus magnetite from the Chester no-rite is similar to the magnertite from the "myrmekitic"intergrowths in the norite; the intercumulus magnet-ite contains 97.2 moL%o (Fe,Mg)FerOo, l.l mol.Vo(Fe,Mg)AlOo, 0.5 molVo (Fe,Mg),TiOo, and 1.2mol.Vo (Fe,Mg)CrrOo. The "myrmekitic" magnetitehas 97.8 mol.Vo (Fe,Mg)FerOo, 0.8 mol.Vo(Fe,Mg)AlrOo, 0.5 mol.Vo (Fe,Mg)rTiOo, and 0.9mol.Vo (Fe,Mg)CrrOo. The intercumulus pleonastecontains 98.8 mol.Vo (Fe,Mg)AlrOr, 0.3 mol.Vo(Fe,Mg)rTiOo, and 0.9 mol.Vo (Fe,Mg)CrrOo; no(Fe,Mg)FerOo is present. These magnetites containless (Fe,Mg)AlrOo and more (Fe,Mg)CrrOo than themagnetites from the ilmenite-spinel websterite xeno-lith.

The coexisting intercumulus magnetite and ilme-nite suggests equilibration at 515oC andf.,,: l0-2r'o(r.e., QFM + 3.0) and the coexisting magnetite and il-menite from the "myrmekitic" intergrowths suggestsequilibration at 480oC andf.,, - 10-23? (1e., QFM +2.0). This seems to indicate that all Fe-Ti oxides ree-quilibrated to similar T-fo, cor.ditions. The co-existing intercumulus magnetite and spinel lie on themagnetite-spinel solvus (Turnock and Eugster, 1962)at the 350oC isotherm. Again, as in the case of theElliott County websterite, the temperature estimatesobtained from the magnetite-ilmenite geothennome-ter-oxygen barometer of Buddington and Lindsley(1964) and Lindsley (1978) are substantially lowerthan estimates obtained from exsolved pyroxenes; thetemperature estimated from the magnetite-spinelsolvus is again lower than the temperature estimatesobtained from the magnetite-ilnenite geothennome-ter-oxygen barometer.

Discussion

Exsolution or co-precipitation?

Nodular clinopyroxene-ilmenite intergrowths fromkimberlite. The nodular clinopyroxene-ilmenite in-tergrowths from kimberlite generally have mineralcompositions that overlap the compositional rangesof discrete megacrysts from the host kimls1fite, al-though the intergrowth compositional ranges aregenerally more restricted than the megacrysts

(Mitchell, 1977). Even though discrete ilmenitemegacryst compositions vary from one kimberlitepipe to another, the ilmenites from the nodular py-roxene-ilmenite intergrowths show major elementcompositional similarities. Although Mitchell (1977)found that both major and trace element data do notindicate any simple relationship between megacrystsand intergrowth phases, this compositional similarityseems to suggest that the nodular pyroxene-ilmeniteintergrowths are genetically related to the discretemegacrysts, or, alternatively, the mineral phases havereequilibrated to the same P-T-fo" conditions. It canbe argued that the ilmenites in the intergrowths havebeen enclosed in the pyroxene and efectively iso-lated, hence their compositions reflect their initialcompositions upon crystallization, barring reequili-bration of the pyroxene-ilmenite Ko.

Gurney et al. (1973) pointed out some very seriousproblems with the models that indicate the fonnationof the nodular pyroxene-ilmenite intergrowths fromkimberlite by either the eutectoidal transformation ofgarnet (Ringwood and Lovering, 1970) or the ex-solution of ilmenite from an "ilmenite-structured"pyroxene (Dawson and Reid, 1970). The main argu-ment against both models is the difficulty each modelhas in explaining the high modal abundance of itne-nite, in light of the fact that at high pressures, pyrox-enes (and garnets) have limited solubility of TiO,(Akella and Boyd, 1973); no such high-TiO, pyrox-enes or garnets have been recovered from kimberliteor found as inclusions in diamonds (Meyer andBoyd, 1972). Gurney and co-workers also arguedagainst the "eutectoidal breakdown of garnet" modelbecause these nodular intergrowths have Ta, Sc, andREE abundances inconsistent with formation from aprecursor garnet.

The graphic (or cuniform) texture of many of theseintergrowths (e.g., Figure l) is difficult to explain interms of exsolution; in exsolution there is little driv-ing force to make an exsolved phase anything butplanar or rod-like in shape. Schulze et al. (1978) ar-gue that the lamellar pyroxene-ilmenite intergrowthsfrom their garnet pyroxenite xenoliths, as well as thepyroxene-ilmenite intergrowths from kimberlite,probably formed by exsolution. Although there arelarge diferences in scale and pyroxene to ilmenite ra-tio between these two occurrences, they stated thattextural similarities suggested the same process---€x-solution. They further state that, although two pyrox-ene solvus temperature estirnates for the host pyrox-e n e s a r e h i g h ( i . e . . , I 1 5 0 " - l 3 5 0 o C ) , t h e l o wtemperature estimates (930'-950'C) based on the

GA RRI S O N A N D TA Y LO R: P YRO XEN E- O XI D E I N TE RGRO WTH S

empirical geothermometer for coexisting ilmeniteand pyroxene (Anderson et al.,1972) argue for a sub-solidus process. It would be difficult for a pyroxene toexsolve a Mg-rich phase (i.e., picroilmenite) at low-T and still retain a Di content suggestive of high-?equilibration. This discordancy could be the result ofsubsolidus reequilibration of the ilmenite and/orproblems with the empirical geothermometer. As dis-cussed earlier, the clinopyroxene-ilmenite geother-mometer of Bishop (1980) provided higher temper-ature estimates for the Elliott County nodularintergrowths than those estimated from the two py-roxene solvus of Lindsley and Dixon (1976).

As pointed out by Boyd (1971), these intergrowthshave textures similar to the eutectic textures pro-duced in alloys by controlled growth. Wyatt (1977)succeeded in experimentally producing a eutectic-like intergrowth of clinopyroxene and ilmenite simi-lar in texture, cgmposition, and crystallogmphy tothe nodular intergrowths from kinberlite. In the nat-ural system that he studied, Wyatt determined thatthe liquidus phase is clinopyroxene which is followedby clinopyroxene + ilmenite. The quench pyroxene-ilmgnils xenolith studied by Rawlinson and Dawson(1979) further supports the co-precipitation of pyrox-ene and ilmenite in natural systems. The composi-tional similarities of phases in the nodular inter-growths of megacryst phases in host kimberlite is alsosupportive of co-precipitation of pyroxene and ilms-nite. In the Elliott County kimberlite and in the Mon-astery Mine kimberlite (Gurney et al.,1979), clinopy-roxene megacrysts general ly have higherequilibration temperatures (by as much as 50oC)than the clinopyroxenes from intergrowths; this isconsistent with the phase relations outlined by Wyatt(1977), although since Wyatt's experiments are an-hydrous, direct temperature comparisons cannot bemade (i.e., the kimberlite system was probably water-rich).

Frick (1973) discussed eutectic crystallization yer-sas cotectic crystallization and concluded that cotec-tic precipitation best explains the intergrowth oc-currences because it is difficult to explain bothorthopyroxene-ilmenite and clinopyroxene-ilmeniteintergrowths as eutectic intergrowths produced fromthe same (proto-kimberlitic) magma, which probablyhas a fairly constant Mg/Fe ratio. The join ilmenite-clinopyroxene is probably pseudobinary rather thana binary eutectic as discussed by Wyatt (1977). Frickexplained the intergrowth texture as the result of"skeletal' ilmenites enclosed by rapidly co-precipi-tating pyroxene.

Experimental and chemical data are over-wheLningly in favor of a co-precipitation origin forthe coarse nodular pyroxene-ilmenite intergrowthsrecovered from kimberlites such as the ElliottCounty kimberlite pipes. The variations in texturefrom coarse lamellar to coarse graphic are probablyrelated to factors such as cooling rate and growthmode (e.9., planar).

"Myrmekitic" pyroxene-oxide intergrowths. Eventhough there are is textural similarity between thenodular pyroxene-ilmenite intergrowths and the"myrmekitic" pyroxene-oxide intergrowths fromcumulate rocks, their mineral compositions do notoverlap. The "myrmekitic" intergrowths are mostcommonly associated with orthopyroxene. Theoxides of the "myrmekitic" intergrowths have a widerange of compositions that overlap the interstitialoxides in their respective host rocks-a situationquite analogous to the similarity of ilnenites fromnodular intergrowths to discrete ilmenite megacrystsfrom their respective host kimberlites. Again, thissuggests a genetic relationship and/or overall reequi-libration of all phases. In the case of the Chestercumulate norite, even the fine lamellar ilnenitewithin pyroxene has a composition similar to the in-terstitial ilemite. This argues that reequilibration isthe more viable alternative for such cumulate rocks.

As in the case of the graphic pyroxene-ilmenite in-tergrowths from kimberlite, it is extremely difficult tovisualize the formation of the irregular, non-oriented"myrmekitic" intergrowths by an exsolution process.Basu and MacGregor (1975) concluded that these"myrmekitic" or "symplectic" intergrowths are theresult of exsolution because in one sample they notedsmall lamellae, interpreted to be exsolved, merginginto the wormy intergrowths. Based on our study ofthe Chester cumulate norite C-14. we believe this tex-ture to be entirely a coincidence; in the Chester no-rite, the fine lamellae are ilmenite, and the "myrme-kitic" intergrowths are dominantly magnetite, andthere appears to be no textural evidence ofa geneticrelationship. The fact, pointed out by Basu and Mac-Gregor, that the lamellae, wonny intergrowths, andinterstitial phases have similar compositions may bea consequence of continued subsolidus reequilibra-tion of all oxide phases.

The "myrmekitic" intergrowths are similar in tex-ture to intergrowths produced in the eutectic crystal-Ezation of alloys, in eutectic crystallization in the sys-tem NaAlSi3O.-SiO, ( i .e. , myrmekite), and ineutectoidal breakdown. The occurrence of a widerange of oxide compositions must be considered in

GA RRI S O N A N D TA YLO R PY RO XE N E. O X I D E I N TE RGRO WTH S

any co-precipitation model invoked to explain theseintergrowths.

An examination of the system MgrSiOo-CaSiOr-SiOr-FerOo (Osborn, 1962) at moderate,fo. revealssome interes,ling phase relations. A cotectic surfaceexists between low-Ca pyroxene and magnetite andbetween diopside and magnetite; a reaction surfaceexists between pyroxene and olivine. Phase relationsin this system show that a liquid of basaltic composi-tion will fractionate by crystallizing phases in the se-quence: olivine, low-Ca pyroxene, low-Ca pyroxene* magnetite,low-Ca pyroxene + diopside + magnet-ite; in hydrous systems amphibole and biotite follow(Osborn, 1962). In systems containing abundantCrrOr, Al2O3, or TiOr, it is possible to have co-precipi-tation of pyroxene and chromite, spinel, or ilmenite;/", probably also controls the composition of theoxide species.

The textures observed in the Chester norite areconsistent with the phase relations discussed above.The orthopyroxene and magnetite (+ ilmenite)"myrmekite" follows olivine (Forn) and plagioclase inthe crystall.ization sequence; it commonly occurs as acorona around olivine. as well as within the centralportions of large orthopyroxene grains quite distantfrom olivine. In many areas where the "myrmekitic"intergrowths are proximal to olivine, the width of thewormy oxide within the intergrowths becomes nar-rower as the otvine grain boundary is approached.This texture suggests, that in these areas, the inter-growth grew toward the olivine, as would be ex-pected if the intergrowth developed by reaction ofolivine with liquid. It does not appear that the occur-rence of these intergrowths is necessarily related tothe back-reaction of olivine with the liquid as wouldbe expected if the liquid path intersected the reactioncurve bounding the fields of low-Ca pyroxene, oli-vine, and magnetite; olivine is commonly surroundedby orthopyroxene, clinopyroxene, or amphibolewithout magnetite-ilmenite and orthopyroxene myr-mekite. Whether the intergrowths form by reaction ofolivine with liquid or by direct precipitation from theliquid is simply a function of the bulk composition ofthe magma. It is possible that within heterogeneouscumulate magmatic systems, both mechanisms maybe operating on a local scale. It is not clear why it isonly along the pyroxene * oxide cotectic surfacesthat the "myrmekitic" intergrowths are developedand not along other such cotectic surfaces; the forma-tion of these intergrowths is probably controlled bysssling rate as this affects the kinetics of crystalliza-tion (i.e., nucleation and growth).

Fine lamellar pyroxene-oxide intergrowths. Thecommon denominator between the many occur-renoes of very fine lamellar intergrowths of oxideswithin pyroxenes is more elusive. Even though bothpyroxenes and oxides have a wide range of composi-tions, the mode of occurrence is the same-all appearto occur as 1-5 trr thick, crystallographically orientedlamellae in pyroxenes from rocks believed to haveundergone extremely slow cooling. The wide range ofpyroxene compositions suggests final equilibrationover a wide range of P-T conditions and initial crys-tallization from liquids with a variety of Mg/Mg+Feratios. Thus, it appears that this phenomenon is notunique to any one specific liquid composition fromwhich the pyroxenes originally crystallized. For ex-ample, they occur in cumulate layers from variouslevels within layered complexes, reflecting crystalli-zation of pyroxene over a wide range of liquid com-positions (Deer and Abbott, 1965). Likewise, the il-menites reflect equilibration over a wide range of ?-/o, conditions; this suggests that/o, may not be a criti-cal factor in the information of these lamellar pyrox-ene-oxide intergrowths, although the possibility ofcontinued subsolidus reequilibration makes this diffi-cult to evaluate.

The fine lamellar intergrowths of pyroxenes andoxides from both the ilmsnils-spinel websterite l-29and Chester cumulate norite C-14 have crystallogra-phic orientations and textural relations consistentwith exsolution; considering the kinetics of oxidecrystallization, it is difficult to imagine such fine-grained lamellae precipitating from a melt. The oc-currence of two spinels (i.e., magnetite and pleonaste)in the pyroxenes in websterite l-29 suggests that thelamellae formed at a temperature below the magnet-ite-pleonaste solvus; for the MgO-free system thissolvus lies below 900oC (Turnock and Eugster,1962). Since the temperature estimates for the buftclinopyroxene and orthopyroxene are 1l30oC andI I 05'-l 140"C, respectively, a subsolidus mechanismis clearly indicated. The temperature estimates forhost clinopyroxene and orthopyroxene of l020oCand l0l5o-1060oc, respectively, and that the twospinel solvus temperature estimates of 500'C wouldseem to suggest that all phases continued to reequili-brate at subsolidus temperatures, although rates anddegree of reequilibration of the pyroxene-oxide Kosare additional unknown variables.

A modelfor oxide exsolutionfrom pyroxene

The exsolution sensu stricto of a spinel phase (cat-ionloxygen :3/4) from a pyroxene (cation/oxygen

GA RRI S O N A N D TA Y LO R: P Y RO X EN E- O X I D E I N TE RGRO T}/TH S

: 4/6) requires that the original pyroxene be non-stoichiometric. The exsolution of ilmenite (cation/oxygen : 4/6) does not require pyroxene non-stoichiometry. The most simple expressions for theexsolution of an oxide phase from pyroxene areshown in equations l-3 in Table 4; each requires themutual solubility of pyroxene and oxide.

Non-stoichiometric pyroxenes have been describedfrom natural rocks (Sobolev et al., 1968) and pro-duced experinentalTy (e.9., Kushiro, 1972). Other oc-currences of non-stoichiometric pyroxenes that lie offthe pyroxene plane in compositional space (1.e., withcation vacancies) are discussed by Robinson (1980).Kushiro (1972) suggested th.at at high temperatures(-1390'C) there is limited solubility (up to 5 moIVo)of olivine (cation/oxygen3/4) in diopside. This diop-side solid solution is a non-stoichiometric pyroxene.Since Kushiro's bompositional data were obtainedwith a microprobe, it is not clear whether it is trulythe Mg,SiOo molecule dissolved in diopside(CaMgSi,O.) or whether the non-stoichiometric py-roxene is actually a solid solution of diopside and asilica deficient "pyroxene component" such asMgMgSiVOu. In this formula, the V indicates a tet-rahedral site vacancy. Sobolev et al. (1968) reported"non-stoichiometric" aluminous pyroxenes whichappeared to have an excess amount of Al; they sug-gested the presence of an "isomorphic admixture ofsome Al-rich molecule". By analogy to the observa-tions of Kushiro (1972), this excess of Al in the py-roxenes could be attributed to limited solid solutionof a spinel molecule (MgAlrO) in the pyroxene (i.e.,this is analogous to Kushiro's olivine molecule); al-ternatively, the pyroxene could have a "silicadeficient" pyroxene molecule such as MgAlrVO6(i.e., a Mg-Tschermak's molecule with tetrahedralvacancy) in solid solution. Equation I in Table 4 il-lustrates the exsolution of spinel (MgAlrO.) from asolid solution of CaMgSirOu and MgAlrO.; equation4 illustrates the exsolution of spinel from a solid solu-tion of diopside and the non-stoichiometric "silicadeficient" MgAlrVO. molecule. Equations 4-8(Table 4) illustrate the exsolution of various spinelspecies from non-stoichiometric pyroxene solid solu-tions containing various "silica deficient" pyroxenemolecules; in each case only pyroxene end-membersdirectly involved in the reactions are shown. Notethat in each case, the vacant site is a tetrahedral siteand that each reaction involves the release of Or. Al-though in Al-rich pyroxenes the non-stoichiometrycould be due to Al-deficient Ml sites. the data ofKushiro (1972) would seem to indicate that the va-

cant sites in non-stoichiometric pyroxenes are in facttetrahedral. If the non-stoichiometry of the pyrox-enes can be attributed to limited solubility of olivinemolecules in pyroxenes, we can write equations suchas equations 9-12 (Table 4) for the exsolution of var-ious spinel phases from olivine-pyroxene solid solu-tions. In each case, a Tschermak's molecule is an in-tegral part of each reaction.

It should be noted that in pyroxenes seaf4iningsite vacancies there is a charge imbalance-an un-stable configuration. It is possible that in pyroxenesthat appear to have tetrahedral site vacancies (1e.,based on microprobe data) charge balance is actuallymaintained by Ho substitution for tetrahedral Si*o asproposed by Sclar et al. (1968). Since the amount ofcharge imbalance required to account for the amountof dissolved oxide species is quite small relative to to-tal pyroxene volume, it is also possible that a smallamount of charge imbalance can be accommodatedby the pyroxene structure.

Since electron microprobe data does not allow usto distinguish whether an olivine or a spinel molecule(cation/oxygen: 3/4) or a "pyroxene molecule withtetrahedral vacancies" is dissolved in the non-stoichiometric pyroxenes, both models remain viablealternatives. The presence of O, as a product of ex-solution reactions 4-8 would seem to require that theformation of oxide lamellae from non-stoichiometricpyroxene is a reduction exsolution process. Since thehost rocks in which these lamellar pyroxene-oxideintergrowths occur appear to have crystallized undermoderate /or, we suggest that reactions 9-12 in-volving otvine-pyroxene solid solutions appear to bemore plausible. In either case, the driving force forthe exsolution of the oxide phase is the tendency ofthe pyroxene toward an ideal stoichiometric compo-sition. Figure 5 schematically displays the exsolutionof two spinels (with cation/oxygen : 3/4) from anon-stoichiometric pyroxene and the movement ofthe pyroxene toward a more stoichiometric composi-tion.

Alternatively, if spinel exsolution from stoichio-metric pyroxene is considered a possibility, the ex-solution of spinel must be accompanied by the ex-solution of an SiO, phase. No SiO, phase wasdetected in any of the pyroxenes in this study, al-though it is possible that this "excess Si" diffused outof the host pyroxene into intergrain areas and sub-sequently was removed. We do not consider this alikely model because it appears more probable thatany phase or phases exsolved from a stoichiometricpyroxene would have pyroxene stoichiometry instead

GA RRI S O N A N D TA YLO R P YRO X EN E. O XI D E I N TERGRO WTH S

Table 4. Possible equations for oxide exsolution from pyroxene

737

(l) {caMcs1206 + McAlAlO4}ss ? CaMcS12O6 + }.tcA1204

(2) {CaMcS1205 * FeFeTloO}"" z CaMcSi2O6 + Fe2T104

(3) {CaMcSl2O6 + FeFeTt206}6s ? CaMsStrOU + 2FeT10'

(4) iCaUcSl2O5 + lreAlAlvo5):s t Cal.{cS12O6 + r4cAL2O4 + 02

(5) {CaMCSl2od + FeAlAlvo5}ss i CaMCSi2o6 + FeA12o4 + 02

(6) {MgMgsl2ob + 2MsCrAlv0U}"" + McMcSl2o6 + {ugcr2o4 + MgAlzo4}ss + 2o2

(7) {FeFeSlrOU + NaTlAlVOU}"" 1 NaAlSlrOU + Fe2TtO4 + 02

(8) {Fel'eslrou + 2FeFe+3Atvoa}"" ? FeFesl2o5 + {FeA12o4 + rer{3oo}"" + zo,

(9) {)dg2sio4 + yCaMBStrOU + zCaAlrSlo.}"" ? {(x+y)CaMgsl2o5 + (z-x)CaAlrSt0.}""

(10) {)dg2sio4 + yMgugs1206 + zMeAlrsto'i 7 t(x+r)ltsltssl2o6 + (z-x)MgAlrSlo.}""

(ll) I2lersroo + 2t'eFe+3ALsiou)"" + 2FeFesl2o6 + {FeA12o4 + FeFe+3oo}""

(r2) {2MerSiOo + 2McCrALSi06}Bs i 21'rdagsrro6 + {}rccr204 + MgA1204}ss

(13) {CaMCSi2O5 + 2FeTtAlrOU}"" ? {CaA12SlO6 + MCA12S1O6}"" + 2FeT10,

(14) {FeFeS12O6 + 2CaTiAlrOU}"" : 2CaAl2SlO6 + 2FeTi03

(15) {2FerS10O + 2NaTiA1Si06}"" * bZ + 2NaAlSl2OU + {2FeT1O, * F.2O3}""

(16) {2FeFeSiroU + 2NaTlAlV0U}"" ? 2NaAlSt2OE, + {2Fer io, * F.203}"" + 3 l2oz

+ d.'lgAL2o4

+ :rMgA12O4

* V denotes a tetrahedral slte vacancy

of spinel stoichiometry. Lamellae of an "Si-richphase" such as plagioclase (cation/oxygen : 5/8)have been described in orthopyroxenes (e.g., Morse,1975); in some examples the amount of plagioclaselamellae is roughly balanced by spinel lamellae.Even in these cases, it is likely that the original py-roxenes had non-stoichiometric compositions in or-der to exsolve plagioclase (cation/oxygen : 5/8) andspinel (cation/oxygen : 3/4).[t is possible that inthese cases in which plagioclase is exsolved, the cat-ion vacancies were also octahedral.

Since ilmenite and pyroxene have the samestoichiometry, it is not necessary that the exsolutionof ilmenite from pyroxene involve pyroxene non-stoichiometry. Equation 3 (Table 4) illustrates thesimple case of ilmenite exsolution from a diopside-il-menite solid solution. Although it has pyroxenestoichiometry, the FeFeTirOu molecule is not gener-

ally consistent with pyroxene crystal chemistry; Ti*4does not enter substantially into tetrahedral sites ofpyroxenes where Si, Al, or Fe*3 is available (Dowtyand Clark, 1973; Tracy and Robinson, 1977). Themost likely site for Ti*o is in a R*'TiAl2Ou moleculesuch as FeTiAl"O" or CaTiAlrOui Ti*o occupies theMl site. Equations 13-14 (Table 4) illustrate ex-solution of ilmenite from a solid solution of pyrox-enes containing various R*2TiAl2O6 molecules. It isalso possible to write a reaction in which ilmenite ex-solves from a non-stoichiometric pyroxene repre-sented as a pyroxene-olivine solid solution, similar toreactions 9-12 discussed above (e.g., equation 15 inTable 4). This reaction differs from reactions 9-12 inthat the pyroxene solid solution contains the NATALmolecule (NaTiAlSiOu) -d requires the addition ofOr; this is in fact oxidation exsolution. Equation 16 il-lustrates the exsolution of ilmenite from a non-

GA RRI S O N A N D TA YLO R P YRO XE N E. O XI DE I N TE RG RO WTH S

stoichiometric pyroxene solid solution containing a"silica deficient" pyroxene molecule; O, is a productof this reduction exsolution reaction. Since pyroxenesthat contain ilmenite exsolution lamellae occur inrocks of similar compositions and formed at a similarrange of p-T-foz conditions as the pyroxenes thatcontain spinel exsolution lamellae, it is tempting tosuggest that pyroxene non-stoichiometry is involved,although we cannot further evaluate which of the re-actions may be involved in the exsolution of ilmenitefrom pyroxene.

Conclusions

The coarse nodular pyroxene-ilmenite inter-growths from kimberlite are products of co-precipi-tation; they are probably genetically related to thecrystallization of the kimberlite megacryst suite froma "proto-kimberlitic" liquid. Cooling rate and growthmode determine whether the intergrowths form la-mellar or graphic textures.

The "nyrmekitic" pyroxene-oxide intergrowths,commonly found in cumulate rocks and mantle pe-ridotites, are produced by (reaction and,/or cotectic)co-precipitation. Magna bulk composition and /o,control the composition of the oxide species whichco-precipitates with pyroxene (1e., Cr-spinel, mag-netite, or ilmenite).

The very fine-grained lamellar pyroxene-oxide in-tergrowths from deep-seated cumulate rocks andmantle rocks are the products of subsolidus ex-solution from high temperature pyroxenes. Spinel ex-solution is the product of exsolution from high tem-perature non-stoichiometric pyroxene; this pyroxenenon-stoichiometry can be attributed to either an "oli-vine or spinel molecule" (cation,/oxygen : 3/4) or a"pyroxene molecule with tetrahedral vacancies" dis-solved in pyroxene. Reactions involving the olivinesolid solution with pyroxene are isochemical, whilethe reactions involving the pyroxene solid solutionwith a "pyroxene molecule with tetrahedral va-cancies" require that the exsolution is a reduction ex-solution process. In either case, the driving force forexsolution of the spinel phase is the tendency of thepyroxene toward an ideal stoichiometric composi-tion. Ilmenite exsolution from pyroxene does notnecessarily require pyroxene non-stoichiometry. Il-menite exsolution from stoichiometric pyroxene in-volves a solid solution a pyroxene containing aR*2TiA12O6 molecule; this reaction is isochemical. Il-menite exsolution from non-stoichiometric pyroxene

involves either oxidation exsolution from a solid solu-tion of olivine and pyroxene containing NATAL orreduction exsolution from a pyroxene solid solutioncontaining a "silica deficient" pyroxene molecule.

AcknowledgmentsThis paper has benefitted from discussions with Drs. E. Essene,

J. Higgins, M. J. Holdaway, O.C.Kopp, D. H. Lindsley, H. Y.McSween, and P. Robinson. Constructive comments by Dr. A. C.Turnock are gratefully acknowledged. Critical reviews by Dn.D. H. Eggler and S. E. Haggerty greatly improved this paper. Thepyroxene short course by the MSA proved invaluable. Dr. H. Y.McSween provided the sample of norite C-14, and J. Agee kindlyprovided analyses of iknenites from the Elliott County graphic in-tergrowths of clinopyroxene-ilmsnif6. A portion of tle publicationcost for tl.is paper were provided by the Exxon Fund of the Dept.of Geological Sciences at U.T. The research presented in this pa-per was supported in part by grants from the Univ. of Tenn. Lib-eral Arts College and the NSF EAR 7906318 (LAT).

References

Akella, J. and Boyd, F. R. (1973) Efect ofpressure on the compo-sition ofcoexisting pyroxenes and garnet in the system CaSiO3-MgSiO3-FeSiO3-CaAlTi2O5. Carnegie Institution of Washing-ton Year Book,72, 523-526.

Albee, A. L. and Ray, L. (1970) Correction factors for electronprobe microanalysis of silicates, oxides, carbonates, phosphates,and sulfates. Analytical Chemistry, 42, 1408-1414.

Anderson, A. T., Braziunas, T. F., Jacoby, J., and Smith, J. V.(1972) Thermal and mechanical history of breccias 14306,14063, 14270, and 14321. Proceedings of the Third Lunar Sci-ence Conference, 8 19-835.

Basu, A. R. and MacGregor, I. D. (1975) Chromite spinels fromultramafic xenoliths. Geochimica et Cosmochimica Acta, 39,937-945.

Bence, A. E. and Albee, A. L. (1968) Empirical correction factorsfor the electron microanalysis of silicates and oxides. Journal ofGeology, 76,382403.

Bishop, F. C. (1980) The distribution of Fe2* and Mg betweencoexisting ilmenite and pyroxene with applications to geo-thermometry. American Journal of Science, 28O, 46-77.

Boyd, F. R. (1971) Enstatite-ilmenite and diopside-ilmenite inter-growths from the Monastery Mine. Carncgie Institution ofWashington Year Book, 70, 134-138.

Boyd, F. R. and Nixon, P. H. (1973) Origin of the ilmenite-silicatenodules in kimberlites from Lesotho and South Africa. In P. H.Nixon, Ed., Lesotho Kimberlites, p.254-268. Lesotho NationalDevelopnent Corporatioq Cape and Transvaal.

Buddington, A. F. and Lindsley, D. H. (1964) Iron-titanium oxideminerals and synthetic equivalents. Journal of Petrology, 5,3 r0-357.

Danckwerth, P. A. and Newton, R. C. (1978) Experimental deter-mination of the spinel peridotite to garnet peridotite reaction intl€ system MgO-Al2O3-SiO2 in the range 900'-11@oC andAl2O3 isopleths of enstatite in the spinel field. Contributions toMineralogy and Petrology, 66, 189-201.

Dawson, J. B. and Reid, A. M. (1970) A pyroxene-iknenite inter-

growth from the Monastery Mine, South Africa. Contributionsto Mineralogy and Petrology, 26,296-301.

Dawson, J. B. and Smith, J. V. (1975) Chromite-silicate inter-growths in upper-mantle peridotites. In L. H. Ahrens, J. B.Dawson, A. R. Duncan, and A. J. Erlank, Eds., Physics andChenistry of the Earth, 9, p. 339-350. Pergamon Press, NewYork.

Deer, W. A. and Abbott, D. (1965) Clinopyroxenes of the gabbrocumulates of the Kap Edvard Hokn Complex, East Greenland.Mineralogical Magazine, 34, I'17 -193.

Dowty, E. and Clark J. R. (1973) Crystal structure refinement andoptical properties of a Ti3+ fassaite from the Allende meteorite.American Mineralogist, 58, 23U242.

Eggler, D. H., McCallum, M. E., and Smith, C. B. (1979) Mega-cryst assemblages in kimberlite from north€rn Colorado andsouthern Wyoming: petrology, geothermometry-barometry, andareal distribution. In F. R. Boyd and H. O. A. Meyer, Eds., TheMantle Sample: Inclusions of Kimberlites und other Volcanics,p.213-226. American Geophysical Union, Washington, D. C.

Frick, C. (1973) Intergrowths of orthopyroxene and ilmenite fromFrank Smith Mine, near Rarkly West, South Africa. Transac-tions of the Geological Society of South Afica, 7 6, 195-200.

Fujii, T. (1976) Solubility of Al2O3 in enstatite coexisting with for-sterite and spinel. Carnegie Institution of Washington YearBook. 75. 566-571.

Garrison, J. R., Jr. and Taylor, L. A. (1980) Megacrysts and xeno-liths in kimberlite, Elliott County, Kentucky: a mantle samplefrom beneath the Permian Appalachian Plateau. Contributionsto Mineralogy and Petrology, 75,2142.

Gurney, J. J., Fesq, H. W., and Kable, E. J. D. (1973) Clinopyrox-ene-iknenite intergrowths from kimberlite: a re-appraisal. In P.H. Nixon, Ed., Lesotho Kimberlites, P. 238-253. Lesotho Na-tional Development Corporation, Cape and Transvaal.

Gurney, J. J., Jakob, W. R. O., and Dawson, J. B. (1979) Mega-crysts from the Monastery kimberlite pipe, South Africa. In F.R. Boyd and H. O. A. Meyer, Eds., The Mantle Sample: In-clusions in Kimberlites and other Volcanics, p.227-243. Ameri-can Geophysical Union, Washington, D. C.

Haggerty, S. E. (1972) Apollo 14: subsolidus reduction and com-positional variations ofspinels. Proceedings ofthe Third LunarScience Conference, 305-332.

Haggerty, S. E., Hardie, R. B. III, and McMahon, B. M. (1979)The mineral chemistry of iknenite nodulc associations from theMonastery diatreme. In F. R. Boyd and H. O. A. Meyer, Eds.,The Mantle Sample: Inclusions in Kimberlites and other Vol-canics, p. 249-256. American Geophysical Union, Washington,D. C.

Haselton, J. C. and Nash, W. P. (1975) Ilmenite-orthopyroxeneintergrowths from the moon and the Skaergaard intrusion.Earth and Planetary Science Letters,26, 287-291.

Herzberg, C. T. (1978) Pyroxene geothermometry and geobarom-etry: experimental and thermodynamic evaluation of some sub-solidus phase relations involving pyrox€nes in the system CaO-MgG-Al2O3-SiO2. Geochimica et Cosmochimica Acta, 42, 945-957.

Ilupin, I. P., Kaminskiy, F. V., and Troneva, N. V. (1973) Pyrox-ene-ilrnenite graphic inclusions from the Mir kimberlite pipe(Yakutia) and their origin. International Geological Reviews,16. 1298-1305.

Kushiro, l. (1972) Determination of liquidus relations in syntheticsilicate systems with electron probe analysis: the system forste-

739

rite-diopside-silica at I atmosphere. American Mineralogist, 57,t26Fr27l.

Lindsley, D. H. (1978) Magnetite-ilmenite equilibria: solutionmodels including MgO and MnO. EOS, 59, 395.

Lindsley, D. H. and Dixon, S. A. (1976) Diopside-enstatite equi-libria at 850" to l400oC, 5 to 35 kb. American Journal of Sci-enc;e,276, 1285-1301.

McCallister, R. H., Meyer, H. O. A., and Brookins, D. G. (1975)"Pyroxene"-ilrnenite xenoliths from ttre Stockdale pipe, Kansas:chemistry, crystallography, and origin. In L. H. Ahrens, J. B.Dawson, A. R. Duncar4 and A. J. Erlank, Eds., Physics andChemistry of the Earth, 9, p. 287-293. Pergamon Press, NewYork.

Meyer, H. O. A. and Boyd, F. R. (1972) Composition and origin ofcrystalline inclusions in natural diamonds. Geochimica et Cos-mochimica Acta. 36. 1255-1273.

Mitchell, R. H. (1977) Geochemistry of magnesian ilmenites fromkimberlites in South Africa and Lesotho. Lithos, 10, 29-37.

Moore, A. C. (1971) The mineralogy of the Gosse Pile UltramaficIntrusion, central Australia. II. Pyroxenes. Journal of the Geo-logical Society of Australia, 18,243-258.

Morse, S. A. (1969) The Kiglapait layered intrusion, Labrador.Geological Society of America Memoir I12.

Morse, S. A (1975) Plagioclase lamellae in hypersthene, Tokkoa-tokhakh Bay, Labrador. Earth and Planctary Science Letters,26,331-336.

Obata, M. (1976) The solubility of Al2O3 in orthopyroxene inspinel and plagioclase peridotites and spinel pyroxenite. Ameri-can Mineralogist, 61, 804-816.

Osborn, E. F. (1962) Reaction series for subalkaline igneous rocksbased on diferent oxygen pr€ssur€ conditions. Americal Miner-alogist, 47, 2ll-226.

Rawlinson, P. J. and Dawson, J. B. (1979) A quench pyroxene-il-menite xenolith from kimberlite: implications for pyroxene-il-menite intergrowths. In F. R. Boyd and H. O. A. Meyer, Eds.,The Mantle Sample: Inclusions in Kimberlites and other Vol-canics, p. 292-299. American Geophysical Union, Washington,D. C.

Ringwood, A. E. and Lovering, J. F. (1970) Significance ofpyrox-ene-ilmenite intergrowths among kimberlite xenoliths. Earthand Planetary Science Letters, 19,371-375.

Robinson, P. (1980) The composition space of terrestrial pyrox-eneF-int€rnal and external limits. In C. T. Prewitt, Ed., Pyrox-enes, Reviews in Mineralogy 7, p. 419493. Mineralogical So-ciety of fimerica, Washington, D. C.

Sclar, C. 8., Carrison, L. C., and Stewart, O. M. (1968) High pres-sure synthesis and stability ofhydroxylated orthoenstatite h thesystem MgO-SiO2-H2O. EOS 49, 356.

Schulze, D. J., Helmstaedt, H., and Cassie, R. M. (1978) Pyrox-ene-ilmenite intergrowths in garnet pyroxenite xenoliths from aNew York kimberlite and Arizona latites. American Mineral-ogist, 63, 258-265.

Smith, C. B., McCallum, M. E., and Eggler, D. H. (1976) Clinopy-roxene-ilmenite intergrowths from the Iron Mountain kimber-lite district, Wyoming. Carnegie Institution of Washington YearBook, 75, 542-5M.

Sobolev, N. V., Jr., Kuznetsova, I. K. and Zytar., N. I. (1968) Thepetrology ofgrospydite xenoliths from the Zagadochnaya kim-berlite pipe in Yakutia. Journal ofPetrology, 9,253-280.

Suwa, K., Yusa, Y., and Kishida, N. (1975) Petrology of peridotitenodules from Ndonyuo Olnchoro, Samburu District, central

GA RRI S O N A N D TA YLO R: P YRO XE N E- O XI DE I NTE RG RO WTH S

7N

Kenya. In L. H. Ahrens, J. B. Dawsor1 A. R. Duncan, and A. J.Erlank, Eds., Physics and Chemistiy of the Earth, 9, p 213-286.Pergamon Press, New York.

Tracy, R. J. and Robinson, P. (1977) Zoned titanian augite in al-kali olivine basalt from Tahiti and the nature of titanium sub-stitution in augite. American Mineralogist, 62,63+645.

Tumock, A. C. and Eugster, H. P. (1962) Fe-AI oxides: phase re-

GA RRI S O N A N D TA YLO R: PY RO XE N E-O XI DE I NTE RGRO WTH S

lations below 1,000oC, Journal ofPetrology, 3, 533-565.Wyatt, B. A. (1977) The melting and crystallizatioir behavior of a

natural clinopyroxene-iknenite intergrowth. Contributi,ons toMineralogy and Petrology, 61, l-9.

Manuscript recei.ved, July 11, 1980;acceptedfor publication, March 13, 1981.