American Mineralogist, Volume 71, pages 900-915, 1986

Hydration of cordierite and hypersthene and a description of the retrogradeorthoamphibole isograd in the Limpopo belf South Africa

Drnx D. vlx RrntvnNDepartment of Geology, Rand Afrikaans University, Johannesburg 2000, South Africa

Ansrnl'cr

The transition from orthopyroxene- to orthoamphibole-bearing assemblages in Mg-richmetapelites across a retrograde M3 orthoamphibole isograd in the Archean Limpopo beltin South Africa is expressed by two hydration reactions that are superposed on M2 coronatextures ofcordierite and hypersthene after garnet hypersthene + quartz + HrO : antho-phyllite, andcordierite + HrO : gedrite + kyanite + quartz-. The temperature ofhydrationat the isograd of less than 625450'C at a minimum pressure of 6 kbar is estimated on thebasis of (l) a comparison of published information on the solvus in the orthoamphibolesystem to the results of the present study, (2) the presence of kyanite coexisting withanthophyllite and gedrite both at and south ofthe isograd, and (3) the application ofthegarnet-biotite thermometer. The estimated temperature is far below the upper thermalstability limit of anthophyllite under conditions where Prro: P-^.

Cordierite at the isograd is partly (or completely) surrounded by gedrite and kyanite,and hypersthene by anthophyllite; coexisting hypersthene + anthophyllite * quartz is re-stricted to a narrow zone defined by the isograd. A modified Redlich-Kwong equation ofstate for nonideal mixing of HrO and CO, was used to calculate the mole fraction of CO,(about 0.8) in the fluid phase necessary to displace the equilibrium curve anthophyllite :

7 enstatite + quarlz + water to the estimated P-Tconditions on the isograd. The presenceof a systematic temperature gradient in the Limpopo belt suggests that the influx of theCOr-rich fluid occurred at a fixed temperature during hydration.

IxrnooucrroN

The Archean high-grade Limpopo belt of southern Af-rica is a polymetamorphic, highly deformed terrane sit-uated between the Rhodesian and Kaapvaal cratons (Fig.l). It has been subdivided into northern and southernmarginal zones, which are separated from a central zoneby major east-northeast-trending parallel ductile shearzones, the Palala shear zone in the south and the Tuli-Sabi cataclastic zone in the north (Mason, 1973; Robert-son and Du Toit, l98l).

The southern marginal zone consists almost entirely oftwo major categories of rocks that represent metamor-phosed and structurally transformed granite- and green-stone-type material of the adjoining Kaapvaal craton (Ma-son, 1973; Robertson and Du Toit, l98l; Du Toit andvan Reenen, 1977; Du Toit et al., 1983; van Reenen,1983a), including gray migmatized tonalitic and trondh-jemitic gneisses (the Baviaanskloof Gneiss) and pelitic,mafic, and ultramafic supracrustals of the BandelierkopFormation that occur as a series of easily recognizablediscontinuous keels surrounded by the BaviaanskloofGneiss (Fig. 2). The Matok charnockitic-granitic plutonand the Schiel alkaline pluton that straddle a retrogradeorthoamphibole isograd (Fig. 2) intruded the metamor-phic rocks at 2620 Ma (Barton et al., 1983). The high-grade rocks were affected by three early folding events,

0003404x/86/0708-o900$02.00 900

o 250 500

kilomeiers

Fig. l. Sketch map showing position of southern marginalzone of Limpopo belt in South Africa.

€o\

ooo

i 6 6

: - 6- €o\

o

9t t<

, n xH

c0

Y N

B q r

F

9 a ob t ra ' t

9t

o Q

5 H6 ( H

o

HR9l cd

90

o E

g o. 9 !

ool)

R . E

c.l

bo

r I

EEEEIHTE i i

N''.aTs \

A - \ f ,o eo f abp "{o " cd s " ' - " 6

902

300 400 500 600 ^

700 800 900

Temperature ("C)

Fig. 3. P-T diagram illustrating uplift and erosion path forsouthern marginal zone of Limpopo belt. Boxes represent esti-mated P-f conditions for Ml, M2, and M3. Also shown are (l)the calculated (endmember) dehydration curve for the reactionanthophyllite: 7 enstatite + quartz + water under conditionswhere P"ro : P,"o, and (2) amount of lowering of/"ro needed todisplace this curve to the M3 P- T estimates. See text for discus-s10n.

and a later deformational event is recognized by post-metamorphic shear zones that cut across earlier structures;along the shear zones the high-gade rocks have been ret-rograded to greenschist-facies assemblages (Du Toit et al.,l 983) .

The goals of this contribution are to (l) describe anisograd established by retrogression of orthopyroxene-bearing, MgO-rich, metapelitic gneisses of the Bandelier-kop Formation in the southern marginal zone intoorthoamphibole-bearing gneisses and (2) estimate P-Tconditions for the retrograde reactions and thereby cal-culate the fluid composition associated with retrogression.

MrrrvronpHrc EvoLUTroN oF THE sourHERNMARGINAL ZONE

The retrograde orthoamphibole isograd is a conspicu-ous regional metamorphic feature of this area. This iso-grad cuts across the structural trend of all lithologic unitsand can be followed for a distance ofat least I 50 km fromwest to east (Fig. 2). It separates an orthoamphibole zonein the south from an orthopyroxene zone in the north andcan be limited to a distance of not more than a few hundredmeters at right angles to its regional trend. The systematicregional distribution of the metamorphic zones and theposition of the isograd in Figure 2 are controlled by aretrograde event (M3) that followed on two earlier gran-ulite events (Ml and M2) before 2500 Ma (van Reenen,1981, 1983a; van Reenen and Du Toit, 1977, 1978; DuToit et al., 1983).

Peak metamorphic conditions of f > 820'C and P >9.5 kbar were attained during the early prograde stage(Ml, Fig. 3) of this extended period of high-grade meta-

ver.r REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT

morphism; this was followed by rapid uplift M2, Fig. 3),which is recorded by corona textures of cordierite andhypersthene after garnet (Fig. 4a). P- Z conditions of thisM2 event are -800'C and 8.6-7.2 kbar (van Reenen,1983a). The retrograde reactions (M3, Fig. 3) discussedin the present paper were superimposed on the M2 as-semblages.

Dnscnrr"rroN oF THE RETRocRADE ISoGRAD

The retrograde isograd, which is only definedwithin thepelitic gneisses, is discussed in terms of changes, acrossthe isograd, in (l) mineral assemblages observed in meta-pelitic gneisses, (2) the chemistry of minerals such as gar-net and biotite; and (3) hydration reactions that are tex-turally preserved in metapelitic gneisses.

Changes in mineral assemblages across the isograd

The orthopyroxene zone, situated to the north of theisograd, is characterized by the ubiquitous presence oforthopyroxene in all suitable rock types, including bandediron formation, ultramafic, mafic, and peltitic gmnulites,as well as some members of the Baviaanskloofgneiss. Themineralogy of representative samples of the metapelticgneisses in the three metamorphic zones is shown in Tablel Hydrous phases, such as biotite in the metapelites andhornblende in the mafic and ultramafic granulites, arepresent in small amounts. Rocks containing exclusivelyhydrous minerals are entirely restricted to narrow shearzones that crosscut all previous structures.

The pelitic gneisses in the orthopyroxene zone are char-acterized by three main petrogtaphic types based on thepresence or absence or cordierite and/or garnet (Table l).These assemblages have been the subject of an investi-gation (van Reenen, I 983a) ofthe phase relations involv-ing coexisting cordierite + garnet * hypersthene + bio-tite in order to determine the distribution of pressure andtemperature within the orthopyroxene zone during thefirst (M2) retrogtade granulite event. It was shown (vanReenen, 1983a) that the phase relations involving theseminerals were controlled by a continuous Fe-Mg reaction(garnet + quartz: cordierite * hypersthene; Figs. 4a,I la), the nature of which in turn was controlled by subtledifferences in bulk chemistry (van Reenen, 1983a, p. 152)during a period of nearly isothermal uplift from granulitefacies conditions.

Compositionally similar rocks in the orthoamphibolezone are characterized by the presence ofcoexisting an-thophyllite + gedrite + kyanite in rocks ofsuitable bulkcomposition (Table l) and by the presence of cordieriteonly as a rare reactant. The position of the retrogradeisograd is defined by coexisting anthophyllite and hyper-sthene in more Fe-rich rocks, or by the presence of hy-persthene, anthophyllite, cordierite, and gedrite in moreMg-rich rocks (Table l).

Modal compositions of representative samples of thepelitic gneisses from the orthoamphibole zone are listedin Table 2. The orthoamphibole-bearingrocks in this table(and in Table l) are subdivided into a number of different

xH2O=O 2

vAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT 903

Fig. 4a. The continuous Fe-Mg reaction gamet (Ga) +quartz : cordierite (C) + hypersthene (I{) in the cordierite-garnetgranulites (Group 2P, Table l) in the orthopyroxene zone formedby the M2 event (sample DV56).

mineralogical types based on subtle differences in bothmineralogy and bulk chemistry. None of these rocks con-tain K-feldspar and all contain quartz. Kyanite is usually,but not always, present in small amounts (Table 2) in theorthoamphibole-bearing samples of Groups lA and 24,as short stubby crystals intimately associated with or-thoamphibole (Fig. 4b). It is generally an inconspicuousmineral except in one sample (M215) where it occurs aswell-defined prismatic grains scattered throughout the rock(Fig. ac). Coexisting orthoamphiboles are discussed be-low. The retrograde M3 activity, based on the persistenceofreactant cordierite, extended at least 5 km to the southof the retrog.rade isograd in Filure 2.

Changes in mineral chemistry across the isogradThe change in the metamorphic conditions across the

isograd is not only reflected in the mineralogy of the peliticgneisses but also by systematic changes in the chemistryof garnet and biotite (Fig. 5) that correspond with theposition of the isograd based on textural evidence. Theobserved increase in the pyrope content ofgarnet and theTi content of biotite in going from the orthoamphibolezone through the isograd to the orthopyroxene zone wasobserved in chemically similar rocks, and this change isso systematic that the position ofthe isograd can be mappedon the basis of the composition of these two minerals.

Hydration of cordierite and hypersthenealong the isograd

The position of the retrograde isograd in Figure 2 isalso defined by hydration reactions involving both cor-dierite and hypersthene. Most hypersthene grains in boththe cordierite-bearing (DV3, DR157,M346 of Group 2Iin Table l) and cordierite-free rocks (DV38 of Group 3Iin Table l) are partly (Fig. 6a) or completely replaced byanthophyllite, and this hydration reaction is restricted tothe retrograde isograd. Coexisting hypersthene, antho-phyllite, and quartz are consequently also restricted to anarrow zone defined by this isograd.

Fig. 4b. Kyanite (K) intergrown with biotite and orthoam-phibole (O) (sample DV43).

Cordierite along the isograd is always intergrown witha relatively coarse grained aggregate ofgedrite (and some-times also of anthophyllite) and kyanite (Fig. 6b). Theaggregate generally appears to have nucleated at cordier-ite-cordierite grain boundaries and to have glown into thecordierite. This is illustrated in Figure 6c, which showsthe incipient hydration ofcordierite in the orthopyroxenezone. The process is identical, but the products are muchfiner grained than those observed along the isograd. Ver-non (1972) described similar hydration processes fromthe Arunta Complex in central Australia. Two generalizedreactions are relevant: hypersthene + quartz + H2O:anthophyllite and cordierite * HrO : gedrite * kyan-ite-t qtnrtz.

Pntnocrrr,prrstnv

Bulk chemical analyses were done by the General Su-perintendence Company, Johannesburg, using standardgravimetric, volumetric, and X-ray fluorescence methods.The chemical compositions of 8 representative samplesof the pelitic gneisses in Table I are given in Table 3.Microprobe analyses were obtained with a nine-spectrom-eter automated enl model seve electron microprobe atthe Anglo American Research Laboratories in Johannes-burg.

Fig. 4c. Coarse-grained kyanite coexisting with orthoamphi-bole (O) (sample M2l5).

6

q o

! o i l

a i l

N f o

O Q

B C

q i l

o d o

0 i

o !

! . !

! i

t l

N

!(,o

I

o

o

t l

A

,o

a z a o

q i h

N > T >

o i l E

E

O F i E

E E I

S E E

a - q r l

t a p ,

r

VAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT

o ^ qi€ " . (

> E E !> l l q o € o l

O p O O + d i O q ^o h + i ! d o

! E N I S O O O 6 o O E Eo o Z a f O S P d ) e e, 4 + a O > i o f d o +! i o u ! > i - L q o o l

o I I E ! \ O 6 0 u 9 0a i r . r g . o o . i r L o YX J r L ^ ! ! U O c O d i q Zo q g t ! 6 o r E ! 6 E p O Yo r o ( t o ! q E ; o i F a EL N o d J i q > o L $ o o o

E + i q o E ! 6 o o o ut o s o o ! ! o q > . E o E E

!

o!

!

o

!

+

q

+

a

N

oI

Eso

t

o

o

n on !

on N n 9@ ^ @ ^ N ^ N ^

F O N O N n O - O N

oq o q o q J d o o o ga P a r a t t T r ,

s o o 0 6 o 0 o oN ! @ ! 9 k n ! h N ! @ L. o . a . a ' o ' . o t o

o v o - o - o v o o - o v

69 ^ hN N

O F O N no

o 0 6 0N L @ L N L

" i o v o v

o+ l L

a d

+ l ++ !

o nN N

o t r+ - 9 < g g d

A

N +N> o

o v c o $c < ! @

o v o F+ ! i d

+ + G

G 6 + Oi g ^ !

e + N i o+ $ A

s +

O n n . -@ g

< !< - A o

6 6o a

g I o + l | + l I

+ + a G A < V@

g g + r t r+

+ + d A q g o^ 6 ^

< @ < + | | g @

+ ^ V I I O + >

= - = 6

+ + q +

o o + l

o + } 4

+ + l +

+ l Q o a

G ^ O

+ r +

^ N

@ < 6N V

< !v + o v

Q A + A

r N O

+

oN

o

go

!o

+ + o + i i d d > + o! <

O O O O J E O

o

o

+

o

o{

!

!

oN

g!

2

904

N

o

ts

q

o

.,

q)

o

U

a

I

ooo

tso

o

F

VAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT 905

Gross+

Spess

t

lrc*-."eil

Mineral chemistry

Microprobe analyses of cordierite, hypersthene, gamet,biotite, anthophyllite, and gedrite from seven represen-tative samples ofthe different mineralogical groups (Tablel) are listed in Tables 4 to 8. The results in these tablesare mean values for at least two spot analyses of the samegrain of the same mineral in the same thin section. Asystematic check for evidence of chemical zonation wasmade only in the case of garnet, where care was taken toobtain point analyses from both the core and the rimcomposition of each analyzed. grain.

Garnets (Table 4) show evidence of chemical zonation,but no detailed zoning profiles were obtained. The generalzoning patterns appear rather irregular, and this is quitedifferent from the situation in the orthopyroxene zonewhere regular zoning patterns (pyrope-rich cores and al-mandine-rich rims) could be ascribed to the resorption ofgarnet during the reaction garnet + quartz : cordierite +hypersthene (van Reenen, 1983a). The combined gros-sular and spessartine content ranges fiom 4.8 to 12.8 molgowith a mean value of 7.6 molo/0.

The mean octahedral site occupancyforall the analyzedbiotites (Table 5) based on 22 oxygens per formula unitis 5.88, and the mean value for octahedral Al is 0.328 forbiotite from the isograd and 0.563 for biotite from theorthoamphibole zone. The Ti-content increases from meanvalues of 0. 199 atoms for biotites in the orthoamphibolezone through 0.351 atoms at the isograd to 0.492 atomsin the orthopyroxene zone (see also Fig. 5). The compo-sition of biotities in the same thin section shows a con-siderable variation that is dependent on the position ofthis mineral with respect to garnet. The most Mg-rich(and Ti-poor) biotites were analyzed from inclusions ingarnet, while the most Fe-rich (and Tirich) biotites occurin the matrix. The bulk chemical control on the compo-

I Bi"tir"l

sition of biotite is illustrated by values of MgO/(MgO +FeO) that increase from a nean of 0.72 in garnet-bearingsamples (Group 24' in Table l) to 0.78 in the garnet-freesamples (Group lA in Table l). These compositional vari-ations are independent of the grade of metamorphism andare similar to those described for biotites from comparablerocks in the orthopyroxene zone (van Reenen, 1983a).

Cordierite is not stable in the orthoamphibole-bearingrocks in the orthoamphibole zone; the analyses in Table6 are, with one exception (DVl), restricted to samplesfrom the isograd. Cordierite in sample DVI is present asa reactant. A MgO(MgO + FeO) value of 0.89 in cor-dierite from sample DVI (garnet-free Mg-rich rocks ofGroup lA in Table l) is much higher than the mean value

Table 2. Modal composition (volume percent) oforthoamphibole gneisses

S a n p l e

60 - Tiq 60 70% Mgo

MsoMole % Atm

Fig. 5. Variation in composition ofgarnet and biotite in pelitic gneisses across iSograd; O: orthopyroxene zone, x : isograd,A : orthoamphibole zone.

r o o o

E OE N

! E

o d

o o @ o

D V . 2 3 1 8 1 9

0 v . 4 3 2 1 2

DV. 53 5 23

0\ / . 2 12 35

D U . 2 6 1 2 2 5

M. 447 9 23

D U . 4 1 I 2 0

rJV. 49 30

DR. 184 32 20

D V . 2 2 4 5

DV. 1 48

r \ r . 215 37

1 7 5

3 0 6

1 1 1

2 5 l i

I

1 3

t 9

3 9

1 5 1 0

1 7 1 3

2 4 2 0

3 1 2 5

31 24

5 2 3

5 5

2 6 2 9

27 26

31 38

4B

1 8 3 2

1 2 2 5

1 4 1 9

x - z 1 %

906 vAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT

Fig. 6a. Hydration of hypersthene (H) along isograd (sampleM346). A : anthophyllite.

of 0.85 for cordierite in the more Fe-rich garnet-bearingrocks (Group 2I, Table l) from the isogad. This chemicaltrend is similar to that displayed by biotite and is also inagreement with similar trends described for cordierite fromthe orthopyroxene zone (van Reenen, 1983a).

All the available analyses for hypersthene are from rocksalong the isograd. The AlrO, content of hypersthene inthese samples (mean, 3.06 wto/0, Table 6) is slightly lowerthan the mean of 5.89 wto/o that was reported for thismineral in similar rocks from the orthopyroxene zone (vanReenen, 1983a). Analyses for adjacent hypersthene andanthophyllite in the same sample on the isograd are iden-tified as such in Tables 6 and 7. The variability in thehypersthene (and orthoamphibole) composition for a spe-cific sample is shown by more than one analysis.

Anthophyllite and gedrite coexist in a variety oftexturesin the pelitic gneisses in the orthoamphibole zone, and acommon occurrence is as large discrete grains with sharpoptical contacts (Fig. 7). In this mode of occurrence thetwo minerals could be distinguished optically-antho-phyllite being colorless and gedrite showing faint pleoch-roism.

Representative analyses for anthophyllite and gedrite,or Al-rich anthophyllite, from rocks along the isograd and

Fig. 6c. Incipient hydration of cordierite in orthopyroxenezone (sample Rl3). K = kyanite.

from the orthoamphibole zone are given in Tables 7 and8. The variability in the orthoamphibole compositionfor a specific sample is shown by more than one analy-sis. The accuracy of these analyses has been evaluatedusing the "residual value" method of Robinson et al.(1971). This method is based on an ideal formulaNa,Rl+(R!1"Ri+XAl*+rSi8-,-,)Orr(OH)r, where x : Aoccupancy and y : octahedral Al + Fe3' * Cr3+ + 2Ti4+.Substitutions of Na (and K) in the A site (;r) and of R3'in the octahedral sites (y) are compensated by substitutionofAl for Si in the tetrahedral sites, so that the sum ofxand y must be equal to the amount of tetrahedral Al asshown in the ideal formula (Robinson et al., 1971, p.l0l0). The "residual value" is (x + /) - Alw, and thisvalue has a mean of -0.062 (with a range from -0. 140to + 0.044) for anthophyllites, and a mean of +0.078(with a range from -0.07 to +0.234) for gedrites in thisstudy. These small values compare favorably with otherorthoamphibole analyses in the literature (e.9., Grant, 198 I ,p. ll29) and indicate either analytical error or small de-viations from the ideal formula. The classification of theorthoamphiboles from this study into anthophyllite, Al-rich anthophyllite and gedrite is shown in Figure 8.

The gedrites from the isograd differ markedly in com-

Fig. 6b. Hydration of cordierite (C) along isograd (sampleM346). G: gedrite.

Fig. 7. Coexisting high-At gedrite (G) and low-Al antho-phyllite (A) in orthoamphibole zone.

.o. t ' ' ' , l .

VAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT

Table 3. Chemical analyses of orthoamphibole gneisses from isograd (DV3 to DV38)and orthoamphibole zone (DV43 to DRl84)

907

S a m p l e D V . ] D R . 1 5 7 l v . 3 8 D V . 4 3 M . 4 4 7 D V . 2 2 D V . 1 D R . 1 8 4

5 i o z

A l z o l

F " z 0 3

l i e O

M g o

M n O

C a O

l ! a . o

^ r o

f i ( ) 2

P ^ O -

i t 2 o *

C O ,

l l , o

l o t a l

6 5 . 2 0

1 ) . 7 0

o . B B

J O A

6 . 3 o

o . 0 B

1 - 3 9

1 . 8 0

1 . 8 2

o . 6 B

o . 0 5

o . 8 0

o . 3 0

o . 1 2

1 8 . 5 o

o . 7 3

B . B B

B . B 6

o . 1 0

o . 6 B

o . B 1

o . 8 2

o . o 7

1 - 1 t l

o . g o

o . 2 0

5 7 . 7 0

1 4 . 7 O

3 . o

1 0 . B o

7 . 6 0

o . 1 2

1 0 0

2 . 5 0

1 . 9 8

o . 6 6

o . 1 . 1

o . g o

o . 4 6

6 5 . 4 o

1,4 - 30

o - 7 4

6 . 6 6

) . ) w

o . 1 0

1 . 9 5

z . 6 0

1 . 8 6

o . 5 8

o . o g

o . g o

o . 4 9

5 7 . 2 6

1 6 . 9 6

2 . j o

6 . j 6

7 . 1 3

o . 1 1

t 1 a

2 - 3 4

1 . 1 8

o . 6 7

o . o 5

1 . 6 9

o . ' t 9

o . 2 8

5 7 . 4 0

r 4 . 7 o

7 . 1 7

1 0 . 5 3

9 - 5 0

o. r.6

1 . 7 8

2 . 2 0

o . 2 9

o . 9 2

o . 1 9

o . B o

o . 3 0

o . 7 3

6 s . ) 6

1 3 . 8 4

7 . 0 7

4 . 9 0

o . 0 5

1 . 8 7

2 . 4 0

o . 2 3

o - 5 8

o . o B

L 0 5

o . o g

o . 1 5

4 9 . 9 7

1 3 . 9 8

I t (

1 7 . 7 6

7 0 . ) 9

o . 2 0

1 . 6 8

o . 6 6

o . 5 2

o . 6 6

o . 1 1

1 - 4 6

o . 6 9

o . 2 6

1 O l . 0 4 9 9 . 2 6 1 o 2 . 5 2 1 o 1 . ' t 7 9 9 . o 5 l o o . 6 7 9 9 . 2 3 9 9 . 4 9' l h e

m i n e r a l o g i c a l s u b d i v i s j - o n o f t h e s e s a m p l e s a r e g i v e n i n T a b l e 1

position from gedrite in chemically similar samples fromthe orthoamphibole zone. They are much more aluminous(Al,Or: 18.42-20.68 wto/o) than gedrite from the or-thoamphibole zone (AlrO, : 13.87-16.00 wto/o) and arealso characterized by higher NarO contents (l.Bl-2.27wto/o against 1.00-1.43 wt9o). The gedrites from the isograd(and especially those from sample DV3) have the highestAlrO. content (20.68 wtTo) of any orthoamphiboles re-ported in the literature (e.g., Rabbitt, 1948: Robinson andJaffe, 1969; Robinson et al., l97l1' Lal and Moorhouse,1969; Gable and Sims, 1969; Beeson, 1978; Spear, 1980).The high NarO content (2.27 wf/A ofthis gedrite is similarto the values of 2.27-2.38 wto/o reported for gedrites insample 73-29D of Spear (1980) and 2.30 wtolo for gedriteof analysis G-23 in Grant (1981).

The gedrites (and anthophyllites) from the orthoam-

8 o + S i i n s t a n d a r d c e l l +

Fig. 8. Classification oforthoamphiboles in the present studyafter Leake (1978).

phibole zone are also enriched in Fe as compared to thosefrom chemically similar rocks along the isog,rad. It istherefore reasonable to suggest that the observed differ-ences in the composition of gedrites are due only to dif-ferences in metamorphic conditions. The compositions ofanthophyllites from the isograd do not show significantdifferences with respect to their NarO and AlrO, contentswhen compared to those from the orthoamphibole zone,although there is a slight increase in AlrO3 and NarO inanthophyllites within this zone.

The presence of a miscibility gap in the anthophyllite-gedrite series is well documented in the literature (Rob-inson et al., l97l; Spear, 1980) and coexisting subsolvusorthoamphiboles have been described by Stout (1970,1971,1972), Robinson et al. (1982), Spear (1978, 1980,1982), and van Reenen (1983b). Figure 9 is a plot of A-site

anthophyllite Alrv tschtrrokite

Fig. 9. Plot of A-site occupancy vs. Alrv for all available or-thoamphibole analyses.

l d t

i n"o'

908 vAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT

Table 4. Representative electron-microprobe analyses of garnet

D V . 3 8 D V . j B D R . 1 5 7

C o r e R i m

D V - I D V - 4 3 M . 4 4 7 M . 4 4 7

C o r e R i m

D R . 1 8 4 D R . 1 i l 4

C o r e R i m

s i 0 2

Atzo. ,

M g o

F e O

C a o

MnO

N a r o

C r 2 0 3

T o t a I

3 9 . z 5 3 9 . 3 3 4 0 - 5 2

2 2 - 3 5 2 2 . 3 9 2 1 . 9 3

9 . o 5 9 . 1 7 7 . 9 )

2 2 . 6 8 2 8 . 6 5 2 7 . 4 5

2 . o 9 7 . 1 5 7 . ' t 4

o . 3 3 o . 2 9 o . 6 0

o . o 1

o . 1 4 o . 1 6 o . 0 5

4 o . 0 6 3 9 - ) 9 3 8 . ) 1

z z . 2 8 2 2 . 1 9 2 1 - 6 8

7 . ) 1 5 - 5 6 6 . 9 5

2 9 - 7 7 3 0 . 7 9 3 0 . 2 1

7 . 2 7 2 . o o 7 . 6 4

o . j 6 1 . 2 7 1 . 7 6

- o . o 1

o . t 4 o . 2 7 o . 7 )

3 8 . 3 0 3 8 . 3 4 3 8 . 1 5

2 1 . 7 8 2 2 . 4 0 2 2 . 0 6

6 . 5 4 7 . o o J - 9 t

2 9 . 3 ) 3 a - . 2 8 3 2 - 5 32 . 8 1 2 . 7 6 2 . o ' l

1 . 0 3 o . 3 1 o . ) 2

o . o 1

o . 1 6 0 . 1 r 2 . O 1

1 o o . 9 o 1 o 1 . 0 9 9 9 . 6 6 r o 7 . 3 9 1 o 1 . 4 8 1 o o . O B 9 9 . 9 5 1 o 1 . 2 0 1 o 1 . 2 7

C a t i o n s p e r 2 4 o x y g e n s

J A

n r a V

5 . 9 9o . o 1

5 . 9 9o . o 1

< o o

o . o 16 . o 6 . 2 2o . o 1

6 . t t 6 . o B 5 . 9 2 5 . 9 4o - o B o . 0 6

6 . O 6 . o 1 6 . 2 2 6 . 1 1 6 . o B 6 . o 6 . o 6 . o b - u

n r V l

Mg

F e

C a

M n

N a

C r

4 . o 1

2 . 0 6

o . 3 4

o . 0 4

o . 0 2

3 . 9 9

1 . 6 2

3 . 9 5o . 2 8

o - 1 5

o . 0 2

4 . o z 3 . 9 r '2 . 0 7 1 - 8 1

3 . 6 5 3 . 5 2o . 1 9 o . 1 9

o . 0 4 0 . o B_ o . 0 1

o . 0 2 0 . o 1

4 . o 1 4 - 0 4

1 . 6 6 1 . 2 8

3 . 8 3 . 9 7o . 2 1 o . 3 3

o . 0 7 o . 1 7- o . o 1

o - 0 2 o . 0 3

4 . o r 4 . o 3 . 9 91 . 5 2 1 . 6 1 1 - 3 7

3 . 8 4 3 . 9 1 4 - 3 3

o . 4 7 0 . 4 6 o . ) 4o . 1 4 o . 0 1 + o . 0 4

o . o o

o . o 2 0 . 0 1 0 . 0 4

5 . 9 9 5 . 9 7 5 . 6 2 5 . 7 6 5 . 7 8 j . 9 9 6 . o 3 6 . o 26 . o z

M g / M g + F e + 2 a t o . ) 6 o . 3 4 o . 3 0 o . 2 4 o . 2 8 o . 2 9 o . 2 ,o . 2 9

occupancy vs. Alw for all the available anthophyllite (in-cluding Al-rich anthophyllite) and gedrite analyses fromthis study. The linear trend observed in this diagram isidentical to that reported by Robinson and Jaffe (1969),Robinson et al. (1971, 1982), James et al. (1978), Spear(1980), and Grant (1981). The trend from anthophylliteto gedrite is very pronounced and is toward the compo-sition of "ideal" gedrite as defined by Robinson et al.(1971). The significance of Al-rich anthophyllite in Fig-ures 8 and 9 is unclear. All three amphiboles were iden-tified as discrete (although not touching) grains in sampleDVI (see Tables 7, 8) and subsequently also in sampleDRl57 from the isograd. Al-rich anthophyllite coexistswith gedrite in sample DRl57, but was not found in con-tact with anthophyllite.

The miscibility gap between anthophyllite and gedriteis also clearly illustrated in Figure 10. It is manifested bya large discontinuity in the A-site occupancy and in the

AlN and Alu contents (Spear, 1980, p. I 109). The depen-dence of the size and shape of the miscibility gap in termsof X." with respect to A-site occupancy and AlIv is obviousfrom Figure 10. The widening of this gap in more Mg-rich compositions, however, is not in agreement with

Spear's data (Robinson et al., 1982, p. 66 and Fig' 29),which showed that the gap should be narrower for Mg-rich compositions and also that the solvus would close ofin Mg-rich compositions at a lower temperature than inFe-rich compositions. The Mg-rich compositions from

this study are from the isograd (Group 2I, Table l), andthe Fe-rich compositionsare from chemically similarlow-er-grade rocks (Group 2A, Table l) from the orthoam-phibole zone.

The distribution of Fe and Mg between coexisting an-thophyllite and gedrite from this study (Fig. l0) is verysimilar to that described by Spear (1980) for coexistingorthoamphiboles from the Post Pond Volcanics in Ver-

VAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT

Table 5. Representative electron-microprobe analyses of biotite

909

D V . 3 g - 1 ' n v . 3 B D R . 1 5 7 D V . l D V . 1 D V - 4 3 1 1 . 4 4 7 D V . 2 2

s i 0 2

A r r o ,T i 0 2

Mgo

I r e O

C a O

IInO

N t r o

nroC r - O ^

4 )

T o t a l

3 6 . 9 3r 7 . o 4

3 . 0 7

7 7 . 9 1

1 1 - ' , t 1

u . 4 )

o a t

o . 5 0

3 r - . 5 4 ) l . 9 7

1 7 . o B 1 7 . 0 6

) . 5 4 ) - 2 o

| ( . t / r / . ) l

1 1 - 4 7 1 1 . 6 7

o . 2 9 o . 3 7

9 . 2 9 9 . 3 8o . J 2 o . 4 3

3 7 . 5 ) 3 u . 9 9 3 7 . 9 1

1 7 - 4 9 1 7 - 5 9 ' , t B . 4 2

3 - 6 8 1 . 7 4 1 - 8 2

1 7 . 0 6 1 8 . 1 7 1 6 . 4 2

1 1 - 6 4 B . g S 7 2 . 7 o

- o . 0 1

o . 0 2

o . 2 8 o . 5 0 o . 5 3

9 . 8 9 B . ) 5 8 . 5 6

o . 7 6 o . 5 2 o . 4 6

3 9 . 6 4 3 7 . 9 51 7 - 8 7 1 7 . 6 8

1 - 7 3 1 . 4 7

1 6 . 7 6 1 7 . 4 7

1 ' t - 6 7 1 1 . 3 7

o . o 2 0 . 0 1

o . 0 3 o . 0 3

o . 5 0 o . 4 4

8 . 9 6 8 . 3 6

o . 2 9 o . 4 o

9 6 . 3 2 9 6 . 9 1 9 7 . 4 5 9 8 . 3 2 9 4 - 7 9 9 6 . 8 4 9 5 . 4 0 9 5 . 2 0

C a t i o n s p e r 2 2 o x y g e n s

S i

A f a v

5 - 3 ' o

2 . 6 4 , q o

) . 4 )

9 C , 6

5 . 3 6 5 . 6 2 5 . 4 62 . 6 4 2 . 3 8 2 . 5 4

5 . 4 9 5 . 5 22 . 5 7 2 . 4 8

B . oB . o8 . oB , oB . oB . oB . oB . o

A l v f

T i

M g

F e

Cr

o . 2 B

o . 3 4

3 . 8 7

1 . ) 5

o . o 6

o . 3 ' l

o . 3 B

3 . 6 9

1 . 3 8o . o 6

o . 3 3

o . 3 5

3 . 7 11 . 4 0o . o 5

o . 5 9

o - 2 0

t . ) )

o . o 5

o . 3 0 o . 6 1

o . 4 0 0 . 1 9

3 - 6 3 3 . 8 9

1 . ) 9 1 . o B

o . o g o . 0 6

o - 5 7 o . 5 6

o . 1 9 o . i 6

) . 6 5 3 . 7 91 - 4 2 ' J . 3 8

o . o 3 o . 0 5

5 . 9 0 ) . o 1 5 . 8 4 5 . B j 5 . 8 3 5 . 8 9 5 . 8 6 5 . 9 4

N a

K

o . 1 3

1 . 7 )

o . o B1 . 7 7

o . 1 0

7 . 7 2

o . 7 5

7 . 5 7

o . o B o . 1 4

1 . B o 7 . 5 4

o - 1 4 0 . 1 2

7 . 6 7 1 . 5 5

1 . 8 6 1 . . 8 2 1 . B B 1 . 6 8 7 . 7 2 1 . 8 1 7 . 6 7

M g , / M g + F " + 2 o . 7 t ' l o . 7 3 o . 7 3 o . ? 2 o . ? B o . 7 z o . 7 3o . 7

I n c o n t a c t w i t h e a r n e t . T h e r e s t a r e f r o m t h e m a t r i x

mont: gedrite is enriched in Fe relative to coexisting an-thophyllite. Total Fe varies very little between antho-phyllite and coexisting gedrite, whereas anthophyllite isalways enriched in Mg relative to gedrite. A similar re-lationship in the orthoamphiboles has been described byKamineni (1975), Spear (1980, Fig. 9), and Grant (1981).

The observed regularity of compositions of mineralssuch as garnet, biotite, and orthoamphiboles across theretrograde isograd implies that mineral assemblages arein a close approach to chemical equilibrium along the

isograd. This is substantiated by the repeated presencealong the isograd ofboth the reactants and the products(hypersthene + anthophyllite + quartz, and cordierite +gedrite + kyanite + quartz) ofthe hydration reactions inthe same samples.

Prusn RELATToNS

The analytical data from Table 3 and Tables 4 to 8 areillustrated in A'FM diagrams (Fig. I l), where A' :

AlrOr-KrO-CaO-NarO, F: total iron as FeO, and M:

910 vAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT

Table 6. Representative electron-microprobe analyses of cordierite and hypersthene

C o r d i e r i t e H y p e r s t h e n e

D R . 1 5 7 D V . 3 D V . 1 D R . 1 5 7 o a . t 5 7 ' D v . 3 * D V . 3 D v . l B + D v . 3 8

s i 0 2

A l ^ o ^

M g o

F e O

MnO

N a r o

4 8 . 9 5

t r . ) l

3 . o oo . o 3o . 1 1

s i 0 2

A l r o l

T i 02

M g o

F e O

C a O

MnO

N a , o

C r , O .

T o t a L

< c A 1

3 . 0 4o . 0 4

2 \ . 4 9

o . o B

o . 1 9

o . o 1

o . 1 1

5 1 . 6 9

3 . 4 5o . 0 2

2 2 . 2 2

o . 0 5

o . 2 2

o . o 1

o . 1 7

\ 8 . 2 6 \ 9 . t B

3 2 . 8 7 3 3 . 2 21 O . 7 3 7 7 . 9 2

3 . 8 3 2 . 6 0

o . 0 2 0 . 0 1

O . 1 0 O . 7 3

5 1 . 7 2 5 1 . 8 4 5 2 . 6 1 5 i . 6 32 . 8 7 2 . 5 3 3 . 2 9 3 . 1 8o . o 2 0 . o 2 - o . 0 2

2 2 . 7 5 2 7 . 3 0 2 3 . 9 7 2 3 . 1 5

2 3 . 5 1 2 5 . 3 7 2 7 . 5 3 2 7 . 4 0

o . o 8 o . 1 0 o . o g o . o 8

o . 2 1 o . 1 9 o . 0 8 o . o g

o . 2 8 0 . 2 0 0 . 1 4 o . 1 4

T o t a l 9 6 . 8 2 9 6 . 2 5 9 7 . 0 7 1 0 2 . 8 0 1 0 0 . 6 1 1 O O . 3 4 7 O 7 . 5 7 7 O 7 . 7 7 9 9 . 6 8

C a t i o n s p e r 1 8 o x y g e n s C a t i o n s p e r s i x o x y g e n s

S i

A 1

M g

F e

C a

4 . 9 9

3 . 9 9

1 . 7 5

o . 2 6

o . 0 2

S i

A l 1 w

1 . 9 7

o . 0 9

1 . 9 1

o . 0 9

7 . 9 1

o . o g

5 . o 2 5 . O

l . 9 8 3 . 9 81 . 6 5 i . 8 1

o . 3 ) o . 2 2

o . 0 2 o . 0 3

1 . 9 3 7 . 9 7 1 . 9 2

o . o 8 o . 0 9 o . o B

2 . O 2 . O O 2 - O O 2 . O O 2 . o o 2 - o o

A l v r

M g

F e

Mn

C r

o . 04

1 . 2 9

o . 5 8

o . o 1

o . 0 6

o . 7 t

o . o 1

o . o 1

o . 0 4 o . O l l

1 . 2 3 1 . 2 O

o . 7 4 0 . 7 6

o . o 1 0 . o 1

o . 0 1 0 . o l

o . o 5 o . 0 6

1 . 3 0 1 . 2 9

o . 6 6 0 . 6 6

o . o ' 1

1 1 . O 1 1 1 . o o 1 1 . 0 4 2 " 0 2 2 . 0 2 2 . 0 3 2 - O 2 2 . o 2 2 . O

M g , / M g + F e o . B 7 o . B l o . 8 9 M g , / M g + F s + 2 o . 6 7 o . 6 4 o . 6 3 o . 6 0 0 . 6 7 0 . 6 6

fnd ica te g ra ins in contac t w i th an thophy l l i te in Tab le 4a l .

MgO. Small amounts of K-feldspar can occnr in assem-blages from the orthopyroxene zone (Table l; van Reenen,1983a, Table l), but this mineral was never recognizedin orthoamphibole-bearing assemblages (Table 2). Ac-

cordingly, most of the minerals in the pelitic rocks of thisstudy can be expressed in A'FM diagrams, which illustratethe dependence of assemblages on the Fe/Mg ratio of therocks (see also Grant, 1981, p. I 129). Biotite has beenomitted from these diagrams because it lies out of theAFM plane owing to Ti substitution (Table 5; van Reenen,1983a, Table 4d).

Initial phase relations before hydration

Mineral assemblages in metapelitic gneisses from theorthopyroxene zone represent the initial conditions as-sociated with isothermal uplift (M2) before the establish-ment of the retrograde isograd during M3 (Fig. 3). Thegeneral phase relations in these metapelitic granulites [withMgO/(MgO + FeO):0.59 - 0.70, Group 2P in Tablell are illustrated in Figure I la. The tie lines in this dia-gram connect the actual compositions of the coexistingminerals in six representative samples from widely dif-ferent areas in this metamorphic zone (van Reenen, 1983a,Fig. 5). The chemographic relations in this diagram illus-trate the continuous Fe-Mg reaction garnet + qvaraz :cordierite * hypersthene, which is petrographically ex-pressed by the presence ofcorona textures (van Reenen,1983a: Figs. 3b, 3c, 3d,, 4a). This reaction records con-

DR157

iIo ^6 "

l ^

00

X F e

Fig. 10. A-site occupancy and AIN vs. X"" for coexisting ged-rite and anthophyllite.

vAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD lN LIMPOPO BELT

Table 7. Representative electron-microprobe analyses of anthophyllite

D V , 3 8 ' n R . r 5 7 ' D R . 1 5 7 D V . l - D V . 1 D V . 4 3 D v . 2 2

9 1 1

s i 0 2

A l 2 o 1

T i O 2

M g o

l e O

C a O

M n O

N a r o

C r . , O .

T o t a l

5 3 . 0 2 5 4 . 4 1 5 3 . 7 5

3 . 9 9 3 . 1 5 3 . 9 6o . o B o . o l o . 0 4

2 2 . 7 0 2 2 . 7 2 2 3 . 4 3

1 7 . 8 5 1 7 . 9 6 1 7 . 5 7

o . 2 7 0 . 1 6 0 . 2 0

o . o B o . 1 4 o . 1 6

o . 2 9 o - 2 2 o . 2 7

o . 1 ? 0 . 1 6 o . 1 5

5 4 . 1 9 5 4 . 5 5

2 . 4 7 3 . 3 6o . 0 2 o . o 7

2 2 . 2 7 2 2 . 9 6

7 8 . 1 5 \ 5 . 0 6

o . 1 6 0 . 2 4

o . o g o . 1 2

o . 1 Z o . z 2

o . 2 0 0 . 4 2

5 1 . 8 3 5 2 . 9 44 . 8 9 3 . 6 7o . 1 1 0 . 1 1

1 9 . 5 1 2 0 . 3 3

2 0 . 1 7 1 9 . 4 6

o . 2 5 o . 3 0

o . 2 7 0 . 2 3

O . l 8 O . 2 7

o . 1 4 0 . 1 7

9 7 . 7 9 9 8 . 9 5 9 9 . 5 1 ) 9 7 . 7 2 9 7 . o 9 7 . 5 5 9 7 . 4 2

c a t i o n s p e r 2 3 o x y g e n s

S i / . > 4

o . 4 5

7 . 6 3o . 3 7

7 . 4 9o . 5 r

7 . 7 o 7 . 7 7 . 4 8 7 . 6 1

o . l o o . 3 0 o . 5 2 o . 3 9

8 . o 8 . oB . oB . oB . o B . o 8 . o

A I V l

Mg

F e

T i

C r

C a

Mn

o . 1 4

4 . 8 7

2 . 0 5

o . o 1

o . 0 2

o . o 3

o . o 2

o . 2 o o . 1 5

4 . 6 8 4 . 7 j

o . o 1

o . 0 2 0 . 0 2

o - 0 3 o . 0 2

o . o l o . 0 2

o . 1 2 o . 2 5

4 . 7 2 4 . 8 3

2 . 1 6 1 . 7 8_ o . o 1

o . 0 2 o . 0 5

O . 0 2 O . O r r

o . 02 0 . o ' 1

o . 3 7 o . 2 3

4 . 2 0 4 . 3 6

2 . 4 3 2 . 3 4

o . o 1 0 . 0 1

o - 0 2 o . 0 2

o . 0 4 o . 0 5

o . o l o . 0 3

M - M .1 4

? . 0 6 7 . o o 7 . o 4 7 . o 4

N a ( A ) o . o B o . 0 6 o . 0 7 o . 0 5 o . 0 6 o . 1 1 o . 0 6

M g , / M g + F e + 2 0 . 6 9 0 . 6 9 o . 7 o 0 . 6 9 o . 7 3 o . 6 3 0 . 6 5

' rnd ica te g ra ins in contac t w i th hypers thene in Tab le 4c .

7 . 0 8

ditions of essentially isothermal (Z : 800"C) uplift (M2,Fig. 3) that raised crustal blocks from depths of at least35 km (10 kbar) to 23 km (7 kbar, van Reenen, 1983a).This same reaction has run to completion in samples withMgO/(MgO + FeO) > 0.70 (Group lP in Table l), there-by giving rise to assemblages of cordierite + hyper-sthene + biotite (+ quartz + plagioclase) without garnet.

Final phase relations after hydration

The orthopyroxene- and cordierite-free assemblages (Fig.I lb), which represent the final conditions in the orthoam-phibole zone after hydration, are characterized by thepresence ofcoexisting anthophyllite + gedrite + kyanite(+ quartz + plagioclase + biotite) in MgO-rich rocks

[MgO/(MgO + FeO) > 0.59, Table l]. Figure llb alsoillustrates the influence of the MgO/FeO ratio of the rockson the composition ofcoexisting minerals in the orthoam-phibole-bearing gneisses. This is shown by comparing thephase relations in sample DVI [MgO/(MgO + FeO) >0.7, Group lA in Table ll with those of samples DV22and DV43 [MgO/(MgO + FeO):0.594.70, Group 2.{

in Table ll. Garnet is also absent from the MgO-rich rocks(DVl, Fig. llb), whereas the FeO-rich assemblages(DRl84, Fig. l lb) contain only garnet, gedrite, and quartz.

Phase relations on the retrograde isograd

Two important mineral assemblages that were recog-nized along the isograd in relatively Al-poor metapeliticgneisses are shown in Figure I lc: garnet * hypersthene *

anthophyllite + quartz -1- plagioclase (+ biotite) in FeO-rich and AlrO3-poor bulk compositions with MeO/MeO +FeO) < 0.59 (DV38, Group 3I in Table l), and the ap-parent low-variance assemblage of hypersthene + antho-phyllite + gedrite + cordierite + kyanite + quartz +plagioclase (+ biotite) in more MgO- and Al'Or-rich bulkcompositions with Meo/(Mgo + Feo) > 0.59 (DV3 andDRl57, Group 2I in Table l). These minerals occur inthe same thin section, but not in direct contact with eachother. Cordierite is intergrown with gedrite and kyanite(Fig. 6b), and hypersthene is partly or completely sur-rounded by anthophyllite (Fig. 6a). The following sub-assemblages (in addition to plagioclase and biotite) are

912 vAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE IsocRAD IN LIMpopo BELT

Table 8. Representative electron-microprobe analyses of Al-rich anthophyllite (first two) and gedrite

D V . 1 D R . 1 8 4 D R . 7 5 7 D R . 1 5 7 D V . 3 D V . 1 D v . 4 l M . 4 4 7 D V - 2 2 Dv .22 DR . I 8r t

s i 0 2

o t r o l

T i o2

Mgo

F e O

C a O

M n O

N a r o

C r r O "

T o t a l

4 9 . 7 ' t 4 6 . 6 8

1 0 . 1 4 7 7 - O 5

1 . o 1 0 . 0 6

7 9 . 5 6 7 7 - 7 7

7 5 . 5 8 1 9 . 7 2

o . 3 4 o . 4 2

o . 7 3 o . 0 5

o . 8 6 o . 9 )

o . 7 5 O . 1 5

4 7 . 6 5 4 4 . 1 4

2 0 . 5 8 7 8 . 4 2

o . 7 7 O . 1 9

1 6 . j 6 1 ? . 3 7

7 5 . 7 9 7 6 . 8 5

o . 2 1 O . 2 7

o . 1 5 o . 7 5

2 . 2 ) 1 . 8 1

0 . 0 6 0 . o 2

4 2 . 4 5 4 3 . 7 ) 4 5 - 1 5

2 0 . 6 8 7 9 . 6 4 1 5 . 4 5

o . 0 4 - o - 2 3

7 5 . 7 9 7 7 . 7 6 7 5 . 5 )

7 6 . 5 3 7 3 . 4 7 1 9 . 8 2

o . 1 4 o . 2 5 O . 4 7

o . 1 0 o . t 4 o . 3 7

2 . 2 7 7 . 7 6 7 . 4 2

o . 0 2 - o . 1 7

4 \ . 2 8 4 6 . 1 8

7 6 . 0 6 1 4 . 6 3

o . 4 6 o . 0 9

75 .95 76 .56

1 8 . 1 8 7 9 . 7 7

o . 3 7 o . 2 9

o . 2 7 O . 3 0

1 . 4 3 7 . 2 9

o . 3 2 O . 1 7

46 .34 45 -62

13 -87 13 -09

o . 0 8 0 . 4 9

76 .68 16 . t z

1 9 . 0 5 7 9 . 6 r

o . 3 4 o . 4 9

o . 2 8 0 . 0 6

1 . 2 5 1 . O O

o . 1 f 0 . 2 6

9 7 . 5 0 9 6 . 7 6 9 7 . 4 7 9 9 . 6 3 9 2 . 4 j 9 6 . 6 9 9 8 . 5 j 9 7 . 3 2 9 9 . 2 8 9 8 . o 1 9 6 . 8 0

C a t i o n s p e r 2 3 o x y g e n s

S i

A l a v

6 . 9 2 6 . 8 67 . 0 3 7 . 7 4

5 - 9 7

2 . 0 3

6 . 2 47 - 7 6

6 . 0 9 6 . 2 1 6 . 4 97 . 9 7 1 . 7 9 7 . 5 7

6 . 4 r 6 . 5 ?7 . 5 9 1 . 4 3

6-66 6-667 . 3 4 7 . 3 4

B . o B . o 8 , o B . o 8 . o 8 . o 8 . o 8 - o 8 . o 8 . o 8 - o

M g

F e

Cr

C a

1"1 n

7 . 4 5

3 . 5 4

7 . 9 0

o . 0 2

o . o 1

o . 0 3

o . 02

' t . 2 9

1 4 1

1 . 9 8o. 02

o. 04o- 02

1 . 7 1

3 - 3 32 . 3 9o. o fo . o 2o. ()6

o. 04

o . 6 7 o . 7 7

4 . 1 6 3 . 7 5r . 8 6 2 . 4 2o . 1 1 0 . 0 1o . o B o . o 2o . 0 5 o . o 7o - o 2 0 . o 1

7 . 5 8 7 . 5 0

) . 2 5 3 . 7 6

1 . 9 8 1 - 5 9

o . o l

o . 0 2 0 . 0 4

o . o 1 0 . 0 2

7 . 1 4 7 . o )

3 . 4 4 j . 5 7

2 . 2 O 2 . 3 5

o - 0 5 o - o 1

o - 0 3 o . 0 2

o.06 () . ( )4

o . 0 3 o . o 4

1 . 0 1 o . 9 2

3 . 5 7 3 . 5 22 . 2 9 2 . 4 Oo . o l o . o 5o . o 2 o . o 3o - o 5 o - o 8o . o 3 o . o 1

M - M -1 4 7 . o o 7 . o 5 6 . 8 j 6 . 9 r 6 . 9 8 6 - 9 i z - o7 . O 6 . 9 9 6 . 9 8 V . o 7

N a ( A ) o . 2 4 O . 2 7 o . 6 2 o . 4 9 0 . 6 3 o . 4 9 o . 4 o o . 4 o 0 . - 3 6 o . 3 5 o - 2 8

M g , / M g + t ' s + 2 o . 6 9 o . 6 1 o . 6 5 o . 6 5 0 . 6 2 o . 7 o o . 5 8 0 . 6 r 0 . 6 0 0 . 6 1 ( ) . 6 0

recognized: hypersthene * anthophyllite + gedrite +gutu, cordierite + gedrite + kyanite + q\artz, and cor-dierite + gedrite + anthophyllite + quartz. The presenceof a possible low-variance assemblage in MgO-rich rockson the isograd (gedrite + anthophyllite * cordierite +kyanite + quartz) is indicated by the dashed tie line be-tween anthophyllite and kyanite in Figure l lc. Garnet ispresent in the more Feo-rich assemblages, but is probablyabsent in the more MgO-rich assemblages. This is indi-cated by the absence oftie lines between garnet and theother minerals in the latter rock type.

Mnr,cldoRpHrc coNDrrroNsMetamorphic temperatures at the isograd and in the

orthoamphibole zone can firstly be estimated by com-paring published information on the solvus in the or-thoamphibole system to the results of the present study.

The orthoamphibole-bearing assemblages from the or-thoamphibole zone have some important distinctions whencompared with similar rocks from elsewhere. Medium-

grade assemblages high in the kyanite-staurolite zone andlow in the sillimanite-staurolite-muscovite zone (T:-530-550'C, Spear, 1980) from central Massachusettsare characterized by the absence ofcoexisting orthoam-phibole + kyanite + sillimanite (Robinson et al., 1982,Fig. 87). Sillimanite coexists with gedrite in higher-gradeassemblages (T: 625-650oC, Spear, 1980, p. I I 16) fromthe middle of the sillimanite-muscovite-staurolite zonefrom "Amphibole Hill," southwestern New Hampshire,but the orthoamphiboles were above the orthoamphibolesolvus (Robinson et al., l97l).

The orthoamphibole-bearing assemblages from the ret-rograde isograd show similarities with samples from thesillimanite-orthoclase zone (650-675"C) of central Mas-sachusetts (Robinson and Tracy, l979,rn Robinson et al.,1982, Fig. 99) where gedrite coexists with hyperstheneand garnet in fairly FeO-rich rocks. Anthophyllite is pres-ent within gedrite as a few colorless patches in sharplydefined contact with the gedrite host (Robinson et al.,1982, p. 190). The coexistence ofhypersthene and silli-

VAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT 9t3

Fig. lla. Topology of the reaction gamet (Ga) * quartz:cordierite (Cor) + hypersthene (Hy) shown in an A'FM diagram.Arrow indicates direction ofreaction toward Fe enrichment dur-ing M2 isothermal uplift.

manite in this rock type is excluded by the tie line betweengedrite and garnet. Although coexisting gedrite + antho-phyllite * hypersthene (which is restricted to the retro-grade isograd in the present study) is not described fromthis locality, it should be a stable assemblage in sampleswith a restricted bulk composition (Robinson et al., 1982,Fie. 99).

The orthoamphibole-bearing assemblages in the silli-manite-orthoclase zone of central Massachusetts formedunder conditions about 50 deg higher than the suite ofentirely hypersolvus orthoamphiboles studied from the"Amphibole Hill" area from southwestern New Hamp-shire (Robinson et al., 1982, p. 190). The opening of theorthoamphibole solvus at these high temperatures isascribed to the high Ti content of the gedrite (with 0.10Ti per formula unit), and the presence of anthophyllitepatches in gedrite are described as relicts from a progradereaction anthophyllite + garnet: orthopyroxene * ged-rite + HrO (Robinson et al., 1982, p. 190). The presenceof an orthoamphibole solvus in samples from the retro-grade isograd in the southern marginal zone, however,cannot be explained by a high Ti content ofthese gedrites(maximum Ti content of gedrite less than 0.05 Ti performula unit). It is possible, therefore, that the metamor-phic temperatures along the retrograde isograd might belower than the estimate (625-650qq for "Amphibole Hill"(Spear, 1980, p. I I 16).

The presence of kyanite in the orthoamphibole-bearinggneisses at the isograd (Fig. 6b) and in the orthoamphibolezone (Figs. 4b, 4c) supports the temperature estimate ofless than 625-650{ forthe metamorphic conditions basedon the presence of the orthoamphibole solvus. This im-plies a minimum pressure of -6 kbarifHoldaway's (1971)AlrSiO,-phase diagram is used (Fig. 3). These conditions

Fig. I lb. Chemographic relations ofthe phases anthophyllite(Ant), gedrite (Ged), kyanite (Ky), and earnet (Ga) in the or-thoamphibole zone as shown in an A'FM diagram. Garnet un-stable in Mg-rich compositions. I : DVl, 2: DV43,3 : DV22'4 : DRl84, A: composition of samPle.

support the presence of a systematic temperature gradientin the southern marginal zone of the Limpopo belt afterrapid uplift (M2) and during the M3 hydrating event (Fig.3). Isobaric cooling from M2 (Fig. 3) gives a temperature(based on the presence of kyanite) of more than 700'C,which is about 100 deg higher than the crest ofthe solvusin the orthoamphibole system (600 + 25'C) as defined bySpear (1980, Fig. la).

Fig. I lc. Chemographic relations ofthe phases cordierite (Cor),gedrite (Ged), kyanite (Ky), anthophyllite (Ant), garnet (Ga), andhypersthene (Hy) on the isograd, as shown in an A'FM diagram.Garnet unstable in Mg-rich compositions (DV3, DRl57)' as in-dicatedby absence oftie lines. I : DV3,2 : DRI57, 3 : DV38,4 : DY34;tand o : composition of samples; other symbols asin Figs. I la and I lb.

9t4 vAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT

Metamorphic temperatures in the different zones (in-cluding the orthopyroxene zone) were also estimated withthe garnet-biotite thermometer (Ferry and Spear, 1978).The partitioning of Fe and Mg between biotite (in thematrix) and the rim composition of garnet yielded tem-peratures that changed systematically from the orthopy-roxene zone (mean value for l0 pairs, 8lOt; range,740-890'C) through the isograd (mean value for 5 pairs, 640C;range, 6 l0-680"C) to the orthoamphibole zone (mean val-ue for 8 pairs, 530"C; range, 510-570"C). These estima-tions cannot be seen as cooling temperatures because thepresence of a systematic temperature gradient (Fig. 3) sug-gests fluid flow at a fixed temperature during hydration.The temperature of 810'C for the orthopyroxene zone,based on the garnet-biotite geothermometer, also agreeswith calculated temperatures of -800.C based on othermethods (van Reenen, 1983a).

The estimated temperature ofless than about 625-650tfor the metamorphic conditions on the retrograde isogradis far below the upper thermal stability limit of antho-phyllite in quartz-bearing rocks (>750.C at pressures >3kbar, Greenwood, I 963) under conditions where the waterpressure eqrnls the total pressure. It follows that the waterpressure must have been much lower than the total pres-sure during the hydration ofcordierite and hypersthene.

Fr-un coMprosrrroN oN THE rsocRADEvidence has been presented that anthophyllite, hyper-

sthene, and quartz might be in equilibrium in metapeliticgreisses along the retrograde isograd. The anthophyllitebreakdown reaction may therefore be used to estimate themaximum fluid composition under conditions where P. :Prournuia. The calculated equilibrium curve anthophyllite :7 enstatite + qtrartz + water, based on the data of Helge-son et al. (1978), is shown as the solid line in Figure 3.Alsoshown on this figure is the box (M3) representing theP-7 estimates for the isograd assemblages. In order todisplace the calculated equilibrium curve to the P-7box,the fugacity of HrO must be lowered. The amount oflowering can be accomplished by diluting the mole frac-tion of HrO in the fluid phase to a value of about 0.2, ascalculated using a modified Redlich-Kwong equation ofstate for nonideal mixing of HrO and CO, (Holloway,1977; Flowers, 1979).

CoNcr-usroNs

The M3 retrograde orthoamphibole isograd in thesouthern marginal zone of the Limpopo belt is expressedin Mg-rich pelitic gneisses by two hydration reactions thatare superimposed on M2 corona textures of cordierite andhypersthene aftergarnet: hypersthene i quartz + HrO:anthophyllite, and cordierite * HrO : gedrite * kyan-ite * quartz. Anthophyllite * hypersthene + quartz co-exist at a temperature of less than 625-650t and a min-imum pressure of6 kbar in a narrow zone defined by theretrograde isograd, whereas chemically similar, but com-pletely retrogressed and recrystallized rocks south ofthe

isograd are characterized by the presence of coexistingorthoamphiboles and kyanite.

The estimated temperature of hydration for the isogradis far below the upper thermal stability limit of antho-phyllite in quartz-bearing rocks under condtions wherethe water pressure equals the total pressure. A modifiedRedlich-Kwong equation of state for nonideal mixing ofHrO and CO, was used to calculate the mole fraction ofH,O of about 0.2 in the fluid phase necessary to displacethe equilibrium curye anthophyllite : 7 enstatite *quartz + water to the estimated P-Z condition on theisograd (Fig. 3). The COr-rich retrograde fluid infiltratedtoward the end of the Limpopo tectonothermal event whilethe rocks in the southern marginal zone still had a tem-perature gradient (Fig. 3). This retrograde event eitheraccompanied, or closely followed, the emplacement of theMatok charnockitic-granitic pluton at 2620 Ma (Fig. 2;Du Toit et al., 1983). The results of a fluid-inclusion studyacross the isograd and possible sources of the COr-richhydrating fluid will be discussed in a paper under prep-

aration.

Acxrvowr,rncMENTS

This study forms part of the South African National Geody-namics Project, and the financial support ofthe CSIR is grateftllyacknowledged. Special thanks are also due to the Anglo AmericanResearch Laboratories in Johannesburg for permission to usetheir microprobe facilities. I thank Cornelis Klein for his criticalcomments on an early draft of this paper and Takashi Miyanofor performing the thermodynamic calculations while on leavefrom the University of Tsukuba, Japan. I am also grateful to J.M. Rice and J. A. Grant for their constructive reviews that im-proved the manuscript considerably. Special thanks go to LincolnHollister for his time and energy in the field and for his criticalcomments on the final draft of the manuscript. Elsa Maritz typedthe manuscript and the figures were drafted by N. Maree.

RnrunnNcns

Barton, J.M., Jr., Du Toit, M.C., van Reenen, D.D., and Ryan,B. (1983) Geochronological studies in the southern maryinalzone of the Limpopo mobile belt, southern Africa. GeologicalSociety ofSouth Africa Special Publication, 8, 55-64.

Beeson, R. (1978) The geochemistry of orthoamphiboles andcoexisting cordierites and phlogopites from south Norway.Contributions to Mineralogy and Petrology, 66, 5-14.

Du Toit, M.C., and van Reenen, D.D. (1977) The southern mar-gin of the Limpopo mobile belt, northern Transvaal, with spe-cial reference to metamorphism and structure. Geological Sur-vey of Botswana Bulletin, 12, 83-97.

Du Toit, M.C., van Reenen, D.D., and Roering, C. (1983) Someaspects of the geology structure and metamorphism of thesouthern marginal zone of the Limpopo metamorphic com-plex. Geological Society ofSouth Africa Special Publication,8 , t2 t - r42 .

Ferry, J.M., and Spear, F.S. (1978) Experimental calibration ofthe partitioning of Fe and Mg between biotite and garnet. Con-tributions to Mineralogy and Petrology, 66, ll3-117.

Flowers, G.C. (1979) Correction of Holloway's(1977) adaptationof the modified Redlich-Kwong equation of state for calcula-tion ofthe fugacities ofmolecular species in supercritical fluidsof geologic interest. Contributions to Mineralogy and Petrol-ogy ,69 ,315-318.

vAN REENEN: ORTHOPYROXENE-ORTHOAMPHIBOLE ISOGRAD IN LIMPOPO BELT 915

Gable, D.J., and Sims, P.K. (1969) Geology and regional meta-morphism of some high-grade cordierite greisses, Front Range,Colorado. Geological Society of America Special Paper 128,87 p.

Grant, J.A. (1981) Orthoamphibole and orthopyroxene relationsin high-grade metamorphism of pelitic rocks. American Jour-nal ofScience. 281. 1127-1143.

Greenwood, H.J. (1963) Synthesis and stability of anthophyllite.Journal of Petrology, 4, 3 17 -351.

Helgeson, H.C., Delany, J.M., Nesbitt, H.W., and Bird, D.K.(l 978) Summary and critique of the thermodynamic propertiesof rock-forming minerals. American Journal of Science, 278-A, r-229.

Holdaway, M. J. ( I 9 7 I ) Stability of andalusite and the aluminiumsilicate phase diagram. American Journal of Science, 27 | , 97 -l 3 l .

Holloway, J.R. (1977) Fugacity and activity of molecular speciesin supercricical fluids. In D.G. Fraser, Ed. Thermodynamicsin geology, 16l-181. Reidel, Dordrecht, The Netherlands.

James, R.A., Grieve, R.A.F., and Pauk, L. (1978) The petrologyof cordierite-anthophyllite gneisses and associated mafic andpelitic greisses at Manitouwadge, Ontario. American JournalofScience, 278,4l-63.

Kamineni, D.C. (1975) Chemical mineralogy of some cordierite-bearing rocks near Yellowknife, Northwest Territories, Can-ada. Contributions to Mineralogy and Petrolo gy, 5 3, 29 3-3 lO.

Lal, R.K., and Moorhouse, W.W. (1969) Cordierite-gedriterocksand associated gneisses ofFishtail Iake, Harcourt Township,Ontario. Canadian Journal ofEarth Sciences, 6,145-165.

Leake, B.E. (1978) Nomenclature ofamphiboles. American Min-eralogist, 63, 1023-1053.

Mason, R. (1973) The Limpopo mobile belt, southern Africa.Royal Society (I-ondon) Philosophical Transactions, 427 3,463-485.

Rabbitt, J.C. (1948) A new study of the anthophyllite series.Amorican Mineralogist, 33, 263-323.

Robertson, I.D.M., and Du Toit, M.C. (1981) Mobile belts. A.The Limpopo belt. In D.R. Hunter, Ed. The Precambrian ofthe Southern Hemisphere, 641-671. Elsevier, Amsterdam.

Robinson, Peter, and Jaffe, H.W. (1969) Chemographic explo-ration of amphibole assemblages from central Massachusettsand southwestem New Hampshire. Mineralogical Society ofAmerica Special Paper 2,251-274.

Robinson, Peter, Ross, M., and Jaffe, H.W. (1971) Compositionof the anthophyllite-gedrite series, comparisons of gedrite-hornblende, and the anthophyllite-gedrite solvus. AmericanMineralogist, 56, 1005-1041.

Robinson, Peter, Spear, F.S., Schumacher, J.C., Laird, J., Klein,

C., Evans, B.W., and Doolan, B.L. (1982) Phase relations inmetamorphic amphiboles: Natural occurrence and theory. InD.R. Veblen and P.H. Ribbe, Eds. Amphiboles: Petrology andexperimental phase relations, l-227 . Mineralogical Society ofAmerica Reviews in Mineralogy, 9b.

Spear, F.S. (1978) Petrogenetic grid for amphibolites from thePost Pond Volcanics, Vermont. Carnegie Institution of Wash-ington Year Book 77, 805-808.

- (1980) The gedrite-anthophyllite solvus and the com-position limits of orthoamphibole from the Post Pond Vol-canics, Vermont. American Mineralogist, 65, I103-1 118.

- (1982) Phase equilibria of amphibolites from the PostPond Volcanics, Vermont. Journal of Petrology, 23, 383-426.

Stout, J.H. (1970) Three-amphibole assemblages and their bear-ing on the anthophyllite-gedrite miscibility gap. (abs.). Amer-ican Mineralogist, 55, 312-313.

-(1971) Four coexisting amphiboles from Telemark, Nor-way. American Mineralogist, 56, 212-224.

-(1972\ Phase petrology and mineral chemistry of coex-isting amphiboles from Telemark, Norway. Journal of Petrol-ogy, 13,99-146.

van Reenen, D.D. (1981) The southern marginalzone. In Barton,J.M., Jr, Ed. Geology ofthe Limpopo belt. Limpopo ExcursionGuidebook, Geological Society ofSouth Afica, 19-44.

- (1983a) Cordierite + garnet * hypersthene + biotite-bearing assemblages as a function of changing metamorphicconditions in the southern marginal zone ofthe Limpopo meta-morphic complex, South Africa. Geological Society of SouthAfrica Special Publication 8, 143-167.

-(1983b) Hydration of cordierite and hypersthene and thenature ofthe orthopyroxene isograd in the Limpopo belt, SouthAfrica. Geological Association of Canada/Mineralogical As-sociation of Canada/Canadian Geophysical Union, Joint An-nual Meeting, Victoria, Program with Abstracts, S, 472.

van Reenen, D.D., and Du Toit, M.C. (1977) Mineral reactionsand the timing of metamorphic events in the Limpopo meta-morphic complex south of the Soutpansberg. Geological Sur-vey of Botswana Bulletin, 12, 107 -128.

- (1978) The reaction garnet + quartz : cordierite + hy-persthene in granulites of the Limpopo metamorphic complexin northern Transvaal. Geological Society ofSouth Africa Spe-cial Publication 4, 149-177.

Vernon, R.H. (1972) Reactions involving hydration of cordieriteand hypersthene. Contributions to Mineralogy and Petrology,35 , 125-137.

MaNuscnrpr REcETvED ApnIl 13, 1984MeNuscnrpr AccEprED Mrncn 18, 1986