<<

Canadian Mineralogist Vol. 15, w. M3-249 (1971)

CALCIUMZONING IN ALMANDINEGARNET, WISSAHICKONFORMATION, PHIIADEIPHIA, PENNSYTVANIA MARIA LUISA CRAWFORD Bryn Mawr College,Bryn Mawr, Pa. 19010 USA

ABsrRAcr INTRoDUcTIoN with clearly developed Ca zoning gen- As our knowledge of the details of erally are associated with zoned plagioclase in chemistry in rnetamorphic rocks continues to samples of interlayered pelites and calcareous sand- grow it is apparent that the calcium content of stones in the northwest of almandine garnets may be used as a geothermo- , Pa. Systematic Ca zoning is absent in meter or a geobarometerin certain metamorphic metamorphio assemblages with hieh variance, al- (Ghent et al. 1975). though the garnet may be zoned in other elements. assemblages 1976; Hensen Apparently calcium zoning in metamorphic alman- In addition, the calcium zoning patterns in gar- dine-rich garnets develops (1) during metamorphic nets usually differ, often markedly, from the reactions involving plagioclase-garnet equilibria; (2) zoning trends shown by Fe, Mg and Mn (Craw- as a rsult of reactions between otler in ford 1974; de B6thune et al. 1968; Linthout & low-vaxiance assemblages; and (3) in volumes of Westra 1968). Most of the detailed studies of rock subject to changes in calcium content due to garnet equilibria have focused on the Fe, Mg and diffusion during metamorlfiic recrystallization. Mn zoning and the chemical interactions be- producing Where several of these mechanisms for tween garnet and otler ferromagnesian phases. Ca zoning operate simultaneously, the zoning pro- 'Where described,but little attention files are complex. the zoning is relatively Ca zoning has been parageneses simple, the trends of the calcium zoniDg itr alman- has been paid to the and dine, when considered in conjunction with the zon. possible reactions whioh govern the variatign of ing in the coexisting plagioclase, can be used in the the calcium in almandine-rich garnet. interpretation of the thermal history of the sample; An advantage to studying compositional va- tho calcium zoning is independent of the mechanisms riation within refractory minerals such as garDet produce which 26ning of Mn, Fe and Mg. is that zoning patterns, where'preserved,provide Souuenr information on the development of the mineral assemblagelrom the moment when those min- Les grenats contetrusdans les 6chantillons de p€- erals first nucleate in the rocks. The thermal lites et de ere.scalcaires interstratifi6s de la forma- tion Wissahickonau nord-ouestde Philadelphie,Pa., histories of most metamorphic rocks can be Etats-Uni$ et montrant une zonation Ca bien d6- placed in one of three getreral categories: rocks veloppde, sont habituellement associ€s au plagio- which were subjected to rapid heating with con- clase zon6. Une zonation syst6matiquedu calcium sideiable overstepping of mineral reactions; est absentedans les associationsm6tamorphiques i rocks in which the mineral assemblageshave variirnce.6lev6e,bien que le grenat puisse€tre zoo6 been affected by extensive retrograde reactions pour d'autre$ 6l6ments. Apparemment, la zonation as the rocks cooled, and those which recrystal- de calcium dans les grenats m6tamorphiquesriches lized under prograde, near-equilibrium condi- en (l) pendant almandin se d6veloppe les rEactions tions and do not show evidence of retrograde m6tamorphiques impliquant un dquilibre entre le effects. The types of calcium zoning in garnets plagioclaseet le grenat (2) comme r6sultat r€- des discussed by actions entre d'autres mineraux en associations! observed in the first case were faible varianie, et (3) dans des volumes de roches Crawford (1974) and in the second by de 86- soumisesi des variations du contenu en calcium en thune et al. {1975) and Hollister (1976). This, raison d€ Ia diffusion pendant la recristallisation paper deals with garnets in rocks of various m6tamoryhique. Les profils de zonation sont trds compositions from the staurolite zone of the complexes quand plusieurs de ogs m6canismescle amphibolite facies of a prograde metamorphic zonation du calcium agissentau m6me endroit. On terrain. utiliso la zonation du calcium dans l'almandin de The'samples are from the Wissahickon For- pa.ir avec la zonation du plagioclasecoexistant, lors, in gue celle-ci est assezsimple, afin d'6tablir l'histoire mhtion exposed along Wissahickon Creek thermique de cet 6chantillon. I-a zonation du Ca northwestern Philadelphi4 Pa. Most of the sec- est ind6pendantedes m€canismesproduisant la zo- tion consistsof alternating pelitic and moderate- nation du Ml, du Fe et du Mg. ly psammitic units several tens of meters thick Clraduit par la R6daction) in the ex,posedsection. Locally, thin bands (2- 2/+3 2M THE CANADIAN MINERALOGIST

10 cm) of calcareous are interlayered mole percent pistacite in the clinozoisite. How- in the pelitic units. The samples were collected evet, these substitutions will shift the equilibrium from two outcrops of pelite and calcareous sand- in opposite directions and probably will tend to stone with tle intention of comparing the gar- cancel each other. In higher grade rocks in this net compositions in rosks of different calcium area, staurolite breaks down in the stability contents. field of kyanite. This suggests a pressure of at Two sets of samples were studied in detail, least 7 kbar for the region, which is compatible one collected shghtly above the staurolite isograd with the observed stability of the clinozoisite* and the other midway between the first appear- muscovite* assemblage. ance of staurolite and the isograd representing One fundamental observation which is of the reaction staurolite*quartz=garnet+kyanite. considerable importance in deciphering the cal- The sampleswere about 1 km apart, perpendicu- cium zoning in almandine is the common asso- lar to the strike of the isograds. In this area the ciation of zoned plagioclase with calcium-zoned exposedwidth of the staurolite zone is 5.5 km. garnet. Thus the calcium content of a metamor- No differences were found in mineral composi- phic almandine garnet is sontrolled by plagio- tion between the two localities. As the assem- clase-garnet equilibria according to reactions blages are in the staurolite stability field, meta- such as: grossular{ALSiOr*quartz=anorthite. morphic temperatures during recrystallization In addition, equilibria involving additional cal- are inferred to be between550" and 650'C (Fig. cium-bearing phases may be involved: zoisite+ 1). Within the calcareous layers muscovite oc- quartz=anorthite*grossular+Hr0, and anorthite **quartz-grossular*COo. To a first ap- proximation therefore, the grossular content of tho garnet will reflect the Na/Ca ratio of the rocks, and secondly the metamorphic history of the sample, which influences the reactions be- tween Ca-bearing phases.

ANar-Yrrcer- PRocEDrtREs In each specimen the garnets were analyzed for Mg, Fe, Mn, Ca, Al and Si with an ARL EMX electron probe microanalyzer. The anal- yses were made along traverses through the center of the grains at 6 and 10 micron intervals 5O0 6q' 7@ fo6 TABLE1, CO}IPOSITIOTIOf GARftETS ILLUSTRATED.II,I FIGI'RES 2-4 Frc. 1. Pressure-temperature field for samples st02 . 4t203 FeOr Mno Mgo c!0 Total studied. Experimentally determined '| equilibrium 37.4 2l.l r-'" core ?7.3 10.0 2.4 3.2 0't.4 curves: (a) chlorite*muscowite=staurolite4bio- rln 37.4 21.3 32.5 3.9 3.8 2.6 101.5 tite+quartz+H0 Gloschek 1969); O) staurolite rh core 37.4 20.s 29.6 6.5 3.6 3.0 t00.9 '" 31.5 3.9 2.2 t0t.7 fguartz-gdrnet+kyadte+HrO and (c) stauro- rlE 37.5 21.2 5,4 ' o @ra 35.5 20.7 fi.7 .2 1.5 4.6 100.2 litefquartz-garnet*silliinanite+H2o (Ganguly ' rln 36.9 21.1 35.4 .2 3.5 3.0 100.2 1972\; (d) mrlscovitefzoisitefquartz-anorthite 2I. l 26.7 3.3 2.3 9.7 rQ0.5 '. core 37.3 *K-feldsparaH0 (Johannes and Orville 1972; rln 38.1 21.3 25.3 2.0 2.6 ll.l 100.4 Hewitt 1973); (e) muscovitelzoisitelquartz= , ore 37.6 21.5 25,5 5.8 9.2 101.2 ' rln 37.6 21.1 30.0 2.2 2.6 7.0 100.5 atrorthite+K- (Ackermand & Karl 1972). 't0l.7 r core 38.0 2l.l 33.0 2.5 1.8 5.3 . duminosilicate stability curyes from .Holdaway - rln 38.5 21.1 31.1 1.5 10t.0 . Q97r). ll|fi$€rs of lons on the basis of l2(0) si., At Fe iln I',l9 {a .vrith '1.82 curs clinozoisite, plagioclase and quartz; ,- ore 2.98 '1.991.99 .68 ,28 .27 'o rln 2.97 2.15 .26 .45 ,22 potassium feldspar was not seen. The break- '1.96 rL core 2.9a 1.98 .a .43 .26 dirrvn of clinozoisite*muscovite*quafiz approx- '" rln 2.98 1.99 2.09 .37 .40 .19 imately parallels the kyanite= isograd , core 2.96 1.98 2.49 .0r .18 .40 ' rlo 2.96 1.99 2.37 .02 .41 ,26 (Ackermand & Karl L972). Data from Johannes '1.98 , core 2.97. 1.78 .22 .27 .83 & Orville (1972) and Hewitt (1973) place this " rlm 3.00 1.97 1.67 .14 .3I .94 re,action about I kbar higher than suggested by , core 2.96 1.99 1.68 .39 .25 .78 ' rln 2.n . 1.98 2.00 .15 .31 .60 (Fi& '1.97 Ackermand & Karl 1). The experimental - core 3.01 2.19 .17 .21 .45 data need to be adjusted to account for 1O-15 " rln 3.03 1.95 2.O5 .10 .38. .47 'Totrl molerpercent albite,in the plagioclase and 10-15 F" reported as Feo CALCIUM ZONING IN ALMANDINE GARNET

(depending on the size of the grain). Similar tra- ()2). Also, they contain minerals which are verses \trere made across zoned plagioclase, al- stable through a wide range of metamorphic though here the stepping interval was 4 microns. conditions: qlJaftz, plagioclase, biotite, musco- The analyses were corrected using a modified vite, and minor garnet. The calciu,m content of Bence-Albee correction procedure. The diagrams the garnet shows no systematic zoning pattern are plotted in such a way as to smooth out sta- within single grains and remains constant be- tistical counting fluctuations and to show only tween garnets in adjacent layers in contrast to acfual variations in the apparent compositions the behavior of Fe, Mg and Mn (Iable l,Fig.2\, of the minerals. Non-statistical brrors in the This reflects an equilibrium composition for the analysescontribute, in part, to the irregularities plagioclase and garnet across a thin-section of the zoning'profiles. However, whenever checks despite variations in the absolute and relative were run by analyzing the same traverse more abundance of other elements. Examination of than once, most of these fluctuatiois were re- the calcium content and zonation in almandine produced. reported by other investigators also shows that calcium content is not sensitive to variations in Low-CercruM ASSEMBLAGES abundance of other divalent cations in garnet The semi-.peliticsamples studied at.s mineldq- (e.g. Crawford 1974; de B6thune et aI. 1975; gically simple, and hence possessa high variance Linthout & Westra 1968).

I6i- - /k -**- q',V._ J"Kv_{^ d-fr S-a ,no

Fr9. l. Zoning profiles across garnet la and 1b from a semipelitic . Chemical analyses for points indicated by arrows reported in Table l.

Flc.3. Zoningprofile across gamet2 from a pelitlc schist. Arrows as in Figure 2. f9..4, 7'onng profiles from core to rim of three garnets: (a) in the center of a calcareous sandstone layer" (b) in the reaction band between the calcareous layer and the pelite and (c) adjacent to the peiito. A:rows as in Figure 2.

Fro 5. ?t g profiles across two_ plagioclase grains: (a) in the reactiou band between the calcaieous layer and the pelite, (b) adjacent to the *tte. THE CANADIAN MINERALOGIST

the Aluminous samples from the Wissahickon consume staurolite. His observationssuggest plagioclase Formation at this metamorphic gnde com- reaction muscovite*chlorite*calcic garnet* monly contain too many phase for divariant *ilmenite = staurolitetbiotite*calcic plagioclase*I&O. Support' equilibrium when judged by phase-rule ,crite- rutilef quartz*sodic ria : kyanite-staurolite-garnet-biotite-muscovite- ing this postulated reaction are observations that quarti-plagioclase-ilmenite-rutile. This type of Ca increases in the rim of garnet 2 and that, at situation is not unusual and has been explained the same point as tle inflection in the Ca,pro- by postulating additional components such as file, the inclusions in the garnet change from Zn in staurolite and Mn in garnet to increase ilmenite to rutile. The reaction consumesplagro- the variance. Garnet 2, illustrated in Figure 3, clase and it is interesting to note that the plagio- is from this type of sample. As can be seen from clase with gamet 2 is sparse and corroded- It is the analysis in Table 1, the Mn content of garnet not possible to further docu'ment this reaction is negligible; the same holds for ilmenite, and in the Fhiladel'phia region due to a horst of the staurolite is free of Zn. Alternatively, the older gneisseswhich intemrpts the metamorphic low variance may indicate that the fugacity of zonation in the Wissahickon schist below the HzO is buffered by ttre mineral assemblage. staurolite isograd. Further west, along Brandy- There is no petrographic reason to suppose wine Creek where lower grade rocks do occut, Psro I Pro- for the in question. In chlorite is abundant in rocks below the stauro- an important particular, the calcareous layers within the lite isograd and biotite becomes pelites all contain clinozoisite. Clinozoisite phase only above that isograd. witl form only in assemblageswith a high mole Cer,cenr,ous Asseltglecns fraction of HzO, so t}te pelites have acted as a .reservoir of HzO, diluting the CO, produced Calcareoussandstone layers within the Wissa- reaction in the calcareous layers by decarbonation re- hickon Formation at this grade show a more actions. Graphite and sulfide minerals are absent zone which separates them from adjacent from these pelitic schists, precluding any sug- pelitic layers. The centers of the calcareous plagioclase gestion of significant C or S components in the iayers consist of clinozoisite, quartz, fluid phase. and garnet. Some layers also contain muscovite; Thi opaque inclusions in garnet show that others are hornblende-bearing. These layers thb garnet nucleated and grew while ilmenite are separated from pelitic schist by a l-5 mm was stable in the matrix. Staurolite and kyanite, band consisting of quartT. plagioclase, and gar- on the other hand, contain only rutile inclusions minor garnet. Within this band some of the and rutile grains are found in the outermost rirn net contains clinozoisite and, occasionally,horn- of the garnet and in the matrix. Matrix rutile is, blende inclusions. The inclusions show the cal- - in turn, rimmed by ilmenite. The mineral rela- careoussandstone pelite interface band formed, tionships suggest that most of tle garnet growth at least in p"rf from a mineral assemblagele- preceded the nucleation of staurolite and kya- sembling the one in the interior of the calcareous sim- nite' when staurolite appeared' ilmenite was re- layers. Subsequent reactions, however, \av9 to placed by rutile. Thus the low vanance Pf!-b{y pifiea the mineralogy of this band relative mech- ieflects a lack of simultaneous equilibrium ihut of the adjacent calcareous layer. The pelitic and among all of the phases in the sample. anism of forniation of bands between 7-oing in garnet 2 reflects the effects of the calcareous layers in schists has been described changing mineral equilibria as the P and T con- bv Vidale & Hewitt (1973). where ditions affecting the sample changed- This is a In the centers of the calcareouslayers garnet direct result of the low variance of the mineral clinozoisite occurs both as inclusions in garnet (Fig. 4a) and assemblage. Figure 3 shows a gradual decrease and in the matrix, both the in Ca content in Ca and Fe and an increasein Mg during most the plagioclase show an inctease the pelite, in those of the growth history of the garnet. At the frori core to rim. Adjacent to garnet or the point where the inclusions change from ilmenite bands with no clinozoisite in the Ca enrichment to rutile there is an inflection in the zoning pro- matrix, the zoning pattern shows by a Cadepleted files and the Ca trend reverses whereas the Mg in the core and rim separated trend flattens out. The inflection in the zoning zone (Fig. 4c). Ca zor'- pattern apparently coincides with a change in In contrast to the unzoned or simple garnet in tle semipeliric ihe mineral assemblage due to a drscontinuous ing pattern observed in garnets and the associated ieaction during prograde metamorphism. Trzci- an? pelitic schists, the band between tle enski (1971 and pers. comm.) has shown in plagibchses in the reaction pelite show fairly com' similar low-variance assemblagesthat plagioclase cal&r"ous layer and the (Fig. clinozoisite in- ir involved in reactions which produce and plex Ca zoning. Here 4b) CALCIUM ZONING II.I ALMANDINE GARNET clusions occur in a high-Ca garnet core. The sents a growth history that reflects changing core is rim,med in turn by low-, high- and low-Ca metamorphic conditions during crystallization. zones. It is possible to predict which garnets will A second assumption is that successivezones have the most complex variation in Ca as the can be correlated betweenplagioclase 4ad imms- adjacent plagioclase shows oscillatory zoning diately adjacent garnet, counting inrrard from consisting of at least four zones Q high-Ca\ 2 the outermost parts of the grains. This assump- low-Ca, Fig. 5a). tion is supported by the observation that the Despite the variable complexity ef Qa sening, number of zones found in each mineral is in zoning of Mg and Mn shows uniform patterns almost every case the same. Finally, if the zon- similar to those of the garnets in the semi- ing in plagioclase and garnet follow the sdme pelitic and pelitic schist regardless of the po- trend (both increasing or both decreasingin Ca sition of the garnets in the rock. Mn is high- content) it is assumedtlat zoning is controlled est in the center and decreasesoutward. The by the overall availability of calcium to the characteristic bell-shaped curve of the Mn zon- growing minerals. On the other hand, if the ing in the core of the grain flattens and may zoning patterns are antithetic, the most reason- show a slight increase at the margin (4b, 4c), able explanation is that the two minerals are re- Inclusions in the garnets are concentrated in acting in exchange equilibrium. The slope of that part of the grain showing the strongest Mn curves on a P-T diagram representing equilibrium zoriing. This is interpreted as a reflection of most between grossular and anorthite are positive rapid growth and simultaneous rapid depletion @oettcher 1970). An increasing temperature of Mn from the matrix for the central part of or decreasingpressure during crystallization will the garnet with numerous inclusions. . therefore result in more calcic plagioclase in Tho Mg content increases from core to rim equilibrium with less calcic garnet for any pla- whereas the Fe profile is more irregular. Where gioclase-grossularreaction. The reverse will be Ca is strongly zoned it is clear that Fe zoning true as the minerals equilibrate during retro- nnirrors Ca zoning.'Fe zoning is also affected by grade metamorphisrn. Of course, both changes the changes in Mn and Mg. It is not useful to in overall Ca content of the local system and discuss Fe/Mg ratios without recognizing the exchange reactions between garnet and plagio- effect, on the Fe content, of Ca as well as Mn. clase may operate.simultaneously and produce Mg values,however, do not seemto show effects very complex relations. due to Ca zoning; the regular increase in Mg In those parts of the samplesstudied in which contents may reflect garnet growth during in- plagioclase and garnet grew in equilibrium with creasing temp€rature conditlons. clinozoisite (center parts of calcareous layers The mineralogical and chemical variations and garnets with clinozoisite inclusions in the observed in calc-silicate layers embedded in cores) both plagioclase and garnet are uniform schist can be explained, in part, by diffusion of in composition (Fig. 4a). Ca released to the calcium down its activity gradient. This kind of system by the breakdown of clinozoisite shows diffusion of non-volatile components, and its as a slight increase in the Ca content of both effects, has been described bv Vidale & Hewitt minerals. Adjacent to the peliteo the core com- (1973) and Thompson (1975i. The presence of positions of garnet and plagioclase are uniform an actively diffusing species will reduce the but are surrounded by a zone of less calcic gar- number of inert components in the region be- net and of more calcic plagioclase(Fig. 4c, 5b). tween the pelite and the center of the calcareous As clinozoisite is not piesent, the assemblageis layer. This will decrease the number of coexist- probably not buffered with respect to calcium. ing phases,as seen in the interface band. Out- Jhs 76ning presumably reflects a tem,perature ward diffusion of calcium during mineral growth increase, for the plagioclase becomes more cal- will also produce a gradient in the composition cic and, in the garnet, the Ca content decreases. of the plagioclaseand the garnet, as is observed" A pressure dectease would produce the .samc The overall calcium content of garnet and pla- zoning pattern. However in the adjacent pelite gioclase is highest in the centers of the calcare- the evidence, cited above, for growth of garnet, ous layers and decreasesas the contact with the followed by staurolite and kyanite, supports a pelite is a,pproached.The variation in composi- model of mineral growth under increasing tem- tion of individual grains will also reflect the perature conditions. In this paxt of the sample changrng Ca content.in the patrix during rnin- the outer rims of both minerals are enriched in eral growth. calcium possibly due to clinozoisite breakdown Interpretation of ihe zoning pattern in the in the calcic central part of the layer and out- .calcareous.layers.'relieson the. assumption tlat ward diffusion of the Ca ions toward the pelite. tho zoning in garnet and plagioclasegrains repre- In samples from the mid-staurolite zone, the THE CANADIAN MINERALOGIST distinct interface band between the calcareous sure increases. Discontinuous reactions, such as and pelitic layers is most well-developed. Here breakdown of clinozoisite, will provide added the zoning pattern of garnet and plagioclase be- Ca to the system and causethe calcium content comes more complex; the number of composi- of both phases to increase. Discontinuous reac- tion zones in both minerals increases (Fig. 4b, tions which do not primarily involve calcic min- 5a). Interpreting the growth history of the min- erals, such as reaction to forn of staurolite, may erals becomes subjective except that they have also produce a distinctive change in the compo- progressed through the same thermal history sition of the calcic phases.This will occur in low- (iacreasing Z) as the garnet in the immediately variance assemblagesin which the compositions adjacent pelite, discussedabove, and the garnets of all minerals are fixed for each P and T. in the more calcareous part of the sandstone Finally, if the chemical gradient is steep enough, layers. In this part of the rock in which diffusion diffusion will cause chemical changes,minored of Ca, as well as other species,was most activen by changes in calcium content, in garnet and as shown by the sirnplified mineralogy, the min- plagioclaseremoved from the immediate vicinity erals grow in an ever-changing chemical system. of the reaction involved in providing the Ca The combination of this effect and possible ex- ions. change reactions among the calcic minerals as they grow could be expected to produce the CoNcr,usroNs complex zoning pattern$ that appear. The analysis of zoning of calcium in these garnetsfiom rocks with similar thermal histories DrscussroN but variable compositions shows that such zon- The origin of zoning in garnet may be attri- ing patterns are best deciphered when related to buted to one or more of four processes: zoning in coexisting plagioclase.The occurrence (a) Preferential fractionation of one element of a similar number of different zones in those into garnet during gfowth to produce a depletion two Ca-bearing minerals permits the assumption of that element in tle matrix and hence a de- that the zonescan be correlated. In high-variance pletion in later garnet zones. This type of zon- and constant-composition assemblagesthere is ing occurs only if the element is preferentially little or no systematic zoning in the calcium incorrporated in garnet relative to all other min- content of garnet and coexisting plagioclase. erals. For calcium it might occur at low meta- However, in sampleswith low variance, or sam- morphic grade, where plagioclaseis still albite or ples in which there has been a sipificant change if the Ca content of the rock is very low and in chemistry due to diffusion during prograde the growth of both garnet and plagioclase re- metamorphism, calcium zoning in garnet and moves the bulk of the Ca from the reacting sys- coexisting plagioclase is evident. A model for tlre tenr- This will produce decreasing Ca zoning development of zoning as a consequenceof con- profiles in both plagioclase and garnet and was tinuous and discontinuous reactions during near- not observed in tlis study. equilibrium prograde metamorphic reactions (b) Zoning due to exchange reactions a$ can explain the zoning patterns observed and changes in temperature and pressure change the provides a mechanism for inferring thermal his- values of pailition coefficients for a pair of min- tory of the sample during the growth of the gar- erals. There is no other common mineral in me- net and plagioclase. tamorphic rocks in which Ca substitutes freely for Fe, Mg, and Mn, hence the grossular compo- ACKNowl.EDGMENTS nent of garnel does not enter into this type of The analytical part of this study was carried reaction. This mechanism of zoning, which is of out on the electron probe microanalyzer at the great importance for Fe and Mg zoning pat- Department of Geology and Geophysics,Prince- terns, does not apply to Ca. ton University. Use of their facilities is grate- (c) Zoning due to continuous and discontinu- fully acknowledged. The paper was written in garnets ous reactions. The analysis above suggests that connection with a symposium on organ- this mechanism provides the reason for most ized by E. D. Ghent and the Mineralogical Ca-zoning in garnet and plagioclaseof the Wissa- Association of Canada. The paper was consider- hickon Formation. Continuous reaction relations ably improved by comments from D. M. Ker- which involve plagioclase and garnet on opposite rick, E. D. Ghent and L. S. Hollister. sidee of the reaction produce antithetic zoning Rrrrnnrrcss patterns in the two minerals. As temperature AcKERMAND,D. & KABr" F. (1972): Experimental increases or pressure decreases, tle plagioclase studies on the formation of inclusions in pla.do- becomes more calcic and the garnet less so. The claso from metatonalites,Hohe Tauern, Austria. reverse is true as temperature decreases or pres- Contr. Mineral. Petrology 85, ll-12. CALCruM ZONING IN ALMANDINE GARNETS 249

BoETTcSE& A. L. (1970): The System CaO-Al2Oe- l. Sci.27l, 97-131. SiOe-H2O at high pressures and temperatures. ./. HoLLrsrER, L. S. (1976): Cordierite-garnet equili- Petrology LL,337-379. bria near Kwinitsa, British Columbia. Geol. As- CRAwFoRD,M.L. (L974): Calcium zoning in alman- soc. Can. - Mineral. Assoc. Can. Program Absn. dine; s model based on plagioclase equilibria. In r, 70, press, Ttr . Manchester University Hoscrrx, G. (1969): The stability of staurolite and 629-6U. chloritoid and their significance in metamorphism or B6rnuxr, P., LenunoN, D. & Bocqunl J. (1975): of pelite rocks. Contr. Mineral. Petology 22, Diffusion processe$ in resorbed garnets. Cbntr. 208-232. Minerul. Petrology 50, L97-204. JoruNnrs, W. & Onvrr-r,r, p. M. (1972): Zur sta- -, Menr[v, H. & THBuxrssEN, K. (196g): bilitiit der Mineralparagenesen Muskovit+Calcit Grenats zon6s de la zone du Mont Rose (Valle *Quartz, Zoisit+Muskovit+Quartz, Anorthit+ K-Feldspat Anzasca, Prov. de Novara, Italie). Schweiz. Min- und Anorthit*Calcit. Fortschl. Min- eral. Petrog. Mitt. 48, 437-454. eral. 50, Berh. l, 4647. LrNrnour, K. & WssrR L. (1968): GeNcur.v, I. (1972): Staurolite stability and related , Compositiou- al zoning in almandine-rich garnets parageneses: theory, experiments, and applications. and its rela- l, Petrology 18, 335-365. tion to the metamorphic history of their host rocks, Konink. Nederl. Akad. Wetensch. Proc., Gnexr, E. D. (1975): Plagioclase-garnet-Al2sioi- Ser. B, 7L,297-312. quailz, a potential geobarometer-geothermometer. TnoursoN, A. B. (1975): Calc-silicate Amer, Mineral. 6L, 7lO-7 14. diffusion zones between marble and pelitic schist. .I. Pe- Hrxsnry B. J. Scrnm, R. & Wooo, B. J. (1925): trology L6, 3l+346. Activity-composition relationships - for T[zcnxsrr, w. E. (l9zl): On the reaction stauro_ grossular garr.et" Contr. Mineral. Petrology 61, lite 1 muscovite +. quartz. Geol. Soc. Amer. r61-166. Program Abstr. 3,735. HEwrrr, D. A. (1973): Stability of the assemblage VIDALE, R. J. & HEwru, D. A. (1973): 1Mobile" muscovite-calcite-quartz. Amer. Minerar. 58, 785- components in tle formation of calc-silicate 79t. batds. Amer. Minerol. 58, 991-997. Hor,oewev, M. J. (1971): Stability of andalusire Manuscript received September 1976, emended Dec- aud the aluminum silicate phase diagram. Amen ember 1976.