CORDIERITE-GARNET GNEISS and ASSOCIATED MICRO- CLINE-RICH PEGMATITE at STURBRIDGE, I,{ASSA- CHUSETTS and UNION, CONNECTICUTI Fnor B.Cnrbn, [

THE AMERICAN MINERALOGIST, VOL 47, IVLY AUGUST, 1962

CORDIERITE-GARNET GNEISS AND ASSOCIATED MICRO- CLINE-RICH PEGMATITE AT STURBRIDGE, I,{ASSA- CHUSETTS AND UNION, CONNECTICUTI Fnor B.cnrBn, [/. S. GeologicalSurttey, Washington,D. C.

Aesrnacr Gneiss of argillaceous composition at Sturbridge, Massachusetts, and at Union, Connecticut, 10 miles to the south, consists of the assemblagebiotite-cordierite-garnet- magnetite-microcline-quartz-plagioclase-sillimanite. The conclusion is made that this assemblagedoes not violate the phase rule. The cordierite contains 32 mole per cent of Fe- end member, the biotite is aluminous and its ratio MgO: (MgOf I'eO) is 0.54, and the garnet is alm6e5 pyr26.agro2.espe1.2.Lenses of microcline-quartz pegmatite are intimately associated with the gneissl some are concordant, others cut acrossthe foliation and banding of the gneiss. The pegmatites also contain small amounts of biotite, cordierite, garnet, graphite, plagioclase, and sillimanite; each mineral is similar in optical properties to the corresponding one in the gneiss. It is suggestedthat muscovite was a former constituent of the gneiss at a lower grade of metamorphism, and that it decomposedwith increasing metamorphism, and reacted with quartz to form siliimanite in situ and at lerst part of the microcline of the gneiss and pegmatites These rocks are compared with similar rocks of Fennoscandia and Canada.

INtnouucrroN Cordierite-garnet-sillimanitegneisses that contain microcline-quartz pegmatiteare found in Sturbridge,Massachusetts, and Union, Connecti- cut. The locality (Fig. 1) at Sturbridgeis on the south sideof the \{assa- chusettsTurnpike at the overpassof the New Boston Road; this is about 1 mile west of the interchangeof Route 15 with the Turnpike. The other locality is about 1] miieseast of the villageof Union, on the north side of Bigeiow Hollow Road, Connecticut Route 198, about 100 feet west of Bigelow Pond. Thesegneisses are part of the widespreadBrimfield Schist (Emerson,l9l7 , p. 68-72;and Rodgersand others,1959, p. 46 and map). The two localitiesare about 10 milesapart.

DrscuprroN ol RocKS The gneissat Sturbridgeis gray, medium- to coarse-grained,and shows a rough compositional banding of mafic and quartzo-feldspathic units from a few millimeters to a few inches in thickness.Irregular folia of intergrown biotite and sillimanite with scattered anhedral garnet and cordieriteare set in a matrix of massiveaggregates of plagioclase,q\artz, microcline,and minor amountsof biotite. Nluscoviteis absent. l,{agnetite, graphite, and pyrite are present in small amounts. Proportions of the minerals vary considerablyfrom one part of the gneissto another.

1 Publication authorized by the Director, U. S. Geological Survey.

907 I'-RED BARKNR

T.,ASSACIIUSEITS OONNECTICIII +NoRrIrI

ffi

Frc. 1. Map showing locations of cordierite-garnet-sillimanite gneiss in Sturbridge, Massachusetts, at A, and in Union, Connecticut, at B.

An analysisof typical gneiss,free of pegmatite,from Sturbridge was madeby chemistsof the U. S. GeologicalSurvey usingthe rapid methods of Shapiroand Brannock (1956): Sio:.. 53.1 N.rO. 1.7 Tioz |.2 KzO 3.5 Alzor. 24.1 HzO 7.+ FzO: 1 4 P:Os 0.06 IreO 8.4 MnO 0.10 Mgo 3.7 co: .. o.o5 CaO. 15 Total IOO 2

The compositionof the gneissis comparableto that of a shale,except for its low content of HzO and its high ratio of FeO: FezOr.The low content of H2O probably is due to dehydration during metamorphism.The original sedimentmay have containedenough ferrous clay to accountfor the 8.4 per cent of FeO, but it is more probable that original Fe2O3,in hydrous iron oxides,was partially reducedto FeO during metamorphism'perhaps by carbonaceousmaterial. The gneissat Union is similar to that at Sturbridge, except that the folia of biotite and sillimanite form only 10 to 20 per cent of the rock; cordierite and garnet ar'e much more abundant; and both garnet and cordieriteoccur in the quartzo-feldspathicportions. There is a gradation CORDI DRI T E-GA RN ET G1I/E1.'S

lrarn

l'rc.2. Mafic band in gneiss from llnion, connecticut. Biotite, bio; cordierite, cdt; garnet, gar; microcline, mic; myrmekite, myr; plagioclase,pl; qtartz, qtzl and sillimanite, sil.

from scatteredgrains and clustersof microclineto small, lenticular and blob-likepegmatites that appearto have replacedthe gneiss.An example of the latter is shownin Fig. 4. rn thin section the folia of biotite and sillimanite in the gneissat union wrap around equant, anhedral garnets and irregular grains of cordierite. small amounts of plagioclaseand quartz are interstitial to these folia; microclinerarely occursinside them, but commonly embays their margins.A typical view is given in Fig.2. The quartzo-feldspathicaggregates are granular mosaics of andesine, qtartz, microcline microperthite, and biotite, with occasional cordierite, garnet, and sillimanite. Rims of myrmekite lie betweenplagioclase and microcline;they are similarto ones in migmatite pictured by Schreyer (1958, Taf. 10, Abb.2). The volu- metric ratio of myrmekite to the adjacent microclineis about 1: 10. The amount of microcline varies greatly from place to place, even as seenin thin section.The relative amounts of microcline and plagioclasecom- monly vary inversely;the feldsparin one thin sectionmay be almost al1 plagioclase,whereas that in a section cut from adiacent rock mav be FRED BARKER

: i', . . r ..,.r, r. ::'::'.... .' ..rrr.InnFt.l,,,.:,,-.'

Frc. 3. Small aggregateof microcline in cordierite-rich gneiss,union connecticut' The grain of myrmekite in right center has only a core remaining of the original grain of andesine.Biotite, bio; cordierite, cdt;garnet, gar;microcline, mic;myrmekite, myr;plagioclase,pl; quartz, qtz, sillimanite, sil; magnetite, mt; and zircon, zr ' Plane light' nearly ali microcline.Some grains of microclineinterfinger into grains of biotite as thin sheetsparallel to the cleavageof the latter. Grains of garnet and cordierite that are embayed by microcline are seenin thin section,as in Fig. 3. Thin shellsof sillimanite are presentbetween much of the microclineand embayedgarnet. DnscnrprroN oF MTNERALS Analysesof biotite,cordierite,and garnetfrom the gneissat Sturbridge are given in Table 1. The formula of the biotite, calculatedon a basisof 24 anions,is: CORDIERITE.GARNET GTZYSS 911

Tegr,r 1. ANu,vsns ol BrorrrB. Connremrr. lm GanNrr rnou Brornr- Conornmm-GenNnr-Me.cNnrrrn-Pllcrocr,asr-Quantz- Srrr,rMANrrE GNrrss, Srunnnrocn, MassecRusners

Biotitel Cordierite2 Garnet2

SiOz 37.3 48.2 39.L TiOz t\ 0.00 0.03 AlzOs r7.3 32.9 22.2 FezO: 0.7r 0.00 0.00 l'eO 15.5 7.4 30.8 Mso 10.3 8.78 6.58 CaO 0.42 0.05 1.04 NurO 0.19 n.d. n.d. KzO 9.6 n.d. n.d. HzO3 J.l 1.66 n.d. F 0.54 n.d. n.d. MnO (0.otq)n 0.06 0.56

Sums 99.3 99.05

1 Analysis by B. L. Ingram. 2 Analysis by C. A. Kinser. 3 Determined by Penfield method. a Determination by semiquantitative spectrographic analysis. 6 Oxygen equivalent for fluorine subtracted from total. n.d.: not determined,

That of cordierite,on a basisof 18 atoms of oxygen,is:

Mgr.soFeo64A14 00Si4 sgOrs' 0.57HzO Schreyerand Yoder (1960,p. 94) have suggestedthat HrO may be present in molecularform in intersticesin the cordieritestructure, and so the formula of cordierite is given with H2O as water of hydration. The formula of the garnet on a basisof 24 atomsof oxygenis:

Fea eMgr s:Cae 17Mn6 67Ala 67Si6e7o2a The molar ratios MgO: (MgOf FeO) of theseminerals are: biotite, 0.54; cordierite,0.68;and garnet,0.28.Formula of the garnetin termsof end numbersis:

alm6e bpyr26 agro2 espet 2 The 0 and 7 indices of refraction in sodium light of biotite at both Sturbridgeand Union are about 1.640.The biotite is pleochroicfrom pale yellowish brown to red-brown.Indices of the cordieritefrom Sturbridge arei ot:1.541,0:1.547, and 7: 1.553,all *0.002; thoseof Union cordierite are similar exceptthat a : 1.542.For garnetat Sturbridge n : t.795, + 0.004,and that from Union is L792. 912 FRED BARKER

The distortion index of the cordierite,as definedby Miyashiro (1957, p. 44-45), is 0.25+0.02. Garnet was *-rayed by R. O. Fournierl the aver- agesof four determinations(hhl: 12.2.0, lI'5'2, 12'6'0, and 10'9'3) from Sturbridgegarnet and of five determinaLions(hkl: 12'2'0, ll'5'2, t2' 6' O,12' 6' 2,and 10.9' 3) of Union garnetwere each o: 11.49 + 0.05A. n{ost grainsof microclineshow grid twinning, but somedo not. About half of the grains of this mineral are microperthitic with fine filaments or irregular patchesof sodic plagioclase.The volume of plagioclasein the perthite rangesfrom a trace to about 20 per cent. Microcline from a large crystal in pegmatite at Sturbridge was homogenizedby heating at 900'C. for 60 hours and then r-rayed. The (201) method with KBrOs as an internal standard was used (Orville, 1958, p. 208) and indicated a compositionof 82 weight per cent of KAISiaOe. The plagioclaseis sodic andesine. Gradationally zoned grains are not uncommon;these have rims of calcicandesine or sodiclabradorite.

l,{rralronpruc PEGMATTTES Lensesof pegmatitefrom a fraction of an inch to about 10feet in thick- nesslie scatteredin the gneiss.The lensescommonly lie parallel to the foiiation of the gneiss,which locally wraps around them. A few lenses, however,transect the foliation. Blobs or blotchesof pegmatite of very irregular shape,from an inch to about 10 feet in maximum dimension, are alsofound in the gneiss.These latter pegmatitesmostly transect the foliation of the gneisswithout displacingit, and many of them decrease in grain size toward the margins and fade into the quartzo-feldspathic aggregatesof the gneiss.Typical examplesare shown in Fig. 4. These blobs are interpreted to have replacedthe gneiss.Both forms of pegmatite consistof about 80 to 95 per cent by volume of microcline,with the remainder qvartz, biotite, and accessoryminerals. Cordierite, garnet, graphite,and rarely plagioclaseand sillimanite are found along the margins of the pegmatite.The optical propertiesof theseminerals are virtually identical to their counterpartsin the gneiss.Muscovite is not present in thesepegmatites. ntlargins of many of the pegmatitesare granulated. In thin section the microcline microperthite, quartz, and cordierite of a mar- ginal specimenare seen to be closely fractured and show pronounced unduiatory extinction. \{uch of the microclinein the larger bodies,however,is in translucentgrains of maximum dimension1 to 4 inches. The pegmatitesdo not show chilled zones,internal zoning,or reaction zoneswith the enclosinggneiss. The samesuite of mineralsof like compo- sitionsis found in the pegmatitesas in the adjacentgneiss. Furthermore, the experimentalstudies of Tuttle and Bowen (1958,p.54-75) on the system KAiSi3O3-NaAlSiBOs-SiO2-HrOindicate that microciine-rich CORDIERI T E-GARN DT GNTEISS

Frc. 4. Specimens of gneiss with small microcline-quartz pegmatites from Sturbridge, Massachusetts (upper), and Union, Connecticut (lower).

pegmatite magma would be far removed from the minimum-melting composition of this system, would be molten at temperatures of about 800"-900o C. (depending on the water pressure), and thus such a melt would be expectedto react strongly with the abundant plagioclaseof the gneiss. Absence of such reaction, together with the field and textural relationsof the pegmatites,suggests that they are not of magmatic,but of metamorphic origin. One probable source of their potassicfeldspar is found in the gneiss itself, in the reaction of muscovite and quartz to sillimanite and microcline. Solutions derived from granitic magmas are anotherpossible, but problematical,source of the potassicfeldspar in the pegmatites. The sillimanite of the gneissis believedto have formed largely from muscovite. The gneiss contains abundant AlrOa and KzO, and is of a proper bulk composition for growth of muscovite at lower metamorphic rank. The transformation of muscovite to sillimanite and potassic feldspar may be representedby the well-known equation (seeHeald, 1950, p.74-75):

MUSCOVITE+ QUARTZ: SILLIMANITE+ K-FELDSPAR(* H,O), wherethe substitutionof Na for K in muscoviteand feldsparis neglected. This reaction has temperature, rock pressure,and activity of HzO'as external variables.At Sturbridge and Union this reaction is believedto have taken place, with sillimanite formed in silu, and with the constitu- 914 FRED BARRER ents of microcline perhaps first dissolved in an aqueous phase, trans- ported a short distance,and then depositedas pegmatite.

Apprrc,luoN oF THE Pn.rsB RulB The fact that gneiss and pegmatite of virtually identical mineralogy formed at Sturbridgeand at Union, 10 milesapart, implies that the grade of metamorphism was the sameat the two localities, and also that equilib- rium was reachedat eachplace. These rocks, as now seen,are believedto representa stable assemblage,one probably formed at or very near the peak of the metamorphism. The phaserule may be applied to theserocks. There are 9 components presentin significantamounts; theseare: Al2O3,CaO, H2O, FeO, Fe2Os, K2O, NIgO, Na2O,and SiOz.The componentHzO probably was mobile to the extent that its activity was constant throughout the rock massunder discussion.The remaining 8 componentswould be "fixed" or "inert" (Thompson,1955, p.81;Korzhinskii, 1959,p. 17).The phaserule thus allows 8 minerals to be stable at arbitrary rock pressure,activity of H2O, and temperature.Eight mineralsare present:biotite, cordierite,garnet, magnetite, sodium-bearing microcline, plagioclase, quartz, and sillimanite. The phaserule is not violated. The assemblagemay be viewed as forming an 8-phaseregion in 8- component space, with the independent variables temperature, rock pressure,and activity of HzO externally applied. Small variations of the 8 "fixed" componentswould not changethe assemblageof minerals,but would merely change their relative proportions. The slightly higher proportions of cordieriteand garnet in the gneissat Union are a reflectionof slightly greater proportions of MgO and FeO than in the gneiss at Sturbridge. The two accessoryphases graphite and pyrite, and their associated componentscarbon and sulfur deservespecial comment. The presenceof graphite and pyrite in the gneisssuggests that reactionsof C and S with oxygen-bearingcomponents to form COzand SOzhave not occurredto any noticeableextent at the presenthigh metamorphicgrade. Any earlier reactionscannot be traced. In applicationof the phaserule C and S are classedas inert components,and the system is consideredclosed to COz and SOz.The activitiesof any COzor SOrformed in theserocks would be both negligiblysmall and variablefrom placeto place.This is in contrast to metamorphism of carbonate-bearingrocks, which have much higher activitiesof COz,and which may have CO2(or C) as a mobile component (Thompson,1955, p. 79;Korzhinskii, 1959, p. 117-135). Theserocks may be partially representedin an AFNI diagram,as shown in Fig.5. The ratios of NIgO:FeO in the biotite, cordierite,and garnet CORDIERITL,GARNETGTZYSS 915 are shown. This diagram, of course,is not a ternary phasediagram, and the four minerals shown on it may coexist stably. The garnet and cordierite are the most magnesianand most ferrous,respectively, that would be expectedin assemblageswith biotite and sillimanite. In an assemblage free of cordierite,with similar externalconditions but different composition, both garnetand biotite would be moreferrous. Similarly, in a garnet- free assemblagethe cordieriteand biotite would be more magnesian.An interestingpaper by Chinner (1959,p. lI2-ll4) discussesthe stability

ailliEanite

garne! biotlte

Frc. 5. AFM diagram for gneissat Sturbridge, Massachusetts,and Union, Connecticut, showing compositional relations of biotite, cordierite, garnet, and sillimanite. Other minerals of the assemblage are graphite, magnetite, sodium-bearing microcline, plagioclase, pyrite, and quartz. of garnet,cordierite, and biotite in regionalversus thermal environments, and shows tie lines of 7 cordierite-garnetpairs on the diagram Mg cordierite-Fecordierite-almandite-pyrope.

OrnBn OccunnBNcns Cordierite-garnet-bearinggneiss similar to the ones describedabove are found in northwesternCanada, Finland, Sweden,and Norway. A gneissnorth of Great Slave Lake, N. W. T., Canada, consistsof andesine,biotite, cordierite, garnet, graphite, microcline, sillimanite, green spinel, and tourmaline (Folinsbee,1940, 1941).The green spinel (pleonaste) presumably is not stable with quartz. Folinsbee alalyzed the cordieriteand garnet, and found that the cordieritecontains 39 mole 916 FRED BARKER per cent of Fe-end member, and the garnet 24 mole per cent of pyrope' Pegmatitesare closelyassociated with this gneissand contain cordierite and garnet. Cordierite-garnetgneisses are found interlayeredwith the granulitesof Finnish Lapland, as describedby Eskola (1952).Some of thesegneisses are migmatitic. A gneissfrom the mouth of sotajoki consistsof the same assemblageas the rocks at Sturbridgeand Union, exceptthat graphite is absent.The cordieritehas 31 mole per cent of Fe-end member, and the garnetis 39 per cent pyrope (Eskola, t952, p.149, 153).Associated rocks contain hypersthene. Cordierite-garnetgneiss is found over a wide areain southernFinland. Hietanen (1943,p. 12-17) describeda biotite-cordierite-garnet-ilmenite- microcline-oligoclase-quartz-sillimanitegneiss from Hii?ihdjdrvi, Kalanti, area,southwestern Finland. The indicesof refractionare given (Hietanen, 1943,p. 12, 14): biotite: ts:'v:7.647 cordierite: q: 1.547 B: r.ssz

sarnet: ;: ;.;;;

Theseindices suggest that the compositionsof these three minerals are similar to those of the cordierite-garnetgneisses of Sturbridge and Union. Hypersthenewas not found in the Kalanti area (Hietanen, 1943). In the Turku areaHietane n (1947, pl 1023-1026, 1 0 71 ) f ound cordierite- garnet rocks like those at Kalanti, to the northwest. Migmatites are abundant here; those in cordierite-garnet gneissconsist largely oI quattz, microcline, and plagioclase,and also contain cordierite and garnet (Hietanen, 1947, p.1071). Theserocks differ from thoseof New England in several respects.First, the Finnish rocks commonly show an inter- iayering of thin bands of cordieritegneiss, garnet gneiss,and cordierite- garnet gneiss.In thin section, cordierite and potassicfeldspar show a reaction texture, having formed biotite and sillimanite' In addition, a secondgeneration, or later growth of cordierite is seen,in which large grains of this mineral contain many inclusions of quartz, and of biotite, in parallel arrangement (Hietanen, 1947, p. 1025-t026). Hypersthene- bearing rocks were found in this area. The West Uusimaa compiex of crystallinerocks' which lies 70 to 130 km east of Turku, has been describedby Parras (1958). Cordierite- garnet gneisseshere are coarsegrained, are intimately mixed with veins and lensesof quartz-microclinepegmatite, and contain layers of garnet- hypersthenegneiss. Specimen no. 53 of Parras (1958, p. 119) consists of the sameassemblage as the gneissaL Sturbridgeand Union. The garnet CORDIE,RITE-GARNETGITZI.'S 917 of no.53 contains26 mole per cent of pyrope; the B-indexof the biotite is 1.644,very closeto the 1.640value of the New England biotite; and the cordieritewas not studied. The associatedplagioclase is Anzo.Speci- men no. 60 of Parrasis the sameassemblage; its plagioclaseis An25,and the other minerals were not studied. Cordierite-garnetgneisses also crop out in southern Finland near Mikkeli and Lake Laatokka (Parras, 1946),and near Porkala (Kranck, 1937,p. 77-78). Parras, Hietanen, and Eskola (seereferences as given) believe that metamorphismof argillaceousrocks has produced the cordierite-garnet gneisses. The Scirmland gneisses,southwest of Stockholm, Sweden, contain cordierite-garnetgneiss, with intimately mixed microcline-richpegmatite (Magnusson,1950, p. 17).J.A.W.Bugge (1943,p. 16) has described similar rocks from the vicinity of Riscir and Kragerd, southern Norway. Anthophyllite, rather than hypersthene,is associatedwith theseSwedish and Norwegian rocks. Several remarks may be made in summary. The association of cordierite-garnet gneiss and microcline-rich pegmatite, both free of pri- mary muscovite, is a common one. The author follows Holmquist's (1921) assertion that these are "venitic" pegmatites, i..e.,formed by segregationof material originally contained in the gneissitself. Forma- tion of microcline-quartzpegmatite in these localities perhaps may be related to the decompositionof muscovite,as suggestedabove for those of Sturbridge and Union. Cordierite-garnet gneissesare associatedwith anthophyllite (gedrite)- bearing rocks in some localities, with hypersthene-bearing rocks in others. The compositions of biotite, cordierite, and garnet in these gneissesare much the samein either case,except for the pyropic garnet described by Eskola. The assemblage biotite-cordierite-garnet-microcline-magnetite-plagioclase-quartz-sillimanite is stable through the range of external conditions (temperature, rock pressure, and activity of HsO) over which the transition: anthophyllite--+hypersthene takes place (in rocks of appropriate composition,of course).Thus this assemblage spans the amphibolite-granulitefacies boundary of Eskola; this point has been discussedby Eskola (1952, p.162-169) and briefly by Fyfe and Turner (1958,p. 159). Much detailed work must be done to determine the extent of stability of biotite and of cordierite in the so- called "granulite" facies.

AcrwowrBoGMENTS The author expressessincere thanks to Drs. Werner Schreyer, Geo- physical Laboratory, Carnegie Institution of Washington, and E-an Zen, rJ. S. Geological Survey, for careful criticism of the manuscrint- 918 I,'RL,D BARKII.R

RBlBrnNces

Buccr, J. A. W. (1943), Geological and petrographical investigations in the Kongsberg- Bamble formation. N or gesGeol. U.nd'ers., 160. CrrrNnnn, G. A. (1959), Garnet-cordierite parageneses.Carnegie Inst. Wask , Year Book, 58, tl2-11+. EutnsoN, B. K. (1917), Geology of Massachusetts and Rhode Island. U. S Geol' SurLey 8u11.,596. Esrora, P. (1952),On the granulitesof Lapland. Am. Iow. Sci., Bowen Volume,133-171. Folrxssnr, R. E. (1940), Gem cordierite from the Great Slave Lake area, N. W T., Canada (abs ). Am. Mineral., 25, 216. --- (1941), The chemicai composition of garnet associatedwith cordierite. Am. Minerol , 26. 50-53 Fvrn, W. S., F. J.TunNan, ANDJ. VERHoocEN (1958), Metamorphic reactions and metamorphic facies.Geol. Soc. Am. Mem.,73. Hurn, M. T. (1950), Structure and petrology of the Lovewell Mountain quadrangle, New Hampshire.Geol. Soc. Am. Bull.,6l, 43-89. HrrreNrN, A. (1943), Uber das Grundgebirge des Kalantigebietes im siidwestlichen Finn- land. Comm. gEoI.Finlond'e Butrl.., l3O. --- (1947), Archean geology of the Turku district in southwestern Finland. Geol'.Soc' Atn. Bull., 58, 1019-1084. Hor-nqursr, P. J. 0921), Typen und Nomenklatur der Adergesteine. GeoI' Fi)ren. Fdrlt, 43,612-631. Konzntxsrrr, D. S. (1959), Pltysiochemical Basis oJ tlrc Analysis oJ the Porogenesis of MineraIs. Consultants Bureau.Inc.. New York. Knnxcr, E. H. (1937), Beitriige zur Kenntnis der Svecofenniden in Finnland. Comm. G6oI.Finland,e BUII-,ll9, 69-91. MncNussox, N. H. (1950), The origin of the Scrmland gneisses.Intel. GeoI.Cong',18th, Great Britoin 1948, Rept , Pt. 3, 14-19. Mrvesnrno, A. (1957), Cordierite-indialite relations. Am. Iour. Sci.,255,43-62- Onvtr-r.r,P.M. (1953),Feldsparinvestigations Carnegielnst.Wosh,Yeor Book,57,206- 209. PARRAS,K. (1946), On the coarse-grainedgarnet-cordierite gneissesof south and southwest Finland. Comm. g4ol.Finlande Bull.,138, l-7. --- (1958), On the charnockites in the light of a highly metamorphic rock complex in southwestern Finland. Cornrn.g6otr Finlande Bull.,l8l. Roocrns, J. R. M. Gerrs, eNl J. L. RosnNrnll (1959), Explanatory text for preliminary geological map of Connecticut, 1956. Conn. Geotr.Not. Hist. SurtteyBull.,84. Scnnrvnn,W. (1958),Die Quarz-Feldspat-Gefugeder migmatischen Gesteinevon Vilshofen an der Donau. NeuesJahrb. Minerol.,92, 147-170. --- AND H. S. Yoorn, Jn (1960), Hydrous Mg-cordierite' CarnegieInsI. Wosh., Year 800k,59,91-94. Snerrno, L. eNo W. W. BuNrocr (1956), Rapid analysis of silicate rocks' L/' S. Geol. Swoey Butrl'.,1036-C, 19-56. TrrouesoN, J. B. (1955),The thermodynamic basisfor the mineral faciesconcept' Am. Jour' Sci., 253, 65-103. Tutrr.r, O. F. ,tNo N. L. BowBN (1958), Origin of granite in the light of experimental stud- ies in the system NaAlSi:OrKAiSirOrSiOrH:O . Geol'S oc. A m' M em.,7 4.

Manuscript receitseil,Notsember 15, 1961.