American Mineralogist, Volume 73, pages 20-47, 1988

Devonian and Carboniferous metamorphism in west-central Maine:The muscovite-almandine geobarometer and the staurolite problem revisited

M. J. HololwlvDepartment of Geological Sciences, Southern Methodist University, Dallas, Texas 75275, U.S.A.

B. L. DurnowDepartment of Geology, Louisiana State University, Baton Rouge, Louisiana 70803, U.S.A.

R. W. HrNroN*The Enrico Fermi Institute, The University of Chicago, Chicago, Illinois 60637, U.S.A.

ABSTRAcT

Important thermal metamorphic events in west-central Maine occurred at 400 Ma (Mr),394-379 Ma(Mr), and325 Ma(Mr). EachiscloselyassociatedwithemplacementofS-typegranites, such that the isograd patterns produced in the surrounding pelitic schists generallyfollow plutonic outlines. From north to south, grade of metamorphism varies from chloriteto sillimanite-K-feldspar-muscovite.

Mineral-chemistry studies on M, and M, indicate the following: (l) Much staurolite-grade chlorite is retrograde as demonstrated by the lack of a consistent relation betweenbiotite composition and presence or absence ofchlorite, given the restricted range ofX"roallowed in these reduced graphitic rocks. Consequently, the first sillimanite-forming re-action for many pelitic schists in Maine does not involve chlorite. (2) Garnet zoningpatterns are prograde in rocks of the staurolite zone and retrograde in rocks of highergrade. (3) Presence of graphite and nearly pure ilmenite suggests low Fe3* in micas andallows for end-member calculations that include a "Ti biotite" (KTiD(Fe,Mg)AlSi3Oro(OH)r)and a "Ti muscovite" (KTi(Fe,Mg)AlSi3O,o(OH)r). (4) Staurolire contains about 3 H (48-oxygen basis) and shows subtle indications of nonideality in Fe-Mg Ko relationships.

Garnet-biotite geothermometry indicates the following average temperatures for the firstoccurrence of minerals: staurolite-510'C, sillimanite-580 "C, sillimanite-K-feldspar-660 "C. The muscovite-almandine-biotite-sillimanite (MABS) geobarometer is calibratedon the basis of an average M. pressure of 3.1 + 0.25 kbar such that staurolite * muscovitebreak down directly to sillimanite + almandine + biotite immediately above the silli-manite-andalusite phase boundary. On the basis of this geobarometer, the M, rocks insidethe second sillimanite isograd crystallized at 3.8 kbar.

The muscovite-quartz dehydration boundary shows excellent agreement with garnet-biotite temperatures and MABS pressure for the second sillimanite isograd in M, ifreasonable models for behavior of fluids in reduced graphitic pelites are assumed. Thestaurolite-quartz dehydration boundary predicts temperatures about 60 oC above the gar-net-biotite temperatures for the first sillimanite isograd in Mr. However, this discrepancycan be partially eliminated by assuming a regular pseudobinary staurolite solution modelwilh WG: I 5 kJ per tetrahedral Fe site. Nonideality in staurolite solid solution may alsoexplain staurolite Ko relationships that involve reversals in Ko values.

The progressive southward increase in pressure from early i|l42(2.35 kbar) to M3 (3.1kbar) to M5 (3.8 kbar) over a75-m.y. period requires a net increase of about 5 km of rockduring a time when much of New England was suffering erosion. This P increase is believedto have been the combination of two efects: (l) as the locus of metamorphic heat movedsouthward, the depth of metamorphism increased by 2 to 3 km over a distance of 125 kmlaterally, implying postmetamorphic tilting of the area and/or differences in original to-pography; and (2) at any given place, P increased with time because of development of awidespread Devonian volcanic highland and correlative emplacement of granites at shal-low levels.

The extensive terrane of Mr, which varies from K-feldspar-bearing to K-feldspar-absentrocks, marks incipient granite melting immediately below the present rocks at 3.8 kbar,660 'C; local ?" variations probably reflect local differences in degree of mobilization ofmagma and fluids. This presumably occurred above the gently north-dipping contact ofthe Sebago batholith exposed south of the M, rocks.

* Present address: Grant Institute of Geology, Edinburgh, Scotland EH93JW.

0003-004x/88/0 l 02-0020$02.00 20

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE 21

INrnooucuoN

West-central Maine provides a unique opportunity tounravel the complexity of Acadian and Hercynian meta-morphic events, as it lies in the transitional zone betweenslightly metamorphosed rocks to the north and high-grademetamorphism to the south. Within this zone, the dis-tinctive overprints of three thermal metamorphic eventshave been preserved. Previously, the metamorphic pe-trology, isograds, mineral reactions, and metamorphichistory of this region, bounded by the cities of Augusta,kwiston, Norway, Rangeley, Kingfield, and Madison (Fig.l), have been summarized by Holdaway et al. (1982) andGuidotti et al. (1983). This area is one of dominantlypelitic metamorphic rocks and S-type granites, mainly ofthe New Hampshire Magma Series, and it represents thetransition from chlorite-zone rocks to the prevailing up-per amphibolite-grade metamorphism that is dominantin southern Maine, southern New Hampshire, and cen-tral Massachusetts. Lithologic units and their ages aresummarized by Holdaway et al. (1982) and referencestherein.

The present report provides data on mineral chemistryand metamorphic conditions for the dominant M. andM, metamorphism of the region covered by Holdaway etal. (1982). Pressure estimates for M, are deduced frombiotite-garnet geothermometry combined with aluminumsilicate stability relations. In addition, the potential for amuscovite-almandine-biotite-sillimanite (MABS) geoba-rometer is examined and applied to estimate the pressureof Mr. Garnet-biotite geothermometry, combined withexperimental stability data on muscovite-quartz andstaurolite-quartz, allows for a test of models for the be-havior of C-H-O fluids in pelitic rocks. These data arethen integrated with the present understanding of themetamorphic history of Maine to provide insight into theconditions and causes of metamorphism in Maine.

These metamorphic rocks are ideal for evaluating andformulating pelitic geothermometers and geobarometersbecause the lack of synchronous deformation allows forthe assumption of nearly constant pressure during eachthermal event, and ubiquitous presence of graphite allowsfor the assumptions of low /o, and low and/or approxi-mately constant Fe3* content in Fe-bearing phases.

M2

Following M,, a widespread thermal event producedstaurolite, staurolite-andalusite, and andalusite in rocksof appropriate composition and grade (terms as in Gui-dotti, 1970). This event is easily recognized in much ofthe area between Farmington, Kingfield, and Rangeley(Fig. l, Holdaway et al., 1982) where retrograde meta-morphism has downgraded staurolite- and andalusite-bearing assemblages and in the process produced distinc-tive pseudomorphs. These authors estimated the timingof this event as >394 + 8 Ma. Recently, Gaudette andBoone (1985) have obtained a whole-rock Rb-Sr age of400 Ma on the Lexington batholith (Fig. l). Dickersonand Holdaway (ms.) have correlated intrusion of this plu-ton with Mr. M, andalusite occurs as a relict mineral inseveral areas most recently affected by M, (Holdaway etal., 1982). A regional study of M, will be the topic of afuture communication.

M3

The most prevalent metamorphism in the area is re-gional-contact metamorphism spatially associated withthe Mooselookmeguntic, Songo, Phillips, Livermore Falls,and Hallowell batholiths or groups of plutons (Fig. l).Whole-rock Rb-Sr ages range from 394 to 379 Ma (Dall-meyer and Vanbreeman, 1978; Moench and Zartman,1976). This metamorphism produced the grade sequencebiotite - almandine - staurolite - sillimanite, all pres-ent with muscovite. Isograds broadly follow plutonic out-crops. No major retrogressive effects have been noted.Grade designations and their mineral assemblages (ab-breviations after Kretz, 1983) for M, and M, are definedby number as follows:

l cht (M3)2 Bt + Chl (M3)3 Grt + Bt + Chl (M3)4 St + Grt + Bt (+ Chl) (M3)5 Sil + st + Grr + Bt (M3)6 Sil + Grt + Bt (M3)6.5 Sil + Grt + Bt (within grade 7) (M')7 Kfs + Sil + Grt + Br (M5)8 Kfs + Sil + Grt + Bt (trace Ms) (M,)

BnrBr suvrnrARy oF METAMOR.HT. E'ENTS All assemblages contain Ms + Qtz * Gr + Pl + Tur.

Hordawav er ar. (1e82) summarized rhe sequence of ;"fill",'JffiiiJff:1":J#:11!ffiTiJi?:"tuf:":metamorphic events of the region considered here' o b::t

n""j-o*r"ts M. isograds for tire staurolite zone (gradesummary of these events, as they are now interpreted, ii'rift--"rite zone (grades 5 and 6), and K-feldspar-sil-follows' hmanite zone (gradei 6.5 and 7 in Mr). Apart from theM, local appearance of M, andalusite, textures and compo-

The first event, slightly older than 400 Ma, produce4 sitions indicate that in grades 4 to 8, all prograde minerals

widespread chlorite-zone metamorphism synchronous equilibrated with M3 or M, conditions'

with S, schistosity (Guidotti, 1970). Much of the areabetween Madison, Farmington, and Kingfield (Fig. l) has M4

suffered only this event, and temperature has not been The recent dating of the Lexington batholith at 400 Masignificantly higher in this area during the subsequent (Gaudette and Boone, 1985) puts the significance of Moevents. for this area in doubt. The Hartland pluton, 15 km north-

22 HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

1. Rangeley2. Phi l l ips3. Kingfield4. Anson5. Rumford6. Dixfield7. Farmington8. Norridgewock9. Bryant Pond

10. Buckfield11. Livermore12. Augusta13. Norway14. Poland15. Lewiston16. Cardiner

Fig. 1. Isograd map for M, and M, in west-central Maine. Isograds shown are M, staurolite (S0, M, sillimanite (Sil), and M,K-feldspar-sillimanite (Kfs-Sil). Whole numbers refer to analyzed specimens. Bold decimal numbers indicate P determinations fromTable l0 in kbar. Pressures in parentheses are those from grade 7 (Table 10) that may represent disequilibrium as discussed in text.Acidic plutonic units-pattern: MB : Mooselookmeguntic batholith, PB : Phillips batholith, LG : Livermore Falls group, HG :Hallowell group, RB : Reddington batholith, LB : l,exington batholith, SkB : Skowhegan batholith, SoB : Songo batholith, SeB :Sebago batholith (Hercynian). Gabbroic plutonic units-stipple. For names of quadrangles, see diagram and list at right. For completedistribution of mineral assemblages, see Holdaway et al. (1982). Modified from Holdaway et al. (1982), Henry (1974), Dutrow(1985), and Evans and Guidotti (1966). Igneous contacts from Osberg et al. (1985). M, metamorphism is not shown and will bediscussed in a future contribution (see also Dickerson and Holdaway, ms.; Holdaway et a1., 1986a). Isograds stop near Skowheganbatholith because its metamorphism involves andalusite and mav be M,.

east ofthe Skowhegan pluton (Fig. l) is the closest datedpluton (whole-rock Rb-Sr, Dallmeyer et" al., 1982) at 362Ma, the age Holdaway et al. (1982) assumed for Mo. Wetentatively conclude that Mo did not affect the area understudy.

Ms

Recent isotopic studies for the Sebago batholith (Fig.l) give Hercynian ages of 325 Ma (Hayward and Gau-dette, 1984) and 324 + 3 Ma (Aleinikoff et al., 1985)

I 2 3 4

5 6 7 8

9 l 0 l l l 2

t 3 t4 l 5 l 6

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRALMAINE z 5

using Rb-Sr whole rock and U-Pb on zircon, respectively.Lux and Guidotti (1985) and Guidotti et al. (1986) havepointed out that much of the higher-grade area to thesouth of the M. plutons has probably equilibrated to thethermal events associated with the emplacement of thisyounger Sebago batholith, thereby preserving only post-M. effects. The sillimanite-K-feldspar isograd (Fig. l)shows a general spatial relation to the Sebago contact.Much of this area is highly migmatitic and pegmatitic,and the age of these igneous rocks may well correlate withthat of the Sebago batholith. Lux and Guidotti (1985) andGuidotti et al. (1986) have suggested that the effects ofthe Sebago event may extend as far as 30 km to the northand certainly are responsible for the sillimanite-K feld-spar isograd. One possible shortcoming of this interpre-tation is the absence of an obvious "hinge" effect (Gui-dotti, 1970), a retrograded zone of M, rocks where lowertemperatures from the Sebago event might have down-graded the rocks far north of the pluton. However, Gui-dotti et al. (1986) have noted some downgrading in theRumford quadrangle. We designate this Hercynian higher-P, higher-T event M, and suggest that those rocks southof the second sillimanite isograd (grades 6.5 and 7) suf-fered its effects. It should be pointed out that M, and M,were originally mapped as a single metamorphic se-quence (e.9., Holdaway et al., 1982). The principal rea-sons for separating M, and M, are the age differencesbetween plutons, the spatial association with the plutons,and important differences in lt and P as discussed below.

The rocks under consideration for this report have suf-fered M. and/or M, depending on their grade and maplocation. M, is discussed in more detail by Holdaway etal. (1986a) for a four-quadrangle area in the Kingfieldregion, by Dickerson and Holdaway (ms.) for the Bing-ham area, north ofthe Kingfield and Anson quadrangles,and by Dutrow and Holdaway (in prep.) for the Far-mlngton area.

ANar-vrrca.l METHoDS

Forthe area of M, and M, in Figure 1,40 polished sectionsof low-variance rocks were prepared for microprobe work. Lo-cations are shown by numbers in Figure l. Microprobe analyseswere done on an automated rcol-rr: microprobe at SMU withKRTsEL automation that uses the conection procedure of Benceand Albee (1968) modified by Albee and Ray (1970). Beam cur-rent was 20 nA, accelerating potential was 15 kV, and beamdiameter was focused to -2 prn for staurolite, garnet, and il-menite and to l0 pm for phyllosilicates and plagioclase. Stan-dards included a number of analyzed silicates and oxides. Foreach mineral, five or six analyses were taken from one portionof the thin section as near as possible to the gamet being ana-lyzed. Analyses were for 30 s or 60000 counts. For staurolite,all major elements were renormalized to standards to minimizedrift, and Al and Fe values were multiplied by constants to bringthe analyses into agreement with previous good-quality micro-probe and wet-chemical analyses, a procedure described byHoldaway et al. (1986b). Ilmenite (Fe, Ti, Mn) and plagioclase(Na, Ca) analyses were also renormalized to standards to mini-mize drift. For garnet, a regularly spaced traverse across one or

two grains was used. The core and intermediate-zone analyseswere grouped and averaged on the basis of chemical similarity.Garnet-biotite geothermometry indicated that all garnet rim

analyses (- 5 pm from rim) showed retrograde temperatures, andthese values are not given here. Biotite analyses were taken from

several grains within 30 to 100 pm ofthe gamet being analyzed.Each biotite was analyzed immediately before or after the co-existing garnet using the same analysis file.

Presence of graphite and/or nearly pure ilmenite (=4o/o he-matite solid solution) in every rock indicates that Fer* should beminor and approximately constant in most of the minerals. Con-sistent with the assumption of low Fe3*, mineral stoichiometrieswere determined by normalizing to the indicated number of ions:mtcas-22 oxygens, chlorite- 28 oxygens, feldspars- 8 oxygens,garnet-8 cations, and ilmenite-2 cations. For staurolite, stoi-chiometry was determined by normalizing (Si + Al) to 25.53 as

discussed below.Ion-microprobe analyses were done on two ilmenite speci-

mens (see Table 8) using an AEI IM-20 ion microprobe at the

University of Chicago. The instrumental configuration was de-scribed by Steele et al. (1981), and the analytical configurationswere similar to those described by Jones and Smith (198a). Theion yields were based on Corning borosilicate glasses, except Li(staurolite), Na (plagioclase), Sc (hibonite), and Nb (assumed tobe the same as Zr). Analyses were made by scanning from mass100 to mass 7, counting for 1 s at each mass. Elements not listedin Table 8 were not detected. These semiquantitative analyseswill be affected by matrix-dependent factors and are estimatedto be between 0.5 and 2 times the correct values.

In order to save space but provide the maximum amount ofdata, the major-element analyses are presented in Appendix Ta-ble I as stoichiometric analyses only, plus the oxide total. Anal-yses are listed in order ofincreasing grade designation (see above).Within each grade, analyses are given in order of increasing tem-perature using a slightly modified Ganguly-Saxena (1984) bio-tite-garnet geothermometer (discussed in the section on geother-

mometry). In the text, tables are provided to illustrate the averagecompositions of minerals with increasing grade.

MrNnru.r, cHEMISTRY AND PETRoLocIC

INTERPRETATION

Petrologic interpretation of the present study area isprovided in a paper by Holdaway et al. (1982). This sec-tion adds aspects to their interpretation that have becomepossible as a result of detailed mineral chemistry.

To a very good approximation, the metamorphic rocksstudied within a given metamorphic event (M. or M')may be regarded as a simple sequence of mineral assem-blages that have similar bulk compositions,/o, controlledat a reducing value, Xpr6 controlled over a nalrow range,and Z (metamorphic grade) as the principal intensivevariable. The thin sections chosen for study were selectedon the basis of minimum variance, and each grade des-ignation (given above) represents a single assemblage.Thus, changes in mineral assemblages and zoning aremainly a response to changes in Z. In support of thissimplified view are (l) the thermal (limited time) natureof the metamorphism, (2) the prevalence of graphite andnearly pure ilmenite, and (3) the fact that, with a fewnoted exceptions, the approach works. The section on

24 HOLDAWAY ET AL.: METAMORPHISM IN WEST.CENTRAL MAINE

TneLe 1. Summary of average muscovite stoichiometry by grade, based on 22 oxygens

5l.Al

'rAlTiFE

Mg>(vi)

KNaBa

>(xii)

Ba-msTi-msphl-annceprlpgms

6.05(2)1.e5(2)

3.74(210.08(1)0.1 2(1 )0 . 1 1 0 )4.050)

1.71(s)0.21(5)0.01(1)1.93(3)

0.0050.0400.025

0.0330.105o.792

0.0050.0550.025

0.0250.0450.845

6.07(3) 6.0s(3)1.s3(3) 1 .s5(3)

3.78(8) 3.82(4)0.03(1) 0.03(1)0.13(4) 0.10(1)0.13(8) 0.09(2)4.07(4) 4.04(1)

1 .61(2) 1 .56(s)0.2s(6) 0.35(e)0.01(0) 0.01(1)1 .s1(3) 1 .93(2)

0.005 0.0050.015 0.0150.035 0.0200.010 0.0200 038 0.0230.145 0j750.752 0.753

6.04(3) 6.04(2)1 .96(3) 1.96(2)

3.80(1) 3.76(4)0.05(1) 0.06(1)0.10(1) 0.11(2)0.10(1) 0.11(2)4.04(1) 4.04(1)

1.se(s) 1.67(8)0.34(6) 0.26(7)0.01(0) 0.01(0)1.s5(2) 1.ss(2)

0.005 0.0050.025 0.0300.020 0.0200.015 0.0200.01 8 0.0150.170 0.1300.747 0.780

6.07(1) 6.041 .97(1) 1 .96

3.71(2) 3.690.08(2) 0.110.13(1) 0 .120.12(1) 0 .134.04(1) 4.05

1.83(3) 1.850.12(2') 0.090.01(0) 0.011.95(1) 1.9s

0.0050.0400.0200.0250.0200.0600.830

Nofe.-Grade 3 is represented by two specimens and grade 8 by one. Value in parentheses is onestandard deviation. See text for grade assemblages. Abbreviations (Kretz, 1983): Ba-ms-"Ba-muscovite"; Ti-ms-"Ti-muscovite"; KTi(Fe,Mg)AlSi"O,lOH)"; phl-ann-phlogopite-annite; ce-cel-adonite; KAI(Fe,Mg)SinO,o(OH),; prl-pyrophyllite: pg-paragonite; ms-mus@vite.

metamorphic conditions demonstrates that these as-sumptions are reasonable.

Muscovite

Analyses for muscovite are provided in Appendix Ta-ble I and summarized by zone in Table 1. All specimenscontain <0.001 Ca and <0.004 Mn ions. Based on anal-yses of 6 specimens over a range of grade, muscovitecontains between 0.02 and 0. l3 wto/o F, corresponding toan average 1.2 + lo/o substitution of OH by F. Thesevalues are comparable to those of Evans (1969).

In both muscovite and biotite, all substitutions involveeither (l) octahedral sites, (2) octahedral and tetrahedralsites, or (3) tetrahedral and l2-fold sites. Nothing in theanalyses suggests the need for any coupled substitutionsinvolving a combination of octahedral and l2-fold sites.The muscovite substitutions can be viewed using the con-cept of exchange components (Thompson, 1982) withmuscovite (KAlrAlSi3Or(OH)r) as the additive compo-nent.

The octahedral site occupancy of all the muscovites isremarkably constant averaging 4.044 + 0.012 calculatedon a 22-oxygen basis. This suggests that (l) any substi-tutions involving octahedral vacancies must be nearlyconstant in amount for all the muscovites and (2) thereis a limited and constant amount of trioctahedral substi-tution. Most studies of muscovite have shown excessoctahedral ions (Guidotti, 1984), and in the present mus-covite this takes the form of about 2 molo/o phlogopite-annite. The substitution involving Ti, since it is variablein extent, cannot involve a change in vacancy content.Holdaway (1980) showed that for pelitic muscovite co-existing with ilmenite and an Al-rich mineral, R2* in-creases and "iAl decreases with increasing Ti, with slopesof 0.98 and, -2.02, respectively, consistent with the sub-

stitution TiR'z*Al ,. These same relations can be seen rnthe present data (Table l) if zones 3 (2 specimens) and 8(l specimen) are ignored.

The average end-member muscovite composition ofeach zone was calculated from the 22-oxygen stoichi-ometry by first dividing by 2 and then using the followingprocedure: (l) "Ba-muscovite"' : Ba1' (2) "Ti-musco-vite,"' KTi(Fe,Mg)AlSi.O,o(OH), : Ti; (3) phlogopite-annire : [>(vi) - 2]; (4) celadonite (ce) : {Fe + Mg -

Ti - 3[>(vi) -2]]; (5) pyrophyllite is the average of [l -

>(xii)l and (Si - ce + Ba - 3) (these difer by between 0and2.5 mol7o); (6) paragonite : Na; (7) muscovite is theremainder. The dominant substitutions in these musco-vites are TiR'z*Al-, for "Ti-muscovite," Rr2+Al ,tr-, forphlogopite-annite, R,*SiAl , for celadonite, trSiIL,Al_,for pyrophyllite, and NaK-, for paragonite.

It should be noted that the celadonite values near 2molo/o are partly a function of the other end membersassumed and the sequence in which they were calculated,and partly a function of the standards used for analysis.Because of the inaccuracy of estimating octahedral ionsabove 2, phlogopite-annite and celadonite contents musthave large errors relative to the other components. How-ever, low celadonite contents are reasonable for the pres-sures at which these muscovites crystallized.

The effects of grade on composition of muscovite arevery subtle and approximately the same as those notedby Guidotti for muscovite in comparable assemblages(1978;1984,p. 407) for the Rangeley area, Maine. Brieflysummarized, they are (l) Ti and R2* increase with grade,(2) "'et decreases with grade, and (3) paragonite content

t "Ba-muscovite," "Ba-biotite," "Ti-muscovite," and "Ti-bio-tite" are given in quotation marks to emphasize that they arehypothetical end-member components and not mineral names.

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRALMAINE

Tnele 2. Summary of average biotite stoichiometry by grade, based on 22 oxygens

G r a d e : 3 4 5 6 6 . 5 7 8

25

SitoAl

'rAlTiFeMgMn

>(vi)

5.42(0) 5.38(4)2.58(0) 2.62(4)

0.84(4) 0.92(3)0.18(0) 0.17(212.51(39) 2.52(30)2.23(421 2.17(3110.02(1) 0.01(1)5.78(1) 5.7s(5)'t.74(4) 1 .68(7)0.04(2) o.o7l2)1.78(s) 1.76(7)

0.53(9) 0.54(6)

0.003 0.0030.090 0.0850.110 0 j200.015 0.01 80.390 0.4240.020 0.0350.372 0.315

5.38(2) 5.37(1)2.62(21 2 63(1)

0.s0(2) 0.88(10)0.23(4) 0.25(5)2.53(15) 2.51(21)2.07(10) 2.Os(21)0.01(0) 0.01(1)5.73(3) 5.71(s)

1.71(21 1 .74(5)0.07(21 0.07(1)1 .78121 1 .81(5)

0.55(3) 0.s5(4)

0.003 0.0030.115 0 .1250.1 10 0.0900.018 0.0190.414 0.4020.035 0.0350.305 0.326

5.35(1) 5.34(2) s.362.65(1) 2.66(2) 2.64

0.78(4) 0.79(3) 0.860.33(4) 0.38(3) 0.332.s5(71 2.53(6) 2.331 .57(121 1 .85(10) 2.O50.02(0) 0.01(1) 0.015.65(4) 5.s7(4) 5.59

1 .80(5) 1 .90(3) 1.860.06(1) 0.03(1) 0.031.82(4) 1.94(2) 1.90

0.57(2) 0.58(2) 0.53

0.003 0.003 0.0030.165 0.190 0.1650.070 0.030 0.0500.004 0.021 0.0350-382 0.353 0.3600.030 0.015 0.0150.346 0.388 0.372

KNa

>(xii)-

Fe/(Fe + Mg)

Ba-btTi-bttlc-mimseas-sidwonphl-ann

Note: Abbreviat ions (Kretz, 1983): Ba-bt-"Ba-biot i te" ; T i -bt-"Ti -b iot i te" ;KTitr(Fe,Mg)AlSi"O,o(OH).; tlc.mi-talc-minnesotaite; ms-muscovite; eas-sid-eastonite-sidero-phyllite, KAI(Fe,Mg),Al,Si,O1o(OH),; won-wonesite; phl-ann-phlogopite-annite.

" Includes 0.005 Ba (see text).

first increases and then decreases as grade rises. Most ofthe changes in muscovite composition reflect the chang-ing saturation levels of "Ti-muscovite" and paragonitecontent. Several of the most subtle changes seen by Gui-dotti, e.g., decreasing Z(vi) and decreasing Si, are not seenin the present muscovites. Diferences in average Si con-tent between the two suites clearly reflect diferences instandardization for analysis.

Biotite

Microprobe analyses for biotite are provided in Ap-pendix Table I and summarizedby zone in Table 2. Allspecimens contain <0.001 Ca ions. Ba in representativebiotite is about half the amount in coexisting muscovite.The Fe3* contents were not measured, but must be lowand nearly constant, and would produce a small amountof ferri-eastonite or oxy-biotite (Dymek, 1983). Analysesof six biotites over a range in grade indicated 0. I 3 to 0.4 Iwto/o F (about 4 times that of coexisting muscovite), cor-responding to an average substitution of 6.9 + 2.4o/o Ffor OH. Values of F for biotites analyzed by Evans (1969)are0.l2 to 0.50 wt0/0. Munoz and Ludington(1974, 1977)have shown experimentally that Mg-bearing biotite par-titions F to a much greater extent than coexisting mus-covrte.

In biotite the best selection for the additive componentis annite. As is the case with all metamorphic biotite (e.g.,Dymek, 1983; Guidotti, 1984), octahedral vacancies arecommon. A plot of >(vi) against Ti (Fig. 2) reveals thatfor the present biotites, most of which coexist with il-menite, the vacancies are mainly associated with Ti con-tent. The sum of )(vi) + Ti for all the biotites is 5.96 +0.04. Clearly most or all of the variation in vacancy con-tent is a linear function of Ti. The relevant substitution

is TiDR'*r, the most important Ti substitution proposedby Dymek (1983). (The remaining 0.04 units of vacancymay be explained by an average muscovite content of 2molo/0.) The end member associated with Ti substitutionis KTitr(Fe,Mg)AlSi,O,o(OH)r, the same as the "Ti-mus-covite" end member. However, in biotite the ions couldbe disordered over both the M(l) and M(2) sites, whereasin muscovite they are presumably all at M(2) sites.

The combination of l2-fold site vacancies and consis-tently lower values for (t"At - 2) than for "iAl indicatesthe presence of significant amounts of talc-minnesotaitecomponent in the biotite. The average end-member com-position ofeach zone was calculated from the 22-oxygenstoichiometry assuming 0.005 Ba ions, dividingby 2,andusing the following procedure: (l) "Ba-biotite" : Ba; (2)"Ti-biorite" : Tl (3) talc-minnesotaite : [1 - >(xii)];(a)muscovite : the average of [3 - >(vi) - Ti] and t/2lviAl -I'Al + 2(xii)l (these differ by < I mol0/o); (5) eastonite-siderophyllite : ("'At - 2ms); (6) wonesite : Na; (7)phlogopite-annite comprises the remainder. The domi-nant substitutions in these biotites are Ti!R1+, for "Ti-biotite," trSiK rAl-r for talc-minnesotaite, Al,trR1J formuscovite, AlrRlisi , for eastonite-siderophyllite, andNaK-, for wonesite. This procedure is comparable to thatused by Dymek (1983) except that Dymek's more oxi-dized rocks necessitated an attempt to calculate Fe3* val-ues, whereas the reasonable assumption of zero Fe3* con-tent for these biotites makes the procedure simpler andperhaps more accurate.

The effects of grade on composition of biotite are morenoticeable than those for muscovite and are essentiallythe same as those observed by Guidotti (1984, p. 446).In summary, the changes with grade are (l) slight de-crease in Si and corresponding increase in toAl, (2) de-

26 HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

TiFig,2. Plot of )(vi) vs. Ti in ions per 22-oxygen formula unit

for biotites. The average sum of)(vi) + Ti is 5.96 + 0.04. Theline is drawn through this intercept with a slope of - I indicatingthat most ofthe Ti substitution in biotite is charge-balanced byvacancies in the octahedral layer. Box is estimated ( I o) analyticalprecrsron.

crease in -A1, (3) a pronounced increase in Ti and cor-responding decrease in 2(vi), (4) an increase in )(xii), and(5) a slight increase in Fe/(Fe + MC). These changes withgrade primarily reflect a progressive increase in "Ti-bio-tite" and progressive decreases in talc-minnesotaite andeastonite-siderophyllite as $ade increases.

Several of the highest-grade rocks contain little or noilmenite. Biotite in grade 7 of the Maine rocks (Fig. l)contains an average of l9 molo/o "Ti-biotite." Biotite ana-lyzed by Tracy (1978) from ilmenite-bearing rocks of the

Quabbin Reservoir area, Massachusetts, also contains 190/o"Ti-biotite" on the average. It appears that despite ab-sence of ilmenite in some rocks, the biotite from Maineis essentially saturated with respect to Ti. Also the Quab-bin Reservoir rocks, which probably formed at somewhathigher P, do not appear to have formed at higher Z orthey would have had higher Ti in biotite coexisting withilmenite. The explanation may lie in the fact that high-grade rocks in both areas appear to involve vapor-absentpartial melting of muscovite-bearing pelites. Accordingto Kerrick (1972), this melting should occur at a nearlyconstant I over a P range of several kilobars.

The increase in average Fe/(Fe + Mg) followed by apossible decrease at the highest grades results from theinterplay of two factors: (l) The reactions summarized byHoldaway et al. (1982) for these rocks (Fig. 3) imply anincrease in Fe/(Fe + Mg) as Fe-rich staurolite reacts awayand then Fe-rich garnet partially reacts with muscovite,but at the highest grades (7, 8) biotite starts to react withsillimanite to grow more Fe-rich garnet and K-feldsparthus reducing Fe/(Fe + Mg) in biotite (Thompson, 1976).(2) As grade increases, biotite and garnet must move clos-er together in Fe/(Fe + Mg) ratio as a consequence of theZ effect on the garnet-biotite exchange reaction (Ferryand Spear, 1978).

Garnet

Stoichiometry of the cores and intermediate zones ofgarnets is provided in Appendix Table I and summarizedby grade in Table 3. Compositional changes of garnetswith grade are as follows: (a) almandine and pyrope con-tents first increase while staurolite is present (grades 4,5), then decrease as garnet begins to react (grades 6,6.5),and then increase again with the onset of new growth(grades 7, 8); (b) spessartine content behaves opposite toalmandine and pyrope; (c) grossular content remains moreor less constant at about 4 molo/o; (d) Fel(Fe + Mg) de-creases with increasing grade. Zoning in garnet is char-acterized (a) by a slight outward decrease in Fe/(Fe +

TneLe 3. Summary of average garnet stoichiometry by grade

almprpspsgrsFe/(Fe + Mg)

atmprpspsgrsFe/(Fe + Mg)r

0.60(7) 0.71(8)0.06(2) 0.0e(2)0.2s(6) 0.15(7)0.04(1) 0.05(2)0.e0(4) 0.89(2)

0.68(12) 0.73(7',)0.08(1) 0.09(2)0.1s(12) 0.13(6)0.04(1) 0.0s(2)0.89(3) 0.89(2)

pro. pro.

Core0.74(4) 0.68(5)0.10(1) 0.10(2)0.12(4) 0.17(3)0.04(2) 0.04(210.88(1) 0.87(3)

Intermediate zone0.75(4) 0.70(s)0.10(1) 0.10(2)0.12(41 0.16(4)0.04(2) 0.04(2)0.88(2) 0.88(2)retro. retro.

0.70(1) 0.74(4) 0.700.11(1) 0 .12(1) 0 .180.16(2) 0.10(s) 0.060.03(1) 0.03(2) 0.060.86(1) 0.86(1) 0.80

0.71(1) 0.74(4) 0.720.1 1(1 ) 0 .1 1 (1) 0 .1 60.15(2) 0.1 1(4) 0.060.03(1) 0.03(2) 0.060.87(1) 0.87(0) 082retro. retro. retro.

/Vote: Abbreviations (Kretz, 1983): alm-almandine; prp-pyrope; sps-spessartine; grs-gros-sular.

. Based on average gamet-biotite temperatures (Table 8), intermediate zones are designated pro-gressive or retrogressive.

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

Temperature

Fig. 3. Tentative P-7 diagram showing divariant reactions involving garnet in pelitic rocks. Lower curves are Fe end-memberboundaries, and upper curves are Mg end-member boundaries. The three AFM model univariant curyes are also shown as narrowdivariant fields. Numbers represent grade designations for M, and Ms, +Mn indicates increase or decrease in Mn from core tointermediate zone ofgarnet, and arrow indicates increase or decrease in 1'from core to intermediate zone. The signs of slopes areconsistent with observations on M. and M, metamorphism. At high or low 4 signs of slopes may change (e.9., Spear and Selverstone,1983). Within the wide fields, the divariant reaction is the only reaction involving Fe-Mg components. The tentative conclusionthat prograde chlorite disappears early in the staurolite zone is discussed in the section on chlorite mineral chemistry and petrologic

interpretation.

27

EJ6o

E(L

Mg) at garnet and staurolite grades (3, 4) and a slightoutward increase in Fe/(Fe + Mg) in all successive gradesand (b) by spessartine content decreasing outward ingrades 3 and 4, remaining nearly constant in grade 5,decreasing outward in grades 6 and 6.5, and increasingoutward in grades 7 and 8.

Prograde changes in garnet composition. The steadychange in average Fe/Mg ratio of garnet with grade canbe seen in the cores, the intermediate zones, and even inthe peak metamorphic portion of the garnet (the inter-mediate zone for grades 3 and 4 and the core for highergrades). With increasing Z, the garnet moves composi-tionally closer to the biotite (K" approaches unity) as aconsequence ofthe Ieffect on the garnet-biotite exchangereaction (Ferry and Spear, 1978).

The progressive changes in average Mn content are ex-plained in Figure 3. Because garnet is the only Mn-richphase in the rocks, Mn content of garnet is an especiallygood monitor of growth or reaction. Garnet decreases inMn as it grows at garnet and staurolite grades (3, 4). Whengarnet reaches staurolite-sillimanite grade (5), it hasreached a minimum Mn content relative to adjacentgrades 4 and 6. Once staurolite disappears, Mn contentincreases in grade 6 as garnet and muscovite react. The

lower average Mn content of grades 6.5 and 7 is incon-sistent with the garnet-muscovite reaction, which shouldincrease Mn. In the narrow T range of zone 7, a simpleZ increase should leave the average garnet more or lessunchanged while muscovite and K-feldspar adjust theirNa content. The lower Mn content of grades 6.5 and 7 isa problem that can be most easily explained by a higherP of crystallization for these rocks (Fig. 3) than for grades3 to 6. The rocks of grades 6.5 and 7 are not only thesouthernmost rocks of the region, but they were meta-morphosed during the younger M, event. A single rockin zone 8 has still lower Mn in the garnet, but this canbe explained in part by the reaction ofbiotite and silli-manite to garnet and K-feldspar in a rock where most ofthe muscovite has reacted away.

Garnet zoning. Most garnets are zoned continuouslyfrom core to rim. The compositional behavior of zonedgarnets (Table 3) can be explained with the aid of Figure3 and the garnet-biotite temperatures (details in a sub-sequent section; see Table 8). Garnet-biotite geother-mometry demonstrates that at garnet and staurolite grade(3, 4), garnet zoning is mainly of a prograde nature, re-flecting a temperature increase, whereas at higher grades,the zoning is dominantly retrograde with the intermedi-

u.e9D9#

28 HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

TeeLe 4. Summary of average staurolite and chlorite stoichiometry by grade

Staurolite

Si"Al'rAlTiFeMgMnZnLi'H> ( R 4 + L i + H ) .Fe/(Fe + Mg)

7.62(6) 7.64(11)0.38(6) 0.36(11)

17.s3(1) 17.53(4)0 .10(1) 0 .11(1)3.31(18) 3.09(37)0.60(10) 0.s3(e)0.06(4) 0.07(210.0s(8) 0.18(14)t0.311 [0.s1][2.e71 t3.04]5.86(4) s.s1(17)0.8s(2) 0.86(1)

D ItoAl

'rAl

Tihe

M9MnCa4vi)Fe/(Fe + Mg)

5.19(5) 5.14(2)2.81(s) 2.86(2)2.92(61 2.93(3)0.01(0) 0.01(1)4.46(81) 4.60(62)4.48(83) 4.36(65)0.04(3) 0.02(2)0 00(0) 0.00(0)

11.90(0) 11.93(2)0.050(9) 0.s1(7)

Note: See text for calculation of stoichiometry of staurolite. Error in "'Al is artificially low becauseof this procedure. Brackets indicate that these entries are based mainly on estimated values.

. For analyses for which Li2O was not determined, Li is estimated as 0.3 atoms (see App. Table 1).

ate zone of the garnet indicating a lower temperature thanthe core. A decrease in Mn content of the garnet fromcore to intermediate zone indicates that the garnet wasgrowing (growth zoning), and an increase in Mn contentindicates that the garnet underwent late-stage resorptionor diffusion zoning (Tracy, 1982).

Growth zoning is indicated in grades 3 and 4 whereincreasing I coincides with decreasing Mn. Garnet grewflrst in response to reaction of chlorite to garnet in thegarnet zone and then as staurolite-biotite reacted to al-mandine-muscovite in the staurolite zone. The consistentdecrease in Mn in zoned garnets of the staurolite zone issupporting evidence that the staurolite-biotite to garnet-muscovite equilibrium has a negative slope (see alsoHoldaway et al., 1982). This reaction was probably notaffected by traces of chlorite in several staurolite-graderocks, because the chlorite is believed to be retrograde,as discussed below. At staurolite-sillimanite grade (5), theconstancy of Mn indicates that garnet was neither beingcreated nor destroyed, but the intermediate zones ofgar-net generally equilibrated at lower temperatures than thecores (see Table 8)., This diffusion zoning must have beenproduced in a retrograde fashion with biotite and silli-manite tending to react to garnet, but with slight stauro-lite production consuming garnet (Fig. 3). In grades 6 and6.5, decrease of Mn with decreasing I in the garnets in-dicates a retrograde growth zoning (Cygan and Lasaga,1982). Retrograde reaction of biotite and sillimanite togarnet and muscovite accounts for the Mn decrease. Ingrades 7 and 8, increase of Mn with decreasing 7 (diffu-sion zoning) results from decrease in garnet content as itreacts with K-feldspar during the early stages of cooling.

'? Retrograde zoning in grade 5 is partly an artifact of the factthat biotite used for core and intermediate-zone I calculation isthe same final biotite. For grades 3 to 6, Fe/(Fe + Mg) in biotiteincreases 0.02 molo/o/K (Tables 2, 8). Less than half of the garnetre-equilibrated with biotite during the (average) 15 .C retrogres-sion. Thus, biotite changed about -0.15 mol0/0, indicating thatactlual T values for grade 5 cores were about 2 'C lower thanmeasured (Ferry and Spear, 1978).

Staurolite

Of the 16 staurolite specimens analyzed in this study(App. Table l, App. l, and Table 4), 4 werc analyzed forH,O (Holdaway et al., 1986c) and 3 were analyzed forLi,O (Dutrow et al., 1986). Holdaway et al. (1986b)showed that, for a larger group of complete stauroliteanalyses, (Si + Al) was nearly constant on a 48-oxygenbasis: for 3l staurolites, (Si + Al) :25.43 + 0.13;for 28of these 3l staurolites not containing high Fe3*, (Si +Al): 25.46 I 0.09; and for 6 Maine specimens that allformed under reducing conditions with graphite, (Si +Al) : 25.53 + 0.03. In order to produce a stoichiometrythat is both accurate and consistent with those specimensanalyzed for water, we normalized the stoichiometry ofthe 12 specimens not analyzed for HrO and LirO to 25.53(Si + Al). The total positive charge was then subtractedfrom 96 to give an approximation of the (H + Li) ions.Through a propagation of errors it was determined thatthe approximate (H * Li) content has an error (lo) of+0.3 ions. Based on the Li analyses of five specimensfrom Maine, the average number of Li ions for the re-maining staurolites can be estimated at 0.3, allowing forreasonable estimates of H ions that agree well with thoseanalyzed for water.

The overall stoichiometry of the staurolite agrees verywell with that given for the 3l staurolites by Holdawayet al. (1986b). Mn and Ti are consistently low, whereasZn and Li are somewhat variable. This variability be-comes more apparent in the specimens from grade 5 wherestaurolite was breaking down by reaction with muscoviteand quartz to sillimanite, garnet, and biotite (e.g., Gui-dotti, 1974). The most unusual staurolite is specimen 6from near East Winthrop (Fig. 1). This staurolite formsabout lolo of the rock but occurs in small euhedral crystalsrather than as remnants in mica pseudomorphs, the morecommon occurrence of the mineral as it reacts out (Hold-away et al., 1982; Guidotti, 1968). It may have been sta-bilized by the - I Li ion per 48 oxygen (see also Dutrowet a l . , 1986).

Based in part on the structure determination of Smith

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE 29

(1968), Holdaway et al. (1986b) suggested that 0.25 Fe3"ions be assigned to staurolite Al octahedral sites, 0.25(Mn + Fe'z*) be assigned to U sites, and the remainingFe, Mg, Zn,Li, Ti, and vacancies be assigned to the fourtetrahedral sites, allowing for the determination of an ap-proximate activity model. Because the present staurolitesgrew under more reducing conditions than the averagestaurolites used for HrO analyses and structure determi-nations, it is reasonable to assume that, for the presentstaurolite, about 0.15 rather than 0.25 Fe3t replace oc-tahedral Al and the remaining Fe is Fe'?*. This value isapproximately equal to the lowest Fe3* determinations ofJuurinen (1956).

Values of Fe/(Fe + Mg) for staurolites range from 0.81to 0.89. Staurolites show no zoning and no consistentpattern of compositional change with grade. If the Li-richspecimen is discarded, the other four staurolites fromgrade 5 range from 0.849 to 0.853, averaging 0.851 +0.002, suggesting that the staurolite breakdown approxi-mates a univariant reaction.

As shown in Figure 4, the staurolite-biotite Ko is ap-parently unaffected by grade whereas the average stau-rolite-garnet Ko appears to increase slightly from grade 4to grade 5. Specimen 6 exhibits anomalous Ko values,i.e., the staurolite Fe/(Fe + Mg) is about 7 molo/o moreFe-rich than would be indicated by the biotite and garnetcompositions in the specimen. Apparently the effect ofhigh Li on staurolite is to impose a positive deviationfrom ideality on the Fe-Mg solid solution or increase pos-sible Fe-Mg immiscibility. Holdaway et al. (1986b) sug-gested that for a given P and T, there are limits on theamount of smaller ions (Zn, Mg, Li, and Ti) that thetetrahedral Fe sites can hold. An increase in Li must de-crease the maximum possible amount of Mg, other thingsbeing equal. This phenomenon is also shown by increas-ing Li with increasing Fe/Mg in staurolite (Dutrow et al.,1986, Fig. 4).

In both the biotite and garnet Ko plots there is a ten-dency for Mg-rich staurolite to exhibit a slightly higherKD than more Fe-rich staurolite. As Mg content increases,the Fe/(Fe + Mg) of staurolite is slightly higher than wouldbe predicted from ideal solid solution, indicating positivedeviation from ideality in staurolite (assuming that thebiotite and garnet Fe-Mg solid solutions are close to ide-al). If some of the Fe is subtracted from staurolite Fevalues to account for the minor U-site and octahedralAl-site occupancy as discussed above, the absolute Kovalues are changed somewhat, but the relationships re-main the same. Zn and Li contents of most of the stau-rolites are low enough to have no measurable effect onKD.

Chlorite

Chlorite is an important mineral in garnet-grade (3)rocks, but forms only <2o/o of 8 of 11 analyzed staurolite-erade (4) rocks. Only well-crystallized chlorite, not seenreplacing biotite, garnet, or staurolite, was chosen foranalysis (App. Table l; Table 4). One exception to this is

4.71t0./t8

o 6,

= 2o

o

66

1

L

o

6

0,6{x,x.o.ffi

5

o.76h0.69

Staurolite FelMg

Fig. 4. Ko Fe-Mg plots for staurolite-biotite and staurolite-garnet (using peak metamorphic zone of garnet). Dots representgrade 4 and circles represent grade 5. Specimen 6, Li-rich stau-rolite, is identified, and plots at garnet Fe/Mg : 4.07. Lines giveaverage Ko for staurolite-biotite and average K" by grade (des-ignated as 4 or 5) for staurolite-gamet, excluding specimen 6.Average Ko and standard deviation are given.

in specimen 94 where chlorite discontinuously surroundsand partly replaces staurolite. Simple stoichiometry in-volving Si, Al, Mg, Fe, and minor Mn characterizes thechlorites. The numbers of Al and Si ions are very nearlyconstant for Al-saturated rock compositions and resem-ble those reported by Novak and Holdaway (1981), Gui-dotti (1974), and Dutrow (1985) for chlorites ofgraphite-bearing pelites from Maine. The present chlorites andthose of Novak and Holdaway (198 l), using a differentmicroprobe and standards, show 0.086 + 0.041 higher"iAl than toAl and slightly low )(vi). To a lesser deg,ree,the same tendency is seen in the data of Guidotti (1974).These observations cannot be explained by Fe3* substi-tuting for "iAl or by intercalated biotite, talc, or wonesitelayers (Veblen, 1983) because all these effects tend to in-crease octahedral R3* or increase Si, which would alsoincrease apparent octahedral Al.

Three possible explanations for these observations aresuggested. (l) They may result from a calibration error inmicroprobe analyses of about lolo relative for SiO, or

JU HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

o 2

L

Io

1

> r uoL

3 g

O.ft9!o.Oln

0.122!O.Ot1

1 1 5 2Chlorite FelMg

Fig. 5. Ko Fe-Mg plots for chlorite-biotite and chlorite-garnet(using peak metamorphic zone ofgarnet). Circles represent grade3 and dots represent grade 4. Lines represent average Ko. AverageK" and standard deviation are given.

AlrO.. Although correcting for such an error would bring"iAl and *Al into agreement, it would lower )(vi) stillmore. (2) The total number of anions per formula unitmay not be stoichiometric. Specifically, the chlorite maycontain a few percent ofintercalated brucite layers (Veb-len, 1983, p. 5'74). To allow for extra brucite layers, theoxygen basis needs to be increased slightly from 28, whichwould also increase >(vi). (3) The chlorite may containvery limited dioctahedral substitution (Eggleston andBailey, 1967; Tompkins et al., 1984). In the absence ofrnu work it is impossible to distinguish between theselast two possibilities.

The main compositional variation in chlorite is in FelMg ratio. Garnet-chlorite K, varies with grade, whereasbiotite-chlorite Ko appears to be constant (Fig. 5). Therecently calibrated garnet-chlorite geothermometer(Dickenson and Hewitt, 1986) gives temperatures thataverage about 5 "C below those of the Ferry and Spear( I 978) garnet-biotite geothermometer (Table 5), suggest-ing that garnet-chlorite K" gives a reasonable estimate ofprograde Z in these rocks.

The AFM univariant assemblage staurolite-garnet-biotite-chlorite-muscovite-quartz occurred over a I interval of about50 'C. In rocks such as these, in which P, Xr,o, and /o, werenearly constant, one might expect significant variations in non-AFM components in one or more minerals to allow for thisassemblage to exist over such a 1'range. Such variations are notseen, and the variations that do occur are random in nature oreasily explained by other reactions (see App. Table 1). A numberofarguments based on the present work and observations else-where suggest that many of the chlorites in the rocks of grade 4are retrograde and result from minor reaction ofbiotite and oth-er minerals to chlorite: (1) Biotite coexisting with staurolite andprograde chlorite should have a restricted range ofFe/(Fe + Mg)that should be lower than that ratio in rocks containing stauro-

TneLe 5. Comparison of garnet-chlorite and garnet-biotite temperatures; calculatedequilibrium constants for Reaction 1

Temperature ("C)

ln ,CctSpeci-

Grade men

3344444444

Average

467443470499498481497513544534495 + 31

0.3660.2140.5030.5260.4330.2520.3780.4690.3780.3020.382 + 0.104

2vb523

6184

1291149427

470 477461 460480 494484 499495 494484 481483 486507 501537 544541 546494 + 27 498 + 27

4614424554U500484494520537527490 + 32

Note. Sequence is that of increasing fgiven in Appendix Table 1, based on Table 8. Abbreviations:C : core; | : intermediate zone.

. Approximate equilibrium constant; values of In ,C for Reaction 1 using activity models given intext.

.. Temperature based on garnet-biotite compositions and Dickenson and Hewitt (1986) geother-mometer modified by M. P. Dickenson (pers. comm., 1987).

t Temperature based on garnet-biotite geothermometry (Table 8). Ferry and Spear (1 978) calibra-tion was used because the garnet-chlorite geothermometer was calibrated with Ferry and Spear(1 978).

HOLDAWAY ET AL.: METAMORPHISM IN WEST.CENTRAL MAINE 3 l

lite-biotite without chlorite (e.g., see AFM diagram of Guidotti,1974). ln grade 4 biotites, Fe/(Fe + Mg) ranges from 0.47 to0.69 in chlorite-bearing rocks, and from 0.47 to 0.57 in rockswithout chlorite. Small differences in X"ro of the fluid phasecould account for some variability but are unlikely to producethe observed degree of randomness during prograde metamor-phism. (2) If chlorite and garnet were an equilibrium pair in thepresent rocks, then the Fe end-member equilibrium (Reaction l)

with 7 (Fig. 5). The composition of retrograde chlorite is con-trolled by the dominant biotite in the rock and the chlorite-biotite KD. In M, (Dickerson, 1984), retrograde chlorite is wide-spread, and almost all chlorite is clearly retrograde. Here thechlorite-biotite Ko is 0.858 + 0.040 compared with 0.869 t0.022 for the present rocks. The close compositional relationshipbetween chlorite and biotite is maintained despite the wide-spread retrogression in Mr.

0.250Fe0'Al ,o?(Alr43sir57)O10(OH)8+0.0l0KerAI, r rFeo,r(AlonrSiro3)Oro(OH),+0.458SiO,chlorite Muscovite Quartz

: 0. 367FerAl.Si3O,, + 0.0 I 1 Ko rrFer rrAlo o7(Alr 3 r Si2 6e)O ro(OH)r +Almandine Biotite

( 1 )HrOFluid

would apply to all rocks of grades 3 and 4 that contain chlorite.Approximate mineral equilibrium constants (Kx) were calculat-ed to take into account dilution of Fe and K by other compo-nents, assuming that Al varies little from the average amounts(Tables 1-4). Activity models based on mineral compositions inReaction I are as follows:

Chlorite ao".n, : [Fel(Fe + Mg + Mn)]+st

Muscovite a,," : [K./(K + Na)]0'g?[Fel(Fe + Mg)lo''Almandine a",- : (XFJ)'

Biotite a""", : [K./(K * Na)]o'at ><

lFe/(Fe + Mg + Mn + 2Ti;1:s:.

The calculated values of Kx are given in Table 5 in order ofincreasing 7 (based on Table 8). The relation between IC and Tfor an equilibrium reaction (Kerrick, 1974) is

AI: R[ln(X*a1,o)AS;'. (2)

Thus as 7 increases and AZ rises relative to the equilibrium Zcalculated for the end-member reaction (e.g., Fig. 3), ln K* mustincrease ifequilibrium prevails. It is clear that prograde 1"is notthe controlling factor for -Kx. Variation in Kx can result fromchanges in the HrO activity as shown by Equation 2, but inreduced graphitic rocks, such variation is not expected to exceed| 5o/o (e.g.,see Table I 2), leading to a maximum variation of 0. 1 6in ln K*. With the possible exception of the first four specimenslisted in Table 5, Kx for Reaction I does not support chlorite asan equilibrium prograde mineral in the staurolite zone. (3) Chlo-rite occurs as small amounts ofwell-crystallizedgrains ofvariablesize mainly cross-cutting the foliation defined by muscovite andsome of the biotite. It commonly occurs near porphyroblasts ofstaurolite or biotite. (4) Aluminum silicate and Mg-bearing bio-tite are a common lower staurolile-zone assemblage elsewhere,suggesting that the alternative staurolite-chlorite assemblage be-comes unstable early in the staurolite zone. Examples are M, inMaine where under reducing conditions, andalusite-biotite be-comes stable immediately after staurolite (Dickerson and Hol-daway, ms.); the Picuris Mountains in New Mexico where an-dalusite-biotite and sillimanite-biotite occur next to normal lowerstaurolite-grade rocks (Holdaway,1978); and the Snake Range,Nevada, where kyanite-biotite occurs above normal staurolite-grade rocks in a sequence ofdownward-increasinggrade (Geving,I 987).

One important question remains to be answered: If much ofthe chlorite of grade 4 is retrograde, why does it give reasonable?'values with the garnet-chlorite geothermometer (Table 5)? Theanswer apparently lies in the constancy of chlorite-biotite KD

Based on these observations, we tentatively concludethat much of the chlorite in grade 4 grew from biotiteand other minerals during cooling or M5 reheating of therocks at temperatures in the range of 450-500 'C andperhaps at pressures higher than those of Mr. Equilibriumretrograde domain assemblages must have been garnet-chlorite-staurolite-biotite, garnet-chlorite-staurolite, gar-net-chlorite-biotite, and staurolite-chlorite-biotite (all as-semblages containing muscovite * quartz). This conclu-sion suggests a re-evaluation of the first sillimanite-forming reaction proposed by Guidotti (1974) for theRangeley area on the western side of the present studyarea and by Holdaway et al. (1982) for the present area.Instead of staurolite-chlorite-muscovite reacting to silli-manite-biotite in normal M, rocks, we suggest that stau-rolite-muscovite reacted to sillimanite-biotite-garnet. Thechlorite-staurolite tie line in AFM projection must breakimmediately after staurolite forms, allowing garnet-chlo-rite-staurolite-biotite to exist over a limited I range, con-sistent with the high AS involved in chlorite dehydration.The staurolite-grade (4) M, rocks of Maine occupy narrowcompositional and pressure ranges such that neither alu-minum silicate3 nor chlorite occur as equilibrium pro-grade minerals. At lower P during Mr, staurolite was moreFe-rich and occurred with andalusite over a range ofgradeas indicated by field relations (Dickerson and Holdaway,ms.). At this lower P the common bulk compositions in-tersect the AFM fields of staurolite-biotite-garnet andstaurolite-biotite-andalusite.

Ilmenite

The only significant components detected in ilmenitewere Fe, Ti, and Mn (App. Table l; Table 6). Ilmenitecontent varies from 89 to 99o/o, pyrophanite (MnTiO)from I to 1.lo/o, and hematite from 0 to 4o/o. The onlyobvious compositional pattern is a strong tendency forthe pyrophanite content to mirror spessartine content ofthe garnet. The lower Mn in the ilmenite combined withmuch lower ilmenite content than garnet in the rocksindicates that only a few percent of the Mn in the rocksis contained in ilmenite and most remaining Mn occursin garnet.

3 But see Note added in proof.

5 Z HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRALMAINE

Tneue 6. Summary of average ilmenite composition by grade

0 s4(5) 0.s7(21 0.e6(2) 0.e4(4) 0.95(1)0.06(5) 0.03(2) 0.04(2) 0.0s(3) 0.04(2)0.00(0) 0.01(1 ) 0.00(0) 0.01(1) 0.01(1)

/Vote.' Abbreviations (Kretz, 1983): ilm-ilmenite; pyr-pyrophanite(MnTiO.); hem-hematite

Careful inspection ofpolished sections ofgrades 6.5 to8 shows that most of the opaque grains are graphite. Inseveral specimens, no ilmenite was detected; and in oth-ers, only trace amounts were seen (see App. Table l). Thisscarcity of ilmenite is not related to Fe/(Fe + Mg) ofbiotite, nor is rutile present in any of the specimens. Pre-sumably, low ilmenite content results from increased Tisolubility in biotite at high grades. The most hematite-rich ilmenite (4o/o in specimen 63) coexists with a traceof graphite. Although graphite is difrcult to recognize be-cause of fine grain size in some lower-grade specimens,its obvious presence in all high-grade specimens com-bined with the low hematite content of all the ilmeniteimplies that graphite is present in all specimens.

On the basis of 2 (Fe,Ti,Mn,Mg) cations, five ilmenitescontain more than l.0l Ti atoms. The theoretical limitof Ti ions in ilmenite is 1.00 since the larger octahedralsites presumably exclude Ti. Ti ions range from 1.013 to1.028, and oxide totals range from 97 .36 to 98.54010. Theremaining ilmenites have oxide totals (corrected for FerO,on basis of hematite content) ranging from 98.49 to100.91o/o. These results suggest that some element maynot have been analyzed in the anomalous specimens.Specimens 6 and 14, the two most extreme cases, wereexamined semiquantitatively with the ion microprobe.The results (Table 7) indicate that no other ion is presentin significant amounts. The high Li of the sraurolite inspecimen 6 is reflected by 167 ppmw of Li in the coex-isting ilmenite. A microprobe wavelength scan of ilmen-ite in specimen 6 revealed only Ti, Fe, and Mn. There-fore, high stoichiometric content of Ti is probably anartifact of the analysis procedure (small grain size, con-duction, thickness ofcarbon coat, etc.).

Other analyzed minerals

Plagioclase is commonly observed in rocks of grades 5and higher (see App. Table l) and is present in smallamounts at lower grades. It is Anrn in one specimen ofgrade 5, An,r-r, in grade 6, An,u-r, in grade 6.5, and An,r_r,in grades 7 and 8. The two most calcic compositions, Ano,and Anr., suggest partial removal of albite componentthrough reaction with muscovite (Tracy, 1978).

Orthoclase in grades 7 and 8 varies from Orro to Orno(see App. Table l). These feldspars are comparable incomposition to those of the southwest part of the studyarea analyzed by Guidotti et al. (1973), e.g., Or* to Ornoin grade 7 schists. Celsian content ofalkali feldspars av-erages about I molo/0.

Most sillimanite coarse enough for analysis is fromgrades 7 and 8. Analyses of sillimanite from six speci-

TABLE 7. Semiquanti tat ive ion-microprobeanalyses of two i lmenites fromgrade 5

Element Sample 14 Sample 6

o.c

ilmpyrnem ?Li

eBe

" E23Na

,nMg27 Al28Si3eK@Ca6Sc46Ti

52CrsFessMn68Srs3Nb

5.41 . 01 . 7

21281554804165

JU

[31.58%]100

1 836.23./"3.15%1

227

1670.20.6

521952824661703455

t31.58%l852037.83'/"3.67%

1 061

Note.' Analyses were normalized to Ti : 31.58%.Analyses are upper limits because background mea-surements were not made. True values are probablybetween 0.5 and 2 times the recorded values becauseof matrix-dependent effects. Unless specified as per-cent, values are given as parts per million (by weight).

mens show no more than trace amounts of Mg, Ti, andMn, and 0.19 to 0.35 wto/o FerOr. The average content ofFerO, is 0.27o/o corresponding to 0.005 Fe atoms per for-mula. This amount of Fe is of no consequence in calcu-lating reaction boundaries from experimental equilibria.

CoNnrrroxs oF MToTAMoRPHISM

Metamorphic conditions were deduced on the basis of(l) estimates of metamorphic temperatures by garnet-biotite geothermometry, (2) metamorphic pressure esti-mates by comparison of the andalusite-sillimanite phasediagram (Holdaway, l97l) with garnet-biotite geother-mometry (grades 5 and 6 of Mr), and (3) estimates ofpressure for grades 6.5 and 7 of M, based on calibrationof the muscovite-almandine-biotite-sillimanite (MABS)geobarometer using the results of (2) above. Finally, Pand estimates for grades 5 and 7 were compared withdehydration equilibria in order to test for accuracy andevaluate fluid-composition models.

Geothermometry

The only reasonably well calibrated geothermometerapplicable to medium- and high-grade pelitic metamor-phic rocks is that based on the garnet-biotite exchangereaction. This geothermometer is ideal for the M, and M,Maine rocks because the biotite/garnet ratio is high andevidence for extensive retrogressive effects is virtually ab-sent. Table 8 provides garnet-biotite temperatures forgarnet core or intermediate zones and surrounding biotitegrains not in physical contact with the garnet, using fourrecent quantitative calibrations: (l) Ferry and Spear(1978), (2) Hodges and Spear (1982), (3) Ganguly andSaxena (1984, 1985), and (4) Ganguly and Saxena (1984,1985) with AlZ,. reduced from 3000 + 500 to 2500 cal.

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

Tnele 8. Garnet-biotite temperatures ('C)

J J

Ferry-Spear Hodges-Spear Ganguly-Saxena Ganguly-Saxena'

cNo. Peak f

2YO

Avg.

1 Aaz3A6 1 A84--129*1145AA6**

9427

Avg.

1375314Ao

23Avg.

Tstf4A471 91125063308140

Avg.

77-377-290t o

coo l

Avg.

738714314596

Avg.

1 1

Avg.$Avg.$$

477460469(1 2)

4924944994944814865015315 1 8544546508(23)

5 1 4540507584572543(34)

494553548551530578c c /

5596 1 1588564(241

598606623605593627609(1 4)

601629621ooo656635(2s)

716

23.8

481462472(131

4654885135 1 44935 1 8537553528564544520(30)

557570543603614577(30)

50657459658s5896066066 1 5654637607(25)

643690686675718682(271

805

25.2

531525528(4)

522519523527540515520560554568589540(2s)

s30ccc

552605594567(31 )

54258657257957759960060662561559s(1 8)

6 1 1620645630620661631 (1 e)

637640631674u464s(17)

682

21.8

467443455(1 8)

4314704994984814975 1 3526502544534500(32)

543556537c / c

598562(25)

487541c / J

c / c

576595587595631625589(27)

624636632636636656637(1 1 )

630665675666712670(29)

779

25.822.5

Grade 3 (M3)490 535479 51548s(8) 525(14)

Grade 4 (M3)521 503509 513513 523510 534493 540509 534524 531cc/ ccv

545 546564 568555 583527(24) 539(24)

Grade 5 (M3)526 552554 568516 576611 600585 616s58(40) s82(261

Grade 6 (M3)514 538582 586568 595562 606538 614586 619576 626578 639635 646600 652581(27) 620(23)

Grade 6.5 (Ms)609 634617 6466316 1 6609

657660657

637 689620(12) 6s7(18)

Grade 7 (M5)614 660654 666633 668675 674662 682648(241 670(8)

Grade 8 (Ms)741 726

24.8 20.7

4794985085155215255285435345 J /

J / b

526(27)

5455595605856055711241

526566578589595607610616628632602(22)

619630637641641669640(1 7)

a0657658663675659(13)

719

21.720.8

506 506507 507507(1) 507(1)

508 508508 508508 508510 515521 521509 525517 528544 544545 5455 5 / C C /

5U 584528(26) 531(24)

522 545546 559537 560591 591583 60s5s6(30) s72(2slI

532 532568 568556 578566 589557 595589 607585 610584 616609 628598 632s7e(18) 603(22)

597 619605 630626 637613 641606 641643 66961s(17) 640(17)

616 640631 657620 658663 663636 675633(19) 659(13)+

675 719++

22.O

506494500(8)

635M7640647652ooo648(1 1 )

20.323.2

Note.'References: Ferry and Spear (1978); Hodges and Spear (1982); Ganguly and Saxena (1984, 1985). A nominal Pof 3500 bars was assumed."4" with sample number indicates presence of M, andalusite. Values in parentheses are one stanclard deviation.

- Ganguly and Saxena with AtyM" reduced from 3000 + 500 to 2500 cal. Peak f italicized and given in last column.-'These three specimens are probably Mr.t lncfusion of three specimens studied by Dutrow (1985) (237, 580'C; 330, 5929;267,613 rc) increases this value to 581(24) qC, presumably a

more accurate value because rof specimen 137 might be retrograde.ft Specimen has textural indications of retrogression; excluded from all averages and from geobarometry.+ This value may be reduced by a maximum of 7'C to compensate for change in biotite composition during early cooling (see t€xt).

f+ May be reduced by a maximum of 20 rc (see text).$ Weighted average of standard deviations, grades 4-7-

$$ Weighted average of peak temperature standard deviations, grades 4-7.

34 HOLDAWAY ET AL.: METAMORPHISM IN WEST.CENTRAL MAINE

Most calibrations after 1978 use Ferry-Spear 1(" data andempirically correct for nonideality, principally from Caand Mn in garnet. A nominal pressure of 3.5 kbar wasused for the calculations.

The quality of these calibrations as they relate to theMaine rocks may be evaluated by comparing the tem-peratures with temperature estimates based on dehydra-tion equilibria (discussed below) and by comparing thescatter of the data and the overlap between grades. Com-parison of the calibrations centers mainly on grades 5 and7 . Each of these zones may represent a relatively naffowZ range (assuming a limited range of XHrq) because eachinvolves a univariant reaction in the model KFMASHsystem. For grade 5 (staurolite breakdown), the modifiedGanguly-Saxena calibration involves the smallest stan-dard deviation, and the Hodges-Spear involves the larg-est (see also Dutrow, 1985). For grade 7 (muscovitebreakdown), the original Ganguly-Saxena calibration ap-pears best, and the Ferry-Spear appears worst. If all thestandard deviations ofgrades 4-7 are averaged, the orig-inal Ganguly-Saxena is slightly lower than the modifiedGanguly-Saxena (Table 8). This may be due to the factthat the Ganguly-Saxena corrections are partly based onbest-fit calculations of data similar to ours.

A partial test of the accuracy of the geothermometricmethods is provided by the two garnet-zone specimens.Stability considerations for staurolite-almandine (e.9.,Ganguly, 1 969) and fluid-composition considerations (e.g.,Holdaway, 1978) suggest that in graphite-bearing rocksthe staurolite isograd should occur at about 520'C at 3-4-kbar pressure. The garnet-zone specimens come fromfield positions very near the staurolite isograd (chloritoiddoes not occur in M.). The high Mn content (up to 28molo/o) of garnet-zone garnets causes the various calibra-tions to spread out over a wide Z range. The modifiedGanguly-Saxena calibration appears to provide the besttemperatures for this zone, while the original Ganguly-Saxena provides slightly higher temperatures that overlapwith several of the staurolite-zone temperatures. It ap-pears that the Ganguly-Saxena calibration wilh AW*,:3000 cal slightly overcompensates for nonideality of spes-sartine component.

Temperatures that estimate peak metamorphic condi-tions are listed in the last column of Table 8. In grades 3and 4, much ofthe zoning is ofa prograde nature, where-as in grades 5 through 8, diffusion zoning and retrogradegrowth zoning are prominent as indicated by intermedi-ate-zone temperatures that are lower than core tempera-tures.4 In grades 5 through 8, the intermediate-zone tem-peratures are more accurate than the core temperaturesbecause the adjacent biotite was equilibrated with inter-mediate-zone compositions last. However, temperatureestimates record an event during the cooling cycle at whichdiffusion between garnet and biotite was frozen in.

4 Note that although final biotite composition was used forcore and intermediate-zone garnet-biotite temperatures, the ?"efect on biotite composition is quantitatively unimportant asshown in footnote 2.

Once K-feldspar had formed in grades 7 and 8, garnetmay have reacted with K-feldspar during retrogrademetamorphism (Fig. 3) to make biotite more Fe-rich thanthat at equilibrium with the garnet cores. Thus the garnetcore temperatures in grades 7 and 8 may be higher thanequilibrium values, whereas the intermediate-zone tem-peratures are more accurate but are retrograde in nature.An estimate of the maximum extent of this efect may bemade by comparing the average difference between coreand intermediate-zone temperatures for grades 6.5 and7. In grade 6.5, LT is 25 oC, and in grade 7 it is 32 'C (ifthe unzoned specimen 145 is excluded). Thus, the grade7 rocks from the same region as grade 6.5 rocks have amaximum Z effect of about 7 oC because of the changein biotite composition. For the single grade 8 specimen,the maximum effect is 20 "C.

Peak temperatures from grade 5 (staurolite breakdown)average 572 + 25 "C. However, the temperature for spec-imen 137 (545 'C) may be retrograde. Addition of threespecimens from the Farmington quadrangle [Dutrow,1985; Table 8, grade 5 (Mr) averagel increases the averageto 581 + 24 "C, which may be a better average for grade 5.

Approximate temperatures are grade 3, < 510; gfade 4,510*560; grade 5, 560-595; grade 6, 595-625; grade 6'5,625-645; grade 7, 645-670; grade 8, >670 "C (Table 8).The largest deviation of any specimen from these rangesis 25 'C which is an approximate indication of (2o) pre-cision. This is in rough agreement with analyses of twodifferent garnet grains in each of three samples (5, 19,and 56) that average a 19'C difference and with analysesof two specimens from 4 m apart in the same outcrop('77-3,77-2, Table 8) that differ by I I "C. A further testof the accuracy of these temperatures is given in a sub-sequent section. It is important to note that grades 6 and6.5, which have the same mineral assemblages, show al-most no overlap in calculated temperatures. In our opin-ion, this is because they crystallized during different eventsat different pressures as discussed below.

Geobarometry

Three geobarometers have possible value in medium-to high-grade pelitic metamorphic rocks: the phengitegeobarometer (e.g., Powell and Evans, 1983); the garnet-aluminum silicate-plagioclase (GASP) geobarometer,most recently calibrated by Newton and Haselton (1981)

and improved by the garnet mixing models of Gangulyand Saxena (198a); and the muscovite-almandine-bio-tite-sillimanite geobarometer (MABS) approximately cal-ibrated by Spear and Selverstone (1983) and Robinson(1933). There are presently large uncertainties in thephengite geobarometer, especially for low-pressure rockswhere the phengite content of muscovite is small andhighly dependent on the method of calculating end mem-bers and activity models (see mineral-chemistry section).Also the method requires the presence of primary chlo-rite, and many of these rocks contain only chlorite inter-preted to be retrograde. The GASP geobarometer suffersfrom the low grossular contents of garnets and variable

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE 35

zoning of plagioclase compositions in these low-pressurerocks. Although in need offurther calibration and testing,the MABS geobarometer may become the best barometerfor pelitic rocks because all the components involved inthe reaction are major components in the minerals.

Pressure of zone 5. The first appearance of sillimanitein M, is by reaction of staurolite with quartz in grade 5.Garnet-biotite temperatures for this zone indicate thatsillimanite began to form at -560 "C (Table 8). M. isdefined as metamorphism in which the aluminum silicateproduced as a staurolite breakdown product was silli-manite. The aluminum silicate diagram of Holdaway(1971) shows that this must have occurred at a pressureabove -3.0 kbar. Because M. immediately followed Mr,and locally the transition from M, to M, is nearly contin-uous (Novak and Holdaway, l98l), it is reasonable toassume that average M, pressure was only slightly abovethis value. There is also an andalusite locality near Far-mington (Dutrow, 1985) that cannot be explained as ear-lier M., but may well represent a slight northward P de-crease during Mr. If the average M, pressure was 3.5 kbar,as determined by Ferry (1976) and generally accepted byother workers in the area, garnet-biotite temperaturesshow that some relict M, andalusites would have oc-curred in the sillimanite P-Z field, without nucleation ofsillimanite. For the purpose of calibrating the MABS geo-barometer we have assumed an average P for M, of 3.1 +0.25 kbar. Although the 3.l-kbar P estimate is adoptedas the average pressure for grades 5 and 6 and is reason-able for grades 3 and 4 (which show close spatial asso-ciation with grades 5 and 6), one cannot assume thatpressure remained constant in grades 6.5 through 8. Thesezones are exposed considerably south of most M3 rocksand suffered later M, effects (Fig. l).

MABS geobarometry. For grades 5 and 6, the presentanalyses and those of Dutrow (1985) in the southern 400/oof the Farmington quadrangle provide a uniquf oppor-tunity to calibrate the geobarometer involving the assem-blage muscovite-almandine-biotite-sillimanite (Fig. 3) thatoccurs in all rocks from grades 5 to 8. The ideal end-member reaction is

*Lo.lfl"'.9*(oH)' + t'l*l1l1f '': **{"1*.o,0(oH),* 1tlf"lR, * t'9;,' (3)

All analyzed M, specimens from grades 5 and 6 were usedwith the exception of one retrograded specimen each fromthe present analyses and from the Dutrow (l 985) analyses.All specimens were assumed to have crystallized at anaverage P of 3. I kbar and peak temperature as defined bythe modified Ganguly and Saxena (1984, 1985) garnet-biotite geothermometer (Table 8).

The equilibrium constantlKx was evaluated using thefollowing activity models to adjust actual mineral com-positions to ideal end-member compositionst ex^ :(XFI)'(rfI)'; a,, : (XXi)'XH"XX'?X'; a^"" : ffi!)3Xft,; a.,, :

l', aoo: l. Fe was assumed to be disordered on the M(l)and M(2) octahedral sites of biotite. Activity coeftcientswere calculated for almandine component following Gan-guly and Saxena (1984) with AW-^:2500 cal. The valueof (rFJ), varies from 0.90 to l.l l. Activity coefficients forK in muscovite were evaluated assuming that pyrophyllitecomponent behaved like muscovite component and ,yf"was based on paragonite component only. The equationof Chatterjee and Flux (1986) for muscovite was used.Biotite was assumed to be ideal. This is partly because ofthe absence of data and partly because the amounts ofmost of the other components were small enough to havenegligible effects on the major components. The possiblenonideality of Al in biotites tends to cancel out becausethe calibration was done on Al-rich biotites. Tetrahedralsites were assumed to be coupled to octahedral or l2-foldsrtes.

The value of AP corrected for observed mineral com-positions is given by

LP: -AV tRZIn Kx, (4)

where AIlis given in Table 9. A plot of M against Zwasused to determine the end-member P-T curve for Reac-tion 3. The resulting equation is

P: l5( r * 968) + AP, (5)

where P is in bars and ?" is in kelvins.Calculated pressures for the rocks from grades 5 and 6

(MJ from Dutrow ( I 98 5) and the present study are shownin Figure 6, whereas all calculated pressures from the pres-ent study are given in Table 10. The approximate con-stancy of pressure for grades 5 and 6 indicates that theslope of about 15 bars/K is reasonable. Attempts weremade to calculate the slope from entropy and volume data(Table 9). For Reaction 3, AS varies between -6.8 and+5.6 J/K depending on the choice of entropy for alman-dine and sillimanite. This would allow a maximum slopeof +3.6 bars/K for the reaction, suggesting that the ther-modynamic data are less precise than the present calibra-tion. If we assume that the best values for sillimanite andalmandine entropy are those of Robie and Hemingway(1984) and Chatillon-Colinet et al. (1983), respectively,and that the muscovite entropy is accurate, then the anniteentropy would need to be about 29 J/K higher than thevalue given by Helgeson et al. (1978) in order to be con-sistent with the l5 bars/K slope.

Several arguments suggest that much of the scatter inFigure 6 (3.09 + 0.38 kbar) results from analytical error,temperature error, and imperfect equilibration rather thanreal variation in P: (l) For grades 5 and 6, the Dutrow(1985) specimens show as much scatter in P as the spec-imens from this report collected over a much wider area.(2) The distribution of P values for M, is neaily randomon the map as shown in Figure l. Considering these ob-servations and the smaller spread of P values for grade6.5 (t0.19), we estimate the (2o) error for the method at0.5 kbar.

Specimens from grade 6.5 (M5) indicate consistent pres-

36 HOLDAWAY ET AL.: METAMORPHISM IN WEST.CENTRAL MAINE

500 550 600 650 700Temperature ('C)

Fig.6. P-T plot for M, metamorphic grades 4,5, and 6 using present data (dots) and those of Dutrow (1985) for southernFarmington quadrangle (circles). "A" indicates M, andalusite present. Temperature is that of the modified Ganguly-Saxena (198a)garnet-biotite geothermometer (Table 8), and pressure is based on the MABS geobarometer (Table l0). Sillimanite-absent rocks areplotted at 3.1 kbar, the average P for M, (see text). Only two specimens (a 6,{ and a 5A containing M, andalusite and M, sillimanite)plot incorrectly in the andalusite field. Much ofthe scatter from the average 3.1 kbar represents analytical error, temperature error,and incomplete equilibration. Temperature error (2o) is +25 'C.

o

Q A

o6o

L

sures of3.8l + 0.19 kbar (Table 10, Figs. l, 7), or about700 bars above the prevailing P during Mr. This is con-sistent with the recent discovery of kyanite in Mr, 35 kmsouthwest of Lewiston (Fig. l) on the south side of theSebago batholith (Thomson and Guidotti, 1986; Husseyet al., 1986). For grades 7 and 8, from the same M, areaas grade 6.5, pressures from all but the lowest-Ispecimenof grade 7 are higher and rather variable (Table 10, inbrackets). Possibly, the presence of K-feldspar, imply-ing partial reaction of biotite and sillimanite to garnet,K-feldspar, and fluid (Fig. 3), disturbs the equilibrationofgarnet and muscovite with biotite and sillimanite. Con-sequently, the geobarometer should be used with cautionin K-feldspar-bearing rocks. An alternative explanationis that the M, rocks share a complicated metamorphichistory, and the grade 6.5 rocks have most completelyequilibrated to a single (presumably the latest) metamor-phic event.

Dehydration equilibria

Univariant dehydration reactions in the modelKFMASH system involving breakdown of staurolite andmuscovite were applied to grades 5 and 7, respectively.Two alternative approaches are (l) to assume that thedetermined P and Z values are correct and solve for Xtroin the fluid, or (2) to assume values for &ro consistentwith presence of graphite in reducing rocks and test forconsistency between P, T, and X"ro. ln our opinion, X"roin graphitic pelites is better constrained than P or T, andthus we have chosen the second approach.

HrO content of fluid. As shown by Ohmoto and Ker-t''ck (1977), the mole fraction of HrO in the fluid ofgraphite-bearing rocks is a slowly varying function of Pand T and a rapidly varying function of the relativeamounts of CO, and CHo in the fluid (or, alternatively,of the6r). We have calculated values of 4H2o at the equil-

TaaLe 9. AS and AVfor Reactions 3. 7. and I

Reaction Irc) P (kbar) aS, (J/K) ay, (J/bar) References, notes'

(3) Alm + Mus

(7) Mus + Qtz(8) St + Qtz

610610610610660660660610

3.84 1254.1253.1

-6.84-3.32

2 . 1 15.63

75.77122.49109.09128.64

2.082

3.550

a , b , da , Da , c , o

a , f , ia , b , d , e , f , g , ha , c , d , f , g , ha , b , d , f , g , h

* a : Helgeson et al. (1 978) (S, y, and Cp), except as noted. b : Metz et al- (1 983) (Alm S); Helgesonet al. (1978) (Alm Cp). c : Chatillon-Colinet et al. (1983) (Alm S); Helgeson et al. (1978) (Alm CJ. d :Robie and Hemingway (1984) (Sil S, Cr. e : Richardson (1966) (St 14. f : Burnham et al. (1969)1n,O % s). g : Hemingway and Robie (1984) (St S, C,). h : See appendix for calculation of chemicaland disorder effect on staurolite. Staurolite formula given in Reaction 8. i : Helgeson et al. (1978)data used tor internal consistencv.

o5

t6

o . t6 6 o o\ 6

>,3

o5

5

. 64 o

6

6

J

t4

5A

Tt. n i

6A

HOLDAWAY ET AL.: METAMORPHISM

TABLE 10. Pressure determinations based on the almandine-muscovite geobarometer

IN WEST-CENTRAL MAINE

600

Temperature ('C)

Fig.1 . Comparison ofaverage Zdetermined by garnet-biotitegeothermometry $oxes) and average 7 determined from dehy-dration equilibria (curves) for Reactions 7 and 8 in grades 7 (M')and 5 (Mr), respectively. Numbers in parentheses indicate modelsfor reaction ofwater with graphite discussed in the text. For grade5, two possibilities are indicated: W" : 0 kJ per staurolite tet-rahedral site and Wc: 14.5 kJ per site as discussed in text. Datafrom Tables 12 and 13.

where Kx is the equilibrium constant based on solid so-lution of muscovite and K-feldspar, AS, is given in Table9, and n is the number of moles of water in the reaction(Kerrick, 1974). At 3.8 kbar, the pressure assumed for allM, in the area (Table l0), the pure phases of Reaction 7are stable at 664"C (Chatterjee and Johannes,1974).Fi-nal results of calculations are given in Table 12 and il-lustrated in Figure 7.

The agreement between garnet-biotite and muscovite-quartz temperatures is excellent. The accuracy of calibra-tion of each geothermometer is estimated to be + 15 oC,

TABLE 1 1 . Mole fractions for water in graphite-bearingoelites

31

r P T - P - 'No. fC) (kba4 No. ('C) (kbar)

Grade 5 (M3)137 545 3.1853 559 3.2814A 560 2.246 591 3.4123 605 3.37

Avg. s72(25) 3.12(50)

Grade 6 (M3)4A 568 2.7747 578 2.8319 589 2.80112 595 3.1950 607 3.5963 610 2.6030 616 2.60I 628 3.31140 632 3.33

Avs. 603(22) 3.00(36)

. Peak garnet-biotite f (Table 8).'. Brackets indicate uncertain values (see text).

ibration Z (calculated from garnet-biotite equilibrium) forReactions 7 and 8 for P-T conditions ofgrades 5 and 7on the basis of two models: (1) The pelitic rocks acted asclosed systems, and graphite reacted with HrO of dehy-dration to produce equimolar amounts of CO, and CHo:

2C + 2HrO: CO, + CHo. (6)

(2) COr, being several times the size of CHo, escaped fromthe rocks more slowly and/or was added to the pelitesfrom the rare metacarbonate rocks of the area. Metacar-bonate rocks equivalent to grades 5 and 6 at the easternedge of the present area contained fluids with Xrro > 0.78(Ferry, 1980). We assume a limiting case of Xco2: l)X.ruThe majority of graphite-bearing pelites with low /o,should have Xnro values between these extremes. Thesources of data and results of these calculations are sum-marized in Table ll. Other fluid species (Hr, HrS, CO,etc.) are assumed to be quantitatively unimportant(Ohmoto and Kerrick, 1977).

Muscovite-quartz reaction. In grade 7, the model uni-variant reaction

*1f,l*y.!o"'' * 319";: YJll;9'. *,lf:"o; . H:3 Q)

was in progress over a relatively narrow ?" range in allthe rocks. Using the activity model discussed above formuscovite, .Xlr" for K-feldspar, and activity coefficientscalculated from Chatterjee and Flux (1986) for muscoviteand Waldbaum and Thompson (1969) for K-feldspar, theZ separation between the experimental curve and naturaloccuffences was calculated. The relationship is

/Vote: Calculated using data of Ohmoto and Kerrick (1977)for thermodynamic data of Reaction 6, Burnham and Wall (un-pub. data) for CO, fugacity coefficients, Burnham et al. (1 969)for H,O fugacity coetficients, Ryzhenko and Volkov (1971) forCH4 fugacity coefiicients, and Kerrick and Jacobs (1981) forH2O-CO, mixing relations. Because of its nonpolar nature andsimilar fugacity coefficient to COr, CH1 was assumed to mixwith the same mixing activity coefficient as COr.

Grade 6.5 (Ms)77-3 619 4.0377-2 630 3.9690 637 3.5276 641 3.7756 641 3.6791 669 3.90

Avg. 640(17) 3.81(19)

Grade 7 (M5)73 640 3.8087 657 14.7'tl143 658 [5.08]145 663 t5.13186 67s [s.93]

Avg. 659(13) 14.93(7711

Grade I (Ms)11 699 [6.24]

E

o

rfC) P(kbar)Model 1

4"oModel 2

dr"o

660 3.8620 3.1580 3.1

0.9030.9020.919

0.8310.8310.860

AI: R?l.ln(Kxa6,o)AS;,, (2)

38 HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

TnaLe 12. Temperatures ('C) of grade 7 (MJ based on mus-covite-quartz reaction and presence of graphite at3.8 kbar

No. K Io_"** I-Model 1t l-Model 2t

Staurolite-quartz reaction. In grade 5 all specimensaontain an assemblage relating to the FASH model re-actron

H.Fer.Al,rnrSiru.Ou, + 4. l66SiO, : 7.527AlrSiOjStaurolite Quartz Sillimanite

+ l.433FerAl,Siro,, + 1.5HrO. (8)Almandine Fluid

The staurolite formula used above is an Fe end-memberformula for graphite-bearing rocks based on the chemicalstudy of Holdaway et al. (1986b). The H content is con-sistent with analyzed H values from the present study(Table 4). As shown in Appendix l, an activity model forstaurolite can be formulated using the results of Holda-way er al. (1986b).

Fe end-member staurolite breaks down along a stability curvebracketed by temperatures of 645 and 660 'C at 3.25 kbar and675 and 690'C (but very close to 690 eC) at 5 kbar (Dutrow andHoldaway, I 983; Dutrow, I 98 5; Dutrow and Holdaway, in prep.).This stability curve for Reaction 8 results from a careful srustudy of reaction products and is consistent with the earlier ex-periments of Richardson (1968) at moderate pressures. Calcu-lated AS values for Reaction 8 (Table 9) were compared withslope calculations based on the experimental results. The exper-iments indicate an average slope between 39 and 117 bars,/K,whereas the Metz et al. (1983) almandine entropy indicates aslope of 34.5 bars/K, and the Chatillon-Colinet et al. (1983)almandine entropy indicates a slope of 30.7 bars/K (Table 9).The agreement is good considering the magnitude oferrors bothin AS and in the experimental slope. A value of 128.64 J/K waschosen for the approximate midpoint between the experimentaland natural Iconditions, consistent with the Metz et al. (1983)almandine entropy. The slope data ar€ not precise enough toselect between the Metz et aI. (1983) and Chatillon-Colinet et al.(1983) almandine entropies, but they do suggest that both entro-pies are reasonable.

Using Equation 2, the almandine activity model discussedpreviously, the staurolite activity model given in Appendix l,and the fugacity ratios of Table I 1, values of 7 were calculatedand are tabulated in Table I 3 for both models of fluid behavior.An initial T of 645 oC was chosen for the experimental end-member reaction at 3.1 kbar, consistent with the experimentsand with the calculated slope values. Despite the improvementin AS, staurolite stability relations, staurolite formula, and gar-net-biotite geothermometry, there remains a 5742'C discrep-ancy between average I calculated from garnet-biotite geother-mometry and that determined from staurolite stability relations(Fig. 7). Such discrepancies have also been noted by many otherworkers (e.g., Pigage and Greenwood, 1982; Mclellan, 1985).Previously, we believed that these discrepancies resulted mainlyfrom problems in the crystal chemistry and stability relations ofstaurolite (Holdaway et al., 1986b). However, it now becomesclear that there must be additional causes, the most likely beingnonideality in staurolite tetrahedral sites.

Nonideality in staurolite solid solution is further evidenced byvalues of Kx for Reaction 8 Oable l3). Since the staurolites allcrystallized over a limited T range, K* should be roughly con-stant, as it is for muscovite (Table l2). Instead (* is high forstaurolites low in Fe and low for staurolites high in Fe. The 1(xvalues indicate that, for three specimens, solid solution reducesequilibrium 7and, for the other five, solid solution increases 7lif water dilution is ignored. The temperatures calculated for the

736 t

14314586

Avg.

1 .1431 . 1661 1081.1251 . 1 7 81.14(3)

640o c /

658663o / c6s9(1 3)

o o /

ooY

664ooo670667(2)

661o50

6576626s9(3)

. le includes activity coefficients lor muscovite and KJeldspar.". Peak garnet-biotite f oable 8).t See text and Table 11 Calculated using Equation 2, initial fof 664 rc

(Chatteriee and Johannes, 1974), and Ag:75.77 UK (Table 9).

suggesting that the averages of the two methods shouldagree within about 2l 'C (1zT5t+ Tf;. Model | (X.o,:X.") yields Z 8 'C higher and model 2 (Xcor: l0X."o)yields I equal to the average garnet-biotite temperaturefor grade 7. The maximum separation between a garnet-biotite and the conesponding muscovite dehydration ?"based on precision alone (950/o confidence level) shouldbe less than2r/TT +T or 27 "C. For model I the max-imum diference between the methods is 27 "C, and formodel2 this maximum is 19 "C (Table l2). This analysissuggests that (l) much of the scatter in garnet-biotite tem-peratures for grade 7 is due to precision error and doesnot represent real variation in 1" and (2) model 2 may beslightly better than model I for muscovite breakdown,although real variation in Xrro is probable. There is noth-ing in the results to suggest lower values of XHro than thoseof model 2 (0.78). Inspection of the values of K* and 7from model 2 (Table 12) shows that the effects of smallamounts of octahedral Fe, Mg, and Ti in muscovit€sroughly cancel out the effects of CO, and CHo in the fluidand greater Na in K-feldspar than in muscovite.

The excellent agreement between the muscovite dehy-dration equilibrium and garnet-biotite temperatures givescredence to the average values of garnet-biotite temper-atures and to the two models for the behavior of HrO inreduced graphitic systems. In addition, the preservationof graphite and reduced nature of the ilmenite indicatethat large volumes of pure HrO have not moved throughthese rocks and that reasonable amounts of CHo re-mained in the equilibrium fluid.

-IT".ioot *orkers (e.g., Speer, 1984, p. 329) have implied that

the data of Velde (1965) require that celadonite component mustreduce the thermal stability of muscovite. 'fhe initial efect ofceladonite component must relate to Reaction 7. Because Al inmuscovite is diluted much more than Al in sillimanite andK-feldspar, muscovite stability must first be increased, reach amaximum, and then be decreased according a reaction such asceladonite ,, + sillimanite: muscovite + biotite. The reactionstudied by Velde is not directly applicable to most pelitic rocksbecause it involves K-rich bulk compositiqns producing K-feld-spar in the absence of aluminum silicate. Velde's study doesdemonstrate that muscovite.celadonite solid solution cannot befar from ideal.

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE 39

most Fe-rich staurolite (specimen 53) most closely approximatethe average garnet-biotite temperature, whereas those for themost Mg-rich staurolites are farther from the average garnet-biotite temperature (Table l3), again suggesting nonideality. Re-cent high-P experimental studies ofRice (1985) can only be ex-plained by nonideal solid solution.

Ifwe exclude specimen 6, a high-Li staurolite, all the stauro-lites of this study contain at least 3.01 Fe atoms. The great ma-jority of staurolites are Fe-rich, and those with high Mg appearto be high-P, high-Z staurolites (e.g., Schreyer et al., 1984).Holdaway et al. (1986b) provide strong evidence, based on high-quality complete chemical analyses, that Fe, Mg, Zn, and Li maybe assigned to the tetrahedral Fe sites. Using this assumption,we postulate a regular-solution (pseudobinary) model for stau-rolite with tetrahedral ions grouped in two categories: Fe andvacancies as one, and Mg, Zn, and Li as the other. The justifi-cations for this are (1) vacancy content oftetrahedral sites mayvary substantially with pre$ence or absence of coexisting R'?*phases and (2) ionic radii and maximum solid solution of Mg,Zn, and Li are comparable (e.g., Holdaway et al., 1986b). Theapproach can only be expected to work well for an (Fe,Mg)-dominated staurolite with minor Li. Zn. and vacancies. A Mar-gules parameter of 15 kJ per tetrahedral site produces a solvusthat passes through 7-X." values for five of the staurolite anal-yses, leaves all the remaining analyses outside ofthe solvus, andhas a crest at 600 'C. From this maximum possible Wo value,activity coefrcients were calculated. The activity model for thepresent situation becomes rl:/ae x ilxls v ux?;ts.

The results (Table 13) show that the use of 15 kJ forWo overconects the Li-rich specimen, suggesting that theFe-Li interaction has a IZo lower than 15 kJ. For the otherstaurolites, a tighter spread in Tresults. For model l, theI averages 32 "C above that indicated by garnet-biotitegeothermometry, whereas model 2 indicates an averagef that is 27 "C above the garnet-biotite average (Fig. 7).Specimens with large diferences in Fe content but similarLi contents (e.g., 53 and l4) show similar calculated tem-peratures. The 27-32 "C difference in average I is veryreasonable in light ofthe various errors involved in garnet-biotite geothermometry, in the oversimplified staurolitesolution model, in estimating Wo, and in the fluid-com-position models.

The behavior of staurolite may be somewhat analogousto that of the muscovite-paragonite system. A (metasta-ble) breakdown curve for the Mg staurolite end memberin equilibrium with muscovite and quartz probably oc-curs at lower temperatures than that for the Fe stauroliteend member. Even though the Mg end member is unsta-ble at low pressures, the Fe limb of the Fe-Mg solvus stillapproximately controls the limiting composition of Fe-rich staurolite in the same way that the muscovite-parag-onite solvus controls the limiting composition of mus-covite even at conditions where paragonite is unstable.Analogous to paragonite, Mg staurolite is stable at highpressures (Schreyer and Seifert, 1969). This model maybe tested by seeking (experimentally and in the field) co-existing Fe- and Mg-rich staurolites at high P and T thatequilibrated below 600 'C.

It is important to distinguish between what can be saidregarding staurolite nonideality with reasonable certaintyand what remains as conjecture. (l) We can be reasonably

TneLe 13. Temperatures ('C) of grade 5 (M.)

W e - O

No K lu-u I(1) T (2)

t4lc: 15 kJisite

r(1) r (2)

137531 4237'.o

330'.23267.'

Avg.Aug +

1 152 5450.953 5591.272 5601.095 s801.246 5910.993 5921.006 6051 1 1 0 6 1 31.10(12) 581(24)

647 642636 631653 648644 6396s2 643639 634639 634646 640645(6) 639(6)643(6) 638(6)

6 1 9 6 1 5613 608609 605618 613[578] [573]609 604609 604612 608608(13) 604(13)613(4) 608(4)

Noter Temoeratures based on staurollte-quartz reaction and presence

of graphite ai 3.1 kbar. Calculated usinE Equation 2 and initial f of 645'C(Dutrow and Holdaway, in prep.). AS. for the calculations was 128.64 J/K

ifaote S;. f (1) - 11st model 1; r(2): from model 2. values given in6rackets correspond to a Li-rich staurolite whose activity coefficient isovercorrected by the pseudobinary model discussed in the text.

' Eouilibrium constant : K*" soecimens 237,330, and 267 are from the Farmington quadrangle

(Dutrow, 1985). See also Table 8, grade 5 (Ms) average.

+ Excluding specimen 6.

certain that staurolite behaves nonideally. The presentmodel distributes the diluting ions Mg, Zn, and Li overabout four sites. No ideal model can be used to accountfor the -50'C difference between garnet-biotite T and Tbased on Reaction 8, when one considers that Fe-Mg K"values between staurolite and garnet are close to unityand Zn and Li in staurolite approximately compensatefor Mn in garnet (note Kx values averaging 1 . l0 in Tablel3). The only way Kx could be reduced slightly would beif the diluting ions in staurolite were distributed over lessthan four sites (presumably 2). This is very unlikely be-cause Mg-rich staurolites contain more than 2 (Mg + Zn)(Schreyer et al., 1984), and synthetic Mg staurolite con-tains substantially more than 2 Mg (Schreyer and Seifert,1969). (2) The assignment of Fe, Mg, Zn, and Li to tet-rahedral sites is slightly less certain, although Holdawayet al. (1986b) have made a strong case for it. (3) Assumingnonideality in the tetrahedral Fe sites, approximationsthat require further testing and refinement are the pseu-dobinary model, the regular-solution assumption, and theexact value of Wo. The value of l/o chosen is the max-imum possible one, and it still doesn't fully compensatefor the I'diference between garnet-biotite geothermome-try and staurolite stability.

The srrggestion of a Margules parameter of -15 kJper site for staurolite may provide an explanation forreversals that occur in Fe-Mg partitioning of staurolitevs. other minerals. Grambling (1983) has observed a re-versal in staurolite-chloritoid partitioning caused by anincrease in Mg. Chloritoid contains more Fe than stau-rolite in very Fe-rich compositions and more Mg thanstaurolite at more Mg-rich compositions. A comparablereversal may occur with staurolite-almandine at muchmore Mg-rich compositions than those observed here(Schreyer et al., 1984). The estimated Z" should be testedagainst these Ko reversals, which may well provide a verysensitive indicator.

40 HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

3.5

2 .5

550 600 650

Temperoture CC)

Fig. 8. Range of P-Zconditions for various metamorphic events in west-central Maine. Approximate age of each event is given.Data arc from Dickerson and Holdaway (ms.) and the present report. The sillimanite-andalusite boundary is from Holdaway (197 l),the first sillimanite curve is the beginning of Reaction 8, and the average sillimanite-K-feldspar curve is the average appearance ofthis assemblage by Reaction 7.

ollJ

E?l etaaaC)

o-

2r500

The results of geothermometry, geobarometry, and de-hydration geothermometry are consistent when staurolitenonideality is taken into account (Fig. 7). These resultsare in agreement with reasonable models for fluids (Xrro:0.78-0.91) in reduced graphite-bearing rocks. In someprevious studies, overestimation of pressure or the stau-rolite problem has led workers to suggest unrealisticallylow values of Xrro.

C.l,usBs oF METAMORpHIC coNDITToNS

The thermal events recognized as Mr, Mr, and M5 pro-vide a discontinuous record of pressure increase during atime interval between 400 and 325 Ma (Fig. 8). The locus

of metamorphism moved southward during this time fromCaratunk, 30 km north-northeast of Kingfield (Dickersonand Holdaway, ms.; Holdaway et al., 1986a), to the Nor-way-Lewiston region, a distance of about 125 km in asouth-southwest direction. In discussing the causes ofmetamorphism, temperature and pressure will be treatedseparately. Although a detailed discussion of M, is not atopic of the present contribution, evidence from othersources will be used here to tie M, into the general pictureof metamorphism lor the region.

Temperature

With the exception of low-grade metamorphism M',all the metamorphic events in the area of Figure I show

,EM3 379-394 m.y.

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE 4 l

a general spatial association with plutonic rocks (Fig. l;Holdaway et al., 1986a). Many of the intrusive rocks ofthe area are sill-like bodies that pass gently beneath thesurrounding metamorphic rocks, commonly with anorthern dip (Hodge et al., 1982; Carnese, 1983). Mod-eling studies of Lux et al. (1986) and DeYoreo (1987)demonstrate that conditions such as those achieved at theculmination of Mr, Mr, and M, require the close prox-imity of magmatic heat. Even though isograds represent-ing substantially increased Z are commonly up to 10 kmand rarely as much as 30 km from the exposed igneousrocks (Fig. l; Holdaway et al., I 986a), the overall igneousspatial association and the gently dipping character ofmany of the igneous rocks imply that the batholithic heatsources lay no more than I to 3 km below the Mr, Mr,and M, isograds. This is shown by Lux and Guidotii(1985) and Guidotti et al. (1986) for the relation betweenthe Sebago batholith and the M, sillimanite-K-feldsparisograd. In areas where isograds lie far from the igneouscontacts and detailed geophysical studies fail to demon-strate that igneous rocks lie at depth, there is the possi-bility that gently dipping plutonic bodies once lay abovethe metamorphic rocks and acted as a heat source formetamorphism.

The conditions of grade 7, M, metamorphism (3.8 kbar,660 "C) are very close to the invariant point at which thevapor-absent melting curve for muscovite-qu artz-plagio-clase occurs in graphitic pelites (Kerrick, 1972; Ohmotoand Kerrick, 1977). In this area (Fig. I in Holdaway etal., 1982) K-feldspar-absent rocks (6.5) are regionally in-terspersed with K-feldspar-bearing rocks (7) that clearlyformed at a higher average ?'than grade 6.5 (Table 8).The large number of small granitic bodies and pegmatitessuggests that incipient melting occurred in the rocks im-mediately below the present exposures, and varying de-grees of upward transmission of magma and heat resulted.The incipient melting of these local magmas and the Se-bago granites at slightly greater depth probably had a buFering effect on the rocks, preventing the total destructionof muscovite at the prevailing P, except in the single grade8 specimen (ll, Table 8) in which muscovite probablygrew after the thermal culmination for the locality. Inaddition to Z differences, minor differences in Xrro be-tween grades 6.5 and 7 are possible.

Pressure

In west-central Maine, pressure increased from northto south and with time. The pressure of 2.35 kbar for thelowest-P portion of M, (Dickerson and Holdaway, ms.),average P of 3.1 kbar for M., and average P of 3.8 kbarfor M, (Fig. 8) correspond to approximate depths of 8.2,10.9, and 13.3 km, respectively, during the three events.Mr, M,, and M, occurred at 400,394-379, and 325 Ma,respectively, and at average distances of about 30, 80,and 1 l0 km south-southwest of Caratunk where the low-est-P M, begins. This increase in P with map positionand time may have been produced by two kinds of P

increase: (1) an increase from north-northeast to south-southwest at any given time and (2) an increase with timeat any given place. Ifone considers only the time, generalspatial distribution, and P-T conditions of the threeevents, either ofthese two causes ofP increase could haveacted alone. Evidence cited below shows that both werenecessary, perhaps each being responsible for about halfof the pressure increase.

Change in P from north-northeast to south-southwestat a given time is difrcult to document quairtitatively,but a number offactors suggest that this has occurred: (l)The northern lobe of the Lexington batholith producedcordierite-andalusite (Mr", -2.35 ! 0.25 kbar), whereasthe central and southern lobes produced staurolite-an-dalusite (M2, -2.7 + 0.25 kbar;Mr", -2.75 + 0.25 kbar)(Dickerson and Holdaway, ms.t Holdaway et al. 1986a).The Gaudette and Boone (1985 and pers. comm.) age of400 Ma is based on all three lobes, and none is distin-guishable from the others beyond the limits of error. (2)Farther south, M, in the Kingfield and Anson quadran-gles produces little or no garnet zone, and andalusite ap-pears immediately after staurolite (Holdaway et a1.,1986a). To the south near Farmington, the gamet zonewidens to about 6 km, and andalusite first appears in theDixfield quadrangle several kilometers west of the firstM, staurolite (Dutrow, 1985). These observations areconsistent with a southward P increase in Mr. (3) Dall-meyer (1979) has shown with incremental Ar-release ageson metamorphic biotites in an area along the eastern edgeof Figure 1 (M, and M5) that ages decrease progressivelyfrom about 330 Ma in the northeast to 230 Ma in thesouthwest. This slower cooling toward the southwest isconsistent with a longer period of erosion to the biotiteblocking isotherm or a higher initial Z, hence the possi-bility of a deeper burial or slower uplift to the southwest.(4) In M, there must have been a P increase southwestacross the Sebago batholith as indicated by P about 3.8kbar on the north side compared with P of 5 kbar or moreon the south side where kyanite routinely occurs withalmandine-rich garnet (Thomson and Guidotti, 1986;Hussey et al., 1986; Carmichael, 1978). These observa-tions require additional documentation, but a case forsouth-southwest increase in P appears to be emerging.This increase (perhaps equivalent to 2-3 km) probablyrepresents a combination of slight tilting during post-metamorphic uplift and higher topography to the south-southwest during metamorphism.

The progressive increase in P with each subsequentevent also requires changes in P with time as shown bythe following evidence: (l) The transition from cordier-ite-andalusite (Mr,) to staurolite-andalusite (Mr, Mr.) oc-curs abruptly at the junction between the northern andcentral lobe of the kxington batholith (Dickerson andHoldaway, ms.; Holdaway et al., 1986a), suggesting a rapidP increase over a small time interval near 400 Ma. Thistransition does not involve a fault, but does involve achange in igneous lithology. (2) In the Rangeley, Rum-ford, Phillips, Dixfield, Farmington, and Augusta quad-

42 HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

rangles (Fig. l), M, (no andalusite) is superimposed onM. (andalusite-bearing), indicating a P increase with timethroughout these areas (Holdaway et al., 1982). (3) In theBryant Pond, Buckfield, Livermore, and Lewiston quad-rangles, there is a much less obvious superimposition ofM5 (-3.8 kbar) on M3 (-3.1 kbar). In part this is lessobvious because of the absence of a P-sensitive phasetransition like andalusite-sillimanite. However, the ap-pearance of retrogressive chlorite in the M, staurolite-grade rocks may be due to this efect.

The progressive increase in P with time requires theaddition of perhaps 2-3 km of rock above the presentlevel of exposure over a 75-m.y. period at a rate morerapid than erosion. This increase in overburden couldhave occurred over the whole region or could have beenadded in larger amounts in the southern and central partsofthe area shown in Figure I than in the northern parts.The rock may well have been added both at the surfaceand at shallow crustal levels. There are several possiblemechanisms whereby P may have increased with time:(1) Marine sedimentary and/or volcanic deposition couldhave continued throughout much of the time in question.This would have resulted from continued subsidence whileplutonic rocks were being emplaced in great volume.However, the likelihood of surface subsidence during sucha time is not great. (2) Nonmarine deposition could havebeen extensive with the build-up of a widespread volca-nic highland. The P increase could have been accom-plished by subsidence at the present level combined withintrusion at shallower levels and extrusion at the surface(see also Holdaway et al., 1982). The shallower igneousrocks would have been correlatives of the batholiths ex-posed at present levels. (3) A final possibility is emplace-ment of crust by overthrusting, presumably from the east.This model has problems because of the general absenceof evidence of Middle Devonian to Mississippian thrust-ing. We prefer the model of emplacement of magmaticrocks at the surface and at levels shallower than presentexposures. It should also be pointed out that the youngestrocks in west-central Maine are the Middle DevonianTomhegan Formation, which is exposed 30 km north ofthe Lexington batholith and contains felsic tuffs, tuffa-ceous sedimentary rocks, and garnet rhyolite (Rankin,1968). It is tempting to speculate that these are the ear-liest remnants of a more widespread volcanic terrane.

CoNcr,usroNs

l. In west-central Maine, M3 Q94-379 Ma) includesthe biotite, garnet, staurolite, staurolite-sillimanite, andsillimanite zones. The younger M5 (325 Ma) includes sil-limanite and sillimanite-K-feldspar within the second sil-limanite isograd.

2. In these graphite-bearing pelitic rocks, the micas arewell expressed in terms of conventional end membersand "Ti-biotite" (KTitr(Fe,Mg)AlSi,O,JOHD or'.Ti-muscovite" (KTi(Fe,Mg)AlSi3Or'(OH)r). Celadonire con-tent of muscovite is very low and has a potentially largeerror. In determining mica end members, no substitution

involving a combination of octahedral and l2-fold sitesis required.

3. In garnet, the progressive changes in Mn contentwith Zmay be used to confirm previously suggested phaserelations between garnet, staurolite, biotite, and silliman-ite. In the staurolite zone, the occurrence ofa wide rangeof biotite compositions with chlorite, along with otherevidence, suggests that chlorite is mainly a retrograde min-eral at this grade. A related observation is that the stau-rolite-chlorite tie line in AFM projection must break earlyin the staurolite zone; rocks of appropriate compositionwould form aluminum silicate immediately after stau-rolite.

4. Staurolite H values are near 3, based on 48 oxygens.Staurolite-garnet and staurolite-biotite KD plots suggestpositive deviation from ideality in Fe-Mg solid solution.

5. Ilmenite is very low in hematite content and coexistswith graphite in most pelitic rocks, supporting low fo,values near QFM.

6. Modified Ganguly-Saxena temperatures indicate thatthe staurolite-sillimanite rocks in M, formed at 581 + 24oC and sillimanite-K-feldspar-muscovite rocks in M,formed at 659 + l3'C.

7. The formation of the first sillimanite from staurolitein M,, directly above the andalusite field, suggests P :

3.1 + 0.25 kbar.The muscovite-almandine-biotite-silli-manite (MABS) geobarometer has been calibrated andyields P : 3.8 kbar for M,.

8. The muscovite breakdown reaction in M, is consis-tent with a Icalculated from garnet-biotite, P: 3.8 kbar,and a graphite-present fluid model in which P.o, is be-tween P."o and l0P.ro. The staurolite breakdown reac-tion in M. is not consistent with garnet-biotite Z, butrequires ? Wo = 15 kJ per tetrahedral Fe site in order tobring the two sets of data into better agreement. This non-rdeality suggests a staurolite solvus at about 600 "C.

9. Heat for Mr, Mr, and M, came from shallow sill-likebatholiths either below or above the present schists. Max-imum temperature in M, leveled out at about 660 "C (3.8kbar) due to bufering caused by incipient vapor-absentmuscovite melting, ubiquitous local pegmatites and gran-ite, and the Sebago batholith below.

10. Pressure increase from M, to M. to M, (Fig. 8)corresponds to about 5 km over a 75-m.y. period. Abouthalf of this is related to a south-southwest movement ofthe locus of igneous activity and metamorphism and as-sociated P increase with map position. The remainder isa P increase with time that may well be associated withemplacement of shallower intrusive rocks and develop-ment of an extensive volcanic highland during the De-vonian and Mississippian.

AcxNowr-BncMENTSWe acknowledge with thanks the assistance of Dwight Deuring on the

electron microprobe. We thank Jo Ann Weber for writing a computerprogram to calculate activity coefficients for almandine component. MaryLee Eggart at LSU drafted the figures. Myrtle Watson is thanked for hercxpert typing of many drafls of the manuscript. M.J.H. acknowledges thesupport of the National Science Foundation (grants EAR-8306389 and

HOLDAWAY ET AL.: METAMORPHISM IN WEST.CENTRAL MAINE 43

EAR-8606489) and B L D the Inslitute for the Study of Earth and Manat SMU R.W H acknowledges support from NASA (grant NA99-51 toR N Clayton). The project has benefited greatly from discussions withC V Guidotli and his colleagues at University of Maine and with DarrellHenry at LSU. The manuscript has been reviewed formally by M. PDickenson III, Jo Laird, and Eileen Mclellan and informally by C VGuidotti and Jinny Sisson

Rnrnnpucns crrpuAlbee, AL, and Ray, L. (1970) Correction factors for electron probe

microanalysis of silicates, oxides, carbonates, phosphates, and sulfatesAnalytical Chemistry, 42, 1408-1414.

Aleinikofl, J.J, Moench, R H, and Lyons, J.B. (1985) Carboniferous U-Pbage of the Sebago batholith, southwestern Maine. Geological Society ofAmerica Bulletin, 69, 990-99 6.

Bence, A.E , and Albee, A.L (1968) Empirical correction factors for elec-tron microanalysis of silicates and oxides Journal of Geology, 7 6, 382-403

Burnham, C W., Holloway, J.R., and Davis, N F (1969) Thermodynamicproperties ofwater to 1,000 "C and 10,000 bars. Geological Society ofAmerica Special Paper I 32.

Carmichael, D.M (1978) Metamorphic bathozones and bathograds: Ameasure of the depth of post-metamorphic uplift and erosion on aregional scale American Journal of Science, 278, 769-797

Carnese, MJ (1983) Gravity studies ofintrusive rocks in west-centralMaine M S thesis, University of New Hampshire, Durhirm, NewHampshire.

Chatillon-Colinet, C., Kleppa, O.J., Newton, R.C., and Perkins, D, III.(1983) Enthalpy of formation of FerAlSi,O,, (almandine) by high rem-perature alkali borate solution calorimetry Geochimica et Cosmo-chimica Acta, 47, 439-444.

Chatterjee, N.D., and Flux, Susanne. (1986) Thermodynamic mixingproperties of muscovite-paragonite crystalline solutions at high tem-peratures and pressures, and their geological applications. Joumal ofPetrology, 27, 677-693.

Chatterjee, N D., and Johannes, Wilhelm (1974) Thermal stability andstandard thermodynamic properties of synthetic 2M,-muscovite,KAlr[AlSi30,o(OH)r] Contributions to Mineralogy and Petrology, 48,89 - l l 4

Cygan, RT., and Lasaga, AC (1982) Crystal growth and the formationof chemical zoning in garnets. Contributions to Mineralogy and Pe-trology, 79, 187-200

Dallmeyer, R D. (1979) Chronology of igneous and metamorphic activityin south-central Maine In J.W Skehan and P.H. Osberg, Eds., TheCaledonides in the U.S A., p. 63-71. Weston Observatory, Weston,Massachusetts.

Dallmeyer, R D , and Vanbreeman, O ( 1978) 40Ar/3eAr and Rb-Sr agesin west-central Maine; their bearing on the chronology of tectono-ther-mal events. Geological Society of America Absfacts with Programs,10 . 38

Dallmeyer, R D., Vanbreeman, O , and Whitney, J A (1982) Rb-Sr whole-rock and 4oArl3eAr mineral ages of the Hartland stock, south-centralMaine: A post-Acadian representative of the New Hampshire PlutonicSeries American Journal of Science, 282, 79-93

DeYoreo, J.J (1987) The role of crustal anatexis and magma migrationin regions ofthickened continental crust. Geological Society ofAmericaAbstracts with Programs, 19, l0-l I

Dickenson, M.P., and Hewitt, D.A (1986) A garnet-chlorite geother-momeler. Geological Society of America Abstracts with Programs, 18,584

Dickerson, R.P (1984) A study of the polymetamorphism and mineralchemistry ofthe Little Bigelow Mountain and Bingham quandrangles,Maine M.S thesis, Soutbem Methodist University, Dallas, Texas

Dutrow, B L (1985) A staurolite trilogy: I. Lithium in staurolite and itspetrologic significance. II An experimental determination ofthe upperstability of staurolite plus quartz. III. Evrdence for multiple metamor-phic episodes in the Farmington quadrangle, Maine. Ph D. thesis,Southem Methodist University, Dallas, Texas.

Dutrow, B.L., and Holdaway, M.J. (1983) The upper stability of stauro-

lite + quartz at medium to low pressures. Geological Society ofAmericaAbstracts with Programs, 15, 563

Dutrow, B L., Holdaway, M J., and Hinton, R W. (1986) Lithium in stau-

rolite and its petrologic significance Contributions to Mineralogy andPetrology, 94, 496-506.

Dymek, R F (1983) Titanium, aluminum, and interlayer cation substi-tutions in biotite from high-grade gneisses, West Greenland American

Mineralogist, 68, 880-899.Eggleston, R.A, and Bailey, S.W. (1967) Structural aspects ofdioctahe-

dral chlorite American Mineralogist, 52, 673-689.Evans. B W. (1969) Chlorine and fluorine in micas of pelitic schists from

the sillimanite-orthoclase isograd, Maine American Mineralogtst, 54,r209 - t 2 t I

Evans, B.W., and Guidotti, C.v. (1966) The sillimanite-potash feldsparisograd in western Maine, U S A Contributions to Mineralogy and

Petrology, 12,25-62.Ferry, J M. (1976) P, T,f"o,, andf"reduring metamorphism of calcareous

sediments in the Waterville-Vassalboro area, south-central Maine.

Contributions to Mineralogy and Petrology, 5'l, l19-143.-(1980) A case study of the amount and distribution of heat and

fluid during metamorphism Contributions to Mineralogy and Petrol-

ogy , 71 , 373 -385 .Ferry, J.M., and Spear, F.S. (1978) Experimental calibration ofpartition-

ing of Fe and Mg between biotite and garnel. Contributions 10 Min-eralogy and Petrology, 66, I l3-1 l7

Fyfe, W.S., Turner, F J., and Verboogen, J (i958) Metamorphic reactionsand metamorphic facies Geological Society of America Memoir 73.

Ganguly, Jibamitra. (1969) Chloritoid stability and related parageneses:

Theory, experiments, and applications. American Joumal of Science,267,910-944.

Ganguly, Jibamitra, and Saxena, S K (1984) Mixing properties of alu-minosilicate gamets: Constraints from natural and experimental data,and applications to geothermobarometry. American Mineralogist, 69,88-97

-(1985) Mixing properties of aluminosilicate gamets: Constraintsfrom natural and experimental data, and applications to geothermo'

barometry: Clarifications. American Mineralogist, 7 0, 1320.

Gaudette, H.E, and Boone, G.M. (1985) Isotopic age of the [rxingtonbatholith; Constraints on timing of Acadian metamorphism in westernMaine Geological Society of America Abstracts with Programs, 17,r 9-20.

Geving, R.L. (1987) A sludy ofthe metamorphic petrology ofthe north-ern Snake Range, east-central Nevada. M.S. thesis, Southern MethodistUniversity, Dallas, Texas

Grambling, J A (1983) Reversals in Fe-Mg partitioning between chlori-toid and staurolite. American Mineralogist 68, 373-388.

Guidotti, C V. (1968) Prograde muscovite pseudomorphs after staurolitein the Rangeley-Oquossoc area, Maine. American Mineralogist, 53,1 368-1 376

-(1970) Metamorphic petrology, mineralogy, and polymetamor-phism in a portion of northwest Maine. New England IntercollegiateGeological Conference, 62nd Annual Meeting, B-2, l-23

-(1974) Transition from staurolite to sillimanite zone, Rangeleyquadrangle, Maine. Geological Society of America Bulletin, 85, 4'75-490.

-(1978) Compositional variation of muscovite in medium- to high-grade metapelites of northwestern Maine. American Mineralogist, 63,878-884

-(1984) Micas in metamorphic rocks. Mineralogical Society ofAmerica Reviews in Mineralogy, 13,357-467.

Guidorti, C.v., Herd, H H, and Turtle, c.L (1973) Composition andstructural state ofK-feldspar from K-feldspar + sillimanite grade rocksin northwestern Maine. American Mineralogist, 58,'7 05-7 | 6.

Guidotti, C.V, Trzcienski, W.E , and Holdaway, M.J (1983) A northernAppalachian metamorphic transect-Eastem townships, Quebec-Mainecoast to the central Maine coast In P.E. Schenk, Ed., Regional trendsin the geology ofthe Appalachian-Caledonian-Mauritanide orogen, p.235-247 Reidel, Boston, Massachusetts

Guidotti, C V , Gibson, D., Lux, D R., DeYoreo, J J., and Cheney, J T(1986) Carboniferous metamorphism on the north (upper) side ofthe

44 HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

Sebago barholith New England Intercollegiate Conference, 78th An-nual Meet ing,306-341.

Hayward, J A., and Gaudette, H E (1984) Carboniferous age of the Se-bago and Effingham plutons, Maine and New Hampshire, GeologicalSociety of America Abstracts with Programs, 16, 22.

Helgeson, H C , Delany, J.M., Nesbitt, H.W., and Bird, D.K (1978) Sum-mary and critique of the thermodynamic properties of rock-formingminerals. American Journal of Science,2'78-A

Hemingway, B.S., and Robie, R.A. (1984) Heat capacity and thermody-namic functions for gehlenite and staurolite: With comments on theSchottky anomaly in the heat capacity of staurolite. American Miner-alogist, 69, 307-3 I 8.

Henry, W.E (1974) Metamorphism of the pelitic schists in the Dixfieldquadrangle, NW Maine. Ph D. thesis, University of Wisconsin, Mad-ison. Wisconsin.

Hodge, D.S , Abbey, D.A , Hasbin, M A., Patterson, J.L., Ring, M.J., andSweeney. J F (1982) Gravity studies ofsubsurface mass distributionsof granitic rocks in Maine and New Hampshire. American Joumal ofScience. 282. 1289-1324

Hodges, K.V, and Spear, F.S. (1982) Geothermometry, geobarometryand the AlrSiO, triple point at Mt. Moosilauke, New Hampshire.American Mineralogist, 67, 1 | l8-l 134.

Holdaway, M J. (1971) Stability ofandalusite and the aluminum silicatephase diagram. American Journal of Scienc e, 2T l, 97 -I 3 l

- (l 978) Significance ofchloritoid-bearing and staurolite-beanng rocksin the Picuris Range, New Mexico. Geologtcal Society of America Bul-letin, 89, 1404-1414

-(1980) Chemical formulae and activity models for biotite, mus-covite, and chlonte applicable to pelitic metamorphic rocks. AmericanMineralogist, 65, 7 I I-7 19

Holdaway, M J., Guidotti, C.V., Novak, J.M., and Henry, W E. (1982)Polymetamorphism in medium- to high-grade pelitic metamorphicrocks, west-central Maine Geological Society of America Bulletin, 93,5'72-584.

Holdaway, M.J., Dickerson, R.P, and Durrow, B L. (1986a) Petrologyand field relations of M2 metamorphism in west-central Maine. NewEngland Intercollegiate Geological Conference, 78th Annual Meeting,240-253

Holdaway, M.J., Dutrow, 8.L., and Shore, Patrick. (1986b) A model ofthe crystal chemistry of staurolite. American Mineralogist, 71, ll42-1 1 5 9 .

Holdaway, M.J., Dutrow, B.L., Borthwick, James, Shore, Patrick, Har-mon, R.S, and Hinton, R W (1986c) H content ofstaurolite as deter-mined by H extraction line and ion microprobe. American Mineralo-g i s t , 7 t , l l 3 5 - l 1 4 l

Hussey, A.M., Bothner, W A , and Thomson, J.A. ( I 986) Geological com-parisons across the Norumbega fault zone, southwestern Maine. NewEngland Intercollegiate Geological Conference, 78th Annual Meeting,53-78

Jones, A P., and Smith, J.V. (1984) Ion probe analysis ofH, Li, B, F, andBa in micas, with additional data for metamorphic amphibole, scapo-lite, and pyroxene. Neues Jahrbuch fiir Mineralogie Monatshefte, 228-240

Juurinen, Aarno (1956) Composition and properties of staurolite. An-nales Academiae Scientiarum Fennicae, Series A, III Geology, 47, l-

fe.nct, O M. (1972) Experimental determination of muscovire + quanzstability with P,,o 1 P,"or American Journal ofScience, 272,946-958.

-(1974) Review of mixed-volatile (H,O-CO,) equilibria. AmericanMineralogist, 59, 729-7 62.

Kerrick, D.M , and Jacobs, G K. (1981) A modified Redlich-Kwong equa-tion for HrO, COr, and H,O-CO, mixtures at elevated pressures andtemperatures. American Journal of Science, 281, 7 35-7 61

Kretz, Ralph. (1983) Symbols for rock-forming minerals. American Min-eralogist, 68, 27 7 -2'1 9.

Lux, D R., and Guidotti, C V. (1985) Evidence lbr extensive Hercynianmetamorphism in western Maine. Geology, 13, 695-700.

Lux, D.R., DeYoreo, J.J., Guidoui, C.V., and Decker, E.R (1986) Roleof plutonism in low-pressure metamorphic belt formation. Nature, 323,794-797

Mclellan, Eileen. (1985) Metamorphic reactions in the kyanite and silli-manite zones ofthe Barrovian type area Journal of Petrology, 26,789-8 1 8

Metz, G.w., Anovitz, L.M , Essene, E J., Bohlen, S.R., Westrum, E.F,and Wall, VJ. (1983) The heat capacity and phase equilibria of al-mandine (abs.). EOS, 64,346-347.

Moench, R.H., and Zaxtman, R.E. (1976) Chronology and styles of mul-tiple deformation, plutonism, and polymetamorphism in the Merri-mack synclinorium of western Maine. In P.C Lyons and A.H. Brown-low, Eds., Studies in New England geology. Geological Society ofAmerica Memoir 146, 203-238.

Munoz, J L, and Ludington, S.D. (1974) Fluorine-hydroxyl exchange inbiotite American Journal of Science. 27 4. 396-413.

-(1971) Fluorine-hydroxyl exchange in synthetic muscovite and itsapplication to muscovite-biotite assemblages. American Mineralogist,62,304-308.

Newton, R.C, and Haselton, H.T. (1981)Thermodynamicsof thegarnet-plagioclase-AlrSi(),-quartz geobarometer. In R.C. Newton et al., Eds.,Advances in physical geochemistry, vol- l, p. lll-147. Springer-Ver-lag, New York.

Novak, J M, and Holdaway, M.J. (1981) Metamorphic petrology, min-eral equilibria, and polymetamorphism in the Augusta quadrangle, south-central Maine. American Mineralogist, 66, 5l-69

Ohmoto, Hiroshi, and Kerrick, D.M. (1977) Devolatilization equilibriain graphitic systems. American Journal of Science, 277, l}l3-1044.

Osberg, P.H., Hussey, A.M., and Boone, G.M. (1985) Bedrock geologicmap of Maine. Maine Geological Survey, Augusta, Maine.

Pigage, L.C., and Greenwood, H J. (1982) Internally consistent estimatesofpressure and temperature: The staurolite problem. American Journalof Science, 282, 943-969.

Powell, Roger, and Evans, J.A ( 1983) A new geobarometer for the assem-blage biotite-muscovite-chlorite-quartz Joumal of Metamorphic Ge-o logy , l , 331 -336 .

Rankin, D W (1968) Volcanism related to tectonism in the Piscataquisvolcanic belt, an island arc of Early Devonian age in north-centralMaine In E-anZen et al., Eds., Studies of Appalachian geology: North-em and maritime, p 355-369. Interscience, New York.

Rice, J M. (1985) Experimental partitioning of Fe and Mg between co-existing staurolite and garnel (abs.). EOS, 66, 1127.

Richardson, S.W (1966) Staurolite. Carnegie Institution of WashingtonYearbook, 65.248-252

-(1968) Staurolite stability in a part of the system Fe-Al-Si-O-H.Joumal of Petrology,9, 461-488

Robie, R A . and Hemingway, B.S ( I 984) Entropies of kyanite, andalu-site, and sillimanite: Additional constraints on the pressure and tem-perature of the AlSiO, triple point. American Mineralogist, 69,298-306.

Robinson, G.R., Jr. (1983) Calibration of the muscovite-biorire-quartz-garnet-aluminosilicate geothermobarometer (abs.) EoS, 64, 35 l.

Ryzhenko, B N , and Volkov, V.P. (1971) Fugacity coefficients of somegases in a broad range of temperatures and pressures. GeochemistryInternational, 8, 468-48 I .

Schreyer, Wemer, and Seifert, F (1969) Compatibility relations of thealuminum silicates in the systems MgO-AlrOr-SiOr-HrO and KrO-MgO-AlrO3-SiOr-H,O at high pressures American Journal of Science, 267,407-443.

Schreyer, Werner, Horrocks, P C, and Abraham, K. (1984) High-mag-nesium staurolite in a sapphirine-garnet rock from the Limpopo belt,southern Africa Contributions to Mineralogy and Petrology, 86, ZO0-207

Smith, J.V. (1968) The crystal structure of staurolite American Miner-alogist , 53, I 139- l 155.

Spear, F.S, and Selverstone, Jane. (1983) Quantitative P-?" paths fromzoned minerals: Theory and tectonic applications. Contributions toMineralogy and Petrology, 83, 348-357.

Speer, J A (1984) Micas in igneous rocks. Mineralogical Society of Amer-ica Reviews in Mineralogy, 13,299-356

Steele, JM, Hervig, R.L. , Hutcheon, ID. , and Smith, JV. (1981) Ionmicroprobe techniques and analysis of olivines and low-calcium py-roxenes. American Mineralogist, 66, 526-546.

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE 45

Thompson, A.B. (1976) Mineral reactions in pelitic rocks: I. Predictionof P-I-X(Fo-Mg) phase relations. American Journal of Science, 276,40t-424

Thompson, J.B. (1982) Composition space: An algebraic and geometricapproach. Mineralogical Society of America Reviews rn Mineralogy,t 0 , l - 3 1 .

Thomson, J A., and Guidotti, C.V. (1986) The occurrence of kyanite insouthern Maine and its metamorphic implications. Geological Societyof America Abstracts with Programs, 17, 59.

Tompkins, LA, Bailey, S.W, and Haggerty, S.E. (1984) Kimberliticchlorites from Sierra Leone, West Africa: Unusual chemistries andstructural polytypes. American Mineralogist, 69, 237 -249.

Tracy, R.J (1978) High grade metamorphic reactions and partial meltingin pelitic schists, west-cenlral Massachusetts. American Joumal of Sci-ence ,278 , 150 -178 .

- ( I 982) Compositional zoning and inclusions in metamorphic min-erals Mineralogical Society of America Reviews in Mineralogy, 10,355-397

Veblen, D.R (1983) Microstructures and mixed layering in intergrownwonesite, chlorite, talc, biotite, and kaolinite. American Mineralogist,68, 566-580

Velde, B. (1965) Phengite micas: Synthesis, stability and natural occur-rence. American Journal of Science, 263, 886-9 I 3.

Waldbaum, D.R., and Thompson, J.B, Jr. (1969) Mixing propenies ofsanidine crystalline solutions: IV. Phase diagrams from equations ofstate. American Mineralogist, 54, 1274-1298.

Zen, E-an. (1981) Metamorphic mineral assemblages of slightly calcicpelitic rocks in and around the Taconic allochthon, southwestern Mas-sachusetts and adjacent Connecticut and New York. U.S. GeologicalSurvey Professional Paper I I 13.

M,tNuscnryr REcETvED Mlncs 20, 1987M,asrrscnrvr AccEprED Ser"rsr4ssr 14. 1987

AppnNorx 1. Srlunor,rrE DATA

Entropy

Hemingway and Robie (1984) have provided entropy data forstaurolite 355-1, originally studied by Zen (1981).

Chenical correction. Holdaway et al. (1986b) have shown thatthis staurolite has the formula Hrrr(FerruMg,orrMnoorfnoorrl-iorru-Croo,rCooorTioos.)Al,?7s7si?2130or.6 In order to correct for thechemical differences between this staurolite and pure Fe "end-member" staurolite (HrFeo.Al,rrrSiruuOor) as used in Reaction8, appropriate amounts of entropy of the following compoundswere added or subtracted from the raw data of Hemingway andRobie (1984): MgO, LirO, pyrophyllite, Cr,Or, ZnO, TiO,, MnO,CoO, SiOr, AlrOr, and fayalite. The molar volumes of the sub-tracted components summed to 3.1 14 J/bar, and the sum for theadded components was 3.055 J/bar. This slight compression wascorrected for by subtracting 1.48 J/K from the entropy (25.1J/K for each J/bar; Fyfe et al., 1958). The total chemical correc-tion was +23.47 - 1.48 : +21.99 J/K.

-u co ir gi*n as 0.02 to account for small amounts of co, v,

and Zr.

Disorder correction. In order to make the disorder correction,two assumptions were made: (l) The staurolite is as ordered aspossible, i.e., minimum corrections were made; (2) the correc-tions that were made are corrections that can be reasonably as-sumed to be constant for Fe "end-member" staurolite and forthe natural staurolites from west-central Maine. The "end-mem-ber" staurolite formula may be rewritten as (HrEr)o(FerrEot).-(A1, ,rFeo ,rEo rr)r(Feo.rEr rr)14.116(Si7 66A10 3o)EOo*. Following themaximum-order structure model of Holdaway et al. ( 1986b) withP(lB), A(3A), U(1) sites partially filled and P(lA), A1(3B), andU(2) sites empty, we have the following disorder terms withAS-o: -R[(Xln X) + (l - X)ln(l - X))] per site: (l) -Fe sitedisorder is -4R[0.025 ln 0.025 + 0.975 ln 0.975] : 0.4676R'(2) P(lB) site disorder involves only 3.95 sites and 2.9 ions be-cause 0.05 H each on P(lA) and P(lB) is coupled to the 0.1vacancy in i"Fe. Disorder is -3.95R[0.734 ln 0.734 + 0.266 ln0.2661:2.2880R. (3) Al(3A) site disorderincludes U(1) disorderto which it is coupled. Disorder is -2R[0.790 ln 0.790 + 0.075ln 0.075 + 0.135 ln 0.1351 : 1.3017R. (4) Si si te disorder is-8R[0.958 ln 0.958 + 0.042 ln 0.042] : 1.4065R. Total disorderentropy is 5.4638R : 45.42 J/K.

Hemingway and Robie's (1984) Table 6, which already has a+ 121.0-J/K chemical and disorder correction, may be adaptedto the present Fe "end-member" staurolite by subtracting 53.59J/K (- 121.0 + 2r.99 + 45.42).

Activity model

Because the above disorder corrections apply to the Fe "end-member" staurolite in hydrothermal experiments and in nature,none ofthem should be taken into account in the activity model.In determining the model and calculating staurolite activities,the following reasonable assumptions were made (Juurinen, 1956;Holdaway et al., 1986b). (1) Assume that the H content of thestaurolites is 3 and fix thei"Fe site vacancy at 0.1 ions pfu. Thusi"Fe occupancy totals 3.9 (to allow this quantity to vary in thesespecimens would produce variations resulting largely from an-alytical error). (2) 0.25 (Mn + Fe) pfu is assigned to U(1) and0.15 Fe pfu is assigned to Al(3A). Where Li is not known, anaverage value of 0.3 pfu is assumed. Remaining Fe, Mg, Zn, andLi are assigned to toFe sites.

The activity model below restricts additional solid solution tothe i'Fe and U(l) sites (i : end-member values):

--l};' x t',YB;l x uXY,!' x uXl:'!'

")G:., t *X2:] x "X3?,', x uXl:lt: tu)(ps x ux|25.

The mole fractions are based on the analytical quantities, per 48oxygens:

0.25 - Mn0.25

The value 0.4 corrects for Mn and Fe in the U(1) and Al(3A)sites (0.25 and 0.15, respectively).

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

AppENDrx rABLE 1. Stoichiometric compositions of minerals"

Grad€Speclmen

s l 5 .420 5 .421 5 .386 5 .405 5 .4r1 5 .404 s .400 5 .362 5 .415 5 .375 5 .279 s ,42s 5 .337 s .384 5 ,394 5 .373 5 .405 5 .353 5 .348 5 .375- . A l 3 .394 3 .452 3 .514 3 .505 3 .530 3 .490 3 .530 3 .548 3 .544 3 .541 3 .559 3 .5?3 3 .525 3 .507 3 .514 3 .515 3 .517 3 .531 3 .591 3 .534! T l 0 .186 0 ,179 0 .166 0 .174 0 .159 0 .144 0 .154 0 .167 0 .161 0 .183 0 .182 0 .163 0 .203 0 .203 0 ,193 0 ,209 0 .278 0 .245 0 .1s8 0 .264E Fe 2 .236 2 ,789 2 .195 2 .439 2 .209 2 ,2s7 2 .69 t 2 .636 2 ,6u 2 .261 2 .183 2 .507 3 . I74 2 .490 2 .522 2 .6U 2 .310 2 .654 2 .848 2 .199; i l s 2 .52? 1 .938 2 .495 2 .252 2 .473 2 .497 1 .979 2 . r01 2 .U7 2 .370 2 . r01 2 .109 1 .416 2 .140 2 ,127 t .947 2 . r57 1 .958 1 .8r1 2 .318- i ln 0.027 0.009 0.013 0.009 0.013 0,018 0.022 0.004 0.004 0.0u 0.005 0.009 0.00't 0.009 0.009 0.018 0.013 0.009 0.009 0.013

K L .752 L . lL2 1 .783 1 .550 1 .637 r .758 1 .697 1 .628 1 .616 1 .774 r ,560 1 .694 1 .694 1 .721 1 .659 t ,706 L .1?4 1 .7r3 1 .698 1 .723l la 0 ,050 0 .027 0 .058 0 .080 0 .101 0 .$1 0 .055 0 .054 0 .085 0 .057 0 ,059 0 ,085 0 .087 0 .088 0 .085 0 .092 0 .031 0 .059 0 .095 0 .057t (0x) 94 .23 93 .15 94 .12 94 .76 95 .27 93 .11 93 .70 93 .72 93 .88 94 .45 93 .98 93 .85 94 .26 93 .87 94 .49 93 .95 93 .68 94 .30 94 ,84 95 .44

7.621 7 .632 7 ,638 7 .578 7 .699 7 .592 7 .521 7 .734 7 .660 7 .581 7 .505 7 .703 7 .s05 7 .617 7 .823 7 .59517.909 17 .898 17 ,911 17 .952 17 .$r 17 .938 17 .891 17 .795 17 .870 17 .949 18 .024 17 .827 18 ,006 17 ,913 17 .898 17 .9350.1 I0 0 .090 0 .114 0 .103 0 .080 0 .097 0 .093 0 .114 0 .086 0 .104 0 .080 0 .106 0 .105 0 .119 0 .083 0 .1153.036 3 .408 3 .300 3 .126 3 ,121 3 .374 3 .383 3 .201 3 .418 3 .422 3 ,662 3 .293 3 .441 3 .008 2 ,ss9 3 ,2490,610 0 .661 0 .790 0 ,623 0 .465 0 .577 0 ,561 0 .583 0 .567 0 .590 0 .443 0 .586 0 .609 0 ,519 0 .374 0 .5500.082 0 .038 0 .114 0 .101 0 .121 0 .029 0 .017 0 ,076 0 .045 0 .040 0 .0r7 0 .045 0 .054 0 .079 0 .101 0 .0540.069 0 .060 0 .037 0 .039 0 .270 0 .064 0 .060 0 .230 0 .037 0 .062 0 .062 0 .083 0 .050 0 .407 0 .202 0 . r73n . d . , n . d . n . d . n . d . n . d . n . d . 0 , 4 1 8 n . d . n . d . n . d . n . d . n , d . 0 . 2 5 9 n . d . 0 . 9 9 3 c n . d .3 .755u 3 .084 2 .677 3 .642 3 .437 3 .342 2 .903 2 .840 3 .272 3 . I85 3 .216 3 .289 2 .969 3 .291 3 ,312 3 .283

10uJT-10U3r100. 10 ry ry ry ee. 84 e93f-l0r.o9- g9-tr]0u.54- ry100. 51 sg;?T- es.87 10lr50-

46

4 t 5 * 51A 52 3A 61A 84 L29 114 5A 65 94 2t 137 53 l4A 6 23 79 4A

s i 6 .090 6 .045 6 .059 6 .002 6 .0r8 6 .049 6 . l l2 6 .063 6 .050 6 .045 6 .080 6 .055 6 .036 5 .083 6 .051 5 ,014 6 .015 6 .013 6 .017 6 .034- . A l 5 .535 5 .792 5 .706 5 .874 5 .826 5 .758 s .675 5 ,777 5 .812 5 .734 5 .704 5 .762 s .81s s .701 5 .737 5 .800 5 .776 5 .785 5 .786 5 .727I T i 0 .031 0 .023 0 .043 0 .023 0 .031 0 .027 0 .027 0 .023 0 .027 0 .043 0 .043 0 .035 0 .020 0 .047 0 .047 0 ,040 0 .043 0 .0s9 0 .062 0 .047E r " 0 .154 0 .103 0 .114 0 .078 0 ,089 o .o9r 0 . rzz 0 ,082 0 .078 o .106 o . toz o . ro t o . t io o .no 0 :098 0 : to6 o . tos o .og l o . l0z 0 .114I Ms 0 . r85 0 .079 0 .122 0 .074 0 ,085 0 .095 0 .110 0 .090 0 .070 0 . l l8 0 .105 0 .093 0 .059 0 .098 0 .098 0 .083 0 . r r7 0 .090 0 .078 0 .1345 K 1 .62 i 1 .599 1 .679 r .444 1 .490 1 ,625 1 .525 1 .533 1 .418 1 .670 1 .662 1 .542 1 .524 r .559 1 .545 1 .594 1 .559 1 .590 1 .540 r .727- ffa 0.252 0.336 0.255 0.488 0.409 0,328 0.264 0.388 0.490 0.248 0.235 0.367 0.406 0.352 0.40{ 0.356 0.247 0.334 0.387 0.225

Ba 0 .008 0 .004 0 ,012 0 .007 0 ,019 0 .004 0 .008 0 ,008 0 .027 0 ,016 0 .012 0 .007 0 .012 0 .008 0 .012 0 .016 0 .016 0 .012 0 .008 0 ,015t (0x) 94 .14 93 .54 94 .71 94 .54 94 .89 93 .50 94 .12 94 ,24 94 .03 94 .56 94 .59 94 ,75 93 .97 94 .09 94 .43 93 .88 94 .88 94 .23 94 .74 94 .s6

a ln 0 .554 0 .554 0 .556 0 .698 0 .687 0 .654 0 .683 0 .73s 0 .810 0 .658 0 ,731 0 .732 0 .84 i 0 .781 0 .754 0 .73r 0 .673 0 .749 0 .763 0 .604n. - p rp 0 .078 0 .051 0 .058 0 .081 0 .107 0 .101 0 .066 0 .081 0 .094 0 .105 0 .078 0 .100 0 .059 0 .109 0 .109 0 ,084 0 .113 0 .105 0 ,065 0 .103F f i sps 0 .333 0 .247 0 .286 0 . r74 0 .157 0 .204 0 .218 0 .128 0 .035 0 .168 0 .124 0 .115 0 .074 0 .075 0 .094 0 .169 0 .144 0 ,105 0 .123 0 .211,i.E srs 0.035 0.048 0.090 0.048 0.038 0.041 0.033 0.055 0,060 0.059 0.067 0.05? 0.026 0.034 0.034 0.016 0.070 0.041 0,049 0.083- - t (0x) 99 .78 98 .49 99 .92 100.45 100.42 99 .40 98 .95 100.73 99 .40 100.39 99 .69 99 .80 100.45 99 .73 100.74 100.15 99 .80 100.37 100.18 100.86

- : a ln 0 ,600 0 .583 0 .671 0 .750 0 .587 0 ,665 0 .583 0 .790 0 .817 0 .666 0 .748 0 .732 0 .862 0 .788 0 .759 0 .733 0 .676 0 .760 0 .769 0 .6321J I p rp 0 .087 0 .057 0 .104 0 .095 0 .107 0 .101 0 .066 0 .084 0 .091 0 .109 0 .084 0 .100 0 .063 0 .100 0 ,104 0 .077 0 .156 0 .099 0 .067 0 .112F E sps 0 .279 0 .210 0 .150 0 .1 i7 0 ,167 0 . i90 0 .218 0 ,067 0 .032 0 ,161 0 .098 0 .116 0 .053 0 .080 0 .092 0 .169 0 ,139 0 ,110 0 .113 0 .184, iE grs 0 .034 0 .050 0 .075 0 .038 0 .038 0 .044 0 .033 0 .059 0 .060 0 .065 0 .059 0 .052 0 .022 0 .032 0 .035 0 .021 0 .059 0 .03r 0 .051 0 .072

t (ox) 99 .83 99 .21 100.41 100.71 100.42 99 .36 98 .95 99 .94 99 .52 100.45 100,52 99 .80 100.32 100.25 100.5s 100.19 100.79 100,49 100,48 100.91

slAI

- _ n! F e

e ; ;Er;6 L l

Ht ( 0 x )

r i lE 0 .906 0 .973 0 .955 0 .974 0 .948 0 .929 P2 pyr 0 .094 0 .027 0 .045 0 ,019 0 .052 0 .0505 Sei 0.000 0.000 0.000 0.000 0.000 0.001t hen 0.000 0.000 0.000 0.007 0.000 0.020i t(0x) 98.44 98.98 98,65 99.52 98.15 98.68

. a n Po a D: o r* r ( o x )

5 .105 5 .142s .814 5 .8280.012 0 .0134.516 5 .8474.484 3.0650,019 0.0060.000 0.000

86.55 86 ,98

0.977 0 .982 0 .963 0 .970 0 .961 0 .994 0 .972 0 .977 0 .941 0 .939 0 .97r 0 .976 0 .9280.013 0 .005 0 .030 0 .022 0 .018 0 .005 0 .0r8 0 .022 0 .0s9 0 .060 0 .027 0 .024 0 .0600.000 0.003 0.000 0.000 0.006 0.000 0.000 0.001 0.000 0.00r 0,00r 0.000 0.0050.010 0.010 0,007 0.008 0.0r5 0.000 0.010 0.000 0.000 0.000 0.00r 0.000 0.007

98.80 99 .24 99 .64 98 .89 99 .41 98 .92 99 ,92 98 .61 97 .92 91 ,36 98 .49 98 .54 98 .99

s l 5 .225 s .157Al 5 .549 5 .808

$ r i 0 .012 0 .013'i Fe 3.887 5.029I l ls s.065 3.8876 l,tn 0.061 0.019

Ca 0.006 0.000l (0x) 85 .89 86 .28

5 . 1 3 1 5 . r 3 r 5 . 1 7 2 5 . 1 5 6 5 . 1 4 3 5 . 1 2 35.801 s .793 5 ,714 5 .770 5 .771 5 .8550.012 0 .012 0 .012 0 .025 0 .012 0 .0054.282 3 .838 3 .99s 4 .87 I 4 .7 t0 4 .7314.697 5 . r07 s .010 4 .030 4 ,296 4 .1930.012 0 .042 0 .037 0 .0s7 0 .012 0 .0060.006 0.000 0.000 0.000 0.000 0.000

87.04 86.85 85.91 85.72 87.04 86.29

0.388 p0 .6r10 .001

100.58

P 0 .3400.5580.002

100.48

HOLDAWAY ET AL.: METAMORPHISM IN WEST-CENTRAL MAINE

APPENDf x r ABLE 1 - Continued

4'l

Gradespeclmn t9

* 6 . 5 rr40 77-3 77-2 90 76 56 91 73 87

Ii43

r 814s 86 11

s i 5 .060 6 .016 6 .014 5 .027 6 .053 5 .0s1 6 .053 6 .037 6 .053 6 .026 5 .037 6 .071 6 .063 6 .035 6 .055Af s .704 s .743 5 ,775 5 .770 5 ,700 5 .597 s .603 5 .716 5 .665 5 .715 5 .706 5 .679 5 .705 5 .680 5 .530

o T i 0 .048 0 .053 0 .063 0 .048 0 .065 0 .055 0 .075 0 .067 0 .081 0 .081 0 .083 0 .072 0 .064 0 .091 0 .103f re 0 .110 0 .115 0 .090 0 .106 0 .115 0 .110 0 .154 0 .115 0 .115 0 . t t0 0 .114 0 .117 0 .110 0 .134 0 .125: l . fq 0 .126 O. lO3 0 .098 0 .099 0 .107 o . l l8 0 .150 0 ,107 0 .119 0 .114 0 .122 0 . t25 0 .098 0 .103 0 .1259 x - r .668 1 .646 1 .585 1 .s85 r , lL i t . i4B t , ' tg2 r .7 r7 1 .702 1 .666 1 .674 L .749 r .698 1 .792 1 .816i Na 0 .2G5 0 .301 0 .326 0 .328 0 .206 0 .20s 0 .163 0 .206 0 .2s8 0 .251 0 .236 0 .137 0 .236 0 .1s0 0 .137

Ba 0.008 0.016 0.008 0,008 o.ol2 0.015 0.020 0.012 0.015 0.020 0.008 0.004 0.000 0.008 0.008t (0x) 93 .95 94 .07 94 .18 93 ,71 93 .81 94 .58 94 .32 93 .81 93 .95 94 .79 94 .35 94 .93 94 .21 94 .42 95 .o4

0.909 0 .978 0 ,895 0 .887 0 .8910.068 0 .017 0 .067 0 .111 0 .0860.001 0.000 0.000 0,000 0.0010.022 0.005 0,037 0.002 0.022

99.92 99 .59 100.97 99 .15 100.29

5t

AIE T iE F eo i l ;

KNA[ ( 0 x )

a l n

E g s p s38 g r s

t (0x)

d a lmE l p rpi 6 s p s3E g t s

- t ( 0x )

S o r: a bl a n5 . .t (0x )

o l ln 0 ,957 0 .969# pyr 0.041 0.028E gei 0.000 0.003! hm 0.002 0.000- t (0x) 99 .49 99 ,55

, an 0 .293 0 .1553 ab 0 .705 0 .832E or 0.002 0.002

t (0x) 100.88 100.39

5.372 5 .361 5 .397 5 .369 5 ,365 5 .361 5 .3763.527 3 .465 3 .727 3 .489 3 .457 3 .525 3 .3960.158 0 .243 0 .2 t7 0 .298 0 .265 0 .257 0 .3262.343 2.533 2.402 2.693 2.725 2.425 2.2982.357 2.054 1.855 1.830 r.A75 2.064 2.2440.009 0 .013 0 .023 0 .004 0 ,013 0 .027 0 .013r .711 1 .709 1 .698 1 .712 1 .795 1 .812 1 .8430.071 0 .072 0 .073 0 .072 0 .054 0 .052 0 .049

94.84 94 .79 93 .93 94 ,69 94 .50 95 .49 94 .96

0.663 0 .709 0 .591 0 .764 0 .711 0 .629 0 .6410.119 0 .099 0 ,096 0 .099 0 .091 0 .101 0 .1310.155 0 .158 0 .181 0 . l l5 0 .150 0 .219 0 .1720.052 0,024 0.032 0.022 0.048 0.051 0.056

100.89 100.04 r00.54 100.06 100.05 101.73 100.94

5.073

0.0740 . 1 3 10.1301.8750.0830.004

94.56

5.374 5 .356 5 ,354 5 .354 5 .343 5 .353 5 .325 5 ,351 5 .3443.397 3 .473 3 .44s 3 .409 3 .4 i2 3 .386 3 .435 3 .4r3 3 .4550.27 ! 0 .292 0 .281 0 .337 0 .360 0 .335 0 .399 0 .419 0 .3562.579 2 .463 2 .529 2 ,531 2 .543 2 .505 2 .658 2 .443 2 .5112.090 2 .0s0 2 ,087 1 .986 1 .967 1 .959 1 .749 1 .878 r .8970.013 0 .013 0 .013 0 .022 0 .018 0 .020 0 .022 0 .022 0 .009r .746 1 .712 1 .738 r .779 1 ,805 1 .834 r .872 r .872 1 .9110.086 0 .057 0 .071 0 .067 0 .045 0 .052 0 .036 0 .044 0 .027

94.35 94 .65 95 .22 94 ,69 94 .85 94 ,73 94 .59 95 ,45 94 .51

0.573 0 .7r0 0 .704 0 .581 0 .685 0 .704 0 .595 0 .675 0 .7310 . 1 1 2 0 . 1 2 1 0 . 1 2 3 0 . 1 1 2 0 . 1 1 2 0 . 1 1 2 0 . 1 0 2 0 . 1 0 8 0 . 1 2 60.185 0 .143 0 ,147 0 .187 0 .177 0 .146 0 ,177 0 .184 0 .0820.029 0 .02s 0 .02s 0 .020 0 .025 0 .038 0 .024 0 .033 0 .051

99.99 10 l .OO 101.02 99 .91 r00 .15 100.37 99 .51 100.32 100.10

5.072 6 .055 5 .072 5 .0385.652 s .531 5 .666 5 .6530.071 0 . r08 0 .060 0 .1110 . r 1 7 0 , 1 2 6 0 . i 3 4 0 . 1 1 80 . 1 0 8 0 . r 1 0 0 , 1 1 8 0 . 1 2 51.940 1 .816 1 .811 1 ,8530.126 0 .130 0 .114 0 .0900.004 0.012 0.008 0.008

94.45 94.38 94.7s 95.26

5,331 5 .349 5 ,312 5 .3593.446 3 .485 3 .457 3 .5050.346 0 .407 0 .364 0 .3312.549 ? .5U 2 .579 2 .3281.898 1 ,679 1 .899 2 .0500.018 0 .013 0 .009 0 .0131,939 1 .895 1 .891 1 .8520.035 0 .036 0 .027 0 .031

94,19 94 .70 95 .25 94 .65

0.751 0 .755 0 .778 0 .7020.130 0 , r14 0 .145 0 . r800.091 0 .099 0 .052 0 .0550.028 0 ,021 0 .0r5 0 .053

99.48 100.04 100.34 100.85

0.674 0 .739 0 .595 0 .785 o .7rg 0 .542 0 .662 0 .701 0 .125 o .7 tg 0 .698 0 .706 0 .719 0 .713 0 .681 0 .734 0 .750 0 .765 0 .783 0 .717o. r rz o .ogo 0 ,084 0 ,096 0 .084 0 .093 0 . r28 0 .106 0 .115 0 .116 0 .112 0 .107 0 .102 0 .097 0 ,101 0 .115 0 .114 0 .114 0 .128 0 .1610.155 0 .139 0 .201 0 .099 0 .148 0 .218 0 .151 0 .165 0 .133 0 .138 0 .170 0 .162 o . l4 l 0 .157 0 .187 0 .089 0 .107 0 .099 0 .073 0 ,062o.oso o :o2G 0 :oz0 0 .ozo 0 .049 0 .047 0 .0s9 0 .029 0 .027 o .oz7 0 .020 0 .025 0 .038 0 .023 0 .032 0 .062 0 .029 0 .021 0 .015 0 .060

101.05 100.52 100,25 100.45 100.30 101.49 roo .68 100.28 100.80 100.92 99 .48 100.54 100.38 100.05 r00 .63 100.32 99 .43 100.04 100.57 101.21

0,840 0 .899 0 ,862 0 .835 0 .883 0 ,8880.152 0 .093 0 .129 0 .146 0 .101 0 .1010.000 0-000 0.000 0.000 0.000 0.0000.008 0 .008 0 .009 0 .0 I8 0 .016 0 .011

100.39 99 .94 99 .58 99 ,30 99 .16 99 .77

0.968 0 .961 0 .9520.032 0 .026 0 ,0270.000 0.003 0.0030,000 0 .010 0 .018

99.45 99.85 100.70

0. r8 i 0 .179 0 .180 P0.817 0 .817 0 .8170,002 0.004 0.003

99.73 99 .92 r00 ,47

0.933 0 .954 P0.066 0.0440.000 0,0000.001 0.002

99.25 98 ,99

0,195 0 .265 0 .157 0 .230 0 .5290.798 0 .732 0 .836 0 .751 0 .4630.007 0.003 0.007 0.009 0,008

100.49 100.35 r00 .53 100.25 100.41

P

P 0 . 1 1 5 0 . 4 7 90.879 0 .5150.006 0.006

100. r0 100.94

0.297 0 .314 0 .3890.701 0 .583 0 .5080.002 0.003 0.003

100.55 100.78 101.09

astoichimtry is based on 22 orygens for nicas, endrember proportions for garnets, feldspars, and ilmnite, 28 orygens for chlorite, and 25.53 (si+Al) for.staurolite. P lndicates nineral-present but not abundant enough for anatyiis. A after nurber indlcates speciren contains tl2 andalusite.ova lues under l ined are aDDrox iMte ca lcu la ted va lues fo r (H+L i ) -and [ (0x) . I f L i i s assuned to be 0 .3 p fu , the reminder l s -an approx imte lnd ica t ion o f H.cll content of thls staul.blite is about 1.39 pfu and variable. The nulser used here is from a mre accurate AA analysis from an otheruise identicalstaurolite from the sare outcrop (Dutror et al., 1986).

Note added in proof. We have recently observed that the four andalusite-bearing rocksof grade 4 (App. Table l, Fig. 1) from the Augusta quadrangle are compositionally distinctfrom all other rocks in grade 4. These andalusite-bearing rocks contain higher Mg/Fe instaurolite, biotite, and garnet and higher spessartine + grossular in garnet, indicating thatthey represent the compositional limit of staurolite stability. Because these specimens arecompositionally unique, the andalusite need not have an M, origin. Other evidence (e'g.,local cordierite) suggests increasing P during metamorphism. Novak and Holdaway ( 198 l)suggested that M, and M3 tend to merge in time and metamorphic conditions in this area.The probability that andalusite formed or re-equilibrated during M, in the Augusta areain no way affects the conditions of metamorphism deduced for the region. This matter willbe considered in detail in a future contribution.