359

The Canadian MineralogistVol. 37, pp.359-373 (1999)

KYANITE IN THE WESTERN SUPERIOR PROVINCE OF ONTARIO:IMPLICATIONS FOR ARCHEAN ACCRETIONARY TECTONICS-

YUANMING PAN'

Department of Geological Sciences, University of Saskatchewan, Saskatoon, Saskatchewan S7N 5E2

MICHAEL E. FLEET

Department of Earth Sciences, University of Westem Ontario, London, Ontario N6A 587

ABSTRACT

Thirteen occunences of kyanite in Archean supracrustal rocks of the western Superior Province of Ontario are grouped intothree distinct lithotectonic associations: I) metapelites close to subprovince and terrane boundaries, II) metapelitei near faultswithin greenstone belts, and III) metamorphosed A1-Si-rich alteration assemblages associated with volcanogenic massive sulfide(VMS) mineralization in greenstone belts. Kyanite of groups I and II is commonly associated with staurolite and texturallypredates the main assemblage of metamorphic minerals (Sil + Grt + Bt + Pl + Qtz), indicating that a medium-P, low-T Barrovianmetamorphism occurred before the main, regional low-P/T metamorphism that characterizes the western Superior Province. Thisearlier Barrovian metamorphism (670-730 MPa, 500-560"C) was followed by significant unloading (up to 12 km at the Quetico-Wabigoon boundary) before the development of the main penetrative deformation and the regional low-P/T metamorphism Theresults of geothermobarometry and the restricted locations of group-I and group-Il kyanite are interpreted to reflect crustal thick-ening as part of accretionary tectonic processes for the collage of individual greenstone belts and subprovinces for the forrnationof the western Superior Province. In particular, the three occurrences of kyanite along the Quetico-Wabigoon boundary supporlthe origin of the Quetico Subprovince as an accretionary wedge during norlhward subduction. Group-III kyanite, on the otherhand, reflects unusual whole-rock compositions and hydrothermal fluids, and may be a potentially useful exploration tool forVMS deposits

Keywords: kyanite, occunences, Barovian metamorphism, Archean, accretionary tectonics, western Superior Province, Ontario.

Sovu,qrnB

Nous nous proposons de regrouper treize indices de kyanite dans les roches arch6ennes d'origine suplacrustale du secteurouest de la Province du Sup6rieur en Ontado en trois associations lithotectoniques distinctes: I) mdtap6lites proches de la bordured'une sous-subprovince ou d'un socle, II) m6tap6lites situ6es prbs de failles dans des ceintures de roches vertes, et III) rochesm6tamorphiques enrichies en Al-Si par alt6ration hydrothermale prbs des gisements de sulfures massifs volcanog6niques dansdes ceintures de roches veftes La kyanite des groupes I et II, gdn6ralement associ6e d la stauroiite, est ant6rieure selon des critdrestexturaux d I'assemblage principal de min6raux m6tamorphiques (Sil + Glt + Bt + Pl + Qtz). Ce fait montre que 1'6pisode dem6tamorphisme barrovien de pression moyenne et de faible temp6rature a pr6c6d6 l'6pisode principal de mdtamorphisme r6gionalde fatble pression et de faible temp6ratue qui caract6rise le secteur ouest de la Province du Sup6rieur Cet 6pisode barrovien(670-730 MPa, 500-560'C) fut suivi par une 6rosion importante (usqu'd 12 km de soulbvement i I'interface entre les socles deQuetico et Wabigoon) avant le d6veloppement de 1'6pisode important de d6formation r6gionale et de m6tamorphisme de faibletemp6rature et de faible pression Les r6sultats d'une analyse g6othermobarom6trique et la rdpartition des indices de kyanite desgroupes I et II assez reskeint refl6teraient un 6paississement de la cro0te suite aux processus tsctoniques accompagnant I'accr6tiond'un collage de ceintures individuelles de roches vertes et de sous-provinces pour former le secteur ouest de la Province duSup6rieur En particulier, trois indices de kyanite provenant de I'interface entre les socles de Quetico et de Wabigoon dtayentI'hypothbse d'une origine de la Sous-province de Quetico comme prisme d'accr6tion lors de la subduction d polarit6 vers le nordEn revanche, les indices de kyanite du groupe III signaleraient la pr6sence de roches de composition particulidre due d I'alt6rationhydrothermale, et seruiraient donc de guides utiles dans les programmes d'exploration pour de gisements de sulfures d'originevolcanog6nique

(Traduit par la R6daction)

Mots-clds: kyanite, indices, m6tamorphisme barrovien, arch6en, tectonique d'accr6tion, Province du Sup6rieur (secteur ouest),Ontario.

* LITHOPROBE contribution number 1030.I E-mail address: yuanming pan@sask.usask ca

360 THE CANADIAN MINERALOGIST

INrnooucrroN

There is a growing consensus among investigators(Card 1990, Thurston & Chivers 1990, Williams 1990,Williams et al. 1991, Percival et al. 1994) that theArchean Superior Province of North America was as-sembled by accretionary plate-tectonic processes, analo-gous to those ofthe modem convergent-plate boundariesof the western Pacific (Hamilton 1988, and referencestherein). Plate-tectonic activity in the Archean is sup-ported by geochronological, structural, and geochemr-cal evidence, and broad-scale metamorphic patterns ofthe Uchi, English River, Wabigoon, Quetico and Wawasubprovinces of the western Superior Province(Thurston & Breaks 1978) and the Abitibi subprovince(Jolly 1978). Calvert et al. (1995) provided compellingevidence for subduction based on seismic reflection pro-files across the Opatica-Abitibi boundary in Quebec.Hamilton (1998), on the other hand, maintained thatplate-tectonic processes did not operate in Archean time.

The western Superior Province, which is character-ized by a well-defined distribution of alternating, east-

west-trending granite-greenstone and metasedimentarybelts (Fig. l), is a classic area where models of platetectonics for the formation of Archean crust were firstintroduced (Krogh & Davis 1977,Langford & Morin1976, Blackburn 1980). Therefore, knowledge of thedistribution of metamorphic grade, conditions of peakmetamorphism, and metamorphic pressure - tempera-ture - timing - deformation history in the Westem Su-perior Province is critical to our understanding of thetectonic evolution of the Archean Superior Province. lnthis contribution, we examine the reported occurrencesof kyanite in the Western Superior Province (Table I,Fig. l) and attempt to provide metamorphic evidencefor and P-T constraints on Archean accretionary tec-tonics responsible for the collage of individual green-stone belts and subprovinces in the Western SuperiorProvince (Percival & Williams 1989, Williams 1990).

DescnrprloN oF KyANITE Occunnntvcss

Thirteen occuffences of kyanite have been reportedin the Archean supracrustal rocks of the Western Supe-

TABLE I SIA4MARY OF KYAMTE OCCI,'RRENCES IN THE WESTERN ST'PERIOR PROVINCE

No Iocation Lithotectonic Association Mineral Assemblage References

Group I: Metapelites nea subprovince bomdaries

I Rainy Lake,Minnesota

Metapelites of the Quetico subprovinmneil the Quetico-Wabigoon boundary

Ky, St, Sil, Grt, Ms, Bt, Pl, Qtz,Ihn

1

Ky, St, Sil, Gt, Ms, Bq Pl, Qtz, Iln

Ky, And, Crd, Sl GG Qt4 Pl, Bt

Tabor (1988)

Moore (1940)

Percival et al (1985)This study

Breaksetal (19E4)ThNton et al (1991)

Ayres (1978)

Patterson et al (1984), Burk et al(1986), Pil & Fleet (1993; 1995)

Frmklin et al (1975)

Lefebue (1982)

Ayres (1978), Amukun (1989)

Osterberg et al (1987)

Breaks (1991)

Schmdl et al (1995)

Burk etal (1986). Kuhs et al (1994)

2 Abbess Lake, Metapelites ofthe Quetico subprovinceAtikoka nedtheQuetico-Wabigoonboudary

3 Raith Metapelites ofthe Quetico subprcvincene{ the Wawa-Quetico boudary

4 North Cdibou metapelites ned the tectonic boundary ofthegreenstone belt Norti Cdibou terue

5? Favourable Lake neil the Bw Head Fault

Group II: Metapelites osociated with faults within greenstone belts

6 WhiG River-Hmlc metapelites of tlle HemlGHeron Bay greetrtoneBlack fuver beltneatheHemloFaultzone

8 Savrut lake Al-Si-rich alteration osociaied with VMS showins

9 Mashall Iake Al-Si-rich alteration osociated with VMS showinss

l0 Onmm Headway-Coulee mdsive sulfide o@urgnce

11 Melchett quartz veins ss@iated with VMS showings

12 Mmitouwadge Al-Si-rich alteration csociated with VMS deposit

13 Heqllo quartz veins at tlle Hemlo gold deposit

Ky, St, Sil, tu, Qtz, Pl, Bt,Ihn

Goup III: Al-Si-rich alteration ssemblages md veins dsociated with VMS md Au deposits in greenston€ belts

7 Sturg@n Iake quartz veins dsociated with VMS deposits Ky, Qtz, Prl, Ms, And

Ky, Sil, And, St, Ms

Ky, Qtz, Sil, Om, Crd

Ky, Qtz, Ard, Ser

Kv, Qtz

Ky, Sil, Ms, Bt, Pl, Qtz

Ky. Otz. Ms. Tur. Py

SeetextfordiscussionontlleFavourableLakeoccunence MineralsymbolsafterKretz(1983) VMS,volcmogenicmdsivesulfidedeposits.

KYANITE IN T}IE WESTERN SUPERIOR PROVINCE

Frc. 1 Occurrences of kyanite in the Western Superior Province [classification ofsubprovinces after Card & Ciesielski (1986); except for the2.9-3.0 Ga North Caribouterrane, after Thurston et al. (1991), dashed linesl. Numbers correspond to Table 1.Group I: l-3) Rainy Lake, Abbess Lake (Atikokan) and Raith areas, respectively, of theQuetico Subprovrnce,4) North Caribou greenstone belt near the northem tectonicboundary of the North Caribou terrane, and 5) Favourable Lake greenstone belt (seetext for discussion). Group tI: 6) the Hemlo - Heron Bay greenstone belt of the Wawasubprovince. Group III: 7-10) Sturgeon Lake, Savant Lake, Marshall Lake and Onamanareas, respectively, of the Wabigoon Subprovince, 1 1) Melchett greenstone belt of theEnglish River subprovince, and 12-13) Manitouwadge greenstone belt and Hemlo golddeposit, respectively, of the Wawa Subprovince.

361

riorProvince (Fig. I, Table 1). Ayres (1978) listedfourofthese; three are located in amphibolite-facies terranesnear subprovince boundaries, and the fourth one is foundin quartz veins associated with metamorphosedvolcanogenic massive sulfide (VMS) deposits. Table Ishows that eight of the nine additional occurrences ofkyanite belong to these two groups, whereas the remain-ing one constitutes a new group in which kyanite oc-curs in metapelites in close proximity to a major fault ingreenstone belt (Table 1, and see below). Also, new datafrom Amukun (1989) and Thurston et al. (1991) war-rant a re-appraisal of two of Ayres's (1978) originaloccurrences (1.e., Favourable Lake, 5 in Fig. 1; MarshallLake, 9 in Fig. 1).

Group I: lqtanite in metapelites near subprovince andterrane boundaries

This group includes three occurrences [Rainy Lake,Abbess Lake (Atikokan), and Raithl in the QueticoSubprovince near the Quetico-Wabigoon boundary(Moore 1940, Percival et al.1985, Tabor 1988; 1-3 inFig. 1), and another one in the North Caribou green-stone belt close to the tectonic boundary of the NorthCaribou terrane (4 in Fig. 1; Thurston e/ al. 1979,1991,Breaks er al. 1984).

The boundary between the metasedimentary QueticoSubprovince and the granite-greenstone WabigoonSubprovince in the Atikokan area (2inFrg. 1) coincides

362 THE CANADIAN MINERALOGIST

; ' : \{:

10 .5 mm I

FIG. 2 Photomicrographs illustrating textural relationships and associated minerals in kyanite-bearing metapelites from theRaith (a--e) and Hemlo (f-g) areas of the Western Superior Province. a) Kyanite poikiloblast (Ky); note that the includedgamet (Grt), albeit partially surrounded by quartz and plagioclase, is locally in direct contact with kyanite. b) Kyanite inclu-sion associated with ilmenite in staurolite porphyroblast (St). c) Kyanite intergrown with staurolite d) Inclusions of kyanite,staurolite (outlined) and biotite in garnet porphyroblast. e) Part of a large garnet porphyroblast (1.2 cm in diameter) illustrating a marked discontinuity between an inclusion-rich core (mainly quartz and plagioclase) and an inclusion-bearing rim(mainly magnetite and ilmenite). 1) Staurolite poikiloblast replaced by biotite + sillimanite assemblage in the hanging-wallmetapelites. g) Kyanite porphyroblast and associated staurolite, discordant with the main foliation (S2; outlined) in the hang-ing-wall metapelites, h) Fibrous sillimanite after kyanite in the hanging-wall metapelites.

' j . : 1 , r I l : . , . lt , i "

KYANITE IN THE WESTERN SUPERIOR PROVINCE JOJ

with the Quetico Fault (a regional-scale dextral shear-zone; Williams 1991) that swings into the QueticoSubprovince just west of Raith (3 in Fig. 1). Moore(1940) reported kyanite in metasedimentary rocks north-west of the Abbess Lake in the Atikokan area (2 inFig. 1), but did not provide any petrographic descrip-tion or information on the mineral assemblage. Percivalet al. (1985) reported relict kyanite in plagioclase froma garnet - sillimanite - biotite - plagioclase - qnaftzschist at Raith. Two of eight metapelitic samples fromthe Raith area collected by the first author contain kya-nite and have a similar assemblage of minerals. Kyanitein these two samples occurs not only as fragments inplagioclase, but also as porphyroblasts or poikiloblasts.The kyanite poikiloblasts are discordant with respect tothe main penetrative foliation and contain inclusions ofgarnet (Fig. 2a), biotite, plagioclase, quartz, muscoviteand ilmenite. Kyanite also has been found as inclusionsin or intergrown with staurolite porphyroblasts (Figs. 2b,c), which have an assemblage of mineral inclusionssimilar to the kyanite poikiloblasts. In addition, kyaniteand staurolite have been found as inclusions in garnetporphyroblasts (Fig. 2d). Sillimanite, not in texturalequilibrium with kyanite (Percival et al. 1985), is in-variably fibrolitic and is present mainly in knots alignedparallel or subparallel to the main penetrative foliation.Other minerals in apparent textural equilibrium withsillimanite include garnet porphyroblasts and biotite,plagioclase and quartz in the matrix. Although a few ofthe largest porphyroblasts of garnet (up to 1.5 cm indiameter) show a textural zonation with a marked dis-continuity in the internal trails of inclusions (Fig.2e;indicative of two episodes of gamet growth), most gar-net porphyroblasts have intemal trails ofinclusions con-sistent with growth during the development of thepenetrative foliation and, therefore, are part of the mainmetamorphic (sillimanite-stable) assemblage.

In the Rainy Lake area, northem Minnesota (1 inFig. 1), the Rainy Lake - Seine Lake Fault marks theboundary between the Quetico Subprovince to the southand the Wabigoon Subprovince (including the RainyLake Wrench Zone, Poulsen 1986) to the north. Tabor(1988) reported that kyanite occurs in metapelites closeto the Rainy Lake - Seine Lake Fault at the northeast-em corner of Kabetogama Peninsula and increases rnabundance from north to south away from the RainyLake - Seine Lake Fault to a maximum at the first an-pearance of sillimanite. Kyanite occurs as porphyro-blasts or poikiloblasts showing preferred orientationparallel to 51, as are internal trails ofinclusions (Tabor1988). Mineral inclusions in kyanite include muscovite,staurolite, biotite and garnet. Tabor (1988) also notedthat some porphyroblasts of kyanite at Rainy Lake arerotated by 52 crenulations and that, where both arepresent, kyanite and sillimanite are not in contact.

The North Caribou greenstone belt is situated closeto the northern tectonic boundary of the 2.9-3.0 GaNorth Caribou terrane (4 in Fig. 1), and consists of four

supracrustal assemblages: the 2980 Ma Aguta Arm as-semblage, the Keeyask assemblage, the McGruer as-semblage, and the Eyapamikama assemblage (Thurstonet al. l99l). Thurston et al. (1979) reported kyanite rnmetasedimentary rocks from the southern part of theNoth Caribou greenstone belt at the ForesterLake area.Breaks e/ al. (1984) reported kyanite in metapelites ofthe Eyapamikama assemblage at the Miskeesik Lakearea (Table 1). Breaks & Bartlett (1991) observed twogenerations of kyanite at Miskeesik Lake: kyanite-l asragged poikiloblasts intergrown with staurolite in ag-gregates that deflect the enclosing phyllosilicate-definedfoliation (c/. Figs. 2f, g below), and kyanite-2 truncat-ing the regional foliation. Breaks & Bartlett (1991) alsonoted the presence of kyanite as an epitactic replace-ment after andalusite at the Miskeesik Lake area. It isnoteworthy that the Miskeesik Lake occurrence of kya-nite is also within 0.5 km of a fault boundary betweenthe Eyapamikama and McGruer assemblages.

Ayres (1978) classified an occurrence of kyanite atthe Favourable Lake greenstone belt (5 in Fig. 1) as amember of Group I, because of its location near the BearHead fault (l.e., the Berens River - Sachigo Subprovinceboundary in Card & Ciesielski 1986; Fig. 1). Thurstonet al. (1991) re-interpreted the Berens River Sub-province to represent a2720Ma magmatic welt duringnorthward subduction at the Uchi - English RiverSubprovince boundary. The Bear Head fault is then nolonger a subprovince boundary, but is a likely ensialicstructure that was active atabout2.'l Ga within the2.9-3.0 North Caribou terrane (Thurston et al. 1991). Un-fortunately, classification of this Favourable Lakeoccurrence ofkyanite into Group II or III is not possibleowing to the lack of petrological data.

Group II: l<yanite in metapelites associatedwith faults within greenstone belts

There is one example of this group: the Hemlo -

Heron Bay greenstone belt of the Wawa Subprovince(6 in Fig. 1; Patterson et al. l984,Bvket al. 1986, Pan& Fleet 1993). The Hemlo - Heron Bay greenstone beltconsists of two supracrustal assemblages (i.e., theHemlo - Black River assemblage to the north and theHeron Bay assemblage to the south) separated by theHemlo Fault Zone, which is part of the regional LakeSuperior Fault Zone (Williams et al. l99l). Figure 3illustrates the close spatial association ofkyanite occur-rences with the Hemlo Fault Zone in the Hemlo - HeronBay greenstone belt. The kyanite-bearing metapelites atthe White River property (6-1 in Fig. 3) are found withina turbiditic metasedimentary sequence (Pan et al. l99l),as are the kyanite-bearing rocks at the Padre Resourcesproperty (6-3 in Fig. 3;Patterson et al. 1984). There aretwo distinct types of kyanite occurrences at the Hemlogold deposit (6-2 in Fig. 3): one in hanging-wall andfootwall aluminous rocks, and the other in quartz veinswithin the orebodies (Burk er al. 1986, Kthns et al.

364 THE CANADIAN MINERALOGIST

FIc. 3. Details of kyanite occurrence #6 of Figure 1. Kyanite in metapelites of the Hemlo- Black River assemblage in the Hemlo - Heron Bay greenstone belt (Group II): 6-1)White River property, 6-2) Hemlo deposit, and 6-3) Padre Resources property nearBlack River. Note the close association with the Hemlo Fault Zone.

1994,Pan & Fleet 1995). We consider these hanging-wall and footwall aluminous rocks at the Hemlo golddeposit to be metapelites because of their associationwith other metasedimentary rocks, similarity in stratig-raphy to metapelites along strike at the White Riverproperty and the Padre Resources property (Pan & Fleet1995), and distinctive geochemical characteristics (Fleetet al. 1997), although Kuhns et al. (1994) interpretedthem to represent metamorphosed rocks that had under-gone porphyry-style argillic alteration.

Kyanite-bearing metapelites at the White River prop-erty and Hemlo gold deposit typically contain kyanite,sillimanite, gamet, staurolite, biotite, muscovite, plagio-clase, and quartz (Burk et al. 1986, Kuhns et al. 1994,Pan & Fleet 1993,1995). Sillimanite, mainly "fibrolite",usually occurs in knots oriented parallel or subparallelto the main penetrative foliation. Kyanite, on the otherhand, typically occurs as prismatic porphyroblasts andpoikiloblasts discordant with respect to the main pen-etrative foliation and is commonly associated with stau-rolite poikiloblasts (Figs. 2f, g). Staurolite poikiloblastsin augen are commonly replaced by a sillimanite + bi-otite + muscovite + plagioclase + quartz assemblage(Fig.2f). The mineral inclusions in kyanite and stauro-lite poikiloblasts are similar and include garnet, plagio-clase, quartz, biotite, ilmenite, zircon, apatite and pyrite.Authors ofprevious studies have emphasized that kya-nite and sillimanite at Hemlo and the White River prop-erty, where they commonly occur together, generally do

not show any replacement textures (Burk el al.1986,Pan & Fleet 1993, Kuhns et al.1994). In this study, ir-regular grain boundaries have been observed on kyaniteporphyroblasts in direct contact with "fibrolite" knots(Fig. 2h), and may be related to resorption of the former.Kyanite and staurolite from the Hemlo gold depositcommonly exhibit extensive alteration to muscovite,chlorite and epidote along grain fractures and bound-aries, but are less affected by such alteration at the WhiteRiver property. It is noteworthy that petrographic de-scriptions of kyanite and associated minerals in alumi-nous rocks (our metapelites) of the Hemlo gold depositby Burk et al. (1986) and Kuhns et al. (1994) are simi-lar to those above, although their interpretations differsignificantly from ours (see below).

Group III: lcyanite in Al-Si-rich alterationas semblage s as s ociate d with mine ralization

This group includes seven occurences: the SturgeonLake (7 in Fig.1; Franklin et al. 1975), the Savant Lake(8 in Fig. 1; Lefebvre 1982), the Marshall Lake (9 inFig. 1; Ayres 1978; Amukun 1989), and the Onamanarea (10 in Fig. l ; Osterberg et al. 1987) of theWabigoon Subprovince, the Melchett greenstone belt ofthe English River Subprovince (11 in Fig.l; Breaks199 I ), the Willecho deposit in the Manitouwadge green-stone belt of the Wawa Subprovince (12 in Fig. l;Schandl et al. 1995), and the Hemlo gold deposit in the

KYANITE IN THE WESTERN SUPERIOR PROVINCE 365

Hemlo - Heron Bay greenstone belt of the WawaSubprovince (13 in Fig. 1; Burk et aL.1986, Kuhns elal. 1994, Pan & Fleet 1995). The first six occurrencesare all associated with base-metal volcanosenic massivesulfide (VMS) deposits or showings. For exampie,Franklin et al. (1975) noted that kyanite at the MattabiCu-Zn deposit ofthe Sturgeon Lake area is restricted tothe massive ores or veins in the immediate footwall,whereas andalusite occurs as porphyroblasts in all rocktypes and in veins. Lefebvre (1982) reported the pres-ence of all three polymorphs of Al2SiO5 in Al-Si-richalteration zones in the Handy Lake metavolcanic rocksat the Savant Lake area. Amukun (1989) also docu-mented all three AlzSiOs polymorphs, staurolite andmuscovite in alteration assemblages in close proximityto base-metal sulfide mineralization at the MarshallLake area At the Manitouwadge greenstone belt,Schandl et al. (1995) reported the presence of relictkyanite in sillimanite knots in altered felsic volcanicrocks from the footwall of the Willecho 3 orebody. Theoccurrence ofkyanite in quartz veins and its associationwith gold mineralization at the Hemlo gold deposit havebeen described by a number of authors (Brxk et al. 1986,Kuhns er al. 1994, Pan & Fleet 1995).

MrNnnar- CHpursrnv AND GEoTIDRMoBARoMETRy

Textural evidence from Group-I and -II occurrencesof kyanite at Raith and Hemlo clearly shows two dis-tinct assemblages indicative of two episodes of meta-morphism (Ml and M2): an earlier one consisting ofkyanite + staurolite + garnet (inclusions in kyanite andstaurolite, Fig. 2a; core of texturally zoned porphyro-blasts, Fig. 2e) + inclusions of biotite, plagioclase,

quartz, and ilmenite, and the main one in the matrix in-cluding sillimanite + gamet + biotite + plagioclase +quartz. In this study, four samples of low-variance, kya-nite-bearing assemblages (two from Raith and two fromthe Hemlo gold deposit) were selected for compositionalanalysis of minerals (see Pan et al. 1994 for analyticaldetails; Table 2) and geothermobarometry Qable 3).Also, geothermobarometric results from kyanite-bear-ing samples of the Rainy Lake area (Tabor 1988) andWhite River property (Pan & Fleet 1993) are includedfor comparison.

The mineral inclusions (i.e.,biotite, plagioclase andgamet) in kyanite and staurolite poikiloblasts from boththe Raith area and the Hemlo gold deposit are composi-tionally distinct from their counterparts in the matrix(Table 2). For example, biotite in kyanite poikiloblastsis higher in Mg/(Mg + Fe) values than its isolated coun-terpart in the matrix, whereas grains of biotite in directcontact with garnet porphyroblasts are intermediate inMg/(Mg + Fe) and apparently have been affected by re-equilibration during retrogression. Similarly, plagio-clase in kyanite and staurolite poikiloblasts is more sodicthan its counterpart in the matrix. Chemical zonation hasnot been detected in the matrix plagioclase of the Raithsamples. However, some isolated grains of plagioclasefrom Hemlo show an outward increase in the anorthitecomponent (up to 5 molVo An). Also, plagioclase inclu-sions in the margin of zoned porphyroblasts of gamet(Fig. 2e) are slightly higher in An than their counter-parts in the core. Garnet inclusions in kyanite and stau-rolite poikiloblasts from both Raith and Hemlo areenriched in Ca relative to the M2 garnet porphyroblasts,whereas the inclusion-rich core ofzoned porphyroblastsof gamet (Fig. 2e) has a Ca content similar to that of the

TABLE 2A COMPOSITIONS OF MINEMLS IN KYANITE-BEARING METAPELITE FROM RAITH (SAMPLE RAITH.I)

Incl$ions in Kymite Mircrals in Matix

sio, (s%o) 37 5Tio, 0 01A,O, 2t 2FeO 33 5McO 3 27I\4nO 212ZnJ ndCaO 277NarO nd&O ndF n dO=ITobl 99 8

3 0 1 102 00622450 391o 144

0 187

120

6 3 5 3 6 1 00 1 9 8 5 2 5

2 2 9 1 8 6 00 1 7 8 4 6 60 1 1 7 0 1 60 0 0 1 0 9 0nd nd 0033 9 0 0 09 2 3 0 2 2 n d0 4 1 9 3 6 n dn d o n d

0999 9t 8 100 2

5 4r9 0 0000224 09963 291 0 0002230 098226t9 0 0060 001 0 019

0 0010 00 063| 797

2 2 0 3 0

3 8 1 3 7 E002 0 0 l

219 21 5329 34 54 0 2 2 4 5I E6 257nd nd

t 2 t t t 1nd nd

Dd nd

nd nd

100 0 100 0

3 022 3 0350 001 0 0012047 20332 t18 23110 476 02930.125 0 175

0 103 0 101

8126 26'700 l l 0 0

1t294 l33t2943 00 5 5 1 00068 00 5200 0 3 3 90 0 6 1 00 0 0 1 70

4 8 0 8 0

3 5 5 3 5 8 4 5 52 3 3 2 2 1 0 0 5

l E 9 1 9 1 3 6 81 9 1 1 8 6 0 8 4t 0 7 1 1 3 0 5 10 a 0 0 2 0 0 5nd trd nd0 0 0 0 10 1 4 0 1 8 0 2 19 1 t 9 2 1 1 0 50 0 00 0 0

9 5 9 9 6 4 9 4 5

5 355 5 352 6 0830264 024t 0 0053 360 3 366 5 7982 405 2 321 0 0942 4 0 1 2 5 2 0 0 1 0 20003 0003 0006

27 9 6020 5 0 0

5 3 3 2 5 6t z t 01 2 7 00 2 8 0241 nd0 0 1 7 t 20 7 I 00 0 3 00 n d0

9 7 6 1 0 0 3

2 8080I 193000

0 1850 1920.023

8 0

SiTiAIF€rytgMnZnCaNaKFo

00 0421 7',7102 2 0

0 00020 052 0 055| 76t | 7950 02 2 0 2 2 0

366 THE CANADIAN MINERALOGIST

TABLE 28 COMPOSITIONS OF MINERALS IN KYANITE-BEARING METAPELITE FROM I]EMLO (SAMPLE IIEMLO.I)

lnclusions in K)@i& Minerals in Matix

sio, (fr %)Tio,ArqFeOMgoMnOZtACaONaroK,OF

Tohl

SiTi

FeMgMnZnCaNaKFo

372 63 00 0 1 0

2 t | 2333 r 2 0236 0602 0no no

2 4 3 4 3 0Dd 935n d 0 1 9nd nd

100 3 100 l

2997 2'7850 000 0 0002003 l2 t42 09t 0 0000284 0 0000 411 0 000

0 210

t20

3 5 6 01 t9 521

1 9 5 0 0 51 8 9 4 5 39 5 5 0 1 90 0 1 1 5 0n d 0 0 10 00 3 2 n d9 5 6 n d0 n d0

95 3 992

5 403 0 0000 215 09973 488 0 0022394 09642 t l2 0 0070 001 0032

0 0000 000 0 0000 04EI E6l0 000220 3 0

2980 3 0040 000 0 0032034 20232t27 2 t090334 02270 400 0 494

0t26 01220 00E 0 000

120

97 9 999

a09t 27460 140 0 000

1t247 12603 268 0 0000 726 0 0000 086 0 0000 0890 000 0 2570000 07220 000 0 0020 000

4 8 0 8 0

) 5 2 3 5 6 4 5 rr 7 0 1 8 1 0 5 8

1 9 8 2 0 2 3 6 81 9 3 1 1 5 0 7 79 0 1 9 6 7 0 5 40t2 020 0nd nd nd

0 0 4 0 0 0 10 41 025 t23936 962 9160 0 4 0 0002

9 5 0 9 5 0 9 4 9

5 377 s 393 6 0130 196 0206 0 0583 564 3 606 5 182246t 2214 0 0862053 2185 01070 016 0025 0 000

0 006 0 000 0 0020 t2r 0 073 0 3r1| 829 1 E63 I 6630 019 0 000 0 0002 2 0 2 2 0 2 2 0

366 37 |0 0 0 5

2 1 2 2 1 23 1 3 3 t 22 ' ,75 1885 80 720nd nd

r 4 4 1 4 00 0 5 0nd nd

nd nd

993 1000

St

2 7 90 u

5 3 4

1 6 E0 3 50 4 1

PI

6 1 90

2 4 1000nds 3 98 4 00 0 3nd

0 2040 8010 0 1 1

8 0

MineralsymbolsafterKrez(19E3)'a'replesenbtheouhportionofM2gmetpo+h)roblatwithmaximmMe/8evalue,'b',outmost mgin of M2 gmet porph)Tobltrst in direct conact with biotite; 'c', isolated grain of brotite; 'd' grair of biotib in direclcon@ with gmd porph,rcblat; w %, weight p€rcmt

garnet inclusions (Table 2). The M2 garnet porphy-roblasts from both Raith and Hemlo exhibit a weak com-positional zonation typical of prograde metamorphism(i.e., increase in Mg and decrease in Fe and Mn) fromcore to rim, except that the outemost margin (<50 pmwide) in direct contact with biotite or chlorite showsreverse trends in Mg, Fe and Mn (Fig. 4), consistent withan exchange of these elements during retrogression.However, the Ca profile of these M2 garnet porphyro-blasts does not show any apparent changes, even wherethey are in direct contact with plagioclase (Fig. 4). Thetwo textural varieties of staurolite (i.e., inclusion in kya-nite and porphyroblasts or poikiloblasts in the matrix)from the Raith area do not show any compositional dif-ferences and are characteized by minor amounts of Zn(rp to 2.4 wt.Vo ZnO; Table 2A). Similarly, minoramounts of Zn (0.4-2.5 wt.% ZnO) are typical of stau-rolite from Hemlo.

The well-calibrated garnet-biotite geothermometerand garnet - Al2SiO5 - qruartz - plagioclase (GASP)geobarometer were selected for temperature and pres-sure calculations to facilitate direct comparison of thetwo episodes of metamorphism (Table 3). In addition tothe experimental calibrations of Ferry & Spear (1978)for the garnet-biotite geothefinometer and Koziol &Newton (1988) for the GASP geobarometer, we havealso used the TWEEQU software package of Berman(1991), with the internally consistent thermodynamicdataset of Berman (1988) and solution models ofBerman (1990) for garnet, McMullin et al. (1991) for

G 3 3o\

=

aI.IJo= 3o

CaO#

Frc 4. Compositional profile for a gamet porphyroblast (8

mm in diameter) from kyanite-bearing metapelite of theRaith area. This garnet porphyroblast most likely grew dur-ingthe M2episode of metamorphism, because its S-shapedinternal trail of inclusions is continuous with the main pen-

etrative foliation.

KYANITE IN THE WESTERN SUPERIOR PROV]NCE

TABLE 3, SUMMARY OF METAMORPHIC PRESSURE-TEMPERATURE RESULTS

36'7

M1 Metamomhism

Sample Gamet Biotite Pl Temperature("C) Pressure (MPa)

X(Alm) X(Pyr) X(crs) X(Sps) X(Mg) X(Fe)

Raith-l 0.751 0132 0.063 0 049 0.453 0.386Raith-2 0.710 0.101 0.072 0117 0397 0.393Hemlo-l 0.696 0 095 0.070 0137 0.382 0.423

X(AD X(TD X(K)

0.123 0 039 0.9660.169 0.041 0.9820157 0.038 0.975

F&S TWQ K&N TWQ

690 670730 690690 670

18 .5 510 56018.1 495 54020.7 515 560

M2 Metamomhism

Biotite Temperature("C) Pressure(MPa)

X(Alm) X(P1r) X(Grs) X(Sps)

Raith-l 0.756 0 165 0.036 0.043Raith-2 0.712 0.125 0.060 0.103Hemlo-l 0.7i2 0.112 0 042 0 134Hemlo-2 0.641 0.167 0 058 0.134

x(Mg) x(Fe) x(Al) x(rD x(K)

0.41s 0.415 0 123 0.046 0 9170.3s4 0.449 0.157 0.040 0.9610 363 0.435 0.167 0.03s 0.9380 477 0.353 0.138 0 033 0.958

F&S TWQ K&N TWQAn

3 5 053.026.23 5.0

330450450520

640 645 320650 670 340590 600 430595 620 450

Retrogression (M3)

Gamet Biotite Temperature ('C)

F&s rwQtX(Alm) X(Pyr) X(crs)

Raith-i 0.802 0.102 0.035Raith-2 0.125 0.092 0.059Hemlo-l 0.714 0.077 0.041Hemlo-2 0.661 0.139 0.042

X(sps) x(Mg) x(Fe) x(AD X(TD

0.061 0 434 0.399 0 r24 0.0430.124 0.431 0.377 0.151 0.0410.167 0.390 0.395 0.178 0 0370.158 0.535 0.298 0.135 0.032

X(K)

0.9710.9770.9620 956

450435430450

460455440455

F&S is the gamet-biotite geothermometer of Ferry & Spear (1978); K&N is the GASP geobarometer of Koziol & Newton (1988)using the solution models ofBerman & Koziol (199i) for gamet and Fuhrman & Lindsley (1988) for plagioclase; TWQ is calculatedftom the TWEEQU method of Berman (1991) using the internally-consistent thermodynamic data set of Berman (1988) and solutionmodels of Berman (1990) for gamet, McMullin et al (1991) for biotite, and Fuhman & Lindsley (1988) for plagioclase. f, assuming aoressure of 200 MPa.

biotite, and Fuhrman & Lindsley (1988) for plagioclase,to evaluate the influence of other components (e.g., Caand Mn in garnet, and A1 and Ti in biotite) on the pres-sure-temperature estimates. These geothermobarom-eters also were chosen to facilitate comparison with theresults ofTabor (1988) for the Rainy Lake area, Percival(1989) for the Raith area, and Pan & Fleet (1993) forthe White River property. For both the Raith area andthe Hemlo gold deposit, pressures and temperatures ofthe M1 metamorphism were calculated exclusively fromthe compositions of mineral inclusions in kyaniteporphyroblasts or poikiloblasts (Table 3). We calculatedthe pressures and temperatures of the M2metamorphismby assuming that the outer portion of the M2 garnetporphyroblasts with the maximum Mg/Fe value (Fig. 4)was in equilibrium with the isolated grains of biotite andplagioclase in the matrix, where sillimanite is the stable

A12SiO5 polymorph (Table 3). In addition, the garnet-biotite geothermometer has been applied to the outer-most margin of gamet porphyroblasts and biotite indirect contact with it to constrain the event of refrogres-sion (M3; Table 3). The lack of variation in Ca in garnetGig. a) where it occurs in direct contact with plagio-clase indicates that the net-transfer reaction involvingthe breakdown of 3 anorthite to give grossular + Al2SiO5+ quartz was not important during the retrogression. Al-ternatively, the retrograde path may have been parallelto the GASP isopleth.

Geothermobarometric results from the Raith areaand the Hemlo gold deposit are given in Table 3 andillustrated in Figure 5. These results confirm the tex-tural relationships for two distinct episodes of metamor-phism and a reffograde event. Moreover, the Raith areaand the Hemlo gold deposit both have undergone an

368

earlier medium P-T (Barrovian-style) metamorphismand a main low-P/T metamorphism (Table 3, Fig. 5),and hence are discussed together below.

Pan & Fleet (1993) reported pressure and tempera-ture conditions of 60G-650 MPa and 500'C for the Mlmetamorphism at the White River property. The gar-net-biotite geothermometer based on the calibration ofFerry & Spear (1978) also yielded 495-515"C for theMl metamorphism at both the Raith area and the Hemlogold deposit (Table 3). However, the temperature esti-mates from the TWEEQU calculation are higher (540-560'C; Table 3). Also, pressure estimates of the Mlmetamorphism from the GASP geobarometer of Koziol& Newton (1988) are slightly higher than those fromthe TWEEQU calculation (Table 3). It is noteworthythat Bauer & Tabor (1991) also recognized an earliermedium-P, low-T metamorphism in the Rainy Lakearea. However, the geothermobarometric results of Ta-bor (1988) are difficult to evaluate, because he did notdifferentiate the two episodes of regional metamor-phism.

Our pressure and temperature estimates of the M2metamorphism at Raith (Table 3) are almost identicalto those obtained by Percival (1989): 650oC from theFerry & Spear gamet-biotite geothermometer and 330

MPa from the Koziol & Newton GASP geobarometer.Similarly, the results of the M2 metamorphism at theHemlo gold deposit are comparable to those at the adja-cent White River property (400-500 MPa and 550-650oC; Pan & Fleet 1993). The temperature estimatesfor the retrogression event (M3) at Raith and the Hemlogold deposit are similar and range from 430 to 460"C0able 3).

Pan & Fleet (1993) also used mineral-inclusiongeothermoba"rometry (St-Onge 1987) to construct ametamorphic P-T path involving both Ml andM2meta-morphism at the White River property. Unfortunately,mineral inclusions in garnet porphyroblasts from theRaith and Hemlo samples examined in this study arenot sufficiently developed to permit use of this method.Nevertheless, the compositional zonation of the M2garnet porphyroblasts (l. e., increase in Mg/Fe from coreto rim except for the outermost margin; Fig. 4) corre-sponds to a temperature increase of about 60"C, assum-ing a constant composition of the coexisting biotite.However, pressure most likely remained constant dur-ing the growth of the M2 garnet porphyroblasts, becausethe temperature increase is offset by an increase in theanorthite component of plagioclase inclusions. Theseresults are in agreement with the calculated metamor-

700

800

1 0 0

b Hemlo

. ,t lI+I

i tvtg

600(u(L

= 500

E 4oo5o

0 3oo

tL 2oo

4oo 500 600 700 400 500 600 700

Temperature (oC)

Frc. 5. Summary of metamorphic P-T conditions calculated from kyanite-bearing metapelites for a) the Quetico-Wabigoonboundary at the Rairh area, and b) the Hemlo gold deposit; also shown is the calculated P-T path from the White River

Foperty (Pan & Fleet 1993; dashed lines). Boundaries of A12SiO5 polymorphs are calculated using the thermodynamicdataset of Berman (1988). Ml is the earlier Barrovian metamorphism, M2 is the regional low-P, medium-T metamorphism,

and M3 is the retrogression.

Raitha

2t _I+I

;M3

KYANITE IN THE WESTERN SUPERIOR PROVINCE 369

phic P-T path from mineral-inclusion geothermo-barometry at the White River property (Pan & Fleet1993; Fig. 5b). Burk et al. (1986) and Kuhns et al.(1994), on the other hand, interpreted kyanite as part ofthe peak metamorphic assemblage and sillimanite as aretrograde phase, although their calculated temperaturesof the kyanite-forming event are noticeably lower thanthose of the sillimanite-forming event. Sillimanite inmany parts of the Western Superior Province is clearlypart of the prograde assemblage of the regional low-P/Tmetamorphism (Ayres 1978, Thurston & Breaks 1978,Percival 1989, Thurston et al. 1991, Williams et al.1991, Pan & Fleet 1993).

DrscussroN

Metamorphic significance of Group-I and -IIoccurrences of lcyanite

Most previous studies have emphasized the regionallow-P/T metamorphism in the Western Superior Prov-ince (Thurston et al. 1991, Williams 1991, Williams eral. 1991, and references therein). Our textural evidencefrom the kyanite occurrences in metapelites and quanti-tative geothermobarometric calculations demonstratethe existence of an early Banovian-style metamorphism(see also Burket a|.1986, Bauer & Tabor 1991, Pan &Fleet 1993) before the regional penetrative deformationthat accompanied the main low-P/T metamorphism. Thegeothermobarometry also reveals a significant decom-pression following the Barrovian metamorphism fromapproximately 700 MPa to about 300 MPa at theQuetico-Wabigoon boundary (Table 3), correspondingto an exhumation of up to l2 km. Exhumation of at least6 km (i.e., from 700 MPa to 400-500 MPq Table 3,Fig. 5b) occurred also in the Hemlo - Heron Bay green-stone belt along the Hemlo Fault Zone.

Barrovian metamorphism is common in Phanerozoicorogenic belts, and is related to large-scale thickeningof the crustfollowed by thermal relaxation (Spear 1993,Windley 1995). High-P/T metamorphism (l.e., blueschistand eclogite facies) that characterizes Phanerozoic sub-duction complexes (Platt 1986, Hamilton 1988), on theother hand, requires anomalously 1ow geothermal gra-dients (Ernst 1972, Windley 1995). The absence ofblueschists and eclogites in Archean terranes has longbeen attributed to either significant higher geothermalgradients in Archean time or to difficulties in preserva-tion (Ernst l912,Windley 1995, andreferences therein).There is ample evidence that the Archean mantle wasconsiderably hotter than today's mantle (e.g., Bickle1978, Martin 1986, Herzberg 1996); therefore, plateproduction and rate of oceanic lithosphere spreadingmay have been greater in the Archean (Bickle 1978,Martin 1986, Condie 1990), resulting in younger, hot-ter, and thicker oceanic lithospheres in Archean subduc-tion zones. Consequently, the thermal regimes inArchean subduction zones must have been significantly

different from their modern counterpafis and were notappropriate for high-P/T metamorphism but, instead,might have yielded medium-P/T (Barrovian) metamor-phism (including kyanite-staurolite assemblages inmetapelites). Therefore, we propose that the Barrovianmetamorphism in the Western Superior Province, asrevealed by the kyanite occurrences in metapelites, maybe directly related to Archean subduction-accretion tec-tonlc processes.

Implications for accretionary tectonics in the westernSuperior Province

According to accretionary plate-tectonic models(e.g., Percival & Williams 1989, Card 1990, Thurston& Chivers 1990, Percival et al.1994), the subprovincesof the Western Superior Province represent volcanicarcs (granite-greenstone subprovinces), accretionarycomplexes (metasedimentary subprovinces), and conti-nent-margin magmatic arcs (plutonic subprovinces),assembled by sequential north-to-south accretion(Krogh & Davis l97l,Cafi 1990, Percival et al. 1994).Specifically, Percival & Williams (1989) interpreted the

Quetico Subprovince as an accretionary wedge in whichsediments of submarine fans and abyssal turbidites werefirst deformed by accretion onto the active Wabigoonarc and later became compressed by docking of theWawa arc (Percival 1989, Williams 1990). Percival &Williams (1989) also suggested a northward subductionon the basis of the dominantly northward-facing stratig-raphy in the Quetico metasedimentary rocks. However,these authors cautioned that the regional low-P/T meta-morphism in the Quetico Subprovince is not consistentwith the thermal structure of accretionary wedges. Wesuggest that the earlier Barrovian metamorphism, andnot the main regional low-P/T metamorphism, formedduring this early stage of the development of the QueticoSubprovince. Several authors (e.g., Tabor 1988, Percival& Williams 1989, Williams 1990) have documentedearly thrust- and nappe-like structures at the Quetico-Wabigoon boundary. Underplating and thrusting ofsediments would be expected to thicken the accretion-ary wedge (Platt 1986) and hence create conditions forBarrovian metamorphism in Archean subduction com-plexes. Platt (1986) also demonstrated that continuedunderplating at the base of the wedge must be compen-sated by extension above, thus providing a mechanismfor bringing up high-P/T metamorphic rocks to upperlevels at the landward margin of the wedge. This com-bination of continued underplating and extension pro-vides a plausible explanation for the observed uplift ofup to 12 km at the Quetico-Wabigoon boundary.

Ptatt (1986) suggested that preferential uplift in ac-cretionary wedges occurs near the landward marginwhere high-P/T metamorphic rocks are commonlyobserved in Phanerozoic accretionary complexes. Therestr icted occurrences of kyanite in the QueticoSubprovince along the northern (Quetico-Wabigoon)

370 THE CANADIAN MINERALOGIST

boundary appears strikingly equivalent to the restrictionof lawsonite- and jadeite-bearing blueschists to the east-ern side of the accretionary prism of the FranciscanComplex in California. This provides further evidencefor northward subduction during formation of theQuetico Subprovince (Percival & Williams 1989). Onthe basis of east-north-east shallowly plunging minerallineations and indicators of dextral shear, Tabor (1988)suggested southward subduction in the western Queticosubprovince in northern Minnesota and proposed a re-versal of the polarity of subduction along the strike ofthe Quetico Subprovince, similar to the Apine Fault inNew Zealand. However, the occurrence of kyanite inthe Rainy Lake area suggests that subduction in thewestern Quetico Subprovince was northward, as it wasthroughout the remainder ofthe subprovince.

The timing of the Barrovian metamorphism at theQuetico-Wabigoon Subprovince boundary has not beendirectly constrained. Percival & Williams (1989) brack-eted the formation of the proposed accretionary wedgefram2750 to 2700 Ma. This was followed by compres-sion as a result of oblique collision with the Wawa arcat27U0-2684 Ma (Corfu & Stou 1986, Percival & Wil-liams 1989, Percival 1989). The M2 metamorphism inthe Quetico Subprovince has been bracketed at approxi-mately 2670 Ma in amphibolite-facies zones (Percival& Sullivan 1988) and 2666-2650 Ma in a granulite-fa-cies zone (Pan et al. 1998), and is interpreted to relateto thermal relaxation following collision with the Wawaarc with additional sources of heat from underplatedmantle-derived magmas (Percival 1989, Pan et al.1994).

It has become clear in recent years that manyArchean greenstone belts may represent a collage oforiginally unrelated greenstone fragments assembled byaccretionary tectonic processes (c/. Williams 1990,Thurston et al. l99l). On the basis of two U-Pb zirconage determinations, Corfu & Muir (1989) suggested thatthe Hemlo - Heron Bay greenstone belt represents a jux-taposition of two originally unrelated fragments (i.e., the277 2 MaHemlo - Black River assemblage and the 2695Ma Heron Bay assemblage) separated by a major struc-tural discontinuity, possibly represented by the HemloFault Zone (see also Williams et al. 1991).If this sug-gestion is valid, the restricted occurrences ofkyanite andinferred earlier Barrovian metamorphism in close spa-tial association with the Hemlo Fault Zone may haverecorded an event of crustal thickening related to small-scale accretion that assembled originally unrelatedslivers of greenstones (e.g., Williams 1990). It is wellknown that accretion of allochthonous terranes (bv col-lision) in the Coast Ranges of British Columbia resultedin intense deformation and Barrovian metamomhismfol lowed by rapid upl i f t ing te.g., Hol l ister 1982).

Pan et al. (1991) had noted that, at the White Riverproperty, the sedimentary and mafic volcanic rocks fromboth sides of the Hemlo Fault Zone are remarkably simi-lar; they suggested that they most likely formed in a

single tectonic environment. However, these lithologiesare merely typical of Archean island arcs. In light of thepresent survey of kyanite occurrences, they could wellhave formed in separate island arcs or indeed could beseparate fragments of the same island arc system. Fur-ther high-precision U-Pb zircon geochronologicalanalyses of volcanic rocks and associated sedimentaryrocks in the Hemlo - Heron Bay greenstone belt areclearly needed to test whether the Hemlo - Black Riverand Heron Bay assemblages represent two related orunrelated fragments of greenstone.

Thurston et al. (199I) proposed the formation of thenorthwestem Superior Province (i.e., the Berens Riverand Sachigo subprovinces in Card & Ciesielski 1986)by a northward accretion onto the North Caribou ter-rane. The occurences of kyanite in the North Caribougreenstone belt, near the northern boundary ofthe NorthCaribou terrane, are consistent with crustal thickeningrelated to this northward accretion. Altematively, theMiskeesik Lake occurrence of kyanite close to a faultboundary between the Eyapamikama and McGruer as-semblages may be related to accretionary processes thatassembled the North Caribou greenstone belt. Furtherresearch of these North Caribou occurrences of kyaniteis needed to establish their relationships with the tec-tonic events, especially in view ofthe possible presenceof multiple generations of kyanite (Breaks & Bartlett1991) .

Comments on the Group-III occurrences of lcyanite

The occurrences of kyanite and other aluminosili-cates in Al-Si-rich alteration assemblages associatedwith VMS and Au mineralization clearly reflect theunusual whole-rock compositions of their host rocks andmay represent potential exploration tools for base-metalmineralization (e.9., Franklin et al. 1975, Amukun1989). However, the relationship of these kyanite oc-curences to tectonic processes is difficult to decipherbecause of the high variance of their mineral assem-blages. In particular, the presence of two or all threeAl2SiO5 polymorphs in these alteration assemblages hasbeen suggested by some to represent metastabllity (e.g.,Lefebvre 1982). Franklin et al. (1975) suggested thatkyanite at the Mattabi deposit formed from the break-down of pyrophyllite and did not require medium-Pmetamorphic conditions. Similarly, Fleet & Pan (1995)interpreted kyanite in the quartz veins of the Hemlo golddeposit to have formed during a late low-grade alter-ation event in a manner equivalent to the formation ofkyanite in veins and segregations in the Lepontine Alps,Switzerland (Kenick 1990). Fleet & Pan (1995) sug-gested that the P-T-t trajectory locally crossed into thestability field of kyanite, and pyrophyllite stability wassuppressed by the high salinity of the fluids. Kyanite inAl-Si-rich alteration assemblages of the Manitouwadgegreenstone belt, on the other hand, occurs as a relicwithin sillimanite (Schandl et al. 1995).It is notewor-

KYANITE IN THE WESTERN SUPERIOR PROVINCE J t l

thy that the Manitouwadge greenstone belt is situatedclosely to the Quetico-Wawa boundary, and the occur-rence of kyanite here may still record crustal thickeningrelated to the docking of the Wawa arc onto the Queticoaccretionary wedge (Percival 1989). Also, Ayres (1978)classified the Marshall Lake kyanite occurrence as amember of Group I because of its location close to theEnglish River - Wabigoon boundary, although petro-graphic or geothermobarometric evidence for an earlierBarrovian metamorphism is still lacking there.

CoNcr-usIoNs

Kyanite in the supracrustal rocks of the ArcheanWestern Superior Province falls into three distinctlithotectonic associations: I) metapelites near sub-province and terrane boundaries, II) metapelites nearfaults in greenstone belts, and III) Al-Si-rich alterationassemblages associated with VMS and Au mineraliza-tion in greenstone belts

Kyanite in Groups I and II indicates an earlierBarrovian metamorphism (670-730 MPa and 500-560"C) before the regional penetrative deformation thataccompanied the regional low-P/T metamorphism. Thepresence of group-I kyanite along the Quetico-Wabigoon boundary and the inferred metamorphic P-Tpath support the origin of the Quetico Subprovince asan accretionary wedge during northward subduction.Group-II kyanite from the Hemlo - Heron Bay green-stone belt may have recorded accretion that assembledoriginally unrelated slivers of greenstones. Our resultssupport the two-stage-accretion model of Williams(1990) for the development of the Western SuperiorProvince.

Kyanite in Group III reflects unusual host-rock com-positions and hydrothermal fluids, and may be a poten-tially useful tool for exploration of volcanogenicmassive sulfide deposits.

Acrwowr-socst\4sNrs

We thank A. Indares, T. Rivers and P. Thurston forthorough reviews and helpful suggestions, and R.F.Martin for editorial assistance. We also thank F.W.Breaks, S. Jackson, A.H. Mumin, J.A. Percival, E.Schandl, H.R. Williams, and E. Zaleski for many stimu-lating discussions on metamorphism and kyanite in thewestern Superior Province, and G. Zhang for the prepa-ration of Figures 1 and 3. F.W. Breaks and E. Schandlalso kindly supplied kyanite-bearing samples from theNorth Caribou and Manitouwadge greenstone belts, re-spectively. Financial support to this study was providedby NSERC Research and Lithoprobe Grants.

RrrsnrNcrs

AMUKUN, S.E. (1989): Precambrian geology of the LittleMarshall Lake area, District of Thunder Bay Ontario Geol.Sun.. Reo. 197.

Avnrs, L.D. (1978): Metamorphism in the Superior Provinceof northwestem Ontario and its relationship to crustal de-velopment. In Metamorphism in the Canadian Shield (J.A.

Fraser & WW. Heywood, eds.) Geol. Surv. Can., Pap'78-10,25-36.

BAUER, R.L. & Teeon, J.R. (1991): Syntectonic, intermediate-pressure regional metamorphism along the northem mar-gin of the Quetico belt in northeastem Minnesota: contrastswith adjacent boundary areas. Inst. I'ake Superior GeoI.Proc 36,2-4.

BERMAN, R G (1988): Intemally-consistent thermodynamicdata for minerals in the system Na2O-K2O-CaO-MgO-FeO-Fe2o3-Al2o3-SiO2-Tio2-Hzo-C.oz. J. Petrol. 29,445-522.

(1990): Mixing properlies of Ca-Mg-Fe-Mn gar-neLs. Am. Mineral. 7 5. 328-344

( 1 99 1 ): Thermobarometry using multi-equilibriumcalculations: a new technique, with petrological applica-tions. Can. Mineral.29, 833-855

&Kono\ A M. (1991): Temary excess propertres

of grossular - pyrope - almandine gamets and their influ-ence in geotherm obarometry . Am. Minetal. 76, 1223-1231

BtcKLs, M.J. (1978): Heat loss from the Earth: a consffaint onArchean tectonics from the relation between geothermalgradients and the rate of plate production. Earth Planet'Sci . Let t 40,301-315

BLACKBURN. C.E (1980): Toward a mobilist tectonic modelfor part of the Archean of northwestem Onraio' Geosct'Can. 7, 64-72

Bnnr.rs, F.W (1991): English River Subprovince. In Geologyof Ontario (P.C. Thurston, H.R. Williams, R.H. Sutcliffe &G.M. Stott, eds.). Ontario GeoI Sun., Spec VoL.4,239'277

& BanrLprr, J.R. (1991): Precambrian geology ofthe Opapimiskan Lake area, District of Kenora, PatriciaPorlion. Ontario Geol Sun., Open-File Rep. 5792.

DEKEMP, 8.A., FINAMoRE, P.F., JoNEs,

G.R. MACDoNALI, A J, SHrcr-ls, H.N. & Walr-lcp, H.(1984): Opapimiskan Lake Project: Precambrian geology,

Quatemary geology, and mineral deposits of the NonhCaribou Lake area, District of Kenora, Patricia PortionOntario Geol. Surv., Misc. Pap. 119,258-273.

Bum, R., HoncsoN, C.J. & Qu.lnrrnvarN, R A. (1986): Thegeological setting of the Teck-Corona Au-Mo-Ba deposit,Hemlo, Ontario, Canada. In Proceedings of Gold'86 (A'J.

Macdonald. ed.). Konsult International, Willowdale, On-tario (311-326)

CALVERT, A J., Slwvnn, E S., Davts, W.J. & LunnBN, J.N.(1995): Archean subduction inferred from seismic imagesof a mantle suture in Superior Province. Nature 375,670-674

3 t l IHE CANADIAN MINERALOGIST

Clno, K.C. (1990): A review of the Superior Province oftheCanadian Shield, a production of Archean accretion.Precambrian Res. 48. 99-156

& Cnsrsr-srr, A. (1986): Subdivisions ofthe Supe-rior Province ofthe Canadian Shield. Geasci Can.13.5-

CoNorB, K.C. (1990): Geochemical character is t ics ofPrecambrian basaltic greenstones. In Early PrecambrianBasic Magmatism (R P. Hall & D J. Hughes, eds.). Blackie,London, U.K. (40-55).

Conpu, F. & Murn, T.L. (1989): The Hemlo - Heron Baygreenstone belt and Hemlo Au-Mo deposit, Superior Prov-ince, Ontario, Canada. 1. Sequence ofigneous activity de-termined by zircon U-Pb geochronology. Chem. Geol 79,183-200.

& Srorr, G.M. (1986): U-Pb ages for latemagmatism and regional deformation in the ShebandowanBelt, Superior Province, Canada. Can. J. Earth Sci.23,1075- 1082.

ERNST, W.G (1972): Occurrence and mineralogical evolutionof blueschist belts with time. Am. J. 'ci. 272, 65i -669.

Fnnnv, J.M. & SpBen, F S. (1978): Experimental calibration ofpartitioning of Fe and Mg between biotite and garnet.Contrib. Mineral. Petrol. 66, 113-ll7.

Ft-psr, M.E. & PAN, YUANMTNc (1995): Magmatism, metamor-phism and deformation at Hemlo, Ontario, and the timingof Au-Mo mineralization in the Golden Giant mine: a dis-cussion. Econ. Geol. 90, 1338-1349.

SELLER, M. & PrN, YueuurNc (1997): Rare earthelements, protoliths and alteration at the Hemlo gold de-posit, Ontario, Canada, and comparison with argillic andsericitic alteration in the Highland Valley Porphyry Dis-trict, British Columbia, Canada. Econ. Geo|.92,551-568

FRANKLTN, J.M., Kaslnoe, J. & PouLsEN, K H (1975): Petrol-ogy and chemistry of the Mattabi massive sulphide deposit.Econ. Geol .70,63-79.

FurnwlN, M.L. & LNnsr-sv, D.H. (1988): Temary-feldsparmodeling and thermometry Am. Mineral. 73,201-215.

HrrunroN, W B ( 1 98 8) : Plate tectonics and island - arcs. Ge oLSoc. Am., Bull 100,1503-1527.

( 1ee8)Geol. Rev.40,

HsRzsrnc, C. (1996): Generation of plume magmas throughtime: an experimental perspective Chem. Geol 126,l-16.

HoI-r-Isren, L.S (1982): Metamorphic evidence for rapid (2mm/yr) uplift of a portion of the Central Gneissic Com-plex, Coast Mountains, B.C. Can. Mineral. 20,319-332.

JoLLy, W.T. (1978): Metamorphic history of the Abitibi belt.1z Metamorphism in the Canadian Shield (J.A. Fraser &

Archean tectonics and magmatism 1nt.t -39.

W.W. Heywood, eds.). Geol. Surtt. Can., Pap.78-10,63-78 .

KERRICK, D.M. (1990): The Al2SiO5 polymorphs. Rev. Mm-eral .22.

Koztor-, A M. & NEwroN, R.C. (1988): Redeterrnination of theanorthite breakdown reaction and improvement of theplagioclase - garnet - Al2SiO5 - q\arlz Eeobmometer Am.Mineral. 73,216- 223.

Knnrz, R. (1983): Symbols for rock-forming minerals. AlnMineral 68,277-279.

KRocH, T.E. & Devrs, G.L (1971): Zircon U-Pb ages ofArchean metavolcanic rocks in the Canadian Shield.Carnegie Inst. Washington, Yearbook 70, 241-242.

KuHNs, R., SAwKrNs, F.J. & Iro, E. (1994): Magmatism, meta-morphism and deformation at Hemlo, Ontario and the tim-ing of Au-Mo mineralization in the Golden Giant mineEcon. Geol 89,720-756.

LANGFoRD, F.F & MoRrN, J.A. (1976): The development ofthe Superior Province of northwestern Ontario by mergingisland arcs. Am J. Sci 276,1023-1034.

Lerravnn, J.J. (1982): Medium grade metamorphism of hy-drothermally altered volcanic rocks in the Savant Lakearea, Ontario. Geol. Assoc. Can. - Mineral. Assoc. Can.,Program Abstr.7, 63.

MARTIN, H ( I 986) : Effects of steeper Archean geothermal gra-dients on geochemistry of subduction zone magmas. Geol-ogy 14,753-756.

McMuLLrN, D W., BnnuaN, R.G & GnrpNwoor, H.J. (1991):Calibration of the SGAM thermobarometer for pelitic rocksusing data from phase-equilibrium experiments and natu-ral assemblages. Can. Mineral.29, 889-908.

MoonE, E S (1940): Geology and ore deposits of the AtikokanNea. Ontario Dep. Mines, 48th Annual Rep., Partll,l-34

Ostlnssnc, S.A., MonroN, R.L & FnaNcrN, J.M (1987):Hydrothermal alteration and physical volcanology ofArchean rocks in the vicinity of the Headway-Coulee mas-sive sulfide occuffence, Onaman area, northwestern On-tario Econ. Geol. 82, 1505-1520.

PeN, YulNrrarNc & FLEET, M.E. (1993): Polymetamorphism inthe Late Archean Hemlo - Heron Bay greenstone belt, Su-perior Province: P-T variations and implications for tec-tonic evolution. Can. J. Earth Sct. 30.985-996.

& _ (1995): The Late Archean Hemlo golddeposit, Ontario, Canada: a review and synthesis. Ore GeolRev.9,455-488

_& He.llrlN, L. (1998): Thermo-tectonicevolution of an Archean accretionary complex: U-Pbgeochronological constraints on granulites from theQuetico Subprovince, Ontario, Canada Precambian Res.92,717-128

KYANITE IN THE WESTERN SUPERIOR PROVINCE 5 l - 1

& SroNB, W.E (1991): Geochemistry ofmetasedimentary rocks in the Late Archean Hemlo - HeronBay greenstone belt, Ontario, Canada: implications for prov-enance and tectonic setting. Precambrian Res 52,53-69

& Wnlraus, H.R. (1994): Granulite-facies metamorphism in the Quetico subprovince, north ofManitouwadge, Ontaio Can. J. Earth Sci 3I,142'1-1439.

PerrsnsoNr, GC, MasoN, J.K. & ScHrsosns, B.R. (1984):Reporl of activities 1983. Thunder Bay resident geologistarea, north central region. Ontario Geol. Sum , Misc. Pap.I t1 , 47 -106.

PERCIVAL, J.A (1989): A regional perspective in the Queticometasedimentary belt, Superior Province, Canada. Can. J.Earth Sci.26,611-693

SrBnN, R.A & Drcpr, M R. (1985): Regional geo-logical synthesis of western Superior Province, Ontario.Geol. Sury. Can , Pap.85-1A, 385-397.

Srulsrr, T., CARD, K.D., MonrsNspN,J K & B6crN, N.J. (1994): Minto Block, Superior Prov-ince: missing link in deciphering assembly of the craton at2 7 Ga Geology 22, 839-842.

& Sur-lrvlN, R.W (1988): Age constraints on thegeological synthesis of westem Superior Province, Canada.Geol- Sun- Can, Pap 88-2,97-108.

& Wrr-lrerr.rs, H.R (i989): The Late ArcheanQuetico accretionary complex, Superior Province, Canada.Geology I7 ,23-25

PLArr, J P (1986): Dynamics of orogenic wedges and theiruplift of high-pressure metamorphic rocks. Geol. Soc Am.,BuIl. 97. 1037-1053.

PoulssN, K.H. (1986): Rainy Lake wrench zone: an exampleofArchean subprovince boundary in northwestern Ontario.1n Workshop on the Tectonic Evolution of GreenstoneBelts. Lunar Planet. Inst, Tech Rep 86-10, 177 -17 9.

ScHarol, E.S., GoRroN, M.P. & WlsreNrvs, H.A (1995):Rare earth element geochemistry of the metamorphosedvolcanogenic massive sul f ide deposi ts of theManitouwadge mining camp, Superior Province, Canada:a potential exploration tool'l Econ. Geol.90,1217-1236

Spsen, F.S. (1993): Metamorphic Phase Equilibria and Pres-sure - Temperature - Time Paths. Mineral. Soc. An,.,Monogr. I

Sr-ONcB, M.R. (1987 ) : Zoned poikiloblastic gamets: P-T pathsand syn-metamorphic uplift through 30 km of structuraldepth, Wopmay Orogen, Canada. J. Petrol 28, l-22.

Teson, J.R. (7988): Deformational and Metamotphic Historyof Archean Rocks in the Rainy Lake District, NorthernMinnesota. Ph D dissertat ion, Univ of Minnesota,Minneapolis, Minnesota.

THURsroN, P.C & BnBers, F.W. (1978): Metamorphic andtectonic evolution of the Uchi - English River sub-provinces 1n Metamorphism in the Canadian Shield (J.AFraser & W.W. Heywood, eds). Geol. Surv. Can., Pap.78-tO.49-62.

& Cnrvsns. K M. (1990): Secular variation ingreenstone sequence development emphasizing SuperiorProvince, Canada. P recamb Re s. 46, 2l-58.

Osrr,raNr, I.A. & SroNs, D. (1991): NorthwesternSuperior Province: review and tenane analysis. In Geol-ogy of Ontar io (P.C. Thurston, H.R. Wi l l iams, R.HSutcliffe & G.M. Stott, eds.). Ontario Geol Surv., Spec.Vol. 4,87-142

Sace, R.P. & Srnacusl, G M. (1979): Geology ofthe Winisk Lake area, District of Kenora (Patricia poftion).Ontario Geol. Sun., Rep 193.

Wu-r-Ierr.ts, H.R. (1990): Subprovince accretion tectomcs in thesouth-central Superior Province. Can J. Earth Sci. 27, 5'70-5 8 1 .

-(1991): Quetico Subprovince. In Geology of Ontario(P C. Thurston, H.R Williams, R.H. Sutcliffe & G.M.Stott, eds.) Ontario Geol. Sum., Spec. VoI.4,383-403.

Srorr, G.M., HBerspn, K.B , Murn, T.L & S.qcr,RP (1991): Wawa Subprovince. 1n Geology of Ontano(P.C. Thurston, H.R. Williams, R.H. Sutcliffe & G.MStott, eds.). Ontario Geol. Surv., Spec. Vol 4,485-539.

WrNor-Bv, B F (1995): The Evolving Continents (third ed ).John Wiley & Sons, New York, N.Y.

Received November 13, 1997, revised manuscript acceptedFebruary 15, 1999.