archaean intracrustal differentiation from partial melting of ......interactions during burial,...

Archaean Intracrustal Differentiation fromPartial Melting of MetagabbroçField andGeochemical Evidence from the Central Regionof the Lewisian Complex, NW Scotland

T. E. JOHNSON1*, S. FISCHER1, R.W.WHITE1, M. BROWN2 ANDH. R. ROLLINSON3

1EARTH SYSTEM SCIENCE RESEARCH CENTRE, INSTITUTE FOR GEOSCIENCES, UNIVERSITY OF MAINZ,

BECHERWEG 21, D-55099, MAINZ, GERMANY2LABORATORY FOR CRUSTAL PETROLOGY, DEPARTMENT OF GEOLOGY, UNIVERSITY OF MARYLAND,

COLLEGE PARK, MD 20742, USA3DEPARTMENT OF GEOGRAPHICAL, EARTH AND ENVIRONMENTAL SCIENCES, UNIVERSITY OF DERBY,

DERBY DE22 1GB, UK

RECEIVED OCTOBER 26, 2011; ACCEPTEDJUNE 12, 2012ADVANCE ACCESS PUBLICATION JULY 27, 2012

The central region of the mainland Lewisian gneiss complex of NW

Scotland is a granulite-facies migmatite terrane.With the exception

of ultramafic and rare calc-silicate rocks, all other lithologies

partially melted during Neoarchaean, ultrahigh-temperature

(Badcallian) metamorphism. The clearest evidence is preserved

within large layered mafic^ultramafic bodies that exhibit macro-

scopic features diagnostic of anatexis. In situ partial melting of

metagabbroic rocks produced patches and sheets of coarse-grained

plagioclase-rich leucosome containing euhedral peritectic clinopyrox-

ene.These leucosomes connect with larger, laterally continuous tonal-

ite or trondhjemite sheets that record segregation and migration of

melt away from the metagabbro source rocks.This melt loss allowed

wide-scale preservation of granulite-facies assemblages within the

residual melanosome.Whole-rock major and trace element geochem-

istry is broadly consistent with the field evidence, but suggests

contamination of the metagabbroic rocks by their host-rocks and a

strong mineralogical control on trace element distributions, the conse-

quence of large diffusive length scales during protracted

ultrahigh-temperature metamorphism.Variations in the trace element

composition of the felsic sheets reflect heterogeneities in the source

rocks, the presence of material entrained from the melanosome and

fractional crystallization dominated by plagioclase.The felsic sheets

are largely cumulate, suggesting loss of the evolved melt fraction to

higher crustal levels. Partial melting of felsic gneisses that surround

the mafic^ultramafic bodies is inevitable at the implied meta-

morphic peak provided they contained hydrous phases, although the

field evidence is largely obscured by later reworking.This study pro-

vides insights into the processes involved in intracrustal differenti-

ation during the Neoarchaean, during which partial melting of

mafic rocks is likely to have made a more significant contribution

than during the Phanerozoic.

KEY WORDS: Archaean; intracrustal differentiation; Lewisian com-

plex; metagabbro; partial melting

I NTRODUCTIONPartial melting and melt loss are the fundamental pro-cesses by which crustal differentiation occurs, creating amore mafic, minimally hydrated and residual lower por-tion, and a more felsic, more hydrated and incompatible

*Corresponding author. E-mail: [email protected]

� The Author 2012. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: [email protected]

JOURNALOFPETROLOGY VOLUME 53 NUMBER10 PAGES 2115^2138 2012 doi:10.1093/petrology/egs046 at U

niversity of Maryland on O

ctober 2, 2012http://petrology.oxfordjournals.org/

Dow

nloaded from

http://petrology.oxfordjournals.org/

element-enriched upper portion (e.g. Brown & Rushmer,2006).The temperature at which rocks begin to melt is gov-erned by the presence or absence of feldspars and/orquartz and the amount of H2O present either as a freevolatile phase or, more commonly, bound within hydrousminerals such as mica and amphibole. With the exceptionof carbonate rocks, other common crustal protoliths (silici-clastic sedimentary rocks and felsic to mafic igneousrocks) contain quartz and/or feldspars, and most containhydrous phases, an amount governed by the original H2Ocontent of the rock (primary) and subsequent fluid^rockinteractions during burial, metamorphism and exhum-ation (secondary). At mid- to lower-crustal pressures,metapelitic rocks may produce significant volumes of leu-cogranitic melt by reactions consuming muscovite at tem-peratures of 7508C or less (e.g. Thompson, 1982; Patin‹ oDouce & Harris, 1998; White et al., 2001). However,wide-scale partial melting of metapelitic rocks, metagrey-wacke and K-rich meta-igneous rocks via reactionsconsuming biotite generally requires granulite-facies con-ditions, with temperatures in excess of 8008C (e.g. Whiteet al., 2001; Watkins et al., 2007; Johnson et al., 2008).Metagabbroic rocks and other K-poor, plagioclase-bearingrocks will melt predominantly via reactions consumingamphibole and plagioclase, and require still higher tem-peratures to produce volumes of melt capable of large-scalesegregation and ascent (e.g. Rushmer, 1991;Wyllie & Wolf,1993; Sawyer et al., 2011).Ultrahigh-temperature (UHT) metamorphic conditions

are first documented in the Neoarchaean rock record(Brown, 2006), of which the granulite-facies central regionof the Lewisian complex of NW Scotland (Fig. 1) is one ex-ample (e.g. Wheeler et al., 2010). Although partial meltingand melt loss have commonly been proposed as mechan-isms to explain aspects of the geochemistry within the cen-tral region, in particular a depletion in some mobile highheat-producing elements and large ion lithophile elements(LILE; K, Rb, Cs, U, Th) (e.g. Moorbath et al., 1969;O’Hara & Yarwood, 1978; Barnicoat, 1983; Cartwright &Barnicoat, 1987), other studies suggest that such featureswere inherited from the source region and that in situ par-tial melting did not necessarily occur (e.g. Park & Tarney,1987; Tarney & Weaver, 1987; Rollinson, 1996; Rollinson &Tarney, 2005). Unambiguous field evidence for anatexis ofthe volumetrically dominant meta-igneous rocks has notbeen demonstrated. To quote Wheeler et al. (2010):‘Evidence of partial melting in the Central District granu-lite facies rocks is sparse, which is problematic given thevery high temperature estimates which surely make melt-ing likely in many lithologies’.In this study, outcrops throughout the central region of the

Lewisian complex have been examined to assess the field evi-dence for partial melting. The evidence presented herefocuses on metagabbroic rocks within large mafic^ultramafic

bodies (Fig. 1), and is based on the recognition of featuresdiagnostic of anatexis to form migmatites (Sawyer, 2008).Bulk-rock major and trace element geochemical data are dis-cussed within the context of the field evidence. We showthat metagabbroic rocks throughout the central regionpartially melted during Badcallian UHT metamorphism,evolving from subsolidus amphibolites to suprasolidus clino-pyroxene-rich granulite-facies migmatites. We suggest thatpartial melting and melt loss were important processes thatdrove intracrustal differentiation within the central regionof the Lewisian complex. These processes must have hadprofound consequences for the compositional, thermal andrheological evolution of the Lewisian continental crust.

REGIONAL GEOLOGYThe Lewisian complex of NW Scotland includes some ofthe oldest rocks in Europe, with protolith ages of around3·1^2·7 Ga (e.g. Hamilton et al., 1979; Whitehouse &Moorbath, 1986; Friend & Kinny, 1995; Whitehouse et al.,1997; Kinny et al., 2005;Whitehouse & Kemp, 2010). It is adeformed and metamorphosed plutonic igneous complex(Peach et al., 1907) that records a protracted history of de-formation, metamorphism and intrusion spanning morethan 1 Gyr (e.g. Wheeler et al., 2010). The rocks are volu-metrically dominated by layered tonalite^trondhjemite^granodiorite (TTG) gneisses, typical of Archaean gneissterranes worldwide (Rollinson & Windley, 1980b;Rollinson, 1996), within which occur abundant sheets andlenses of metamorphosed mafic^ultramafic rocks and raremica-rich rocks (Peach et al., 1907).The mainland Lewisian complex comprises a

granulite-facies central region structurally bounded byamphibolite-facies regions to the north and south (Peachet al., 1907; Sutton & Watson, 1951; Fig. 1), and the centralregion is thus interpreted as representing a deeper crustallevel (Sheraton et al., 1973; Park, 2005). In general, rockswithin the central region preserve assemblages consistentwith peak pressures of 8^12 kbar and temperatures inexcess of 9008C (e.g. Cartwright & Barnicoat,1987; Sills &Rollinson, 1987; Johnson & White, 2011) that developedduring the c. 2·7 Ga Badcallian metamorphic event (Park,1970; Corfu et al., 1994; Zhu et al., 1997; Whitehouse &Kemp, 2010), although the occurrence of a granulite-faciesmetamorphic event at 2·7 Ga has been questioned by someworkers (Love et al., 2004; Kinny et al., 2005). Centralregion gneisses are depleted in silica, H2O, U,Th and cer-tain LILE (K, Rb, Cs) relative to those in the northernand southern regions (e.g. Sheraton et al., 1973; Tarney &Windley, 1977; Rollinson & Windley, 1980a; Weaver &Tarney,1981; Fowler,1986; Cohen et al.,1991), although therehas been considerable disagreement as to the origin of thisdepletion. Several studies have concluded that, with the ex-ception of ultramafic and rare calc-silicate rocks, all rocksmust have partially melted with loss of melt purging the

JOURNAL OF PETROLOGY VOLUME 53 NUMBER 10 OCTOBER 2012

2116

at University of M

aryland on October 2, 2012

http://petrology.oxfordjournals.org/D

ownloaded from


Fig. 1. Simplified geological map of the mainland Lewisian complex showing the locations [UK grid references] of those large mafic^ultra-mafic bodies studied in detail, located at or near: (1) Cnoc Gorm [NC 167 498]; (2) Scourie [NC 143 447]; (3) Badcall [NC 146 417]; (4) GormChnoc [NC 218 448]; (5) Ben Dreavie [NC 263 390]; (6) Ben Strome [NC 253 356]; (7) Drumbeg [NC 114 333]; (8) east of Clachtoll [NC 058273]; (9) SE of Achmelvich [NC 071 235]; (10) Strathan [NC 092 201]; (11) inlier at Auchiltibuie [NC 035 082]; (12) Gruinard Bay [NG 956 907].The traditional subdivisions into the northern, central and southern regions are shown along with the Gruinard, Assynt and Rhiconich terranesof Kinny et al. (2005). LSZ, Laxford shear zone; SL, Strathan line; GB, Gruinard Bay. Modified afterWhitehouse & Kemp (2010).

JOHNSON et al. PARTIAL MELTING, CENTRAL LEWISIAN COMPLEX

2117

at University of M



ownloaded from


rocks of these mobile elements (Moorbath et al., 1969;O’Hara&Yarwood,1978; Pride &Muecke,1980; Barnicoat,1983; Cartwright & Barnicoat, 1987; Cohen et al., 1991). Incontrast, other workers have suggested that anatexis andmelt loss is inconsistent with the trace element geochemistryand that in situ partial melting did not play a major role inthe evolution of the central region (Weaver & Tarney,1981;Tarney & Weaver, 1987; Park & Tarney, 1987; Rollinson,1996), with mobile element depletion argued to haveoccurred prior to metamorphism during dehydration andmetasomatism of a mafic source region at depth (Rollinson,1996; Rollinson & Tarney, 2005).Highly deformed, layered, large maficûltramafic

bodies, some several hundred metres in thickness, arewidely distributed within the central region (e.g. Peachet al., 1907; O’Hara, 1961; Bowes et al., 1964; Sills et al., 1982;Fig. 1). In the north of the central region these bodiesdefine a linear belt extending for more than 12 km thatmay represent an accreted marginal ocean floor assem-blage (Davies, 1974; Park & Tarney, 1987; Goodenoughet al., 2010). The bodies comprise clinopyroxene-rich meta-gabbro, commonly with abundant garnet and/or amphi-bole, and contain layers of metamorphosed ultramaficrocks. Layers of meta-anorthosite are reported from sev-eral of the bodies (Bowes et al., 1964; Davies, 1974; Weaver& Tarney, 1980). Extreme disruption of metagabbroicrocks has been previously described (e.g. Davies, 1974;Weaver & Tarney, 1980), and feldspar-rich segregations insome mafic rocks have been interpreted as the products ofpartial melting (Barnicoat, 1983; Cartwright & Barnicoat,1987; Cartwright & Valley, 1992). However, mica-richrocks spatially associated with the maficûltramaficbodies and interpreted to be of metasedimentary originare the only rock types in which compelling evidence forpartial melting has been documented (Cartwright &Barnicoat, 1986, 1987).Felsic sheets, centimetres to a few metres thick, occur

throughout the central region. The sheets are generallysill-like but locally cross-cut the TTG gneisses, maficûltramafic bodies and mica-rich rocks (Cartwright &Valley, 1992; Rollinson, 1994). The felsic sheets preservegranulite-facies mineral assemblages, but are themselvesdeformed, and are inferred to have been emplaced at orclose to the Badcallian metamorphic peak (Rollinson &Windley, 1980b; Cartwright & Barnicoat, 1987). Felsicsheets within the large maficûltramafic bodies are mostcommonly tonalitic to trondhjemitic in composition,whereas those within the surrounding TTG gneisses andmica-rich rocks are generally granodioritic to granitic(Barnicoat, 1983; Cartwright & Valley, 1992). The origin ofthe felsic sheets has been a longstanding debate (Cart-wright & Rollinson, 1995). Largely on geochemicalgrounds, some workers have suggested that the felsicsheets were derived locally from partial melting of the

host TTG gneisses (Pride & Muecke, 1982; Barnicoat,1983; Cartwright, 1990; Cohen et al., 1991), whereas othershave argued that they were derived from partial meltingof an LILE-depleted mafic source at depth and are unre-lated to the high-grade metamorphism (Rollinson &Windley, 1980b; Rollinson & Fowler, 1987; Rollinson, 1996;Rollinson & Tarney, 2005).

F I ELD EV IDENCEThe large layered maficûltramafic bodies are volumetric-ally dominated by medium- to coarse-grained granoblasticmetagabbro with thin (centimetres to several decimetresthick) layers of metamorphosed ultramafic rocks rangingin composition from metaperidotite to metapyroxenite.Relict magmatic layering is commonly preserved in themetagabbros, although the contacts between layers are dif-fuse and irregular. Single layers are characterized bywidely varying proportions of plagioclase, clinopyroxene,garnet, hornblende and orthopyroxene (Fig. 2a). At themargins of the large maficûltramafic bodies, the meta-gabbroic rocks are strongly deformed and extensively tocompletely retrogressed to amphibolite-facies assemblagesrich in green hornblende, the result of late Archaean toPalaeoproterozoic reworking (e.g. Goodenough et al.,2010). However, the cores of these bodies exhibit relativelylow strain and preserve near-pristine granulite-faciesassemblages.Metagabbroic rocks are meso- to melanocratic and,

rarely, leucocratic. The majority contain abundant clino-pyroxene with variable proportions of plagioclase andorthopyroxene. Peak metamorphic brown^green horn-blende is abundant in many metagabbro layers. Layerscontaining garnet are generally volumetrically subordin-ate to garnet-absent metagabbro, but in some of thebodies (e.g. Cnoc Gorm; Fig. 1) garnetiferous metagabbrodominates.Outcrops of metagabbro throughout the central region

preserve features diagnostic of anatexis (Sawyer, 2008)and may be described as migmatitic.They comprise a mix-ture of pale plagioclase-rich quartzo-feldspathic leuco-some, representing the former sites of melt segregationand/or accumulation, and dark plagioclase-deficient mela-nosome, representing the crystal-rich residuum fromwhich melt has been extracted. The relative proportion ofleucosome and melanosome varies widely.Where leucosome contents are low the rocks are meta-

texites, containing small cuspate patches and veins ofcoarse-grained plagioclase-rich, quartz-bearing leuco-some, within which large grains of euhedral^subhedralclinopyroxene are common (Fig. 2b and c). In stromaticvariants, these clinopyroxene-bearing pegmatitic leuco-somes (hereafter Cpx-pegmatites) are generally discon-tinuous, oriented subparallel to the foliation and variablydeformed within this fabric (Fig. 2c). Thin leucosome


2118

at University of M



ownloaded from


Fig. 2. Field relations of migmatized metagabbros within the large mafic^ultramafic bodies. (a) Primary magmatic layering preserved in meta-gabbro. Single layers are rich in plagioclase (pale), clinopyroxene (� hornblende) and garnet (Ben Strome). (b) Patch metatexite containing ir-regular patch of Cpx-leucocome (Clachtoll). (c) Stromatic metatexite comprising discontinuous stroma of Cpx-pegmatite in whichclinopyroxene is strongly flattened subparallel to the subhorizontal foliation. The leucosome is associated with a diffuse mafic selvedge(Achmelvich). (d) Highly migmatized metagabbro comprising dark clinopyroxene-rich melanosome and pale plagioclase-rich leucosome(Cpx-pegmatite). The left-hand side dominantly comprises Cpx-pegmatite containing deformed peritectic clinopyroxene and schlieren andschollen of melanosome, all of which occurs within a large interboudin partition. On the right, thin stromatic leucosome veins feed into thislarger Cpx-pegmatite accumulation (Achmelvich). (e) Parallel-sided, cross-cutting sheets of Cpx-pegmatite. The later, shallowly inclined sheetexhibits a thin but pronounced mafic selvedge (Cnoc Gorm). (f) Close-up of part of (e). Peritectic clinopyroxenes within the plagioclase- andquartz-rich Cpx-pegmatite sheet are surrounded by a thin rind of hornblende, suggesting only limited retrograde reaction with melt. Themafic selvedge (M) at the top of this sheet is indicated (Cnoc Gorm). (g) Contact between Cpx-pegmatite (bottom) and a felsic sheet (top).The Cpx-pegmatite, which contains abundant peritectic clinopyroxene, merges into the felsic sheet with petrographic continuity. TheCpx-pegmatite is associated with a diffuse mafic selvedge and interconnects with thin veins of leucosome in the melanosome. The felsic sheet isaround 20 cm thick, is deficient in mafic components and has a sharp contact with the clinopyroxene-rich melanosome away from the contactwith the Cpx-pegmatite (Badcall). (h) Contact between Cpx-pegmatite (top) and felsic sheet (bottom). Similar relations described in (g) areshown.The bifurcation of the felsic sheets towards the right should be noted. S. Fischer for scale (Badcall). (i) Alignment of large euhedral^sub-hedral peritectic Cpx (bottom left to top centre) within clinopyroxene-rich melanosome. This relationship is interpreted to representnear-complete extraction of melt from a Cpx-pegmatite (Gorm Chnoc). (j) Metagabbro layer rich in clinopyroxene and garnet, in whichgarnet is relatively evenly distributed and surrounded by a plagioclase-rich corona, interpreted to record high-temperature decompression.Evidence for partial melting in such layers is cryptic, and these may represent protoliths that were not hydrated prior to metamorphism(Scourie). (k) Contact between highly residual garnet-rich melanosome (left) and a Cpx-pegmatite (right). A mafic selvedge is developed atthe contact. Garnet within the melanosome occurs as large porphyroblasts and grain aggregates surrounded by plagioclase-rich coronas. Thehighly heterogeneous distribution of garnet is much more common than, for example, that shown in (j). The Cpx-pegmatite contains abundantperitectic clinopyroxene and garnet entrained from the melanosome (Cnoc Gorm). (l) Schollen diatexite comprising chaotic mix of peritecticclinopyroxene-bearing leucosome, melanosome and selvedge (Cnoc Gorm).


2119

at University of M



ownloaded from


veins and patches interconnect with larger sheets andpatches of Cpx-pegmatite, the latter commonly occurringwithin interboudin partitions (Fig. 2d).Larger sheets of Cpx-pegmatite are centimetres to

decimetres thick and may be continuous on an outcropscale (Fig. 2e). They are commonly subparallel to the fo-liation and compositional layering, but may cross-cutthese early structures at a moderate to steep angle.Chilled margins are not evident, and a pronouncedmafic (clinopyroxene- and/or amphibole-rich) selvedge

is common at the margins of the Cpx-pegmatites (Fig.2e^g). Euhedral^subhedral clinopyroxene grains withinthe Cpx-pegmatites are an order of magnitude largerthan those within the surrounding melanosome and com-monly exhibit replacement at their margins by amphi-bole (Fig. 2f). In strongly deformed sheets, clinopyroxenegrains are flattened and may be completely pseudo-morphed by amphibole (e.g. Fig. 2c and d).Cross-cutting relations imply more than one generationof Cpx-pegmatite sheet (Fig. 2e).

Fig. 2. Continued.


2120

at University of M



ownloaded from


The Cpx-pegmatite sheets are petrographically continu-ous with (i.e. merge into without any discernible contact)larger quartzofeldspathic felsic sheets (Fig. 2g and h),most of which are less than a metre thick. Like theCpx-pegmatites, the felsic sheets are generally subparallelto the foliation, but may cross-cut the foliation at a lowangle; bifurcation of larger felsic sheets is not uncommon(Fig. 2g and h). Both the Cpx-pegmatites and felsic sheetscontain identical mineral assemblages. However, the felsicsheets are finer-grained, contain a higher ratio of quartzto feldspar and are deficient in mafic components relativeto the Cpx-pegmatites, although schlieren of hornblendeand/or biotite and rare grains of clinopyroxene occur inmost. Some felsic sheets may have a diffuse mafic selvedgeat their margins. Where largely unaffected by retrogrademetamorphism, both the Cpx-pegmatites and felsic sheetscontain bluish opalescent quartz. Although contact rela-tions are commonly obscured, decimetre- to metre-scalefelsic sheets feed into and merge with larger felsic bodies,the best example of which occurs close to the top of theGorm Chnoc body (Fig. 1) and is in excess of 40m across.This sheet-like body is broadly concordant with the foli-ation and is composite, comprising numerous single felsicsheets some of which are pegmatitic. However, thesesheets are distinct from largely undeformed coarser-grained, pink (K-feldspar-rich) Palaeoproterozoic granitesheets that become abundant a few hundred metres furtherto the north (e.g. Sutton & Watson, 1951).The metagabbro melanosome is generally rich in clino-

pyroxene with or without orthopyroxene, plagioclase,hornblende, garnet and oxide phases (mainly magnetite).Retrograde hornblende replacing clinopyroxene is com-mon. Where melt loss was near complete from garnet-absent metagabbro, the residuum is almost pure (clino)pyroxenite. Planar accumulations of coarse-grainedeuhedral^subhedral clinopyroxene within finer-grainedmelanosome are interpreted to represent former Cpx-pegmatite channels, through which most or all of the meltwas extracted leaving the coarse peritectic clinopyroxenebehind (Fig. 2i).The melanosome of many metagabbro layers addition-

ally contains abundant garnet, which may occur as distrib-uted smaller grains, as large porphyroblasts or as clustersof porphyroblasts up to several centimetres in diameter(Fig. 2j^l). Although in some cases garnet is distributedmore or less homogeneously throughout garnetiferouslayers (Fig. 2j), in most cases its distribution is extremely ir-regularçin places garnet is the dominant mineral and inothers it occurs as an accessory phase (Peach et al., 1907).Garnet porphyroblasts commonly show marginal replace-ment by coronas rich in plagioclase with orthopyroxeneand Fe^Ti oxides, the result of high-temperature retro-grade decompression (e.g. O’Hara & Yarwood, 1978;Savage & Sills, 1980; Barnicoat, 1983; Johnson & White,

2011; Fig. 2j^l). Almost all garnet is confined to the melano-some, although some grains and grain clusters are en-trained within Cpx-pegmatite and/or felsic sheets,particularly at their margins (Fig. 2k and l). Some layersor patches of melanosome comprise 490 vol. % garnetand clinopyroxene (Fig. 2j and k). In many examples therocks are texturally heterogeneous on all scales andpreserve little evidence for coherent relict layering. Theserocks are schollen and schlieren diatexites comprisinga chaotic mix of leucosome, melanosome and selvedge(Fig. 2l).Coherent ultramafic layers may extend for many tens of

metres, although the proportion of ultramafic material isgenerally small. Ultramafic layers are metaperidotite tometapyroxenite in composition, in which the lowest vari-ance assemblages contain olivine, clinopyroxene, orthopyr-oxene, hornblende, spinel and magnetite (e.g. Johnson &White, 2011). The ultramafic layers show no evidence forpartial melting, and represent a refractory palaeosome(Sawyer, 2008).Additional field evidence for the features discussed above

is given in Supplementary Data (SD) Fig. S1 (availablefor downloading at http://www.petrology.oxfordjournals.org/).

PETROGRAPHYMetagabbroMetagabbroic rocks within the large mafic^ultramaficbodies have a medium- to coarse-grained granoblasticmicrostructure and comprise assemblages of clinopyroxeneand plagioclase with or without orthopyroxene, horn-blende, garnet, quartz, spinel, magnetite, ilmenite, biotite,apatite and sulphide minerals (Fig. 3). Most samples ex-hibit a well-annealed microstructure with straight grainboundaries and 1208 triple junctions. The grain size ofminerals generally ranges from 0·5 to 2·0mm and rocksare typically equigranular, although in many of the garne-tiferous samples garnet forms porphyroblasts that may beseveral centimetres in diameter.Within granulite-facies metagabbro, clinopyroxene is

ubiquitous and forms pale green subhedral grains,commonly 1^2mm across, which may or may not containfine exsolution lamellae of orthopyroxene and which arecommonly variably altered around their margins tofine-grained green hornblende (Fig. 3a). Plagioclase (ande-sine^labradorite) forms stubby subhedral to anhedralgrains, generally less than 1·0mm in diameter, whichare commonly partially sericitized (Fig. 3a^d). Mostplagioclase lacks any distinct compositional zoning, butconcentrically zoned grains may occur. Pleochroic ortho-pyroxene grains may be in excess of 2mm across (Fig. 3b)but are generally smaller (51mm; Fig. 3b^d) and are


2121

at University of M



ownloaded from

http://petrology.oxfordjournals.org/cgi/content/full/egs046/DC1


http://www.petrology.oxfordjournals.org/

http://www.petrology.oxfordjournals.org/


commonly partially altered to biotite, serpentine and, lesscommonly, orthoamphibole. Green^brown hornblende iscommon and forms slightly elongate subhedral to anhedralgrains 1^2mm in length (Fig. 3b and c). Garnet formslarge anhedral^subhedral (rarely euhedral) grainscommonly surrounded by a corona rich in plagioclasewith or without orthopyroxene, green^brown hornblendeand magnetite that separates garnet from clinopyroxene(Fig. 3c and d).Where present, opaque phases form subhe-dral equant grains that are generally less than 0·5mmacross (Fig. 3a, c and d). Magnetite is the dominant oxidephase, although ilmenite occurs in many samples; sulphideminerals are also common. Fe^Ti oxides may be sur-rounded by a thin rind of green hornblende, biotite or, in

garnet-bearing samples, garnet (Fig. 3a and d). Biotiteoccurs only as small anhedral grain clusters replacingorthopyroxene and/or Fe^Ti oxides (Fig. 3a). Quartz isfound only rarely in garnet-absent meta-leucogabbro, andin some cases is interpreted to represent annealed veinlets.Although many samples share a common mineral as-

semblage, the relative proportion of minerals varies greatlyfrom sample to sample (Fig. 3). Clinopyroxene is the mostabundant phase in most metagabbro samples, with the ex-ception of rare meta-leucogabbro and garnet^orthopyroxene-rich rocks, with some samples containing450 vol. % clinopyroxene. Orthopyroxene contents rangefrom zero to around 20 vol. %. The proportion of plagio-clase varies widely, from in excess of 50% in rare

Fig. 3. Petrography of granulite-facies metagabbros illustrating the variety of assemblages and modal abundances of phases (scale bar repre-sents 1mm). (a) Plagioclase-rich sample dominated by clinopyroxene and plagioclase. Clinopyroxene contains fine exsolution lamellae of ortho-pyroxene and is altered around its margins to hornblende. The 1208 triple junctions between clinopyroxene grains should be noted. Grains ofFe^Ti oxides (mainly magnetite) are variably replaced by biotite (Badcall). (b) Plagioclase-poor sample comprising roughly equal amounts ofclinopyroxene, orthopyroxene and green^brown hornblende.This sample contains no Fe^Ti oxide phase (Ben Dreavie). (c) Garnetiferous meta-gabbro rich in clinopyroxene and hornblende. Garnet is anhedral and surrounded by an irregular corona rich in plagioclase. Small grains oforthopyroxene and rounded magnetite are common (Ben Dreavie). (d) Coarse-grained metagabbro comprising490 modal % garnet andclinopyroxene. Garnet is surrounded by a corona containing plagiclase, orthopyroxene and magnetite (Scourie). Thin sections (a)^(c) are30 mm thick, whereas (d) is around 50 mm thick.


2122

at University of M



ownloaded from


meta-leucogabbro samples (Fig. 3a) to 1^2 vol. % inmeta-melanogabbro (Fig. 3d). In general, plagioclase con-tents are below 30 vol. %. Primary (green^brown) horn-blende contents vary widely from zero to 30 vol. %.Garnet is absent from many rocks but may make up450vol. % in others. Samples volumetrically dominated bygarnet and clinopyroxene are common (Fig. 3d).

Felsic sheets and Cpx-pegmatitesFelsic sheets within the large mafic^ultramafic bodies aredominated by quartz and plagioclase (oligoclase^ande-sine) that together constitute490 vol. % of most samples.The proportion of these two minerals varies considerably,from samples containing roughly equal proportions ofquartz and plagioclase to others that are almost all plagio-clase. Quartz occurs as single anhedral grains or elongatedaggregates of subgrains. Single grains or grain aggregatesare generally 1^2mm, but may be in excess of 5mmacross. Plagioclase grains are generally subhedral to anhe-dral and of a similar grain size to quartz. Single plagio-clase grains are variably sericitized and commonlycontain larger randomly oriented crystals of muscoviteand/or epidote. In some cases rounded exsolved blebs ofK-feldspar occur within plagioclase (i.e. antiperthite).Many plagioclase grains are unzoned or exhibit weak con-centric compositional zoning. However, in plagioclase-richsamples, subhedral to euhedral more anorthite-rich coresare commonly overgrown by a pronounced morealbite-rich rim (Fig. 4a). The felsic sheets generally containa few per cent mafic minerals, most commonly retrogradebiotite, hornblende, epidote and/or chlorite. In rare casesclinopyroxene is preserved. Magnetite, zircon and rarerutile are accessory phases. K-feldspar is absent from mostfelsic sheets. Where present, it occurs as rare anhedralinterstitial grains (Fig. 4b). Sample GC59, collected fromthe 40m wide felsic sheet running through the top of theGorm Chnoc body, is notably different, containingabundant K-feldspar (as microcline), minor muscoviteand much less plagioclase than other felsic sheets (Fig. 4c).This sample is mineralogically and texturally a(micro)granite.The Cpx-pegmatites are broadly petrographically simi-

lar to the felsic sheets but are generally much coarser-grained, have a high ratio of plagioclase to quartz and con-tain a higher proportion of mafic minerals (Fig. 4d).Plagioclase is commonly sericitized. Single grains may be410mm across and are commonly antiperthitic, con-taining abundant rounded blebs of exsolved K-feldspar(Fig. 4d). Quartz is much rarer than in the felsic sheets,commonly forming55 vol. % of samples.Where fresh, de-formed grains of clinopyroxene may be several centimetresin length (Fig. 4d), although these are commonly flattenedand completely replaced by aggregates of acicularhornblende.

WHOLE -ROCK GEOCHEMISTRYAnalytical methodsThe analysed sample set comprises 33 metagabbroic rocks,11 samples from felsic sheets and six samples fromCpx-pegmatites. The metagabbroic rocks are subdividedinto garnet-absent (n¼17) and garnetiferous (n¼16) vari-ants. All chemical analyses were undertaken at the Depart-ment of Geosciences, Johannes Gutenberg University,Mainz. Representative whole-rock samples were firstcrushed with a hydraulic press and then milled in a ringand puck agate mill until a powder with a grain size55 mm was obtained. Whole-rock major and selectedminor and trace elements (Cr, Ni, Sc, Pb, Rb) were deter-mined by X-ray fluorescence (XRF) using a PhilipsMagiXPRO spectrometer. For major element analysis, aglass bead was produced by 14 times dilution of the rockpowder with Li2B4O7 powder and subsequent melting ina Pt cup followed by quenching in a Pt coquille. For traceelement analysis, 6 g of the rock powder was mixed with atwo-component epoxy and pressed to a powder presstablet for about 20 s and afterwards dried at 608C.Remaining trace elements (Ba, Th, U, Nb, Ta, La, Ce, Pr,Nd, Sr, Sm, Hf, Zr, Ti, Eu, Gd, Dy, Ho, Y, Er, Yb, Lu)were analysed by laser ablation inductively coupledplasma mass spectrometry (LA-ICP-MS) using an Agilent7500ce ICP-MS system fitted with a NewWave UP-213 LAsystem, in which rock powders were first melted to formglass beads on an iridium strip heater in an argon atmos-phere (Nehring et al., 2008). Before each analysis, back-ground measurements were performed for 60 s. Theconcentration of Ca in each sample as determined byXRF was used for internal standardization. Externalstandardization as well as instrument performance andstability were monitored by repeat measurements ofUSGS reference glass BCR-2G. The precision on BCR-2Gwas better than 10% for all elements and accuracies arewithin 5% except for Zr and Hf (10%). Detection limitsare lower than 0·05 mg g�1 except for Ba (0·1 mg g�1). Datareduction was carried out using the software Glitter.Table 1 shows analyses of representative samples.

MetagabbrosAlthough there is considerable compositional overlap,major element variation diagrams show that garnet-absentand garnet-present metagabbro exhibit some compos-itional distinctions. With the exception of a single leuco-gabbro sample that contains abundant veins of quartz,garnet-absent metagabbros have SiO2 contents rangingfrom 43 to 54wt % whereas garnetiferous samples havelower values (41^49wt %). Overall, garnet-absent sampleshave lower concentrations of Al2O3 (9^18wt %) andFeO* (5^17wt %, where FeO*¼ total Fe expressed asFeO) and higher Na2O (0·9^4·0wt %), K2O (0·2^1·2wt%) and XMg (0·58�0·21; 2s) compared with garnetiferous


2123

at University of M



ownloaded from


samples [Al2O3 10^24wt %; FeO* 9^19wt %; Na2O0·7^2·1wt %; K2O 0·2^2·4wt %; XMg 0·52�0·18 (2s);Fig. 5]. Concentrations of CaO (5·7^15·1wt %), TiO2

(0·2^2·1wt %) and MgO (3·7^14·9wt %) show widevariability but are similar in both types. The concentra-tion of P2O5 in most samples is 50·05wt % althoughfive samples show much higher values (40·10 to 0·77wt%) consistent with a high modal abundance of apatite.There are no strong compositional trends for any majorelement when plotted against SiO2. Al2O3 and Na2Oshow a weak negative correlation with MgO, and SiO2

shows a weak positive correlation against XMg. Plotsversus XMg show that the metagabbros are generallydepleted in SiO2 and TiO2 and enriched in FeO* relativeto mid-ocean ridge basalt (MORB) and Archaean tholei-itic basalts (Fig. 5a and b).

Figure 6a and b shows concentrations of selected traceelements in metagabbro samples normalized to primitivemantle values (McDonough & Sun, 1995), in which theelements are ordered from left to right by increasingcompatibility in oceanic basalts (Hofmann, 1988). Shownfor reference is a compositional field representative ofArchaean tholeiites based on analyses (n¼ 72) from theliterature (Kerrich et al., 1999; Polat, 2009; Ordo¤ n‹ ez-Caldero¤ na et al., 2011).In general, the metagabbro samples are depleted in the

moderately compatible high field strength elements(HFSE) Hf, Zr and Ti relative to Archaean tholeiites,with some samples close to primitive mantle values. Ti/Zrratios are highly variable with values of 43^225 (aver-age¼121) in the garnet-absent metagabbros and 80^690(average¼ 242) in garnetiferous samples, even excluding

Fig. 4. Petrography of Cpx-pegmatites and felsic sheets (scale bar represents 1mm). (a) Plagioclase-rich felsic sheet. Many plagioclase grainscontain euhedral^subhedral more anorthite-rich cores surrounded by more albite-rich rims.The contact between the core and rim compositions(arrowed) is sharp. These microstructures are consistent with fractional crystallization of plagioclase (Cnoc Gorm). (b) Rare example ofK-feldspar (microcline) within a felsic sheet. The microcline occurs as angular interstitial grains suggesting late crystallization from melt(Badcall). (c) (Micro)granite sheet comprising subequal quantities of quartz and K-feldspar (microcline) with minor plagioclase from thelarge 40m wide felsic sheet running through the Gorm Chnoc body (Gorm Chnoc). (d) Cpx-pegmatite comprising large flattened peritecticclinopyroxene and sericitized plagioclase grains. Clinopyroxene is partially replaced by hornblende at its margins. Plagioclase contains abun-dant rounded exsolved K-feldspar blebs (Badcall).


2124

at University of M



ownloaded from


Table 1: Representative major (weight per cent) and trace (parts per million) element analyses of the metagabbroic rocks,

felsic sheets and Cpx-pegmatites

Lithology: Cpx-pegmatite Felsic sheet

Sample: L10/AU89 L10/GC66 L10/ST40 L10/ST49 L10/AU88 L10/BC27 L10/BD78 L10/115

Locality: Achiltibuie Gorm Chnoc Strathan Strathan Achiltibuie Badcall Ben Dreavie Stoer

SiO2 53·25 55·59 49·64 67·46 71·41 74·39 60·43 69·17

TiO2 0·19 0·94 0·36 0·42 0·28 0·06 0·52 0·11

Al2O3 17·26 23·2 14·5 16·38 15·67 14·29 18·18 17

FeO* 7·23 3·7 11·42 2·56 1·69 1·4 3·69 0·94

MnO 0·13 0·1 0·19 0·05 0·02 0·02 0·09 0·02

MgO 7·33 1·35 8·07 1·32 0·76 0·75 2 0·53

CaO 9·14 7·04 9·31 4·29 3·07 1·88 6·4 3·96

Na2O 3·39 5·78 3·11 4·31 4·94 5·09 5·38 4·83

K2O 0·34 0·24 0·85 0·53 0·9 0·71 1·24 0·92

P2O5 0 0·07 0·01 0·09 0·02 0 0·26 0·1

LOI 1·31 0·58 4·86 1·19 0·63 1·43 1·31 1·32

Sum 99·57 98·59 102·32 98·6 99·39 100·02 99·5 98·9

Cs 0·1 0·09 0·04 0·1 0·09 0·78 0·12 0·13

Rb 4 3 4 3 12 8 9 12

Ba 88·7 84·5 270·2 590·5 1078·7 160·4 455·7 817·7

Th 0·03 0·06 0·19 0·02 0·74 0·06 0·44 0·26

U 0·02 0·03 0·07 0·01 0·11 b.d. 0·42 0·05

Nb 0·62 7·57 2·18 2·48 1·49 0·2 3·81 0·37

Ta 0·06 0·49 0·06 0·05 0·04 0·03 0·1 0·04

La 4·95 5·48 19·04 10·17 15·09 5·31 19·91 25·67

Ce 9·24 9·75 51·77 19·3 25·71 6·99 42·85 43·44

Pb 6 4 5 8 6 5 16 16

Pr 1·09 1·03 5·8 2·01 2·11 0·54 5·05 4·39

Nd 4·9 4·1 21·26 8·14 6·84 1·78 20·43 16·33

Sr 319·8 395·1 199·1 555·6 382·9 361·7 854·8 510·8

Sm 1·22 0·72 3·46 1·48 0·75 0·24 3·38 2·25

Hf 0·56 0·12 0·62 0·26 3·58 0·18 3·75 0·67

Zr 13 3·6 19·4 10·1 134·8 4·8 198·5 29·9

Eu 0·51 0·88 0·91 0·83 0·69 0·51 1·31 1·3

Gd 1·17 0·51 2·86 0·9 0·6 0·26 2·11 1·29

Dy 1·07 0·36 2·49 0·43 0·24 0·1 1·27 0·54

Ho 0·2 0·06 0·5 0·07 0·05 b.d. 0·21 0·09

Y 5·54 1·84 13·03 1·86 1·18 0·39 6·02 2·37

Er 0·65 0·19 1·42 0·19 0·1 b.d. 0·61 0·19

Yb 0·64 0·17 1·34 0·15 b.d. 0·1 0·57 0·13

Lu 0·09 0·03 0·19 0·02 0·04 b.d. 0·09 0·02

Sc 16 4 44 8 3 1 9 4

Ni 170 19 92 26 13 52 59 53

Cr 114 68 39 22 14 15 31 12

(continued)


2125

at University of M



ownloaded from


Table 1: Continued

Lithology: Metagabbro (no garnet) Metagabbro (garnetiferous)

Sample: L10/

AU83

L10/

BC29

L10/

BD77

L10/

GC58

L10/

ST48

L10/

TB21

L10/

ST50

L10/

AU84

L10/

BS8

L10/

TB19

L10/

DB7

L10/

GC63

SC01 L10/

ST47

Locality: Achiltibuie Badcall Ben Gorm Strathan Cnoc Strathan Achiltibuie Ben Cnoc Drumbeg Gorm Scourie Strathan

Dreavie Chnoc Gorm Strome Gorm Chnoc

SiO2 51·13 47·43 49·12 48·52 48·81 52·07 53·26 48·86 41·41 46·49 46·27 47·45 45·74 46·05

TiO2 0·52 1·46 0·42 0·28 1·03 0·28 0·82 0·56 2·08 0·52 0·99 0·37 0·87 1·03

Al2O3 13·31 14·89 9·73 14·88 17·21 15·45 17·27 11·52 14·46 14·74 14·89 20·13 14·2 14·48

FeO* 11·43 14·86 11·73 6·92 10·22 7·42 6·38 10·6 19·38 14·82 13·44 9·31 14·53 14·56

MnO 0·25 0·29 0·21 0·15 0·19 0·15 0·11 0·14 0·3 0·25 0·26 0·22 0·26 0·26

MgO 8·69 7·13 14·87 9·78 5·72 9·13 3·68 12·84 7·72 11·44 7·3 6·53 9·02 9

CaO 9·2 10·03 10·33 14·39 12·67 8·84 11·72 11·68 12·93 9·92 13·11 11·5 12·99 11·64

Na2O 2·13 2·29 1·34 1·47 2·24 3·03 3·8 1·41 0·79 0·75 1·34 2·05 1·15 0·93

K2O 0·6 0·31 0·52 0·58 0·24 1·17 0·34 0·44 0·1 0·27 0·6 0·09 0·06 0·58

P2O5 0·02 0·09 0·01 0 0·04 0·01 0·04 0·02 0·03 0·02 0·05 0·01 0·03 0·05

LOI 1·99 0·51 0·53 1·9 0·19 1·44 0·57 0·54 -0·34 0·09 0·74 0·75 0 0·32

Sum 99·27 99·29 98·81 98·87 98·56 98·99 97·99 98·61 98·86 99·31 98·99 98·41 98·85 98·9

Cs 0·25 0·31 0·12 1·18 0·04 2·62 0·01 0·03 0·06 0·24 0·15 0·08 0·01 0·05

Rb 8 6 6 21 4 75 1 5 6 6 15 3 0 10

Ba 207·1 92·6 54·8 77 82·3 100·3 71·5 66·1 17·7 38·2 196·2 22·1 6·7 196

Th 0·32 0·29 0·46 0·39 0·05 0·37 0·04 0·18 b.d. 0·64 b.d. 0·02 0·01 0·29

U 0·2 0·11 0·14 0·14 0·02 0·11 0·02 b.d. 0·03 0·15 b.d. 0·01 0·03 0·05

Nb 1·84 4·96 4·93 0·62 2·86 0·44 2·24 1·22 3 1·09 1·8 0·51 1·07 1·91

Ta 0·11 0·3 0·46 0·03 0·15 0·09 0·14 0·06 0·15 0·06 0·09 0·04 0·04 0·09

La 8·86 9·74 7·89 1·63 3·2 8·5 2·97 3·87 0·7 4·86 1·23 1·23 1·3 3·92

Ce 21·65 23·78 23·31 3·05 8·45 17·32 7·77 9·4 3·04 13·02 4·56 3·75 4·79 10·41

Pb 4 4 6 5 1 11 1·69 3 3 3 1 4 0·22 4

Pr 2·48 3·25 3·13 0·41 1·25 1·89 1·18 1·16 0·64 1·82 0·81 0·63 0·89 1·43

Nd 9·88 14·91 12·94 2·08 6·62 7·01 6·53 4·71 4·26 8·41 5·1 3·49 5·37 7·16

Sr 187·9 180·1 50·3 150·2 130 281·3 156·4 106·4 54·5 91·4 154·8 204·8 51·6 71·1

Sm 1·9 3·72 2·9 0·76 2·18 1·63 2·32 1·24 2·02 2·1 2·04 1·34 2·1 2·37

Hf 0·93 1·6 1·36 0·32 1·09 0·59 1·41 0·71 0·69 0·83 1·19 0·42 0·79 1·17

Zr 27·9 48 30·5 8·8 32·6 18·9 46·5 19·4 17 29·7 34 12·1 20·5 30·9

Eu 0·69 1·11 0·7 0·32 0·77 0·49 0·76 0·48 1·08 0·72 0·77 0·48 0·76 0·85

Gd 2·15 4·43 2·86 1·01 2·86 1·51 2·88 1·42 3·38 2·24 3·06 1·89 2·94 3·46

Dy 2·27 5·13 2·92 1·27 3·43 1·56 3·57 1·37 5 2·3 3·89 4·01 3·93 4·16

Ho 0·5 1·11 0·6 0·28 0·74 0·32 0·73 0·27 1·2 0·46 0·85 1·07 0·82 0·93

Y 12·92 28·66 17·35 7·06 18·69 8·74 18·61 6·21 30·18 12·02 21·14 29·65 21·74 23·97

Er 1·41 3·13 1·73 0·78 2·12 0·85 2·11 0·65 3·51 1·31 2·38 4·07 2·5 2·69

Yb 1·55 3 1·75 0·77 2·11 0·92 2·02 0·53 3·7 1·28 2·42 5·96 2·62 2·75

Lu 0·23 0·47 0·26 0·12 0·29 0·13 0·3 0·06 0·58 0·18 0·35 1 0·4 0·42

Sc 48 45 48 49 49 42 46 40 65 36 48 48 58 56

Ni 97 122 298 88 191 94 153 404 84 230 197 26 112 133

Cr 486 170 1879 60 291 67 292 1377 139 260 213 36 366 321

LOI, loss on ignition; b.d., below detection limit.


2126

at University of M



ownloaded from


the samples containing high P2O5 contents, which havevalues that are both anomalously high and low comparedwith the other samples. Relative to Archaean tholeiites,several garnetiferous samples and four garnet-absentsamples are depleted inTa and Nb, and several have con-centrations below primitive mantle values. Most garneti-ferous samples are also depleted in U and Th, althoughconcentrations vary by almost two orders of magnitude.Most garnet-absent samples have higher concentrations ofU and Th that are at the lower end of those recorded byArchaean tholeiites, and concentrations of these elementsare far less variable. Both variants of metagabbro showLILE (Cs, Rb, Ba, Pb, Sr and K) concentrations that arecomparable with those of Archaean tholeiites and up to100 times primitive mantle values.Figure 7a and b shows the abundances of REE in

garnet-absent and garnetiferous metagabbros respectively,normalized to CI chondrite values of McDonough & Sun(1995). Samples with high P2O5 contents are shown(dashed) but are not discussed in detail, as their traceelement concentrations are anomalous with respect tomost of the other data. Garnet-absent samples are

generally weakly fractionated and all have (La/Lu)N 41.Most have (La/Lu)N ratios below three; three sampleswith higher values (4·0, 6·1 and 7·8) are within a fewmetres of contacts with large felsic sheets or TTG gneiss.Garnet-absent samples have flat middle to heavy REE(MREE and HREE respectively) patterns, with (Gd/Lu)N ratios of 1·0^1·9 in all samples but one, values similarto or less than those recorded by Archaean tholeiites.Garnet-absent samples show both small positive and nega-tive europium anomalies (Eu/Eu*¼ 0·6^1·4), although themajority, particularly those with higher overall REE abun-dances, are negative. Most garnet-absent samples havelight REE (LREE) concentrations that lie within the fieldof Archaean tholeiites, with patterns that are weakly frac-tionated with (La/Sm)N ratios41. The REE patterns forgarnet-absent samples are almost indistinguishable fromthose of the hornblende-metagabbro suite (HMS) fromthe Gruinard Bay area in the southernmost central regionreported byWhitehouse et al. (1996) (Fig. 7a and b).In comparison, garnetiferous samples have REE pat-

terns that are generally flat with (La/Lu)N51 in mostsamples. Concentrations are mostly at the lower range of

SiO2

35

45

55

65

0.2 0.4 0.6 0.8X(Mg)

garnet-absent metagabbro

garnetiferous gabbro

wt%

MORB

Archaean tholeiites

Oceanic island basalts

Island-arc volcanics

(a) garnet-absent metagabbro

garnetiferous gabbro

(b)

0.1

1

10

0.2 0.4 0.6 0.8X(Mg)

wt%

TiO2

MORB

OIB

AT

IAV

(c)

Al2O3

10

20

30

40

40 50 60 70 80

cpx-pegmatite

metagabbro (no g)

metagabb (with g)felsic sheets

SiO2 (wt%)

wt%

An

Ab

plagioclase

quartz

cpx-pegmatite

metagabbro (no g)

metagabb (with g)felsic sheets

(d)

CaO

0

4

8

12

16

40 50 60 70 80

SiO2 (wt%)

wt%

Fig. 5. Major element variation diagrams. (a) SiO2 and (b) TiO2 plotted against XMg [¼ molar Mg/(MgþFe)] showing the compositions ofthe metagabbros relative to Archaean tholeiites (AT), MORB, ocean island basalts (OIB) and island arc volcanics (IAV) of Arndt et al. (1997);(c) Al2O3 and (d) CaO plotted against SiO2, showing the composition of the metagabbros, felsic sheets and Cpx-pegmatites.


2127

at University of M



ownloaded from


those recorded in Archaean tholeiites. The patterns arestrikingly similar to that for normal (N)-MORB, albeit atlower overall abundances. LREE concentrations are not-ably lower than those in garnet-absent samples, and onlythree samples exhibit (La/Sm)N 41; all of these samplesare close to a contact with either TTG gneiss or a largefelsic sheet. Concentrations of MREE and HREE are simi-lar to those of garnet-absent metagabbro, although twosamples that are very rich in garnet (450 vol. %) show ex-treme HREE enrichment [(Gd/Lu)N¼ 0·2^0·3]. Otherthan a single sample of leucosome-rich diatexite, garneti-ferous samples do not have pronounced Eu anomalies.

Felsic sheets and clinopyroxene pegmatitesThe felsic sheets within the large mafic^ultramafic bodiesare intermediate to acidic in composition, with SiO2 con-tents in the range 58^76wt %. A moderate to stronglinear trend exists for Al2O3 (14^25wt %; Fig. 5c), TiO2

(0·1^0·5wt %), FeO* (0·3^3·7wt %), MgO (0·2^2·0wt%) and CaO (2·1^7·5wt %; Fig. 5d), which are negativelycorrelated against SiO2, although samples with the lowestSiO2 values commonly depart from these trends. BothTiO2 and FeO* are positively correlated with MgO.Concentrations of Na2O (3·1^8·9wt %) and K2O(0·5^4·7wt %), and values of XMg (0·2^0·5), show fargreater scatter. The smaller number of analyses of

Cpx-pegmatite samples show a significant range of majorelement concentrations, in part owing to the difficulties inobtaining samples large enough to be representative.Nevertheless, clear trends exist for several major elements.SiO2 ranges from 45 to 60wt % and shows a positive cor-relation with Al2O3 (15^25wt %; Fig. 5c) and Na2O (2·0^7·7wt %) and a negative correlation with CaO (5·7^9·5wt %; Fig. 5d), FeO* (0^11wt %) and MgO (0^13wt%). Al2O3 and Na2O are negatively correlated, and CaOand FeO* positively correlated, with MgO. For mostmajor elements the Cpx-pegmatites plot in a field inter-mediate between the felsic sheets and the metagabbros(Fig. 5c and d).When plotted on a normative feldspar classification dia-

gram (Barker, 1979), the felsic sheets straddle the tonalite^trondhjemite join and lie within the range of experimentalglass compositions produced by partial melting ofArchaean tholeiite (Winther, 1996; Fig. 8), although thisfield is large. A single analysis from the large compositesheet running through the Gorm Chnoc body (sampleGC59; star in Fig. 8) lies outside this general trend andwithin the field of granite. The Cpx-pegmatites showgreater scatter and are generally richer in An, all plottingwithin the compositional field of tonalite (Fig. 8). Theobserved abundances of quartz, plagioclase and clinopyr-oxene in the Cpx-pegmatites and felsic sheets correspond

0.1

1

10

100

1000

Cs Rb Ba Th U Nb Ta K Pb Sr Hf Zr Ti Y

garnet-absent metagabbro(a)

0.1

1

10

100

1000


garnetiferous metagabbro(b)

ave. crust ave. crust

0.1

1

10

100

1000


TTG gneiss

metagabbros

felsic sheets(c)GC59

0.1

1

10

100

1000


TTG gneiss

metagabbros

cpx pegmatites(d)

N–MORB

Archaean tholeiite

N–MORB

Archaean tholeiite

Fig. 6. Primitive mantle normalized patterns for selected trace elements ordered by compatibility in oceanic crust. The fine dashed lines in (a)and (b) are those samples with high P2O5 contents. The bold dashed lines in (b) are samples rich in garnet. The compositional field forArchaean tholeiites is a compilation of data (n¼ 72) from Kerrich et al. (1999), Polat (2009) and Ordo¤ n‹ ez-Caldero¤ na et al. (2011).


2128

at University of M



ownloaded from


well to the calculated CIPW normative abundances, inwhich the Cpx-pegmatites contain an average (in vol. %)of 1·7% normative quartz, 72% plagioclase and 7% diop-side, whereas the felsic sheets contain an average of 23%quartz, 67% plagioclase and 0·9% diopside.Figure 6c and d shows primitive-mantle normalized

trace element abundance patterns for the felsic sheets andCpx-pegmatites, on which the patterns defined by themetagabbros, 21 analyses of TTG gneisses bordering themetagabbros and the average continental crustal compos-ition of Rudnick & Gao (2003) are shown for reference.With the exception of the single sample of the large com-posite sheet running through the Gorm Chnoc body(GC59), the felsic sheets have the lowest concentrations ofTi and Yof all samples. Abundances of Hf, Zr, Th and Uare highly variable and span the range recorded in themetagabbros, although several samples have the highestconcentrations. Concentrations of Pb and K are at theupper range or in excess of those in the metagabbros. Formost samples, Ba and Sr concentrations are the highest ofall measured samples. A pronounced negative anomalyexists forTa and Nb, with concentrations of these elements,

as well as Cs and Rb, similar to those in the metagabbros.Sample GC59 has the highest concentrations of Rb, Kand Pb, and the lowest Sr contents of all the felsic sheets.Although two samples show the lowest measured values ofHf and Zr, the Cpx-pegmatites generally fall within thecompositional range defined by the felsic sheets, but showsignificant scatter. In general, both the felsic sheets andCpx-pegmatites show a good correspondence to the TTGgneisses, although several samples are relatively depletedin Nb,Ta and HFSE.With the exception of Ba, Pb and Sr,most samples are depleted relative to the average continen-tal crustal composition.Figure 7c and d shows chondrite-normalized abun-

dances of REE in the felsic sheets and Cpx-pegmatites.The REE profiles for both are fractionated with enrich-ment in LREE relative to HREE, which is most pro-nounced for the felsic sheets. (La/Lu)N ratios are 22^155for the felsic sheets and 5^19 for the Cpx-pegmatites.LREE abundances in both overlap those in garnet-absentmetagabbro samples, but are generally higher than in gar-netiferous samples. Although sample GC59 shows a weaknegative Eu anomaly, most felsic sheets show positive Eu

1

10

100

1000

La Ce Pr Nd Sm Eu Gd Dy Ho Er Yb Lu

metagabbro (no g)(a)

1

10

100

1000


(b) metagabbro (with g)

0.1

1000

1

10

100


(c) setitamgep xpcsteehs cislef

0.1

1000

1

10

100


(d)

GC59

TTG gneiss

metagabbros

TTG gneiss

metagabbros

Archaean tholeiite

HMS

Archaean tholeiite

HMS

N-MORBN-MORB

Fig. 7. Chondrite-normalized REE patterns. The fine dashed lines in (a) and (b) are those samples with high P2O5 contents. The bold dashedlines in (b) are samples rich in garnet. The compositional field for Archaean tholeiites is a compilation of data (n¼ 72) from Kerrich et al.(1999), Polat (2009) and Ordo¤ n‹ ez-Caldero¤ na et al. (2011). HMS, hornblende-metagabbro suite of Whitehouse et al. (1996).


2129

at University of M



ownloaded from


anomalies (Eu/Eu*¼ 0·83^9·5), which are most pro-nounced for those samples with the lowest overall abun-dance of REE. Similar features are exhibited by theCpx-pegmatites, albeit with more variability. HREE con-centrations are significantly lower in most samples offelsic sheets with respect to both the metagabbros andCpx-pegmatites.

DISCUSS IONField and petrographic evidenceMany of the important field relations described above wereeloquently summarized in the 1888 Geological Surveyreport (Peach et al., 1888, p. 388). Describing the Archaeancrystalline rocks, this report stated:‘A remarkable feature of these original gneisses is the oc-

currence among them of numerous masses of highly basicigneous rocks. . .. These patches of non-foliated igneousrock are intersected by veins of grey pegmatite varying inthickness from a few inches to several yards, consistingmainly of felspar and quartz usually opalescent. Occasion-ally a small quantity of pyroxene or hornblende is

associated with the quartz and felspar. . .. the pegmatitesmerge into the grey highly quartzose bands, consistingmainly of opalescent quartz and felspar. The conclusion is,therefore, obvious that the original types of gneiss in thewest of Sutherland have been formed out of eruptive basicrocks and the pegmatites developed in them prior to thefoliation.’Although the consensus of current opinion would hold

that the basic rocks are of intrusive rather than extrusiveorigin and are now granulite-facies metamorphic rocks(e.g. O’Hara, 1961; Bowes et al., 1964; Davies, 1974; Sillset al., 1982; Johnson & White, 2011), this passage containsthe critical observations regarding lithological relation-ships, which may be interpreted to suggest that: (1) theCpx-pegmatites (grey pegmatites) developed within, andfrom, the metagabbros (basic igneous rocks); (2) theCpx-pegmatites merge into the felsic sheets (highly quart-zose bands), suggesting they were contemporaneous andare genetically related; (3) all the rock types were presentunder granulite-facies conditions. The last sentence of thequotation shows that Peach and colleagues consideredthat the original types of gneiss (TTG gneisses) were

rObA

An

gran

odio

rite

tona

lite

trondhjemite

granite

cpx pegmatites

felsic sheets (this study)

Wolf & Wyllie1994

Rushmer 1991

Winther 1996Watkins et al. 2007

SC4

SC5

GC59

felsic sheets (Rollinson 1994)

Fig. 8. Normative feldspar compositions of the felsic sheets and Cpx-pegmatites. Superimposed are the fields of experimental glasses producedby partial melting of amphibolites (diagonal-line shading) and TTG gneisses (stippled). The small dots show the normative composition offelsic sheets from Rollinson (1994).


2130

at University of M



ownloaded from


formed from (partial melting of) the basic rocks, an ideathat is consistent with geochemical data (e.g. Rollinson &Fowler, 1987).The field evidence for in situ partial melting of metagab-

bro within the large maficûltramafic bodies throughoutthe central region is incontrovertible. However, micro-structural evidence is less clear owing to slow coolingrates (e.g. O’Hara & Yarwood, 1978; Barnicoat, 1987) andreworking experienced by all rocks during the protractedpost-Badcallian retrograde evolution. Mineral assemblageswithin the ultramafic layers at Scourie are consistent withpressures of 7^10 kbar and temperatures of 900^9508C(Johnson & White, 2011), conditions under which am-phibolites will melt by fluid-absent reactions consuminghornblende and plagioclase to produce peritectic clinopyr-oxene and up to 40 vol. % melt of tonalitic to trondhjemiticcomposition (e.g. Rushmer, 1991; Wyllie & Wolf, 1993).Pegmatitic leucosomes containing euhedral clinopyroxene,which are abundant in the metagabbroic rocks within thelarge maficûltramafic bodies (Fig. 2b^g), provide arecord of these melting reactions. The clear spatial associ-ation of peritectic clinopyroxene (the solid product of in-congruent melting) with the leucosome (crystallized fromthe melt) is critical to an interpretation of in situ melting,as opposed to an origin via injection of melt from a moredistal source. The inferred degree of partial melting basedon field evidence varies from a few volume per centwithin layered metatexites to420 vol. % or �20 vol. %in highly disrupted diatexites (Sawyer, 2008), and willhave been largely a function of bulk composition, in par-ticular the degree to which the metagabbros were hydratedprior to UHT metamorphism.Driven largely by deformation, melt produced in situ

would have moved down gradients in pressure, withmelt pathways controlled both by pre-existing structuralfabrics and syn-anatectic deformation (e.g. Brown et al.,2011). Melt would have drained into larger-scale (deci-metre- to metre-scale) channels now represented by thefelsic sheets, which provide a record of melt migrationaway from its source. Where melt extraction from theCpx-pegmatites into the felsic sheets was particularly effi-cient, trails of euhedral peritectic clinopyroxene recordthe channels through which melt drained (Fig. 2i). Thelarge modal abundance of plagioclase and the generaldeficiency in quartz and K-feldspar in many of the felsicsheets suggest that they are unlikely to represent primarymelt compositions. Euhedral cores of anorthite-richplagioclase suggest that the felsic sheets are plagioclasecumulates.Overall, the loss of melt resulted in a melanosome

depleted in plagioclase and hydrous phases and, in mostcases, rich in clinopyroxene, with or without garnet, horn-blende and orthopyroxene. Such melt loss is a prerequisitefor the preservation of anhydrous or near anhydrous

granulite-facies assemblages during cooling (White &Powell, 2002).Where contacts are exposed, decimetre- to metre-scale

felsic sheets merge into larger composite felsic bodies,some tens of metres across. Many of the felsic sheets form-ing the 40m wide composite body near the top of GormChnoc are granitic, containing abundant K-feldspar andminor plagioclase (Fig. 4c). Although highly disaggregatedby later deformation, this composite body may be continu-ous with similar sheet-like bodies occurring, for example,near the top of the Cnoc Gorm body, and may representthe major pathways that channelled evolved melt fractionsto higher crustal levels.With one exception (east of Clachtoll; Fig. 1), there is

little evidence for a clear spatial association betweengarnet and leucosome. Garnet is either confined to themelanosome or entrained as grains or grain clusterswithin the felsic sheets (Fig. 2j^l). The implication is thatgarnet first grew in subsolidus rocks at lower temperatures(e.g. amphibolite-facies conditions producing garnet am-phibolites). As temperatures rose beyond the solidus,either garnet was not a product of the melting reaction or,if it were, additional garnet growth occurred onpre-existing grains within the melanosome.Although the evidence presented here concentrates on

the large maficûltramafic bodies, smaller bodies of meta-gabbro, ranging from a few centimetres to several metresacross, are ubiquitous within the central region. Thesesmaller bodies, though generally lacking garnet, showidentical field relations to those described from the largerbodies, consistent with all such compositions havingundergone extensive partial melting (Fig. S1cê,Supplementary Data).

GeochemistryBulk-rock geochemistry shows that the metagabbroic rocksspan a wide compositional range, although they exhibitno strong major element correlations. In general, the meta-gabbros are depleted in Ti, Zr and Hf relative toArchaean tholeiites (Figs 5a), and it is likely that this is aprimary feature of the magmas. The relative depletion inSiO2 (Fig. 5a) may also be primary, but is also consistentwith loss of a felsic melt fraction. The negative Ta^Nbanomalies that characterize all samples are likely to reflectcharacteristics of the source region, most probably frac-tionation of amphibole and/or rutile during generation ofthe original gabbroic melt (e.g. Foley et al., 2002).However, in most other respects the metagabbros are com-positionally similar to Archaean tholeiites (Condie, 1976;Sills et al., 1982; Arndt et al., 1997; Figs 6 and 7).Concentrations of LILE (Cs, Rb, Ba, K, Pb) within the

metagabbros are similar to those recorded in Archaeantholeiites (Fig. 6a and b). This is inconsistent with a modelof partial melting and melt loss, in which these strongly in-compatible (and mobile) elements should have been


2131

at University of M



ownloaded from




partitioned into the melt, depleting the residue (e.g. Xiong,2006). This may reflect a primary (i.e. pre-metamorphic)enrichment of these elements in the mafic magmas, but ismore likely to reflect contamination of the metagabbrosby LILE from the surrounding felsic TTG gneisses and/ormicaceous rocks during emplacement and crystallizationand/or during the long-lived UHT granulite-facies meta-morphic event (e.g. Fowler, 1986; Whitehouse et al., 1996;Rollinson & Gravestock, 2012).Although there are differences in the major and trace

element composition of garnet-absent versus garnetiferoussamples, in many cases it is difficult to determine the rela-tive contribution to these differences from primary (mag-matic) versus secondary (metamorphic including partialmelting) processes. However, the layered nature of thelarge maficûltramafic bodies (Fig. 2a) demonstrates thatprimary compositional variations existed prior to meta-morphism. In both variants, MREE and HREE patternsare generally flat and at concentrations at the lower rangeof, or below, those of Archaean tholeiites. However, garne-tiferous samples are characterized by lower LREE concen-trations and (La/Sm)N51, whereas garnet-absent samplesare relatively enriched in LREE and have (La/Sm)N41.The LREE-enriched patterns of garnet-absent samples,which are strikingly similar to the suite of coarse-grainedhornblendites and metagabbros from the Gruinard area[the ‘HMS’ suite of Whitehouse et al. (1996)], might be in-terpreted to suggest derivation from an LREE-enrichedsource (similar to E-MORB), although this is consideredto be an unlikely explanation. In many cases bothLREE-enriched (garnet-absent) and LREE-depleted (gar-netiferous) layers occur within the same maficûltramaficbody, and although more than one source region is pos-sible, it is more likely that these bodies evolved from asingle source. Samples of both garnet-absent and garneti-ferous metagabbro within a few metres of the host TTGgneisses show the highest enrichment in LREE, suggestingthat the enrichment is probably due to contaminationwith neighbouring rocks. Importantly, enrichment in bothLILE and LREE is also evident in the trace element pat-terns for ultramafic layers (not shown) within the largemaficûltramafic bodies.Various lines of evidence suggest a strong mineralogical

control on trace element distributions. Samples showinganomalous trace element concentrations in bothgarnet-absent and garnetiferous metagabbros are thosewith the highest P2O5 concentrations (Figs 6 and 7), con-sistent with a (relatively) high abundance of apatite thatwould exert a strong control on trace element concentra-tions, particularly for the REE (Nehring et al., 2010).Garnetiferous samples generally have lower XMg, SiO2,Na2O, K2O, Th, U, Nb, Ta and LREE, and higher Al2O3

and FeO* compared with garnet-absent samples. In gen-eral, the samples containing the highest modal abundance

of garnet show the most extreme enrichment or depletionin these elements (e.g. Figs 6 and 7). Field evidence suggeststhat garnet first grew early with respect to the UHTgranulite-facies peak, prior to the onset of melting.The ini-tial subsolidus growth of garnet in some layers but notothers must have been due to variations in the originalmajor element bulk composition of the layers, with garnetmost probably growing in those layers with (relatively)low XMg and high Al2O3. Although the original chemicalcomposition of the layers controlled where garnet grew, astemperatures (and diffusion rates) rose garnet controlledthe distribution of particular trace elements within thegarnetiferous layers. For example, the depletion in LREEof garnetiferous samples [(La/Sm)N51] relative togarnet-absent samples [(La/Sm)N41] reflects the very lowpartition coefficients for LREE in garnet relative to theother major phases (e.g. Nehring et al., 2010). The lowestconcentrations of Ce, Pr and Nd occur within the samplecontaining the highest modal abundance of garnet, al-though the upturn of the REE pattern for this sample to-wards La supports LREE mobility (Fig. 7b). Althoughvariations in HREE concentrations in garnetiferous sam-ples may in part be due to variations in the modal abun-dance of garnet, this cannot account for similar variationsin garnet-absent samples, which must reflect differences inthe original compositions of the garnet-absent layers.Garnet also has partition coefficients for Nb and Ta thatare significantly lower than for clinopyroxene and horn-blende, which might explain the lower concentrations ofthese elements in garnetiferous samples, although vari-ations in the source composition cannot be discounted.The geochemistry of the felsic sheets shows that several

major elements are strongly correlated with SiO2 (Fig. 5cand d) and that they have strongly fractionated REE pat-terns. The trace element patterns are broadly similar tothe surrounding TTG gneisses (Figs 6 and 7), althoughthe felsic sheets have low relative concentrations of Nb,Ta, HFSE and MREE^HREE and exhibit pronouncedpositive Eu anomalies (Fig. 7). The low relative concentra-tions of HFSE, Nb andTa in the felsic sheets mirror similardepletions in their metagabbroic parents. Like the TTGs,the felsic sheets are depleted in LILE (Fig. 6), consistentwith loss of an evolved melt fraction and/or depletion ofthese elements in the source rocks. However, there is noevidence for primary LILE depletion in the metagabbros,suggesting that melt loss is a more likely explanation. Theextreme variation in concentrations of Th, U, Hf and Zrwithin the felsic sheets (Fig. 6c) is likely to reflect variableretention of zircon as it crystallized from the melt on cool-ing, whereas large variations in HREE contents (Fig. 7c)probably reflect variable entrainment of garnet from themelanosome.Excluding the single granitic sample (GC59), the felsic

sheets have normative feldspar proportions that lie within


2132

at University of M



ownloaded from


the field of experimental glasses of Winther (1996) derivedfrom fluid-absent partial melting experiments on hydratedArchaean tholeiites (Fig. 8), although this field is largeand includes experiments performed under widely rangingP, T conditions and with highly variable H2O contents.However, on field and petrographic grounds (Fig. 4a)many of the felsic sheets are inferred to be plagioclase cu-mulates. Fractional crystallization of plagioclase with orwithout quartz in the felsic sheets is supported by themajor element data, with samples defining a linear arrayin SiO2 vs Al2O3 space that projects from intermediateto albite-rich plagioclase compositions towards quartz(Fig. 5d), and by their positive europium anomalies(Fig. 7c). These samples also contain high contents of Sr(up to 855 ppm) that are at the upper range or in excessof those measured in theTTG gneisses, whereas concentra-tions of K are similar in both (Fig. 6c). Sample GC59shows the lowest Sr and highest K and Rb concentrations,suggesting that it may represent an evolved melt fraction,consistent with the data of Rollinson (1994).The normativefeldspar composition of the Cpx-pegmatites is also broadlyconsistent with an origin via in situ partial melting of meta-gabbro, although many of the geochemical data reflect thefact that samples comprise a mixture of peritectic clinopyr-oxene (and/or hornblende) and a coarse-grained leuco-some dominated by plagioclase. The REE patterns for theCpx-pegmatites are similar to those of the felsic sheets,with the flatter LREE patterns in the Cpx-pegmatitesbeing due to the presence of clinopyroxene and/or horn-blende. The tonalite^trondhjemite felsic sheets occurringwithin the large mafic^ultramafic bodies are composition-ally distinct from those occurring within theTTG gneissesaround Scourie, which range in composition from tonaliteto granite (Rollinson, 1994; Fig. 8), suggesting a differentorigin.

REE modellingFigure 9 shows the results of REE modelling to test the hy-pothesis that the felsic sheets evolved via partial meltingof the metagabbros followed by segregation and fractionalcrystallization of this melt. For simplicity, the starting com-position for the modelling is a flat REE pattern, typical ofArchaean tholeiite (Fig. 7), at 15 times chondrite valuesçalthough the metagabbro samples show significant REEvariation, the mean of the data (excluding those sampleswith high P2O5 contents) yields a weakly fractionatedREE pattern with (La/Lu)N¼1·31 and a mean concentra-tion 15 times chondrite values. The modelling uses threemodal compositions (A, B and C) for the solid residuethat broadly correspond to those of the samples shown inFig. 3b, c and d, respectively. In terms of the volume percent of the phases plagioclase, clinopyroxene, hornblende,orthopyroxene and garnet, these are, respectively: 10%,50%, 20%, 20%, 0% for Composition A; 10%, 40%,30%, 10%, 10% for Composition B; 10%, 45%, 0%, 5%,

40% for Composition C (Fig. 9). REE partition coefficientsfor clinopyroxene, hornblende, orthopyroxene and garnetare taken from Foley (2008) and Nehring et al. (2010).Partition coefficients for plagioclase were calculated for amelt SiO2 content of 65wt %, appropriate for fluid-absentmelting of amphibolite at 10 kbar (Wolf & Wyllie, 1994),using the equations of Be¤ dard (2006).Figure 9a^c shows the results of equilibrium (batch)

melting of compositions A, B and C respectively at meltfractions (F) of 5, 10, 20 and 40%. The garnet-absentsample, A, generates melts with unfractionated MREE^HREE and moderately enriched LREE. Composition Bwith minor garnet shows similar LREE enrichment butmoderate depletion in MREE^HREE. The garnet- andclinopyroxene-rich sample C generates highly fractionatedREE patterns with La/Lu ratios of 66 at the lowest (5%)melt fractions. None of the modelled REE patterns predictany Eu anomaly in the melt. The composition of the resi-due (grey shaded areas in Fig. 9a^c) shows an increasingdepletion in LREE and enrichment in HREE withincreasing amounts of garnet in the residue. The relativeLREE enrichment exhibited by many of the garnet-absentsamples (Fig. 7a) implies either a different starting com-position for these rocks or LREE enrichment owing tointeraction with surroundingTTG gneisses.Figure 9d^f shows the composition of plagioclase crys-

tallizing from the (segregated) liquid produced by 5%partial melting of the starting composition coexistingwith the solid residue compositions A, B and C and withvarying amounts of melt remaining (75%, 50%, 25%,10% and 5%); Fig. 9g^i shows the same information butfor plagioclase crystallizing from the 20% melt fraction.In both cases the appropriate melt compositions are alsoshown. In all cases, the REE patterns for plagioclase crys-tallizing earlier (i.e. with more melt remaining) showlower overall concentrations and more pronounced Euanomalies than those crystallizing later (i.e. with lessmelt remaining). These patterns are strikingly similar tothe REE patterns of the felsic sheets shown in Fig. 7c(superimposed on Fig. 9), lending support to the stated hy-pothesis. Although the model considers only fractionalcrystallization of plagioclase, the addition of quartz as afractionating phase has no effect on the modelled REEpatterns.The fit of the model data to the measured composition of

the felsic sheets, particularly with respect to HREE, isgenerally best for composition A, consistent with thefield evidence that garnet-absent metagabbro is volumet-rically more abundant than garnetiferous metagabbro.Additionally, the fit is better for fractional crystallizationfrom low initial melt fractions, consistent with efficientdeformation-driven segregation of small batches of meltaway from the metagabbro melanosome (e.g. Rosenberg& Handy, 2005). Importantly, the modelled patterns show


2133

at University of M



ownloaded from


good correspondence to the measured data only providedmost of the evolved melt fraction was lost from the system.Although the modelled compositions of the plagioclase

cumulate broadly correspond to the measured values ofthe felsic sheets, several of the variables are poorly con-strained and the modelling does not account for accessoryphases in the residue or material entrained from the resi-due into the melt, all of which may have a significanteffect on the modelled patterns. Of particular importanceis the starting composition, mineralogical composition ofthe residue and the initial melt fraction produced.Although the starting composition assumes a flat REE

pattern, a slightly fractionated array with relative LREEenrichment is consistent with literature data on Archaeantholeiites and the data presented here. The presence andabundance of major minerals within metagabbro samplesis highly variable (Fig. 3), and will lead to a similarlywide variation in the modelled trace element compositionof the melt and residue. In addition, within any particularlayer or outcrop of metagabbro there is no way of reason-ably estimating the proportion of melt that it producedbased, for example, on the abundance of leucosome, giventhat much of the melt is inferred to have drained into thefelsic sheets. The fertility of any particular rock will have

Fig. 9. Results of REE modelling. (a^c) Compositions of melts produced by batch melting of a starting composition with 15 times chondritevalues with melt fractions of 5, 10, 20 and 40%. The grey fields in show the modelled compositional field of the bulk residue for CompositionsA, B and C (minerals and abundances indicated). (d^f) Composition of plagioclase crystallizing from a melt generated from 5% batch meltingof the starting composition with 75, 50, 25, 10 and 5% of this melt remaining. (g^i) Composition of plagioclase crystallizing from a melt gener-ated from 20% batch melting of the starting composition with 75, 50, 25, 10 and 5% of this melt remaining. The grey fields in (d)^(i) showthe measured compositional range of the felsic sheets. (See text for more details.)


2134

at University of M



ownloaded from


largely been a function of its pre-anatectic H2O content,which is unconstrained. Finally, it is likely that the felsicsheets drained melt from several metagabbroic sourcerocks all with different compositions, and that these meltchannelways were continually, or periodically, replenished.As such they may be rich in solid phases crystallizingearly from successive melt batches but over a protractedtime period, rather than from a single batch of melt.With respect to accessory phases, apatite will predomin-

antly enrich the residue and deplete the melt in LREE(e.g. Nehring et al., 2010), although the calculated norma-tive proportion of apatite in most metagabbro samples issmall (50·1 vol. %) and will have little effect. The mostabundant accessory phase in most of the metagabbros ismagnetite, which constitutes from zero to a few volumeper cent of samples. Published REE distribution coeffi-cients for magnetite are scarce and, although generallyincreasing towards HREE, vary by several orders of mag-nitude (from50·01 to41; Nielson et al., 1992). Rutile mayalso strongly influence REE patterns, although no rutilehas been found in any of the studied metagabbro samples.There is abundant evidence within the felsic sheets for en-trainment of solid phases from the melanosome (Fig. 2kand l). Most importantly, entrainment of garnet willresult in an increase in HREE concentrations of the felsicsheets relative to the model values (e.g. Stevens et al., 2007;Lavaure & Sawyer, 2011).

CONCLUSIONS(1) Metagabbroic rocks throughout the central region ofthe Lewisian complex partially melted via reactions con-suming amphibole and plagioclase during UHT meta-morphism. The Cpx-pegmatites within the metagabbromelanosome provide an in situ record of these meltingreactions.(2) Differences in the geochemistry of the metagabbroic

rocks, particularly garnet-absent versus garnetiferous vari-ants, might suggest derivation from more than one source(one enriched and one depleted in LREE). However, thedata suggest significant mobility of LILE and LREE be-tween the metagabbros and their felsic host-rocks.(3) The geochemistry of the tonalitic or trondhjemitic

felsic sheets is consistent with their derivation fromthe metagabbroic rocks, and the sheets have inheritedmany geochemical characters of their source rocks. Thefelsic sheets record segregation and migration of meltaway from the residual metagabbro source and subsequentcrystal fractionation, dominated by plagioclase, of theescaped melt.(4) The strong inferred mineralogical control on trace

element distributions suggests large diffusive length scalesconsistent with slow heating and cooling rates and thepresence of melt. Better quantifying the pressure^tempera-ture^time evolution of rocks within the central region

remains critical to understanding their tectonothermalevolution.(5) At UHT conditions, partial melting of the large

volume of TTG gneisses was inevitable and consistentwith their residual major element and LILE-depletedbulk composition.

ACKNOWLEDGEMENTSWe thank D. Jacob, M. Barth and N. Groschopf for helpwith analyses, and Axel Zirkler and Annemarie Militzerfor sample preparation. We are grateful to Kent Condiefor providing a spreadsheet of geochemical data onArchaean tholeiites. Many thanks to KathrynGoodenough and MartinWhitehouse for their helpful andperceptive reviews. This is a contribution to IGCP Project482599çThe Changing Early Earth.

FUNDINGThis work was supported by an internal grant to T.E.J.jointly funded by the University of Mainz and theGeocycles Earth System Science Research Centre.

SUPPLEMENTARY DATASupplementary data for this paper are available at Journalof Petrology online.

REFERENCESArndt, N. T., Albarede, F. & Nisbet, E. G. (1997). Mafic and ultra-

mafic magmatism. In: De Wit, M. J. & Ashwal, L. D. (eds)Greenstone Belts. Oxford: Clarenden Press, pp. 233^254.

Barker, F. (1979). Trondhjemite: Definition, environment and hypoth-eses of origin. In: Barker, F. (ed.) Trondhjemites, Dacites and Related

Rocks. Amsterdam: Elsevier, pp. 1^12.Barnicoat, A. C. (1983). Metamorphism of the Scourian Complex,

NW Scotland. Journal of Metamorphic Geology 1, 163^182.Barnicoat, A. C. (1987). The causes of the high-grade metamorphism

of the Scourie complex, NW Scotland. In: Park, R. G. &Tarney, J. (eds) Evolution of the Lewisian and Comparable High Grade

Terrains. Geological Society, London, Special Publications 27, 73^79.Be¤ dard, J. H. (2006). Trace element partitioning in plagioclase feld-

spar. Geochimica et Cosmochimica Acta 70, 3717^3742.Bowes, D. R., Wright, A. E. & Park, R. G. (1964). Layered intrusive

rocks in the Lewisian of the North-West Highlands of Scotland.QuarterlyJournal of the Geological Society of London 120, 153^192.

Brown, M. (2006). Duality of thermal regimes is the distinctive char-acteristic of plate tectonics since the Neoarchean. Geology 34,961^964.

Brown, M. & Rushmer, T. (2006). Evolution and Differentiation of the

Continental Crust. Cambridge: Cambridge University Press, 553 p.Brown, M., Korhonen, F. J. & Siddoway, C. S. (2011). Organizing melt

flow through the crust. Elements 7, 261^266.Cartwright, J. (1990). Prograde metamorphism, anatexis, and retro-

gression of the Scourian complex, north-west Scotland. In:Asworth, J. R. & Brown, M. (eds) High-temperature Metamorphism

and Crustal Anatexis. London: Unwin^Hyman, pp. 371^399.


2135

at University of M



ownloaded from



Cartwright, I. & Barnicoat, A. C. (1986). The generation ofquartz-normative melts and corundum-bearing restites by crustalanatexis: petrogenetic modelling based on an example from theLewisian of North-West Scotland. Journal of Metamorphic Geology 4,79^99.

Cartwright, I. & Barnicoat, A. C. (1987). Petrology of Scourian supra-crustal rocks and orthogneisses from Stoer, NW Scotland: implica-tions for the geological evolution of the Lewisian complex. In:Park, R. G. & Tarney, J. (eds) Evolution of the Lewisian and

Comparable High Grade Terrains. Geological Society, London, Special

Publications 27, 93^107.Cartwright, I. & Rollinson, H. R. (1995). Discussion on: Origin of

felsic sheets in the Scourian granulites: new evidence from rareearth elements, Published in Scottish Journal of Geology, Vol. 30,1994, pp. 121^129. ScottishJournal of Geology 31, 91^94.

Cartwright, I. & Valley, J.W. (1992). Oxygen-isotope geochemistry ofthe Scourian complex, northwest Scotland. Journal of the GeologicalSociety, London 149, 115^125.

Cohen, A. S., O’Nions, R. K. & O’Hara, M. J. (1991). Chronology andmechanism of depletion in Lewisian granulites. Contributions to

Mineralogy and Petrology 106, 142^153.Condie, K. C. (1976). Trace-element geochemistry of Archean green-

stone belts. Earth-Science Reviews 12, 393^417.Corfu, F., Heaman, L. M. & Rogers, G. (1994). Polymetamorphic evo-

lution of the Lewisian complex, NW Scotland, as recorded by U^Pb isotopic compositions of zircon, titanite and rutile. Contributionsto Mineralogy and Petrology 117, 215^228.

Davies, F. B. (1974). A layered basic complex in the Lewisian, south ofLoch Laxford, Sutherland. Journal of the Geological Society, London130, 279^284.

Foley, S. F. (2008). A trace element perspective on Archean crust for-mation and on the presence or absence of Archean subduction.Geological Society of America Special Papers 440, 31^50.

Foley, S., Tiepolo, M. & Vannucci, R. (2002). Growth of early contin-ental crust controlled by melting of amphibolite in subductionzones. Nature 417, 837^840.

Fowler, M. B. (1986). Large-ion lithophile element characteristics of anamphibolite facies to granulite facies transition at Gruinard Bay,North-west Scotland. Journal of Metamorphic Geology 4, 345^359.

Friend, C. R. L. & Kinny, P. D. (1995). New evidence for protolith agesof Lewisian granulites, northwest Scotland. Geology 23, 1027^1030.

Goodenough, K. M., Park, R. G., Krabbendam, M., Myers, J.,Wheeler, J., Loughlin, S. C., Crowley, Q. G., Friend, C. R. L.,Beach, A., Kinny, P. D. & Graham, R. H. (2010). The LaxfordShear Zone: an end-Archaean terrane boundary?. In: Law, R.,Butler, R. W. H., Holdsworth, R. E., Krabbendam, M. &Strachan, R. (eds) Continental Tectonics and Mountain Building: The

Legacy of Peach and Horne Geological Society, London, Special Publications

335, 103^120.Hamilton, P. J., Evensen, N. M., O’Nions, R. K. & Tarney, J. (1979).

Sm^Nd systematics of Lewisian gneisses: implications for theorigin of granulites. Nature 277, 25^28.

Hofmann, A.W. (1988). Chemical differentiation of the Earth: the re-lationship between mantle, continental crust, and oceanic crust.Earth and Planetary Science Letters 90, 297^314.

Johnson,T. E. & White, R.W. (2011). Phase equilibrium constraints onconditions of granulite-facies metamorphism at Scourie, NWScotland. Journal of the Geological Society, London 168, 147^158.

Johnson, T. E., White, R. W. & Powell, R. (2008). Partial melting ofmetagreywacke: a calculated mineral equilibria study. Journal ofMetamorphic Geology 26, 837^853.

Kerrich, R., Polat, A.,Wyman, D. & Hollings, P. (1999).Trace elementsystematics of Mg-, to Fe-tholeiitic basalt suites of the Superior

Province: implications for Archean mantle reservoirs and green-stone belt genesis. Lithos 46, 163^187.

Kinny, P. D., Friend, C. R. L. & Love, G. J. (2005). Proposalfor a terrane-based nomenclature for the Lewisian GneissComplex of NW Scotland. Journal of the Geological Society, London162, 175^186.

Lavaure, S. & Sawyer, E. W. (2011). Source of biotite in the WulumaPluton: Replacement of ferromagnesian phases and disaggregationof enclaves and schlieren. Lithos 125, 757^780.

Love, G., Kinny, P. & Friend, C. (2004). Timing of magmatism andmetamorphism in the Gruinard Bay area of the Lewisian GneissComplex: comparisons with the Assynt Terrane and implicationsfor terrane accretion. Contributions to Mineralogy and Petrology 146,620^636.

McDonough,W. F. & Sun, S.-s. (1995). The composition of the Earth.Chemical Geology 120, 223^253.

Moorbath, S., Welke, H. & Gale, N. (1969). The significance of leadisotope studies in ancient, high-grade metamorphic basement com-plexes, as exemplified by the Lewisian rocks of NorthwestScotland. Earth and Planetary Science Letters 6, 245^256.

Nehring, F., Foley, S. & Ho« llta« , P. (2010).Trace element partitioning inthe granulite facies. Contributions to Mineralogy and Petrology 159,493^519.

Nehring, F., Jacob, D. E., Barth, M. G. & Foley, S. F. (2008). Laser-ablation ICP^MS analysis of siliceous rock glasses fused on an iri-dium strip heater using MgO dilution. Microchimica Acta 160,153^163.

Nielson, R. L., Gallahan, W. E. & Newberger, F. (1992).Experimentally determined mineral^melt partition coefficientsfor Sc, Y and REE for olivine, orthopyroxene, pigeonite, mag-netite and ilmenite. Contributions to Mineralogy and Petrology 110,488^499.

O’Hara, M. J. (1961). Zoned ultrabasic and basic gneiss masses in theearly Lewisian metamorphic complex at Scourie, Sutherland.Journal of Petrology 2, 248^276.

O’Hara, M. J. & Yarwood, G. (1978). High pressure^temperaturepoint on an Archaean geotherm, implied magma genesis by crustalanatexis, and consequences for garnet^pyroxene thermometry andbarometry. Philosophical Transactions of the Royal Society of London,

Series A 288, 441^456.Ordo¤ n‹ ez-Caldero¤ na, J. C., Polata, A., Fryera, B. J. & Gagnona, J. E.

(2011). Field and geochemical characteristics of Mesoarchean toNeoarchean volcanic rocks in the Stor� greenstone belt, SWGreenland: Evidence for accretion of intra-oceanic volcanic arcs.Precambrian Research 184, 24^42.

Park, R. G. (1970). Observations on Lewisian chronology. Scottish

Journal of Geology 6, 379^399.Park, R. G. (2005). The Lewisian terrane model: a review. Scottish

Journal of Geology 41, 105^118.Park, R. G. & Tarney, J. (1987). The Lewisian complex: a typical

Precambrian high-grade terrain?. In: Park, R. G. & Tarney, J.(eds) Evolution of the Lewisian and Comparable High Grade Terrains

Geological Society, London, Special Publications 27, 13^25.Patin‹ o Douce, A. E. & Harris, N. B. W. (1998). Experimental con-

straints on Himalayan anatexis. Journal of Petrology 39, 689^710.Peach, B. N., Horne, J., Gunn, W., Clough, C. T., Hinxman, L. &

Cadell, H. M. (1888). Report on the Recent Work of theGeological Survey in the North-west Highlands of Scotland,based on the Field-notes and Maps (Read April 25, 1888). QuarterlyJournal of the Geological Society of London 44, 378^441.

Peach, B. N., Horne, J., Gunn,W., Clough, C. T., Hinxman, L.W. &Teall, J. J. H. (1907). The Geological Structure of the North West

Highlands of Scotland. Memoirs of the Geological Survey of Great Britain.


2136

at University of M



ownloaded from


Polat, A. (2009). The geochemistry of Neoarchean (ca. 2700 Ma) tho-leiitic basalts, transitional to alkaline basalts, and gabbros, WawaSubprovince, Canada: Implications for petrogenetic and geody-namic processes. Precambrian Research 168, 83^105.

Pride, C. & Muecke, G. K. (1980). Rare earth element geochemistryof the Scourian complex N.W. Scotlandçevidence for the gran-ite^granulite link. Contributions to Mineralogy and Petrology 73,403^412.

Pride, C. & Muecke, G. K. (1982). Geochemistry and origin of graniticrocks, Scourian Complex, NW Scotland. Contributions to Mineralogy

and Petrology 80, 379^385.Rollinson, H. (1996). Tonalite^trondhjemite^granodiorite magmatism

and the genesis of Lewisian crust during the Archaean. In:Brewer, T. S. (ed.) Precambrian Crustal Evolution in the North Atlantic

Region. Geological Society, London, Special Publications 112, 25^42.Rollinson, H. (1994). Origin of felsic sheets in the Scourian granulites:

new evidence from rare earth elements. Scottish Journal of Geology30, 121^129.

Rollinson, H. R. & Fowler, M. B. (1987). The magmatic evolution ofthe Scourian complex at Gruinard Bay. In: Park, R. G. &Tarney, J. (eds) Evolution of the Lewisian and Comparable High Grade

Terrains. Geological Society, London, Special Publications 27, 57^71.Rollinson, H. & Gravestock, P. (2012).The trace element geochemistry

of clinopyroxenes from pyroxenites in the Lewisian of NWScotland: insights into light rare earth element mobility duringgranulite facies metamorphism. Contributions to Mineralogy and

Petrology 163, 319^335.Rollinson, H. R. & Tarney, J. (2005). Adakitesçthe key to under-

standing LILE depletion in granulites. Lithos 79, 61^81.Rollinson, H. R. & Windley, B. F. (1980a). Selective elemental

depletion during metamorphism of Archaean granulites,Scourie, NW Scotland. Contributions to Mineralogy and Petrology 72,257^263.

Rollinson, H. R. & Windley, B. F. (1980b). An Archaeangranulite-grade tonalite^trondhjemite^granite suite from Scourie,NW Scotland: Geochemistry and origin. Contributions to Mineralogy

and Petrology 72, 265^281.Rosenberg, C. L. & Handy, M. R. (2005). Experimental deformation

of partially melted granite revisited: implications for the continen-tal crust. Journal of Metamorphic Geology 23, 19^28.

Rudnick, R. L. & Gao, S. (2003). Composition of the continentalcrust. In: Holland, H. D. & Turekian, K. K. (eds) Treatise on

Geochemistry,Vol. 3. Amsterdam: Elsevier, pp. 1^63.Rushmer, T. (1991). Partial melting of two amphibolites: contrasting

experimental results under fluid-absent conditions. Contributions toMineralogy and Petrology 107, 41^59.

Savage, D. & Sills, J. D. (1980). High pressure metamorphism in theScourian of NW Scotland: Evidence from garnet granulites.Contributions to Mineralogy and Petrology 74, 153^163.

Sawyer, E. W. (2008). Atlas of Migmatites. Canadian Mineralogist, Special

Publication 9, , 371 p.Sawyer, E. W., Cesare, B. & Brown, M. (2011). When the continental

crust melts. Elements 7, 229^234.Sheraton, J.W., Tarney, J.,Wheatley, T. J. & Wright, A. E. (1973). The

structural history of the Assynt district. In: Park, R. G. &Tarney, J. (eds) The Early Precambrian of Scotland and Related Rocks of

Greenland. Keele: University of Keele, pp. 31^43.Sills, J. D. & Rollinson, H. R. (1987). Metamorphic evolution of the

mainland Lewisian complex. In: Park, R. G. & Tarney, J. (eds)Evolution of the Lewisian and Comparable High GradeTerrains. Geological

Society, London, Special Publications 27, 81^92.Sills, J. D., Savage, D.,Watson, J. V. & Windley, B. F. (1982). Layered

ultramafic^gabbro bodies in the Lewisian of northwest

Scotland: geochemistry and petrogenesis. Earth and Planetary

Science Letters 58(3), 345^360.Stevens, G., Villaros, A. & Moyen, J.-F. (2007). Selective peritectic

garnet entrainment as the origin of geochemical diversity inS-type granites. Geology 35, 9^12.

Sutton, J. & Watson, J. (1951). The pre-Torridonian metamorphic his-tory of the Loch Torridon and Scourie areas in the North-WestHighlands, and its bearing on the chronological classification ofthe Lewisian. Quarterly Journal of the Geological Society of London 106,241^307.

Tarney, J. & Weaver, B. L. (1987). Geochemistry of the Scouriancomplex: petrogenesis and tectonic models. In: Park, R. G. &Tarney, J. (eds) Evolution of the Lewisian and Comparable

High Grade Terrains. Geological Society, London, Special Publications 27,45^56.

Tarney, J. & Windley, B. F. (1977). Chemistry, thermal gradients andevolution of the lower continental crust. Journal of the Geological

Society, London 134, 153^172.Thompson, A. B. (1982). Dehydration melting of pelitic rocks and the

generation of H2O-undersaturated liquids. American Journal of

Science 282, 1567^1595.Watkins, J., Clemens, J. & Treloar, P. (2007). Archaean TTGs

as sources of younger granitic magmas: melting of sodic metatona-lites at 0·6^1·2 GPa. Contributions to Mineralogy and Petrology 154,91^110.

Weaver, B. L. & Tarney, J. (1980). Rare earth geochemistry ofLewisian granulite-facies gneisses, northwest Scotland:Implications for the petrogenesis of the Archaean lower continentalcrust. Earth and Planetary Science Letters 51, 279^296.

Weaver, B. L. & Tarney, J. (1981). Lewisian gneiss geochemistry andArchaean crustal development models. Earth and Planetary Science

Letters 55, 171^180.Wheeler, J., Park, R. G., Rollinson, H. R. & Beach, A. (2010). The

Lewisian Complex: insights into deep crustal evolution. In:Law, R., Butler, R. W. H., Holdsworth, R. E., Krabbendam, M.& Strachan, R. (eds) ContinentalTectonics and Mountain Building:The

Legacy of Peach and Horne. Geological Society, London, Special Publications

335, 51^79.White, R. W. & Powell, R. (2002). Melt loss and the preservation of

granulite facies mineral assemblages. Journal of Metamorphic Geology

20, 621^632.White, R.W., Powell, R. & Holland,T. J. B. (2001). Calculation of par-

tial melting equilibria in the system Na2O^CaO^K2O^FeO^MgO^Al2O3^SiO2^H2O (NCKFMASH). Journal of Metamorphic

Geology 19, 139^153.Whitehouse, M. J. & Kemp, A. I. S. (2010). On the difficulty of

assigning crustal residence, magmatic protolith and metamorphicages to Lewisian granulites: constraints from combined in situ

U^Pb and Lu^Hf isotopes. In: Law, R., Butler, R. W. H.,Holdsworth, R. E., Krabbendam, M. & Strachan, R. (eds)Continental Tectonics and Mountain Building: The Legacy of Peach and

Horne. Geological Society, London, Special Publications 335, 81^101.Whitehouse, M. J. & Moorbath, S. (1986). Pb^Pb systematics of

Lewisian gneissesçimplications for crustal differentiation. Nature319, 488^489.

Whitehouse, M. J., Fowler, M. B. & Friend, C. R. L. (1996).Conflicting mineral and whole-rock isochron ages from the LateArchaean Lewisian complex of northwestern Scotland:Implications for geochronology in polymetamorphic high-gradeterrains. Geochimica et Cosmochimica Acta 60, 3085^3102.

Whitehouse, M. J., Claesson, S., Sunde, T. & Vestin, J. (1997). Ionmicroprobe U^Pb zircon geochronology and correlation ofArchaean gneisses from the Lewisian Complex of Gruinard Bay,


2137

at University of M



ownloaded from


northwestern Scotland. Geochimica et Cosmochimica Acta 61,4429^4438.

Winther, K. T. (1996). An experimentally based model for the originof tonalitic and trondhjemitic melts. Chemical Geology 127, 43^59.

Wolf, M. B. & Wyllie, P. J. (1994). Dehydration-melting of amphiboliteat 10 kbar: the effects of temperature and time. Contributions to

Mineralogy and Petrology 115, 369^383.Wyllie, P. J. & Wolf, M. B. (1993). Amphibolite dehydration-melting:

sorting out the solidus. In: Prichard, H. M., Alabaster, T.,

Harris, N. B. W. & Neary, C. R. (eds) Magmatic Processes and

Plate Tectonics. Geological Society, London, Special Publications 76,405^416.

Xiong, X.-L. (2006). Trace element evidence for growth of early con-tinental crust by melting of rutile-bearing hydrous eclogite. Geology34, 945^948.

Zhu, Z. K., O’Nions, R. K., Belshaw, N. S. & Gibb, A. J. (1997).Lewisian crustal history from in situ SIMS mineral chronometryand related metamorphic textures. Chemical Geology 136, 205^218.


2138

at University of M



ownloaded from


archaean intracrustal differentiation from partial melting of ......interactions during burial,...

Documents