| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 6 | Pages 995-1029 | 2003
© Oxford University Press 2003
Processes and Conditions During Contact Anatexis, Melt Escape and Restite Formation: the Huntly Gabbro Complex, NE Scotland
1 DEPARTMENT OF EARTH SCIENCES, UNIVERSITY OF MANCHESTER, MANCHESTER M13 9PL, UK
2 SCHOOL OF EARTH SCIENCES AND GEOGRAPHY, CEESR, KINGSTON UNIVERSITY, PENRHYN ROAD, KINGSTON-UPON-THAMES KT1 2EE, UK
E-mail: Giles.Droop{at}man.ac.uk
RECEIVED DECEMBER 15, 2001; ACCEPTED DECEMBER 2, 2002
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONThe Huntly Gabbro is one of a suite of large, Ordovician, syn-orogenic, mid-crustal, layered, mafic intrusions, emplaced into Proterozoic metaclastic rocks of NE Scotland soon after the thermal peak of static, high-T, low-P regional metamorphism. This gabbro and its associated contact metamorphic rocks illustrate a variety of processes operating during contact anatexis and subsequent melt segregation and extraction. These processes may closely mirror those occurring at much larger scales in the deep crust during high-grade regional metamorphism and the generation of granitic magmas. The emplacement of the Huntly mafic magma resulted in high-grade contact metamorphism and, locally, anatexis of metapelites, leading to the formation of migmatites. The migmatites and country-rock schists were studied to establish the physical conditions of metamorphism and anatexis, the nature of the melting reactions, the compositions of the melts produced, and the extent to which melting was a closed- or open-system process. The country-rock schists immediately to the south of the Huntly Complex contain mineral assemblages characteristic of the regional andalusite zone. Thermobarometry of an andalusite schist yields regional metamorphic conditions of 537 ± 42°C and 0·27 ± 0·12 GPa, consistent with previously published P–T estimates. The contact metamorphic rocks include sillimanite hornfelses, metatexites and diatexites. The metatexites consist of cordierite–K-feldspar hornfels melanosomes and K-feldspar-rich garnetiferous leucosomes. The diatexites consist of schollen of fine-grained granoblastic hornfels and metatexite suspended in igneous-textured matrix rocks composed of abundant sub/euhedral garnet, cordierite, plagioclase and, locally, orthopyroxene, with minor interstitial biotite, K-feldspar and quartz. The hornfels melanosomes and schollen retained their structural integrity during partial melting, but the matrix rocks did not. In the highest-grade diatexites, the assemblage Grt + Opx + Crd + Hc + Pl characterizes both the hornfels schollen and the sub/euhedral minerals of the matrix rocks. Application of phase equilibria to Opx-bearing rocks yields estimated peak-metamorphic conditions of 900 ± 50°C, 0·45 ± 0·1 GPa and aH2O < 0·3. The pressure estimate implies an emplacement depth of
16 ± 3 km. The prograde P–T path of contact-metamorphic rocks had a low, positive d P/d T slope, indicating that the gabbro intrusion increased the lithostatic load on the country rocks by overplating. Pseudomorph textures involving Al-silicates provide strong evidence that the diatexites evolved from andalusite schists via a sillimanite hornfels stage. Mineralogical changes reflect a sequence of dehydration reactions, followed by fluid-absent partial-melting reactions involving biotite breakdown. It was the fluid-absent reactions that generated the sub/euhedral minerals in the diatexites, as peritectic phases. In many of the highest-grade diatexites, quartz-free anhydrous solid assemblages were produced via reactions such as Bt + Sil + Grt = Crd + Hc + Kfs + melt and Bt + Grt + Crd = Opx + Hc + Kfs + melt. Whole-rock major-element geochemical studies indicate that the Opx–Crd hornfelses and diatexite matrices are depleted in Si and K relative to their schist protoliths. Mass-balance calculations indicate that (1) the Opx–Crd hornfels xenoliths represent solid Ca-, Mg-, Fe-, Al-, Na-rich residues left after extraction of 60% melt; (2) the Opx-bearing diatexite matrix rocks are also residual, and represent restite-enriched crystal–liquid mushes left after extraction of 55% melt; (3) the Opx-free diatexite matrix rocks probably represent restite-enriched mushes that retained a higher proportion of residual melt; (4) the anatectic melts were of H2O-undersaturated, peraluminous, low-Ca, potassic granite composition. Melt compositions and proportions were confirmed experimentally by partially melting local metapelite samples at 900 °C and 0·5 GPa, under fluid-absent conditions. The similarity between the compositions of calculated and experimental melts implies that the melts underwent little or no fractional crystallization before their expulsion. At many localities, a large proportion of the melt escaped from the sites of its generation. Segregation of melt from restite was probably achieved through gravity-driven processes. The fugitive granitic melts did not mix with the gabbroic magma to any great extent, although contamination of mafic magma occurred locally, leading to the generation of biotite-bearing gabbros. The fugitive melts probably contributed to the contemporaneous ‘Grampian’ suite of S-type granites. The mid-Ordovician middle crust of NE Scotland was thus a site of crustal differentiation. The results demonstrate that crustal fusion and magma production can occur without significant chemical interaction between the mantle-derived heat source and the crustal melts, and that melt extraction can occur in the absence of regional tectonic deformation. KEY WORDS: migmatite; diatexite; fluid-absent melting; melt segregation; peraluminous granite
|
INTRODUCTION |
|---|
The differentiation of the continental crust is widely held to involve the partial melting of mid- to deep-crustal rocks, at high metamorphic grades, with the upward escape of granitic melts, leaving behind high-grade restitic residues (Clemens, 1990; Thompson, 1990; Vielzeuf et al., 1990). Intracrustal heat sources are generally inadequate to power these processes, and major contributions of mantle heat seem to be required (England & Thompson, 1986). Such mantle-derived heat is most easily supplied through intrusion of gabbroic magmas into the crust. Exposed examples of contact anatexis surrounding gabbroic intrusions are therefore of considerable interest. This paper describes a particularly instructive example of this, and explores the nature of the petrogenetic processes involved.
Partial melting phenomena are commonly recognized in the inner parts of the thermal aureoles of large mid- and upper-crustal intrusions, where the wall rocks are of suitable composition (e.g. Ashworth & Chinner, 1978; Pattison & Harte, 1988; Grant & Frost, 1990; Finger & Clemens, 1995; Holness & Clemens, 1999). In the context of crustal differentiation, a key question is the extent to which the anatectic melts migrate away from, or remain within, their sites of generation in such aureoles. In principle, several potential fates can be envisaged for contact-anatectic melts; they can (1) crystallize in situ (e.g. Pattison & Harte, 1988); (2) intrude the chilled margin of the pluton, forming back-veining complexes; (3) invade the magma of the pluton, either to mix with it (if miscible) to produce a hybrid magma (e.g. Greenfield et al., 1996) or mingle with it (if immiscible or inefficiently comminuted) to form enclaves; (4) migrate elsewhere in the inner aureole, either by flowing along fractures or by percolating along grain-boundaries; (5) leave the aureole to intrude cooler country rock at higher crustal levels. In situ crystallization is most likely to occur where the heat-supplying pluton is at relatively low temperature (i.e. a granitoid magma) and generally produces relatively small volumes of anatectic melt by fluid-saturated reactions (e.g. Holness & Clemens, 1999). Such reactions tend to have negative dP/dT slopes, causing near-solidus fluid-saturated melts to solidify immediately if they begin to ascend (Brown & Fyfe, 1970). Conversely, melts are likely to be able to leave the aureole to intrude cooler country rocks only if the heat-supplying pluton is relatively large, hot (i.e. mafic) and produces relatively large volumes of low-viscosity, fluid-undersaturated anatectic melt, by fluid-absent reactions. Such reactions have positive dP/dT slopes, allowing the melts to ascend for appreciable distances without immediately freezing (Brown & Fyfe, 1970; Clemens et al., 1997; Clemens & Droop, 1998).
If it occurs on a large enough scale, the segregation and ascent of anatectic melt from its source region must result in crustal differentiation, with enrichment of the source areas in refractory (restitic) components, and of higher crustal levels in the complementary ‘granitophile’ components. It has been suggested that the intraplating and underplating of the crust by large volumes of mafic magma and the ensuing partial melting of fertile crustal rocks is the major process by which mobile granitoid magmas are produced and the continental crust grows in volume (e.g. Clemens, 1990; Rudnick, 1990; Vielzeuf et al., 1990; Stevens & Clemens, 1993). In a deep-crustal context, petrological and geochemical evidence for granitic melt extraction and restite formation is well documented (e.g. the granulite-facies ‘stronalites’ of the Ivrea Zone, northern Italy; Schnetger, 1994). However, so far, there has been little to suggest that these processes could also operate at mid-crustal levels. In this paper, we examine the thermal effects of a large gabbro intrusion and present evidence that it caused substantial restite formation, at mid-crustal levels. We demonstrate that the country-rock metapelites underwent very high degrees of partial melting and that, locally, a large proportion of the melts segregated from the solid residua, and indeed escaped the source region.
|
GEOLOGICAL SETTING |
|---|
The Huntly Gabbro in NE Scotland (Fig. 1) is one of a suite of large, syn-orogenic, layered, mafic intrusions of Middle Ordovician age, locally known as the ‘Newer Gabbros’ (Read, 1919, 1923, 1961). These were emplaced into Dalradian (Late Proterozoic) metaclastic rocks during, or soon after, the thermal peak of the Grampian regional metamorphism (Fettes, 1970).
|
The classic Buchan-type regional metamorphism in NE Scotland is characterized by a sequence of biotite, cordierite, andalusite and sillimanite zones in the metapelites (Read, 1923a, 1952). The metamorphism occurred statically under conditions of high temperature and low pressure (Chinner, 1966; Harte & Hudson, 1979; Hudson, 1980, 1985). The regional isograds describe a broad horseshoe pattern around a central area of low-grade rocks (Read, 1952), probably as a result of post-metamorphic folding (Treagus & Roberts, 1981).
The Newer Gabbro intrusions themselves display rhythmic and cryptic layering, and consist mainly of cumulate-textured peridotites (now serpentinized), troctolites, norites, gabbros and, locally, more differentiated monzogabbros (Wadsworth et al., 1966; Clarke & Wadsworth, 1970; Ashcroft & Boyd, 1976; Wadsworth, 1982, 1986, 1988; Munro, 1984). The gabbros post-date at least one phase of regional ductile deformation in the Dalradian country rocks, but were themselves affected by post-magmatic shearing, particularly along their contacts (Ashcroft & Munro, 1978; Boyd & Munro, 1978; Ashcroft et al., 1984). Stewart & Johnson (1960) suggested that the various gabbro bodies represent the dismembered remains of a single lopolith. However, as the details of the modal layering differ from one body to another (Munro, 1984), it is more likely that the intrusions evolved independently by in situ crystal fractionation of separate magma batches (Wadsworth, 1986). Pankhurst (1970) obtained a whole-rock Rb–Sr isochron date of 486 ± 17 Ma for the Newer Gabbros, but more precise U–Pb mineral ages of 468 ± 8 Ma and 470 ± 9 Ma have been obtained by Rogers et al. (1994) and Dempster et al. (2002), respectively.
Many of the Newer Gabbros possess well-developed thermal aureoles (e.g. Stewart, 1946; Gribble, 1966; Allan, 1970; Fettes, 1970; Ashworth, 1975), some of which have locally modified the regional isograd pattern (e.g. Fettes, 1970). Thus, rather than having been entirely responsible for it, the exposed gabbro bodies appear to have locally augmented the regional metamorphism (Pankhurst, 1970). This seems to be a common feature of plutonometamorphism (Powers & Bohlen, 1985; Wickham, 1987; Finger & Clemens, 1995). Pressure estimates of 0·4–0·5 GPa from the thermal aureoles (Ashworth & Chinner, 1978; Droop & Charnley, 1985) indicate that the Newer Gabbros were emplaced at a depth of 15–18 km. The pressure increase recorded by the overprinting of andalusite by kyanite in schists structurally beneath the gabbros (Chinner & Heseltine, 1979; Chinner, 1980) has been attributed to the extra load imposed by the intrusions (Dempster et al., 1995) or to post-intrusive thrusting (Baker, 1987).
Outcrops of xenolithic rocks (diatexitic schollen migmatites), composed of a cordierite-bearing igneous matrix and xenoliths of silica-poor aluminous hornfels, are closely associated with several of the Newer Gabbros, both in their inner aureoles and within the gabbros themselves. Watt (1914) and Read (1923a, 1923b, 1935), who first described these xenolithic complexes, believed that the igneous matrix rocks, which they termed ‘cordierite norites’, had been produced by contamination of mafic magma by assimilation of pelitic wall rocks. Chinner & Schairer (1962), Gribble & O'Hara (1967) and Gribble (1968, 1970) later demonstrated that formation of cordierite norites by bulk assimilation of metapelite by mafic magma could not occur by a process of down-temperature evolution, owing to the existence of the An–Opx–Tri thermal divide in the CMAS system. Instead, they argued that the ‘cordierite norites’ were derived by partial melting of Dalradian metapelites, without significant mixing with the Newer Gabbro magmas, a view supported by their relatively high 87Sr/86Sr initial ratios (0·720–0·730; Pankhurst, 1969).
Granitoid plutons are volumetrically abundant in NE Scotland (Fig. 1). Of these, the ‘Grampian Granites’ (Brown, 1991) are late-tectonic peraluminous granites of Ordovician age (Pankhurst, 1974), which have yielded U–Pb mineral ages of 475 ± 5 Ma (Pidgeon & Aftalion, 1978), 470 ± 1 Ma (Kneller & Aftalion, 1987) and 467 ± 6 Ma (Oliver et al., 2000). The coincidence between these ages and those of the Newer Gabbros indicates that there was a major thermal event in the Buchan area at that time, involving high-T, low-P regional metamorphism, gabbro intrusion, crustal melting and the generation of granite magmas (Rogers et al., 1994).
|
THE HUNTLY COMPLEX |
|---|
The Huntly Complex is a poorly exposed, pear-shaped outcrop of gabbros and associated rocks, measuring
10 km x 9 km, intruding broadly north–south-trending Dalradian metasediments of the Argyll and Southern Highland Groups (Fig. 2). The southern contact cuts andalusite-porphyroblastic metapelites of the Boyndie Bay Group and turbiditic metapelites and psammites of the Whitehills Group. The western contact abuts biotite-rich garnet-, staurolite- and sillimanite-bearing schists and gneisses of the Portsoy Group. The mafic igneous rocks crop out in two major areas, one to the west, comprising mainly layered cumulate troctolites, and one in the east–central part, comprising mainly gabbronorites and norites (Munro, 1970; Weedon, 1970). The cumulate layering dips steeply and youngs to the east (Shackleton, 1948), indicating that the western margin of the complex was the original floor of the intrusion. The thermal aureole surrounding the Huntly Complex has not been mapped. This is due partly to the general paucity of exposure and partly (at least around the western margin) to excision of the aureole by faulting (Munro, 1970) along a NNE-trending zone of intense strain (Baker, 1987) known as the ‘Portsoy–Duchray Lineament’ (Harte, 1988). However, closely associated with the mafic igneous rocks are numerous outcrops of migmatite (‘xenolithic complexes’) (Fig. 2), which undoubtedly represent high-grade contact metamorphic rocks. Many of these appear to be arranged around the margins of the complex, particularly on the eastern side, but isolated outcrops of migmatite also occur within the gabbros (Fig. 2) (Munro, 1970). Judging by the trends in strike of the country rock units, it is likely that the migmatites in the southern part of the complex (Huntly–Dunbennan Hill) were derived from schists of the Boyndie Bay and Whitehills Groups. Two kilometre-scale granites also occupy marginal positions in the complex (Fig. 2). The more northerly of these granites (Avochie) is foliated (Munro, 1970) and may belong to the Grampian suite.
|
|
PETROGRAPHY OF METAPELITES |
|---|
In this section we describe the field relations, mineralogy and textures of pelitic country rocks and migmatites from the southern and western parts of the Huntly Complex. Mineral assemblages of representative samples are listed in Table 1.
|
Regional metamorphic rocks
The regional metamorphic rocks exposed around the south and west margins of the Huntly Complex span a wide range in mineralogy, texture and grade. Biotite-slates of the lowermost Macduff Slate crop out at St. Michael's Well (British National Grid reference NJ 519359) and Coynachie (NJ 491344), south of the area covered by Fig. 2. These rocks are fine-grained cleaved metasiltstones with millimetre- to centimetre-scale compositional layering (bedding) defined by alternating quartz-rich and quartz-poor layers, and an oblique penetrative cleavage defined by aligned biotite, muscovite and chlorite flakes. The rocks belong to the Buchan biotite zone.
Coarser-grained pelitic mica schists and subordinate semipelites (metagreywackes) and quartzites of the Boyndie Bay Group occur further west, in the vicinity of Clashmach Hill (NJ 498385). The schists contain the assemblage Ms + Bt + Qtz + Pl ± And ± St ± Grt + Ilm. Aligned muscovite flakes measuring 500 µm x 50 µm define the schistosity, whereas more equant, randomly orientated biotite flakes and granoblastic quartz and plagioclase make up the rest of the matrix. Andalusite occurs as sub-rectangular poikiloblasts, up to 6 mm long, containing quartz inclusions. Staurolite forms subhedral micropoikiloblasts up to 0·3 mm long. Rare, tiny (150 µm diameter) subhedral garnets occur in some samples. These rocks belong to the Buchan upper andalusite zone [the staurolite zone of Hudson (1980)].
West of the Huntly Complex, rocks of the Portsoy Group were sampled at Cairnie (NJ 482446). Many samples (e.g. Sin1, Sin101) are medium-grained plagioclase- and quartz-rich garnet–biotite gneisses containing minor staurolite or fibrolitic sillimanite. The rocks possess a biotite ± sillimanite schistosity, parallel to a layering defined by variations in modal biotite proportion. Somewhat finer-grained garnet–mica schists (e.g. Sin2) lack modal layering and contain essential muscovite, which, together with biotite, defines the schistosity. The garnets form anhedral inclusion-free crystals up to 0·4 mm across. These schists and gneisses, which belong to the lower sillimanite zone, appear to belong to a facies series of Barrovian (medium P/T) type.
Non-migmatitic hornfels
A loose block (>1 m diameter) of non-migmatitic cordierite–sillimanite hornfels (Dun1), representing a contact-metamorphosed metapelite that did not undergo partial melting, was found on the eastern side of the River Deveron (NJ 499422) near Dunbennan Hill in the SW part of the complex. The rock contains scattered subhedral garnet porphyroblasts up to 3 mm across and abundant large (8 mm x 2 mm x 2 mm) randomly orientated bundles of sub-parallel prismatic sillimanite crystals, set in a fine-grained granoblastic matrix of biotite, cordierite, plagioclase, quartz and ilmenite (Fig. 3a). Cordierite is locally replaced by retrograde chlorite + andalusite. The sillimanite aggregates are elongated and roughly square in cross-section, suggesting that they are pseudomorphs after andalusite. Similar sillimanite pseudomorphs after andalusite have been described by Vernon (1987).
|
Migmatites
The majority of contact-metamorphic rocks in the Huntly Complex are migmatites, and those described below correspond mainly to the ‘cordierite-bearing migmatites’ of Ashworth (1976). The following descriptions, although overlapping in part with those of Ashworth (1976), concentrate on new observations. Migmatite terminology used in this paper is from Ashworth (1985).
Metatexites
Large loose blocks of stromatic metatexite are abundant in the vicinity of Cumrie and Cormalet Hill in the north central part of the Complex (Fig. 2). These rocks typically consist of blue–grey cordierite hornfels melanosomes and veins of garnetiferous leucosome, the latter typically occupying 20–40% of the rock. The hornfels melanosomes are fine grained and locally contain garnet porphyroblasts. The leucosomes are mostly <1 cm thick and tend to form sub-parallel arrays, following the primary lithological layering (bedding), even where the latter is tightly folded. In most metatexites, however, branching, anastomosing and crosscutting veins also occur, indicating the former presence of a melt phase. Leucosome veins have locally coalesced to form irregular granitic patches (>5 cm across), studded with large garnets.
Most melanosomes are Opx-free Crd–Kfs hornfelses (e.g. Cor2, Cor6 and Cor11a; Table 1). Cordierite, plagioclase, K-feldspar and ilmenite all occur as equant, anhedral grains and form a typical granoblastic polygonal matrix (Fig. 3b). Randomly orientated, thin, red–brown biotite flakes are partially replaced by fibrolitic sillimanite in some rocks. Quartz, where present, is usually subordinate, interstitial and texturally associated with garnet. In some quartz-free hornfelses, small granules of brown hercynite occur in the matrix. The K-feldspar is commonly microperthitic and locally replaced by myrmekitic quartz–plagioclase intergrowths. Garnet, where present, typically forms euhedral porphyroblasts, up to 5 mm in diameter, containing rounded inclusions of ilmenite, biotite and, in some rocks, hercynite, near its rims. In some hornfelses, however, the garnets have atoll-like or spongy textures and are intergrown with relatively coarse-grained quartz and K-feldspar. Apart from the fact that they appear to be isolated and spheroidal, rather than planar, these quartz–K-feldspar–garnet aggregates closely resemble the leucosomes described by Powell & Downes (1990) from pelitic granulites at Broken Hill and, like them, probably represent localized sites of melting.
The leucosomes are dominated by quartz, K-feldspar and garnet with subordinate biotite, plagioclase and cordierite (e.g. Cor11b: Table 1). The garnet is typically inclusion free, and forms either irregular grains (Fig. 3b) or large (up to 1 cm diameter) euhedral crystals. The K-feldspar forms blocky, subhedral crystals of coarse-grained orthoclase microperthite, to which quartz is locally interstitial.
Garnet is a common solid product of incongruent, biotite-consuming, fluid-absent melting reactions in aluminous rocks (Waters & Whales, 1984; Grant, 1985a, 1985b; Conrad et al., 1988; LeBreton & Thompson, 1988; Vielzeuf & Holloway, 1988; Waters, 1988; Powell & Downes, 1990; Vielzeuf & Montel, 1994; Carrington & Harley, 1995; Patiño Douce & Beard, 1996; Stevens et al., 1997; Pickering & Johnston, 1998). The ubiquity and abundance of garnets in the leucosomes provides strong evidence that this type of reaction was responsible, at least in part, for the migmatization of the Huntly rocks.
Diatexite matrices (Grt tonalites and Crd norites)
Schollen diatexites are common throughout the Huntly Complex (Fig. 2). On a mesoscopic scale, these rocks typically consist of 70–95% of medium- to coarse-grained, igneous textured mobilisate (‘matrix’) with subordinate irregular hornfels or metatexite schollen up to 1 m long. Most of the schollen are pelitic, but other refractory rock types, such as quartzite and calc-silicate, occur locally.
The diatexite matrix rocks have pelitic or semi-pelitic mineralogies but igneous textures. Opx-free varieties [‘granitoids’ of Ashworth (1976)] are well represented in the Cormalet–Cumrie area. The Grt tonalite sample Cor7 (Table 1; Fig. 3c), for example, consists of abundant large (1 cm diameter) euhedral, inclusion-free garnets and sporadic, blocky, euhedral, inclusion-free cordierites in a medium- to coarse-grained matrix of biotite, quartz and euhedral plagioclase. Much of the quartz is clearly interstitial to garnet, cordierite and plagioclase (Fig. 3c). Small patches of fibrolite occur locally, near the cordierite and garnet.
Opx-bearing varieties [‘cordierite norites’ of Read (1923, 1935); ‘noritoids’ of Ashworth (1976)] crop out within the complex at Battlehill Quarry (NJ 539395), and in the River Deveron at Castle Bridge (NJ 533409) and north of Dunbennan Hill (NJ 499422) (Fig. 2). At Battlehill Quarry, such rocks form discordant bodies with sharp contacts against biotite–hornblende gabbro, micronorite and Opx–Crd hornfels. They are typically medium-grained rocks of mafic igneous appearance, rich in prominent, randomly orientated crystals of cordierite, orthopyroxene, garnet and plagioclase. The cordierite is fresh and forms subhedral prismatic crystals with well-developed simple or sector twinning. The orthopyroxene is markedly pleochroic and typically forms elongated subhedral prisms (Fig. 3d), locally enclosing rounded plagioclase inclusions. Garnet, commonly less abundant than Opx and Crd, forms subhedral grains that are either inclusion free or contain scattered ilmenite and biotite inclusions. Plagioclase crystals are subhedral and commonly show normal zoning, with sharply demarcated cores and rims, and locally also oscillatory zoning. In samples from Castle Bridge (e.g. CB1) biotite forms thick, subhedral books up to 1·5 mm across, but in others it appears to be mainly interstitial. In sample BHQ1 it forms large poikilitic patches, up to 8 mm across, enclosing plagioclase and cordierite. Quartz is generally present only in minor amounts and commonly forms small cuspate grains interstitial to garnet, cordierite, orthopyroxene and plagioclase (Fig. 3d). These quartz grains texturally resemble the ‘melt pools’ figured by Sawyer (2001). Samples from Castle Bridge contain modally abundant quartz, in part clearly interstitial to the ferromagnesian minerals. K-feldspar (orthoclase microperthite) is present in only a few cordierite norites and is generally poikilitic or interstitial to plagioclase and ferromagnesian minerals. Dark green hercynite is a constituent of many Crd norites. In samples from Battlehill Quarry hercynite is a minor phase and tends to occur only as small inclusions within cordierite, although in some samples (e.g. BQ38, BQ202) it is also enclosed by orthopyroxene, plagioclase and garnet. Hercynite is modally abundant in many Crd norites from Castle Bridge and Dunbennan Hill, where it forms oblong clusters, measuring up to 5 mm x 2 mm x 2 mm, of 0·01–0·1 mm granules intergrown with fine-grained cordierite. These hercynite clusters locally have square cross-sections and, in one sample (Dun4), are locally cored by corroded aggregates of prismatic sillimanite (Fig. 4a), indicating that they are pseudomorphs after that mineral. This texture also occurs in Crd-rich Grt tonalite sample Fow1. Hercynite has not been found in contact with quartz in any Huntly diatexite matrix rock. Retrograde textures include symplectic intergrowths of biotite and quartz replacing orthopyroxene and fine-grained biotite rimming garnet (Ashworth, 1976).
|
The igneous textures of the diatexite mobilisates clearly indicate that they contained a melt phase during crystallization (see also Kenah & Hollister, 1983; Vernon & Collins, 1988; Harte et al., 1991). These rocks resemble cumulates in that the abundant, large, subhedral to euhedral cordierites, orthopyroxenes, garnets and plagioclase cores clearly crystallized early, whereas the plagioclase rims and the modally minor, interstitial or poikilitic quartz, K-feldspar and biotite clearly crystallized later. The question arises as to whether the large crystals precipitated from the melt as phenocrysts or represent the solid products of incongruent melting reactions. We favour the latter interpretation because (1) the temperatures required to completely melt aluminous metapelites are higher, and presumably therefore less readily attained, than those necessary for partial melting, (2) cordierite, garnet and orthopyroxene are known to form during incongruent melting of biotite (e.g. Grant, 1985a, 1985b), and (3) the Opx–Crd hornfelses (see below) contain similar assemblages to some of the cordierite norites but have textures indicative of solid-state equilibration.
Crd–Kfs hornfels schollen
Crd–Kfs hornfels schollen (e.g. Cor8) within garnet granitoids at Cormalet closely resemble hornfels melanosomes in metatexites, both texturally and mineralogically.
Opx–Crd hornfels schollen and xenoliths
Opx–Crd hornfelses occur in the southern and western parts of the Huntly Complex as schollen within ‘cordierite norites’ and as screens in gabbroic rocks. At Battlehill Quarry, sharp contacts are exposed between biotite–hornblende gabbro and screens of Opx–Crd hornfels up to 6 m thick. The attitude of the lithological layering varies dramatically from one screen to the next, indicating that the screens are large xenoliths. A 3 m thick xenolith of Opx–Crd hornfels is in sharp contact with biotite gabbro in a small quarry on the NW side of Battlehill Woods (NJ 511398).
The hornfelses are fine-grained, layered, dark bluish grey cordierite-rich rocks, commonly bearing small (1–2 mm) garnet porphyroblasts. The matrices typically consist mainly of small equant grains of orthopyroxene, plagioclase and twinned cordierite, accompanied by abundant granules of ilmenite and, in many samples, dark green hercynite. The grains have random orientations and a well-developed granoblastic polygonal texture (Fig. 4b), indicating a high degree of textural equilibration. Plagioclase is only rarely zoned. Cordierite tends to have a slightly larger grain size than Opx and Pl and locally forms subhedral porphyroblasts. Plagioclase is absent from samples BQ41 and BQ101, but its place is taken by anorthoclase. Some samples (e.g. BQ17) contain sporadic 2 mm x 1 mm composite plagioclase–K-feldspar intergrowths, which may be exsolved ternary feldspars. Quartz is usually absent. Garnet occurs mainly as subhedral porphyroblasts (Fig. 4b), locally with inclusions of ilmenite and hercynite but, in some samples, also forms tiny euhedral crystals (Fig. 4b) and anhedral rims around ilmenite grains. Hercynite is in contact with cordierite, garnet, orthopyroxene and plagioclase but not quartz. Biotite is rare or absent.
Some hornfelses (e.g. HBHQ5, 9996, 10035, BW202) contain exceptionally high modal amounts of cordierite (>60%), with subordinate plagioclase, orthopyroxene, garnet, ilmenite and hercynite (Fig. 4c), and are referred to here as ‘cordieritites’.
Many Opx–Crd hornfelses possess compositional layering, manifest as variations in both modal mineralogy and mean grain size. Relatively coarse-grained layers tend to be more leucocratic than the finer-grained layers; they lack hercynite, and contain appreciable amounts of interstitial K-feldspar, quartz and biotite. Garnets in such layers are either irregular porphyroblasts or atoll-type grains filled with relatively coarse biotite, K-feldspar and quartz. These layers have textures reminiscent of the leucosomes in the metatexites, but are less clearly demarcated from melanosomes. Distinct leucosomes do, however, exist. A 6 mm thick vein of medium-grained Grt–Crd–Kfs rock cuts the cordieritite of sample 9996. This igneous-textured leucosome consists of abundant, euhedral to subhedral, twinned cordierites (with inclusions of ilmenite and hercynite) and rare garnets enclosed by coarse poikilitic K-feldspar (Fig. 4d). In many respects this rock resembles the diatexite matrix rocks described above.
Disregarding interstitial, leucosome and poikilitic minerals, the characteristic peak-metamorphic mineral assemblage of the majority of Opx–Crd hornfelses is Crd + Opx + Grt + Hc + Pl + Ilm. Fine-grained granoblastic domains clearly indicate crystallization in the solid state, but the local presence of relatively coarse igneous-textured leucosome-like patches suggests the former presence of melt locally.
|
MINERAL CHEMISTRY |
|---|
Major-element compositions of minerals in selected samples of regional and contact metamorphic rocks (Tables 2–4) were determined by energy-dispersive spectrometry (EDS) using a converted Geoscan electron microprobe operating at 15 kV, 3 nA specimen current on Co metal, and 40 s count time per analysis. X-ray spectra were processed using Link Systems ZAF4/FLS software. In addition, F and Ba contents in biotites in two samples were determined by wavelength-dispersive spectrometry using a Cameca Camebax electron microprobe with fluorite and celsian standards, respectively. Ferric iron contents in garnets and hercynites were calculated by specifying fixed cation-to-oxygen ratios of, respectively, 16:24 and 3:4 (Droop, 1987).
|
|
|
Regional metamorphic rocks
Minerals were analysed in two garnet–biotite gneisses (Sin1, Sin2) and one garnet–mica schist (Sin101) from Cairnie, and three mica schists (Bog1, Clas66 and Clas112) from Clashmach Hill.
Garnet
Garnet crystals in the Cairnie rocks are virtually unzoned. Those in Sin 1 and Sin101 have compositions close to Alm70Py23Sps05Grs01Adr01, whereas those of Sin2 cluster around Alm76Py13Sps08Grs03. The garnets in Clas66 are also unzoned, but differ from those at Cairnie in their high Mn and low Mg contents; their compositions are close to Alm57Py08Sps31Grs02Adr02.
Staurolite
Staurolite in Sin101 has an Mg-number value [= 100 Mg/(Fe2+ + Mg)] ranging from 21 to 26 over several grains and contains a small amount of Zn [0·10–0·19 atoms per 48(O)]. The staurolite in Clas66 has Mg-number ranging from 14 to 16 and contains up to 0·19 a.p.f.u. (atoms per formula unit) Zn.
Andalusite
Andalusite in Clas66 contains detectable Fe and Mg impurities (up to 0·01 a.p.f.u. of each).
Biotite
AlVI contents of all analysed regional-metamorphic biotites are closely grouped, mostly between 0·75 and 1·0 atoms per 22(O). Biotites from Cairnie have higher Mg-number values (50–62) and lower Ti contents (0·17–0·25 a.p.f.u.) than those from Clashmach Hill (0·29–0·40 and 0·26–0·40, respectively).
Muscovite
Regional muscovites are all mildly phengitic, with 6·06–6·14 Si and 0·15–0·56 Fe + Mg per 22(O), and contain some Na in their A sites [Na/(Ca + Na + K) = 0·06–0·16 in the Clashmach Hill schists and 0·19–0·26 in Sin2].
Plagioclase
Plagioclases in both Sin1 and Sin2 range from An17 to An23. K contents are below the detection limit. Rims and cores of individual grains have similar compositions. Plagioclases in Clas66 are also uniform, with compositions lying in the range An17–21.
Contact metamorphic rocks
Analyses of coexisting minerals were obtained from the Sil–Crd hornfels (Dun1), a Crd–Kfs hornfels (Cor8), six Opx–Crd hornfelses (Cor10, BQ17, BQ41, BHQ5, Pir1 and Pir4), and six cordierite norites (BHQ1, BQ38, Dun4, Dun12, CasB2i and CasB5). With a few exceptions, minerals in individual rocks are unzoned and show small within-sample variation, assumed to reflect a close approach to chemical equilibrium.
Garnet
Garnets in Sil–Crd and Crd–Kfs hornfelses display slight zonation, with broad cores of uniform composition (approximately Alm73Py17Sps05Grs04Adr01 in Dun1 and Alm79Py16Sps01Grs04 in Cor8) and rims up to 1 mm wide, displaying a smooth increase in Fe and decrease in Mg (Fig. 5a), indicating retrograde diffusional exchange.
|
Garnets in Opx–Crd hornfelses are remarkably unzoned (Fig. 5b) and show little within-sample variation. The Mg-number is typically between 24 and 29, with XCa between 0·02 and 0·05.
Garnets in cordierite norites are also typically unzoned (Fig. 5c) and similar in composition to those of Opx–Crd hornfelses. The Mg-number is generally between 26 and 30, Mn contents are low (XMn mostly between 0·004 and 0·02) and XCa ranges from 0·027 to 0·068.
Orthopyroxene
The Mg-number and 
contents of orthopyroxenes from all the Opx–Crd hornfelses are similar, ranging from 40 to 50 and from 0·07 to 0·18, respectively. Cation totals cluster around 4·00 per 6(O), suggesting that Fe3+ contents are very low.
Orthopyroxenes from cordierite norites are generally very similar to those of Opx–Crd hornfelses; Mg-number generally varies from 45 to 52, except in CasB2i, in which it varies from 50 to 58, and is typically between 0·05 and 0·18. 
is low, rarely exceeding 0·01.
Cordierite
The Mg-number values of the two analysed Opx-free hornfelses are similar, ranging from 60 to 68 in Cor8 and from 59 to 71 in Dun1. The cordierites in the Opx–Crd hornfelses are slightly more magnesian, Mg-number varying from 63 to 72 in BQ17, from 61 to 64 in BQ41, and from 61 to 66 in BHQ5 and Pir1. The Mg-number values of cordierites in the cordierite norites overlap substantially with these ranges (62–67 in BHQ1, 72–73 in CasB2i, 63–67 in BQ38 and 71–73 in CasB5). No significant zoning has been detected.
Thermogravimetric determination of total volatile content was carried out for cordierites in two samples: cordierite norite BQ38, and cordieritite 10035. Fifteen milligram aliquots of powdered sample were heated under Ar (to prevent oxidation) and held at 900°C for at least 30 min. The measured mass losses were 1·07% for BQ38 and 0·65% for 10035. H2O and CO2 are known to be able to occupy the channel site in cordierite (e.g. Schreyer & Yoder, 1964; Johannes & Schreyer, 1981; Vry et al., 1990) and are thus the volatile species most likely to have been present. The 1·07% weight loss for BQ38 is equivalent to a maximum XH2O value of 0·36 or a maximum XCO2 value of 0·15. For 10035, the data imply a maximum XH2O value of 0·22 or a maximum XCO2 value of 0·092.
Biotite
Biotites in the Sil–Crd and Crd–Kfs hornfelses have AlVI contents of between 0·67 and 1·46 a.p.f.u. [22(O)]. The Mg-number values range from 44 to 51 in Cor8 and from 54 to 55 in Dun1. Ti contents are relatively low (0·16–0·54 a.p.f.u.). Biotites in contact with garnet generally have higher Mg-number than those remote from garnet, complementing the Fe-rich garnet rims, and confirming that there has been retrograde Fe–Mg exchange.
Biotites in Opx–Crd hornfelses are less aluminous than those of Opx-free hornfelses, with AlVI contents of between 0·24 and 0·35. The Mg-number ranges from 48 to 57, and Ti contents are high (0·57–0·75 a.p.f.u.). Biotites in cordierite norites have similar Mg-number but slightly higher AlVI contents. Again, biotites closest to garnet are richest in Mg. Ba and F were determined in the biotites of two cordierite norites; XF in biotites in BHQ1 range from 0·017 to 0·061, and those in BQ38 from 0·014 to 0·052. Ba contents in both samples are <0·5 wt %.
Hercynite and sillimanite
All analysed spinels are rich in the FeAl2O4 component, with low Cr [0·01–0·05 per 4(O)] and Zn (0·03–0·07 a.p.f.u.), and barely detectable Ti. Calculated Fe3+ contents are <0·12 a.p.f.u. The Mg-number values of hercynites in Opx–Crd hornfelses range from 26 to 27 (BQ17), from 16 to 25 (Pir1) and from 18 to 32 (BHQ5); those of hercynites in cordierite norites are similar, ranging from 22 to 29. Sillimanites in Dun1 and Cor8 are virtually pure, with no significant Fe3+ or Mn3+ substitution.
Feldspars
Three compositional types of feldspar occur in the thermally metamorphosed rocks: plagioclase, K-feldspar and anorthoclase. The plagioclases in the Sil–Crd and Crd–Kfs hornfelses are andesines; those of Cor8 show mild normal zonation from An46 cores to An34 rims. The compositions of small plagioclase grains in Opx–Crd hornfelses Pir1 and Pir4 are tightly clustered between An34 and An38; those of large plagioclases in BQ17 are more variable (An35–40Or02–08) (Fig. 6a). The plagioclases in the cordierite norites are commonly strongly zoned, with distinct An-rich cores and Ab-rich rims. In BQ38, for example, compositions range from An59 to An41 (Fig. 6b) and in Dun12 they range from An74 to An49. Oscillatory zoning occurs within both core and rim regions. Such zoning patterns are normally present only in plagioclases from igneous rocks, rimward enrichment in Na being commonly interpreted as reflecting decreasing temperatures during crystallization. These zoning patterns therefore strongly suggest that the plagioclases in Crd norites crystallized in the presence of melt, with rims crystallizing during cooling.
|
K-feldspars in hornfelses Cor8 and BQ17 have compositions Or81–85 and Or87–90, respectively, whereas those in cordierite norite BHQ1 are Or82–88. The composition of the anorthoclase in BQ41 is variable within the thin section, ranging in a linear fashion mainly between Or50Ab43An07 and Or24Ab66An10. No twinning or optical zoning is apparent in either type of feldspar.
Ilmenite
All analysed ilmenites are close to the FeTiO3 end-member in composition. Fe2+ contents invariably exceed 0·96 atoms per 3(O), the main diluent being Mg.
|
P–T–aH2O CONDITIONS OF METAMORPHISM |
|---|
In this section, mineral equilibria are used to estimate the peak P–T conditions of Grampian regional metamorphism in the country rocks of the Huntly Complex, and the peak P–T–aH2O conditions of contact metamorphism in hornfelses and migmatites. The results provide information on (1) the thermal state of the country rocks before intrusion, (2) the crustal level of emplacement of the gabbros, (3) the conditions of migmatization, and (4) the P–T path followed by the rocks.
P–T conditions of Grampian regional metamorphism
Peak-metamorphic conditions of regional rocks to the west and south of the Huntly Complex were estimated by applying version 2.75 (1998) of the program THERMOCALC (Powell & Holland, 1988; Holland & Powell, 1998), in ‘average P–T mode’, to coexisting minerals in schists Clas66 and Sin2. Activity models used were Berman (1990) for grossular, almandine and pyrope components in garnet, Elkins & Grove (1991) for anorthite in plagioclase, Holland & Powell (1990) for muscovite and ferroceladonite in muscovite [with non-ideal parameters for alkali-sites from Chatterjee & Flux (1986)], Holland & Powell (1990) for annite, phlogopite and eastonite in biotite, and ideal ionic mixing for staurolite. Andalusite, quartz and H2O activities were assumed to be unity.
On the basis of five independent equilibria, the best-fit results for andalusite schist Clas66, from the south of the complex, are T = 537 ± 42°C and P = 0·27 ± 0·12 GPa (Fig. 7), with a fit index ( f ) of 0·88, which, passing the 
2 test, indicates that the averaging is reasonable with 95% confidence. These results are consistent with those of Hudson (1985) for andalusite-zone rocks on the Banff coast. The pressure value implies a depth of burial of 10 ± 4 km before gabbro intrusion. Five independent equilibria in sillimanite-zone schist Sin2, from the west of the complex, gave a best fit of 637 ± 31°C and 0·65 ± 0·13 GPa ( f = 0·25). The pressure difference between localities on the south and west of the complex is consistent with the abrupt change in regional metamorphic character across the Portsoy–Duchray Lineament noted by Harte & Hudson (1979) and Beddoe-Stephens (1990).
|
P–T conditions of contact metamorphism
Peak contact-metamorphic conditions in the non-migmatitic hornfels and the highest-grade (Opx-bearing) migmatites were estimated using THERMOCALC and independently calibrated thermobarometers. The latter include the Grt–Bt (Bhattacharya et al., 1992), Grt–Crd (Nichols et al., 1992), Grt–Opx (Bhattacharya et al., 1991) Fe,Mg-exchange geothermometers, the Opx(Al)–Grt geothermometer (Harley & Green, 1982) and the Grt–Opx–Pl–Qtz geobarometer (Bhattacharya et al., 1991). For the Grt–Opx–Pl–Qtz geobarometer, end-member activities were calculated using the quaternary mixing model of Berman (1990) for garnet, Elkins & Grove (1991) for plagioclase, and Wood & Banno (1972) for orthopyroxene. Results from most of these calibrations are summarized in Table 5.
|
Non-migmatitic hornfels
Application of Fe,Mg-exchange geothermometers to Sil–Crd hornfels sample Dun1 yields temperatures in the vicinity of 600°C. In ‘average P–T mode’, THERMOCALC yields best-fit conditions of T = 628 ± 75°C and P = 0·51 ± 0·08 GPa (f = 0·93) for this rock at an arbitrary aH2O value of 1·0.
High-grade hornfelses and diatexite matrix rocks
Application of Grt–Opx thermometers to Opx-bearing contact rocks yields temperatures mostly in the range 850–950°C (Table 5). Fe,Mg-exchange thermometers involving biotite and cordierite, on the same samples, generally give significantly lower temperatures, reflecting the susceptibility of such exchange equilibria to retrograde re-equilibration (e.g. Frost & Chacko, 1989; Fitzsimons & Harley, 1994). The anorthoclases of sample BQ41 plot mostly between the 900°C and 970°C isotherms on the 5 kbar ternary feldspar solvus of Elkins & Grove (1991), in agreement with the Grt–Opx thermometry.
Application of the Grt–Opx–Pl–Qtz geobarometer to cores of coexisting minerals in Opx–Crd norites yields pressures [at temperatures given by Grt–Opx (Fe,Mg) thermometry] ranging from 0·39 to 0·57 GPa (±>0·1 GPa) and averaging 0·49 GPa, consistent with the results of Ashworth & Chinner (1978) and Droop & Charnley (1985). However, these pressures must be treated with caution because the quartz in these rocks is interstitial and probably crystallized after the garnet, orthopyroxene and plagioclase. As an independent check, P–T conditions were also calculated using equilibria among end-member components in the quartz-absent restitic assemblage Grt + Opx + Crd + Hc + Pl in Crd norites and Opx–Crd hornfelses. The following ideal mixing models were used: for cordierite aFeCrd = (XFe)2 · Xv and aMgCrd = (XMg)2 · Xv, where Xv is the mole fraction of channel-site vacancies, and for hercynite aHc = XFe · (XAl)2 and aSpl = XMg · (XAl)2. The Xv values for Crd norites and Opx–Crd hornfelses were calculated from the maximum H2O contents permitted by the gravimetric data in cordierites from BQ38 (0·36 H2O a.p.f.u.) and 10035 (0·22 H2O a.p.f.u.), respectively. Best-fit results, calculated using THERMOCALC in ‘average P–T mode’, are mostly in the range 900–950°C, 0·4–0·5 GPa ( f < 1·2) (Table 5), in agreement with results of Grt–Opx–Pl–Qtz barometry.
The present thermobarometric results for Opx-bearing rocks confirm earlier pressure estimates (0·45 ± 0·1 GPa) but indicate that peak temperatures were in the region of 900 ± 50°C, higher than previously thought.
H2O activity conditions of contact metamorphism
Having estimated peak P–T conditions in the Opx-bearing rocks, dehydration equilibria can now be used to estimate aH2O conditions (Lamb & Valley, 1988). Crd norite BHQ1 contains biotite, quartz, orthopyroxene and K-feldspar, for which the equilibrium
 | (1) |
0·3. At 830°C, arguably a more realistic temperature for the crystallization of interstitial quartz and K-feldspar and poikilitic biotite around orthopyroxene, lower aH2O values are implied (0·06 and 0·2 for Mg- and Fe-end-member versions, respectively). The data of Clemens & Wall (1981) yield aH2O values of 0·33–0·42 for K-feldspar crystallization at the expense of Opx and melt over a temperature range of 845–820°C at 0·5 GPa in an S-type granitic system. The low volatile contents of cordierites also provide evidence for low values of aH2O in the highest-grade rocks. Application of the H2O–cordierite data of Harley et al. (2002) to BQ38 and 10035 cordierites at 900°C and 0·45 GPa yields maximum aH2O values of 0·27 and 0·13, respectively. The H2O contents of granitic melts under those conditions, calculated using the 900°C, 0·5 GPa cordierite–melt partitioning data of Carrington & Harley (1996), are 4·5 and 2·9 wt %, respectively, significantly lower than those of H2O-saturated eutectic haplogranitic melt (Holtz & Johannes, 1994).
|
REACTION HISTORY |
|---|
In this section we use textural evidence and established chemographic relations to reconstruct the mineralogical changes that took place as the Dalradian metapelites, in the vicinity of the Huntly Complex, were thermally metamorphosed and partially melted by the gabbro intrusion.
Subsolidus reactions
The sillimanite pseudomorphs after andalusite in Dun1 provide textural evidence for the conversion of regional andalusite-bearing schists, such as outcrop south of the Huntly Complex, into Sil–Crd hornfels. The simplest set of KFMASH discontinuous reactions by which the dominant regional And + St + Bt + Ms + Pl + Qtz assemblage could have been converted into Grt + Crd + Sil + Bt + Pl + Qtz is as follows:
 | (2) |
 | (3) |
 | (4) |
Reactions (3) and (4) must have occurred in the stated sequence, but the relative timing of reaction (2) is unconstrained.
Partial melting reactions
The assemblage biotite + Al-silicate + quartz, which characterizes many regional schists and the non-migmatitic hornfels, does not appear to occur in equilibrium in the migmatites. The widespread occurrence of garnetiferous leucosomes in the metatexites (Fig. 3b) and of large sub/euhedral garnets and cordierites in the garnet tonalites (Fig. 3c) suggests that this assemblage was destroyed by the incongruent, fluid-absent melting reactions
 | (5) |
 | (6) |
There is no textural or mineralogical evidence for any further melting reactions in the metatexites and garnet tonalites. However, the crystallization of orthopyroxene and hercynite in the cordierite norites and Opx–Crd hornfelses requires that further melting reactions occurred, and indicates that these rocks experienced even higher grades of metamorphism.
In relatively siliceous, low-alumina rocks, sillimanite is likely to be the first mineral to be consumed in reaction (6). Further heating would then result in incongruent, fluid-absent melting of biotite + garnet + quartz (Grant, 1985a, 1985b; Holland et al., 1996):
 | (7) |
In relatively aluminous, silica-poor rocks, quartz may be exhausted before biotite or sillimanite in reaction (6). In these rocks, instead of reaction (7), further heating would result in incongruent melting of biotite + sillimanite + garnet to produce a quartz-free solid product assemblage (Pattison & Tracy, 1991; Srogi et al., 1993):
 | (8) |
On further heating, any biotite remaining after reactions (7) and (8) would ultimately be consumed by reaction with garnet and cordierite (Pattison & Tracy, 1991):
 | (9) |