Pdf/142/1/7/4888516/Gsjgs.142.1.0007.Pdf by Guest on 27 September 2021 8 D

J. geol. Soc. London, Vol. 142, 1985, pp. 7-28, 11 figs, 5 tables. Printed in Northern Ireland

A petrogenetic grid for pelites in the Ballachulish and other Scottish thermal aureoles

D. Pattison & B. Harte Grant Institute of Geology, University of Edinburgh, King’s Buildings, West Mains Road, Edinburgh EH9 3JW SUMMARY:The Ballachulish ‘Granite’ is acomposite Devonian intrusive complex surrounded by a distinctive thermal aureole developed in regionally deformed and metamorphosed Dalradian sediments on the W coast of Scotland. The most abundant rocks in the aureole are pelites, which show aprogression of assemblages from quartz-muscovite-chlorite upto a variety of high-grade assemblages involving combinations of cordierite, corundum, spinel and, rarely, hypersthene and garnet. Metamorphiczones have been mappedaround the granite, which are defined by the following reactions going upgrade:

(1) Mu + Chl + Q = Cd + Bi + V Mu+Bi+Q=Cd+Kf+V (2) or { Mu+Cd=Q+Bi+As+V (3) Mu+Q=As+Kf+V Q+Bi+As=Cd+Kf+V (4) { or Mu+Cd=Bi+As+Kf+V (5) Mu=Cor+Kf+V (6) Bi+As=Cor+Kf+Cd+V The restricted occurrences of assemblages involving spinel, hypersthene and garnet do not allow higher grade zones to be mapped. Variations in the reaction sequence as a consequence of bulk compositionalfactors, in particular the development of quartz-bearing versus quartz-absent assemblages, are described. Details of the mineral assemblages from Ballachulish are combined with high-grade assemblage datafrom the Belhelvie,Lochnagar and Comrie aureoles to constructa comprehensiveschematic petrogenetic grid. The grid involves the minerals quartz, chlorite, muscovite, biotite, cordierite, alumino-silicate, K-feldspar, corundum, spinel, hypersthene and garnet, whose assemblagerelationships are modelled in the system K,0-FeO-MgO-A1203- Si0,-H20 (KFMASH).

This paper has two principal objectives: to summarize assemblages. For natural mineral assemblages, this is the sequence of pelitic assemblages developed in the accomplished by examining their variations in rocks of thermal aureole of the Ballachulish ‘Granite’ (Argyll- similar bulk composition at different grades of meta- shire),and to combine the assemblage information morphism. Consideration of the assemblages within a from Ballachulish with that from other Scottish modelframework of components and phases (con- thermal aureoles to develop a comprehensive schema- forming tothe Gibbsphase rule), followed by tic petrogenetic grid for thermal aureoles in general. systematization of the reactionboundaries between A great deal of work has gone into the construction different assemblages, using Schreinemakers’ analysis, of petrogenetic grids for regional metamorphic settings producesa schematic ‘grid’ or Schreinemakers’ net. atpressures above about 3 kbar (e.g. Albee 1965; Such a grid incorporates a great deal of relative P-T Harte & Hudson 1979; Carmichael,unpubl.). Much information. less precise is the knowledge of relative and absolute In the first part of the paper we present a detailed stabilities of commonmineral assemblages in low- analysis of the distribution of mineral assemblages pressure geological settings, such as thermal aureoles. found in thethermal aureole of the Ballachulish Another feature of existing petrogenetic grids is that ‘Granite’,Argyllshire, and then use these data quartz-absent assemblages have been largely neg- together with data from the aureoles of the Lochnagar lected, even though they are frequently developed in granite, Aberdeenshire, Belhelvie Gabbroic Complex, thermal aureoles. Aberdeenshire, and Carn Chois Diorite Complex In constructing such a grid, the most important step (Comrie Aureole), Perthshire, to construct a compre- is to establish the relative stabilities of mineral hensive schematic petrogenetic grid. The grid incorpo-

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 8 D. Pattison & B. Harte

rates bothquartz-bearing and quartz-absentreaction the Lochaber and Ballachulish subgroups (see Table 1 sequences, both of which are well developed in the and Fig. 1). The pelites that enter the aureole range aureoles. from semi-pelites to black graphitic, sulphide-bearing The minerals involved includechlorite, quartz, slates. muscovite, biotite,cordierite, K-feldspar, alumino- Structurally, the Loch Leven area has undergone at silicate, corundum,spinel, hypersthene and garnet, least five deformational episodes during the Cambro- which may be modelledin the system K20-FeO- OrdovicianCaledonian Orogeny (Treagus 1974). In Mg0-A1203-Si02-H20 (KFMASH). the vicinity of the aureole, structures are largely of the Calibrating this schematicframework in P-T-X first deformation (D1), as manifested in themajor space involves both the direct experimental determina- NE-SW trending foldsand slides (see Fig. 1). tion of the P-T-X variations of specific reactions and Associated with the major D1 structures is a penetra- the use of chemical thermodynamics. In this paper we tive slaty cleavage developed in the phyllites, and a arenot concerned with the details of calibration, spaced cleavage in the dolomites and quartzites. Later focusing instead onthe construction of a schematic deformationepisodes are indicated by small-scale grid based on careful analysis of awide range of refolding and crenulation cleavages. naturalmineral assemblages which we hope will provide a solid framework for reliable calibration and Regional metamorphism other refinements. One major regional-metamorphic episode, approx- imately coeval with the D1-deformation, has affected Geological setting of the the rocks of the Ballachulish area. The grade increases Ballachulish area towards the SE, from a zone of chlorite-muscovite- quartz in the NW, into abroad zone where biotite Stratigraphy and structure appears sporadically, and finally into a garnet zone. The garnet-bearingzone is first developed in the The Ballachulish ‘Granite’, one of several composite pelites on the SE margin of the large Leven Schist belt calc-alkaline intrusions of Devonian age in the Scottish that runs through the centre of the map area (see Fig. highlands, intrudes lower Dalradian metasediments of l), and ceases abruptlyat the Ballachulish Slide.

TABLE1: Stratigraphic units in the Ballachulish area

Group Subgroup Formation Description Middle Dalradian BlairAtholl Cui1 slateBayDark greyslate Appinphyllite Grey phyllitewith flaggy interbeds ______------Appinlimestone* Cream-coloured, banded limestone (Tiger rock) White dolomite Appin quartzite White, gritty, feldspathic, current-bedded Lower Dalradian Ballachulish quartzite (Appin Group) StripedFinelyinterbedded quartzite and black slate transition zone Ballachulish Black graphitic,sulphide bearing slate slate

Ballachulish Dark pure limestone limestone Calcareous schist Light-coloured limestone Leven schist Grey phyllites and semipelites Lochaber Glen CoeGlen Fine-grained,well-bedded quartzite quartzite * This boundary is not very clear. Often the limestone units and the phyllite are interbedded, and the limestone is not always the first unit above the Appin quartzite.

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid foraureoles thermal 9

Garnet is absent in the Ballachulish Slate and Appin (sometimes with clinozoisite in the slightly more Phyllite to the SE of the slide, but reappears when the calcareous rocks); they are not significant components Leven Schist again outcrops. The localization of of the principal phases discussed below. This is also garnet-bearing assemblages to the Leven Schist sug- true of minorcomponents such as FezO3, P205 and gests strong bulk compositional control. Ti02, which if present in sufficient quantity in the bulk composition give rise to additional phases (magnetite, The igneous complex apatite, rutile or ilmenite). The main exception is probably TiOz inbiotite, where Ti02 may reach In addition tothe main intrusive complex at 5 wt%. However, Ti02 is not an important constituent Ballachulish, there are numerous small mafic igneous of theother silicates and all rockscontain rutile or bodies which occur nearby.These include explosion ilmenite; thus, ignoring both TiOz and the oxide phase breccias and a seriesof coarsely crystalline pyroxene f does not affect the variance of individual assemblages. olivine * amphibole f biotite pipes (the Appinite In general, Thompson’s (1957) arguments for ignor- Suite, Bowes & Wright 1967). These bodies represent ing minorphases and components apply well, and the earliestphases of igneous activity in thearea, suggest that KFMASH is a reasonable model system. predating the main intrusive complex. Within this system may be represented all the principal The Ballachulish ‘Granite’ itselfis acomposite phases which are abundant in the rocks andlor show intrusive complex classified by Read (1961) as one of significant changes indistribution with grade. These the ‘Newer granites’ which occur throughout the principal phases include quartz, chlorite, muscovite, Scottish Caledonian Orogen. Essentially, the complex biotite, cordierite, K-feldspar, AlZSiO5 minerals, may be divided intoan outer tonalitelquartzdiorite corundum, spinel,hypersthene and garnet.In addi- which is cross cut by acentral pink granitelquartz tion, at the time of metamorphism, it is assumed that monozonite (Streckeisen 1967). Both phases have no there were fluid phasespresent: hydrousa fluid penetrativepost-emplacement tectonic fabric. The (volatile phase) at low grades and a silicate melt (melt outer tonalitecontains numerous xenoliths of phase) at high grades. sedimentaryand igneous origin which are found The occurrence of a volatile phaseat low and mainly at the margins of the body. On a large scale, middle grades is indicated by the marked decrease in intrusive contacts are discordant to bedding (see Fig. hydrous minerals (e.g. chloriteand muscovite) with 1). increasing grade (cf. Thompson 1957); in high-grade The lack of tectonic fabric in the complex and the rocks, biotite (and less importantly, cordierite) are the discordantcontacts suggest that the intrusion of the only remaining hydrousphases. While it is possible complex was a late eventduring the Caledonian that the rocks were ‘dry’ (fluid-absent) at the begin- Orogeny.This is consistent with radiometric K-Ar ning of prograde metamorphism, theonset of de- dates frombiotite andhornblende of theouter hydration reactions (even if prematurely promoted by tonalite, which give a lower Devonian closure age of an activity of water less than unity) will have rapidly 400 * 20 Ma (Brown et al. 1965). A detailed study of introduced a hydrous vapour phase. The presence of the field relations, petrology and geochemistry of the graphite in the slatesand minor carbonate in some igneous complex is currently being conducted by S. marginal low-grade phyllites will introduce primarily Weiss (Minera1og.-Petrograph. Institut, Munchen). CH4 and CO2to the volatile phase (Ohmoto & Recentstudies by one of us (D.R.M.P.) and S. Kerrick 1977). Referring to Fig. 3 of Ohmoto & Weiss have revealed a large NE-SW-trending trans- Kerrick (1977), the maximum XHZOof a fluid in current fault, which post-dates the intrusive complex equilibrium with graphite lies between 0.8 and 0.95 in (see Fig. 1).This fault has a sinistral strike-slip the range 450-650”C and 34kbar, the approximate displacement of at least 700m, and may be traced for P-T conditions of low and middle grade in the aureole over 50 km. (see latersection). It is therefore assumed thatthe volatile phase was close to pure H20 in most pelitic rocks. Chemographic analysis In situ partial melting is clearly suggested in high grade rocks by the occurrence of unusual textures and The model chemical system and fluid structures, and the segregation of leucocratic phases, in particular K-feldspar (this is covered in more detail phases in a later section). With the onset of melting it is likely The principal chemographic relations of the mineral that the hydrous volatile phase dissolves in the silicate assemblages in the pelitic and semi-pelitic rocks of the melt, rather than forming a separate phase. Thus, it Ballachulish aureole may be expressed using Thomp- seems a reasonable approximation to assume that all son’s (1957) K~0-Fe0-Mg0-A1z03-SiOz-H~0 mineral assemblages coexist with only one fluid phase, (KFMASH) system. Na20 and CaO are notlarge bulk which is a melt at the highest grades and a hydrous components and are mainly contained in plagioclase volatile phase at lower grades. This assumption is used

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 FIG. 1. Geological map of the Ballachulish area. The map is a composite from several sources: Bailey & Maufe (1960) (Sheet 53, B.G.S.), general geology, primarily outside the thermal aureole; Bowes & Wright (1967), general geology of Ardsheal peninsula; Weiss (unpubl.), details of the main igneous complex; this study, details of the stratigraphy in the aureole, regional metamorphism and all thermal aureole zones. The thermal aureole zones are constrained by about 300 specimens in addition to field evidence. Zone I1 is not visible in all areas. Zones Va and Vb are only developed clearly in the small syncline on the NE margin of the igneous complex, where there are sufficient quartz-absent assemblages to permit the mapping of separate zones. Gt, Hy and Sp indicate localities of garnet, hypersthene and spinel-bearing assemblages respectively.

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogeneticgrid foraureoles thermal 11

in allocating degrees of freedom tothe mineral We have chosen two principal projections: the AKF assemblages, and although a simplification, it provides for low-grade,quartz-bearing assemblages, and the a reasonable basis for the general paragenetic analysis AFS forhigher grade rocks that all contain K-feldspar. given here, which relatesa wide array of pelitic In applying both of these projections, we assume the assemblages from thermal aureoles. presence of a single H20-bearing fluid phase(see With the appearance of melt as the fluid phase, the above). The AKFS tetrahedron and the two projec- assumption of similar water activities in most rocks tions from quartz and K-feldspar, to yield the AKF probably becomes untenable. Water activity must be and AFS triangles, respectively, are illustrated in Fig. expected to be buffered by various melting reactions 2. (e.g. Powell 1983). However,for the highest grade Because theAKF and AFS projections do not mineral assemblages we are not particularly concerned separate Fe0 and MgOcomponents, a number of with their zonal distribution, but most especially with common assemblages plot as four-phaseassemblages in their variety. We shall show that this may be these ternary projections; for example, Mu-Chl-Cd- accounted for relatively simply without detailed analy- Bi, Mu-Cd-Bi-Kf,Mu-Cd-Bi-As, (all with quartz sis of the precise melt relationships. and hydrous fluid) in AKF, and Mu-Bi-As-Cd and It is interesting to note thatlow variance 'univariant" Bi-As-Cd-Cor (with K-feldspar and hydrous fluid) in assemblages are much more common amongst the AFS(see Fig. 3). Being in the six-component highest grade (melt phase involved) assemblages than KFMASH system, the crossed tie-line relationships in amongst the lower grade(volatile phase involved) theseprojections therefore indicate divariant (con- assemblages. This probably relates to the buffering of tinuous)equilibria, not univariant (discontinuous) water activity along melt reactions, and the fact that equilibria (which require seven phasestogether in most melting, and reaction, occurs at the intersections equilibrium). of melting curves and dehydration curves in T-XH,o space (see Powell 1983.) Metamorphic zones Graphical projections Metamorphic zones in the pelites have been mapped The wide variety of pelitic assemblages developed in aroundthe granite complex (see Fig. 1). These are the Ballachulish aureole make it impossible to repre- based on field evidence and thin-section petrography; sentthem all with a single composition-paragenesis thelatter is particularly important owing tothe diagram. The absence of quartz and/or muscovite in fine-grained nature of the hornfelses. many rocks at medium and high gradesmakes the The zones in Fig. 1 essentially occur below the onset common AFM projections invalid, while at low grades of in situ partialmelting in theaureole. Reactions where quartz and muscovite are present, most of the above the onset of melting are abundant, but are not reactions involve only two ferromagnesianphases readily mappable owing to their development in small (cordieriteand biotite), which means thatthe AFM areas adjacent to the igneouscontacts; they will be diagram is not particularly informative. examined separately in a later section.

Key to Fig. 1. SEDIMENTARYUNITS INTRUSIVE ROCKS THERMALAUREOLE ZONES B CulBay Slate Granite I Mu Chl Q m AppinPhyllite Tonalite ---I_II ~ppinLimestone m MaficStock ___--- Cd Bi [ Appin Quartzite Mu Bi Q Ill Mu Cd Drift CdKf " " I ' ' QBiAs ' m I TransitionSeries B REGIONALSTRUCTURE IVb M; Q IVa AND METAMORPHISM *As,Kf Ballachulish Slate Fault/Slide Ballachulish Limestone 4 Lithologic Contact m Leven Schist 19) Glen CoeQuartzite Granite/Tonalite

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 12 D.Pattison & B. Harte

FIG. 2. The AKFS tetrahedron and the AKF and AFSprojections, using the idealized mineral formulae from Table 2. Note that the low silica content of eastonitic biotites leads them to plot outside the AFS triangle. In subsequent figures, for simplicity of presentation, we have plotted biotite on the A-F side of the triangle near the F apex.

Each zone, proceeding upgrade, is briefly described. whole numbers which indicate the approximatepro- Critical low variance assemblages are implicitly fo- portions of phasesproduced and consumed in each cused on, since these define the reaction sequence, reaction-in reality,these stoichiometric coefficients although higher variance assemblages are more abun- vary with the exact composition of the biotite (and dant. with the compositions of all the other minerals). The stoichiometric coefficients of the reactions have All of the reactions involve cordierite, which been calculated assuming the ideal mineral composi- containsa variable amount of water in itsstructure tions of Table 2. Biotite is assumed to be onthe (Newton & Wood 1979). The amount of water in the annite-phlogopite join, althoughactual biotite com- cordierite, n, multiplied by the stoichiometric coef- positions fall between this join and the siderophyllite- ficient of cordierite in the reaction, is subtracted from eastonite join. However, by using thisassumption, or added to themoles of water released in the reaction stoichiometric coefficients are kept to relatively small ,calculated assuming anhydrous cordierite, depending uponwhether the cordierite is onthe vapour- producing or vapour-consuming side of the reaction. TABLE2: Abbreviations and idealized mineral formulae of pelitic minerals used in the graphical projections Zone I (marginal zone) Quartz Q SiOz Chlorite Chl (Fe,Mg)d&Si301d0H)8 Pelitic rocksoutside andon the margin of the Muscovite Mu KA13Si,0,0(OH)Z aureole show regional metamorphic assemblages Biotite Bi K(Fe.Mg),AIS~301n(OH)z containing: muscovite-chlorite-quartz f biotite f -K(% Mg),sA1ZSi~.SOI(I(OH)2 garnet f epidote-zircon-apatite-rutile f carbonate k Cordierite Cd(Fe, Mg)ZA14Si5018. nHzO graphite f sulphide. It is difficult to interpret textural- K-feldspar Kf KA1Si30, ly whether biotite is of contact or regional metamor- Al-silicate AIZSi05 As phic origin. In a number of specimens on the margins Corundum Cor A1,03 of the aureole, biotite clearly cross-cuts the regional Spine1 SP (Fe,MdA1204 schistosity, suggesting a late origin, but whether this is Hypersthene Hy(Fe,Mg)Si03 Garnet (Fe,Gt M&Al2Si3OIZ in response to the intrusion or is the result of a late regional metamorphic growth is difficult to assess.

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid foraureoles thermal 13

Owing to the complexities and variability of biotite- irregularly surrounded by very fine (<0.5 mm) K- forming reactions (see Mather 1970), these scattered feldspargrains, and contains fewer inclusions of occurrenceshave not beenincorporated intothe muscovite and quartz. In the surroundingmatrix, aureole reaction sequence. muscovite decreases in abundance, and K-feldspar is also intergrown with the micas and quartz. The model reaction that accounts for the incoming Zones I1 and (lower cordierite zones) I11 of K-feldspar with cordierite is another continuous The first mineralchange that most clearly and reaction: consistently delineates the boundary of thethermal metamorphism aroundthe Ballachulish ‘Granite’ is 6Mu + 2Bi + l5Q = 3Cd + 8Kf + (8 - 3n)HzO (2b). the first appearance of cordieritein the slates and As one proceedsupgrade in zone IVb,the pelites phyllites (see Figs 1 and 3). The cordierite first appears become progressively moreindurated as micas are as 1-2mm ovoid spots irregularly scattered on consumed to produce cordierite and K-feldspar. cleavage or beddingsurfaces, more easly visible in In the Ballachulish slate, upgrade of zone 111, handspecimen than in thinsection. The cordierite andalusite appears to give the assemblage Mu-Bi-Q- ‘ovoids’ usually have their long axis (C-axis) oriented Cd-As(zone IVa). Althoughandalusite may be parallel tothe schistosity, while onthe cleavage identified in the field, this change is less obvious in the surfaces, the ovoid shapes are randomly oriented. As field than the appearance of K-feldspar (zone IVb), one proceeds upgrade, the cordierite increases in size because the spotted slates retain much of their fissility and abundance, and the phyllite becomes less fissile. and general appearance from zone 111. In thin section, In thin section, the cordierite occurs as roughly the andalusite grows in small (1-3 mm) prisms inter- ovoid poikiloblastic growths that irregularly overgrow grown or surrounded with biotite and spatially related the schistosity and exhibit patchy extinction. The to cordierite. The continuous reaction that introduces cordierite is packed with fine inclusions of muscovite, andalusite is quartzand matrix accessories, and is frequently associated spatially with biotite, which occurs in all 2Mu + 3Cd = 7Q + 8As + 2Bi + 3nH20 (2a). cordierite-bearing rocks. The small amount of waterreleased dueto the The inferred model reaction that introduces cordier- breakdown of cordierite represents all of the released ite is vapour in this reaction,and therefore strongly in- Mu + Chl + 2Q = Cd + Bi + (4 - n)HzO (1). fluences the reaction topology. In summary, zones IVa and IVb are not sequential Being a six-phase reaction in a model six-component in grade, but are defined by mineral assemblages system,this therefore representsadivariant (con- which may occur in rocks at the same grade due to tinuous) reaction, and occurs over a range of pressure differences in bulk composition (fuller explanation is and temperature. In some areas around the igneous provided in the section on the schematic grid). complex, this full divariant assemblage occurs over a broad enough area to be mapped as a separate zone (Zone 11, Figs 1 and 3). Muscovite + quartz breakdown Upgrade of this zone, chlorite is consumed (no The next important reaction that occurs upgrade of specimens have been found where either muscovite or bothzones IVa andIVb is the muscovite-quartz quartz has been consumed and chlorite remains). The breakdown: assemblage Mu-Bi-Cd-Q is abundant and persists over a wide area on the ground (Zone 111, Figs 1 and Mu+Q=As+Kf+HZO (3). 3). In the field, theappearance of andalusite with K-Feldspar is very clear: in rocks containing the zone IVa assemblage (Mu-Cd-Bi-Q-As), the appearance Zones IVaand IVb (upper corderite of K-feldspar in addition to andalusiteproduces an zones) abrupt loss of fissility, due to the loss of muscovite, Upgrade of zone 111, pelites enter zone IV, which while in the alreadytough cordierite-K-feldspar varies mineralogically and texturally as a function of hornfelses of zone IVb, andalusite appears as conspi- bulk rock composition. cuous prisms, usually between 5 mm and 10 mm, and In much of the aureole (particularly outside of the sometimes up to 5 cm in length. In thin section, the Ballachulish Slate (see Fig. l)),the cordierite-spotted reaction is texturally manifested by the intergrowth phyllites rapidly lose their fissility over a distance of a and/or fringing of andalusite with K-feldspar. In all few tens of metres,and texturally become classic specimens, minor sillimanite is present as fibrolite, fine-grained,tough hornfelses (zone IVb).The cor- which often grows in cordierite or nucleates on zircon dieritein thin section is intergrown with andoften or some otherinert accessory mineral;texturally,

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 14 D. Pattison & B. Harte

I N V

Ll Iu+CdtQ+BitAa+VQ+Bi+Aa+Cd+Kf+V I I l APPROXIMATE W -- _-. ' i. ONSET OF MELTING l K FF S I A A I A Fe +Mg Mg +Fe Quartz present ----- Quartz absent Nb l

Ereasing temperature I L==-

MARGINAL ZONE LOWER CORDERITE ZONE MUSCOVITE + QUARTZSTABLE ZONES I AbSi05+K-FELDSPAR STABLE ZONES I I

FIG. 3. Chemographic projections of metamorphic zones up to the onset of partial melting, corresponding to the zones in Fig. 1. Stippled ellipses represent the qualitative bulk composition ranges of the pelites, based on observed mineral assemblages. For low-grade, quartz-bearing zones, the AKF projection is used; for higher grade zones, the AFS projection is used (see Fig. 2). To demonstrate thesame reaction and bulk compositionrange in the two projections, both are included in zone IVb. TheFe-Mg arrows in the divariant reaction zones indicate the direction of migration ofFe-Mg tie-lines as one proceeds upgrade through the zones (see also Figs 4 and 7). V refers to hydrous vapour.

andalusite appears tobe in equilibrium with the rest of The muscovite-quartz breakdownreaction has a the assemblage. critical effect on the assemblages developed in the rest Assuming the muscovite to pure be of theprograde sequence, because pelites that pass KA12Si3A1010(OH)2, this reaction is a degenerate through this reaction either remain quartz-bearing or univariant(discontinuous) reaction in KFMASH, become quartz-absent (silica undersaturated), depend- against which reactions 2a and 2b terminate (see Figs ing upon whetherquartz or muscovite is first con- 3, 4 and 5). Assuming thatthe prograde P-T sumed in reaction 3. trajectory in the aureole is isobaric, the occurrence of both divariant assemblages 2a and 2b necessitates that theprograde path passes throughthe portion of Zone V (K-feldspar + AI2SiO5 zone) univariantreaction 3 that joins the Fe and Mg end-member invariant points produced at the intersec- Quartz-bearing assemblages tion of reactions 2a, 2b and 3 (see Fig. 4). All pelites entering reaction 3, whether from zone, IVa or IVb, In most of the pelites in the Ballachulish aureole, will have the sameassemblage Mu-Bi-Q-Cd-Kf-As- quartz persists and muscovite is consumed at reaction (H20). Three specimens have been found that contain 3, leaving the assemblage Q-Cd-Bi-As-Kf-(H20), this rare univariant assemblage. which defines the divariant KFMASH reaction(see

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid for thermal aureoles 15

FIG. 4. Schematic P-T diagram for mineral zones developed in the Ballachulish aureole up to the onset of partial melting (see Fig. 1 and 3). Zones and ornaments are the same as those used in Fig. 1. Dashed lines represent quartz-absent reactions. For divariant Fe-Mg reactions, the Mg end-member lines have been labelled. V refers to hydrous vapour. This diagram, particular to Ballachulish, is a restricted part of the complete Schreinemakers’ net shown in Fig. 5 and rationalized in Table 3 (see footnote 0 of Table 3 for the restrictions). The reactions have been schematically orientated in P-T space to be consistent with experimental data on some of the reactions (see Fig. 11). The relative positions of the KMASH and KFASH end-member reactions that bound the divariant fields are based on experimental data and probe data from this aureole and various other localities (see footnote? of Table 3 and Fig. 7 for a brief discussion of reaction 1). For divariant reactions 1 and 2a, metastable as well as stable parts are shown in order to simplify the diagram. The complete arrangement of stable reactions is shown by the [Cor, Kf] invariant point in Fig. 5. Figs 3 and 4), in hand specimen, in thin section there is often a clear textural association of biotite,andalusite and K- 6As + 2Bi + 9Q = 3Cd + 2Kf + (2 - 3n)H20 (4a). feldspar, which tend to grow together. This association Texturally, this zone is not distinctive either in the is in contrast to the andalusite-K-feldspar intergrowth field or in thin section,although this divariant that texturally characterizesreaction 3, so that in assemblage is abundant and persists in quartz-bearing several specimens the occurrence of both associations rocks up to the onset of partial melting. clearly demonstrates the progression of the specimen through the two reactions. The above divariant assemblage defines zone Va in Zone Va (quartz-absent assemblages; silica-undersaturated rocks, which occurs at the same muscovite zone) grade as quartz-bearing rocks of zone V (see Figs 3 In thosepelites where quartz is consumed at and 4). reaction 3, the assemblage Mu-Cd-Bi-As-Kf-(HzO) occurs, which defines the divariant reaction (see Figs 3 Zone Vb (quartz-absent assemblages; and 4, corundum zone) 9Mu + 3Cd = 2Bi + 15As + 7Kf (7 3n)HzO (4b). The next important changeupgrade, in silica- + + undersaturated rocks, is the appearance of corundum, While the progress of this reaction is not conspicuous which marks the onset of zone Vb (see Figs 1, 3 and

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 16 D. Pattison & B. Harte

4). Inthe field, corundum may be identified with occurs atdifferent grades in different bulk composi- difficulty, occurringas small (1-2mm)rounded, tions(see also the final section),a partial-melting knobby grains that weather above the pelitic matrix. ‘isograd’ has not been mapped. Andalusitecrystals are longer andmore distinctly prismatic. In thinsection, corundum usually appears Schematic petrogenetic grid for as0.5-2mm high-relief mottled blobs,often intergrown with K-feldspar and usually altering to musco- metamorphic zones below the vite. Occasionally the corundum forms well-developed onset of partial melting stubby, barrel-shaped, 0.5-1 mm hexagonal prisms. The white mica occurring in corundum-bearing The successive reactionzones mapped aroundthe rocks is usually secondary,but three specimens Ballachulish ‘Granite’up tothe onset of partial contain corundum coexisting with fine-grained, schis- melting,described in the previous sections, are all tosity-parallel muscovite that is interpretedto be related by the reaction net of Fig. 4. The solid phases primary, along with biotite, cordierite, K-feldspar, and involved includechlorite, muscovite, quartz, biotite, alumino-silicate. This univariant assemblage therefore cordierite, K-feldspar, alumino-silicate and corundum; contains the sub-assemblagecharacteristic of the a hydrous vapour phase is also assumed to have been KFMASH degenerage discontinuous reaction, present, as discussed above. Thisrepresents nine phases in a six-component system (KFMASH). Quartz Mu = Cor + Kf + H20 and muscovite are always present in low grade Continuous reaction 4b terminates against this uni- assemblages, and K-feldspar is present in all assemb- variant curve (see Fig. 4). lages above reaction 3 (breakdown of Mu + Q). Immediately upgrade of reaction 5, all quartz-absent Using the general exclusions thatCor + Q and specimenscollected belong to a muscovite-absent Cor + Chl cannot occur, only one KFMASH invariant field, zone Vb, corresponding to thedivariant assemb- point, [Cor], is theoretically possible. Assuming pure lage of the continuous reaction, muscovite, three of the univariant reactions associated with the [Cor] invariant point degenerate into reaction 2Bi + 15As = 3Cd + 9Cor + 2Kf + (2 - 3n)HzO (6). 3 (Mu + Q = Kf + As + V) and the invariant point has to lie on this reaction line; if muscovite is not pure Texturally, this reaction is suggested by the inclusion the invariant point will still lie close to the degenerate of biotite-andalusite-K-feldspar intergrowths, and reaction(see below). Two of theother univariant occasionally euhedral corundum prisms, in cordierite reactionsassociated with [Cor] are(Cor, Mu) and poikiloblasts. (Cor, Q). Theseboth involve the stability of the assemblage Chl + Kf + As in which the occurrence of Partial melting and high grade assemblages chlorite (and particularly the association Chl-Kf) suggests temperatures below reaction 3 (Mather 1970), Asone proceeds upgrade in zones V and Vb, a whilst simultaneously the Kf + As association cannot variety of features suggest that partial melting has be stable below reaction 3. For these reasons it seems occurred inthe peliticrocks. Some of the rocks, likely thatthe invariantreactions (Cor, Mu) and particularly onthe westernside of theaureole, (Cor, Q), and therefore the invariant point [Cor], are develop a migmatitic appearance, and the small-scale metastable, and thatthe only possible stable sedimentary and regional metamorphic layering be- KFMASH univariantreactions are those listed in comes disrupted. Table 3. The exclusion of KFMASH invariant points Elsewhere in the aureole, partial melting is inferred still allows invariantpoints to occur in the five fromthe scattered appearance of K-feldspar-rich component, KMASH and KFASH, chemical systems blebs, lenses and laminae surrounded by selvedges of and a net of these invariant points and their related biotite andcordierite, that are both parallel toand univariant reaction curves is given for either KFASH cross-cut the earlierprimary structures. Sometimes or KMASH in Fig. 5. It is the duplication forboth theycontain bentand brecciatedfragments of the end-member systems of invariantand univariant originalsedimentary layers. The variable size and reactions present in Fig. 5 which gives the basic form distribution of the K-feldspar-dominated segregations of Fig. 4. TheKFMASH univariantreactions link makesthem unlikely to beinjections fromthe corresponding invariant points in the two end-member intrusive igneous complex. systems. Partialmelting fundamentally alters the reaction In the Ballachulish aureole, no development is seen sequence: the activity of water will be lowered, and of thepotential low-grade assemblages involving the interaction of melt phase, of progressively chang- Chl + Kf and Chl + As of Fig. 5 andTable 3; the ingcomposition, with the rest of a high grade lowest grade aureole assemblages all involve Cd + Bi. assemblage is potentially very complicated.Because The schematicpetrogenetic grid of Fig. thus4 initial melt is difficult to identify, and partial melting illustrates Cd + Bi assemblages from low grades up to

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid for thermal aureoles 17 TABLE3: Summary of KFMASH reactions below the onset of partial melting

PHASES: Mu ChlQ Bi Cd Kf AsCor V (9) COMPONENTS: KFMASH (6) EXCLUSIONS: Q-Cor, Chl-Cor OTHER RESTRICTIONS: V assumed present in all assemblages Q and Mu present inall assemblages below reaction 3 Kf present in all assemblages above reaction 3 INVARIANTPOINTS: 0' UNIVARIANT CURVES (invariant pts in KFASH or KMASH if not degenerate): 4t Name Phases present KFMASH reaction [Cor,A] MuChl Q Bi Cd Kf V Mu+Chl+Q+Kf=Bi+Cd+V [Cor, Kf] Mu ChlQ Bi Cd As V Mu + Chl+ Q = Bi + As +Cd + Vf [Cor, Chl]Mu Q Bi CdAs Kf V Mu + Q = As + Kf + V (3) degenerate [Q, Chl]Mu Bi CdAs Cor Kf V Mu = Cor + Kf + V(5) degenerate DIVARIANT CURVES (univariant curves in KFASH or KMASH): 136 K f, As] [Cor, Kf, Mu+Chl+Q=Cd+Bi+V (1) [Cor,Chl]MuKf, + Cd = Bi + As + Q + V (2a) [Cor, Kf, Mu, Bi] Chl+As+Q=Cd+V [Co r, Kf, Cd] Kf, [Cor, Mu+Chl=Bi+As+Q+V [C or, As, [Cor, Chl] Mu+Bi+Q=Cd+Kf+V [C or, As, [Cor, Mu] Chl+Kf=Bi+Cd+Q+V [C or, As, [Cor, Bi] Mu+Chl+Q=Cd+Kf+V [C or, As, Cd] As, [Cor, Chl+Kf=Mu+Bi+Q+V [Co r, Chl, Bi, Cd] Bi, Chl, [Cor, Mu+Q=As+Kf+V (3) [Cor, Chl, Mu] Chl, [Cor, Bi+Q+As=Cd+Kf+V(44 [Cor,Chl, MuQ] + Cd = Bi + As + Kf (4b)+ V Chl, As, Cd, Si] Cd, As,[Q, Chl, Mu=Cor+Kf+V (5) [Q , Chl,[Q, Mu] Bi + As = Cor + Cd + Kf + V (6)

~~ ~ KFMASH univariant curves (invariant pts in KFASH or KMASH) are designated by the phases absent. Reaction curves are numbered as in Figs 4 and 5. V refers to hydrous vapour. * Given only the above exclusions, it is possible that one invariant point is stable: [Cor]. This invariant point is presumed to be metastable because it would have to occur along the P-T line of reaction 3) assuming pure muscovite (Fig. 5) and the distribution of other Fe-Mg equilibria suggests that this is impossible (see text for more complete discussion). t This assumes that muscovite is pure KA13Si,010(OH)2, which makes the [Cor, Cd], [Cor, Bi] and [Cor, Chl] univariant reactions degenerate and identical. The [Cor, Mu] and [Cor, Q] reactions are absent for the same reason as the [Cor] invariant point (see text). $The arrangement of phasesaround this reaction has been the cause of some discussion(see Guidotti et al. 1975; Harte & Hudson 1979; Carmichael et al. 1978), depending upon the AFM position of chloriterelative to cordierite and biotite. This arrangement favours the arrangement of Guidotti et al. (1975) and the results of this study (see Figs 4 and 7). 9 Thenumbered reactions are seen in the Ballachulish aureole. Only cordierite- bearing reactions are considered (see text). Chl-As and Chl-Kf have not been seen in cordierite-bearing specimens, and cordierite always coexists with biotite. temperature conditionswhere evidence of melt is prograde sequence of assemblages. The large range of seen.Both Mg and Fe end-memberreactions are assemblages described by Tilley (1924) in the Comrie shown as well as linking KFMASH univariantreac- aureole all fit into various of the tri-, di- and univariant tions(Table 3). Theimportant KFMASH divariant fields of this grid. (continuous) reactions seen within the mapped zones The two univariantcurves, reactions 3 and 5, are thereforeappear as fields bounded by Mg and Fe bothdegenerate reactions involving muscovite, end-member reactions in Fig. 4. assumed to be pure KA12Si3A1010(OH)2.If muscovite Most of the reaction curves were orientated in the is considered more realistically to have aceladonite schematic grid in Fig. 4 from experimental data (see component in it,then a small amount of biotite or Fig. ll), while the other curves (reactions 4b and 6) cordierite would have to be produced on the (As + weredrawn to be consistent with the observed KO-side of reaction 3 to account for Fe-Mg released

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 18 D. Pattison & B. Harte IP

FIG.5. Schematic KFASH or KMASH Schreinemakers’ net for phases that occur below the onset of partial melting (refer to Table 3). The net is roughly orientated to be consistent with experimental data on some of the reactions (see Fig. 11). Reaction numbers are the same as those used elsewhere in the paper.

fromthe breakdown of the muscovite. This would schematic T-XFe-Mg diagram (Hensen 1971) may be result in biotite-absent and cordierite-absent divariant constructed that more clearly demonstrates this (see fields being generatedfrom univariantreaction 3 in Fig. 7). addition to reactions 2a, 2b, 4a and 4b (see Fig. 6). All chlorite-bearing pelites pass through reaction 1. Joining the Mg and Fe end-membez invariant points Proceeding up-temperature, Fe-rich pelites will enter would bea non-degenerate, seven-phase KFMASH reaction 2a, while Mg-rich pelites will enter reaction univariantcurve involving cordieriteand biotite in 2b. Above reaction 3, the determiningfactor as to addition tothe five other phases involved in the which of reactions 4a or 4b is entered is not the FeMg pure-muscovite reaction 3. ratio of the pelites,but the presence or absence of The very small amount of Fe and Mg contained in quartz. muscovite,relative tothat contained in biotite and As thepelites pass through the upgradesequence of cordierite, means that these three reactions would all continuousreactions, the FeMg ratios of the ferro- lie very close to degenerate reaction 3 in P-T space, so magnesian minerals will progressively change. The that inpractice the assumption of muscovite being cordierite and biotite in the pelites that enter reaction pure introduces little error. Similar arguments apply to 2a will become more Mg-rich as they pass through the degenerate univariant reaction 5. reaction, while those that enter reaction 2b will move Several of the reactions overlap in P-T space, for towards more Fe-rich compositions. Above reaction 3, examplereactions 2a and 2b (see Fig. 4). Higher cordierite and biotite that pass through reaction 4a will pressures would favourreaction 2a, while lower become more Fe-rich, while those in reaction 4b will pressures would favour reaction 2b. However, because become more Mg-rich. Thepattern will continue the P-T path in the aureole is assumed to be isobaric, upgrade. Specimens that have been collected from all bulkcomposition must control which of these two of the divariantzones show Fe-Mg compositional reactions is entered as one proceedsupgrade from trendsthat confirm these idealized progressions; zone 111. Using the assumption of uniform pressure, a further details will be given elsewhere.

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid for thermal aureoles 19

,/ ON As A Bi, / V Kf ’ Cd P

\ \ \ \ \ \M \ 0 \ 0 0 \ / \ /’ / \ / /I / /

\V \ \ \

FIG. 6. Illustration of the effect on reaction 3 (Mu + Q = As + Kf + V) if muscovite is considered to contain a celadonitic component. If muscovite is not pure KA13Si3010(OH)2, then reaction 3 is no longer degenerate; the Fe-Mg released by the muscovite breakdown will be taken up by a ferro-magnesian phase (here, cordierite or biotite), producing narrow divariant fields. Aseven-phase, KFMASH univariantcurve will join the two end-members’ invariant points. These curves have been drawn assuming (Mg/Fe)Cd > (Mg/Fe)Mu > (Mg/Fe)Bi, consistent with most of theprobe data from Ballachulish. In P-T space,these reactions will occur slightly up-temperature of the pure-muscovite degenerate reaction. This is because muscovite, on the low-T side of the reaction, has its activity reduced by the addition of a small celadonitic component, which displaces the curve up-T. (A similar,but opposing effect, is the distribution of Na betweenK-feldspar and muscovite. These opposing tendencies further justify the simplification that muscovite is pure KA12Si3010(OH)2.)

Apartfrom chloritein reaction 1, the only two tion of pelites in thermal aureoles into characteristic ferromagnesianminerals involved in the reaction massive hornfelses. sequence are cordierite and biotite. These participate in continuous reactions going upgrade that essentially drive their FeMg ratios back and forth. High-grade assemblages from Another conspicuous feature of this reactionse- Ballachulish and other quence is the accumulative production of vast amounts Scottish thermal aureoles of K-feldspar. Every reaction, apart from reactions 1 and 2a, producesK-feldspar, essentially as the K- Assemblages from zone V in quartz-bearing rocks and bearing product of micas breaking down, The replace- M, in quartz-absent rocks persist into the region where ment of mica by K-feldspar,coupled with the partial melting is inferred to have occurred. accumulation of cordierite as the main Fe-Mg phase, Further assemblages are sparsely developed close to are the principal causes of the progressive transforma- the igneous contacts, and in xenoliths and enclaves in

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 20 D. Pattison & B. Harte

APPROXIMATEMELTING BOUNDARY

FIG. 7. Schematic T-XF,-M, diagram for the Ballachulish aureole along the isobaric trajectory shown in Fig. 4. Dashed lines represent quartz-absent reactions. Reaction 1 is drawn with chlorite having an MgEe ratio between cardierite and biotite, and on the Fe-side of the cordierite-biotite join in an AFM diagram (see footnotet, Table3). Pelites with Fe-Mg bulk composition (A) encounter reactions 1, 2a and 3. Quartz-bearing rocks enter reaction 4a; quartz-absent rocks enter reaction 4b, pass outof this zone (with the destructionof muscovite), pass over reaction5 and enter reaction 6. Pelites of bulk composition (b) encounter reactions 1, 2b and 3. Above reaction 3, the pelites go through the same sequence as bulk composition (A). Pelites of bulk composition (C) go through the same sequence up to reaction 3 asbulk composition (B). Abovereaction 3, quartz-bearingrocks enter reaction 4a; quartz-absent rocks encounter reactions 4b, 5 and 6.

the igneouscomplex (see Fig. 1). Their limited In five locations (Sp localities in Fig. l), dark green occurrence makesit difficult tomap out zones like pleonaste spine1 appears in the assemblage Bi-As- those of the lower grade assemblages. Cor-Cd-Kf-Sp-inferred melt as 0.2-0.5 mm irregular Inan enclavein the S of the complex (Gt, Hy aggregates, usually growing incordierite and associ- locality in Fig. l), high-grade garnet and garnet- ated with corundum. hypersthenehave been found in quartz- Inorder to put these high-grade assemblages in biotite-cordierite-K-feldspar assemblages which are better context, high-grade mineral assemblages from inferred to have coexisted withmelt. The garnet threeother well-described Scottish thermal aureoles appears as 1-3 mm red crystals with rectilinear have been examined. These include the aureole of the parting. In thinsection, thegarnet is euhedral, Carn Choisquartz-diorite (Comrie aureole),Perth- unaltered, and contains minor inclusions of cordierite, shire, described by Tilley (1924) in one of the classical biotite and quartz. Hypersthene has beenfound in one papers on hornfelses in thermal aureoles, the Belhel- specimen, occurring as strongly pleochroic, 0.5-2 mm vie aureole,Aberdeenshire, described by Stewart euhedral grains, coexisting with the above minerals. (1946) andDroop & Charnley(1985), and the

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid for thermal aureoles 21

1 assemblage:Quartz-bearing I MutCdeBitAst K1 1 Bi+AszCortCdtKf I

Highestgrade Quartz-absentassemblage Bi-As-Cor-Cd-Sp-Kf -melt L-BALLACHULISH :-- l I l FIG. 8. High-gradeassemblages from Ballachulish. All projections areAFS (see Fig.2), with stippled bulk composition ellipses the same as those in Fig. 3. A melt phase is assumed to be present with assemblages above the muscovite-consuming reactions.

Lochnagar aureole,Aberdeenshire, described by the four aureoles is listed in Table 4, while in Fig. 8, Chinner (1962). the progression of assemblages found at Ballachulish Thesame collection of minerals (as distinct from above reaction 3 (breakdown of Mu + Q) is illustrated mineral assemblages) found at Ballachulish are found using the AFS projection. at Lochnagar, while at Comriegarnet is absent. At Belhelvie a few specimenscontain the additional Schematic petrogenetic grid for high-grade phaseanthophyllite, but these assemblages lack K- assemblages feldsparand arenot considered here. Typically, biotite,cordierite and K-feldspar are presentin all A schematic petrogenetic grid that encompasses all other assemblages (in addition toan inferred melt reported assemblages from the above four aureoles phase). The complete set of assemblages from each of has been constructed in Figs 9 and 10. The minerals

TABLE4: List of high-grade assemblages from the Comrie, Ballachulish, Lochnager and Belhelvie aureoles Comrie Ballachulish Lochnagar Belhelvie AS-Q 4a As-Q 4a Sp-Cor 7 Sp-Cor 7 As-Cor 6 As-Cor 6 SpAsCor 8 SpAsCor 8 Sp-CorSp-Cor 7 7 As-Sp 9 As-Sp 9 Sp-As-Cor 8 Sp-As-Cor 8 Gt-Q 11 Gt-AS-Q10 As-Sp 9 Gt-Q-HyHy-Q 12 13 Gt-Q 11 13 Gt-Hy Hy-Q 13 16 Gt-Q-Hy 12 Hy-Sp 18 HY-SP 18 Gt-AS-Sp 15 Gt-Sp-Hy 19 Gt-Sp 17 Hy-Sp 18 Hy-Sp Gt-Sp-Hy 19 All assemblages contain biotite, cordierite, K-feldspar and an inferred melt phase. Numbers refer to the reaction fields in Fig. 10.

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 22 D.Pattison & B. Harte involved include quartz,corundum, spinel,hyper- stableinvariant points and eight univariant curves sthene,garnet, alumino-silicate, cordierite,biotite, could exist. K-feldspar and an inferredmelt phase. This represents Aschematic Schreinemakers’ net was constructed 10 phasesin the six-component KFMASH system. relating the five KFASH(or KMASH)invariant Mineral pairs that do not coexist under relatively low points (Fig. 9).The net is simplified to show only pressure conditions include Cor-Q, Hy-As, Hy-Cor thosereactions in which biotite, cordierite and and Gt-Cor. K-feldspar arepresent, since these minerals are The association Sp + Q does not appear to bein present in all the assemblages from Ballachulish and equilibrium in the aureole assemblages, for although Lochnagar, and the vast majority of assemblages from Sp and Q may occur in the same rock they are never in Comrie and Belhelvie. The linking reactions(uni- contact. The stability of the hercynite (FeAI,O,) variantin KMASHKFASH,or divariant in spinel end-member with quartz is well documented KFMASH)were orientated from available ex- (e.g.Holdaway & Lee 1977), whilst the magnesian perimental data (see Fig. 11 and the next section) and (MgAI2O4) end-member does not appear to be stable from the following constraints based on molar entropy with quartz. Verylittle compositional or assemblage and volume relations: biotite is likely to be on the low data is available on the limits of stability of spinel solid temperature side of the reaction;cordierite onthe solution, (Fe, Mg)A1204,with quartz. To avoid undue low-pressure side; garnet on the high P-T side. speculation and resultant complication we have there- The orientationof the net satisfies the relative grade fore excluded the co-existence of SpQin constructing requirements at Ballachulish that amongst these high Figs 9 and 10, because this seems to be thecase for the grade assemblages, the relatively lowest-grade quartz- bulkcompositions and P-T range considered in this bearing assemblage is Cd-Bi-Kf-As-Q-(melt) and, paper. To combine our Schreinemakers’ net with the quart-absent assemblage is Cd-Bi-Kf-Cor-As- higher-pressure net of Hensen & Green (1973) (with (melt). Sp-Q stability), it would be necessary to evaluate The end-member KFASH (or KMASH)grid of Fig. Sp-Q stability more carefully, as well as taking into 9 is expanded in Fig. 10 to the KFMASH system, so account the stability of Hy + As at higher pressures. that invariantpoints in Fig. 9 becomes univariant The above restrictions result in no stable KFMASH curves and univariant curves become divariant fields. invariant points and only five stable univariant curves Plotting all of the recorded assemblages from the four in KFMASH (invariantpoints in KMASH or aureoles on Fig. 10 provides a comprehensive repre- KFASH)--seeTable 5. If inFe-rich compositions, sentation of most of the di- and univariant fields hercynite-quartz is considered to be stable, then two established in the grid. The two lowest-grade divariant

t T

FIG. 9. Schematic KMASH or KFASH Schreinemakers’net for high gradeassemblages that contain biotite, cordierite, K-feldspar and inferred melt (refer to Table 5). The rationale for the construction and orientation of the net is discussed in the text. Reaction numbers are the same as those used elsewhere in the paper.

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid for thermal aureoles 23

Belhelvie

Ballachulish Lochnagar

Comrle

W I FIG. 10. Schematic KFMASH reaction net for high-grade assemblages in the Ballachulish, Comrie, Lochnagar and Belhelvieaureoles that containbiotite, cordierite, K-feldspar and melt (not shown).This diagram is essentially expanded from the end-member net of Fig. 9, such that invariant points become univariant curves, and univariant curves become divariant fields. For divariant reactions, the Mg-bounding curves are labelled. The pressures shown for the different aureoles in relation to the reactions are only relative and are positioned to illustrate most clearly how pressure may be the cause of differences between the aureoles. To further this objective and in contrast to Fig. 4, this diagram is drawn with Fe- and Mg-bounding reactions that reflect the common range of pelitic compositions, and are therefore not strict KFASH and KMASH end-member bounding reactions. This allows the inferred pressure differences between the aureoles, as manifested by their different mineral assemblages, to be moreclearly shown (see text for further discussion). Assemblages from each of the aureoles are plotted on the diagram (refer to Table 4). Only reactions 4a and 10 have been experimentally examined (Holdaway & Lee 1977) and this, coupled with the uncertain effects of melt and varying water activity, means that orientation of the curves may vary quite broadly. Univariant curves have been arbitrarily drawn parallel to each other.

fields from the grid in Fig. 10 (reactions 4a and 6) are the four. The lack of any garnet in the assemblages the two highest-grade reactions from the grid of Fig. from Comrie suggests that it is at the lowest pressure &this provides the link between the two grids, and of the four, while the assemblages from Ballachulish consequently establishes a complete prograde reaction and Lochnagar,although they vary, do not in net for the Ballachulish aureole. themselves delineate any pressure difference between A number of differencesemerge between the the two. assemblages fromeach of the fouraureoles. The While it is possible that all of the assemblage presence of coexisting garnet and alumino-silicate is variation between the aureoles is strictly a product of restricted to Belhelvie, which as illustrated in the grid FeMg bulk compositional variation, it seems unlikely. in Fig. 10, suggests that it is at the highest pressure of If this were the case, then it would follow from Fig. 10

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 24 D. Pattison & B. Harte

TABLE5: Summary of univariant reactions and invariant points for high grade assemblages

-~ ~ PHASES: Cd Q Bi As Kf SpCor Hy Gt (melt) (10) COMPONENTS:KFMASH (6) EXCLUSIONS:Q-Cor,Hy-As, Hy-Cor, Gt-Cor, Sp-Q* OTHER RESTRICTIONS: Bi-Cd-Kf-Melt present in all assemblages INVARIANTPOINTS: 0 UNIVARIANT CURVES (invariant pts in KFASH or KMASH): 5 Name Phases present [Q-Hy-Gt] As-Cor-Sp[Q-Hy-Gt] (8) [Sp-Cor-Hy] As-Q-Gt [Sp-Cor-Hy] (10) [Sp-Cor-As]Hy-Q-Gt + Bi-Kf-Cd-melt (12) [Q-Cor-Hy] Gt-As-Sp [Q-Cor-Hy] (15) [Q-Cor-As] Gt-Hy-Sp (19) * If, in Fe-rich compositions, SpQI are assumed to co-exist, then: INVARIANTPOINTS: 2 [Cor-Hy] [Cor-As] UNIVARIANTCURVES: 8 [Q-Hy-Gt] [Cor-Hy-As] [SpCor-Hy] [Cor-As-Gt] [SpCor-As] [Cor-Hy-GtJ [Q-Cor-Hy] [Q-Cor-As]

~~ ~~ KFMASH univariant curves (invariant pts in KMASH or KFASH) are designated by phases absent and are numbered as in Figs 9 and 10. that all reported assemblages from Comrie, none of difficult mineral to handle thermodynamically because which contain garnet, are more Mg-rich than those of of its variable water content (Newton & Wood 1979). Ballachulish andLochnagar, which in turn must be A self-consistent thermodynamicapproach to accur- more Mg-rich than all Belhelvie assemblages which ately orientate the grid in P-T space, using existing contain coexisting garnet,cordierite, sillimanite and thermodynamic data for cordierite, results in reaction quartz. topologies that contradict existing experimentaland There is no evidence of such consistent geochemical field data(e.g. Droop & Treloar 1981). Small variation betweenthe peliticrocks thatenter the deviations in the water content of cordierite (in different aureoles. Three of the aureoles (Comrie, response to some or all of PtotalrPHZO and T) affect Belhelvie and Lochnagar) are developed in Middle both its thermochemical properties and the amount of and Upper Dalradian stratigraphicunits which have fluid released or consumed in cordierite-bearing similar and not widely varying MgO: Fe0 ratios reactions, which together have significant effects on (Atherton & Brotherton 1982). FeMg ratios of garnet the position and slope of these reactions (e.g. reaction and cordierite coexisting with sillimanite at Belhelvie 2a). (Droop,unpubl. data) are not significantly different In spite of the theoretical difficulties associated with from those of coexisting garnet and cordierite (without cordierite, some consistency hasbeen established sillirnanite) at Ballachulish. Furthermore, this pattern experimentallybetween Mg end-member cordierite- is consistent with independent geobarometry deter- bearing reactions. Figure 11 shows experimentally minations of4.5-6 kbar forBelhelvie (Droop & determined reaction positions that are relevant to the Charnley 1985; Ashworth & Chinner 1978), 4.2 kbar grid. for Lochnagar(Ashworth & Chinner 1978) and Reaction 3, thedegenerate univariantcurve to tentative 2.5-4 kbar for Ballachulish (based on some which reactions 2a, 2b, 4a and 4b connect, has been preliminary geobarometry not reported in this paper); calibrated by Chatterjee & Johannes (1974). Hold- no pressures have yet been estimated for Comrie. It away & Lee (1977) conductedexperiments on the would also appearthat the Ballachulish aureole bounding Feend-member reaction of continuous reached lower peak temperatures than the other three reaction 4a: aureoles. Sill + Fe-Bi + Q = Fe-Cd + Kf + V. P-T calibration of the grid P-XFeMg experiments on this reaction,at a fixed temperature, allowed them to calculate the position of Reliable calibration of the gird in P-T-XFe-Mg space is the Mg end-memberreaction. The intersection of a complicated task for a number of reasons. Cordier- these end-member reactions with reaction 3 estab- ite, one of the principal ferromagnesian minerals that lishes the positions of theend-member [Cor, Chl] occurs throughout the entire reaction sequence, is a invariant point (terminology as in Fig. 5).

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid for thermal aureoles 25

FIG. 11.Selected calculated and experimentally determined reactions. Solid lines,experimentally determined reactions;dashed lines; calculatedreactions. Alumino-silicate data:R, Richardson ef al. (1969); H, Holdaway (1971). Reaction 1, Seifert (1970); 2a, Seifert (1970); 2b, Seifert (1976); 3, Chatterjee & Johannes (1974); 4aM, 4aF, Holdaway & Lee (1977); 5, Chatterjee & Johannes (1974); 10, Holdaway & Lee (1977); 13, thermodynamic calculation, using thermochemical data from Helgeson et al. (1978) and fluid fugacities from Kerrick & Jacobs (1981); 20, Richardson (1968), Holdaway & Lee (1977); 21, Tuttle & Bowen (1958); 22, Seifert (1976); 23, Shaw (1963). The reaction positions are ‘preferred fits’ of the authors (op. cif.)through their experimental brackets. Reactions 2a and 4aM, which are drawn to intersect with reaction 3 at the KMASH invariant point [Cor, Chl], are in slightly different positions than in the cited papers, but are still within the experimental brackets.

Seifert (1970) and Bird & Fawcett (1973) indepen- same [Cor, Chl] invariant point, neither substantiate dentlyexperimentally bracketed the position of the norcontradict the position of the invariantpoint Mg end-member reaction of continuous reaction 2a: because, as Seifert noted,the slope and position of this reaction and reaction (3) are very similar above Mu + Mg-Cd = Mg-Bio + Sill + Q + V. 3 kbar. Seifert’s ‘best-estimate’ reaction position passes near From the general agreement between these various the intersection of reaction3 andthe 4aMg experiments, the Mg end-member [Cor, Chl] invariant end-member reaction (within the experimental brack- point lies roughly at 5.3 kbar and 700°C. The intersec- ets, it can be drawn to pass through it). tion of theChatterjee & Johannes (1974) dataon Experiments by Seifert (1976) onthe Mg end- reaction 3 and the Holdaway & Lee (1977) data on member reaction 2b, reaction 4a place the Fe end-member [Cor, Chl] point roughly at 1.5 kbar, 580°C. Mu + Mg-Bio + Q = Mg-Cd + Kf + V, The calibration of reaction 20 (Fe-Cd = Alm + Sill + Bio) in Fig. 11 is particularly relevant tothe another of the reactions that is generated from the high-grade reaction grid. In Figs 9 and 10 this

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 26 D. Pattison & B. Harte degenerate curveintersects with reaction 4a at participation of these phases will promote melting at KFASH invariantpoint [Hy,Cor, Sp], which ex- lower temperatures (Seifert 1976). Although in theory pandedinto KFMASH becomesunivariant reaction there will be a whole family of melting curves, each 10. Because virtually all of the assemblages in the four particular to adifferent mineral assemblage (fora aureoles containbiotite and K-feldspar,reaction 20 detailed discussion of this subject, seeThompson has not been included in the grid, although it provides (1982)), a good experimental approximation to these a useful constraint on the P-T orientation and position curves is the Mg-Cd + Kf + Q + V e L melting of the grid. curve (reaction 22) of Seifert (1976). These experimental studies suggest that the stable Further discussion of the details of the partial biotitecomposition was close tothe eastonite- melting at Ballachulish, which are presently being siderophyllitejoin (asopposed to the phlogopite- investigated, is beyond the scope of this paper. annitejoin), even when, in the Mg system, One of the striking features of pelitic reaction stoichiometric phlogopite was synthesized and inserted sequences in thermal aureoles, in contrast to regional asstarting material (Seifert 1970, 1976)-thisis the terranes, is theabundance of quartz-absent assemb- reason for using “g-Bio’ rather than ‘Phl’ in the lages, particularly before the onset of partial melting. written reactions. Because of the divergence of dehydration-curves and Dotted lines in Fig. 11 are the AI-silicate polymor- melting curves (seeFig. ll), pelites in low pressure phictransitions of Richardson et al. (1969) and aureoles may pass through reactions such as the Holdaway (1971). The relative merits and petrological muscovite-quartz breakdown (reaction 3) before melt- implications of these experiments, inparticular the ing occurs, and this allows thedevelopment of andalusiteto sillimanite transition, havebeen the silica-undersaturated assemblages which may contain focus of sustained debate ever since their publication such minerals as corundum and spine]. At higher (see, for example, Anderson et al. 1977; Greenwood pressure, meltingcommences before or with the 1972; Day & Kumin 1980; Harte & Hudson 1979). muscovite-quartz breakdown reaction, so that musco- The A g S transition, which involves the two poly- vite is involved in melting reactionsand thus never morphs present in the Ballachulish aureole, is char- reacts to produce corundum. acterized by particularly small changes in AS and AV, and is thus susceptible to kineticproblems. Inthe Conclusions Ballachulish aureole,andalusite alone is present in divariantreaction 2a,but in all assemblages onor The prograde sequence of pelitic mineral assemblages above reaction 3 and well intothe melting zone, from theaureole of the Ballachulish ‘Granite’ has fibrolitic sillimanite coexists with andalusite. In beendescribed. Combining this information with volume, andalusite is more abundant, and texturally it assemblage descriptionsfrom other Scottish thermal appears to be the phase in equilibrium with the rest of aureoles, two schematic petrogenetic grids have been the assemblage (e.g. intergrown with K-feldspar above constructed: one for assemblages essentially below the reaction 3, intergrown with biotite and K-feldspar in onset of partial melting (Figs 4 and 5) and another for and above reaction 4). Sillimanite occurs irregularly, high-grade assemblages (Figs 9 and 10). By combin- aslong prisms and patchesin cordierite, fibrous ing these two nets, a complete prograde reaction grid ‘sprays’ around accessory minerals, and as small is established, and its approximate P-T calibration is prisms in K-feldspar-biotite zones at high grade. Only given in Fig. 11. in the very highest grade specimensdoes sillimanite The KFMASH net in Fig. 10 accounts for virtually occur without andalusite. The textural ambiguity and all of the high-grade assemblages found in four thebroad zone of overlap shows thatthe simple separate aureoles developed in different pelitic units at coexistence of the two polymorphs in an assemblage is differentpressures, which suggests that it should be of minimal significance in delineating its P-T condi- useful for many thermal aureoles. Systematic examina- tions of formation. tion of low-grade assemblages in aureoles other than Three meltingcurves are includedin Fig. 11. The Ballachulish and Comrie is lacking, so thatthe ‘minimum granitic melt’ curve(reaction 21) is of KFMASH reaction net in Fig. 4 is not as broadly volumetrically minor importance in this aureole be- based. With further information on low grade rocks cause plagioclase is minor in abundance in the pelites, from otheraureoles, however, it should also be and therefore the amountof melting that occurs at this possible to expand Fig. 4 to takeaccount of all the reaction will be small and difficult to identify. lower temperatureand higherpressure reactions of Specimens from Ballachulish that show partial melt Fig. 5. segregations are strongly dominated by K-feldspar, suggesting that a more relevant melting curve, in the ACKNOWLEDGEMENTS. D.R.M.P.acknowledges the Associa- simplified KASH system, is Kf + Q + V L (reaction tion of Commonwealth Universities for funding his PhD at 23). Other phases, such as cordierite, alumino-silicate Edinburgh through a Commonwealth Scholarship. We thank and biotite are also included in these segregations; the Neil Hudson for a careful and helpful review of the paper,

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 A petrogenetic grid foraureoles thermal 27 and Roger Powell for his review of the paper and previous D.R.M.P.thanks Stefan Weiss, Munich, for stimulating discussions with D.R.M.P. duringdevelopment of the work. discussion andthe exchange of informationon our two We would also like tothank F. Seifert,head of a projects,and Beth,Alan and Russell Bolton,and the Logan multidisciplinary Deutsche Forschungsgemeinschaft group family in Duror forputting up with him in the field over two studying variousaspects of the Ballachulish complex,for summers. Finally, we thank Heather Hooker and DianaBaty. inviting us toBonn formeetinga in December 1982. References ALBEE, A. L. 1965. Apetrogenetic grid for the Fe-Mg BIRD, D. K. 1978. Summary and critique of the silicates of pelitic schists. Am. J. Sci. 263, 512-36. thermodynamic properties of rock formingminerals. ANDERSON,P. A. M., NEWTON,R. C. & KLEPPA,0. J. 1977. Am. J. Sci. 278A, 1-229. The enthalpy change of the andalusite-sillimanite reac- HENSEN,B. J. 1971. Theoreticalphase relations involving tion and the AlzSi05 diagram. Am. J. Sci. 277, 585-93. cordierite and garnet in the system MgO-Fe@A120,- ASHWORTH,J. R. & CHINNER, G. A. 1978. Coexisting garnet SO2. Contrib. Mineral. Petrol. 33, 191-214. and cordierite in migmatites from the Scottish Caledo- HENSEN,B. J. & GREEN, D. H.1973. Experimental study of nides. Contrib. Mineral. Petrol. 65, 379-94. the stability of cordierite and garnet in pelitic composi- ATHERTON,M. P. & BROTHERTON,M. S. 1982. Major tions at high pressures and temperature. 111. Synthesis of element composition of the pelites of the Scottish experimental data and geological applications. Contrib. Dalradian. Geol. J. 17, 185-221. Mineral. Petrol. 38, 151-66. BAILEY,E. B. & MAUFE,H. B. 1960. The geology of Ben HOLDAWAY,M. J. 1971. Stability of andalusite and the Nevis and Glen Coe. (Explanation of sheet 53). Mem. aluminum-silicate phasediagram. Am. J. Sci. 271, Geol. Surv. U.K. (Scotland). 97-131. BIRD, G. W. & FAWCETT,J. J. 1973. Stability relations of - & LEE, S. M. 1977. Fe-Mg cordierite stability in Mg-chlorite, muscovite and quartz between 5 and 10 kb high-grade pelitic rocks based on experimental, theore- water pressure. J. Petrol. 14, 415-28. tical and natural observations. Contrib. Mineral. Petrol. BOWES,D. R.& WRIGHT,A. E. 1967. The explosion-breccia 63, 175-98. pipes nearKentallen, Scotland, and their geological KERRICK, D. M. & JACOBS,G. K. 1981. A modified setting. Tram. R. Soc. Edinburgh, 67, 109-43. Redlich-Kwong equation for H20, CO2 and H20-C02 BROWN,P. E., MILLER, J. A.& GASTY,R. L. 1965. Isotopic mixtures at elevated pressures and temperatures. Am. J. ages of late Caledonian granitic intrusions of the British Sci. 276, 883-916. Isles. Proc. Yorks. geol. Soc. 36, 251-76. MATHER,J. D. 1970. The biotite isograd and the lower CARMICHAEL,D. M,, MOORE,J. M. & SLAPPEN,G. B. 1978. greenschist facies in the Dalradian rocks of Scotland. J. Isogradsaround the Hastings metamorphic 'low'. In: Petrol. 11, 253-75. CURRIE, A. L. & MACKASEY,W. 0. (eds) Toronto '78 NEWTON,R. C. & WOOD, B. J. 1979. Thermodynamics of Field Trips Guidebook, Geol. Assoc. Can. 325-46. water in cordierite and some petrologic consequences of CHATTERJEE,N. D. & JOHANNES,W. 1974. Thermal stability cordierite as a hydrous phase. Contrib. Mineral. Petrol. andstandard thermodynamicproperties of synthetic 68, 391-405. 2M1-muscovite, KA12[AISi3010(OH)Z].Contrib. Mineral. OHMOTO,H. & KERRICK,D. 1977. Devolatilization equilib- Petrol. 48, 89-114. ria in graphitic systems. Am. J. Sci. 277, 1013-44. CHINNER,G. A. 1962. Almandine in thermal aureoles. J. POWELL,R. 1983. Fluids and melting under upper amphibo- Petrol. 3, 310-40. lite facies conditions. J. geol. Soc. London, 140, 629-34. DAY, H. W. & KUMIN,H. J. 1980. Thermodynamic analysis READ, H.H. 1961. Aspects of Caledonian Magmatism in of the aluminum silicate triple point. Am. J. Sci. 280, Britain. Liverpool Manchester geol. J. 2, 653-83. 265-87. RICHARDSON,S. W. 1968. Staurolite stability in a part of the DROOP, G.T. R. & CHARNLEY,N. 1985. Comparative system Fe-AI-Si-0-H. J. Petrol. 5, 467-88. geobarometry of pelitic hornfelses associated with the -, GILBERT,M. C. & BELL, P. M. 1969. Experimental newer gabbros. J. Keol. Soc. London, 142, 53-62. determination of kyanite-andalusite and andalusite- -& TRELOAR, P. J.1981. Pressures of metamorphism in sillimanite equilibria: the aluminum silicate triple point. the thermal aureole of the Etive Granite Complex. Scott. Am. J. Sci. 267, 259-272. J. Geol. 17, 85-102. SEIFERT,F. 1970. Low-temperature compatibility relations of GREENWOOD,H. J. 1972. Aliv-Siiv disorderin sillimanite and cordierite in haplopelites of the system K20-MgO- its effecton phase relations of thealuminum silicate A1203-Si02-H20. J. Petrol. 11, 73-99.

minerals. Mem. geol. Soc. Am. 132, 553-71. ~ 1976. Stability of the assemblage cordierite + K- GUIDOTTI,C. V., CHENEY, J.T. & CONATORE,P. D. 1975. feldspar + quartz. Contrib. Mineral. Petrol. 57, 179-85. Coexisting cordierite + biotite + chlorite from the Rum- SHAW,H. R. 1963. The four-phasecurve sanidine-quartz- ford quadrangle, Maine. Geology, 3, 147-8. liquid-gas between 500 and 600 bars. Am. Mineral. HARTE,B. & HUDSON,N. C. F. 1979. Pelite facies series and 48, 883-96. thetemperatures and pressures of Dalradian meta- STEWART,F. H. 1946. The gabbroic complex of Belhelvie in morphism in E. Scotland. In: HARRIS,A. L., HOLLAND, Aberdeenshire. Q. J. geol. Soc. London, 102, 465-98. C.H. & LEAKE, B. E. (ed) The Caledonides of the STRECKEISEN,A. L. 1967. Classification and nomenclature of BritishIsles-Reviewed. Spec. Publ. geol. Soc. London, igneous rocks. NeuesJahrb. Mineral. Abhandlungen, 8, 323-37. 107, 144-214. HELGESON,H. C., DELANEY, J. M,, NESBITT,H. W. & THOMPSON,A. B. 1982. Dehydration melting of pelitic rocks

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021 28 D.Pattison & B. Harte

andthe generation of H20-undersaturated granitic li- TREAGUS,J. E. 1974. A structural cross-section of the Moine quids. Am. J. Sci. 282, 1567-95. and Dalradian rocks of the Kinlochleven area, Scotland. THOMPSON,J. B., JR. 1957. The graphical analysis of mineral J. geol. Soc. London, 130, 525-44. assemblages in pelitic schists. Am. Mineral. 42, 842-58. TUITLE, 0. F. & BOWEN,N. L. 1958. Origin of granite in TILLEY,C. E. 1924. Contact metamorphism in theComrie light of experimentalstudies. Mem.geol. Soc. Am. 74, area of thePerthshire Highlands. Q. J. geol.Soc. 153 pp. London, 80,22-71.

Received 21 January 1984; revised typescript accepted 5 Jum 1984.

Downloaded from http://pubs.geoscienceworld.org/jgs/article-pdf/142/1/7/4888516/gsjgs.142.1.0007.pdf by guest on 27 September 2021