Disequilibrium in the Ross of Mull Contact Metamorphic Aureole

Disequilibrium in the Ross of Mull ContactMetamorphic Aureole, Scotland: aConsequence of Polymetamorphism

J. WHEELER*, L. S. MANGANy AND D. J. PRIOR

DEPARTMENT OF EARTH AND OCEAN SCIENCES, LIVERPOOL UNIVERSITY, LIVERPOOL L69 3GP, UK

RECEIVED JUNE 12, 2000; ACCEPTED SEPTEMBER 8, 2003

The Ross of Mull pluton consists of granites and granodiorites and

intrudes sediments previously metamorphosed at amphibolite facies.

The high grade and coarse grain size of the protolith is responsible for a

high degree of disequilibrium in many parts of the aureole and for some

unusual textures. A band of metapelite contained coarse garnet, biotite

and kyanite prior to intrusion, and developed a sequence of textures

towards the pluton. In Zone I, garnet is rimmed by cordierite and

new biotite. In Zone II, coarse kyanite grains are partly replaced by

andalusite, indicating incomplete reaction. Coronas of cordierite þmuscovite around kyanite are due to reaction with biotite. In the higher-

grade parts of this zone there is complete replacement of kyanite and/or

andalusite by muscovite and cordierite. Cordierite chemistry indicates

that in Zone II the stable AFM assemblage (not attained) would

have been cordierite þ biotite þ muscovite, without andalusite. The

observed andalusite is therefore metastable. Garnet is unstable in

Zone II, with regional garnets breaking down to cordierite, new biotite

and plagioclase. In Zone III this breakdown is well advanced, and

this zone marks the appearance of fibrolite and K-feldspar in the

groundmass as a result of muscovite breakdown. Zone IV shows garnet

with cordierite, biotite, sillimanite, K-feldspar and quartz. Some

garnets are armoured by cordierite and are inferred to be relics. Others

are euhedral with Mn-rich cores. For these, the reaction biotite þsillimaniteþ quartz! garnetþ cordieriteþ K-feldsparþ melt is

inferred. Using a petrogenetic grid based on the work of Pattison

and Harte, pressure is estimated at 3�2 kbar, and temperature at theZone II---III boundary at 650�C and in Zone IV as at least 750�C.

KEY WORDS: contact metamorphism; disequilibrium

INTRODUCTION

Contact metamorphic aureoles provide the best-constrained settings in which to interpret metamorphic

textures and assemblages (Kerrick, 1991). In principle,knowledge of the shape and size of the intrusion,together with the temperatures of the magma (at thetime of intrusion) and the surroundings, allows predic-tion of the T---t history at any point. In practice, thereare many assumptions involved in such predictions, butthese can be tested and refined with reference to aureoleassemblages. Contact metamorphic aureoles commonlyshow evidence for disequilibrium; even in inner zones,where high temperatures are attained and reaction ratesare expected to be fast, the timescale for heating andcooling is shorter than that for regional metamorphism.For instance, the Christmas Mountains aureole ( Joesten,1983) affects limestones with chert nodules; thesenodules are rimmed by contact metamorphic assem-blages, but did not react completely. Other aureoles,such as Ballachulish in Scotland (Voll et al., 1991),appear to have pelites equilibrated at the peak tempera-ture attained in each zone. Here, the initial fine grainsize of the regionally metamorphosed chlorite slates mayhave aided fast reaction. In quartzites in the same aur-eole there is evidence for disequilibrium during partialmelting (Holness & Clemens, 1999), highlighting theimportance of water supply if input of H2O is requiredfor reaction. In general, dry coarse-grained rocks are lesslikely to equilibrate and therefore carry more informa-tion on the kinetics of reactions during contactmetamorphism. In this paper, we describe how coarse-grained amphibolite-facies pelites responded to contactmetamorphism in the aureole of the Ross of Mull pluton,Scotland. We show how disequilibrium is prevalent inthe low-grade zones, and is marked even at the highestgrades.

JOURNAL OF PETROLOGY VOLUME 45 NUMBER 4 PAGES 835–853 2004 DOI: 10.1093/petrology/egg113

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

yPresent address: Pilkington Technology Center, Hall Lane, Lathom

L40 5UF, UK.

Journal of Petrology 45(4) # Oxford University Press 2004; all rights

reserved

Dow

nloaded from https://academ

ic.oup.com/petrology/article-abstract/45/4/835/1529671 by guest on 04 April 2019

REGIONAL SETTING OF THE ROSS

OF MULL PLUTON

TheCaledonides in Scotland include a large area of region-ally metamorphosed rocks: the Moinian Supergroup,deposited in the Precambrian (Harris, 1995). It was region-ally metamorphosed prior to closure of the Iapetus suture,at c. 500Ma, although the timing of that event remainscontroversial. The Moinian Supergroup also appears tohave a Precambrian metamorphic history (Harris, 1995).The protoliths were dominated by clastic rocks, mainlyarkoses and sandstones with subordinate pelites and thinlimestones, but also included sheet-like bodies of basicrocks. Most of these rocks are now at amphibolite facies,the more psammitic protoliths giving rise to biotite schistsand gneisses, and the pelites containing more variedmineral assemblages. After the closure of the IapetusOcean, a suite of late-orogenic granites and intermediateplutonic rocks was intruded into these already deformedandmetamorphosed sediments. The Ross ofMull pluton isone such, intruded at 414� 3Ma (Halliday et al., 1979) andranging frommonzogranite to granodiorite in composition(British Geological Survey, 1999).The island of Mull contains a small isolated area of

Moinian metasediments (Bosworth, 1910; Holdsworthet al., 1987). These are bounded on the NE by down-faulted Tertiary flood basalts, on the west by the Ross ofMull pluton, and to the north and south by sea (Fig. 1).The Ross of Mull pluton is itself bounded to the westmainly by sea, but small islands just off the east shore ofthe island of Iona are granite (Potts et al., 1995). Thenearby country rocks on Iona are psammites and pelitesof the Iona group, which are not correlated with theMoinian Supergroup-----faulting and shearing in theSound of Iona are postulated to account for the (pre-intrusion) juxtaposition of different units (Potts et al.,1995). Hornfelsing of the Iona rocks (British GeologicalSurvey, 1999) shows that they were affected by the Rossof Mull pluton. Within the Moinian inlier the dominantrock type is a biotite---quartz---feldspar schist or gneisssometimes with garnet, interpreted as meta-arkose,psammite or pelitic variants of these depending on abund-ances of feldspar and mica. Basic sheets (now garnetamphibolites), calc-silicate lenses and pelites (nowgneisses with some or all of garnet, biotite, kyanite andmuscovite) are subordinate. Upright folds in the inliertrend NE---SW. There are two main outcrops of peliticlithology: the main one is a band that trends NE---SWacross the middle of the inlier [within the ArdalanishStriped and Banded Formation of Holdsworth et al.(1987)]; the other occurs in the NW of the inlier adjacentto the pluton. The response of these metapelites to thecontact metamorphism is the focus of this study. Becausethe main pelite band is oblique to the roughly north---south-trending pluton contact, it exhibits diverse contact

metamorphic mineral assemblages that partly overprintthe regional metamorphic assemblages. The thermalmetamorphic assemblages were first described by Bos-worth (1910), Cunningham-Craig (1911) and MacKenzie(1949), but there are no more recent published studies.The aureole was mentioned briefly by Kerrick (1990, p.278) but it should be noted that the scale bar on his fig.9.12 was mistakenly labelled as ‘10 km’ whereas it is infact 1 km.The intrusion has an east---west outcrop width of 7 km

(British Geological Survey, 1999). Gravity modelling(Beckinsale & Obradovich, 1973) has been interpretedto show that it is an east-dipping slab with a true thicknessof 3�4 km, but other cross-sections suggest a more vertical-sided body with true width of the order of 7 km (BritishGeological Survey, 1999). There is little evidence of plas-tic deformation of the country rocks during intrusion:instead, rafts of country rock quartzite, dismemberedbut unrotated, and mappable ‘ghost stratigraphy’ withinthe granite suggest that the pluton attained its final posi-tion by stoping. Although dominated by granitoids, theintrusion is not all granite. Mafic enclaves in the westprovide evidence for magma mingling, and there aremore dioritic parts to the intrusion. It is possible, then,that the thermal aureole was due, in part, not to thegranite but to more mafic and therefore hotter magmas.In subsequent sections we first outline the variety of

regional metamorphic mineral assemblages found in themain pelite band at its NE end, where there is no visibleevidence of contact effects. We then document the newminerals and their textural relationships observed as thepluton is approached, and diagnose the main reactionsthat are giving rise to these. Subsequently we discussmineral chemistry in more detail for each of the zonesdefined to produce a more constrained picture of theP---T conditions, with reference to published petrogeneticgrids. Finally, this framework allows us to discuss theeffect of the protolith on reaction kinetics.

REGIONAL METAMORPHISM

In detail, the pelite band shows a variety of bulk-rockchemistries that give rise to different mineral assemblages(Mangan, 1996). Four regional metamorphic assem-blages have been identified (Table 1). Assemblage 1 isthe divariant Grt---Bt---Ky---Ms---Qtz assemblage inKFMASH [all abbreviations after Kretz (1983)] andassemblages 2, 3 and 4 are related trivariant assemblages,each with one phase from the above list absent, reflectingvariable bulk compositions. Assemblage 2 lacks musco-vite (so is poor in K), assemblage 3 lacks garnet (low Fe/Mg) and assemblage 4 lacks kyanite (poor in Al). Allassemblages contain plagioclase and rutile reflecting thenon-KFMASH components Na, Ca and Ti. Garnet istypically 2---5mm across, biotite 2mm and bladed

JOURNAL OF PETROLOGY VOLUME 45 NUMBER 4 APRIL 2004

836

Dow



kyanite porphyroblasts range up to 5 cm � 1 cm. Micasgenerally form a strong foliation. Significant amountsof tourmaline are found in places within the kyanitegneisses-----there is no textural evidence that suggests thatthis is related to contact metamorphism. Euhedral tour-malines are in places included within kyanite grains andtherefore the tourmaline is considered to have crystallizedduring regional metamorphism, possibly as a result ofmetasomatism. The mineral chemistry of the regionalassemblages provides a template for understanding theirmodification as the aureole is approached. The completedataset of analyses from 29 samples outside and inside theaureole may be downloaded from the Journal of Petrology

website at http://www.petrology.oupjournals.org/.

Table 2 shows selected analyses but only from inside theaureole. Figure 2 plots mineral compositions and tie-linesbetween minerals for several samples from outside theaureole. The absence of crossing tie-lines shows that theseregional assemblages are compatible at a single pressureand temperature.The regionally metamorphosed rocks show evidence

for disequilibrium: plagioclase varies slightly in composi-tion within individual rocks, and garnet is occasionallypartly replaced by retrogressive chlorite. Bearing this inmind, we have constrained the regional conditions usingFe---Mg exchange between garnet and biotite (Ferry &Spear, 1978), giving temperatures in the range610---700�C for a nominal pressure of 8 kbar. There is

38 39

Ardalanishbeach

Torr Mor

BendoranCottage

22

21

20

19

1836 37 38 39

23

LochAssapol

Ross of Mull Pluton

1 KM

N

DunFuinn

Bunessan

ISLEOFMULL

10 kmRoss of Mull

Iona

70411Kb

KcKf2

Ky61,62

Ky54,57

Kf1

Ky52

70413

Bc

117B, RM5001,Rm5002

Ct5

Zon

e I

Zon

e II

Zon

e II

I

Zone IV

PortUiskenbeach

KgBg,Bh

Bf7

Be

70414

Ky1-22

Ky46

Ky49,51Ky45

Ky39

Ky25,26

Ky23

Fig. 1. Geological setting of Ross of Mull pluton and approximate location of isograds, shown as black roughly north---south-trending lines. Thickgrey lines indicate kyanite-bearing bands that terminate at inferred faults shown by black dashed lines. Isograds can be defined only within thesebands, and then only approximately. Dots indicate sample localities: those labelled are referred to in this paper or the Electronic Appendix; otherswere discussed by Mangan (1996).

WHEELER et al. THE ROSS OF MULL PLUTON

837

Dow



http://www.petrology.oupjournals.org/

Table1:Keyfeaturesofregionalandcontactmetam

orphicassemblages

Ass.

Reg

ional

ZoneI

ZoneII

ZoneIII

ZoneIV

1GrtBtKyMs

Fine-grained

MsreplacingKy;

sporadic

Chl;e.g.Ky2,Ky5,

Ky16,Ky21,Ky25,Ky39,Ky46

2GrtBtKy

Fine-grained

MsreplacingKy;

sporadic

Chl;e.g.Ky13,Ky19,Ky45

Kyrimmed

byCrd

þMs;

e.g.Ky23

Grtrimmed

byCrd

þPlþ

Bt;

Kyrimmed

byCrd

þMs;

And

patch

esin

Ky;

Rtrimmed

byIlm

;

e.g.70411-10(Fig.3a),

70411-2

(Fig.3c),Kb

Grtpseudomorphed

byCrd

þPlþ

Btþ

Ilm;

Andrimmed

byCrd

þKfs

þBt;

e.g.70413-10(Fig.3e),70413-13(Fig.3f)

Inclusion-freeGrt;relict

Btco

ntainsSilneedles;

Silprism

s

inmatrix;

Andrimmed

byCrd;

3BtMsKy

e.g.Ky49,Ky51

Kyrimmed

byCrd

þMs;

Kyrimmed

byCrd

þMs;

Andpatch

esin

Ky;

e.g.Bf7

(Fig.3g

)Kfs

inmatrix;

Ilm;e.g.117B

(Fig.3h

),Ct5

(Fig.3i,j),70414

e.g.Kc,Kf2,Ky61,Ky62,Ky54

(Fig.3b

),Kg(Fig.3d

)

4GrtBtMs

e.g.Ky26

Grtrimmed

byCrd

þPlþ

Bt;

Rtrimmed

byIlm

;

e.g.Kf1,Ky52,Ky57

Grtpseudomorphed

byCrd

þPlþ

Btþ

Ilm;

relictBtco

ntainsSilneedles;

prism

atic

And

inmatrix;

Kfs

inmatrix;

Msalso

presentas

largeirregularmasses;

e.g.Bc,Be,Bg

Underlined

arethekeyKFMASH

mineralsch

aracterizingeach

regional

assemblage.

Empty

fieldsindicatenoinform

ation.Allrocksco

ntain

quartz

andplagioclasein

additionto

thetabulatedphases,an

dmay

containrutile.In

boldaresamplenumbersforwhichch

emicaldataareprovided

hereorintheElectronicAppen

dix.In

italicare

samplenumbersrelatingonlyto

photomicrographs.Aquestionmarkbefore

asamplenumber

indicates

that

thereisuncertainty

abouttheoriginalregionalassemblage;

inZoneIV

samplestheregionalassemblages

cannotbeiden

tified

.


838

Dow



no ilmenite in the rocks that have not been thermallyoverprinted, but rutile is present. The mineral pairrutile þ garnet can be used to give a minimum pressureestimate based on the GRAIL geobarometer (Bohlenet al., 1983). The most Fe-rich regional garnet found hasFe/(Fe þ Mg) ¼ 0�84 and this gives P 4 8 kbar for anominal T of 650�C. The pressure-sensitive equilibriumbetween grossular, anorthite, plagioclase and quartz(GASP; Hodges & Crowley, 1985) could be applied inprinciple, but plagioclase chemistry is too variable to beused for an equilibrium calculation. We have foundstaurolite (rarely) in the aureole rocks, but also far fromthe intrusion, as documented by Holdsworth et al. (1987).It is unlikely to be a contact metamorphic mineral, andtextures suggest that it was part of the regional assem-blage, and cofacial with the garnet amphibolite sheets.

CONTACT METAMORPHISM

Four metamorphic zones (I---IV) have been distinguishedusing the pelite outcrops; unfortunately, these cannot bemapped in the surrounding biotite schists (Table 1, Fig. 1).In the lower-grade zones, relict regional mineral assemb-lages can still be identified, but at the highest grade this isnot possible. Textures characterizing the zones include:

Zone I: cordierite---muscovite coronas round kyanite;Zone II: cordierite---muscovite coronas round kyanite;

cordierite---biotite---plagioclase---ilmenite inand around garnet;partial replacement of kyanite by andalusite;

Zone III: K-feldspar, andalusite and fibrolite;Zone IV: cordierite around garnet; euhedral garnet;

K-feldspar; prismatic sillimanite.

Table 2: Representative mineral compositions for Zones I---IV

Cordierite (to 18 O) Garnet (to 12 O)

Zone: II II III III IV IV II II III IV IV IV

Sample: Ky57 Kf2 Bc 70413 Ct5 Ct5 Ky57 Ky57 Bc Ct5 RM 5002 RM 5001

Reg. assem.: 4 3 4 2 ? ? 4 4 4 ? ? ?

Textural

setting:

Corona

round Grt

Corona

round Ky

Corona

round Grt

Corona

round Grt

Corona

round Grt

Matrix

decussate

Relict

Grt core

Relict

Grt edge

Relict

Grt core

Grt (with

Crd corona)

Euhedral

core

Euhedral

rim

SiO2 48.13 48.76 47.48 48.11 46.73 47.35 36.25 36.83 37.45 36.57 36.51 37.70

TiO2 0.02 0.04

Al2O3 32.24 32.99 32.16 32.27 32.11 32.48 21.00 20.73 20.82 21.05 20.91 21.75

FeO 9.98 7.00 11.21 10.00 10.10 12.09 30.17 31.88 34.34 34.80 21.99 32.97

MnO 0.30 0.38 0.50 0.23 0.37 0.22 5.23 2.83 2.21 3.63 13.21 4.82

MgO 7.09 8.95 6.46 7.09 7.03 6.08 2.58 3.36 3.01 3.27 0.86 2.23

CaO 0.08 0.08 0.04 2.86 3.29 2.39 1.17 6.27 2.11

Na2O 0.80 0.46 0.20 0.64 0.53 0.00 0.19 0.37

K2O 0.02 0.00

Total 97.74 98.96 98.29 98.56 97.15 98.79 98.09 99.25 100.41 100.86 99.75 100.09

Si 5.00 4.98 4.98 5.01 4.94 4.95 2.98 2.98 3.01 2.95 2.97 2.97

Ti

Al 4.01 3.97 3.97 4.01 4.00 4.01 2.03 1.98 1.97 2.00 2.00 2.03

Fe 0.86 0.60 0.98 0.87 0.89 1.06 2.07 2.16 2.31 2.34 1.50 2.23

Mn 0.03 0.03 0.04 0.03 0.03 0.02 0.36 0.19 0.15 0.25 0.91 0.33

Mg 1.07 1.36 1.01 1.07 1.11 0.95 0.32 0.41 0.36 0.39 0.11 0.27

Ca 0.01 0.01 0.25 0.29 0.21 0.10 0.47 0.18

Na 0.09 0.13 0.11 0.03 0.06

K

Total 10.97 11.11 11.08 10.99 11.12 11.10 8.01 8.06 8.03 8.03 7.95 8.01

Mg no. 55 69 51 55 56 47 13 16 13 14 7 11

alm 69.0 70.8 76.2 76.0 50.1 74.1

spss 12.0 6.2 5.0 8.1 30.5 11.0

py 10.7 13.4 11.9 12.7 3.5 9.0

grs 8.3 9.5 6.9 3.2 15.9 6.0


839

Dow



We now describe these and other textures and interpretthem in terms of simplified stoichiometric reactionsbetween mineral end-members. Our main assumptionin doing this is that coronas represent reaction products,with the central mineral being a reactant. Ambiguityarises for complex coronas where ‘distant’ mineralsmust have been involved as reactants. This first stage ofanalysis allows us to approximate P---T conditions in theaureole. We give several balanced reactions that sum-marize what is happening in the aureole. Some of thesereactions appear as equilibria on P---T grids; others aremetastable, partly because the regional assemblages wereso far from equilibrium when contact metamorphismoccurred. We return in a subsequent section to a moredetailed discussion of the reactions and textures.

Zone I

In rocks bearing regional kyanite, this mineral is rimmedby relatively fine-grained cordierite andmuscovite aggreg-ates. Because all such rocks contain biotite, the principalreaction involved is

Ky þ Bt þ Qtz þ eW ! Crd þ Ms: ðR1Þ

Here we use the symbol eW to indicate that some watermight be required, depending on the non-stoichiometricwater content of the cordierite, but apart from this thereaction is water-conserving. We will use the symbol Wwhen phases containing different amounts of stoichio-metric water are involved. Rarely, some garnet break-down products are observed, but these are more

Biotite (to 22 O) Muscovite

Zone: II II II II III III III III IV II II III

Sample: Ky57 Ky57 Kf2 Kf2 Bc Bc 70413 70413 Ct5 Kf2 Ky62 Bc

Reg. assem.: 4 4 3 3 4 4 2 2 ? 3 3 4

Textural

setting:

Relict

foliated

Corona

round Grt

Relict

foliated

Corona

round Ky

Relict

foliated

Decussate Relict Corona

round Grt

Decussate

matrix

Corona

round Ky

Relict

foliated

Relict

foiated

SiO2 36.30 35.11 38.24 37.60 35.34 33.64 37.26 35.56 34.84 45.15 46.24 46.32

TiO2 2.32 2.13 1.68 1.56 2.85 2.25 1.52 3.75 3.12 0.40 0.00 0.18

Al2O3 18.35 18.56 18.77 19.72 20.15 19.53 19.02 18.30 19.73 36.16 36.01 36.48

FeO 17.49 18.76 12.40 15.94 21.88 21.61 17.89 19.80 21.76 0.70 0.83 1.28

MnO 0.13 0.00 0.02

MgO 10.22 9.47 15.45 12.46 7.53 7.17 11.64 8.40 7.47 0.49 1.44 0.61

CaO 0.19 0.08 0.17 0.02 0.10 0.12 0.00 0.07

Na2O 0.48 0.38 0.45 0.67 0.38 0.48 0.45 0.36 0.52 1.69 1.64 0.94

K2O 9.06 8.94 8.50 8.52 9.36 8.73 7.50 9.34 8.60 8.56 8.82 10.20

Total 94.22 93.35 95.47 97.05 97.67 93.60 95.28 95.73 96.16 93.15 94.98 96.10

Si 5.52 5.44 5.56 5.47 5.30 5.28 5.54 5.41 5.30 6.09 6.12 6.11

Ti 0.27 0.22 0.16 0.18 0.32 0.27 0.17 0.28 0.36 0.04 0.00 0.02

Al 3.29 3.39 3.22 3.38 3.58 3.63 3.33 3.58 3.54 5.74 5.62 5.67

Fe 2.23 2.43 1.51 1.94 2.75 2.84 2.13 2.61 2.77 0.08 0.09 0.14

Mn 0.02 0.02

Mg 2.32 2.19 3.35 2.72 1.68 1.68 2.58 1.95 1.69 0.10 0.28 0.12

Ca 0.03 0.01 0.03 0.02 0.02 0.00 0.01

Na 0.14 0.11 0.13 1.71 0.11 0.15 0.13 0.09 0.15 0.44 0.42 0.24

K 1.76 1.77 1.58 1.58 1.71 1.65 1.42 1.82 1.67 1.47 1.49 1.72

Total 15.52 15.56 15.52 15.50 15.55 15.59 15.40 15.49 15.49 13.96 14.03 14.02

Mg no. 51 47 69 58 38 37 55 43 38 56 76 46

Data obtained using a modified Cambridge Instruments Geoscan Electron Microprobe with a Link Analytical energy dispersivesystem at the Department of Earth Sciences, Manchester University, except for the two ‘RM’ garnet analyses, which arereproduced from Brearley & Champness (1986). Bottom rows indicate per cent of components in garnet. For more details ofmineral chemistry, see the Journal of Petrology website at http://www.petrology.oupjournals.org/ and Mangan (1996).

Table 2: continued


840

Dow



http://www.petrology.oupjournals.org/

characteristic of Zone II and are, therefore, discussed inthe next section.

Zone II

As in Zone I, when kyanite is present, it shows cordierite---muscovite coronas or more pervasive replacement bythose minerals (Fig. 3b). Regional biotite is always presentand we infer the operation of (R1). Closer to the plutonwithin Zone II, it is evident from field and thin-sectionobservations that these coronas are thicker; at the hillockof Dun Fuinn (Fig. 1) some kyanites are almost entirelypseudomorphed by the new minerals. This pattern of

increasing corona width towards the pluton is discussedlater. The cordierites in these coronas are commonlypinitized, but some robust analyses of cordierite wereobtained (Fig. 4; Table 2). Although the coronas aredominated by muscovite and (pinitized) cordierite, theseare intergrown with small amounts of decussate biotiteinferred to be new. This has a different composition fromthe large aligned biotite flakes inherited from the regionalmetamorphism.In addition, in the field, blue kyanite blades are seen to

contain irregular masses of flesh-pink granular andalusite(Fig. 3c). This is evidence for the reaction

Ky ! And: ðR2ÞAs the pluton is approached within Zone II, it is evidentfrom field observations that andalusite is more abundant:it occurs in a greater proportion of relict kyanites, andoccupies, on average, more of each relict kyanite [Fig. 3d;see also fig. 112 of Yardley et al. (1990)]. This patternaccords with the prediction that higher peak temperat-ures will lead to a greater overall transformation, for anythermally activated reaction. In thin section, thesemasses of andalusite show undulose extinction (Fig. 3d).The kyanite cleavage is subdued or lost within thesemasses. The kyanites are commonly slightly kinked, butthis is probably due to late regional deformation, as it isobserved far from the pluton. Andalusite with strongundulose extinction is sometimes enclosed withinundeformed kyanite. Although the undulose extinctiongives the appearance of subgrains, we infer that it is notdue to imposed deformation, because then the kyanitewould also be strained. Instead, it is a consequence of themechanism of andalusite growth.Another texture characteristic of Zone II is rutile

rimmed by ilmenite. Figure 4 shows that new biotitesare more Fe-rich than old biotites so this mineral cannotbe the source for the Fe in ilmenite. In addition, rutilesare not rimmed in Assemblage 3 rocks in which garnet isabsent. This suggests the operation of

Alm þ Rt ! Ilm þ Qtz þ ‘Al2SiO5’: ðR3Þ

This resembles the GRAIL reaction used in geobaro-metry (Bohlen et al., 1983) but here we do not suggest itnecessarily led to the production of more aluminiumsilicate as a phase, but instead that the Al and Si couldhave been involved in other reactions discussed below.Zone II shows large regional garnets rimmed by aggreg-

ates of cordierite, relatively fine-grained biotite, plagio-clase, ilmenite and some muscovite (Fig. 3a). These arepresent in all garnet-bearing lithologies, both in the pres-ence and absence of regional kyanite. The plagioclases inthese coronas are more calcic than those in the ground-mass. Cordierites are not always pinitized and the rela-tionship between their compositions and those of newand relict biotite is shown in Fig. 4. The corona biotites

A

F M

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8-. 8

- .6

- .4

- .2

.2

.4

.6

.8

Biotite

Kyanite

Assemblage (3)Bt + Ky

?

Garnet

?

Muscovite

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

?Garnet

Biotite

FeO MgO

?

Assemblage (1)Grt + Bt + Ky

Assemblage (4)Grt + Bt

Assemblage (2)Grt + Bt

Assemblage (1)Grt + Bt + Ms

(b)

(a)

K2O

Assemblage (3)Bt + Ms

Regionalassemblagesoutside aureole

+ Muscovite+ Plagioclase+ Quartz+ Water

+ Kyanite+ Plagioclase+ Quartz+ Water

Fig. 2. (a) Thompson AFM diagram for regional assemblages. This hasbeen drawn by projecting from pure muscovite; this is not completelyvalid, as the actual white mica is somewhat phengitic. Garnet containssome grossular, which is implicitly projected from, and biotite containsTi, implying an implicit projection from rutile. The diagram does,however, illustrate the compatibility between the assemblages andforms a useful template for displaying disequilibrium assemblages inthe aureole. Pale grey areas indicate trivariant assemblages; dark greyarea indicates a quadrivariant assemblage. (b) Projection from kyanite(and quartz, water) onto KFM plane in KFMASH for Ky-bearingassemblages, including assemblage 2, which cannot be shown on theThompson AFM diagram.


841

Dow



Crd + Pl

Grt

Crd

Ilm

New Bt

+ Ms

Ky

Ky

AndAnd

Crd Ms

And

2 mm

And

Reg Bt

(a) (b)

(c) (d)

(e) (f)

Fig. 3. Textures of aureole rocks (see Table 1 for setting in terms of original assemblage and aureole zone). Width of each photomicrograph is3�95mm. (a) Sample 70411-10, plane-polarized light (PPL). Garnet rimmed by new biotite, cordierite and ilmenite (note also biotite in cracks). (b)Sample Ky54, PPL. Pseudomorphed elongate kyanite, showing only two high-relief relics at either end. (c) Sample 70411-2, cross-polarized light(XPL). Andalusite ‘island’ within kyanite; note fingers of andalusite parallel to (001) of kyanite. (d) Sample Kg, XPL. Andalusite pseudomorph afterkyanite. Andalusite shows complex extinction, and a corona of cordierite and muscovite. (e) Sample 70413-10, PPL. Garnet completelypseudomorphed by plagioclase and cordierite (colourless), biotite, and ilmenite (black). (f) Sample 70413-13, PPL. Prismatic andalusite cross-cuttingregional biotite foliation, surrounded by and including flakes of regional biotite. (g) Sample Bf7, PPL. Fibrolite mats associated with and probablyreplacing biotite. (h) Sample 117B, PPL. Garnet associated with fibrolite (top right), prismatic sillimanite and other colourless phases as shown. (i)Sample Ct5, PPL. Small anhedral garnet partly rimmed by cordierite. (j) Sample Ct5, XPL. Same field of view as (i). (k) Field photograph of kyaniteblades (pale blue) partly or completely pseudomorphed by andalusite (pale pink), Zone II, Dun Fuinn. Coin for scale, diameter 2 cm. Yellow, greyand white patches are lichen.


842

Dow



are consistently more Fe-rich, Al-rich and Ti-poor thanregional biotites from the same thin section. They are alsomore Fe-rich than the new cordierites, which is to beexpected if these two new phases are in equilibriumwith each other. Both cordierite and plagioclase could

be produced, in kyanite-bearing assemblages, by

Grt þ Ky þ Qtz þ eW ! Crd ðR4Þ

Grs þ Ky þ Qtz ! An: ðR5Þ

(h)Grt

Crd

Kfs

Fib

(g)Sil

(j)(i)Crd

Grt

(k)

And

Ky

Fig. 3. Continued.


843

Dow



Here, Grt is used in a general way for any compositionbetween Fe and Mg end-members. The same texturesare, however, seen in kyanite-absent assemblages, andhere muscovite could be the Al source:

Grt þ Ms þ Qtz þ eW ! Crd þ Bt ðR6ÞGrs þ Grt þ Ms ! An þ Bt: ðR7Þ

In some Grt---Bt---Ms assemblages, large oriented mus-covites appear pristine, which means it is problematic tointerpret this as a reactant for the garnet breakdown-----infact, some new muscovite appears in coronas roundgarnet. This can be explained, as the presence ofilmenite in the garnet coronas suggests operation of (R3).The released Al could be consumed in reactionsequivalent to (R4) and (R5), giving, for example,

Alm þ Rt þ Bt þ eW !Ilm þ Qtz þ Crd þ Ms:

ðR8Þ

In other words, a reaction between garnet, rutile andbiotite would be sufficient to explain the development ofthe coronas without the participation of either regionalkyanite or muscovite.

Zone III

This zone is characterized by the coexistence of alumi-nium silicate with K-feldspar. K-feldspar appears,

together with cordierite, in coronas around andalusitemasses (pseudomorphs after kyanite), and within relictregional biotites and muscovites (although it is not inter-preted as relict itself ). Andalusite is present not only asirregular masses with undulose extinction in pseudo-morphs after kyanite, but also as prismatic grains in thematrix, commonly cutting across foliated (relict regional)biotite and including that phase (Fig. 3f ). The prismaticmatrix andalusite is, in places, slightly salmon pink andpleochroic in thin section, indicating some Fe3þ and/orMn3þ (Deer et al., 1982). The andalusite in pseudo-morphs has no detectable levels of those elements. Thepresence of K-feldspar as well as this matrix andalusite(which is not associated with kyanite) suggests

Ms þ Qtz ! Kfs þ And=Sil þ W: ðR9ÞTwo other reactions that can produce K-feldspar, andthat may have operated locally, are

Ms þ Bt þ Qtz ! Kfs þ Crd þ W ðR10Þ

Ky=And=Sil þ Bt þ Qtz ! Kfs þ Crd þ W:

ðR11ÞPrismatic matrix andalusite could also have beenproduced by

Crd þ Ms ! And þ Bt þ Qtz þ eW: ðR12ÞFibrolite is present in Zone III as mats within relict

regional biotite (Fig. 3g). It is again not straightforward toidentify the reaction producing this texture. Reactions(R10) and (R11) both indicate biotite as a reactant butthese do not produce Al2SiO5. Fibrolite is, despite thelarge differences in Al between the two phases, foundwithin biotite in many metamorphic terrains and this isprobably due to ease of nucleation there (Chinner, 1961;Kerrick, 1990, p. 276). It has been proposed that fibroliteis produced from biotite by leaching of K, etc. (Kerrick,1987), but we find no evidence for fluid flow here. Wefavour a model for fibrolite and matrix sillimanite basedon reaction (R9), where biotite is involved as a preferrednucleation site (see Carmichael, 1969).Aggregates of cordierite, plagioclase, biotite and ilmen-

ite are interpreted as pseudomorphs after garnet (Fig. 3e).In this zone, then, regional garnet and kyanite havealmost completely disappeared; aluminium silicate andK-feldspar are stable, but regional muscovite and biotitepersist. Figure 5 shows coexisting cordierite and biotitecompositions in Zone III.

Zone IV

The main band of pelites does not extend all the way tothe pluton in the SW of the Moinian inlier and is prob-ably faulted out (Fig. 1). To the NW, however, outcropsof similar pelites are found adjacent to, and as largexenoliths or roof pendants within, the pluton. These areassigned to Zone IV. Many of the textures in Zone IV

Key

A

F M

- .4

- .2

.2

.4

.6

.8

Kyanite

Biotite

KbKf2

Kc

Ky61

Cordierite

Ky62

Precursorregionalminerals

Newthermalminerals

Kf1Ky52

Ky57

Garnet

Regionalassemblage

(2)

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

+ Muscovite+ Plagioclase+ Quartz+ Water

Zone II

Regionalassemblage

(4)

Regionalassemblage

(3)

Regional assemblage (2)



Tie lines between new zone (II) minerals(dashes indicate cordierite compositioninferred from pinite)

Fig. 4. AFM diagram for Zone II. Pale grey areas indicate wheretrivariant regional assemblages would have plotted prior to contactmetamorphism. Filled symbols indicate relict regional minerals; opensymbols indicate new phases in the same rock. Arrows show how old andnew biotites differ in each sample.


844

Dow



resemble those in Zone III but in addition, prismaticsillimanite occurs [see MacKenzie & Guilford (1980,p. 15)], as do small garnets. Prismatic sillimanite isfound in two textural settings: in 1---2 cm andalusite-rich‘knots’ clearly visible in outcrop, and in the matrix.The knots themselves carry rare kyanite patches. Ourfavoured explanation of the knots is that they are pseudo-morphs after kyanite. Although they are equant ratherthan having the elongate outline of regional kyanite, wecan think of no other way to obtain such large patches ofaluminium silicate, and the rarely preserved kyanite mustsurely be a relict regional phase. Sillimanite may beproduced by the direct reaction

And ! Sil ðR13Þas well as by reaction similar to that in Zone III[reaction (R9)]. The small garnets can be markedlyeuhedral; this is visible in the field in the outcrops c. 20meast of Bendoran Cottage and recorded in thin section[e.g. Brearley & Champness (1986, fig. 1b)]. These donot show signs of reaction rims or growth of newminerals along cracks. The small garnets are interpretedas new contact metamorphic products, from one of thereactions

And=Sil þ Bt þ Qtz ! Grt þ Crd þ Kfs þ W

ðR14ÞAnd=Sil þ Bt þ Qtz ! Grt þ Crd þ Kfs þ L:

ðR15ÞIt should be noted that the reaction can be balancedusing either water (W) or melt (L) on the right-hand side.The picture is complicated by the occurrence, also inoutcrops east of Bendoran Cottage and in a roof pendantwest of there, of anhedral garnets with various coronasincluding biotite (Fig. 3h) or commonly entirely

cordierite. We deduce these are relict regional garnet.In Zone III regional garnets are commonly entirelypseudomorphed, and one might expect the same inZone IV. However, given the complexities of kinetics,this prediction is not robust. We propose that theseanhedral garnets are resorbed regional metamorphicgarnets. The resorption may have remained incompletebecause, for example, some coronas are single crystals ofcordierite. Corona growth would then have involvedlattice diffusion through cordierite, which is much slowerthan grain boundary diffusion in the polycrystallinecoronas of Zone II. These issues are discussed further ina subsequent section.The presence of two or three Al2SiO5 polymorphs

indicates the extent of disequilibrium. Further evidencefor this is found in the variable chemistry of phases suchas cordierite. Figure 6 shows that cordierites adjacent tonew garnets have distinct chemistries from those furtheraway in the matrix, which are more Fe-rich.

PRESSURE---TEMPERATURE

CONDITIONS IN THE ROSS OF

MULL PLUTON AUREOLE

P---T grid to be used

The extent of disequilibrium means that it would not beeasy to use our data to propose or modify P---T grids andequilibria related to low-pressure metapelite assemblages.Instead, we use the grid that has been refined over many

A

.1

.2

.3

.4

.6

.7

.8

.9

Sillimanite

.5

F M

70414 Ct5Smallgarnets

70414Ct5

Relic decussate biotite

Tie lines betweenrelict Zone IIIminerals

Stable mineralsin Zone IV

Tie lines between coexistingZone IV minerals

Key

Relic zone (III)minerals

+ K-feldspar+ Plagioclase+ Quartz+ Water

Zone IV

Cordierite

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Fig. 6. AFM diagram (projected from K-feldspar) for two samples fromZone IV (70414 and Ct5). It should be noted that, although tie linescross, all four of the marked phases (Crd, Grt, Sil, Bt) are present in boththe slides.

A

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

.1

.2

.3

.4

.6

.7

.8

.9

Biotite

.5

70413Bg

BeBc

Increasing proximityto granite contact MF

Cordierite

+ K-feldspar+ Plagioclase+ Quartz+ Water

Zone III

Fig. 5. AFM diagram (projected from K-feldspar) for Zone III, show-ing analyses from four rocks (70413, Bc, Be and Bg).


845

Dow



years by Pattison and co-workers (Pattison & Harte,1997), which we will refer to as the PH97 grid (Fig. 7).We remark on specific equilibria now, and give reasonsfor this choice.

The stability limits of Aluminiumsilicate þ Biotite þ Qtz in KFMASH

Powell & Holland (1990) and Xu et al. (1994) calculatedKFMASH P---T grids using an internally consistent ther-modynamic database (Holland & Powell, 1990). Thesegrids predict that, although biotite---sillimanite---quartzcan be stable, the assemblage biotite---andalusite---quartzis unstable in KFMASH, being represented instead bycordierite---muscovite---staurolite or, at higher tempera-tures, cordierite---muscovite---garnet. However, biotite---andalusite---quartz occurrences are common in manymetamorphic terrains and this is acknowledged as aproblem with the thermodynamic database (Powell &Holland, 1990). Occurrences of muscovite---cordierite---staurolite---biotite, which, if in equilibrium, wouldindicate the stability of cordierite---muscovite---staurolite,are instead argued to be polymetamorphic (Pattison et al.,1999). Therefore data from natural systems are morecompatible with the PH97 grid.

The equilibrium Crd þ Ms ¼ And þBt þ Qtz þ eW (R12)

This is of great importance in understanding low-pressure assemblages in metapelites. There is no doubtthat the right-hand side is favoured for more Fe-richcompositions, but alternative slopes for isopleths of thisreaction have been proposed. Using data from manymetamorphic aureoles, Pattison et al. (2002) proposed anegative slope, so that the stability field of aluminiumsilicate plus biotite increases with rising temperature.However, Powell & Holland (1990) and Xu et al. (1994)calculated a positive slope for the Mg end-member reac-tion using an internally consistent thermodynamic data-base (Holland & Powell, 1990). A more recent version ofthat database, and other databases, yields different posi-tions but similar positive slopes (Pattison et al., 2002).Thermodynamic data have been revised in accordancewith field constraints (Pattison et al., 2002) and a newKFMASH grid has been calculated from these data.The predicted grid for subsolidus phase relationships inthe absence of graphite, and using the Pattison (1992)triple point, was given by Pattison et al. (2002, fig. 10a).This is very similar to the PH97 grid, except that theequilibrium (R12) in their diagram is contoured in terms

Mg-C

rdK

fsQ

tzW

Mg-C

rdK

fsQ

tzW

Ms C

hl Q

tzCrd

Bt W

Ms B

t Qtz

(R10)

Crd

Kfs

WM

s Qtz

Als

Kfs

W

Bt Als Qtz W(R12)

Ms Crd

Ky

And

Ky

Sil

Bt Sill Q

tz

Crd Kfs L

Bt S

il Q

tzG

rt C

rd K

fs L

L

Pl K

fs Qtz W

6

5

4

3

2

1

500 600 700 800

P(kb)

Ms

Chl

Qtz

Bt A

ls W

Ms C

hl Q

tzC

rd B

t Als

W Grt K

fs L

Bt C

rd Qtz

(a)

(b)

(c)

Grt K

fs L

Bt Sil Q

tz(R

16)

SilAnd

LI & II III IV

?

T (˚C)

(R15)

(R9) (R13)

Mg65

Mg55

Mg45

Fig. 7. P---T grid (Pattison & Harte, 1997). Black lines and labels indicate univariant KFMASH equilibria; black dashed lines and italic labels,isopleths for divariant KFMASH reactions; Mgxy, cordierite Mg number for some isopleths; grey dotted lines, alternative positions for the And¼ Silequilibrium (Richardson et al., 1969; Holdaway, 1971). Isopleths for (R16) are sketched in for completeness. Thick grey lines with arrows indicateapproximate ranges of peak temperature for zones I---IV in the Ross of Mull aureole, assuming P ¼ 3�2 kbar.


846

Dow



of biotite Mg-number rather than cordierite Mg-number.Using a cordierite---biotite KD of 0�57 (Pattison et al., 2002)shows that the isopleths in the two versions of the grid arenot greatly dissimilar. The PH97 grid has the advantagethat the isopleths are extended to different equilibria athigher temperatures. In summary, with regard to boththis equilibrium and that discussed in the previous sec-tion, data from natural systems are more compatible withthe PH97 grid.Using the PH97 grid, then, the divariant assemblage

Bt---And---Qtz---Crd---Ms is expected to show a cordieritecomposition varying (becoming more Mg-rich) with tem-perature (at fixed pressure) but not with bulk-rock chem-istry; trivariant assemblages such as Bt---Qtz---Crd---Msare expected to show cordierite compositions varyingwith bulk-rock chemistry but not temperature.

The equilibrium Bt þ And/Sil þ Qtz ¼Crd þ Kfs þ W (R11)

The PH97 grid predicts that, just up T of the muscovitedehydration breakdown line (R9), for a given P and T

there is a specific cordierite composition buffered byAl2SiO5, K-feldspar, biotite, quartz and water. This buf-fered cordierite composition becomes more Fe-rich as Trises at fixed P in accordance with the sliding reaction(R11):

Bt þ And=Sil þ Qtz ! Crd þ Kfs þ W:

For a particular P and T, cordierites more Mg-rich thanthis buffered composition may be in equilibrium withonly two of sillimanite, biotite and K-feldspar. Inunbuffered (trivariant) assemblages, for a given bulkchemistry, the cordierite composition would not vary atall as T varies, if we acknowledge the negligible T

dependence of Fe---Mg exchange between biotite andcordierite (Pattison et al., 2002). Thus: the divariantassemblage Bt---And---Qtz---Crd---Kfs is expected to showa cordierite composition varying (becoming more Fe-rich) with temperature (at fixed pressure) but not withbulk-rock chemistry; trivariant assemblages such asBt---Qtz---Crd---Kfs are expected to show cordieritecompositions varying with bulk-rock chemistry but nottemperature.

The equilibrium Bt þ Sil þ Qtz ¼ Grt þCrd þ Kfs þ W and related dehydrationmelting (R14) and (R15)

This KFMASH univariant equilibrium appears on manygrids (Xu et al., 1994; Pattison & Harte, 1997; Spear et al.,1999). At high pressures and temperatures it is metastablewith respect to the dehydration melting reaction (R15) inwhich melt rather than water is produced (e.g. Holland

et al., 1996; Cenki et al., 2002). In some studies Kfs appearson the left (low-T ) side of that reaction (Carrington &Harley, 1995), although this does not affect the inter-pretations we give here. White et al. (2001) presented agrid for partial melting equilibria in NCKFMASH. Theaddition of Na and Ca makes the equilibrium (R15)divariant:

Bt þ Sil þ Qtz � Pl ¼ Grt þ Crd þ Kfs þ L:

However, this and other divariant fields that correspondto KFMASH univariant melting reactions are narrowand close in position to the KFMASH univariant lines.This is illustrated in, for example, fig. 7 of White et al.(2001), where the divariant field of Bt---Sil---Qtz---Grt---Crd---Kfs---Pl---L is shown spanning just a few degrees at c.740�C (at 5 kbar). Although this grid is very sophist-icated, it is based on a thermodynamic dataset thatpredicts that equilibrium (R12) will favour muscovite þcordierite as T rises (see, for example, the subsolidus partof White et al.’s fig. 7), which contradicts field data asargued in the previous section. The univariant line (R15)on the PH97 grid will be taken as an adequateapproximation to the divariant region that would bepresent in NCKFMASH. We have sketched in isoplethsfor where the divariant reaction

Bt þ Sil þ Qtz ¼ Grt þ Kfs þ L ðR16Þ

would appear on the PH97 grid-----the slopes areconstrained topologically [see Cenki et al. (2002),although those workers showed Kfs on the low-T sideof this equilibrium].

Assumptions involved

We will assume that the peak temperature at each pointwas attained at roughly the same pressure. This will not betrue if the pluton and its aureole were being exhumedquickly, or if it has been substantially tilted, but we haveno evidence for either of these scenarios. Although insome models the pluton is portrayed as a dipping sheet(as mentioned above), there is no suggestion it has beentilted. The errors involved here are likely to be smallerthan those arising from other uncertainties. Our problemis to distinguish what parts of each texture are likely torepresent equilibrium assemblages for each zone and touse a petrogenetic grid to constrain P and T using theseassemblages, bearing in mind that grids are themselvessubject to uncertainty andmust be assessed pragmatically.The previous section indicates various types of disequi-

librium in every zone:(1) regional relict minerals may persist, probably in

disequilibrium, but are relatively easy to recognize;(2) new minerals are likely to be in equilibrium with

each other in a given corona but in the lower-grade


847

Dow



zones, with separate coronas around different regionalporphyroblasts, equilibrium between coronas is notguaranteed;(3) not only regional metamorphic minerals, but also

minerals formed on the prograde contact metamorphicpath may persist; for example, in Zone IV, prismaticsillimanite must be the peak Al2SiO5 polymorph, yetandalusite, itself thermal, remains.The presence of thermal andalusite implies pressure

below the Al2SiO5 triple point, itself placed at 3�8 kbar(Holdaway, 1971), 5�2 kbar (Richardson et al., 1969) orsomewhere between these values on the PH97 grid [point(a) in Fig. 7]. As discussed above, andalusite is found inZone III not only in kyanite pseudomorphs but also aseuhedral grains in the matrix. The latter are interpretedas being a product of the breakdown reaction (R9). Silli-manite is, however, also present in Zone III and inter-preted as due to the equivalent to (R9). We infer that themuscovite breakdown reaction began by producingandalusite (R9) and then, as temperature rose, producedsillimanite instead. This sequence suggests that pressurewas below that of the intersection of (R9) with the And ¼Sil line [point (b) in Fig. 7], placing us within FaciesSeries 1 of Pattison & Tracy (1991); however, we arguebelow that pressure was in fact a little higher, placing theaureole just within Facies Series 2.

Zone II: extent of disequilibrium

The most obvious evidence for disequilibrium is the coex-istence of andalusite with relict kyanite: the transformation

is incomplete for kinetic reasons. Additional evidence fordisequilibrium comes from considering the minerals Crd,Ms, And, Bt and Qtz. In all rocks from Zone II thatcontain relict kyanite, new cordierite, muscovite and anda-lusite are present, together with relict biotite and varyingamounts of new biotite. This suggests that although kyan-ite (and, where present, regional garnet) breakdown is notcomplete, the new stable assemblage would have beenCrd---Ms---And---Bt---Qtz. Zone II is rather wide and withinit the peak temperature must have varied. If this divariantassemblage is in equilibrium throughout Zone II then, asdiscussed in the ‘cordierite equilibria’ section above,cordierite composition should vary systematically withposition (specifically, more Mg-rich compositions are pre-dicted for higher-T assemblages closer to the pluton).However, the variation of cordierite compositions withposition in Zone II (Fig. 8) is not systematic. Moreover,the isopleths in Fig. 7 show that high pressures are requiredto stabilize Mg-rich cordierite (XMg ¼ 0�7, for example)in equilibrium with aluminium silicate, biotite, quartz andeither muscovite or K-feldspar. Such pressures wouldmean that andalusite would never be a stable polymorphin the aureole and we reach a contradiction. Finally, themore Mg-rich cordierites from Zone II in Fig. 4 are fromcoronas round relict kyanite (with variable amounts oftransformation to andalusite) whereas the more Fe-richcordierites are from coronas round garnet.We propose that the range in cordierite and new biotite

compositions in Zone II is purely a function of bulk-rock(or local textural domain) composition, not of meta-morphic grade. We suggest that, given more time, each

Approximate distance (in map view) from pluton (m)

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 200 400 600 800 1000 1200 1400 1600

Zone IIZone IIIIV

70411

Ky52

Ky57

Kc

Kf2

Ky61

70413

BgBc

Be

Ct570414

Crd

Mg

num

ber cr

d +

bt +

and

+ m

u

crd

+ bt

+ a

nd/s

ill +

kfs

crd

+ an

d/si

ll +

kfs

+ gr

t

Fig. 8. Plot of actual cordierite compositions from all samples as a function of approximate orthogonal distance from the pluton contact, togetherwith a schematic indication of expected spatial variations in cordierite composition if that mineral were buffered by the assemblages indicated inboxes.


848

Dow



of the analysed rocks from Zone II would have equili-brated to one of the two trivariant assemblagesCrd---Ms---Bt---Qtz or Crd---Ms---And---Qtz. Which onewould have been attained depends on whether the bulk-rock chemistry lies above or below the cordierite line inFig. 4. For rocks whose bulk chemistry lies below thecordierite line, the andalusite forming from kyanite ismetastable-----it would eventually have reacted with biotiteand been completely consumed. We next argue that thethicknesses of coronas in Zone II show that aluminiumsilicate would have been completely consumed at equilib-rium, for the analysed Zone II rocks. We may imagine arock with a particular M/FM ratio of biotite in contactwith regional kyanite, and quartz. Let us suppose that thisrock is taken to a variety of peak temperatures but atmuch lower pressures than those of the regional event,and the two grains react to reach equilibrium. At suffi-ciently low temperatures (the precise conditions being afunction of the M/FM ratio), the equilibrium assem-blage would be Crd---Ms---Bt: the regional kyanitewould be completely replaced. At intermediate tempera-tures, And---Crd---Ms---Bt would be stable. Some kyanitewould react, but the rest would transform to andalusite.At high temperatures, And---Bt (or Sil---Bt) would bestable. The kyanite would transform to andalusite butno other reaction need occur. In other words, we wouldexpect less Crd-producing reaction occurring at highergrades, and narrower coronas. In fact, there is a generaltrend towards wider coronas and, in the higher-gradeparts of Zone II, regional kyanites are commonly com-pletely pseudomorphed by cordierite---muscovite aggre-gates. This trend in corona widths is then an effect ofthermally activated reaction kinetics. Even though thedriving force (free energy reduction) for reaction is less athigher grades, diffusion must be involved and is so sensi-tive to temperature that this kinetic effect dominates.

Zone III: possible equilibrium

In Zone III the rocks contain, amongst other phases,sillimanite, K-feldspar, cordierite, new biotite and quartz.We propose that the systematic variation with position ofcordierite and biotite chemistry (Figs 5 and 8) supportsthe hypothesis that these five phases were in equilibrium,rather than the alternative that the cordierite chemistry iscontrolled by bulk-rock chemistry and at least one of theother phases listed is in disequilibrium. A previous sectiondiscussing (R11) emphasized that a systematic variationof cordierite towards more Fe-rich compositions is to beexpected with rising temperature.The cordierite composition at the lowest temperature

within Zone III (XMg ¼ 0�55) would then indicate theintersection of that cordierite isopleth with the muscovitebreakdown line. This gives point (c) in Fig. 7, at 3�2 kbar.At this pressure, muscovite would break down just outside

the andalusite stability field (using the And ¼ Sil equilib-rium line of Fig. 7). This is an apparent contradiction tothe argument that matrix andalusite in Zone III wasproduced by muscovite breakdown, and therefore atpressures below that of point (b). There are several waysto explain this. First, the position of the And¼ Sil line stillremains debatable, and experimental studies give widelyvarying predictions (Richardson et al., 1969; Holdaway,1971). Second, in Zone III, sillimanite is present only asfibrolite. This can only be stable at higher temperaturesthan coarse-grained, defect-free sillimanite (Salje, 1986;Penn et al., 1999), and andalusite could have crystallizedmetastably if defect-free sillimanite could not nucleate.Third, the matrix andalusite is sometimes coloured, indi-cating presence of Fe3þ and/or Mn3þ that will enlarge itsstability field at the expense of that of sillimanite. Ifnucleation of coarse-grained sillimanite were inhibited,the line Ms þ Qtz ¼ And þ Kfs þ W would extendmetastably to higher pressures, but close to the stableequilibrium line Ms þ Qtz ¼ Sil þ Kfs þ W. TheXMg ¼ 0�55 isopleth would intercept this line at approx-imately 3�2 kbar and 650�C.

Zone IV: significance of garnet

Zone IV contains anhedral garnets with cordierite co-ronas. The chemistry of these garnets is broadly compar-able with that recorded from relict garnets in other zones,and they are interpreted as relics. Euhedral garnets areinterpreted as new because of their morphology. On thePH97 grid, garnet can appear up-grade in two ways.One is by the divariant reaction (R16), but this can onlyaccount for the appearance of garnet in Fe-rich cordierite-absent assemblages, so cannot apply here (see, e.g.Pattison & Harte, 1997, fig. 8). Alternatively, garnetappears by the univariant dehydration melting reaction(R15), predicted to be at 775�C if P was 3�2 kbar.There are three points to discuss in relation to the

proposed reaction. First, euhedral garnets have Mn-richcores that would stabilize garnet at lower temperaturesthan those shown on the PH97 grid (which includes onlyKFMASH equilibria). Second, there is limited field evid-ence for large-scale partial melting in Zone IV. Third, thegrid indicates that, near the dehydration melting reac-tion, cordierite should have evolved towards muchmore Mg-poor compositions than are actually seen. Weaddress these points below.

Effect of Mn on garnet stability

Given the disequilibrium present at all grades, the factthat Na and Ca will also have an effect, and uncertaintiesin thermodynamic data, we cannot be definite aboutthe effect of Mn in garnet on (R14) and (R15). InMnKFMASH, there will be a divariant field of Bt þSil þ Qtz þ Grt þ Crd þ Kfs þ L, stabilizing garnet at


849

Dow



lower temperatures than for the univariant line. At3�2 kbar the univariant dehydration reaction (R14) ismetastable with respect to the dehydration melting line(R15). However, in MnKFMASH the assemblage Bt þSil þ Qtz þ Grt þ Crd þ Kfs þ W will have a divariantfield stretching down temperature from the position ofmetastable (R14). To gain some insight regarding theeffect of Mn we calculated the amount required to stabil-ize garnet in the assemblage Bt þ Sil þ Qtz þ Grt þCrdþKfs þW, which is divariant in MnKFMASH. Weassumed Mn was present only in garnet, and used thethermodynamic database of Holland & Powell (1998).Because, as explained above, this dataset does not repro-duce key features of the PH97 grid, we use this modelonly to estimate how much a given Mn content in garnetwill shift the equilibrium (R14) down temperature, not topredict its actual position. The modelling was performedusing the Vertex software package (e.g. Connolly &Petrini, 2002). Figure 9a shows how garnet compositionis predicted to vary. A spessartine content of 30% couldstabilize garnet 30�C below the univariant (R14).Figure 9b is a sketch of how Mn would vary for garnetin equilibrium with melt, based on the premise that theassemblage in equilibrium with W will be stable at lowertemperatures than that in equilibrium with L, and thatKFMASH univariant (R15) is at lower T than (R14). Ifgarnet were in equilibrium with L not W then a 30%spessartine content would stabilize it only less than 30�Cbelow the univariant line (R15). The PH97 grid showsthat W cannot be in equilibrium with plagioclase þK-feldspar þ quartz above c. 670�C. We do not think itis possible that all of zones III and IV were squeezed intothe small temperature interval 650---670�C, and thereforeuse the PH97 grid to predict that melt was likely to havebeen present in Zone IV. Thus new garnet cores grew ata minimum temperature of 745�C and rims, with lowerMn, are likely to have formed less than 20�C below(R15), in other words at a minimum of 755�C.In summary, Mn in garnet could have allowed growth

to begin perhaps 30�C below the temperature of theunivariant (R15), but T must have risen to not far belowthat KFMASH univariant as Mn-poor rims developed.

Evidence for melting

The previous section argues that melting is likely to haveoccurred during garnet formation in Zone IV. In thefield, near the pluton, various textures involving granit-oids and country rock can be described as migmatitic.Some of these involve decimetre- to metre-scale sharp-edged granitoid sheets, which are clearly intrusive (e.g. onthe south coast at 367179, and in outcrops east ofBendoran cottage). Elsewhere (e.g. between Bendorancottage and the main pluton), smaller leucosome patcheswith less distinct boundaries are present. These could be

due to anatexis, but are not markedly developed in theoutcrops that exhibit the most euhedral garnets. Wepropose that melt produced by anatexis permeatedsome distance, into nearby country rocks or into bodiesof intruded magma. In the exposures east of BendoranCottage, there are several metre-scale fingers of granitepresent, so migration over large distances is not required.If melt did not move at all, it would have back-reactedwith other phases on cooling. This would have involvedconsumption of garnet and destruction of the euhedraloutlines. We cannot envisage how the prograde, euhedralgarnets could have survived if melt had remained presenteverywhere it had been generated.

Cordierite compositions in Zone IV

Extrapolation of cordierite isopleths on the PH97 gridshows that cordierites should evolve to markedly

P = 3.2 kb

670 680 690 700 710 720 730 740 750

Grt with L

Grt with W

KFMASH dehydration melting T

KFMASH dehydration T (metastable)

T (˚C)

T (˚C)

0

10

20

30

40

50

60

70

80

Spes

sart

ine

%Sp

essa

rtin

e %

(b)

(a)

Fig. 9. (a) Mn content of garnet in the MnKFMASH assemblage Bt þSil þ Qtz þ Grt þ Crd þ Kfs þ W, as a function of temperature forP ¼ 3�2 kbar. This is not intended to be consistent with Fig. 7 (see textfor details). (b) Schematic plot of how the equivalent assemblage withL instead of W would buffer Mn in garnet.


850

Dow



Mg-poor compositions before reaction (R15) is encoun-tered. Such compositions have not been measured, buteven in a single thin section cordierite shows a variety ofcompositions (Table 2) and is clearly not in equilibrium.We suggest that cordierite compositions readjusted to anextent on the retrograde path in Zone IV.

DISCUSSION

Unusually marked disequilibrium and the role of water

Wehave outlined various arguments that demonstrate theextent of disequilibrium in this aureole: they are based on adirect interpretation of textures and on relating assem-blages and chemistry to a published petrogenetic grid. Atleast two unusual factors contribute to the extent of dis-equilibrium, both aspects of the high grade of the protolith.(1) Coarse grain size means that diffusion distances

are longer than in a fine-grained protolith.(2) The protolith, being high grade, did not carry

many hydrous minerals. Some prograde reactions,however, required input of H2O. We have notmeasured the water content of cordierite, but assumethere must be some. In many aureoles, previously low-grade protoliths undergo dehydration during heating.In contrast, assuming the K/H ratio in regional biotiteand new muscovite is the same, contact metamorphicreactions such as (R1) in the Ross of Mull aureolewould require addition of H2O. This might be ratelimiting. There are three possible sources of H2O inaddition to that in biotite. First, there are small amountsof chlorite in the regional assemblages. This relates toretrogression after the peak of regional metamorphism,and the breakdown of this chlorite could plausiblyprovide the relatively small amounts of H2O requiredfor reactions such as (R1). Second, H2O will have beenreleased in the higher-grade parts of the aureole (e.g.from breakdown of muscovite in Zone III). Third, watercould have come from the crystallizing pluton. How-ever, in Zone IV water activity is predicted to be lessthan unity, if the garnet-producing reaction is correctlyidentified. This is evidence against water from thepluton having permeated through the wall rocks visibleat present. In the aureole as a whole, there is not muchmacroscopic or microscopic evidence for large-scalefluid movement. Because water as a phase is generallyconsidered to speed up reaction rates, even if it is notrequired to balance the reaction, its absence here isperhaps another reason for slow reaction rates.

Relative diffusion rates

Textures give some insight into the relative diffusion ratesfor different elements. This assessment, although qualita-tive, is based on the premise that nucleation, andattachment or detachment of atoms at interfaces, is not

rate-controlling. The Ky ! And transformation is anexception: as the two minerals are adjacent, no diffusionis involved; the transformation must have been controlledby interface kinetics.The reaction (R1) involves both kyan-ite and biotite as reactants. Coronas are observed aroundkyanite, not around biotite; biotites not adjacent to kyan-ites are unaffected. This suggests that diffusion of K, Mgand Fe (supplied from biotite) was easier than diffusionof Al from kyanite. Other metamorphic studies have alsoinferred that Al was a slow-diffusing species. Rutile in thematrix, distant from ferromagnesian phases, is rimmedby ilmenite, again suggesting that Fe could diffuse oversignificant distances, with Ti relatively immobile.The slow rate of Al diffusion would inhibit garnet

breakdown by (R4)---(R6), all of which involve transportof Al from other phases. Instead, the reaction scheme forgarnet breakdown in Zone II was dominated by reactionwith rutile. This can be thought of as (R3) releasing‘Al2SiO5’ from garnet with (R4) and (R5) locally consum-ing this within the garnet pseudomorph, or, effectively,

Ca,Fe,Mg Grt þ Rt þ eW ! Crd þ Pl þ Ilm:

Rutile inclusions in garnet are commonly rimmed by orcompletely replaced by ilmenite, in contrast to those inkyanite that are well preserved. This corroborates thesuggested reaction scheme.

Intrusion temperature and heat transfer

At what temperature were the country rocks when theRoss of Mull pluton intruded? K---Ar ages from the intru-sion of 423� 4Ma (biotite) and 416� 4Ma (hornblende)might in principle be used to constrain its cooling history(Beckinsale & Obradovich, 1973); however, the biotiteages, being older than those of hornblende, are probablyaffected by excess argon. The hornblende age overlapswith the Rb---Sr intrusion age of 414 � 3Ma (Hallidayet al., 1979), indicating only that the intrusion cooledthrough c. 500�C soon after emplacement. An aureolepressure of 3�2 kbar gives a depth estimate of c. 12 km(for a nominal density of 2800 kg/m3). A high regionalgeothermal gradient of 30�C/km would imply an ambi-ent temperature of 350�C, which we regard as an upperlimit. Simple thermal models of intrusion show that thepeak aureole temperature, adjacent to the intrusion, willbe comparable with or less than the average of intru-sion and wall rock temperatures (Furlong et al., 1991). AZone IV temperature of 750�C would require a magmatemperature of 1150�C, which is far too hot for a graniticmagma. Although the intrusion is composite, with moremafic components implying the involvement of hottermagmas, these are only exposed in the SW (British Geo-logical Survey, 1999) and there is no evidence for themnear the studied aureole rocks. Instead, magma mayhave passed through the currently exposed level for a


851

Dow



prolonged period, so that the wall rocks became hotterthan a purely conductive model would predict. Latentheat of crystallization will also have served to raise wallrock temperatures substantially. Fluids expelled from themagma, however, are not thought to have passed throughthe exposed aureole rocks, and cannot therefore havecontributed to heat transfer. It is likely that fluids werelost at structurally higher levels in the pluton system,which have now been removed by erosion.

CONCLUSIONS

The Ross ofMull aureole affected initially coarse-grained,dry rocks that had previously been regionally metamor-phosed. Four metamorphic zones have been defined, butnone show equilibrium textures or assemblages.Zone I shows regional kyanite rimmed by cordierite

and muscovite formed from reaction with biotite. Zone IIshows the same texture with a greater extent of replace-ment; in addition, kyanite is partly replaced by andalusite.The extent of replacement increases towards the pluton.Garnet is partly replaced by cordierite, plagioclase andnew biotite. Incomplete reaction is due to slow diffusionthrough coarse-grained material, and limited supply ofH2O. The andalusite, although new, is itself metastable.We show that, for a range of bulk compositions, theequilibrium AFM assemblage would have been cordieriteand biotite, without andalusite. Diffusion was too slowto allow complete reaction of kyanite/andalusite tocordierite and muscovite.Zone III shows the occurrence of K-feldspar and fibro-

lite, as a result of the breakdown of muscovite. Garnetsmay be completely replaced by cordierite, plagioclaseand new biotite, and kyanite is sporadically preserved.Zone IV, which occurs adjacent to the pluton and in raftswithin it, shows prismatic sillimanite as well as fibrolite. Insome locations anhedral garnets surrounded by cordieriteare probably regional relics. Others nearby are euhedralwith Mn-rich cores. We suggest that the euhedral garnetsare formed by the reaction of biotite with sillimanite toproduce cordierite, K-feldspar, garnet and melt. The Mncontent is not sufficient to greatly reduce the temperatureof this reaction. Preservation of euhedral garnet formssuggests that the melt moved away to some extent, other-wise it would have back-reacted with garnet.This aureole is unusual for showing all three alumi-

nium silicate polymorphs, and for other markedlydisequilibrium assemblage and textures. This characteris due at least in part to the coarse-grained and dry natureof the protolith. In contrast, many other Caledonianintrusions such as Ballachulish (Pattison & Harte, 1997)and Fanad (Edmunds & Atherton, 1971) affect low-gradeprotoliths, including chlorite slates and schists. Such rockswould not require input of water to prograde and, beingfine grained, would react relatively fast. The extent of

disequilibrium in the Ross of Mull aureole makes itunusually difficult to determine metamorphic conditions,but especially interesting for the study of reaction kinetics.

ACKNOWLEDGEMENTS

L.M. acknowledges receipt of an NERC Ph.D. student-ship. We thank I. Fitzsimons, L. Kriegsman and G. Wattfor detailed and constructive comments on an earlierversion of this manuscript. We thank M. Wilson for hereditorial work, and in particular for accommodating thelong revision time (which was entirely the responsibility ofthe first author). S. Harley dealt with the initial version ofthe manuscript.

SUPPLEMENTARY DATA

Supplementary data for this paper are available onJournal of Petrology online.

REFERENCESBeckinsale, R. D. & Obradovich, J. (1973). Potassium---argon ages for

minerals from the Ross of Mull, Argyllshire. Scottish Journal of Geology

9, 147---156.

Bohlen, S. R., Wall, V. J. & Boettcher, A. L. (1983). Experimental

investigations and geologic applications of equilibria in the system

FeO---TiO2---Al2O3---SiO2---H2O. American Mineralogist 68,

1049---1058.

Bosworth, T. G. (1910). Metamorphism around the Ross of Mull

granite. Quarterly Journal of the Geological Society, London 66, 376---399.

Brearley, A. J. & Champness, P. E. (1986). Magnetite exsolution in

almandine garnet. Mineralogical Magazine 50, 621---633.

British Geological Survey (1999). Ross of Mull Sheet. 1:50000 Provisional

Series 43S. Keyworth: British Geological Survey.

Carmichael, D. M. (1969). On the mechanism of prograde meta-

morphic reactions in quartz-bearing pelitic rocks. Contributions to

Mineralogy and Petrology 17, 173---186.Carrington, D. P. & Harley, S. L. (1995). Partial melting and phase

relations in high-grade metapelites: an experimental petrogenetic

grid in the KFMASH system. Contributions to Mineralogy and Petrology

120, 270---291.Cenki, B., Kriegsman, L. M. & Braun, I. (2002). Melt-producing and

melt-consuming reactions in the Achankovil cordierite gneisses,

South India. Journal of Metamorphic Geology 20, 543---561.Chinner, G. A. (1961). The origin of sillimanite in Glen Clova, Angus.

Journal of Petrology 2, 312---323.

Connolly, J. A. D. & Petrini, K. (2002). An automated strategy for

calculation of phase diagram sections and retrieval of rock properties

as a function of physical conditions. Journal of Metamorphic Geology

20(7), 697---708.

Cunningham-Craig, E. H. (1911). South part of the Ross of Mull and

the adjacent islands. In: Cunningham-Craig, E. H., Wright, W. B. &

Bailey, E. B. (eds) The Geology of Colonsay and Oronsay with Part of the Ross

of Mull. Memoir of the Geological Survey of Scotland, 75---100.

Deer, W. A., Howie, R. A. & Zussman, J. (1982). Rock-Forming Minerals,

Volume 1A: Orthosilicates. Harlow: Longman.

Edmunds, W. M. & Atherton, M. P. (1971). Polymetamorphic

evolution of garnet in the Fanad aureole. Donegal, Eire. Lithos 4,

147---161.


852

Dow



Ferry, J. M. & Spear, F. S. (1978). Experimental calibration of the

partitioning of Fe and Mg between biotite and garnet. Contributions to

Mineralogy and Petrology 66, 113---117.

Furlong, K. P., Hanson, R. B. & Bowers, J. R. (1991). Modelling

thermal regimes. In: Kerrick, D. M. (ed.) Contact Metamorphism.

Mineralogical Society of America, Reviews in Mineralogy 26, 437---506.

Halliday, A. N., Aftalion, M., Van Breemen, O. & Jocelyn, J. (1979).

Petrogenetic significance of Rb---Sr and U---Pb isotope systems in the

400Ma old British Isles granitoids and their hosts. In: Harris, A. L.,

Holland, H. D. & Leake, B. E. (eds) Caledonides of the British Isles-----

Reviewed. Geological Society, London, Special Publications 8, 653---661.

Harris, A. L. (1995). Nature and timing of orogenesis in the Scottish

Highlands and the role of the Great Glen Fault. In: Hibbard, J. P.,

van Staal, C. R. & Cawood, P. A. (eds) Current Perspectives in the

Caledonian---Appalachian Orogen. Geological Society of Canada Special Paper

41, 65---79.Hodges, K. V. & Crowley, P. D. (1985). Error estimation and empirical

geothermometry for pelitic systems. American Mineralogist 70,

702---709.

Holdaway, M. J. (1971). Stability of andalusite and the aluminium

silicate phase diagram. American Journal of Science 271, 97---131.

Holdsworth, R. E., Harris, A. L. & Roberts, A. M. (1987). The

stratigraphy, structure and regional significance of the Moine rocks

of Mull, Argyllshire, W. Scotland. Geological Journal 22, 83---107.

Holland, T. J. B. & Powell, R. (1990). An enlarged and updated

internally consistent thermodynamic dataset with uncertainties and

correlations: the system K2O---Na2O---CaO---MgO---MnO---FeO---Fe2O3---Al2O3---TiO2---C---H2---O2. Journal of Metamorphic Geology 8,

89---124.

Holland, T. J. B. & Powell, R. (1998). An internally consistent

thermodynamic data set for phases of petrological interest. Journal of

Metamorphic Geology 16(3), 309---343.

Holland, T. J. B., Babu, E. & Waters, D. J. (1996). Phase relations of

osumilite and dehydration melting in pelitic rocks: a simple

thermodynamic model for the KFMASH system. Contributions to

Mineralogy and Petrology 124(3---4), 383---394.

Holness, M. B. & Clemens, J. D. (1999). Partial melting of the Appin

Quartzite driven by fracture-controlled H2O infiltration in the

aureole of the Ballachulish Igneous Complex, Scottish Highlands.

Contributions to Mineralogy and Petrology 136(1---2), 154---168.

Joesten, R. (1983). Grain growth and grain-boundary diffusion in

quartz from the Christmas Mountains Texas contact aureole.

American Journal of Science 283A, 233---254.

Kerrick, D. M. (1987). Fibrolite in contact aureoles of Donegal,

Ireland. American Mineralogist 72, 240---254.

Kerrick, D. M. (1990). The Al2SiO5 Polymorphs. Mineralogical Society of

America, Reviews in Mineralogy 22, 406 pp.

Kerrick, D. M. (1991). Overview of contact metamorphism. In:

Kerrick, D. M. (ed.) Contact Metamorphism. Mineralogical Society of

America, Reviews in Mineralogy 26, 1---12.

Kretz, R. (1983). Symbols for rock-forming minerals. American

Mineralogist 68, 277---279.

MacKenzie, W. S. (1949). Kyanite gneiss within a thermal aureole.

Geological Magazine 86, 251---255.

MacKenzie, W. S. & Guilford, C. (1980). Atlas of Rock-Forming Minerals

in Thin Section. Harlow: Longman.

Mangan, L. S. (1996). Equilibrium and disequilibrium aspects of

contact metamorphism: the Ross of Mull granite aureole. Ph.D.

thesis, Liverpool University.

Pattison, D. R. M. (1992). Stability of andalusite and sillimanite and the

Al2SiO5 triple point: constraints from the Ballachulish aureole,

Scotland. Journal of Geology 100, 423---446.

Pattison, D. R. M. & Harte, B. (1997). The geology and evolution of

the Ballachulish Igneous Complex and aureole. Scottish Journal of

Geology 33, 1---29.Pattison, D. R. M. & Tracy, R. J. (1991). Phase equilibria and

thermobarometry of metapelites. In: Kerrick, D. M. (ed.) Contact

Metamorphism. Mineralogical Society of America, Reviews in Mineralogy 26,

105---206.

Pattison, D. R. M., Spear, F. S. & Cheney, J. T. (1999). Poly-

metamorphic origin of muscovite plus cordierite plus staurolite

plus biotite assemblages: implications for the metapelitic petroge-

netic grid and for P---T paths. Journal of Metamorphic Geology 17(6),

685---703.

Pattison, D. R. M., Spear, F. S., Debuhr, C. L., Cheney, J. T. &

Guidotti, C. V. (2002). Thermodynamic modelling of the reaction

muscovite plus cordierite ! Al2SiO5 plus biotite plus quartz plus

H2O: constraints from natural assemblages and implications for the

metapelitic petrogenetic grid. Journal of Metamorphic Geology 20(1),

99---118.

Penn, R. L., Banfield, J. F. & Kerrick, D. M. (1999). TEM investigation

of Lewiston, Idaho, fibrolite: microstructure and grain boundary

energies. American Mineralogist 84, 152---159.Potts, G. J., Hunter, R. H., Harris, A. L. & Fraser, F. M. (1995).

Late-orogenic extensional tectonics at the NW margin of the

Caledonides in Scotland. Journal of the Geological Society, London 152,

907---910.Powell, R. & Holland, T. J. B. (1990). Calculated mineral equilibria in

the pelite system, KFMASH (K2O---FeO---MgO---Al2O3---SiO2---

H2O). American Mineralogist 75, 367---380.

Richardson, S. W., Gilbert, M. C. & Bell, P. M. (1969). Experimental

determination of kyanite---andalusite and andalusite---sillimanite

equilibria: the aluminium silicate triple point. American Journal of

Science 267, 259---272.Salje, E. (1986). Heat capacities and entropies of andalusite and

sillimanite: the influence of fibrolitisation on the phase diagram of

the Al2SiO5 polymorphs. American Mineralogist 71, 1366---1371.

Spear, F. S., Kohn, M. J. & Cheney, J. T. (1999). P---T paths from

anatectic pelites. Contributions to Mineralogy and Petrology 134(1), 17---32.

Voll, G., T€oopel, J., Pattison, D. R. M. & Seifert, F. (1991). Equilibrium

and Kinetics in Contact Metamorphism: the Ballachulish Igneous Complex and

its Aureole. Berlin: Springer.

White, R. W., Powell, R. & Holland, T. J. B. (2001). Calculation of

partial melting equilibria in the system Na2O---CaO---K2O---FeO---

MgO---Al2O3---SiO2---H2O (NCKFMASH). Journal of Metamorphic

Geology 19(2), 139---153.

Xu, G., Will, T. M. & Powell, R. (1994). A calculated petrogenetic grid

for the system K2O---FeO---MgO---Al2O3---SiO2---H2O, with particu-

lar reference to contact-metamorphosed pelites. Journal of Metamorphic

Geology 12, 99---119.

Yardley, B. W. D., MacKenzie, J. K. & Guilford, C. (1990). Atlas of

Metamorphic Rocks and their Textures. Harlow: Longman.


853

Dow



Disequilibrium in the Ross of Mull Contact Metamorphic Aureole

Documents

Transcript of Disequilibrium in the Ross of Mull Contact Metamorphic Aureole