Contrib Mineral Petrol (1996) 125: 26–44 C Springer-Verlag 1996

Jagmohan Singh? Wilhelm Johannes

Dehydration melting of tonalites.Part II. Composition of melts and solids

Received: 27 December 1995y Accepted: 7 May 1996

Abstract Dehydration melting of tonalitic composi-tions (phlogopite or biotite-plagioclase-quartz assem-blages) is investigated within a temperature range of700–10008C and pressure range of 2–15 kbar. The solidreaction products in the case of the phlogopite-plagio-clase(An45)-quartz starting material are enstatite, clino-pyroxene and potassium feldspar, with amphiboles oc-curring occasionally. At 12 kbar, zoisite is observed be-low 8008C, and garnet at 9008C. The reaction products ofdehydration melting of the biotite (Ann50)-plagioclase(An45)-quartz assemblage are melt, orthopyroxene,clinopyroxene, amphibole and potassium feldspar. Atpressures. 8 kbar and temperatures below 8008C, epi-dote is also formed. Almandine-rich garnet appearsabove 10 kbar at temperatures$ 7508C. The composi-tion of melts is granitic to granodioritic, hence showingthe importance of dehydration melting of tonalites forthe formation of granitic melts and granulitic restites atpressure-temperature conditions within the continentalcrust. The melt compositions plot close to the cotecticline dividing the liquidus surfaces between quartz andpotassium feldspar in the haplogranite system at 5 kbarandaH2O 5 1. The composition of the melts changes withthe composition of the starting material, temperature andpressure. With increasing temperature, the melt be-comes enriched in Al2O3 and FeO1MgO. Potash in themelt is highest just when biotite disappears. The amountof CaO decreases up to 9008C at 5 kbar whereas at highertemperatures it increases as amphibole, clinopyroxeneand more An-component dissolve in the melt. The Na2Ocontent of the melt increases slightly with increase intemperature. The composition of the melt at tempera-tures . 9008C approaches that of the starting assem-blage. The melt fraction varies with composition andproportion of hydrous phases in the starting composition

as well as temperature and pressure. With increasingmodal biotite from 20 to 30 wt%, the melt proportionincreases from 19.8 to 22.3 vol.% (8508C and 5 kbar).With increasing temperature from 800 to 9508C (at5 kbar), the increase in melt fraction is from 11 to 25.8vol.%. The effect of pressure on the melt fraction is ob-served to be relatively small and the melt proportion inthe same assemblage decreases at 8508C from 19.8 vol.%at 5 kbar to 15.3 vol.% at 15 kbar. Selected experimentswere reversed at 2 and 5 kbar to demonstrate that nearequilibrium compositions were obtained in runs oflonger duration.

Introduction

Magmas formed at mid to deep crustal levels are typical-ly water undersaturated and may be generated by dehy-dration melting of phase assemblages including hydrousminerals like micas (mainly biotite and muscovite) andamphiboles (mainly hornblende) (Wyllie 1977; Clemensand Wall 1981; Clemens 1984; Clemens and Vielzeuf1987; Wall et al. 1987; Stevens and Clemens 1993). Inrecent years, much of the experimental work has beendevoted to the dehydration melting of such composi-tions. These experimental studies can be divided into twogroups; one dealing with dehydration melting of rockscontaining hornblende (amphibolites) as the OH-bearingmineral (see Wyllie and Wolf 1993, for review), and oth-ers with biotite as the main “water-container” (see e.g.,Vielzeuf and Montel 1994; Patin˜o Douce and Beard1995).

The experimental investigations on the biotite-bear-ing systems have been either devoted to stability of bi-otite (phlogopite) (Wood 1972; Bohlen et al. 1983; Mon-tana and Brearly 1989; Peterson and Newton 1989;Vielzeuf and Clemens 1992), or to its fertility as a meltproducer (Le Breton and Thompson 1988; Rutter andWyllie 1988; Vielzeuf and Holloway 1988; Patin˜o Douceand Johnston 1991; Skjerlie and Johnston 1993; Skjerlieet al. 1993; Patin˜o Douce and Beard 1995). The investi-

J. Singh (✉) ? W. JohannesInstitut für Mineralogie, Universität Hannover, Welfengarten 1,D-30167 Hannover, Germany

Editorial responsibility: W. Schreyer

27

gations falling in the second category were started eitherwith a pelitic composition (Le Breton and Thompson1988; Vielzeuf and Holloway 1988; Patin˜o Douce andJohnston 1991), or a tonalitic composition (Rutter andWyllie 1988; Skjerlie and Johnston 1993; Patin˜o Douceand Beard 1995), or both investigated side by side (Skjer-lie et al. 1993). The proportion of melt observed wasdirectly related to the amount of hydrous minerals in thesource material, temperature and amount of water thatcan enter the melt.

Le Breton and Thompson (1988) observed an increasein proportion of melt from 6.1 to 16.5 vol.% at 8608C and10 kbar with an increase in biotite proportion from 28 to55 wt%. Rutter and Wyllie (1988) and Vielzeuf and Hol-loway (1988) detected a steplike increase in the melt pro-portion with corresponding increase in temperature.Patino Douce and Johnston (1991) observed a continuousincrease in melt fraction, e.g., at 10 kbar, increase inmelt from 18 to 60 vol.% with a temperature increasefrom 825 to 9758C. Biotite was completely decomposedat 9758C and the resultant melt productivity curve flat-tened out. The effect of pressure was said to be notstrong. “… the decrease (in melt proportion) at constanttemperature over this pressure range (7–13 kbar) is of atthe most 15 vol.% (absolute)” (Patin˜o Douce and John-ston 1991).

During the dehydration melting experiments on a F-rich tonalitic gneiss, Skjerlie and Johnston (1993) ob-served very low melt proportions (1–10 wt%) before thetemperatures reached 9508C at 6 and 10 kbar pressures.The reason given for very low fractions of melt was highF content of biotite in the starting composition (0.47wt%).

When the metapelite and the F-rich tonalitic gneisswere run side-by-side at 10 kbar, the abundance of meltin the tonalitic gneiss increased by¥10 times and in themetapelite by¥25 vol.% compared to when each wasrun alone (Skjerlie et al. 1993). Skjerlie et al. (1993)suggested that “the large increase in melt fraction intonalite is mainly a result of increased Al2O3 activity inthe melt, which lowers the temperature of the biotite de-hydration melting reaction”. The increase in abundanceof melt in metapelite was suggested to be due to “trans-port of plagioclase component in the melt from thetonalite-layer to the pelite-layer”. Therefore, it was con-cluded that the “rocks (e.g., F-rich tonalitic gneiss)which are poor melt producers on their own can becomevery fertile if they occur in contact with rocks that con-tain components that destabilize the hydrous phase(s)and facilitate dehydration melting” (Skjerlie et al. 1993).

Recently, Patin˜o Douce and Beard (1995) investigateddehydration melting of biotite gneiss (biotite-plagio-clase-quartz-ilmenite) and quartz amphibolite (horn-blende-plagioclase-quartz-ilmenite). The bulk composi-tions of starting materials consisted of biotite (Ann45)and hornblende [Fey(Fe1Mg) 5 0.40], plagioclase(An38), quartz and ilmenite. The melts produced in theexperimental investigation range in composition fromgranitic to granodioritic. No significant difference was

observed between the solidi for dehydration melting ofbiotite- and hornblende-bearing quartz-saturated rocksof comparable Fey(Fe1Mg) ratio (see Patin˜o Douce andBeard 1995, for details). They concluded that the dehy-dration melting “atT # 9008C will generate similar meltfractions (up to¥20%, depending on pressure) for bothtypes of lithologies”, whereas at temperatures$ 9508C,the biotite gneisses can generate 2–3 times more meltthan quartz amphibolites.

Except for the dehydration melting of a metapeliticassemblage consisting of biotite1 plagioclase1 kyanite1 quartz, with and without garnet and K-feldspar, inves-tigated by Le Breton and Thompson (1988), and of syn-thetic biotite gneiss and hornblende gneiss investigatedby Patino Douce and Beard (1995), all other investiga-tions dealing with dehydration melting of biotite-bearingrocks are till now mainly restricted to melting of rockprotoliths with complex mineralogies and containingmore than one hydrous mineral. Besides biotite, the otherhydrous mineral was muscovite in the case of meta-pelites (Vielzeuf and Holloway 1988; Patin˜o Douce andJohnston 1991) or hornblende in tonalites (Rutter andWyllie 1988; Skjerlie and Johnston 1993). The startingrocks originated from high grade metamorphic terranes,with hydrous minerals (in all cases) occurring already ina partially dehydroxylated state. This lead to a consider-able increase of dehydration melting temperatures anddecrease of melt proportions (see discussion in part I). Inone case (Skjerlie and Johnston 1993) the melt propor-tion was as low as 1–2 wt% at 6 kbar and a temperatureas high as 9008C. Due to the presence of more than onehydrous mineral in the source rock, the contribution ofone of them in the production of melt could not clearly bedifferentiated from that of the other. Therefore, we de-cided to investigate the simple tonalitic compositionsphlogopite-plagioclase-quartz and biotite-plagioclase-quartz.

This contribution is mainly devoted to the composi-tion of partial melts and coexisting phases. Reversed ex-periments were performed at 2 and 5 kbar in an effort toestablish equilibrium. Solidus curves for dehydrationmelting of biotite-plagioclase-quartz assemblages arepresented in part I of our investigation on dehydrationmelting of tonalites.

Starting material and experimental procedures

Starting materials in most runs were fine grained mixtures ofsynthetic phlogopite or synthetic biotite (Ann50) having grain sizesup to 10mm, natural plagioclase (20–80mm) of intermediate com-position (An45 Ab52Or3) from Tellnes (South Norway) and naturalquartz (5–20mm) from Göschenen (Swiss Alps). Details of thebiotite syntheses and chemical composition of the plagioclase isgiven in part I. In all except two experiments, the ratio betweenbiotite, plagioclase and quartz was kept constant (2: 5:3), as theyoccur commonly in this ratio in natural tonalites. The compositionof the 2: 5:3 mixture of phlogopite1 plagioclase (An45) 1 quartz(A) and biotite (Ann50) 1 plagioclase (An45) 1 quartz (B) is givenin Table 1. The fine grained mixtures were placed in a 1.5 cm32.8 mm Ag75Pd25 tube, kept overnight open at 1108C and then

28

Table 1 Compositions of starting materials and average composi-tion of tonalites Results

The run conditions and phase assemblages observed inruns performed at and above 8008C are compiled in Table2. For results obtained below 8008C see part I of thisinvestigation. Selected compositions of solids and meltsare given in Tables 3 and 4.

The phase relationships of the two compositions stud-ied, phlogopite 1 plagioclase1 quartz and biotite(Ann50) 1 plagioclase1 quartz, are shown in Figs. 2 and3, respectively. The solidus curves discussed in part I arealso included.

Occurrence and composition of solid reaction products

Clinopyroxene

Clinopyroxene is always observed in the hypersolidusruns with the phlogopite-plagioclase-quartz assemblage.It ranges in composition from augite to diopside(Wo25–50) (Table 3). Clinopyroxene is also detected athypersolidus conditions in the Ann50-bearing assem-blages, but disappears along with potassium feldspar atapproximately 8508C (Fig. 3). Rutter and Wyllie (1988)observed clinopyroxene only after breakdown of horn-blende. Patin˜o Douce and Beard (1995) also observedclinopyroxene in most experiments with the syntheticquartz amphibolite, but with the synthetic biotite gneissonly at 15 kbar and temperatures$ 9508C.

Orthopyroxene

Orthopyroxene is also observed in most hypersolidus ex-periments (see Fig. 1a, b) except for a run performed at12 kbar and 7008C with the phlogopite-plagioclase-

wt% Tonalite A Tonalite B Average(20% Phl150% (20% Ann50150% compositionAn45130% Qtz) An45130% Qtz) of tonalitesa

SiO2 67.51 66.63 66.15TiO2 – – 0.62Al2O3 16.00 15.75 15.56FeO 0.11 4.76 4.64MnO – – 0.08MgO 5.79 2.60 1.94CaO 4.66 4.66 4.65Na2O 2.99 2.99 3.90K2O 2.50 2.27 1.42Total 99.56 99.66 98.96

a Total Nockolds (1954)

immediately welded shut. In addition to the experiments with finegrained mixtures, a few experiments were performed with a singleplagioclase crystal surrounded by quartz-biotite powders (for de-tails of this method see part I).

The experiments were conducted with cold seal pressure ves-sel, internally heated gas pressure apparatus and piston cylinderapparatus. Temperatures, measured with calibrated Ni-CrNi ther-mocouples are accurate within+ 108C. Pressures were measuredwith strain gauge manometers, and are accurate within+ 200 barsat 5 kbar for the hydrothermal and gas pressure apparatus andwithin + 500 bars in piston cylinder apparatus. ThefO2 in the coldseal pressure vessel and internally heated gas pressure apparatuswas controlled by NiNiO buffer (f H2

NNO at 8008C and 5 kbar5 21bar). ThefO2 in piston cylinder lies close to that of CoCoO buffer(f H2

CCOat 8008C and 10 kbar5 305 bar). The experimental productswere polished, mounted on an epoxy and then again polished toprepare thin sections for petrographic and microprobe investiga-tions.

To model the change in composition and proportion of melt,time-, temperature- and pressure-dependent run series were per-formed. The experimental products were analyzed with a Cameca-Camebax microprobe at operating conditions of 15 kV accelerat-ing potential with an 18 nA beam current. The analyses at lowerbeam currents did not yield better results. Large pools of melt wereanalyzed by defocussing the electron beam to approximately20 mm, and by taking shorter times of measurement, 2 secondseach for K2O and Na2O and 5 seconds each for other oxides. Thewater contents of the melts were calculated as the difference be-tween the sum of all oxides and 100. This value was corrected byapplying the correction proposed by Scaillet et al. (1995) based ondifference between the wt% water obtained from microprobe dataagainst wt% water present in synthetic glass samples. The propor-tions of melts were obtained by taking pictures with the Cameca-Camebax microprobe equipped with a slow scan frame grabber(Point Electronic, Halle, Germany) which captures high resolutionTIFF-images. For each run product nearly 50 pictures of the size10243 1024 pixels (8 bit) were taken. These pictures show dis-tinct grey levels for different phase compositions. For modal anal-ysis, grey level thresholds corresponding exclusively to the inter-esting phase (in this case melt) were selected. Wherever grey levelcontrast is not high, filters were used to enhance the picture qual-ity. The point counting was carried out using the computer soft-ware SIS (analySIS, Münster, Germany). The uncertainties in thegiven melt fractions are estimated to be of the order of+ 5 vol.%on the basis of uncertainties in point counting.

Fig. 1a–f Back scattered electron (BSE) images of experimentalrun products:a Obtained at 8508C and 5 kbar, after 6 days withstarting composition biotite (Ann50) 1 plagioclase (An45) 1quartz (melt melt of granitic composition,opx orthopyroxene,En47 Fs52 , pl plagioclase, An63 , qtzquartz).b Obtained at 8508Cand 15 kbar, after 6 days. Starting assemblage: biotite (Ann50) 1plagioclase (An45) 1 quartz (cpxclinopyroxene, Wo43 En30 Fs27 ,grt garnet, Alm55 Pyr17 Gr28 , opx orthopyroxene, En51 Fs45 , plplagioclase, An48 , qtzquartz,meltmelt of granitic composition).c Obtained at 7708C, 5 kbar, after 4 days and started with a singlecrystal of plagioclase (An45), surrounded by titanian biotite(Ann50) and quartz. Composition of product plagioclase deter-mined at various points is as indicated (start plstarting plagioclase(An45), bt biotite, Ann41 , melt melt of granitic composition).dObtained at 8108C and 5 kbar, after 6 days started with a singlecrystal of plagioclase (An45), natural phlogopite (Kragerö) andquartz. White pointsare locations where microprobe analyseswere carried out. The variation in composition of melt observed atthe data points is depicted in Fig. 6. (En enstatite,phl phlogopite,meltmelt of granitic composition,new plnew plagioclase).e Of areversal performed at 2 kbar (see text for details) (bt biotite, Ann48,opx orthopyroxene, En51 Fs47 , pl plagioclase, An45 , qtz quartz).f Of a reversal performed at 5 kbar (see text for details) (bt biotite,Ann48 , opx orthopyroxene, En51 Fs46 , pl plagioclase, An46 , qtzquartz)

a

b

c f

e

d

29

30

Table 2 Experimental results[bt biotite, cpx clinopyroxene,en enstatite,opx orthopyrox-ene,amphamphibole,hblhornblende,grt garnet,phlphlogopite,pl plagioclase,kfsk-feldspar,qtz quartz,m melt]

P (kbar) T (8C) Duration Phases observed

(a) Starting composition: phl1plAn451qtz:

2 790 10 en1cpx1phl1kfs1pl1qtz1m2 810 10 en1cpx1hbl1phl1pl1qtz1m2 850 6 en1cpx1phl1pl1qtz1m2 875 6 en1cpx1phl1pl1qtz1m2 950 6 en1cpx1pl1qtz1m5 790 10 en1cpx1phl1kfs1pl1qtz1m5 810 12 en1cpx1amph1phl1pl1qtz1m5 830 10 en1cpx1amph1phl1pl1qtz1m5 850 6 en1cpx1amph1phl1pl1qtz1m5 900 4 en1cpx1phl1pl1qtz1m5 950 6 en1cpx1pl1qtz1m5 1000 6 en1cpx1pl1m8 800 6 en1cpx1kfs1phl1pl1qtz1m8 850 6 en1cpx1phl1pl1qtz1m

10 800 4 en1cpx1kfs1phl1pl1qtz1m10 850 6 en1cpx1kfs1amph1phl1pl1qtz1m10 900 6 en1cpx1phl1pl1qtz1m10 950 6 en1cpx1pl1qtz1m12 900 6 grt1en1cpx1pl1qtz1m15 800 8 en1cpx1hbl1pl1phl1qtz1m15 850 6 en1cpx1hbl1pl1phl1qtz1m

(b) Starting composition: btAnn501plAn451tz:

2 750 4 opx1cpx1kfs1bt1pl1qtz1m2 800 4 opx1cpx1kfs1bt1pl1qtz1m2 850 4 opx1hbl1pl1qtz1m5 800 4 opx1cpx1hbl1kfs1bt1pl1qtz1m5 850 6 opx1cpx1hbl1kfs1pl1qtz1m5 900 4 opx1hbl1pl1qtz1m5 850a 4 opx1cpx1kfs1pl1bt1qtz1m8 850 6 opx1cpx1hbl1pl1kfs1qtz1m

10 850 6 grt1opx1cpx1pl1kfs1qtz1m10 900 6 grt1opx1pl1qtz1m12 900 12 grt1opx1pl1qtz1m15 800 8 grt1opx1cpx1hbl1kfs1pl1bt1qtz1m15 850 6 grt1opx1cpx1hbl1kfs1pl1bt1qtz1m

(c) Starting composition: btAnn701plAn451qtz:

2 800 4 opx1kfs1bt1pl1qtz1m2 850 4 opx1cpx1kfs1pl1bt1qtz1m5 830 4 opx1hbl1kfs1bt1pl1qtz1m5 900 4 opx1cpx1pl1m8 800 4 cpx1hbl1pl1bt1qtz1m8 850 4 grt1opx1cpx1hbl1kfs1pl1bt1m9 800 4 grt1opx1cpx1hbl1kfs1pl1bt1m

10 800 4 grt1opx1kfs1pl1bt1m15 800 8 grt1opx1cpx1hbl1pl1bt1qtz1m

a With 30wt%bt145wt%pl125wt%qtz

quartz assemblage and at 7008C (8–12 kbar) with thebiotite (Ann50)-plagioclase-quartz assemblage. The or-thopyroxene composition depends on the starting com-position. It is almost pure enstatite in the case of phlogo-pite-bearing assemblage, and the ferrosilite componentincreases with increase in annite component of startingbiotite (Table 3). The orthopyroxenes remain stable with-in the investigated temperature range. In the biotite(Ann50)-plagioclase-quartzassemblage, the enstatite com-ponent of orthopyroxene increases continously with tem-perature (Table 3). At very high temperatures (. 9008C),crystals of orthopyroxene containing up to 13.5 wt%Al2O3 were also observed (Table 3). Patin˜o Douce andBeard (1995) also observed orthopyroxene with Al2O3

ranging from 13.1 to 16.2 wt% as a product of dehydra-tion melting of biotite gneiss at 9258C and 3 kbar.

Amphiboles

The amphiboles are observed at almost all pressure-tem-perature conditions, but are sporadic in occurrence andvery few in number. The composition of amphiboles de-pends on the starting composition. The amphiboles areclassified after Hawthorne (1981) and the nomenclatureis given in Table 3 along with the structural formulae ofamphiboles. Some of the amphiboles belonging to theFe-Mg group are optically recognised as monoclinic

31

Fig. 3 Pressure-temperature diagram showing the phase relation-ships in the case of the biotite(Ann50)-plagioclase-quartz assem-blage as starting material (for determination of solidus and lowtemperature 12 kbar runs see part I). Abbreviations: see Fig. 2

Fig. 2 Pressure-temperature diagram depicting the phase rela-tionships in phlogopite-plagioclase-quartz assemblage as startingmaterial (for determination of solidus see part I). (Abbreviations:phl phlogopite,bt biotite, pl plagioclase,qtz quartz,en enstatite,opx orthopyroxene,cpx clinopyroxene,kfs potassium rich alkalifeldspar,amphamphibole,zo zoisite,ep epidote,grt garnet) Alkali feldspars are one of the major products at hy-

persolidus conditions in the investigated system. Theyrange in composition from Or82Ab18 to almost purepotassium feldspar (Or¥98) (Table 3). The alkalifeldspar-out curve lies at 8008C at 2 and 5 kbar, and above8508C at higher pressures (. 10 kbar) for the phlogopite-plagioclase-quartz assemblage (Fig. 2). In the run prod-ucts of the biotite(Ann50)-plagioclase-quartz assem-blages, alkali feldspar is observed at somewhat highertemperatures (compare Figs. 2 and 3). This is consistentwith the results of Vielzeuf and Clemens (1992) on dehy-dration melting of phlogopite-quartz assemblages, whosuggested a restricted temperature interval over whichphlogopite and alkali feldspar exist together. The largerstability field of alkali feldspar at higher pressures(. 8 kbar) is due to the higher H2OyK2O of the melt ascompared to the lower pressures (Carrington and Watt1995; Patin˜o Douce and Beard 1995).

Plagioclase

Chemical diffusion is very slow in plagioclase feldspars(Baschek and Johannes 1995) and may prohibit a closeapproach to equilibrium compositions in melting experi-ments. In order to get information on the kinetics of pla-gioclase-biotite-quartz melting we performed a singlecrystal experiment at 7708C (5 kbar) with plagioclase(An45) surrounded by quartz and biotite (Ann50). Theresult obtained after 4 days is presented in Fig. 1c. Thenumbers given in Fig. 1c show a continuous increase of

(cummingtonite type, see Table 3), by their oblique ex-tinction and optically positive character. None of the pre-vious investigations dealing with dehydration melting ofbiotite-bearing assemblages reported formation of am-phiboles except for Vielzeuf and Montel (1994) who ob-served gedrite in two runs at 8 kbar. Hornblende was alsoobserved by other authors in their run products (Rutterand Wyllie 1988; Skjerlie and Johnston 1993; Patin˜oDouce and Beard 1995). However, in these cases horn-blende formed part of the starting material.

Alkali Feldspar

Most of the previous investigators either did not reportthe occurrence of alkali feldspar in experiments on dehy-dration melting of rocks with micas and excess quartz(Vielzeuf and Holloway 1988; Peterson and Newton1989; Patin˜o Douce and Johnston 1991), or reported it inonly some of the experiments (Le Breton and Thompson1988; Vielzeuf and Clemens 1992; Vielzeuf and Montel1994). Peterson and Newton (1989) even suggested thatthe alkali feldspar may be a reactant instead of being aproduct. Patin˜o Douce and Beard (1995) observed for-mation of alkali feldspar during crystallization experi-ments with a starting material of glass of biotite gneiss,and neoblastic alkali feldspar was also reported in dehy-dration melting experiments at 12.5 and 15 kbar.

32

Table 3 Representative com-positions of mineral phasesat some selected experimen-tal conditions [FeO* all ironas FeO,FM molFeO*y(FeO*1MgO)]

(a) Clinopyroxene:

Starting composition (i) phl1plAn451qtz

Wt% 2 kbar 5 kbar 8 kbar 10 kbar

8108 C 8508 C 9508 C 8108 C 8308 C 9508 C 8008 C 8508 C 8508 C 9008 C

SiO2 60.14 55.28 57.83 58.71 60.41 58.04 56.47 55.88 55.98 52.03Al2O3 31.13 0.25 1.60 1.73 0.66 2.91 0.81 1.51 33.26 1.92FeO* 0.32 1.84 0.67 0.99 1.68 0.95 0.09 0.25 0.44 8.53MgO 25.26 18.86 19.83 24.29 24.23 18.07 18.32 19.15 18.35 13.64CaO 11.73 23.14 21.17 12.39 11.98 20.89 24.40 22.44 20.88 22.86Na2O 0.15 0.08 0.21 0.34 0.28 0.39 0.30 0.58 0.89 0.35K2O 0.38 0.08 0.35 0.23 0.12 0.61 0.24 0.10 0.43 0.02Total 99.11 99.53 101.67 98.67 99.36 101.84 100.63 99.91 100.23 99.35Wo 0.25 0.45 0.43 0.26 0.25 0.45 0.49 0.46 0.45 0.47En 0.75 0.52 0.56 0.72 0.73 0.54 0.51 0.54 0.55 0.39Fs 0.00 0.03 0.01 0.02 0.02 0.01 0.00 0.00 0.00 0.04

Wt% 12 kbar 15 kbar (ii) btAnn501plAn451qtz

9008 C 8008 C 8508 C 5 kbar 10 kbar 12 kbar 15 kbar

8008 C 8508 C 8508 C 9008 C 8008 C 8508 C

SiO2 52.55 55.34 55.32 51.88 51.49 52.92 52.09 50.82 49.88Al2O3 5.44 6.00 8.70 2.22 0.32 1.63 1.70 2.04 6.82FeO* 0.86 0.84 0.95 9.02 25.46 9.03 8.75 10.93 14.75MgO 16.32 15.33 15.66 13.56 15.06 14.15 13.69 12.46 9.12CaO 23.65 21.09 17.93 22.72 6.84 22.21 23.06 23.22 17.93Na2O 0.62 1.99 1.74 0.46 0.11 0.32 0.35 0.42 1.22K2O 0.06 0.38 1.35 0.00 0.46 0.09 0.03 0.04 0.13Total 99.49 100.97 101.65 99.87 99.37 100.35 99.67 99.93 99.85Wo 0.50 0.49 0.44 0.47 0.14 0.45 0.47 0.48 0.43En 0.48 0.50 0.54 0.39 0.44 0.40 0.39 0.35 0.30Fs 0.02 0.01 0.02 0.14 0.42 0.15 0.14 0.17 0.27

(b) Orthopyroxene:

(i) phl1plAn451qtz

Wt% 2 kbar 5 kbar 8 kbar 10 kbar

8108 C 8508 C 9508 C 8108 C 8308 C 8508 C 9508 C 8508 C 8008 C 8508 C

SiO2 60.84 60.05 60.70 59.51 60.85 59.05 52.07 60.07 59.38 58.12AlsO3 0.57 0.61 0.90 0.58 0.90 2.12 13.41 0.20 0.73 2.05FeO* 0.51 0.26 1.39 0.30 0.41 1.59 1.02 0.88 0.04 1.60MgO 38.17 38.85 36.97 40.44 38.48 37.04 34.16 38.96 38.91 37.07CaO 0.53 0.34 0.98 0.48 1.11 0.95 0.77 0.62 0.43 0.92Na2O 0.08 0.02 0.05 0.00 0.03 0.04 0.06 0.00 0.06 0.06K2O 0.18 0.05 0.14 0.02 0.05 0.27 0.04 0.06 0.76 0.03Total 100.88 100.18 101.12 101.33 101.83 101.16 101.53 100.89 100.41 99.83En 0.98 0.99 0.96 0.99 0.98 0.96 0.97 0.98 0.99 0.96Fs 0.02 0.01 0.04 0.01 0.02 0.04 0.03 0.02 0.01 0.04

An content from the unchanged core to the outer borderof the reaction zone. The change of An is especiallysmall in the outer part of the reaction zone and indicatesthat almost homogeneous plagioclase can be expected inthe products of runs performed at and above 8008C withthe fine grained starting mixtures (the grain size of theplagioclase in our fine grained mixtures is# 10 mm, seeabove).

Plagioclase is observed at all investigated pressure-temperature conditions. However, it changes completely

or partly its composition during partial melting. At lowtemperatures (few degrees above the solidus) plagioclasemay be replaced along the margins by potassiumfeldspar (see Part I). In all runs performed at higher tem-peratures ($ 7808C) and below 10 kbar, plagioclase be-comes enriched in calcium as compared to the startingcomposition, and plagioclase with up to 85 mol% anor-thite component has been observed (Table 3). At pres-sures of 10 kbar and above, the plagioclase compositionof the run product is close to that of the starting material

33

Table 3 (continued)Wt% 10 kbar 12 kbar 15 kbar (ii) btAnn501plAn451qtz

9008 C 9508 C 9008 C 8008 C 8508 C 2 kbar 5 kbar

8508 C 8008 C 8508 C

SiO2 59.63 64.76 58.75 58.11 58.55 52.27 53.07 52.26Al2O3 3.85 2.70 4.08 2.99 2.88 0.59 2.43 0.54FeO* 1.64 1.08 1.79 0.45 1.40 24.40 25.98 29.51MgO 35.69 32.87 36.04 37.87 37.65 21.41 17.79 17.30CaO 0.72 0.78 0.71 0.31 0.20 1.55 1.19 0.25Na2O 0.10 0.12 0.10 0.08 0.04 0.01 0.05 0.01K2O 0.35 0.00 0.06 0.41 0.58 0.03 0.33 0.23Total 101.98 102.31 101.53 100.22 101.20 100.26 100.84 100.10Wo 0.00 0.00 0.00 0.00 0.00 0.03 0.02 0.00En 0.96 0.97 0.96 0.99 0.98 0.59 0.54 0.51Fs 0.04 0.03 0.04 0.01 0.02 0.38 0.44 0.49

Wt% 5 kbar 10 kbar 12 kbar 15 kbar

9008 C 900→ 8008 C 7508 C 9008 C 9008 C 8008 C 8508 C

SiO2 51.21 51.54 50.00 52.57 53.37 50.04 45.59Al2O3 1.62 1.39 7.06 1.38 4.44 2.21 10.31FeO* 28.92 28.11 25.46 22.84 26.14 32.02 29.26MgO 17.45 17.50 15.47 21.78 15.75 14.09 11.53CaO 1.19 1.25 1.55 0.63 2.14 1.73 4.16Na2O 0.01 0.05 0.35 0.02 0.11 0.16 0.04K2O 0.03 0.14 0.14 0.02 0.07 0.06 0.00Total 100.44 99.98 100.03 99.22 102.02 100.31 100.89Wo 0.02 0.03 0.04 0.01 0.06 0.04 0.00En 0.51 0.51 0.50 0.62 0.49 0.42 0.37Fs 0.47 0.46 0.46 0.37 0.47 0.54 0.53

(c) Alkalifeldspar:

(i) phl1plAn451qtz (ii) btAnn501plAn451qtz

Wt% 8 kbar 10 kbar 5 kbar 10 kbar 12 kbar 15 kbar

8508 C 8008 C 8508 C 8008 C 8508 C 9008 C 8508 C 9008 C 8008 C 8508 C→ 8008 C

SiO2 65.96 66.41 64.02 65.92 65.21 64.31 65.51 66.55 63.85 66.55Al2O3 18.47 18.61 19.55 18.58 18.17 18.61 19.01 20.26 18.89 19.41FeO* 0.05 0.05 0.00 0.15 0.83 0.18 0.48 0.21 0.31 0.86MgO 0.37 0.97 0.00 0.00 0.03 0.00 0.00 0.01 0.01 0.38CaO 0.00 0.00 0.28 0.10 0.01 0.25 0.00 0.68 0.09 0.57Na2O 0.71 0.96 2.68 0.10 0.01 0.25 1.34 3.14 1.41 2.35K2O 13.67 14.88 11.89 14.22 15.05 13.83 15.18 10.81 14.83 10.62Total 99.23 101.98 98.42 99.99 99.85 99.26 101.52 101.66 99.38 100.74Ab 0.07 0.09 0.25 0.10 0.03 0.19 0.12 0.30 0.13 0.24Or 0.93 0.91 0.73 0.90 0.97 0.80 0.88 0.67 0.87 0.73An 0.00 0.00 0.02 0.00 0.00 0.01 0.00 0.03 0.00 0.03

(Table 3). However, there is one exception, in a run per-formed at 9008C and 12 kbar with phlogopite-plagio-clase-quartz assemblage, plagioclase with only 27 mol%An is observed coexisting with grossular garnet (Tables 2and 3).

Le Breton and Thompson (1988), Vielzeuf and Hol-loway (1988) and Patin˜o Douce and Beard (1995) alsoobserved more sodic plagioclase at high pressures($ 10 kbar). Skjerlie and Johnston (1993) detected onlya slight increase in the CaO content of plagioclase at veryhigh temperatures (1050–10758C) at 10 kbar. Vielzeuf

and Montel (1994) observed a significant increase in theOr content of the residual plagioclase like in our runswith single plagioclase crystals and pure iron-magne-sium biotites.

Garnet

Almandine-rich garnet forms at and above 10 kbar in theAnn50-tonalitic composition (Alm51–Alm56) (Fig. 1b)and already at 8 kbar and 8508C in iron-richer (Ann70-

34

Table 3 (continued) (d) Plagioclase:

(i) phl1plAn451qtz

Wt% 2 kbar 5 kbar 8 kbar 10 kbar

8108 C 8508 C 9508 C 8108 C 8308 C 9508 C 8508 C 8508 C 9008 C 9508 C

SiO2 55.85 53.47 49.36 49.81 51.88 48.62 58.47 57.17 58.45 57.77Al2O3 27.70 29.46 32.69 31.81 29.45 31.88 27.37 27.23 27.17 26.86FeO* 0.03 0.27 0.19 0.14 0.19 0.19 0.17 0.16 0.19 0.15MgO 0.01 0.15 0.02 0.00 0.00 0.45 0.02 0.02 0.03 0.02CaO 11.40 12.92 16.19 15.52 12.80 16.62 9.14 9.12 9.07 9.17Na2O 2.77 3.31 4.77 2.26 4.08 1.21 4.04 6.27 6.17 6.50K2O 2.10 0.76 0.12 0.32 0.23 0.56 0.48 0.52 0.41 0.47Total 99.94 100.34 100.90 99.86 98.63 99.54 99.69 100.49 101.49 100.94Ab 0.27 0.30 0.21 0.20 0.36 0.11 0.43 0.54 0.54 0.55Or 0.03 0.05 0.00 0.02 0.01 0.04 0.03 0.03 0.02 0.02An 0.60 0.65 0.79 0.78 0.63 0.85 0.54 0.43 0.44 0.43

Wt% 12 kbar 15 kbar (ii) btAnn501plAn451qtz

9008 C 8008 C 8508 C 5 kbar 10 kbar 12 kbar 15 kbar

9008 C 8508 C 9008 C 9008 C 8008 C 8508 C

SiO2 53.69 57.05 57.82 56.89 56.65 56.31 57.17 56.48 57.60Al2O3 28.88 26.95 27.42 27.06 27.60 28.27 28.05 27.65 27.52FeO* 0.03 0.20 0.24 0.30 0.29 0.26 0.26 0.41 0.36MgO 0.02 0.01 0.01 0.00 0.01 0.02 0.05 0.01 0.04CaO 13.53 8.98 9.37 9.79 9.96 10.60 9.91 9.98 9.19Na2O 2.89 6.26 6.17 5.49 5.64 5.52 5.61 6.04 6.41K2O 0.68 0.49 0.55 0.57 0.41 0.41 0.55 0.35 0.46Total 99.72 99.94 101.58 100.09 100.56 101.39 101.60 100.92 101.58Ab 0.69 0.54 0.53 0.49 0.48 0.49 0.49 0.51 0.43Or 0.04 0.03 0.03 0.03 0.02 0.03 0.03 0.02 0.03An 0.27 0.43 0.44 0.48 0.50 0.48 0.48 0.47 0.54

(e) Garnet:

(i) phl1plAn451qtz (ii) btAnn501plAn451qtz

Wt% 12 kbar 10 kbar 12 kbar 15 kbar

9008 C 8508 C 9008 C 7508 C 9008 C 8508 C

SiO2 39.88 38.88 38.67 37.51 38.82 38.35Al2O3 22.49 21.79 21.72 21.37 21.44 21.02FeO* 0.47 25.37 25.65 23.92 24.74 25.38MgO 0.80 5.76 5.72 0.62 5.84 4.43CaO 36.65 8.15 8.57 16.71 9.40 10.23Na2O 0.00 0.02 0.00 0.05 0.00 0.02K2O 0.00 0.05 0.05 0.01 0.08 0.00Total 100.39 100.02 100.38 100.19 100.32 99.93Alm 0.01 0.55 0.55 0.52 0.52 0.55Py 0.03 0.22 0.22 0.02 0.22 0.17Gr 0.96 0.23 0.24 0.46 0.26 0.28

bearing) tonalitic composition (Alm63–Alm68). The al-mandine garnet does not show any marked zonation atany of the investigated conditions. The almandine com-ponent of garnet depends on the composition of the start-ing assemblage. Pyrope was not observed for the startingcomposition phlogopite-plagioclase-quartz within theinvestigated pressure-temperature range, because it isstable only at higher pressures (. 15 kbar, Schreyer

1988). In the case of this assemblage, grossular was ob-served at 9008C and 12 kbar (Table 3).

Formation of garnet in garnet-free starting composi-tions was also reported at high pressure ($ 10 kbar) byLe Breton and Thompson (1988), Rutter and Wyllie(1988), Skjerlie and Johnston (1993) and Patin˜o Douceand Beard (1995). During dehydration melting of meta-greywackes consisting of 32 wt% plagioclase (An22), 25

35

Table 3 (continued)(f) Phlogopitesybiotites:

(i) phl1plAn451qtz

Wt% 5 kbar 8 kbar

7508 C 7708 C 7908 C 8108 C 8308 C 8508 C 9008 C 9508 C 8008 C 8508 C

SiO2 43.85 41.38 41.84 41.29 41.03 46.02 42.78 45.64 46.01 48.10Al2O3 12.32 17.70 17.31 16.89 17.72 13.51 15.93 12.60 12.93 12.29FeO* 0.00 2.91 2.96 3.44 1.95 0.75 2.43 0.52 0.19 0.54MgO 26.72 23.06 24.10 22.86 24.69 28.15 22.97 29.55 24.76 24.85CaO 0.00 0.31 0.02 0.02 0.03 0.08 0.14 0.16 0.05 0.03Na2O 0.14 0.70 0.44 0.51 0.59 0.28 0.59 0.28 0.20 0.23K2O 10.26 8.78 8.89 9.11 8.92 8.72 8.50 6.72 9.35 9.28Total 93.29 94.86 95.57 94.14 94.94 95.51 193.33 95.48 93.49 95.34

Wt% 10 kbar 15 kbar (ii) btAnn501plAn451atz

8008 C 8508 C 9008 C 8008 C 8508 C 5 kbar 15 kbar

8008 C 8508 C 9008 C 8008 C 8508 C→ 8008 C

SiO2 44.55 43.76 46.30 43.88 43.30 41.51 37.83 37.49 38.24 37.84Al2O3 11.71 13.25 12.76 14.19 16.12 12.30 19.66 14.59 15.13 19.36FeO* 0.04 0.45 0.30 1.49 0.81 21.46 20.19 20.40 20.08 17.00MgO 26.90 26.69 29.71 25.36 24.90 12.61 8.52 12.22 12.09 12.64CaO 0.00 0.15 0.08 0.04 0.06 0.36 0.07 0.11 0.11 0.08Na2O 0.15 0.49 0.28 0.40 0.33 0.18 0.13 0.32 0.31 0.32K2O 9.93 8.38 7.56 9.30 9.26 6.53 10.57 8.97 8.01 8.64Total 93.28 93.17 98.65 94.67 94.78 94.94 95.03 94.17 93.97 95.89FM 0.49 0.57 0.48 0.48 0.43

(g) Amphiboles: structural formulae (based on 24 O atoms) and nomenclature of some selected amphibole compositions (Mag-cummmagnesio-cummingtonite,Mag-hblmagnesio-hornblende,Ferri-actino ferri-actinolite,cummcummingtonite,Ferri-tsch ferri-tscher-makitic hornblende)

Cations (i) phl1plAn451qtz (ii) btAnn501plAn451qtz

2 kbar 5 kbar 9 kbar 15 kbar 2 kbar 5 kbar 15 kbar

8108 C 8108 C 8308 C 8508 C 9008 C 8008 C 8008 C 8508 C 8508 C 8008 C 8008 C 8508 C

Si 7.95 6.27 6.23 7.03 7.31 7.71 7.19 6.83 7.17 7.21 7.59 6.29Al IV 0.05 1.73 1.77 0.97 0.69 0.29 0.81 1.17 0.83 0.79 0.39 1.71AlVI 0.13 1.83 2.07 0.83 0.39 0.22 0.63 1.01 0.49 0.33 – 1.18Fe* 0.04 0.15 0.16 0.12 0.26 0.10 0.13 0.16 1.88 3.07 3.22 2.16Mg 5.05 3.58 3.16 5.71 4.31 4.85 3.96 4.32 2.80 3.26 2.71 1.90Ca 1.69 0.72 0.77 0.03 1.82 1.70 1.86 0.03 1.81 0.52 1.16 1.62Na 0.11 0.10 0.44 0.05 0.01 0.30 0.14 1.11 0.30 0.03 0.22 0.51K 0.07 1.08 0.94 1.01 0.09 0.20 0.60 1.01 0.09 0.06 0.07 0.33Nomenclature Tremolite Mag- Mag- Mag- Tremolite Tremolite Mag- Mag- Mag- Cumm- Ferri- Ferri-

cumm cumm cumm hbl cumm hbl actino tsch tsch

wt% biotite (Ann55) and 41 wt% quartz, Vielzeuf andMontel (1994) observed formation of garnet at pressuresas low as 5 kbar.

Zoisite and epidote

At hypersolidus conditions, zoisite first appears at12 kbar in phogopite-bearing, and epidote appears above8 kbar in the Ann50-bearing tonalite (see part I). Epidote

contains a considerable amount of Fe2O3 (1–4.1 wt%,Table 2). The formation of zoisite and epidote is relatedto the breakdown of plagioclase at high pressures (Gold-smith 1981; Johannes 1984).

Biotite

The phlogopite is stable up to 9008C at 2 to 10 kbar, andup to 8508C at 15 kbar (Fig. 2). With increasing tempera-ture, phlogopite becomes deficient in potassium (Fig. 4).

36

Fig. 4 Change in amount of K2O (wt%) of phlogopite with in-crease in temperature at 5 kbar. Starting composition: phlogopite1 plagioclase (An45) 1 quartz. The data for 750 and 7708 C arefrom single crystal experiments described in part I

Fig. 5 Variation in amount of Al2O3 of phlogopite with increase intemperature at 5 kbar in run products with phlogopite1 plagio-clase (An45) 1 quartz assemblage as starting composition. Thedata for 750 and 7708C are from single crystal experiments de-scribed in part I

At 7708C and 5 kbar, the Al2O3 of phlogopite shows nochange, which may be due to unfavorable reaction kinet-ics (Fig. 5). At higher temperatures, phlogopite becomesperaluminous. The Al2O3 content of phlogopite revertsback toward its ideal value (12.2 wt%) before it breaksdown at 9508C and 5 kbar. Aluminium replaces mainlymagnesium, whereas potassium is replaced by sodiumand calcium.

Similarly, biotites show a remarkable change in com-position with increase in temperature. The annite com-ponent of the Fe-bearing biotite decreases with the onsetof melting as well as increase in temperature and it be-comes peraluminous before it breaks down. With in-creasing temperature, Le Breton and Thompson (1988),Patino Douce and Johnston (1991), Skjerlie and Johnston(1993) and Patin˜o Douce and Beard (1995) also observeda decrease in FeO and increase in MgO, TiO2, Al2O3,Na2O and F in biotite.

Oxides

We did not observe formation of magnetite during dehy-dration melting experiments on assemblages containingbiotites of the phlogopite-annite join. However, ilmenitewas observed in experiments with biotite or phlogopitecontaining appreciable amounts of TiO2.

The nature of oxide phases depends upon the compo-sition of the starting material, initial oxidation state ofbiotite and thefO2 prevailing during the experiments

(Vielzeuf and Montel 1994; Patin˜o Douce and Beard1995). The Fe-Ti oxides are the major oxide mineralsobserved during dehydration melting experiments re-ported till now, and their composition varies in runs per-formed in different vessels.

Vielzeuf and Montel (1994) observed magnetite andyor Ti-magnetite in almost all the runs in cold seal andinternally heated vessels, whereas in runs performed inpiston-cylinder apparatus they are absent and insteadpyrrhotite was present. Patin˜o Douce and Beard (1995)observed an increase in amount of hematite in ilmenitefrom 1–7 mol% in runs in the piston cylinder apparatusto 20–63 mol% hematite in runs in internally heated ves-sels. They could correlate the change in oxide composi-tion to the change infO2 conditions between both experi-mental setups. Some part of the melting observed at lowtemperatures by these authors was ascribed to “oxida-tion-melting” leading to formation of magnetite and melt(e.g., Skjerlie and Johnston 1993; Vilezeuf and Montel1994). The experiments were not buffered due to ab-sence of a free vapor phase, but later on considerableinfluence of change infO2 on the composition of runproducts was observed (see Patin˜o Douce and Beard1995).

Reversals at 2 and 5 kbar

Peterson and Newton (1989) and Vielzeuf and Clemens(1992) reversed melting experiments on the phlogopite-

37

quartz assemblage. Peterson and Newton (1989) per-formed reversals at both water-saturated and dehydrationmelting conditions. In the case of the water-saturatedsystem, they performed reversal experiments at 10, 15,and 18 kbar by holding the assemblage at or approxi-mately 108C above the temperatures of previously ob-served melting, followed by lowering the temperaturebelow those of initial melting. They could demonstratereversibility by recrystallization of the partial melt.

In the case of dehydration melting experiments, suchreversals did not give the desired results. Therefore, inanother series of experiments, Peterson and Newton(1989) held the phlogopite-quartz assemblage at 10 kbarand 9508C for 80 h to produce substantial melt. Thecharges in the same run were then held at a lower temper-ature for 200 h. No phlogopite was found to grow above8508C. The temperature of unreacted starting mixture at10 kbar is 7908C. Therefore, the authors delineated atemperature bracket between 790 and 8508C at 10 kbar.

Vielzeuf and Clemens (1992) have termed such rever-sals as “pseudo-reversals” because high temperaturephases (in this case enstatite) also persist along withphlogopite and quartz at lower temperatures. Like Peter-son and Newton (1989), they brought the temperature to9108C at 10 kbar for 7 or 8 days, and then dropped it tolower values. At 8758C, there was no optical evidence ofphlogopite, although some rare phlogopite crystals weredetected with the help of scanning electron micropscopeand microprobe. When the temperature was dropped to8508C, extensive phlogopite growth was observed.Therefore, they located the bracket at 10 kbar between849 and 8768C.

Vielzeuf and Montel (1994) and Patin˜o Douce andBeard (1995) reversed the biotite stability successfullyby crystallizing it from glasses prepared by melting therespective starting materials.

In the present study, two experiments were reversed at2 and 5 kbar with biotite (Ann50)-plagioclase-quartz as-semblage as starting material. The 2 kbar run was firstheld at 8508C for 4 days, and then brought down isobar-icallly to 7508C and held there for 4 days. The runproduct showed all phases present in the direct 7508Crun. Extensive growth of biotite could be observed (seeFig. 1e). The 5 kbar run was held at 9008C for 6 days andthen brought down to 8008C and again held for 6 days.The end products were the same as in the prograde 8008Crun except for clinopyroxene and potassium feldspar.Some biotites are well recrystallized and show plates ap-proximately 10mm thick (Fig. 1f). The new biotite isperaluminous (Al2O3 5 13.3–14.6 wt%) and deficient inK2O (7.8–9 wt%) , which is almost identical to the biotiteobserved in the prograde runs (Table 3).

Composition of melt

As the compositions of solid products vary with the com-position of starting materials and temperature and pres-sure, so do the composition of the melts. A systematic

Fig. 6 Composition of melt changing with increase in distancefrom plagioclase in the run product of a single-crystal plagioclase(An45) surrounded by a mixture of natural phlogopite (Kragerö)and quartz performed at 8108C and 5 kbar, after 6 days

investigation was carried out to elucidate the variation ofcomposition and amount of melt by performing composi-tion-, temperature- and pressure-dependent runs.

First of all a single crystal experiment [a plagioclasecrystal (An45) was surrounded by fine grained naturalphlogopite (Krägero) and quartz] was performed at8108C and 5 kbar for 6 days in order to obtain informationon the influence of diffusion on melt composition.Thebackscattered electron image of the run product is shownin Fig. 1d, and the chemical composition of the melt as afunction of distance from the plagioclase crystal is pre-sented in Fig. 6. The oxides Al2O3, CaO, Na2O and K2Odecrease with increasing distance from the crystal,whereas SiO2 increases. Both FeO and TiO2 remain al-most constant. The data indicate slow diffusion rateswithin the whole system and slow approach to equilibri-um composition. They also show that the physical stateof the starting material influences attainment of equi-librium compositions. Therefore, most experiments wereperformed at higher temperature (800–10008C) with finegrained mixture as starting materials. The distances be-tween the reacting minerals are much shorter than the

38

Table 4 Compositions of themelts at some selected experi-mental conditions for differentstarting compositions. Eachanalysis represents an averageof 5 to 15 analyses. The num-bers refer to the data points inFigs. 7 and 8. (C corundum,Pyx pyroxene). FeO* all ironas FeO

(i) phl1An451qtz

Wt% 2 kbar 5 kbar

1 2 3 4 57708 Ca 8108 C 8508 C 7908 Ca 8108 C

SiO2 68.14 70.93 68.65 70.62 69.58Al2O3 14.86 11.83 14.17 13.43 15.71FeO* 0.07 0.14 0.32 0.34 0.23MgO 0.39 0.59 0.47 0.37 0.60CaO 2.45 0.99 1.52 1.63 2.68Na2O 2.80 2.07 2.65 3.21 2.88K2O 4.86 5.59 5.77 3.70 3.38Total 93.57 92.14 93.55 93.36 94.66H2O 6.43 7.86 6.45 6.64 5.34

CIPW norms

Qtz 29.26 37.31 28.79 35.72 34.97Or 30.72 35.88 36.48 23.42 21.12Ab 25.29 18.99 23.94 29.08 25.68An 13.00 5.33 8.07 8.97 14.08C 0.55 0.60 0.83 1.14 2.14Pyx 1.18 1.88 1.88 1.68 2.02

Wt% 5 kbar 8 kbar 10 kbar

6 7 8 9 10 11 12 138308 C 8508 C 9008 C 9508 C 8008 C 8508 C 8008 C 8508 C

SiO2 70.76 71.02 68.44 66.80 66.60 66.68 69.42 68.96Al2O3 14.72 13.02 15.04 16.62 15.45 15.44 14.19 14.61FeO* 0.15 0.12 0.11 0.18 0.28 0.29 0.22 0.32MgO 0.20 0.73 1.96 2.87 2.08 2.37 2.38 2.52CaO 1.68 1.17 3.33 4.12 3.13 3.39 2.80 3.21Na2O 3.23 2.52 2.94 3.00 3.26 3.70 2.66 3.20K2O 4.27 5.69 4.37 3.73 4.08 5.05 3.23 3.30Total 95.01 94.27 96.19 97.32 94.88 96.92 94.90 96.12H2O 4.99 5.73 3.81 2.67 5.12 3.08 5.10 3.88

CIPW norms

Qtz 33.33 32.77 25.80 22.41 23.16 16.66 33.60 27.95Or 26.57 35.70 26.87 22.67 25.46 30.82 20.10 20.31Ab 28.71 22.59 25.83 26.05 29.08 32.29 23.71 28.14An 8.78 6.16 15.47 21.02 16.23 10.87 14.62 16.35C 1.81 0.60 – 0.14 – – 1.29 –Pyx 0.79 2.17 6.02 7.71 6.07 9.37 6.67 7.25

profile presented in Fig. 6 (by a factor of approximately100) and it is assumed that melt analyses compiled inTable 4 and depicted in Figs. 7 and 8 represent near equi-librium compositions.

A plot of the normative ternary feldspar componentsAb-An-Or (NaAlSi3O8-CaAl2Si2O8-KAlSi3O8) demon-strates that all investigated melts are of granitic or grano-dioritic composition (see Fig. 7). This is in good agree-ment with results published recently by Patin˜o Douceand Beard (1995).

The normative components Qz(SiO2)-Ab-Or of theanalyzed melts plot very close to the cotectic line at5 kbar andaH2O 5 1 after Holtz et al. (1992a) (see Fig.8a,b). The melt compositions obtained in the biotite(Ann50)-plagioclase-quartz starting assemblage areslightly poorer in Qz at high pressures (nos. 21–22),whereas all of the 5 kbar compositions (nos. 15–20) lie

on the Qz side of the cotectic line (Fig. 8b). This differ-ence is due to differences in the total pressure. All datapoints plot on the right side of the H2O-saturated eutecticpoint. This is attributed to low aH2O in the melt and addi-tion of CaO component to the haplogranite system. Lowwater activity and addition of CaO moves the minimummelt composition to the right (James and Hamilton 1969;Holtz et al. 1992a), whereas a slightly peraluminouscharacter of the melts at 5 kbar (Table 4) moves the cotec-tic line upward (Holtz et al. 1992b).

The composition of the melts changes with variationsin the composition of the starting material, temperatureand pressure. These three factors are discussed below indetail. The change in composition of the melt with in-crease in temperature and pressure is discussed only inthe case of the biotite (Ann50)-plagioclase-quartz assem-blage.

39

Table 4 (continued) (ii) Ann501An451Qz

Wt% 2 kbar 5 kbar

14 15 16 17 188508 C 8008 C 8508 Cb 8508 Cb 8508 Cc

SiO2 69.55 68.94 68.44 68.32 69.34Al2O3 14.44 14.94 14.15 14.22 13.31FeO* 2.60 2.39 2.74 3.70 4.28MgO 0.53 0.78 1.26 1.72 1.84CaO 2.71 4.01 3.77 3.26 2.58Na2O 3.03 2.68 2.61 2.47 2.20K2O 3.78 3.23 3.09 3.79 4.55Total 96.64 96.97 96.06 97.48 98.10H2O 3.36 3.03 3.94 2.52 1.90

CIPW norms

Qtz 29.73 30.33 30.62 27.51 27.81Or 23.13 19.68 19.03 23.00 27.43Ab 26.51 23.36 22.96 21.42 18.95An 13.90 19.78 18.45 16.60 13.06C 0.44 – – 0.11 0.06Pyx 6.29 6.86 8.94 11.37 12.69

Wt% 5 kbar 10 kbar 15 kbar

19 20 21 229008 C 9508 C 8508 C 8508 C

SiO2 69.28 66.68 64.60 67.23Al2O3 14.47 15.46 16.19 14.91FeO 3.18 3.56 2.50 2.73MgO 1.37 1.60 1.17 1.58CaO 2.79 3.96 4.59 3.45Na2O 2.45 3.15 3.13 3.32K2O 3.94 2.64 3.47 4.26Total 97.48 97.05 95.65 97.48H2O 2.52 2.95 4.35 2.52

CIPW norms

Qtz 30.04 25.17 21.01 20.99Or 23.91 16.09 21.46 25.85Ab 21.24 27.43 27.66 28.78An 14.21 20.26 20.73 13.49C 1.11 0.21 – –Pyx 9.49 10.85 9.14 10.89

Effect of starting composition

At temperatures around 8508C and 5 kbar pressure, themelt in the case of the biotite (Ann70)-plagioclase-quartzassemblage contains more normative corundum(C 5 1.42 at 8308C and 5 kbar), as compared to melt inthe the biotite (Ann50)-plagioclase-quartz assemblage(C 5 0.11 at 8508C, 5 kbar) (Table 4). The melts in thephlogopite-plagioclase-quartz assemblage containC 5 1.81 and 0.60 at 8308C and 8508C, respectively at5 kbar (Table 4). They plot mostly within the granite fieldof the Ab-Or-An diagram (Fig. 7b). At 8508C and 5 kbar,with increase in proportion of biotite (Ann50) from 20 to30 wt%, the FeO, MgO and K2O of the melts increase,whereas CaO contents decrease (see points 17 and 18 inTable 4 and Figs. 7b and 8b).

Effect of temperature

Net compositions of the melts change with temperature.The first melts observed contain more water compared tomelts observed at higher temperatures (Table 4). The ini-tial melts are also relatively rich in normative Qz (Table4). With increasing temperature, potassium feldspar, bi-otite and more quartz dissolve in the melt and it becomesenriched in K2O and SiO2 until temperatures of 9008Care reached at 5 kbar (Fig. 9). On the contrary, the CaOcontent of the melt decreases up to 9008C and then slight-ly increases as more An component dissolves in the meltat high temperature (Fig. 9). The Na2O content of themelt increases very slightly and the Al2O3 abundanceincreases continuously with increasing temperature.With increasing temperature, the melt also becomes en-riched in FeO and MgO (Fig. 9). This is attributed to

40

Table 4 (continued) (iii) Ann701An451qtz

Wt% 2 kbar 5 kbar 8 kbar 9 kbar 10 kbar

23 24 25 26 27 28 29 308008 C 8508 C 8308 C 9008 C 8008 C 8508 C 8008 C 8008 C

SiO2 70.80 69.09 68.51 70.15 69.63 70.20 69.53 70.35Al2O3 13.58 14.11 15.31 14.32 16.12 15.30 14.30 14.20FeO* 2.17 2.18 2.26 3.01 1.73 2.24 1.65 1.76MgO 0.46 0.23 0.51 0.36 0.22 0.41 0.42 0.22CaO 2.14 1.91 3.12 2.61 2.60 2.31 2.62 2.55Na2O 2.32 2.99 2.98 2.30 2.64 2.10 2.25 2.08K2O 3.35 5.42 3.10 5.72 3.74 5.52 3.61 3.64Total 94.82 95.93 95.79 98.47 96.68 98.08 94.38 94.80H2O 5.18 4.07 4.21 1.53 3.32 1.92 5.62 5.20

CIPW norms

Qtz 39.46 25.87 31.32 26.91 33.76 30.01 37.11 39.10Or 20.90 33.42 19.14 34.36 22.88 33.29 22.62 22.71Ab 20.68 26.34 26.29 19.74 23.08 18.10 20.15 18.54An 11.20 9.40 16.17 11.98 13.35 11.69 13.78 13.35C 2.35 0.00 1.42 0.00 3.09 1.68 2.02 2.30Pyx 5.41 4.97 5.66 7.00 3.85 5.23 4.32 3.98

a Results with single crystal experimentsb Melt composition of run repeated at samePT conditionsc With 30 wt% bt145 wt% pl125 wt% qtz

breakdown of biotite and other ferromagnesium mineralssuch as hornblende and pyroxenes at higher tempera-tures. The melt compositions at temperatures. 9008Capproach the composition of the starting material (com-pare data given in Tables 1 and 4, and star and point no. 20in Fig. 7b).

During dehydration melting of the F-rich tonaliticgneiss, Skjerlie and Johnston (1993) observed increasesin FeO, MgO, TiO2, K2O and F with increase in tempera-ture, whereas CaO was observed to remain constant ex-cept above 10508C.

Effect of pressure

With increase in pressure, the melt at 8508C in the caseof the biotite (Ann50)-plagioclase-quartz assemblage be-comes slightly depleted in K2O and SiO2, and other ox-ides show a slight increase (Table 4). Skjerlie and John-ston (1993) also observed an increase in FeO and CaO,and decrease in K2O of melt compositions in dehydrationmelting experiments on tonalitic gneiss with increasingpressure from 6 to 10 kbar. Due to very few and smallvolumes of melt at 8508C at high pressures (especially at15 kbar), no clear compositional trends can be deducedon the basis of the present study. Patin˜o Douce and Beard(1995, in press) have investigated the effect of pressureon composition of melts in detail. According to them,crystallization of garnet during dehydration melting ofless aluminous metagreywackes and tonalitic gneissesrequires incongruent breakdown of anorthite in order tosupply the necessary Al. This reaction liberates Na2O,and hence increases the Na2OyK2O ratio of the melts.

Proportion of melt

In a series of time-dependent runs with the bio-tite (Ann50)-plagioclase-quartz assemblage at 8508Cand 5 kbar, an increase in proportion of melt from 15.5to 19.8 vol.% was observed when the run durationwas increased from 3 to 6 days. An increase in runduration from 6 to 12 days does not make a big dif-ference; the increase in melt proportion observed is upto 21.5 vol.%. Therefore, all other runs to investigatemelt proportions were performed for a duration of 6days.

Besides run duration, the melt proportion also de-pends upon the amount and composition of the hydrousphase in the starting composition, and on temperatureand pressure. At 8508C and 5 kbar, the melt fraction in-creased from 19.8 to 22.3 vol.% when the proportion ofbiotite (Ann50) was increased from 20 to 30 wt%. At theapplied pressure-temperature conditions (8508C,5 kbar), biotite (in the assemblage with 30 wt% biotite) isstill present whereas biotite in the assemblage with 20wt% biotite has already disappeared. When the 20 wt%biotite (Ann50) was replaced by the same proportion ofphlogopite, the melt fraction observed is similar (22vol.%). Although some phlogopite is still present at8508C, it has to be considered that the H2O content of thestarting phlogopite is higher than that of the starting bi-otite (4.3 wt% instead of 3.5 wt%).

In a series of temperature-dependent runs with thebiotite (Ann50)-plagioclase-quartz assemblage at 5 kbar,an increase in melt proportion from 11 vol.% at 8008C to25.8 vol.% at 9508C was observed (Fig. 10). The largestincrease detected was between 8008C (11 vol.%) and8508C (19.8 vol.%) when the biotite disappeared. Due to

41

Fig. 7a, b Normative Ab, Or, An values of melts for:a startingcomposition: phlogopite1 plagioclase (An45) 1 quartz (circles);b starting composition: biotite (Ann50)-plagioclase-quartz (dots)and biotite (Ann70)-plagioclase-quartz (triangles). The f ields forvarious rock types are delineated after Barker (1979).Star in brepresents Ab, Or, An values of starting composition biotite(Ann50)-plagioclase (An45)-quartz

Fig. 8a, b Normative Qz, Ab, Or values of melts for:a startingcomposition: phlogopite1 plagioclase (An45) 1 quartz;b startingcomposition: biotite (Ann50)-plagioclase-quartz (dots) and biotite(Ann70)-plagioclase-quartz (triangles). The cotecticlines for thehaplogranite system at 5 kbar andaH2O 5 1 are marked after Holtzet al. (1992a)

this, at temperatures above 8508C the temperature-meltproportion curve flattens out (Fig. 10).

With the increase in pressure from 5 to 15 kbar, themelt proportion decreases from 19.8 vol.% at 5 kbar to18.5 vol.% at 8 kbar and further to 16.6 vol.% at 10 kbar.With further increase in pressure to 15 kbar, a melt frac-tion of 15.3 vol.% is observed (Fig. 11). This is a result oftwo opposing effects: dehydration melting begins at low-

er temperatures at high pressures (see Part I) and on theother side the proportion of the partial melt decreaseswith increasing pressure at given constant proportion ofwater (Holtz and Johannes 1994). Patin˜o Douce (inpress) and Patin˜o Douce and Beard (in press) observedthat the starting assemblages giving rise to garnet at highpressures as a product of incongruent melting of plagio-clase will liberate albite causing increases in the melt

42

Fig. 9 Variation of melt com-positions with increasing tem-perature at 5 kbar in the caseof runs with biotite (Ann50)-plagioclase-quartz as startingcomposition

Fig. 11 Variation in melt proportion with pressure at 8508C. Thestarting composition is biotite (Ann50) 1 plagioclase (An45) 1quartz

Fig. 10 Increase of melt proportion with temperature at 5 kbar inbiotite (Ann50)-plagioclase- quartz starting assemblage

fraction at high pressures. This may be a plausible reasonfor the not so significant effect of pressure on melt pro-portions in our experiments.

Melt proportion: comparison with literature

Clemens and Vielzeuf (1987) modeled melt proportionsat 5 and 10 kbar formed by dehydration melting of pelitesand quartzofeldspathic rocks. One of their quartzo-feldspathic rocks consists of biotite, plagioclase andquartz (QF-B, Clemens and Vielzeuf 1987), which is

quite similar to the tonalitic composition biotite (Ann50)-plagioclase-quartz in this investigation. At 5 kbar,Clemens and Vielzeuf (1987) deduced melt proportionsfor such a rock for given amounts of water. For instance,the melt proportion deduced for a quartzofeldspathicrock with 0.7 wt% water (i.e., 20 wt% Ann50) was nearly30 vol.% at 8508C. This melt proportion is somewhathigher as compared to our data at 5 kbar (19.8 vol.%melt, see Fig. 10).

At 10 kbar, Clemens and Vielzeuf (1987) estimatedmelt proportions of nearly 15 vol.% at 8508C and22.5 vol.% at 9008C, which are quite close to the

43

Concluding remarks

Dehydration melting of tonalites leads to formation ofmelts ranging in composition from granodiorite to gran-ite (Fig. 7a, b). The residual phases orthopyroxene,clinopyroxene, plagioclase and K-feldspar dominate atlow pressures (# 10 kbar). At pressures above 9 kbar inthe starting composition biotite (Ann50) 1 plagioclase(An45) 1 quartz, garnet is also observed, whereas, epi-dote is formed first above 8 kbar and orthopyroxene isabsent at low temperatures (# 7008C). In the case ofbiotite (Ann70) 1 plagioclase (An45) 1 quartz, garnetappears at an even lower pressure of 8 kbar (at 8508C).Garnet and zoisite are first observed at 12 kbar in thecase of phlogopite1 plagioclase (An45) 1 quartz assem-blage (Fig. 2). The orthopyroxene in the case of the iron-free assemblage was enstatite, whereas ferrosilite com-ponent in the iron-bearing assemblages increased withincrease in the annite component in the starting assem-blage. Therefore, depending on the composition of thestarting material, a pyroxene-rich restite at lower pres-sures and a garnet1 epidote-bearing restite at higherpressures is observed at different levels in the continentalcrust. This has important implications for the formationof granulites rich in pyroxenes at shallower depths andthose consisting of garnet and epidote at greater depths.The difference infO2 in low and high pressure experi-ments with cold seal pressure vessels and internally heat-ed pressure apparatus (withfO2

controlled by NiNiObuffer) and piston cylinder apparatus (withfO2

close toCoCoO buffer) did not have any observable influence onthe stability of garnet as also observed by Patin˜o Douceand Beard (in press).

The increasing pressure does not have a significantinfluence on the composition of melts in the investigatedcompositions. With increasing pressure, the melt be-comes granodioritic (Fig. 7a, b), i.e., an increase in theAb and An component. Patin˜o Douce (in press) and Pati-no Douce and Beard (1996) observed a similar effect onthe composition of melts in dehydration melting experi-ments on model greywacke and metapelites. At 5 kbar,with increasing temperature, the composition of the meltapproaches that of the starting assemblage (compare starand point no. 20 in Fig. 7b).

The experimental runs were reversed at 2 and 5 kbar toestablish equilibrium. Reversals performed by firstcrossing the stability field of biotite and then loweringthe temperature into the biotite stability field proved tobe successful. The composition of plagioclase in a singlecrystal run was observed to change from an unchangedcore to An-richer rim. At the rim of the plagioclase crys-tal, the composition was observed to remain constant,thus indicating an approach to equilibrium.

Most of the runs were performed within the tempera-ture range 750–9008C, a range commonly reached dur-ing high grade metamorphism in the lower continentalcrust. An amount of 11 vol.% of melt was already ob-served at a temperature of 8008C and a pressure of 5 kbar.This is more melt at these pressure-temperature condi-tions than so far reported in the literature on dehydration

Fig. 12 Comparison of melt fractions obtained at 10 kbar duringdehydration melting of tonalites and biotite (Ann50) 1 plagioclase(An45) 1 quartz assemblage.R & WRutter and Wyllie (1988),S &J Skjerlie and Johnston (1993)

values observed in our runs (16.6 and 22 vol.%, repec-tively).

Rutter and Wyllie (1988) and Skjerlie and Johnston(1993) reported different melt proportions in the dehy-dration melting experiments on tonalites due to differentcompositions of the starting materials. Rutter and Wyllie(1988) conducted dehydration melting experiments on atonalite consisting of 12.5 vol.% biotite and 9 vol.%hornblende, and Skjerlie and Johnston (1993) used atonalitic gneiss containing 20 wt% biotite and 2 wt%hornblende. In both cases the proportion of hydrous min-erals is close to that in the present investigation (20 wt%biotite).

Rutter and Wyllie (1988) observed a steplike increasein the melt proportion at 10 kbar between 850 and 9008C,reaching 22 vol.% at 9008C, when the biotite disappeared(see Fig. 12). The second step was defined by the disap-pearance of hornblende at 10008C and an increase in meltproportion to 35 vol.%.

Skjerlie and Johnston (1993) determined very lowamounts of melt until temperature crossed 9508C and thebiotite disappeared at 6 and 10 kbar. The reason givenwas the high thermal stability of biotite due to its highfluorine content (0.47 wt%). At 10 kbar, hornblende dis-appeared between 950 and 9758C, whereas biotite is ob-served till 10008C.

A comparison between the melt fractions obtained byRutter and Wyllie (1988), Skjerlie and Johnston (1993)and in the present investigation at 10 kbar is shown inFig. 12. Our melt fractions at 8508C are higher than theothers. At 9008C, our melt proportion lies close to that ofRutter and Wyllie (1988).

44

melting of tonalites (Rutter and Wyllie1988; Skjerlie andJohnston 1993; Patin˜o Douce and Beard 1995, in press).

The melts observed plot near the cotectic line dividingthe liquidus surfaces between quartz and potassiumfeldspar in the haplogranitic system at 5 kbar andaH2O 51 (Fig. 8). Compared to the water-saturated eutecticcomposition of the system Qz-Ab-Or, the composition ofmelts in the present investigation at 5 kbar are richer inSiO2 and have a higher OryAb ratio (Fig. 8). The higherOryAb ratio may be explained by the presence of Ancomponent, and by the fact that the system is water un-dersaturated.

The results obtained in the present study support themodel that tonalites may be the protoliths for graniticrocks intruded into the upper continental crust and thepyroxene-rich granulites, now being main constituentsof the lower continental crust.

Acknowledgements The authors are thankful to Jürgen Koepkefor help during microprobe analyses and Carsten Wolke in ex-periments with piston cylinder apparatus. Thanks are also due toCatherine McCammon who carried out the Mössbauer spec-troscopy of synthetic biotites and Otto Diedrich for preparing thethin sections. The manuscript benefited from reviews by A.D.Johnston and many critical remarks by A.E. Patin˜o Douce. Thiswork was supported by DAAD and Deutsche Forschungsgemein-schaft (SFB 173).

References

Barker F (1979) Trondhjemite: definition, environment and hy-potheses of origin. In: Barker F (ed) Trondhjemites, dacites,and related rocks. (Developments in petrology, 6). Elsvier,Amsterdam, pp 1–12

Baschek G, Johannes W (1995) The estimation of NaSi-CaAl in-terdiffusion rates in peristerites by homogenization experi-ments. Eur J Mineral 7: 295–307

Bohlen SR, Boettcher AL, Wall VJ, Clemens JD (1983) Stabilityof phlogopite-quartz and sanidine-quartz: a model for meltingin the lower crust. Contrib Mineral Petrol 85: 270–277

Carrington DP, Watt GR (1995) A geochemical and experimentalstudy of the role of K-feldspar during water undersaturatedmelting of metapelites. Chem Geol 122: 59–76

Clemens JD (1984) Water contents of intermediate to silcic mag-mas. Lithos 17: 273–287

Clemens JD, Vielzeuf D (1987) Constraints on melting and mag-ma production in the crust. Earth Planet Sci Lett 86: 287–306

Clemens JD, Wall VJ (1981) Origin and crystallization of someperaluminous (S-type) granitic magmas. Can Mineral 19: 111–131

Goldsmith JR (1981) The join CaAl2Si2O8-H2O (anorthite-water)at elevated pressures and temperatures. Am Mineral 66: 1183–1188

Hawthorne FC (1981) Crystal chemistry of the amphiboles. In:Veblen DR (ed) Amphiboles and other hydrous pyriboles-min-eralogy. (Reviews in mineralogy, 9A) Mineral Soc Am, Wash-ington, DC, pp 1–102

Holtz F, Johannes W (1994) Maximum and minimum water con-tents of granitic melts: implications for chemical and physicalproperties of ascending magmas. Lithos 32: 149–159

Holtz F, Pichavant M, Barbey P, Johannes W (1992a) Effects ofH2O on liquidus phase relations in the haplogranite system at 2and 5 kbar. Am Mineral 77: 1223–1241

Holtz F, Johannes W, Pichavant M (1992b) Peraluminous granites:the effect of alumina on melt composition and coexisting min-erals. Trans R Soc Edinburgh Earth Sci 83: 409–416

James RS, Hamilton DL (1969) Phase relations in the systemNaAlSi3O8-KAlSi3O8-CaAl2Si2O8-SiO2 at 1 kilobar waterpressure. Contrib Mineral Petrol 21: 111–141

Johannes W (1984) Beginning of melting in the granite systemQz-Or-Ab-An-H2O. Contrib Mineral Petrol 86: 264–273

Le Breton N, Thompson AB (1988) Fluid-absent (dehydration)melting of biotite in metapelites in the early stages of crustalanatexis. Contrib Mineral Petrol 99: 226–237

Montana A, Brearley M (1989) An appraisal of the stability ofphlogopite in the crust and in the mantle. Am Mineral 74: 1–4

Nockolds SR(1954) Average chemical composition of some ig-neous rocks. Bull Geol Soc Am 65: 1007–1032

Patino Douce AE (1996) Effects of pressure and H2O content onthe compositions of primary crustal melts. Trans R Soc Edin-burgh 3rd Hutton Symp (in press)

Patino Douce AE, Beard JS (1995) Dehydration melting of biotitegneiss and quartz amphibolite from 3 to 15 kbar. J Petrol36: 707–738

Patino Douce AE, Beard JS (1996) Effects ofP, f(O2) and MgyFeratio on dehydration melting of model greywackes. J Petrol (inpress)

Patino Douce AE, Johnston AD (1991) Phase equilibria and meltproductivity in the pelitic system: implications for the origin ofperaluminous granitoids and aluminous granulites. ContribMineral Petrol 107: 202–218

Peterson JW, Newton RC (1989) Reversed experiments on biotite-quartz-feldspar melting in the system KMASH: implicationsfor crustal anatexis. J Petrol 97: 465–485

Rutter MJ, Wyllie PJ (1988) Melting of vapour-absent tonalite at10 kbar to simulate dehydration melting in the deep crust.Nature 331: 159–160

Scaillet B, Holtz F, Rouer O, Pichavant M (1995) Determiningtotal water contents in silicate glasses with the electron mi-croprobe: an experimental calibration of the by-differencemethod (unpublished)

Schreyer W (1988) Experimental studies on metamorphism ofcrustal rocks under mantle pressures. Mineral Mag 52: 1–26

Skjerlie KP, Johnston AD (1993) Fluid-absent melting behaviourof an F-rich tonalitic gneiss at mid-crustal pressures: implica-tions for the generation of anorogenic granites. J Petrol34: 785–815

Skjerlie KP, Patin˜o Douce AE, Johnston AD (1993) Fluid absentmelting of a layered crustal protolith: implications for the gen-eration of anatectic granites. Contrib Mineral Petrol 114: 365–378

Stevens G, Clemens JD (1993) Fluid-absent melting and the rolesof fluids in the lithosphere: a slanted summary? Chem Geol108: 1–17

Vielzeuf D, Clemens JD (1992) The fluid-absent melting ofphlogopite1 quartz: experiments and models. Am Mineral77: 1206–1222

Vielzeuf D, Holloway JR (1988) Experimental determination ofthe fluid-absent melting relations in the pelitic system. ContribMineral Petrol 98: 257–276

Vielzeuf D, Montel JM (1994) Partial melting of metagreywackes.1. Fluid-absent experiments and phase relationships. ContribMineral Petrol 117: 375–393

Wall VJ, Clemens JD, Clarke DB (1987) Models for granitoidevolution and source compositions. J Geol 95: 731–749

Wood BJ (1972) The reaction phlogopite1 quartz5 enstatite1sanidine1 H2O. In: McKenzie WS (ed) Progress in experi-mental petrology: third progress report of research supportedby NERC. Nat Environ Res Counc Publ Ser D 6, pp 17–19

Wyllie PJ (1977) Crustal anatexis: an experimental review. Tec-tonophysics 43: 41–71

Wyllie PJ, Wolf MB (1993) Amphibolite dehydration melting:sorting out the solidus. In: Pritchard HM, et al (eds) Magmaticprocesses and plate tectonics. Geol Soc Spec Publ 76: 405–416