Skip Navigation


Journal of Petrology Advance Access originally published online on July 8, 2004
Journal of Petrology 2004 45(9):1777-1797; doi:10.1093/petrology/egh033
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
45/9/1777    most recent
egh033v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (7)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by BUICK, I. S.
Right arrow Articles by GIBSON, R. L.
Right arrow Search for Related Content
GeoRef
Right arrow GeoRef Citation
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Journal of Petrology 45(9) © Oxford University Press 2004; all rights reserved

The Role of Water Retention in the Anatexis of Metapelites in the Bushveld Complex Aureole, South Africa: an Experimental Study

I. S. BUICK1,*, G. STEVENS2 and R. L. GIBSON3

1 DEPARTMENT OF EARTH SCIENCES, LA TROBE UNIVERSITY, BUNDOORA, VIC. 3086, AUSTRALIA
2 DEPARTMENT OF GEOLOGY, UNIVERSITY OF STELLENBOSCH, STELLENBOSCH, PRIVATE BAG X1, MATIELAND 7602, SOUTH AFRICA
3 SCHOOL OF GEOSCIENCES, UNIVERSITY OF THE WITWATERSRAND, PRIVATE BAG 3, WITS 2050, JOHANNESBURG, SOUTH AFRICA

RECEIVED MAY 22, 2003; ACCEPTED MARCH 26, 2004


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 LOCAL GEOLOGY
 EXPERIMENTAL AND ANALYTICAL...
 STARTING MATERIALS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Highly restitic metapelites occur at the contact of the Rustenburg Layered Suite (Bushveld Complex). On the basis of previous experimental studies, the high (~60%) degrees of melting required to form these restites via fluid-absent processes could not have occurred at the maximum temperatures estimated in the aureole (~800°C). We have investigated their formation by partial melting experiments using a medium-grade and a low-grade metapelitic hornfels from the aureole that are isochemical except for water content (~1·2 and ~4·4 wt % H2O, respectively). The partial melting experiments (700–1000°C at 3 kbar) demonstrate strongly contrasting behaviour for the two samples. For the higher-grade starting composition only minor melting occurred up to 800°C (<~13%); extensive melting (>50%) required a temperature of 1000°C. In contrast, for the lower-grade starting composition, extensive melting (~50–65%) and the presence of a similar restite assemblage to that observed in the aureole rocks was achieved in all experiments run at temperatures ≥750°C. In the absence of fluid infiltration, the results suggest that during rapid heating in the innermost parts of high-temperature contact aureoles, extensive partial melting may occur at relatively low temperatures if the fluid from the low-grade protolith is still resident in the system. This locally renders metapelitic rocks less dehydrated and, hence, more fertile than if heating had been slower.

KEY WORDS: low-pressure; melting experiments; metapelites


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
LOCAL GEOLOGY
 EXPERIMENTAL AND ANALYTICAL...
 STARTING MATERIALS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The nature of partial melting of initially low-grade or unmetamorphosed pelitic rocks in high-temperature contact metamorphic aureoles around mafic intrusions may be expected to differ significantly from that produced during regional metamorphism in the middle and lower crust. In particular, heating to peak temperatures above the solidus will be much more rapid, and the duration of high-temperature metamorphism and anatexis much shorter, than during regional high-grade metamorphism. For example, conductive thermal models for heating adjacent to high-temperature intrusions predict that country rocks within tens to hundreds of metres from such intrusions reach peak temperatures above the wet solidus for quartzofeldspathic rocks on a time scale of ~102–104 years (Kerrick et al., 1991Go; Johnson et al., 2003Go; Dutrow et al., 2004Go). This suggests that processes such as reaction overstepping during heating, leading to the progress of metastable sub-solidus and melting reactions (Tracy & McLellan, 1985Go; Rubie & Brearley, 1987Go), may be important in high-temperature contact aureoles (Kerrick et al., 1991Go).

The relationships between peak metamorphic temperature, heating rate and the duration of the thermal peak create potentially complex partial melting scenarios that may change as a function of distance from the intrusion. In the absence of fluid infiltration, the more slowly heated, outermost parts of anatectic zones around mafic intrusions would be expected to experience limited melting at the wet-pelite solidus, followed by incongruent melting of biotite-bearing assemblages. For low-pressure aureoles (<4 kbar), melting is unlikely to involve muscovite because of its prior sub-solidus breakdown (Holland & Powell, 2001Go). Overall, the melting history would be similar to that experienced in regional low-P–high-T environments. In contrast, in the innermost, very rapidly heated parts of the aureole: (1) overstepping of muscovite sub-solidus dehydration reactions and the persistence of muscovite above the wet-pelite solidus may allow metastable incongruent melting of muscovite (Rubie & Brearley, 1987Go; Brearley & Rubie, 1990Go); and/or (2) fluids released through prograde devolatilization reactions that had not had time to leave the system could flux extensive melting at the wet-pelite solidus.

In natural rocks there is likely to be an interplay between the possibilities discussed above. For example, rapidly heated rocks that start melting via fluid-present processes at the wet-pelite solidus may also contain metastable muscovite that ultimately melts via a metastable incongruent reaction. The composition of most metapelites will ensure that fluid from sub-solidus reactions will be consumed first by melting reactions before the rocks become quartz- or mica-absent above the wet-pelite solidus. This will ensure that further melting will occur through quartz-consuming, fluid-absent incongruent melting reactions of muscovite (metastable) or biotite. The variation in heating rates within high-temperature contact aureoles (Kerrick et al., 1991Go) might be expected to cause melting histories that differ significantly with distance from the contact.

In this study, we present the results of an experimental investigation of low-pressure (3 kbar) melting in metapelitic rocks designed to investigate these potential variations, and to explain the development of relatively low-temperature (~800°C), quartz-free, cordierite- and spinel-rich restitic metapelites developed at the contact of the Rustenburg Layered Suite, Bushveld Complex, South Africa.


   LOCAL GEOLOGY
TOP
 ABSTRACT
 INTRODUCTION
 LOCAL GEOLOGY
EXPERIMENTAL AND ANALYTICAL...
 STARTING MATERIALS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The ~2·06 Ga Rustenburg Layered Suite (RLS) comprises a 6–8 km thick, shallow-dipping sheet-like body that intruded with a slight discordance into the late Archaean–early Palaeoproterozoic Transvaal Supergroup. It was rapidly emplaced as a series of discrete pulses of ultramafic and mafic magma that ranged in temperature from 1300°C to 1160°C (Cawthorn & Walraven, 1998Go). Prior to emplacement, the Transvaal Supergroup sediments experienced lower greenschist-facies burial metamorphism (Engelbrecht, 1990Go). The combined effects of rapid intrusion and hot magmas have produced an extensive metamorphic aureole in the Transvaal Supergroup floor rocks beneath the RLS. Contact metamorphic effects are generally most pronounced in the rocks of the upper Transvaal Supergroup (Pretoria Group). However, in the NE lobe of the RLS, around Potgietersrus (Fig. 1a), the RLS has also strongly contact metamorphosed floor rocks of the lower Transvaal Supergroup (Chuniespoort Group). The intensity of syn-emplacement deformation and fabric development beneath the RLS varies considerably. Around the western and northeastern lobes of the RLS, contact metamorphic rocks are hornfels at all grades (Nell, 1985Go; Engelbrecht, 1990Go). In contrast, medium- to high-grade contact metamorphism in floor rocks beneath the eastern lobe of the RLS was accompanied by syn-emplacement deformation (Uken & Watkeys, 1997Go).



View larger version (72K):
[in this window]
[in a new window]
  		Fig. 1. (a) Geology of the northeastern lobe of the RLS and its country rocks. The open star shows the locality of Timeball Hill cordierite-rich restitic rocks at the RLS contact. (b) Geology of the eastern lobe of the RLS and its country rocks, showing the locality of starting compositions BV73 and BV77. Also shown are 500°C, 600°C and 700°C isotherms in the floor rocks (Sharpe & Chadwick, 1981Go).

 
Beneath the eastern lobe of the RLS, Johnson et al. (2003)Go modelled partial melting in floor rock metapelites of the Silverton Formation (Pretoria Group; Fig. 1b) as varying from minor (<5 vol. %) fluid-present melting at ~650–680°C in the lowermost part of the anatectic zone (~350 m below the RLS) to modest (<19–25 vol. %) dehydration-melting of biotite-bearing assemblages at higher temperatures (~750°C at ~100 m below the RLS). Around the NE lobe (Fig. 1a) there is little evidence of partial melting in the floor rocks, except directly at the contact with the RLS. Here, Nell (1984Go, 1985)Go described the occurrence of quartz-absent, Al–Fe–Mg-rich and Ca–Na–K-poor rocks from the Timeball Hill Formation (Pretoria Group). These rocks typically contain the assemblage cordierite + spinel + sillimanite ± corundum ± K-feldspar. On the basis of comparisons with lower-grade, stratigraphic equivalents, Nell (1984Go, 1985)Go suggested that these rocks were Timeball Hill metapelites whose bulk composition had been modified by the extraction of ~60 wt % of a silicate melt of peraluminous granitic composition. The fugitive granitic melt is not preserved at this locality, but elsewhere there is evidence of veining or contamination of the RLS by peraluminous granitic material (Engelbrecht, 1990Go; Johnson et al., 2003Go). The assemblages developed in the Timeball Hill restites are mineralogically distinct from quartz-absent, ultrahigh-temperature mullite- and/or sapphirine-bearing metapelitic xenoliths in the RLS (Willemse & Viljoen, 1970Go).

Metamorphic conditions in the aureole have been the subject of a number of studies. Pressure estimates are typically in the range ~2·0 to ~3·5 kbar (Nell, 1985Go; Engelbrecht, 1990Go; Johnson et al., 2003Go). Beneath the western lobe of the RLS, maximum temperatures of ~760–780°C were estimated by Engelbrecht (1990)Go from garnet–cordierite–orthopyroxene-bearing hornfelses using garnet–cordierite Fe–Mg exchange thermometry. Around the eastern lobe, Johnson et al. (2003)Go estimated similar temperatures of 750 ± 100°C (at 3 kbar) using garnet–cordierite Fe–Mg exchange thermometry from metapelites 100 m below the RLS.

Beneath the NE lobe, Nell (1985)Go used an empirical calibration of the cordierite–spinel Fe–Mg exchange thermometer to estimate temperatures at the RLS contact from (1) the Timeball Hill cordierite–spinel–sillimanite–corundum rocks (described above), and (2) Fe-rich hornfelses in the Penge Iron Formation (Chuniespoort Group), which contain the assemblage olivine–orthopyroxene–cordierite–spinel (Nell, 1984Go). The former yielded temperatures of 864 ± 50°C, and the latter 803 ± 34°C (Nell, 1985Go). However, using the more recent experimental calibration of this thermometer (Nichols et al., 1992Go) and representative electron microprobe data from Nell (1984)Go, we obtained temperatures of 700–810°C (at 3 kbar) for five Timeball Hill cordierite–spinel–sillimanite–corundum-bearing restites. We have analysed additional cordierite–spinel pairs from one of these samples (PH169; Nell, 1984Go), as well as another sample (R495-69) collected for this study from the same locality. At 3 kbar, the averages of five cordierite and spinel compositions (analytical methods described below) for each of the two samples yield temperatures of ~730°C and ~740°C, respectively. Nell (1985)Go also estimated temperatures of ~760°C from ferruginous Penge hornfelses that contain the assemblage quartz–orthopyroxene–garnet–cordierite–plagioclase using garnet–cordierite Fe–Mg exchange thermometry. Using representative mineral compositions from Nell (1984)Go, we have calculated temperatures of 715–830°C at 3 kbar for seven of these samples using garnet–orthopyroxene thermometry corrected for late Fe–Mg exchange (Pattison et al., 2003Go).

In summary, a number of different geothermometers applied to the highest-grade aureole rocks, including those from the RLS contact, yield maximum temperatures ≤800°C. Previous partial melting experiments at low to medium pressure predict only low to moderate degrees (~10–30%; Patiño Douce & Johnston, 1991Go; Patiño Douce & Beard, 1995Go; Montel & Vielzeuf, 1997Go; Stevens et al., 1997Go) of melting of metapelites at these temperatures. Given this, it is unclear how the high degrees of partial melting required to produce the restitic Timeball Hill cordierite-rich rocks were attained. Although infiltration of water-rich fluids into the metapelites could have driven extensive water-saturated partial melting (Cartwright & Buick, 1998Go), there is no oxygen isotope evidence to support this in the aureole (Johnson et al., 2003Go). Alternatively, extensive melting at the RLS contact may have resulted from geologically instantaneous heating of initially low-grade metapelites to anatectic temperatures, before either fluid evolved from sub-solidus reactions had escaped the system or low-temperature hydrous minerals had been able to break down through sub-solidus dehydration reactions.

In this study we test these possibilities through partial melting experiments using two metapelitic rocks from the aureole that are isochemical other than for water content. One sample is a medium-grade rock equilibrated at temperatures of 600–650°C that in the experiments should mimic the behaviour of metapelites that heat slowly enough to lose most of their prograde dehydration water prior to crossing the wet-pelite solidus. The other sample is a lower-grade equivalent, which will be used to assess the general potential for ‘instantaneous melting’ at the contacts of mafic intrusions for generating local, high melt volumes.


   EXPERIMENTAL AND ANALYTICAL METHODS
TOP
 ABSTRACT
 INTRODUCTION
 LOCAL GEOLOGY
 EXPERIMENTAL AND ANALYTICAL...
STARTING MATERIALS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Experiments were conducted at the University of Stellenbosch in an internally heated, small-volume Holloway-type gas vessel using high-purity argon as the pressure medium. The 5 W furnace used had a hot spot of ~2 cm. When loaded with the 10 mm i.d. silica glass tubes used as sample carriers, the temperature variation within the hot spot was typically less than 5°C, at pressure and with the vessel mounted horizontally. Up to four capsules between 12 and 15 mm long were loaded into the glass tube side-by-side and two type-K Inconel sheathed thermocouples situated at either end of the capsules monitored the temperature gradient over the sample volume. Small degrees of vessel tilt were applied to keep the temperature gradient through the sample volume as low as possible during experiments. Temperature measurement of the primary thermocouple and control of the furnace was achieved using a Temperature Controls multifunction solid-state temperature controller with integral ice point acting through a 1 kW thyristor unit. The temperature of the second thermocouple was measured using a similar device. Temperatures were readily maintained within 1°C of the set point for the duration of an experiment. Confining pressures were monitored via a Heise Bourdon tube gauge and pressure was stable to ±100 bar during the course of each experiment.

Samples were ground for up to 2 h with a mechanical mortar and pestle in an attempt to produce average grain sizes of ~5 µm in the powders. Despite the long grinding time, the existence of minerals of high hardness, as well as the lubricating effect of the substantial amount of mica in the samples ensured that some larger (20–50 µm) particles of hard minerals were always present. Sample powders were dried at 200°C overnight and kept under vacuum in a desiccator prior to loading into gold capsules. The capsules (typically 12 mm x 5 mm x 5 mm) were constructed from 0·2 mm thick high-purity annealed gold plate. Approximately 100–200 mg of sample powder was sealed into the capsules using an arc welder. The capsules were then weighed and vacuum tested in a water bath for leaks. The duration of individual experiments varied between 138 and 232 h. On completion of the experiments, the run products were removed from the capsules and mounted in epoxy resin on glass slides that were then polished and carbon coated for scanning electron microscopy (SEM).

Starting materials were analysed by X-ray fluorescence (XRF) analysis at the University of Stellenbosch. Imaging of the run products and analysis of the phase compositions was accomplished using a LEO 1430VP scanning electron microscope at the University of Stellenbosch. Phase compositions were quantified by energy dispersive spectrometry (EDS) analysis using an Oxford Instruments 133 keV detector, Link Semquant software and natural mineral standards. Beam conditions during the analyses were 20 kV and 1·5 nA, with a working distance of 13 mm. Pure Co, as well as Ti and Fe in ilmenite, were used periodically during the analyses to correct for detector drift. Glass analyses were conducted using a Hexland cryo-stage cooled by liquid nitrogen to minimize problems with migration of Na2O away from the beam. Mineral abbreviations follow Kretz (1983)Go apart from the following: Fe-Ts, Fe tschermakite component of muscovite; Gl, glass; Liq, silicate liquid; Ulv, ulvöspinel.


   STARTING MATERIALS
TOP
 ABSTRACT
 INTRODUCTION
 LOCAL GEOLOGY
 EXPERIMENTAL AND ANALYTICAL...
 STARTING MATERIALS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The bulk compositions, mineral modes and estimated water contents of the two samples are given in Table 1. The higher-grade starting composition (BV73) is a fine-grained, medium-grade cordierite–andalusite–biotite hornfels from the Silverton Formation (Pretoria Group) and was sampled from the contact aureole beneath the eastern lobe of the RLS (Fig. 1b). This sample was taken at a metamorphic grade interpreted to be below the first occurrence of partial melting in the inner aureole (assumed to be approximated by the 700°C isotherm in Fig. 1b). The lower-grade starting composition (BV77) is a very fine-grained muscovite–chlorite hornfels that was sampled ~30 km south along strike from BV73 (Fig. 1b).


View this table:
[in this window]
[in a new window]
  		Table 1: Bulk composition of protoliths BV77 and BV73, and average compositions of Timeball Hill Formation low-grade metapelites and cordierite-rich restites from the RLS contact (Nell, 1985Go)

 
On an anhydrous basis, the bulk compositions of BV73 and BV77 are nearly identical (Table 1), with similar ASI values [molar Al/(Ca + Na + K); 3·05–3·25] and XMg [molar Mg/(Mg + Fe); 0·61–0·59]. For the experiments, rocks from the Silverton Formation were preferred to those from the Timeball Hill Formation because the latter is much more compositionally variable than the former (Nell, 1984Go). Notwithstanding this, the chemical composition of BV77 is similar to that of average low-grade Timeball Hill metapelites (Nell, 1984Go; Table 1).

In thin section BV77 contains Qtz + Ms (Ms73–78Pg20–23 Fe-Ts2–4) + Chl [XMg = Mg/(Mg + Fe) = 0·38–0·40] + Pl (Ab62An37Or1) + Ilm [XMn = Mn/(Mn + Mg + Fe2+) = 0·14–0·15; Table 2]. It contains randomly oriented laths of fine-grained (~10 µm long) muscovite and chlorite, and coarser-grained (20–25 µm long) porphyroblasts of ilmenite. Apatite and zircon occur at trace levels. BV73 contains the assemblage Qtz + Ms (Ms82–83Pg15–16Fe-Ts2–3) + Bt [XMg = Mg/(Mg + Fe) = 0·34–0·36; 0·29–0·31 Ti cations per 22 oxygen formula unit] + And + Crd (XMg = 0·51; 0·17–0·19 cations Na per 18 oxygen formula unit) + Pl (Ab76–77An21–23Or0–1; Table 2) + Ilm. Two compositionally distinct types of ilmenite occur; relatively Mn-rich as in BV77 (XMn = 0·13–0·14) and Mn-poor (XMn = 0·04). The sample contains millimetre-diameter poikiloblasts of cordierite and subordinate andalusite, set in a matrix of randomly oriented muscovite and biotite, plagioclase and quartz. Both cordierite and andalusite contain inclusions of biotite, ilmenite and quartz. Monazite, zircon and apatite occur in trace abundance.


View this table:
[in this window]
[in a new window]
  		Table 2: Composition of minerals in the starting bulk compositions BV73 and BV77, expressed as structural formulae

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 LOCAL GEOLOGY
 EXPERIMENTAL AND ANALYTICAL...
 STARTING MATERIALS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The results of 3 kbar melting experiments undertaken at 700°C, 750°C, 800°C, 900°C and 1000°C for starting materials BV73 and BV77 are described below, and are summarized in Table 3. Textural relationships are shown in Figs 2 and 3, and the compositions and proportions of experimentally produced silicate glasses, and mineral reactants and products are shown in Tables 4–7 and Figs 4–10.



View larger version (158K):
[in this window]
[in a new window]
  		Fig. 2. BSE images of 3 kbar partial melting experiments using the BV73 bulk composition at (a) 750°C, (b)–(d) 800°C, (e) 900°C and (f) 1000°C. Textural relationships are described in the text. Subscripts ‘1’ and ‘2’, where present, refer to minerals that occur as both relict grains from the hornfels protolith (1) and as newly formed products of the experiments (2).

 


View larger version (100K):
[in this window]
[in a new window]
  		Fig. 3. Back-scattered electron (BSE) images of 3 kbar partial melting experiments using the BV77 bulk composition at (a) 700°C, (b) 750°C, (c) 800°C, (d) 900°C and (e) 1000°C. Textural relationships are described in the text.

 

View this table:
[in this window]
[in a new window]
  		Table 3: Summary of the 3 kbar experiments undertaken using the BV73 and BV77 starting compositions

 

View this table:
[in this window]
[in a new window]
  		Table 4: Average glass compositions, recalculated on an anhydrous basis

 

View this table:
[in this window]
[in a new window]
  		Table 5: Representative compositions for new minerals formed in the BV73 experiments (expressed as structural formulae)

 

View this table:
[in this window]
[in a new window]
  		Table 6: Representative compositions for new minerals formed in the BV77 experiments (expressed as structural formulae).

 

View this table:
[in this window]
[in a new window]
  		Table 7: Modal mineral and melt proportions in partial melting experiments using bulk compositions BV73 and BV77

 


View larger version (15K):
[in this window]
[in a new window]
  		Fig. 4. Normative Ab–An–Or contents of silicate glasses produced in the melting experiments compared with the classification of Barker (1979)Go.

 


View larger version (20K):
[in this window]
[in a new window]
  		Fig. 5. Plot of (a) aluminium saturation index (ASI) vs temperature (°C), and (b)–(d) major element composition of glasses produced during partial melting of BV73 and BV77 vs temperature (°C). Error bars are 1{sigma} uncertainties on average values; if not indicated, they are smaller than the size of the plot symbol.

 


View larger version (24K):
[in this window]
[in a new window]
  		Fig. 6. Projection of new feldspar compositions on the Ab–Or–An plane. The projections of the feldspar solvus at 750–1000°C are from Fuhrmann & Lindsley (1988)Go.

 


View larger version (17K):
[in this window]
[in a new window]
  		Fig. 7. Muscovite compositional variation in terms of (a) Fe + Mg + Mn vs Ti and (b) Ti + Aliv vs Si + Alvi cations per 22 oxygen formula unit.

 


View larger version (14K):
[in this window]
[in a new window]
  		Fig. 8. Biotite compositional variation in terms of (a) XMg vs temperature and (b) Ti + AlVI vs Fe + Mg cation contents per 22 oxygen formula unit.

 


View larger version (9K):
[in this window]
[in a new window]
  		Fig. 9. The composition of complex and Al-rich spinels expressed within the Fe3+–Al–Ti ternary system.

 


View larger version (21K):
[in this window]
[in a new window]
  		Fig. 10. Variation in the abundance of silicate glass (quenched melt) expressed in wt % as a function of temperature for partial melting experiments using protoliths BV73 and BV77. Curved lines fitted to the data for both experiments are hypothetical melt production fits based on the assumption that melt productivities are primarily controlled by water availability. The shaded region shows maximum temperature estimates for contact metamorphism in the aureole obtained from Engelbrecht (1990)Go and Johnson et al. (2003)Go, and recalculated using the compositional data of Nell (1984Go, 1985)Go as discussed in the text.

 
Phase and textural relationships
BV73 (1·2 wt % H2O)
The 700°C experiment produced no glass and no new minerals. In the 750°C experiment, very minor amounts of glass were present. This suggests that the solidus for the sample is located at slightly below 750°C at 3 kbar.

In the 750°C experiment, glass occurs as thin (<1–3 µm wide) films around quartz, andalusite, biotite, plagioclase and poikiloblastic cordierite that were inherited from the hornfels precursor. These minerals occur as anhedral grains draped by films of glass, but with little evidence of embayment, recrystallization or new mineral growth. Muscovite is present in the sample as slightly embayed anhedral grains rimmed by thin (1–3 µm) melt films (Fig. 2a). In general, the largest melt accumulations are in the vicinity of muscovite grains. Muscovite has been observed in the run products in direct contact with quartz, as well as separated from quartz by only a thin glass film.

In the 800°C experiment, muscovite is absent and (10–20 µm wide) films of glass are developed around quartz, andalusite, plagioclase, biotite and poikiloblastic cordierite (Fig. 2b–d). Biotite occurs as anhedral grains that are studded with typically sub-micron grains of Fe3+-rich spinel. Although most original biotite grains were ~5–20 µm long, a number of coarser grains (~50 µm long) managed to survive the grinding process. Within pools of glass, relict plagioclase is commonly overgrown by a shell of K-feldspar. Euhedral, inclusion-free K-feldspar (Fig. 2b), as well as a compositionally distinct new generation of plagioclase also occur. Very locally, relatively large pseudomorphs were observed. They are complex intergrowths of euhedral, 10–15 µm long K-feldspar, <5 µm long needles of sillimanite, and interstitial pockets of silicate glass (Fig. 2b). These phases are inferred to have replaced coarse-grained (~50 µm long) muscovite flakes. New cordierite occurs as euhedral (~7–12 µm) crystals (Fig. 2d) within silicate glass. Ilmenite and rare grains of Al-rich hercynitic spinel also occur locally within the melt pools.

The 900°C run product (Table 3) is similar to that at 800°C, with the exception that neither acicular sillimanite nor K-feldspar is observed. Both relict poikiloblastic and finer-grained, new, euhedral cordierite occur, with many grains of new cordierite containing older, compositionally distinct, cores (Fig. 2e). Both ragged plagioclase inherited from the hornfels, and euhedral, new, compositionally distinct plagioclase also occur (Fig. 2e). Compared with the 800°C experiment the proportion of new cordierite and glass appear to have increased, and the proportion and grain size of relict cordierite, biotite, plagioclase and quartz appear to have decreased.

In the 1000°C experiment abundant new euhedral cordierite (≤15–20 µm) occurs together with Al-rich spinel (2–15 µm), embayed quartz (<10–15 µm) and rare, acicular orthopyroxene (2–4 µm x 10–15 µm) in a glass matrix. Relics of reactant cordierite were also observed as irregular, Fe-rich cores to the new cordierite (Fig. 2f). Locally, ~30 µm x 50 µm aggregates of relatively coarse-grained (<7–15 µm diameter) spinel and glass occur (Fig. 2f).

BV77 (4·4 wt % H2O)
Melting experiments on the muscovite–chlorite hornfels produced a very small proportion of glass at 700°C (Fig. 3a), suggesting that the solidus for this sample was just below 700°C at 3 kbar. In contrast, large proportions of glass formed in all experiments from 750°C to 1000°C (Fig. 3b–e). Only the 700°C experiment contained muscovite. It additionally contained pools of glass of 1–3 µm diameter (Fig. 3a) in association with rare, new, very fine-grained (1–2 µm long) biotite, euhedral, 2–5 µm diameter new cordierite and K-feldspar, partially resorbed 4–7 µm quartz and both relict and compositionally distinct new plagioclase (see below).

All higher-temperature experiments (Fig. 3b–e) contained abundant, small (<5–8 µm), euhedral cordierite crystals, and less common grains of poikiloblastic ilmenite (5–15 µm; Fig. 3d) that are texturally (but not compositionally) similar to that in the starting bulk composition. These experiments also contained euhedral, 1–5 µm grains of Fe3+–Ti–Al spinel (Fig. 3c–e), whereas ~1–2 µm diameter crystals of Al-rich hercynitic spinel occurred only in the 1000°C experiment. Irregular, globular and corroded grains of quartz (Fig. 3c) occurred in the 750–900°C experiments, and needles of sillimanite (<4–6 µm long) were found in glass-rich domains only in the 750°C experiment (Fig. 3b).

Gass chemistry
Mean compositions of silicate glasses from the run products over the temperature interval 750–1000°C for BV77, and 800–1000°C for BV73, are listed in Table 4. Small amounts of glass also occurred in the 700°C BV77 and 750°C BV73 experiments, but could not be analysed. Care was taken to avoid glass analyses contaminated by minerals, and the mean values are typically averages of 5–10 analyses. Absolute 1{sigma} standard deviations about the means are generally small; typically ≤0·25 wt % for major elements other than SiO2 (e.g. Al2O3 and K2O), and ≤0·15 wt % for elements with oxide abundances of 0–2 wt %. Mean SiO2 contents have 1{sigma} uncertainties of 0·19–0·77 wt %, with most being ≤0·5 wt %. In some experiments, MgO contents were below the detection limits of the EDS system (~0·2 wt % in typical glass compositions from this study).

All the glasses are granitic (Fig. 4) and peraluminous (Fig. 5a) in composition. All except for the BV73 900°C experiment have aluminium saturation index [ASI; molar Al2O3/(CaO + Na2O + K2O)] values >1·1 and could be classed as having S-type granite characteristics (Chappell & White, 1974Go). Both sets of glasses are corundum normative (BV73 normative corundum 0·8–2·1%; BV77 normative corundum 2·5–6·1%; Table 4). Normative corundum contents for BV73 show no consistent variation with rising temperature; those from BV77 increase with rising temperature (Table 4). Even though the low-temperature experiments for the two starting compositions contained very different proportions of minerals and glass, glass compositions are similar (Fig. 5). Glasses from BV73, which contained quartz at all temperatures, have ASI values that are constant within error (ASI = 1·06–1·18; Fig. 5a). Glasses from BV77, which became quartz-absent in the highest-temperature experiment, show a considerably wider range (ASI = 1·21–1·63), with the ASI value low and approximately constant in the 700–800°C experiments, and increasing with rising temperature from 800°C to 1000°C (Fig. 5a). With rising temperature, glasses from BV73 show general trends of decreasing SiO2, Al2O3 and K2O, near-constant Na2O, and increasing MgO, FeO and TiO2 contents (Fig. 5b–d). Glasses from BV77 show similar trends (Fig. 5b–d), with the exception that SiO2 contents remain almost constant from 700°C to 900°C before decreasing at 1000°C (Fig. 5b).

Mineral chemistry
Feldspars
Relict plagioclase (Ab72–75An24–25Or01–03) from the hornfels precursor occurs in all the run products for BV73 up to 900°C. New plagioclase is more anorthitic, and becomes more anorthitic and orthoclase-rich with rising temperature: Ab62–69An26–29Or05–08 (800°C) to Ab43–44 An49–46Or08–10 (900°C; Table 5). New plagioclase compositions are similar to those in experiments in the An–Ab–Or ternary system at comparable temperatures (Fuhrmann & Lindsley, 1988Go; Fig. 6). New K-feldspar in the same experiments is Ab19–20An00–03Or76–79 (800°C; Table 5).

In the 700°C BV77 experiment, both plagioclase identical to that in the precursor and new, more Ca-rich plagioclase (Ab64–68An28–30Or04–06; Table 6) are present. The new K-feldspar is similar in composition to that in the 800°C BV73 experiment (Ab23An3Or74; Table 6). The compositions of new plagioclase and K-feldspar are also consistent with equilibration at elevated temperatures (Fig. 6).

Muscovite
Muscovite is present in the BV73 experiments at 700 and 750°C and in the BV77 experiment at 700°C. As the metamorphic grade of the starting rocks is very different, the composition of the muscovite in BV73 and BV77 differs significantly. Muscovite in BV73 is significantly more Ti-rich and has lower Si + Alvi than that in the lower-grade BV77 (Fig. 7). Total Fe + Mg + Mn for both starting muscovites is similar, at ~0·1 p.f.u., and is higher for all experimental muscovite (Fig. 7a). No obvious systematic correlation exists between total Fe + Mg + Mn and Ti, such as might be expected if Ti and the divalent cations were being incorporated into muscovite via a coupled substitution such as Ti + Fe = Alvi (Guidotti, 1978Go). Rather, Ti appears to be incorporated through the substitution Ti + Aliv = Alvi + Si (Guidotti, 1978Go) (Fig. 7b). In Fig. 7b, the 700°C muscovite compositions from the low- and medium-grade compositions converge, suggesting that they have (at least) partially equilibrated at this temperature.

Biotite
Biotite occurs in the 700°C, 750°C, 800°C and 900°C BV73 experiments as relict grains. In the 800°C and 900°C experiments it also occurs as new, euhedral and very small crystals (<1 µm wide; ~5 µm long; Fig. 2c) that were too small to analyse by SEM-EDS. Relict biotite grains have, however, adjusted their compositions, and these change systematically with rising temperature of the experiments. Overall, with rising temperature of the experiment, biotite becomes more Mg-rich and Al-poor (Table 5; Fig. 8), and we assume that the most Mg-rich and Al-poor biotite grains, which are typically the smallest, are the most compositionally equilibrated for any given experiment (see Patiño Douce et al., 1993Go; Montel & Vielzeuf, 1997Go; Stevens et al., 1997Go). Given this, the best estimates for the XMg of biotite in equilibrium with melt at 750°C, 800°C and 900°C are ~0·48, ~0·51 and ~0·60, respectively (Fig. 8). For any given experiment, biotite shows a relatively large range in octahedral Al (AlVI), and a somewhat smaller range in Ti contents (typically ~0·35 and 0·20–0·25 cations p.f.u., respectively; Table 5; Fig. 8). Although with rising temperature there is a general trend of increasing Ti and decreasing AlVI contents, respectively, there is considerable overlap between data from experiments at different temperatures. For all experiments, biotite compositions show the same linear relationship between Fe + Mg and Ti + AlVI cations p.f.u. (Fig. 8). A similar result was noted by Stevens et al. (1997)Go, who ascribed the compositional variation to a coupled substitution involving vacancies ({square}) of the form 0·5Ti + 0·66AlVI + 0·83{square} = 2(Fe,Mg).

Only two acceptable analyses of biotite could be obtained from the very small biotite crystals produced in the BV77 700°C experiment; their composition varies significantly (XMg = 0·31 and 0·39; 0·13 and 0·18 Ti atoms p.f.u.).

Cordierite
Relict poikiloblastic cordierite in all BV73 experiments has a similar composition (XMg = 0·50–0·51) to that from the hornfels precursor. Analysing newly formed cordierite in both BV73 and BV77 was difficult because of the small size of overgrowths (BV73) or new grains (BV77).

In general, cation totals for all cordierite crystals analysed were above the ideal value of 11 per 18 oxygen formula unit. The degree to which the cation total exceeds 11 cations appears to be a function of the total alkali (Na + K) content; cation totals of >11 p.f.u. were obtained even from coarse-grained cordierite in the BV73 hornfels precursor. This suggests that alkali elements are not bound to the cordierite silicate framework, but instead occur in channelways, where they are charge balanced by volatile species (Vry et al., 1990Go). New cordierite overgrowths in the BV73 800°C experiment have similar XMg to the poikiloblastic cordierite (XMg = 0·51–0·53); those in the 900°C and 1000°C experiments have similar compositions (XMg = 0·61–0·63; Table 5) that are considerably more Mg-rich than relict poikiloblastic cordierite.

Newly formed cordierite in the BV77 experiments was very difficult to analyse because of its extremely small grain size, with the result that most cordierite analyses had greater than the ideal five Si atoms per 18 oxygen formula unit, and correspondingly lower Al contents than ideal. Given the lack of other high-Si/low-Al phases in the experimental charges, such anomalous Si and Al values could have been caused by sub-micron-scale inclusions of glass, which could not be observed by back-scattered electron (BSE) imaging (see Montel & Vielzeuf, 1997Go).

Regardless of the cause of the high Si and low Al values, relatively large departures from ideal structural formulae (i.e. deviation of ≤0·2 cations Si above the nominal five per 18 oxygen formula unit) were not accompanied by significant differences in XMg ({Delta}XMg ≤ 0·01–0·02) suggesting that, even for these analyses, the calculated XMg values are fairly robust. Least contaminated analyses are relatively Fe-rich at 700°C (XMg = 0·43) and Mg-rich at 1000°C (XMg = 0·65–0·67); XMg values vary little from 750°C (0·59–0·61) to 900°C (0·59–0·62; Table 6). This lack of compositional variation is mirrored by almost constant melt proportions between 750°C and 900°C (see below).

Orthopyroxene
Orthopyroxene occurred only in the 1000°C BV73 experiment, and has a lower XMg (0·48–0·50) than coexisting cordierite (0·61–0·63; Table 5). Orthopyroxene crystals have variable Al2O3 contents (2·8–6·4 wt %). Al contents are homogeneous on an intra-grain scale, but differ considerably between crystals. Variable, non-equilibrium Al2O3 contents appear to be a general feature of orthopyroxenes produced in experimental studies of this type (see Montel & Vielzeuf, 1997Go; Stevens et al., 1997Go). All orthopyroxene grains analysed are Ca- and Na-free and contain 0·4–0·6 wt % TiO2 and 0·3–0·4 wt % MnO.

Ilmenite
All ilmenite analyses contain very small amounts of SiO2 (typically <1 wt %). Ilmenite in the BV77 precursor muscovite–chlorite hornfels has XHem [Fe3+/(Fe3+ + Fe2+)] = 0·0 and is Mn-rich. In contrast, ilmenite in the BV77 700°C and 750°C experiments has a range of compositions (XHem = 0·0–0·07 and 0·12–0·36, respectively), and is Mn-poor (Table 6). Ilmenite in higher-temperature experiments shows higher and more uniform XHem values, i.e. 0·38–0·40 (800°C), 0·53–0·57 (900°C) and 0·54–0·57 (1000°C). It also becomes slightly more Mg-rich with rising temperature, from XMg = 0·00–0·01 at 700°C to 0·04–0·07 at 1000°C (Table 6).

Ilmenite also occurs in the 750–900°C BV73 experiments, where it increases in XMg (from 0·05 to 0·12) and XHem (from 0·21 to 0·44) with rising temperature (Table 5). In the BV73 1000°C experiment, ilmenite has a moderate haematite component (XHem = 0·36) and is relatively Mg-rich (XMg = 0·11).

Spinel group minerals
Two types of spinel occur and, in some experiments, coexist. Analysing both of the spinels was particularly difficult because of their very small grain size. As a result, all analyses show small Na, K and Si contents that cannot be accounted for in the structural formulae, and the data discussed are for least contaminated analyses.

‘Complex’ spinels are solid solutions between magnetite, ulvöspinel and hercynite–spinel components, and occur in all BV77 experiments, where they coexist with ilmenite and, at 1000°C, additionally with Al-rich spinel (hercynite–spinel solid solutions). Complex spinel in the BV77 experiments becomes more Al-rich with rising temperature, i.e. from Mag80–81Ulv5–6Spl+Hc14–15 at 750°C to Mag55–54Ulv12–13Spl+Hc33–34 at 1000°C (Fig. 9; Table 6). Al-rich spinel in the 1000°C experiment has the composition Mag25–26Ulv2Spl+Hc72–73 and XMg = 0·24–0·25 (Table 6).

Similar relationships between Al-poor and Al-rich spinel, and ilmenite occur in BV73. Complex spinel occurs in the 800 and 900°C experiments, whereas Al-rich spinel occurs in the 800–1000°C experiments. Al-rich spinel was too small to be successfully analysed in the 800°C experiment, but in the 900°C and 1000°C experiments it has a similar composition (Mag10–11Ulv0–1Spl+Hc88–89; XMg = 0·24–0·27; Table 5).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 LOCAL GEOLOGY
 EXPERIMENTAL AND ANALYTICAL...
 STARTING MATERIALS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Approach to equilibrium
The mineral and melt compositions described above result solely from partial melting experiments and hence the melting behaviour of the two bulk compositions was not reversed. However, the following lines of evidence suggest that the experimental results represent an approach to equilibrium: (1) the analysed glass compositions for any given experiment are very uniform and granitic (Table 4), as expected; (2) the glass compositions show systematic and predictable changes as a function of temperature and buffering mineral assemblage; (3) newly formed minerals are euhedral and compositionally unzoned; (4) the compositions of minerals in the experiments differ from those of the same minerals in the starting compositions (e.g. plagioclase, biotite and cordierite in BV73; ilmenite in BV77), and change systematically with rising temperature; (5) the proportion of melt and product and reactant minerals changes with rising temperature in both sets of experiments in a systematic manner (see below).

Melt and mineral proportions
Mineral and melt proportions (in wt %) for all the experiments in which melt compositions could be determined were calculated by a constrained least-squares mass balance approach, with the goodness-of-fit monitored through the sum of squared residuals (r2). For the two experiments in which glass was observed but could not be analysed (700°C BV77 and 750°C BV73), melt proportions were constrained by image analysis of BSE images and found to be very small (<5%). The results of mass balance calculations for glass proportions are shown in Fig. 10 and summarized in Table 7. For experiments involving BV77, from 750°C to 900°C the experiments have very similar modal mineral and glass proportions, with glass contents between 48 and 50 wt %. The 1000°C experiment had an appreciably different mode; it was characterized by significantly higher proportions of glass (~64 wt %), lacked quartz and had correspondingly lower cordierite contents than in the lower-temperature experiments.

In the BV73 experiments, mass balance calculations were hampered by the mineralogical complexity of the samples (e.g. the occurrence of new and old generations of plagioclase and cordierite) and because in some cases new minerals observed by BSE imaging could not be analysed because of their extremely small grain size. To facilitate calculations, the following assumptions were made: (1) in the 800°C experiment, new cordierite appeared to have essentially the same composition as that of the precursor poikiloblastic cordierite; therefore, the composition of this cordierite was used in the calculations; (2) the composition of Al-spinel in the 800°C experiment was used in modal calculations even though it was, compared with spinel used in calculations for other experiments, badly contaminated by Si (SiO2 contents of ~3·5 wt %). Using these assumptions, mixing calculations for all samples returned very small values of r2. The resulting calculated modes generally agree with visual inspection of BSE images, i.e. with rising temperature, the proportions of Qtz, Bt, Sil/And and Pl decrease, and the proportions of glass (Fig. 10), new Crd and oxide phases (in particular, Spl) increase (Table 7).

Glass proportions in the BV73 experiments appear to increase in a simple linear manner with rising temperature (Fig. 10). However, this is likely to be a function of the size of the temperature step between experiments. As shown by a number of previous experimental studies (see Stevens et al., 1997Go; Patiño Douce & Harris, 1998Go) melt production should be non-linear because of the substantial increases in melt volume within narrow temperature bands as the ultimate thermal stability of muscovite and biotite is exceeded. Melt productivity for BV77 is non-linear with rising temperature, as predicted by these studies (Fig. 10). We would expect that melt proportions in BV73 would also have increased non-linearly with rising temperature in Fig. 10 had more experiments been undertaken in the interval between 800°C and 900°C.

Glass proportions in the 750–800°C BV73 experiments are similar to the maximum melt proportions predicted from the calculations of Johnson et al. (2003)Go for these temperatures. Glass proportions in the two higher-temperature BV73 experiments converge towards the glass proportions in BV77. Only at an interpolated temperature in excess of ~950°C does the glass proportion in BV73 approach that found in the BV77 experiments undertaken at temperatures comparable with maximum temperatures in the RLS aureole (~800°C). This suggests that, unless peak temperatures in the RLS aureole have been greatly underestimated, the experimentally determined melt productivity of BV73 cannot account for the amount of inferred granitic melt component lost from pelitic rocks immediately adjacent to the RLS.

Melting reactions
Observed textural relationships in the experimental charges, and calculated changes in mineral–glass modes allow some general constraints to be placed on melting relationships. For BV73, calculated modal biotite contents in the lower-temperature, above-solidus (750°C and 800°C) experiments are only slightly lower than those of the hornfels precursor, suggesting that it was unlikely to have been involved significantly in low-temperature melting. The two lowest-temperature experiments have muscovite present either as a sub-solidus phase (700°C) or as relic, partially melted, crystals (750°C). In both experiments, it is locally in contact with quartz, despite the fact that the experimental temperatures were 100–150°C higher than the calculated position of the quartz-saturated muscovite sub-solidus breakdown reaction in the KNASH system (Holland & Powell, 2001Go)

(1)
It is possible that muscovite is present in both experiments metastably (Rubie & Brearley, 1987Go; Brearley & Rubie, 1990Go). Alternatively, muscovite in these experiments contains significant Ti and Fe components (consistent with incorporation of Ti into the mica structure by substitutions such as Ti + Aliv = Alvi + Si; Guidotti, 1978Go) that may enhance the thermal stability of muscovite. Patiño Douce & Harris (1998)Go documented muscovite dehydration-melting between 750°C and 800°C at 6 kbar. This is significantly above the appropriate KNASH incongruent melting reaction

(2)
Muscovite in these experiments was also Ti-bearing, possibly indicating that, as with biotite (e.g. Stevens et al., 1997Go), the presence of Ti significantly enhances the thermal stability of muscovite. The 800°C BV73 experiment contains intergrowths of K-feldspar, melt and sillimanite that are inferred to be pseudomorphs after muscovite, consistent with reaction (2).

With rising temperature from 800 to 900°C, the modal abundances of quartz, biotite and aluminosilicate decrease, whereas glass, new cordierite and spinel increase, and biotite and plagioclase systematically change composition. K-feldspar, present at 800°C, disappears by 900°C. This suggests an overall multivariate reaction of the form

(3)
The occurrence of K-feldspar as a product in lower-temperature experiments, but as a reactant at higher temperatures, has been noted in other experimental studies (e.g. Pickering & Johnston, 1998). The small temperature window for K-feldspar stability in these experiments compared with higher-pressure experiments is consistent with the expansion of the K-feldspar + melt stability field with increasing pressure (Patiño Douce & Beard, 1995Go).

The 900°C experiment retains traces of Al2SiO5; however, the 1000°C experiment lacks Al2SiO5, biotite and plagioclase, and contains orthopyroxene and spinel. Given the traces of Al2SiO5 present at 900°C, it is likely that further dehydration melting of Bt + Qtz + Al2SiO5 exhausted Al2SiO5 at some temperature above 900°C. A number of experimental studies have shown that the production of orthopyroxene-bearing assemblages at low to moderate pressures requires dehydration-melting of Bt + Qtz ± Pl, rather than Bt + Qtz + Sil ± Pl (e.g. Montel & Vielzeuf, 1997Go; Stevens et al., 1997Go), and that melting of the former occurs at higher temperatures than that of the latter. Therefore, the occurrence of orthopyroxene in the 1000°C experiment suggests that, after exhaustion of sillimanite by reaction (3), a subsequent reaction of the form

(4)
occurred, and exhausted the remaining biotite and plagioclase.

The melting history of BV77 is more obscure because the sample melted extensively in all but the lowest-temperature (700°C) experiment. The incipient melting in the 700°C experiment suggests that the 3 kbar wet solidus for this sample was only slightly below 700°C. In the 700°C experiment, the lack of chlorite coupled with the very low abundance of melt, the relatively low abundance of muscovite compared with the precursor (Tables 1 and 7), and the abundance of cordierite and K-feldspar, suggest that the latter have grown at the expense of the former under sub-solidus conditions. The lack of sillimanite suggests that reaction (1) was overstepped. The fluid released from sub-solidus dehydration was trapped within the capsule, which was found to contain free water on opening. No feldspars or muscovite occur in the higher-temperature experiments, suggesting that between 700°C and 750°C congruent wet melting of Ms + Kfs + Qtz + Pl occurred via the metastable projection of the reaction

(5)
(Huang & Wyllie, 1973Go).

Water-saturated melting is consistent with the hydrous nature of the protolith, which contained 28 wt % chlorite and 30 wt % muscovite. Given ideal mineral stoichiometries, this translates to 4·4 wt % H2O in the bulk rock [the loss on ignition (LOI) was 4·3%]. The 3 kbar experimental data of Holtz et al. (1995)Go for haplogranitic melts suggest that 8 wt % H2O would be the water saturation limit of the melts in this study over almost the entire experimental range. Thus, assuming chlorite subsolidus breakdown and sufficient quartz and feldspar, BV77 has the potential to saturate ~40 wt % melt close to the wet solidus based on H2O derived from chlorite breakdown alone. This minimum constraint is consistent with the melt proportion calculated from the 750°C experiment (Table 7). Additional melting would have been possible from fluid released from muscovite sub-solidus dehydration.

Between 750°C and 900°C, glass proportions remain essentially constant in BV77 (Fig. 10). Over the same temperature interval, the lack of feldspar in the sample and the fact that melt saturation water concentrations are almost constant at 3 kbar (Holtz et al., 1995Go), prevented significant additional melting. The transition from the 900°C to 1000°C experiment involved the exhaustion of quartz, a significant decrease in the abundance of cordierite and an increase in its XMg, an increase in the abundance of glass and oxide phases (spinel minerals), and a significant shift in the chemistry of the glass towards more Fe- and Mg-rich compositions. Overall, changes in modal abundance across this temperature interval are consistent with a multivariate melting reaction of the form

(6)


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 LOCAL GEOLOGY
 EXPERIMENTAL AND ANALYTICAL...
 STARTING MATERIALS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
The 3 kbar melting experiments presented here have shown that, unless metamorphic temperatures at the contact of the RLS have been significantly underestimated, the amount of melt produced is far too low to account for the occurrence of highly restitic metasedimentary rocks found at the contact of the RLS. The evidence for partial melting at the RLS contact is much better modelled by the melting behaviour of the low-grade precursor (BV77), chosen to mimic instantaneous heating of relatively low-grade metasediment at the contact of the RLS. The locally high-volume melting recorded at the RLS contact could also potentially have been facilitated by local influx of water, derived either from dehydration of lower-grade country rocks or from the crystallization of the RLS itself. However, oxygen isotope data from extensively melted Silverton Formation rocks are inconsistent with either significant infiltration of fluid derived from dehydration of lower-grade pelitic rocks into the anatectic zone or fluid derived from or equilibrated with the RLS (Johnson et al., 2003Go).

Although it is unlikely that melting behaviour of the type that we have explored here is generally representative of partial melting in the outermost anatectic zones surrounding high-temperature mafic intrusions where heating may occur more slowly, such a process may be unavoidable in the innermost parts of these zones, where heating of country rocks to above-solidus conditions occurs very rapidly, and metastable melting and/or fluid retention into the melting interval should be expected.


   ACKNOWLEDGEMENTS
 
I.S.B., G.S. and R.L.G. acknowledge support from the Australian Research Council (Senior/Professorial Research Fellowships and IREX Grant), the National Foundation for Research (South Africa), and the University of the Witwatersrand, respectively. SEM images were collected with assistance from Niel Steenkamp. Johan Nell provided samples from his M.Sc. thesis for further analysis. Constructive reviews by Giles Droop, Max Schmidt and Kjell Skjerlie are gratefully acknowledged.


   FOOTNOTES
 

* Corresponding author. Present address: School of Geosciences, Monash University, Melbourne, Vic. 3800, Australia. Fax: +61-3-99054903. E-mail: Ian.Buick{at}sci.monash.edu.au


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 LOCAL GEOLOGY
 EXPERIMENTAL AND ANALYTICAL...
 STARTING MATERIALS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Barker, F. (1979). Trondhjemite: definition, environment and hypotheses of origin. In: Barker, F. (ed.) Trondhjemites, Dacites and Related Rocks. Amsterdam: Elsevier, pp. 1–12.

Brearley, A. J. & Rubie, D. C. (1990). Effects of H2O on the disequilibrium breakdown of muscovite + quartz. Journal of Petrology 31, 925–956.[Abstract/Free Full Text]

Cartwright, I. & Buick, I. S. (1998). Oxygen isotope resetting, partial melting, and fluid flow in metamorphic terrains. Terra Nova 10, 81–85.[CrossRef][Web of Science]

Cawthorn, R. G. & Walraven, F. (1998). Emplacement and crystallization time for the Bushveld Complex. Journal of Petrology 39, 1669–1687.[CrossRef][Web of Science]

Chappell, B. J. & White, A. J. R. (1974). Two constrasting granite types. Pacific Geology 8, 173–174.

Dutrow, B. L., Foster, C. T., Gable, C. W. & Travis, B. J. (2004). Metamorphic heating rates in contact aureoles: consequences of intrusion depth and size, geothermal gradient and host rock permeability. Geological Society of America, Abstracts with Programs 35, 396.

Engelbrecht, J. P. (1990). Contact metamorphic processes related to the aureole of the Bushveld Complex in the Marico District, western Transvaal, South Africa. South African Journal of Geology 93, 339–349.[Abstract]

Fuhrmann, M. L. & Lindsley, D. H. (1988). Ternary-feldspar modelling and thermometry. American Mineralogist 73, 201–215.[Abstract]

Guidotti, C. V. (1978). Compositional variation of muscovite in medium- to high-grade metapelites of northwestern Maine. American Mineralogist 63, 878–884.[Abstract]

Holland, T. J. B. & Powell, R. (2001). Calculation of phase relations involving haplogranite melts using an internally consistent dataset. Journal of Petrology 42, 673–683.[Abstract/Free Full Text]

Holtz, F., Behrens, H., Dingwell, D. B. & Johannes, W. (1995). Water solubility in haplogranitic melts. Compositional, pressure and temperature dependence. American Mineralogist 80, 94–108.[Abstract]

Huang, W. L. & Wyllie, P. J. (1973). Melting relations of muscovite granite to 35 kbar as a model for fusion of metamorphosed subducted oceanic sediments. Contributions to Mineralogy and Petrology 24, 1–14.

Johnson, T., Brown, M., Gibson, R. L., Buick, I. S. & Cartwright, I. (2003). Partial melting of metapelitic rocks beneath the Bushveld Complex, South Africa. Journal of Petrology 44, 789–813.[Abstract/Free Full Text]

Kerrick, D. M., Lasaga, A. C. & Raeburn, S. P. (1991). Kinetics of heterogeneous reactions. In: Kerrick, D. M. (ed.) Contact Metamorphism. Mineralogical Society of America, Reviews in Mineralogy 26, 583–672.

Kretz, R. (1983). Symbols for rock-forming minerals. American Mineralogist 68, 277–279.[Abstract]

Montel, J.-M. & Vielzeuf, D. (1997). Partial melting of metagreywackes, Part II. Compositions of minerals and melts. Contributions to Mineralogy and Petrology 128, 176–196.[CrossRef][Web of Science]

Nell, J. (1984). Geochemical and thermodynamic controls on the formation of mineral assemblages from the metamorphic aureole of the Bushveld Complex in the Potgietersrus area. M.Sc. thesis, University of Pretoria.

Nell, J. (1985). The Bushveld metamorphic aureole in the Potgietersrus area: evidence for a two-stage metamorphic event. Economic Geology 80, 1129–1152.[Abstract/Free Full Text]

Nichols, G. T., Berry, R. F. & Green, D. H. (1992). Internally consistent gahnitic spinel–cordierite–garnet equilibria in the FMASHZn system: geothermobarometry and applications. Contributions to Mineralogy and Petrology 111, 362–377.[CrossRef][Web of Science]

Patiño Douce, A. E. & Beard, J. S. (1995). Dehydration-melting of biotite gneiss and quartz amphibolite from 3 to 15 kbar. Journal of Petrology 36, 707–738.[Abstract/Free Full Text]

Patiño Douce, A. E. & Harris, N. (1998). Experimental constraints on Himalayan anatexis. Journal of Petrology 39, 689–710.[CrossRef][Web of Science]

Patiño Douce, A. E. & Johnston, A. D. (1991). Phase equilibria and melt productivity in the pelitic system: implications for the origin of peraluminous granitoids and aluminous granites. Contributions to Mineralogy and Petrology 107, 202–218.[CrossRef][Web of Science]

Patiño Douce, A. E., Johnston, A. D. & Rice, J. M. (1993). Octahedral excess mixing properties in biotite: a working model with applications to geobarometry and geothermometry. American Mineralogist 78, 113–131.[Abstract]

Pattison, D. R. M., Chacko, T., Farquhar, J. & McFarlane, C. R. M. (2003). Temperatures of granulite facies metamorphism: constraints from experimental phase equilibria and thermobarometry corrected for retrograde exchange. Journal of Petrology 44, 867–900.[Abstract/Free Full Text]

Rubie, D. C. & Brearley, A. J. (1987). Metastable melting during the breakdown of muscovite + quartz at 1 kbar. Bulletin de Mineralogie 110, 533–549.

Sharpe, M. R. & Chadwick, B. (1981). The geometry and origin and structures of certain Transvaal Sequence rocks within and adjacent to the eastern compartment of the Bushveld Complex. Institute of Geological Research on the Bushveld Complex (University of Pretoria) Research Report 27, 19.

Stevens, G., Clemens, J. D. & Droop, G. T. R. (1997). Melt production during granulite-facies anatexis: experimental data from ‘primitive’ metasedimentary protoliths. Contributions to Mineralogy and Petrology 128, 352–370.[CrossRef][Web of Science]

Tracy, R. J. & McLellan, E. L. (1985). A natural example of the kinetic controls on compositional and textural equilibration. In: Thompson, A. B. & Rubie, D. C. (eds) Metamorphic Reactions: Kinetics, Textures and Deformation. New York: Springer, pp. 118–137.

Uken, R. & Watkeys, M. K. (1997). Diapirism initiated by the Bushveld Complex, South Africa. Geology 25, 723–726.[Abstract/Free Full Text]

Vry, J. K., Brown, P. E. & Valley, J. W. (1990). Cordierite volatile content and the role of CO2 in high-grade metamorphism. American Mineralogist 75, 71–88.[Abstract]

Willemse, J. & Viljoen, E. A. (1970). The fate of argillaceous material in the gabbroic magma of the Bushveld Complex. In: Visser, D. J. L. & von Gruenewaldt, G. (eds) Symposium on the Bushveld and Other Layered Intrusives. Geological Society, South Africa, Special Publication 1, 336–366.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Eur J MineralHome page
D. HARLOV, A. RENZULLI, and F. RIDOLFI
Iron-bearing chlor-fluorapatites in crustal xenoliths from the Stromboli volcano (Aeolian Islands, Southern Italy): an indicator of fluid processes during contact metamorphism
European Journal of Mineralogy, April 1, 2006; 18(2): 233 - 241.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
45/9/1777    most recent
egh033v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (7)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by BUICK, I. S.
Right arrow Articles by GIBSON, R. L.
Right arrow Search for Related Content
GeoRef
Right arrow GeoRef Citation
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?