Cenki 2002

J. metamorphic Geol.
, 2002, 20, 543–561
Melt-producing and melt-consuming reactions in the Achankovil

cordierite gneisses, South India
B. CENKI,1 L. M. KRIEGSMAN2,* AND I. BRAUN1
1
Mineralogisch-Petrologisches Institut, Universität Bonn, Poppelsdorfer Schloß, 53115 Bonn, Germany
(cenki@uni-bonn.de; ingo.braun@uni-bonn.de)
2
Department of Geology, University of Turku, 20014 Turku, Finland
ABSTRACT Migmatitic cordierite gneisses within the Achankovil Zone (AZ) of southern Pan-African India record
melt-producing and subsequent melt-consuming mineral reactions. Early mineral assemblages Bt-Sil-Qtz
and Bt-Sil-Spl, deduced from inclusion textures in garnet prophyroblasts, break down via successive
dehydration melting reactions to high-T phase assemblages (e.g. Grt-Crd-Liq, Opx-Liq, Spl-Crd-Liq).
Later back reactions between the restite and the in situ crystallizing melt resulted in thin cordierite
coronas separating garnet from the leucosome, and partial resorption of garnet to Opx-Crd or Crd-
Bt-Qtz symplectites. Leucosomes generally display a moderate (low-strain gneisses) to strong (high-
strain gneisses) depletion of alkali feldspar attributed to mineral-melt back reactions partly controlled by
the degree of melt segregation.
Using a KFMASH partial petrogenetic grid that includes a melt phase, and qualitative pseudosections
for microdomains of high and low Al ⁄ Si ratios, the successive phase assemblages and reaction textures
are interpreted in terms of a clockwise P–T path culminating at about 6–7 kbar and 900–950 C. This
P–T path is consistent with, but more detailed than published results, which suggests that taking a melt
phase into account is not only a valid, but also a useful approach. Comparing P–T data and lithological
and isotopic data for the AZ with adjacent East Gondwana fragments, suggests the presence of a
coherent metasedimentary unit exposed from southern Madagascar via South India (AZ) and Sri Lanka
(Wanni Complex) to the Lützow–Holm Bay in Eastern Antarctica.
Key words: anatectic migmatites, cordierite gneisses, dehydration melting, Gondwana, KFMASH
petrogenetic grid, South India.
Meen, 1995; Bartlett et al., 1998) and those in the

INTRODUCTION
Wanni Complex of Sri Lanka (Milisenda et al., 1994).
Gondwana reconstructions generally locate South However, since geochemical, petrological, geochrono-
India and Sri Lanka close to East Africa, Madagascar logical and structural data are still limited for the
and East Antarctica (e.g. Lawver & Scotese, 1987; Achankovil Zone, any Gondwana reconstruction
Powell et al., 1988). Similarities in tectonic style, aiming to link the South Indian Granulite terrane with
degree of metamorphism and age patterns suggest that the lower crustal complexes of Madagascar, Sri Lanka
these high-grade metamorphic terranes shared a com- and East Antarctica must remain speculative.
mon tectonothermal evolution during the Pan-African The first reports on the petrography and the petro-
orogeny (Hiroi & Motoyoshi, 1990; Yoshida et al., logy of the cordierite gneisses from the Achankovil
1992; Shiraishi et al., 1994; Kriegsman, 1995). A com- Zone were given by Sinha-Roy et al. (1984) and San-
mon feature of these terranes is the occurrence of tosh (1987). Thermobarometric calculations yielded
cordierite-rich gneisses either along shear zones peak P–T conditions of 5.5–7 kbar and 700–800 C
(Madagascar: Nicollet, 1985; South India: Santosh, and fluid inclusion studies revealed that high-grade
1987) or in specific lithotectonic units (Wanni metamorphism occurred in the presence of a CO2-
Complex, Sri Lanka: Prame & Pohl, 1994; Raase & dominated fluid phase. The coronitic growth of low-P
Schenk, 1994; Lützow-Holm Bay, Antarctica: Hiroi & assemblages around high-P minerals was interpreted in
Motoyoshi, 1990). Preliminary data also suggest a terms of rapid uplift and strong isothermal decom-
similarity in Nd model ages for cordierite gneisses in pression (Santosh, 1987). In a recent study, Nanda-
the Achankovil Zone (Harris et al., 1994; Brandon & kumar & Harley (2000) derived P–T estimates of
6.5–7.0 kbar and 900 ± 20 C from Grt-Opx ther-
mobarometry applied to charnockites and cordierite
*National Museum of Natural History, PO Box 9517, NL-2300 gneisses. They concluded that temperatures were sig-
RA Leiden, the Netherlands. nificantly higher close to the boundary of the Madurai
Blackwell Science Inc., 0263-4929/02/$15.00 543
Journal of Metamorphic Geology, Volume 20, Number 6, 2002
544 B. CENKI ET AL.
Block than in the central part of the KKB. According composition, and (iii) the influence of melt presence on
to their investigations, the post-peak metamorphic the reconstruction of P–T conditions.
P–T path was characterized by an initial stage of limited
decompression which was followed by strong decom-
GEOLOGICAL SETTING
pression after temperatures had dropped below 800 C.
This study reports additional information on the The Achankovil Zone forms the boundary between the
peak P–T conditions and the P–T path of cordierite- Late Proterozoic Kerala Khondalite Belt (KKB) in the
rich gneisses in the Achankovil Zone, based on south and the Madurai Block (MB) in the north. Both
microtextural and petrological investigations of suc- terranes most likely formed part of East Gondwana
cessive phase assemblages recorded in these rocks. A and were subjected to polyphase ductile deformation
key difference with earlier studies (Sinha-Roy et al., and metamorphism during the Pan-African Orogeny at
1984; Santosh, 1987) is that we emphasize the presence c. 550 Ma (Buhl et al., 1983; Choudhary et al., 1992;
of melt and its involvement in both prograde and ret- Soman et al. 1983; Miller et al., 1997; Bartlett et al.,
rograde metamorphic reactions. Most reports on 1998) which in the KKB reached granulite facies con-
migmatitic granulites conclude that dehydration melt- ditions (Chacko et al., 1987, 1996; Braun et al., 1996;
ing reactions were responsible for the production of an Nandakumar & Harley, 2000).
in situ melt, which upon crystallization formed leuco- The Achankovil Zone is a NW–SE trending lin-
some (e.g. Powell & Downes, 1990; Fitzsimons, 1996; eament of <50 km width and easily identified in
Greenfield et al., 1998; Kriegsman & Hensen, 1998; LANDSAT images (Drury & Holt, 1980; Drury et al.,
Berger & Kalt, 1999). However, only a few of these 1984). However its structural character is not well
studies (e.g. Jones & Brown, 1990; Kohn et al., 1997; known (Radhakrishna et al., 1990) and there are
Spear et al., 1999) take into account that melt- contradictory results on the shear sense (Sacks et al.,
consuming reactions may have operated after the peak 1997; Rajesh et al., 1998). Because of this and the
of metamorphism. results of our own field studies which provided very
Melt segregation in migmatites during and after little indication of strong shear movement, we avoid
partial melting, particularly under static conditions, is the term ÔShearÕ, and simply call this the ÔAchankovil
commonly limited (Brown et al., 1995). Most of the Zone (AZ)Õ.
melt may therefore still be present during decom- Lithologically, the AZ consists of garnet-biotite
pression or cooling, allowing partial back reaction gneisses, metapelitic Grt-Bt-Sil ± Crd-gneisses (khon-
between in situ crystallizing melt and the restite dalites), charnockites and cordierite gneisses. The
(Ashworth, 1985; Waters, 1988; Clemens & Droop, latter have not been reported from the KKB itself and
1998; Spear et al., 1999). The understanding of these seem to be restricted to the AZ. They form discon-
processes is fundamental for interpreting reaction tinuous, km-size lenses within an 8–10 km wide zone
textures and for the metamorphic evolution of along the northern edge of the AZ (Sinha-Roy et al.,
anatectic migmatites, because the presence of melt 1984; Santosh, 1987). Besides the petrological work of
may affect the derivation of P–T paths in two ways Santosh (1987) only a few geochronological and geo-
(Kriegsman & Hensen, 1998): (i) the slopes (dP ⁄ dT) chemical investigations have been carried out on these
of most dehydration reactions steepen considerably rocks. Single zircon evaporation ages (Bartlett et al.,
when melt rather than aqueous fluid are produced, 1998) yielded a bimodal distribution of Early Pro-
indicating that many corona textures in migmatites do terozoic (1802 ± 16 Ma) and Pan-African ages
not require substantive changes in pressure, but may (530 ± 21 Ma). This distribution is similar that
result simply from cooling after the initial heating obtained from the KKB (Bartlett et al., 1998; Braun
step; (ii) P–T points derived from thermobarometry et al., 1998) and reflects two stages of high-grade
using only solid–solid equilibria may be erroneous metamorphism (Bartlett et al., 1998).
when some of the phases involved were not solid at Nd model ages of the cordierite gneisses range
the time of reaction, but were components of a melt between 1.5 and 1.2 Ga (Brandon & Meen, 1995;
phase with unknown activity. Therefore, the applica- Bartlett et al., 1998) and are interpreted to result
tion of classical thermobarometers involving only from mixing of source rocks of Early and Late
solid phases (e.g. Grt-Opx-Pl-Qtz) to unravel the P–T Proterozoic age (Bartlett et al., 1998). They are sig-
evolution of anatectic migmatites, in which melt is nificantly younger than those obtained from gneisses
omnipresent, may be problematic. and charnockites of the adjacent KKB, ranging
The Achankovil cordierite gneisses are a good between 3.0 and 2.0 Ga (Brandon & Meen, 1995;
example of anatectic migmatites where melting Bartlett et al., 1998; Köhler, pers. comm., 1999).
occurred during or close to the thermal peak of Similar contrasting Nd model ages are also known
metamorphism and where a significant melt fraction from Sri Lanka, where a unit with cordierite gneisses
remained in situ during the high-T segment of the P–T (Wanni Complex) shows significantly younger Nd
path. They are therefore highly suitable to study (i) the model ages (between 1.8 and 1.1 Ga) than the
interaction between melt and minerals (ii) the effects of adjacent Highland Complex, ranging between 3.0
these processes on mineralogy, mineral and melt and 2.0 Ga (Milisenda et al., 1988, 1994).
MELT-PRESENT REACTIONS IN THE ACHANKOVIL GNEISS 545
Fig. 2. Outcrop photograph, showing high abundance of

leucosome garnet and/or cordierite melanosome and variable
strain. Coin diameter is 2 cm.
Fig. 1. Sketch map of the northwestern edge of the AZ and the

Inclusion textures and porphyroblasts
neighbouring Madurai and Trivandrum Blocks. Stars indicate In the high-strain samples, melanocratic aggregates or layers are
the sample locality (Vazhamuttam) and other occurrences of usually elongated and oriented parallel to the gneissic foliation. Most
cordierite gneisses within the AZ. Inset shows a simplified geo- widespread are aggregates of subhedral bronze orthopyroxene with
logical map of South India and the location of the study area opaque inclusions, pale pink garnet with rare biotite inclusions,
(modified after Raith et al., 1997). MSZ: Moyar Shear Zone; apatite and green spinel, the last of which always forms intergrowths
BSZ: Bhavani Shear Zone; PCSZ: Palghat Cauvery Shear Zone; with opaques (ilmenite and rare magnetite). The shapes of ortho-
AZ: Achankovil Zone; KKB: Kerala Khondalite Belt; NH: pyroxene, garnet and spinel match each other but never touch: they
Nilgiri Hills; BRH: Biligirrangan Hills; ShH: Shevaroy Hills. are invariably separated by plagioclase.
Other aggregates are elongated along the foliation and consist of
ilmenite ⁄ spinel intergrowths (c. 0.5 · 1.5 mm) and rare corundum,
Samples of metapelitic cordierite gneiss from the enclosed by porphyroblastic cordierite. Spinel and cordierite rarely
Achankovil Zone were collected in three active quar- show small biotite inclusions (<100 lm), whereas small (<50 lm)
ries close to Vazhamuttam village (N 09 13.692¢; anhedral garnet inclusions are only found in cordierite. Furthermore,
there are some elongated layers composed of biotite crystals asso-
E 76 47.392¢), which is located near the north-western ciated with plagioclase and porphyritic garnet. Inclusions in garnet
tip of the AZ (Fig. 1). Strongly migmatitic cordierite are small (<50 lm) and comprise fibrolitic sillimanite, which is
gneisses are the dominant rock type in this area, which elongated along the gneissic foliation, as well as spinel, rounded
delimits the AZ to the Madurai Block in the north, biotite (Fig. 3a,b), opaques and alkali feldspar. Some fine-grained
layers are rich in biotite, plagioclase and opaques and display equi-
but have also been found elsewhere in the AZ (Fig. 1). librium textures. In addition, one section showed layers where sub-
However, a contact between cordierite gneisses hedral orthopyroxene, associated with an opaque phase and
and other rock types has not yet been observed. plagioclase, occurs in contact with cordierite.
In the low-strain gneisses, patches of Fe-Mg phases are closely
linked or completely surrounded by the leucosome, which we inter-
PETROGRAPHY pret as in situ crystallized melt. Subhedral garnet (1–8 mm in size) is
usually creamy coloured and hosts various groups of inclusions
Macroscopic sample description (fibrolitic sillimanite, biotite, spinel (Fig. 3b); biotite, quartz, fibro-
litic sillimanite (Fig. 3a; also found in some cordierite porphyro-
Cordierite gneisses at Vazhamuttam are highly migmatitic rocks blasts); rare cordierite, plagioclase, spinel + ilmenite; rare cordierite,
(Fig. 2). Melanosomes contain either cordierite or garnet or cor- plagioclase, quartz; biotite, alkali feldspar, opaque; plagioclase,
dierite-garnet intergrowths. The individual layers are 1–20 cm spinel without ilmenite; ilmenite, alkali feldspar, with two-phase
thick and alternate with quartzofeldspathic leucosomes (1–5 cm inclusions of biotite + plagioclase).
thick) or granite sheets (up to 1 m thick). Locally, orthopyroxene, Orthopyroxene (0.1–1 mm) occurs as isolated crystals in the leu-
magnetite and spinel crystals are up to 3 cm and usually associ- cosome. Locally they form discontinuous rims on garnet or fully
ated with cordierite and garnet. The gneisses display varying surround opaques or form coarse intergrowths with cordierite
degrees of deformation and can be subdivided into high- and low- (Fig. 3c). Regardless of its mode of occurrence, orthopyroxene is
strain varieties. High-strain gneisses have a conspicuous stromatic partly to fully chloritized.
texture. The difference in bulk chemistry as reflected by the indi- Spinel–ilmenite (± rare magnetite) aggregates are present in the
vidual leuco- and melanocratic layers may reflect original varia- leucosome or as inclusions in cordierite or garnet. They lack inclu-
tions in protolith composition. These primary compositional sions except for rare biotite + plagioclase. Cordierite porphyroblasts
heterogeneities were subsequently intensified or modified when the (0.5–2 cm) contain small (0.1–1 mm) inclusions of rounded plagio-
gneisses were subjected to metamorphism, migmatization and melt clase, alkali feldspar, opaque, spinel + ilmenite aggregates, two-phase
segregation at different scales. In low-strain gneisses melt segre- alkali feldspar + biotite inclusions, or quartz. Locally, cordierite
gation was less intense, hence, the stromatic texture is poorly porphyroblasts enclosing spinel + ilmenite aggregates mimic origin-
developed. Instead, the contacts between cordierite-garnet-rich ally euhedral garnet. Generally, some relic garnet is left (Fig. 3d).
melanosomes and the quartzofeldspathic leucosomes are rather Monazite, zircon and apatite form inclusions in porphyroblasts
diffuse. and are present in leucosomes.
546 B. CENKI ET AL.
Symplectites, coronas and embayments interface between partly resorbed garnet and the breakdown pro-
ducts shows different shapes: saw-tooth, smooth or euhedral
In the low-strain samples, partial resorption of garnet and cordi- contacts, and this may represent various stages of textural
erite is widespread and gives rise to a variety of textures. Garnet re-equilibration.
commonly has a strongly embayed shape and is surrounded by In the high-strain samples, these features are also present, but are
symplectites and coronas. Strongly embayed garnet is separated less common. Locally, orthopyroxene is replaced by biotite in the
from spinel-ilmenite aggregates by cordierite and, more rarely, by presence of plagioclase (Fig. 3i), indicating partial resorption. Gar-
cordierite-biotite intergrowths (Fig. 3e). Locally, cordierite breaks net often shows strong resorption and the development of embay-
down to intergrowths of biotite ± plagioclase ± quartz. Some ments filled with biotite and plagioclase. Quartz occurs nearby
garnet is rimmed by cordierite and cordierite-quartz single or (<500 lm) but is never in contact with garnet. Generally, these fea-
double coronas, while others show embayments filled by large tures are more developed at the garnet faces perpendicular to the
biotite and plagioclase (0.1–0.5 cm). Often it is surrounded by fine foliation. Fine-grained symplectites with biotite-plagioclase and
(at the garnet interface) to coarse grained (at the leucosome inter- cordierite-quartz are found at the contact between cordierite-rich
face) orthopyroxene · cordierite symplectites (towards the garnet) layers and leucosome. These intergrowths are usually limited to the
(Figs 3f & 4a). Small (10 lm) or large (500 lm) plagioclase crys- outermost margins of the layers (500 lm), and only locally reach the
tals are often present as well. Locally, coarse-grained orthopy- interior parts.
roxene is observed at the edges of large garnet crystals (Fig. 3g).
Orthopyroxenes contain euhedral garnet inclusions and they share
crystal faces, but are not intergrown. Locally, at the contacts Leucosomes
between garnet and alkali feldspar-plagioclase leucosome, the fol-
lowing sequence of mineral associations has developed: anhedral, In high-strain samples, the leucosome is generally composed of
coarse-grained cordierite; cordierite-quartz symplectites; rare bio- 60 vol.% quartz (0.5–5 mm), 30 vol.% plagioclase (500 lm) and
tite-cordierite symplectites; biotite-plagioclase-quartz symplectites 10 vol.% perthitic alkali feldspar. At their grain boundaries, finer
in which biotite and plagioclase are separated by a thin (<10 lm) grained crystals of the same phases form a mosaic texture. The modal
quartz rim (Figs 3h & 4b,c). Both cordierite and garnet are often composition of the leucosome is different near the spinel-cordierite
separated from the leucosome by small plagioclase crystals. The layers, where it is composed of approximately equal amounts of
Fig. 3.
(e) (f)
Grt
Grt Pl
Bt Opx
Crd
Crd
Opx
Pl
Ilm 4a Crd
Spl Chl
Grt
150 m 100 m
Kfs (g) (h)

Grt Grt
Bt
Kfs
Pl
Ilm Opx
Pl
Ilm
Opx Bt
Grt 4c
Crd Kfs
Crd + Qtz
Pl Bt + Qtz
Opx 4b Pl
Grt Kfs
Grt
Pl
500 m 250 m
(i)
Fig. 3. Photomicrographs documenting the textural and miner-

alogical evolution of the cordierite gneisses during migmatiza-
Pl tion. (a–d): Textures related to melt-producing reactions. a & b
show relics of the pre-migmatitic phase assemblage (a: Bt-Sil-
Qtz; b: Bt-Sil-Spl) which occur as inclusions in garnet. (c) Opx-
Crd pseudomorph after garnet formed through reaction (3). (d)
Spl-Crd aggregate after garnet break down through reaction (4).
(e–i): Textures related to melt-consuming reactions. (e) Open
embayments in garnet at the contact with Spl-Ilm aggregates,
Bt Opx filled by cordierite and locally Crd-Bt intergrowths which were
formed from reaction (5). (f) Opx-Crd symplectites resorbing
garnet according to reaction (6a). (g) Garnet breakdown to
orthopyroxene and plagioclase through reaction (6c).
(h) Development of multiple coronas consisting of anhedral
cordierite, Crd-Qtz symplectites, and Bt-Pl-Qtz symplectites
around garnet according to reaction (7a) or reactions (7b) and
150 m (7c). (i) Orthopyroxene breakdown to biotite through reversal of
reaction (2).
quartz, alkali feldspar and plagioclase. Here, relatively fine-grained In low-strain samples, leucosomes generally show a coarse-grained
plagioclase is concentrated at the melanosome interface. Only where (1–5 mm) subequigranular texture consisting of 40–60 vol.% anhe-
cordierite-quartz symplectites are present, is quartz from the leuco- dral quartz, 30–50 vol.% sericitized and antiperthitic plagioclase,
some in contact with the envelope of the symplectites. In all 10–20 vol.% porphyroblastic and perthitic alkali feldspar, and minor
other cases, quartz is located further away (>500 lm) from the biotite. Locally, near cordierite-dominated melanosomes, the leuco-
symplectites. some contains 50 vol.% of large (1–10 mm) perthitic alkali feldspar
548 B. CENKI ET AL.
Fig. 4. BSE images showing the development of reaction textures during garnet breakdown. (a) (magnification of Fig. 3f). Grt rimmed
by Opx-Crd-Pl intergrowths. (b & c) (magnification of Fig. 3h). Coarse-grained cordierite formed during the first steps of reaction
between garnet and the melt which now forms the Kfs-Pl-rich leucosome (right part of the picture). Later, symplectites of Crd-Qtz,
Crd-Bt and finally Bt-Qtz developed. (d) Typical crystallisation texture of a eutectic granitic melt in the leucosome.
with rare inclusions of biotite, opaques, plagioclase and quartz, and Kretz, 1983). For orthopyroxene it is Mg ⁄ (Mg +
25 vol.% each of quartz and plagioclase. Fe2+ + AlVI.); for garnet it is XMg ¼ Mg ⁄ (Mg +
All leucosomes show typical crystallization textures (Fig. 4d) far
away (>1 mm) from garnet: imbricated small, equidimensional Fe + Mn + Ca), and similar definitions for XFe, XMn,
plagioclase, anhedral quartz and porphyritic, spindle-perthitic and XCa. The Fe3+ content of spinel and ilmenite was
alkali feldspar, and graphic textures. Locally, biotite is present as estimated stoichiometrically.
well. Garnet chemistry (Table 1) is dominated by alman-
dine (XFe ¼ 0.52–0.59), followed by pyrope (XMg ¼
MINERAL CHEMISTRY 0.29–0.39), spessartine (XMn ¼ 0.04–0.07) and gros-
sular (XCa < 0.04). Tetrahedral alumina, as well as the
Mineral compositions were analysed with a Cameca andradite component, are insignificant within analy-
Electron microprobe (Camebax Microbeam) at the tical errors. Garnet is surprisingly homogeneous
University of Bonn. Operating conditions were (Fig. 5a) when it is considered that the rocks show
15 kV (accelerating voltage) and 15 nA (specimen considerable complexity in garnet growth and break-
current), counting times were 20 s (peak) and 10 s down reactions.
(background). Data processing was carried out using Orthopyroxene has a hypersthene composition
the PAP correction program (Pouchou & Pichoir, (XMg ¼ 0.52–0.59) and low contents of Ti, Mn
1984). and Ca. The total Al ⁄ 2 (in cations) varies from
In the following, XMg represents Mg ⁄ (Mg + Fe) 0.10 to 0.17 p.f.u. It is usually higher for
for all phases except orthopyroxene and garnet (Opx porphyroblastic orthopyroxene and orthopyroxene
and Grt, respectively; mineral abbreviations after from core regions of coarse grained symplectites,
Table 1. Representative chemical compositions of major minerals of the cordierite gneisses.

Garnet Orthopyroxene
Coarse symplectite Incongruent Cordierite Symplectite
Core Rim Core Rim Fine symplectite Aggregate Coarse Fine Incongruent Aggregate
SiO2 37.45 38.01 38.96 37.09 49.99 49.62 49.76 46.85 49.89 47.69 49.89
TiO2 0.02 0.03 0.05 0.04 0.13 0.05 0.18 0.08
Al2O3 21.73 22.04 21.92 21.69 34.03 33.95 33.89 7.33 5.57 7.82 4.59
FeO 25.99 26.55 26.12 28.09 4.00 3.73 4.56 23.42 20.59 22.51 24.02
MgO 10.49 9.60 9.34 7.37 11.69 11.78 10.99 20.91 23.45 20.55 20.67
MnO 2.29 2.65 2.63 3.02 0.15 0.12 0.15 0.68 0.54 0.70 0.81
CaO 1.05 1.23 0.98 1.37 0.02 0.00 0.03 0.07 0.10 0.11 0.08
S 99.03 100.12 100.00 98.68 99.87 99.21 99.38 99.40 100.20 99.55 100.14
normalized to 12 oxygen normalized to 18 oxygen normalized to 6 oxygen
Si 2.92 2.94 3.00 2.95 4.97 4.96 4.98 1.78 1.84 1.79 1.87
Ti 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Al 2.00 2.01 1.99 2.03 3.99 4.00 4.00 0.33 0.24 0.35 0.20
Fe2+ 1.70 1.72 1.68 1.87 0.33 0.31 0.38 0.74 0.64 0.71 0.75
Mg 1.22 1.11 1.07 0.87 1.73 1.75 1.64 1.18 1.29 1.15 1.16
Mn 0.15 0.17 0.17 0.20 0.01 0.01 0.01 0.02 0.02 0.02 0.03
Ca 0.09 0.10 0.08 0.12 0.00 0.00 0.00 0.00 0.00 0.00 0.00
XFe 0.54 0.55 0.56 0.61 XFe 0.16 0.15 0.19

XMg 0.39 0.36 0.36 0.29 XMg 0.84 0.85 0.81 XFe 0.33 0.29 0.32 0.35
XCa 0.03 0.03 0.03 0.04 XMg 0.52 0.59 0.52 0.55
XMn 0.05 0.06 0.06 0.07 XAl 0.16 0.12 0.16 0.10
XMg ¼ Mg ⁄ (Mg + Fe) XMg ¼ Mg ⁄ (Mg + Fe + Al).
Table 1. (Cont’d).
Spinel Magnetite Ilmenite K-feldspar Plagioclase Biotite
Core Inclusion Inclusion Matrix Inclusion Symplectite Matrix Inclusion Core
TiO2 0.04 0.08 0.00 0.24 47.68 SiO2 64.04 58.45 59.50 57.07 SiO2 38.75 37.10 38.33
Al2O3 59.76 61.16 60.30 0.59 0.03 Al2O3 17.84 23.70 25.75 26.32 TiO2 1.85 4.36 5.87
Cr2O3 0.06 0.42 0.04 0.07 0.00 CaO 0.11 5.92 7.11 8.20 Al2O3 14.52 14.80 14.97
FeO 26.31 22.53 26.27 31.00 49.35 Na2O 1.64 8.58 7.51 7.08 FeO 7.57 11.26 10.45
Fe2O3 68.91 K2O 14.73 0.14 0.32 0.29 MgO 21.06 17.66 16.49
MgO 10.80 11.12 9.13 0.06 1.14 S 98.34 96.79 100.19 98.96 Na2O 0.05 0.24 0.07
MnO 0.23 0.18 0.20 0.03 1.05 normalized to 8 oxygen K2 O 10.63 9.81 10.22
ZnO 2.53 3.27 3.41 0.00 0.04 Si 3.00 2.69 2.65 2.58 F 3.91 0.85 1.69
S 99.73 98.76 99.35 100.89 99.28 Al 0.98 1.29 1.35 1.40 S 98.34 96.08 98.08
normalized to 4 oxygen, 3 cations 3 oxygen, 2 cations Ca 0.01 0.29 0.34 0.40 normalized to 22 oxygen
Ti 0.00 0.00 0.00 0.00 0.90 Na 0.15 0.77 0.65 0.62 Si 5.66 5.47 5.55
Al 1.91 1.96 1.95 0.01 0.00 K 0.88 0.01 0.02 0.02 Ti 0.20 0.48 0.64
Cr 0.00 0.01 0.00 0.00 0.00 Al 2.50 2.57 2.56
Fe2+ 0.51 0.47 0.55 0.43 0.83 Fe2+ 0.92 1.39 1.27
Fe3+ 0.09 0.04 0.05 0.86 0.20 Xab 0.14 0.72 0.64 0.60 Mg 4.59 3.88 3.56
Mg 0.44 0.45 0.37 0.00 0.04 Xor 0.85 0.01 0.02 0.02 Na 0.01 0.07 0.02
Mn 0.01 0.00 0.00 0.00 0.02 Xan 0.01 0.27 0.34 0.38 K 1.98 1.84 1.89
Zn 0.05 0.07 0.07 0.00 0.00 F 1.81 0.40 0.77
XMg 0.83 0.74 0.74
Xmag 0.05 0.02 0.03 Xhem 0.10 XMg ¼ Mg ⁄ (Mg + Fe)

Xhc 0.46 0.45 0.53 Xilm 0.83
Xgah 0.05 0.07 0.07 Xpyro 0.02
Xspl 0.44 0.45 0.37 Xgeik 0.04
XMg 0.42 0.47 0.38
XMg ¼ Mg ⁄ (Mg + Fe); uncorrected for Fe3+.
whereas it is lower in orthopyroxene from fine fluids which is in line with the results from SIMS
grained symplectites. analyses (0.10–0.64 wt% H2O & 0.63–0.85 wt% CO2)
Cordierite is usually highly magnesian, with XMg of cordierite from these lithologies (Nandakumar &
between 0.80 and 0.85. The Na is always below the Harley, 2000).
detection limit. Totals range from 98 to 100 wt%, Spinels are hercynite-rich with an XMg of 0.38–0.47.
indicating the presence of minor amounts of channel The ZnO content is 2.2–3.9 wt%, giving a gahnite
550 B. CENKI ET AL.
component of 4–7 mole%. Ferric iron has been 1.00
calculated using stoichiometric constraints giving (a)

maximum values for the magnetite component of 0.80
4.5 mole%. Ilmenite compositions are close to the Ti
end-member and contain low or negligible amounts of
Mn, Mg and Fe3+. Opaques containing only Fe 0.60 X Fe
appear to be exclusively magnetite when the total wt%
is recalculated for Fe3+. X Mg
0.40
Alkali feldspar composition in leucosomes is K-rich
(Or78)92Ab21)7An1)2). Plagioclase has andesine com-
position (Ab59)71An37)29Or0)4), with an anorthite 0.20
content independent of the textural position of the X Ca
X Mn
crystals. However, the albite content is slightly higher
in plagioclase inclusions and symplectites and lower 0.00
0 100 200 300 400 500 600 700 m 800
for plagioclase in the leucosome.
Biotite shows vacancies in the octahedral sites, that A (b)
may be due to the high Ti content. XMg is low (0.65– incongruent phases
coarse grained symplectite
0.73) for leucosome and symplectitic biotite, but phases (core of crystals)
usually higher (0.67–0.79) for biotite inclusions in coarse grained symplectite
phases (rim of crystals) Crd
garnet. The fluorine content can be very high (up to fine grained symplectite
3.9 wt%), as well as TiO2 (up to 5.9 wt%). These phases
values are lower for leucosome biotite than for inclu-

sions and symplectitic biotite.
Grt
Coexisting garnet, orthopyroxene and cordierite
have similar XMg values, irrespective of the textural
position (porphyroblastic, symplectite core or rim) and
of the different mineral generations (Fig. 5b). There is
Opx
more spread in biotite compositions, with higher XMg,
F and TiO2 for inclusions and lower values in other F M
textures. The only remarkable chemical variation is Fig. 5. (a) Flat garnet profile, showing the absence of any
shown by the Al content in orthopyroxene (4.5–8 wt% noticeable compositional zoning. (b) AFM Compositional data
Al2O3) which is considered insensitive to retrograde for contiguous grains of Grt, Opx and Crd, showing their very
re-equilibration (Harley & Green, 1982; Harley, 1984; small variation in chemical composition.
Fitzsimons & Harley, 1994). A more detailed discus-
sion of these results is given in the ÔThermobarometryÕ melt segregation which was significantly higher in the
section. high-strain gneisses and therefore prevented intense
mineral-melt back reaction in these rocks. Those fea-
tures indicate that the behaviour of the melt phase
REACTION SEQUENCE
largely influences the growth or resorption of minerals
The nonequilibrium mineralogy of the cordierite as well as their compositional adjustment during the
gneisses comprises: garnet, cordierite, orthopyroxene, multistage reaction history of anatectic migmatites and
spinel, biotite, sillimanite, ilmenite, magnetite, that understanding the impact of the melt phase plays
corundum, quartz, alkali feldspar, plagioclase and a critical role in the interpretation of the metamorphic
chlorite. Many of these minerals are present in two evolution of these rocks.
or more generations. All the assemblages can be The highly magnesian and aluminous metapelitic
described in terms of melt-producing reactions that cordierite gneisses of the AZ are composed of numer-
also produce incongruent phases, followed by (retro- ous small-scale domains of different bulk compositions
grade) melt-consuming back reactions between the (Stüwe & Powell, 1989; Stüwe, 1997). Depending on
porphyroblastic (incongruent) phases (garnet, cordi- the amount of melt generation and extraction, each of
erite, spinel, orthopyroxene) and the in situ crystallizing these domains developed in a different way upon
melt. cooling and each contain potential information on
In the high-strain samples, melanosomes are clearly metamorphic conditions and physico-chemical pro-
separated from each other by leucocratic layers, and cesses which were effective at various stages of the
the breakdown of incongruent phases due to back metamorphic history of the rocks.
reaction with the in situ melt is strongly limited. In
comparison, low-strain rocks display an homogeneous
Melt-producing reactions
and random distribution of all phases and a wide-
spread replacement of the incongruent minerals. These One of the earliest assemblages recorded in the inves-
contrasting textural features reflect the difference in tigated rocks comprises biotite, quartz, alkali feldspar
and fibrolitic sillimanite, which occur as inclusions suggests that retrograde reactions with the in situ
(10–200 lm) in garnet porphyroblasts (Fig. 3a), con- crystallizing melt during cooling have modified the
sistent with the KFMASH divariant dehydration melanosome (see Kriegsman & Hensen, 1998; Spear
melting reaction: et al., 1999; Kriegsman, 2001). Most garnet grains
partially break down and are rimmed by coronas
Bt+Sil+Qtz+Kfs (+Pl)=Grt+Liq (+Ilm). ð1aÞ
(50 lm) that consist of either anhedral cordierite or
Other rocks from the same locality, that probably have fine grained cordierite + quartz ± plagioclase sym-
a higher bulk XMg, show cordierite porphyroblasts plectites. Locally, biotite is present as inclusions in
instead of garnet. They contain the same inclusions, cordierite and at cordierite margins (Fig. 3e). Spinel
except sillimanite, and an appropriate reaction leading grains also show evidence for breakdown to cordierite.
to cordierite growth is: Hence, a trivariant reaction explaining these observa-
tions is:
Bt+Sil+Qtz(+Pl)=Crd+Kfs+Liq(+Ilm). ð1bÞ
Another early assemblage is composed of coexisting Spl+Grt+Liq=Crd+Bt (+Pl-rich leucosome).
euhedral spinel (c. 4 wt% ZnO) and fibrolitic silli- ð5aÞ
manite. Both occur as small (<20 lm) inclusions in
porphyroblastic garnet (Fig. 3b) and may have It is more common, however, to see garnet grains
formed from the breakdown of Zn-bearing stauro- partly replaced by cordierite without involving spinel
lite. or spinel grains partly replaced by cordierite in the
Large (0.2–1 mm), subhedral orthopyroxene grains absence of garnet. This implies the local operation of
in the quartzofeldspathic matrix may have formed reactions with a higher variance:
through the multivariate KFMASH melting reaction:
Grt+Liq=Crd+Bt (+Pl-rich leucosome). ð5bÞ
Bt+Qtz (+Pl) = Opx+Liq (+Ilm). ð2Þ
Spl+Liq=Crd+Bt (+Pl-rich leucosome). ð5cÞ
Smaller (500 lm) anhedral orthopyroxene crystals
are closely associated with cordierite porphyroblasts. Abundant coarse (200–400 lm) to fine (10–20 lm)
They either form elongate aggregates in the foliation grained orthopyroxene-cordierite symplectites resorb-
or locally mimic a former garnet shape suggesting ing garnet (Figs 3f & 4a) can be further interpreted as a
that they are pseudomorphs after garnet (Fig. 3c). reaction between garnet and the crystallizing in situ
Recalculating the garnet composition by combining melt according to
the XMg values of cordierite (0.80) and orthopyrox- Grt+Qtz+Liq-I=Opx+Crd (+Pl)+Liq-II. ð6aÞ
ene (0.57) results in an anomalously Mg-rich garnet
(XMg ¼ 0.62), which is considered too magnesian The only leucosome phase closely associated with these
for normal metapelitic rocks. We therefore sug- symplectites and which might have crystallized from
gest that biotite was involved as a reactant in the the in situ melt (Liq-I) is plagioclase, leaving behind an
reaction: alkali-rich granitic liquid (Liq-II). Biotite is absent and
can therefore not explain the lack of a K-feldspar
Bt+Grt+Qtz(+Pl)=Opx+Crd+Liq(+Ilm). ð3Þ component in the leucosome. Leucosome of granitic
composition only occurs in the outermost part
A second generation of spinel is always associated with (>1 mm) of the symplectitic envelope.
ilmenite (and locally with rare magnetite), and both Locally, coarse-grained orthopyroxene is observed
probably originated from the breakdown of Ti-rich at the edges of large garnet crystals (Fig. 3g). Ortho-
biotite. Together with cordierite porphyroblasts they pyroxene and garnet share crystal faces, but are not
commonly form cm-scale layers enclosing small relics intergrown. Garnet relics inside orthopyroxene suggest
of garnet (up to 20 lm), biotite, quartz or, rarely, corun- that it formed from garnet breakdown. At a first
dum. Spinel is never in contact with the leucosome. In glance, this could be due to the solid–solid reaction:
low-strain parts, aggregates of spinel and porphyrob-
lastic cordierite form pseudomorphs after garnet Grt+Qtz=Opx+Pl. ð6bÞ
(Fig. 3d) and commonly separate partially resorbed
garnet from the leucosome, which can be explained by However, there are several arguments to invoke the
the reaction: involvement of a melt phase in the reaction. First,
the volume ratio Pl ⁄ Opx of the products of reaction
Bt+Grt(+ Sil+Pl)=Spl+Crd+Liq (+Ilm). ð4Þ (6b) should be about 0.7, whereas the observed ratio
is about 1.5. Second, extensive leucosome is present
next to the reaction rim. Third, garnet in these
Melt-consuming reactions
samples has very low Ca, which is normal for
The melanosomes contain a high modal abundance of metapelites, but recalculating the original grossular
incongruent phases, which are often embayed and component before the inferred reaction progress
partly consumed in the vicinity of the leucosome. This would give XGrs > 0.05. Hence, we infer that a
552 B. CENKI ET AL.
more likely reaction (a higher variance one closely Late fluid release from the crystallizing melt
linked to reaction 6a) was: according to reaction (8) may possibly explain the
partial chloritization of the Fe–Mg phases, notably
Grt+Liq=Opx+Pl. ð6cÞ of Opx. Calculations with Thermocalc indicate that a
Orthopyroxene is often embayed and partly replaced possible reaction, Opx + Crd + V ¼ Chl + Qtz,
by large biotite crystals (up to 200 lm; Fig. 3i), con- occurs at 695 and 610 C (at 6 kbar) in MASH and
sistent with a reversal of reaction (2). FASH, respectively, which is indeed close to the wet
In some places garnet is partly resorbed and displays solidus.
embayed grain boundaries. These small (10–100 lm) The incomplete nature of retrograde reactions
embayments are filled with symplectic intergrowths between solid phases and the crystallizing melt can be
with modal compositions changing systematically from explained by a combination of several mechanisms:
the garnet boundary towards the leucosome: anhedral shielding of anhydrous minerals by first crystallization
cordierite (Fig. 3h), cordierite-quartz (Figs 3h & 4b), products, leading to physical separation, kinetic inhi-
rare biotite-cordierite (Figs 3h & 4b), and biotite- bition, and diffusive loss of the remaining H2O (after
plagioclase-quartz symplectites (Fig. 4c) in which bio- Waters, 1988). Shielding by crystallization products is
tite and plagioclase are separated by a quartz rim a common feature in corona-forming metamorphic
(<10 lm). We infer operation of the melt-consuming reactions and in migmatites. It reflects the combined
reaction: and opposing effects of increasing diffusion path
lengths between products during progressive reaction
Grt+Liq=Crd+Bt+Qtz. ð7aÞ and exponentially decreasing diffusion rates during
Alternatively, the textures could be explained by a cooling (Kriegsman, 2001).
succession of reactions, in which the melt composition
changed gradually. First, garnet breakdown could
PARTIAL PETROGENETIC GRID
have resulted in a single cordierite rim, possibly due to
AND PSEUDOSECTIONS
a multivariant reaction:
Grt+Liq-I=Crd+Liq-II. ð7bÞ Choice of phases and chemical system
In addition to the melt phase inferred to have been
followed by the growth of cordierite–quartz symplec-
tites. Subsequently, cordierite may have reacted with present under high-grade conditions, nine main min-
erals are in the samples: garnet, cordierite, ortho-
the melt to form biotite–quartz symplectites:
pyroxene, spinel, biotite, sillimanite, quartz, alkali
Crd+Liq-I=Bt+Qtz. ð7cÞ feldspar and plagioclase. Minor phases are magnetite,
ilmenite and corundum. At least six major components
Waters (2001) argued that symplectitic quartz-bearing are necessary to describe the system: K2O, FeO, MgO,
intergrowths in migmatites develop as solid-state Al2O3, SiO2, H2O (KFMASH). MnO and ZnO are
replacements, but several observations suggest that only present in small quantities in garnet and hercy-
here a melt was involved in their production. First, nite, respectively, and Fe3+ is negligible in ilmenite,
where retrograde symplectites formed, the leucosomes spinel and garnet. Thus, these components are not
do not have a granitic composition. In some places, the taken into account. Na2O and CaO are major com-
only leucosome phase is plagioclase, indicating that the ponents of plagioclase and of the inferred melt phase,
other melt components have been consumed. Second, TiO2 is present in biotite and ilmenite, and biotite also
some outermost biotite-bearing intergrowths in direct contains F. Hence, the proper chemical system to
contact with the leucosome resemble a graphic texture, describe all phases and reactions reasonably well is
indicating that they formed by crystallization of a NCKFMASHT.
eutectic melt. In these intergrowths, biotite is always Plagioclase is considered as a melt component that
separated from the leucosome by a <10 lm quartz passively crystallized from the melt and was not
rim. This could either imply that biotite reacted with actively involved in restite-melt back reactions. The
the last melt phase, or it is a nucleation effect. We grids (Fig. 6a,b) and pseudosections (Fig. 7a,b) are
therefore propose that such intergrowths develop therefore drawn for a KFMASH system and omit
during the final stages of back reaction with in situ plagioclase and ilmenite. K-feldspar may be a product
crystallizing melt, at a stage where diffusion rates have or reactant during prograde and retrograde melting
already slowed down considerably. reactions, depending on the H2O ⁄ K2O ratios of biotite
Away from the back reaction textures in the mel- and melt (Carrington & Watt, 1995), which are diffi-
anosome, the melt exhibits a granitic composition and cult to assess from thin section and microprobe studies
shows graphic intergrowth textures involving quartz, alone. K-feldspar is more likely to be a product when
plagioclase and K-feldspar (Fig. 4d). Hence, the last biotite + plagioclase are reactants during partial
melt possibly crystallized at the solidus via: melting. As we omit plagioclase in these petrogenetic
grids, Kfs is considered as a reactant and reactions
Liq=Qtz+Kfs+Pl+V Bt. ð8Þ have been balanced following Carrington & Watt
(Crd)
BtQtz(GrtKfs)
OpxLiq
[Spl V]
(Opx)
(Crd)
A
Bt
Gr tzKf
Op r t C
Cr
Bt
tLi
BtSilQtzKfs
(a) (b)
dQ
P
x
q
G
(Crd)
GrtLiq
Si d K
Grt Spr Sil
l L fs
(kbar) (Sil) rt (Kfs)
s
Gr tzG
iq
r
Spl Crd tS lQ q (Kfs)
Sp pr Q t Si dLi
l C tz B Cr [Qtz V]
px t
r
Liq
Cr QtzG
rd
SplL fs
iq
l (Kfs)
tSilK
tSi
l Q Sil
dO
Gr iq
Bt
(Bt)
Op Bt rdL
GrtB
Fig. 6b
Sp rd
tz
xS lC
tC
GrtS Sp
Bt il L Crd ilL
BtK iq
Gr
Gr iq (Opx)
tC (Liq) fs
rd
GrtSplL
(Liq)
BtCrdK
Kf (Bt)
rn s
8 dC (Liq)
r (Bt)
Crd Kfs
iq
t C Sil
fs
BtQtz(C
Gr Spl (Grt)
Liq
OpxLiq fs)
Spl rt Sil
(Grt) (Sil)
Gr Crd
Bt G
(Crd)
Op
rdK
tz
tL BtK
x
rd
iq
S pl Q
tV
Grt C
Kfs
Liq z
(Kfs)
t
Grt l Kfs Q
rt B
tz Kfs
fs
Bt
Liq
Opx
Liq
Qtz
Liq
Sil (Grt) (Liq)
il G
Kf
Crd
sG Liq
Opx Crd
Bt Grt Q
i
S il
fs S
Bt S
rt C
Crd
(Bt)
rd
V
K
Bt
Qt
Grt
Bt
Qtz
zK Li
6 fs
Gr q (Grt)
rd Kfs
Qt
rd Kfs
tC
l Liq
zK
fs
rd
l Liq
tz
tz V
B
tK
V
Liq rt C
fs
l Q fs
(Kfs) Q
Bt Grt C
Opx Sp
Si d K
tz
lO d
Sil
Gr
q
Sp rt Cr
rd) Liq
Bt Grt C
lQ
Opx Sp
Li
B C r
G
Sil
t
rd
Gr
G
P
Kf t
sV
rd Gr
lC
Bt
rd
Qtz
Bt (Grt)
l C il
[Spl Opx]
tz
sV
Opx (C
Sp
Sp Bt S
Kf
D
V
x C rt Q
sS Liq
Kf
il S SO
Op t G
rd
pl
Crd
B
LID
V
T ( C) T
C
US
800 900 1000 1100 [Qtz Opx] E

Fig. 6. General partial petrogenetic grid (a) for K2O-FeO-MgO-Al2O3-SiO2-H2O (KFMASH) system. Melting curves are after
Carrington & Harley (1995) and Schreinemakers analysis (see text). (b) Divariant bundles linking univariant lines. Liq ¼ melt;
V ¼ fluid.
(1995) and Carrington & Harley (1996). Moreover, observed divariant reactions. Two of the KFMASH
Kfs singularities have not been treated in detail. invariant points (labelled with the absent phases) were
Ilmenite is considered as a breakdown product of taken from the experimental data of Carrington &
Ti-bearing biotite. It therefore always appears on the Harley (1995): [Spl V] and [Qtz V]. These points
opposite side of reactions involving biotite. Involving are connected via the univariant reaction Bt +
plagioclase + ilmenite would lead to grids and pseu- Grt + Crd + Kfs ¼ Opx + Sil + Liq (A). This
dosections that are qualitatively similar to those in and three other KFMASH univariant dehydration
KFMASH, although shifted to different positions in melting reactions are taken from Carrington & Harley
P–T space. Hence, we have chosen to interpret the (1995), with the slopes modified according to the data
microtextural observations using a simplified, semi- of Holland et al. (1996): Bt + Sil + Qtz + Kfs ¼
quantative KFMASH partial petrogenetic grid Grt + Crd + Liq (B), Bt + Grt + Qtz +
(Fig. 6a,b) and pseudosections (Fig. 7a,b), and to Kfs ¼ Opx + Crd + Liq (C), Bt + Sil + Grt +
discuss qualitatively the effects of Na, Ca and Ti, as Kfs ¼ Spl + Crd + Liq (D). The solidus is taken
well as fluids in cordierite, on the exact P–T locations from Ebadi & Johannes (1991) and H2O as a free
in a later section. vapour phase is considered to be restricted to the
subsolidus side. We follow Kriegsman & Hensen
(1998) in locating the P–T intersection of reaction (B)
Construction of a partial KFMASH grid
with the solidus at 6 kbar, 800 C. This location is the
Leaving apart plagioclase + ilmenite, as discussed average position of the intersection with the biotite
above, implies that the only major phases that need to dehydration curve Bt + Sil + Qtz + Kfs ¼ Grt +
be represented on the partial grid are: garnet, cordi- Crd + V, using data in Thompson (1976), Spear &
erite, orthopyroxene, spinel, biotite, sillimanite, quartz, Cheney (1989) and Xu et al. (1994).
melt and K-feldspar. As we wish to employ a grid that The fifth dehydration melting reaction, Bt +
shows the solidus, fluid (V) must be added to the sys- Grt + Crd + Kfs ¼ Opx + Spl + Liq (E), pro-
tem. For each invariant point, seven phases (+Kfs) are jects from invariant point [Qtz V] to lower pressure
stable. For our purposes, it is sufficient to construct and has a similar slope to reaction C. Divariant bun-
enough stable univariant lines to accommodate all dles around univariant KFMASH reactions B, C, D
554 B. CENKI ET AL.
Grt Bt Qtz Kfs Liq (a) (b)

(Crd)
(Crd) (Crd)
Bt Sil Kfs Qtz

Bt Qtz Grt Opx Liq
Bt Qtz (Grt Kfs)
Gr t Liq
Bt Sil Kfs Qtz
(Kfs)
Opx Liq
Gr t Liq
(Crd)
Liq
Grt Bt Sil Kfs Liq
(Gr t)
Grt Bt Spl Sil Liq
s
(Crd)
il Kf
Bt Qtz
1
Liq
Bt S
il
Opx
1a 1b tS
Gr
S pl
(Kfs)
t C z Kfs
Bt
Gr t
2 Liq
Liq
d
3 Gr t Cr
Qt
Bt Sil Qtz Kfs l
Sp
Kfs
Sil L 4
rd
Gr
Liq s
Grt Opx Qtz Crd Liq iq
Sil
pl Kf
tL
Liq
Gr l Qtz
iq
Bt Sil Qtz Kfs
d S rt
Gr
Bt
Cr Sil G
rd
Si
tz
tC
tz Grt Spl Crd Bt Liq

(Kfs) Q
Crd
Bt
Bt
higher XMg Q
rd) Liq
Kfs
fs
lK
6
Grt Spl Liq

Bt
Si
z
Cr
Bt
Qt
iq
dL
dQ
Opx (C
Bt (Grt)
7
fs
lK
Bt Sil Crd Kfs Liq
Cr
tz
Si
(Grt)
iq
K
dL
fs
Gr 5
Bt
tQ
rt
Cr
tz
Li
Crd Kfs
O
Liq
q
Kfs
px
(Grt)
Bt Crd Kfs
Cr
pl Liq
d
Op
iq
Bt Grt Crd
Bt
Kfs)
xC
P
Kf
Opx Spl L
Opx Liq
rd
s
Kf (Bt)
Bt Sil Crd Kfs

Bt Grt
Bt Qtz (Crd
s
Opx S
P Bt Crd Qtz Kfs Liq
Spl Liq
Opx Crd Grt Liq Kfs
Opx Crd Qtz Kfs Liq

T T (Grt)
(Grt) Bt Crd Spl Kfs Liq
Fig. 7. Partial KFMASH qualitative pseudosections for bulk compositions with relatively high (a) and low (b) silica contents.
Based on the grid and divariant bundles in Fig. 6(a) and (b). Bold lines: KFMASH univariant reactions; normal lines: KFMASH
divariant reactions (reactions in bold: phases that are exhausted in each particular pseudosection; reactants in brackets: possible
degeneracies); italic boxes: KFMASH trivariant fields.
and E were constructed by balancing the different which provided constraints on the slopes of other
reactions using matrix analysis on a spreadsheet in a reactions through Schreinemakers analysis. Note that
five component end-member system, and then ordering the water content of cordierite has a strong influence
them in agreement with Schreinemakers’ rules and on the pressure location of this curve, but not on its
labelling them with the absent phase (Zen, 1966). We slope. The pressure location of this line will be
used the average melt composition of Carrington & discussed in a later section.
Harley (1995), biotite with 1.65 Al p.f.u. on 11 oxygen,
and partition coefficients for water between cordierite
Pseudosections
and water-undersaturated melt from Carrington &
Harley (1996). As some rock domains are quartz-saturated, whereas
We have drawn all K-feldspar-saturated divariant other domains ran out of quartz during partial melt-
reactions and three K-feldspar-absent ones (Fig. 6) and ing, it was not possible to describe all the reactions in
have linked those bundles that share a common reac- one pseudosection. Two qualitative pseudosections
tion. Several degeneracies exist in the system. For (Hensen, 1971) are sufficient to describe the most
example, melting reaction (2) can be degenerate in important microtextural features: one for quartz-rich
KFMASH and balances for specific biotite and domains (now orthopyroxene-bearing patches; Fig. 7a)
melt compositions. It can also be considered as a and one for quartz-poor domains (now spinel-bearing
quadrivariant reaction. Hence, melt production via patches; Fig. 7b). These patches are regarded as rep-
reaction (2) and melt consumption via its reversal can resenting different domains of effective bulk compo-
occur at different pressure, as shown in Fig. 7(a), which sition (Stüwe, 1997), that change with space and time
was constructed for compositions rather close to during the rock history.
degeneracy. K-feldspar singularities have not been
drawn and K-feldspar is considered as a reactant in the
DESCRIPTION OF THE P–T PATH
univariant reactions considering appropriate H2O ⁄ K2O
ratios for biotite and melt (Carrington & Watt, 1995).
Qualitative description of the P–T path
The slope of divariant reaction Grt + Qtz + L ¼
Opx + Crd + Kfs was calculated with THERMOCALC Based on the inferred KFMASH pseudosections, the
(Holland & Powell, 1990) for a bulk XMg of 0.47, interpretation of assemblages and reaction textures in
these Mg–Al rich cordierite gneisses yields a clockwise decompression path accompanied by heating. Both
P–T path. Heating led to partial melting via reactions paths are consistent with the qualitative P–T path
1–4. The proposed reaction sequence is consistent with derived here.
experiments (Vielzeuf & Holloway, 1988; Vielzeuf &
Montel, 1994) at increasing temperature, but in some
Semi-quantitative P–T estimations
samples quartz was exhausted before the last stages of
partial melting (reaction 4), as also described by Since all reactions described above include a melt phase,
Montel et al. (1986) and Kriegsman & Hensen (1998). quantitative P–T conditions are difficult to estimate
In silica-rich domains (Fig. 7a), sillimanite probably owing to the lack of thermodynamic data for the melt
ran out during the first melting reaction (1a), which phase, and to the difficulties of estimating the original
produced garnet and, in more Mg-rich rocks, cordierite melt composition. However, some important petrolo-
(1b) (see Fig. 7a). By contrast, quartz ran out during gical constraints give at least a confident P–T window.
reaction (1) in domains with relatively low silica con- Orthopyroxene and sillimanite do not coexist, and so
tents (Fig. 7b). This established the mineralogical dif- the pressure did not exceed 8 kbar (Hensen & Green,
ference between the two domains, which determined 1971; Bertrand et al., 1991; Carrington & Harley, 1995).
subsequent reactions. In both domains, sillimanite does Orthopyroxene and spinel formed through different
not reappear along the retrograde path, which provides dehydration melting reactions, and the assemblage
important constraints on the P–T path (see below). Opx + Spl + Liq was not stable, which provides an
Partial melting reactions (2) and (3) only operated upper temperature limit of 940 C at 8 kbar.
in silica-rich systems (Fig. 7a), whereas reaction (4) is The P–T intersection of univariant curve Bt +
restricted to silica-poor domains (Fig. 7b). These Sil + Qtz + Kfs ¼ Grt + Crd + Liq (B) with the
three reactions can be accommodated along an biotite dehydration curve Bt + Sil + Qtz ¼
overall heating path, possibly with minor decom- Grt + Crd + Kfs + V at the solidus is located at
pression at peak temperature. Several divariant and about 6 kbar, 800 C (Kriegsman & Hensen, 1998;
trivariant melt-consuming reactions (reactions 5–7) using data in Thompson, 1976; Spear & Cheney,
then led to partial resorption of incongruent phases, 1989; and Xu et al., 1994). From this it follows that
either at the contact with the melt (now leucosome), all reactions observed in our samples must have
or at grain boundaries via melt infiltration. Melt- occurred at P > 6 kbar and T > 800 C (Fig. 6).
consuming reactions 5–7 require a combination of Hence, a first estimate is 6–8 kbar and 800–940 C,
cooling and decompression (Fig. 7a,b), which is con- similar to the data previously presented for the AZ
sistent with an uplift path. Garnet breakdown to and the KKB further south (Braun et al., 1996;
orthopyroxene + plagioclase via reaction (6b) or (6c) Chacko et al., 1996; Nandakumar & Harley, 2000).
probably occurred in the same P–T domain as reac- The previous estimates are restricted to the system
tions (2) and (3), possibly during decompression KFMASH. However, biotite contains significant
rather than during heating. Crystallization at the amounts of Ti and also of F (Table 1), both of which
solidus via reaction (8) represents the last cooling stabilize it to higher temperatures (e.g. Dooley &
stage, at unknown pressure. Patiño-Douce, 1996). By contrast, the presence of
In both pseudosections, the succession of assemb- plagioclase (An30) shifts all equilibria to lower tem-
lages defines a clockwise P–T path. The two pseudo- perature. We therefore assume, that the competing
sections overlap in P–T space and the slopes of effects of Ti, F, Na and Ca balance each other and will
individual P–T vectors in each of them are consistent. neither greatly shift the inferred P–T path nor signifi-
The absence of retrograde sillimanite gives an cantly affect our main conclusions. This is corrobor-
important constraint on the P–T path. It may indicate ated by thermobarometry (see below).
that the retrograde path is steeper than reaction B as its The KFMASH invariant points and univariant
reversal never occurred and passed between invariant curves are experimentally located in P–T space. In
points [Spl Opx] and [Qtz Opx] (see Fig. 6b), both addition, as the divariant network is fixed by Sch-
located close to 6 kbar and 800 C, unless kinetic reinemaker analysis, the location of one other curve
factors inhibited further reaction between in situ melt is enough to locate all curves in Fig. 8. In an attempt
and melanosome phases. to calibrate the pseudosections in P–T space, the
In the trivariant field for the assemblage Grt + location of the upper pressure limit of the divariant
Qtz + Opx + Crd + Liq, it is possible that the con- field for reaction Grt + Qtz + Liq ¼ Opx +
tinuous reaction Grt + Qtz + Liq ¼ Opx + Crd Crd + Kfs was calculated, using the thermodynamic
operated. As cordierite contains channel fluids, reaction data in Thermocalc (Holland & Powell, 1990) and
progress requires that some H2O is taken from the melt melt activity models modified from Holland et al.
phase. Whether this is possible or not, depends on the (1996). The water content of cordierite, crucial for
exact slope of the P–T path. Assuming that the slopes of the pressure location of this curve and for balancing
solidus and liquidus curves in Holtz & Johannes other reactions, has been estimated using a value of
(1994) are also valid for KFMASH melts, it would 3.2 for the ratio [wt% H2O in melt] ⁄ [wt% H2O in
require either near-isothermal decompression or a cordierite] (after Carrington & Harley, 1995) yielded
556 B. CENKI ET AL.
pressures between 7 and 8 kbar. However, using 10
P (kbar)
isopleths for the same reaction given by Spear et al.
(1999) gives slightly lower pressures of 6–7 kbar.
Hence the P–T path of the Achankovil cordierite
gneisses is composed of a clockwise loop at 6–8 kbar O
px
and 850–950 C followed by a near isothermal G
rt
Cr
Si
lL
decompression stage. d
K
fs
iq
Qtz
Thermobarometry
Liq
Kfs
Thermometers based on Fe–Mg exchange for ultra-
Crd
Sil
Li fs
K
high temperature metamorphism are often reset during
Bt
Grt
q
il
tS
Opx Spl Liq

s
Bt Grt Crd Kf
rd
cooling and the calculated temperatures are often
Gr
lC
Bt
Sp
underestimated (Frost & Chacko, 1989; Fitzsimons &
Harley, 1994). This is also indicated by the application
fs Qtz
Liq
of Grt–Opx Fe–Mg exchange thermometers (Harley, 6
Opx Crd
1984; Bhattacharya et al., 1991) to the Achankovil
Bt Grt K
cordierite gneisses Table 2, which gave too low tem-
peratures of 700–800 C at the inferred pressure of
7 kbar (Fig. 8). The only calibrated line that can be
used with some confidence is for Al in orthopyroxene
coexisting with garnet (Harley & Green, 1982). This T (˚C)
method gave peak temperatures of about 940 C at 4
7 kbar for incongruent pyroxene and coarse-grained 700 900
symplectites and 50–150 C less for fine-grained sym-
plectites (Fig. 8). Fig. 8. Thermobarometry based on Al in Opx coexisting with
Grt (shaded area: Harley & Green, 1982) and Fe-Mg exchange
Reaction (6b) would normally provide an inde- between Grt and Opx (Harley, 1984, vertical lines, and Bhatta-
pendent barometer (e.g. Newton & Perkins, 1982; charya et al. (1991), horizontal lines). For each calibration used,
Perkins & Chipera, 1985). However, its application the upper temperature box is the range of results for incongruent
to the AZ samples is not warranted, because of phases and coarse-grained symplectites, the lower temperature
(i) uncertainty in the presence of quartz at the time box is the range for fine-grained symplectites. Key KFMASH
invariant points and univariant lines from Fig. 6 are shown for
comparison, as well as the inferred P–T path.
Table 2. Results of Grt and Opx thermobarometry. of orthopyroxene and plagioclase growth and (ii) the
Fine symplectite Coarse symplectite
very low Ca content of garnet.
min. value max. value min. value max. value Thermobarometry using Al in orthopyroxene gives
HG82 Grt-rim Grt-core Grt-core Grt incongruent results that are consistent with the semiquantitative
Xgr 0.037 0.036 0.028 0.028
KFMASH grid (Figs 6 & 7). This suggests that the
Xalm 0.587 0.578 0.591 0.585 combined effect of Ti, F, Na and Ca balance and that
Xprp 0.376 0.386 0.381 0.387 the P–T shift due to these elements is negligible in our
Fe ⁄ Mg 1.559 1.497 1.550 1.513
Opx-rim Opx-rim Opx-core Opx incongruent samples. Hence, the KFMASH grid and pseudosec-
XFe 0.356 0.366 0.380 0.381 tions give reliable constraints on the shape of the P–T
XMg 0.644 0.634 0.620 0.619
XAlM1 0.120 0.123 0.129 0.173
path and on the absolute P–T conditions.
Fe ⁄ Mg 0.554 0.577 0.612 0.614
Calculated temperatures (C) at 7 kbar

DISCUSSION
T-HG 863 874 885 945
Grt-rim Grt-rim Grt-rim Grt-core
Xgr 0.035 0.036 0.033 0.029
The investigated cordierite gneisses of the AZ allow
XFe 0.587 0.578 0.591 0.565 the study of the effects of contemporaneous partial
XMg 0.378 0.386 0.377 0.406 melting and deformation in anatectic migmatites and
XFe ⁄ XMg 1.551 1.497 1.568 1.390
Opx-rim Opx-rim Opx-rim Opx-core
particularly the influence of deformation on melt
XFe 0.330 0.366 0.366 0.386 segregation and melt–mineral back reaction proces-
XMg 0.670 0.634 0.634 0.614 ses. Migmatization of the gneisses was triggered by
XFe ⁄ XMg 0.492 0.577 0.578 0.628
several biotite dehydration-melting reactions. These
Calculated temperatures (C) at 7 kbar
T-H 658 752 728 841
operated at different P–T conditions and, due to
T-B 686 768 747 842 primary chemical and mineralogical heterogeneity, in
different parts of the rocks. Syn-anatectic deforma-
T-HG ¼ T after Harley & Green (1982).
T-H ¼ T after Harley (1984).
tion was focused in selected domains resulting
T-B ¼ T after Bhattacharya et al. (1991), corrected calibration. in high-strain gneisses. They are characterized by
intense segregation and substantial loss of in situ Similarly, thermobarometry in granulites has often
melt. By contrast, melt escape in domains of low- been based on solid-solid equilibria alone. However,
strain was limited and most of the liquid crystallized even though a reaction like Grt + Qtz ¼ Opx + Pl
in situ. is well calibrated (e.g. Newton & Perkins, 1982;
A well known phenomenon of anatectic migmatites Perkins & Chipera, 1985), it cannot be applied when
is that leucocratic layers rarely match the modal plagioclase and quartz were components of the melt
composition of a granitic minimum melt. In most of phase (cf. Kriegsman & Hensen, 1998). In this study,
the AZ cordierite gneisses, the leucosomes display a the inferred P–T path is based on melt-present reac-
moderate (low-strain gneisses) or strong (high-strain tions, and its shape therefore deviates significantly
gneisses) depletion of alkali feldspar and thus have a from those of earlier studies.
granodioritic or tonalitic composition. In the high- Reliable P–T constraints on assemblages involving
strain gneisses, the shift towards a quartz and plagio- melt using the semiquantitative KFMASH petro-
clase dominated mineralogy could be explained by genetic grid (e.g. Fig. 7) and the Al-in-Opx thermo-
syn-anatectic deformation and extraction of rather barometer (Harley & Green, 1982) give maximum
large volumes of a Kfs-rich granitic melt. In the low- values of 940 C and 7 kbar. These are in fairly
strain gneisses, however, melt segregation is expected good agreement with the estimates of Nandakumar
to be very limited and this model is unsatisfactory. & Harley (2000), who obtained 6.5–7.0 kbar and
Similar discrepancies in migmatites elsewhere have 900 ± 20 C from garnet–orthopyroxene thermo-
been attributed to (i) large equilibration volumes barometry on charnockites and cordierite gneisses
(Powell & Downes, 1990); (ii) local segregation of from the same area. The data presented in this study
selected melt components (Harte et al., 1991); (iii) the thus provide further support that high to ultra-high
cumulate nature of leucosome (Ellis & Obata, temperature conditions prevailed in southernmost
1992); (iv) incorporation of restitic (e.g. plagioclase: India during the Pan-African orogeny. We did not
Johannes, 1985) or incongruent phases (e.g., garnet: find relics of a high-pressure stage, e.g. the Opx–Sil
Powell & Downes, 1990) into the leucosome; assemblage reported by Santosh (1987). In our view,
(v) consumption of alkali feldspar by later reactions orthopyroxene and sillimanite were not coexisting
and the crystallization of mica instead (Ashworth, phases, and the pressure estimates of 7–8 kbar
1985). The AZ cordierite gneisses, notably the derived from this and other studies represent maxi-
low-strain samples, show intense mineral-melt back mum values for Pan-African metamorphism in the
reaction textures between incongruent phases (garnet, region.
orthopyroxene, cordierite) and in situ crystallizing Incongruent phases produced along the prograde
melt, which may have prevented the crystallization of path and partly consumed on the retrograde path may
alkali feldspar and quartz and instead promoted bio- exhibit very typical retrograde compositional gradients
tite crystallization. Hence, the combination of plagi- (e.g. Mn enrichment towards garnet rim: Brown, 1983;
oclase-rich leucosomes and retrograde biotite growth Spear et al., 1999; Kohn & Spear, 2000). However,
fits Ashworth’s mechanism (1985). such zoning is absent from our samples (Fig. 5a). As
The first P–T path for cordierite gneisses from the temperatures from Fe–Mg exchange between garnet
Achankovil Zone was presented by Santosh (1987), and orthopyroxene are 100–150 C lower than those
who inferred a moderate to rapid isothermal decom- from Al-in-Opx thermometry, we conclude that the
pression at a maximum temperature of c. 750 C which Fe–Mg system was reset during cooling. In view of the
was followed by near-isobaric cooling at c. 5 kbar. very homogeneous XMg values for all ferromagnesian
Further south in the KKB, Satish-Kumar & Harley phases (garnet, cordierite, orthopyroxene, spinel) at
(1998) proposed an isobaric cooling path in the tem- this stage, the resetting must have been pervasive at
perature range of 840–750 C for calc-silicate samples. temperatures of about 650–700 C, i.e. still above the
These findings were supported and refined by Nanda- solidus. This indicates a stage of protracted heating
kumar & Harley (2000) who concluded that the initial during the Pan-African orogeny, which is supported by
part of the post-peak P–T path was dominated by the results of U–Pb monazite dating from granitic
near-isobaric cooling until temperatures of 775– garnet-biotite gneisses of the KKB (Braun & Bröcker,
800 C and pressures of 5.5–6 kbar were reached. pers. comm., 2000).
A subsequent stage of decompression to pressures of The Kerala Khondalite Belt (KKB) has several
c. 3 kbar at temperatures still above 700 C was characteristics in common with the Wanni Complex
inferred from garnet breakdown to cordierite and to (WC) of Sri Lanka, such as the presence of graphite in
orthopyroxene + plagioclase. veins (KKB: Soman et al., 1986; WC: Silva, 1985); and
A common feature to these published P–T paths is evidence for arrested charnockitization, commonly
that they are based solely on solid-solid reactions, even associated with late shear zones (KKB: Hansen et al.,
though the presence of a melt phase is acknowledged. 1987; Santosh et al., 1990; Raith & Srikantappa, 1993;
However, the presence of melt has a strong influence WC: Burton & O’Nions, 1990). The Achankovil Zone
on reaction line slopes and could greatly affect the (AZ) also has characteristics in common with the WC:
shape of the P–T path (Kriegsman & Hensen, 1998). (i) the presence of late-tectonic alkali-granites (Santosh
558 B. CENKI ET AL.
et al., 1989) with ages of c. 550 Ma (K-Ar Hbl AZ: et al., 1994; Brandon & Meen, 1995; Bartlett et al.,
Soman et al., 1983; U–Pb zircon data, WC: Baur et al., 1998; WC: Milisenda et al., 1994). As the AZ seems to
1991 and Hölzl et al., 1994); (ii) the abundance of share more characteristics with the WC than does the
cordierite gneisses in both high-grade domains (AZ: KKB, we propose to treat the AZ gneisses as a sep-
Sinha-Roy et al., 1984; Santosh, 1987; this paper; WC: arate tectonostratigraphic unit, rather than as a sub-
Prame & Pohl, 1994; Raase & Schenk, 1994); (iii) unit within the KKB. This view is supported by the
similar P–T paths (WC: Raase & Schenk, 1994; observation that cordierite gneisses have not been
AZ: our data); similar Nd model ages (AZ: Harris found in the KKB.
Cordierite gneisses similar to those in the AZ and
WC also occur along shear zones in southern Mada-
gascar (Nicollet, 1985; Markl et al., 2000) and in
400 km Lützow-Holm Bay, Eastern Antarctica (Hiroi et al.,
1990) and the latter ones show a P–T path
(Motoyoshi et al., 1989) that is virtually identical to
the one recorded in the WC (Raase & Schenk, 1994).
Hence, it seems warranted to postulate the existence
of a largely metasedimentary unit with Nd model ages
ranging between 2.0 and 1.0 Ga, containing cordierite
gneisses that record a clockwise P–T path culminating
N at c. 7 kbar, 850–950 C, which is exposed from
southern Madagascar via South India and Sri Lanka
to the Lützow-Holm Bay in Eastern Antarctica
(Fig. 9).
Madagascar Indian
Archean craton
CONCLUSIONS
Cordierite gneisses from the Achankovil Zone in South
Palghat-Cauvery
shear zone system India show a variety of partial melting reactions fol-
lowed by back reaction between restite and in situ
Eastern Ghats
crystallizing melt. The textural and chemical evolution
of individual layers is strongly influenced by syn-
.
anatectic deformation, which controlled melt segrega-
tion and therefore the operation or inhibition of back
reaction. A common feature of all domains is that,
Madurai either due to segregation or back reaction processes,
block
Kerala Wanni the leucosomes rarely have preserved the original melt
Khondalite
Napier
composition. In addition, thermobarometers involving
Belt phases (quartz, K-feldspar, plagioclase) that are melt
Complex
components should be applied with caution. The
Achankovil Zone
application of a KFMASH petrogenetic grid to
melt-present assemblages offers the possibility to
obtain semiquantitative information on the absolute
Gunnerus ridge
Sri Lanka Rayner Complex P–T conditions and on the shape of the P–T path
during their development. The resulting clockwise P–T
path culminated at c. 6–8 kbar and 900–950 C.
Lützow-Holm Complex ACKNOWLEDGEMENTS

We thank Drs G.R. Ravindra Kumar, S. Sukumaran,
S. Karmakar, B. Gopakumar and M. v.d. Gugten for
their valuable support during field work. BC warmly
Yamato-Belgica
Sør Rondane Complex
thanks C. Nicollet. B.C and I.B. acknowledge financial
Mountains support by the German Science Foundation (DFG)
and the German Academic Exchange Office (DAAD).
LMK acknowledges financial support by the Dutch
Fig. 9. Gondwana fit of India, Antarctica, Madagascar and Dr Schürmann Foundation and the Finnish Academy
Sri Lanka showing trend lines of upright folds (short black lines)
and dominant stretching lineations (short arrows) and possible of Science. N. Hudson and D. Waters are thanked
margins of a late sinistral strike-slip belt affecting a large part of for their constructive reviews. D. Robinson is thanked
the Mozambique Belt (after Kriegsman, 1995; and refs. therein). for additional comments and editorial handling.
REFERENCES Within East Gondwana (eds Santosh, M. & Yoshida, M.).

Gondwana Research Group Memoir, 3, 157–165. Field Science
Ashworth, J. R., 1985. Introduction. In: Migmatites (ed. Publishers, Osaka.
Ashworth, J. R.), pp. 36–85. Blackie and Son, Glasgow. Chacko, T., Ravindra Kumar, G. R. & Newton, R. C., 1987.
Bartlett, J. M., Dougherty-Page, J. S., Harris, N. B. W., Metamorphic P-T conditions of the Kerala (South India)
Hawkesworth, C. J. & Santosh, M., 1998. The application Khondalite Belt: a granulite facies supracrustal terrain.
of single zircon evaporation and Nd model ages to the Journal of Geology, 96, 343–358.
interpretation of polymetamorphic terrains: an example from Choudhary, A. K., Harris, N. B. W., Van Calsteren, P. &
the Proterozoic mobile belt of south India. Contributions to Hawkesworth, C. J., 1992. Pan-African charnockite formation
Mineralogy and Petrology, 131, 181–195. in Kerala, South India. Geological Magazine, 129, 257–264.
Baur, N., Kröner, A., Todt, W., Liew, T. C. & Hofmann, A. W., Clemens, J. D. & Droop. G. T. R, 1998. Fluids, P-T paths and
1991. U-Pb isotopic systematics of zircons from prograde and the fates of anatectic melts in the Earth’s crust. In: Generation
retrograde transition zones in high-grade orthogneisses Sri of Granitic Rocks and Deep Crustal Processes (eds Clemens,
Lanka. Journal of Geology, 99, 527–545. J. D. & Hutton, D. H.). Lithos, 44, 21–36.
Berger, A. & Kalt, A., 1999. Structures and melt fractions as Dooley, D. F. & Patiño-Douce, E., 1996. Fluid-absent melting of
indicators of rheology in cordierite-bearing migmatites of F-rich phlogopite + rutile + quartz. American Mineralogist,
the Bayerische Wald (Variscan Belt, Germany). Journal of 81, 202–212.
Petrology, 40, 1699–1719. Drury, S. A., Harris, N. B. W., Holt, R. W., Reeves-Smith,
Bertrand, P., Ellis, D. J. & Green, D. H., 1991. The stability of G. W. & Wightman, R. T., 1984. Precambrian tectonics and
sapphirine-quartz and hypersthene-sillimanite-quartz assem- crustal evolution in South India. Journal of Geology, 92, 1–20.
blages: an experimental investigation in the system FeO-MgO- Drury, S. A. & Holt, R. W., 1980. The tectonic framework of the
Al2O3-SiO2 under H2O and CO2 conditions. Contributions to South Indian craton: a reconnaissance involving LANDSAT
Mineralogy and Petrology, 108, 55–71. imagery. Tectonophysics, 65, T1–T5.
Bhattacharya, A., Krishnakumar, K., Raith, M. & Sen, S. K., Ebadi, A. & Johannes, W., 1991. Beginning of melting and
1991. An improved set of a-X parameters in Fe-Mg-Ca composition of first melts in the system Qz-Ab-Or-H2O-CO2.
garnets and refinement of the orthopyroxene-garnet thermo- Contributions to Mineralogy and Petrology, 106, 286–295.
meter and the orthopyroxene-garnet-plagioclase-quartz baro- Ellis, D. J. & Obata, M., 1992. Migmatite and melt segregation
meter. Journal of Petrology, 32, 629–656. at Cooma, New South Wales. Transactions of the Royal
Brandon, A. D. & Meen, J. K., 1995. Nd isotopic evidence for Society of Edinburgh, Earth Sciences, 83, 95–106.
the position of southernmost Indian terranes within East Fitzsimons, I. C. W., 1996. Metapelitic migmatites from Bratt-
Gondwana. Precambrian Research, 70, 269–280. strand Bluffs, East Antarctica – metamorphism, melting
Braun, I., Montel, J.-M. & Nicollet, C., 1998. Electron micro- and exhumation of the mid crust. Journal of Petrology, 37,
probe dating of monazites from high-grade gneisses and 395–414.
pegmatites of the Kerala Khondalite Belt, southern India. Fitzsimons, I. C. W. & Harley, S. L., 1994. The influence of
Chemical Geology, 146, 65–85. retrograde cation exchange on granulite P-T estimates and a
Braun, I., Raith, M. & Ravindra Kumar, G. R., 1996. convergence technique for the recovery of peak metamorphic
Dehydration-melting phenomena in leptynitic gneisses and conditions. Journal of Petrology, 35, 543–576.
the generation of leucogranites: a case study from the Kerala Frost, B. R. & Chacko, T., 1989. The granulite uncertainty
Khondalite Belt, southern India. Journal of Petrology, 37, principle: Limitations on thermobarometry in granulites.
1285–1305. Journal of Petrology, 97, 435–450.
Brown, M., 1983. The petrogenesis of some migmatites from the Greenfield, J. E., Clarke, G. L. & White, R. W., 1998. A
Presqu’ile de Rhuys, Southern Brittany, France. In: Migma- sequence of partial melting reactions at Mt Stafford, central
tites, Melting and Metamorphism (eds Atherton, M. P. & Australia. Journal of Metamorphic Geology, 16, 363–378.
Gribble, C. D.), pp. 174–200. Shiva Geology Series. Shiva Hansen, E. C., Newton, R. C., Prame, W. K. B. N. & Ravindra
Publications, Nantwich. Kumar, G. R., 1987. Charnockite in the making in southern
Brown, M., Averkin, Y. A., McLellan, E. L. & Sawyer, E. W., India and Sri Lanka. Contributions to Mineralogy and
1995. Melt segregation in migmatites. In: Mechanisms and Petrology, 96, 225–244.
Consequences of Melt Segregation from Crustal Protoliths (eds Harley, S. L., 1984. An experimental study of the partitioning of
Brown, M., Rushmer, T. & Sawyer, E. W.). Journal of Fe and Mg between garnet and orthopyroxene. Contributions
Geophysical Research, B 100, 15 655–15 679. to Mineralogy and Petrology, 86, 359–373.
Buhl, D., Grauert, B. & Raith, M., 1983. U-Pb zircon dating of Harley, S. L. & Green, D. H., 1982. Garnet-orthopyroxene
Archaean rocks from the south Indian craton: results from the barometry for granulites and peridotites. Nature, 300, 697–701.
amphibolite to granulite facies transition zone at. Kabbal Harris, N. B. W., Santosh, M. & Taylor, P. N., 1994. Crustal
Quarry, Southern Karnataka Fortschritte der Mineralogie 61, evolution in South India: Constraints from Nd isotopes.
Beiheft, 1, 43–45. Journal of Geology, 102, 139–150.
Burton, K. W. & O’Nions, R. K., 1990. The timescale and Harte, B., Pattison, D. R. M. & Linklater, C. M., 1991. Field
mechanism of granulite formation at Kurunegala Sri Lanka. relations and petrography of partially melted pelitic and
Contributions to Mineralogy and Petrology, 106, 66–89. semi-pelitic rocks. In: Equilibrium and Kinetics in Contact
Carrington, D. P. & Harley, S. L., 1995. Partial melting and Metamorphism: the Ballachulish Igneous Complex and
phase relations in high-grade metapelites: An experimental its Aureole (eds Voll, G., Töpel, J., Pattison, D. R. M. &
petrogenetic grid in the KFMASH system. Contributions to Seifert, F.), pp. 181–209. Springer-Verlag, Berlin.
Mineralogy and Petrology, 120, 270–291. Hensen, B. J., 1971. Theoretical phase relations involving
Carrington, D. P. & Harley, S. L., 1996. Cordierite as a monitor cordierite and garnet in the system MgO-FeO-Al2O3-SiO2.
of fluid and melt H2O contents in the lower crust: An Contributions to Mineralogy and Petrology, 33, 191–214.
experimental calibration. Geology, 24, 647–650. Hensen, B. J. & Green, D. H., 1971. Experimental study of the
Carrington, D. P. & Watt, G. R., 1995. A geochemical and stability of cordierite and garnet in pelitic compositions at
experimental study of the role of K-feldspar during water- high pressures and temperatures. I. Compositions with excess
undersaturated melting of metapelites. Chemical Geology, 122, aluminosilicate. Contributions to Mineralogy and Petrology,
59–76. 33, 309–330.
Chacko, T., Lamb, M. & Farquhar, J., 1996. Ultra-high Hiroi, Y. & Motoyoshi, Y., 1990., eds. Study of Geologic
temperature metamorphism in the Kerala Khondalite Belt. Correlation Between Sri Lanka and Antarctica (1988–89).
In: The Archaean and Proterozoic Terrains in Southern India Chiba University, Japan.
560 B. CENKI ET AL.
Holland, T. J. B., Babu, E. V. S. S. K. & Waters, D. J., 1996. Phase Bay region, East Antarctica. In: Evolution of Metamorphic
relations of osumilite and dehydration melting in pelitic rocks – Belts (eds Daly, S., Cliff, R. A. & Yardley, B. W. D.), 325–329.
A simple thermodynamic model for the KFMASH system. Geological Society Special Publication, 43. Blackwell, Oxford,
Contributions to Mineralogy and Petrology, 124, 383–394. UK.
Holland, T. J. B. & Powell, R., 1990. An enlarged and updated Nandakumar, V. & Harley, S. L., 2000. A reappraisal of the
internally consistent thermodynamic dataset with uncer- Pressure-Temperature Path of Granulites from the Kerala
tainties and correlations: the system K2O-Na2O-CaO- Khondalite Belt, southern India. Journal of Geology, 108,
MgO-MnO-FeO-Fe2O3-Al2O3-TiO2-SiO2-C-H2-O2. Journal 687–703.
of Metamorphic Geology, 8, 89–124. Newton, R. C. & Perkins, D. III, 1982. Thermodynamic
Holtz, F. & Johannes, W., 1994. Maximum and minimum water calibration of geobarometers based on the assem-
contents of granitic melts: implications for chemical and blages garnet-plagioclase-orthopyroxene (clinopyroxene)-
physical properties of ascending magmas. In: Crustal Anatexis quartz. American Mineralogist, 67, 203–222.
and Ascent of Felsic Magmas (ed. Brown, M.). Lithos, 32, Nicollet, C., 1985. Les gneiss rubanés à cordierite et grenat
149–159. d’Ihosy: un marqueur thermo-barométrique dans la Sud de
Hölzl, S., Hofmann, A. W., Todt, W. & Köhler, H., 1994. U-Pb Madagascar. Precambrian Research, 28, 175–185.
geochronology of the Sri Lankan basement. In. Tectonic, Perkins, D. III & Chipera, S. J., 1985. Garnet-orthopyroxene-
Metamorphic and Isotopic Evolution of Deep Crustal Rocks, quartz barometry: refinement and application to the English
with Special Emphasis on Sri Lanka (eds Raith, M. & Hoernes, River subprovinces and the Minnesota River Valley. Con-
S.). Precambrian Research, 66, 123–149. tributions to Mineralogy and Petrology, 89, 69–80.
Johannes, W., 1985. The significance of experimental studies for Pouchou, J. L. & Pichoir, F., 1984. A new model for quantitative
the formation of migmatites. In: Migmatites (ed. Ashworth, X-ray microanalysis, Part I: Application to the analysis of
J. R.), pp. 63–85. Blackie and Son, Glasgow. homogeneous samples. La Recherche Aerospatiale, 3, 167–192.
Jones, K. A. & Brown, M., 1990. High-temperature ÔclockwiseÕ Powell, R. & Downes, J., 1990. Garnet porphyroblast-bearing
P-T paths and melting in the development of regional leucosomes in metapelites: mechanisms, phase diagrams, and
migmatites; an example from southern Brittany, France. an example from Broken Hill, Australia. In: High-Tempera-
Journal of Metamorphic Geology, 8, 551–578. ture Metamorphism and Crustal Anatexis (eds Ashworth, J. R.
Kohn, M. J. & Spear, F. S., 2000. Retrograde net-transfer & Brown, M.), pp. 105–123. The Mineralogical Society Series
reaction insurance for pressure-temperature estimates. Geol- 2, Unwin-Hyman, London.
ogy, 28, 1127–1130. Powell, C. McA., Roots, S. R. & Veevers, J. J., 1988. Pre-
Kohn, M. J., Spear, F. S. & Valley, J. W., 1997. Dehydration breakup continental extension in East Gondwanaland and the
melting and fluid recycling during metamorphism: Rangeley early opening of the eastern Indian Ocean. Tectonophysics,
Formation, New Hampshire, USA. Journal of Petrology, 38, 155, 261–283.
1255–1277. Prame, W. K. B. N. & Pohl, J., 1994. Geochemistry of pelitic and
Kretz, R., 1983. Symbols for rock-forming minerals. American psammopelitic Precambrian metasediments from southwes-
Mineralogist, 68, 277–279. tern Sri Lanka: implications for two contrasting source
Kriegsman, L. M., 1995. The Pan-African event in East terrains and tectonic settings. In: Tectonic, Metamorphic and
Antarctica: a view from Sri Lanka and the Mozambique Belt. Isotopic Evolution of Deep-Crustal Rocks, with Special
Precambrian Research, 75, 263–277. Emphasis on Sri Lanka (eds Raith, M. & Hoernes, S.).
Kriegsman, L. M., 2001. Partial melting, partial melt extraction, Precambrian Research, 66, 223–244.
and partial back reaction in anatectic migmatites. In: Prograde Raase, P. & Schenk, V., 1994. Petrology of granulite-facies
and Retrograde Processes in Migmatites (ed. Kriegsman, metapelites of the Highland Complex Sri Lanka: implications
L. M.). Lithos, 56, 75–96. for the metamorphic zonation and the P-T path. In: Tectonic,
Kriegsman, L. M. & Hensen, B. J., 1998. Back reaction between Metamorphic and Isotopic Evolution of Deep-Crustal Rocks, with
restite and melt: Implications for geothermobarometry and Special Emphasis on Sri Lanka (eds Raith, M. & Hoernes, S.).
pressure-temperature paths. Geology, 26, 1111–1114. Precambrian Research, 66, 265–294.
Lawver, L. A. & Scotese, C. R., 1987. A revised reconstruction Radakrishna, T., Mathai, J. & Yoshida, M., 1990. Geology and
of Gondwanaland. In: Gondwana Six: Structure, Tectonics, structure of the high-grade rocks from Punalur-Achankovil
and Geophysics (ed. McKenzie, G. D.), pp. 17–23. Geophysical sector, south India. Journal of the Geology Society of India, 35,
Monograph, 40. American Geophysical Union, Washington 263–272.
DC. Raith, M., Karmakar, S. & Brown, M., 1997. Ultra-high-
Markl, G., Bäuerle, J. & Grujic, D., 2000. Metamorphic temperature metamorphism and multistage decompressional
evolution of Pan-African granulite facies metapelites from evolution of sapphirine granulites from the Palni Hills Ranges,
southern Madagascar. Precambrian Research, 102, 47–68. southern India. Journal of Metamorphic Geology, 15, 379–399.
Milisenda, C. C., Liew, T. C., Hofmann, A. W. & Köhler, H., Raith, M. & Srikantappa, C., 1993. Arrested charnockite
1994. Nd isotopic mapping of the Sri Lanka basement: formation at Kottavattam, southern India. Journal of
update, and additional constraints from Sr isotopes. In: Metamorphic Geology, 11, 815–832.
Tectonic, Metamorphic and Isotopic Evolution of Deep-Crustal Rajesh, H. M., Santosh, M. & Yoshida, M., 1998. Dextral Pan-
Rocks, with Special Emphasis on Sri Lanka (eds Raith, M. & African shear along the southwestern edge of the Achankovil
Hoernes, S.). Precambrian Research, 66, 95–110. shear belt, South India: constraints on Gondwana reconstruc-
Milisenda, C. C., Liew, T. C., Hofmann, A. W. & Kröner, A., tions: a discussion; with reply by Sacks, P. E., Nambiar,
1988. Isotopic mapping of age provinces in Precambrian high- C. G. & Walters, L. Journal of Journal of Geology, 106,
grade terranes: Sri Lanka. Journal of Geology, 96, 608–615. 105–114.
Miller, J. S., Santosh, M., Pressley, R. A., Clements, A. S. & Sacks, P. E., Nambiar, C. G. & Walters, L. J., 1997. Dextral
Rogers, J. J. W., 1997. A Pan-African thermal event in Pan-African shear along the southwestern edge of the
southern India. Journal of Southeast Asian Earth Sciences, 14, Achankovil Shear Belt, South India: constraints on
127–136. Gondwana Reconstructions. Journal of Geology, 105,
Montel, J. M., Weber, C. & Pichavant, M., 1986. Biotite- 275–284.
sillimanite-spinel assemblages in high-grade metamorphic Santosh, M., 1987. Cordierite gneisses of south Kerala, India.
rocks: occurences, chemographic analysis and thermobaro- Petrology, fluid inclusion and implications on uplift history.
metric interest. Bulletin de Mine´ralogie, 109, 555–573. Contributions to Mineralogy and Petrology, 97, 343–356.
Motoyoshi, Y., Matsubara, S. & Matsueda, H., 1989. P-T Santosh, M., Harris, N. B. W., Jackson, D. H. & Mattey, D. P.,
evolution of the granulite-facies rocks of the Lützow-Holm 1990. Dehydration and incipient charnockite formation: a
phase equilibria and fluid inclusion study from South India. Stüwe, K., 1997. Effective bulk composition changes due to
Journal of Geology, 98, 915–926. cooling; a model predicting complexities in retrograde reaction
Santosh, M., Iyer, S. S., Vasconcellos, M. B. A. & Enzweiler, J., textures. Contributions to Mineralogy and Petrology, 129,
1989. Late Precambrian alkaline plutons in southwest India: 43–52.
geochronologic and rare-earth element constraints on Pan- Stüwe, K. & Powell, R., 1989. Metamorphic segregations
African magmatism. Lithos, 24, 65–79. associated with garnet and orthopyroxene porphyroblast
Satish-Kumar, M. & Harley, S. L., 1998. Reaction textures in growth: two examples from the Larsemann Hills, East
scapolite-wollastonite-grossular calc-silicate rock from the Antarctica. Contributions to Mineralogy and Petrology, 103,
Kerala Khondalite Belt, southern India: evidence for high- 523–530.
temperature metamorphism and initial cooling. Lithos, 44, Thompson, A. B., 1976. Mineral reactions in pelitic rocks. I.
83–99. Prediction of P-T-X (Fe-Mg) phase relations. II. Calculation
Shiraishi, K., Ellis, D. J., Hiroi, Y., Fanning, C. M., Motoyoshi, Y. of some P-T-X (Fe-Mg) phase relations. American Journal of
& Nakai, Y., 1994. Cambrian orogenic belt in East Antarctica Science, 276, 401–454.
and Sri Lanka: Implications for Gondwana assembly. Journal of Vielzeuf, D. & Holloway, J. R., 1988. Experimental determina-
Geology, 102, 47–65. tion of the fluid-absent melting relations in the pelitic system.
Silva, K. K. M. W., 1985. Geological structure of Sri Lanka – Contributions to Mineralogy and Petrology, 98, 257–276.
A new perspective. In: Recent Advances in the Geology of Vielzeuf, D. & Montel, J.-M., 1994. Partial melting of
Sri Lanka (eds Dissanayake, C. B. & Cooray, P. G.), pp. 101– metagreywackes; Part 1, Fluid-absent experiments and phase
108. International Center for Training and Exchanges in relationships. Contributions to Mineralogy and Petrology, 117,
the Geosciences, Orleans, Occasional Publications, 6. 375–393.
Sinha-Roy, S., Mathai, J. & Narayanaswamy, 1984. Structure Waters, D. J., 1988. Partial melting and the formation of
and metamorphic characteristics of cordierite-bearing gneiss granulite facies assemblages in Namaqualand, South Africa.
of South Kerala. Journal of the Geological Society of India, 25, Journal of Metamorphic Geology, 6, 387–404.
231–244. Waters, D. J., 2001. The significance of prograde and retrograde
Soman, K., Lobzova, R. V. & Sivadas, K. M., 1986. Geology, quartz-bearing intergrowth microstructures in partially melted
genetic types and origin of graphite in South Kerala, India. granulite-facies rocks. In: Kriegsman, L. M., ed. Prograde and
Economic Geology, 81, 997–1002. Retrograde Processes in Migmatites, Lithos, 56, 97–110.
Soman, K., Narayanaswamy & Van Schmus, W. R., 1995. Xu, G., Will, T. M. & Powell, R., 1994. A calculated
Preliminary U-Pb zircon ages of high-grade rocks in southern petrogenetic grid for the system K2O-FeO-MgO-Al2O3-SiO2-
Kerala, India. Journal of the Geological Society of India, 45, H2O, with particular reference to contact-metamorphosed
127–136. pelites. Journal of Metamorphic Geology, 12, 99–119.
Soman, K., Santosh, M. & Golubyev, V. N., 1983. Early Yoshida, M., Funaki, M. & Vitanage, P. W., 1992. Proterozoic
Paleozoic I-Type granite from central Kerala and its bearing to Mesozoic East Gondwana: the juxtaposition of India-Sri
on possible mineralizations. Indian Journal of Earth Science, Lanka-Antarctica. Tectonics, 11, 381–391.
10, 137–141. Zen, E.-An, 1966. Construction of pressure-temperature dia-
Spear, F. S. & Cheney, J. T., 1989. A petrogenetic grid for pelitic grams for multicomponent systems after the method of
schists in the system SiO2-Al2O3-FeO-MgO-K2O-H2O. Schreinemakers: a geometric approach. United States Geo-
Contributions to Mineralogy and Petrology, 101, 149–165. logical Survey Bulletin, 1225, 1–56.
Spear, F. S., Kohn, M. J. & Cheney, J. T., 1999. P-T paths from
anatectic pelites. Contributions to Mineralogy and Petrology,
134, 17–32. Received 12 March 2001; revision accepted 13 January 2002.

Cenki 2002

Uploaded by

Copyright:

Available Formats

Cenki 2002

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cenki 2002

Uploaded by

Copyright:

Available Formats

J. metamorphic Geol.

, 2002, 20, 543–561

Melt-producing and melt-consuming reactions in the Achankovil

Meen, 1995; Bartlett et al., 1998) and those in the

Fig. 2. Outcrop photograph, showing high abundance of

Fig. 1. Sketch map of the northwestern edge of the AZ and the

Kfs (g) (h)

Fig. 3. Photomicrographs documenting the textural and miner-

Table 1. Representative chemical compositions of major minerals of the cordierite gneisses.

normalized to 12 oxygen normalized to 18 oxygen normalized to 6 oxygen

XFe 0.54 0.55 0.56 0.61 XFe 0.16 0.15 0.19

XMg ¼ Mg ⁄ (Mg + Fe) XMg ¼ Mg ⁄ (Mg + Fe + Al).

XMg 0.83 0.74 0.74

Xmag 0.05 0.02 0.03 Xhem 0.10 XMg ¼ Mg ⁄ (Mg + Fe)

XMg ¼ Mg ⁄ (Mg + Fe); uncorrected for Fe3+.

component of 4–7 mole%. Ferric iron has been 1.00

calculated using stoichiometric constraints giving (a)

values are lower for leucosome biotite than for inclu-

800 900 1000 1100 [Qtz Opx] E

Grt Bt Qtz Kfs Liq (a) (b)

Bt Sil Kfs Qtz

Bt Qtz (Grt Kfs)

Bt Sil Qtz Kfs

tz Grt Spl Crd Bt Liq

Grt Spl Liq

Bt Sil Crd Kfs Liq

Bt Sil Crd Kfs

P Bt Crd Qtz Kfs Liq

Opx Crd Qtz Kfs Liq

pressures between 7 and 8 kbar. However, using 10

Opx Spl Liq

Calculated temperatures (C) at 7 kbar

Lützow-Holm Complex ACKNOWLEDGEMENTS

REFERENCES Within East Gondwana (eds Santosh, M. & Yoshida, M.).

You might also like