Reaction textures

In this section, the mutual relations between the different phases and their compositions are described (see Tables 26). In view of their diversity and complexity, the oxide intergrowths are discussed separately.

Table 2. Representative mineral compositions from the members of the layered magmatic complex

Ferromagnesian layers in the migmatitic khondalite show mineralogical variations even within a few centimetres. Two associations are identified on the basis of characteristic minerals. Association I can be distinguished from the other by the presence of quartz and absence of porphyroblastic corundum. The mineralogy of the two associations is given below:

Association I: garnet-spinel (with magnetite)-sillimanite-quartz-(Fe-Ti) oxides-perthite-plagioclase-biotite-sulphide minerals;

Association II: garnet-spinel (with magnetite)-corundum-sillimanite-(Fe-Ti) oxide-perthite-plagioclase-biotite-sulphide minerals.

In most places the mutual relations among the phases are partially to completely obliterated as a result of superimposed deformation. Replacement of the anhydrous phases, such as garnet, spinel and perthite, by biotite + quartz, biotite + quartz + sillimanite, or biotite + sillimanite is frequently observed. This retrograde biotite defines a schistosity which often swerves around and replaces the porphyroblastic phases. This schistosity is parallel to the regional pervasive foliation S₂. However, a few zones of relatively low strain preserve pre-deformational reaction textures that are described below.

Two generations of garnet, one porphyroblastic [hereafter garnet (1)] and the other coronal [hereafter garnet (2)] occur in both the associations. In association I, garnet (1) includes one or more of the phases biotite, sillimanite, quartz and plagioclase. Garnet (1) is frequently embayed by lobate spinel. These spinel grains are rimmed against quartz by garnet (2) and sillimanite of various thickness (Fig. 3a). Srogi et al., (1993) interpreted a similar texture as two-stage garnet growth, i.e. before and after the formation of spinel. In association II, biotite and sillimanite are the dominant inclusions in garnet (1), quartz being present in only a few garnet (1) grains. In places, small granular inclusions of spinel with one or both of the phases biotite and sillimanite occur in the core of garnet (1) (Fig. 4a). Commonly, garnet (1) forms clusters with porphyroblastic corundum and spinel [hereafter, corundum (1) and spinel (1)] (Fig. 4b). Garnet (1) is often replaced by coarse intergrowths of spinel and sillimanite at the contact of corundum (1) (Fig. 4a).

Figure 3. Photomicrographs illustrating reaction textures in association I. (a) Garnet (1) (G) is embayed by spinel (Sp) which is in turn rimmed by garnet (2) and sillimanite (S) against quartz (Q). Coarse alkali feldpar (F) and biotite (B) inclusion in garnet (1) are also seen. Bar represents 200 µm. (b) Porphyroblastic ilmenite (black) is rimmed by an intergrowth of garnet (2) and granular ilmenite against quartz (Q). [Also note the lamellar intergrowth of corundum (C) and ilmenite (black).] Bar represents 200 µm.

Figure 4. Photomicrographs illustrating reaction textures in association II. (a) Garnet (1) (G) is replaced by a coarse intergrowth of sillimanite (S) and spinel (Sp) against corundum (1) (C). Note the inclusion of granular early spinel and biotite in garnet (1). F, alkali feldspar; B, retrograde biotite. Bar represents 200 µm. (b) Cluster of garnet (1), corundum (C) and spinel (Sp). F and B are retrograde alkali feldspar and biotite. Bar represents 200 µm. (C) Coarse spinel (grey base) with deformed dendritic magnetite lamellae (white). A slab of corundum (C) in spinel. [Note the localization of magnetite granules (M) along the contact of spinel and the corundum lath.] Bar represents 100 µm. (d) Coarse aggregate of spinel (Sp) and ilmenite (I). The dendritic lamellae of magnetite in spinel and fine Ti-haematite lamellae (bright) in ilmenite are also seen. [Note the sharp contact among the oxide grains.] Outline of the pristine spinel is still present (see text). Sillimanite (S) and garnet (G) are similar to those in (a). Bar represents 50 µm. (e) Composite aggregate of corundum (C), spinel (Sp) and ilmenite (I). Corundum forms lamellar intergrowth with ilmenite. Bar represents 50 µm. (f) Garnet (2) (G) and sillimanite (S) separating spinel (Sp) from ilmenite (I). Bar represents 50 µm.

Spinel is hercynitic and has three modes of occurrence. Small granular spinel occurs as inclusions along with biotite with or without sillimanite in the core of garnet (1) in association II (Fig. 4a). This spinel is likely to have formed before garnet (1). The most abundant variety is porphyroblastic [i.e. spinel (1)] that embays garnet (1) in both the associations. This variety frequently forms coarse aggregates with ilmenite and invariably shows crystallographically controlled intergrowths with magnetite in both the associations. Details of the oxide intergrowths will be discussed separately. In association I, spinel (1) is separated from quartz by either garnet (2) or a compound corona of sillimanite and garnet (Fig. 3a). In association II, spinel (1) often forms coarse intergrowths with sillimanite laths that replace garnet (1) (Fig. 4a).

Coarse ilmenite in association I is rimmed by garnet (2) against plagioclase and quartz. Garnet (2) is found to be intergrown with granular ilmenite (Fig. 3b).

Two generations of biotite are identified. Primary biotite that existed before garnet (1) mostly occurs as inclusions in the latter. In none of the associations does primary biotite coexist with sillimanite or quartz in the matrix. Secondary biotite is profusely developed in zones of intense deformation and defines the S₂ fabric. This biotite replaces garnet (1) and other anhydrous phases along cracks and fractures. In places, delicate intergrowths of sillimanite and secondary biotite are observed.

Corundum occurs in a variety of forms. Porphyroblastic corundum [i.e. corundum (1)] coexists with garnet (1) and spinel (1) in association II. Corundum also forms complex intergrowths with spinel and ilmenite in this association.

K-feldspar in both the associations is perthitic with numerous blebs and/or lamellae of plagioclase. It is mostly concentrated in the leucocratic bands along with quartz and plagioclase. A few grains of plagioclase are antiperthitic.

Sillimanite exhibits several textural varieties. The earliest occurs as inclusions in garnet (1) in both the associations. In association I sillimanite rims spinel (1) against quartz (Fig. 3a). In association II, haphazardly oriented coarse blades of sillimanite are intergrown with spinel and replace garnet (1) (Fig. 4a).

Oxide minerals in the study rocks show spectacular intergrowths. Spinel (1) in both the associations forms dendritic intergrowths with magnetite of variable thickness and length (Fig. 4c). The coarsest magnetite lamellae are invariably deformed to various extents and have tensional cracks. In places laths of corundum and/or sillimanite are also present in spinel (1) (Fig. 4c). In places, granules of magnetite often occur along the spinel-corundum contacts (Fig. 4c). The volume proportion of magnetite in spinel (1) never exceeds 15%. Magnetite never forms free grains in the matrix. This feature, together with rimming of spinel (1) by coronitic garnet, suggests that magnetite exsolved after the formation of the coronal garnet (see Waters, 1991). Coarse ilmenite grains invariably show fine lamellae of Ti-haematite (Fig. 4d). This feature is common in many magmatic and high-grade metamorphic rocks and is formed as a result of unmixing from a homogeneous Fe₂O₃-rich ilmenite during cooling (Haggerty, 1976). Spinel (1) in both associations often forms coarse aggregates with ilmenite (Figs 3 and 4). The oxide-oxide contacts of the aggregates are straight. In most places, these aggregates preserve the shape of the original spinel grain, suggesting that they are the products of unmixing of a homogeneous spinel rich in TiO₂ (Haggerty, 1976; Waters, 1991). As the solubility of ilmenite in the spinel structure or vice versa is insignificant, the pristine spinel must have been rich in Fe₂TiO₄ (ulvöspinel). This component upon oxidation produced the present oxide aggregates [the granular exsolution of Buddington & Lindsley, (1964); see Haggerty, (1976) and Waters, (1991)]. The volume proportion of ilmenite in most of these aggregates ranges from 20% to >35%, suggesting a significant proportion of Fe₂TiO₄ component in the original spinel (>19 mol %, Table 3). This is much higher than most of the other reported occurrences [see Waters, (1991) and Fitzsimons, (1996)]. Some composite oxide aggregates show spectacular intergrowths of corundum-spinel-ilmenite (Fig. 4e). In these aggregates corundum forms lamellar intergrowths with ilmenite which have sharp and straight boundaries with spinel. The spinel shows dendritic magnetite lamellae whereas ilmenite exsolves Ti-haematite lamellae. Mutual relations of the oxide phases in such composite grains also suggest the former existence of a homogeneous Fe-Ti-Mg-Al spinel. However, the corundum-ilmenite lamellar intergrowths also occur without spinel in places (Fig. 3b). Unlike other reported spinel-bearing occurrences (see Waters, 1991), the Fe-Ti-Al oxide aggregates are not rimmed by continuous garnet coronas and, in places, spinel and ilmenite are separated by garnet (2) and/or sillimanite (Fig. 4f).

Table 3. Representative spinel compositions

Mineral chemistry

Chemical composition of the phases was determined with a CAMECA MICROBEAM electron microprobe at the University of Bonn. The instrument was operated with 15 kV accelerating voltage, 1-2 mm beam diameter and 15 nA current. However, for area scans the beam diameter was broadened up to 50 mm. At least four area scans were measured for each reintegration. Natural mineral standards were used and the raw electron probe microanalysis (EPMA) data were corrected by the PAP procedure (Pouchou & Pichoir, 1984).

Spinel is primarily an MgAl₂O₄-FeAl₂O₄ solid solution with X_Mg [= Mg/(Mg + Fe²⁺)] varying from 0·38 to 0·43 (Table 3). ZnO content in spinel is always low (<1 wt %) except in the small spinel granules in garnet (1), which contain higher amounts of ZnO (3·5 wt %, Table 3). The spot analyses of this mineral are always more magnesian than the coexisting garnet (Table 7). The presence of dendritic magnetite in spinel and the granular exsolution of ilmenite indicate that the pristine spinel was richer in Fe and Ti. The composition of the pristine spinel is not only crucial to understanding the peak thermal history of the host rock, but also has an important bearing on the topology of the system FeO-MgO-Al₂O₃-SiO₂ (FMAS) [see Waters, (1991)]. Estimation of the pristine spinel composition in the studied rock is fraught with difficulties, such as: (1) the wide modal variation of ilmenite in the oxide aggregates; (2) the presence of variable contents of magnetite and Ti-haematite lamellae in spinel and ilmenite, respectively; (3) the absence of a continuous coronal phase, which makes it difficult to estimate the proportion of the different oxides in an aggregate; (4) the involvement of corundum lamellae in the intergrowth. The last problem has been circumvented by choosing aggregates devoid of corundum lamellae for integrated analyses. In view of point (3), we have chosen aggregates that contain a smaller proportion of ilmenite (20-25 vol. %) than the others. Bulk compositions of such aggregates are likely to provide a conservative estimate of Fe₂TiO₄ content in the pristine spinel. For reconstruction of the pristine spinel composition, the procedure mentioned by Waters, (1991) was followed. The representative reintegrated compositions are presented in Table 3. The reintegrated compositions show significant ulvöspinel component (up to 17 mol %) and X_Mg is always lower than in coexisting garnet (1) (Table 3). This corroborates an earlier observation that in Fe-rich rocks, such as the present ones, spinel is always less magnesian than the coexisting garnet (Sengupta et al., 1991; Waters, 1991; Nichols et al., 1992; Fitzsimons, 1996). The implication of this will be discussed later.

Garnet (1) in both the associations is almandine rich with significant proportions of pyrope (32-38 mol %, Table 4). Grossular and spessartine contents are always low, and together never exceed 4 mol % (Table 4). A slight increase in almandine content towards the rims (3-4 mol %) is noted at contacts with spinel. Coronal garnet has a similar composition to that of the porphyroblastic garnet rims (Table 4).

Table 4. Representative garnet compositions

Reintegrated compositions of porphyroblastic ilmenite show significant Fe₂O₃ contents (>12 mol %, Table 5). However, granular ilmenite intergrown with coronal garnet shows lower proportions of Fe₂O₃ than spot analyses of porphyroblastic ilmenite (Table 5). The lamellar intergrowth between ilmenite and corundum suggests solubility of significant Al₂O₃ (>10 mol % calculated from image analysis) in the ilmenite structure.

Table 5. Representative compositions of magnetite (mt), ilmenite (ilm), biotite (bt) and sillimanite (sill)

Biotite included in garnet is phlogopite rich with significant TiO₂ (5·49 wt %) and F (1·38 wt %) (Table 5). Retrograde biotite has higher F (3·32 wt %) but lower TiO₂ (1·96 wt %) (Table 5).

The result of the reintegrated alkali feldspar and the antiperthite analyses by EPMA area scans are presented in Table 6. The reintegrated perthite compositions show up to 11 mol % anorthite (Table 6). The anorthite content of the plagioclase lamellae varies between 20 and 25 mol %. Porphyroblastic plagioclase shows no compositional zoning and varies in composition within a restricted range (An_35-30; Table 6).

Table 6. Representative feldspar compositions

Corundum and magnetite are nearly pure phases, but sillimanite contains up to 1·5 wt % of Fe₂O₃ (Table 5).

Reaction history

In association I, inclusions of biotite, sillimanite and quartz in garnet (1) suggest operation of the dehydration-melting reaction

biotite + sillimanite + quartz -> garnet + liquid.

(a)

Embayment of garnet (1) by lobate spinel (1) is intriguing. In domains free from complex spinel-ilmenite intergrowths, the likely reaction is

garnet + sillimanite -> spinel + quartz

(b1)

(Fig. 3a). This reaction proceeds to the right with rising temperature. However, the formation of spinel + ilmenite aggregates in places merits discussion. The textural and compositional data (discussed above) suggest that these aggregates are the breakdown product of a former Ti-rich spinel. As neither of the reactants in reaction (b1) can contribute Fe₂TiO₄ component, a Ti-saturating phase (such as biotite) must also be involved in these domains. It is likely that subsequent to the reaction (a) these domains were impoverished in silica and the residual biotite became more Ti rich, and ultimately reacted to form the complex Fe-Al-Ti spinel at higher temperature through the reaction

Ti-biotite + sillimanite -> complex Fe-Al-Ti spinel + liquid.

(b2)

Formation of coronal garnet and/or compound corona of sillimanite and garnet over the lobate spinel (Fig. 3a) suggest that reaction (b1) proceeded in the reverse direction during cooling. The free silica required for this reaction was possibly derived from the crystallization of the granitic liquid.

Intergrowth of garnet (2) and Fe₂O₃-poor granular ilmenite around Fe₂O₃-rich porphyroblastic ilmenite at the contact of plagioclase can be explained by the de-oxidation reaction of the type

Fe2O3-rich ilmenite + plagioclase + quartz -> Fe2O3-poor ilmenite + garnet + O2

(c)

(Fig. 3b). This reaction proceeds to the right during cooling and is reported from many high-grade metapelites [see Ellis & Green, (1985) and Dasgupta et al., (1995), and references therein].

In association II, stability of corundum (1) and garnet (1) and inclusion of biotite and sillimanite in the latter suggest the reaction

Ti-biotite + sillimanite -> garnet (1) + corundum (1) + liquid.

(d)

However, coexistence of garnet (1), corundum (1) and spinel-ilmenite aggregates suggests the reaction of the type

biotite + sillimanite -> garnet (1) + Ti-rich spinel + corundum (1) + liquid

(e)

(Fig. 4b). The significant TiO₂ content in the included biotite suggests that the Ti component of the oxide aggregates was derived from biotite. This reaction was also inferred by Powers & Bohlen, (1985).

The Zn-rich spinel inclusions in the cores of garnet (1) are problematic. They can form either by dehydration of staurolite or as a result of early melting of biotite in the ZnKFMASH system via the reaction

biotite + sillimanite + quartz ± vapour -> Zn-rich spinel + liquid

(Montel et al., 1986). However, the textural information is insufficient to choose between the two alternatives. Nevertheless, the presence of this type of spinel and biotite inclusions in garnet (1) suggests the reaction

spinel + biotite + sillimanite -> garnet (1) x+ liquid.

(f)

This reaction occurs at lower temperature than the biotite melting reactions (d) and (e) at pressures higher than the KFMASH univariant reaction (Spr, Opx, Crn, Qtz) (see Fig. 6, below) and is noted in many high-grade terranes (e.g. Fitzsimons, 1996).

The coarse spinel-sillimanite intergrowth replacing garnet (1) (Fig. 4a) can be explained by the reaction

garnet (1) + corundum (1) -> spinel + sillimanite.

(g)

This reaction has been experimentally calibrated by Shulters & Bohlen, (1989) and it proceeds to the right with rising temperature.

Evolution of the oxide aggregates

It was discussed earlier that textural features and compositional characteristics of the oxide aggregates indicate early stabilization of a Ti-rich spinel. In corundum-free aggregates, this Ti-spinel produced ilmenite_ss (FeTiO₃-Fe₂O₃) and spinel_ss (FeAl₂O₄-MgAl₂O₄-Fe₃O₄) through oxidation of the Fe₂TiO₄ component in spinel,

Fe2TiO4 + O2 -> (FeTiO3 + Fe2O3)ss+ Fe3O4 (partitioned in aluminous spinel)

(h)

(see Haggerty, 1976).

Experimental and theoretical studies in the system Fe-Mg-Ti-Al-O (Sack & Ghiorso, 1991) indicate that spinel with >10 mol % Fe₂TiO₄ is stable only at temperatures of the order of 1000°C or more. This implies that reaction (h) occurred at very high temperature.

The ilmenite-corundum intergrowth in some of these aggregates requires an additional reaction of the type

FeAl2O4 + O2 Fe3O4 (partitioned in spinel)+ Al2O3 (in ilmenite)

(i)

(Fig. 4e).

The presence of abundant corundum lamellae in ilmenite indicates significant solubility of Al₂O₃ in ilmenite. Experimental data of Hauck, (1981) show that only at temperatures higher than 1100°C can the ilmenite structure incorporate a substantial amount (>10 mol %) of Al₂O₃. Separate rimming of ilmenite and spinel by coronal garnet (Fig. 4f) suggests that formation of the latter mineral occurred subsequent to the reactions (h) and (i).

The development of magnetite lamellae of different thickness in spinel and Ti-haematite lamellae in ilmenite (Fig. 4c and d) indicate progressive supersaturation in Fe₃O₄-rich spinel and Fe₂O₃-rich ilmenite solid solutions during cooling [see Haggerty, (1976)]. Available experimental data and theoretical analysis of the spinel_ss (FeAl₂O₄-MgAl₂O₄-Fe₃O₄) suggest that unmixing of spinel_ss (with up to 15 mol % magnetite) occurred at T < 800°C [see Nell et al., (1989)]. Formation of discrete corundum laths in spinel and the occurrence of magnetite granules along the contact (Fig. 4c) suggest that reaction (i) also occurred at lower temperatures subsequent to the separation of ilmenite from spinel (Clarke et al., 1989).

Petrogenetic grid

The metamorphic evolution of the studied metapelitic rocks can be portrayed in a petrogenetic grid in the system KFMASH with TiO₂, Fe₂O₃, ZnO and F as extra components. Considering the fact that the Fe-Ti-Al oxide aggregates are products of oxy-exsolution of Fe₂TiO₄-rich spinel, these extra components do not form any new phase. Hence, the effect of incorporation of these non-KFMASH components will only be to increase the variance of the KFMASH assemblages. Excepting the one reported by Srogi et al., (1993), all other published petrogenetic grids for the KFMASH system or its subsystem FMAS consider either corundum- or quartz-bearing assemblages and hence are not suitable for the present study (see Grant & Frost, 1990; Hensen & Harley, 1990; Dasgupta et al., 1995). The petrogenetic grid of Srogi et al., (1993), on the other hand, does not include high-temperature phase relations. For this reason, a new petrogenetic grid has been constructed combining the available experimental data and the constraints imposed by the natural assemblages (Fig. 5).

We begin our analysis in the simplified KFMASH system and the effects of the non-KFMASH components will be discussed subsequently. Textural relations suggest that the early melting reactions with biotite evolved in the KFMASH system. Following the elimination of biotite and extraction of the melt, the subsequent mineral reactions evolved in the simplified FMAS sub-system. For quartz-bearing assemblages, the geometrical relations in the KFMASH and FMAS systems and their interrelationships have been elegantly presented by Hensen & Harley, (1990). The topology presented in Fig. 5, although broadly similar to that of Hensen & Harley, (1990), differs in the relative position of garnet and spinel in some of the reactions. Hensen & Harley, (1990) considered the FMAS topology of Hensen, (1986), which is valid for X_Mg(Spinel) > X_Mg(Garnet). However, in Fe-rich bulk compositions, such as the present one, the relationship is reversed (Grant & Frost, 1990; Clarke & Powell, 1991; Sengupta et al., 1991; Waters, 1991; Fitzsimons, 1996; references cited therein). As a consequence, garnet and spinel change sides in some univariant reactions to maintain reaction stoichiometry. Waters, (1991) showed that in the simple FMAS system (i.e. the biotite-free portion of Fig. 5) spinel + cordierite + quartz will be stabilized only in very iron-rich bulk compositions and cordierite with X_Mg up to 0·4 can coexist with spinel. However, in most of the natural spinel + cordierite-bearing assemblages, the X_Mg of cordierite exceeds 0·6 (Waters, 1991). Waters explained this anomaly by the up-pressure shift of the reaction spinel + cordierite + quartz garnet + sillimanite because of incorporation of non-KFMASH components such as Fe₂O₃, TiO₂, and ZnO in spinel. Geometric relations further indicate that a topological inversion would occur when such a shift of spinel-bearing equilibria causes the invariant point [Spl, Bt, Crn] to lie within the stability field of spinel (Hensen & Harley, 1990; D. J. Waters, personal communication, 1998). This will generate the new invariant points [Opx, Crn, Bt], [Spr, Crn, Bt], [Grt, Crn, Bt] and [Crd, Crn, Bt]. However, the [Spl, Spr, Crn] invariant point of Hensen & Harley, (1990) will remain essentially unaltered. This is shown in Fig. 5, where the last two invariant points are omitted for the sake of clarity. Experimental investigations of Annersten & Seifert, (1981) showed that the assemblage spinel-orthopyroxene-sillimanite-quartz, which is the key assemblage for the inverted topology, is stabilized as a result of significant incorporation of Fe₃O₄ at high f O₂. This, together with the observation that this four-phase assemblage mostly occurs in high-f O₂ granulites, led Hensen, (1986) to conclude that f O₂ is the only controlling factor for the topological inversion. Subsequent work of Sengupta et al., (1991), Waters, (1991), Nichols et al., (1992) and Fitzsimons, (1996), although supporting the contention of Hensen, (1986), also emphasizes the role of other non-FMA components such as ZnO and TiO₂ in stabilizing the spinel-bearing assemblages and consequent topological inversion. Using the internally consistent data set of Holland & Powell, (1990), D. J. Waters (personal communication, 1998) showed that spinel should contain nearly 20 mol % non-KFMASH components to cause the topological inversion and that the absolute P-T positions of [Spr, Crn, Bt] and [Crd, Crn, Bt] are determined by the amount of non-FMA components in spinel. With increasing impurities, the sapphirine- and cordierite-absent invariant points would move towards [Spl, Spr, Crn], thereby causing contraction of the triangular area bounded by these invariant points (Fig. 5). Incorporation of F and Ti in biotite would further shrink this triangle, but towards higher temperatures (Fig. 5). We have chosen the inverted topology because of the presence of significant non-KFMASH components in early spinel and the high f O₂ conditions (indicated by the presence of ferrian ilmenite). To explain the corundum-bearing reactions, another invariant point [Spr, Qtz, Opx] is created (Fig. 6). The geometry around this invariant point is similar to that of Srogi et al., (1993), except for some spinel-garnet equilibria for reasons stated above. The topological constraints of Fig. 6 indicate that this invariant point will lie in the triangular area defined by the three corundum-absent invariant points and should be bounded by the (Opx) and (Sill) absent reactions emanating from the invariant point [Spr, Spl, Crn]. Also shown in Fig. 6 are some of the bivariant reactions relevant to this study.

Figure 5. Schematic petrogenetic grid in the system KFMASH. Filled circles and thin lines are the invariant points and related univariant reactions similar to the `low f O<sub>2</sub>' grid of Hensen, (1986). [Note that Spl and Grt change their side in some of the reactions (see text).] Filled squares and thick lines are the invariant points and the related univariant reactions in the inverted topology [similar to `high f O₂ ' grid of Hensen, (1986)]. a-b-c , 1-2-3 and solid arrowheads indicate the shifts of some `low f O₂' reactions [e.g. (Sil, Crn, Spr, Bt) and (Opx, Crn, Spr, Bt)] as a result of presence of non-FMAS components in spinel (see text). Open arrowhead indicates the displacement of the reactions for Ti and F in biotite (see text). Some uni- and invariant reactions are omitted for clarity. Mineral abbreviations are after Kretz, (1983).

Figure 6. Schematic partial petrogenetic grid in the system KFMASH involving Qtz and Crn. The thick and thin lines are the univariant and some bivariant reactions, respectively. The stability fields of the assemblages Grt + Spl + Crn + Sil and Grt + Spl + Crd + Sil are shaded. The P-T coordinates of the Crn absent invariant points are taken from Dasgupta et al., (1995), and the same coordinate for [Spr, Opx, Qtz] has been deduced in this study (see text). The pressure-temperature trajectory constrained from the textural relations is shown by thick and hollow line with arrowhead. Univariant reactions: 1, Bt + Sill Grt + Cd + Crn + L; 2, Bt + Sill + Q Opx + Cd + L; 3, Opx + Sill + Q Gt + Cd; 4, Sp + Cd + Q Opx + Sill; 5, Gt + Cd Sp + Opx + Q; 6, Gt + Sill Cd + Sp + Q; 7, Gt + Sill Sp + Cd + Crn; 8, Bt + Gt + Sill Sp + Cd + L; 9, Bt + Sill + Q Gt + Cd + L; 10, Bt + Sp + Q Gt + Cd + L; 11, Bt + Sill + Q Sp + Cd + L.

Figure 7. Photomicrographs illustrating reaction textures in mafic-ultramafic rocks. (a) Thin garnet corona (black) along the contacts of orthopyroxene (O), clinopyroxene (Cp) and plagioclase (P) in gabbronorite. Bar represents 200 µm. (b) Coarse spinel (Sp) showing granular exsolution of magnetite (M) and forming a granoblastic mosaic with orthopyroxene (O) and plagioclase (P). [Note the symplectitic intergrowth of orthopyroxene and the secondary spinel. Also note the thin garnet (G) corona on spinel.] Secondary amphibole (A) replacing orthopyroxene and plagioclase. Fine opaque phases in amphibole are ilmenite and magnetite. Bar represents 200 µm.

In the P-T region of biotite-dehydration melting, K-feldspar can be either a reactant or a product depending upon the H₂O/K₂O ratio in the melt and biotite (Carrington & Watt, 1995). If we consider the F contents of biotite inclusions to be representative of compositions at the time of melting, it is likely that the H₂O/K₂O ratio of biotite was higher than that of the coexisting melt. This places K-feldspar on the reactant side of the dehydration melting reactions. However, the topology presented in Fig. 6 would change only marginally with a change of side for K-feldspar in the melting reactions (Fitzsimons, 1996).

The thermobaric evolution of the study assemblages can be better explained if the P-T coordinates of the invariant points shown in Fig. 6 are available. The experimental data of Carrington & Harley, (1995) for the system KFMASH tightly constrains the P-T coordinates of [Spr, Spl, Crn, Bt] at 9 ± 0·5 kbar and 900°C, which is also corroborated by the thermodynamic analysis of Holland et al., (1996) (their [Os] invariant point). The presence of Ti and F in biotite would shift the invariant point towards higher temperature (Fig. 5). Considering the TiO₂ and F contents of biotite inclusions (Table 5), a minimum of 30°C up-temperature shift of this invariant point is expected from the experimental data of Hensen & Osanai, (1994). On the basis of P-T estimates of natural assemblages, Dasgupta et al., (1995) showed that the invariant points [Opx, Spr] (equivalent to [Opx, Spr, Crn] in Fig. 6) and [Spr, Bt] (equivalent to [Spr, Bt, Crn] in Fig. 6) occur at 5 ± 1 kbar, 750-850°C and ~9 kbar and 950°C, respectively. Using the thermodynamic data of Holland et al., (1996), D. J. Waters (personal communication, 1998) calculated the P-T coordinates of these two invariant points with 20 mol % of non-aluminous components in spinel. Although the calculated P-T coordinates of [Spr, Crn, Bt] match perfectly with those of Dasgupta et al., (1995), the invariant point [Opx, Spr, Crn] is placed at nearly the same pressure but at a temperature ~100°C lower (Fig. 5). However, if the shift of the biotite-melting reaction (Spr, Opx, Crn, Spl) for these non-aluminous components is calculated using the experimentally determined ones of Carrington & Harley, (1995), the resultant P-T coordinates of [Spr, Opx, Crn] match perfectly with those of Dasgupta et al., (1995). Holland et al., (1996) also argued that the reactions involving biotite calculated from their data set are always placed at temperatures 80-100°C lower than the experimentally determined positions of Carrington & Harley, (1995), and attributed this discrepancy to the poorly constrained melt parameters used in their data set. In view of this, we have adopted the P-T coordinates of these invariant points from Dasgupta et al., (1995).

The P_{[Spr,Opx,Qtz]} can be estimated using the topological constraints P_{[Opx,Spr,Crn]} < P_{[Spr,Opx,Qtz]} < P_{[Spl,Spr,Crn]}. The P coordinates of the bounding invariant points give a large pressure bracket of 5-9 kbar. However, for spinel containing >20 mol % non-aluminous components and biotite with appreciable Ti and F, the pressure bracket will be narrower. Powers & Bohlen, (1985) and Srogi et al., (1993) described an assemblage that equilibrated close to P_{[Opx,Spr,Crn]}. Geobarometry in these areas indicates P ~ 7 ± 1 kbar. Recently, Dasgupta et al., (1997) have reported the cordierite-corundum-spinel-sillimanite assemblage, which lies below the biotite-absent univariant curve emanating from this invariant point (Fig. 6). Those workers estimated a pressure of 7 kbar using different barometric estimates. Considering the nearly flat slope of the biotite-absent reaction and barometric estimates for the above areas, the P_{[Spr,Opx,Crn]} is placed at 7 ± 1 kbar.

Several important points emerge from the topology presented in Fig. 6:

(1) dehydration melting of biotite + sillimanite + quartz can produce spinel + cordierite or garnet + cordierite depending on the depth of melting, i.e. at pressures lower and higher than P_{[Spr,Opx,Crn]}, respectively. This is consistent with the widespread occurrence of spinel + cordierite in low-pressure contact aureoles and of garnet + cordierite-bearing assemblages in medium- and high-pressure (>5 kbar) regionally metamorphosed pelites (Grant & Frost, 1990; Harley et al., 1990; Dasgupta et al., 1995; Fitzsimons, 1996; references therein).

(2) In quartz-undersaturated sillimanite-rich metapelite (with Ti- and F-rich biotite), complete elimination of biotite would indicate very high temperatures, i.e. above 900°C.

(3) In silica-undersaturated iron-rich metapelites, corundum-bearing assemblages are potentially valuable depth indicators. The assemblage garnet + corundum + spinel + sillimanite occurs exclusively at pressures >P_{[Spr,Opx,Qtz]} and is only expected in deep crustal rocks, whereas the cordierite + corundum + sillimanite + spinel-bearing assemblages occur at lower pressures. This is consistent with the preponderance of the latter assemblage in upper- and mid-crustal contact aureoles [reviewed by Barton et al., (1991) and Dasgupta et al., (1997)].

The mineral assemblages and the inferred mineral reactions in the studied rocks can now be interpreted in terms of this petrogenetic grid. Textural relations presented above suggest that biotite in both the associations was eliminated through a number of dehydration-melting equilibria within the temperature interval of T_{[Spr,Opx,Crn]} and T_{[Spr,Opx,Qtz]} (Fig. 6). The appreciable amount of Fe₂TiO₄ (>19 mol %, Table 3) in the pristine spinel indicates that the peak metamorphic assemblages were stabilized beyond T_[Spr,Crn,Bt] (i.e. >950°C). Non-appearance of other characteristic high-temperature assemblages such as sapphirine + quartz, osumilite + garnet ± sapphirine ± orthopyroxene can be attributed to the Fe-rich bulk composition of the rocks (Hensen & Harley, 1990). The occurrence of corundum + garnet + sillimanite + spinel and absence of cordierite in the quartz-undersaturated association II indicate that the prograde reactions occurred at high pressures, i.e. P > P_{[Spr,Opx,Qtz]}. Textural features further suggest that all the biotite-absent reactions were intersected twice from opposite directions, i.e. during the prograde and retrograde stages of metamorphism. In the absence of orthopyroxene and sapphirine no upper pressure limit can be constrained from the petrogenetic grid. However, the geobarometric estimates, discussed below, indicate that cooling occurred at pressures above P_[Spr,Bt,Crn]. Taken together, all this evidence indicates an anticlockwise heating-cooling path at pressures above P_{[Spr,Opx,Qtz]} (Fig. 6).

Geothermobarometry

The absence of orthopyroxene and cordierite excludes application of most of the cation-exchange thermometers in the study rocks. However, oxide and feldspar thermometers can be applied to constrain the peak temperature of metamorphism. The reintegrated spinel compositions (with >19 mol % Fe₂TiO₄) requires temperatures in excess of 1000°C (Sack & Ghiorso, 1991). This is a conservative estimate as the aggregates with minimum modal ilmenite were considered. Temperatures in excess of 1000°C are also required to dissolve significant amounts of Al₂O₃ (10 mol % or more) in the ilmenite structure (Hauck, 1981). Bulk compositions of feldspars when projected on the ternary feldspar isotherms at 10 kbar of Hayob et al., (1989) indicate temperatures in the range of 950-1000°C. These too are conservative estimates in view of the likely partitioning of some Ca from alkali feldspar to porphyroblastic plagioclase during cooling [see Kroll et al., (1992)]. Such an extreme temperature is consistent with the predicted temperature (>900°C) for the instability of the assemblage Ti-rich biotite + sillimanite ± quartz from the petrogenetic grid (Fig. 6). Host-lamellae pairs in alkali feldspars yield distinctly lower values (650-700°C), which are obviously cooling temperatures.

Peak metamorphic pressure, on the other hand, can be estimated using the pressure-sensitive equilibria involving garnet-plagioclase-sillimanite-quartz (GPSQ, Koziol & Newton, 1988), garnet-corundum-spinel-sillimanite (GCSpS, Shulters & Bohlen, 1989), and garnet-sillimanite-spinel-quartz (GSSpQ, Bohlen et al., 1986). The spinel-bearing equilibria are sensitive to the input spinel composition and temperature [see Shulters & Bohlen, (1989)]. To obtain a conservative pressure estimate, we have used spot analyses of spinel and 950°C (the minimum estimates of the peak temperature). Application of the GPSQ equilibrium to the studied rocks is fraught with uncertainties inherent in the low grossular content of garnet (Table 4). Notwithstanding the uncertainties associated with these barometers, the three equilibria yield metamorphic pressures mostly in the range of 9-10 kbar (Table 7). These pressure values are consistent with the constraints imposed by the petrogenetic grid (i.e. stability of garnet + corundum and absence of cordierite in association II). The estimates match well with the results from the more precise GOPS and GCPS barometry in the enclosing mafic rocks (Table 7).

Table 7. Results of thermobarometry

P-T HISTORY OF THE LAYERED BASIC INTRUSION

The mafic members of the complex are olivine- to quartz-normative low-K tholeiites with mg-number varying from 30 to 74. Magnesian varieties are olivine-normative (1-12% normative olivine), whereas the Fe-rich rocks are quartz-normative (4-6% normative quartz). The compositional range overlaps with the experimental run composition (No. R 698) of Irving, (1974) and the tholeiitic composition of Green & Ringwood, (1967). This allows direct application of their experimental phase diagrams to this rock suite. Representative compositions of the different phases in the members of the layered complex are presented in Tables 2 and 7. The mafic rocks consist of orthopyroxene, clinopyroxene, highly calcic plagioclase and ilmenite. Olivine is conspicuously absent and coronal garnet developed only in the Fe-rich rocks. Orthopyroxene and clinopyroxene often contain coarse to fine (100) lamellae of Ca-rich and Ca-poor pyroxene, respectively. Reintegrated pyroxene compositions yield temperatures of 1000-1100°C when plotted on the projected isotherm of Davidson & Lindsley, (1989). The calculated temperatures are also consistent with the solidus temperature of tholeiitic magma at 8-10 kbar (Green & Ringwood, 1967; Irving, 1974). Non-appearance of olivine even in olivine-normative rocks indicates that the depth of emplacement of the magma was higher than the stability field of olivine + plagioclase. The phase diagram of Irving, (1974) suggests a pressure of magmatic crystallization above 6 kbar at 1100°C. However, non-appearance of garnet constrains an upper pressure limit of 11 kbar at the same temperature (Green & Ringwood, 1967; Irving, 1974). In Fe-rich rocks, thin coronas of garnet or garnet-quartz symplectite locally occur along the contacts of pyroxene-plagioclase-ilmenite (Fig. 7a). This garnet is likely to have formed during subsolidus, near-isobaric cooling of the rocks [see Ellis & Green, (1985)].

Further evidence in support of such deep-level intrusion comes from the mineralogy of the layers and/or xenoliths of clinopyroxenite and hercynite-bearing orthopyroxenite. Clinopyroxenite is composed dominantly of cumulus clinopyroxene with extremely calcic plagioclase and few grains of orthopyroxene as intercumulus phases. This suggests crystallization from the enclosing mafic to anorthositic melt of clinopyroxene followed by orthopyroxene and plagioclase. Available experimental data on basaltic rocks of varied bulk compositions suggest that clinopyroxene can be the sole liquidus phase only at pressures >8 kbar (e.g. Fram & Longhi, 1992).

The hercynite-bearing orthopyroxenite is dominated by spinel and orthopyroxene with one or more of the phases plagioclase, clinopyroxene or olivine. Olivine and plagioclase never coexist and amphibole is the common retrograde phase that replaces the pyroxenes. Spinel occurs in two modes. Coarse spinel forms granoblastic mosaics with orthopyroxene. This spinel shows granular exsolution of magnetite (Fig. 7b). The other type of spinel forms myrmekitic intergrowths with orthopyroxene and, less frequently, with clinopyroxene (Fig. 7b). Thin rims of garnet are locally present around pyroxenes, hercynite and plagioclase. Symplectitic intergrowths of orthopyroxene and hercynite similar to this are common in many lower-crustal xenoliths and are interpreted to have formed by either of the two processes: (a) replacement of olivine + plagioclase or (b) direct crystallization of aluminous pyroxene and subsequent subsolidus exsolution in the spinel peridotite-gabbro stability field (e.g. Green & Ringwood, 1967; Ellis & Green, 1985; Ballhaus & Berry, 1991; Dasgupta et al., 1993). Mutual exclusion of olivine and plagioclase tends to support the second alternative. The mineralogical evolution can be presented schematically as follows: Stage 1: Al-Opx(1) -> Al-Opx(2) + Spinel; Stage 2: Al-Opx(2) -> Garnet. The estimated P-T conditions of the garnet-bearing mafic rocks cluster around 8-9 kbar and 670-720°C (Table 7). The temperature estimates are obviously cooling temperatures reflecting the blocking of Fe-Mg exchange between garnet and pyroxenes [see Ellis & Green, (1985)]. Distinctly lower pressures (4-5 kbar at 700°C) are obtained from the alumina solubility in orthopyroxene coexisting with garnet in the hercynite-bearing orthopyroxenite (Harley, 1984). Fitzsimons & Harley, (1994) showed that the formulation of Harley, (1984) is strongly sensitive to retrograde Fe-Mg resetting, which often leads to underestimation of the equilibrium pressure in the natural assemblages.

Summarizing, the mineralogical and phase compositional attributes constrain the emplacement and fractionation of the layered magmatic suite to have occurred above 8 kbar and at temperatures >= 1100°C. Subsequent near-isobaric cooling produced coronal garnet in the Fe-rich mafic rocks and exsolution of spinel from the aluminous pyroxene in the hercynite-bearing orthopyroxenite.

Comparison with the northern sector of the Eastern Ghats Belt and East Antarctica

Available petrological information from several parts in the northern sector of the Eastern Ghats Belt indicates at least two phases of granulite facies metamorphism [reviewed by Dasgupta & Sengupta, (1995)]. The earlier one is characterized by UHT metamorphism followed by near-isobaric cooling. Relics of prograde metamorphism, albeit documented from only few places, suggest that the UHT metamorphism occurred on an anticlockwise P-T trajectory (Sengupta et al., 1990; Dasgupta et al., 1995; Mukhopadhyay & Bhattacharya, 1997). On the other hand, Lal et al., (1987), interpreting the reaction textures of a suite of Mg-Al granulites from the Paderu area, construed a moderate to steeply decompressive retrograde path from the peak temperature. Although the rocks do not preserve any information on the prograde path, those workers preferred a `clockwise' path on the basis of the retrograde P-T trajectory. However, the reaction textures of Lal et al., (1987) have been reinterpreted by Harley, (1989) and Pal & Bose, (1997), who postulated a near-isobaric cooling from the peak temperature followed by a steeply decompressive P-T path. Interpreting the reaction textures from the granulites of the Chilka Lake area, Eastern Ghats Belt, Sen et al., (1995) showed a complex retrograde P-T trajectory with a steep decompression from the peak temperature followed by near-isobaric cooling. Although some decompression from the peak temperature is also possible along an anticlockwise P-T trajectory [see path 6 in fig. 3D of Harley, (1989)], in the absence of any relics of prograde metamorphism, the nature of the P-T evolution in the Chilka Lake area remains unclear.

Except for the depth of metamorphism, the metamorphic history construed in this work has striking resemblance to that of the Ongole area (Dasgupta et al., 1997) in the southern sector of the belt (Fig. 1). These P-T paths are very similar to the first granulite facies event recorded in most places in the northern sector. However, in the absence of definitive geochronological support, their temporal equivalence cannot be invoked at the present state of information. Subsequent to this event, the rocks in the northern sector witnessed another granulite facies metamorphism at distinctly lower peak metamorphic temperature (750-850°C at ~8 kbar), followed by a steeply decompressive retrograde P-T segment [650-700°C and 4-5 kbar; reviewed by Dasgupta & Sengupta, (1995) and schematically shown in Fig. 8]. In fact, this is the more dominant petrological overprint in the northern sector, and has thoroughly reworked the older near-isobarically cooled granulites [see Dasgupta & Sengupta, (1995)]. The available geochronological data indicate the last granulite facies metamorphism to be at ~1000 Ma in this sector. These data include the U-Pb cooling ages of zircon, allanite, monazite, perrierite and titanite (Grew & Manton, 1986; Aftalion et al., 1988; Paul et al., 1990; Mezger et al., 1996; Mezger & Cosca, 1999). Except for zircon, the blocking temperatures of all other dated minerals are lower than the peak metamorphic temperature of the second granulite facies event [see Mezger et al., (1996)]. The ~1000 Ma zircons were also collected from granitoid rocks that intruded the older granulite basement (Grew & Manton, 1986; Aftalion et al., 1988; Paul et al., 1990). The only exception to this is the 863 Ma discordant U-Pb age of a titanite from the Rajamundry area (Mezger & Cosca, 1999). This age is distinctly younger than the ~1000 Ma `Rayner Orogeny'. Mezger & Cosca, (1999) have argued that the titanite from the northern sector shows resetting to various degrees subsequent to the ~1000 Ma metamorphism, and have attributed this discordance to a late thermal overprint at ~550 Ma. The effect of this last event was not strong enough to completely homogenize the ~1000 Ma memory of the titanite. In view of these observations it is reasonable to conclude that the second granulite facies event in the northern sector is characterized by a steeply decompressive retrograde P-T path that occurred at ~1000 Ma. Therefore, this second granulite facies event in the northern sector of the Eastern Ghats Belt has remarkable similarity to the Rayner complex exposed along the McRobertson Land and Kemp coast of East Antarctica (see Fig. 8). In contrast to the northern sector, a decompressive metamorphic overprint is, so far, not recognized in the southern sector, although only a few areas have been investigated (Fig. 8). The available U-Pb cooling ages of minerals also indicate no significant high-grade metamorphic event subsequent to ~1600 Ma (in Kondapalle) and ~1300 Ma (in Ongole) except for late shear zones. These narrow shear zones record a mild Grenvillian thermal imprint that could reset the ⁴⁰Ar-³⁹Ar radioclock (1100 Ma) of amphibole but not the U-Pb systematics of allanite and monazite. Geochemical data are meagre in the southern sector. However, on the basis of the available petrological features together with the absence of a Grenvillian granulite facies event in the two areas of the southern sector, it is tempting to compare the metamorphic history of this segment of the Eastern Ghats Belt with the Napier complex of East Antarctica. This would provide an important petrological constraint to delineate the Napier-Rayner terrane boundary in the Eastern Ghats Belt provided that the former terrane is correlatable with the latter belt.

Figure 8. Metamorphic pressure-temperature trajectories of different domains in the Eastern Ghats Belt of India and Enderby Land, East Antarctica. The last granulite facies event in the different domains of both continents is in bold. The quoted ages are related to these last events (see text). The similarity in the metamorphic P-T histories of the northern and southern segments of the Eastern Ghats Belt to those of the Napier and Rayner complexes, respectively, should be noted. Thin broken lines are the proposed extension of the Napier-Rayner boundary into the Eastern Ghats Belt: C, Chetty, (1995); G, Grew et al., (1988). Ara, Araku-Anatagiri (Sengupta et al., 1990); Gar, Garbhum (Dasgupta et al., 1992); Sal, Salur (Mukhopadhyay & Bhattacharya, 1997); Chil, Chilka Lake (Sen et al., 1995); Ana, Anakapalle (Dasgupta et al., 1993); Vis, Visakhapatnam (Kamineni & Rao, 1988); Raj, Rajamundri (Dasgupta et al., 1995); Chi, Chimakurthi (Ongole) (Dasgupta et al., 1997); Kon, Kondapalle (this work). All other abbreviations are as in Fig. 1.

The Napier-Rayner boundary in the Eastern Ghats-a metamorphic perspective

In East Antarctica, the ~1000 Ma Rayner complex (along the McRobertson Land and Kemp coast) with its characteristic decompressive P-T path and moderate P-T conditions (7-8 kbar, ~800°C) contrasts sharply with the Archaean Napier complex with its distinctive UHT metamorphism and retrograde near-isobaric cooling P-T path [reviewed by Harley & Hensen, (1990)]. Thus, the Napier-Rayner boundary (N-R) near the Edward VIII Gulf can be viewed as a boundary separating two blocks with contrasting thermal histories and retrograde P-T paths. This boundary plays an important role in the assembly of the East Gondwana supercontinent. Recognizing the importance of locating the continuation of the N-R boundary in the Eastern Ghats Belt, several attempts have been made in this direction (e.g. Grew et al., 1988; Chetty, 1995). The peak thermal conditions and the retrograde P-T loops are considered to be important characteristics in identifying the Napier and Rayner terranes, and hence, delineating their boundary in East Antarctica (Harley & Hensen, 1990). None of these studies, however, considered using the metamorphic impress to identify this boundary in the Eastern Ghats Belt. Grew & Manton, (1986) first documented an ~1000 Ma age for the metamorphic and magmatic events in the Anakapalle area, Eastern Ghats Belt (Fig. 8). Combining these age data with the ~1900 Ma allanite age of Vinogradov et al., (1964) from an unmetamorphosed apatite-magnetite vein from the adjoining Kashipatnam area, Grew et al., (1988) construed the N-R boundary to pass south of Anakapalle (Fig. 8). Recently, Kovach et al., (1997) and Mezger & Cosca, (1999) redetermined the age of the same vein (dating the zircon) and obtained a Pan-African age (~500 Ma). Chetty, (1995) delineated this boundary to the north of Anakapalle on the basis of intercontinental lineament correlations (Fig. 8). However, there is no lithological break across his inferred boundary, and this, together with the Grenvillian age of metamorphism of granulites at Anakapalle (Grew & Manton, 1986), argues against such a correlation (see Fig. 8). Recently, Dasgupta et al., (1995) documented polyphase granulite facies metamorphism in which the last event shows a steep decompression from 8 kbar, 800°C to 5 kbar, 650°C from the Rajamundry area lying north of but close to the Godavari rift (Fig. 8).

Available geochronological data (discussed above) indicate the last high-grade metamorphism to have occurred at ~1000 Ma. Thus, the petrological signatures and the geochronological data indicate that the effect of `Rayner orogeny' in the Eastern Ghats Belt can be traced south of Anakapalle at least as far as the Godavari Rift (Rajamundry area) (Fig. 8). This asserts that the continuation of the N-R terrane boundary into the Eastern Ghats Belt should pass between Rajamundry and Kondapalle. It is interesting to note that the Godavari rift passes through this zone and, in the reassembled Indo-Antarctic block, this rift can be linked with the concealed East Enderby rift zone that lies close to the N-R boundary in Enderby Land (Fig. 8). Therefore, the idea of considering the Godavari rift as the N-R boundary in the Eastern Ghats Belt is tempting. Implicit in this assumption is the fact that the Napier complex of East Antarctica and the segment of the south Indian shield lying south of the Godavari rift have shared a similar geological history since Archaean time. In the Napier complex, detailed geochronological evidence indicates that the UHT metamorphism occurred within a narrow time span between 3000 and 2900 Ma [reviewed by Harley & Hensen, (1990)]. Notwithstanding the similarity of the UHT metamorphism and the retrograde P-T trajectory between the Napier complex and the southern segment of the Eastern Ghats Belt, the timing of such high-grade metamorphism is unknown in the latter segment. This opens up two possibilities: (1) the UHT metamorphic events in these crustal domains are coeval. This means that the Napier complex and the southern segment of the Eastern Ghats are fragments of the same Archaean block. (2) The UHT metamorphic events in the two domains are temporally unrelated. This means that these domains represent two different crustal segments that were welded together during the ~1000 Ma tectono-thermal event. Recently, Bhaskar Rao et al., (1996) have interpreted Sm-Nd whole-rock data on layered complexes at Sittampundi and Bhavani Sagar to document a ~2900 Ma high-grade metamorphic event within the Bhavani-Cauvery shear zone, southern India. In the assembled Indo-Antarctic landmass, this important crustal-scale shear belt juxtaposes against the Napier complex [see Fig. 1 and Bhaskar Rao et al., (1996)]. This piece of evidence, although not conclusive, hints at the first option. On the other hand, the western part of the northern sector exposes a zone dominated by basic to enderbitic granulites [the western charnockite zone of Nanda & Pati, (1989)] that shows similarities withthe igneous association of the southern sector (e.g. atKondapalle and Ongole). Similarly, a suite of supracrustal granulites occurs in the easternmost parts of the southern sector. In the absence of geochronological data, however, the correlation of the magmatic and the supracrustal rock units across the Godavari rift remains unclear. It also needs to be ascertained through detailed geochronological work whether these supracrustal granulites of the southern sector have been involved in the ~1000 Ma tectono-thermal event. If so, then the Napier complex would represent an isolated nucleus separated by the ~1000 Ma mobile belts. A comprehensive petrological study and, in particular, modern isotope dating of the two segments of the Eastern Ghats Belt is, therefore, needed to allow their crustal evolution to be deciphered and to allow more definite terrane correlations in East Gondwana.