Large garnet poikiloblasts hosted by leucosome in metapelitic gneiss from Broken Hill reflect complex mineral-melt relationships. The spatial relationship between the leucosomes and the garnet poikiloblasts implies that the growth of garnet was strongly linked to the production of melt. The apparent difficulty of garnet to nucleate a large number of grains during the prograde breakdown of coexisting biotite and sillimanite led to the spatial focussing of melting reactions around the few garnet nuclei that formed. Continued reaction of biotite and sillimanite required diffusion of elements from where minerals were reacting to sites of garnet growth. This diffusion was driven by chemical potential gradients between garnet-bearing and garnet-absent parts of the rock. As a consequence, melt and peritectic K-feldspar also preferentially formed around the garnet. The diffusion of elements led to the chemical partitioning of the rock within an overall context in which equilibrium may have been approached. Thus, the garnetbearing leucosomes record in situ melt formation around garnet porphyroblasts rather than centimetrescale physical melt migration and segregation. The near complete preservation of the high-grade assemblages in the mesosome and leucosome is consistent with substantial melt loss. Interconnected networks between garnet-rich leucosomes provide the most likely pathway for melt migration. Decimetre-scale, coarse-grained, garnet-poor leucosomes may represent areas of melt flux through a large-scale melt transfer network.

The loss of a metamorphic fluid via the partitioning of H 2 O into silicate melt at higher metamorphic grade implies that, in the absence of open system behaviour of melt, the amount of H 2 O contained within rocks remains constant at temperatures above the solidus. Thus, granulite facies rocks, composed of predominantly anhydrous minerals and a hydrous silicate melt should undergo considerable retrogression to hydrous upper amphibolite facies assemblages on cooling as the melt crystallizes and releases its H 2 O. The common occurrence of weakly retrogressed granulite facies assemblages is consistent with substantial melt loss from the majority of granulite facies rocks. Phase diagram modelling of the effects of melt loss in hypothetical aluminous and subaluminous metapelitic compositions shows that the amount of melt that has to be removed from a rock to preserve a granulite facies assemblage varies markedly with rock composition, the number of partial melt loss events and the P-T conditions at which melt loss occurs. In an aluminous metapelite, the removal of nearly all of the melt at temperatures above the breakdown of biotite is required for the preservation of the peak mineral assemblage. In contrast, the proportion of melt loss required to preserve peak assemblages in a subaluminous metapelite is close to half that required for the aluminous metapelite. Thus, if a given proportion of melt is removed from a sequence of metapelitic granulites of varying composition, the degree of preservation of the peak metamorphic assemblage may vary widely.

Eclogites from the south Tianshan, NW China are grouped into two types: glaucophane and hornblende eclogites, composed, respectively, of garnet + omphacite + glaucophane + paragonite + epidote + quartz and garnet + omphacite + hornblende (sensu lato) + paragonite + epidote + quartz, plus accessory rutile and ilmenite. These eclogites are diverse both in mineral composition and texture not only between the two types but also among the different selected samples within the glaucophane eclogite. Using thermocalc 3.1 and recent models of activity–composition relation for minerals, a P–T projection and a series of P–T pseudosections for specific samples of eclogite have been calculated in the system Na2O–CaO–FeO–MgO–Al2O3–SiO2–H2O (NCFMASH) with quartz and water taken to be in excess. On the basis of these phase diagrams, the phase relations and P–T conditions are well delineated. The three selected samples of glaucophane eclogite AK05, AK11 and AK17 are estimated to have peak P–T conditions, respectively, of 540–550 °C at c. 16 kbar, c. 560 °C at 15–17 kbar and c. 580 °C at 15–19 kbar, and two samples of hornblende eclogite AK10 and AK30 of 610–630 °C and 17–18 kbar. Together with H2O‐content contours in the related P–T pseudosections and textural relations, both types of eclogite are inferred to show clockwise P–T paths, with the hornblende eclogite being transformed from the glaucophane eclogite assemblage dominantly through increasing temperature.

Granulite facies metapelites of the Mather and Filla Paragneisses within the Rauer Group, east Antarctica, possess markedly different compositions. The metamorphic evolution of the two metapelite types has been interpreted as temporally distinct, with the Rauer Group preserving at least two distinct granulite facies tectonothermal episodes. Calculated P–T pseudosections and orthopyroxene Al content indicate the revised maximum‐preserved P–T conditions within the Mather Paragneiss to lie in the vicinity of 950–975 °C and 10–10.6 kbar, less extreme than previous estimates. The range of possible P–T paths for the Mather Paragneiss consistent with mineral textural relationships and pseudosections contoured for mineral proportion are significantly shallower (dP/dT) than previous estimates. A near‐isothermal decompression P–T path, and extreme peak metamorphic conditions, are not necessary to explain the development of preserved mineral reaction textures. The Filla Paragneiss contains pelitic assemblages less amenable to rigorous quantitative analysis. Nevertheless, possibilities for the shared or otherwise metamorphic evolution of the Mather and Filla Paragneisses may be postulated on the basis of calculated pseudosections in the context of existing geochronology for the Rauer Group and preserved microstructures. A shared evolution, most likely during Pan‐African granulite facies metamorphism, is plausible and consistent with mineral assemblage development, geochronology and microstructures. A revised interpretation of the Rauer Group's preserved metamorphic evolution may warrant the revision of existing tectonic models, applicable also to the remainder of Prydz Bay. More generally, the employed approach may incite a revision of peak P–T and P–T paths in other granulite facies terranes.

The Koralpe of the Eastern European Alps experienced high-temperature/low-pressure metamorphism (650°C and 6.5 kbar) during the Permian and eclogite facies metamorphism (700°C and 14 kbar) during the Eo-Alpine (Cretaceous) metamorphic event. In the metapelitic Plattengneiss shear zone that constitutes much of the Koralpe, the second metamorphism caused only partial re-equilibration of the assemblages formed during the first metamorphism. It is shown here that the Eo-Alpine mineral assemblage, garnet + biotite + muscovite + plagioclase + quartz (with or without kyanite), formed under low water activity conditions that are consistent with the level of dehydration that occurred during the Permian event. This implies that the rocks were essentially closed-system from the peak of the Permian metamorphism through the Eo-Alpine event. The evolution of water content of the rocks is traced through time: that prograde dewatering during the Permian metamorphic event terminated at the metamorphic peak with a water content around 3-4 mol.%. The water content remained then constant and led to water-undersaturation during the subsequent Eo-Alpine metamorphism. From the water content and activity evolution a post-peak isothermal decompression path close to the solidus is inferred for the Eo-Alpine event.