Granites and related volcanic rocks of the Lachlan Fold Belt can be grouped into suites using chemical and petrographic data. The distinctive characteristics of suites reflect source-rock features. The first-order subdivision within the suites is between those derived from igneous and from sedimentary source rocks, the I- and S-types. Differences between the two types of source rocks and their derived granites are due to the sedimentary source material having been previously weathered at the Earth's surface. Chemically, the S-type granites are lower in Na, Ca, Sr and Fe3+/Fe2+, and higher in Cr and Ni. As a consequence, the S-types are always peraluminous and contain Al-rich minerals. A little over 50% of the I-type granites are metaluminous and these more mafic rocks contain hornblende. In the absence of associated mafic rocks, the more felsic and slightly peraluminous I-type granites may be difficult to distinguish from felsic S-type granites. This overlap in composition is to be expected and results from the restricted chemical composition of the lowest temperature felsic melts. The compositions of more mafic I- and S-type granites diverge, as a result of the incorporation of more mafic components from the source, either as restite or a component of higher temperature melt. There is no overlap in composition between the most mafic I- and S-type granites, whose compositions are closest to those of their respective source rocks. Likewise, the enclaves present in the more mafic granites have compositions reflecting those of their host rocks, and probably in most cases, the source rocks.

S-type granites have higher δ18O values and more evolved Sr and Nd isotopic compositions, although the radiogenic isotope compositions overlap with I-types. Although the isotopic compositions lie close to a mixing curve, it is thought that the amount of mixing in the source rocks was restricted, and occurred prior to partial melting. I-type granites are thought to have been derived from deep crust formed by underplating and thus are infracrustal, in contrast to the supracrustal S-type source rocks.

Crystallisation of feldspars from felsic granite melts leads to distinctive changes in the trace element compositions of more evolved I- and S-type granites. Most notably, P increases in abundance with fractionation of crystals from the more strongly peraluminous S-type felsic melts, while it decreases in abundance in the analogous, but weakly peraluminous, I-type melts.

Blocks of medium-grained granodiorite to 4 m, and minor diabase, quartz diorite, granite, aplite and granophyre, are common in ejecta of the ∼6,900 yrBP calderaforming eruption of Mount Mazama. The blocks show degrees of melting from 0–50 vol%. Because very few have adhering juvenile magma, it is thought that the blocks are fragments of the Holocene magma chamber's walls. Primary crystallisation of granodiorite produced phenocrystic pl + hyp + aug + mt + il + ap + zc, followed by qz + hb + bt + alkali feldspar (af). Presence of fluid inclusions in all samples implies complete crystallisation before melting. Subsolidus exchange with meteoric hydrothermal fluids before melting is evident in δ18O values of −3·4+4·9‰ for quartz and plagioclase in partially melted granodiorites (fresh lavas from the region have δ18O values of +5·8−+7·0‰); δ18O values of unmelted granodiorites from preclimatic eruptive units suggest hydrothermal exchange began between ∼70 and 24 ka. Before eruption, the granitic rocks equilibrated at temperatures, estimated from Fe-Ti oxide compositions, of up to ∼1000°C for c. 102–104 years at a minimum pressure of 100-180 MPa. Heating caused progressive breakdown or dissolution of hb, af, bt, and qz, so that samples with the highest melt fractions have residual pl + qz and new or re-equilibrated af + hyp + aug + mt + il in high-silica rhyolitic glass (75-77% SiO2). Mineral compositions vary systematically with increasing temperature. Hornblende is absent in rocks with Fe-Ti oxide temperatures >870°C, and bt above 970°C. Oxygen isotope fractionation between qz, pl, and glass in partially fused granodiorite also is consistent with equilibration at T≥900°C (Δ18Oqz.pl = +0·7 ± 0·5‰). Element partitioning between glass and crystals reflects the large fraction of refractory pl, re-equilibration of af and isolation or incomplete dissolution of accessory phases. Ba and REE contents of analysed glass separates can be successfully modelled by observed degrees of partial melting of granodiorite, but Rb, Sr and Sc concentrations cannot. Several samples have veins of microlite-free glass 1–5 mm thick that are compositionally and physically continuous with intergranular melt and which apparently formed after the climactic eruption began. Whole-rock H2O content, microprobe glass analysis sums near 100% and evidence for high temperature suggest liquids in the hotter samples were nearly anhydrous. The occurrence of similar granodiorite blocks at all azimuths around the 8 × 10 km caldera implies derivation from one pluton. Compositional similarity between granodiorite and pre-Mazama rhyodacites suggests that the pluton may have crystallised as recently as 0·4 Ma; compositional data preclude crystallisation from the Holocene chamber. The history of crystallisation, hydrothermal alteration, and remelting of the granitic rocks may be characteristic of shallow igneous systems in which the balance between hydrothermal cooling and magmatic input changes repeatedly over intervals of 104-106 years.

Like many granites, the Late Cretaceous intrusives of the eastern Mojave Desert, California, have heretofore provided useful but poorly focused images of their source regions. New studies of lower crustal xenoliths and inherited accessory minerals are sharpening these images.

Xenoliths in Tertiary dykes in this region are the residues of an extensive partial melting event. Great diversity in their composition reflects initial heterogeneity (both igneous and sedimentary protoliths) and varying amounts of melt extraction (from <10% to >70%). Mineral assemblages and thermobarometry suggest that the melting event occurred at T ≥ 750°C at a depth of about 40 km. Present-day Sr, Nd, and Pb isotopic ratios indicate a Mojave Proterozoic heritage, but unrealistic model ages demonstrate the late Phanerozoic adjustment of parent/daughter ratios. A link between these xenoliths and the Late Cretaceous granites, though not fully documented, is probable; in any case, they provide invaluable clues concerning a crustal melting event, recording information about nature of source material (heterogeneous, supracrustal-rich), conditions of melting (moderately deep, moderately high T, accompanied by partial dehydration), and melt extraction (highly variable, locally extensive).

The Old Woman-Piute granites contain a large fraction of inherited zircon and monazite. A SHRIMP ion probe investigation shows that these zircons record a Proterozoic history similar to that which affected the Mojave region. Zonation patterns in zircons, and to a lesser extent monazites and xenotimes, document multiple phases of igneous, metamorphic, and sedimentary growth and degradation, commonly several in a single grain. Low Y in portions of the cores of inherited zircons and monazites and in monazites and outer portions of zircons from the xenoliths appear to indicate growth in equilibrium with abundant garnet.

Fluid dynamic modelling of crystallising calc-alkalic magma bodies has predicted that differentiated liquids will ascend as boundary layers and that accumulation of these buoyant liquids near chamber roofs will result in compositionally stratified magma chambers. This paper reports physical features in La Gloria Pluton that can be interpreted as trapped ascending differentiated liquids. Leucogranitic layers decimetres thick, which are locally stratified, are trapped beneath overhanging wall contacts. The same felsic magmas were also preserved where they were injected into the wall rocks as dykes and as large sill complexes. These rocks do not represent differentiated magmas produced by crystallisation along the exposed walls because the felsic layers occur at the wall rock contact, not inboard of it. Rather, we speculate that evolved felsic liquids are generated by crystallisation all across the deep levels of chambers and that initial melt segregation occurs by flowage of melt into tension fractures. Melt bodies so formed may be large enough to have significant ascent velocities as diapirs and/or dykes. The other way in which the leucogranite occurrence is at variance with the convective fractionation model is that the ascending liquids did not feed a highly differentiated cap to the chamber, as the composition at the roof, although the most felsic in this vertically and concentrically zoned pluton, is considerably more mafic than the trapped leucogranitic liquids. This suggests that these evolved liquids were usually mixed back into the main body of the chamber. Backmixing may be general in continental-margin calc-alkalic magmatic systems, which, in contrast to those in intracontinental settings, rarely produce volcanic rocks more silicic than rhyodacite. That the highly differentiated liquids are preserved at all at La Gloria is a result of the unusual stepped nature of the contact and the entirely passive mode of emplacement of the pluton, which, in contrast to ballooning in place, does not result in wall zones being “scoured”.

Cretaceous and Cainozoic granites and rhyolites in the northwestern U.S.A. provide a record of silicic magmatism related to diverse tectonic settings and large-scale variations in crustal structure. The Late Cretaceous Idaho Batholith is a tonalitic to granitic Cordilleran batholith that was produced during plate convergence. Rocks of the batholith tend to be sodic (Na2O > K2O), with fractionated HREE, negligible Eu anomalies, and high Sr contents, suggesting their generation from relatively mafic sources at a depth sufficient to stabilise garnet. In contrast, Neogene rhyolites of the Snake River Plain, which erupted in an extensional environment, are potassic (K2O > Na2O), with unfractionated HREE patterns, negative Eu anomalies, and low Sr contents, suggesting a shallower, more feldspathic source with abundant plagioclase. Eocene age volcanic and plutonic rocks have compositions transi- tional between those of the Cretaceous batholith and the Neogene rhyolites. These data are consistent with a progressively shallowing locus of silicic magma generation as the tectonic regime changed from convergence to extension.

Two models for the heating responsible for granite generation during convergent deformation may be distinguished on the basis of the length- and time-scales associated with the thermal perturbation, namely: (1) long-lived, lithospheric-scale heating as a conductive response to the deformation, and (2) transient, localised heating as a response to advective heat sources such as mantle-derived melts. The strong temperature dependence of lithospheric rheology implies that the heat advected within rising granites may affect the distribution and rates of deformation within the developing orogen in a way that reflects the thermal regime attendant on granite formation; this contention is supported by numerical models of lithospheric deformation based on the thin-sheet approximation. The model results are compared with geological and isotopic constraints on granite genesis in the southern Adelaide Fold Belt where intrusion spans a 25 Ma convergent deformation cycle, from about 516 to 490 Ma, resulting in crustal thickening to 50–55 km. High-T metamorphism in this belt is spatially restricted to an axis of magmatic activity where the intensity and complexity of deformation is significantly greater, and may have started earlier, than in adjacent low-grade areas. The implication is that granite generation and emplacement is a causative factor in localising deformation, and on the basis of the results of the mechanical models suggests that granite formation occurred in response to localised, transient crustal heating by mantle melts. This is consistent with the Nd- and Sr-isotopic composition of the granites which seems to reflect mixed sources with components derived both from the depleted contemporary mantle and the older crust.

The Cooma Complex of southeastern New South Wales comprises an andalusite-bearing S-type granodiorite surrounded by migmatites and low-pressure metamorphosed pelitic and psammitic sediments. The migmatite formed by the melting reaction:

Biotite + Andalusite + K-feldspar + Quartz + V = Cordierite + Liquid

at about 350–400 MPa , 670-730°C.

The melanosome consists of biotite + cordierite + andalusite + K-feldspar + plagioclase + quartz + ilmenite, whereas the leucosome consists of cordierite + K-feldspar + quartz with extremely rare biotite and plagioclase. In a closed system, freezing of the leucosome melt patches should have resulted in cordierite back-reaction with melt to produce biotite and andalusite. The virtually anhydrous mineralogy of the leucosome patches, lack of cordierite reaction and the absence of biotite selvedges at the leucosome-melanosome contacts, indicates that the melt did not completely solidify in situ. These observations can be explained by an initial peritectic melting reaction in the migmatite being arrested from back-reaction upon cooling because of the removal of hydrous melt, enabling leucosome cordierite to escape back-reaction. We propose that the melanosome is the residue of partial melting but that the leucosome patches do not represent frozen melt segregations but rather the liquidus minerals (cumulates) which precipitated from the melt.

In the restite-rich granodiorite from the core of the Cooma Complex, cordierite of similar composition to that in the migmatite has reaction rims of biotite and andalusite and there are coexisting biotite and andalusite in the matrix. The granodiorite consisted of about 50 wt% melt together with resite biotite, quartz and plagioclase, which can possibly be identified in the surrounding migmatite. Previous work suggested that the Cooma Granodiorite can be derived from a mixture of the surrounding metasediments which are of similar composition in the high and low-grade areas surrounding the granodiorite. Re-examinatibn of those data shows that the high-grade metasediments are more An-rich than the low-grade rocks. The Cooma Granodiorite is very similar to the high-grade rocks in terms of Or-Ab-An ratio. This suggests derivation of the Cooma Granodiorite from the high-grade rocks and not from the relatively An-poor low-grade rocks which are typical of exposed sediments in the Lachlan Fold Belt. It is most likely that the granodiorite and envelope of high-grade rocks have been emplaced into the compositionally different lower grade rocks from slightly greater depths.

It should be possible to infer the thermal state of the source terrane for granitic bodies, provided we have independent means to establish the chemical nature of this terrane. The chemical nature of the granitic rocks, including their degree of hydration, implies the solidus temperature. The concentration of the heat-producing radioactive elements in the granite (K, U, and Th) probably provides an upper estimate of their concentration in the source rock, which is an important thermal parameter. The depth and ambient temperature of the country rock into which the granite magma intruded provide useful boundary conditions for the thermal regime at the crustal level of anatexis. These constraints in turn form the bases for estimating the subcrustal thermal flux as well as the effective thermal interface for enhanced heat flow from below that resulted in anatexis. These inferences, in combination with other field-based parameters such as uplift rates and permissible time lapses for the geological events, permit realistic thermal modelling for the formation of granitic batholiths. The procedure is applied to the Late Cretaceous Pioneer and Boulder batholiths in southwestern Montana, U.S.A. The modelling results suggest that mantle upwelling, not subduction or thrust loading, caused anatexis. The isotopic chemistry of the granitic rocks rules out direct mixing of mantle magma, and field relations rule out crustal thinning as causes for partial melting.

Felsic I-type granites and associated volcanic rocks of Carboniferous age are extensively developed over an area of 15,000 km2 in northern Queensland. These granites have been subdivided into four supersuites: Almaden, Claret Creek, Ootann and O'Briens Creek.

Granites of the Almaden Supersuite are intermediate to felsic (56-72% SiO2) and are characterised by high K2O, K/K(K + Na), Rb, Rb/Sr, Th, U and relatively low Ba and Sr. The Claret Creek Supersuite granites are a little more felsic (65-77% SiO2), and are chemically distinctive, having higher A12O3, CaO, Na2O and Sr, and lower K2O, Rb, Th and U than granites of the Almaden Supersuite.

Granites of the Ootann and O'Briens Creek supersuites all contain more than 70% SiO2 and these comprise more than 90% of the total area of granites. These two supersuites are characterised by low Sr, Sr/Y and large negative Eu/Eu*, with the more evolved rocks becoming strongly depleted in TiO2, FeO* MgO, CaO, Ba, Sr, Sc, V, Cr, Ni, Eu, CeN/YN and K/Rb, and enriched in Rb, Pb, Th, U and Rb/Sr. Granites belonging to the O'Briens Creek Supersuite contain significantly higher abundances of HFSE, HREE and F (0·2-0·5 wt%) than those of the Ootann Supersuite, and as such have developed some characteristics of A-type granites.

Geochemical and isotopic properties suggest that all granites are of crustal derivation. The granites of all supersuites have very similar initial 87Sr/86Sr and εNd of 0·710 and −7·0–−8·0, respectively, except where they outcrop within Proterozoic country rocks, when they have more evolved εNd (−8·0–−11·0). Depleted-mantle model ages cluster around 1·5 Ga. The isotope systematics and geochemistry indicate that these granites were not derived from the equivalents of any exposed country rocks.

Models for the petrogenesis of these granites all appear to require the involvement of a long-lived and isotopically homogeneous crustal protolith, that most probably underplated the crust in the Proterozoic. Granites of the two more felsic supersuites were either derived by varying degrees of partial melting from this protolith of andesitic to dacitic composition, and/or were produced by a two-stage process by remelting of intermediate rocks similar in composition to the mafic end-members of the Almaden Supersuite. The resulting primary partial melts for the Ootann and O'Briens Creek supersuites underwent extensive, high-level, feldspar-dominated, crystal fractionation.

The Layos Granite forms elongated massifs within the Toledo Complex of central Spain. It is late-tectonic with respect to the F2 regional phase and simultaneous with the metamorphic peak of the region, which reached a maximum temperature of 800–850°C and pressures of 400–600 MPa. Field studies indicate that this intrusion belongs to the “regional migmatite terrane granite” type. This granite is typically interlayered with sill-like veins and elongated bodies of cordierite/garnet-bearing leucogranites. Enclaves are widespread and comprise restitic types (quartz lumps, biotite, cordierite and sillimanite-rich enclaves) and refractory metamorphic country-rocks including orthogneisses, amphibolites, quartzites, conglomerates and calc-silicate rocks.

These granites vary from quartz-rich tonalites to melamonzogranites and define a S-type trend on a QAP plot. Cordierite and biotite are the mafic phases of the rocks. The particularly high percentage of cordierite (10%–30%) varies inversely with the silica content. Sillimanite is a common accessory mineral, always included in cordierite, suggesting a restitic origin. The mineral chemistry of the Layos Granite is similar to that of the leucogranites and country-rock peraluminous granulites (kinzigites), indicating a close approach to equilibrium. The uniform composition of plagioclase (An25), the high albitic content of the K-feldspar, the continuous variation in the Fe/Mg ratios of the mafic minerals, and the high Ti content of the biotites (2.5–6.5%) suggest a genetic relationship.

Geochemically, the Layos Granite is strongly peraluminous. Normative corundum lies between 4% and 10% and varies inversely with increase in SiO2. The CaO content is typically low (<1.25%) and shows little variation; similarly the LILE show a limited range. On many variation diagrams, linear trends from peraluminous granulites to the Layos Granite and associated leucogranite can be observed. The chemical characteristics argue against an igneous fractionation or fusion mechanism for the diversification of the Layos Granite. A restite unmixing model between a granulitic pole (represented by the granulites of the Toledo Complex) and a minimum melt (leucogranites) could explain the main chemical variation of the Layos Granite. Melting of a pelitic protolith under anhydrous conditions (biotite dehydration melting) could lead to minimum-temperature melt compositions and a strongly peraluminous residuum.

For the most mafic granites (61–63% SiO2), it is estimated that the trapped restite component must have been around 65%. This high proportion of restite is close to the estimated rheological critical melt fraction, but field evidence suggests that this critical value has been exceeded. This high restite component implies high viscosity of the melt which, together with the anhydrous assemblage of the Layos Granite and the associated leucogranites, indicates H2O-undersaturated melting conditions. Under such conditions, the high viscosity magma (crystal-liquid mush) had a restricted movement capacity, leading to the development of parautochthonous plutonic bodies.

Approximately 10-15 vol% of the Neogene Hoyazo dacite consists of Al-rich restite rock inclusions (A12O3 = 20–45%) and monocrystal inclusions derived therefrom. Restite material and dacitic melt were formed syngenetically from a (semi-)pelitic rock sequence by means of anatexis. Restite rock fragments and dacite show similar high δ18O values (13–16‰) corresponding to those found for sedimentary material. Striking monocrystal restite inclusions in the dacite rock are graphite crystals measuring a few hundred μm, 0.5–10 mm blue cordierite crystals and 2–10 mm ruby red crystals of almandine-rich garnet (1.1 ± 0.2 vol%). Although the almandine crystals are perfectly euhedral, they are identical in every respect to the crystals found in the Al-rich restite rock inclusions and cannot be crystallisation products of the magmatic melt. The dacite also contains many inclusions of quartz gabbroic and basaltoid material which contains inclusions identical to the restite material found in the dacitic glass base. Many basaltoid inclusions show well-developed chilled borders. These inclusions may represent a more mafic magma of deeper origin which mixed with some dacite magma before mingling into it.

Thermal, mechanical and chemical exchange occurs between felsic and mafic magmas in dynamic magma systems. The occurrence and efficiency of such exchanges are constrained mainly by the intensive parameters, the compositions, and the mass fractions of the coexisting magmas. As these interacting parameters do not change simultaneously during the evolution of the granite systems, the exchanges appear sequentially, and affect magmatic systems at different structural levels, i.e. in magma chambers at depth, in the conduits, or after emplacement. Hybridisation processes are especially effective in the plutonic environment because contrasting magmas can interact over a long time-span before cooling. The different exchanges are complementary and tend to reduce the contrasts between the coexisting magmas. They can be extensive or limited in space and time; they are either combined into mixing processes which produce homogeneous rocks, or only into mingling processes which produce rocks with heterogeneities of various size-scales. Mafic microgranular enclaves represent the most common heterogeneities present in the granite plutons. The composite enclaves and the many types of mafic microgranular enclaves commonly associated in a single pluton, or in polygenic enclave swarms, are produced by the sequential occurrence of various exchanges between coexisting magmas with constantly changing intensive parameters and mass fractions. The complex succession and repetition of exchanges, and the resulting partial chemical and complete isotopic equilibration, mask the original identities of the initial components.

Igneous charnockites are characterised by distinctively high abundances of K2O, TiO2, P2O5 and LIL elements and low CaO at a given SiO2 level compared to metamorphic charnockites, and I-, S- and A-type granites. They form a distinctive type of intrusive igneous rocks, the Charnockite Magma Type (CMT or C-type), which generally lack hornblende and consist of pyroxene, alkali feldspar, plagioclase, quartz, biotite, apatite, ilmenite and titanomagnetite. Although this mineral assemblage superficially resembles that of metamorphic charnockites, magmatic charnockites are characterised by inverted pigeonite, exceptionally calcic alkali feldspar, potassic plagioclase, and coexisting opaque oxides, all with crystallisation temperatures of 950-1050°C. Apatite is a ubiquitous phase which, together with the very high concentrations of Zr and TiO2 over a wide silica range, is consistent with the derivation of the Charnockite Magma Type by very high temperature partial melting and fractionation.

The credibility of intrusive charnockites as a magmatic type has historically foundered because of their apparent restriction to granulite belts and the absence of any reported extrusive equivalents. We report examples of volcanic rocks, of various ages, with the same distinctive major and trace element compositions, mineral assemblages and high temperatures of crystallisation as intrusive chamockites.

The Charnockite Magma Type is considered to be derived by melting of a hornblende-free or poor, LILE-enriched fertile granulite source which had not been geochemically depleted by a previous partial melting event but which was dehydrated in an earlier metamorphism. Whereas H2O-saturated melting produces migmatites or "failed" granites, and vapour-absent melting of an amphibolite can produce I-type granites, according to this model the vapour-absent melting of a hornblende-free or hornblende-poor granulite at even higher temperatures produces charnockites.

The 1·65–1·54 Ga rapakivi granites of southeastern Fennoscandia represent the silicic members of a bimodal magmatic association in which the mafic members are tholeiitic diabase dykes and minor gabbroic-anorthositic bodies. They are metaluminous to slightly peraluminous A-type granites and occur as high-level batholiths and stocks in an E-W-trending belt extending from Soviet Karelia to southwestern Finland. The Soviet Karelian granites were emplaced into the contact zone between Archaean craton and Svecofennian juvenile 1·9Ga-old crust, while the Finnish granites were intruded into the Svecofennian crust. Deep seismic soundings show that the rapakivi granites and the contemporaneous, mainly WNW or NW-trending diabase dyke swarms are situated in a zone of relatively thin crust. Below the Wiborg Batholith there exists a domal structure in the lithosphere in which a transitional zone (mafic underplate) occurs between the crust and the mantle.

The Nd isotopic evolution of the rapakivi granites (εNd(T) −3·1—−0·2) corresponds to the evolution of the 1·9Ga-old Svecofennian crust, as do their Pb isotopic compositions. This implies that the Finnish granites represent anatectic melts of the Svecofennian crust. In contrast, the Soviet Karelian granites show isotopic composition indicative of substantial incorporation of Archaean lower crust material. Petrochemical modelling of one of the Finnish batholiths shows that its parental magma could have been generated by c. 20% melting of a granodioritic source and that fractional crystallisation was important during the subsequent evolution of this magma.

The rapakivi granites are redefined as A-type granites that show the rapakivi texture at least in larger batholiths. The field, geochemical, and seismic data indicate that the classical Finnish rapakivi granites were generated in an anorogenic extensional regime by partial melting of the lower/middle crust. The melting, and possibly also the extensional tectonics, were related to upwellings of hot mantle material which led to intrusion of mafic magmas at the base and into the crust.

Early Proterozoic rapakivi intrusions in S Greenland occur as thick sheets which have ramp–flat geometry and were intruded along the median planes of active ductile extensional shear zones. These shear zones and their intrusions were linked via transfer zones in a major three-dimensional framework. At high structural levels (c. 6 km) the rapakivi intrusions developed thermal aureoles which overprint the regional assemblages, whereas at deeper levels in the regional structure they are contemporaneous with regional metamorphism. Thermobarometry on the regional and contact assemblages indicates low pressure granulite facies conditions (200–400 MPa, 650°-800°C) suggesting very high thermal gradients. The rapakivi suite and associated norites have low initial 87Sr/86Sr together with positive εNd values, indicating the involvement of predominantly young crust and/or mantle component in the generation of the igneous suite. It is considered that the voluminous norites are closely related to the mafic melts which underplated the juvenile crust to trigger the generation of the monzonitic rapakivi suite. Taken together, the data are consistent with a model of Proterozoic lithospheric extension, thinning of relatively juvenile continental crust and compression of mantle isotherms, resulting in high crustal heat flow, mafic underplating, and crustal melting with emplacement of magmas along a linked network of extensional shear zones.

Most zoned plutons described in the geological literature have mafic rims and felsic cores and are referred to as “normally zoned”, whereas relatively few “reversely zoned” intrusions (felsic rims and mafic cores) have been described. That unusual zonation pattern has been variously attributed to in situ processes or to the reordering of an underlying, vertically stratified, magma chamber either by intrusion through an orifice or by emplacement of composite diapirs. The Turtle Pluton is an early Cretaceous, reversely zoned, intrusion that is divided into four facies: a Rim Sequence that is graditionally zoned from bt + ilm + muse monzogranite to hb + bt + mt + sph granodiorite; a Core Facies of more homogeneous hb + bt + mt + sph granodiorite to quartz monzodiorite; between these two facies, a structural discontinuity termed the Schlieren Zone; and an Eastern Facies of monzogranite to granodiorite. Field relationships, distribution of strain, and geochemical and isotopic studies (including a range of initial87Sr/86Sr from 0·7085–0·7065) suggest that the reverse zonation of the Turtle Pluton is the result of sequential emplacement of two diapirs each derived from the same underlying, vertically stratified, magma chamber, and that the Rim Sequence zonation is chiefly the result of mixing of intermediate and felsic magmas from distinct sources accompanied by minor fractional crystallisation.

The Criffell pluton in southwestern Scotland (397 Ma, a Newer Granite of late Caledonian age) is concentrically zoned with outer granodiorites of typically I-type aspect passing into inner granite with more evolved characteristics. The zonation is examined in terms of the compositional surfaces of bulk parameters such as SiO2 and Rb/Sr and compositional variation is best modelled as multi-pulse, there being greater variation in bulk composition between pulses than within pulse. Published variations in Sr, Nd and O isotopes reflect the derivation of the pulses from separate and isotopically distinct sources. Other evidence for open-system behaviour includes mingling with mafic magmas to form enclaves, whereas closed-system behaviour is indicated by restite separation in the early granodiorites, and fractional crystallisation in the late granites. A dominant infracrustal I-type magma formed the first pulse followed by magma derived from more evolved crustal rocks (mainly metasediments of varying ages and maturities). Experimental fluid-absent melting of amphibolite and metapelite at about 900°C has shown that significant quantities of melt can be generated, respectively with I-type and S-type characteristics. Despite having similar bulk compositions, these melts have very different viscosities and densities for the same H2O contents (ηS-type>ηI-type and ρS-type≤ρI-type). It is argued that the rheological controls on magma escape from the source region along complex and tortuous pathways favour the more fluid I-type melts over the more viscous (and only slightly less dense) S-type melts. This constraint could have the effect of reversing the expected buoyancy-driven emplacement sequence, and may represent an alternative rheological differentiation mechanism for the formation of some zoned plutons.

Granites and their associated comagmatic felsic volcanic rocks occur in most Proterozoic provinces of Australia. Using multi-element, primordial-mantle-normalised abundance diagrams and various petrological characteristics, Australian Proterozoic granites can be subdivided into five groups: (i) I-type, Sr-depleted, Y-undepleted, restite-dominated, (ii) I- type, Sr-depleted, Y-undepleted, fractionated, low in incompatible elements, (iii) I-type Sr-depleted, Y-undepleted, enriched in incompatible elements (anorogenic granites), (iv) I-type, Sr-undepleted, Y-depleted, (v) S-type, Sr-depleted, Y-undepleted. The four Sr-depleted groups dominate, and group (iv) is of very limited extent. A comparison of these Proterozoic granites with Australian and Papua New Guinean granites of other time periods shows that these characteristic Sr-depleted Y-undepleted patterns are also dominant in early Palaeozoic granites. They are significantly different from those of granites in modern island arcs associated with subduction, and with most granites from Archaean terranes, where the multi-element diagrams are dominated by Sr-undepleted, Y-depleted patterns.

The Sr-depleted, Y-undepleted patterns are thought to indicate source regions that contained plagioclase but not garnet, whilst the Sr-undepleted, Y-depleted patterns are taken to correspond with the presence of garnet, but not plagioclase, in the source rocks. The Sr-depleted, Y-undepleted patterns also only occur in regions where the lower crustal structure is dominated by an underplated mafic layer with a P-wave velocity of 7·2-7·-4 km/s. In contrast, in regions where the granites are dominated by Sr-undepleted, Y-depleted patterns, such as in the Archaean and in Cainozoic island arcs, this intermediate velocity layer is not present, and the crust-mantle boundary is very sharp.

Two other distinctive compositional changes have been noted among the I-type granites of different age. Firstly, Na is highest in Archaean and Cainozoic granites, and lowest in early Proterozoic granites; Palaeozoic and Mesozoic granites have intermediate values. Secondly, late Archaean and Proterozoic granites are the most enriched in K, Th and U, while the Cainozoic and early Archaean tonalites are the most depleted; Palaeozoic and Mesozoic granites again contain intermediate amounts of those elements.

Late Archaean granitic rocks from the southern Yilgarn Craton of Western Australia have a close temporal relationship to the basaltic and komatiitic volcanism which occurs within spatially associated greenstone belts. Greenstone volcanism apparently began ∼2715 Ma ago, whereas voluminous felsic magmatism (both extrusive and intrusive) began about 2690 Ma ago. A brief but voluminous episode of crust-derived magmatism ∼2690-2685 Ma ago resulted in the emplacement of a diverse assemblage of plutons having granodioritic, monzogranitic and tonalitic compositions. This early felsic episode was followed immediately by the emplacement of mafic sills, and, after a further time delay, by a second episode of voluminous crust-derived magmatism dominated by monzogranite but containing plutons covering a wide compositional range, including diorite, granodiorite and tonalite. The products of this 2665–2660 Ma magmatic episode now form a significant fraction of the exposed southern Yilgarn Craton. Later magmatism, which continued to at least 2600 Ma ago, appears largely restricted to rocks having unusually fractionated compositions.

The magmatic sequence basalt-voluminous crust-derived magmatism-later diverse magmatism, is interpreted in terms of a dynamically-based model for the ascent of the head of a new mantle plume. In this model basalts and komatiites are derived by decompression melting of rising plume material, and the crust-derived magmas result after conductive transport of heat from the top of the plume head into overlying continental crust. This type of magmatic evolution, the fundamentally bimodal nature of the magmatism, the presence of high-Mg volcanics (komatiites), and the areal extent of the late Archaean magmatic event, are all suggested to be characteristic of crustal reworking above mantle plumes rather than resulting from other processes, such as those related to subduction.