A High-K, Mantle Derived Plutonic Suite from 'Linga', near Arequipa (Peru)

The Linga Group consists of a suite of Cretaceous high-K calc-alkalinerocks intruded into 2?0 Ga old basement in S.E. Peru. The rocksrange in composition from gabbros, through diorites and grandioritesto granites. Microprobe, major and trace element and isotopedata suggest the suite evolved by fractional crystallization,with plagioclase as the dominant phase and with surprisinglylittle interaction with the pre-existing crust. The rocks yieldan Rb/Sr whole rock isochron of 68 ? 3 Ma with an initial Sr-isotoperatio 0.70516 ? 8, and from three Nd-isotope analyses initialNd = – 1?4 to – 2?0. ¹⁸ O increases from 5?0 permil in a gabbro to 7?0 per mil in a granite, and models arepresented which suggest that the suite evolved from parentalmagmas which had ¹⁸O = 5?8 to 6?0 per mil, 1.5 per cent K₂O,63 p.p.m. Rb, 582 p.p.m. Sr, 35 p.p.m. Ce, 0.38 p.p.m. Ta, ⁸⁷Sr/⁸⁶Sr= 0.7052, and ¹⁴³Nd/¹⁴⁴Nd = 0.51247. Trace element considerationsimply that these magmas contain contributions from incompatibleelement enriched upper mantle material and from a componentassociated with subduction. The latter is thought to reflectthe preferential mobilization of alkali and alkaline earth elementsby H₂O-rich fluids from the subducted slab, and preliminarycalculations indicate that it was responsible for {small tilde}45 per cent of the Sr and {small tilde} 80 per cent of the Kin the more primitive Linga rocks. However, the isotope datasuggest that while the fluids may be from the subducted slab,the elements in this component may also have been mobilizedfrom the overlying mantle wedge.     

Oxygen and Hydrogen Isotope Geochemistry of the Mourne Mountains Tertiary Granites, Northern Ireland

An oxygen and hydrogen isotopic study of minerals and wholerocks from the granites of the Mourne Mountains Tertiary complex,and related rocks, shows that whereas a significant circulationof meteoric water was associated with the complex, it had onlyminor and localized effects on the granites themselves. TheSilurian slate and greywacke country rocks, which would havehad ¹⁸O_(SMOW) values of +10 to +20 before the Tertiary igneousevents, have been depicted ¹⁸O to values of –40 to –05Tertiary acid minor intrusions outside the main granite massesare also ¹⁸O depleted. ^l8O whole-rock data on the granites showa range of +6.0 to +9.5, and include values significantly higherthan most of those obtained for the granites of the Tertiarycentral complexes of Skye, Mull, and Ardnamurchan. Many of thelowest whole-rock ¹⁸O values are found in samples where theminerals are not in isotopic equilibrium. The mineral oxygenisotopic data can be explained in terms of localized interactionwith meteoric water, resulting in preferential ¹⁸O depletionin feldspar(s) and biotite, with quartz being much less affected.The granites all show low values of D_(SMOW) for biotite andamphibole separates (–137 to –104). The lowest valuesoccur close to the margins of the plutons, near internal contactsor near greisen localities, and these probably reflect limitedinteraction with meteoric water. The higher D values are fromsamples which show evidence of chloritization. This processappears to have occurred both during interaction with meteoricwater, and also during autometasomatism by an exsolved magmaticfluid in other parts of the plutons, including central locationswhere there is little or no evidence for the penetration ofmeteoric water. Granite samples which exhibit near-equilibriumoxygen isotope fractionations for constituent minerals are characterizedby magmatic O-isotopic compositions. The G2 granite, the largestpluton of the eastern centre, has a magmatic ¹⁸O_(SMOW) valueof {small tilde}+95; intrusions G3 (eastern centre) and G4(western centre) both have ¹⁸O_(SMOW) values of {small tilde}+90.The other two main intrusive phases have distinctly lower ¹⁸O_(SMOW)values: {small tilde}+75 for Gl (the least fractionated graniteof the Mourne Mountains central complex), and from +75 to +85for G5. The oxygen isotopic data rule out simple partial meltingof the country rocks as the origin of the granites and alsopreclude an origin by closed-system fractional crystallizationof basaltic magma typical of that represented by Tertiary basicigneous rocks of the region. ^* Present address: NERC Isotope Geosciences Laboratory, Keyworth, Nottingham BG12 5GG, UK Present address: School of Engineering Technology, Georgian College, Barrie, Ontario, L4M 3X9, Canada     

Dehydration--Melting Phenomena in Leptynitic Gneisses and the Generation of Leucogranites: a Case Study from the Kerala Khondalite Belt, Southern India

Pan-African high-grade metamorphism in the Kerala KhondaliteBelt (South India) led to the in situ formation of garnet-bearingleucosomes (L1) in sodic quartz—alkali feldspar—biotitegneisses. Microtextures, mineralogy and the geochemical characteristicsof in situ leucosomes (L1) and gneiss domains (GnD) indicatethat the development of leucosomes was mainly controlled bythe growth of garnet at the expense of biotite. This is documentedby the selective transfer of FeO, MgO, , Sm and the heavy rareearth elements into the L1 domains. P-T constraints (T>800C,P>6kbar, aH₂O0.3) suggest that the leucosomes were formedthrough complete melting of biotite in fluid-absent conditions,following the model reaction Biotite+Alkali feldspar+QuartzlGarnet+Ilmenite+Melt.The fraction of melt generated during this process was low (<10vol.%). The identical size of the leucosomes as well as theirhomogeneous and isotropic distribution at outcrop scale, whichlacks any evidence for melt segregation, suggest that the migmatiteremained a closed system. Subsequent to migmatization, the leptyniticgneisses were intruded by garnet-bearing leucogranitic melts(L2), forming veins parallel and subperpendicular to the foliation.The leucogranites are rich in potassium (K₂O5.5 wt%), (Ba400p.p.m.) and Sr (300 p.p.m.), and exhibit low concentrationsof Zr (40 p.p.m.), Th (<1 p.p.m.) and (<10 p.p.m.). Thechondrite-normalized REE spectra show low abundances (La_N20,Lu_N3) and are moderately fractionated (La_N/Lu_N7). An Eu anomalyis absent or weakly negative. The higher ⁸⁷Sr/⁸⁶Sr ratio at550 Ma (0.7345) compared with the migmatite (0.7164) precludesa direct genetic relationship between leptynitic gneisses andleucogranites at Manali.Nevertheless, the chemical and mineralogicalcompositions of the leuocogranites strongly favour a derivationthrough fluid-absent biotite melting of isotopically distinctbut chemically comparable Manali-type gneisses. The undersaturationof Zr, Th and REE, a typical feature of leucogranitic meltsgenerated during granulite facies anatexis of psammo-peliticlithologies and attributed to disequilibrium melting with incompletedissolution of accessory phases (zircon, monazite), is weaklydeveloped in the leucogranites of Manali.It is concluded thatthis is mainly due to the sluggish migration of the melts instatic conditions, which facilitated equilibration with therestitic gneisses. ^*Fax: 0228-732763; e-mail: ingo.braun{at}uni-bonn.de     

Geochemical and Isotopic (O, Nd, Pb and Sr) Constraints on A-type Granite Petrogenesis Based on the Topsails Igneous Suite, Newfoundland Appalachians

The voluminous, bimodal, Silurian Topsails igneous suite consistsmainly of ‘A-type’ peralkaline to slightly peraluminous,hypersohnis to subsolvus granites with subordinate syenite,onzonite and diabase, plus consanguineous basalts and highsilicarhyolites. _Nd(T) values from the suite range from –1.5to +5.4; most granitoid components exhibit positive _Nd(T) values(+1.1 to +3.9). Granitoid initial ⁸⁷Sr/⁸⁶Sr and most ¹⁸ O valuesare in the range expected for rocks derived from mantle-likeprotoliths (0.701–0.706 and +6 to +80/). Restricted ²⁰⁷Pb/²⁰⁴Pbvariation is accompanied by significant dispersion of ²⁰⁶Pb/²⁰⁴Pband ²⁰⁸Pb/²⁰⁴Pb. Superficially, petrogenesis by either direct(via fractionation from basalt) or indirect (via melting ofjuvenile crust) derivation from mantle sources appears plausible.Remelting of the granulitic protolith of Ordovician are-typegranitoids can be ruled out, because these rocks exhibit negative_Nd(T) and a large range in ²⁰⁷Pb/²⁰⁴Pb. Geochemical and isotopicrelationships are most compatible with remelting of hybridizedlithospheric mantle generated during arc-continent collision.A genetic link is suggested among collision-related delaminationor slab break-off events and emplacement of ‘post-erogenic’granite suites. A-type granites may recycle previously subductedcontinental material, and help explain the mass balance notedfor modern arcs. However, they need not represent net, new,crustal growth. KEY WORDS: A-type granites; juvenile crust; isotopes; Newfoundland ^*Telephone: (613) 995-4972. Fax: (613) 995-7997. e-mail: jwhalen{at}gsc.emr.ca     

Contemporaneous Convergent Margin and Intraplate Magmatism, North Island, New Zealand

A convergent margin magma series with characteristic low Nband Ta abundances and enrichments in alkalis and alkaline earthsis intercalated with typical intraplate alkalic basalts in aback-arc setting, 200–250 km above the Wadati-Benioffzone on the North Island, New Zealand. These two contrastingmagma types, together with late-stage K-rich maflc lavas, wereerupted over a short time period (1{dot}60–2{dot}74 Ma)and constitute the Alexandra Volcanics. Field relationshipsindicate that these diverse magma types were contemporaneous,and thus their mantle source regions coexisted, in a singletectonic environment. The convergent margin magma series forms a linear chain of stratovolcanoesaligned at right angles to the present subduction zone. Closed-systempolybaric fractional crystallization models can explain theevolution from ankaramites to transitional olivine basalts toolivine tholeiites to high-Al basalts to medium- and high-Kandesites. The most primitive lavas have geochemical (high LIL/LREEand LIL/HFS element ratios) and Sr, Nd, and Pb isotopic compositionstypical of convergent margin magmas. Calculated source compositionssuggest that three components are involved: a MORB component,a component derived from subducted oceanic crust, and a contributionfrom subducted sediments. The alkalic basalts occur as dispersed monogenetic volcanoesand are intercalated with the larger convergent margin stratovolcanocs.These basalts are enriched in LILE, LREE, Nb, and Ta, and havelow Ba/Nb and Ba/La ratios, all of which are characteristicof ocean island (intraplate) basalts (OIBs). Their relativelyhigh _Nd (+5{dot}5 and low ⁸⁷Sr/⁸⁶Sr(0{dot}703l–0{dot}7036)are also typical of OIBs. These alkalic magmas were derivedfrom the underlying continental lithospheric mantle that hasbeen enriched by upward-migrating silica-undersaturated melts,probably including volatiles, from the low- velocity zone. Asubducted slab component is not required to account for theirincompatible element enriched character. The K-rich mafic lavas, basanites, and absarokites are volumetricallyminor and cap the largest of the stratovolcanoes, Pirongia.The basanites have geochemical and isotopic compositions whichsuggest they are mixtures of multiple source components, includingthe alkalic and convergent margin region.     

The Petrology and Origin of the Tertiary Lundy Granite (Bristol Channel, UK)

The British Tertiary Volcanic Province (BTVP) comprises within-platecentral igneous complexes associated with plateau lavas andregional dyke swarms. Lundy is the southernmost complex of theBTVP and comprises granite ({small tilde}90%) emplaced intodeformed Devonian sedimentary rocks within the Hercynian Cornubiangranite province of southwest England. The complex is intrudedby a northwest-southeast trending dyke swarm. In common withother BTVP igneous complexes, Lundy is associated with positivegravity and magnetic anomalies which are interpreted in termsof the presena of an underlying basic intrusion at shallow depth,with a volume exceeding that of the overlying granite. The Lundy intrusion is a coarse-grained megacrystic granitecontaining up to 20% alkali feldspar megacrysts in a coarse-grainedgroundmass composed of alkali feldspar, quartz, lithium-bearingmuscovite, and ‘biotite’ (lithian siderophyllite),with a range of aaxssory minerals. The main granite has a coarse-grained(locally miarolitic) pegmatitic facies and is intruded by thinsheets and veins of fine-grained aplite and microgranite. Themineralogy indicates crystallization of the Lundy granite froma highly fractionated H₂O- and halogen-rich magma at a relativelyshallow crustal level. The main Lundy granite is a peraluminous leucogranite with Na₂O=3–4%,K₂O{small tilde}5%, low TiO₂, MeO, CaO, Zr, and Sr, and highRb and Rb/Sr in comparison with many other peralurninous granites,including those from the Cornubian batholith and the BTVP. Anew Rb-Sr whole-rock isochron for the granite yields an ageof 58?7?1?6 Ma with an initial ⁸⁷Sr/86Sr of 0?715?0?006. _Ndvalues for the granite (–0?9 to –1?9) plot betweencontemporaneous mantle (positive _Nd and Cornubian granites (_Nd=ca.–11). The trace element data (Rb, Y, Nb) show affinities with syn-collisionand within-plate granites. As the Sr isotope data indicate amajor crustal component, and the Nd isotope data suggest bothmantle and crustal components, we propose that the Lundy graniteis derived from a parental magma comprising crustal components(derived from a similar source to that of the Cornubian granitebatholith) and a mantle-derived component (derived from a differentiateof contemporaneous basaltic magma This magma experienced fractionalcrystallization of plagioclase, alkai feldspar, Fe-Mg minerals,and REE-bearing accessory minerals before emplacement, and theLundy granite experienced further in situ fractional crystallization,associateded with crustal contamination by the Devonian shaleafter emplacement.     

Genetic Significance of Multiple Enclave Types in a Peraluminous Ignimbrite Suite, Lachlan Fold Belt, Australia

The Violet Town Volcanics (Lachlan Fold Belt, Australia) arean S-type ignimbrite suite containing microgranitoid enclaves,basaltic andesite enclaves and enclaves of high-silica rhyolite.The microgranitoid enclaves are similar to those in peraluminousgranites. They typically have lower initial ⁸⁷Sr/⁸⁶Sr and higherNd than the host, and represent globules of a mafic, mantle-derivedmagma, which was hybridized by mixing and diffusional exchangewith the host magma. The basaltic andesite enclaves were incorporatedinto the ignimbrite as xenoliths, but their parental magma mayhave been similar to that of the microgranitoid enclaves. Theyare isotopically less depleted than other mantle-derived rocksfrom the Lachlan Fold Belt, reflecting contamination by crustalmaterial, or derivation from less depleted mantle sources. Thehigh-silica rhyolite enclaves, previously interpreted to berelated to the ignimbrite by crystal fractionation, have Ndvalues up to 3 units higher than their host, and cannot be relatedto their host by crystal fractionation or assimilation-fractionalcrystallization (AFC) processes. The coexistence of S-type magmasand mantle-derived magmas suggests that the latter may haveplayed a role in the Palaeozoic magmatism of the Lachlan FoldBelt, acting as a heat source for melting and perhaps also contributingchemical components to the crustally derived magmas. KEY WORDS: enclaves; magma mingling; magma mixing; S-type ^*Present address: Department of Geology and Geophysics, University of Adelaide, Adelaide, S.A. 5005, Australia. Telephone: +-61-8-3035973. Fax: +-61-8-3034347. e-mail: melburg{at}geology.adelaide.edu.au     

Multiple Isotopic Components in Quaternary Volcanic Rocks of the Cascade Arc near Crater Lake, Oregon

Quaternary lavas and pyroclastic rocks of Mount Mazama, CraterLake caldera, and the surrounding area have variable Sr, Nd,and Pb isotopic compositions. High-alumina olivine tholeiites(HAOT) have ⁸⁷Sr/⁸⁶Sr ratios of 0.70346–0.70364; basalticandesite, 0–70349–0.70372; shoshonitic basalticandesite, 0.70374–0.70388; and andesite, 0.70324–0.70383.Dacites of Mount Mazama have ⁸⁷Sr/⁸⁶Sr ratios of 0.70348–0.70373.Most rhyodacites converge on 0.7037. However, rhyodacite ofthe caldera-forming, climactic eruption has ⁸⁷Sr/⁸⁶Sr=0.70354because of an admixed low-⁸⁷Sr/⁸⁶Sr component. Andesitic tomafic-cumulate scoriae of the climactic eruption, and enclavesin preclimactic rhyodacites, cluster in two groups but shownearly the entire ⁸⁷Sr/⁸⁶Sr range of the data set, confirmingpreviously suggested introduction of diverse parental magmasinto the growing climactic chamber. Pb and Nd isotope ratiosdisplay less variation (²⁰⁶Pb/²⁰⁴Pb= 18.838–18.967, ²⁰⁷Pb/²⁰⁴Pb=15.556–15.616,²⁰⁸Pb/²⁰⁴Pb=38.405–38.619; _Nd= +3.9 to +6.1) and generallycovary with ⁸⁷Sr/⁸⁶ Sr ratios. Radiogenic isotope data fromCrater Lake plot with published data for other Cascade volcanoeson isotope ratio correlation diagrams. The isotopic data for the Crater Lake area require sources ofprimitive magmas to consist of depleted mantle and a subductioncomponent, introduced in variable quantity to the depleted mantlewedge. Variable degrees of melting of this heterogeneous mantle,possibly at different depths, produced the diversity of isotopiccompositions and large-ion lithophile element (LILE) abundancesin primitive magmas. Trace element ratios do not indicate presenceof an ocean island basalt (OIB) source component that has beenreported in lavas of some other Cascade volcanoes. Crustal contamination may have affected isotope ratios and LILEconcentrations in evolved HAOT, where initial LILE concentrationswere low. Contamination is more difficult to detect in the calcalkalinelavas because of their higher LILE concentrations and the smallisotopic contrast with likely contaminants, such as mid- tolower-crustal rocks thought to be equivalents of igneous rocksof the Klamath Mountains and associated lower crust. Crustalassimilation appears to be required for calcalkaline rocks onlyby ¹⁸O values, which vary from lows of +5.6 to + 6.0% in HAOTand primitive basaltic andesites to a high of +7.0% in dacite,a range that is too high to be explained by plagioclase-dominatedclosed-system fractional crystallization. Elevated ¹⁸O valuesof differentiated lavas may be attributed to interaction withrelatively ¹⁸O-rich, ⁸⁷Sr-poor crustal rocks. Variably fused granitoid blocks ejected in the climactic eruption,and rarely in late Pleistocene eruptive units, have ¹⁸Opl of–3.4 to +6.5% and ¹⁸O_qz of –2.2 to +8.0% but haveSr, Nd, and Pb isotope ratios similar to volcanic rocks (e.g.⁸⁷Sr/⁸⁶Sr0.7037). Rb and Sr data for glass separates from granodioritessuggest that the source pluton is Miocene. Glass from granodioritehas ⁸⁷Sr/⁸⁶Sr ratios as high as 0.70617. Oxygen isotope fractionationbetween quartz, plagioclase, and glass indicates requilibrationof O isotopes at magmatic temperatures, after ¹⁸O/¹⁶O had beenlowered by exchange with meteoric hydrothermal fluids. Unmeltedgranodiorite xenoliths from pre-climactic eruptive units have¹⁸O values that are consistent with onset of hydrothermal exchangeearly during growth of the climactic magma chamber. Assimilationof such upper-crustal granodiorite apparently lowered ¹⁸O valuesof rhyodacites without significantly affecting their magmaticcompositions in other ways.     

Element Mobility During Incipient Granulite Formation at Kabbaldurga, Southern India

At Kabbaldurga, infiltration of carbonic fluids along a systemof ductile shears and foliation planes has led to partial transformationof Archaean grey biotite–hornblende gneiss to coarse-grainedmassive charnockite at about 2.5 b.y. ago. The dehydration ofthe gneiss assemblage was induced by a marked metasomatic changeof the reacting system from granodioritic to granitic, and obviouslytook place under conditions of an open system at 700–750?C and 5–7 kb. Extensive replacement of plagioclase (An_16–30)by K-feldspar through Na, Ca–K exchange reactions withthe ascending carbonic fluids led to strong enrichment in K,Rb, Ba, and SiO₂, and to a depletion in Ca. Progressive dissolutionof hornblende, biotite, magnetite, and the accessory mineralsapatite and zircon resulted in a marked depletion in Fe, Mg,Ti, Zn, V, P, and Zr. Most important is the recognition of REEmobility: with advancing charnockitization, the moderately fractionatedREE distribution patterns of the grey gneisses (La_N270; La_N/Yb_N= 5–20; Eu_N27; Eu/Eu^* = 0.6–0.3) give way to stronglyfractionated REE patterns with a positive Eu-anomaly (La_N200;La_N/Yb_N = 20–80; Eu_N22; Eu/Eu^* = 0.6–1.8). The systematicdepletion especially in the HREE is due to the progressive dissolutionof zircon, apatite (and monazite), which strongly concentratethe REE. Stable isotope data (¹⁸O of 6.9–8.0 per mille for gneissesand charnockites; ¹³C of –8.5 and –6.5 per millefor late carbonate) indicate a magmatogenic source for the carbonicfluids. In contrast to the currently favoured derivation ofcarbonic fluids by decarbonation of the upper mantle or degassingof underplated basaltic intrusions, it is discussed here thatabundant fluid inclusions in lower crustal charnockites providedan extensive reservoir of ‘fossil’ carbonic fluids.Shear deformation has tapped this reservoir and generated thechannel-ways for fluid ascent. Charnockitization of the Kabbaldurgatypethus appears to be a metasomatic process which is tectonicallycontrolled and restricted to the crustal level of the amphiboliteto granulite transition.     

Petrogenesis of Cogenetic Silica-Oversaturated and -Undersaturated Syenites by Periodic Recharge in a Crustally Contaminated Magma Chamber: the Kangerlussuaq Intrusion, East Greenland

The Palaeogene Kangerlussuaq Intrusion (50 Ma) of East Greenlanddisplays concentric zonation from quartz-rich nordmarkite (quartzsyenite) at the margin, through pulaskite, to foyaite (nephelinesyenite) in the centre; modal layering and igneous laminationare locally developed but there are no internal intrusive contacts.This is an apparent violation of the phase relations in Petrogeny'sResidua System. We propose that this intrusion is layered, gradingfrom quartz syenite at the bottom to nepheline syenite at thetop. Mineral and whole-rock major and trace element data andSr–Nd–Hf–Pb isotope data are presented thatprovide constraints on the petrogenesis of the intrusion. Radiogenicisotope data indicate a continuously decreasing crustal componentfrom the quartz nordmarkites (⁸⁷Sr/⁸⁶Sr = 0·7061; _Ndi= 2·3; _Hfi = 5·2; ²⁰⁶Pb/²⁰⁴Pb_meas = 16·98)to the foyaites (⁸⁷Sr/⁸⁶Sr = 0·7043–0·7044;_Ndi = 3·8–4·9; _Hfi = 10·7–11·1;²⁰⁶Pb/²⁰⁴Pb_meas = 17·78–17·88); the foyaitesare dominated by a mantle isotopic signature. The average Mg-numberof amphibole cores becomes increasingly primitive, varying from26·4 in the nordmarkites to 57·4 in the pulaskites.Modal layering, feldspar lamination and the presence of hugebasaltic xenoliths derived from the chamber roof, now restingon the transient chamber floor, demonstrate bottom-upwards crystallization.The intrusion cannot, therefore, have formed in a system closedto magmatic recharge. The lack of gneissic xenoliths in thenordmarkites suggests that most contamination took place deeperin the crust. In the proposed model, the nordmarkitic magmaformed during crustal assimilation in the roof zone of a large,silica-undersaturated alkali basaltic/basanitic, stratifiedmagma chamber, prior to emplacement in the uppermost crust.The more primitive syenites, terminating with foyaite at thetop of the intrusion, formed as a consequence of repeated rechargeof the Kangerlussuaq Intrusion magma chamber by tapping lesscontaminated, more primitive phonolitic melt from deeper partsof the underlying chamber during progressive armouring of theplumbing system. KEY WORDS: Kangerlussuaq; East Greenland; syenite; crustal contamination; magma mixing     

A Melt Inclusion Record of Volatiles, Trace Elements and Li-B Isotope Variations in a Single Magma System from the Plat Pays Volcanic Complex, Dominica, Lesser Antilles

Glass inclusions in plagioclase and orthopyroxene from daciticpumice of the Cabrits Dome, Plat Pays Volcanic Complex in southernDominica reveal a complexity of element behavior and Li–Bisotope variations in a single volcanic center that would gounnoticed in a whole-rock study. Inclusions and matrix glassesare high-silica rhyolite with compositions consistent with about50% fractional crystallization of the observed phenocrysts.Estimated crystallization conditions are 760–880°C,200 MPa and oxygen fugacity of FMQ + 1 to +2 log units (whereFMQ is the fayalite–magnetite–quartz buffer). Manyinclusion glasses are volatile-rich (up to 6 wt % H₂O and 2900ppm Cl), but contents range down to 1 wt % H₂O and 2000 ppmCl as a result of shallow-level degassing. Sulfur contents arelow throughout, with <350 ppm S. The trace element compositionof inclusion glasses shows enrichment in light rare earth elements(LREE; (La/Sm)_n = 2·5–6·6) and elevatedBa, Th and K contents compared with whole rocks and similaror lower Nb and heavy REE (HREE; (Gd/Yb)_n = 0·5–1·0).Lithium and boron concentrations and isotope ratios in meltinclusions are highly variable (20–60 ppm Li with ⁷Li= +4 to +15 ± 2; 60–100 ppm B with ¹¹B = +6 to+13 ± 2) and imply trapping of isotopically heterogeneous,hybrid melts. Multiple sources and processes are required toexplain these features. The mid-ocean ridge basalt (MORB)-likeHREE, Nb and Y signature reflects the parental magma(s) derivedfrom the mantle wedge. Positive Ba/Nb, B/Nb and Th/Nb correlationsin inclusion glasses indicate coupled enrichment in stronglyfluid-mobile (Ba, B) and less-mobile (Th, Nb) trace elements,which can be explained by fractional crystallization of plagioclase,orthopyroxene and Fe–Ti oxides. The ⁷Li and ¹¹B valuesare at the high end of known ranges for other island arc magmas.We attribute the high values to a ¹¹B and ⁷Li-enriched slabcomponent derived from sea-floor-altered oceanic crust and possiblyfurther enriched in heavy isotopes by dehydration fractionation.The heterogeneity of isotope ratios in the evolved, trappedmelts is attributed to shallow-level assimilation of older volcanicrocks of the Plat Pays Volcanic Complex. KEY WORDS: subduction; volcanic arcs; igneous processes; melt inclusions; SIMS; trace elements; lithium and boron isotopes; diffusion     

Petrogenesis of the Swaziland and Northern Natal Rhyolites of the Lebombo Rifted Volcanic Margin, South East Africa

The Jozini and Mbuluzi rhyolites and Oribi Beds of the southernLebombo Monocline, southeastern Africa, have geochemical characteristicsthat indicate they were derived by partial melting of a mixtureof high-Ti/Zr and low-Ti/Zr Sabie River Basalt Formation types.Compositional variations within the different rhyolite typescan largely be explained by subsequent fractional crystallization.The Sr- and Nd-isotope composition of the rhyolites is uniqueamongst Gondwana silicic large igneous provinces, having _Ndvalues close to Bulk Earth (–0·94 to 0·35)and low, but more variable, initial ⁸⁷Sr/⁸⁶Sr ratios (0·7034–0·7080).Quartz phenocryst ¹⁸O values indicate that the rhyolite magmashad ¹⁸O values between 5·3 and 6·7, consistentwith derivation from a basaltic protolith with ¹⁸O values between4·8 and 6·2. The low-¹⁸O rhyolites (< 6·0)come from the same stratigraphic horizon and are overlain andunderlain by rhyolites with more ‘normal’ ¹⁸O magmavalues. These low-¹⁸O rhyolites cannot have been produced byfractional crystallization or partial melting of mantle-derivedbasaltic material. The rhyolites have low water contents, makingit unlikely that the low ¹⁸O values are the result of post-emplacementalteration. Modification of the source by fluid–rock interactionat elevated temperatures is the most plausible mechanism forlowering the ¹⁸O magma value. It is proposed that the low-¹⁸Orhyolites were derived by melting of earlier altered rhyolitein calderas situated to the east, which were not preserved afterGondwana break-up. KEY WORDS: rhyolite; Lebombo; stable and radiogenic isotopes; low-¹⁸O magmas; partial melting     

An Oxygen and Hydrogen Isotope Study of the Skaergaard Intrusion and its Country Rocks: a Description of a 55 M.Y. Old Fossil Hydrothermal System

Oxygen isotope analyses have been obtained on rocks and coexistingminerals, principally plagioclase and clinopyroxene, from about400 samples of the Skaergaard layered gabbro intrusion and itscountry rocks. The ¹⁸O values of plagioclase decrease upwardin the intrusion, from ‘normal’ values of about+6.0 to +6.4 in the Lower Zone and parts of the Middle Zone,to values as low as –2.4 in the Upper Border Group. The¹⁸O depletions of the plagioclase all took place under subsolidusconditions, and were produced by the Eocene meteoric-hydrothermalsystem established by this pluton. Clinopyroxene, which is moreresistant to ¹⁸O exchange than is plagioclase, also underwentdepletion in ¹⁸O, but to a lesser degree (¹⁸O = +5.2 to +3.5).The ¹⁸O-depleted rocks typically show reversed ¹⁸Oplag–pxfractionations, except at the top of the Upper Zone, where thepyroxenes are very fine-grained aggregates pseudomorphous afterferrowollastonite; these inverted pyroxenes were much more susceptibleto subsolidus ¹⁸O exchange (¹⁸O = +3–9 to +0.7). D/H analysesof the chloritized basalt country rocks and of the minor quantitiesof alteration minerals in the pluton (D = –116 to –149)confirm these interpretations, indicating that the rocks interactedwith meteoric groundwaters having an original D –100.and ¹⁸O –14. Low D values ( –125) were also foundthroughout the biotites of the Precambrian basement gneiss,requiring that small amounts of water penetrated downward todepths of at least 6 to 10 km. These values, together with thelack of ¹⁸O depletion of the gneiss, imply that the overallwater/rock ratios were very small in that unit (<0.01), andthus that convective circulation of these waters was much morevigorous in the overlying highly jointed plateau basalts (¹⁸O –4.0 to +4–0) than in the relatively impermeablegneiss (¹⁸O +7–3 to +7–7). This contrast in permeabilitiesof the country rocks is also reflected in the distribution of¹⁸O values in the pluton; the plagioclases with ‘normal’¹⁸O values all lie stratigraphically beneath the projectionof the basalt-gneiss unconformity through the pluton. Elsewhere,the ¹⁸O depletions are correlated with abundance of fracturesand faults, particularly in the NE portion of the intrusion,where the Layered Series is very shallow-dipping and permeablebasalts underlie the gabbro. The transgressive granophyres in the lower part of the intrusivehave ¹⁸O values identical to those of the basement gneiss, indicatingthey were probably formed by partial melting of stoped blocksof gneiss. In the upper part of the intrusion these granophyredikes have ¹⁸O values similar to the adjacent host gabbro; thissuggests that much of the hydrothermal alteration occurred aftertheir emplacement. However, because of the rarity of low-temperaturehydrous alteration minerals, it is also clear that most of theinflux of H₂O into the layered gabbro occurred at very hightemperatures (>400–500 °C). Prior to flowing intothe gabbro, these fluids had exchanged with similar mineralassemblages in the basaltic country rocks, explaining the lackof chemical alteration of the gabbro. Xenoliths of roof rockbasalt and of Upper Border Group leucogabbro were strongly depletedin ¹⁸O by the hydrothermal system prior to their falling tothe bottom of the magma chamber and being incorporated in thelayered series. This proves that the hydrothermal system wasestablished very early, at the time of emplacement of the Skaergaardintrusion. However, no measurable ¹⁸O depletion of the gabbromagma could be detected, indicating that very little H₂O penetrateddirectly into the liquid magma, in spite of the fact that ahydrothermal circulation system totally enveloped the magmachamber for at least 100, 000 years during its entire periodof crystallization. Only as crystallization proceeded was thehydrothermal system able to collapse inward and interact withthe solidified and fractured portions of the gabbro. Neverthelesssome H₂O was clearly added directly to the magma by dehydrationof the stoped blocks of altered roof rock. It is also plausiblethat small amounts of meteoric water diffused directly intothe magma, most logically in the vicinity of major fracturezones that penetrated close to, or were underneath, the late-stagesheet of differentiated ferrodiorite magma. It is suggestedthat such influx of meteoric waters was responsible for manyof the gabbro pegmatite bodies that are common in the MarginalBorder Group; also, such H₂O might have produced local increasesin Fe⁺³/Fe⁺² in the magma that in turn could explain some ofthe asymmetric crystallization effects in the magma chamber.Local lowering of the liquidus temperature would also occur,perhaps leading to topographic irregularities on the floor ofthe magma chamber (e.g. the trough bands?).     

Isotopic and Geochemical Investigation of the Chilwa Island Carbonatite Complex, Malawi: Evidence for a Depleted Mantle Source Region, Liquid Immiscibility, and Open-System Behaviour

Initial Nd, Pb, and Sr isotopic data from carbonatites and associatedintrusive silica-undersaturated rocks from the early Jurassic,Chilwa Island complex, located in southern Malawi, central Africa,suggest melt derivation from a Rb/Sr- and Nd/Sm-depleted butTh/Pb- and U/Pb-enriched mantle source. Initial ¹⁴³Nd/¹⁴⁴Nd(0.51265–0.51270) isotope ratios from the Chilwa Islandcarbonatites are relatively constant, but their initial ⁸⁷Sr/⁸⁶Sr(0.70319–0.70361) ratios are variable. The ¹⁸O_smow (9.53–14.15%0)and ¹³C_PDB (–3.27 to –1.50%0) isotope ratios ofthe carbonates are enriched relative to the range of mantlevalues, and there is a negative correlation between ¹⁸O andSr isotope ratios. The variations in Sr, C, and O isotopic ratiosfrom the carbonatites suggest secondary processes, such as interactionwith meteoric groundwater during late-stage carbonatite activity.The initial ¹⁴³Nd/¹⁴⁴Nd (0.51246 0.51269) and initial ⁸⁷Sr/⁸⁶Sr(0.70344–0.70383) isotope ratios from the intrusive silicaterocks are more variable, and the Sr more radiogenic than thosefrom the carbonatites. Most of the Pb isotope data from Chilwa Island plot to the rightof the geochron and close to the oceanic regression line definedby MORBs and OIBs. Initial Pb isotopic ratios from both carbonatites(²⁰⁷Pb/²⁰⁴Pb 15.63–15.71; ²⁰⁶Pb/²⁰⁴Pb 19.13–19.78)and silicate rocks (²⁰⁷Pb/²⁰⁴Pb 15.61–15.72; ²⁰⁶Pb/²⁰⁴Pb18.18–20.12) show pronounced variations, and form twogroups in Pb-Pb plots. The isotopic variations shown by Nd, Pb, and Sr for the ChilwaIsland carbonatites and intrusive silicates suggest that thesemelts underwent different evolutionary histories. The chemicaldata, including isotopic ratios, from the carbonatites and olivinenephelinites are consistent with magmatic differentiation ofa carbonated-nephelinite magma. A model is proposed in whichdifferentiation of the carbonatite magma was accompanied byfenitization (metasomatic alteration) of the country rocks bycarbonatite-derived fluids, and subsequent alteration of thecarbonatite by hydrothermal activity. The chemical and isotopicdata from the non-nephelinitic intrusive silicate rocks reveala more complex evolutionary history, involving either selectivebinary mixing of lower-crustal granulites and a nephelinitemagma, or incremental batch melting of a depleted source andsubsequent crustal contamination.     

Geochemical Constraints on the Petrogenesis of Diamond Facies Pyroxenites from the Beni Bousera Peridotite Massif, North Morocco

The petrogenesis of pyroxenite layers within the Beni Bouseraperidotite massif is investigated by means of elemental andNd-Sr-Pb-O-S isotope analyses. The light rare earth element(LREE) depleted nature of many of the pyroxenites, their widevariation in composition, and lack of correlation between incompatibleelements and fractionation indices preclude them from representingcrystallized melts from a peridotitic source. The physical characteristicsof the pyroxenites and their large (greater than a factor of20) range in Ni rule out partial melting as the cause of theirpetrological and geochemical diversity. Major and compatibletrace element geochemistry is consistent with formation of mostof the pyroxenite suite via high-pressure crystal segregationin magma conduits intruding the peridotites. These magmas crystallizedclinopyroxene, orthopyroxene, and garnet. The pressure of crystallizationis constrained to be above {small tilde}45 kbar from the presenceof graphitized diamonds in pyroxenite layers. Lack of correlationbetween fractionation indices and highly incompatible elementsand the wide variation in incompatible element abundances suggestthat the suite did not form from genetically related magmas.The presence of positive and negative Eu anomalies (Eu/Eu* =0•54–2•0) in pyroxenites which crystallizedat pressures much greater than the plagioclase stability field({small tilde} 45 kbar) suggests that the parental magmas originatedfrom precursors which formed in the crust. Oxygen isotope compositionsof coexisting minerals in the pyroxenites indicate high-temperatureequilibration but ¹⁸O values vary from +4•9 to + 9•3,ruling out their derivation from the host peridotites or othernormal mantle sources. The extreme O-isotope variation, togetherwith ³⁴S values of up to + 13 in sulphides included within CPXstrongly suggests that the melts from which the pyroxenitescrystallized were derived from hydrothermally altered, subductedoceanic lithosphere. Extreme initial radiogenic isotope variationin the pyroxenites (_Nd + 26 to –9 , ⁸⁷Sr/⁸⁶Sr 0•7025–0•7110,²⁰⁶Pb/²⁰⁴Pb 18•21–19•90) support such an originbut also require a component with ancient, high U/Pb and Th/Pbin their source to explain the high 7/4 and 8/4 values of somepyroxenites. This component may be subducted hemi-pelagic sediment.Further evidence for a sediment component in the pyroxenitesis provided by isotopically light carbon in the graphite pyroxenites(¹³C–16 to – 28). Parentdaughter isotopes in thepyroxenites are strongly decoupled, making estimation of formationages speculative. The decoupling occurred recently (<200Ma), probably as a result of partial melting associated withdiapiric upwelling and emplacement of the massif into the crustfrom the diamond stability field. This late partial meltingevent further depleted the pyroxenites in incompatible elements.The variably altered nature of the subducted protolith and complexhistory of trace element fractionation of the pyroxenites haslargely obscured geochemical mixing trends. However, Nd–Pbisotope systematics indicate that incorporation of the componentwith high U/Pb–Th/Pb occurred relatively recently (<200Ma) for some pyroxenites. Other pyroxenites do not show evidencefor incorporation of such a component and may be substantiallyolder. Tectonic, geophysical, and isotopic constraints indicateformation of the pyroxenites in the mantle wedge above a subductingslab during the Cretaceous. Physical and chemical evidence forhigh-pressure fractionation seen in most of the pyroxenitesprecludes them from simply representing ancient subducted oceaniclithosphere, thinned by diffusion. However, the petrologicaland isotopic diversity of the massif support the concept ofa ‘marble cake’ mantle capable of producing theobserved geochemical diversity seen in oceanic magmas. *Present Address: Department of Terrestrial Magnetism, 5241 Broad Branch Road, N.W., Washington, DC 20015 Present address: Department of Geological Sciences, 1066 C.C. Little Building, University of Michigan, Ann Arbor, Michigan 48109     

Fluid--Rock Interaction during Low-Pressure Polymetamorphism of the Reynolds Range Group, Central Australia

Marbles and metapelites from the Reynolds Range Group (centralAustralia) were regionally metamorphosed at low pressure duringM₂ at 1.6 Ga, M₂ ranged in grade from greenschist to granulitefacies along the length of the Reynolds Range, and overprinted1.78 Ga granites and their contact aureoles in the ReynoldsRange Group metasediments. At all M₂ grades the marbles andmetapelites have highly variable oxygen isotope ratios [marbles:¹⁸O(carb) 14–20%; metapelites: ¹⁸O 6–14%). Similarly, 1.78 Ga granites have highly variable oxygen isotope ratios(¹⁸O 5–13%), with the lowest values occurring at thegranite margins. In all rock types, the lowest oxygen isotopevalues are consistent with the infiltration of channelled magmaticand/or meteoric fluids. The variable lowering of oxygen isotopevalues resulted from pre-M₂ contact metamorphism and fluid—rockinteraction around the 1.78 Ga granites. In contrast, mineralassemblages in the marbles define a trend of increasing X_CO2with increasing grade from <0.05 (greenschist facies) to0.7–1.0 (granulite facies). This, together with the lackof regionally systematic resetting of oxygen isotope ratios,implies that there was little fluid—rock interaction duringprograde regional metamorphism. KEY WORDS: low pressure; polymetamorphism; fluids; stable isotopes; petrology ^*Corresponding author Fax: 61–3–94791272. e-mail: geoisb{at}lure.latrobe.edu.au     

Geochemical Constraints on the Origin of the Late Archean Skjoldungen Alkaline Igneous Province, SE Greenland

The present work reports the first broad geochemical investigationof the recently discovered late Archean (2700 Ma) Skjoldungenalkaline igneous province (SAP) in southeast Greenland. Therocks studied range in composition from ultramafic to felsicand comprise pyroxenites, hornblendites, hornblende noritesand diorites, monzonites, syenites, and nephelinitic rocks andcarbonatites. Various lithologic units from the host Archeangneissic basement are also investigated. The magmatic rocksshow remarkably coherent major element, trace element, rareearth element (REE), and Sr and Nd isotope systematics, suggestinga petrogenetic relationship. The most important geochemicalfeatures are high normative proportions of nepheline, forsteriteand albite, low TiO₂ (<15 wt %) and moderate FeO (total)(<12 wt %) contents, enrichments in large ion lithophileelements (LILE) and light rare earth elements both absoluteand relative to high field strength elements (HFSE) that displaylarge negative anomalies, and generally low to moderate abundancesof compatible elements. Field relations and REE and compatibleelement systematics among Skjoldungen rocks suggest that maficand ultramafic hornblende-rich samples may represent cumulatelithologies of the regional parental magma. On the basis ofmineral data, this is deduced to have had mg-number of 064,shoshonitic affinities (K₂O15 wt %), been close to silica saturationand volatile rich. Major element, trace element and REE systematicsfurther suggest that felsic intrusions are related to the maficregional parental magma through extensive olivine, hyperstheneand hornblende fractionation. Lack of correlation between La/Yband other critical trace and REE ratios indicates that apatite,zircon and titaniferous minerals were not important cumulusphases at advanced stages of evolution. The measured Sm–Ndwhole-rock isochron age is 2716 23 Ma (2 error) [mean squareof weighted deviates (MSWD) = 14], whereas linear regressionof the Sr isotope data yields an age of 26047 Ma (2 error)(MSWD = 22•2). The age obtained by Nd isotopes is corroboratedby U–Pb zircon results (2698 7 Ma), suggesting thatthe Sm–Nd system remained closed since crystallization.By contrast, the 100 Ma younger age obtained by Sr isotopessuggests that the Rb–Sr system has been disturbed. Initial¹⁴³Nd/¹⁴⁴ Nd ratios span a narrow range corresponding to _Nd(27Ga) =+074 to –109, whereas initial _Sr values at 27Ga cover a comparatively larger interval from –10 to +20.Neither initial _Nd nor initial _Sr values conform to previouslysuggested mantle depletion curves and no meaningful correlationexists between Nd and Sr isotopes for the Skjoldungen magmaticrocks as a whole. Although compositionally heterogeneous, theanalyzed suite of samples from the host agmatitic basement isextremely homogeneous with respect to age, with T^CHUR crustalresidence times around 2700–2800 Ma confirming limitedavailable isotopic evidence. Large-scale assimilation of Archeancrust or recycling of sediments derived from the local basementinto the mantle source fails to explain adequately negativeNb anomalies and low ^Nd signatures characteristic of the Skjoldungenintrusions. Rather, the nearchondritic isotopic compositionof Nd in the Skjoldungen samples together with the decoupledLILE and HFSE enrichment and slightly positive _Sr values areconsidered to reflect characteristics of the mantle source ina subduction zone environment. The geodynamic site hosting theSkjoldungen province thus may be an early manifestation of modern-styleplate tectonics. KEY WORDS: Skjoldungen province; Greenland; Archean; alkaline igneous rocks; geochronology; geochemistry ^*Corresponding author. Present address: Ecole Normale Suprieure de Lyon, 46 AlLe d'Italie, 69364 Lyon Cedex 07, France     

Magmatic Differentiation at an Island-arc Caldera: Okmok Volcano, Aleutian Islands, Alaska

Okmok volcano is situated on oceanic crust in the central Aleutianarc and experienced large (15 km³) caldera-forming eruptionsat 12 000 years BP and 2050 years BP. Each caldera-forming eruptionbegan with a small Plinian rhyodacite event followed by theemplacement of a dominantly andesitic ash-flow unit, whereaseffusive inter- and post-caldera lavas have been more basaltic.Phenocryst assemblages are composed of olivine + pyroxene +plagioclase ± Fe–Ti oxides and indicate crystallizationat 1000–1100°C at 0·1–0·2 GPain the presence of 0–4% H₂O. The erupted products followa tholeiitic evolutionary trend and calculated liquid compositionsrange from 52 to 68 wt % SiO₂ with 0·8–3·3wt % K₂O. Major and trace element models suggest that the moreevolved magmas were produced by 50–60% in situ fractionalcrystallization around the margins of the shallow magma chamber.Oxygen and strontium isotope data (¹⁸O 4·4–4·9,⁸⁷Sr/ ⁸⁶Sr 0·7032–0·7034) indicate interactionwith a hydrothermally altered crustal component, which led toelevated thorium isotope ratios in some caldera-forming magmas.This compromises the use of uranium–thorium disequilibria[(²³⁰Th/ ²³⁸U) = 0·849–0·964] to constrainthe time scales of magma differentiation but instead suggeststhat the age of the hydrothermal system is 100 ka. Modellingof the diffusion of strontium in plagioclase indicates thatmany evolved crystal rims formed less than 200 years prior toeruption. This addition of rim material probably reflects theremobilization of crystals from the chamber margins followingreplenishment. Basaltic recharge led to the expansion of themagma chamber, which was responsible for the most recent caldera-formingevent. KEY WORDS: Okmok; caldera; U-series isotopes; Sr-diffusion; time scales; Aleutian arc     

The 1982-83 Eruption at Galunggung Volcano, Java (Indonesia): Oxygen Isotope Geochemistry of a Chemically Zoned Magma Chamber

Mount Galunggung is a historically active volcano in southwesternJava that has erupted four times in the last two centuries.During the most recent event, which occurred during a 9–monthinterval in 1982– 83, some 305 10⁶ m³ of medium–K,calc–alkaline magma was erupted. This eruption was unusualbecause of its duration, the diversity of eruption dynamicsand products, and the range of lava compositions produced. Thecomposition of juvenile material changed gradually during thecourse of the eruption from initial plagioclase (An_60–75)and two–pyrozene bearing andesites with 58% SiO₂ to finalplagioclase (An_85–90), diopside, and olivine (Fo_85–90)bearing primitive magnesium basalts with 47% SiO₂ Mineralogicaland compositional relationships indicate a magmatic evolutioninvolving differentitation of high–Mg parental melt. Theeruptive volumes of 35 10⁶ m³ andesite, 120 10⁶ m³ maficandesite, and 150 10⁶ m³ basalt are consistent with the ideathat the 1982– 83 eruption progressively tapped and draineda magma chamber that had become chemically stratified throughextensive crystal fractionation. Separates of plagioclase and pyroxene have ¹⁸O( SMO W) rangesof + 5. 6 to + 6.0 and + 5.3 to + 5.6, respectively, with ¹⁸O_plag–pxvalues of + 0.4 to + 0.6o, indicating internal O–isotopeequiliburium at temperature of 1100–850 C. The magenesianbasalts have magmatic ¹⁸O/ ¹⁶O ratios similar to those of mid–oceanridge basalt, and the O–isotope ratios of compositionallyevolved derivative melts show no evidence for contaminationof the galunggung magmas by ¹⁸O–rich crust during differentiation.Andesites and transitional mafic and sites have a more variableO–isotope character, with laves and phenocrysts havingboth higher and lower ¹⁸O values than observed in the parentalmagnesium basalts. These features are interpreted to reflectintramagma chamber processes affecting the upper portions ofthe differentiating Galunggung magma body before the 1982–83eruption.     

A Strontium and Neodymium Isotopic Investigation of the Fongen--Hyllingen Layered Intrusion, Norway

The Fongen—Hyllingen Intrusion, situated 60 km SE of Trondheim,Norway, is a synorogenic layered mafic intrusion of Caledonianage . The intrusion is divided into four evolutionary stages based on cryptic variations: StageI—a basal reversal; Stage II—unchanged mineral chemistryor slight normal evolution; Stage III—a gradual regression;Stage IV— a strong normal fractionation trend Magma replenishmentdominated during most of the crystallization, i.e. during StagesI, II and III Replenishing magma was more dense than resident,evolved magma, and continuing influx eventually caused a compositionallystratified magma column to form. Cryptic lateral variation isan important feature in the southern part of the complex andformed by in situ crystallization from a stratified magma alongan inclined floor, where modal layering formed parallel to thecrystallization front. Initial Sr- and Nd-isotopic ratios inthe cumulates vary as a result of assimilation of country rockand subsequent mixing between uncontaminated, replenishing magmaand contaminated, resident magma. The parental magma had a moderatelydepleted isotope composition, relative to Bulk Earth, with _Nd=584and Sr_i=070308, whereas the main contaminant was a partialmelt of metapelitic country rock with _Nd=-874 and Sr_i=07195(Sr_i is the initial ⁸⁷Sr/⁸⁶Sr). Sr_i in the analysed cumulatewhole-rock samples ranges from 070308 to 070535 and initial_Nd ranges from. 158 to 584. There is a strong correlationbetween mineralogical composition and isotopic trends in mostof the cumulates: the most primitive samples are the least contaminated,as reflected by relatively high ed and low Sr,, and more evolvedsamples have progressively lower eNi and higher Sry A gradualregression of several hundred metres thickness characterizesStage III; stratigraphically upwards mineral compositions becomemore primitive and isotope compositions more depleted (higher_Nd and lower Sr_i), implying a process of. progressive mixing-inof replenishing, primitive and uncontaminated magma. Magma influxin Stage III took place by fountaining, whereas magma additionwas more tranquil in the earlier stages. The fountaining influxentrained resident, relatively evolved and contaminated magma,resulting in a hybrid magma which ponded at the floor. Duringprolonged magma addition with concomitant crystallization, thelowermost magma layer was replaced by progressively more primitivehybrid magma, creating a gradual regression in the crystallizingcumulate sequence. A detailed two-dimensional study revealslateral variations in mineral compositions both at the baseand top of Stage III, whereas lateral variations in Sr- andNd-isotopic compositions are present at the top, but not atthe base. This implies that the lowest crystallizing part ofthe magma column was essentially isotopically homogeneous, butcompositionally stratified, before influx in Stage III. Isotopicgradients in the magma were strong close to the roof, wheremost of the assimilation occurred, and decreased downwards,merging into isotopically homogeneous magma. This stratifiedsystem was destroyed by turbulent mixing between replenishingand resident magma during fountaining influx in Stage III, anda new stratification was established with both an isotopic anda compositional gradient. After the final influx, crystallizationcontinued in an essentially closed system, in which the remainingmagma column eventually became homogenized, as magma layersmixed when their densities converged owing to release of buoyant,residual liquid during fractional crystallization. Corresponding author