Melting of a Hydrous Mantle: I. Phase Relations of Natural Peridotite at High Pressures and Temperatures with Controlled Activities of Water, Carbon Dioxide, and Hydrogen

Four natural peridotite nodules ranging from chemically depletedto Fe-rich, alkaline and calcic (SiO₂=43?7–45?7 wt. percent, Al₂O3=1?6O–8?21 wt. per cent, CaO=0?70–8?12wt. per cent,alk=0?10–0?90 wt. per cent and Mg/(Mg+Fe²⁺)=0?94–0?85)have been investigated in the hypersolidus region from 800?to 1250?C with variable activities of H₂O, CO₂, and H₂. Thevapor-saturated peridotite solidi are 50–200?C below thosepreviously published. The temperature of the beginning of meltingof peridotite decreases markedly with decreasing Mg/(Mg+Fe)of the starting material at constant CaO/Al₂O₃. Conversely,lowering CaO/Al₂O₃ reduces the temperature at constant Mg/(Mg+Fe)of the starting material. Temperature differences between thesolidi up to 200?C are observed. All solidi display a temperatureminimum reflecting the appearance of garnet. This minimum shiftsto lower pressure with decreasing Mg/(Mg+Fe) of the startingmaterial. The temperature of the beginning of melting decreasesisobarically as approximately a linear function of the mol fractionof H₂O in the vapor (X_H2O). The data also show that some CO₂may dissolve in silicate melts formed by partial melting ofperidotite. Amphibole (pargasitic hornblende) is a hypersolidus mineralin all compositions, although its P/T stability field dependson bulk rock chemistry. The upper pressure stability of amphiboleis marked by the appearance of garnet. The vapor-saturated (H₂O) liquidus curve for one peridotiteis between 1250? and 1300?C between 10 and 30 kb. Olivine, spinel,and orthopyroxene are either liquidus phases or coexist immediatelybelow the temperature of the peridotite liquidus. The data suggest considerable mineralogical heterogeneity inthe oceanic upper mantle because the oceanic geotherm passesthrough the P/T band covering the appearance of garnet in variousperidotites. The variable depth to the low-velocity zone is explained byvariable a_H2O conditions in the upper mantle and possibly alsoby variations in the composition of the peridotite itself. It is suggested that komatiite in Precambrian terrane couldform by direct melting of hydrous peridotite. Such melting requiresabout 1250?C compared with 1600?C which is required for drymelting. The genesis of kimberlite can be related to partial meltingof peridotite under conditions of (). Such activities of H₂Oresult in melting at depths ranging between 125 and 175 km inthe mantle. This range is within the minimum depth generallyaccepted for the formation of kimberlite.     

Melting of a Hydrous Mantle: II. Geochemistry of Crystals and Liquids Formed by Anatexis of Mantle Peridotite at High Pressures and High Temperatures as a Function of Controlled Activities of Water, Hydrogen, and Carbon Dioxide

The system peridotite-H₂O–CO₂ serves as a simplified modelfor the phase relations of mantle peridotite involving morethan one volatile component. Run products obtained in a studyof phase relations of four mantle peridotites in the presenceof H₂O- and (H₂O+CO₂)- bearing vapors and with controlled hydrogenfugacity (f_H2) at high pressures and temperatures have beensubjected to a detailed chemical investigation, principallyby the electron microprobe. Mg/(Mg+Fe) of all phases generally increases with increasingtemperature and with increasing Mg/(Mg+Fe) of the starting material.This ratio appears to decrease with increasing pressure forolivine, and for amphibole coexisting with garnet. Decreasingf_H2 from that of IW buffer to that of MH buffer decreases Mg/(Mg+Fe)of the partial melt from approximately 0-85 to approximately0.50, whereas the Fo content of coexisting olivine increasesslightly less than 3 per cent and the Mg/(Mg+Fe) of clinopyroxeneincreases about 4 per cent. However, the variations in Fo contentof olivines are within those observed in olivines from naturalmantle peridotite. The chemistry of other silicate mineralsdoes not significantly reflect variations of f_H2. Consequently,the peridotite mineralogy and/or chemistry is not a good indicatorfor the f_H2 conditions during crystallization. All crystalline phases, except amphibole, and to some extentgarnet, show increasing Cr content with increasing temperatureand increasing Cr content of the starting material, resultingin a positive correlation with Mg/(Mg+Fe). Partial melts aredepleted in Cr₂O₃ relative to the crystalline phases. High Mg/Mg+Fe)and Cr₂O₃ are thus expected in crystal residues after partialmelting. The absolute values depend on degree of melting andthe composition of the parent peridotite. Liquids formed by anatexis of mantle peridotite are andesiticunder conditions of X_H2O^v > 0.6 to at least 25 kb total pressureand to more than 200?C above the peridotite solidus. This observationsupports numerous suggestions that andesite genesis in islandarcs may result from partial melting of underlying peridotitemantle. In contrast to basaltic rocks, the absence of amphibole(paragasitic hornblende) does not affect the silica-saturatednature of the liquids. Increasing K₂O content of the startingmaterial (up to 1 wt. per cent K₂O) results in increasing potassiumcontent of the amphibole (1 wt. per cent K₂O) as well as theappearance of phlogopite. The liquid under these conditionsis relatively K₂0-poor (less than 1 wt. per cent K₂O). Partial melts are olivine normative with X_H2O 0.5, and initialliquids contain normative ol and ne at X_H2O 0.4. The alkalinityof these liquids increases with decreasing X_H2O below valuesof 0.5. The (ol+opx)-normative liquids resemble oceanic basaltswhereas (ol+ne)-normative liquids resemble olivine nepheliniteand melilite basalt. Low aHlo and high a_Co2 conditions may bethose under which kimberlites and related rocks are formed inthe mantle.     

Melting of a Hydrous Mantle: II. Geochemistry of Crystals and Liquids Formed by Anatexis of Mantle Peridotite at High Pressures and High Temperatures as a Function of Controlled Activities of Water, Hydrogen, and Carbon Dioxide

The system peridotite-H₂O-CO₂ serves as a simplified model forthe phase relations of mantle peridotite involving more thanone volatile component. Run products obtained in a study ofphase relations of four mantle peridotites in the presence ofH₂O- and (H₂O+CO₂)-bearing vapors and with controlled hydrogenfugacity (f_H2) at high pressures and temperatures have beensubjected to a detailed chemical investigation, principallyby the electron microprobe. Mg/(Mg+Fe) of all phases generally increases with increasingtemperature and with increasing Mg/(Mg+Fe) of the starting material.This ratio appears to decrease with increasing pressure forolivine, and for amphibole coexisting with garnet. DecreasingfH₂from that of IW buffer to that of MH buffer decreases Mg/(Mg+Fe)of the partial melt from approximately 0?85 to approximately0?50, whereas the Fo content of coexisting olivine increasesslightly less than 3 per cent and the Mg/(Mg+Fe) of clinopyroxeneincreases about 4 per cent. However, the variations in Fo contentof olivines are within those observed in olivines from naturalmantle peridotite. The chemistry of other silicate mineralsdoes not significantly reflect variations of fH₂. Consequently,the peridotite mineralogy and/or chemistry is not a good indicatorfor the fH₂ conditions during crystallization. All crystalline phases, except amphibole, and to some extentgarnet, show increasing Cr content with increasing temperatureand increasing Cr content of the starting material, resultingin a positive correlation with Mg/(Mg+Fe). Partial melts aredepleted in Cr₂O₃ relative to the crystalline phases. High Mg/(Mg+Fe)and Cr₂O₃ are thus expected in crystal residues after partialmelting. The absolute values depend on degree of melting andthe composition of the parent peridotite.     

Chemistry of Kornerupine and Associated Minerals, a Wet Chemical, Ion Microprobe, and X-Ray Study Emphasizing Li, Be, B and F Contents

Kornerupine and associated minerals in 31 samples of high-graderocks relatively rich in Al and Mg were analysed by wet chemistry,ion microprobe mass analyser, electron microprobe and X-raypowder diffraction. For 11 samples of kornerupine and threesamples of biotite (F only) analysed by both wet chemical andion microprobe methods, the best agreement was obtained forB₂O₃, whereas the ion microprobe Li₂O values were systematicallysomewhat higher than the wet chemical values. The wet chemicalmethods give Li₂O=0–0?19 wt.%; BeO=0–0?032 wt.%;B₂O₃=0–4?01 wt.%; and F=0?07–0?77 wt.% in kornerupine,whereas ion microprobe analyses on other kornerupines give valuesup to 0?35 wt.% Li₂O, O066 wt.% BeO, and 4?72 wt.% B₂O₃. Thesum B+Al+Fe³⁺+Cr is close to 6?9 atoms per 22 (O, OH, F) or21?5 (O) in kornerupine. In general, Li/Fe ratios decrease as follows: kornerupine ?sapphirinebiotite> Crd (Na<0?03 per 18 oxygens)>tourmaline, garnet,orthopyroxene. However, for cordierite with Na>004, Li/Fedecreases as follows: cordierite>kornerupine. Sapphirineand sillimanite are the only associated minerals to incorporatesignificant boron (0?1–0?85 wt.% B₂O₃) and then only whenthe single site for B in kornerupine is approaching capacity.Sillimanite B₂O₃ contents increase regularly with kornerupineF. Fractionation of fluorine increases as follows: kornerupine<biotite<tourmaline,and K^krn-BtD=(F/OH)^Krn/(F/(OH)^Bt (assuming ideal anion composition)increases with biotite Ti. Kornerupine B₂O₃ content is a measureof B₂O₃ activity in associated metamorphic fluid, whereas sillimaniteB₂O₃ content increases with temperature, exceeding 0?4 wt.%whenT=900?C at very low water activities. New data on 11 kornerupines and literature data indicate thatthe unit cell parameters a, c, and V decrease with increasingB content and b, c, and V increase with increasing Fe³⁺ content.In Fe³⁺-poor kornerupines, b increases with Mg and with (Mg+ Fe²⁺) but the effect of Mg on b via the substitution ^VIMg+^IVSi=^VIAl+^IVAloverwhelms the effect of Fe²⁺=Mg substitution.     

The Solubility of Alumina in Orthopyroxene Coexisting with Garnet in FeO-MgO--Al2O3--SiO2 and CaO--FeO--MgO--Al2O3--SiO2

The pressure-temperature-compositional (P-T-X) dependence ofthe solubility of Al₂O₃ in orthopyroxene coexisting with garnethas been experimentally determined in the P-T range 5–30kilobars and 800–1200 ?C in the system FeO—MgO—Al₂O₃—SiO₂(FMAS). These results have been extended into the CaO—FeO—MgO—Al₂O₃—SiO₂(CFMAS) system in a further set of experiments designed to determinethe effect of the calcium content of garnet on the Al₂O₃ contentsof coexisting orthopyroxene at near-constant Mg/(Mg + Fe). Startingmaterials were mainly glasses of differing Mg/(Mg + Fe) or Ca/(Ca+ Mg + Fe) values, seeded with garnet and orthopyroxene of knowncomposition, but mineral mixes were also used to demonstratereversible equilibrium. Experiments were performed in a piston-cylinderapparatus using a talc/pyrex medium. Measured orthopyroxene and corrected garnet compositions werefitted by multiple and stepwise regression techniques to anequilibrium relation in the FMAS system, yielding best-fit,model-dependent parameters G^o_y= –5436 + 2.45T cal mol^–1,and W^M1_FeA1= –920 cal mol^–1. The volume change ofreaction, V^o, the entropy change, S^o₉₇₀ and the enthalpy changeH^o_1,970, were calculated from the MAS system data of Perkinset al. (1981) and available heat capacity data for the phases.Data from CFMAS experiments were fitted to an expanded equilibriumrelation to give an estimate of the term W^ga_CaMg = 1900 ? 400cal/mole cation, using the other parametric values already obtainedin FMAS. The experimental data allow the development of a arnet-orthopyroxenegeobarometer applicable in FMAS and CFMAS: where This geobarometer is applicable to both pelitic and metabasicgranulites containing garnet orthopyroxene, and to garnet peridoditeand garnet pyroxenite assemblages found as xenoliths in diatremesor in peridotite massifs. It is limited, however, by the necessityof an independent temperature estimate, by errors associatedwith analysis of low Al₂O₃ contents in orthopyroxenes in high-pressureor low-temperature parageneses, and by uncertainties in thecomposition of garnet in equilibrium with orthopyroxene. Ananalysis of errors associated with this formulation of the geobarometersuggests that it is subject to great uncertainty at low pressuresand for Fe-rich compositions. The results of application ofthis geobarometer to natural assemblages are presented in acompanion paper.     

Petrology of the Younger Andesites and Dacites of Iztacc?huatl Volcano, Mexico: II. Chemical Stratigraphy, Magma Mixing, and the Composition of Basaltic Magma Influx

The Younger Andesites and Dacites of Iztacc?huatl volcano, Mexico,constitute a medium-K calcalkaline rock suite (58–66 wt.per cent SiO₂) characterized by high Mg-numbers (100Mg/(Mg+0?85Fe²⁺=55–66) and relatively high abundances of MgO (2?5–6?6wt. per cent), Ni(17–158 p.p.m.), and Cr (42–224p.p.m.). Chemical stratigraphy plots of eruptive sequences indicatethe existence of a plexus of long-lived dacite magma chambersperiodically replenished by influxes of basaltic magma ascendingfrom depth. Short-term geochemical evolution after batch influxwas dictated by magma mixing and eventual dilution of the basalticcomponent by ‘quasi-steady state’ hornblende dacitemagma. The chemical data support textural and mineralogicalevidence for rapid homogenization of originally diverse magmasby convective blending of residual liquids accompanied by dynamicfractional crystallization (Nixon, 1988). Internally-consistent mixing calculations used to derive thecomposition of basaltic magma influx incorporate analyticaluncertainties and the observed range of salic end-member compositions.Mafic end-members are basalts to basaltic andesites (52–54wt. per cent SiO₂) with Mg-numbers (73–76), MgO (9–11wt. per cent), Ni (250 p.p.m.), and Cr (340–510 p.p.m.)concentrations, and liquidus olivine compositions (Fo_90–88),appropriate for unfractionated partial melts of mantle peridotite.The majority of model compositions are Ol-Hy-normative, similarto those of primitive basaltic lavas on the flanks of Iztacc?huatland in the Valley of Mexico. However, calculated magma batchesrange from weakly Qz-normative to strongly Ne-normative. Bothcalculated and analyzed basaltic compositions are distinguishedby highly variable abundances of alkalies and incompatible traceelements, notably Rb, Ba, Sr, P, Zr, and Y. Initial ⁸⁷Sr/⁸⁶Sr ratios for Iztacc?huatl lavas (0?7040–0?7046;n=24) are comparable to those for primitive basaltic rocks (0?7037–0?7045;?=4) and indicate that (1) mantle source regions are isotopicallyheterogeneous; and (2) contamination of iztacc?huatl magma chambersby radiogenic crustal rocks was not a significant factor inthe evolution of calc-alkaline andesites and dacites. The replenishment of Iztacc?huatl dacite reservoirs by Ne-normativemagmas late in the history of cone growth precludes exhaustionof mantle source regions by progressive partial melting. Thewaning stages of volcanic activity at Iztacc?huatl appear toreflect the inability of dense basaltic influxes to successfullypenetrate a large high-level chamber of low density hornblendedacite magma.     

Melting and Thermal Reconstitution of Pelitic Xenoliths, Wehr Volcano, East Eifel, West Germany

Pelitic xenoliths derived from amphibolite grade basement rocksoccur within a Pleistocene, trachytic, pyroclastic unit of theWehr volcano, East Eifel, West Germany: With increasing temperatureand/or prolonged heating at high temperature, quartz-plagioclaseand micaceous layers of the xenoliths have undergone meltingto form buchites and thermal reconstitution by dehydration reactions,melting and crystallization to form restites respectively. Thexenoliths provide detailed evidence of melting, high temperaturedecomposition of minerals, nucleation and growth of new phasesand P-T-f_o2 conditions of contact metamorphism of basement rocksby the Wehr magma. Melting begins at quartz-oligoclase (An_17·3Ab_82·3Or_0·4-An_20·0Ab_78·1Or_1·9)grain boundaries in quartz-plagioclase rich layers and the amountof melting is controlled by H₂O and alkalis released duringdehydroxylation/oxidation of associated micas. Initially, glasscompositions are heterogeneous, but with increasing degreesof melting they become more homogeneous and are similar to S-typegranitic minimum melts with SiO₂ between 71 and 77 wt. per cent;A/(CNK) ratios of 1·2–1·4; Na₂O < 2·95and normative corundum contents of 1·9–4·0per cent. Near micas plagioclase melts by preferential dissolutionof the NaAlSi₃O₈ component accompanied by a simultaneous increasein CaAl²Si²O⁸ (up to 20 mol. per cent An higher than the bulkplagioclase composition) at the melting edge. With increasingtemperature the end product of fractional melting is the formationand persistence of refractory bytownite (An_78–80) in thosexenoliths where extensive melting has taken place. Initial stage decomposition of muscovite involves dehydroxylation(H₂O and alkali loss). At higher temperatures muscovite breaksdown to mullite, sillimanite, corundum, sanidine and a peraluminousmelt. Mullite (40–43 mol. per cent SiO₂) and sillimanite(49 mol. per cent SiO₂) are Fe₂O₃ and TiO₂ rich (up to 6·1–0·84and 3·6–0·24 wt. per cent respectively).Al-rich mullite (up to 77 wt. per cent Al₂O₃) occurs with corundumwhich has high Fe₂O₃ and TiO₂ (up to 6·9 and 2·1wt. per cent respectively). Annealing at high temperatures andreducing conditions results in the exsolution of mullite fromsillimanite and ilmenite from corundum. Glass resulting fromthe melting of muscovite in the presence of quartz is peraluminous(A/(CNK) = 1·3) with SiO² contents of 66–69 percent and normative corundum of 4 per cent. Sanidine (An_1·9Ab_26·0Or_72·1-An_1·3Ab_15·9Or_82·9)crystallized from the melt. Dehydroxylation and oxidation of biotite results in a decreaseof K₂O from 8·6 to less than 1 wt. per cent and oxidetotals (less H₂O + contents) from 96·5 to 88·6,exsolution of Al-magnetite, and a decrease in the Fe/(Fe + Mg)ratio from 0·41 to 0·17. Partial melting of biotitein the presence of quartz/plagioclase to pleonaste, Al-Ti magnetite,sanidine(An_2·0Ab_34·9Or63·1) and melt takesplace at higher temperatures. Glass in the vicinity of meltedbiotite is pale brown and highly peraluminous (A/CNK = 2·1)with up to 6 wt. per cent MgO+FeO_{(total iroq)} and up to 10 percent normative corundum. Near liquidus biotite with higher Al₂O₃and TiO₂ than partially melted biotite crystallized from themelt. Ti-rich biotites (up to 6 wt. per cent TiO₂) occur withinthe restite layers of thermally reconstituted xenoliths. Meltingof Ti-rich biotite and sillimanite in contact with the siliceousmelt of the buchite parts of xenoliths resulted in the formationof cordierite (100 Mg/(Mg+Fe+Mn) = 76·5–69·4),Al-Ti magnetite and sanidine, and development of cordierite/quartzintergrowths into the buchite melt. Growth of sanidine enclosedrelic Ca-plagioclase to form patchy intergrowths in the restitelayers. Cordierite (100 Mg/(Mg+Fe+Mn) = 64–69), quartz,sillimanite, mullite, magnetite and ilmenite, crystallized fromthe peraluminous buchite melt. Green-brown spinels of the pleonaste-magnetite series have awide compositional variation of (mol. per cent) FeAl₂O₄—66·6–45·0;MgAl₂O₄—53·0–18·7; Fe₃O₄—6·9–28·1;MnAl₂O₄—1·2–1·5; Fe₂TiO₄—0·6–6·2.Rims are generally enriched in the Fe₃O₄ component as a resultof oxidation. Compositions of ilmenite and magnetite (single,homogeneous and composite grains) are highly variable and resultfrom varying degrees of high temperature oxidation that is associatedwith dehydroxylation of micas and melting. Oxidation mainlyresults in increasing Fe³⁺, Al and decreasing Ti⁴⁺, Fe²⁺ inilmenite, and increasing Fe²⁺, Ti⁴⁺ and decreasing Fe³⁺ in associatedmagnetite. A higher degree of oxidation is reached with exsolutionof rutile from ilmenite and formation of titanhematite and withexsolution of pleonaste from magnetite. Ti-Al rich magnetite(5·1–7·5 and 8·5–13·5wt. per cent respectively) and ilmenite crystallized from meltsin buchitic parts of the xenoliths. Chemical and mineralogic evidence indicates that even with extensivemelting the primary compositions of individual layers in thexenoliths remained unmodified. Apparently the xenoliths didnot remain long enough at high temperatures for desilicationand enrichment in Al₂O₃, TiO₂, FeO, Fe₂O₃, and MgO that resultsby removal of a ‘granitic’ melt, and/or by interactionwith the magma, to occur. T °C-f_o2 values calculated from unoxidized magnetite/ilmenitegive temperatures ranging from 615–710°C for contactmetamorphism and the beginning of melting, and between 873 and1054°C for the crystallization of oxides and mullite/sillimanitefrom high temperature peraluminous melts. f_o2 values of metamorphismand melting were between the Ni-NiO and Fe₂O₃-Fe₃O₄ buffer curves.The relative abundance of xenolith types, geophysical evidenceand contact metamorphic mineralogy indicates that the xenolithswere derived from depths corresponding to between 2–3kb P_load = P_fluid. The xenoliths were erupted during the latestphreatomagmatic eruption from the Wehr volcano which resultedin vesiculation of melts in partially molten xenoliths causingfragmentation and disorientation of solid restite layers.     

Peralkalinity, Al{rightleftharpoons}Si Substitution, and Solubility Mechanisms of H2O in Aluminosilicate Melts

The solubility mechanisms of H₂O in peralkaline sodium aluminosilicatequenched melts (anhydrous NBO/T = 0.5) have been studied withRaman spectroscopy as a function of Al/(Al + Si) (0–0–3)and H²O content (0–7.5 wt.%). The coexisting structuralunits in the anhydrous quenched melts are TO2 (Q4), T2O5(Q3),and TO3 (Q2). In Al-free Na₂Si₄O₉ (NS4) melt, H₂O forms complexes with Na⁺(Na–OH bonds) and with Si⁴⁺ (Si–OH bonds). MolecularH₂O is also detected. TO3 structural units are not detectedin this composition. In the H₂O concentration range between0 and 4 wt.%, there is an approximately 20% increase in NBO/Tresulting from the increased abundance ratio, T₂O₅/TO2. Withfurther increments in water activity, the NBO/T of hydrous NS4melt is reduced. The depolymerization results from hydroxylationof the silica tetrahedra, whereas polymerization is due to formationof complexes with Na–OH bonding. In Al-bearing compositions on the Na₂Si₄O₉–Na₂(NaAl)₄O₉–join, there is evidence for Al–OH bonding in additionto Na–OH and Si–OH bonds. Among these complexes,the relative abundance of those with Si–OH bonds diminisheswith increasing Al/(A1 + Si), whereas complexes with Al–OHand Na–OH bonds become more important. Complexes withNa–OH bonds dominate for H₂O4 wt.%, whereas complexeswith Al–OH dominate at higher water content. The threestructural units, TO₃, T₂O₅, and TO², were observed in bothanhydrous and hydrous peralkaline sodium aluminosilicate melts.Their abundance varies, however, with the H₂O concentrationin the melts. The NBO/T decreases to a minimum (a 30–50%lowering of NBO/T relative to anhydrous materials) for low H₂Ocontents (3–4 wt.% H₂O), and increases as the H₂O contentis increased further.     

Petrology and Petrogenesis of Clinopyroxene-Rich Tholeiitic Lavas, Merelava Volcano, Vanuatu

The mineralogy, petrography and geochemistry of a suite of clinopyroxene-richolivine tholenite lavas from Merelava island, Vanuatu are described.Located at the southern end of the Northern Trough back-arcbasin, this suite displays all the characteristics of primitiveisland arc lavas: flat REE patterns, depleted HFSE, enrichmentin K-group elements relative to LREE, highly calcic plagioclase(to An_{9 3} and Cr-rich spinels (cr-number80) Analysis of groundmasscompositions demonstrates that the variation in MgO within thelava suite (from 13?7 to 4?3% MgO) represents only a small departurefrom a liquid line of descent. Some of the more primitive lavas contain low-Al₂O₃ clinopyroxenemegacrysts (mg-number = 100Mg/(Mg+Fe^{2 +} and ultramafic xenoliths,the latter ranging from fine-grained, tectonite wehrlites andchnopyroxene-bearing harzburgites, to coarse-grained cumulatewehrlites. The cumulate nodules, megacrysts and phenocrysts are shown tobe co-magmatic, and an empirical compositional relationshipis demonstrated for equilibrium olivine-clinopyroxene pairs,covering the observed fractionation range (mg-number_Cpx=0?6375mg-number_O1 + 35?3). On the basis that the most primitive olivine(mg-number _{91 7}) is close to the liquidus composition, thiscompositional relationship demonstrates that clinopyroxene (mg-number=94,and containing no Fe³⁺) was also a liquidus phase. Clinopyroxeneswith mg-number>94 are the product of local oxidation duringmixing of primitive, relatively reduced magmas, and more evolved,oxidized magmas. This mixing also gave rise to relatively narrow,reversely zoned, internal rims on many clinopyroxene and olivinephenocrysts, cumulus crystals, and clinopyroxene megacrysts. Fractionation modelling demonstrates that the most differentiatedsample with 19 wt.% Al₂O₃ can be derived from the most primitivesample with 10?3% Al₂O₃ by removal of 48% crystals of clinopyroxeneand olivine in the proportions 73:27 Plagioclase is a late crystallizingphase and has an insignificant role in the fractionation process. The parent melt composition (mg-number=77) is deduced from themost primitive olivine composition and the liquid line of descent,and is shown to contain equal amounts of MgO and CaO (137 wt.%),a high CaO/Al₂O₃ ratio of 1?3 and an unusually low Ni contentof 137 ppm. Data from published high pressure (8–20 kb)experiments on melting of peridotite and pyrolite do not providean explanati in for the large normative diopside component inthis parent melt (38 mol.%), and a hypothesis is proposed wherebyhigh degrees of melting of refractory Iherzolite or harzburgite+acomponent of lower crustal pyroxenite and/or wehrlite takesplace at the base of the crust (5–55 kb). At this depth,and initially under hydrous conditions, high degrees of meltingwould progressively eliminate orthopyroxene and then clinopyroxeneto produce a dunite residue. The liquid produced near the pointof clinopyroxene elimination would be compatible with the highCaO and Sc contents, and high Sc/Ni, Cr/Ni and D1/Hy ratiosof the lavas, and the refractory nature of the phenocrysts.     

Water-Saturated and -Undersaturated Melting of Metaluminous and Peraluminous Crustal Compositions at 10 kb: Evidence for the Origin of Silicic Magmas in the Taupo Volcanic Zone, New Zealand, and Other Occurrences

The melting relations of two proposed crustal source compositionsfor rhyolitic magmas of the Taupo Volcanic Zone (TVZ), New Zealand,have been studied in a piston-cylinder apparatus at 10 kb totalpressure and a range of water activities generated by H₂O-CO₂vapour. Starting materials were glasses of intermediate composition(65 wt.% Si0₂ representing a metaluminous ‘I-type’dacite and a peraluminous ‘S-type’ greywacke. Crystallizationexperiments were carried out over the temperature range 675to 975?C, with aH₂O values of approximately 1?0, 0?75, 0?5,and 0?25. Talc-pyrex furnace assemblies imposed oxygen fugacitiesclose to quartz-fayalite-magnetite buffer conditions. Assemblages in both compositions remain saturated with quartzand plagioclase through 675–700?C at high aH₂O, 725–750?Cat aH₂O0?5, and 800–875?C at aH₂O0?25, corresponding to<60–70% melting. Concentrations of refractory mineralcomponents (Fe, Mg, Mn, P, Ti) in liquids increase throughoutthis melting interval with increasing temperature and decreasingaH₂O. Biotite and hornblende are the only mafic phases presentnear the solidus in the dacite, compared with biotite, garnet,gedritic orthoamphibole, and tschermakitic clinoamphibole inthe greywacke. Near-solidus melting reactions are of the type:biotite + quartz + plagioclase = amphibole ? garnet, potentiallyreleasing H₂O for dehydration melting in the greywacke, butproducing larger amounts of hornblende and releasing littleH₂O in the dacite. At aH₂O0?25 and temperatures 825–850?C,amphibole dehydration produces anhydrous mineral phases typicalof granulite fades assemblages (clinopyroxene, orthopyroxene,plagioclase?quartz in the dacite; garnet, orthopyroxene, plagioclase?quartzin the greywacke) coexisting with melt proportions as low as40%. Hornblendce-saturated liquids in the dacite are weaklyperaluminous (0?3–1?6 wt.% normative C—within therange of peraluminous TVZ rhyolites), whereas, at aH₂O0?25 andtemperatures 925?C, metaluminous partial melt compositions (upto 1?8 wt.% normative Di) coexist with plagioclase, orthopyroxene,and clinopyroxene. At all water activities, partial melts ofthe greywacke are uniformly more peraluminous (1?5–2?6wt.% normative C), reflecting their saturation in the componentsof more aluminous mafic minerals, particularly garnet and Al-richorthopyroxene. A metaluminous source for the predominantly Di-normativeTVZ rhyolites is therefore indicated. With decreasing aH₂O the stability fields of plagioclase andquartz expand, whereas that of biotite contracts. These changesare reflected in the proportions of normative salic componentsin partial melts of both the dacite and greywacke. At high aH₂O,partial melts are rich in An and Ab and poor in Or (trondhjemitic-tonalitic);with decreasing aH₂O they become notably poorer in An and richerin Or (granodioritic-granitic). These systematic variationsin salic components observed in experimental metaluminous tostrongly peraluminous melts demonstrate that a wide varietyof granitoid magmas may be produced from similar source rocksdepending upon P-T-aH₂O conditions attending partial melting.Some peraluminous granitoids, notably trondhjemitic leucosomesin migmatites, and sodic granodiorites and granites emplacedat deep crustal levels, have bulk compositions similar to nearsolidus melt compositions in both the dacite and greywacke,indicating possible derivation by anatexis without the involvementof a significant restite component.     

Immiscible Transition from Carbonate-rich to Silicate-rich Melts in the 3 GPa Melting Interval of Eclogite + CO2 and Genesis of Silica-undersaturated Ocean Island Lavas

We explore the partial melting behavior of a carbonated silica-deficienteclogite (SLEC1; 5 wt % CO₂) from experiments at 3 GPa and comparethe compositions of partial melts with those of alkalic andhighly alkalic oceanic island basalts (OIBs). The solidus islocated at 1050–1075 °C and the liquidus at 1415 °C.The sub-solidus assemblage consists of clinopyroxene, garnet,ilmenite, and calcio-dolomitic solid solution and the near solidusmelt is carbonatitic (<2 wt % SiO₂, <1 wt % Al₂O₃, and<0·1 wt % TiO₂). Beginning at 1225 °C, a stronglysilica-undersaturated silicate melt (34–43 wt % SiO₂)with high TiO₂ (up to 19 wt %) coexists with carbonate-richmelt (<5 wt % SiO₂). The first appearance of carbonated silicatemelt is 100 °C cooler than the expected solidus of CO₂-freeeclogite. In contrast to the continuous transition from carbonateto silicate melts observed experimentally in peridotite + CO₂systems, carbonate and silicate melt coexist over a wide temperatureinterval for partial melting of SLEC1 carbonated eclogite at3 GPa. Silicate melts generated from SLEC1, especially at highmelt fraction (>20 wt %), may be plausible sources or contributingcomponents to melilitites and melilititic nephelinites fromoceanic provinces, as they have strong compositional similaritiesincluding their SiO₂, FeO^*, MgO, CaO, TiO₂ and Na₂O contents,and CaO/Al₂O₃ ratios. Carbonated silicate partial melts fromeclogite may also contribute to less extreme alkalic OIB, asthese lavas have a number of compositional attributes, suchas high TiO₂ and FeO^* and low Al₂O₃, that have not been observedfrom partial melting of peridotite ± CO₂. In upwellingmantle, formation of carbonatite and silicate melts from eclogiteand peridotite source lithologies occurs over a wide range ofdepths, producing significant opportunities for metasomatictransfer and implantation of melts. KEY WORDS: carbonated eclogite; experimental phase equilibria; partial melting; liquid immiscibility; ocean island basalts     

Petrology of Biotite-Cordierite-Garnet Gneiss of the McCullough Range, Nevada II. P-T-aH2O Path and Growth of Cordierite During Late Stages of Low-P Granulite-Grade Metamorphism

Thermodynamic calculations based on addition of mass balanceequations to the Gibbs Method (Spear, 1986) are used to modelthe cordierite-producing reaction in pelitic gneiss from theMcCullough Range, southern Nevada. Calculations which treatthe model paragenesis as a system open to transfer of H₂O areconsistent with textural relations. Results indicate that cordieritegrew by the continuous net-transfer reaction: 0?76 BIO+1?72 SILL+3? 55 QTZ+0?27 PLG+0?005 GRT +0?06Al₂R²⁺_–1Si_–1[BIO]1?02 KSP+0?76 H₂O +0?30 FeMg_–1[CRD]+0?15FeMg_–1[BIO]+0?0005 FeMg_–1[GRT] +0?005 CaNaAl_–1Si_–1[PLG] with decreasing P, decreasing T, and increasing a_H2O The steepretrograde dP/dT path for these low-pressure granulites contrastswith isobaric cooling paths typical of higher pressure granulites,and suggests uplift and erosion were active during Proterozoicgranulite-grade metamorphism in this area.     

Volcan Popocatepetl, Mexico. Petrology, Magma Mixing, and Immediate Sources of Volatiles for the 1994-Present Eruption

Volcán Popocatépetl has been the site of voluminousdegassing accompanied by minor eruptive activity from late 1994until the time of writing (August 2002). This contribution presentspetrological investigations of magma erupted in 1997 and 1998,including major-element and volatile (S, Cl, F, and H₂O) datafrom glass inclusions and matrix glasses. Magma erupted fromPopocatépetl is a mixture of dacite (65 wt % SiO₂, two-pyroxenes+ plagioclase + Fe–Ti oxides + apatite, 3 wt % H₂O, P= 1·5 kbar, f_O2 = NNO + 0·5 log units) and basalticandesite (53 wt % SiO₂, olivine + two-pyroxenes, 3 wt % H₂O,P = 1–4 kbar). Magma mixed at 4–6 km depth in proportionsbetween 45:55 and 85:15 wt % silicic:mafic magma. The pre-eruptivevolatile content of the basaltic andesite is 1980 ppm S, 1060ppm Cl, 950 ppm F, and 3·3 wt % H₂O. The pre-eruptivevolatile content of the dacite is 130 ± 50 ppm S, 880± 70 ppm Cl, 570 ± 100 ppm F, and 2·9 ±0·2 wt % H₂O. Degassing from 0·031 km³ of eruptedmagma accounts for only 0·7 wt % of the observed SO₂emission. Circulation of magma in the volcanic conduit in thepresence of a modest bubble phase is a possible mechanism toexplain the high rates of degassing and limited magma productionat Popocatépetl. KEY WORDS: glass inclusions; igneous petrology; Mexico; Popocatépetl; volatiles     

Melting of a Hydrous Mantle: III. Phase Relations of Garnet Websterite+H2O at High Pressures and Temperatures

A garnet websterite nodule from the Honolulu volcanic series,Oahu, Hawaii, has been melted in the presence of nearly pureH₂O. The solidus is intermediate between that of peridotiteand gabbro. The curve displays a temperature minimum around20 kb reflecting the breakdown of plagioclase. The Iiquidusis between 1130 ?C and 1150 ?C between 10 and 20 kb vapor pressure.Amphibole (pargasitic hornblende) has an extensive stabilityfield, reaching a maximum temperature about 20 ?C below thegarnet websterite liquidus at 15 kb and a maximum pressure of27.5 kb at 950 ?C. The amphibole-out curve intersects the soliduswith a positive slope. Liquids formed by partial melting of garnet websterite are quartz-normativewithin the stability field of amphibole, but become olivine-normative(tholeiitic) with increasing temperature. Amphibole and clinopyroxeneare enriched in Tschermak's molecule at higher temperatures,pargasite content of amphibole increases with increasing pressure. A garnet websterite-rich upper mantle containing modal olivineyields quartz-normative (13–16 per cent), aluminous (21–4wt. per cent A1₂O₃) melts at 17 P 10 kb and in the presenceof nearly pure H₂O. However, the presence of amphibole controlsthe liquid composition, a situation not found for liquids formedfrom wet peridotite. In contrast to many basalt liquids, liquidof garnet websterite composition cannot fractionate to andesiteby precipitation of amphibole, as amphibole is not a liquidusphase.     

Experimental Phase Relations in the System Tonalite-Peridotite-H2O at 15 kb; Implications for Assimilation and Differentiation Processes near the Crust-Mantle Boundary

Experiments at 15 kb in the tonalite-peridotite-H₂O system provideinformation on some of the phase equilibrium factors that mayinfluence reaction and assimilation processes between quartznormativemagmas and ultramafic rocks in the deep crust and upper mantle.Experiments were done with 5 or 10 wt.% H₂O added to powderednatural samples of tonalite, and mixtures of tonalite with 5or 10 wt.% peridotite added (TP5 and TP10, respectively). Theliquidus phase relations of these starting compositions wereinvestigated between 850 and 1100?C at 15 kb, using gold capsulesso that iron loss to the sample containers was not a problemand meaningful glass and mineral analyses could be obtained.Experiments on the tonalite alone show either liquidus garnet,for samples with 5% H₂O added, or liquidus hornblende, for sampleswith 10% H₂O. In contrast, orthopyroxene is the sole liquidusphase, irrespective of water content, in experiments using startingmixtures of 5 or 10 wt.% peridotite added to tonalite. Glassanalyses of partially crystallized tonalite define a crystallizationpath diverging significantly from the calc-alkaline trend towardshigher Ca/(Mg + Fe) in the CaO–(MgO + FeO)–?SiO₂triangle. In contrast, glasses from partially crystallized mixturesof tonalite with 5 or 10 wt.% peridotite added define a liquidtrend close to natural calc-alkaline compositions in terms ofCa/(Mg + Fe). Of more general significance, the proximity ofa field ofliquidus orthopyroxene on the high (Mg + Fe) sideof compositions along the calc-alkaline trend serves to limitthe Mgenrichment of such melts by interaction with ultramaficrocks. Unless heat is added to the system, reaction of tonaliticcomposition melts with ultramafic rocks will produce only slightlyMg-enriched melts: increasing degree of reaction simply resultsin further precipitation of orthopyroxene + garnet ? clinopyroxeneonce melt compositions reach the orthopyroxene field boundary.     

An Experimental Study of the Influence of Oxygen Fugacity on Fe-Ti Oxide Stability, Phase Relations, and Mineral--Melt Equilibria in Ferro-Basaltic Systems

Equilibrium crystallization experiments at atmospheric pressureand over a range of oxygen fugacity (fO₂) have been carriedout on a ferro-basaltic composition similar to liquids proposedto have been parental to much of the exposed portion of theSkaergaard intrusion. Before Fe-Ti oxide saturation the liquidline of descent is little affected by fO₂. However, the appearancetemperatures of the magnetite-ulvspinel solid solution (Mt)and the ilmenite-haematite solid solution (Ilm) depend stronglyon fO₂. Above the fayalite-magnetite-quartz (FMQ) buffer Mtis the first oxide phase to appear on the liquidus, but belowthe FMQ buffer Ilm is the first oxide to crystallize. The appearancetemperature of Mt is 1100C at FMQ and the Mt liquidus slopeis 30C/log fO₂ unit between FMQ–;2 and FMQJ+1. The Ilmliquidus is at 1100C between FMQ and FMQ–2, but movesto lower temperature at higher fO₂ where Mt is the first oxidephase. The results indicate that the ferric iron content ofMt-saturated melts varies linearly with inverse temperature,and that Ilm saturation is closely related to melt TiO₂ content.Mt saturation produces an immediate enrichment of SiO₂ and depletionin FeO^* in the melt phase, whereas Ilm saturation produces similarenrichment in SiO², but inn enrichment may continue for 10Cbelow the ilmenite liquidus. The experimental liquids reacha maximum of 18 wt% FeO^*, at 48 wt% SiO₂ for ilmenite-saturatedmelts at low f_O2, more differentiated melts having lower ironand higher silica. Cotectic proportions, derived from mass balancecalculations, are in good agreement with data from natural samplesand other experimental studies. Olivine resorption is inferredat all fO₂, with the onset of resorption occurring 10C higherthan the appearance of magnetite. The effect of fO₂ on silicatemineral compositions, and partitioning of elements between coexistingmineral-melt pairs, is small. Thermodynamic considerations suggestthat variations of Fe-Mg partitioning between the iron-richolivines, pyroxenes and melts produced in this study may beexplained by known non-idealities of Fe-Mg mixing in the crystallinephases, rather than nonidealities in the coexisting melts. Theseexperiments also provide insights into many features commonto natural tholeiitic series of volcanic and plutonic rocks,and provide experimental data required for modelling of fractionalcrystallization and crystallization closed to oxygen, processeswhich are not easily investigated experimentally. KEY WORDS: ferro-basalt; Fe-Ti oxides; oxygen fugacity; Skaergaard intrusion; iron enrichment ^*Corresponding author. Present address: Bayerisches Geoinstitut, Univerritt Bayreuth, D-95440 Bayreuth, Germany     

Crystallization of Chromite and Chromium Solubility in Basaltic Melts

The equilibrium between chromite and melt has been determinedon four basalts at temperatures of 1200–1400?C over arange of oxygen fugacity (f_o2) and pressures of 1 atm and 10kb. The Cr content of chromite-saturated melts at 1300?C and1 atm ranges from 0?05 wt.% Cr₂O₃ at a log f_o2= –3 to1?4 wt.% at a log f_o2=–12?8. The Cr²⁺/Cr³⁺ of melt increaseswith decreasing f_o2 and is estimated by assuming a constantpartitioning of Cr³⁺ between chromite and melt at constant temperature.The estimated values of Cr²⁺/Cr³⁺ in the melt are at f_o2 valuesof 4–5 orders of magnitude lower than the equivalent Fe²⁺/Fe³⁺values. The Cr/(Cr+Al) of chromite coexisting with melt at constanttemperature changes little with variation of f_o2 below log f_o2=–6.Five experiments at 10 kb indicate that Cr₂O₃ dissolved in themelt is slightly higher and the Cr/(Cr + Al) of coexisting chromiteis slightly lower than experiments at 1 atm pressure. Thus variationin total pressure cannot explain the large variations of Cr/(Cr+ Al) that are common to mid-ocean ridge basalt (MORB) chromite. Experiments on a MORB at 1 atm at f_o2 values close to fayalite-magnetite-quartz(FMQ) buffer showed that the Al₂O₃ content of melt is highlysensitive to the crystallization or melting of plagioclase,and consequently coexisting chromite shows a large change inCr/(Cr + Al). It would appear, therefore, that mixing of a MORBmagma containing plagioclase with a hotter MORB magma undersaturatedin plagioclase may give rise to the large range of Cr/(Cr +Al) observed in some MORB chromite.     

Granulite Metamorphism, Fluid Buffering, and Dehydration Melting in the Madras Charnockites and Metapelites

Interlayered and cofolded charnockites and metapelites of thetype charnockite area near Madras were metamorphosed under granulitefades conditions. Fe-Mg partitioning between orthopyroxene,garnet, and biotite indicates that chemical equilibrium wasapproached under similar P-T conditions in the two rock suites.Several geothennometers and geobarometers give P-T values whichconverge at 750–800?C and 6.5–7.5 kb. Computations utilizing data from high pressure phase equilibriumexperiments of Bohlen et al. (1983a) and Wones & Dodge (1977)point to several significant relations regarding the behaviourof H₂O during the granulite metamorphism. a_H2O values, computedfrom Bohlen et al.'s (1983a) reversal data and the a-X modelfor phlogopite after Bohlen et al. (1980), show distinctly lowermagnitudes in metapelites (0.10–0.16) than in charnockites(0.23–0.34). No systematic spatial gradients exist withinthe charnockites or metapelites, and a_H2O has similar valuesin metapelite exposures widely separated in the field. Theseimply an internal, rather than an external (e.g., by CO2 influx),control of the fluids. Applying the algebraic method developed by Rumble (1976), Gibbsanalysis in the system K₂O-MgO-FeO-Fe₂O₃-Al₂O₃-SiO₂-TiO₂-H₂Oshows that the chemical potentials of H₂O and to O₂, as monitoredagainst biotite composition and , exhibit gradients with respect to X_Mg in the two rock suites under isothermal-isobaricconditions. µ_H2O was found to decrease with X_Mg^bt in both,while µ_O2 increases with decreasing X_Mg^bt in metapelitesbut increases sympathetically with X_Mg^bt in charnockites. Thesefindings point out again that µ_H2O and µ_O2 wereinternally buffered. The absence of graphite in the metapelites,at an estimated f_O2 = 10^–14.7 b, also argues against anexternal influx of CO₂ and, inter alia, supports internal buffering.A complementary enquiry into variations of a_TIO2 reveals aninverse relation between a_TIO2 and a_H2O, suggesting a similarcontrol for a_TIO2. The inferences from biotite dehydration equilibria, when combinedwith the P-T data and with several field and chemical featuresof these rocks noted earlier (Sen, 1974), make dehydration meltinga distinct possibility for the Madras rocks. It is argued thatthe low a_H2O and high a_TIO2 ({small tilde} 0.9) observed inthe metapelites have been caused by a greater extent of meltingin the precursors of metapelites, which were more hydrous thanthose of charnockites, coupled with preferential partitioningof Ti into the residual rocks—thus strengthening the casefor dehydration melting.     

Thermodynamic Mixing Properties of Muscovite-Paragonite Crystalline Solutions at High Temperatures and Pressures, and their Geological Applications

Reversed Na-K exchange data between mica and a 2 molal aqueous(Na,K)Cl fluid (Flux & Chatterjee, 1986) have been employedto model the thermodynamic mixing behaviour of muscovite-paragonitecrystalline solutions on the basis of the Redlich-Kister equation.For these binary micas, G^ex_m may be expressed as where A=11222+1.389 T+0.2359 P, B=–1134+6.806 T–0.0840 P, and C=–7305+9.043 T, with T in K, P in b, G^ex_m, A, B, and C in joules/mol. G_m^ex is well constrained between 450 and 620?C, and may be extrapolatedbeyond that range with caution. The calculated solvi are skewedtoward the paragonite end member. In the range up to 15 kb,the critical temperature, T_c and the critical composition, X_cmay be expressed as a function of P by the relations: and with P indicated in bars. Calculated phase relations of muscovite-paragonite crystallinesolutions have been depicted in terms of the system KAlSi₃O₈-NaAlSi₃O₈-Al₂O₃-SiO₂-H₂O.These data may be applied to appropriate assemblages involvingmica, alkali feldspar, an Al₂ polymorph, and quartz to estimateP, T and a_H2O conditions of their equilibration. In principle,the muscovite limb of the solvus may be used to obtain geothermometricdata for coexisting muscovite-paragonite pairs, provided theequilibrium pressure is independently known. However, such applicationmust be restricted for the present to micas on the ideal muscovite-paragonitejoin. Mica-alkali feldspar-Al₂SiO₅-quartz or mica-plagioclase-Al₂SiO₅-quartzassemblages may be used to deduce a_H2O in the coexisting fluid,if P, and T of equilibrium are independently known. Examplesof such geological applications are given.     

Petrology and Petrogenesis of Cumulate Peridotites and Gabbros from the Marum Ophiolite Complex, Northern Papua New Guinea

The Marum ophiolite complex in northern Papua New Guinea includesa thick (3–4 km) sequence of ultramafic and mafic cumulates,which are layered on a gross scale from dunite at the base upwardsthrough wehrlite, lherzolite, plagioclase lherzolite, pyroxenite,olivine norite-gabbro and norite-gabbro to anorthositic gabbroand ferrogabbro at the top. Igneous layering and structures,and cumulus textures indicate an origin by magmatic crystallizationin a large magma chamber(s) from magma(s) of evolving composition.Most rocks however show textural and mineralogical evidenceof subsolidus re-equilibration. The cumulate sequence is olivine and chrome spinel followedby clinopyroxene, orthopyroxene and plagioclase, and the layeredsequence is similar to that of the Troodos and Papuan ophiolites.These sequences differ from ophiolites such as Vourinos by thepresence of cumulus magnesian orthopyroxene, and are not consistentwith accumulation of low pressure liquidus phases of mid-oceanridge-type olivine tholeiite basalts. The cumulus phases show cryptic variation from Mg- and Ca-richearly cumulates to lower temperature end-members, e.g. olivineMg_93–78, plagioclase An_94–63. Co-existing pyroxenesdefine a high temperature solidus with a narrower miscibilitygap than that of pyroxenes from stratiform intrusions. Re-equilibratedpyroxene pairs define a low-temperature, subsolidus solvus.Various geothermometers and geobarometers, together with thermodynamiccalculations involving silica buffers, suggest the pyroxene-bearingcumulates crystallized at 1200 °C and 1–2 kb pressureunder low f_O2. The underlying dunites and chromitites crystallizedat higher temperature, 1300–1350 °C. The bulk of thecumulates have re-equilibrated under subsolidus conditions:co-existing pyroxenes record equilibration temperatures of 850–900°C whereas olivine-spinel and magnetite-ilmenite pairs indicatefinal equilibration at very low temperatures (600 °C). Magmas parental to the cumulate sequence are considered to havebeen of magnesian olivine-poor tholeiite composition (>50per cent SiO₂, 15 per cent MgO, 100 Mg/(Mg + Fe²⁺) 78) richin Ni and Cr, and poor in TiO₂ and alkalies. Fractionated examplesof this magma type occur at a number of other ophiolites withsimilar cumulate sequences. Experimental studies show that suchlavas may result from ial melting of depleted mantle lherzoliteat shallow depth. The tectonic environment in which the complexformed might have been either a mid-ocean ridge or a back-arebasin.