Origin of Chromitites in Layered Intrusions: Evidence from Chromite-Hosted Melt Inclusions from the Stillwater Complex

Origin of chromitites in layered intrusions: Evidence from chromite-hosted melt inclusions from the Stillwater Complex

Carl Spandler* Research School of Earth Sciences, Australian National University, Canberra 0200, Australia John Mavrogenes Research School of Earth Sciences, and Department of Earth and Marine Sciences, Australian National University, Canberra 0200, Australia Richard Arculus Department of Earth and Marine Sciences, Australian National University, Canberra 0200, Australia

ABSTRACT 1993); and (5) assimilation of country rock by primitive magma (Ir- Chromitites from layered ma®c intrusions are of great eco- vine, 1975; Kinnaird et al., 2002). nomic importance, yet the origin of these deposits remains enig- The antiquity and slow cooling of most chromitite-bearing layered matic. We describe multiphase silicate inclusions trapped within intrusions have hindered our understanding of the genesis of chromi- chromite grains from the G chromitite seam of the Stillwater Com- tites and layered intrusions in general. The primary petrologic and geo- plex, Montana, United States. These inclusions are interpreted to chemical evidence that are crucial for determining the evolution of represent melt trapped during chromite growth and hence provide cumulate rocks are often removed during subsolidus reequilibration and information on chromitite formation. Most reheated inclusions subsequent hydrothermal alteration or metamorphism. Furthermore, the have variable quench textures and chemical compositions that are compositions of the parental magmas of these intrusionsÐcritical in- consistent with variable degrees of mixing between a high-Mg ba- formation for understanding chromitite formationÐremain poorly saltic parental magma and a Na-rich trondhjemitic melt. The tron- constrained. dhjemite is suggested to derive from partial melting of ma®c or Melt inclusion studies have been extensively used for investigat- ing volcanic systems, yet have largely been ignored by researchers of metasedimentary country rocks. Based on these inclusions, we out- layered intrusions. Trapped within cumulus minerals, melt inclusions line a model for chromitite formation involving ponding of a new may remain unaffected by postcrystallization alteration and hence may pulse of primitive magma at the roof of the Stillwater magma be useful for determining parent magma compositions (Spandler et al., chamber, followed by localized partial melting and assimilation of 2000) or for unraveling complex processes such as magma mixing or the country rock. The newly formed hybrid melts become oversat- assimilation. In this paper we examine multiphase inclusions within urated in chromite, leading to extensive chromite crystallization. chromite from the G chromitite seam of the Stillwater Complex, Mon- Chromitite horizons are proposed to form from dense chromite- tana. These inclusions are interpreted to be crystallized melt inclusions rich plumes that periodically sink down from the roof zone to settle that were trapped during chromite growth, and hence they provide fun- out as layers at the basal cumulate mush zone. Numerous radio- damental information on the composition of the cumulate-forming genic isotope studies, together with the widespread occurrence of magmas and, more signi®cantly, allow us to establish the processes similar multiphase inclusions in chromite from other cumulate responsible for chromitite formation. complexes, indicate that assimilation of country rock by primitive magma may be a critical mechanism for forming chromitites in GEOLOGICAL SETTING AND SAMPLE DESCRIPTION many layered intrusions. The 2.7 Ga Stillwater Complex is composed of a Ͼ6-km-thick sequence of ma®c and ultrama®c cumulates emplaced into the Archean metasedimentary rocks of the Beartooth Mountains, Montana, United Keywords: chromitite, melt inclusions, Stillwater Complex, parent States. Mineralogical and isotopic variations through the sequence in- magmas, layered intrusions. dicate that several magma types and multiple magma injections were responsible for cumulate formation (McCallum, 1996). The peridotite INTRODUCTION zone near the base of the complex includes at least 20 cyclic units Layered ma®c intrusions not only represent natural laboratories consisting of olivine, olivine-orthopyroxene, and orthopyroxene cu- for studying processes of magmatic differentiation and assimilation mulates (Raedeke and McCallum, 1984). Chromitite layers that occur within the crust, but may also contain extensive precious and base near the base of many of the cyclic units are sequentially labeled from metal mineralization. Chromite-rich seams (chromitites) within layered A (lowermost) through K (uppermost). The G and H chromitites are intrusions, such as the Bushveld and Stillwater Complexes, host the the thickest and most economically important seams (Campbell and majority of the world's Cr reserves and may contain signi®cant plati- Murck, 1993). num group element (PGE) mineralization. These chromitite horizons In this study we examine a sample of the main G chromitite seam have been subject to extensive prior study, yet their origin and evo- collected from above the Benbow Mine head frame. The sample con- lution remain highly debated. Chromium is relatively immobile during sists of 1±2 mm cumulus chromite grains set in a matrix of foliated hydrothermal processes, and chromite is only a minor phase produced serpentinite. Polished sections of the sample reveal that isolated mul- during closed-system cotectic crystallization of ma®c parent magma tiphase inclusions or inclusion clusters occur within the core zones of at least 20% of the chromite grains. Similar inclusions in ultrama®c (e.g., Campbell and Murck, 1993). Therefore, exceptional magmatic zone chromites were described by Jackson (1961) and Page (1971). It processes are required for chromitite formation. Most commonly pro- is well known that chromitite layers undergo recrystallization during posed triggers for extensive chromite crystallization include: (1) a pres- cooling (e.g., Campbell and Murck, 1993). Nonetheless, high- sure change in the magma chamber (Cameron, 1977); (2) a change in resolution backscattered-electron imaging of the chromite grains (see oxygen fugacity of the magma (Ulmer, 1969); (3) interaction of sea- GSA Data Repository1) reveals the location of the original magmatic water or alkaline ¯uids and primitive magma (Talkington et al., 1984; Whittaker and Watkinson, 1984); (4) mixing of primitive magma with 1GSA Data Repository item 2005173, analytical techniques, Table DR1, fractionated residual magma (Irvine, 1977; Campbell and Murck, representative compositions of chromite and inclusion phases, and Table DR2, representative homogenized melt inclusion compositions, is available online at www.geosociety.org/pubs/ft2005.htm, or on request from editing@geosociety *Corresponding author. E-mail: [email protected]. .org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.

᭧ 2005 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. Geology; November 2005; v. 33; no. 11; p. 893±896; doi: 10.1130/G21912.1; 4 ®gures; Data Repository item 2005173. 893 Figure 1. A±B: High-resolution backscattered-electron (BSE) images of chromite grains with original magmatic grain boundaries and melt inclusions evident. Melt inclusions are clearly contained within original grain boundaries. Bright halos around melt inclusions are Fe-rich chromite zones formed by Fe-Mg-Mn exchange between melt inclusions and host spinel (see text). C: BSE image of melt inclusion prior to rehomogenization with numerous daughter minerals exposed.

grain boundaries, and in almost all cases, the inclusions are within the remelting. Complete remelting and rehomogenization of all inclusions original chromite crystals (Figs 1A, 1B). was only achieved at 1450 ЊC, so we only present data for inclusions We examined more than 100 individual inclusions ranging in size quenched at 1450 ЊC. Three types of homogenized inclusions were from 10 to 100 ␮m. All inclusions have rounded or negative crystal observed by SEM. Roughly half of the inclusions were quenched to a shapes and consist of two or more phases. Enstatite and aspidolite (Na- homogeneous glass (type 1; Fig. 2A), whereas most other inclusions phlogopite) are always present, and in all cases compose the majority quenched to glass intergrown with spinifex-textured olivine (type 2; of the inclusions (Fig. 1C). Other common phases include magnesio- Fig. 2B). Both type 1 and 2 inclusions may have contained one or katophorite (Na-Ca amphibole), albite, and diopside. Tiny chalcopyrite more vapor bubbles, but the vapor phase composed Ͻ5% of these crystals were identi®ed in only two of the inclusions. Between inclu- inclusions in all cases. Type 3 inclusions composed Ͻ5% of the total sions there is little variation in mineral compositions, but mineral pro- inclusion population and consisted of a large vapor bubble with a small portions may be highly variable. Representative compositions of these rim of melt (Fig. 2C). minerals and the host chromite are presented in Table DR1 (see foot- Representative compositions of the homogenized inclusion are note 1). The ferromagnesian inclusion minerals are close to the pure presented in Table DR2 (see footnote 1). All inclusions have signi®- ϩ Mg end-member compositions, and mica and amphibole in the inclu- cantly higher FeO MgO and lower Al2O3 compared to typical ba- sions are signi®cantly more Na rich than primary amphibole and phlog- saltic melts. Type 2 inclusions have the highest ferromagnesian con- opite from the surrounding cumulates (Page and Zientek, 1987). The tents. Type 1 and 2 inclusions vary in composition, but generally have composition of the host chromite is typical of high-temperature mag- relatively low SiO2, very low CaO, and high Na2O contents. Chlorine matic chromite, and is similar to chromite reported from other sections concentrations are elevated compared to typical oceanic basalts (Boud- of the G chromitite (Campbell and Murck, 1993). reau et al., 1997). Melt compositions from the type 3 inclusions are We interpret the inclusions as crystallized silicate melt inclusions, generally similar to the type 1 and 2 inclusions, except they contain based on their containment within magmatic chromite, their morphol- lower Na2O and much higher CaO concentrations. ogy, their similar multisilicate mineral composition, and their behavior In general, the chemical compositions of the type 1 and 2 inclu- on reheating. In order to establish the chemical composition of these sions are consistent with the assemblage of daughter minerals in the trapped melts, we attempted to rehomogenize the inclusions by heating unheated inclusions. However, the daughter minerals are lacking in Fe, clean chromite separates enclosed in platinum capsules in a 1 atm fur- whereas the homogenized inclusions contain high Fe contents. This nace for 1 h, followed by rapid quenching. Oxygen fugacity was set discrepancy is caused by extensive exchange of divalent Fe, Mg, and at fayalite-magnetite-quartz buffer conditions, and temperatures varied Mn between the melt inclusions and chromite host during the reho- from 1300 to 1450 ЊC. After quenching, the chromite grains were mogenization procedures (e.g., Danyushevsky et al., 2000). Moreover, mounted in epoxy and polished to expose inclusions. Inclusions were the Mg-rich compositions of the daughter minerals are regarded as an then examined and analyzed for major elements using an energy dis- artifact of divalent cation exchange with the chromite host during the persive spectrometer-equipped scanning electron microscope (SEM). slow cooling of the cumulate sequence. Consequently, the MgO, FeO, Instrument speci®cations and analytical techniques are given in the and MnO contents of the reheated inclusions are not considered to be GSA Data Repository (see footnote 1). representative of the original trapped melts, although the total atomic proportion of Fe ϩ Mn ϩ Mg in the inclusions should have remained COMPOSITION OF THE MELT INCLUSIONS constant. Therefore, meaningful interpretation of the inclusion com- In experiments conducted at 1300±1400 ЊC, some inclusions were positions can be made by recasting the element concentrations as atom- homogenized and quenched to glass, but many only underwent partial ic percent (Table DR2).

Figure 2. Backscattered- electron images of melt inclusions after reheating to ؇C. A: Glassy type 1 1450 inclusions. B: Spinifex- textured type 2 inclusions. C: Vapor-rich type 3 inclusions.

894 GEOLOGY, November 2005 Figure 3. Elemental variation plots for type 1 and 2 rehomogenized melt inclusions. All elements are expressed as atomic percent, except Cl, which is in ppm.

Collectively, the type 1 and 2 inclusions span a continuous com- Figure 4. Stillwater magma chamber during formation of peridotite positional range from high Fe ϩ Mg ϩ Mn, low Si melts through to zone. Chromitite formation is shown to form after in¯ux of primi- ϩ ϩ tive magma, partial melting of roof rocks, magma mixing, and relatively high Si, Al, and Na, and relatively low Fe Mg Mn chromite crystallization and accumulation. melts (Fig. 3). There is a slight trend of decreasing K with decreasing Fe ϩ Mg ϩ Mn, while Ca and Cl contents do not correlate with any other element. hydrothermal alteration or metamorphism. The enrichment of Ca in the small volumes of melt trapped with the ¯uid indicates that the ¯uid DISCUSSION was probably Ca rich, with Ca partitioning into the melt during the Origin of the Melt Inclusions rehomogenization procedures. Exsolution of a Ca-rich ¯uid during The melt inclusion compositions described here are unlike any country-rock assimilation and chromite crystallization would also ac- magma that has been proposed as parental to any layered intrusion. count for the very low CaO contents of the type 1 and 2 inclusions. We interpret the type 1 and 2 inclusion suites to represent products of variable mixing between two end-member components. The Model for Chromitite Formation ferromagnesian-rich end member is suggested to be high-Mg basaltic Because the melt inclusions represent samples of the melt respon- magma that may be parental to much of the ultrama®c series of the sible for chromite crystallization, they provide important information Stillwater Complex (McCallum, 1996). The presence of hydrous on the mechanisms of chromitite formation. We have shown that the daughter minerals and high Cl contents in even the most primitive melt inclusions do not represent mixing of primitive magma with ¯uid or inclusions supports the suggestions of Boudreau et al. (1997) that this residual fractionated magma, so these processes are discounted as parent magma may have contained a signi®cant volatile component. mechanisms of formation of the G chromitite. Instead, localized assim- The Si-rich end member is interpreted to be silicate melt rather than a ilation of country rock by parental magma is the proposed trigger for Cl-bearing ¯uid because of the relatively low vapor component, the supersaturation of chromite. lack of correlation between Na and Cl, and excellent correlation be- We propose a model for chromitite formation (Fig. 4) similar to tween Na, Al, and Si (Fig. 3). Furthermore, this component cannot that outlined by Irvine (1975) and Kinnaird et al. (2002). Batches of represent residual magma produced by parent magma fractionation, high-Mg basaltic parent magma are expected to have periodically in- because there is an opposing relationship between incompatible ele- jected into the Stillwater magma chamber during accumulation of the ments K and Na. The end member is expected to be trondhjemitic in peridotite zone (Raedeke and McCallum, 1984). We suggest that the composition with very high Na and Na/K contents. The source of this high temperature (Ͼ1400 ЊC) of the parent magma allowed it to ascend component is suggested to be partial melt of crystallized ma®c wall to the roof of the chamber. Metasedimentary country rocks or prior rock or metasedimentary country rock to the complex. Partial melts of crystallized ma®c rocks at the roof of the chamber underwent partial hornfels facies country rock have previously been described from the fusion to form high-Na trondhjemitic liquids. The original roof rocks basal zone of the complex (Page, 1979). Therefore, the type 1 and 2 to the Stillwater Complex are not preserved, but evidence of country- inclusions are interpreted to represent disequilibrium melt compositions rock melting at the roof of other similar layered intrusions has been formed by variable degrees of mixing between partially melted country well documented (e.g., Irvine, 1975). Mixing between the trondhjemite rock and primitive parent magma. We expect that even the most prim- and parent magma at the roof of the magma chamber led to localized itive melt inclusions contain at least 10% of the trondhjemitic melt hybridization and rapid cooling of the melt. Irvine (1975) demonstrated component. Conditions for entrapment of melt inclusions are most fa- that contamination of basaltic or picritic melts with even small amounts vorable close to the wall rock±parental magma interface (Danyushev- of silica and alkalies may suppress olivine crystallization, leaving chro- sky et al., 2004), so it is likely that the type 1 and 2 inclusions only mite as the only crystallizing phase. Therefore, the magma mixing and represent localized melt contamination processes. Similar processes associated cooling would lead to extensive chromite crystallization and were invoked by Danyushevsky et al. (2004) to explain unusual melt may also have caused exsolution of a minor ¯uid phase. Rapid chro- inclusion compositions trapped in forsteritic olivine phenocrysts from mite growth promoted entrapment of samples of the hybrid magma and volcanic rocks. exsolved ¯uid as inclusions. Rapid crystallization and cooling would Crystallized equivalents of the type 3 inclusions have not been also promote convective overturn of the parental magma at the roof of identi®ed, so the nature of these inclusions remains poorly constrained. the chamber (Marsh, 1988), allowing large volumes of primitive mag- Homogenized type 3 inclusions are dominated by a bubble (Fig. 2C) ma to interact with the country rock over a relatively short time, and and are contained within primary chromite. Thus they represent a ¯uid hence, extensive chromite accumulation. Due to density contrasts, the phase trapped during chromite growth, rather than during subsequent accumulating chromite is expected to have periodically sunk through

GEOLOGY, November 2005 895 the underlying magma as plumes (Marsh, 1988) to settle out within inclusions in primitive olivine phenocrysts: The role of localized reaction the cumulate mush near the base of the chamber (Fig. 4). Regular processes in the origin of anomalous compositions: Journal of Petrology, v. 45, p. 2531±2553, doi: 10.1093/petrology/egh080. seismic events may have had an important role in initiating the plume Horan, M.F., Morgan, J.W., Walker, R.J., and Cooper, R.W., 2001, Re-Os iso- migration and causing liquefaction of the cumulate mush, which al- topic constraints on magma mixing in the peridotite zone of the Stillwater lowed development of laterally continuous chromitite layers within the Complex, Montana, USA: Contributions to Mineralogy and Petrology, mush, as suggested by Nex (2004). This model is consistent with the v. 141, p. 446±457. Irvine, T.N., 1975, Crystallization sequences in the Muskox intrusion and other recent recognition of numerous low-angle unconformities in the peri- layered intrusionsÐII. Origin of chromitite layers and similar deposits of dotite zone cumulate sequence (Cooper, 1997), the lack of composi- other magmatic ores: Geochimica et Cosmochimica Acta, v. 39, tional variations in cumulus minerals that would be expected for in p. 991±1020, doi: 10.1016/0016-7037(75)90043-5. situ cumulate formation (Raedeke and McCallum, 1984; Campbell and Irvine, T.N., 1977, Origin of chromite layers in the Muskox intrusion and other Murck, 1993), and radiogenic isotope data of Stillwater chromitites that stratiform intrusions: A new interpretation: Geology, v. 5, p. 273±277, doi: 10.1130/0091-7613(1977)5Ͻ273:OOCLITϾ2.0.CO;2. require the addition of crustal contaminants in the magma chamber Jackson, E.D., 1961, Primary textures and mineral associations in the ultrama®c (Lambert et al., 1994; Horan et al., 2001). zone of the Stillwater Complex, Montana: U.S. Geological Survey Pro- The evidence presented here concerning chromitite formation in fessional Paper 358, 106 p. the Stillwater Complex is also relevant for mineralization in other mag- Kinnaird, J.A., Kruger, F.J., Nex, P.A.M., and Cawthorn, R.G., 2002, Chromitite formationÐA key to understanding processes of platinum enrichment: In- matic complexes. Chromite-hosted inclusions consisting of unusual al- stitution of Mining and Metallurgy Transactions, Section B, Applied Earth kaline mineral assemblages have been reported in chromitites from Science, v. 111, p. B23±B35. other layered intrusions (e.g., McDonald, 1965; Irvine, 1975; Talking- Lambert, D.D., Walker, R.J., Morgan, J.W., Shirey, S.B., Carlson, R.W., Zientek, ton et al., 1984). Na-phlogopite is almost exclusively found only as M.L., Lipin, B.R., Koski, M.S., and Cooper, R.L., 1994, Re-Os and Sm- inclusions in chromitites (e.g., Costa et al., 2001). Like the Stillwater Nd isotope geochemistry of the Stillwater Complex, Montana: Implica- tions for the petrogenesis of the J-M Reef: Journal of Petrology, v. 35, chromite inclusions described here, it is likely that most of these in- p. 1717±1753. clusions represent hybrid melts trapped during chromite formation. Marsh, B.D., 1988, Crystal capture, sorting, and retention in convecting magma: Based on this premise, and consistent with numerous isotopic studies Geological Society of America Bulletin, v. 100, p. 1720±1737, doi: that necessitate crustal assimilation into the parental magmas of chrom- 10.1130/0016-7606(1988)100Ͻ1720:CCSARIϾ2.3.CO;2. McCallum, I.S., 1996, The Stillwater Complex, in Cawthorn, R.G., ed., Layered itites (e.g., Schoenberg et al., 1999; Kinnaird et al., 2002), we propose intrusions: Amsterdam, Elsevier Science, p. 441±483. that processes similar to the model outlined here may cause chromitite McDonald, J.A., 1965, Liquid immiscibility as one factor in chromitite seam formation in many cumulate complexes. Mixing of crustally contami- formation in the Bushveld Igneous Complex: Economic Geology and the nated and mantle-derived magma is also invoked as the mechanism Bulletin of the Society of Economic Geologists, v. 60, p. 1674±1685. responsible for sul®de liquid saturation and PGE mineralization in lay- Nex, P.A.M., 2004, Formation of bifurcating chromitite layers of the UG1 in the Bushveld Igneous Complex, an analogy with sand volcanoes: Geolog- ered intrusions (Schoenberg et al., 1999, and references therein). ical Society [London] Journal, v. 161, p. 903±909. Country-rock assimilation and magma mixing are commonly re- Page, N.J., 1971, Sul®de minerals in the G and H chromitite zones of the garded to be critical processes for mineralization and cumulate for- Stillwater Complex, Montana: U.S. Geological Survey Professional Paper mation in layered intrusions, yet postmagmatic recrystallization of 694, 20 p. Page, N.J., 1979, Stillwater Complex, MontanaÐStructure, mineralogy and pe- these rocks often removes evidence of these processes. As a ®nal point, trology of the basal zone with emphasis on the occurrence of sul®des: we stress the potential bene®t of studying melt inclusions contained in U.S. Geological Survey Professional Paper 1038, 69 p. cumulus minerals for understanding complex magmatic and cumulate Page, N.J., and Zientek, M.L., 1987, Composition of primary postcumulus am- processes. phibole and phlogopite within an olivine cumulate in the Stillwater Com- plex, Montana: U.S. Geological Survey Bulletin 1674, p. A1±A35. ACKNOWLEDGMENTS Raedeke, L.D., and McCallum, I.S., 1984, Investigations in the Stillwater Com- The Electron Microscope Unit, Australian National University, is ac- plex: Part II. Petrology and petrogenesis of the ultrama®c series: Journal knowledged for use of the scanning electron microscopes. We thank Greg Yax- of Petrology, v. 25, p. 395±420. ley and Nick Tailby for comments on an early version of the manuscript, and Schoenberg, R., Kruger, F.J., Nagler, T.F., Meisel, T., and Kramers, J.D., 1999, Norman Page and two anonymous reviewers for constructive and critical PGE enrichment in chromitite layers and the Merensky Reef of the west- reviews. ern Bushveld Complex; a Re-Os and Rb-Sr isotope study: Earth and Plan- etary Science Letters, v. 172, p. 49±64, doi: 10.1016/S0012-821X(99) REFERENCES CITED 00198-3. Boudreau, A.E., Stewart, M.A., and Spivack, A.J., 1997, Stable Cl isotopes and Spandler, C.J., Eggins, S.M., Arculus, R.J., and Mavrogenes, J.A., 2000, Using origin of high-Cl magmas of the Stillwater Complex, Montana: Geology, melt inclusions to determine parent-magma compositions of layered intru- v. 25, p. 791±794, doi: 10.1130/0091-7613(1997)025Ͻ0791:SCIAOOϾ sions: Application to the Greenhills Complex (New Zealand), a platinum- 2.3.CO;2. group-minerals±bearing island-arc intrusion: Geology, v. 28, p. 991±994, Cameron, E.N., 1977, Chromite in the central sector of the eastern Bushveld doi: 10.1130/0091-7613(2000)028Ͻ0991:UMITDPϾ2.3.CO;2. Complex, South Africa: American Mineralogist, v. 62, p. 1082±1096. Talkington, R.W., Watkinson, D.H., Whittaker, P.J., and Jones, P.C., 1984, Campbell, I.H., and Murck, B.W., 1993, Petrology of the G and H chromitite Platinum-group minerals and other solid inclusions in chromite of ophiol- zones in the Mountain View area of the Stillwater Complex, Montana: itic complexes: Occurrence and petrological signi®cance: Tschermaks Mi- Journal of Petrology, v. 34, p. 291±316. neralogische und Petrographische Mitteilungen, v. 32, p. 285±301. Cooper, R.W., 1997, Magmatic unconformities and stratigraphic relations in the Ulmer, G.C., 1969, Experimental investigations of chromite spinels, in Wilson, peridotite zone, Stillwater Complex, Montana: Canadian Journal of Earth H.D.B., ed., Magmatic ore deposits: Economic Geology Monograph 4, Sciences, v. 34, p. 407±425. p. 114±131. Costa, F., Dungan, M.A., and Singer, B.S., 2001, Magmatic Na-rich phlogopite Whittaker, P.J., and Watkinson, D.H., 1984, Genesis of chromitite from the in a suite of gabbroic crustal xenoliths from VolcaÂn San Pedro, Chilean Mitchell Range, central British Columbia: Canadian Mineralogist, v. 22, Andes: Evidence for a solvus relation between phlogopite and aspidolite: p. 161±172. American Mineralogist, v. 86, p. 29±35. Danyushevsky, L.V., Della-Pasqua, F.N., and Sokolov, S., 2000, Re- Manuscript received 31 May 2005 equilibration of melt inclusions trapped by magnesian olivine phenocrysts Revised manuscript received 7 July 2005 from subduction-related magmas: Petrological implications: Contributions Manuscript accepted 13 July 2005 to Mineralogy and Petrology, v. 138, p. 68±83. Danyushevsky, L.V., Leslie, R.A.J., Crawford, A.J., and Durance, P., 2004, Melt Printed in USA

896 GEOLOGY, November 2005