A parameterized model for REE distribution between low-Ca pyroxene and basaltic melts with applications to REE partitioning in low-Ca pyroxene along a mantle adiabat and during pyroxenite-derived melt and peridotite interaction

Low-Ca pyroxenes play an important role in mantle melting, melt-rock reaction, and magma differentiation processes. In order to better understand REE fractionation during adiabatic mantle melting and pyroxenite-derived melt and peridotite interaction, we developed a parameterized model for REE partitioning between low-Ca pyroxene and basaltic melts. Our parameterization is based on the lattice strain model and a compilation of published experimental data, supplemented by a new set of trace element partitioning experiments for low-Ca pyroxenes produced by pyroxenite-derived melt and peridotite interaction. To test the validity of the assumptions and simplifications used in the model development, we compared model-derived partition coefficients with measured partition coefficients for REE between orthopyroxene and clinopyroxene in well-equilibrated peridotite xenoliths. REE partition coefficients in low-Ca pyroxene correlate negatively with temperature and positively with both calcium content on the M2 site and aluminum content on the tetrahedral site of pyroxene. The strong competing effect between temperature and major element compositions of low-Ca pyroxene results in very small variations in REE partition coefficients in orthopyroxene during adiabatic mantle melting when diopside is in the residue. REE partition coefficients in orthopyroxene can be treated as constants at a given mantle potential temperature during decompression melting of lherzolite and diopside-bearing harzburgite. In the absence of diopside, partition coefficients of light REE in orthopyroxene vary significantly, and such variations should be taken into consideration in geochemical modeling of REE fractionation in clinopyroxene-free harzburgite. Application of the parameterized model to low-Ca pyroxenes produced by reaction between pyroxenite-derived melt and peridotite revealed large variations in the calculated REE partition coefficients in the low-Ca pyroxenes. Temperature and composition of starting pyroxenite must be considered when selecting REE partition coefficients for pyroxenite-derived melt and peridotite interaction.     

变质玄武岩体系相平衡及矿物 熔体微量元素分配：限定TTG/埃达克岩形成条件和大陆壳生长模型

埃达克质岩石是高Na、Al和Sr、低Y和HREE以及Nb、Ta亏损的钠质花岗质岩石,奥长花岗岩-英云闪长岩-花岗闪长岩(TTG)是早期(太古宙)大陆壳主要组分,成分与埃达克质岩石相似,这些成分独特的岩石总体上认为是俯冲洋壳、下地壳和拆沉的下地壳中变质玄武岩部分熔融的产物。文中综述我们近年来在变质玄武岩体系相平衡和矿物-熔体微量元素分配实验研究成果:相平衡实验和熔体微量元素特征研究表明,变质玄武岩部分熔融过程中金红石是导致TTG/埃达克岩浆Nb、Ta亏损的必要残留矿物,从而否定了前人“TTG由无金红石的角闪岩熔融产生”的观点;证实金红石仅仅在压力1.5GPa以上才能稳定存在,从而限定TTG/埃达克岩熔体必定产生在大约50km以上,表明TTG/埃达克岩是在相对较深的含金红石榴辉岩相条件下熔融产生的。矿物(石榴子石、角闪石,单斜辉石和金红石)-熔体微量元素分配系数测定和部分熔融模拟结果进一步限定俯冲洋壳和下地壳起源的TTG/埃达克岩浆由含金红石角闪榴辉岩熔融产生,而拆沉下地壳起源的埃达克岩浆的产生要求软流圈地幔高温,由无水或含有少量含水矿物的榴辉岩熔融产生。     

Distribution of REE between clinopyroxene and basaltic melt along a mantle adiabat: effects of major element composition, water, and temperature

The distribution of rare earth elements (REE) between clinopyroxene (cpx) and basaltic melt is important in deciphering the processes of mantle melting. REE and Y partition coefficients from a given cpx-melt partitioning experiment can be quantitatively described by the lattice strain model. We analyzed published REE and Y partitioning data between cpx and basaltic melts using the nonlinear regression method and parameterized key partitioning parameters in the lattice strain model (D ₀, r ₀ and E) as functions of pressure, temperature, and compositions of cpx and melt. D ₀ is found to positively correlate with Al in tetrahedral site (Al ^T ) and Mg in the M2 site (Mg^M2) of cpx and negatively correlate with temperature and water content in the melt. r ₀ is negatively correlated with Al in M1 site (Al^M1) and Mg^M2 in cpx. And E is positively correlated with r ₀. During adiabatic melting of spinel lherzolite, temperature, Al ^T , and Mg^M2 in cpx all decrease systematically as a function of pressure or degree of melting. The competing effects between temperature and cpx composition result in very small variations in REE partition coefficients along a mantle adiabat. A higher potential temperature (1,400°C) gives rise to REE partition coefficients slightly lower than those at a lower potential temperature (1,300°C) because the temperature effect overwhelms the compositional effect. A set of constant REE partition coefficients therefore may be used to accurately model REE fractionation during partial melting of spinel lherzolite along a mantle adiabat. As cpx has low Al and Mg abundances at high temperature during melting in the garnet stability field, REE are more incompatible in cpx. Heavy REE depletion in the melt may imply deep melting of a hydrous garnet lherzolite. Water-dependent cpx partition coefficients need to be considered for modeling low-degree hydrous melting.     

Basic phase relations of polybaric batch, percolative and critical melting

Summary ?Partial melting of the mantle is polybaric which implies that the phase relations change during partial melting. In addition to the pressure the composition of the melt depends on the melting mode. Various melting models have been suggested. Here the basic phase relations of polybaric batch, percolative, and critical melting are considered, using a simple ternary system. The percolative melts are in equilibrium with their residua, but differ somewhat in composition from those of batch melting. Critical melting is a fractional type of melting where the residuum contain interstitial melt. The critical melts differ in composition from batch melts. The linear trends of peridotites from ophiolites show that the extracted melts had nearly constant compositions, and therefore were extracted within a small pressure interval. A comparison between the trends of mantle peridotite and experimental batch melts suggests strongly that the melt extracted from the peridotites are in equilibrium with their residua. This could suggest that either batch or percolative melting are relevant melting modes for the mantle. However, isotopic disequilibria favor instead a critical mode of melting. This inconsistency can be avoided if the ascending melts are accumulated within a source region and equilibrate with the residuum before the melt is extracted from the source region. The evidence for equilibrium suggests that multisaturation of tholeiitic compositions in PT-diagrams is relevant for estimating pressure and temperature of generation of primary tholeiitic magmas. Received September 2, 2001; revised version accepted March 20, 2002     

Rare earth element concentrations in a suite of basanitoids and alkali olivine basalts from Grenada,Lesser Antilles

A suite of basanitoids and alkali olivine basalts from Grenada, Lesser Antilles were analyzed for rare earth elements. The REE concentrations of these rocks are characterized by a small variation in the heavy REE (7 to 9 times chondrite) and a large variation in the light REE (17 to 93 times chondrite). Among the possible mechanisms to account for the REE variations, fractional crystallization processes at low and high pressures, and partial melting processes (both batch melting and fractional melting) were examined, using the partition relationships of REE among silicate minerals and melts. It is suggested that the observed REE variations are best explained by variable degrees of batch partial melting, in which garnet is present as one of the solid phases through 2 to 17% melting of a garnet lherzolite parent rock.     

Platinum-group element systematics in Mid-Oceanic Ridge basaltic glasses from the Pacific, Atlantic, and Indian Oceans

The concentrations of Ir, Ru, Pt and Pd have been determined in 29 Mid-Oceanic Ridge basaltic (MORB) glasses from the Pacific (N = 7), the Atlantic (N = 10) and the Indian (N = 11) oceanic ridges and the Red Sea (N = 1) spreading centers. The effect of sulfide segregation during magmatic differentiation has been discussed with sample suites deriving from parental melts produced by high (16%) and low (6%) degrees of partial melting, respectively. Both sample suites define positive and distinct covariation trends in platinum-group elements (PGE) vs. Ni binary plots. The high-degree melting suite displays, for a given Ni content, systematically higher PGE contents relative to the low-degree melting suite. The mass fraction of sulfide segregated during crystallization (X_sulf), the achievement of equilibrium between sulfide melt and silicate melts (R_eff), and the respective proportions between fractional and batch crystallization processes (S_b) are key parameters for modeling the PGE partitioning behavior during S-saturated MORB differentiation. Regardless of the model chosen, similar sulfide melt/silicate melt partition coefficients for Ir, Ru, Pt and Pd are needed to model the sulfide segregation process, in agreement with experimental data. When corrected for the effect of magmatic differentiation, the PGE data display coherent variations with partial melting degrees. Iridium, Ru and Pt are found to be compatible in nonsulfide minerals whereas the Pd behaves as a purely chalcophile element. The calculated partition coefficients between mantle sulfides and silicate melts (assuming a PGE concentration in the oceanic mantle at ∼0.007 × CI-chondritic abundances) increase from Pd (∼10³) to Ir (∼10⁵). This contrasting behavior of PGE during S-saturated magmatic differentiation and mantle melting processes can be accounted for by assuming that Monosufide Solid Solution (Mss) controls the PGE budget in MORB melting residues whereas MORB differentiation processes involve Cu-Ni-rich sulfide melt segregation.     

Phase relations of osumilite and dehydration melting in pelitic rocks: a simple thermodynamic model for the KFMASH system

 Thermodynamic modelling of (1) osumilite solid solutions and (2) dehydration melting in pelitic compositions within the KFMASH system is quite successful in reproducing the invariant and univariant reactions determined in experimental studies. Even though rather preliminary, such melt thermodynamic models may be very useful in interpolating and extrapolating the limited information available from a small number of experimental runs. These methods allow the compositions of all phases to be monitored as a function of pressure, temperature and equilibrium phase assemblage for any desired bulk composition. Locating the higher variance phase fields (e.g. quadrivariant, quinivariant) is often difficult or impossible by inspection, but is made relatively easy using thermodynamic software such as thermocalc. In the KFMASH system the calculated partition of Fe and Mg between osumilite, garnet, cordierite, orthopyroxene and biotite are shown to be in good agreement with experimental and natural data and allow reliable calculation of mineral compositions coexisting with quartz-saturated and H₂O-undersaturated melts for a variety of bulk compositions. These phase diagram calculations allow quite tight limits to be placed on the pressure, temperature and water activity conditions which accompanied metamorphism of natural osumilite occurrences in Nain, Namaqualand, and Rogaland. At fixed bulk composition, the initial melting of pelites by dehydration of biotite can occur via univariant, divariant or trivariant equilibria depending upon pressure of metamorphism. Of particular interest is that, for low pressures or more magnesian bulk compositions, fluid-absent melting begins by generating liquid from the high-variance assemblage biotite+cordierite+K-feldspar+ quartz. This type of modelling allows investigation, at least qualitatively, of the fine scale details of melt production as a function of changes in pressure, temperature and bulk composition. Received: 29 November 1995 / Accepted: 22 April 1996     

REE Abundaces in Megacrysts and Host Basalts：REE Behavior of Magma at High Pressures

Based on REE abundances in megacrysts and host basalts and their equilibrium conditions,it has proved that megacrysts may have been produced from the magma derived from the host rocks or of more basic composition.The REE ratios of megacrysts to host rocks may be taken as partition coefficients when both are equilibrium with each other.The crystal fractionation of megacrysts has caused the evolution of REE in the magma.It is obvious that some host basalts are the product of magma evolution after crystal fractionation.According to REE abundances in the host rocks and the partition coefficients between crystal and liquid,the history of crystal fractionation of magma can be traced.     

Stoichiometric control of crystal/liquid single-component partition coefficients

Single-component (e.g., single-element) partition coefficients for silicate melts and the minerals that crystallize from them typically show large variations with temperature and melt composition. It is shown that, in particular cases, these variations can be calculated from two-component distribution coefficients that are practically constants, given the structural formula of the mineral and the melt composition.     

Distribution of groups I and II elements between liquidus phases of fluorine-saturated Si-Al-Na-K-Li-H-O system

Experimental data allow modeling the behavior of the named elements during formation of fluorine- saturated leucocratic rocks of silicic and alkaline compositions. The distribution of alkaline and alkaliearth elements is discussed at equilibrium between the silica-alumina melt with fluoride phases (crystalline and liquid) and with feldspar. Cryolite crystals form during saturation of silica-alumina melt of normal alkalinity with fluorine. Continuous solid solution of sodium-potassium cryolite is stable at 800°C. The equilibrium between melt and crystals continues up to the maximum molar fraction of 0.1 lithium end member in cryolite, at which two fluoride phases (crystalline and liquid) coexist with the silica-alumina melt of fixed composition. Separation of salt melts during late differentiation stages of granite and alkaline rocks is a regular process continuing the natural evolution of ore-magmatic systems. At equilibrium of two liquid phases, the silica phase is relatively enriched in potassium, and the fluoride phase is substantially enriched in sodium. This detected effect is the only currently possible mechanism for the occurrence of the potassium differentiation trends of granite melts. All effects related to crystallization cause enrichment in sodium. In other cases (with Ca, Sr, Mg, Rb, and Cs), separation of the second liquid phase acts in the same direction and enhances the action of crystallization. Comparison between partition coefficients allows derivation of the following affinity rows of alkaline elements for fluoride melt: Li > Na > K > Rb≈Cs and Mg > Ca > Sr > Ba. Hence, the known rule for joining strong bases with strong acids and weak bases with weak acids is fulfilled.     

Trace element partitioning between mica- and amphibole-bearing garnet lherzolite and hydrous basanitic melt: 1. Experimental results and the investigation of controls on partitioning behaviour

Thirty five minor and trace elements (Li, Be, B, Sc, Cu, Zn, Ga, Ge, As, Rb, Nb, Mo, Ag, Cd, In, Sn, Sb, Cs, Ba, La, Ce, Nd, Sm, Tb, Ho, Tm, Lu, Hf, Ta, W, Tl, Pb, Bi, Th and U) in experimentally produced near-liquidus phases, from a primitive nelpheline basanite from Bow Hill in Tasmania (Australia), were analysed by LAM ICP-MS. A number of halogens (F, Cl and I) were also analysed by electron microprobe. The analyses were used to determine mineral/melt partition coefficients for mica, amphibole, garnet, clinopyroxene, orthopyroxene and olivine for conditions close to multiple saturation of the basanite liquidus with garnet lherzolite (approximately 2.6 GPa and 1,200°C with 7.5 wt% of added H₂O). A broader range of conditions was also investigated from 1.0 GPa and 1,025°C to 3.5 GPa and 1,190°C with 5–10 wt% of added H₂O. The scope and comprehensiveness of the data allow them to be used for two purposes, these include the following: an investigation of some of the controlling influences on partition coefficients; and the compilation of a set partition coefficients that are directly relevant to the formation of the Bow Hill basanite magma by partial melting of mantle peridotite. Considering clinopyroxene, the mineral phase for which the most data were obtained, systematic correlations were found between pressure and temperature, mineral composition, cation radius and valence, and ΔG ^coulb (the coulombic potential energy produced by substituting a cation of mismatched valence into a crystallographic site). ΔG ^coulb is distinctly different for different crystallographic sites, including the M2 and M1 sites in clinopyroxene. These differences can be modelled as a function of variations in optimum valence (expressed as 1 sigma standard deviations) within individual M1 and M2 site populations.     

The dependence of partition coefficients on differences of ionic volumes in crystal-melt systems

Plots of partition coefficients versus differences of ionic volumes have been constructed from reported values in olivine-, pyroxene-, hornblende-, biotite-, and plagioclase-groundmass systems. The trend behaviour of partition coefficients as a function of the differences of ionic volumes is largely mineralogically controlled. The slopes of the trend lines can be related to the volume work resulting either from (i) expansion of the host lattice surrounding the substituting ions or (ii) collapse of the surrounding lattice if small ions are incorporated. When the volume work is in the order of the enthalpy of melting, ion selectivity ceases.If melts change from basic to highly acidic the chemical control of element partitioning by complex formation in the fluid phase is recognizable. Mineralogical versus chemical control is discussed by means of the trends of the partition coefficients of REE.The proposed ln D vs. r ³ plots improve the physical interpretation of element partition, delineates methods for interpolation or extrapolation of unknown data. Deviations of extreme data from linear relations demonstrate the constraints of our knowledge on diadochy. Rules of element partition as recently proposed by Philpotts (1978) are critically discussed.     

LA-ICPMS analyses of silicate melt inclusions in co-precipitated minerals: Quantification, data analysis and mineral/melt partitioning

We present a new approach to determine the composition of silicate melt inclusions (SMI) using LA-ICPMS. In this study, we take advantage of the occurrence of SMI in co-precipitated mineral phases to quantify their composition without depending on additional sources of information. Quantitative SMI analyses are obtained by assuming that the ratio of selected elements in SMI trapped in different phases are identical. In addition Fe/Mg exchange equilibrium between olivine and melt was successfully used to quantify LA-ICPMS analyses of SMI in olivine. Results show that compositions of SMI from the different host minerals are identical within their uncertainty. Thus (1) the quantification approach is valid; (2) analyses are not affected by the composition of the host phase; (3) the derived melt compositions are representative of the original melt, excluding significant syn- or postentrapment modification such as boundary layer effects or diffusive reequilibration with the host mineral. With this data we established a large dataset of mineral/melt partition coefficients for the investigated mineral phases in hydrous calc-alkaline basaltic-andesitic melts. The clinopyroxene/melt and plagioclase/melt partition coefficients are consistent with the lattice strain model of Blundy and Wood [Blundy, J., Wood B., 1994. Prediction of crystal-melt partition-coefficients from elastic-moduli. Nature372, 452-454].     

Trace element distribution coefficients in alkaline series

Mineral/groundmass partition coefficients for U, Th, Zr, Hf, Ta, Rb, REE, Co and Sc have been systematically measured in olivine, clinopyroxene, amphibole, biotite, Ti-magnetites, titanite, zircon and feldspars, in basaltic to trachytic lavas from alkaline series (Velay, Chaîne des Puys: Massif Central, France and Fayal: Azores). Average partition coefficients are denned within the experimental uncertainty for limited compositional ranges (basalt-hawaiite, mugearites, benmoreite-trachyte), and are useful for trace element modelling. The new results for U, Th, Ta, Zr and Hf partition coefficients show contrasting behaviour. They can thus be used as “key elements” for identifying fractionating mineral phases in differentiation processes (e.g. Ta and Th for amphibole and mica).Partition coefficient may be calculated using the two-lattice model suggested by Nielsen (1985). Such values show a considerably reduced chemical dependence in natural systems, relative to weight per cent D values. The residual variations may be accounted for by temperature or volatile influence. This calculation greatly enhances modelling possibilities using trace elements for comparing differentiation series as well as for predicting the behaviour of elements during magmatic differentiation.     

Melting equilibria in multicomponent systems and liquidus/solidus convergence in mantle peridotite

The melting of undepleted mantle peridotite proceeds through a temperature interval which decreases with increasing pressure. If liquidus and solidus actually meet in the range 100–150 Kb, as suggested by Herzberg (1983), peridotite must transform there directly to a melt of its own composition. Thermodynamic analysis shows that such a liquidus/solidus meeting would be very unlikely in a system as chemically complex as mantle peridotite and would require that unanticipated phase equilibrium relations suppress all incongruent melting behavior. But Takahashi and Scarfe's (1985) preliminary experiments suggest that the upper mantle itself may indeed have a special composition with respect to phase equilibrium relations between liquids and solids at very high pressure. If so, mantle peridotite composition cannot be generated as a crystal accumulate or melting residue, because these two popular theories of origin are difficult to reconcile with a supposed eutecticlike composition. If upper mantle peridotite were itself a solidified liquid composition produced either as a partial melt or, more likely, as a crystallization residue of some more primitive melt composition representative of the whole mantle, an approach of liquidus to solidus might be expected at high pressure although the phase relations of Herzberg (1983) and Herzberg and O'Hara (1985) remain implausible.     

Nature of the chemical heterogeneity of the continental lithospheric mantle

The paper reports data on the chemical composition of mantle peridotite xenoliths from kimberlites and alkaline basalts that represent the continental lithospheric mantle (CLM) beneath Early Precambrian and Late Proterozoic-Cenozoic structures, respectively. In order to identify compositional trends during the melting of primitive material and propose the most reliable criteria for constraining the conditions of this process and its degree, we analyzed literature data on the melting of spinel and garnet peridotites within broad temperature and pressure ranges. It was determined that the degree of melting (F%) of pristine peridotite of composition close to that of the primitive mantle (PM) can be deduced from the Mg/Si and Al/Si ratios in the residue; an equation was proposed for evaluating F from the Mg/Si ratio. The Ca/Al ratio of residues at low (1–1.5 GPa) pressures and degrees of melting from 2–3 to 20–25% increases several times but decreases with increasing F at pressures higher than 3 GPa. The Na partition coefficient between melt and residue decreases at increasing pressure and approaches one at a pressure close to 20 GPa. Residues after low-degree melting are strongly depleted in Ti, Zr, Y, and Nb but are enriched in Cr. The application of these criteria to the composition of xenoliths brought to the surface from the mantle occurring beneath tectonic structures of various age led us to conclude that compositional heterogeneities of CLM (particularly the variations in the concentrations of major and certain siderophile elements) are controlled, first of all, by the melting of the mantle source material. These processes occurred under various thermodynamic conditions (T, P, and $ f_{O_2 } $ f_{O_2 } ) and differed in their intensity, and this predetermined the compositional diversity of the residual mantle material (its concentrations of Mg, Al, Si, Ca, Na, K, Ni, Co, V, and Cr). Our results are principally consistent with the hypothesis of the global magmatic ocean. It is thought that the early phases of its consolidation were variably controlled by the fractionation of minerals, for example, majorite. Moreover, heterogeneities in the distribution of siderophile elements could be partly predetermined by changes in the properties of these elements at ultrahigh temperatures and pressures. The processes of partial melting were the most intense during the early evolution of the mantle (perhaps, in the Early Precambrian), and hence, the mantle has different chemical composition beneath Archean cratons and Phanerozoic foldbelts.     

Trace element behaviour during anatexis in the presence of a fluid phase

If anatexis takes place in the presence of an immiscible volatile-rich fluid phase, the behaviour of a trace element depends not only on partition coefficients D^{i ?} between mineral i and silicate liquid, but also on coefficients D^{i f} between mineral i and fluid. The limited experimental data available, as well as theoretical arguments, suggest that for common minerals these coefficients differ, in some cases (REE) notably.Theory has been developed to permit calculation of concentration variations with the fraction F of rock melted, if D-values are known. In its present state this theory neglects the influence of T, P and composition variations in liquid and fluid on the D-values: if such variations can be expressed as functions of F, appropriate modifications can be made. The theory also neglects the effects of T and P on the solubility of fluid in liquid and the resulting modifications to phase petrology, by assuming simply that a given mass of rock melts in the presence of a mass proportion v of fluid. By choosing different values of v, from 0 (dry melting) to 100 (large fluid excess), the response of the trace element concentrations can be followed, beginning with concentration c_o in the unmelted rock. Such treatment is highly idealised, but serves to indicate some limits on what can occur.     

Partition coefficients for olivine-melt and orthopyroxene-melt systems

Thermodynamic analysis shows that olivinemelt and orthopyroxene-melt partition coefficients for many elements should be approximately linear functions of D_Mg. These simple relationships can be combined with the constraint of mineral stoichiometry to allow the direct calculation of partition coefficients for these elements if the major element chemistry of the melt phase is known. A large dataset of published and unpublished experimental mineral-melt pairs for compositions in the range komatiite to andesite has allowed the determination of the empirical constants required for this calculation. The precision of these parameterisations is demonstrated by comparing the values calculated with those observed. Comparison of phenocryst-matrix partition coefficients with those measured from experimental mineral-melt pairs demonstrates that experimentally determined partition coefficients are equivalent to those in magmatic processes. There are therefore no significant kinetic factors precluding magmatic partitioning being reproduced on an experimental timescale. The model provides a set of simple tests for equilibrium and enables the chemical evolution of a magma fractionating olivine or orthopyroxene to be modelled. An empirical equation for distinguishing orthopyroxene from other low-Ca pyroxenes in chemical analyses of experimental runs is also presented.     

Some trace element relationships among liquid and solid phases in the course of the fractional crystallization of magmas

The classical equations relating the trace element concentrations of the liquid and solid phases coexisting in the simple fractional crystallization of a parental magma have been put in a simple graphical form, which allows rapid analysis of the possible genetic relationships in a given rock suite. The effects of an incomplete separation between the two phases are taken into account. The approach does not require the use of otherwise estimated partition coefficients. Trace element data concerning the minerals of cumulates, where available, may provide an independent estimation of the effective mineral-liquid partition coefficients. With reasonable assumptions, this approach may even be applied to plutonic rocks. Interpretation of the published rare earth element data from the Southern California Batholith by this procedure suggests that a tonalitic parental magma could generate a granodioritic liquid by crystallizing 40–50 wt % of a solid residue of gabbroic composition, in agreement with Larsen's (Mem. Geol. Soc. Amer. 29, 1948) calculations. The calculated mineral-liquid partition coefficients for the REE fall in the range of published phenocryst-groundmass values for acidic volcanic rocks.