REE, Ba, and Sr abundances and Sr, Nd, and Ce isotope ratos of ODP Hole 110-504B basalts

Abundances of rare earth elements (REE), Ba, and Sr and isotopic ratios of Sr, Nd, and Ce were determined for six samples of basalts drilled at Hole 504B on Leg 111 of the Ocean Drilling Program. Analyses found that these basalts are the most depleted in Sr, Ba, and light REE among mid-ocean ridge basalts (MORB); Ba depletion is especially notable. On the other hand, Sr, Nd, and Ce isotopic ratios for basalts from Hole 504B are within the range of typical MORB values.

Geochemistry of Hawaiian volcanics

The "Ko'olau" component of the Hawaiian mantle plume represents an extreme (EM1-type) end member of Hawaiian shield lavas in radiogenic isotope space, and was defined on the basis of the composition of subaerial lavas exposed in the Makapu'u section of Ko'olau Volcano. The 679 m-deep Ko'olau Scientific Drilling Project (KSDP) allows the long-term evolution of Ko'olau Volcano to be reconstructed and the longevity of the "Ko'olau" component in the Hawaiian plume to be tested. Here, we report triple spike Pb isotope and Sr and Nd isotope data on KSDP core samples, and rejuvenation stage Honolulu Volcanics (HV) (together spanning ~2.8 m.y.), and from ~110 Ma basalts from ODP Site 843, thought to be representative of the Pacific lithosphere under Hawai'i. Despite overlapping ranges in Pb isotope ratios, KSDP and HV lavas form two distinct linear arrays in 208Pb/204Pb-206Pb/204Pb isotope space. These arrays intersect at the radiogenic end indicating they share a common component. This "Kalihi" component has more radiogenic Pb, Nd, Hf, but less radiogenic Sr isotope ratios than the "Makapu'u" component. The mixing proportions of these two components in the lavas oscillated through time with a net increase in the "Makapu'u" component upsection. Thus, the "Makapu'u" enriched component is a long-lived feature of the Hawaiian plume, since it is present in the main shield-building stage KSDP lavas. We interpret the changes in mixing proportions of the Makapu'u and Kalihi components as related to changes in both the extent of melting as well as the lithology (eclogite vs. peridotite) of the material melting as the volcano moves away from the plume center. The long-term Nd isotope trend and short-term Pb isotope fluctuations seen in the KSDP record cannot be ascribed to a radial zonation of the Hawaiian plume: rather, they reflect the short length-scale heterogeneities in the Hawaiian mantle plume. Linear Pb isotope regressions through the HV, recent East Pacific Rise MORB and ODP Site 843 datasets are clearly distinct, implying that no simple genetic relationship exists between the HV and the Pacific lithosphere. This observation provides strong evidence against generation of HV as melts derived from the Pacific lithosphere, whether this be recent or old (100 Ma). The depleted component present in the HV is unlike any MORB-type mantle and most likely represents material thermally entrained by the upwelling Hawaiian plume and sampled only during the rejuvenated stage. The "Kalihi" component is predominant in the main shield building stage lavas but is also present in the rejuvenated HV. Thus this material is sampled throughout the evolution of the volcano as it moves from the center (main shield-building stage) to the periphery (rejuvenated stage) of the plume. The presence of a plume-derived material in the rejuvenated stage has significant implications for Hawaiian mantle plume melting models.

Geochemical composition and crystallization conditions of melts from the Mid Atlantic Ridge near Ascension Island (7-11°S)

Phase equilibria simulations were performed on naturally quenched basaltic glasses to determine crystallization conditions prior to eruption of magmas at the Mid-Atlantic Ridge (MAR) east of Ascension Island (7°11°S).The results indicate that midocean ridge basalt (MORB) magmas beneath different segments of the MAR have crystallized over a wide range of pressures (100-900MPa). However, each segment seems to have a specific crystallization history. Nearly isobaric crystallization conditions (100-300MPa) were obtained for the geochemically enriched MORB magmas of the central segments, whereas normal (N)-MORB magmas of the bounding segments are characterized by polybaric crystallization conditions (200-900MPa). In addition, our results demonstrate close to anhydrous crystallization conditions of N-MORBs, whereas geochemically enriched MORBs were successfully modeled in the presence of 0.4-1wt% H2O in the parental melts.These estimates are in agreement with direct (Fourier transform IR) measurements of H2O abundances in basaltic glasses and melt inclusions for selected samples. Water contents determined in the parental melts are in the range 0.04-0.09 and 0.30-0.55 wt% H2O for depleted and enriched MORBs, respectively. Our results are in general agreement (within ±200MPa) with previous approaches used to evaluate pressure estimates in MORB. However, the determination of pre-eruptive conditions of MORBs, including temperature and water content in addition to pressure, requires the improvement of magma crystallization models to simulate liquid lines of descent in the presence of small amounts of water. KEY WORDS: MORB; Mid-Atlantic Ridge; depth of crystallization; water abundances; phase equilibria calculations; cotectic crystallization; pressure estimates; polybaric fractionation

Petrology of basalts from DSDP Hole 83-504B

Basalts from Hole 504B, Leg 83, exhibit remarkable uniformity in major and trace element composition throughout the 1075.5 m of basement drilled. The majority of the basalts, Group D', have unusual compositions relative to normal (Type I) mid-ocean ridge basalts (MORB). These basalts have relatively high mg values (0.60-0.70) and CaO abundances (11.7-13.7%; Ca/Al = 0.78-0.89), but exhibit a marked depletion in compatible trace elements (Cr and Ni); moderately incompatible trace elements (Zr, Y, Ti, etc.); and highly incompatible trace elements (Nb, LREE, etc.). Petrographic and compositional data indicate that most of these basalts are evolved, having fractionated significant amounts of plagioclase, olivine, and clinopyroxene. Melting experiments on similar basalt compositions from the upper portion of Hole 504B (Leg 70; Autio and Rhodes, 1983) indicate that the basalts are co-saturated with olivine and plagioclase and often clinopyroxene on the 1-atm. liquidus. Two rarely occurring groups, M' and T, are compositionally distinct from Group D' basalts. Group T is strongly depleted in all magmaphile elements except the highly incompatible ones (Nb, La, etc.), while Group M' has moderate concentrations of both moderately and highly incompatible trace elements and is similar to Type I MORB. Groups M' and T cannot be related to Group D' nor to each other by crystal fractionation, crystal accumulation, or magma mixing. The large differences in magmaphile element ratios (Zr/Nb, La/Yb) among these three chemical groups may be accounted for by complex melting models and/or local heterogeneity of the mantle beneath the Costa Rica Ridge. Xenocrysts and xenoliths of plagioclase and clinopyroxene similar in texture and mineral composition to crystals in coarse-grained basalts from the lower portion of the hole are common in Hole 504B basalts. These suggest that addition of solid components either from conduit or magma chamber walls has occurred and may be a common source of disequilibrium crystals in these basalts. However, mixing of plagioclase-laden depleted melts (similar to the Costa Rica Ridge Zone basalts) with normal MORB magmas could provide an alternate source for some refractory plagioclase crystals found out of equilibrium in many phyric MORB. The uniformity of major element compositions in Hole 504B basalts affords an ideal situation for investigating the effects of alteration on some major and trace elements in oceanic basalts. Alteration observed in whole-rock samples records primarily two events - a high-temperature and a low-temperature phase. High-temperature phases include: chlorite, talc, albite, actinolite, sphene, quartz, and pyrite. The low-temperature phases include smectite (saponite), epistilbite or laumontite, and minor calcite. Laumontite may actually straddle the gap between the low- and high-temperature mineral assemblages. Alteration is restricted primarily to partial replacement of primary phases. Metamorphic grade, in general, increases from the top to the bottom of Hole 504B (Legs 69, 70, and 83) as seen in the change from a smectiteto- chlorite-dominated secondary mineral assemblage. However, a systematic progression for the interval recovered during Leg 83 is not apparent. Rather, the extent of alteration appears to be a function of the initial texture and fracture density. Variations in whole-rock major and trace element concentrations cannot be attributed convincingly to any differences in alteration observed. Compositional characteristics of the secondary minerals indicated that extensive remobilization of elements has not occurred; local redistribution is suggested in most cases. Thus, the major and trace element signature of these basalts remains effectively the same as the original composition prior to alteration.

(Table 1) Platinum group element abundanves in ODP Leg 135 samples

Ocean Drilling Program Leg 135 backarc basin lavas are characterized by anomalously high Au contents (1.0-11.4 ppb) and strongly fractionated relative platinum group element (PGE) abundances (Pd/Ir ratio, approximately 100). The Rh and Ir contents are very low, ranging from below detection (approximately 0.02 ppb) to 0.08 ppb. The Pd and Pt contents range from <0.3 to 4 ppb. Rh, Pd, and Pt values are consistently and significantly higher in Site 836 and 839 samples relative to those from Sites 834 and 835. Major, trace, and rare earth element (REE) data suggest Sites 836 and 839 have a more pronounced arc signature than Sites 834 and 835. No correlation exists between noble metal abundance and indices of alteration or fractionation (e.g., loss on ignition (LOI), Mg#, and Cr or Ni contents), suggesting that measured values and ratios are primary and reflect characteristics of the mantle source. The evaluation of Leg 135 noble metal data with respect to potential mantle-source components is hindered by the lack of data on magmas derived from such sources. However, analyses of the limited available data for the different magma types suggest that the characteristic enrichment of Leg 135 lavas in Au, relative to Pd and Cu, cannot be derived solely from simple MORB-type or ocean-island-type mantle, or mantle depleted by a previous melt extraction event. The Au-enriched signature of the Lau basin lavas could, however, be produced through the addition of a sedimentary component from the downgoing slab. Separation of Au from the PGE occurs within oceanic hydrothermal systems and gold values of the resultant precipitates are 2-3 orders of magnitude higher than other oceanic crustal components. Even small additions of this component from the downgoing oceanic crust to a supra-subduction zone mantle melt could account for the high mean Au/Pd ratios of the Leg 135 samples (Sites 834 and 835, Au/Pd = 5.04; Sites 836 and 839, Au/Pd = 2.26).

Chemistry of bulk-rocks and minerals of ODP Hole 120-748C

Five hundred meters of a unique Upper Cretaceous Cr-rich glauconitic sequence (Unit III) that overlies a 3-m-thick alkali-basalt flow with underlying epiclastic volcanogenic sediments was drilled at ODP Leg 120 Site 748. The Cr-rich glauconitic sequence is lithostratigraphically and biostratigraphically divided into three subunits (IIIA, IIIB, IIIC) that can also be recognized by the Cr concentration of the bulk sediment, which is low (<200 ppm) in Subunits IIIC and IIIA and high (400-800 ppm) in Subunit IIIB. The Cr enrichment is caused by Cr-spinel, which is the only significant heavy mineral component beside Fe-Ti ores. Other Cr-bearing components are glauconite pellets and possibly some other clay minerals. The glauconitic sequence of Subunit IIIB was formed by reworking of glauconite and volcanogenic components that were transported restricted distances and redeposited downslope by mass-transportation processes. The site of formation was a nearshore, shallow inner shelf environment, and final deposition may have been on the outer part of a narrow shelf, at the slope toward the restricted, probably synsedimentary, faulted Raggatt Basin. The volcanic edifices uncovered on land were tholeiitic basalts (T-MORB), alkali-basaltic (OIB) and (?)silicic volcanic complexes, and ultramafic rocks. The latter were the ultimate source for the Cr-spinel contribution. Terrestrial aqueous solutions carried Fe, K, Cr, Si, and probably Al into the marine environment, where, depending on the redox conditions of microenvironments in the sediment, green (Fe- and K-rich) or brown (Al-rich) glauconite pellets formed. The Upper Cretaceous glauconitic sequence at Site 748 on the Southern Kerguelen Plateau constitutes the transition in space and time from terrestrial to marine, from magmatically active subaerial to magmatically passive submarine conditions, and from a tranquil platform to active rifting conditions.

Trace element contents in zircons from oceanic crust

We present newly acquired trace element compositions for more than 300 zircon grains in 36 gabbros formed at the slow-spreading Mid-Atlantic and Southwest Indian Ridges. Rare earth element patterns for zircon from modern oceanic crust completely overlap with those for zircon crystallized in continental granitoids. However, plots of U versus Yb and U/Yb versus Hf or Y discriminate zircons crystallized in oceanic crust from continental zircon, and provide a relatively robust method for distinguishing zircons from these environments. Approximately 80% of the modern ocean crust zircons are distinct from the field defined by more than 1700 continental zircons from Archean and Phanerozoic samples. These discrimination diagrams provide a new tool for fingerprinting ocean crust zircons derived from reservoirs like that of modern mid-ocean ridge basalt (MORB) in both modern and ancient detrital zircon populations. Hadean detrital zircons previously reported from the Acasta Gneiss, Canada, and the Narryer Gneiss terrane, Western Australia, plot in the continental granitoid field, supporting hypotheses that at least some Hadean detrital zircons crystallized in continental crust forming magmas and not from a reservoir like modern MORB.

Geochemistry of rare earth elements in basalts from Walvis Ridge

Selected basalts from a suite of dredged and drilled samples (IPOD sites 525, 527, 528 and 530) from the Walvis Ridge have been analysed to determine their rare earth element (REE) contents in order to investigate the origin and evolution of this major structural feature in the South Atlantic Ocean. All of the samples show a high degree of light rare earth element (LREE) enrichment, quite unlike the flat or depleted patterns normally observed for normal mid-ocean ridge basalts (MORBs). Basalts from Sites 527, 528 and 530 show REE patterns characterised by an arcuate shape and relatively low (Ce/Yb)N ratios (1.46-5.22), and the ratios show a positive linear relationship to Nb content. A different trend is exhibited by the dredged basalts and the basalts from Site 525, and their REE patterns have a fairly constant slope, and higher (Ce/Yb)N ratios (4.31-8.50). These differences are further reflected in the ratios of incompatible trace elements, which also indicate considerable variations within the groups. Mixing hyperbolae for these ratios suggest that simple magma mixing between a 'hot spot' type of magma, similar to present-day volcanics of Tristan da Cunha, and a depleted source, possibly similar to that for magmas being erupted at the Mid-Atlantic Ridge, was an important process in the origin of parts of the Walvis Ridge, as exemplified by Sites 527, 528 and 530. Site 525 and dredged basalts cannot be explained by this mixing process, and their incompatible element ratios suggest either a mantle source of a different composition or some complexity to the mixing process. In addition, the occurrence of different types of basalt at the same location suggests there is vertical zonation within the volcanic pile, with the later erupted basalts becoming more alkaline arid more enriched in incompatible elements. The model proposed for the origin and evolution of the Walvis Ridge involves an initial stage of eruption in which the magma was essentially a mixture of enriched and depleted end-member sources, with the N-MORB component being small. The dredged basalts and Site 525, which represent either later-stage eruptives or those close to the hot spot plume, probably result from mixing of the enriched mantle source with variable amounts and variable low degrees of partial melting of the depleted mantle source. As the volcano leaves the hot spot, these late-stage eruptives continue for some time. The change from tholeiitic to alkalic volcanism is probably related either to evolution in the plumbing system and magma chamber of the individual volcano, or to changes in the depth of origin of the enriched mantle source melt, similar to processes in Hawaiian volcanoes.

Utility of thallium and geochemical analysis of Mariana arc sediments from ODP Hole 129-801A and 129-801B

We provide the first exploration of thallium (Tl) abundances and stable isotope compositions as potential tracers during arc lava genesis. We present a case study of lavas from the Central Island Province (CIP) of the Mariana arc, supplemented by representative sedimentary and altered oceanic crust (AOC) inputs from ODP Leg 129 Hole 801 outboard of the Mariana trench. Given the large Tl concentration contrast between the mantle and subduction inputs coupled with previously published distinctive Tl isotope signatures of sediment and AOC, the Tl isotope system has great potential to distinguish different inputs to arc lavas. Furthermore, CIP lavas have well-established inter island variability, providing excellent context for the examination of Tl as a new stable isotope tracer. In contrast to previous work (Nielsen et al., 2006b), we do not observe Tl enrichment or light epsilon 205Tl (where epsilon 205Tl is the deviation in parts per 10,000 of a sample 205Tl/203Tl ratio compared to NIST SRM 997 Tl standard) in the Jurassic-aged altered mafic ocean crust subducting outboard of the Marianas (epsilon 205Tl = - 4.4 to 0). The lack of a distinctive epsilon 205Tl signature may be related to secular changes in ocean chemistry. Sediments representative of the major lithologies from ODP Hole Leg 129 801 have 1-2 orders of magnitude of Tl enrichment compared to the CIP lavas, but do not record heavy signatures (epsilon 205Tl = - 3.0 to + 0.4), as previously found in similar sediment types (epsilon 205Tl > + 2.5; Rehkämper et al., 2004). We find a restricted range of epsilon 205Tl = - 1.8 to - 0.4 in CIP lavas, which overlaps with MORB. One lava from Guguan falls outside this range with epsilon 205Tl = + 1.2. Coupled Cs, Tl and Pb systematics of Guguan lavas suggests that this heavy Tl isotope composition may be due to preferential degassing of isotopically light Tl. In general, the low Tl concentrations and limited isotopic range in the CIP lavas is likely due to the unexpectedly narrow range of epsilon 205Tl found in Mariana subduction inputs, coupled with volcaniclastic, rather than pelagic sediment as the dominant source of Tl. Much work remains to better understand the controls on Tl processing through a subduction zone. For example, Tl could be retained in residual phengite, offering the potential exploration of Cs/Tl ratios as a slab thermometer. However, data for Tl partitioning in phengite (and other micas) is required before developing this application further. Establishing a database of Tl concentrations and stable isotopes in subduction zone lavas with different thermal parameters and sedimentary inputs is required for the future use of Tl as a subduction zone tracer.

Major, trace and rare earth element concentration of ODP Site 210-1277 basalts, Newfoundland margin

Drilling of the distal Newfoundland margin at Ocean Drilling Program Site 1277 recovered part of the transition between exhumed sub-continental mantle lithosphere and normal mid-ocean-ridge basalt (N-MORB) volcanism perhaps related to the initiation of seafloor spreading, which may have occurred near the Aptian/Albian boundary, coincident with the final separation of subcontinental mantle lithosphere. Subcontinental mantle lithosphere was recovered near the crest of a basement high, the Mauzy Ridge. This ridge lies near magnetic Anomaly M1 and is inferred to be of Barremian age. The recovered section is dominated by serpentinized spinel harzburgite, with subordinate dunite and minor gabbroic intrusives, and it includes inferred high-temperature ductile shear zones. The serpentinite is capped by foliated gabbro cataclasite that is interpreted as the product of a major seafloor extensional detachment. The serpentinized harzburgite beneath is highly depleted subcontinental mantle lithosphere that was exhumed to create new seafloor within the ocean-continent transition zone. After inferred removal of overlying brittle crust, the detachment was eroded, producing multiple mass flows that were dominated by clasts of serpentinite and gabbro in a lithoclastic and calcareous matrix. Basaltic lavas were erupted spasmodically, mainly as sheet flows, with subordinate lava breccia, hyaloclastite, and possible pillow lava. The sedimentary-volcanic succession and the exhumed mantle lithosphere experienced later high-angle extensional fracturing and probably faulting. Extensional fissures opened incrementally and were filled with silt-sized carbonate, basalt-derived clastic sediment, and hyaloclastite, forming neptunian dykes and geopetal structures. Chemical analysis of representative basalts for major elements and trace elements were made using a high-precision, high-accuracy X-ray fluorescence method (utilizing increased count times) and by whole-rock inductively coupled plasma-mass spectrometry that yielded additional evidence for rare earth elements. The analyses indicate N-MORB to slightly enriched compositions. The MORB was produced by relatively high degree melting of a fertile mantle source that differed strongly from the cored serpentinized peridotites. The basalts exhibit a distinct negative Nb anomaly on MORB-normalized plots that can be explained by prior extraction of melt from upper mantle that had previously been affected by subduction, possibly during closure of the Iapetus or Rheic oceans. In the proposed interpretation, mantle lithosphere was exhumed to the seafloor and experienced mass wasting to form serpentinite-rich mass flows. The interbedded MORB records the beginning of a transition to "normal" seafloor spreading. This interpretation takes into account drilling results from the Iberia-Galicia margin and the Jurassic Alps-Apennines.

Mineral composition of mid-ocean-ridge basalts from ODP Leg 187 sites

The chemical compositions of olivine, plagioclase, pyroxene, and spinel in lavas collected during Ocean Drilling Program Leg 187 in the Australian Antarctic Discordance, Southeast Indian Ridge (41°-46°S, 126°-135°E) were analyzed, and modeling of the theoretical equilibrium petrogenetic conditions between olivine and melt was conducted. The cores of larger olivine phenocrysts, particularly in the isotopic Indian-type mid-ocean-ridge basalt (MORB), are not equilibrated with melt compositions and are considered to be xenocrystic. Larger plagioclase phenocrysts with compositionally reversed zonation are also xenocrystic. The compositions of primary magma were calculated using a "maximum olivine fractionation" model for primitive MORB that should fractionate only olivine. Olivine compositions equilibrated with calculated primary magma and compositions of calculated primary magma suggest that (1) isotopic Pacific-type MORB is more fractionated than Indian-type MORB, (2) Pacific-type MORB was produced by higher degrees of partial melting than Indian-type MORB, and (3) primary magma for Indian-type MORB was segregated from mantle at 10 kbar (~30 km depth), whereas that for Pacific-type MORB was segregated at 15 kbar (~45 km depth).

Geochemistry of basalts from ODP Hole 109-648B

Legs 106-109 achieved the first basaltic bare-rock drill hole, on a small volcano (Serocki volcano) located on the flanks of the rift valley in the MAR about 70 km south of the Kane fracture zone. Because of severe technical difficulties only 50.5 m of basalt below seafloor was recovered. Geochemical analysis shows that the recovered basalts display typical N-MORB characteristics as expected in this segment of the Mid-Atlantic ridge. The lava flows display rather equivalent geochemical characteristics all over the drilled section and show the imprint of a previous magmatic differentiation suffered by the magmas before their emission, indicative of a fractional crystallization of plagioclase-bearing cumulates. The incompatible and alkali element content of these 648B magmas is very low, a feature which resembles those of other N-MORB. The geochemical characteristics of these basalts look closely similar to those of basalts from the same flow line emitted respectively 10 m.y. (Hole 395, Legs 45-46), and 110 m.y. (Hole 417A, Legs 51-53) ago, supporting the persistence in this ridge segment of a mantle source with depleted characteristics over the last 110 m.y., but with some variations in the degree of depletion of the source along this period. Although these rocks appear fresh, the imprint of an incipient low temperature alteration can be noticed in a few samples, as evidenced by slight increases of alkali, U elements, and 87Sr/86Sr isotopic compositions.

Sr, Nd, and Pb isotope geochemistry of ODP Hole 104-642E

ODP Leg 104 recovered 914 m of volcanics at Site 642 on the Vøring Plateau in the Norwegian Sea. The upper series of these volcanics correlates with seaward-dipping seismic reflectors (DRS), and is tholeiitic in character. The lower series underlies the DRS and is broadly andesitic in character. Rb-Sr, Sm-Nd, and Pb isotopic analyses show that upper series samples have isotopic features characteristic of MORB, except for one dike sample that has a Pb isotopic composition that may indicate interaction of its parent magma with older continental crust. The five most silicic samples from the lower series, which occur high up in the sequence, define a 63 ± 19 Ma Rb-Sr whole-rock isochron age, and have an initial 87Sr/86Sr of 0.7116 ± 0.0004. Other lower series samples have lower initial 87Sr/86Sr, but all are greater than any upper series rock. The combined evidence of initial 87Sr/86Sr, initial epsilon-Nd values, Sm-Nd model ages, Pb isotopic compositions, and petrographic features clearly indicate that lower series rocks were derived, at least in part, from continental crustal source materials. That the DRS is underlain by rocks of continental character is an important observation, constraining models for the development of DRS-type passive continental margins.

Geochemistry and oxidation state of Lau Basin basalts

The backarc glasses recovered during Ocean Drilling Program Leg 135 are unique among submarine tholeiitic glasses with respect to their oxygen fugacity and sulfur concentrations. Unlike mid-ocean-ridge basalt glasses, fO2 in these samples (inferred from ratios Fe3+/Fe2+) is high and variable, and S variations (90-1140 ppm) are not coupled with FeO concentration. Strong correlations occur between the alkali and alkaline-earth elements and both fO2 (positive correlations) and S concentrations (negative correlations). Correlations between fO2 and various trace elements are strongest for those elements with a known affinity for hydrous fluids (perhaps produced during slab dehydration), suggesting the presence of a hydrous fluid with high fO2 and high alkali and alkaline earth element concentrations in the Lau Basin mantle. Concentrations of S and fO2 are strongly correlated; high fO2 samples are characterized by low S in addition to high alkali and alkaline earth element concentrations. The negative correlations between S and these trace elements are not consistent with incompatible behavior of S during crystallization. Mass balance considerations indicate that the S concentrations cannot result simply from mixing between low-S and high-S sources. Furthermore, there is no relationship between S and other trace elements or isotope ratios that might indicate that the S variations reflect mixing processes. The S variations more likely reflect the fact that when silicate coexists with an S-rich vapor phase the solubility of S in the silicate melt is a function of fO2 and is at a minimum at the fO2 conditions recorded by these glasses. The absence of Fe-sulfides and the high and variable vesicle contents are consistent with the idea that S concentrations reflect silicate-vapor equilibria rather than silicate-sulfide equilibria (as in MORB). The low S contents of some samples, therefore, reflect the high fO2 of the supra-subduction zone environment rather than a low-S source component.