Three lava samples were selected for experimental study. They include Unit 41 from the Middle Series at Site 917 (Sample 152-917A-27R-2 [Piece 4, 38-43 cm]), Unit 12B from oceanic succession at Site 918 (Sample 152-918D-108R-1, [Piece 2B, 54-58 cm]), and Unit 1 from Site 989 (Sample 163-989B-10R7 [Piece 5A, 55-59 cm]). All samples in the following are referred to by their respective section identifications. The samples are all from relatively evolved, hypersthene normative basaltic flows with Mg/(Mg+Fe^total) atomic ratios of <0.54 (Table 1). The phase relations for the Site 989 sample were determined as part of a companion experimental study of nucleation and crystallization kinetics (Lesher et al., Chap. 12, this volume). The phase compositions for these experiments are reported in this study. We also relate the new data to our previous results on the magnesium-rich flows from Site 917, which included Unit 13 from the Upper Series (Section 152-917A-11R-4) and Unit 84 from the Lower Series (Section 152-917A-86R-7) (Thy et al., 1998).

The selected sample from the Middle Series (Section 152-917A-27R-2) is compositionally a basaltic andesite with 5.6 wt% MgO, while the two other samples (Sections 152-918D-108R-1 and 163-989B-10R-7) are basalts with MgO contents of 8-8.2 wt% (Table 1). The basaltic andesite sample (Section 152-917A-27R-2) contains small amounts (~2%) of plagioclase (An_53-65) and trace amounts of augite phenocrysts in a groundmass of fine-grained pyroxene, plagioclase, and mesostasis (Larsen, Saunders, Clift, et al., 1994; Demant, 1998). The sample from the oceanic succession (Section 152-918D-108R-1) is aphyric with <1% plagioclase phenocrysts (An_72-86) and a fine-grained groundmass of plagioclase, augite, Fe-Ti oxide minerals, and mesostasis (Larsen, Saunders, Clift, et al., 1994). The near vent lava from Leg 163 (Section 163-989B-10R-7) is aphyric with a groundmass assemblage of plagioclase, augite, Fe-Ti oxide minerals, and mesostasis (Duncan, Larsen, Allan, et al., 1996; Lesher et al., Chap. 12, this volume). All samples are altered by very low-grade metamorphism and contain secondary assemblage of zeolites and clay minerals (Duncan, Larsen, Allan, et al., 1996; Demant et al., 1998). Despite the secondary replacements, the samples are relatively fresh and, in view of their aphyric to sparsely phyric nature, likely represent magmatic liquid compositions.

The experimental techniques used in the present study are identical to those used by Thy et al. (1998), and only the main points need to be given. The rock samples were ground in a tungsten-carbide shatterbox and subsequently by hand in an agate mortar under acetone to an estimated average grain-size below ~10 µm. The powder was mixed with polyvinyl alcohol and pressed into pellets and dried. These pellets were broken into ~50-mg pieces and fused to an 0.004-in Fe-Pt alloy suspension wire prepared by iron electroplating (Grove, 1981). The powder from the sample taken from Section 163-989B-10R-7 was pressed into 50-mg pellets without using an organic binder. The experimental charges were suspended in a 1-atm. vertical quench furnace at the run temperatures. Temperature was monitored by a Pt/₉₀Pt₁₀Rh thermocouple (S-type) calibrated against the melting point of gold. The furnace atmosphere was controlled to the fayalite-magnetite-quartz oxygen buffer using a CO-CO₂ gas mixture and monitored by a solid ZrO₂-ceramic oxygen probe calibrated against the Ni-NiO reaction. Run duration varied from 8-239 hr, generally increasing with decreasing melting temperature. The experimental products were quenched in air. A summary of the experimental conditions and results are given in Table 2.

The phase compositions of the experimental products were determined using a Cameca SX50 electron microprobe operated with an accelerating voltage of 15 kV and a beam current of 10 nA. Details of the analytical procedures can be found in Thy et al. (1998). A focused beam was used for minerals, while glasses were normally analyzed with a 10-µm broad beam to minimize volatilization of sodium. The small residual volumes of glass in some low-temperature experiments prohibited the use of a broad diameter beam, and these, therefore, were analyzed with a narrow focused beam. An internal glass standard prepared from international rock standard W-1 (Govindaraju, 1989) was analyzed concurrently and used to evaluate analytical precision and accuracy (Thy et al., 1998, table 1). The analyses were screened by compositional and stoichiometric criteria and are considered representative for the experiments. The average glass and mineral compositions for the individual experiments are reported in Table 3, Table 4, Table 5, and Table 6. The iron in liquids has been redistributed between Fe₂O₃ and FeO appropriate for the fayalite-magnetite-quartz oxygen buffer using the equations of Kilinc et al. (1983).

The standard deviations of the replicate analyses are reported in Table 3, Table 4, Table 5, and Table 6, and reflect the homogeneity of the phases. In particular, the results for plagioclase and augite for the low-temperature experiments reflect a high degree of heterogeneity and incomplete reaction of the starting material. Glass and olivine, on the other hand, are generally homogeneous.

The modal proportions of the experimental phases were estimated by weighted mass balance using the compositions of the phases determined by electron microprobe (Bryan et al., 1969). All of the oxides were assigned a weighting factor of 1.0, except SiO₂ and Al₂O₃, which were weighted by 0.4 and 0.5, respectively. Losses of sodium to the furnace gas and under the electron microbeam during analysis were estimated by including Na₂O as a phase in the mass balance calculation. Table 2 presents the analyzed modes and shows that <20% by weight of Na₂O is unaccounted for in a given experimental charge.