An olivine and spinel phyric from Okmok has Fo92.9 olivine cores, Mg/(Mg+Fe2+)(Mg♯)=0.79 after correction for olivine accumulation, and a calculated liquidus temperature of 1409¿ C. A basalt which is also olivine and spinel phyric is different from the picrite in most aspects of trace element and isotope geochemistry and has Mg♯=0.72 and a liquidus temperature 1286¿ C. Five basalts are fractionated but have trace element and isotope chemistries in between those of the two inferred primary magmas. All the lavas have Pb, Nd, Sr, and O isotopic ratios, and interelement high field strength element (HFSE) ratios similar to enriched oceanic basalts (EMORB) which are transitional between depleted mid-ocean ridge basalts. (NMORB) and ocean island basalts (OIB), although elemental concentrations of HFSE are lower in the primary Okmok magmas than in primary EMORB. Ratios of heavy rare earth element (HREE), and light REE (LREE), large ion lithophile element (LILE) to HFSE are progressively higher than EMORB ratios.

Those samples most removed from the EMORB signature have the lowest 206Pb/204Pb. These features suggest progressive enrichment of an EMORB-like mantle with an incompatible element-rich component. In the final mantle source, 80--90% of the LILE and 30--40% of the LREE have been added. Given this extent of metasomatism and the isotopic ratios of the lavas, we see little evidence for significant sediment involvement and suggest that the metasomatic fluid is derived from a lower portion of the slab which has undergone little isotopic exchange with seawater. The high liquidus temperatures of the primary liquids, HFSE with concentrations similar to NMORB and relative abundances similar to EMORB, low Ti/V, and high TiO2/Ni in the primary magmas all require high temperatures and high degrees of partial melting of the mantle source. The temperatures required are sufficiently high that induced counterflow and upward migration of hot (1500¿--1600¿ C) mantle is required. Any melting mechanism involving static mantle, or vertically moving diapirs, is specifically prohibited under Okmok. We propose that decompression of the upward moving limb of the induced counterflow is the dominant melting mechanism and that the melting zone is localized by impingement on the subforearc mantle wedge, that is, the melting zone (and thus the location of the volcanic front) is localized for physiocal rather than physichemical reasons. Variable mixing between the static and counter-flowing mantle and a variety of metasomatic fluids yieldss a range of primary magma compositions under Okmok and presumably under arcs in general.