This paper presents the results of the first quantitative petrological modelling of abyssal peridotites. The mantle beneath a ride may be considered as two regions: (1) the melting region between the solidus (P-o) at which upwelling mantle begins to melt, and the depth (P-f) at which melting stops because of conductive cooling to the surface; (2) the thermal boundary layer between P-f and the base of crust. In the melting region, decompression near-fractional melting is characterized by the reaction aCpx + bObx + cSpl = dOl + 1Melt, i.e. clinopyroxene, orthopyroxene and spinel melt whereas olivine crystallizes as melting proceeds. In much of the pressure range (P-o less than or equal to 25 kbar), orthopyroxene contributes more than clinopyroxene to the melt (i.e. b > a) during decompression melting, which is unexpected from isobaric melting experiments, but is constrained by the incongruent melting of Opx double right arrow Ol + SiO2 with decreasing pressure. The melting reaction also explains the so-called local trend of mid-ocean ridge basalt (MORB) chemistry characteristic of slow-spreading ridges. Melts produced over a wide region and depth range in the mantle will ascend and migrate laterally towards the axial zone of crustal accretion. These melts cool and crystallize olivine as they pass through previously depleted residues in the thermal boundary layer. This explains why abyssal peridotites have excess olivine relative to simple melting residues. The greater the ambient extent of mantle melting, the more melt is produced in the mantle. Thus, greater extents of melting lead to more olivine (up to 50% of the rock mass in abyssal peridotites) crystallization at shallow levels. Additional important implications are: (1) neither MORB melts nor the bulk igneous crust is compositionally comparable with partial melts produced in peridotite melting experiments because primary mantle melts crystallize olivine back in the mantle; (2) diffusive porous flow is the primary mode of melt migration even at very shallow levels because excess olivine is observed on thin-section scales in abyssal peridotites; (3) low-pressure melt equilibration during ascent is inevitable because the melting reaction preserved in residual peridotites requires continuous solid-liquid equilibration, and because olivine crystallization in the thermal boundary layer is the natural consequence of melt ascent and cooling; (4) perfect fractional melting is unlikely because melt porosity (a few percent?) in the melting mantle is required by the melting reaction, whole-rock major element data and other observations; (5) compositional variations of both MORB and abyssal peridotites are consistent with varying extents (similar to 10-22%) of mantle melting beneath global ocean rides.
