Spiegelman and Kenyon [1992] explored the controls of melt channel spacing, relative velocities of migrating melt and upwelling mantle, and solid state diffusion rates on the extent to which melts moving through a mantle matrix re-equilibrate. They demonstrated that for geologically reasonable choices for velocities and diffusion rates, the extent to which melts re-equilibrate as they percolate through mantle beneath a spreading center depends on the melt channel spacings, and that spacings of the order of a few grain sizes will preclude re-equilibration. In an effort to explain ``strong trace element evidence for melting at great depths, where garnet is a stable phase in mantle peridotite'' in many MORB, Hart [1993] showed that a fractal magma ``tree'' may provide a network in which magma rapidly loses diffusive chemical contact with its host matrix. Such a magma tree would be capable of extracting and focussing melts from a great depth without resulting in chemical re-equilibration. In an effort to model U-series disequilbrium in MORB, Spiegelman and Elliott [1993] showed that melt transport with interaction between the solid and liquid can have large effects on the abundances of short-lived radionuclides by changing the residence times of parent and daughter elements in the melting regime.
Plank and Langmuir [1992] showed that while changing the shape of the melting regime has little effect on the trace element characteristics of the oceanic crust, changing the way the mantle flows through the melting regime and the relationship between melt fraction and pressure during adiabatic melting are both important parameters for controlling the resulting trace element composition of the oceanic crust. Forsyth [1994] explored relationships between crustal thickness, degree of partial melting and average depth of melting resulting from simple models of fractional melting induced by passive upwelling beneath ocean ridges.