The ocean-floor spreading hypothesis requires that island arc areas be loci of descending mass transport and destruction of oceanic crust. All island arc areas are characterized by volcanic activity and in some, unusually high heat flow values have been measured for hundreds of kilometers behind the arc. Downward mass motion is associated presumably with the sinking of cool and thus relatively denser matter, and high temperature phenomena normally would not be expected where this was occurring. The only feasible means of generating heat in such areas appears to be by frictional dissipation along the seismically active zone of movement (the Benioff zone) between the moving oceanic crust and upper mantle, and the relatively stationary material that it underrides. It has been shown elsewhere that dissipative heating is of the right order of magnitude to explain the observations, but the required shear stresses are rather high; in this paper a detailed thermal structure for the Benioff zone and the regions on either side is presented. The cold descending material heats very slowly and should form a thin cold slab extending to depths of more than 500 km. Within the zone of movement, temperatures are buffered by the melting temperatures of the various components of the oceanic crust that partially fuse to give the main members of the calc-alkaline igneous suite. The thermal structure of the region above the Benioff zone is subject to uncertainty because processes other than lattice conduction (for instance, magmatic activity) are probably involved in the transfer of heat to the surface.
Observed rates of magmatic activity in island arcs require a relatively small degree of fusion of the descending material and probably only the oceanic crust is involved. The siliceous and presumably water-bearing sedimentary material of the oceanic crust is the most important source of material added to island arcs; growth occurs either by physical addition of sedimentary material, which, for mechanical reasons, may not be carried down, or by partial fusion of sedimentary material along the fault zone and by rise of the resultant magmas to the surface. Rates of growth suggested by these processes are compatible with the known ages of arcs. Older arcs should have larger crustal cross-sectional areas than younger ones, if both have been continuously active at similar rates.