The conclusion by Tanner that trenches and island arcs are caused by primary regional tension is contradicted by various lines of evidence. Earthquake data show these features located above Benioff zones where complex tectonic processes take place, including tensional as well as compressional and strike-slip motions; but regional geologic data show compression to be dominant, as in the South American Andes facing the Pacific Ocean.

Diapirs, including those responsible for volcanoes, are caused by rise of buoyant plastic or viscous material when the overlying denser material is stressed beyond the yield point. This process can take place in areas subject to compression as well as in areas subject to tension; in fact, shale diapirs and associated mud volcanoes are found almost exclusively in areas subject to lateral compressive stress at depth. Volcanoes along the inside of island arcs are in stress fields similar to those of the Andes where compression is dominant.

The conclusion that there is no downgoing slab in Benioff zones is based on a model of vertical density distribution invalid for the upper mantle because it makes insufficient allowance for the character of the low-velocity seismic zone in the asthenosphere. More realistic models show density inversions in the upper mantle and consequent gravitational instability which favors convection. A diagrammatic model has been constructed on this basis to demonstrate the feasibility of the convection process. Assumptions made in constructing this model generally are accepted as reasonable. Actual conditions are obviously of vastly greater complexity and involve different conditions than visualized here.