This paper presents the results of some speculations concerning the existence and geologic effects of convection‐currents in the Earth's substratum, the so‐called silicate shell extending according to the best evidence, some 3,000 km below the bottom of the Earth's outer crust. The possibility of the existence of such currents has been amply demonstrated by the calculations of Pekeris [see 1 of “References” at end of paper] and others. It is quite true that the mere existence of an unstable temperature‐gradient will not of itself cause convection provided the substratum has a fundamental strength, but, as Pekeris has shown, a horizontal temperature‐variation of several tens of degrees would set up sufficient stress to overcome the fundamental strength. In the present work we have made a somewhat more careful estimate of the stresses existing in an elastic body due to various types of thermal perturbations, and the results confirm Pekeris almost exactly, namely, the maximum stress is about 5 kg/cm2 per degree of horizontal temperature‐difference, so that even with an assumed strength of 100 kg/cm2, convection would start with a temperature‐contrast of about 20°, such as could easily be set by irregularities in the cooling of the Earth.