The experimental observations according to which the reabsorption of the metal begins in the neighborhood of the metal compound interface is explained simultaneously by the germination and the growth of pores and cavities, and by a mechanism analgous to the mill of Frank and Read. The authors' interpretation rests on the same bases as that of the Kirkendall effect, and the same theoretical calculations can be used. The Kirkendall effect is connected with a diffusion in transitory behavior, whereas the authors' phenomenon is observable in permanent behavior. Its effects are therefore much more considerable.