Abstract
Afonin, Gal'perin and Gurevich have shown that in disordered metals quantum interference between two multiple-scattering paths can occur, in spite of inelastic collisions, if the energy loss in the latter is small enough. This paper restates and seeks to explain the evidence given by Mott that quantum interference is absent in liquid metals in the regime where, if present, it should be observable, namely that for which conductivity is less than the Ioffe-Regel value of about 0·3 e2/ha, 3000ω−1 cm−1. Our explanation assumes that collisions in liquids are inelastic, with a transfer of energy to the ions of order hw (that of a short-wave phonon), and depends essentially on the fact that the Fermi surface is very diffuse, which means that a large number of inelastic collisions occur in the quantum interference process. These considerations in no way invalidate the Ziman theory of the resistivity.