Abstract
It is suggested that for electrons in a non-crystalline solid the random-phase model is correct for the wavefunctions at a mobility edge when this lies in midband, and can be modified to apply to the case when it lies near the band edge.