Abstract
The concept of the anisotropic relaxation time is critically examined. It is shown that an anisotropic relaxation time can be defined in a rigorous mathematical way with no approximation. This anisotropic relaxation time is related to the relaxation of the electronic distribution toward equilibrium at the moment all applied fields are turned off. The anisotropic relaxation time is not to be interpreted as a true time constant for exponential decay of the electronic distribution, but is instead a weighted average of the true characteristic time constants for the electron system. It is pointed out that two conflicting definitions of the anisotropic relaxation time are currently in use and that this has generated some confusion and inconsistency in the literature.