Abstract
It is shown that in predicting the temperature dependence of multiphonon absorption one must take the temperature dependence of the phonon frequencies into account. Inclusion of the latter leads to results in improved agreement with experiment, as compared to those of Hardy and Agrawal.