The main results of some previous work of the author on this subject are rederived in a more rigorous fashion. In particular, no appeal is made to perturbation theory or to the author's "damping operator" formalism, and closer attention is paid to the question of proper averaging over configurations of absorbers. All effects due to translational motion of absorbers are neglected, but the dipolar "resonance" interactions are included. It is shown that the formal results of the previous work, which permit a detailed calculation of the absorption line shape, are valid under the following assumptions: (a) The average optical behavior of the gas is describable in terms of a refractive index; (b) there is no correlation between the positions of different absorbers, i.e., the gas is ideal. However, it is also shown that the cutoff procedure for handling a divergent integral which appears in the theory must be modified. The modified cutoff procedure leads to a large increase in the theoretically predicted linewidth, in qualitative agreement with experiment.