Abstract
The Mulliken and the Sklar approximations for difficult molecular integrals have for some years been the subject of several investigations. So far, however, we are not provided with clear insight into these approximate methods and, accordingly, with sufficiently reliable criteria for the choice between them for each individual case.
In this paper a possible answer to the problem is presented by examining carefully the accuracies of these two methods. In addition, a new approximate method is proposed, which is useful when both the Mulliken and the Sklar approximations break down seriously.