The object of the paper is to determine the extent to which the differences between the ultimate rates of mortality for the whole life and endowment assurance classes are attributable to class selection.

The effect on the ultimate rates of mortality of the different distributions of the exposed to risk in the two classes is examined on the basis of common “select” rates of mortality derived from the combined experience.

While the effect is found to be appreciable, there remains a substantial proportion of the differences between the actual ultimate rates of mortality unaccounted for, and the conclusion is reached that this unaccounted balance constitutes a measure of the effect of class selection.

Graduations of the separate class experiences with reference to the official combined rates are given in the paper.