I have somewhere read that the experience of life assurance companies furnishes the only certain means of ascertaining the law of mortality, since by this means only is it possible to keep in view through a lengthened period and trace up to death, each of a selected number of persons. As far as the possibility is concerned, we may admit that the law of mortality of the members may be deduced in this manner by proper methods. It is important, however, to bear in mind that, in the case of assurances, we have to deal with selected lives; and the influence of this selection, though it may be disregarded in the later years of assurance, is of considerable importance in the earlier years, so that, unless special precautions be taken, the mortality of the general population cannot be correctly obtained. The mortality may be more nearly determined if an account be kept of the rejected proposals, as is done in England, and the deaths noted which happen among those cases. This, I certainly think, should be done universally, as it must be of practical use to the companies to know how far the rules by which they are guided in accepting or rejecting proposals are justified. No such account is kept, as far as I know, either in Germany or Austria. We may otherwise eliminate the influence of selection, by omitting from the calculation the earlier years of assurance, during which this influence is felt.