The practice of making a comparison between the mortality actually experienced by a life assurance society during a given period of time, and that which might have been expected (to use the term in its professional sense) according to the mortality table on which the periodical valuations are based, has now become so general, that no apology is needed for drawing further attention to the subject, if it can be shown that the method usually employed at present is not altogether satisfactory.

What is the best method to use in the graduation of a mortality table, although not a new question, is one which always must have a practical interest for the profession; and as the various methods of graduation which have been devised from time to time can all boast of their respective admirers, the subject is one which cannot easily be disposed of at present. In the following paper it is not proposed to open up any discussion on the general principles involved, nor is it proposed to go into any detail as to how far these general principles have been carried out in the various well known methods of graduation. The purpose of the paper is of a somewhat different nature.

In the first part of this paper, it has been shown to what extent the reserve under a single policy is affected by various methods of graduating the mortality experience on which such reserve is based. We will now endeavour to obtain some notion of the financial effect upon the total reserve made by an office at a periodic valuation of its liabilities.

A satisfactory anser to the question I hav placed at the hed of this articl can only be got by means of an examination of statistics carefully compiled. If we coud obtain statistics shoing the rate of mortality from small-pox among two bodies of persons, similarly circumstanst as to ages, occupations, conditions of life, and other things; and differing only to this extent, that all the persons in the one body had been vaccinated, and none at all in the other body; then an examination of such statistics woud giv a conclusiv anser to the question. But such statistics ar clearly unattainabl. Those that ar forthcoming wil be more or less deficient in the accuracy we hav supposed; and our conclusions will therefore only hav a greater or less degree of probability. acording as the statistics ar more or less complete and accurat.