16

sidered safe or prudent to count on a materially lower mortality experience than obtains in the general population.

If, then, there could be an accurate count kept of the persons in a nation entering upon a certain year of age and then an accurate count of those who die during that year of age, a reasonably reliable death-rate for that age could be computed. This would have to be repeated for each age and, when the work was completed, we would have the death-rate for each age. From this we could readily construct a mortality table, beginning with the assumption of any large number of persons, newly born, that we choose, and diminishing that number annually by applying these respective death-rates.

The ideal method of making a mortality table, if it were possible, would be to take a large number of per-sons under observation from their birth and keep ac-count of the diminution of their number each year by death. This is not practicable, however, for one could not be sure to keep them all under observation. But, if one has already ascertained, by observing a large number at each age, what the rate of mortality is at each age, he can, as has been said, construct a table which will show how the lives would fail, quite as well as if he had been able to keep the large number of persons under observation from their births.

The first trace of anything like a mortality table dates back to about 364, When Ulpian, a Roman Praetorian prefect, introduced a table of probable life, in order to enable annuities upon lives to be valued. Upon what data it was based is unknown.

John de `'Fitt, the famous Grand Pensioner of Holland