If the recorded numbers of the population and of the deaths were free from irregularities and errors, the rates of mortality directly derived from them would, if plotted graphically, be capable of being represented by a smooth curve. These condi tions are never fulfilled in actual experience and it is necessary to submit the data, or the rates derived from the data, to a process of adjustment, technically known as graduation.
Sex is another element to be taken into consideration. Female mortality is generally lighter than that of males. Certain special classes of lives are subject to abnormal mortality rates. Race, climate, social status, occupation, degree of urbanization or density of population, housing conditions and other aspects of environment, geographical situation, types of various diseases, are also factors which affect vitality. Of the utmost importance in connection with the experience of assured lives is the variation of mortality according to duration of assurance. Before a life assurance policy is issued, the proposer is usually required to undergo a medical examination or at least to furnish satisfac tory evidence of good health. At entry into insurance the policy holder is, therefore, a "select" life. Select rates of mortality vary according to age at entry and duration of contract, and a "select table" is arranged in a form showing the select rates converging towards and finally running into the "ultimate" rates which vary only with age. When the effects of selection are ignored and all the data irrespective of duration are combined in one mortality experience, the table is designated an "aggregate" table.
Separate tables have sometimes been constructed by excluding the first five years' experience from the data obtained for the aggregate tables, and are called "truncated aggregate" tables.
When it is desired to compare the results of different investi gations of mortality experience, several criteria may be adopted, e.g., (a), the rates of mortality at selected ages throughout the
table, (b) the number of survivors at selected ages, (c) the prob ability of surviving a specified term of years from the attainment of selected ages, and (d) the expectation of life at selected ages. Of these criteria (c) is, perhaps, most generally satisfactory.
One feature which has been disclosed by successive investiga tions of the mortality experience of lives of the same class is the progressive improvement in vitality, the later investigations al most invariably revealing the lighter rates of mortality at all ages. There are no indications of any retardation in this tendency, and in several instances endeavours have been made to forecast its effects, in the one case by a definite modification of the rates of mortality directly deduced from the experience, and in the other by a suitable adjustment of the annuity values.
The Northampton Table was constructed by a method which gave unduly heavy rates of mortality, particularly at the higher ages. It was, however, the only authoritative measure of mortality available for many years, and the assurance companies which perforce had to adopt it as the basis of their scales of premium were enabled to accumulate large profits. The Government adopted it in 18°8 for the grant of annuities, and thereby incurred a loss estimated at two million pounds before discarding it 20 years later. It was gradually superseded by the Carlisle Table, formed by Joshua Milne. This table was a great improvement on previous ones, and the extensive monetary tables which were based upon it are even now referred to.
The compulsory registration of births, marriages and deaths was introduced into England and Wales in 1837. The statistics which have thus become available, taken in conjunction with the successive decennial censuses, have led to the publication of a most valuable series of English Life Tables. The earliest tables were prepared by Dr. William Farr, whose services in the de velopment of the science of vital statistics are pre-eminent.