THERE is a well-known saying of Lord Kelvin that "When you can measure what you are speaking about and express it in numbers, you know something about it, but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind." Those who were interested in mortality in the seventeenth, eighteenth and early nineteenth centuries were in the unsatisfactory situation which Kelvin had in mind; they could not measure what they were speaking about. The means of measurement were provided outstandingly by William Farr, who "found medical statistics in such a state that only men of genius like Graunt could use them at all without reaching false conclusions, and when he [Farr] retired from the public service, left them so organized and arranged that any reasonably intelligent man could draw accurate conclusions from thern'Yl), On the foundations laid by Farr many men have built. It is unnecessary-indeed, it would be inappropriate-for me to attempt to trace the development of mortality statistics down to the present day. I hope, however, that I may take this opportunity of referring to the work of the present Medical Statistician of the General Register Office, Dr. Percy Stocks, upon whose studies in mortality statistics so much reliance is placed by all whose work requires them to venture into this field. An immense amount of information, remarkably detailed when one realizes the small number of facts which are recorded on death certificates, has been and still is extracted by Dr. Stocks year after year, and the tables he produces have formed the basis of published works by many people-and perhaps one or two reputations.Mortality statistics, however, greatly as they have grown in accuracy and in range of detail in the past century, do not and, by definition, cannot remain the only source of knowledge concerning the prevalence of disease in the community. They are the most selective of all medical statistics in the sense that they relate only to illnesses and injuries which have a fatal issue, and these are but a tiny proportion (perhaps not one-half of one per cent.) of the gross total of separate illnesses and injuries suffered by the population as a whole during a single calendar year. Moreover, the prevalence of fatal illness or injury among the older members of the population is relatively much greater than it is among the young and the middle-aged: the risk of a fatal issue must increase (for many diseases) with advancing years. Mortality statistics, therefore, are "overloaded" in the upper age-brackets. They are probably overloaded in the lowest age-brackets; also, since the newborn have at least one thing in common with the aged, namely, both are separated by many years from their prime of life. In particular (if a single point of detail may be raised here), mortality statistics bearing upon the occupational distribution of disease are subject to a special disability. Many old people-I do not know how many-at the time of death are following occupations quite diffe...