factors had been interpolated in one or both groups. The "law of average" would so demand, and proof of the fact is to be had by a study of the mortality statistics of England or any other country which has tabulated its deaths by occupations or other sharplydefined groupings for long series of years. On examination of the mortality experience of the two sections of the English life company above presented it will be found that the difference between the death-rates of the abstainers and non-abstainers was by no means constant, except in that it constantly and materially varied, as is shown by the following tabulation of the ratios of actual to expected mortality in the two sections and the excess and percentage of excess of the General Section's ratios, by five-year periods: This tabulation reveals some surprising variations in the fiveyear differences between the two sections' ratios of actual to expected mortality, the percentage of the excess of the General Section's ratio over the Temperance Section's ratio of actual to expected having jumped from 25.1 in 1866-70 to 45.2 per cent. in 1871-75, or nearly doubled. It dropped from 42.9 per cent. in 187680 to 29.4 per cent. in 1881-85, or nearly one-third, and in the last three five-year periods was approximately only one-half as large as in 1871-75, and continuously much lower for the fifteen years than in the five-year period immediately preceding that epoch. Why this pronounced shifting of the difference between the ratios of the two sections, and the marked decrease in the last fifteen years, if even the moderate use of alcohol to which non-abstaining accepted life insurance risks are presumably restricted ordains a decidedly increased death-rate for that class as compared with total abstainers? If the body of surviving non-abstaining policyholders were to increase their imbibitions as they grew older, might it not be confidently expected that the excess of their death-rate over that of the total abstainers would show an increase especially as the total abstainers could not very well lower their own death-rate by drinking less than nil? As the above tabulation shows, for the last fifteen years the difference between the death-rates of the two sections has markedly decreased, instead of increased. NO PRACTICABLE MEASURE FOR A GENERAL DEATH-RATE FOR NON-ABSTAINERS FROM ALCOHOL In a word, this last tabulation clouds in mystery the real significance of the oft-cited and allegedly conclusive figures of the United Kingdom Provident Institution's mortality experience with its abstaining and non-abstaining classes of policyholders. I can conceive of but one of two explanations of the surprising variations in the difference between the death-rates of the two sections, namely: either the group of non-abstaining policyholders in the General Section is not a sufficiently homogeneous body to promise a constant difference between its death-rate and that of the group of total abstainers-which in at least one way certainly is a homogeneous body or else the number of lives thus grouped is too small to warrant safe averages. Whichever of these possible explanations is accepted, it would seem that the 45-year experience of the United Kingdom Provident with groups of abstaining and non-abstaining policyholders by no means affords any reliable measure of the alleged decreased longevity of non-abstainers. And, as the greater presumably includes the less in this case as in all cases, if the larger and longer experience of the United Kingdom Provident Institution, of London, covering forty-five years, does not afford such a measure, of what value are the materially smaller and shorter experiences of the Abstainers and General Insurance Company, of Birmingham, the Sceptre Life Association, Limited, of London, the Scottish Temperance Life Assurance Company, Limited, of Glasgow, the Manufacturers' Life Insurance Company, of Toronto, and the Australasian Temperance and General Life Assurance Society, which were also cited by Dr. McMahon in his paper before the Association of Life Insurance Medical Directors? To be sure, the comparative figures for all of these companies show a considerably higher ratio of actual to expected claims in their General Sections than in their Temperance Sections. But, as no two of the comparisons for long stretches of years even approximately concur in fixing the supposed margin of difference, these supplemented data only tend to confirm the doubt as to the real value of any life insurance experience on these lines. In so far as life insurance experience in the United States is concerned, no companies of any importance now maintain separate sections of abstaining and nonabstaining policyholders, or ever have maintained such sections for sufficient time to afford any evidence of the slightest value. In so far as any life insurance company's mortality experience can be cited by the spokesmen for the temperance cause as alleged mathematical proof and measure of the difference between the deathrates of the two types of men loosely classified as drinkers and nondrinkers, the comparatively broad 45-year experience of the United Kingdom Provident Institution undoubtedly takes precedence of all other supposed evidence on these lines. But, looking at this supposed "evidence" from a purely statistical viewpoint, for the reasons already enumerated I cannot regard the experience in question as of any value as a measure of the presumptive difference between the death-rates of the two groups of men in the world at large. At most, as I see it, it merely corroborates in a general way the common belief that there is a difference between the death-rates of these two types of men, as there doubtless is between any two large groups of men of widely different habits, amusements, and activities. But the margin of difference between the recorded death-rates of the abstaining and non-abstaining groups of policyholders in the oldest and largest of the life companies maintaining such classes so sharply varies, even when carefully computed by five-year periods, it would seem impossible to believe that the two groups are sufficiently well-defined to be fairly comparable. If the good and bad qualities of the two artificially-separated groups of life insurance risks, or the physiological credits and debits, so to speak. were reasonably constant factors, might it not naturally be expected that the difference between the mortality of the two groups -however large or small it might be-would also be fairly constant? That the difference by five-year periods is anything but constant I have already shown by the tabulations of the mortality experience of the Temperance Section and General Section of the company in question. To the eye untrained in reading and interpreting figures, however, mere tabulations are generally more or less dumb mysteries, and by way of simplifying in graphic form which any man can instantly read, the widely fluctuating differences between the mortality experiences of the two groups in question, I have translated the tabulations in question into the chart which accompanies this paper. In this plain-spoken chart the excess of the five-year ratios of actual to expected mortality in the General Section over the corresponding ratios of the Temperance Sections is traced by the decidedly-wavy plotted line which records the evervarying percentage of that excess, and the heavy plotted horizontal line showing the average excess for forty-five years by its marked contrast with the other, up-hill-and-down-dale, plotted line, sharply emphasizes the fact that the excess was far from being a fixed quantity. It would seem that this chart could scarcely fail to raise a serious question as to the real value of the United Kingdom Provident mortality comparisons in the mind of any impartial student of the alcoholic problem who may happen to glance at it. But I gravely doubt if the graphic showings of the chart will have any significance in the eyes of the over-zealous type of Prohibition advocates who apparently have no patience with any evidence which does not tend to support their contentions. So deeply immersed are they in the movement in which they have enlisted that they are scarcely qualified for jury-duty in cases in which any phase of the alcoholic problem is involved. In his notable address before the Insurance and Actuarial Society of Glasgow, on February 1, 1909, under the title of "Some Observations on the Comparative Death-Rates of Abstainers and NonAbstainers in Life Insurance Companies," Dr. Ebenezer Duncan, F.F.P.S.G., J.P., remarked: "I have found some strong teetotallers genial, generous, and lovable men, but with characteristic failings, and one of the worst of them is a habit of exaggeration and intemperance in their statements about moderate and temperate indulgence in beverages containing a percentage of alcohol and about the evils, often imaginary, of such moderate indulgence. I 45% 40% 1866 1871- 1876- 1881- 1886- 1891- 1896- 1901-1906- 45% 40% A GRAPHIC RECORD OF THE FIVE-YEAR FLUCTUATIONS IN THE DIFFER- IN THE GENERAL AND TEMPERANCE SECTIONS OF THE |