of the actual difference between the death-rates of users and nonusers of alcohol is concerned. There is quite an array of life insurance figures which conclusively prove a difference between the death-rates of persons insured in the abstaining and non- abstaining classes, but after a thorough study of all available data on these lines I am satisfied that no competent and unbiased person who has carefully examined these figures will conclude that they by any means prove either one of the two points so vehemently advanced by the Prohibition spokesmen; namely, (1) that they establish the actual difference between the death-rates of drinkers and non-drinkers, and (2) that they afford a mathematical measure of the mortality directly chargeable to the use of alcohol. Of course I do not for a moment expect that I can convince my Prohibitionist friends of either of these negative deductions. It is no easy matter to prove any negative, and to establish to the satisfaction of the average Prohibitionist the unsoundness of any of his pet arguments is hopeless. I am, however, quite ready to rest with any unbiased reader, or competent student of statistics, the soundness of my deductions from the evidence below presented.

At the twenty-second annual meeting of the Association of Life Insurance Medical Directors of America, held in New York City in October, 1911, Dr. T. F. McMahon, medical director of the Manufacturers' Life Insurance Company, of Toronto, presented a paper on "The Use of Alcohol and the Life Insurance Risk," and included in this paper certain figures showing the mortality experience of some English life insurance companies with separate classes of so-called abstaining and non-abstaining policyholders, which are discussed on subsequent pages of this paper. Dr. McMahon's address was followed by a general discussion of the subject by members of the Association of Medical Directors, and in part the report of the discussion reads as follows:

Dr. Dwight (Medical Director of the New England Mutual Life) I have made some figures in the experience of Company which I think may be worth while calling to your attention. I am sure that we have all been interested in the figures that have been given, and I am sure that there can be no question as to the accuracy of such figures, but I am equally sure that if such figures are published without some qualification, they are apt to give a false impression, and they are apt to be accepted by certain portions of the community at an unfair valuation. I am sure that most of you gentlemen appreciate the importance of having your classes

homogeneous, and this particular investigation shows, I think, the value. There are many more factors, I think, than the one question as to whether an individual drinks or not, which should be taken into consideration. I imagine the mortality of the total abstainer in the Presbyterian clergyman class and the total abstainer in the bartender class would be quite different, and included in these large groups are a great many cases which are influenced by habitat, other habits, occupation, residence, type of life, and many other factors.

I am not going to take up much of your time, but I have before me figures on habits in our Company for the past sixty years, both on liquor and tobacco. I bring them up to show two or three rather interesting points. In the first place our figures on the habits as to the use of alcohol agree almost exactly with those which have been given us. We divide them into four classes. No man acknowledges, or very few men acknowledge, that they are excessive in the use of liquor. We divide them into the Total Abstainer, Rarely Use, Temperate, and Moderate. By "rarely use" we mean the man who says that he perhaps twice a year at a dinner drinks two glasses of champagne. While we have every reason to believe that this is an honest statement, it is not quite fair to say he is a total abstainer. Otherwise they are divided by the individual's own statement-in many instances incorrect. In many instances the total abstainer is a total abstainer because he has to be and has been advised to be, but take them as they run, we find the total abstainer with a mortality of 59% of the American Table-these are rough figures; Rarely Use, 71%; Temperate, 84%; and Moderate, 125%. If we let these figures stand as they are, it is accumulative evidence to demonstrate that we ought all to stop drinking the next minute, and that we ought to divide our applicants into abstainers and non-abstainers; but let us see what the effect of tobacco is. We find that the total abstainer from tobacco has a mortality of 57% as against 59% when he is a total abstainer from alcohol; that the Rarely Use is 72% as against 71% from alcohol; that the Temperate is 84% as against 84% from alcohol; and that the Moderate is 93% as against 125% for alcohol. In other words, the mortality on the total abstainer from alcohol is almost absolutely the same as that on the total abstainer from tobacco. It may be said that a man is usually a total abstainer from both, but he is not necessarily so. It does mean, I think, that we are describing the same kind of groups, the same type of man, the same conservative type of man in the total abstainers from both kinds. Then to show you the numbers—we find that out of 180,000 cards, 42,700 were total abstainers from alcohol-at least they said they were-while the total abstainers from tobacco were 41,100-almost exactly the same number, 13,000 rarely used tobacco, while 20,000 rarely used alcohol. I only bring them up as they are, simply as a suggestion that before we accept all total abstainers on a different premium or before we talk very much about doing it, we had better be

sure that it is all due to total abstinence from alcohol or tobacco, and we had better go a good deal further in our investigations in our attempt to make these groups homogeneous, before we here as an Association or as individuals advocate very strongly the wide separation between total abstainers and those who rarely use.

Dr. Emery (Manhattan Life Insurance Company)—I would like to make a suggestion for statistics in the future, and that is to find out how much a man eats, for oftentimes it is much more important. A man who is an overeater is not likely to live as long as some excessive drinkers. In other words, a man's habits of eating, and whether he performs the daily functions of life regularly and properly, is what we should know, and if we can only know that, I think we would do a great deal more than we do now, and do more for our companies. It is most important. The man who

is known as a club-man, a free liver, is generally an early dier (pp. 473-6, Abstract of the Proceedings of the Association of Life Insurance Medical Directors of America, from the Seventeenth to and including the Twenty-second Annual Meeting, New York, 1912).

Dr. Dwight's observations clearly stated the danger of taking too seriously the figures of the handful of English life companies. which maintain separate classes for abstaining and non-abstaining policyholders, and succinctly named some of the many all-important factors of longevity which are generally overlooked in considering experience with these classes. Coming as they do from so professionally conservative an observer as the Medical Director of one of the oldest life insurance companies in the country, they are deserving of most thoughtful consideration on the part of every man who reads them and is desirous of getting as near as possible to the real truth of the much-discussed question of the relative death-rates of users and non-users of alcohol. For these reasons I have reproduced them in their entirety as reported in the Proceedings of the Association, and have italicized some of the more striking passages.

A CLASSIFICATION RESTING ON BIASED STATEMENTS OF APPLICANTS

In so far as the remarks relate to the general question, they would seem to rest on a bed-rock basis, and to be entirely without the range of dissent or criticism. But in so far as they recite the supposed experience of the physician's own life insurance company, they are manifestly open to the criticism which in my judgment fairly applies to the returns of all life insurance companies which have made any attempt to classify their policyholders on the

basis of their individual drinking or non-drinking habits. On just what grounds does that classification practically rest? In at least the vast majority of cases, merely on the prejudiced statement made by the policyholder to the medical examiner when he was applying for his life insurance in the company which afterward accepted him. To be sure, of late years, and especially in the case of applications for large policies, many, if not most, of the more important companies send inspectors out to secure supplemental information regarding applicants for policies and in some cases may have investigations of a policyholder's standing and habits made one, two, or three years after the policy was written. But, in the main, practically all of the so-called statistics of life insurance experience as to allegedly abstaining and non-abstaining policyholders solely rest on the statements made by the policyholders at the time of applying for their insurance.

Of what real value in any scientific investigation of the mortality rates of these two groups of policyholders is evidence so gathered? To all intents and purposes, every applicant for life insurance knows at least enough of the attitude of life insurance companies to feel that he is on the defensive, so to speak, when he is undergoing his medical examination, and to be cautious in his answers to any of the examiner's questions as to his habits which might cause his rejection by the company to which he has applied. In short, the applicant is necessarily a biased witness, and even if he does intend to tell the whole truth about his habits, previous sicknesses, etc., he is apt to take a decidedly optimistic view of any possible defects in his qualifications as a first-class life insurance risk.

When the questions as to drinking habits are reached in the examination, ninety-nine per cent. of the applicants who admit drinking at all are reasonably certain to classify themselves as "moderates," or words to that effect, and if asked for details as to the number of drinks which they average for each day are absolutely certain not to overstate their daily quantum. In the vast majority of cases, they undoubtedly understate the daily average, either deliberately or unconsciously. In view of these well-known

facts, how much importance is to be attached to the classification by Dr. Dwight's company, or any other company having only similar evidence or lack of it-of its army of policyholders in such groups as "rarely use," "temperate," and "moderate?" Probably no two men's honest notions of the precise meanings of these terms

in so far as drinking is concerned would exactly tally; daily rations of alcohol which one man might honestly regard as extremely "temperate" undoubtedly would seem decidedly excessive to another type of man, and so on up and down the line.

From my point of view a rating on any such basis of policyholders as to their individual drinking habits would be practically valueless, even though they were to be polled on this question once a year during the entire duration of their policies. As a matter of fact, they are actually polled but once, at the time their policies are taken out, and generally in their early years, and of course their habits may radically change for better or for worse before their deaths are entered up in the mortality of the "total abstainer," "rarely use," "temperate," or "moderate" classes to which they were assigned many years before, and in which they have continuously since been rated. And the same drawbacks on reliable records operate in the case of the policyholder's habits respecting the use of tobacco.

Furthermore, a moment's consideration of the figures stated by Dr. Dwight as showing his Company's supposed experience with these several classes will suffice to raise the most serious question as to their probable value. For instance, the total abstaining class is credited with only 59 per cent. of the expected mortality according to the American Experience Table of Mortality, whereas, the "rarely use" class, which Dr. Dwight defines as the "man who says he perhaps twice a year at a dinner drinks two glasses of champagne," is charged with a mortality of 71 per cent. of the expected, or an increase of 20.3 per cent. over the mortality of the class which does not drink at all. Would even the most enthusiastic Prohibitionist seriously think of contending that two glasses of champagne twice a year would make the user almost one-quarter again as bad a life insurance risk as was the total abstainer, other things being even? I think not. To go a step farther in the demonstration of the absurdity of figures so calculated, note that according to Dr. Dwight's returns, the mortality of even the "moderate" users of alcohol accepted by his Company was apparently 125 per cent. of the expected mortality among acceptable life insurance risks, or, in other words, 111.9 per cent. in excess of the mortality of the total abstainers, and 76.1 per cent. in excess of that of the men who drank only two glasses of champagne twice a year at dinners. I venture to believe that no further riddling of these