EFFECT OF ALCOHOL ON FOWL 169 



would be altogether premature to draw any conclusion in regard 

 to the matter until more extended data are at hand. At present 

 I desire merely to put on record the facts now available. 



During the period covered by the present report none of the 

 male birds, either alcoholic or control, died. 



The superior mortality record of the treated birds, while a 

 side issue to the main genetic interest of the study, has some 

 interest on its own account in connection with the general prob- 

 lem of the effects of alcohol upon the organism. There is a wide- 

 spread popular opinion that life insurance statistics have 'proved' 

 that even the most moderate use of alcohol definitely and meas- 

 urably shortens human life. In common, as I suppose, with 

 most persons who have made no special personal investigation 

 of the original literature on the subject I had supposed this 

 statement to be true. The present results were, however, so 

 clear-cut in the opposite direction that my curiosity was aroused 

 to examine critically the actuarial evidence. The results were 

 somewhat astonishing. The evidence on which the current 

 statements are based would not be accepted by anyone trained 

 in the critical valuation of statistical and biological evidence as 

 'proving' anything. All of the various actuarial investigations 

 of the question which have been made, including Moore's analy- 

 sis of the experience of the United Kingdom Temperance and 

 General Provident Institution, McClintock's review of the ex- 

 perience of the Mutual Life of New York, Phelps' study of the 

 experience of the Northwestern Mutual Life, and the widely 

 quoted Medico-Actuarial Mortality Investigation, based on the 

 mortality experience of 43 American life insurance companies, 

 appear to suffer, in greater or less degree, from the following 

 defects, which entirely invalidate them for the purpose of deter- 

 mining critically and scientifically the effect of alcohol in differ- 

 ent dosages upon human longevity: (1) The numbers dealt with 

 are small. (2) There is no evidence of any sort or kind as to 

 how much alcohol the subjects of the investigations consumed 

 except their own statements on the subject made at the time 

 insurance was appUed for. (3) No allowance is, or can be, 

 made for the influence of almost numberless other factors which 

 may differentially influence the mortality in the groups com- 



