140 Effect of Alcohol on Psycho-Physiological Functions. 



APPENDIX III. 



OBSERVATIONS ON THE AVERAGES OF DATA FOR NORMAL AND ALCOHOL DAYS 



PRESENTED BY DODGE AND BENEDICT. 



The following discussion has absolutely no bearing or modifying effects upon 

 any of the comparisons which Dodge and Benedict have made by what they 

 term "the method of differences." It applies only to their discussion of aver- 

 age values, which they have used as supplementary values in expressing the 

 effect of alcohol on the various processes and occasionally have limited their 

 comparisons to them. A common footnote which accompanied their tables 

 states that "the values for the first period of the alcohol experiment were 

 obtained before the alcohol was given, and are therefore not included in the 

 averages." No specific statement is made as to whether the values for the 

 first period on normal days were included or omitted in the averages for these 

 days, which were to be compared with the average values for the alcohol days, 

 but as their footnote implies and a recomputation of any of the data in their 

 tables will show, the first-period values were included in the averages for the 

 normal days. The alcohol and normal days would have been more directly 

 comparable had the averages in both cases been computed without including 

 the first period, which is termed " normal of the day," and is really preliminary. 

 The point is a minor one, and produces no qualitative change in the conclusions 

 offered by Dodge and Benedict, but as it is one the careful critic may notice, 

 the following recomputations of the Dodge and Benedict data are presented : 



The pulse data obtained during mental and physical activity are given in 

 Dodge and Benedict's table 40, pages 212 to 222 of their report. The mean 

 variations between the lengths of individual pulse cycles for the respective 

 periods are regarded as significant measures and are further presented in their 

 tables 44 and 45. They form the basis of discussion of these data, with the 

 conclusion that the pulse after alcohol shows a smaller mean variation and 

 hence a decreased responsiveness of the cardio-inhibitory mechanism (see 

 pages 238 and 239 of their report). A recalculation of the averages and the 

 mean variations in their table 40 will show that the mean variation for the first 

 period of an experimental day is usually smaller than the average mean varia- 

 tion for the other periods, i. e., the heart rate was usually faster. If, therefore, 

 we omit the first period in the averages for the normal day, as was done in the 

 averages for the alcohol days, the final computed decrease of the mean varia- 

 tions, as shown in their table 45, will be 22 per cent in place of 19 per cent. 

 The method of presentation has to an appreciable extent masked the effect of 

 the alcohol in this instance. . 



The summary of the association pulse data in table 38, page 209 of their 

 report, is an apparent exception to the usual method of computing the aver- 

 ages; although no statement is made in the text, the averages forboth normal 

 and alcohol days apparently include the first-period values. This is the only 

 observed instance in Dodge and Benedict's computations where the average 

 for the alcohol experiment has included the normal of the day. These aver- 

 ages, as presented in their table 38, are therefore directly comparable, but 

 since the first period has been included, the alcohol section of the experiment 

 is not so well isolated. If we accept the data in table 36, pages 202 to 205, as 

 correct, then it is a simple matter to reconstruct table 38, as the preliminary 

 data for it are given in the extreme right-hand column of table 36. In part I 

 of table 30, Dodge and Benedict's table 38 is reprinted as published. Unfor- 

 tunately this contained several errors. One of these errors is large, i. e., the 

 average for Subject VI, Normal i and n, should be 806 instead of 876. In 

 part ii the errors have been corrected, the method of computation employed 

 being that used in the original table. In part in of table 30 the averages do 



