28 PSYCHOLOGICAL EFFECTS OF ALCOHOL. 



We have raised serious objection to expressing the effect of alcohol 

 by any difference in the measurements of the experimental processes 

 before and after the dose is administered. 1 Such an expression would 

 fail to show any accidental inhibition of normal changes, while it would 

 improperly include all changes due to other intercurrent tendencies, 

 such as the results of enforced quiet, of repetition, of regular daily 

 rt^thms, etc. There are even greater objections to expressing the 

 effect of alcohol by the difference in the averages of measurements 

 which are made on normal and on alcohol days respectively. Such 

 expressions would improperly include all the accidental peculiarities 

 of the condition of the subject on the respective daj^s, such as changes 

 of general well-being, fatigue, and mild infections, all the seasonal 

 rhythms, climatic changes, etc. It is obvious, however, that if the 

 number of experiments were sufficiently large, and if they were spread 

 over a number of years, the accidental errors in this method of ex- 

 pression would tend to balance. 



The real statistical problem is to find an impartial expression for the 

 effects of alcohol from a relatively small number of experiments on a 

 subject on any experimental day. It should tend to exclude the short 

 rhythmic and arrhythmic changes on the one hand, and the longer 

 changes in general condition on the other, leaving as exclusively as 

 possible just those precise changes that are occasioned by the experi- 

 ment itself. We believe that on the whole these requirements are best 

 met by the average differences between the " normal of the day" and 

 subsequent measurements on normal and alcohol days respectively. 



The normal of a day, it will be remembered, is the result of meas- 

 urements which are obtained during the first period of each session. 

 The first period in our experiments was always normal, even on days 

 when an alcohol dose was subsequently administered. The normal 

 of any day should consequently represent the general condition of the 

 subject on that particular day. The average differences between the 

 normal of the day and subsequent measurements on a normal day 

 should represent the normal rhythmic and arrhythmic tendencies of an 

 experimental session. Deviations of the average difference after alco- 

 hol from a normal average difference should come as near as possible 

 to expressing the actual change produced by the alcohol alone. 



In all our statistical expressions, then, these average differences 

 between the first and subsequent periods are of primary importance. 

 In our tables they are commonly accompanied by a statement of 

 average measurements, but the latter are regarded as of relatively little 

 importance. They are only given as details that may be of interest 

 to some future investigator who may be measuring similar processes. 



The effect of alcohol on the average differences may be expressed 

 either in absolute units or in percentiles. The percentile expression is 



18. 



