PLAN OF THE INVESTIGATION. 29 



probably more useful for comparing one individual, dosage, or process 

 with the others. The main objection to the percentile is that it elimi- 

 nates every vestige of the units in which the measurements were 

 actually made. It is useful for comparative purposes to know the 

 percentage of change. It is also important to know these same changes 

 in terms of the unit of measurement. Our summaries will contain 

 both values. 



The methods for computing these various values are not especially 

 significant. It is important only that they be uniform and clearly 

 understood. The average difference of a day's measurements is obtained 

 as follows: 



Av . D . = (1-2) + (1-3) + (1-4) + (I-*) 



n 



That is, the sum of the algebraic results of subtracting the various 

 subsequent measurements from the normal of the day is divided by 

 the number of periods. If the Av. D. has a minus sign it shows 

 that the measured values are larger as the session progresses. Con- 

 versely, if the Av. D. has a positive sign it shows that on the average 

 the subsequent measurements are less than the normal of the day. 

 The effect of alcohol as expressed in Av. D. is computed as follows: 



Effect of alcohol = the Av. D. on alcohol days minus 

 the Av. D. on normal days. 



If the effect of alcohol has a plus sign, then the +Av. D. on alcohol 

 days is greater than the +Av. D. on normal days, or the latter has a 

 negative sign. Similarly, mutatis mutandis, if the effect of alcohol has 

 a negative sign. It should be noticed that the sign of the effect in all 

 cases is a result of the statistical procedure. It does not indicate 

 whether the effect of alcohol increases or decreases the sensitivity of a 

 process unless it is interpreted in the light of its origin. 



The effect of alcohol as expressed in percentiles retains the same sign as 

 when the effect is given in the units of measurement. It is computed 

 by dividing the latter by the average of the relevant normals of the day. 



For the convenience of the reader, these various mathematical 

 expressions are commonly reinterpreted as they occur in the tables and 

 the text. 



DOSAGE. 



Neither in the experimental literature nor in the theoretical dis- 

 cussions is there any uniform standard of alcohol dosage. Probably 

 the most satisfactory arrangement of the dosage, in man as in animals, 

 would be according to some definite percentage of the mass of the blood. 

 This would appear to be necessary in all attempts to measure individual 

 differences. In view of our relatively simpler problem, we chose to 

 follow the easier traditional usage in these experiments, and administer 

 the alcohol in fixed doses for all subjects. The quantity of alcohol in 



