208 PSYCHOLOGICAL EFFECTS OF ALCOHOL. 



tendency of the pulse during the experiment, the course of the pulse 

 in any single experiment would be subject to all possible accidental 

 disturbances. For example, if the post-stimulation acceleration hap- 

 pened to coincide with the inspiration acceleration it would be too 

 large. Conversely, if it happened to coincide with the expiration 

 depression, it would be too small, perhaps even negative. Fortunately, 

 the experimental rhythm, 10" between stimuli, is quite different from 

 the respiration rhythm, and it seemed possible, consequently, to elabo- 

 rate the data by using the statistical device which is commonly known 

 as the sliding average to eliminate in part the shorter rhythms, while 

 leaving the longer rhythm relatively undisturbed. For example, a 

 supposititious series of pulse-records may be thus elaborated. We may 

 suppose a pulse-sequence which shows a respiratory pulse-rhythm cor- 

 responding to the series, 95, 90, 95, 100, 95. If the experimental accel- 

 eration were 10, and the stimulus occurred at 90, the series would read, 

 95, 90, 85, 90, 85; in which case the apparent post-stimulus acceleration 

 would be only 0.05", or only 50 per cent of the hypothetical acceleration. 

 If, on the other hand, the stimulus coincided with the value 100, the 

 line would read, 95, 90, 95, 100, 85, 90, etc., and the apparent post- 

 stimulation acceleration would be 0.15" or 150 per cent of the hypo- 

 thetical stimulation. The operation of a sliding average of 3 would 

 transform our supposititious pulse-rhythm to 93, 95, 97, 95, and the 

 consequent disturbance of the experimental change occurring at any 

 point would be correspondingly reduced. 



The pulse data for the Kent-Rosanoff association series, A and B, 

 were elaborated in this way by substituting the sliding average of three 

 for each measured pulse-length. From the elaborated table the post- 

 stimulation accelerations were computed for each experiment. It is 

 these values which are used in the correlation measurements of pulse 

 and association character as described in Chapter IV. It is obvious 

 that such elaboration of the pulse data does not entirely eliminate the 

 lesser rhythm that it mitigates, and that it leaves all of the larger but 

 probably slower and less disturbing rhythms untouched. Each meas- 

 urement, consequently, has a relatively large probable error. But the 

 errors were accidentally distributed, and any regular or close connec- 

 tion between an association category and an exaggerated pulse-accel- 

 eration should appear as a general tendency in the correlation, if it 

 existed. 



In addition to the effect of alcohol on the post-stimulation pulse- 

 acceleration, our data permit us to study the more general effect of the 

 alcohol doses on the course of the pulse from period to period throughout 

 the 3-hour experimental sessions. A summary of the average duration 

 of the pulse-cycles on normal and on alcohol days is given in the first 

 part of table 38, together with the average differences between the 

 normal of the day and subsequent pulse-cycles. In the second part of 



