Statistical Method. 29 



the eyelash cast a shadow upon the slit of a camera, behind which a 

 sensitive plate could be made to move in a horizontal direction. The 

 shadow was so placed that the usual closure of the lid would not com- 

 pletely withdraw it from the camera slit, causing a break in the picture. 

 Thus the conditions for recording the amplitude of movement were 

 standardized. The stimulus for the wink reflex was produced by the 

 sharp snap of a spring hammer against a sounding board. This latter 

 member has a short projection reaching to the slit of the photographic 

 camera, and the vibration of this pointer at the moment of sound regis- 

 tered the instant of stimulus on the photographic record. Two stimuli 

 were used, separated by an interval of approximately 0.5 second, which 

 was within the relatively refractory period. The instruction was to 

 count backward verbally, not too rapidly, from a number which was 

 given by the operator just prior to the stimulus. Some time after the 

 subject had commenced to count, the plate was noiselessly released and 

 its movement opened in turn two circuits, which released the spring 

 hammers and delivered the stimuli. The photographic curves were 

 read for latency and amplitude. Four such records were made for 

 each period. 



The ten measurements just described completed the cycle of observa- 

 tions taken during an experimental period (approximately 50 minutes) .* 

 The periods followed each other without intermission, excepting the 

 slight break which always came between periods 1 and 2 when the dose 

 was given. Theoretically, for purposes of comparison the same num- 

 ber of periods should be taken each day and the periods should be of 

 equal length; this ideal of uniformity is hard to attain with a complicated 

 group of measurements. Much depends upon the facility with which 

 the subject can adjust himself to the apparatus and perform his part. 

 During the six days there was no accidental interruption to cause the 

 omission of any measurement or enforce a delay of more than ten 

 minutes for the adjustment or repair of apparatus. The number of 

 experimental periods completed on the respective days was as follows: 

 June 29, four periods; June 30 and July 1, five periods; July 2, 3, and 

 4, six periods. 



STATISTICAL METHOD OF PRESENTATION OF RESULTS. 



In comparing normal and alcohol experiments we are interested in 

 any differences which may exist between the two groups of data, and in 

 the "probable correctness" of such differences as evidence of an alcohol 

 effect. Any group of measurements is most conveniently described in 

 terms of (1) central tendency and (2) variability. In the following 



1 Dodge and Benedict's periods were approximately 30 minutes in length, for, as stated in 

 their report (pp. 14 and 15), in their experiments all of the different measurements (reflexes, 

 reactions, coordinations, etc.) were not made on the same day, as was the case in this second 

 group on Subject VI. 



