112 PSYCHOLOGICAL EFFECTS OF ALCOHOL. 



reaction experiments. But that association would have been disas- 

 trous to the present experiments. It would have been equally possible 

 to give words by a dictaphone, as was suggested by some friendly critics 

 before the experiments began. But there is no natural impulse to talk 

 back to a dictaphone, none at least to respond to its pronouncements 

 by an associated word. Still more serious than the psychological "set," 

 is the confusion of the intercurrent noises and the instrumental elisions 

 of sound which may be variously important in stimulus words of 

 different lengths. Moreover, it takes practice to become a good dicta- 

 phone operator, and even the best must constantly depend on recon- 

 structing the sound from the sense. This is naturally impossible with 

 isolated words. Actual experiments with a typical series of words 

 recorded on the dictaphone showed enormous individual variations in 

 the number of errors. One subject failed in about 80 per cent of the 

 trials. Not even a practiced operator understood them all. 



It would have been entirely possible to record the moment of reaction 

 by our speech-reaction key. We tried it. But, owing to the muffling 

 of the sounds by the diaphragm, it proved to be utterly impossible for 

 the operator to be sure what was the response of the subject. At 

 present, at least, there appears to be no means for mechanizing the 

 timing device without jeopardizing the main technical requirement of 

 the experiment the clear mutual understanding of operator and 

 subject. 



For convenience of identification on the record, the stimulus words 

 were given in groups of 5. Between each group of 5 words a blank line 

 was run on the record without reaction. After the first 25 words of each 

 series an interval of a few seconds was allowed for resetting the markers. 

 This divided the graphic record further into halves. Each half con- 

 sisted of 5 groups of 5 records each. Thus the subsequent correlation 

 of each record with its appropriate association was a simple and accu-. 

 rate process. 



The pulse-records and pneumographic records were superposed on 

 the reaction-records by the following arrangements : After the mechan- 

 ical pulse-wave had been transformed into an electric impulse by the 

 mercury-cup device, which is described on page 191, the electric cir- 

 cuit was carried directly to the same duplex marker that recorded the 

 latency of the response. Coincident with the association latency 

 records, then, and on the same record line, appears a continuous record 

 of the length of the concurrent pulse-waves. Thus the pulse-lengths at 

 any part of the reaction process may be read directly from the records. 



The pneumograph records were made by using a second mercury-cup 

 device to transform the mechanical action of respiration to electrical 

 waves which caused a marker to touch the record during each inspira- 

 tion only. This recorded only the respiration rhythm, not its depth, 

 but it sufficed to show that the pulse-rhythm of the experiments is 



