172 KNIGHT DUNLAP 



tion-timing is at fault, and as the stop-watch is much used in 

 this work, it has seemed to me worth while to investigate the 

 possible errors in the use of the watch for association-timing. 



8even advanced students volunteered to carry out experi- 

 ments by the following method. The chronoscope and voice- 

 keys were set up with proper accessories for the association 

 reaction. Sensitive voice keys, responding accurately, were 

 used, set so that experimenter and reactor faced each other 

 directly across an ordinary table. The experimenter spoke 

 the stimulus-word in each case against the stimulus key, and 

 manipulated the stop-watch. An assistant managed the chron- 

 oscope, and recorded the readings, which the experimenter did 

 not see. In this way, for each reaction, we obtained a stop- 

 watch record, and simultaneously, a chronoscope record of 

 the same reaction. The six experimenters and several other 

 persons served as reactors at different times. 



The reactions were taken in series of fifty, using word-lists 

 that had been prepared by Dr. Loring for other purposes. As 

 the absolute reaction values are immaterial these word-lists 

 are omitted from this account. Twenty series were obtained, 

 with the results presented in Table 1. 



In Table 1 we can readily discern certain general facts. (1) 

 The average stop-watch time is in every case longer than the 

 average chronoscope time. This would be expected from the 

 tendency to start the watch approximately with the stimulus 

 word (synchronizing reaction), and to stop it after the beginning 

 of the response word (serial reaction). (2) The mean variations 

 of the corresponding averages show considerable disagreement. 

 Experimenters B, D, and E in each series reduce with the stop- 

 watch the actual (chronoscope) variation. Experimenter G in 

 each increases the variation. Experimenters A and C reduce it 

 in three series and increase it in one. Experimenter F reduces 

 the variation in one series and increases it in the other. The 

 differences between the average times for stop-watch and 

 chronoscope, and the divergences of mean variations, do not 

 show in full measure the actual discrepancies between the 

 individual chronoscope times and the stop-watch times, since, 



