174 



KNIGHT DUNLAP 



In the first column of Table 2 is given a stop-watch reading. 

 In the second column the number of times that reading occurred 

 in the series designated is given. In the third column is given 

 the range of the chronoscope readings corresponding to the 

 stop-watch readings, and in the fourth column is given the 

 average of these chronoscope readings. For example: in the 

 second series of Experimenter E the reading 1.2 occurred 13 

 times; the chronoscope measures of the same 13 reactions ranged 

 from 0.940 to 1.3; and the average of the 13 chronoscope read- 

 ings was 1.083. 



TABLE 2 

 Experimenter E; second series 



Experimenter A; fourth series 



CONCLUSIONS 



Stop-watch records of association reactions are highly un- 

 reliable, when taken by experimenters who have not had long 

 practice in such work. The experimenter tends (1) to lengthen 

 the record (a matter of slight importance), and (2) to reduce 

 the variations in the records by setting up a reaction-habit of 

 relatively constant period. This latter is a highly important 

 matter, especially in association tests of normal persons, for in 

 these cases not only are the variations, (and not the absolute 

 averages,) of importance, but also, these variations are small. 



While these conclusions are so far valid for experimenters 

 without long practice, there is every reason to believe that with 



