DOCKERAY : PHYSICAL FATIGUE AND MENTAL EFFICIENCY. 209 



effort was made to make the order as irregular as possible, and 

 at the same time sound each sounder approximately an equal 

 number of times. It was found, however, that sounder number 

 three (operated with third finger) got the preference, followed 

 by two, one and four. There was also a difference in the diffi- 

 culty with which the sounders were distinguished. Numbers 

 two and three were most difficult and most often confused with 

 each other. 



As might be expected, some subjects could distinguish the 

 sounders more easily than others. Thus the test was found to 

 be very easy for some subjects and difficult for others. Neces- 

 sarily it would be as poor a test for either extreme. The sound- 

 ers could not be adjusted for each subject. The method finally 

 hit upon was the introduction of a fifth sounder which should 

 sound simultaneously with each of the others, and the intensity 

 of which could be adjusted readily. The best means of mak- 

 ing this adjustment was found to be to adjust it so as to ob- 

 scure almost completely the difference in the other sounders. 

 Then, by inserting one or more thin papers under the hammer, 

 the intensity could be reduced and the differences in the other 

 sounders made more distinguishable. For some subjects two 

 papers were used ; for others one paper was sufficient, and for 

 one subject, De., no paper at all was necessary. 



The average number of clicks in each four-minute period 

 to which the subject should react is given in the first 

 column of Table Ilia, followed in the second column by the 

 average number of times he did not react, and in the third 

 column the average number of times that he reacted wrongly. 

 Though in reality both failures to react and wrong reactions 

 are errors, for the sake of brevity we will speak of the former 

 as "omissions" and the latter as "errors." The quality of the 

 work done is expressed in this series negatively, that is, an 

 increase in per cent means a decrease in efficiency, as it is the 

 mistakes instead of the successes that are recorded. The per 

 cent of omissions and the per cent of errors, given in the 

 fourth and fifth columns, are based on the first column, though 

 the per cent of errors might just as reasonably have been based 

 upon the average of all sounders, that is, 232. The totals 

 given in Table Ilia are computed directly from the correspond- 

 ing columns of the four periods into which the tests were di- 

 vided, while the results shown in Table Illb were determined 



