488 H. M. JOHNSON 



SUMMARY 



In all 18,300 discriminative reactions are presented as obtained 

 on four subjects under three lighting conditions. 



For all the subjects, the time required for making a discrimina- 

 tive reaction is longer for the condition under which the surround- 

 ings are dark than for the condition under which the surroundings 

 are three-fourths as bright as the stimulus. The differences 

 between the averages vary between 2 per cent and 20 per cent 

 for the several reactors; and in seven sets out of eight the prob- 

 ability is many billion to one against the differences being due 

 to chance. The one exception is one series given by subject C, 

 whose preference is for. the dark. In this case there was no 

 reversal of the effect, but merely a reduction of the difference 

 to a value which is four times its probable error. 



All the subjects required a longer time for discriminative 

 reaction when the surroundings were 2.25 times as bright as the 

 stimulus than when they were 0.75 times as bright as the stimulus. 

 The magnitude of the difference varies between 4 per cent and 

 10 per cent for the different subjects and in every case its value 

 with respect to its probable error indicates a probability of many 

 billion to one against the difference being due to chance. 



For three of the four subjects the retardation due to excessively 

 bright surroundings is less than that due to dark surroundings. 

 The difference for subject E is between 3 and 5 times its probable 

 error, and therefore reasonably probable; for subject M it is 

 from 19 to 24 times its probable error, and therefore almost 

 absolutely certain; while for subject A it varies between 5 and 9 

 times its probable error. One set of subject C shows a reversal 

 of this effect and the other set is ambiguous. 



For all the subjects, the dispersion of the results is greater 

 when the surroundings are dark than when they are 75 per cent 

 as bright as the stimulus. The magnitude of the difference 

 varies between 15 per cent and 54 per cent for the different sub- 

 jects. For subject C the difference is approximately 5 to 6 

 times its probable error; and for the other subjects, from 8 to 

 17 times its probable error. 



