360 H. M. JOHNSON 



H . M. J. preferred the light, but usually felt quite certain that 

 his work in the dark was the more rapid, though requiring a 

 greater effort. G. H. and P. W. C. gave similar descriptions. 



7. L., L. S., L. T. T. and C. C. preferred the dark, and were 

 certain that their performance was better in the dark than in the 

 light. They thought that they were "less distracted" in the 

 dark. 



H. D. preferred the light, and was sure that her performance 

 was better in the light than in the dark. 



F. G., F. F., and E. M. had no preference, and their judgments 

 of performance varied inconsistently. 



G. P. L. was greatly prejudiced in favor of the dark. He 

 habitually works in the photographic darkroom without using 

 the ruby lamp, preferring to rely on touch alone than to utilize 

 " unsatisfactory and inadequate illumination." He was certain 

 at all times that he could work better without the distraction of 

 "light that he couldn't see by," and predicted that his scores 

 would justify his views. I think it unlikely, however, that he 

 consciously failed to exert a maximal effort at any time; and as 

 a comparison of the opinions of the other observers with their 

 scores will show, one's personal estimates of time under the 

 conditions of the experiment are quite unreliable. 



Most of the subjects, it will be noted, believed their perform- 

 ance in the dark to be superior to their performance in the light. 



RESULTS 



The daily results are presented in tables 4 to 9, according 

 to subjects. They are summarized for the several subjects in 

 table 2. The performance of the group as a whole is shown 

 statistically in table 3 and graphically in Fig. 1. 



The character of the results necessitated their subjection to a 

 somewhat unusual form of statistical treatment, in order to ex- 

 hibit them adequately. A brief account may be of interest. 



The subject's daily performance (P) under either condition 

 is measured by the number of cards correctly sorted per unit of 

 time. The unit chosen was 1000 seconds, for convenience of 



