154 VINNIE C. HICKS 



upon dealing with a large group of animals. However, the 

 defect of irregularity cannot be urged against the time criterion 

 any more than it can against that of error according to our 

 results. 



A curve combining the advantageous features of the time and 

 error criteria may be urged for two reasons: — -ist. While time 

 does represent the fact of distance elimination, yet, inasmuch 

 as the two are independent variables and independent to a 

 considerable degree in the first stage of learning, it is evident 

 that time cannot portray this eliminative process as adequately 

 as the distance curve itself. 2nd. Probably one of the most 

 important causes for the divergence of time and distance is the 

 difference of behavior between species of animals. Obviously, 

 such a factor is not operative in our experiment. That a choice 

 of a criterion may depend upon the animal used was suggested 

 in the first part of the paper. We have no experimental evidence 

 in support of this contention. The peculiarities of behavior 

 of the dancing mouse as opposed to that of the rat are apparently 

 such as to demand a greater emphasis upon the distance cri- 

 terion. If this conception be valid, it is evident that a combi- 

 nation curve constructed from the time and distance criteria 

 will offer practical advantages for comparative purposes espe- 

 cially. Curve I of figure 5 represents a combination of time, 

 error and distance in which each has been given an equal value. 

 The error and distance curves, however, represent the same 

 factor, so that a combination constructed on this basis gives 

 too much importance to distance. Moreover, the distance cri- 

 terion is impracticable for ordinary use. Hence we venture to 

 suggest that the most practicable and the most representative 

 curve for general comparative purposes will be obtained by a 

 combination of the time and error criteria, provided that our 

 standard of error unity is utilized. Such a combination is prac- 

 tically represented by curve H of figure 5, w^hich gives half 

 value to time and a quarter each to distance and error. 



CONCLUSIONS. 



The distance and error criteria are fundamentally alike, in 

 that both represent the factor of distance elimination. The 

 distance curve is the better representative of the progressive 

 approximation of the act towards automatic accuracy. It 



