270 



E. C. MACDOWELL AND E. M. VICARI 



(the difference is 4.13 times its probable error) and omitting the 

 first day (2.95 times the probable error). In the other groups 

 of trials there is no significant difference (table 18). 



TABLE 17 

 Showing the significance of the differences between the averages of the number of errors 

 made by the tests and controls in each of the six groupings of the trials. These 

 differences are taken from the averages for all strains and both sexes in table 16. 

 Plus signs indicate that the tests made more errors than the controls. There is a 

 strong tendency for the tests to make more errors, and the differences in the averages 

 are in most cases significant 



First half training . . . . 

 Second half training . . 



Omitting first day 



All training 



Retention 



Training and retention 



TABLE 18 

 Standard deviations of the averages of tests and controls based on the number of errors 

 per trial in each of the six groupings of trials; calculated from the averages for 

 each rat given in table C in the appendix. When the standard deviation of the 

 tests is greater than that of the controls, the difference is called plus; when the 

 controls have a greater standard deviation the difference is marked minus. The 

 only significant differences are in the direction of greater variability in the average 

 number of errors on the part of the tests 



First half training . . . . 

 Second half training . . 



Omitting first day 



All training 



Retention 



Training and retention 



+1.82±0.44 

 -0.25±0.20 

 +0.62±0.21 

 +0.10±0.26 

 -0.05±0.18 

 +0.29±0.19 



DIFF./P. E. 



4.13 

 1.23 

 2.95 

 1.38 

 0.27 

 1.52 



b. Each type of error separately. The different types of errors 

 seem to provide an excellent opportunity for a more analytical 

 study of the behavior of the test and control rats; it seemed 

 possible that the differences in the time and distance records 

 might be accompanied by differences in the kind as well as the 



