240 E. C. MACDOWELL AND E. M. VICARI 



time than the controls in running the maze. In certain cases 

 the tests took less time, but these differences tend to be smaller 

 than the differences in the other direction. In the first half of 

 the training period the differences are statistically significant. 



Still further comparisons have been made, using the time for 

 each successive trial instead of total time for each day. That 

 is, averages of the first trials of all the rats in all the different 

 combinations of sexes and strains have been calculated, and so 

 for each of the thirty-six trials. This study gives results in full 

 agreement with those obtained from the study of averages 

 based on the total time for each day. The variability is naturally 

 increased, but the number of points of comparison has been 

 tripled; even without this evidence the conclusion is so unques- 

 tionable and the tables involved are so extensive that these sum- 

 maries are not presented. 



c. Variahility of the tests vs. controls, judged by time. In table 

 5 are given the standard deviations of the averages for each rat 

 for each of the six different groups of trials, when the males and 

 females in all strains are put together. The tests have higher 

 standard deviations in all groups of trials except in retention 

 and, when the 'incompletes' and 'incompletes' and 'failures' are 

 added, in the second half of training. The difference is more 

 than three times the probable error of the difference in four of 

 the six groups of trials when the 'completes' are alone; adding 

 the 'incompletes' does not change this; but when the 'failure' 

 rats are also added, none of the groups of trials give differences 

 between the standard deviations that are significant; i.e., three 

 times their probable error. 



Table 6, which gives the standard deviations for each day, 

 shows that the tests are significantly more variable on the first 

 four days of training and on the fourth day of retention; with the 

 'failures' included, the tests are significantly more variable on 

 the first, third, and fourth days of training and on the fourth 

 day of retention. Although the differences in the standard 

 deviations are not significant when the first twelve trials are 

 taken together (first half of training) and the 'failures' included, 



