266 



E. C. MACDOWELL AND E. M. VICARI 



only half as large as its probable error (0.44 times) ; in retention 

 the difference is about twice as large as its probable error (2.07 

 times). 



No claim, on the basis of the statistical significance, can be 

 made that the tests are really slower, but the greater number of 

 comparisons do go in one direction, and so may appear to have 

 some weight apart from the significance of the individual differ- 

 ences considered separately. This evidence is offset by the fact 

 that the exceptions are strongly centered in one strain; they 

 are not distributed at random as though due to irregular causes. 

 The frequency distributions of each rat in the group of 'completes' 



TABLE 15 



Showing the significance of the difference in the speed {centimeters per second) of 

 the tests and controls; sexes and strains together. The differences favor the 

 controls, but are not significant 



upon which these comparisons are based are given in figure 13 

 for training and retention. 



The averages of the speed for each day separately of training 

 and retention have been studied. When the sexes are put 

 together all but one day in strain A, and two days in strain L 

 show the controls with higher speeds; but every day in strain C 

 shows the tests with higher speeds. This same general difference 

 between the strains holds when each sex is averaged alone. So 

 it appears that the averages for all training and retention repre- 

 sent differences that hold day by day for each strain, and they 

 are not due to irregular circumstances effective only on some 

 particular day or days. The findings are so unquestionably 

 negative that it has not seemed necessary to present the tables. 



This study of speed was undertaken to see if there was any 

 difference in the general nature of the motor activity of the 

 tests and controls apart from their success on the maze. In 



