258 



E. C, MACDOWELL AND E. M, VICARI 



separately in the separate strains. The fifty-seven 'plus' ratios 

 have an average of 1.503; the fifteen 'minus' ratios have an aver- 

 age of 1.203. The difference between these averages (0.300 ± 

 0.048) is 6.2 times its probable error. Besides a larger number 

 of the ratios being 'plus,' it appears that the average size of the 

 'plus' ratios is significantly greater than the average size of the 

 'minus' ratios. 



DISTANCE PER DAY 



FREQUENCY DISTRI BUTIONS OF RATIOS 



FREQUENCIES 



MINUS RATIOS 



PLUS RATIOS 



1.0 1.2 1.4 



SCALEOF RATIOS 



Fig. 11 Frequency distributions of the ratios of tests vs. controls, based on 

 thu averages of the distance for each day, with sexes and strains treated sepa- 

 rately; the ratios are taken from table 9. The size of the ratios is shown on the 

 base line, the vertical scale gives the frequencies; the broken line shows the 

 'minus' ratios (controls taking more time); the solid line, the 'plus' ratios (tests 

 taking more time); 'completes' and 'incompletes' included, but not 'failures.' 



c. Variability of the tests vs. controls as judged by distance. 

 Table 11 gives the standard deviations for the averages for each 

 rat for each of the six groupings of the trials (see the distributions 

 of the averages in figure 8) . For the first half of training, omit- 

 ting the first day, training, and training and retention, the standard 

 deviations for the tests are higher; for the second half training 

 and retention the standard deviations for the controls are slightly 

 higher. This shows an agreement between the size of the means 

 and the standard deviations, for it is in these last two groups of 

 trials that the control averages are not significant. However, 

 the differences between these standard deviations are in no group 



