ALCOHOLISM AND BEHAVIOR OF WHITE RATS 



249 



of the tests and controls are compared by the ratios of one to 

 the other and not by their differences; a 'plus' ratio indicates 

 that the test average is higher and has been divided by the 

 control average; a 'minus' ratio indicates that the control average 

 is higher and has been divided by the test average. 



In the whole table there are sixty-seven 'plus' ratios against 

 five 'minus' ratios. The highest 'minus' ratio is 1.204; forty- 

 nine of the 'plus' ratios are higher than this. The males give 

 'plus' ratios in all cases. The females give the five 'minus' 

 ratios; strain A seems responsible for these. When the sexes 



TABLE 8 



Showing the significance of the differences between the distance averages, based on 

 the averages of each rat for the different groupings of the trials. The differences 

 are taken from the averages for sexes and strains together in table 7. Plus signs 

 indicate that the averages of the tests are higher than the controls; in five of the six 

 groupings the differences are probably significant, indicating that the distances 

 covered by the tests are significantly greater 



First half training . . . . 

 Second half training . . 



Omitting first day 



All training 



Retention 



Training and retention 



+54.2±9.2 

 +13.9±5.1 

 +26.4±5.3 

 +36.2±6.2 

 + 7.6±4.2 

 +26.0±5.2 



DIFF./P. E. 



5.9 

 2.7 

 5.0 

 5.8 

 1.8 

 5.0 



are combined all the ratios become plus. In table 8 are given 

 the differences between the test and control averages which 

 include all the rats of both sexes and all strains, the probable 

 errors of the differences, and the differences in terms of their 

 probable errors. 



In four of the six groups of trials the differences are well above 

 three times their probable errors (5.9, 5.0, 5.8, and 5.0 times, 

 respectively). For the second half of training the difference is 

 only shghtly below three times its probable error (2.7 times), 

 but for retention the difference is only 1.8 times its probable 

 error. For distance, as for time, the difference in favor of the 

 controls is fully significant during the initial period of rapid 



