ALCOHOLISM AND BEHAVIOR OF WHITE RATS 



237 



Above the line are plotted the 'plus ratios' (those that indicate 

 that the test average was higher than the control average); 

 below the line are plotted the 'minus ratios' (those indicating 

 that the control averages were higher than the tests). The 

 heavy hne connects the plotted ratios for the group of rats 

 excluding the 'failure' rats; the light line includes the 'failure' 

 rats. Obviously, the strong tendency in each strain is for the 

 hne connecting the ratios to lie well above the base line of equal- 

 ity; in other words, the test averages tend to be higher (that is, 

 take more time) than the control averages. 



The following tabulations are given in consideration of the 

 size of the ratios : 



The number of ratios over '-plus' 1.500 



Excluding 'failure' rats, 42.5 per cent of the 'plus' ratios are 

 above 1.500; 25 per cent of the 'minus' ratios are above 1.500; 

 17.5 per cent more of the 'plus' ratios are over 1.500. It is clear 

 that besides being more frequent there is a larger proportion of 

 high 'plus' ratios than high 'minus' ratios. This is brought out 

 more plainly in the following graph (fig. 5), which gives the distri- 

 butions of the ratios on each day for the sexes and strains sepa- 

 rately (that is, not using ratios involving the repetition of data). 

 The 'plus' ratios are shown in the solid line and the 'minus' 

 ratios by the broken line. The mean of the 'plus' ratios is 

 higher than the mean of the 'minus' ratios, but the difference 

 has questionable statistical significance as tested by its probable 

 error. (Average of 'plus' ratios = 1.685; average of 'minus' 

 ratios = 1.478; difference 0.207 ± 0.09 = 2.30 times the prob- 

 able error.) 



