300 THE AMERICAN NATURALIST [Vol.LYI 



increase the number of correct first choices more rapidly 

 than do the tests, and as the difference between the av- 

 erages increases it becomes statistically significant. 



Granting that the controls are adequate, the data on 

 behavior indicate that a modification has been brought 

 about by the alcohol; the generation showing the least 

 absohite difference in maze-behavior is shown to be defi- 

 nitely modified when the tests are made on a multiple- 

 choice apparatus. 



Fektility 



Compared with the difficulty of measuring the behavior 

 tendencies of rats, the measure of fertility is very simple 

 and definite. However, the great amount of time required 

 by the behavior studies prevented the collection of many 

 of the available data on the purely physiological side. 

 As a result of this, instead of the long list of criteria 

 of fertility that have been given by other authors, it is 

 possible to give only two with any degree of accuracy 

 and completeness. These are: the number of rats in a 

 litter, and the number of litters. A more detailed report 

 on the data leading to the following conclusions may be 

 found elsewhere (MacDowell, '22a). 



