SEX-RATIO OF THE DEER-MOUSE. 



155 



ment on the significance of these figures. Most recent experi- 

 ments with other animals have given negative results as regards 

 the effect of diet upon the sex ratio. 



THE YEAR. 



We have already referred to the surprising fact that the sex 

 ratios of mice born in different years may differ widely and sig- 

 nificantly from one another. The following table gives the figures 

 for all the years comprised in our records, with the exception of 

 1914, for which the data are very meager: 



As judged by the customary standard, some of these differences 

 are highly significant. That, for example, between 1916 and 

 1917 is 54. 80 9. i, the difference being six times its probable 

 error. The chances that this result was due to "accident" (i.e., 

 that no difference would have been found if our numbers had 

 been indefinitely great) are only about 23 in a million. The dif- 

 ference between the ratio for 1916 and the ratio for the entire 

 period (97.37 1.93) is 3.5 times its P. E., that between 1917 

 and the general ratio being 5.3 times its P. E. The difference 

 between the largest figure (125.36) and the second smallest 

 (87.68) is 3.5 times its P. E., that between the smallest figure 

 (70.56) and the second largest figure (118.10) is 4.3 times its 

 P. E. The difference between the second largest and the second 

 smallest ratios is, however, only 2.5 times its P. E. Finally the 

 difference between the ratios for the two years during which the 

 greatest number of births occurred (1918 and 1921) is 14.42 

 5.2, i.e., it is 2.8 times its P. E. 



The foregoing probabilities are, to be sure, not wholly cumula- 

 tive, but the conclusion seems hardly avoidable that certain of 

 them are real in the sense of not being due to errors or random 



