136 F. B. SUM-NER, M. E. MCDANIEL AND R. R. HUESTIS. 



as a whole were rather confusing. We have, however, combined 

 the data for small broods (one to three individuals) and for 

 large broods (four to nine individuals) respectively (Fig. 5). 

 This gives us two groups, each comprising about half of our mate- 

 rial. The former group exhibits a very pronounced biennial 

 rhythm with maxima in April and August. 1 The curve for the 

 latter group/however, although based on somewhat greater num- 

 bers than the first, agrees only in having a well-marked fall maxi- 

 mum (September-October), with lower ratios in the summer and 

 winter. There is but slight evidence of a spring maximum, and 

 there are other irregularities in the " curve " for this large sub- 

 division of our material. In general, none of these differences are 

 of probable statistical significance. The high point in January 

 cannot be regarded seriously, owing to the small number of indi- 

 viduals (53) for that month. 



It must be conceded that the differences in the seasonal cycle, 

 shown by these two groups of mice, seriously weaken our evi- 

 dence for the existence of significant seasonal differences of any 

 sort. For it does not seem likely that mice belonging to broods 

 of different sizes actually behave differently in this respect. 2 It 

 seems more likely that, for present purposes, we have two random 

 samples of the population. 



However, the fact stands that considerable differences have 

 been found in the sex ratios of Peromyscus at different times of 

 the year, whether or not these differences are due to " chance " 

 ( i.e., errors of random sampling) . Let us grant, for the moment, 

 that the differences are not accidental. It then remains to con- 

 sider whether they are due to season per sc. It will be shown 

 below that our hybrids as a class give considerably higher sex 

 ratios than do the mice of pure strain. Also, quite independently 

 of this last, it will be shown that mice born in different years 

 differ widely in their sex ratios. The question arises : Can it be 

 possible that the seasonal differences which we find are due either 

 to the unequal distribution of hybrid and pure-bred mice through- 



1 The difference between the figures for April and June is more than four 

 times its probable error. 



2 Even this is not impossible, however, for there is some evidence (see 

 below) that smaller and larger broods differ in their mean sex ratios. 



