SEX-RATIO OF THE DEER-MOUSE. 159 



" significant " figures may result from chance. An illustration of 

 this fact, drawn from the experience of the senior author, is 

 worth recording here. Two series of mice had been subjected to 

 a difference of treatment which even the most ardent believer in 

 " sex determination " by physical agencies would not think of as 

 being effective in this connection. The difference was that in one 

 series the right sciatic nerve w r as cut, in the other the left. The 

 first 9 broods of " right " parentage consisted of 14 females and 

 no males. During the same period, 5 broods were born of 

 " left " parentage, consisting of 10 males and 4 females. Had the 

 attempt been made to influence sex by experimental procedure, 

 the result would have seemed highly satisfactory. Thus, the 

 chance of obtaining 14 individuals which were exclusively females 

 is only one in 16,384, and the "significance" of the results is 

 increased when we consider the high preponderance of males in 

 the contrasted group. Experiences such as this lead one to de- 

 mand higher statistical probabilities than are frequently accepted 

 as convincing. 



SUMMARY. 



Data have been presented, based upon over 4,600 deer-mice of 

 known sex, which were born and reared in captivity, under tem- 

 perature conditions not far different from those existing in nature. 

 The following results seem to be of most importance. 



1. In size the broods ranged from i to 9, the mean of the 

 1,567 broods being 3.22. 



2. The sex ratio for the entire lot was 97.37 1.93. When 

 we include only those broods in which no deaths are known to 

 have occurred (nearly nine tenths of the whole) the figure be- 

 comes 98.01 2.07. Broods known to have been incomplete give 

 a ratio of 93.08 5.25. In order to eliminate the effect of sea- 

 sonal differences, the mean of the monthly means has been com- 

 puted. This is 95.65. None of these figures can be regarded 

 as differing significantly. 



3. The probable errors employed throughout this paper are 

 based upon a formula different from that which has been used by 

 various previous writers. As a result, the errors here given are 

 about twice as great as would formerly have been computed, and 



