SEX-RATIO OF THE DEER-MOUSE. 125 



only 305 broods, or less than 20 per cent., showed any mortality 

 whatever between the time of the first count (16 days or less) 

 and the time of marking (60 days). The number of young of 

 unknown sex which died during this interval was 384 or about 

 seven and a half per cent, of the total. 



Thus it is not possible that a differential death rate (unless 

 very marked) would seriously affect the sex ratio of the sur- 

 vivors. We have, however, fully considered the possible influence 

 of such a differential death rate, and have endeavored to determine 

 its degree, if actual. This has been done by comparing the sex 

 ratios in broods in which deaths (of undetermined ' sex) are 

 known to have occurred with broods in which no deaths are 

 known to have occurred (see below). Unfortunately the figures 

 are available for only a very small number of offspring of known 

 sex (40) which died from natural causes before the " sexing" of 

 the broods to which they belonged. 



As already stated, the total number of broods recorded is 1,567, 

 comprising 5,050 young, or an average of 3.22 mice per brood. 

 According to sex, these were distributed as follows: 



Males . 2,295 



Females 2,357 



Sex undetermined (dead, killed and escaped) 398 



Total 5,050 



For those of known sex the sex ratio (number of males per 

 hundred females) is 97.37=!= J-93- 1 



i In computing the probable errors we have employed a formula furnished 

 us by Dr. Raymond Pearl, viz.: 



/> 



67.45 d + R) \-, 



\ n 



in which R is the number of males divided by the number of females, and n 

 the total number of individuals concerned. 



Our colleague Dr. G. F. McEwen has computed a somewhat simpler form- 

 ula for the sex ratio : 



'/> a 

 -6745 -\|~ X4, 



in which p and q are the percentages of males and females respectively. This 

 gives approximately the same values as Dr. Pearl's formula when the sex 

 ratios do not depart widely from 100. Neither formula is accurate when the 

 departures are very wide. 



It must be stated that the probable errors employed in this paper are about 



