PEARL— SEX RATIO IN DOMESTIC FOWL. 



421 



the percentage of J'c? to total number of chicks was somewhere be- 

 tween lo per cent, and 19.9 per cent. Other entries are to be cor- 

 respondingly read. 



The first thing w^hich strikes one's attention in examining these 

 tables is that extreme values of the sex ratio (below 20 and above 

 80 say) occur relatively frequently only in small families. If the 

 families are very small (Table III.) extreme values of the sex ratio- 

 become actually more frequent than medium values. ' The greater 

 frequency of extreme sex ratios in small families is obviously w^hat 

 would be expected on merely arithmetic grounds. Thus to take the 

 data of Table III. We find from the original records that there 

 were 54 families of i, 53 of 2, and 54 of 3 each contributing to this 

 table. Suppose males and females were equally likely to occur (t. 

 e.^ i?^=5o); then according to the laws of chance, the totals of 

 Table III. would be expected to be as shown in Table IV. These 

 are compared with the actually observed totals. 



•TABLE IV. 



Comparing Totals of Table III., with Chance Distribution of Same 

 Number of Families, on the Assumption that R-, = so. 



While this is by no means a perfect fit of the observations by 

 the chance distribution, the latter is close enough to the former to 

 indicate clearly the essentially chance determined character of the 

 observed distribution. The resemblance would be still closer if we 

 took a value of R ^ for the computation more nearly in accord with 

 the actual fact than is 50, the value actually used. 



There is no need to pursue this point further, as it will be evi- 

 dent to anyone who will examine Tables I., II. and III., in the light 

 of the points just made, that we cannot draw any conclusions of 

 critical value regarding the normal variation of the sex ratio in the 

 fowl, at least, except on the basis of families containing at least 10 

 individuals each. 



