422 



PEARL— SEX RATIO IN DOMESTIC FOWL. 



We may next consider the mean sex ratio, dealing separately 

 with each of the three groups. In calculating these means, and the 

 other variation constants, it was not assumed^ as is ordinarily done, 

 that each class centered at the mid-point of the strip of base on 

 which its frequency stands. To have done so would have involved 

 a considerable error. Instead the actual centering point for each 

 class was determined from the individual records. The results are 

 shown in Table V., and from this table one can see how large the 

 error involved in the usual statistical assumption would have been. 

 The reason for the error is, of course, purely arithmetical, and 

 arises from the fact that in small groups, such as the families here 

 dealt with, only certain percentage values are possible. 



Using the values of Table V., we get, by ordinary methods, the 



TABLE V. 



Showing the Actual Centering Points of the Several Classes in Tables 



I., II., AND in. 



mean sex ratios exhibited in Table VI. We shall deal at this point 

 only with the total distribution of Tables I., II., and III. 



TABLE VI. 

 Mean Sex Ratio of the Domestic Fowl. Various Breeds. 



c?' cf per 

 Group. •''^rf- 1,000 cf cf 



Families of 10 and over (Total Table I.) 48.57 + 0.28 944 



Families of 4 to 9 inclusive (Total Table II.) 49-39 i -§4 976 



Families of i to 3 inclusive (Total Table III.) 55.07 + 2.11 1226 



Families of 4 and over (Tables I. and II. combined) 48.80+ .33 953 



Families of all sizes (Tables I., II., and III. combined) 49.45 978 



