80 
On the Inheritance of the Sex-ratio 
(ii) Data drawn from The Whitney Family of Connecticut and its A ffiliations 
(1649—1878), by S. Whitney Phoenix, 3 vols., Newport, 1878. This work con- 
tains a very great deal of genealogical information with regard to American 
families connected nearly or remotely with the Quaker Family of Whitney. In 
this case no family in both generations of less than four members was used to 
determine the sex-ratio. 2197 such families were extracted. 
(iii) Data drawn from The General Studbook, 20 vols., J. E. and T. P. Wetherby. 
In this case 1000 thoroughbred mares were taken at random and the sex-ratio of 
their produce and that of their dam calculated. Both mother and daughter 
must have had at least eight foals to be included in the list. 
Some word must be said as to what has been understood by sex-ratio in the 
course of the work. It has been taken to represent the fraction : number of male 
offspring divided by total number of offspring. This point must not be forgotten 
in the following investigation. Thus, in dealing with the father's sibship, there 
must always be one male, and, considering the average size of human families, it is 
extremely unlikely that the sex-ratio as defined above should fall between "00 — "05, 
it would in fact require at least 20 children. Again, in the mother's sibship there 
is always one female, and thus it is unlikely that the sex-ratio should fall between 
■95 and I'OO ; this would again require at least 20 children. It will be seen that 
in the sex-ratio of the offspring's sibship we have a tendency for the frequency to 
lump up in these terminal groups, although their range is only half that of the 
other elementary frequency groups. This is almost entirely due to families in 
which there ai-e no males or no females. Undoubtedly certain individuals tend to 
produce offspring all of one sex, either ^jer se or because they are mated with a 
special type of consort. The latter reason seems the more probable, because, in the 
case of thoroughbred horses, where the matings change there appears to be no similar 
tendency for produce all of one sex to occur. A special study of cases in man and 
other animals in which for the same mating there is constancy of sex would be 
very instructive. It is probably due, as the tendency shows no sign in our tables of 
inheritance, to some physical characteristic of the individual which remains wholly 
dormant until it is affected by a corresponding characteristic in the mate. In work- 
ing the moments and products, the frequency has been centered at the middle of the 
elementary range. This is probably not true in the case of the extreme elements 
in the offsprings' sibships in Tables I. and II., but tlie actual centering was found 
to have little influence on the correlations, and made no modification in the funda- 
mental significance of the results. Sheppard's corrections were used. 
In Tables I. and II. the sex-ratio of a family is correlated with that of the 
father's and mother's sibships respectively. It may be said : Why not, when 
dealing with the latter sibship, leave out father and mother successively in 
calculating the sex-ratio? A somewhat similar method has been adopted by 
Francis Galton for another purpose, and is justified in his case if we may assume 
that the chance of male or female is practically one-half for each family. It does 
