_vice versa_. (5) These quantities get fixed by natural selection
in a single race which always lives in the same environment, i.e., the
doses of maleness and femaleness in a given sex always bear practically
the same relation to each other. Hence the types are fixed and uniform.
(6) But different races are likely to have a different strength of
chemical sex-doses, so that when they are crossed, the ratios of
maleness to femaleness are upset. Often they are almost exactly equal,
which produces a type half male and half female--or 2/3, or 1/3, etc.
The proof of this theory is that it solved the problems. Goldschmidt
was able to work out the strengths of the doses of each sex in his
various individuals, and thereby to predict the exact grade of
intersexuality which would result from a given cross.

Riddle's work on pigeons [5,6] brings us much nearer to man, and
suggests the results noted by both Goldschmidt and Lillie. As in the
Free-Martin cattle, there is an apparent reversal of the sex
predisposition of the fertilized egg. As in the gypsy moths, different
grades of intersexes were observed. In the pigeons, it was found that
more yolk material tended to produce a larger proportion of females. The
most minute quantitative measurements were made of this factor, to
eliminate any possibility of error.

The chromosome mechanisms practically force us to suppose that about
half the eggs are originally predisposed to maleness, half to
femaleness. A pigeon's clutch normally consists of two eggs, one with a
large yolk and one with a small yolk. But the half-and-half numerical
relation of males to females varies considerably--i.e., not all the
large-yolked eggs produce females or all the small-yolked ones males.

Wild pigeons begin the season by throwing a predominance of males, and