590 
GEOFFREY SMITH. 
It must be remembered, moreover^tliat sex is not necessax’ily 
a simple unit character, inherited in its entirety as such ; 
thus sexual characters fall into two main divisions, primary 
and secondary, and the latter again may variously affect any 
of the organs or parts of the bod}\ A^ie will give reasons, 
however, in the next section, for assuming the existence of a 
sexual formative substance, male or female, which controls 
the development of both primaiy and secondary sexual 
characters, and for the present it is assumed that the male 
and female modifications of this substance are the allelo- 
morphs which segregate in the manner described above, and 
give rise to the half-hybrid nature of sex. 
2. On the Correlation between Primary and Secondary 
Sexual Characters. 
Various definitions have been given of primary and 
secondary sexual characters. In these studies the term 
“primary sexual characters” is applied to those characters 
which affect the differentiation of the actual generative organ, 
testis or ovary, in which the ova and spermatozoa are pro- 
duced, while all those sexual characters are considered 
secondary which affect the otlier parts of the body, e.g. 
generative ducts, copulatory or any other organs, external or 
internal, which differ in the two sexes. 
The fact that there is a physiological correlation between 
the state of development of the secondary sexual characters 
and the primary rejiroductive gland has been vaguely 
recognised from time immemoidal. Thus the knowledge that 
the castrated males of the liuman sjiecies and of many races 
of domestic animals show in various degrees an arrested 
development of the secondary sexual characters goes back to 
periods long antecedent to scientific biology. Put despite 
this long familiarity with certain fundamental facts, there 
does not exist even at the [iresent time a clear conception of 
the nature and limits of this correlation. Careful observation 
and a certain amount of experimental work have revealed a 
