The Evidence from Sexual Characters. 239 



occurs in several species of the dragon flies of the genus 

 Agrion, in which a certain number of females are of an 

 orange color, and thus differ from the males and ordi- 

 nary females. 



He suggests that this is probably a case of reversion, 

 for in the true Libellulse, whenever the sexes differ in 

 color, the females are always orange or yellow, so that, 

 supposing Agrion to be descended from some primordial 

 form having the characteristic sexual colors of the typi- 

 cal Libellulae, it would not be surprising that a tendency 

 to vary in this manner should occur in the females 

 alone. 



This explanation seems to apply to several of the re- 

 corded cases of female polymorphism, but not to all, and 

 we must acknowledge that in these cases, the female 

 shows, in a far greater degree than the male, a tendency 

 to deviate from the primitive form of the species, and to 

 give rise to new race modifications. 



We have already called attention to the fact that 

 among the Crustacea there are many cases of male poly- 

 morphism, and many cases of the same kind are known 

 among male insects; as well as many cases, besides those 

 I have mentioned, of female polymorphism. 



In many of the social insects we have most profound 

 structural modifications, and most complex instincts, 

 which can only have arisen in females; and as allied spe- 

 cies of social insects differ from each otlier in characters 

 which are confined to the females, we must acknowledge 

 that in these forms there is no lack on the part of this sex 

 of a power to give rise to hereditary race modifications. 



That facts of this kind present a serious diflSculty 1 

 cannot deny, but we must recollect thi-t our hypothesis 

 does not demand that the power to transmit variations 

 should be confined exclusively to males, but simply that 



