322 Heredity. 



The remarkable history of reciprocal crosses is, on the 

 whole, exactly what we should expect, and although 

 there are many difficulties, they are no greater than 

 the complexity of the subject would lead us to anticipate. 



The study of variation brings out a number of second- 

 ary laws, all of which might have been derived from our 

 view of the nature of heredity. 



The law that sexual offspring are more variable than 

 those produced asexually has just been discussed, and it 

 is clearly in perfect accordance Avith our view. 



Another most interesting and remarkable law — that 

 changed conditions do not act directly, but that they 

 cause subsequent generations to vary — receives a simple 

 explanation as soon as we recognize that variation is due 

 to the transmission of gemmules, not to the direct modi- 

 fying influence of external conditions. 



We can also understand why variation should itself 

 be hereditar\', why specific characters should be more 

 variable than generic characters, why species of large 

 genera should vary more than species of small genera, 

 why a part developed to an unusual degree or in an un- 

 usual way should also be extremely variable, and why 

 secondary sexual characters should show a marked ten- 

 dency to vary. 



The study of secondary sexual characters aids us, like 

 the study of hybrids and of variation, to analyze or dis- 

 entangle the influences of the two sexes in heredity. 

 These characters,- therefore, possess especial interest in 

 connection with our subject. They are found, upon ex- 

 amination, to present many striking peculiarities which 

 might have been directly deduced from our view of the 

 nature of heredity. 



As gemmules which are formed in the male body are 

 much more likely to be transmitted to descendants, and 



