REPRODUCTION 209 



sities of environment may readily modify developmental pro- 

 cesses and should teach us caution in using such characters for 

 the interpretation of natural affinities. It is perfectly obvious 

 that the structural and physiological features by which species 

 may differ in their development do not always correspond to 

 the system, based on a knowledge of the adult, on which a 

 natural classification should rest. Forms undergoing meta- 

 morphosis have had a developmental history of their own, and 

 larval forms such as tadpoles are outside the cycle of recapitu- 

 lation, the ontogeny being broken by the intercalation of the 

 larval phasis. 



14 



