552 BERTRAM G. SMITH 



different conditions. A strict application of the biogenetic law 

 would imply that the early stages of development are exclusively 

 palingenetic; but from the earliest stages we find adaptations 

 (e.g., the presence of yolk) that are prospective in their signifi- 

 cance and have to do with distinctively larval phases, while the 

 larval characters may be highly coenogenetic. It is the failure 

 to distinguish between palingenetic and coenogenetic features 

 of development that is mainly responsible for bringing the recap- 

 itulation theory into disrepute. With a clearer recognition of 

 the limitations in this field of study, we may yet question whether 

 the reaction against the validit}^ of the recapitulation theory 

 has not gone too far. As compared with anatomy and paleontol- 

 ogy, embryology is doubtless of little service in connecting up the 

 great groups of animals ; yet it is indispensable in the solution of 

 many special problems. 



In a comparative study of the breeding habits (Part I, Smith 

 '12) we must remember that we are dealing with characters that 

 occur very late in the ontogeny, having to do with the adult 

 rather than with the developing animal. Consequently such 

 characters are of no more phylogenetic value than the habits 

 in general and the family, generic and specific morphological 

 characteristics of the adult; they are exceedingly plastic and of 

 value for comparison only within a very limited range of forms. 



A phylogenetic interpretation of the methods of fertilization 

 in urodeles has been given in a previous paper (Smith '07 b) ; it 

 is here referred to with the remark that in view of the conclusions 

 reached in the present paper regarding the trend of evolution 

 in the urodeles, the series should be reversed. If we could go 

 back far enough in the phylogenj^ of the vertebrates, doubtless 

 we should find external fertilization to be the primitive condition 

 (e.g., as in Ampbioxus) ; but it is entirely possible that in Crypto- 

 branchus the method is secondarily acquired. 



As we should expect, the brooding habit of Cryptobranchus 

 is very similar to that of the closely-related Amphiuma. The 

 absence of a brooding habit in Necturus is noteworthy. Brood- 

 ing habits very similar to that of Cryptobranchus are found in 

 Desmognathiis and Plethodon, but with this important differ- 



