424 PROCEEDINGS OF THE NATIONAL MUSEUM. vol. 53. 



seems to have originated eastward and gone around the mountains 

 in Georgia and across them in northern Virginia and Pennsylvania. 

 This convergence from two sides may account for the apparent 

 absence of fusca in western Kentucky, western Tennessee, and 

 northern ^lississippi. 



As to the seventh criterion, Desmognathus is found in the moun- 

 tains of North CaroUna in all situations; from far out in streams as 

 large as the Linville River to under the bark of dead trees 5 feet 

 from the ground. 



The eighth criterion is satisfied by the southward increase in the 

 size oi fusca already referred to. If we arrange the species in the 

 following order: 



J 7 ^ ^- 7 /• (auriculata-hrimleyorum, 



quadramaculata-monticola-fuscai , . . , ^ , 



[carolinensis-ocfirophaea, 



we shall get by starting with quadramaculata definite and continuous 

 variation in several characters; in size and importance of the dorsal 

 larval areas, ending with the striped ocJiropliaea; in the size and 

 importance of the lateral larval areas, ending with hrimleyorum; in 

 the hooking of the male mandible, ending with ochropJiaea; in the 

 shortening of the parasphenoid patches ending with hrimleyorum; in 

 the loss of the male vomerines and the shortening of the series in 

 both sexes (not much change in this character or in that of discrep- 

 ancy in size between the sexes takes place after fusca is reached) ; in 

 the weakening of the hmbs and the lengthening of the body, ending 

 with hrimleyorum; in terrestrialism , ending with ocJirophaea. 



In the series ending with hrimleyorum there is geographic continuity 

 from the southern Alleghenies to Arkansas, but in the series ending 

 with ocJiropTiaea we apparently leave the mountains to develop /wsca 

 and enter them again to obtain carolinensis. 



As an hypothesis to account for this pecuhar distribution the fol- 

 lowing is put forward with considerable diffidence, as it rests on no 

 fossil evidence: 



C. C. Adams states that during the last glaciation there existed 

 south of the ice first a tundra belt, then a coniferous forest, and lastly 

 a deciduous forest. Now monticola fmc] fusca are in general restricted 

 to deciduous forests. We may place their glacial distribution as the 

 glacial deciduous forest — monticola in the uplands and fusca in the 

 lower land. 



Possibly at this time auriculata and hrimleyorum were developed 

 in the extreme south, but more probably their differentiation is post- 

 glacial. 



During the last glaciation specimens oi fusca invaded the glacial 

 coniferous forest, acquiring less aquatic habits and an adaptation 

 to a lower temperature, thereby or therewith beiag converted into 

 carolinensis. 



