446 DUNN. 



On Hondo, pcropus of the high interior is very similar to the Korean 

 //. leechii, while the very specialized form on Yezo, rctardatus, has 

 evident affinities with pcropus. These two may well be the oldest 

 Hynobius stock of Japan, as none of the other groups have reached 

 Yezo. 



A rather isolated group, (stejnegeri of Kiusiu, kimurai of Hondo, 

 and naevius of Hondo and Shikoku), has affinities remotely with 

 leechii rather than with any of the other Japanese forms, and may 

 represent another invasion. The Formosan species, sonant, is much 

 like naevius. It may be a member of the naevius group, or the like- 

 ness may be parallelism. Indeed this series of four species may have 

 come from some mainland form rather different from leechii. 



A third group, the animals of the islands in the Korean straits, 

 connects leechii with nebulosus of Kiusiu, with randenburghi of Hondo, 

 and with the as yet undescribed form from Shikoku. Apparently 

 this group represents the most recent invasion. 



Pachypalaminus bovlengeri, is, save for its peculiar feet, very like 

 Hynobius vandenburghi of the same general region. This may be 

 taken in the absence of other evidence to indicate genetic relationship. 



Batrachupcrus sineyisis has the same foot modification, and what 

 are essentially weakened prevomers, and weakened vomerine denti- 

 tion. It has probably arisen from some form of Hynobius, many of 

 its characters, as color, loss of the fifth toe, and development of the 

 labial folds, suggesting H. keyserlingii. 



Ranodon sibiricus has the same dentition as Batrachupcrus but none 

 of the other characters of that animal. It has probably arisen inde- 

 pendently from some species of Hynobius. H. turkestanicus, as the 

 only Hynobius of that region, may be the closest relative. 



Onychodactylus, with two closely allied species, is developed for 

 mountain brook life, and the lungs have disappeared. The prevomers 

 are shortened, and larvae and breeding adults have claws. Otherwise 

 it is not very different from Hynobius, from some species of which it 

 has doubtless arisen. No modern species of Hynobius, however, is at 

 all close to Onychodactylus. 



The course of evolution of the family has been about as follows: 

 The primitive stock was a series of essentially similar species spread 

 over northern Asia. Owing, perhaps, to different immigrations, either 

 in space, or in time, or in both, more than one form of Hynobius is 

 found in Hondo, in Shikoku, and in Kiusiu. From this series of 

 essentially similar species modified forms have arisen in several places, 

 (Semiryechensk, Sze-chuan, Ussuri, Hondo). All these modified 



