466 BI^LLETTN 58, UNITED STATES NATIONAL MUSEUM. 



The snakes of this genus are distinguished from AgJcistrodon at 

 the first glance by the absence of the reguUir shields on top of the 

 head, which are here replaced by scales more or less similar to those 

 of the body. 



It has not been demonstrated conclusively that they are gener- 

 ically identical with the numerous American pit-vipers of a similar 

 head scutellation, and which are usually known as Tngonoce2)halus 

 or Bothrops. The South American genus Lachesis is sufficiently 

 characterized by the peculiar scutellation of the tail. 



It is not even certain that all the snakes here enumerated as Tri- 

 meresurus really belong to the same genus. There are various indi- 

 cations that T. gramineus represents a group of species which might 

 profitably be separated from the others As a matter of fact, until 

 the cranial structure of all the various forms wliich make up the 

 bulk of the Crotalidse is known, generic combinations must be very 

 uncertain in this family. 



KEY TO THE SPECIES OF TRIMERESURUS OCCURRING IN FORMOSA AND THE RIU KlUS. 



a ^ A squarish loreal shield between nasal and upper preocular; nasal divided or semi, 

 divided; tail normal; color brownish. 

 b^ A single elongate subocular; more than 175 ventrals; more than 65 subcaudals. 

 c^ Less than 30 scale rows. 



d^ More than 195 ventrals T. mucrosquamatus , p. 467. 



d^ Less than 195 ventrals T. elegans. p. 470. 



c ^ More than 30 scale rows T. flavoviridis, p. 475. 



&^ A series of several small suboculars; less than 155 ventrals; less than 60 subcau- 

 dals T. okinavensis, p. 479. 



a- All elongate can thai shield between nasal and upper preocular; nasal flat, undi- 

 vided; tail prehensile; color green T. gramineus, p. 480. 



The five forms here recognized occur within our territory only in 

 Formosa and the Riu Kiu Ai'chipelago. T. gramineus extends from 

 the eastern Himalayas to Formosa, but has not as yet been found in 

 the islands to the northeast of the latter. It represents a group of 

 species not very closely allied to the other four. These fall again 

 into two well-defined groups, viz, T. olcinavensis on the one hand, 

 and T. mucrosguainatus , elegans, and Jlavoviridis on the other. The 

 last three are plainly of common origin, and T. mucrosquamatus, which 

 is supposed to occur in Assam as well as in Formosa, may conse- 

 quently be regarded as the mother species," T. elegans having developed 

 from it on the southern group of the Riu Kius by a decrease in the 

 number of scales and scutes, while T. Jlavoviridis which became iso- 

 lated in the middle and northern groups developed in the other direc- 

 tion, viz, by an increase in scale rows. T. okinavensis, finally, has 

 thus far only been found in the central and northern groups, and 

 no near relative is known either from the southern Riu Kius or from 



o See pp. 467-468 for a discussion of this species and its relation to T. jerdonii. 



