CllOCODiLiANS, LIZARDlfl, AND SNAKES. 705 



family heads, which are uot, with the exceptiou of the Houuilopsinae, 

 detiiied. It is not certain, therefore, that their contents are in all cases 

 properly limited or distributed. It remains a desideratum to discover 

 the characters of the natural divisions of the Colubridie, if any there 

 be. " The characters presented by Dumeril and Bibron, and by Giin- 

 tlier, are important but insufficient. For the definition of the genera 

 distinct characters exist, although the subject is one of much difficulty. 

 The object of definition being, as I imagine, precision, and the conse- 

 quent increased facility of determination, I have emjiloyed all struc- 

 tural characters whatever, and only neglected them where it is evident 

 that they are inconstant within the limits of a species. I find of the 

 greatest importance the grooved or nongrooved characters of the pos- 

 terior teeth, and the absence or number of the scale pits. The division 

 or nondivisiou of the anal scutum is also of much importance, although 

 in a very few genera (for example Xenodoti) it is not constant. Rely- 

 ing, as the system always must, on exact characters, I have not allowed 

 considerations of "physiognomy" to change a result where it contiicts 

 witli structure, which is, however, rarely the case. The tendency of some 

 authors to neglect characters and to depend on " physiognomy " destroys 

 precision and explains nothing, besides rendering identification of 

 species most laborious, resting as it must in that case on purely empir- 

 ical methods. I also do not use as generic characters the number of 

 rows of scales, or of labial scuta, believing that these are only available 

 in the distinction of species. 



In a few instances 1 have not been able to examine the skeletons of 

 genera of doubtful position, so that their reference to a family division 

 may yet have to be altered. I have, however, studied the fine series 

 in the museums of Paris and Washington, besides a considerable num- 

 ber in my own collection. For the characters of many genera which I 

 have not seen I have had to rely on the descriptions of others, espe- 

 cially those of the recently issued second edition of the Catalogue of 

 Snakes, in the British Museum, by Boulenger. 



III. PHYLOGENY. 



The earliest snakes known to paleontologists have been found in 

 the Upper Cretaceous beds of Europe and North America, those of the 

 former region being a little the older. Some of these are allied to 

 the Boas, while others are said by Sauvage to display characters of the 

 Epanodonta. But few specimens are known, and these are vertebriii, 

 so that their characters remain uncertain. Marsh has described some 

 ophidian vertebrie from the Laramie beds of the Rocky Mountain 

 region. I have described the genus Ilelayras from rather numerous 

 vertebra from the Puerco bed of New Mexico. An interesting pecul- 

 iarity of this genus is the imperfection of the zygantrum in some ver- 

 tebra' and the robustness of the neural si)ine. Next in time are the 

 Pahvophids, vertebra; of which Owen first discovered in the English 

 NAT MUS 1)8 45 



