190 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. 



not only would the forms of each group fail to intergrade with each 

 other in their variations, but that it would be extremely improbable, 

 in view of their geographic location, that they were at all closely 

 related. As it is, each group is formed of a line of directly related 

 forms, the extremes of which are very distinct, hut those which are geo- 

 graphicalhj nearest together are but slightly different from each other. 



As previously stated, I believe the tests used above are valid, 

 and when applied show the utter untenableness of the hypothesis 

 put forward by Cope and Brown that the center of dispersal is in 

 southeastern United States and that sirtalis is the ancestral form. 

 From the latter standpoint the groups which we have distinguished 

 in the genus must be ignored and the forms connected in a manner 

 entirely antagonistic to the criterion that the lines of dispersal shall 

 be indicated by continuity in variation and be in harmony with 

 geograpluc probabilities. Note in the diagram of affinities given by 

 Brown (1904, 471) that the lines of genetically related forms fullfil 

 the requirements of neither of the above tests, and that, admitting 

 the validity of our own groups, we can get, at most, but three lines 

 radiating from eastern North America, and one of these does not 

 reach the southeastern States! If the criteria discussed above are 

 valid I believe that they indicate unquestionably that the center of 

 origin is in northern Mexico, and I may test the other criteria that 

 have been formulated by different writers. 



Two other criteria of Adams may be considered together. They 

 are: That the center of origin will be marked by a "dominance or 

 great abundance of individuals" and by a minimum ''dependence 

 upon a restricted habitat." Neither of these postulates is true in 

 this genus, for, although much yet remains to be discovered on the 

 habitat relations and abundance of the different forms, garter-snakes 

 are not noticeably abundant but rather rare in the desert and coastal 

 regions of northern Mexico and southern New Mexico and Arizona, 

 while in many places in North America they are almost increditably 

 numerous.'* Again, on the plateau of Mexico and in southern New 

 Mexico and Arizona, which forms the home of three of the groups, 

 the conditions are very arid, and these snakes are evidently confined 

 principally to the immediate vicinity of mountain streams and similar 

 environments where water is most abundant. On the other hand, 

 in the forested regions of North America, notably in eastern United 

 States, the forms, although showing habitat preferences, are of quite 

 general distribution. This is exactly what one might expect in view 

 of the preference shown by these snakes for a moist habitat. It has 

 been objected that this habitat preference argues against the con- 

 clusion that the genus originated in northern Mexico. I can not 

 see the force of this argument, for it seems to me that partially aquatic 



a See Van Denburgh (1897, 212) and Ruthven, p. 73, in this paper 



