188 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. 



species occur in the garter-snakes, I liave proof that different Hnes of 

 dispersal have been followed; and it will furthermore be recalled 

 that, as I have defined these groups, they are each composed of forms 

 which are distributed linearly. When I find, then, that nowhere 

 else in the range are all of the groups represented, I am justified in 

 concluding that the genus has had its origin in a general way in 

 northern Mexico. 



As a corollary to this criterion is Adams's first test, namely, that 

 the center of origin is the location of the greatest differentiation of a 

 type. If the lines of dispersal converge toward or radiate from the 

 center, in this general region and here only, it should be possible to 

 find representatives of each group or lines of directly related forms. 

 This is exactly what occurs in the genus Thamnoph'u^. In no region 

 except Mexico can there be found representatives of all of the four 

 groups of garter-snakes, wliile at least to the northward representa- 

 tives of but one or two groups are found toward the outlying parts 

 of the range. 



The application of either of the above tests, however, can not, in 

 my opinion, be trusted to indicate more than the general region in 

 which the center of origin of the garter-snakes is located, for in a 

 constricted area like Mexico and Central America the groups differ- 

 entiated in one part must of necessity remain together geograph- 

 ically until they emerge into a more extensive land area, such as North 

 and South America. This is exactly what occurs, for the four groups 

 occur throughout Mexico, although, as will be shown later, the}^ prob- 

 ably became differentiated in a particular part of this general region. 



Another test that has been suggested is the one laid down by Allen 

 that synthetic forms will be found at the center of origin. It is one 

 that naturally follows an acceptance of the theory of single centers 

 of origin, and is closely bound up with the first two criteria cited — ■ 

 the convergence of lines of dispersal and the location of the greatest 

 differentiation of a type. At the center of origin of the genus of 

 garter-snakes, then, theoretically all four groups should be repre- 

 sented, and the representatives of these groups should be more closely 

 related than any other representatives of the groups. This point 

 is difficult to test, owing to the fact that the different groups as 

 they diverge genetically and often geographically may converge in 

 appearance. This reduplication of forms in widely separated areas 

 occurs to such an extent in the garter-snakes as to make it often 

 difficult or even impossible to determine specimens without a knowl- 

 edge of the locality in which they were taken. This parallelism is 

 brought about by the reduction in size which takes place as the 

 groups depart from the center, as this reduction in size is accom- 

 panied by a reduction in scutellation. Thus every group has forms 

 which are practically identical in the number of scales in each series. 



