VARTATTONS OF GARTER-SNAKES. 189 



This is best shown in the end forms. Thus in hutleri (IncHana, Ohio, 

 etc.), ordinoides (coast of Washington, Oregon, and northern CaH- 

 fornia), scalaris (Veracruz), and sumichrasti (Veracruz) the scutel- 

 lation is often identical, while the coloration in but one of these, 

 scalaris, can be relied upon as a differential character. As stated 

 above, the similarity of such forms often makes it difficult to dis- 

 tinguish between parallel and synthetic types, but an exact knowl- 

 edge of the value of differential characters and of geographic distri- 

 bution and trends of variation will allow the distinction to be inade. 

 While it is thus impossible to accept similarity in scutellation be- 

 tween the forms of different groups as always indicative of nearness 

 of relationship, it is significant that where the groups come together 

 geographicall}^ their representatives also come to resemble each 

 other closely in scutellation. Thus, in megaloys, angustirostris, 

 eques, and yroximus the labials are practically the same in number, 

 and the number of ventral scutes is as nearly the same as can be 

 expected for so variable a series. It is true that the usual dorsal 

 scale formula in megalo'ps and angustirostris is 21-19-17, while for 

 proximus and eques it is 19-17, but it must be remembered that it 

 is nowhere higher than 19-17 in the respective groups to which 

 eques and 'prox/imus belong, so that these forms resemble megalo'ps 

 and angustirostris more closely than the two latter are resembled 

 by any other forms of the Sauritus and Sirtalis groups. While, 

 therefore, similarity in scutellation does not necessarily in itself 

 (especially at the limits of the range of the groups) imply close rela- 

 tionship, if we grant the evidence of the other criteria that the center 

 of dispersal is in Mexico, there is no objection from this standpoint, 

 for angustirostris, megalops, eques, and proximus give every indica- 

 tion of being more closely related to each other than any other four 

 forms in the different groups. If this is true it restricts the center 

 of origin of the genus to northern Mexico. 



Two other criteria of Atlams may be considered together, i. e., 

 "continuity and direction of individual variation or modification 

 radiating from the center of origin along the highways of dispersal" 

 and "■ direction indicated by biogeographical affinities." Evidently, it 

 seems to us, both of these criteria are of importance in any investiga- 

 tion of this kind, for, while they are not concerned primarily with the 

 conditions at the center, they are supreme tests of the genuineness of 

 the lines of dispersal, which must be determined Ix^fore the center 

 can be located. The tests can be reformulated as follows: Are the 

 variations in the forms along any supposed lines of dispersal con- 

 tinuous, and is it probable geographically that the forms are related 

 in the supposed manner? For instance, I might divide the garter- 

 snake into several groups on the basis of similarity in the number 

 of scale rows, but if I then applied these tests we would find that 



