16 BULLETIN 61, UNITED STATES NATIONAL MUSEUM. 



perpetual danger is encountered of correlating it with another in 

 which the labials may also be said to be 7, occasionally 8, but in which 

 this variation is geographic. As another example of this the usual 

 formula given of the ventral scutes of different forms usually includes 

 the extreme variation in number known in the group. It may be 

 easily seen that this also offers no clue to affinities, for any geographic 

 variation in the average that may exist and might serve to connect 

 different forms is in this way completely obscured. 



The nature of the variations may be determined of course by 

 taking the mean of the variation for the different sexes in different 

 localities throughout the range, it being necessary to distinguish 

 between the sexes, for if there is considerable sexual variation the 

 predominance of either in the series from particular localities may 

 obscure any geographic differences that may be present. This 

 requires a large series of specimens from each locality, however, which 

 we regret to say have rarely been available in this investigation. 

 However, notwithstanding the fact that sufficient material has not 

 been available in some instances to establish the actual extent of the 

 range of variation of the sexes in the different localities, we believe 

 that the data obtained is in nearly every case sufficient to indicate 

 the nature and trend of the modifications exhibited by the different 

 groups. 



Variation in scutellation. — It is not surprising to find that it is 

 largely upon the basis of the scutellation that most of the forms of 

 Thamnoi)his have been defined, for the number of scale rows, labial, 

 ventral, and subcaudal scutes have long been regarded by herpetolo- 

 gists as available specific characters. Indeed, as will be brought out 

 in the following discussion, these traits are, within limits, sufficiently 

 stable to admit of their employment as diagnostic characters in the 

 different forms of garter-snakes. Inasmuch, however, as similarity 

 in the number of scales has been held to indicate genetic relation- 

 ships, it is necessary to investigate the variations which they exhibit. 



The number of rows of dorsal scales has undoubtedly been con- 

 sidered the most important s^^stematic character in the determination 

 of the different forms of this genus, as well as in many other groups 

 of snakes. When an individual snake is examined these rows are 

 found to be arranged in longitudinal series, running the entire length 

 of the body, from the head to the tip of the tail. The scales of the 

 different rows are alternately arranged, so that transverse as well as 

 longitudinal series are formed. 



Variation in number of dorsal scale rows. — The number of longi- 

 tudinal scale rows on the body, i. e., anterior to the vent, is normally 

 an odd number, as the series are usually paired on either side of a 

 median dorsal row. If a specimen of some species (sirtalis, ^or exam- 

 ple) be examined, it will be found that on the anterior half of the 



