18 BULLETIN 17 5, UNITED STATES NATIONAL MUSEUM 



ever, the range of individual variation is so great in these characters 

 that there is a wide overlap between most of the forms, and thus they 

 generally can be used as diagnostic specific criteria only in combination 

 with other less variable scale and pattern characters. For the genus 

 as a whole, the range of variation in ventrals is from 205 to 262, for 

 caudals from 47 to 84, while in c. deserticola alone the ventrals vary 

 from 214 to 259, and in only 25 specimens of d. deppei the caudals 

 vary from 52 to 79. 



A sexual difference in the numbers of ventrals and caudals is 

 apparent in every form. The number of ventrals on the average, and 

 generally in the extremes as well, is higher in females than in males 

 (fig. 7), while the number of caudals is higher in males (fig. 8). These 

 sexual differences are doubtless correlated with the dissimilar repro- 

 ductive functions of the two sexes. Thus, the larger number of 

 ventrals, as well as the larger number of scale rows, in females is 

 correlated with the longer body necessary for the accommodation of 

 the eggs, while the larger number of caudals in males is correlated with 

 the use of this appendage as a clasping organ in copulation. The 

 variation in the number of caudals is relatively considerably greater 

 than that in the number of ventrals. Thus, the average difference 

 in the number of ventrals is from 1 to 10 in the genus, for caudals 

 from 2 to 12. 



In every form for which the number of specimens available was 

 large enough to furnish conclusive evidence, there is a marked corre- 

 lation between the geographic variation apparent in the number of 

 scale rows and that observable in the ventrals and caudals. The 

 tendency is always toward decrease away from the center of distri- 

 bution for the form, furnishing further evidence for the theory that 

 there is a general trend toward dwarfing within most of the forms of 

 the genus. In the subspecies of melanoleucus the situation is appa- 

 rently different, but the series of specimens for these forms are too 

 small to warrant any conclusions being based upon them. 



In general, when any distinct geographic variation in the pro- 

 portionate tail length is observable, it is the opposite of that evident 

 in the numbers of scale rows, ventrals, and caudals in the same form. 

 This is doubtless to be explained by the close correlation between 

 caudals and tail length and by the fact that, since geographic varia- 

 tion is so much more marked in ventrals than in caudals, a general 

 geographic decrease simultaneously in both would result in a general 

 increase in the ratio between tail length and total length. In every 

 form the average proportionate tail length is greater in males than 

 in females. In the genus as a whole the tail length varies from 

 0.100 to 0.185 of the total length. 



Variation in numbers of labials and oculars. — The individual varia- 

 tion in the numbers of labials and oculars in most forms is very 



