REVISION OF THE KI:N^G SNAKES. 9 



be constant for the genus and to be distinctively, although not greatly, 

 different from that in TliamnopJiis. 



As in any study of this kind, it becomes essential to distinguish 

 between several kinds of variation, particularly individual, sexual, 

 and geographic. Since in the present study a deficiency of material 

 has been a great handicap, usually not more than supportive reliance 

 can be placed upon statistical results. It is hoped, however, that 

 conclusions have been sufficiently conservative to allow for the 

 greater quantity of material that will be available to later workers. 



Variation in dorsal scale rows. — The dorsal scales are arranged in 

 alternate series throughout the length of body, an even number of 

 rows on each side of a middorsal series. The number of rows varies 

 in the genus from a maximum of 27 to a minimum of 15; but in no 

 single individual is the variation less than 2 nor more than 6 (with 

 the exception of a few examples of elapsoides to be presently men- 

 tioned). If we take an average example, say, a hoylii, and c unt 

 the number of rows of scales behind the neck, we will find there are 

 23 ; if now we trace the sixth row backward (counting upward from 

 the ventral plates) we wall find that at some point past the middle 

 of the body this row is lost or fused with the next row below, and 

 since this occurs almost simultaneously on both sides of the body, 

 the number of rows changes here to 21; if now we follow the fifth 

 row backward, we will find that this row, toward the posterior end 

 of the body, is either dropped or fused with the row below it, leaving 

 19; we call the dorsal scale formula in this case, therefore, 23-21-19. 

 If we take any other form in the genus that makes the change from 

 23 to 21 rows, we will find that this change always takes place by a 

 loss of the sixth row, due either to diminution and disappearance 

 or to fusion with the row below. Likewise any form or individual 

 in the genus that makes the change from 21 to 19 rows always does 

 so by loss of the fifth row or by its fusion with the fourth. We may 

 express these facts in figures as follows : 



23-21-19 

 6 5 



in which the upper line denotes the number of scale rows around the 

 body, and the figures of the lower line denote which row, counting 

 from the ventral plates upward, is involved in the change indicated 

 by the figures above it. From a study of all the forms in the genus 

 it is found that the change from any particular number of rows to 

 the next lower number always involves the same row, counting 

 upward from the ventrals. Expressing these results in the same 

 manner as above, we get the following: 



27-25-23-21-19-17-15 

 7 6 6 5 5 4 



