16 BULLETIN 114, UNITED STATES NATIONAL MUSEUM. 



row and its final disappearance or its fusion with the next lower 

 scute. Usually the latter increases in size as the upper decreases, 

 but reduction in number is very evidently due to a flattening of the 

 head, and while the single anterior temporal is frequently larger than 

 either when both are present, it is generally narrower than the com- 

 bined width of the two, and in elapsoides the space between the upper 

 labials and the parietals has become decidedly narrowed (fig. 7). 



The loreal plate in the great majority of the forms of the genus is dis- 

 tinctly longer than high and nearly quadrilateral in outline (fig. 3). It 

 may be absent by anomaly in any form and is commonly absent in elap- 

 soides, its place being taken by a downward extension of the pre- 

 frontal (fig. 7). In some its shape has been noticeably altered; in 

 getulus and hoTbrooki it is commonly nearly square or higher than 

 long (fig. 4); in pyrrhomelaena its length is increased and its lower 

 posterior angle is strongly acute; in micropJiolis its shape is dis- 

 tinctly atypical. The assumption is inevitable that the ancestral 

 type has this scute longer than high and with nearly right angles. 



Variation in dentition. — Dental characters are of uncertain value 

 in a work of this kind. No relationships have been discovered by 

 their means that were not already evident from external characters. 

 The variability of the teeth is greater than that of the head scutes, 

 necessitating the examination of many skulls to determine the ex- 

 tremes and averages of the characteristics. The chief value in the 

 present study has been the confirmation derived from dental charac- 

 ters of the validity of the major groupings within the genus. 



Variation in color pattern. — ^There is no evidence of a fundamental 

 pattern from which all forms of the genus have been derived. On 

 the other hand, there are at least three styles of patterns, and, if we 

 include mexicana and alterna, probably two more. On the color 

 patterns alone the major groups in the genus can be recognized. 

 Two of the groups in particular {triangulum and getulus) show most 

 remarkable evolution in pattern. When we see the changes that may 

 take place in the color pattern within a single group of closely related 

 forms, we conclude that scant dependence can be placed upon differ- 

 ences and resemblances in pattern until these are supported by struc- 

 tural evidence. Furthermore, where structural differences are slight 

 or unreliable the color pattern may furnish conclusive evidence as to 

 relationships. 



SUBDIVISIONS OF THE GENUS LAMPROPELTIS. 



The many forms of king snakes fall naturally into three groups 

 (exclusive of the aberrant and little-known forms, alterna and mexi- 

 cana), recognizable on structural characters as follows: 



