VAEIATIONS OF GARTER-SNAKES. 147 



be considered as the center of dispersal for elegans, a fact that explains 

 the wider separation of the high plateau and Rocky Mountain indi- 

 viduals of elegans from hammondi as shown by the fact that the 

 former apparently never have two preoculars. The hypothesis may 

 be further tested by examining the Pacific coast form of this group. 



I have already referred to the fact that elegans apparently tends 

 to become reduced as to the number of scale rows on the western slope 

 of the Sierra Nevada-Cascade ranges. This reduction is hardly to 

 be detected when true elegans specimens (those with 19-21-19-17 or 

 21-19-17 rows and from the slope of the mountains) are alone con- 

 sidered, but when those from the foot of the mountains are considered 

 the reduction is strikingly evident. So pronounced is this that in 

 the interior valleys between the Sierra Nevada-Cascade and the coast 

 ranges from British Columbia to Kern County the form becomes 

 transformed into another (ordinoides) , with a dorsal scale formula 

 as small as any in the genus. This subject will be reverted to in 

 the discussion of the next form, and it is only necessary to point out 

 here (1) that the intergradation between the two forms is perfect, 

 and (2) that it takes place rapidly. Material is not available to fix 

 the limits of the intermediate region, but, as shown by the distribu- 

 tion of the two forms, it may be placed tentatively as the long west- 

 ward slope of the Sierra Nevada-Cascade range. It should be noted 

 further, however, that specimens from this intermediate region 

 (Fresno and Oakland, California) also exhibit occasionally two 

 preoculars, as does also true ordinoides (see p. 152), thus showing that 

 the trait occurs quite close to the range of hammondi. 



The average number of ventrals is about 169, even in northern 



New Mexico and Arizona, where it approaches the range of marcianus. 



This, with the smaller dorsal scale formula, seems to us to indicate 



that elegans has been derived from some form with a larger number of 



ventral plates than marcianus, in which the average number falls 



between 155-160. 



ORDINOIDES.a 



Description. — West of the Sierra Nevada-Cascade range, in Cali- 

 fornia, Oregon, and Washington, there occurs a form of garter-snakes 

 that has caused herpetologists considerable embarrassment. As here 



tt Thamnophis ordinoides (Baird and Girard), Proc. Acad. Nat. Sci. Phila., 1852, 

 p. 176. Includes E. leptocephalus Baird & Girard, E. cooperi and E. atrata Kenni- 

 coTT, E. elegans vidua Cope, T. rubristriata and T. leptocephalus ohjmpia Meek, and 

 in part the E. elegans infernalis, and E. hiscutata of various authors, but not of the 

 original describers. It seems that most writers have been vmcertain as to just what 

 form this name should apply. This uncertainty has been brought about by the 

 description of the form given by Baird and Girard in 1853 (1853, 33-34), which applies 

 equally to parietalis, elegans, and the present form, and indeed of the three specimens 

 listed the first is a parietalis, and the second a specimen of the present form with 

 21-19-17 rows. These specimens are labeled types of ordinoides, but it should be 

 recalled that ordinoides was described in 1852 and on the basis of a specimen from 



