ventral surface. 5) The dorsal stripe consistently involves the middorsal scale row and the adjacent half of the next row on each side, and is bright yellow in fitchi, but in parietalis it may be slightly wider, may be duller with more dusky suffusion, and its edges may be less sharply defined.

Intermediate and Atypical Populations

Of many specimens examined from eastern Oregon, Idaho, Utah, Wyoming and Colorado, few were typical of either parietalis or fitchi. Many were intermediate in some respects or showed a composite of characters of the two subspecies. No well-defined belt of intergradation exists, but the transition extends over more than a thousand miles, with local populations somewhat isolated and slightly differentiated along divergent lines. In view of this situation some plausibility could be claimed for any of several dividing lines between the subspecies. However, by far the most logical division is the Continental Divide; south of the Teton Range it constitutes a broad barrier separating eastern and western populations. Across Montana and Canada also it constitutes a more or less formidable barrier, with high altitudes and cold climates that probably are limiting to garter snakes. With few exceptions the snakes from east of the Continental Divide are more nearly like parietalis in the sum of their characters whereas those from west of the Divide are more nearly like fitchi.

In the Teton Range and in Yellowstone National Park these garter snakes occur in headwater streams up to the Continental Divide. KU 27956 from Two Ocean Lake 3-1/2 miles northeast of Moran, Teton County, Wyoming, agrees in its characters with fitchi, having the red lateral marks small and inconspicuous, discernible only on the anterior half of the body. The dorsolateral area is dark, almost black. The ventrals lack dark markings.

In Utah, populations of sirtalis occur in the drainages of the Bear, Weber and Sevier rivers and other smaller streams of the western half of the state. Obviously the species invaded Utah from the north, probably at a time when Lake Bonneville, the predecessor of the present Great Salt Lake, drained into the Snake River of Idaho. Van Denburgh and Slevin (1918:190) separated from their western "concinnus" and "infernalis" and allocated to parietalis the populations of Utah and southeastern Idaho, but presumably these authors were not familiar with typical parietalis of the Mid-west. Subsequent authors (Wright and Wright, 1957:834; Stebbins, 1954:505; Conant, 1958:328) have followed this arrangement. A re-examination of specimens from Utah, including living individuals collected at Bear Lake in the summer of 1959, indicates that they should be assigned to fitchi rather than to parietalis.

Likewise various specimens from the drainage basin of the Snake River in Idaho are predominantly fitchi in the sum of their characters, although they differ from that subspecies in its most typical form and resemble parietalis in some respects. KU 23133 from two miles east of Notus, Canyon County, Idaho, has the red crescents on the lower part of the sides (between scale rows six and seven) consistently developed on the anterior half of the body. KU 21873, a large female from Bannock County, Idaho, is exceptional in having small lateral black spots on the ventrals, resembling parietalis most closely in this respect. Also, it has the red lateral crescents unusually well developed; the first three series are conspicuous, those of the fourth series are consistently developed, and those of the fifth series show occasionally.

Forty-five specimens in the University of Colorado Museum from northwestern Colorado were subjected to pattern analysis. In three specimens the dorsolateral black area between the dorsal stripe and the lateral stripe on each side has no markings, and in eight others there is only an occasional fleck or crescent on the skin between the sixth and seventh scale rows. All others have the normal complement of dorsolateral crescents or flecks between the scales of rows three and four, four and five, and five and six. But, these specimens vary in extent of development of the crescents in the upper half of the dorsolateral area on each side—between scale rows six and seven, seven and eight, and eight and nine. Only six snakes show traces of the crescents of the uppermost series (between scale rows eight and nine). Development of these crescents is variable but in all the specimens the crescents are confined to the anterior half of the body. The crescents between rows six and seven and between seven and eight are present in 20 specimens and in ten of these the crescents are conspicuous and regularly arranged, often meeting and consequently form H-shaped markings. In most of the snakes the crescents are best developed in the second fifth of the body and disappear posteriorly. In five of the twenty, crescents between rows six and seven are fairly regular, but those between rows seven and eight are few and appear only sporadically. In eight specimens there are no crescents between either rows seven and eight or eight and nine. In eight others the crescents between rows six and seven are likewise absent, and only the crescents between rows three to six are present.

These specimens from Colorado also differ from typical parietalis in having the black spots on the anterolateral edges of the ventrals less developed. In three of the 45 these spots are lacking entirely and in four others they are few and small. In the majority of specimens the spots are from 1/4 to 1/5 the length of the ventrals. In approximately one-third of the specimens the spots are absent posterior to mid-body. In five specimens obtained at Sheridan Lake, Pennington County, South Dakota, in the Black Hills in August, 1960, dorsolateral areas are dark with red crescents small and inconspicuous, and with black spots either lacking from the ventrals or only faintly developed. In two specimens from Sundance, Crook County, northeastern Wyoming, the red crescents are small and inconspicuous also. In one of these specimens, KU 28028, small black spots are present in the corners of the ventrals, but in the other, KU 23654, the spots are absent.

In having the dorsolateral area consistently black, with the three uppermost series of red crescents reduced or absent, and in having the ventral black spots reduced or absent, these specimens from Colorado, Wyoming, and South Dakota differ from more eastern and more typical parietalis, and tend toward fitchi, even more strongly than some Idaho specimens tend toward parietalis. Nevertheless, all things considered, the Continental Divide is the most logical boundary between the two subspecies, even though occasional individuals and even local populations deviate from the general trend of characters from east to west.

Acknowledgments

Dr. Doris M. Cochran of the United States National Museum kindly furnished information concerning the type specimen of Eutainia dorsalis formerly in the National Museum collection but now lost. Dr. James S. Findley of the University of New Mexico and Dr. Ralph J.