VARIATIONS OF GARTER-SNAKES. 79 



ward in the valleys of the streams, as tiiey are sunk into the plains 

 by the increase in elevation, in a manner somewhat analogous 

 to the extension of the range of the prairie types into the plains 

 region (Pound and Clements, 1900, 74-78). Of course, the records 

 available are too meager to ascertain the actual state of affairs, but 

 it may be significant that within the region where the plains reach a 

 height of 5,000 feet the Colorado and Montana localities at which 

 radix has been taken (Pueblo, Fort Collins, and Greeley, Colorado, 

 and Threeforks, Montana) are all in the valleys of the Arkansas, 

 Platte, and Missouri rivers, wliich have cut their channels well below 

 the higher levels of the plateaus. 



It is unfortunate that reliance can not be placed on the locality 

 given for three specimens of radix in the U. S. National Museum 

 (Bridgers Pass), but the locality given is probably but a general one, 

 so that we have no evidence at present as to what extent, if any, this 

 form has pushed westward through this gap in the Rockies. The rec- 

 ord suggests, however, that radix may follow the tributaries of the 

 North Platte well across the Laramie plains, that form a break in the 

 continental divide at this point. 



To the eastward of the plains region the records indicate that radix 

 occurs throughout the prairie region. The northern limit to which 

 it is actually known to occur may be represented by a line drawn 

 from Pembina, North Dakota, southward through Fort Snelling, 

 Minnesota, and Madison, Wisconsin, to Racine, on Lake Michigan. 

 This line corresponds quite closely to that of the common boundary 

 between the northern coniferous forest and the prairie in these States, 

 and it might be concluded, since radix is a prairie form, that this 

 approximately represents the actual northern limit in this region. 

 There caii be little doubt, I think, that the margin of the prairie in 

 Minnesota and Wisconsin does determine the northern limit of the 

 principal distribution of radix in this region, but as the tension line 

 between the forest and prairie is not a sharp one, but marked by a 

 transition zone of brush prairie and open woods, while for a consider- 

 able distance within the forest area proper there are tongues and 

 patches of prairie conditions, it is very probable that radix will be 

 found in these habitats somewhat within the forest area. If a line 

 be drawn from Lake in the Woods to Mille Lac, Minnesota, and from 

 this point through Dunn, Eau Claire, and Jackson counties, Wisconsin, 

 to Racine, it will roughly intlicate the northern limit of the outlying 

 habitats related to the prairie, and I believe that radix will not be 

 found to occur much beyond this line. 



On the south and east the prairie peninsula is also limited by the 

 forest, and its border is the tension line between the prairie and the 

 deciduous forest of southeastern United States. As mapped by Pound 

 and Clements (1900), this boundary lies mostly to the north of Mis- 



