Cedar Vale, Chautauqua County, Kansas, but the specimen from there must now be assigned to cansensis on geographic grounds. Probably the specimen was not obtained from Cedar Vale itself for the habitat is not suitable there. Numerous specimens are known from 3 mi. W Cedar Vale, in Cowley County, Kansas, all of which are assigned to cansensis. Osgood's recorded locality is situated between this locality and the type locality of cansensis, which is 4 mi. E Sedan, Chautauqua County, Kansas. The distribution of cansensis also is shown in Fig. 1.

The probable geographic range of P. boylii is based on trapping data (see Fig. 1). The brush mouse is confined to systems of wooded cliffs in Kansas. The two subspecies seem to be separated by more than 80 miles of grasslands. Blair (1959) has postulated that in the northeastern part of its range P. b. attwateri is represented by disjunct, relict populations formed by diminishing montane or cool, moist environmental conditions. He has implied that the critical climatic change occurred during post-Wisconsin times, and that the isolation of these populations occurred so recently that no morphological differentiation has resulted in them. Inasmuch as the species is widely distributed in México, the southwestern United States, and in California, and has been recorded from the Pleistocene of California (Hay, 1927:323), it is reasonable to suppose that the species immigrated into Kansas from the southwest and that the immigration was in a generally northward or eastward direction. If long tail and large eyes are specializations for a scansorial mode of life (discussed below), then P. b. cansensis must be considered more primitive than P. b. attwateri for the eyes are less protuberant and the tail is shorter in P. b. cansensis than in the latter. I suggest that P. b. cansensis occurred in what is now known as Kansas before P. b. attwateri entered this area by way of the Ozark Mountains. The occurrence of a mouse of "the truei or boylei group" (Hibbard, 1955:213) in southwestern Kansas in the Jinglebob interglacial fauna of the Pleistocene adds little to support the thesis outlined above, but is not inconsistent with the thesis. Incidentally, the geographic distribution of P. boylii may differ somewhat from that shown by Blair (1959:fig. 5); whereas he has mapped the distribution of P. boylii to show disjunctivity in P. b. attwateri and homogeneity in the distribution of other subspecies of the brush mouse to the westward and southward, disjunctivity actually occurs frequently also in the western and southern subspecies.

In Kansas the brush mouse is confined to systems of