Distribution and Taxonomy of Blarina 103 



104), tail length 19.7 (14-27), and hind foot length 11.2(10.0-12.5). 



Although specimens from central Tennessee are scarce, preliminary 

 analysis of cranial measurements indicated that specimens from this 

 area might be intermediate in size. A stepwise discriminant analysis was 

 conducted on 265 B. carolinensis and 193 B. brevicauda with complete 

 measurement data. The five previously described cranial measurements 

 were used in this analysis and the specimens represent localities 

 throughout the Southeast. Twenty-one central Tennessee specimens 

 were then compared to these target samples. Seven of the central Ten- 

 nessee specimens lacked condylobasal length and maxillary breadth 

 measurements and two others lacked only maxillary breadth measure- 

 ments due to breakage. Missing data were estimated for the nine speci- 

 mens using the REGR option in the PAM subroutine of the Biomedical 

 Computer Programs (Brown and Dixon 1979). In order to obtain a 

 visual representation, the first two canonical variables were computed 

 and plotted as described by Rao (1952) and used by Lawrence and Bos- 

 sert (1967, 1969), Gipson et al. (1974), Kirkland and Van Deusen (1979), 

 Parkinson (1979), Diersing (1980), and others. The Biomedical program 

 PAM was used for these calculations. 



RESULTS 



The number of specimens examined from any one county varied 

 from one to seventy-five. Small ranges in cranial measurements from 

 large series suggest that small samples, other than from near the zone of 

 contact, can usually be considered representative of the local popula- 

 tion. Only slight overlap was found between the cranial measurements 

 of all B. brevicauda and B. carolinensis (Table 1). Cranial measurements 

 (mm) of two very recently weaned B. brevicauda from Alabama were 

 condylobasal length 21.0 (broken); cranial breadth 11.2, 11.9; interorbi- 

 tal breadth 5.5, 5.8; and maxillary toothrow length 8.6, 8.9, maxillary 

 breadth 7.6, 7. 9. The lower range of each of these measurements is as 

 great as or greater than the upper range of the same measurements from 

 a mixed age sample of B. carolinensis from the Southeast (Table 1). 



Perimeters of the extreme ranges of canonical variables for B. brev- 

 icauda and B. carolinensis are shown in Figure 1 to be nonoverlapping. 

 Individual specimens from central Tennessee are identified in this figure 

 by the first letter or letters of the county in which they were collected 

 (Anderson, Davidson, Franklin, Hickman, Marion, Putnam, Warren 

 and Wayne). 



The most striking pattern is the clustering of the Hickman, Putnam 

 and Warren county specimens well outside the range of both B. brevi- 

 cauda and B. carolinensis canonical clusters. The Marion County spec- 

 imen appears properly identified as a B. carolinensis, although it is 

 located on the edge of this distribution. Seven Anderson and one Frank- 



