Sorex Mensural Discrimination 17 



statistical procedures in distinguishing S. cinereus and S. longirostris 

 from the southeastern United States, where the two species show a 

 broad area of sympatry, on the basis of standard body and cranial 

 measurements. 



MATERIALS AND METHODS 



We used univariate and multivariate statistics to examine 200 

 museum specimens for morphological variation. To provide for robustness 

 in our analysis and include any differences due to clinal variation, 

 we selected 50 specimens of each species from the southern portion 

 of its range in Georgia, North Carolina, and South Carolina and another 

 50 specimens from central and northern Virginia. A priori identifications 

 were based on specimen tag information. In addition, we used six 

 additional specimens of each species not used in the model building 

 process to test the model. These were measured to the nearest 0.01 

 mm with dial calipers under a disecting microscope. Specimens examined 

 are listed in the Appendix. 



Menzel measured the cranial characters to the nearest 0.01 mm 

 with a Wild M400 Stereo microscope. Images were received by an 

 Optronics VA-470 video camera and transferred to a 486 PC utilizing 

 Analytical Imaging Concepts (Irvine, California) imaging software and 

 stored in the TIF format. To assess the repeatability of the video measure- 

 ment system, a set of 10 specimens were measured three times each. 

 The set of 10 specimens was measured, then the order was randomized, 

 and the set was measured again, and finally the order was again randomized 

 and remeasured. Although video images could be stored for re-examination, 

 each specimen was rescanned and the system was recalibrated prior 

 to each remeasurement. 



Eleven cranial characters (Table 1) were measured on all individuals: 

 condylobasilar length (CBL), cranial breadth (CB), length of unicuspid 

 toothrow (LUT), length of 1st unicuspid (LU1), breadth of 1st unicuspid 

 (BUI), length of 3rd unicuspid (LU3), breadth of 3rd unicuspid (BU3), 

 length of 4th unicuspid (LU4), breadth of 4th unicuspid (BU4), length 

 of unicuspids 3 and 4 (LU34), and breadth across 2nd molars (BM2). 

 External body measurements (total length, tail length, and hind foot 

 length) and sex were recorded when available from specimen tags; 

 body length was calculated by subtracting tail length from total length. 

 Each specimen was assigned to one of 12 age classes based on the 

 criteria of Rudd (1955). 



Statistical analyses were performed with Systat 5.1a (Wilkenson 

 1989) and SPSS 4.01 (Norusis 1990). Univariate normality and homogeneity 

 of variance were tested by inspection of plotted residuals and Bartlett's 



