64 Christopher W. Brown and Carl H. Ernst 



zero. The canonical variable is then evaluated at the group mean for 

 each specimen, and all cases are plotted in a histogram to demonstrate 

 separation of distinct groups. Table 3 lists the constant and canonical 

 coefficients of the discriminating characters for each analysis. Figure 6 

 shows a comparison of the histogram from each discriminant analysis. 



Analysis was performed at the George Mason University Comput- 

 ing Services on the Cyber 170-720 computer system. The P-series of the 

 BMDP (Dixon and Brown 1979) was used, as were all default proce- 

 dures, except the second factor analysis, in which four factors were 

 requested. 



Specimens Examined: 

 Carnegie Museum (CM): S 9130; 36497, 40186, 40187, 40192, 54721, 

 91446, 91447, 91482-91484, 91582, 91583, 91677, 92053, 92056, 92057, 

 92063, 92065 



North Carolina State Museum (NCSM): 2347, 5744, 8035, 8041, 8121, 

 8520, 8725, 9638, 9655, 9772, 9879-9885, 9888, 10229, 10779, 10920, 

 11017, 11259, 11874, 11875, 12011, 12061, 12108, 12112, 12113, 12263, 

 12266, 12795, 12857, 12894, 12911, 13899, 14011, 14111, 14141, 15678, 

 15793, 15926, 16657, 16711, 17056, 17059, 17105, 17150, 19241, 19359, 

 19595, 19641,21808 



National Museum of Natural History (USNM): 8372, 9973, 10519, 

 14755 (2 specimens), 17959, 19970, 20651, 29362, 44313, 49958, 101858, 

 102714, 107879, 108687, 110487, 127601, 129094, 129759, 130167, 

 130168, 139618-139620, 145377, 156804,210092,218911. 



RESULTS 



Four factors accounting for 52% of the variance were produced in 

 the factor analysis of the first 13 characters. The variation in characters 

 CS and R correlated most closely with factor 1; ADS and DSM with 

 factor 2; RIL with factor 3; and LSL, RSL with factor 4. Character 

 variation that correlated less than 0.500 with any factor was not consi- 

 dered significant. The factor loadings and eigenvalues are summarized 

 in Table 2, and estimated factor scores for the 80 specimens used are 

 plotted in Figure 4. 



In the factor analysis of all 19 characters, 4 factors were requested 

 to limit the number produced. These accounted for a cumulation of 52% 

 of the variance, but the characters SVL, TL, HL, and POS correlated 

 most closely with factor 1; CS, R, and TL with factor 2; ADS and DSM 

 with factor 3; and MS with factor 4. These results are summarized in 

 Table 2. Factor scores for the 70 specimens used are plotted in Figure 5. 



Fig. 4. Scatterplots of estimated factor scores for specimens from the factor 

 analysis of the first 13 characters. Solid circles represent one or more specimens 

 of Crotalus h. horridus; open circles represent specimens of C. h. atricaudatus. 

 Tail measurements (characters CS and R) correlated most closely with factor 1; 

 dorsal scale rows (ADS and DSM) correlated with factor 2; infralabials (RIL) 

 correlated with factor 3; and supralabials (LSL and RSL) correlated with factor 

 4. 



