In the caudal -peduncle character, the 

 Hudson and Delaware samples proved to be 

 similar in both slope and intercept (table 2). 

 Therefore, least caudal -peduncle depth cannot 

 be used as a criterion for separating the Hudson 

 and Delaware populations . The Hudson and 

 James collections were compared for homogene- 

 ity of slopes, and the resulting "F" value was 

 not significant at the 1 -percent level, but the 

 intercepts were found to be highly significant 

 (table 3) . 



The collections from the James River 

 were significantly different in caudal peduncle 

 - standard length relation from the Rappahan- 

 nock (table 4), York (table 5), Potomac (table 6) 

 and Albemarle Sound samples (table 7). Using 

 the same relation the Potomac (table 8) and 

 York (table 9) samples were found to be signif- 

 icant from the Rappahannock's. Therefore, 

 when the regression of caudal peduncle on 

 standard length is used as a criterion, Chesa- 

 peake Bay appears to be composed of heterogene- 

 ous populations. 



In the Santee -Cooper System, the 

 population below the dam is seen to be deeper 

 in caudal peduncle depth than the upstream 

 samples (table 12). The upstream -downstream 

 comparison of this character resulted in an "F" 

 value that was significant at the 1 -percent level 

 (table 10). A comparison of the Albemarle 

 Sound collections with those of Santee -Cooper 

 (below dam) showed that the two samples were 

 not homogeneous in caudal peduncle - standard 

 length relation (table 11). 



In summary, the regression of caudal 

 peduncle depth on standard length differs sig- 

 nificantly between most areas; the homogeneity 

 of the Hudson and Delaware collections was the 

 only exception. 



Prepelvic di stance - - The overall covari- 

 ance procedure testing the homogeneity of 

 regression coefficients, for the regression of 

 prepelvic distance on standard length, resulted 

 in a highly significant "F" value (table 1). This 

 significant value points out that the samples 

 used in this study do not belong to one homogene- 

 ous population. 



Using the prepelvic distance - standard 

 length relation as an index, the Hudson River 

 samples were compared with both the Delaware 

 and the James collections . Tables 2 and 3 show 

 that these comparisons were highly significant. 



With the exception of the James River 

 samples, the prepelvic distance - standard 

 length relation remained constant throughout the 

 specimens studied from Chesapeake Bay. The 

 James River collections diverged somewhat and 

 were found to be highly significant from the York 

 (table 5) and Rappahannock (table 4) samples . 

 A similarity in both slope and intercept was 

 found between the James and Potomac samples 

 (table 6) as well as the James and Albemarle 

 collections (table 7). No significance was en- 

 countered when the Potomac samples were com- 

 pared with those of the Rappahannock (table 8) 

 and also when the latter were compared with the 

 York River specimens (table 9). 



The Santee -Cooper collections proved to 

 be homogeneous in the prepelvic distance - 

 standard length relation as the upstream versus 

 downstream comparison was not significant in 

 either slope or intercept (table 10). The Santee - 

 Cooper (below dam) samples were found to be 

 similar in prepelvic distance with the Albemarle 

 Sound collections (table 11). 



In conclusion, the prepelvic distance - 

 body depth relation remains relatively constant 

 between areas of close geographic proximity. 

 The ranking numbers of table 12 may lead to 

 confusion because nonsignificance is present be- 

 tween some locations, however, the numbers do 

 show a slight north -south trend. 



Predorsai distance- - The homogeneity of 

 slopes within the nine location regressions, for 

 the regression of predorsai distance on standard 

 length, was tested by the appropriate covariance 

 procedure. The resulting "F" value was found 

 to be significant at the 1 -percent level (table 1) . 

 Comparisons between individual locations were 

 made to determine the cause of this heterogeneity. 



The regression of predorsai distance on 

 standard length for the Hudson and Delaware 

 samples was homogeneous in slope, but signif- 

 icantly different in intercept (table 2); this is 



