was also compared with the composite sample 

 of year classes from the Hudson River; t was 

 significant at the 5 percent level (table 35B). 



Results from the above tests of gill 

 raker counts point to a wide difference between 

 the Upper Chesapeake Bay and California pop- 

 ulation at the present time . The latter is much 

 more closely related to the Hudson River popula- 

 tion on the basis of this character. 



A comparison within populations from 

 Canada, Delaware River and Albemarle Sound . -- 

 Samples are available from the St. Lawrence 

 River, the Miramichi River in New Brunswick 

 and the Philip River in Nova Scotia. A com- 

 parison of the samples collected from the 

 Miramichi River in 1952, 1953, and 1955, was 

 made using chi- square as the test criterion; 

 the result based on the total number of gill rak- 

 ers was significant at the 5 percent level (table 

 36A) . However, no significant difference was 

 found when the same test was run based on the 

 number of gill rakers on the upper arm (table 

 36B). This indicates that there was little differ- 

 ence among these year classes . 



Samples collected in 1955 from the 

 Miramichi and Philip Rivers were compared by 

 at- test; the result using the total number of 

 gill rakers was not significant (table 37A). A 

 t - test using the number of gill rakers on the 

 upper arm was run comparing the collections of 

 1952, 1953 and 1955 of the Miramichi with the 

 1944 collection from the St. Lawrence River; 

 t was highly significant (table 37B). At- test 

 based on the total number of gill rakers was 

 run comparing samples of the three year classes 

 from the Miramichi River with those from the 

 Philip River; t was not significant (table 37C). 

 The results indicate that there is a difference 

 between the St. Lawrence River and the other 

 two rivers, the Philip and the Miramichi . 



An analysis of variance using the total 

 number of gill rakers was conducted for three 

 different year classes of specimens collected 

 from the Delaware River; the resulting F - 

 value was not significant (table 38) and indicates 

 that the samples studied are homogeneous. 



In the Albemarle Sound, gill raker counts 

 were made on several year classes and age 



groups. An analysis of variance using the total 

 number of gill rakers was run on these samples 

 and F proved to be highly significant (table 39) . 

 A chi -square test was run on the same data based 

 on the number of gill rakers on the upper arm; 

 it also was significant at the one percent level 

 (table 40). Samples from the 1953 year class 

 were compared by an analysis ofvariance; the 

 F was highly significant (table 41). The results 

 indicate that in the Albemarle Sound differences 

 exist among and within yeaj clarses. 



An overall comparison of several 

 populations. - - In order to get an overall picture 

 of the populations based on gill raker counts, 

 the means of the total number of gill rakers on 

 the first left branchial arch are shown in table 

 42. The locations are given from north to south 

 along the Atlantic Coast. A similar table was 

 made for the number of gill rakers on the upper 

 arm of the first left branchial arch (table 43) . 

 The means shown in these tables include differ- 

 ent year classes and age groups and present a 

 graphical comparison of the areas studied. 



A series of t - tests based on the total 

 number of gill rakers (unless otherwise indicated) 

 were carried out from observations of tables 42 

 and 43 . The following comparisons were made 

 in order to determine the relationship of samples 

 from different localities along the Atlantic Coast . 

 A sample from the Miramichi River in New 

 Brunswick was compared with the Santee- Cooper 

 River System sample for the 1953 year class; 

 no significance resulted (table 44A) . A test be^ 

 tween the Miramichi River sample versus the 

 Upper Chesapeake Bay subpopulation for the 1955 

 year class was highly significant (table 44B) . By 

 observation, there is a significant difference be- 

 tween the samples from the James and Miramichi 

 Rivers for the 1955 year class. The result of a 

 t - test between the York -Rappahannock sub- 

 population and the Miramichi River sample for 

 the 1955 year class was not significant (table 44C). 

 A comparison between the York -Rappahannock 

 subpopulation versus the Santee-Cooper River 

 System sample for the 1955 year class was sig- 

 nificant at the 5 percent level (table 45 A). How- 

 ever, a similar test based on the number of gill 

 rakers on the upper arm was not significant 

 (table 45B) . Table 45C shows that heterogeneity 

 exists between the Chesapeake Upper Bay sub- 

 population and the Santee-Cooper River System 



10 



