in the case of intergradation between infraspec- 

 ific categories. The characters of the popula- 

 tions of the rivers of eastern Georgia are as 

 yet unknown. 



The two races which have been studied 

 most are those of the Hudson River and Chesa- 

 peake Bay. Tagging studies by Merriman 

 (1941: 44) and Vladykov and Wallace (1952: 172) 

 indicate little exchange between the striped 

 bass populations in Chesapeake Bay and Albe- 

 marle and Pamlico Sounds, North Carolina. 

 Some meristic data also indicate that they differ, 

 the latter are separable from their nearest 

 neighboring population in the James River on 

 the basis of dorsal fin spines. Final decision 

 on the status of this population is deferred until 

 the completion of studies now under way. 



MATERIALS 



This study is based entirely on young 

 striped bass which are those in their first sum- 

 mer or fall (to January 1). It seems certain 

 that these young were spawned in the river in 

 which they were found. The pelagic larvae are 

 swept back and forth by tidal currents . When 

 they are approximately 13 to 19 mm . in stand- 

 ard length the young are first found in small 

 schools over the beaches . Counts were made 

 on young which were 35 mm. or more in stand- 

 ard length; all had the definitive number of fin 



3/ 

 rays (table 1) ._ ' It is assumed that no sexual 



dimorphism exists for the characters enumer- 

 ated. 



Series were collected expressly for this 

 study. Most were taken from July through 

 October and an attempt was made to get mater- 

 ial from a wide geographic range in each river. 

 When possible, collections were made at differ- 

 ent times at each locality and some composite 

 samples, especially those from the Hudson 

 River, included as many as 10 samples. 

 Samples were taken with seines, the most ef- 

 fective of which was a 25 foot by 5 foot knotted 

 bag seine with one -fourth inch bar mesh. Var- 

 ious shore habitats were sampled over a period 

 of several years but the only types to yield 

 satisfactory numbers were sand, or sand-gravel 

 beaches located on points which were swept by 



tidal currents; 25 to 100 young were often 



3/ The tables appear at the end of this paper. 



collected in a single seine haul . Few young 

 were tatcen by trawling off shore near such 

 points although yearling striped bass were found. 

 Adjacent coves with silt bottom, with or without 

 aquatic vegetation, were seined but yielded few 

 striped bass. The quiet cove was the favored 

 habitat of the related white perch, Roccus 

 americanus (Gmelin) . Young striped bass were 

 often associated with young American shad, 

 Alosa sapidissima (Wilson), while white perch 

 were frequently found with the glut herring, 

 Alosa aestivalis (Mitchill) . 



METHODS 



Counts were made of spines in the first 

 dorsal fin, and of the soft rays in the second 

 dorsal and anal fins. In counting the dorsal and 

 anal soft rays, the method of Hubbs and Lagler 

 (1949: 10) was followed; the last two fin -ray 

 bases were counted as one. All dorsal spines 

 were enumerated. Because three persons counted 

 it is possible that this is a source of bias despite 

 the effort to assure agreement on methods . All 

 dorsal spine counts were rechecked by the author. 



Because the distributions of each of the 

 three characters are primarily binomial the chi- 

 square test was used (Snedecor, 1946) . In setting 

 up the data in tables 2-22 for this test, those data 

 for the numbers of first dorsal fin spines were 

 lumped as 9 or fewer vs . 10 or more; for soft 

 rays in the second dorsal, as 11 or fewer vs. 12 

 or more; for anal soft rays, as 10 or fewer vs. 

 11 or more; for the character index, as 13 or 

 fewer vs . 14 or more . It may be of interest to 

 report that the analysis of variance and the t-test 

 gave the same indication of significance and at 

 the same level in every case except for the data 

 in table 18 where the F-test gave an indication of 

 difference at the 5 percent level whereas the chi- 

 square indicates a difference at the 1 percent 

 level. The results of the tests (X and t values) 

 are included in the tables; the abbreviation N.S. 

 indicates that there was no significance at the 

 5 percent level . 



A meristic or character index is used in 

 studying correlated characters (tables 21 and 22) . 

 The computations (adding and subtracting) are 

 performed for the data on each specimen and 

 frequency distributions are then constructed. 

 Some use has been made of the "percentage of 



87 



