Jersey. Additional collections were 

 made on cruises of the Alhatross III 

 in 1955 and 1956. Care was taken to 

 correctly identify the silver hake and 

 not include the American or offshore 

 hake, Merluccius albidus, in this study. 

 The location and description of each 

 sample are shown in figure 1 and table 1. 



The characters measured conformed 

 to the descriptions of Hubbs and Lagler 

 (1947) and are listedin table 2. The fork 

 length was used as the independent 

 variable in all comparisons; all other 

 characters measured were employed as 

 dependent variables. The statistical 

 methods outlined in Snedecor ( 1946) for 

 the covariance analysis and a T-test 

 described by Ouenouille ( 1950) to deter- 

 mine the significance of the distance 

 between regression lines were used. 



The original data were plotted to 

 determine the relationship of the vari- 

 ous dependent variables on the inde- 

 pendent variable. A linear relationship 

 was found to exist between variables 

 for silver hake from 26 to 38 centime- 

 ters. Accordingly, only those fish in the 

 sample between 26 and 38 centimeters 

 in length were used in this study and 

 approximately 25 fish were collected in 

 each 1 -centimeter size group at each 

 sampling location. 



■ •FROM COMMERCIAL CATCH 1955 

 A - FROM COMMERCIAL CATCH 1956 

 • -ALBATROSS HI CRUISES 65 - 6G 

 O - ALBATROSS EI CRUISES 61-82 



75' 74' Tf ?;• Tf TCT »»• 6<* ST* «• 6S' «" 



Figure 1. --Sampling locations and source of samples for 

 the silver hake used in this study. 



CXienouille outlines two assumptions 

 that must be fulfilled in order to use 

 the T-test between regression lines: 

 (1) The variance of the samples must be 

 comparable, when the larger variance 

 of the two is divided by the smaller, the 

 resulting F-ratio should not be signifi- 

 cant; (2) the regression coefficients 

 must not be significantly different. The 

 above assumptions wfere fulfilled for 

 head length and length of the pelvic fin 

 only; all other parameters failed to 

 meet these assumptions. 



If conditions 1 and 2 are met then the 

 variances and coefficients may be 

 pooled and the two regression lines 

 reconnputed using a mutual regression 

 coefficient. The difference between the 

 lines may then be computed simply by 

 subtracting values of -one formula from 

 that of the other. The difference is then 

 tested to determine if it is significantly 

 different from zero. .An example of the 

 application of this test using data from 

 the present study is given in table 3. 



To further determine whether or not 

 there are significant differences be- 

 tween the silver hake collected from the 

 Gulf of Maine and those from Rhode 

 Island to New Jersey, covariance analy- 

 ses of eight morphometric characters 

 were calculated for each area. These 

 analyses are summarized in table 4. 

 Those characters showing a difference 

 at the 1 -percent level or higher were 

 head length, length of the pelvic fin, 

 length of the first ventral fin, and eye 

 diameter. These four characters were 

 used to separate the groups. The T-test 

 was applied to only two of the charac- 

 ters, head length and length of the 

 pelvic fin. The other two characters, 

 eye diameter and length of the first 

 ventral fin, were not applicable since 

 the variance of the samples was too 

 great. 



RESULTS 



Following Marr's (1957) suggestion, 

 the term "group" will be used to define 

 the various fractions of the population 

 since the distinctive characteristics 

 cannot be classed as being either 

 phenotypic or genotypic. 



