Figure 1. — The northwest Allanlic area, considered in length-weight regression analyses for squid, showing (A) survey strata (strata numbers circled) and ICNAK 

 (International Commission For the Northwest Atlantic Fisheries) Divisions (such as 5Zw) and (B) geographical areas. 



range. The length data, therefore, do not represent an unbiased 

 subsample of the survey catches. 



Frozen samples were thawed prior to analysis. Dorsal mantle 

 length was measured from the apex of the tail fin to the 

 anterodorsal protuberance, to the nearest millimeter (Fig. 2); 

 total weight was measured to the nearest gram; and sex, 

 maturity, and stomach content information was recorded. All 

 data were audited and stored on computer files for statistical 

 analysis. 



The form of the length-weight relationships was assumed to 

 be: 



W=ALi> 



where W= total weight (g), 



L = dorsal mantle length (cm), 

 A and b = coefficients of regression. 



Least squares regressions were fitted to the linearized form of 



this function: 



Y = a + bx 



where Y = log^ ff, 

 X = loge L, 

 a = log^ A, 

 b = coefficient of regression. 



Various regressions were fitted, with the Statistical Package 

 for the Social Sciences (SPSS) (Nie et al. 1975) SCAT- 

 TERGRAM subprogram, to combinations of the data, il- 

 lustrating effects of sex, season, year, and area differences on 

 the length-weight relationship. Pearson correlation coefficients 

 (r) were calculated for each regression to measure the strength of 

 the relationship and the goodness of the fit of the calculated 

 regression line to the empirical data. 



One-way analyses of covariance were conducted using the 

 Biomedical Computer Programs (BMDP) (Dixon and Brown 



