7 . Seasonal trends were not present. 



8. The best available equations for con- 

 verting dressed to live weights utilize 

 the mean ratios of round to gutted 

 lengths at the mean length of the 

 samples. 



Estimating equations and standard errors 

 for scrod and large haddock from Georges 

 Bank and from the Nova Scotian area are set 

 forth in Table 5. A length-weight conversion 

 table based on these equations is given in 

 Appendix Table A10. It will be noted that all 

 four equations estimate very similar weights for 

 the same length. The loss of precision in using 

 the total regression equations rather than using 

 the separate equations derived from a sample 

 from each trip is estimated in Table 6. The 

 highest of these ratios of respective mean 

 squares indicates a 43 percent loss. However, it 

 would be impractical to try to obtain a 

 regression equation for each trip landed, and 

 for past data, this, of course, is impossible. 

 There is no apparent statistical justification for 



using finer breakdowns into year or area strata, 

 and samples for each month are not available. 

 Such differences that may actually be present 

 between these categories were obscured by the 

 large variation among samples. 



The differences found in the length-weight 

 regressions between Georges Bank and the 

 areas off Nova Scotia considered in this paper 

 agree with other evidence on the separation of 

 these stocks of haddock. Grosslein (1962) 

 reported that tag returns indicated a small 

 degree of movement between these two re- 

 gions. Hennemuth et al. (1964) found growth 

 rates of haddock collected from southern and 

 central Nova Scotia to be similar to each other 

 but differing from those on Georges Bank. 



In view of the large sampling error, the use 

 of length-weight regressions to compute the 

 numbers of fish in the catch is inefficient. 

 Since for this purpose what is needed is the 

 average weight per fish in the length-frequency 

 samples, a better procedure would be to obtain 

 the total weight of all fish measured and divide 

 by the number of fish to calculate the average 

 weight per fish in each sample. 



Table 5.— Regression statistics for haddock length-weight estimating equations ( log units). 



Description 



Equation 



Standard 



error of 



Y 



Large haddock from Georges Bank 



1 Y= -10.0580 + 2.8053X 



±0.0014 



Scrod haddock from Georges Bank 



2 Y= -9.2184 + 2.5864X 



±0.0027 



Large haddock from Nova Scotia area 



3 Y= -10.6191 + 2.9389 X 



±0.0027 



Scrod haddock from Nova Scotia area 



Y= -9.4570 + 2.6362X 



±0.0043 



1 Antilog e ofa = 0.00004284 



2 Antilog e ofa = 0.00009920 



3 Antilog e ofa = 0.00002444 



4 Antilog e ofa = 0.00007814 



