Table 1.— Number of trips sampled for haddock length-weig 



ht study 













Region 



Year 



Jan. 



Feb. 



Mar. 



Apr. 



May 



June 



July 



Sept. 



Dec. 



Western 



1931 



1/1* 3/1 



— 



— 



— 



2/1 



3/1 



— 



/I 



Georges 



1932 



2/2 - 



— 



— 



— 



5/1 



1/1 



— 



— 



Bank 



1933 



— — 



1/ 



— 



— 



— 



— 



— 



— 





1942 



— — 



/l 



— 



— 



— 



— 



— 



— 



Eastern 



1931 



5/ 



— 



— 



— 



4/ 



— 



3/3 



/I 



Georges 



1932 



1/1 



— 



1/ 



— 



— 



1/1 



— 



— 



Bank 



1941 



— — 



— 



— 



— 



— 



— 



— 



2/1 





1942 



— — 



3/3 



5/ 



— 



— 



— 



— 



— 



Browns 



1931 



— — 



— 



— 



1/ 



— 



— 



— 



— 



Bank 



1932 



- — 



— 



1/ 



— 



— 



— 



— 



— 



and 



1933 



— — 



2/ 



— 



— 



— 



— 



— 



— 



La Have 



1942 



— — 



2/1 



1/ 



— 



— 



— 



— 



— 



Bank 



1955 



— — 



1/1 



1/1 



— 



— 



— 



— 



— 



Western 



1931 



— — 



— 



— 



— 



— 



1/ 



— 



2/1 



Bank of 



1941 



— — 



— 



— 



— 



— 



— 



— 



1/ 



Nova 



1942 



- — 



— 



1/1 



— 



— 



— 



— 



— 



Scotia 





















large market category/scrod market category 



When utilizing covariance analyses it is 

 always possible that the difference is not due 

 to the factor examined, for example area, but 

 to some other factor. One possible confound- 

 ing factor could be the different size of fish 

 within the market category being examined 

 contributing to differences in length-weight 

 equations. The mean In length of the samples 

 are given in Appendix Table Al and visual 

 examination of these values does indicate large 

 differences in the size of the different samples. 



ESTIMATION OF SAMPLING VARIATION 



Subsamples 



The samples used in these pooled analyses 

 were known to consist of fish from several 

 carts for each trip. However, the data for each 

 cart (subsample) were not recorded separately. 



In April, 1942, landings of five trips from 

 eastern Georges Bank were sampled in an 

 attempt to measure variation within trips, i.e., 

 among subsamples. These samples were taken 

 over a 10-day period from landings of boats 

 fishing in the same section of eastern Georges 

 Bank in depths of 45 to 55 fathoms. Each 



subsample was composed of 25 fish taken from 

 a single cart, and from four to eight subsamples 

 were taken from each trip. All of these fish 

 were in the large size category. 



The analysis of covariance among subsam- 

 ples is presented in Appendix Table A2. There 

 was a significant difference among the adjusted 

 means of the subsamples. The mean square 

 among samples (trips) was not significant. 



The differences found between subsamples 

 could have been the result of varying lengths of 

 time or the position that the fish were kept in 

 the hold. Also, each part may have contained 

 fish caught in different sections of the general 

 area that the boat fished in. 



The mean square for among subsamples is 

 twice as large as that among samples. The 

 assumptions of the model would be violated if, 

 in fact, the difference was significant. The 

 inverted F-ratio (0.01222/0.0065 = 1.88), with 

 58 and 8 degrees of freedom does not, in fact, 

 exceed the tabular F at the 5 percent prob- 

 ability level. 



We may conclude that sample to sample 

 variation is negligible. This is not surprising 

 because the short time period and restricted 



