SEA-FISHERIES LABORATORY. 105 



" constants " are avoided. Further, the measurements of 

 length, and the deduction of modal sizes apply to samples 

 — that is all the fish taken in one haul of the trawl-net. 



What is a " representative sample "? It would be of 

 service if there were some criterion or test which might 

 be applied to a series of measurements of length, and 

 which would enable us to say whether or not the sample 

 catch represented truly the whole population of fish 

 resident on the sea bottom over which the trawl was 

 dragged. Unfortunately no such criterion exists, and 

 one can only say that a sample catch is representative if, 

 on trawling again and again, one gets similar results. 

 But one cannot do this. Biologists and actuaries, who 

 have to deal with statistical material, are able to make 

 conclusions as to the accuracy of the data by employing 

 frequency-curves. But it will be seen that this test is not 

 possible in the consideration of the statistics dealt with 

 in this Report. Nevertheless it is possible to determine 

 roughly whether or not a sample catch with a particular 

 net is, or is not, normal for the fishing ground and season 

 of the year considered. An accident to the trawl — the 

 tearing of the net, or the partial stoppage of the meshes 

 by accumulations of sea-weed or jellyfishes — is usually 

 indicated by the measurements. But one cannot be quite 

 sure whether or not such unusual measurements may not 

 after all truly represent the fish population on the sea 

 bottom. 



2. THE TABLES AND GRAPHS. 



The measurements represent the mean lengths in 

 centimetres of the fishes. The limits of error arc +0'5. In 

 a large catch, as for instance in some of Mr. Eccles' 

 figures where there may be several hundreds of fish in 

 some cm. groups, the limits of error are much smaller 



