■ r SEA-FISHERIES LABORATORY. 249 



crater wliicli it is desired to investigate will capture all 

 the organisms, great or small, present in this water, and 

 the use of such a net is, of course, impracticable. Neither 

 could we attempt successfully to obtain a representative 

 sample of all the plankton organisms present by using a 

 series of nets of different meshes, since, just as in the case 

 of the trawl-nets, there is nothing to show how many hauls 

 of each kind of net should be made so that approximately 

 similar samples of all the kinds of organisms in the water 

 might be obtained. The difficulty is, in fact, a much 

 more serious one in the case of plankton fishing than in 

 the case of trawling for the investigation of length fre- 

 quencies in the fish population ■ inhabiting the ground 

 sampled ; for the error involves the numbers of kinds, or 

 species, actually present. It is impossible that any com- 

 bination of ordinary nets can actually obtain a true 

 sample of all the species present except by chance, while 

 if the combination be changed, the relative numbers of 

 the species obtained must also change. 



The table on p. 269 giving the values of the co- 

 efficient k continues the investigations of former years. 

 These figures have been calculated by equating the 



r u 



expression / ax s dx to the sum of the average 



J u 



weights of the sample, the series being continuous 

 between the limits L x and L 2 . The method assumes that 

 the function ax 3 represents the increase of weight with 

 increasing length, x being the length, and a a constant. 

 This relation is, as I showed in last year's report, not 

 quite accurate, but the error involved by employing it is 

 probably quite insignificant. The object in making these 

 evaluations of the coefficient k was originally the deter- 

 mination of the differences in " condition " of plaice on 

 different grounds ; and it was assumed that k would be 



