CA:^ADIAy FISHERIES EXPEDITIOy, 19Vrl5 113 



two categories, the one embracing a sample from the spawning grounds taken during 

 the spawning season (the spring fishery of the west coast) and the other including 

 immature fat herring taken in autumn on the north coast of Norway. 



It will be noticed that the individuals found among the grown fish in 1908 have 

 a considerably greater t^ than the autumn-caught immature fish from the north, a 

 similar difference is also observable in 1909. In 1910, however, the curve for the ma- 

 ture fish exhibits a different course, and is here distinctly bimodal ; during the autumn 

 of the same year, only a few specimens of this year-class were found among the im- 

 mature fish, and the following year none. After 1910, the curve maintains its pecu- 

 liar form, with the two modes, the position of which, as will be seen, corresponds to 

 the two modes in the two pairs of curves for 1908 and 1909. 



Similar conditions will be found to apply in the ease of the remaining growth 

 dimensions (^i t2 ti fs) save that the original difference between the two categories is 

 not so great as to produce a final bimodal curve, but only a compound curve. 



From the foregoing, it will be seen that the biological process which is termed 

 the dissociation of year-classes is a phenomenon which must be taken into consideration 

 in statistical investigations of the growth (and age) of herring. The phenomenon 

 itself would seem, as far as can be judged up to the present, to be of considerable 

 dimensions, and to be closely related to other phenomena such as maturation, migra- 

 tion, accumulation of fat, etc. It is therefore, in my opinion, worthy of closer study, 

 in the prosecution of which growth measurements may be of assistance, in view of the 

 relation between dissociation and size (growth). 



In the case of investigations dealing with the problem of growth during different 

 years, or in different waters, the phenomenon will appear as a complication, a new 

 and variable factor. It will therefore be necessary, in such investigations, to procure 

 material, the elements of which are as far as possible comparable. And as circums- 

 tances are, the best material would be that consisting of older fish, as offering greater 

 facilities for the procuring of representative samples. 



With regard to the question as to number and size of the samples, it will be seen 

 from the foregoing that various points need to be taken into consideration here. The 

 number of samples, and their size (i. e., the number of individuals contained) must 

 be determined according to the peculiarities of the stock to be dealt with, or of the 

 water under investigation, or of the problems which it is desired to solve. In the 

 case of the immature Norwegian herring, for instance, where, as we have seen, the 

 year-classes occur in several combinations, and dissociated according to size, a large 

 number of samples will be necessary; owing to the small number of year-classes re- 

 presented, however the samples need not be particularly large. The mature Norwe- 

 gian herring, on the other hand, in 1914 might, as we are now able to see, have been 

 dealt with through fewer samples than were collected, owing to a comparative stabi- 

 lity in the distribution of the year-classes with a single dominant group, which mark- 

 ed this year, as compared with the variability of the situation in 1916, where several 

 new year-classes cropped up. 



A water area containing several different tribes of herring will in particular re- 

 quire a larger number of samples than one where simpler conditions prevail. 



In dealing with problems which demand that the values operated with (e. g.. the 

 mean values for growth dimensions) shall be accurate, i. e.. with the smallest possible 

 degree of accidental error, it may be necessary to make the sample larger. 



It is impossible to lay down any definite rule or system for determining this ques- 

 tion, as the circumstances to be considered vary from place to place and from time 

 to time. In most cases, however, we may say that given a certain niimber of speci- 

 mens to be examined, there will be more chance of obtaining the best material when 

 these are distributed among a largen number of smaller samples, then if they are 

 massed in one or a few large ones. In the Norwegian investigations among the 

 grown herring, the material was at first collected by taking a few samples each year, 



