132 DEPARTMEyr OF THE J^AYAL SERVICE 



XII.— GROWTH. 



COMPARISON OF GROWTH IX SAMPLES OF SIMILAR AGE COMPOSITION. 



A number of samples were taken from the material available, for further treat- 

 ment with regard to growth. In making this selection, several points were taken into 

 consideration. In the first place, it was desirable to include samples from as many 

 areas possible. In addition to which, it would naturally be interesting to investigate 

 in this respect such samples as might be said to occupy a peculiar ]X)sition in the 

 material, e.g., the one exceptional sample from Newfoundland. It was necessary, 

 moreover, to select those offering the best scale material, the quality of this being of 

 more importance in growth measurements than in age determinations. And finally, 

 it was recognized that the growth material should include, as far as possible, material 

 illustrative of the manner in which growth proceeds during the different seasons of the 

 year. With regard to this seasonal growth of the herring, there is but little inform- 

 ation to be gleaned from the present material, some facts may, however, be brought 

 to light by examination of the samples containing young fish. 



The nature and quantity of the material will to a certain extent determine M'hat 

 problems may be taken up for consideration in these growth investigations: it is 

 undoubtedly best suited for a study of the growth in the different waters. This pro- 

 blem, then, will receive the greatest share of attention in the following examination, 

 and we shall, bearing in mind the results of the investigations as to age composition, 

 endeavour to arrive at a solution of the following questions: — 



1. Are those samples which have been found to be of uniform character, as 

 regards age composition, likewise uniform or similar with respect to growth 

 of the individuals ; and 



2. Are those samples or groups of the same, which differ in respect of age- 

 composition from the remainder,, likewise different from these in point of 

 individual growth? 



The two questions taken as one may be formulated thus : Can the grouping accord- 

 ing to similarity and dissimilarity of age-composition, as drawn up in the foregoing 

 chapter, be further supported and rendered more distinct by consideration of the 

 growth of the fish? 



In making comparisions of this nature, it would of course be an advantage to 

 have an opportunity of examining one or more year-classes common to all samples. 

 This will be done when we come to compare samples of like character as regards age- 

 composition ; having regard, however, to the great differences in this respect which 

 exist between the samples from the different waters, a very extensive mass of material, 

 with very large samples, would have been necessary in order to ensure that each sam- 

 ple contained a sufficient number of the particular year-classes. In the material as 

 it stands, we find, for instance, that a year-class which is well represented in tlie sam- 

 ples from Northumberland strait (the 1903 year-class) appears but very poorly so in 

 the Newfoundland samples. By proceeding according to this method, we should, in 

 comparing two samples from different localities in many, or possibly most cases, ac- 

 tually be comparing, pairs of values of which one only could be taken as accurate, 

 having been based on a great number of single observations, while the other would be 

 subject to a higher degree of error, being based on only a few observations. The error 

 in a difference between two values depends upon the separate errors in the two values 

 themselves, and will always be greater than the larger of these, with the method 

 referred to; therefore many comparisons would infallibly give unreliable results from 

 a statistical point of view. In the following comparisons, the greatest imiiortance 



