CAXAfnw Fisin:i{n:>i ExrEDniox, lOi'rJS 



153 



The curves for increment need no further comment, the differences hetween the 

 Newfoundland samples and the remainder having been sutftciently indicated in the 

 foregoing. The length curves show how the Xewfoundland herring, starting modestly 

 enough, ends, owing to the favourable growth of the later years, with an average length 

 greater even than that of the fish from the Magdalen islands and Xorthumberland 

 strait (fig. 40). Fig. 41 shows that the somewhat slower growth of the Newfoundland 

 fish during the first two years enables the Atlantic herring, and, to a lesser degree, also 

 those from North Sydney, to maintain a part of the distance gained during the first 

 years, albeit the differences become more and more reduced with increasing age. 



11. Samples from Magdalen islands compared with the remainder. — The samples 

 from Magdalen islands have already been compared with those from Newfoundland, 

 and were found to differ greatly from these with regard to growth. It now remains to 

 compare them with the samples from Northumberland strait. North Sydney (includ- 

 ing the exceptional sample from St. George's bay) and with the Atlantic herring. 



Table 40 shows the samples from ^lagdalen islands compared with those from 

 Northumberland strait, by means of total averages for the 1903 year-class in both 

 waters. The table is arranged in exactly the same manner as table 32. 



Table 40. — Growth of herring from Magdalen Islands and Northumberland Strait 

 compared by help of total averages for the year-class 1903. 



Locality. 



Magdalf n Islands . . . 

 Xiirthnmberland Strait 



Difference (D) 



D 



d 



10- 



10-181 

 OO7I 



7 2fi 

 6 4.^ 

 081 



0.351 4-77 



4 62 

 4 94 

 0-32 



2-67 



tia 



73 

 72 

 01 



0-50 



It will be seen that the growth dimension t„ is the only one exhibiting a difference 

 which can really be called significant. Possibly 1 and t^ also differ; otherwise the 

 differences are unimportant in proportion to their errors. In order to test the one 

 really serious difference, the two samples from the Magdalen islands were compared 

 with each of the five from Northumberland strait, as regards the first five growth' 



dimensions. Table 41 shows the values of — here found. 



a 



Table 41. — Distribution of values of fraction -j- arising by comparison between each 



of the samples from Magdalen Islands and Northumberland Strait (year-class 

 1903). 



It will be noticed that the differences for t„ are greater than three times their 

 error in five out o? ten cases and exceed twice the error in nine, whence it would seem 

 6.5.51— n 



