THE VARIATION, RACES AND MIGRATIONS OF THE MACKEREL. 283 



magnitude of the mean numbers of dorsal finlets ; the inverse order 

 of magnitude for the second dorsal values is given in brackets. 



It will be seen that there is a conspicuous lack of correlation between 

 the mean values for the various local groups of European fish. Kerry 

 and the North Sea, which have the highest average number of finlets, 

 ought to have the lowest number of dorsal rays, instead of which these 

 values are among the highest observed. On the other hand, the 

 combined Brest and Scilly values occupy approximately their correct 

 positions in the sequence. 



It is clear, therefore, that the means cannot be regarded as correctly 

 representative of the local groups of fish, and that no racial affinities or 

 differences can be based on the figures as they stand. In view of the 

 indubitable correlation between the two characters under discussion, I 

 am inclined to attribute the inaccuracy of the means to the slight 

 amount of variation in these characters among European fish, which 

 renders necessary a much larger number of data than tliose at my 

 disposal. A sufficient increase in the number of observations for each 

 region would render the means more truly representative, and ought 

 to reveal the correlation that must be exhibited before the values can be 

 regarded as reliable. 



Indirectly this explanation indeed can be shown to be correct by 

 amalgamating those groups which we have already seen upon other 

 grounds to be closely related. The order is the same as in the preceding 

 list. 



This table shows that when the local groups of fish are amalgamated 

 in the manner described, the means for finlets and second dorsal fin 

 are distributed in the order demanded by the correlation of the two 

 characters. 



This result is not obtainable by any other mode of amalgamation. 

 Thus, if the Kerry and North Sea groups are merged together, on the 



NEW SERIES. — VOL. V. NO. 3. T 



