258 THE VARIATION, RACES AND MIGRATIONS OF THE MACKEREL. 



of the bars was 26'66, a number well within the range of the means 

 for North Sea and Channel samples, viz., from 26"50 to 26'94. 



§ 2. The Bars which cross or meet the Lateral Line. The frequency of 

 the variations in the number of these bars is given in Table C (p. 291). 



It will be noticed that, in spite of the greater exactitude with which 

 the number of these bars can be determined, the range of variation is 

 distinctly greater. In the case of the entire series the range covered only 

 11 bars ; in the present case the range is from 12 to 25, and covers conse- 

 quently 14 bars. When it is remembered that in the former case the 

 whole side, and in the present case only a part of the side of the fish, 

 was under examination, the observed difference in the range of variation 

 must be regarded as significant. I attribute the difference to the fact 

 that the field occupied by the bars in the first case is invariable, being 

 the whole side of the fish, while that which provides the material for 

 variation in the present case is in itself variable. The anterior boundary 

 is fixed by the shoulder-girdle, as in the entire series ; but the posterior 

 boundary of the field depends on the curvature of the lateral line in 

 that part of the body. The place where the bars cease to meet the 

 lateral line generally coincides with an abrupt downward bend in the 

 course of the line ; but, as the bend is sometimes absent, or takes place 

 in front of, or behind, its usual position, the number of bars which 

 meet the line is correspondingly increased or reduced. 



The number of bars affected by this irregularity in the curvature 

 of the lateral line is not great, and the error which is introduced by 

 it into the records of the variability of the number of bars consequently 

 tends to become smaller and smaller as the number of observations is 

 increased. But, as the mean values in Table C show, it is sufficient 

 to break down the high uniformity of results between the unit- 

 samples of any one region which was exhibited in Table B. 



Excluding the two small samples of 12 fish from Scilly, the modal 

 number is seen to be usually 18 ; but the exceptions to this rule are 

 more frequent in the present case than for the entire series, being 6 

 instead of 3. 



The means for the samples from Lowestoft, Eamsgate, and Plymouth 

 vary from 18-03 to 18-58 ; for Scilly and Brest from 18-64 to 18-65, and 

 for the Irish samples from 18-15 to 18-62. The American mean 

 is 18-88. 



If we exclude all samples of less than 90 fish, the means for the 

 North Sea and Channel vary from 1804 to 18-43, those for the other 

 localities being unaffected. 



Thus the American mean, as before, is the highest, and the lowest 

 means are found among the North Sea and Channel samples, as was 

 also the case for the entire series of bars; but the contrast between 



