182 



THE RATE OF GROWTH 



[CH. 



curve of growth by any means. Still more obvious is it that the 

 several year-classes (if such they be) do not tally with the age- 

 composition of any ordinary population, nor agree with any ordinary 

 curve of mortality. But even if we had ten separate year-groups 

 represented here, which I most gravely doubt, all that we know of 

 the selective action of the drift-net forbids us to assume that we 

 are deahng with a fair random sample of the herring population; 

 so that, even though the number of rings did enable us to distinguish 

 the successive broods, we should still have no right to assume that 



Fig. 46. 



1 2 3 4 5 6 



Years 

 Mean curve of growth of Scottish (East Coast) herring. 



these annual broods actually combine in the proportions shewn, 

 to form the composite population. 



It is held by many (in the first" instance by Einar Lea) that we 

 may deduce the dimensions of a herring at each stage of its past 

 life from the corresponding dimensions of the rings upon its scales. 

 Some such relation nmst obviously exist, but it is an approximation 

 of the roughest kind. For it involves the assumption not only that 

 the scales add ring to ring regularly year to year, and that fish 

 and scale grow all the while at corresponding rates or in direct 

 proportion to one another, but also that the scale grows by mere 



