(B. ) If, on the other hand, age is nore highly correlated v;ith 

 body length than v.lth scale length, we shovild expect some degree of 

 positive correlation between age and body length among fish v:ith 

 identical (or nearly so) scale length. To examine this possibility, 

 scales of SS9 fish taken in the fall commercial fishery for adults, 

 comprising four year classes, and ranging from 190-219 ram. in body 

 length, v/ere divided into 10 mm. classes according to scale length. 

 The coefficient of correlation betv/een body length and age v.':'. thin 

 each of these scale-length classes was then calculated. These cal- 

 culations (Table IV) show significant correlations betv.'een these 

 two quantities, with a low probability (P), averaging less than one 

 in one huncired, of arriving at such high values by chance alone, if 

 there were no correlation. These results could not have been modi- 

 fied by any influence of scale length, for the design of the analysis 

 precluded that. 



TABLE IV. -Correlation (r) Between Age and Body Length for 

 Fish Having Identical Scale Length Classification. 



i/ Number of specimens 



(C.) The results reached under A and 3 are conclusive; but, 

 nevertheless, consideration might be given to the possibility that 

 age is correlated in like degree with both bo(3y length and scale length. 

 Then the coefficient of correlation for age on body length with fish 

 having identical (or nearly so) scale length, and that for age on 

 scale length with fish having identical (or nearly so) body length 

 should be about equal. Then, if in our age determination, the authors 

 had been influenced by the a., .e of the scales instead of the number 

 of age marks, the correlation coefficient for age on body length should 



44 



