306 BULLETIN OF THE BUREAU OF FISHERIES 



3 per cent (when 1 year intervenes) to 20 per cent (when 7 years intervene). Tlie 

 calculated values were based on the scale radius. 



Jarvi, however, ignored the fact that the discrepancies were greater in his young 

 fish than in the old for corresponding numbers of intervening years. In his table 18 

 he considers only the 3-year and older fish. His 2-year fish, for instance, show a 

 deviation of 1.5 centimeters instead of 0.5 centimeter in the calculations of the first 

 year of life. 



Jarvi later (1924) discovered that the errors in computations varied somewhat 

 with the races of coregonids. The correction of 0.5 centimeter per year, referred to 

 above, applies only to individuals of a slow-growing race (Keitelesee, Pielavesi). 

 In a fast-growing race (Nilakka) the corrective factor must be doubled (1 centimeter 

 per year). And, further, if the anteroposterior diameter of the posterior field is 

 employed, the error in each case is reduced by approximately one-half. 



Lee (1920) treated the scale lengths (radii) and fish lengths of different species 

 statistically and concluded that the growth increment of scales is, on the average 

 for each species, a constanfproportion of the growth increment of the fish, but that 

 the length of the fish and the length of the scale are not proportional to each other. 

 The "phenomenon of apparent change in growth rate" is partly due to the method 

 of calculation, which ignores late scale formation, and is partly due to the segregation 

 of fish according to size. Lee expresses Fraser's (1916) correction for the fonner 



factor in the form of a formula, Zi = C+ -f^ (L-C), in which C is the length of the fish 



when scales first appear, L the length of the fish at death, V the scale dimension, 

 Li the computed length at the end of the fii-st year, and Vi the scale dimension to 

 the first annulus. 



Rich (1920) concluded, from a study of series of fry and yearlings of the chinook 

 salmon, that "the increase in the number of rings on the scnles and the increase in 

 the length of the anterior radii are proportionate to the increase in length of the 

 fish [p. 53]." 



Birtwistle (1921) recorded for the herring "the width of the respective summer 

 zones on the scales as percentages of the measured part of the scale — that is, the total 

 distance between the 'base line' and the outer edge of the striated portion of the 

 scale" — and found that the corresponding percentages decreased as older fish were 

 employed, and that it seemed "as if the whole scale shrinks up in the older fish and 

 shrinks the more, the older the fish is." 



Miss Shorriff (1922) obtained a fonnula {L = AV^ + BV+ C), which presumably 

 expresses mathematically the growth relation between the body, L, and scales, V, in 

 the marine herring. 



H. Thompson (1923, 1924) finds that in the haddock, fish and scales grow very 

 nearly proportionally. The disparity between empirical and calculated sizes is due 

 to the shoaling of better grown young haddock with the less well grown older fish, 

 and to the employment of scales other than the largest (on flanlc) on the body. "The 

 size of the first platelet is proportionally smaller than that of the fish by about J^ 

 centimeter, which must be added to the calculated first year size. If scales are taken 

 from other parts of the body, where they are even later in appearing (than on the 

 fiank), the error may increase to 2}^ centimeters [1923]." The author finds (H. 



