124 KENDALL: NEW ENGLAND SALMONS. 



computing the percentile proportions and their averages, fractions were carried out to 

 two decimal places, then reduced to whole numbers b}^ regarding fractions of 0.5 per 

 cent and up as one per cent and disregarding all fractions below 0.5 per cent. The per- 

 centile proportions do not include all that might have been made, but they comprise 

 most of those used in technical descriptions. Three ratios were employed: (1) to length 

 of head, comprising seven dimensions pertaining to the head; (2) to standard length, 

 including 16 proportions, which include the head, fins, and body dimensions; (3) to the 

 length of caudal peduncle, i.e., its least depth to the distances from the adipose fin and 

 anal fin to upper and lower base of the caudal fin, respectivel3^ The percentages of the 

 first and third sets of ratios are naturally higher than those of the others. 



The greatest variation is exhibited by those structures and dimensions which undergo 

 the most change incident to growth and the breeding conditions, for the proportions 

 vary according to the changes. Therefore a percentile proportion of one variable to 

 another signifies very httle per se beyond the fact that the indicated proportion obtains 

 in that particular instance. A difference found by comparison of such a percentile 

 proportion of one individual with that of the same structure in another individual does 

 not necessarily signify that the structure or dimension actually differs in size or extent 

 wliich a difference in percentile, proportions seems to indicate. For example: given two 

 salmon in which the percentile proportions of the length of the eyes to the length of the 

 head differ by a certain per cent, the difference does not necessarily mean that the eye 

 of one is larger than the other. The eyes may be exactly the same size and still differ in 

 that ratio to the length of the head, due to the fact that the head of one is longer than 

 that of the other. The way percentile proportions varj^ may be illustrated as follows: 



Nail (1930, p. 303) indicates that the maxillary of the salmon increases its proportional 

 length gradually. He says that in a salmon 4 inches long, the maxillary extends back to 

 the middle of the pupil; at 6 inches, it extends to the posterior edge of the pupil; at 2 

 or 3 pounds, it extends back to the posterior edge of the eye; in larger fish, it extends 

 sUghtly beyond the posterior edge of the eye. This is a definite comparison of extent in 

 relation to a fixed point. But if the length of the maxillary is considered relatively to 

 the length of the head the percentile proportion may remain the same for the several 

 sizes of fish. Or it may progressively mcrease in proportion by actual increase of length 

 of the maxillary, the relative length of the head remaining the same. Again the per- 

 centile proportion would be greatly increased by growth of the maxillary and a pro- 

 portional shortening of the head. Or, still again, if, as in the breeding season, the head 

 is considerably increased in length, the percentile proportion of the maxillary would be 

 reduced. Thus the percentile proportions are relative. 



The averages of percentile proportions in tables 21 and 22 are of this nature. The in- 

 crease of one may be reflected in the decrease or even an unchanged proportion of 

 another. Therefore in determining the relative proportions of dimensions, the ratios 

 should be of those dimensions which are modified the least under conditions of growth, 

 sex, and breeding condition, or else of a large number of fish of the same age, sex, season, 

 and breeding condition. The first alternative is out of the question, for it is yet to be 



