• Two badly damaged larvae (estimated lengths. 6.6 and 7.9 mm.) were not included in the series. 



^ The distance from tip to anterior edge of mesethmoid, measured along the dorsal profile of the snout. 



' Length of larval pectoral fin in parentheses. 



nearest 0.1 mm. at low magnification (7X 

 objective lens) of a dissecting microscope, but 

 that of the smaller dimensions, such as snout 

 length, orbit diameter, premaxillary length, 

 body depth, etc., were made to the nearest 0.01 

 mm. at high magnification (SOX objective lens). 

 At the higher magnification each division on the 

 disc micrometer represented 0.02 mm. 



Plots of body part length against standard 

 length indicated varying degrees of nonlinear 

 growth in premaxillary length, snout length, 

 head length, and in body depth. Because snout 

 and head length were found to be affected 

 greatly by a rapid growth of the premaxillary, 

 snout-less-premaxillary length and head-less- 

 snout length were plotted (fig. 4) and compared 

 against standard length. 



The relation of these body parts to standard 

 length is best described by the allometric 

 growth equation, 



Y = bXa 

 where Y is the body part, X is the standard 



length, a is the ratio of the instantaneous 

 growth rates of Y and X, and b, is a constant 

 of proportionality, sometimes referred to as the 

 "initial growth constant" (Simpson, Roe, and 

 Lewontin, 1960). This equation when trans- 

 formed to logarithms 



Log Y = Log b + a Log X 

 is linear. 



Because not all the points for each body part 

 could be fitted by a single curve, they were sepa- 

 rated into several groups, and regression lines 

 were fitted to each group separately. The 

 original data are plotted in figure 4, and the 

 constants for the allometric growth equations 

 are given in table 4. The subdivision of the 

 data was made after they had been tested for 

 deviations from simple allometry as suggested 

 by Richards and Kavanaugh (1945). Such sub- 

 division of data seems justified on the basis of 

 the tests and on the premise that in nature 

 relative grovii^h may deviate considerably from 

 simple allometry, often revealing the existence 



U.S. FISH AND WILDLIFE SERVICE 



307 



