508 



FISHERY BULLETEST OF THE FISH AND WILDLIFE SERVICE 



100 



200 



400 



500 



Figure 9. 



300 

 WEIGHT ( POUNDS) 



-Mean and standard deviation from regression of the weight of marlins at 250 cm. (left) and 300 cm. (right) 

 fork lengths. (Names in parentheses indicate source of data in the literature.) 



growth. In the POFI data, head length plotted 

 against fork length shows slight positive allometry 

 in amj)la and slight negative allometry in audax. 

 The condition in marlina is intermediate, but too 

 few measurements are available to be conclusive. 

 Therefore, regression methods are indicated for all 

 species. 



When we compare the POFI data with those 

 published by Gregory and Conrad (1939) and 

 Conrad and LaMonte (1937), we find good agree- 

 ment between samples of the same species except 

 that marlina from the central Pacific have some- 

 what longer heads than from the New Zealand- 

 Australia sample (fig. 11). However, the number 

 of samples is so small and the overlap is so great 

 that we consider this difference to be only racial. 

 The differences between species, too, are so slight 

 that the character is almost useless for diagnostic 

 purposes. 

 Length of snout 



Length of snout from front of orbit was used by 

 Jordan and Evermann (1926) as well as by Nichols 



and LaMonte (1941) in an attempt to separate 

 these species of fish, no doubt because of the gen- 

 eral impression that marlina has the shorter and 

 stouter spear and audax and ampla the longer and 

 slenderer ones. When snout length was compared 

 with head length we found no evidence of allo- 

 metric growth; hence, we can compare snout 

 lengths by simple ratio. When this is done (fig. 

 12) for the published data and the POFI data we 

 find that appearances as to snout length are 

 misleading, for all samples of all three species 

 show remarkably similar ratios with the overlap 

 among species and between samples almost 

 complete in all cases. Spear stoutness was not 

 investigated because of the small amount of 

 data. Also, measuring the breadth and width 

 at the tip of the mandible, as we did, is not 

 satisfactory because of the allometric growth of 

 tlie mandible in ampla (see next section). 

 Length of mandible 



When this character is plotted from our POFI 

 measurements (fig. 13), we find a strikingly 



