oratory, Honolulu) (table 6). 



All available data are included in table 7, 

 although the comparison is restricted to indi- 

 viduals between 7.0 and 17.0 mm. because the 

 full range of sizes for the three species was not 

 identical. The ranges of ratios of premaxillary 

 length to head length and the ratios of pre- 

 maxillary length to standard length are distinc- 

 tive for each species, and without overlap 

 among the three. The ratio of premaxillary 

 length to head length is largest in wahoo and 

 smallest in skipjack. The lower and upper lim- 

 its of the range for sierra mackerel (0.221:1- 

 1.250:1) are about 1.71 and 1.54 times greater, 

 respectively, than those for skipjack (0.129:1- 

 0.163:1), and the lower and upper limits of 

 the range for wahoo (0.332:1-0.413:1) are 2.57 

 and 2.53 times greater, respectively, than those 

 for skipjack. Logarithmic plots of the pre- 



T.\BLK 6. — Measurements iif standard length, head length, and 

 prcmaiillory length, and ratios of jtremaxillary length to 

 standard letigth and premaxillary length to head length of 

 larval and juvenile wahoo, sierra mackerel, and skipjack tuna 



maxillary length against standard length (fig. 

 5) show this more clearly. 



STANDARD LENGTH (MM. I 

 Figure 5. Scatter diagram of premaxillary lengths 

 plotted against standard length of larval and juvenile 

 wahoo, sierra mackerel, and skipjack tuna. Line 

 segments are regression lines fitted by the least- 

 squares method. 



These large differences in the ratios of pre- 

 maxillary length to head length suggest that 

 the size of the premaxillary may be of generic 

 significance. This view is contrary to Conrad's 

 (1938) opinion that the long, pointed premaxil- 

 lary, thought to be so characteristic of the 

 genus Acanthocybinm, "is an unfortunate illu- 

 sion." Conrad based his opinion upon the close- 

 ness of the ratios of premaxillary length to 

 skull length of Scomber and Acanthocyhium, 

 0.23:1 and 0.24:1, respectively; however, a 

 careful check of his ratio for Scomber indicates 

 a possible error. 



Conrad defined premaxillary length as "the 

 length of premaxillae anterior to the dermeth- 

 moid" and length of skull as the length "from 

 anterior tip of premaxillae to posterior tip of 

 supra-occipital crest." He did not give the 

 species name or the source of his Scomber 

 measurements, but judging from his numerous 

 citations of Allis (1903) concerning this genus, 

 it is very likely that the measurements were 

 obtained from the figure of S. scomber by Allis. 

 Using the reference points defined by Conrad, 



310 



U.S. FISH AND WILDLIFE SERVICE 



