SPEARFISHES OF THE CENTRAL PACIFIC 



503 



80 



70 



60 



50 



o 



< 

 q: 

 O 



H 



a 

 uj 40 



I 



en 



30 



20 



■• MAKAIRA AMPLA 

 • o MAKAIRA AUDAX 

 -i ISTIOMPAX MARLINA 

 o TETRAPTURUS ANGUSTIROSTRIS 



200 

 FORK LENGTH (CM.) 



FiniRE 4. — Relation of length of pectoral fin to fork length. (Measurements by POFI have been supplemented by 

 measurements of the Hawaiian Division of Fish and Game (DFG) on specimens between 150 and 200 centimeter 

 fork length.) 



about one-fourth of tlic spread of the distribution. 

 Therefore, we consider that it is satisfactory to 

 compare pectoral fins by using the ratio, or 

 percentage, of fin length to fork length for speci- 

 mens of more than 200 cm. fork length. (Figiure 

 4 demonstrates, however, that this character is 

 of no value for separation of species.) 



An example, in which a considerable amount of 

 allometry is to be found, is that of the greatest 

 depth of body plotted against fork length, again 

 from the data collected by POFI (fig. 5). There 

 is quite obviously a considerable positive allom- 

 etry in atulax and amjda — which is as expected 

 from the observations that these species tend to 

 become more humpbacked in the larger iiuli- 

 viduals. We, therefore, conclude tliat if we use 

 the relative depth of the body we must use 

 regression analysis. Straight-line regressions are 

 satisfactory, for there is no visible curvUineaiuty 

 within the range of oiu- data. Another obvious 



435062 O— 58 2 



conclusion is that other measurements may be 

 compared to the depth of the body in a simple 

 ratio only if they happen to grow proportionately 

 to it. 



Using the graphic technique, we have decided 

 that the following body-part relationships are 

 sufficiently isometric over the range of our 

 samples to permit the use of simple ratios for 

 comparing species: (1) Tip of the snout to the 

 anterior edge of the orbit in relation to the length 

 of the head; (2) height of the anterior lobe of the 

 first dorsal to fork length; (3) length of pectoral 

 to fork length; (4) caudal spread to fork length; 

 and (5) height of the anterior lobe of the first anal 

 to the height of the anterior lobe of the first 

 dorsal. It is necessary to use regression analysis 

 for the relation between the greatest depth of the 

 body and the fork length, the head length and 

 the fork length, the height of the anterior lobe of 

 the first dorsal and the greatest body depth, and 



