and Holmberg (1950). Range of skull lengths, in 

 mm., in parentheses. 



Thunnus alalunga 19 Atlantic (125-167), 26 Pacific 



(99-152), 2 Indian (146-157) 

 Thuntnis albacares 21 Atlantic (98-196), 26 Pacific (49- 



149), 12 Indian (101-127) 

 Thunnus atlanlicus 26 (51-111) 

 Thunnus ohesus 21 Atlantic (119-215), 5 Pacific (97- 



237), 4 Indian (112-178) 

 Thunnus maccoyii 9 Aastralia (111-219), 4 South 



Africa (128-218), 4 SE. Pacific (207-238) 

 Thunnus thynnus thynnus 21 Atlantic (76-335), 1 South 



Africa (322) 

 Thunnus thynnus orientalis 40 (34-294), 1 SE. Pacific 



(290) 

 Thunnus tonggol 2 East Australia (122-128), 6 Indian 



(56-99). 



Details were corroborated by numerous partial 

 dissections of all species. We also examined many 

 additional skulls, postcranial skeletons, and radio- 

 graphs. 



We are grateful to Daniel M. Cohen, J. A. F. 

 Garrick, Brian J. Rothschild, Donald W. Strasburg, 

 Frank H. Talbot, and Stanley H. Weitzman for 

 making comments on various drafts of the manu- 

 script. Mildred H. Carrington and Gale G. Pasley 

 made most of the figures from our sketches. Alice 

 Holland typed many drafts of the manuscript . John 

 E. Fitch made it possible to use the plate blocks from 

 Godsil and Byers (1944) for figures 20 and 21. 



METHODS 



Dissections. — Although numerous partial dissec- 

 tions of both pi'eserved and fresli specimens were 

 made, the most thorough work was done with fresh 

 specimens on shipboard or, more often, with frozen 

 specimens in the laboratory. After the fish thawed, 

 colored latex was injected into the arteries and veins 

 through the lateral cutaneous branches that had been 

 exposed by removing the thick skin. behind the 

 pectoral fin base. Often the injection mass did not 

 reach the posterior ends of the cutaneous vessels or 

 the posterior commissure, but these vessels ordinarily 

 could be followed rather easily. In the other direc- 

 tion, the injection mass seldom penetrated beyond 

 the liver, partly because the deeper regions were not 

 completely tliawed. After the latex had set, the 

 lateral cutaneous system was studied. The ventral 

 wall of the body cavity was then removed and the 

 viscera drawn in situ. The ventral organs weie 

 then turned aside or removed to expose the swim- 



bladder, and this, in turn, was removed and the 

 dorsal fibrous connective tissue cut to expose the 

 kidney and ureters. The most difficult aspect of the 

 dissection involved exposing and tracing the anterior 

 arteries, which lie so far forward and are so deep 

 that they are difficult to reach without mutilating 

 the branchial region. Mter the appropriate obser- 

 vations were completed, the specimen was fleshed 

 and the skeleton cleaned. 



Counts and tneasurements. — Most external counts 

 and measurements were made on fresh or fi'ozen speci- 

 mens, some on preserved material, according to the 

 methods described by Marr and Schaefer (1949), with 

 the following exceptions. Our fork length is what 

 they called "total length." We measured length of 

 bony orbit rather than diameter of iris; in our com- 

 parisons, therefore, we used only our own data for 

 this character. (We do not recommend this pro- 

 cedure for future workers.) We did not measure 

 "pectoral insertion to insertion first dorsal," or 

 "length longest dorsal finlet," but we made the fol- 

 lowing measurements not mentioned by Marr and 

 Schaefer: Snout to insertion of pectoral fin. maximum 

 width of body, pelvic fin length, insertion of pelvic 

 fin to vent, tip of depressed pelvic fin to vent (all of 

 which are self-explanatory), snout length (snout tip 

 to front edge of bony orbit), and interorbital width 

 (least distance between dorsal rims of bony orbits 

 formed by frontal bones). 



Measurements made by the same worker of the 

 same specimen before and after freezing may diiTer 

 enough to negate a morphometric difference between 

 species, therefore, all morphometric characters should 

 be used only with rather wide margins for error. 



Skeletons. — Skull length was measured from the 

 anterior tip of the vomer to the lower posterior end of 

 the ankylosed first vertebral centrum. Individual 

 bones of the skull, pectoral girdle, and pelvic girdle 

 were compared simultaneously in all species, and 

 measurements were made only when proportional 

 differences were suspected. The only significant differ- 

 ences requiring mea.surements for definition were 

 among those already pointed out by Godsil and Byers 

 (1944) or Godsil and Holmberg (1950). Dial calipers 

 were used in most cases, and articulation cartilages 

 were removed from skull bones. The methods of 

 mensuration follow. 



Anterior articulating head of hyomandibula (See 

 fig. G). Length (B) was measured from the end of 

 the horizontal articulation surface (pterotic head) 

 to the most anterior point of the anterior articulation 



ANATOMY AND SYSTEMATICS OF TUNAS 



67 



