CABANGIDAE 31 



No auditory bulla apparent externally. 



Temporal crests well developed, and usually reaching the edge of 

 the cranium above the orbital cavity. 



Myodrome large and (except in Trachinotus) opening to the ex- 

 terior posteriorly. 



Opisthotic sometimes interposed more or less between the exoccip- 

 ital and pterotic, but usually scarcely at all, and never to the extent 

 that it is in the Scombridaa. 



Basisphenoid present and with a process descending to the para- 

 sphenoid. Eye with a bony sclerotic case not so well developed as in 

 most of the Scombridae. 



Suborbital ring complete and with a sensory tube; a suborbital 

 shelf variously developed or absent.* 



Nasals well developed and projecting. 



Premaxillaries protractile in all but the Scombroidinae ; their as- 

 cending processes of varying length from moderately short to very 

 long. 



Maxillary with a supplementary element on its upper posterior 

 edge in all but Oligoplites and Trachinotus. 



Preopercle unarmed except in the very young. 



Head of hyomandibular, where it joins the cranium, more or less 

 divided into two parts. 



Four pairs of superior pharyngeals present (three with teeth). 



Three basibranchials, the first in front of the arches and extend- 

 ing under the glossohyal. 



Clavicle more nearly vertical than in the Scombridae, extending 

 farther above the pectoral fin, and not so far anteriorly; never project- 

 ing much in front of hypocoracoid at its lower end. 



Actinosts, 4; moderately short; 2^ or S 1 /^ of them on hypercora- 

 coid. 



Postclavicle in two parts. 



*The worthlessness of the character of the presence or absence of the suborbital 

 shelf, which has been used in taxonomy, is shown by Oligoplites and Scomberoides 

 two genera of undoubted close relationship. In the latter genus the shelf is well 

 developed and in the former it is absent. It may, furthermore, even vary within a 

 genus. In Trachinotus kennedyi, from the Pacific Coast of Central America (five 

 specimens examined from 8 to 20 inches in length) there is no suborbital shelf de- 

 veloped. While in T. rhodopus, from the Pacific Coast of Central America (five 

 specimens examined from 6 to 12 inches in length), and in T. glaucus and T. caro- 

 linus from Florida (one specimen of each 12 and 18 inches in length) there is a 

 well developed suborbital shelf. It is not coossified with the suborbital chain but 

 connected by tissue. In T. ovatus from Formosa, however, (a specimen 30 inches 

 in length) the shelf is large and firmly coossified with the suborbital bone that 

 bears it. 



