504 WILLIAM K. GREGORY 



by the presence of but a single long dorsal fin, the spinous 

 portion often much reduced, or else formed of numerous short 

 spines. From the Scombriformes it may be distinguished by the 

 presence of the oesophageal pouches or when these are absent by 

 the gill raker-like knobs below the pseudobranchias. The small 

 cycloid scales (when present) of the Stromateidae, Icosteidas, 

 resemble those of many Scombriformes and of the Lampridae, 

 while the scales of theTetragonuridae, which are described as hard, 

 bony, adherent, ciliated, and grooved or strongly keeled, may 

 perhaps be compared with the cycloid, heavily ridged or keeled 

 scales of the Bramidae. The pelvic bones are free from the clavi- 

 cle in Tetragonuridae, Icosteidas, and in some Stromateidae, in 

 others more closely, but still movably, attached by ligament 

 (Boulenger). On the assumption that this loose attachment 

 is "a primitive character and not the result of specializa- 

 tion, such as occurs in some cases among true Acanthoptery- 

 gians," Boulenger, as we have said, removes the group from 

 the Scombriformes to the Percesoces. This enables him to 

 improve the technical definition of the Scombriformes, and, 

 furthermore, the Nomeiformes may well be remotely related to 

 the true Percesoces, by inheritance of the typical formulae of 

 10 + 14 in the vertebras and I, 5 in the Ventral fins. But this 

 does not seem to lessen the phylogenetic significance of the 

 many resemblances of this group to the Scombriformes. Con- 

 vergence plus the inheritance of primitive characters hardly 

 seem enough to account for the detailed resemblances between 

 the Butter-fish, Rhombus (Poronotus) triacanthus of the family 

 Stromateidae and the Common Pampano (Trachinotus carolinus) 

 of the family Carangidae, or the osteological similarities between 

 the Tetragonuridae and the Bramidae. If, as Boulenger holds, 

 the families in question cannot be regarded as Scombriformes 

 without rendering it impossible to define that group, then I 

 suggest that they be segregated as a division, Nomeiformes, of the 

 suborder Percomorphi, coordinate with and in the neighborhood 

 of the Scombriformes. If the Upper Cretaceous genera Omosoma 

 and Platycormus are correctly referred to this group, the Nomei- 

 formes are older than any known Scombroids, and as old as any 

 other known Acanthopterygians. 



