622 PROCEEDINGS OF THE NATIONAL MUSEUM. vol. xxv. 



the Gasterosteoidea). The Synentog^naths are at once separated from 

 the Hemibranchs b}' the united inferior pharyngeals and the absence 

 of fin spines, as well as by numerous minor characters. 



The Hemibranchs show relationship with the Percesoces^ in having 

 the parapoph3^ses developed on all the abdominal vertebra?: in having 

 the supraclavicle, when present, small; in not having the exoccipitals 

 united above the basioccipital; and in having the supraclavicle, when 

 present, reduced in size. Fistularla and Aulostomus have processes 

 running backward from the epiotics, which are strikingly similar to 

 the epiotic processes possessed by all the Percesoces. Though they 

 serve the same purpose (that of muscle attachments) in both, the}^ are 

 somewliat different in character, being in the Percesoces processes 

 from and of the epiotics. while in the Aulostomoidea they are joined 

 by ligaments to the epiotics. The Hemiliranchs easily stand apart 

 from the Percesoces in having no opisthotics and usually no parietals; 

 in having the post-temporals simple, not typically forked; and in hay- 

 ing the postclavicle composed of a single piece when present (com- 

 posed of two pieces in the Percesoces). 



Dr. Gill in his excellent paper," though evidently having few inter- 

 nal characters at his command, has left little to be added in the arrange- 

 ment of the Hemibranchiate families. 



Gasterostem and closely related genera are the most generalized of 

 the Hemibranchs, The}^ are the only ones in the group having the 

 following typical characters: Anterior vertebrse unmodilied; suspen- 

 sorium and mouth normal; ribs typical; post-temporal approaching 

 the normally forked condition, and parietals present (the last .a super- 

 family character.) 



Dr. GilP has pointed out how the tube-mouthed forms have de- 

 scended in an unljroken line from Gasterosteu.s through Sj>/nac/na and 

 the family Aulorhynchidi\3, these constituting the superfamily Gas- 

 terostoidea. 



The Gasterostida^ and Aulorhynchida? should perhaps l)c regarded 

 as a single family, but following the lead of the above authority, they 

 are here kept separate, though the latter family is regarded "simply 

 as a convenient one at the most." 



Between the other families occur wider gaps, that are more or less 

 difficult to span. 



The Fistularid}^ and Aulostomida? are well placed in the same super- 

 family. They show in a marked degree how two families of undoubt- 

 edly close relationship having many characters in common can still 



' 111 each of these characters Sphyrsena is excepted. It seems to be a very much 

 more generaUzed form than the other members of the Percesoces. 



''On the Mutual Relations of the Hemibranchiate Fishes, Proc. Acad. Nat. Sci. 

 Phila., 1884, p. 154. 



3 Proc. Acad. Nat. Sci. Phila., 1862, p. 233. 



