494 WILLIAM K. GREGORY 



Percesosces, and thus represents a considerable advance upon the 

 Isospondyli. These are the loss of the mesocoracoid, the lack of 

 open communication between the swim bladder (when present) , 

 and the gut, the separation of the parietals by the supraoccipital. 

 Archaic isospondylous characters are the abdominal or 

 subabdominal position of the pelvic fins (when present), 

 and the suspension of the shoulder girdle from the cranium 

 by a bony posttemporal. The latter bone is simple, non- 

 furcate, and immovably attached to or even fused with the 

 cranium. 



Progressive characters are the following: (i) In the ancestral 

 types (the Sticklebacks which, as shown by Gill 1 lead beautifully 

 into the Aulorhynchidse) the pelvis is either free or attached to the 

 backwardly produced coracoids (hypocoracoids) , but this con- 

 nection may be secondarily lost in the more specialized forms 

 through partial atrophy of the shoulder girdle; (2) the coracoids 

 (hypocoracoids) are much enlarged, forming so-called "infracla- 

 vicular plates " often enameled externally; (3) the anterior 

 vertebrae (except in Gasterosteids) are more or less modified or 

 coalesced, often forming a long tube; (4) the branchial arches 

 are always more or less reduced; (5) the branchial lamellae are 

 pectinated (Hemibran chii), or produced into tufts (Lophobranchii) ; 

 (6) the dorsal fin often has a spiny portion consisting of free spines, 

 and the anal fin also occasionally develops a spine; (7) the snout 

 in the Gasterosteidae is either conical or but slightly tubiform, 

 but in all the higher forms it is fully tubiform, the small mouth 

 being terminal and bounded solely by the premaxillaries ; (8) 

 the scales are small (Fistulariidae) , reduced (Aulorhynchidae) , 

 or absent (Gasterosteidae), progressively superseded by bony 

 scutes; the latter process culminates in the complete bony 

 cuirass of Amphisile, which is fused with the enlarged ribs and 

 other portions of the endoskeleton. 



In discussing the probable affinities of the Hemibranchii and 

 the Percesoces (excepting Sphyraena) , Dr. Starks 2 enumerates the 



1 Gill "On the Mutual Relations of the Hemibranchiate Fishes," Pror.. 

 Acad. Nat. Sci. Phila., 1884, p. 154. 



-"The Shoulder Girdle and Characteristic Osteology of the Hemi- 

 branchiate. Fishes. " Proc. U. S. Nat. Mus., Vol. XXV, 1902, p. 622 



