224 



TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY 



but only a general and ambiguous resemblance between the latter and Mugil (Fig. 138) 

 and Sphyrcena (Fig. 141) of the true Percesoces. Tate Regan (1929, p. 317) also notes 

 that " the large number of branchiostegals (9-1 S), the structure of the mouth, the absence of 

 spinous rays, the truly abdominal pelvic fins, etc., indicate their derivation from the Isospon- 

 dyli." Evidence from the branchiostegal rays is also cited by Hubbs (1919, p. 66) as 

 follows: "The branchiostegals of the Synentognathi (Belonidae, Scombresocidae, Hemirham- 

 phidse, Exocoetidse) are wholly similar to those of the typical Isospondyli; they are rather 

 numerous (ten in Euleptorhamphus), but not constant in number, flat, imbricate plates; 

 the uppermost skirting the lower margins of the opercula, and all with their lower edges 

 exposed. The characters of the branchiostegal rays of the Synentognathi strongly confirm 

 Regan's view that the resemblance between these fishes and the Percesoces is purely fic- 

 titious; the group should be placed among the typical soft-rayed fishes." 



The otoliths of the Synentognathi, according to G. Allan Frost (1926r, pp. 471-473), 

 retain elopine and clupeoid features in the details of the sagitta, while some species show 

 resemblances to the forms of the orders Salmopercae and Apodes. On the other hand, 

 these otoliths differ considerably from those of the Mycrocyprini. These facts, added to 

 other evidence, seem to justify Regan in placing the Synentognathi ahead of the Micro- 

 cyprini and Percomorphi in his ordinal classification. 



Thoracostei (Sticklebacks, Tube-mouths, Sea-horses, etc.) 



The skull (Fig. 102) of the rough-tailed stickleback {Gasterosteus trachurus, No. 86, 

 Brit. Mus. Nat. Hist.) represents a primitive stage in this group before the elongation of 

 the preorbital part of the face. The mouth is fairly small, the quadrate-articular joint 



P°P top 



Gasterosteus aculeatus 



Fig. 102. Gasterosteus aculeatus. 



lying well in front of the preorbital border. The ascending processes of the premaxillas 

 are of moderate length, the small maxillae are closely fastened at the distal end to the pre- 

 maxillae and by ligament to the side of the small mandible, so that the mouth must be pro- 

 trusile, at least to some extent. The lacrymal is fairly large and the third suborbital 

 over-arches the cheek and extends downward and backward to make contact with the pre- 

 opercular. This analogy with the Scorpaenoids is hardly borne out by other skull charac- 

 ters. The opercular lacks a spine and the skull is devoid of crests and is not percoid in 



