GREGORY: FISH SKULLS 223 



influenced by the very high position and large size of the pectoral fins. As a result the 

 supracleithrum has become reduced, while the enlarged posttemporal collaborates with the 

 backwardly-produced pterotic in forming a firm strut for the pectoral girdle. Meanwhile 

 the epiotic has been reduced, especially in Tylosurus, and crowded in against the small but 

 well braced supraoccipital. The latter is firmly embraced by the ends of the broad frontals, 

 which are the dominant bones of the skull-roof; the parietals, at least in this form, having 

 been eliminated entirely. 



The Synentognathi are unquestionably a natural group. Their characteristic fused 

 lower pharyngeals, which form a toothed plate opposing the more or less enlarged upper 

 pharyngeals, are merely analogous with those of the wrasses and many other groups of the 

 spiny-finned series, but may possibly be related by community of origin with the suturally- 

 united dentigerous lower pharyngeals of some of the cyprinodonts. Boulenger (1910, p. 

 636) indeed regards the Scombresocidse (gars, etc.) as being "somewhat related to the 

 Cyprinodonts." A review of the family characters of the two groups as listed by several 

 authors leads at first to the impression that the synentognaths and cyprinodonts are the 

 divergent offshoots of a broad common stock, the cyprinodonts becoming specialized in 

 many features. The practice of internal fertilization, so highly developed in the cyprino- 

 donts, is also inherited by Hemirhamphus of the synentognaths (Boulenger, op. cit., p. 638). 



A comparison of the surface characters of the skulls of Fundulus, representing the 

 cyprinodonts, and of Tylosurus, Hemirhamphus and Halocypselw;, representing the synen- 

 tognaths, reveals a sufficient degree of similarity to suggest at least superordinal relation- 

 ships. The operculars are large, convex and smooth, without spikes. The not large 

 supraoccipital is in contact with the frontals; the skull-top more or less flat without sharp 

 crests; opposite "prefrontals" nearly in contact medially; mesethmoid more or less disc-like 

 in top view (Starks, 1926a, pp. 205-211); orbits fairly large; lacrymal prominent, over- 

 lapping maxilla, other suborbitals absent or represented by narrow rim; suspensorium 

 produced forward, mouth pointing more or less upward, rather broad transversely; pre- 

 maxillae more or less beak-like, excluding maxillae from gape; presence of proethmoids. 



The gars, needle-fishes and flying fishes were referred by Boulenger to the "suborder 

 Percesoces," which comprised some twelve families of very diverse-looking fishes. These 

 were grouped together because collectively they showed transitional characters between the 

 soft-rayed fishes with open duct of the swim-bladder, and the spiny-finned groups with 

 closed swim-bladder. But the five characters cited by Boulenger in the definition of the 

 order are each in itself subject to wide variation among teleosts as a whole. Realizing 

 this, Boulenger (1910, p. 636) remarks: "Although this suborder is perhaps only an artificial 

 association, it must be borne in mind that notwithstanding the very wide divergence which 

 exists between the first and last families, and however dissimilar their members may appear 

 to be at first sight, a gradual passage may be traced connecting the most aberrant types." 



Tate Regan (1929), however, does not accept this view of the case and distributes 

 the contents of Boulenger's "Percesoces" under a number of different orders. He follows 

 earlier authors in treating the skippers and flying-fishes as a distinct order, Synentognathi, 

 while he refers the typical Percesoces (Sphyraenidae, Atherinidae, Mugilidae, etc.) to the 

 Percomorphi. 



In this connection the surface views of the skull (Figs. 100, 101) reveal a close resem- 

 blance in family heritage between the needle-fish Tylosurus and the flying-fish Halocypselus, 



