GREGORY: FISH SKULLS 371 



condition it is shriveled through the loss of the cartilage. Starks (1926a, p. 298) notes that 

 the bones of the ethmoid region are very thin surface ossifications, filled with cartilage and 

 overlying masses of cartilage. "There appears to be no endosteal bone in the ethmoid 

 region whatever. ... A dried cranium becomes very much distorted." The flattish skull 

 roof is crowded with irregular thin-walled cells, and is without crests. The fairly large 

 orbits are directed laterally. The skull top is blunt anteriorly and widens posteriorly 

 toward the attached posttemporal horns. The ethmovomer block forms a small promi- 

 nence in the top view, but in the side view is very shallow, due to the reduction of the 

 mesethmoid which, according to Starks (1926a) is a thin simple plate lying below the 

 frontals between the upper end of the projecting prefrontals. This reduction may well 

 be correlated with the marked reduction of the ascending processes of the premaxillae. 

 The mandibular teeth are small. The lower anterior border of the lacrymal bears a small 

 downwardly directed spike followed after an interval by a much larger one. The opercular 

 is small and without spikes. There is no " suborbital stay " connected with the preopercular. 

 The broadly crescentic, large preopercular bears six radially directed points on its raised 

 outer surface, which are faintly suggestive of the eight radially diverging processes on the 

 preopercular of Uranoscopus, but the opercular region differs widely from that of Trachinus. 



Tate Regan (1913a, p. 136) puts the Trichodontidae as "Division 7, Trichodontiformes" 

 of the suborder Percoidea, "differing from the Perciformes in the pectoral fin skeleton." 

 Jordan (1923, p. 203) holds that the trachinoid fishes follow the Trichodontidae in natural 

 sequence, "a fact not to be shown in a linear series." Starks (1923, p. 265), in considering 

 the relationships of the uranoscopoid fishes, wrote that Trichodon (as compared with 

 Parapercis and Bathymaster) "with its much reduced mesethmoid, its sphenotic extending 

 inwards to the parietal and separating the frontal from the pterotic, its widely separated 

 ectethmoids and its large opisthotic, shows a somewhat closer relationship with the Urano- 

 scopoids, but not nearly close enough to be admitted into the group." But in his recent 

 work on "The Primary Shoulder Girdle of the Bony Fishes" (1930, p. 75) he notes that 

 "the shoulder girdle [of Trichodon] is strikingly like that of some cottoid fishes," which, 

 as he shows, are widely different from those of the uranoscopoid and notothenioid groups. 

 Consequently it is at present doubtful whether the skull characters which Trichodon shares 

 with the uranoscopids are enough to indicate a real relationship. 



The otolith of Trichodon trichodon is described by Frost (1928a, p. 455) as resembling 

 in shape that of Iniistius, an aberrant member of the division Labriformes of Tate Regan's 

 classification of the Percoidea. It differs widely from those of the Trachiniformes. 



Xenopterygii (Cling-fishes) 



Some very unusual specializations from the percomorph ground-plan are embodied in 

 these little fishes, which have the pelvic fins modified into a sucking-disc but are certainly 

 not related to the gobies and other forms with similar pelvic sucking-discs. 



The skull (Figs. 249, 250) is extraordinarily specialized in many directions. The short 

 stoutly-built jaws bear forwardly inclined incisors resembling somewhat those of man and 

 evidently adapted for .nipping. Replacing teeth lie in the alveoli. The short dentary 

 forks over the massive articular, which bears a transverse hinge-like joint with the quadrate. 

 The ascending process of the articular is vertical and very stout, indicating powerful ad- 

 ductor muscles. These were braced by a strong backwardly directed process of the 

 20 



