268 



TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY 



the Percesoces, affords a graphic demonstration of unity of subordinal plan, with divergences 

 in proportionate development of parts, in the several families. While Sphyrana is extremely 

 long-headed, in Atherinopsis the skull is of normal percoid proportions, and in Mugil the 

 skull is short and relatively very wide in front, the last characteristic being plainly corre- 

 lated with the thickening of the snout and with the peculiar characters of the premaxillae 



5 phot 



Polydactylus sp. 



P. octonemus 



Fig. 144. Polydactylus sp. 



already noted. The peculiar bristle-like or branching prolongations of the epiotics and 

 the failure of the supraoccipital crests to project above the level of the skull roof are con- 

 spicuous evidences of unity of origin of all three families, when taken in connection with 

 the pervasive resemblances in basic features. 



Polynemids. — Most authorities either refer the Polynemidae to the Percesoces or place 

 them next to that group. Tate Regan, for instance (1929) gives them a rank coordinate 

 with the "Mugiloidea" as a division of the percoid suborder of the Percomorphi. 



The skull of Polynemus (Fig. 144) does not closely resemble those of either of the three 

 main types of Percesoces. The side view gives the impression of a forward and upward 

 displacement of the large eyes, involving the forward shifting of the sphenotic and of the 

 attached anterior head of the hyomandibular, combined with a marked lengthening of the 

 pterygoid and a downward and backward displacement of the quadrate-articular joint to 

 a point almost directly beneath the mid-point of the hyomandibular. The posterior end 

 of the upper jaw followed the quadrate backward and dragged, as it were, the front end of 

 the premaxilla backward beneath the growing snout, which was covered by the enlarged 

 down-turned nasals. Thus was produced a convergent resemblance to the head and jaws 

 of primitive palseoniscoids. Meanwhile the opercular region remained fairly normal, the 

 serrate preopercular being suggestive of percoid relationship. 



