GREGORY: FISH SKULLS 325 



plates is a concomitant and perhaps a cause of spinescence in this family as well as in some 

 others. 



The second most conspicuous character of typical scorpsenids is the presence of a 

 "suborbital stay," consisting of three flat suborbital bones (Fig. 201) conjoined with each 

 other and extending from the preorbital projection of the lacrymal backward to the anterior 

 border of the preopercular, to which the posterior edge of the elongate third suborbital is 

 more or less securely fastened. How did this contact between the third suborbital and the 

 preopercular arise.'' Is it a souvenir of the remote epoch when the ancestral percomorph 

 was short-bodied somewhat like the Cretaceous berycoids, when also the posterior end of 

 the third suborbital was in contact with the preopercular.^ Did the subsequent antero- 

 posterior lengthening of the body as a whole affect the suborbital bar without breaking 

 its contact with the preopercular.'' The latter hypothesis now appears to me likely, in spite 

 of the fact that in many respects even the most primitive known scorpaenoid skull seems 

 to have been derived from a form not earlier than the primitive bass type, by which stage 

 the body presumably had already passed from the very short to the normal bass type in 

 which the third postorbital was already moved away from the preopercular. The adhesion 

 between the suborbital bar and the preopercular may easily have arisen at the time of the 

 crowding of the suborbitals against the preopercular in embryonic and larval stages and 

 through the subsequent retention of the adhesion in the adult. This early crowding, as 

 well as the subsequent lengthening of the space between the orbit and the preopercular, is 

 well shown in the larvae and fry of the rosefish (Sebastes marinus) as figured by Bigelow and 

 Welsh (see below, 326). The backward growth of the third suborbital bone also may have 

 been favored by a tendency of the suborbital canal (Fig. 203) to send a horizontal branch 

 backward toward the preopercular branch of the canal system, especially as the lateral 

 line canals are so vigorously developed in the skull of scorpaenids, where they form prominent 

 channels and tubes of bone. In the percomorph Cirrhitus (Fig. 135) the posterior tip of the 

 third suborbital is fastened to the upper bar of the pr^eopercular, which is near to the orbit. 

 Such a connection might later have moved downward to near the middle of the rim of the 

 preopercular. 



However it may have arisen, there seems little doubt that the stiff "suborbital stay" 

 as a whole serves to protect the jaw muscles, the eyes, and to some extent the delicate 

 palatopterygoid tract, while also bracing the spike-bearing preopercular and opercular 

 arches. 



As to the posterior and upper elements of the circumorbital plates, I could not find 

 them in the fragments of a dried Sebastes marinus skull, but Allis (1909, p. 97) states that in 

 Sebastes dactylopterus there are two postorbitals, which are delicate semi-cylindrical bones 

 bounding the hind edge of the orbit and transmitting the main infraorbital canal from the 

 second suborbital to the postfrontal (sphenotic). These are also seen in his figure of 

 Scotpeena scrofa (cf. Fig. 203). 



The premaxillae and maxillae of Sebastes and Scorpcena (Fig. 203) are completely percoid 

 in fundamental plan. Between the ascending and articular processes of the premaxillae 

 is a large median cartilage, the rostral cartilage of Allis, which slides on the dorsally-keeled 

 mesethmoid and is received between the diverging anterior horns of the mesethmoid. In 

 Scorpana scrofa Allis also figures the elaborate system of check-ligaments (Fig. 203) by 

 which the premaxillae are prevented from dorsal and lateral dislocation and which also ties 



