GREGORY: FISH SKULLS 



339 



one between the supraoccipitals and the other two between the supraoccipltals and the 

 epiotics. The scorpsenoid heritage of this skull is seen in many characters, as of the lacry- 

 mal and suborbital stay, small preopercular spikes, etc. 



In a very wide-skulled species of Myoxocephalus (Fig. 214) from Alaska (presumably 

 M. jaok) the strongly-raised, fine trabeculse of the skull roof show a tendency to give off 

 very numerous small points on ridges radiating from the center of the bone. A continuation 

 of this tendency might give rise to the finely-papillate or villous surface of the cranial armor 

 of the gurnards (Triglidae). In another Myoxocephalus skull the radiating ridges are 

 equally evident but are not subdivided into papillae; in a third species the trabeculse do not 

 project above the surface and there are no villi. Here then we have in allied species 

 interesting differences in bone texture which imply corresponding differences in the under 

 side of the skin. 



A strange modification of the cottid type is seen in the skull of Enophrys {Aspicottus) 

 bison (Fig. 216). Here the ectosteal bones of the face and head are thickened and their 



Penstedion cataphractum 



Fig. 220. Peristedion. After Allis. 



surface is studded with a carpet of minute horny pustules or denticles arranged on radiating 

 ridges from the growth-centers and covering even the two large spikes on the preopercular 

 and opercular. These pustules or villi are much more numerous and uniform in size than 

 those described above as occurring in Myoxocephalus jaok. They are usually at right 

 angles to the direction of the ridges or trabeculse, like beads of water on a thin wire. 



The third suborbital is pointed posteriorly and in the dried skull diverges widely from 

 the preopercular. The endochondral quadrate and metapt^rygoid bones show the usual 

 zonal structure but are very thin and translucent with fine, delicate, radiating fibres. 



Cyclopterids. — Nothing could well be more unlike the bristling, aggressive scorpaenoids 

 than the obese lump-fishes, Cyclopteridae, yet the possession of a well developed suborbital 

 stay (Fig. 217) seems to offer a reliable indication of their taxonomic position. Indeed 

 Boulenger (1910, p. 698) states that the Cyclopteridae are very closely related to the Cot- 

 tidse, with which they are connected through Psychrolutes, and that "it is even doubtful 

 whether they deserve to be separated from them." 



The peculiar genus Caracanthus (Micropus), which is widely distributed among the 

 18 



