GREGORY: FISH SKULLS 343 



Myoxocephalus jaok and other cottoids suggest that in more primitive conditions the bone 

 was porous with minute tubes opening on the exterior. Then the radiating trabeculae 

 began to assert themselves, the pores became open channels between the trabeculae, and 

 with further increase in the growth pressure the trabeculae began to sprout into irregularly 

 cuspidate edges as in Myoxocephalus jaok. Finally the almost uniform distribution of the 

 villi or denticles was attained, as in the triglids. In Peristedion (Fig. 220) the growth 

 momentum of the lacrymal bones has resulted in a pair of long processes which project far 

 in front of the nasals. This, together with the reduction of the jaws, produces a somewhat 

 sturgeon-like appearance of the head and is doubtless associated with the habit of resting 

 on the bottom. 



In spite of these and other specializations, the distribution of the lateral line canals on 

 the cheeks and skull roof assists us in homologizing the several elements with those of 

 normal scorpaenolds {cf. AUis, 1909, Taf. VI). 



On the whole, the skull of Prionotus, as well as those of other triglids, is remarkably 

 heavy, especially in its outer parts, while it is equally delicate and thin in the well-covered 

 parts, including the jaws. In many, but not in all Prionotus skulls, there is a pair of peculiar 

 dense swellings on the lower part of the prefrontals. The base of the cranium is smooth 

 and compactly built, with three wholly distinct and well-rounded occipital condyles, 

 essentially as in Cottidae and Scorpaenidae. 



The flying-gurnard Dactylopterus volitans is the type of a very peculiar family whose 

 connection with the scorpaenoid group is indicated by the presence of a suborbital stay, 

 by the forward extension of the lacrymal, and perhaps other characters. But. the skull 

 (Figs. 222, 223) has become so highly specialized that the more precise relationships of this 

 family to the other scorpaenoids is uncertain. Comparison of the skull-top of Dactylopterus 

 volitans and of Trigla hirundo as figured by Allis (1909, Taf. VIII) reveals immense dif- 

 ferences in the pattern and in the arrangement of the individual elements. In Dactylopterus 

 the plates above the shoulder-girdle have been expanded into a huge neck shield, which is 

 conjoined by immovable sutures with the expanded cranial roof. The differences in rela- 

 tions of all the bony elements and the course of the lateral line canals between Dactylopterus 

 and Trigla seem in fact to be irreconcilable with the idea of a near relationship between 

 the two families. This negative conclusion is strengthened by the recent paper of Starks 

 on the shoulder-girdle of the teleost fishes (1930) in which it is shown (p. 71) that the pectoral 

 girdle of Dactyloptena orientalis is extremely different from those of the cottoid and triglid 

 fishes, so that Starks notes its typical percoid characters and says that it is "difficult to 

 understand why this evidently aberrant trigloid form should possess the typical percoid 

 shoulder-girdle of the main line of descent, rather than that of its immediate relatives of 

 the family Triglidae, which have the cottoid shoulder-girdle." 



The otoliths of the scorpaenoids, as studied by G. Allan Frost (1929^, pp. 257-263), 

 "show a strong, affinity with those of the suborder Percoidea, and, although in some cases 

 considerably modified, resemble either the Percid or the Labrid type. ... In the family 

 Triglidae, the otoliths are high in shape and present certain Labrid features; they may 

 usually be distinguished by the contorted appearance of the sulcus, due to the uneven 

 position of the upper and lower angles, and to the elevation of the cauda. In the families 

 Scorpaenidae, Cottidae, and Dactylopteridae, the otoliths resemble the percid type." With 

 some exceptions the otoliths vary from the primitive, slightly elongate percid type of 



