108 TRANSACTIONS OF THE AMERICAN PHILOSOPHICAL SOCIETY 



The palate (Fig. 9) is characterized by the presence of two chief rows of teeth respec- 

 tively on the outer or secondary and on the inner or primary upper jaw, and especially by 

 the very short parasphenoid, which, unlike that of later fish, did not extend back under the 

 occiput. According to Watson (1926, pp. 245-249), the massive palatoquadrate arch is 

 ossified from about seven centers, the various regions giving rise to the palatal plate, true 

 pterygoid (ectopterygoid), epipterygoid, three suprapterygoids (including the metaptery- 

 goid) and the quadrate. The arrangement of the numerous elements of the outer and inner 

 upper jaws is strikingly similar in fundamental plan with that of the primitive Amphibia 

 on the one hand and of primitive actinopterygian ganoids on the other (Watson, 1926, p. 

 198; 1925, p. 854; 1928, p. 62). There can therefore be little doubt that the premaxilla, 

 maxilla, palatine, ectopterygoids, etc. of the crossopterygians are correctly equated with 

 those of the earliest amphibians and earliest actinopterygian ganoids. 



The hyomandibular in the rhipidistians Eusthenopteron (Bryant), Rhizodopsis (Watson 

 and Day, 1916) and Megalichthys (Watson, 1926, p. 250) articulates above with the otic 

 region of the neurocranium and at least in Eusthenopteron bears a process for the opercular. 

 As noted by Bryant (p. 12) from excellent material, "it resembles very much its homolog 

 in the Palseoniscidae." In the Triassic coelacanth Wimania, as described by Stensio 

 (1921, p. 74), the main suspensor of the quadrate was the metapterygoid, which articulated 

 with the side of the roof of the cranial vault. The hyomandibular (pharyngohyal) seems 

 to have been cartilaginous, but the epihyal and ceratohyal were well ossified. Allis (1928, 

 p. 207) infers that " fVimania is more primitive in several respects than the much earlier 

 Rhipidistia," but Stensio (1921, p. 135) concludes that "... as far as we can decide from 

 the facts now known, the Coelacanthids ought to be taken as a highly specialized group, 

 the ancestors of which are to be sought for among the primitive Rhipidistids." 



In the coelacanth Macropoma, according to Watson (1921, p. 336), "there was a com- 

 plete loss of the hyomandibular as a supporting element of the jaw. This loss is an exact 

 parallel to that which has occurred in Tetrapods and Dipnoi." At the same time Watson 

 adduces strong evidence that the coelacan-ths were derived from the osteolepids. Thus it 

 seems that there is evidence that the coelacanths are much further from the actual ancestors 

 of the teleosts than were the rhipidistids. Hence I doubt the propriety of assuming the 

 conditions of the "pharyngohyal" of the coelacanth Wimania as structurally ancestral to 

 different types of hyomandibular in Dipnoi, recent Chondrostei, certain palaeoniscids and 

 certain teleosts (Allis, 1928, p. 218). Edgeworth (1926) also adduced much embryological 

 evidence for the old view that the hyomandibular of shark, ganoid and teleost is the same 

 element throughout, a view which may be provisionally adopted in the present work. 



POLYPTERINI 



Polypterus is a heavily armored ganoid with a depressed reptilian-looking head not un- 

 like that of Ophiocephalus. It is plainly predaceous in habit and its skull on the whole 

 recalls that of the Upper Devonian Crossopterygian Eusthenopteron. However, according 

 to Goodrich (1907, 1909, 1928), Polypterus does not belong with the Palaeozoic crossoptery- 

 gians at all, but represents a wholly different branch, allied rather with the primitive 

 actinopterygian ganoids. Allis (1922) in a beautiful monograph on the cranial anatomy of 

 Polypterus has compared the morphology of the skull with that of Amia, and In spite of the 

 great gap between them has been able to establish the homologies of most of the elements. 



