TRIASSIC FISHES FROM SPITZBERGEN 
i3 9 
As we know, there is no cheek-plate"’ in the Actinopterygians that can be directly 
compared with the squamosal of the Rhipidistids. Watson and Day (1916) have shown, 
as has been pointed out above, that in primitive forms among the Rhipidistids the 
squamosal consists of several independent elements, and I now think that this fact can 
guide us in judging of the conditions in the Actinopterygians. Thus one may venture 
the supposition that all the homologues of the elements forming the squamosal in the 
Rhipidistids have never been fused with one another into one bone-plate in the 
Actinopterygians, but that they were either distributed among the surrounding bones 
or with partial transformation had been fused into certain larger units which afterwards 
preserved their mutual independence or else were finally more and more reduced. In the 
Palaeoniscids, Platysomids and Catopterids it is conceivable that the original squamosal 
elements were divided up between the maxillary and the preopercular and that they 
have even possibly provided material for the so-called postorbitals; in certain Proto- 
spondyli, such as the Semionitids and Eugnathids, they all probably provided material 
for the majority of the so-called postorbitals (suborbitals); — the ventral one, or 
some of them, can probably, as we have seen, be considered as corresponding to the 
quadratojugal; in the Teleosts the homologues of the squamosal elements usually 
seem to be reduced. Sometimes, however, it is possible that parts of them may persist 
in the plates of the infraorbital chain, in the cases when these plates are much developed 
and cover the cheek between the orbit and the preopercular to any great extent. 
The large pterygoid of the Coelacanthids together with the quadrate and the 
metapterygoid usually resemble the suspensor)r apparatus in the Teleosts. Owing to 
this fact it has hitherto been thought that the hyomandibular was intimately fused with 
the palatoquadrate, a view that I have been able entirely to refuse in this work. 
In the presence of a strong urohyal the Coelacanthids resemble the teleosts 
(Goodrich, 1909, p. 350) and, as I shall show below, also certain Palaeoniscids (Acro- 
rhabdus). 
As far as can be judged at present from the skeleton of the head, the fossil 
Crossopterygians have undoubtedly a more primitive appearance than the Actinopterygii 
(Abel, 1919, p. 155). *) These latter appeared as early as in the middle Devonian well 
differentiated from the Crossopterygians and so far no Crossopterygian form is known 
from which the Actinopterygians could have arisen. Since, however, as we have seen, 
it is evident that the more primitive forms among the Actinopterygians resemble the 
Crossopterygians in certain r.espects with regard to the skeleton of the head, and in a 
number of other respects can from a theoretical point of view easily be derived from a 
Crossopterygian-like type, it seems not improbable that they originate from some early 
Devonian or Upper Silurian form, from which the Crossopterygians have also been 
developed (cf. Abel, 1919, p. 184; Stromer, 1920). 
It seems fairly certain that this common ancestral form resembled the Crosso¬ 
pterygian type with regard to the cranial skeleton, and several authors (Abel, 1919, 
pp. 185 —186; Braus, 1901, p. 287; Gegenbaur, 1873, 1876, 1894, 1898; Goodrich, 1909, 
pp. 106—109; Stromer, 1920, p. 17; Jaekel, 1909a, b; Simroth, 1891, pp. 842—351) have 
3 ) An opposite view is maintained by Regan (1904, p. 347) and several other authors. 
18* 
