GREGORY: FISH SKULLS 131 



teus vertebrae seem quite easily derivable from the generalized condition of the column in 

 the Semionotidse, especially in the light of the comparison of skull patterns. 



MacTOsemiids. — The incompletely known skull pattern (Fig. 26) reveals a few impor- 

 tant morphological details. In this Triassic ganoid the eyes had already assumed the 

 dominance characteristic of teleosts in comparison with sharks. The large backwardly- 

 positioned eyes contrast widely with small forwardly-placed eyes of the primitive chon- 

 drosteans. The base of the skull was stiffened by a strong parasphenoid. The suspen- 

 sorium was inclined forward, the preopercular being sharply bent and ridged in contrast 

 with the gently curved cheek-plate preopercular of palaeoniscids. The jaws are nearly as 

 short as in the Semionotidae and there is a sharp ascending ramus of the mandible. Both 

 the small premaxillae and short maxillae bear pointed teeth. The opercular is short and 

 deep. The skull gives the impression of having been derived from one that was shorter 

 and deeper. The elongate dorsal fin of the Macrosemiidae suggests a secondary elongation 

 of the body from a more deep-bodied ancestral stock, perhaps not far from the Permian 

 Acentrophorus. 



Amioidei 



In the Triassic of Europe appears the first known of the great group of amioids, which 

 as a whole stand between the Palaeozoic ganoids and the late Mesozoic teleosts. These 

 primitive amioids have inherited from some older ganoid stock the complete ground-plan 

 of their skeletons, the microscopic structure of the scales, the abbreviate heterocercal tail, 

 the general pattern of the skull bones. But they are far more progressive than any known 

 earlier ganoid group in the swift-swimming type of body with its large forked hemi- 

 heterocercal tail, as well as in numerous details of skull structure. During the Jurassic 

 and Cretaceous periods they exhibit a wide adaptive radiation into many different body- 

 forms. Most of these end in specialized side lines; one gives rise to the existing Amia, 

 while the leptolepids finally attain the teleost grade and appear to be directly ancestral 

 to the order Isospondyli at least. 



As to the exact point of origin of this entire series, even the prolonged researches of 

 Smith Woodward, Watson and others have failed to reveal it. On the whole, evidence 

 seems to indicate that in spite of the marked difference in habits and habitus the primitive 

 pike-like amioids may have been derived from a relatively short-bodied, small-mouthed 

 form like the Permian Acentrophorous rather than directly from the Devonian pike-like 

 palaeoniscids. This view has indeed been more or less independently suggested by Smith 

 Woodward (1895, p. ix), Tate Regan (1923a, p. 456; 1929, p. 312) and the present writer 

 (1923, p. 239). 



Eugnathids. — Whatever its precise derivation may nave been, the skull (Fig. 27) of 

 Eugnathus (A. S. Woodward, 1895, Pis. IV, V) gives the impression that the eye has 

 migrated backward, thus giving rise, through some non-Lamarckian principle, to confusion 

 and irregularity in the circumorbital plates, while the snout has grown forward, the two 

 movements conditioning the marked elongation and downward pitch of the lacrymal and 

 antorbital elements, a feature unknown in earlier ganoids, but highly characteristic in the 

 amioids and teleosts. The long preorbital plate is subdivided longitudinally into two 

 parallel plates, of which the dorsal corresponds to the antorbital, the lower to the lacrymal 

 of Amia. A close scrutiny of the original specimen shows that at least parts of the suture 

 5 



