ESPECIALLY MYCTOPHOIDS 237 
because of the enormous coronoid process of the mandible of Leptolepis, and the 
practically edentulous condition of the jaws asa whole. Other leptolepids however, 
e.g. Leptolepis coryphaenoides (Bronn) (Text-fig. 99D) and Leptolepis normandica 
Nybelin (Text-fig. g9C) have a lower coronoid and are provided with rather more 
teeth. The Pholidophoridae also exhibit a relatively large coronoid and small 
teeth (except Pholidophorus similis, Saint-Seine 1949, although Saint-Seine’s 
reconstruction of this fish is to be viewed with some reservation). This does not 
imply that the salmoniforms were not derived from some pholidophorid, as the jaws 
are a particularly labile entity capable of easy modification. Thus it is conceivable 
that a leptolepid such as Leptolepis normandica may have provided the ancestral 
salmoniform stock. 
Possibly a trivial character, but one worth mentioning, is the suprapreopercular 
bone bearing the upper part of the preopercular sensory canal (formed from the 
fragmentation of the upper part of the preoperculum). This bone is seen in the 
salmonid Salvelinus (Norden, 1961), certain Gonorhynchiformes and many Ostari- 
ophysi (Greenwood, ef al., 1966). Greenwood, ef al. (1966) have indicated the 
extreme primitiveness of these three groups and place them near to the base of the 
protacanthopterygian radiation. Greenwood, ef al. (1966 : 382) describe the 
ostariophysan suprapreoperculum as “‘ an ossicle above the uppermost part of the 
preoperculum and partially surrounding the sensory canal’’. It seems probable 
that since it is a canal bearing bone, the suprapreoperculum would have been present 
in the ancestral form. Fragmentation of the preoperculum occurs throughout the 
Chondrostei, including the Parasemionotidae (see Gardiner, 1967a : 200). In most 
cases it is the non-canal bearing anterior part of the preoperculum which fragments to 
produce the suborbital series. Less frequently the canal bearing portion of the 
preoperculum also fragments to give a smaller suprapreoperculum dorsally (Gardiner, 
1967a, fig. 2A). Such a condition is seen in Pavasemionotus laborde: Piveteau (Text- 
fig. g9A) and the totally unrelated chondrostean Boreosomus gilliott Priem. No so 
far described pholidophorid has a suprapreoperculum (see Nybelin, 1966). However, 
Nybelin (1962) has shown in both Leptolepis coryphaenoides and Leptolepis normandica 
a large suborbital, and has indicated that beneath the suborbital of the latter 
there is a separate suprapreoperculum. If this in fact is such a variable feature then 
no importance can be attached toit. But if both suborbital and suprapreoperculum 
are formed from the fragmentation of the preoperculum, as Gardiner (1967a : 199) 
has shown, it is hard to believe that one fragment could completely cover the other, 
especially as the covered fragment is canal bearing. Nevertheless it has already 
been seen that on scale structure the leptolepids are off the ancestry of the salmoni- 
forms (Gardiner, 1960). A suprapreoperculum is not known in any other hale- 
costome. It would seem unlikely that further fragmentation of the preoperculum 
within the halecostomes or early teleosts would have taken place since by the 
halecostome grade of organization the preoperculum had become more and more 
intimately associated with both the jaw musculature and the underlying palate 
(Gardiner, 1967a). 
Thus it may be that some as yet unknown pholidophorid which retained the 
parasemionotid suprapreoperculum gave rise to the salmoniforms, gonorhynchi- 
