STRDCTUKE OF CERTAIN PALAEOZOIC DIPNOI. 
169' 
wliole hinder border of the skull. It springs from the three hinderinost 
roof-bones, not at their actual edge, but from a little way forward on the 
under surface, and along this line the extreme hind border of the bones is 
frequently cracked and bent down in fossilization. It is possible that in life 
this thin flange stood vertically, more or less at right angles to the skull-roof, 
and overlapped the hinder surface of the neural cranium. 
From the outer hind corner of the tabular forward to the outermost point 
of the squamosal the edge of the skull-roof is rounded off and usually lobed 
and corrugated, except for a smooth and very definite notch in the border of 
the squamosal. It is along this part of the roof that the operculum is hinged 
Fig. 5. 
Sagenodns. Civcumorbitals, X about f. 
Outer faces on tlie left, inner on the right. 
(see fig. 18, p. 183). The notch in the squamosal lodges the anterior knob of 
the operculum, and the two bones forming the hinge-line behind the notch 
bear on their under surface a groove, bounded internally by a ridge, which 
engages a flange borne by the operculum along its upper border. The 
posterior knob of the operculum, more or less strongly developed, fits against 
the hinder end of the smaller corrugated bone close to the corner of the 
tabular. 
From the outer point of the squamosal forwards the edge of the skull-roof 
everywhere shows articular faces, with articular laminae where the bones 
are well preserved. It is here that the orbitals and other marginal bones 
were attached. As a rule, however, the marginal bones, and the nasals as 
well, have been lost, even where the main part of the skull-roof has held 
together. Probably this main part (represented in the outlines in fig. 2, 
p. 165), was flat or nearly so, whereas the bones bordering it would be 
