7 8 
ERIK A : SON STENSIO 
may sometimes be reduced to a certain extent or be quite absent as is the case in 
W. sinuosa, so that there is a large interspace between the postorbital and the opercular. 
To judge from what has been mentioned, this whole state of affairs seems to support 
the idea that the dorsal portion of the squamoso-preopercular, which in primitive 
Coelacanthids projects upwards dorsally behind the postorbitale, has in Mesozoic ones 
become detached from its connection with the other parts of the bone. It may then 
either have persisted as an independent element (Sassenia) or have coalesced with the 
postorbital (Diplurus, Undina, Macropoma, etc.) or else have been reduced (Wimania). 
Besides the jugal and the postorbital the cheek in the Rhipidistids in general is 
covered with a fairly varying number of bones. Thus Holoptychius is said to have five, 
Glyptopomus, Tristichopterus, Rhizodopsis and Megalichthys to have three, and Osteolepis to 
have two, while in Diplopterus, Gyropthychius and Eusthenopteron only one large plate seems 
to occur (Traquair, 1875a, PL XXXII, figs, x, ii; 1881, p. 172, fig. 2, pp. 176—177; 1896; 
Watson and Day, 1916, pp. 1—27; Goodrich, 1919; Woodward, 1891b, pp. 3 ig— 32 o; 1898a, 
pp. 70, 72, 75—77; Wellburn, 1900, pp. 62— 63 ; Pis. XIII, XV; Birks, 1916, pp. 307—327). 
It is pretty clear, as Watson and Day (1916) have shown, that we are here concerned 
with various more or less advanced stages of fusion in the different forms and that 
forms with a larger number of these bones ought in this respect to be taken as more 
primitive than those with a smaller number. In Holoptychius, Glyptopomus, Tristichopterus 
and Rhizodopsis one can regularly identify the preoperculum, quadrato-jugal and squamosal; 
in Holoptychius, however, if Watson and Day’s observations are correct, the last-mentioned 
bone consists of three independent plates; in Osteolepis, according to Goodrich, a small 
independent preoperculum seems to be present besides the large postorbital plate, where¬ 
as this small plate was not observed by Watson and Day (1916). 
If now we examine the Rhipidistids that have a preoperculum, quadrato-jugal and 
.a squamosal, we may state the following facts about these bones: the preoperculum is 
situated close to the anterior margin of the operculum, along which it projects more or less 
far upwards in different forms; usually it is high and narrow. In front of it follows the 
squamosal, which is attached anteriorly to the jugal and the postorbital, ventrally to the 
maxillary and the quadrato-jugal, posteriorly to the preoperculum and to a little plate 
following in front of the anterior upper end of the operculum, or, if this is united with 
the operculum, to the antero-dorsal part of the latter. It reaches upwards right to the 
cranial roof, and seems to be pierced by the jugal canal, which enters it from the jugal 
and subsequently passes backwards and downwards towards the postero-ventral corner. 
The quadrato-jugal forms the postero-ventral part of the cheek and is attached dorsally 
to the squamosal and the preoperculum, anteriorly to the maxillary, and posteriorly with 
its upper part to the preoperculum. The remaining part of its posterior margin, like its 
whole ventral one, is free. In Osteolepis the postero-ventral part of the jugal canal 
pierces the preoperculum (Goodrich, 1919); in other forms we do not know for certain 
the relation of this canal either to the preoperculum or to the quadrato-jugal, but it seems 
most probable that the last-mentioned bone has been situated in front of this canal. 
If now we try to apply the experience gained by our examination of the region 
of the cheek in the Rhipidistids to the conditions found in the Coelacanthids it is evident 
that both the squamoso-preoperculum and the preoperculo-quadratojugal would comprise 
