i32 
ERIK A : SON STENSIO 
The chorda dorsalis also in both cases probably continued into the neurocranium and 
at any rate pierced the occipital region. Cope ( i 883, p. 628) described — also in 
Megalichthys — «ossified parachordals» occupying the basal parts of the neurocranium 
and, if I interpret his brief description correctly, these parachordal ossifications should 
correspond most closely to the paired prootico-opisthotic in the Coelacanthids. The 
autosphenotic is absent in Coelacanthids and probably in Rhipidistids as well. In 
Megalichthys Watson (1912 a, pp. 9 — 10) mentions a basisphenoid which seems to agree 
in all essentials with that of the Coelacanthids. As in these, it consequently corresponds 
to the postero-caudal part of the large sphenoid in Dictyonosteus. It does not seem 
improbable that in addition a large sphenoid similar to the one in this form was 
developed in a number of Devonian forms as well. The conditions in Dendrodus, if 
Rohon’s figures (PI. 1, figs. 1, 9) are correct, would, inter alia, support this view. It is also 
noteworthy that in Dictyonosteus the homologues of the alisphenoid of the Coelacanthids 
forms an integral part of the sphenoid without any signs of fusion with the membrane 
bones in the cranial roof. 
Both in Dictyonosteus and in Megalichthys the basisphenoid has distinct basipterygoid 
processes with which the palatoquadrate probably articulated (. Dictyonosteus ) or was 
fixed ( Megalichthys according to what is indicated by Watsons statement 1912 a, p. 9).^ 
The basipterigoid processes in Dictyonosteus are situated, as I have pointed out, not close 
to the basis cranii but somewhat higher up and thus show in this respect a certain 
resemblance to those in the Coelacanthids (specially those in Axelia). As far as one thus 
can see, the Coelacanthids and the Rhipidistids agree, inter alia, in having an articulation 
or a connection developed between the palatoquadrate and the posterior part of the 
orbito-temporal region, a fact that is of great importance for our judgement of the 
relations of these fishes. 
The ethmoidal region is certainly preserved in both Dendrodus and Cricodus, but, 
as Rohon’s (1889) interpretation of it is incorrect throughout (cf. Traquair 1889), no 
great reliance can be placed in his figures either. In Dictyonosteus we find this region 
practically entirely occupied by a large paired ossification, which to judge from its 
extent, ought to comprise the homologues of both the preethmoid and the exethmoid. 
The shape and extent of the cranial cavity and the position of the nerve foramina 
seem to show strikingly great agreements in Coelacanthids and Dictyonosteus, as I have 
had occasion to point out in detail above and as is shown in my proviously published 
work on Dictyonosteus (1918 c). 
The parasphenoid agrees in Dictyonosteus, Sauripterus (Eastman, 1917, PI. 7, fig. 8, 
and possibly Glyptopomus too with the same bone in the Coelacanthids. On the' other 
hand it seems, if we can believe the figures published by Rohon 1889, that it was 
strikingly narrow in Dendrodus. 
With regard to the membrane bones of the cranial roof the Coelacanthids can, as 
we have already seen above, be derived from a Rhipidistid-like type. Their supratemporo- 
extrascapular has arisen by fusion of the homologues of the supratemporal and a part 
of the lateral extrascapular plate of the Rhipidistids, and in a similar way their 
parieto-intertemporal is formed by the fusion of the homologues of the parietal and the 
intertemporal, and their fronto-dermosphenotic by the fusion of the homologues of the 
