TRIASSIC FISHES FROM SPITZBERGEN 
95 
The parasphenoid in its anterior halves is flat and lacks the upward-curving 
lateral parts, which otherwise are so usual in the Coelacanthids. Such upward-curving 
lateral parts are, on the contrary, to be found in its posterior half (text figs. 3 g, 41; 
PL 1 3 ; PI. 14, fig. 5), where most posteriorly they are considerably high. A transverse 
section through the parasphenoid posteriorly (text 
fig. 41; PI. 14, fig. 5) shows plainly that these up¬ 
ward-curving parts have likewise been thick and 
strong and were formed of typical radiate bone 
substance of the same appearance as in the ventral 
processes ( y) of the basisphenoid. The horizontal part 
lying between them also has a similar structure in 
its dorsal half. The upward-curving parts, together 
with the ventral processes of the basisphenoid which 
rest against their dorsal surface, form lateral boundaries 
of the anteriorly open canal (in) which, as I have 
tried to show before, should correspond to a myo- 
dome. Dorsally this canal is bounded anteriorly by 
the ventro-lateral parts of the basisphenoid lamellae 
and behind them by the basisphenoid corpus; ven- 
trally by the median horizontal part of the parasphe¬ 
noid. Anteriorly the canal in question is broad and 
rather high and has a typical six-sided transverse 
section, posteriorly it diminishes rapidly in height 
so that at last only a low cavity remains. That the 
canal has not had its present circumference in 
reality, but has been partly filled by cartilage, is 
seen from the appearance of the surrounding bones. 
While the parasphenoid in W. sinuosa was a 
dermal bone in the whole of its extension, there is 
much that suggests that in A. robusta it was partly 5stQ p__^j ; 
developed as a primary bone in the posterior part. 
This view is favoured in the first place by the 
characteristic radiate structure of the bone in the 
upward-curving parts and partly also in the hori¬ 
zontal part lying between them, further by the 
way in which these upward-curving parts unite 
with the basisphenoid, and lastly by the relatively 
important extension in the way of height in the 
upward-curving parts at the expense of the ventral processes of the basisphenoid. 
It is uncertain whether vomeres were to be found. If such was the case they ought, 
however, to have partly covered the anterior end of the parasphenoid bone with their posterior 
parts, and for the rest have borne the same relation to the preethmoidea as in Amia (Allis). 
If we now turn to the cranial roof we naturally find in A. robusta, in respect to 
its membrane bones, that taken on the whole it is in agreement with that in W. sinuosa. 
'-Stemp.exr 
Text fig. 43. Axelia robusta n. sp. 
Dermal cranial roof from the dorsal side. The 
sculpture is not drawn. Sensory canals with dotted 
lines and shading. From P. ig 5 . V,. 
Ext, Extrascapulars, Fr.dsph, Fronto-dermospheno- 
tic; Ir, interrostral; Na. ant, nasalo-antorbital; 
Pa. it, parieto-intertemporal,' Plr, postrostral ; 
R, rostrals; So, supraorbitals; Sscap, suprascapular; 
Stemp.ext, supratemporo-extrascapular; na x , anterior 
nasal aperture; na„ posterior nasal aperture; p, pores 
of the sensory canal (belonging to the supraorbital 
