| 
_ BARYONYX WALKERI 
“brae become progressively more elongate transversely (See Table 
) 1, p. 30.) This progressive diminution of the vertebrae in a forward 
| direction culminates in a remarkably small axis, its length being 
‘only 7.8% of the estimated length of the skull (comparable figures 
‘for Allosaurus, Ceratosaurus, Coelophysis, Deinonychus and 
Dilophosaurus are 13.6%, 9.8%, 15.7%, 10.8% and 10.6%). Each 
/centrum is constricted in its middle so that it has a waisted or “hour- 
'glass’ shape, but in fact there is no deepening of the floor of the 
‘neural canal within each vertebra like that found in some other 
/archosaurs, e.g. “Mandasuchus’ (Charig MS 1956). The centra are 
‘strongly opisthocoelous, the concavity of each posterior face ar- 
‘ticulating with a great ball-like convexity on the anterior face of the 
next centrum behind. There are no ventral keels on the centra. 
However, a faint ventral ridge is present on the posterior half of the 
/axial centrum and on Ce8; the latter occurrence suggests that the 
undersides of the unknown Ce7 and Ce9 may have been similarly 
ridged. 
Models were made of the missing cervical vertebrae (namely 
Ce4, Ce7 and Ce9), their dimensions and characters being estimated 
‘as intermediate between the actual vertebrae before and behind in 
the series as ordered by us. When all eight vertebrae (Ce2-Ce9), 
actual or modelled, were lined up in what we judged to be reasonable 
natural articulation between the opposing faces of the centra and 
‘between the relatively huge zygapophyses, then the neck was more 
or less straight; there was no sigmoidal curvature of the neck such as 
jis found in many other archosaurs. More precisely, the two faces of 
each centrum lie parallel to each other; they are not at an angle to 
each other as in Allosaurus, Ceratosaurus, Deinonychus and many 
other archosaurs, in which the necks must have curved strongly 
upwards. 
A large parapophysis projects laterally from an anteroventral 
position on the side of the centrum (except on the axis, see below) 
and retains that position throughout the series. The distance be- 
tween the paired parapophysial facets, seen from below, increases 
as we pass backwards down the neck. This widening is due to the 
‘increasing diameter of the centra and affords useful confirmation of 
the correctness of our ordering. Immediately posterodorsal to the 
/parapophysis lies a pleurocoel, the size of which increases back- 
‘wards down the series; the left and right pleurocoels of each 
‘centrum appear not to communicate with each other across the 
midline. In Ce3 each side has two adjacent pleurocoels, separated 
by a septum. 
Although in some instances the centrum and neural arch appear to 
: be firmly sutured together, this is not always the case. In Ce5 the two 
2lements were preserved adjacent but separate; they were somewhat 
distorted, and had to be taken apart in the laboratory and re-set. In 
Ce8 the centrum and neural arch were preserved completely sepa- 
ate in the same block. In the other vertebrae the line of the 
“neurocentral suture is clearly visible. Itruns more or less straight and 
orizontal and lies about one-third of the way up the sides of the 
‘neural canal. The diapophysis is on the side of the anterior half of the 
neural arch, directly above the parapophysis and the pleurocoel (in 
eontrast toAllosaurus). Itis flattened dorsoventrally, projects ventro- 
jaterally and scarcely varies through the series. There are no laminae 
fadiating from the diapophysis connecting it with the zygapophyses, 
which (as already stated) are remarkably large. The postzygapophyses 
‘lo not diverge more widely from the midline and from each other 
than do the prezygapophyses. 
| The neural spines are low, transversely thin, inclined very slightly 
backwards, and without spine tables. In Ce5-Ce8 the anterior edge 
pf the neural spine bifurcates ventrally into two ridges that diverge 
jowards the left and right prezygapophyses respectively. The poste- 
slor edge bifurcates similarly into two ridges that likewise run down 
31 
towards the epipophyses. In both cases, just below the point of 
bifurcation and between the ridges, a massive, rugose, bony protu- 
berance is developed in the midline, projecting respectively anteriorly 
and posteriorly, and clearly distinct from the neural spine itself. 
These projections, which doubtless served for the attachment of 
interspinous ligaments, vary considerably in size and form (see Fig. 
20). Epipophyses are massively developed, but by contrast, there are 
no hyposphenes (except in the axis) or hypantra. 
The first two cervical vertebrae, the atlas and axis, differ greatly 
from the typical members of the series that follow them and must be 
described individually. The atlantal pleurocentrum forms the odon- 
toid peg that protrudes from the front of the axis. Although it was 
found firmly attached to the dorsal third of the anterior face of the 
axial pleurocentrum, it is evident that it is not fused thereto; indeed, 
on the right-hand side it is somewhat displaced from its natural 
position. It is of fairly typical form: a hemispherical protruding peg 
with a flattish semicircular dorsal surface, convexly rounded in 
front, below, and on either side. Posteriorly these lower surfaces 
flare out like a collar to form a crescentic articulating surface for the 
(missing) atlantal intercentrum. At the base of the anterior surface is 
a small median depression which may well be the notochordal pit. 
The atlantal intercentrum is missing entirely. It may be presumed 
that it fitted into the middle third of the anterior face of the axial 
pleurocentrum. 
The right atlantal neurapophysis (the right half of the neural arch 
of the atlas vertebra) appears in dorsolateral view as a large, roughly 
trapezoidal, plate-like element that arises dorsally by the shorter of 
its two parallel sides from what might best be described as a 
footplate below. It widens dorsomedially towards the longer parallel 
side, which is virtually straight, but at the same time it curves round 
more horizontally so that it would have approached the left element 
in the midline to form a roof over the most anterior part of the neural 
canal. It also becomes much thinner as it approaches its medial 
margin, which is so thin as to be almost blade-like. The anterolateral 
part of the smooth external (dorsolateral) surface is strongly convex 
in all directions. The internal (ventromedial) surface, which is 
correspondingly concave and almost as smooth, bears a distinct 
facet close to its posteroventral edge which articulated with the axial 
prezygapophysis. The anteroventral margin is concave in profile, 
smooth, and thickly rounded; it forms the rim of the anterior opening 
of the neural canal, from which the spinal cord passed forwards into 
the foramen magnum. The posteroventral margin is thin dorsally but 
somewhat thicker ventrally. 
The base of this trapezium is expanded in two directions into what 
we have referred to above as a ‘footplate’: a medial process that 
extends towards the (missing) atlantal intercentrum, and a lateral 
process that forms a triangular wing when viewed from above or 
below. Neither of these processes has been perfectly preserved, both 
being a little eroded towards their ends. To the best of our knowledge 
no such lateral process has been reported in any other theropod. 
The axial intercentrum is represented by the usual crescentic 
wedge. Allowing for some distortion, its posterior face fits fairly 
well into the ventral third of the anterior face of the axial pleurocen- 
trum. Its anterodorsal face, which must have articulated with the 
missing atlantal intercentrum, is extended forwards into a protuber- 
ant lip. Between these two faces, beneath the lip, is the ventral 
surface, slightly concave from front to rear; on either side it curves 
upwards and becomes narrower. 
The axis itself, with the odontoid attached to its anterior face, is 
almost perfectly preserved. It is very small compared to the pre- 
sumed size of the skull. It shows strong opisthocoely (like the other 
cervicals), and its anterior face possesses a pair of facets, situated 
ventrolaterally, which articulated with the axial intercentrum. There 
