BARYONYX WALKERI 
displaced to varying extents, and in yet others they were found 
completely separate. Their rugose sutural surfaces on either side 
were unmistakable and generally complete. This lack of co-ossifica- 
tion in the vertebral column confirms our observations on the skull 
and our suggestion that this particular individual might well be — 
_ despite its great size — a juvenile. 
| The neurocentral suture itself is invariably straight. In D1 the 
_ centrum and neural arch were found in their natural relative position. 
_ In D2 they are more or less in their proper position but have come 
, slightly adrift. In D3 the neural arch has been tipped forwards by 
about 45° so that, though still in articulation anteriorly, posteriorly it 
_ is separated from the centrum by a wide gap. In D5, D6, D11 and 
_ D14 the two components had parted completely but have been re- 
| assembled in the laboratory. 
The neural spines increase steadily in height (see Table 1, p. 30) 
from 87 mm in D1 to 197 mm in D11 and 201 mm incomplete in 
D14. (There is one exception to this in the case of DS, which at 96 
mm complete is shorter than either D2 or D3.) D1 is very slender in 
t lateral view, and D2 and D3 are successively more slender still, all 
three measuring considerably less in an axial direction than their 
transverse thickness. D5, however, reverses this trend, and from 
} there on, while the transverse thickness remains much the same, the 
_ axial length increases steadily. Thus, in D14 the axial length is four 
_ and a half times the transverse thickness. To summarize, the neural 
_ spines change from a short stout rod at the front of the series to a tall 
broad plate at the back. 
There are prominent rugosities on the neural spines of D1-DS5, 
projecting in front of and behind the spine itself and of more or less 
uniform width along its height. They represent the ossified portions 
of the interspinous ligaments. These have the effect of giving the 
spine a cruciform cross-section in D1, an appearance which is 
| 
| 
deep anteroposteriorly; each rugosity projects dorsally above the top 
of the neural spine proper so as to give a pronounced saddle-shaped 
appearance in lateral view. In DS there are still slight rugosities, the 
anterior being better developed than the posterior; the anterior 
margin of the spine slopes obliquely posterodorsally, the posterior 
_|margin is more or less vertical, and the short dorsal margin is 
_ transversely narrow, horizontal, and almost flat. Behind this point 
| there are no more rugosities. 
i In D6 the anterior margin of the neural spine curves forwards 
| towards the top; the posterior margin is again vertical, in the same 
_ plane as the posterior face of the centrum. The top of the spine, 
_ however, is much wider than in DS, and in lateral view is pitched like 
a roof, its apex being appreciably behind its mid-point. D7 is very 
| similar, a little wider still, but with the apex of the roof-like dorsal 
_ Margin exactly central. The neural spine of D11 is again very similar 
_ but a great deal wider, with a more steeply pitched ‘roof’ and with a 
_)median flange-like projection on its posterior margin; its widest 
_ point is therefore at about half height. In D14 the anterior edge of the 
_/neural spine is badly damaged but the base of a forward projection 
_ temains; more dorsally there is a very pronounced posterior projec- 
tion overhanging the postzygapophyses and high above them. This 
last dorsal vertebra is the only one in which the full height of the 
_ neural spine is not preserved, but, even so, it is higher than any of the 
_jothers. There are no spine tables anywhere in the series. 
| Another variable character is the angle of inclination (to the 
horizontal) of the zygapophysial facets (see Table 2), which are 
jusually tilted so as to face dorsomedially (prezygapophyses) or 
| ventrolaterally (postzygapophyses). Again, it is not possible to make 
'|precise measurements, but our somewhat crude figures give a good 
jindication of general trends. The inclination tends to decrease as we 
'|pass backwards down the series. Note that there can be large 
_ accentuated in D2. In D3 the cross is transversely wider but not so 
3) 
variations between different parts of the column (7°—32°), between 
adjacent vertebrae (20°—27°), and even between the pre- and 
postzygapophyses of the same vertebra (7°-18.5°). Nevertheless, it 
might be supposed that the postzygapophysial angle of a given 
vertebra should match the prezygapophysial angle of the vertebra 
behind, with which it articulated. In four of the six cases where there 
are two adjacent vertebrae with measurable facets there is indeed a 
close correspondence, with discrepancies of 0.5°, 1°, 1.5° and 2°; 
another is not too big, at 4.5°; but the worst case is a discrepancy of 
13.5°, between the axis and the supposed third cervical, which leads 
us to suspect that the latter may be badly distorted. 
Table 2 Inclination of the zygapophysial facets. 
A B 
Ce2 (axis) - 18.5° 
Ce3 sy? AOS? 
Ce5 24° I@.5° 
Ce6 20° ils 
Ce8 Zils PISS 
D1 - 20° 
D2 24.5° Die 
D3 28° 30° 
DS 14° 1? 
D6 10.5° 7-3! 
D7 15.5 i 
D10 ilpe - 
D11 WS a 
D14 Ti 18.5° 
unnumbered caudal - Ail? 
A, angle of inclination of prezygapophysial facets (facing dorsomedially); B, angle 
of inclination of postzygapophysial facets (facing ventrolaterally). 
The angle given is the angle between the zygapophysial facet and the horizontal 
plane, rounded off to the nearest half-degree. Where measurements can be made on 
both members of a pair of facets the figure quoted is the mean of the two. 
In all but the most anterior dorsals of Baryonyx a hyposphene and 
a hypantrum are well developed. The hyposphene is a large solid 
median projection between and below the postzygapophyses. Seen 
from behind it is triangular, with the apex pointing upwards, and on 
either side a large semilunar facet is directed dorsolaterally. The 
hyposphene as a whole fitted neatly into the gap between the 
prezygapophyses of the succeeding vertebra, where its facets articu- 
lated with the hypantrum; this latter comprises a pair of similar 
semilunar facets on the medial sides of the prezygapophysial pedun- 
cles, directed ventromedially. 
In D1 the prezygapophyses are not preserved, in D2 they are very 
widely separated, and in D3 slightly less so. It is obvious, therefore, 
that hypantra were absent. In all three of those vertebrae the 
hyposphene, if present at all, is the merest rudiment and could not 
have been functional. In contrast to this, both hyposphene and 
hypantrum are well developed in D5 and — as far as the state of 
preservation will permit us to determine — from there on throughout 
the rest of the dorsal series to D14. Although the hyposphene of D14 
is broken off, enough of it remains to indicate unequivocally that 
there must have been a hypantrum on the first sacral vertebra. 
In vertebrae D6-D10 (and perhaps D11) there is a pair of conical 
pits (peduncular fossae) on the anterior face of the neural arch, 
immediately above the opening of the neural canal and medial to the 
bases of the prezygapophyses. In D6 they are fairly widely spaced, 
circular and shallow. In D7 they are closer together and deeper. In 
D10 they are very close together, larger and deeper still. Similar pits 
are present in Allosaurus in vertebrae D9-D14, D12 excepted; they 
are developed best in D9. 
