BARYONYX WALKERI 
this means that they must have evolved independently in Baryonyx 
and in the other forms concerned. Examples are: 
5. Pronounced subnarial gap, indicating a possibly mobile premax- 
illary-maxillary joint. This gap (renamed the subrostral notch in 
Baryonyx, see p. 14) occurs also in certain ceratosaurs, namely 
Coelophysidae and Dilophosaurus, and was therefore consid- 
ered by Holtz (1994a: 1104) tobe anunambiguous synapomorphy 
| for the Coelophysoidea. Rowe & Gauthier (1990: 154) claimed 
that articulated material shows the premaxillary-maxillary joint 
| of ceratosaurs to bea firm junction, as is also the case inBaryonyx. 
| 77. Nasals narrow. This occurs also in Dromaeosaurus, 
Archaeopteryx, Tyrannosaurus, Troodon and Ornithomimus. 
13. Parietals projected dorsally. This occurs also in Ceratosaurus 
and Abelisaurus. 
__ We considered also the partial analysis of Sereno er al. 1994 (in 
_ which the synapomorphies are listed only under “References and 
| Notes’). Those authors did consider Baryonyx, though not as a 
| separate O.T.U. but as part of the taxon Spinosauridae (see above, p. 
| 56). Their cladogram (Fig. 46) shows the Spinosauridae as the sister- 
group of Torvosauridae, arising from node 4 (unnamed, our 
/ numbering); those two O.T.U.s together as the sister-group of 
Afrovenator, arising from node 3 (Torvosauroidea); and the 
_ Torvosauroidea as the sister-group of the Neotetanurae. The 
| Torvosauroidea and the Neotetanurae together constitute the 
_ Tetanurae. We replaced Spinosauridae as an O.T.U. by the genus 
| Baryonyx on its own, in an otherwise unmodified data-matrix, then 
_ tan the programme as described above (p. 57). This produced a 
single most-parsimonious tree of 58 steps, C.I. 93%, R.I. 90%, 
| identical to the tree figured by Sereno et al. except in that Baryonyx 
. occupies the position that was previously occupied by Spinosauridae. 
_ This cladogram requires critical scrutiny. Sereno et al’s (1994) 
piinking of the Spinosauridae (which included Baryonyx; 
_ Spinosauroidea sensu this work) with the Torvosauridae 
| (Torvosaurus + Eustreptospondylus) to form an unnamed taxon is 
| supported by only two synapomorphies: 
"1. Radius less than 50% of humeral length. This is true of Baryonyx 
(49%), but in Spinosaurus neither humerus nor radius is known. 
(1) Neotheropoda 
Fig. 46 Cladogram of the Theropoda, according to Sereno et al. (1994). 
59 
N 
Manual digit I ungual elongate (three times height of proximal 
articular end). This is essentially true of Baryonyx, where the 
length of the ungual is more than four times the height of the 
proximal end: length of ungual measured around the outside of 
the curve 310 mm, height of end 73 mm. Again, the manus is 
unknown in Spinosaurus. 
Thus, while there is no evidence either way for the condition in 
Spinosaurus, it seems that Baryonyx and Torvosaurus do share two 
characters. However, such similarities in relative proportions could 
easily have developed homoplastically and therefore lack 
phylogenetic significance. Further, while the reduction or absence 
of a quadrate foramen links Torvosaurus with Afrovenator (see 
below), that same foramen is prominent in Baryonyx; indeed, it is 
more of a fenestra than a foramen. Thus there does not seem to be an 
adequate demonstration of a close phylogenetic link between 
Baryonyx and the Torvosauridae. 
Sereno etal. (1994) also claimed that this unnamed Torvosauridae 
+ Spinosauridae taxon was the sister-group of their new Saharan 
theropod Afrovenator, ranking the combined taxon as a superfamily 
and naming it Torvosauroidea (though those same authors, in their 
1996 publication, called it Spinosauroidea). This was based on five 
synapomorphies, all characters of the skull: 
1. Anterior ramus of maxilla as long anteroposteriorly as tall. In 
Baryonyx the whole of the maxilla as preserved is anterior to the 
antorbital fenestra and may therefore be presumed to be part of 
the ‘anterior ramus’ of Sereno ef al.; in which case it is several 
times longer than tall. In Spinosaurus practically nothing of the 
maxilla is preserved and this character-state is therefore un- 
known. 
Anterior ramus of lacrimal dorsoventrally narrow. In Baryonyx 
this structure, more usually referred to as the nasal or dorsal 
ramus, shows a condition intermediate between those of 
Torvosaurus and Allosaurus. Nothing is known of the condition 
in Spinosaurus. 
3. Lacrimal foramen small and positioned at mid-length along the 
jugal ramus. This is confusing in so far as there are three lacrimal 
foramina in Jorvosaurus (Britt 1991: 17-19); Britt’s photograph 
of the lacrimal in medial view clearly shows that the large one, 
which he called the posterior lacrimal foramen, lies one-third of 
the way down the orbital side of the jugal ramus. In Baryonyx also 
there are three foramina, one large and, close together, two very 
small ones; the large one lies beneath the junction of the two 
rami, i.e. at the dorsal extremity of the jugal ramus. The corre- 
sponding character-state is unknown in Spinosaurus. This 
proposed synapomorphy is altogether unclear. 
4. Ventral process of postorbital broader transversely than 
anteroposteriorly. This character-state is not found in Baryonyx 
and is unknown in Spinosaurus. 
5. Quadrate foramen reduced or absent. Baryonyx has a large, 
prominent quadrate foramen between the quadrate and the 
quadratojugal. The condition is again unknown in Spinosaurus. 
N 
In summary, two of the relevant character-states (nos | and 2) may 
be present in Baryonyx but it is difficult to have much confidence in 
them. For the other three (nos 3, 4 and 5) the character is either 
unclear or in the opposite state. All five characters are unknown in 
Spinosaurus; and none appears in the abbreviated description and 
diagnosis of Afrovenator abakensis published by Sereno et 
al. (1994: 270). We can find no justification for the placing together 
of these alleged sister-groups. 
The net result of our re-working of the analysis of Sereno et al. 
