56 
and on the referred Moroccan fossil). He believed that both genera 
might be in the same family, but concluded that, for the time being, 
the Spinosauridae and the Baryonychidae should be regarded as 
separate but closely related. We subsequently argued (Charig & 
Milner 1990: 131-133, 139) that the Moroccan maxilla resembled 
Baryonyx more closely than it did Spinosaurus, and we stated 
unequivocally that it should be referred to the Baryonychidae as 
‘baryonychid gen. et sp. indet.’. 
Buffetaut (1992), however, listed certain objections to our argu- 
ments: 
1. We had claimed that his 1989 illustrations of the Moroccan 
maxilla showed a ‘peg-like anterior extremity’ and noted also that 
the posterodorsal rim of the fragment was the ‘anteroventral 
portion of the rim of the external naris’, despite the fact that 
neither of those structures was mentioned in the legend or in his 
description. Buffetaut subsequently stated (1992) that no evi- 
dence of either structure could be seen on the specimen; it seems 
that we had misinterpreted as true edges the broken edges of the 
fragment (as seen in the illustrations). 
2. We had wrongly accused Buffetaut of misinterpreting the broken 
posterodorsal rim of the fragment as evidence of an antorbital 
fenestra, failing to note his comment (1989: 82) that “no evidence 
of an antorbital fenestra is visible’. 
. We stated (1990: 32) that the characters common to the Moroc- 
can specimen and to Spinosaurus ‘can hardly be considered on 
their own as diagnostic of a particular genus or even a particular 
family’, but we nevertheless included most of them in our list of 
the important characters which, in combination, typify Baryonyx. 
These statements, however, do not really contradict each other; 
characters that are insufficiently diagnostic on their own might 
yet be helpful in combination with others. 
Ww 
Moreover, further preparation and research have shown that our 
interim accounts of the partly developed Baryonyx contained some 
errors, e.g. we had stated wrongly (1990) that the animal possessed 
at least one elongated neural spine. 
The matter was broken down by Buffetaut into two separate 
problems. Should Spinosaurus and Baryonyx be included in the 
same family? (If the answer is yes, then the family would have to be 
called by the older family name, viz. Spinosauridae.) The resolution 
of this problem depends only on material that may be included with 
near-certainty in one genus or the other, i.e. only on the two 
holotypes; the Moroccan maxilla is wholly irrelevant. Secondly, is it 
possible to tell whether the Moroccan maxilla is more similar to 
Baryonyx or to Spinosaurus, and if so, which? 
Buffetaut (1992) was right in pointing out the incorrectness of the 
similarities that had caused us to ally the Moroccan maxilla with 
Baryonyx rather than with Spinosaurus (Charig & Milner 1990); 
without those alleged similarities the only structures on which the 
Moroccan maxilla may be compared with either Baryonyx or 
Spinosaurus are the teeth. Table 3 summarizes all relevant compari- 
sons between the teeth of those three forms (as far as is possible with 
the extremely limited material at our disposal). 
On teeth alone it can be seen that the Moroccan maxilla resembles 
Spinosaurus more closely than it does Baryonyx, which is what 
Buffetaut (1989) had originally claimed. 
Sereno et al. (1994: 270) also proposed a unique family relation- 
ship between Baryonyx and Spinosaurus, based on a cladogram 
containing 44 characters in 9 terminal taxa, including the 
Spinosauridae (specifying Baryonyx and Spinosaurus) and 
Afrovenator, a new theropod of Early Cretaceous age. This relation- 
ship was based on the following proposed synapomorphies: 
A.J. CHARIG AND A.C. MILNER 
Table 3. Comparisons of teeth of Spinosaurus with those of related 
forms. 
Baryonyx Moroccan maxilla Spinosaurus 
All teeth: 
curvature recurved, slightly ? hardly recurved 
but consistently at all 
serrations 7 to the mm none none 
fluting lingual sides of not mentioned sometimes very 
faint, fine vertical 
striping on enamel 
towards base 
crowns fluted 
Upper teeth only: 
spacing widely spaced widely spaced dh 
interdental plates on both none ‘ 
pmx and mx 
labial edge of Wavy in wavy in 
maxilla ventral view ventral view ? 
Lower teeth only: 
number 32 ? not more than 15 
spacing close together ? well separated 
interdental plates present ? absent (fide Stromer 
and Buffetaut) 
1. Elongate prenarial snout. In Baryonyx the whole of the snout is 
elongate; in Spinosaurus it is unknown. 
2. Specialized anterior dentition, manifested by increase in number 
of premaxillary teeth. Baryonyx has 6/7 premaxillary teeth, | 
which is 1/2 more than the 5 usual in theropods; but, in any case, | 
their number is unknown in Spinosaurus. 
3. Specialized anterior dentition manifested also by terminal rosette. 
We coined that term for the horizontally expanded tip to the upper | 
jaw in Baryonyx, almost completely surrounded by dental al- | 
veoli. The upper jaw of Spinosaurus is unknown except for a | 
small fragment of maxilla. 
4. Increase in height of dorsal neural spines. This was mistakenly 
reported in our 1990 publication and is incorrect. In contrast, the | 
dorsal neural spines of Spinosaurus are greatly elongated (hence | 
the generic name). 
Thus, of the four allegedly synapomorphic character-states, three | 
are unknown in Spinosaurus and may therefore be discounted. The | 
last is present in Spinosaurus but absent in Baryonyx. This means | 
that the synapomorphies of Sereno er al. (1994), considered on their | 
own, provide no justification for placing the two genera concerned in 
the same family. 
Consequently, we are now in agreement with Buffetaut in so far as 
we accept a particular relationship between Spinosauridae and) 
Baryonychidae. However, in the absence of decisive evidence in) 
either direction, the following factors predispose us against 
synonymising the two families: 
1. The material of Baryonyx is far more complete than that of 
Spinosaurus, and Baryonyx would therefore make a much more 
informative type-genus for the family in which it is included. More} 
importantly, all of Stromer’s (1915) Spinosaurus material, housed 
in Munich Museum, was destroyed in an Allied bombing raid in| 
1944, and is therefore no longer available for comparisons; nor is 
there any other material that could be designated as a neotype. 
2. The most important distinguishing feature of the Spinosauridae, | 
as suggested by the name, is the elongation of the neural spines |~ 
on the vertebrae. Baryonyx has no such elongated spines. The 
presence or absence of such spines may have little or no 
