epoccipitals are characteristic of Anehieerarops (Langston 1959). The epoccipital is almost equi- lateral and is thicker in the center of its base than the adjacent frill bone.

Vascular grooves occur on the frills of most ceratopsians and are apparently somewhat variable within species. Triceratops, Arrhino- eerarops, and Anetueerarops generaHy show the most heavily sculptured surfaces. But Arrh~no- cerarops and Torosaurus do not have large triangular epoccipitals, and the surface sulci of Triceratops are less regularly disposed than those of Anchieeratops. Triceratops epoccipitals seem to attach to facets along the edges of the frill and are not large cap-like osteoderms fused to epocci- pital processes of the frill bones as in Anehi- eerarops. Although its size agrees better with A. /ongirostris Sternberg than with A. Ornarus Brown, the position of this fragment in the frill in unknown, and there is no basis for specific assignment.

Two sections of ceratopsian horns have been found at Scabby Butte. The larger one (NMC 10645) was obtained on the surface several hundred metres southeast of Site 2. It is 8.6 cm long, has a greatest proximal (Yanteroposterior) diameter of 10.8 cm, a distal diameter of 8.4 cm, and an oval transverse section. The horn was gently curved, but I am uncertain in which direction. A cavity occurs in the base of the specimen, and, lacking any trace of a 'median' suture, it is most likely a brow horn. Generic identification of this fragment by itself would be impossible, but since it occurred in beds con- taining the hornless Pachyrhinosaurus and the long-horned Anchieerarops, it may reasonably be referred to the latter genus.

1595

LANGS TON: CERATOPSIAN DINOSAURS

:representing epoccfipital processes. These are oval and arranged so that their long axes cross the edge of the frill obliquely, rising posteriorly. They are separatIId by wide and similarly oriented (oblique) inotches in the edge of the friH. Epoccipital bones do not seem to be present. The parietal thins rapidly toward the edge of the intrapaeietal fenestra, but no edges of this opening are preserved. As usual in cer- atopsian frill bones„ the surface is deeply marked by irregular grooves that are more numerous and deeper on the underside than above. Two possible arrangements of this specimen in the frill are suggested in Fig. 13.

At first glance the second frill (NMC 10644) appears quite diffetent. It bears no spikes, and whereas the edge olI the first specimen is broadly curved, the second; is almost straight (Fig. 14). But both bones have the same unusual epocci- pital development, similar thickening of the edges, and rapid thinning of the parietal bone medial to this. Fuitthermore, NMC 10644 con- tains a large sinus with an internal ridge some- what similar to that in NMC 9602. This specimen may therefore represent the same part of the frill as NMC 9602, with the spike replaced by an epoccipital process, This differs from other pro- cesses, being twice as long and having its principle axis parallel to that of the frill instead of obliquely inclined.

These specimens] appear to represent frills of similar size, so the differences noted are aberrant, perhaps attributable to individual ( asexual) variation. In this connection, the variability in the development of' the forwardly hooked pro-

)X) )I
CO

...:.~ I J (i,.

0
Vl

4
Fro. 13. Ccratopsian frill, NMC 9602, from Site 2,
tentatively referred to Pachyrhinosaurus, anterior aspect.

Two possible orientations of this specimen are shown
as explained in the text. A-E, sections through frill at
points indicated. Scale: 10cm.

narrowly separated at its base from a larger
process, now lost. Thickening of the frill beneath
the spikes is comparable to that seen in Centro-

0 c)

saurus adjacent to the frill hooks. This is the
< 8 thickest part of the frill and has an oval cross-
section with a maximum diameter of 6.9cm.

8 ~R The thickened region contains a huge sinus
surrounded by broken edges. This opening is
about 15.7 cm wide and 5 cm high and is partly
divided into four subequal pockets by two broad
intersecting ridges within. If the interparietal bar
is correctly identified, the cavity extends across
the midline of the skull and was bilaterally
asymmetrical. The bony floor beneath the cavity
is expanded downward, and its ventral (i.e.„

0

6~

V

oS
ui >

rn )

Q)

W)

O P
rA

v c)
4

V

A ~ B C Q) D ~ E

size clearly removes any possibility that the
other bones might belong to the same individual
as this vertebra.

Ceratopsian Frills

Parts of two ceratopsian frills of unusual form were found at Site 2. Both specimens are of the same sort, though they differ considerably in form and probably represent two individuals. The better specimen (NMC 9602) is believed to be part of the right parietal (Fig. 13). Viewed from above, it is reminiscent of the parietal of Styraeosaurus, for it bears at its edge a huge posterolaterally (or laterally) directed spike. Just medial to this spike is a smaller, hooked process suggestive of the medially turned 'hooks' on Centrosaurus frills (see Lambe 1904, Pl. I). The large spike has a preserved length of 23 cm, but some of the tip is missing. It is essentially straight, curving only a little toward the side distally. The base has an oval section 12.5 cm wide and 7.4 cm high. Most of the spike appears to have been accidentally flattened dorsoven- trally and, in its normal condition, would probably have maintained these proportions throughout most of its length. The surface is roughened, and several wide, roughly parallel grooves extend unbranched almost the full length of the process. Firmly fused to the edge of the frill, the base is expanded in such a way as to suggest that this spike is of osteodermal origin and not an outgrowth of the parietal bone. It is uncertain if the hooked process is completely preserved; what is seen may be only a thick splinter adhering to the edge of the frill after the major part of another large spike had been broken away. It may also be a discrete structure