Chasmosaurus russelli Sternberg, 1940

Maidment, Susannah C. R. & Barrett, Paul M., 2011, A new specimen of Chasmosaurus belli (Ornithischia: Ceratopsidae), a revision of the genus, and the utility of postcrania in the taxonomy and systematics of ceratopsid dinosaurs, Zootaxa 2963, pp. 1-47 : 5-39

publication ID

https://doi.org/ 10.5281/zenodo.278172

DOI

https://doi.org/10.5281/zenodo.4584095

persistent identifier

https://treatment.plazi.org/id/03E33412-F770-7317-FF66-FE43E8E856DC

treatment provided by

Plazi

scientific name

Chasmosaurus russelli Sternberg, 1940
status

 

Chasmosaurus russelli Sternberg, 1940

Mojoceratops perifania Longrich, 2010: 683 .

Amended diagnosis. Taxon displaying the combination of characters unique to the genus Chasmosaurus along with the following features: (1) parietal posterior bar bearing a median emargination and is broadly arched either side so that it forms an ‘M’ shape with the parietal median bar; (2) each side of the posterior parietal bar bears three roughly equally sized epiparietals (after Godfrey & Holmes, 1995). Both characters are autapomorphic for C. russelli within Chasmosaurus .

Holotype. CMN 8800, comprising a partial skull lacking the lower jaw and part of the rostral, part of the jugals from both sides, part of the right quadrate, squamosal, and parietal (Sternberg 1940).

Distribution. Lower beds of the Dinosaur Park Formation, Alberta, Canada ( Ryan & Evans 2005).

Referred specimens. AMNH 5656; CMN 2280, 41933; TMP 1983.25.0 1

Description of NHMUK R4948. Skull. The skulls of ceratopsid dinosaurs are often found articulated with the individual bones fused together and sutures between the elements entirely obliterated. The skull of NHMUK R4948 was disarticulated when discovered, and it has not been reconstructed, presenting the rare opportunity to examine the morphology of a number individual skull bones from multiple views. This lack of fusion, together with several postcranial characters (see below), suggests that the specimen was not osteologically mature at the time of death. Skull, lower jaw and braincase measurements are given in Table 1 View TABLE 1 .

Rostral. The rostral is a single median element fused anteriorly to the premaxilla forming a beak-like tip to the snout, and its presence constitutes a ceratopsian synapomorphy (e.g. Sereno 1986, 1999). In NHMUK R4948 the element is preserved in articulation with the premaxilla ( Fig. 1 View FIGURE 1 ), but the sutures between them are clear. In lateral view the element has an elongate dorsal process that contacts the anterodorsal premaxilla along a flattened ventrally facing facet, and an elongate posterior process (broken on the right side, but better preserved on the left) that extends along the ventral margin of the premaxilla ( Fig. 1 View FIGURE 1 D). The ventral margin of the rostral is concave upwards in lateral view, as in all ceratopsids ( Godfrey & Holmes 1995). The surface of the rostral is rugose, pitted and grooved, suggestive of a horny covering in life. In ventral view, the rostral is deeply concave and drawn to a posterior point on the midline where it intervenes for a short distance between the premaxillae.

Premaxilla. The premaxillae appear to be fused on the midline but the dorsal half of the left element is not preserved and has been reconstructed ( Fig. 1 View FIGURE 1 ). The dorsal process of the premaxilla is thickened transversely and the surface is rugose, as in other specimens of Chasmosaurus (ROM 843, 839; CMN 2280; TMP 1981.19.175) and Vagaceratops (CMN 41357). At the dorsal end of the process a flat facet is present laterally for the nasal ( Fig. 1 View FIGURE 1 B); the suture between the two elements interdigitated so that the nasal overlapped the premaxilla laterally but a tongue from the fused premaxillae extended ventrally between the two lateral processes of the nasals on the midline (in contrast to the situation described in Agujaceratops [ Forster et al. 1993; Godfrey & Holmes, 1995], but similar to that seen in other specimens of Chasmosaurus [e.g. ROM 839; CMN 2280; UALVP 40] and Vagaceratops [CMN 41357]). The premaxilla forms the anterior margin of the external naris. The snout was transversely narrow in this area and the left and right premaxillae closely approach each other forming a shallow premaxillary fossa ( Fig. 1 View FIGURE 1 B; Godfrey & Holmes 1995). Anteriorly, the fossa is pierced by a large fenestra, the edges of which are broken and have been reinforced with plaster ( Fig. 1 View FIGURE 1 B). A smaller foramen pierces the fossa just ventral to the larger foramen on the left side ( Fig. 1 View FIGURE 1 D), but this area is covered with plaster on the right. A similar foramen was also observed in ROM 839 ( Chasmosaurus sp.) and is present in Agujaceratops and other ceratopsids ( Lehman 1989). A thin flange that arises from the ventromedial premaxillary palate extends posteriorly into the external naris ( Fig. 1 View FIGURE 1 B), as in Chasmosaurus ( Forster et al. 1993) and Agujaceratops ( Lehman 1989; Forster et al. 1993). A similar flange is present in Utahceratops and Kosmoceratops but it is less developed and does not extend along the entire anterior margin of the external naris (Sampson et al. 2010). The ventral margin of the premaxilla is edentulous, a feature characteristic of all ceratopsids (e.g. Sereno 1986, 1999), and transversely expanded relative to the premaxillary body above it. Anteriorly, the ventral margin bears a flattened facet laterally and a deep groove ventrally, both for the articulation of the rostral ( Fig. 1 View FIGURE 1 B). A similar lateral facet can be seen in UALVP 40 ( Chasmosaurus sp.). The transversely thickened margin extends posteriorly to form a process that extends under the external naris, forming its ventral margin. Posteriorly the process turns dorsally and there is a grooved and ridged facet for the maxilla ventrally ( Fig. 1 View FIGURE 1 D). A flattened facet is present dorsally for the nasal ( Fig. 1 View FIGURE 1 B). The distal end of the process is broken off on both sides. In ventral view the ventral margin of the premaxilla curves medially to form a rudimentary premaxillary palate, but it does not meet its counterpart on the midline. This represents an unusual state in ceratopsids, but the space between the premaxillae in ventral view is infilled with plaster, and it may be that the medial margins are broken. The palate is pierced by two foramina, the anterior of which is larger and extends forwards as a shallow canal towards the rostral suture.

Maxilla. Both maxillae are preserved ( Fig. 2 View FIGURE 2 ). The maxilla is roughly triangular in lateral view. The anteriormost portion of the maxilla is reconstructed on both sides, but a groove is present anterodorsally on the left maxilla, into which the nasomaxillary process of the premaxilla would have fitted ( Fig. 2 View FIGURE 2 G). The lateral surface of the maxilla is flat anteriorly, but moving posteriorly a sharp ridge arises that extends posterodorsally resulting in medial offset of the tooth row ( Fig. 2 View FIGURE 2 C, G), as in other specimens of Chasmosaurus (ROM 843, 839; UALVP 40; TMP 1981.19.175) and other chasmosaurines (e.g. Agujaceratops, Lehman 1989 ; Utahceratops, Kosmoceratops, Sampson et al. 2010; Triceratops, Hatcher et al. 1907 ). Dorsal to this ridge the lateral surface of the maxilla is rotated to face dorsolaterally. The facet for the nasal is a dorsally facing surface that forms the apex of the maxilla. The nasal would have overlapped the maxilla dorsolaterally ( Fig. 2 View FIGURE 2 C). The ridge that extends across the maxilla, offsetting the tooth row medially, ends posteriorly in a fluted and grooved facet for the lacrimal and jugal ( Fig. 2 View FIGURE 2 C); between the lacrimal–jugal facet and the nasal facet is a shallow groove marking the position of the small antorbital fenestra ( Fig. 2 View FIGURE 2 C), which is similar to that observed in ROM 839 ( Chasmosaurus sp.). Ventral to the ridge the surface of the maxilla curves concavely towards the tooth row and is irregularly pierced with foramina ( Fig. 2 View FIGURE 2 C, G), as in other specimens of Chasmosaurus (e.g. ROM 843; ROM 839) and other chasmosaurines (e.g. Agujaceratops, Lehman 1989 ; Triceratops, Hatcher et al. 1907 ). Below the ridge, the posterior margin of the maxilla describes an arc, concave anteriorly, adjacent to which the coronoid process of the lower jaw and its associated musculature would have projected when the skull and lower jaws were in articulation. The posteroventral process of the maxilla bears a shallowly concave facet dorsolaterally that may have been for the ectopterygoid ( Fig. 2 View FIGURE 2 G; Godfrey & Holmes 1995). The tooth row dominates the maxilla in medial view. No teeth are preserved, but 28 alveoli are present in both maxillae; two additional sockets are reconstructed anteriorly on the right, while three additional sockets are reconstructed on the left. Twenty-seven teeth are present in TMP 1981.19.175 ( Chasmosaurus sp.), and tooth count is known to vary ontogenetically in chasmosaurines ( Forster et al. 1993). Anteriorly, the medial surface of the maxilla is gently concave dorsal to the tooth row ( Fig. 2 View FIGURE 2 H). Posteriorly, a large foramen pierces the medial side of the maxilla and posterior to this a thin plate of bone projects dorsally ( Fig. 2 View FIGURE 2 H). Dorsal to the foramen the thin plate of bone forms the ventral margin of the antorbital fenestra. Posteroventrally, just dorsal to the tooth row, is a fluted, ridged and grooved facet for the pterygoid ( Fig. 2 View FIGURE 2 H).

Nasal. The nasals are firmly fused to each other on the midline ( Fig. 3 View FIGURE 3 ). Anteriorly a short, transversely compressed process forms the articular surface for the premaxilla, which the nasal overlapped laterally ( Fig. 3 View FIGURE 3 B). Immediately posterior to the premaxillary facet the nasal horn core rises dorsally. A very prominent groove extends transversely around the front of the horn core ( Fig. 3 View FIGURE 3 D), fading on the lateral surface, and this may indicate the junction with the epinasal. That the suture between the epinasal and nasals is still visible may indicate that NHMUK R4948 was not fully osteologically mature at time of death (Horner & Goodwin 2006). A very prominent suture is also visible between the nasals in ventral view. The horn core is relatively short (15.5 cm from the dorsal margin of the external naris). It is oval in transverse cross-section with its long axis anteroposterior at its base ( Fig. 3 View FIGURE 3 E, F; in some specimens of Chasmosaurus the horn core is rounded at the base, but this seems to be intraspecifically variable: Godfrey & Holmes 1995), flattened on top and projects dorsally (in some specimens the horn core projects posterodorsally, another feature that is apparently intraspecifically variable: Godfrey & Holmes 1995). The morphology of the nasal horn core is rather different from that of Kosmoceratops and Agujaceratops , in which the horns are transversely compressed ( Forster et al. 1993; Sampson et al. 2010). Its surface is rugose and pitted in NHMUK R4948, suggestive of a horny covering in life ( Fig. 3 View FIGURE 3 ). A ridge extends up the horn core posteriorly, but disappears before the tip. Ventral to the horn core, the nasal describes an arc forming the posterior margin of the external naris in lateral view, as in other ceratopsids ( Longrich 2010). Ventrally there is a tongue-and-groove joint for articulation with the premaxilla ( Fig. 3 View FIGURE 3 B, D), as in CMN 1254 ( Chasmosaurus sp., Lambe 1915) and Vagaceratops (CMN 41357), although this has been partly reconstructed on both sides. Part of the posteroventral nasal has also been reconstructed, and the articular surface for the prefrontal and frontal is not preserved.

Circumorbital region. The bones of the circumorbital region, including the lacrimal, jugal, postorbital and supraorbital are fused together on the right side ( Fig. 4 View FIGURE 4 ; on the left only the jugal and part of the postorbital are preserved: Fig. 5 View FIGURE 5 ). The orbit is circular. Anteroventrally the lacrimal presumably forms the margin of the orbit, although sutures between the lacrimal and other elements cannot be seen. Medially the boundary between the lacrimal and the jugal appears to be defined by a very prominent ridge that bears a grooved facet for the maxilla ( Figs 4 View FIGURE 4 D, 5H). At least one supraorbital element likely formed the anterior and dorsal margins of the orbit ( Forster 1996). In ROM 843 ( C. belli ), ROM 839 and TMP 1981.19.175 a deep groove is present dorsally above the orbit that separates two areas of rugosity, plausibly suggesting a suture between two supraorbital elements, as in a variety of other ornithischians ( Maidment & Porro 2010). No such groove is observable in NHMUK R4948, however, although this area is heavily repaired with plaster. Anterior to the orbit the anterior supraorbital (palpebral of Godfrey & Holmes 1995) is rugose, bearing a large number of horizontally extending grooves and ridges ( Fig. 4 View FIGURE 4 C), as seen in other specimens of Chasmosaurus (ROM 843, 839; CMN 2245, 2280; UALVP 40; TMP 1981.19.175) and in some specimens of Agujaceratops ( Lehman 1989) , Kosmoceratops and Utahceratops (Sampson et al. 2010). The horn core is located anterodorsal to the orbit and projects anteriorly ( Fig. 4 View FIGURE 4 ). It is oval in cross-section at its base, very short (5.5 cm tall from the dorsal margin of the orbit), and flattened on top. Medially it is partially reconstructed. Supraorbital horn size is variable in specimens referred to Chasmosaurus ( Lull 1933; Godfrey & Holmes 1995) and other ceratopsids, such as Agujaceratops ( Lehman 1989) . The postorbital forms the posterodorsal margin of the orbit. It is formed of two distinct surfaces in lateral view: one surface faces dorsally and formed the skull roof posterior to the orbit, while one surface faces laterally and formed the posterior margin of the orbit ( Fig. 5 View FIGURE 5 A– D). Anteriorly, the left element bears a concave groove above the orbit for the posterior supraorbital (a neomorphic element: Maidment & Porro 2010; Fig. 5 View FIGURE 5 C). Posterior to this the orbital margin is slightly thickened and rugose. The medial edge of the dorsally facing surface extends posteromedially in a dorsoventrally deep, straight, flattened facet for the frontal, and ends in a medially facing concave surface that may have contributed to the frontoparietal fontanelle ( Fig. 5 View FIGURE 5 D). A deep, triangular pit below this surface indicates the facet for the laterosphenoid ( Fig. 5 View FIGURE 5 D), as seen in Agujaceratops ( Lehman 1989) . The posteromedial surface is broken on both sides, but posterolaterally there is a fluted and grooved interdigitating series of facets for the squamosal and parietal, and it is clear that these bones and the postorbital overlapped each other along this complicated suture ( Fig. 4 View FIGURE 4 C), as in CMN 1254 ( Chasmosaurus sp., Lambe 1915). The suture between the postorbital and jugal cannot be seen, but presumably extended from the lateral temporal fenestra to the posterior margin of the orbit, as in other ornithischians. The jugal forms the ventral margin of the orbit and a series of radial grooves extend across the lateral surface in this area ( Fig. 4 View FIGURE 4 C). The posteroventral process of the jugal extends posterior to the orbit, as in other chasmosaurines (e.g. Agujaceratops, Forster et al. 1993 ; Kosmoceratops, Utahceratops; Sampson et al. 2010; Vagaceratops, Holmes et al. 2001). The ventral margin of the process describes a smooth, dorsally concave curve; the posterior surface of the jugal forms the anterior margin of the lateral temporal fenestra. A pyramidal epijugal is situated at the distal end of the posterior process ( Figs 4 View FIGURE 4 C, 5G), as in other specimens of Chasmosaurus (e.g. CMN 2245, Lambe 1914a; AMNH 5401, Brown 1933; ROM 843; TMP 1981.19.175) and the suture between it and the jugal is clear in both lateral and medial view. It is 5 cm tall and 6.5 cm wide anteroposteriorly at its base. Its surface is pitted and rugose and it projects laterally. In medial view, ventral to the orbit, the jugal bears a smooth oval facet for the maxilla with a ridge extending posterodorsally across it. Immediately posterodorsal to the facet on the left side is a small foramen with a canal opening towards the maxillary facet ( Fig. 5 View FIGURE 5 H). The distal end of the posterior process of the jugal bears a two-pronged facet in medial view for the quadratojugal, which was overlapped by the jugal. A deep groove extends up the centre of this facet ( Figs 4 View FIGURE 4 D, 5H).

Quadrotojugal. Both quadratojugals are preserved: the right is almost complete while the left is incomplete posteromedially and has been reconstructed ( Fig. 6 View FIGURE 6 ). In lateral view the quadratojugal is roughly triangular with the apex pointing ventrally. The posterodorsal margin forms a portion of the ventral margin of the lateral temporal fenestra, a feature present in CMN 1254 ( Chasmosaurus sp., Lambe 1915) and UALVP 40 ( Chasmosaurus sp.), which are both juveniles. In some Chasmosaurus specimens, however the squamosal overlaps the top of the quadrate and quadratojugal in lateral view, contacting the jugal and excluding the quadratojugal from the margin of the lateral temporal fenestra ( Lambe 1914a; Godfrey & Holmes 1995; CMN 2280; CMN 2245; ROM 839). Several lines of evidence, such as the lack of fusion of some skull elements, a visible epinasal suture, lack of fusion of the ilia to the sacrum and visible neurocentral sutures on dorsal vertebrae (see below) suggest NHMUK R4948 was not osteologically mature at the time of death and the contribution of the quadratojugal to the margin of the lateral temporal fenestra may well vary ontogenetically. A well-developed ridge extends from the dorsal surface ventrally and projects strongly laterally; it reaches an apex ventrally ( Fig. 6 View FIGURE 6 D). Anterior to this ridge is a facet for the posterior process of the jugal; it is similar in morphology to the poorly preserved quadratojugal of Vagaceratops (CMN 41357) and that of Triceratops ( Hatcher et al. 1907) . The facet is deeply grooved and large, occupying half the quadratojugal; the apex of the ridge apparently helped to support the epijugal ossification: it is rugose and contributes posterolaterally to the epijugal when the elements are articulated ( Fig. 6 View FIGURE 6 D, F). Posterior to the jugal facet and ridge the surface of the quadratojugal is smooth. The posterior margin is transversely thick and bears a slightly concave facet for the quadrate ( Fig. 6 View FIGURE 6 E). This facet rotates from facing wholly posteriorly more dorsally to facing medially on the ventral quadratojugal. In this area the facet is very rugose and lumpy, as it is in Vagaceratops (CMN 41357). In medial view there is a flattened oval facet dorsally for the squamosal, indicating that at least part of the squamosal underlapped the quadratojugal ( Fig. 6 View FIGURE 6 E).

Quadrate. Both quadrates are preserved but have been repaired with plaster ( Fig. 7 View FIGURE 7 ). The quadrate is anteroposteriorly compressed and rectangular in lateral outline with the long axis trending dorsoventrally. Medially the pterygoid process is short dorsoventrally and incomplete on both sides; in posterior view a ridge extends ventrally down the medial surface of the quadrate and curves medially along the pterygoid process, marking the facet for the pterygoid ( Fig. 7 View FIGURE 7 F, H), as in Vagaceratops (CMN 41357). The lateral surface of the quadrate is the anteroposteriorly broadest part of the element. It is composed of three surfaces. The quadratojugal facet occupies the ventral third of the lateral surface, as in Agujaceratops ( Lehman 1989) and Vagaceratops (CMN 41357). Ventrally it is rugose, as is the corresponding surface on the quadratojugal, and moving dorsally it becomes smoother and rotates onto the anterior surface of the bone. Dorsal to this facet, which is defined laterally by a ridge trending posteroventrally, the middle part of the lateral surface would be visible when articulating with other skull elements in lateral view. Dorsally, again separated by a ridge, is a flattened facet for the squamosal ( Fig. 7 View FIGURE 7 E). Distally, the articular condyle of the quadrate is anteroposteriorly broad. In posterior view, a series of striations and low rugosities positioned dorsolaterally may indicate facets for the paroccipital processes ( Fig. 7 View FIGURE 7 F, H). Similar features were also observed in Vagaceratops (CMN 41357).

Squamosal. Both squamosals are preserved. They are shaped like elongate scalene triangles with the apex pointing posteriorly ( Fig. 8 View FIGURE 8 ). Anteriorly the left element preserves a fluted and grooved facet for the postorbital, which apparently overlapped the squamosal along much of the length of the contact ( Fig. 8 View FIGURE 8 F), as in Agujaceratops ( Lehman 1989) and other ceratopsids. Medially, an anterodorsally trending ridge offsets part of the postorbital facet laterally relative to the rest of the medial surface, forming an articular facet for the paroccipital process ( Fig. 8 View FIGURE 8 D). A second facet is present ventrally, probably for the head of the quadrate. The part of the squamosal that contributes to the lateral temporal fenestra is not preserved on either side. Ventral to the postorbital facet, the squamosal arcs posteroventrally. The lateral surface of the squamosal is flat anteriorly and becomes concave posteriorly, as in other specimens referred to Chasmosaurus (ROM 843, 839; CMN 2245, 2280), and bears numerous pits and grooves ( Fig. 8 View FIGURE 8 B). The medial margin of the squamosal, where it articulates with the parietal, is slightly transversely thickened. A large foramen pierces the surface of the left element, but this is not present on the right side; similar asymmetry is seen in ROM 843 and ROM 839 ( Chasmosaurus ; Godfrey & Holmes 1995), Utahceratops (Sampson et al. 2010) and other ceratopsids ( Tanke & Farke 2007). Medially the squamosal is gently convex. A groove extends all the way along the margin medially ( Fig. 8 View FIGURE 8 F) marking the facet for the parietal and suggesting the lateral parietal bar was complete, a feature that, although thought to be diagnositic of Chasmosaurus belli by Godfrey and Holmes (1995), is also present in other chasmosaurines (e.g. Kosmoceratops [Sampson et al. 2010]; Vagaceratops [ Holmes et al. 2001]; Torosaurus [ Hatcher et al. 1907]), some specimens of Chasmosaurus russelli (CMN 2280), and is known to vary within individuals in Pentaceratops ( Lehman 1989) . The ventral margin of the squamosal bears a series of fused pyramidal episquamosals with oval bases. On the right side seven are preserved, while on the left there are six, although there is space for one more. Sutures between episquamosals and the squamosals are very clear, suggesting one may not have been entirely fused and has been lost from the left side. Episquamosal number is known to vary between six and ten in specimens of Chasmosaurus ( Lull 1933; Godfrey & Holmes 1995; Farke 2011). The episquamosals are generally asymmetrical, with the apex nearer to their anterior margin than their posterior margin. The fourth episquamosal on the right side has a double apex ( Fig. 8 View FIGURE 8 B), and in medial view a groove separates the peaks, suggesting that this episquamosal was originally formed from two ossifications that have coalesced. This condition has not, to our knowledge, been reported before in ceratopsids.

Parietal. The parietal is preserved in six pieces that fit together and represent all parts of the element ( Fig. 9 View FIGURE 9 ). Anteromedially the parietal forms a transversely broad, dorsally convex ridged and grooved platform that extended anteriorly to form the posterior margin of the frontoparietal fontanelle ( Fig. 9 View FIGURE 9 B). Laterally, the sides of this platform are smooth and concave and form the medial margins of the supratemporal fenestrae, which were confluent with the parietal fenestrae, as in other specimens of Chasmosaurus and other ceratopsids (Farke 2010). Ventrally, this part of the parietal bears three rounded fossae. The median bar is complete but in three pieces, and as preserved is 86 cm long anteroposteriorly. It is roughly triangular in cross-section with the apex pointing ventrally and the triangle base forming a flat dorsally-facing surface that bears a series of longitudinal striations. Laterally the sides of the median bar bear irregular grooves and ridges that are asymmetric on either side. The posterior parietal bar lies at right angles to the median bar and there is no ‘U’-shaped emargination where the two meet ( Fig. 9 View FIGURE 9 ; Fig. 10 View FIGURE 10 ), in contrast to Agujaceratops ( Lehman 1989) , Chasmosaurus russelli ( Fig. 10 View FIGURE 10 D; CMN 2280) and Utahceratops (Sampson et al. 2010), but similar to other specimens of Chasmosaurus belli including the holotype (CMN 491; Lambe 1902), AMNH 5401 ( Brown 1933), ROM 843 ( Fig. 10 View FIGURE 10 A) and CMN 2245 ( Fig. 10 View FIGURE 10 B), Vagaceratops ( Holmes et al. 2001), Torosaurus ( Hatcher et al. 1907) and Kosmoceratops (Sampson et al. 2010). The dorsal surface of the junction between the bars is crossed with sinuous grooves that probably represent blood vessel channels ( Fig. 9 View FIGURE 9 B). The posterior bar, which is completely preserved on the left side, is rounded in cross-section but ventrally bears a thin anteriorly directed flange that is reconstructed and repaired with plaster along much of its length; how much of this feature is real is difficult to determine. The posterior bar of ROM 843 ( C. belli ) is also rounded in cross-section ( Fig. 10 View FIGURE 10 A), as is that of Agujaceratops ( Lehman 1989) , and TMP 1983.25.1 ( C. russelli ; Fig. 10 View FIGURE 10 E) but that of CMN 2245 ( C. belli ; Fig. 10 View FIGURE 10 B) is ‘C’-shaped, being strongly concave anteriorly, and this suggests the shape of the posterior bar is variable in Chasmosaurus and should not be used as a diagnostic feature of the genus (contra Longrich 2010). Epiparietals are not fused to the posterior bar and are not preserved, but the fusion of ossifications to the parietal may vary ontogenetically in Chasmosaurus ( Lehman 1989) as they do in Triceratops (Horner & Goodwin 2006) . The internal surface of the corner of the posterior bar and lateral bar is flattened, while the external surface, which forms the posterolateral corner of the frill, bears a facet for an epiparietal that was not fused and has not been preserved. The lateral bar is compressed dorsolaterally-ventromedially. Laterally it bears an elongate facet for the squamosal ( Fig. 9 View FIGURE 9 B, D), while medially there is a distinct groove below a low ridge. This ridge is better developed in CMN 2245 ( C. belli ), in which the lateral bar is triangular in cross-section with the apex pointing medially. The middle section of the lateral bar is not preserved in NHMUK R4948, but anteriorly the lateral bar flares medially as a thin flange to form the anterolateral corner of the parietal fenestra. The surface bears a large number of blood vessel grooves, a feature present in other Chasmosaurus specimens ( Lull 1933; ROM 843; CMN 2280) and is broken medially.

Braincase. The braincase is complete and, rarely for chasmosaurines, is disarticulated from the rest of the skull ( Figs 11 View FIGURE 11 , 12).

Individual braincase elements are fused and the sutures between them have been obliterated.

The paroccipital process is presumably formed by the fused exoccipital and opisthotic, and as in many adult reptiles, no suture is visible between them ( Romer 1956). In posterior view ( Figs 11 View FIGURE 11 E, 12A), NHMUK R4948 has an elongate paroccipital process that projects posterolaterally. It is deep dorsoventrally and flares distally; although it is incomplete distally on both sides, it does not appear to flare to the same degree as it does in the centrosaurine Diabloceratops ( Kirkland & Deblieux 2010: fig. 8.8). The dorsal part of the paroccipital process is gently excavated producing a shallow, posterodorsally facing sulcus, which probably served as an insertion area for the m. rectus capitis posterior (Fig. 12A; Tsuihiji 2010). As the dorsal margins of the paroccipital processes are poorly preserved, it cannot be determined if posttemporal foramina were present or absent. The morphology of this area is similar to that seen in other Chasmosaurus specimens (ROM 839; CMN 2245; UALVP 40; TMP 1981.19.175), Centrosaurus ( Dodson et al. 2004) , Diabloceratops ( Kirkland & Deblieux 2010) and Triceratops ( Hatcher et al. 1907) . The anterior and distal parts of the paroccipital process are probably formed by the opisthotic ( Romer 1956). The anterior surface of the paroccipital process bears a number of fine transverse ridges and striations. At the distal end of the left paroccipital process a diagonal ridge extending dorsomedially–ventrolaterally is present and this may represent a facet for articulation of the quadrate ( Figs 11 View FIGURE 11 D, 12C).

Ventral to the paroccipital process, a deeply concave fossa is present on each side of the occipital condyle. A second, slightly smaller fossa positioned lateral to this large fossa is preserved on the left side of the condyle; this is one of several asymmetries in the braincase and is probably preservational in origin (Fig. 12A). The larger fossa probably surrounded the exits for the vagus (X), accessory (XI) and hypoglossal nerves (XII), as it does in Triceratops ( Forster 1996) , Agujaceratops ( Lehman 1989) , Pentaceratops ( Lehman 1993) , and Pachyrhinosaurus ( Witmer & Ridgely 2008) , although the actual foramina are not visible due to the presence of matrix. Primitively in dinosaurs, cranial nerves IX to XI exited the braincase via the metotic foramen ( Sampson & Witmer 2007), which represents the remains of the embryonic metotic fissure in reptiles and is located between the otic capsule and the occipital pillar of the embryological chondrocranium, which ossifies to form the opisthotic ( Starck 1979). However, in neotheropods, the vagus canal, which minimally transmitted the vagus and accessory nerves, was diverted posteriorly to open on to the occiput in conjunction with the hypoglossal nerve ( Sampson & Witmer 2007). This derived condition, which is not present in non-ceratopsid neoceratopsians (e.g. Forster 1996; Makovicky 2001) appears to have occurred convergently in ceratopsids ( Forster 1996; Witmer & Ridgely 2008) and thus represents a ceratopsid synapomorphy.

In adult ceratopsids the exoccipitals meet dorsally above the foramen magnum, excluding the supraoccipital from its margin ( Hatcher et al. 1907: fig. 6; Goodwin et al. 2006). In NHMUK R4948 these sutures cannot be seen, and they were not observed in other Chasmosaurus specimens (ROM 839; CMN 2245; UALVP 40; TMP 1981.19.175). The foramen magnum is deeper dorsoventrally (3.4 cm) than it is wide transversely (3.1 cm) and it is angled posterodorsally so that it is not clearly visible in posterior view ( Fig. 11 View FIGURE 11 A). In articulated specimens, this appears to be because the occipital condyle is angled slightly ventrally relative to the horizontal axis of the skull (ROM 839; CMN 2245; Diabloceratops, Kirkland & Deblieux 2010 ). In Triceratops ( Hatcher et al. 1907: fig. 6; Forster 1996) Agujaceratops ( Lehman 1989: fig. 6), Centrosaurus ( Dodson et al. 2004: fig. 23.4), Diabloceratops ( Kirkland & Deblieux 2010) and Chasmosaurus sp. (UALVP 40; TMP 1981.19.175) the ventral margin of the supraoccipital articulates with the exoccipitals, forming a vertical and slightly posteriorly inclined wall. These taxa also possess a prominent ridge separating two fossae that were probably occupied by a venous sinus (Tshihiji 2010). The ridge extends ventrally down the midline of the supraoccipital. The dorsal margin of the occiput is broken in NHMUK R4948 ( Fig. 11 View FIGURE 11 A), but it differs from the aforementioned taxa in being angled anterodorsally and in lacking a prominent midline ridge. In Diabloceratops , the exoccipitals form a thin, caudally-extending shelf which projects over the dorsal margin of the foramen magnum ( Kirkland & Deblieux 2010); this feature is not present in NHMUK R4948.

The occipital condyle is composed of the exoccipitals laterally and the basioccipital ventrally in ceratopsids ( Hatcher et al. 1907: fig. 7; Lehman 1989: fig. 6; Goodwin et al. 2006: fig. 9) but no sutures can be identified in this area in NHMUK R4948. The condyle is larger than the foramen magnum (6.7 cm tall dorsoventrally; 6.5 cm wide transversely), has a circular outline in posterior view, and a short neck that curves down to the basal tubera, narrowing in width slightly as it does so. The basal tubera are separated in posterior view by a shallow notch. The basal tubera are extremely large in comparison with other ornithischians, suggesting enlarged nuchal musculature (m. longissimus capitis, pars transversalis cervicis and m. iliocostalis capitis, Tsuihiji 2010) and are transversely broad plates in posterior view. In lateral view, they expand strongly anteroposteriorly towards their distal ends and project ventrally. In ventral view the distal ends of the tubera have an oval outline, with the long axis of this oval orientated transversely; a small foramen opens between them and comparison with crocodiles suggests this is for the basilar artery (Fig. 12B, Hopson 1979).

In lateral view ( Figs 11 View FIGURE 11 C, D; 12C, D), a large foramen is present at the base of the paroccipital process immediately anterior to the basal tubera. This represents the fenestra ovalis, which would have received the footplate of the stapes, and it may also have transmitted the glossopharangeal nerve (IX; Sampson & Witmer 2007; Witmer & Ridgely 2008). Anterior to the fenestra ovalis another large foramen pierces the braincase and is identified as the trigeminal foramen, transmitting the maxillary (V2) and mandibular (V3) parts of the trigeminal nerve (V). In birds V2 &3 exit the braincase via separate foramina ( Hartwig 1993), but in dinosaurs the foramina are coalesced, as they are in crocodiles ( Hopson 1979; Starck 1979). The trigeminal foramen is located within the prootic and its anterior boundary is formed by the laterosphenoid ( Starck 1979). The fenestra ovalis and trigeminal foramen are separated by a sharp ridge of bone, which is often termed the crista prootica (e.g. Forster 1996). This ridge arises from the posteroventral margin of the trigeminal foramen, extends dorsally to a point level with the dorsal margin of the foramen and then turns posteriorly, to extend dorsal to the fenestra ovalis. The ridge continues posteriorly, extending on to the anterior surface of the paroccipital process, fading out approximately one-third of the way along the process (Fig. 12C, D).

Anterior to the basal tubera the basisphenoid expands ventrally to form the basipterygoid processes, which are separated from each other by a deep fossa (Fig. 12B). The right process is broken, but the left is complete. It is triangular in lateral view with the apex tapering ventrally, but in anterior view it expands slightly transversely. A small foramen positioned ventral to the trigeminal foramen in the concavity formed between the basal tubera and basipterygoid process is present on the left side of the braincase (its presence on the right cannot be established externally due to breakage). A shallow groove extends from it posteriorly towards the basal tubera and it probably represents the entrance for the internal carotid artery ( Sampson & Witmer 2007; Witmer & Ridgely 2008). A similar feature in the same location was also interpreted as the foramen for the internal carotid in Diabloceratops ( Kirkland & Deblieux 2010) .

FIGURE 12. Interpretive drawings of the braincase of NHMUK R4948, Chasmosaurus belli in A, posterior, B, ventral, C, left lateral and D, right lateral view. Abbreviations: II, foramen for exit of optic nerve; III, foramen for exit of oculomotor nerve; IV, foramen for exit of trochlear nerve; V1, foramen for exit of trigeminal nerve, ophthalmic branch; V2 &3, foramen for exit of trigeminal nerve, maxillary and mandibular branches; X–XII, foramen for exit of vagus, accessory and hypoglossal nerves; aob, antotic buttress; ba, foramen for basilar artery, bt, basal tuber; bv, foramen for blood vessel; bpt p, basipterygoid process; crp, crista prootica; dep, preservational depression; fo, fenestra ovalis; ica, internal carotid artery; oc, occipital condyle; poc, paroccipital process; poc-q f, facet for quadrate on paroccipital process; sul, sulcus. Hatching indicates broken bone. Scale bar equal to 10 cm.

In other dinosaurs the foramen for the facial nerve (VII) pierces the braincase in the region between, or slightly ventral to, fenestra ovalis and the trigeminal foramen and a small facial foramen has been identified in all ceratopsians whose braincases have been studied ( Hay 1909; Brown & Schlaikjer 1940; Maryańska & Osmólska 1975; Lehman 1989, 1993, 1996; Forster 1996; Chinnery & Weishampel 1998; Makovicky 2001; Witmer & Ridgely 2008; Kirkland & Deblieux 2010). However, this opening cannot be identified in NHMUK R4948.

Anterior to the trigeminal foramen, the lateral wall of the braincase is pierced by numerous small foramina (Fig. 12C, D). Several are only present on the right side of NHMUK R4948, but this asymmetry is probably preservational. A ridge extending dorsally along the laterosphenoid probably corresponds to the structure termed the antotoic buttress by Godfrey and Holmes (1995) and the laterosphenoid buttress by Kirkland & Deblieux (2010). It is pierced ventrally by a foramen opening anteriorly that probably represents the exit for the ophthalmic branch of the trigeminal nerve (V1). A similar structure in the same location was identified as V 1 in Diabloceratops ( Kirkland & Deblieux 2010) . Ventromedial to the V1 foramen a second small foramen is present. Previous workers have identified this as the opening for the oculomotor nerve in other ceratopsian braincases (III; Brown & Schlaiker 1940; Lehman 1989, 1996; Makovicky 2001). These foramina mark the anterior extent of the basisphenoid and its suture with the orbitosphenoid above and parasphenoid below ( Norman 1986). Further anteriorly, three small foramina lie adjacent to each other. Previous work has suggested that the foramen anterodorsal to the exit of cranial nerve III is the exit for the trochlear nerve (IV; Hay 1909; Brown & Schlaikjer 1940; Lehman 1989, 1996; Forster 1996; Makovicky 2001; Dodson et al. 2004) while the opening anterior to that for cranial nerve III (which also lies closest to the midline) is the exit for the optic nerve (II; Hay 1909; Brown & Schlaikjer 1940; Lehman 1989, 1996; Forster 1996; Makovicky 2001; Dodson et al. 2004), which lies on the suture between the parasphenoid below and orbitosphenoid above ( Norman 1986). The dorsalmost foramen in this area is located within the orbitosphenoid and probably represents a blood vessel canal, possibly the orbitocerebral vein ( Witmer & Ridgely 2008). Ventral to these foramina the braincase is broken and has been reconstructed with plaster, potentially filling the space that would have marked the exit for the olfactory tracts. The interpretation of these cranial foramina differs slightly from the interpretation of the braincase of the centrosaurine Diabloceratops by Kirkland and Deblieux (2010), in which the foramen for cranial nerve III lies anterior and dorsal to V1, and corresponds to the foramen here interpreted as the exit for cranial nerve II. The opening for cranial nerve II in Diabloceratops is interpreted as the opening for IV, while the opening for IV in Diabloceratops is interpreted as a blood vessel pit herein. Our interpretation of braincase foramina is more similar to other work on ceratopsian braincase anatomy ( Hay 1909; Brown & Schlaikjer 1940; Makovicky 2001; Witmer & Ridgely 2008).

In dorsal view ( Fig. 11 View FIGURE 11 A), the supraoccipital and posterior part of the frontals have been broken, but anteriorly there is a large concave fossa present between the laterosphenoids, presumably lying within the remnants of the broken frontals, and this probably represents part of the supracranial sinus, a feature present in many ceratopsids ( Farke 2006; 2010).

Lower jaw. Predentary. The predentary is arrow-head shaped in dorsal view. Its dorsal surface is deeply concave, and the lateral margins (triturating surfaces) are transversely broad, orientated horizontally and flat to slightly concave, joining anteriorly in a point that is curved dorsally in lateral view ( Fig. 13 View FIGURE 13 ). Posteroventrally the predentary is also drawn into a midline point, and in posterior view two deeply concave facets extend from the posteroventral midline point to the dorsal surface for articulation with the dentaries. Ventrally the surface of the predentary is convex, pitted and grooved, suggestive of a horny covering in life. It is extremely similar to the predentaries of Triceratops and Agujaceratops ( Hatcher et al. 1907; Lehman 1989) in all respects. A distinct groove extends anteroposteriorly down the midline ventrally.

Posterior lower jaw elements. The posterior elements of the lower jaw, including the surangular, articular and prearticular are preserved on both sides and are fused together; sutures are visible in places ( Figs 14 View FIGURE 14 , 15 View FIGURE 15 ). Additionally the coronoid process of the left dentary is present ( Fig. 14 View FIGURE 14 K, L). Godfrey and Holmes (1995) suggested that coossification of the posterior jaw elements never occurred in Chasmosaurus , however, this is clearly not the case in NHMUK R4948, where these elements are in articulation, have not been displaced and are fused together. In lateral view ( Figs 14 View FIGURE 14 B, 15B) the surangular is rectangular, with the long axis angled obliquely anterodorsally; it is similar to that of Triceratops ( Hatcher et al. 1907) . The surface is rugose and posterolaterally it bears striations and small lateral tumescenses. The angular would have been located ventral to the surangular, as in Triceratops ( Hatcher et al. 1907) and ventrally, the surangular bears a deeply concave facet into which the angular would have articulated, wrapping underneath the posterior lower jaw ( Fig. 14 View FIGURE 14 B, 15B). Anterolaterally there is an oval facet for articulation with the dentary ( Fig. 14 View FIGURE 14 B), also observed in CMN 2245 ( C. belli ). This differs from the condition in Vagaceratops, where the dentary appears to articulate along a blunt, anteriorly-facing facet (CMN 41357). The anterior parts of both surangulars are broken. On the right a large foramen extends posteriorly from the margin of the adductor fossa into the body of the surangular; this area is damaged on the left side and infilled with plaster. In posterior view ( Figs 14 View FIGURE 14 F, 15J), the surangular extends medially as a dorsoventally thin process to articulate with the prearticular, as in Vagaceratops (CMN 41357). Two small foramina are present posterolaterally; foramina in similar locations were illustrated in Triceratops by Hatcher et al. (1907) and are present in Vagaceratops (CMN 41357).

The prearticular forms the posteromedial and medial portions of the posterior part of the lower jaw ( Figs 14 View FIGURE 14 D, 15D). Posteromedially the prearticular contacts the surangular, and in ventral view ( Figs 14 View FIGURE 14 J, 15H) a suture can be traced extending anteriorly along the ventromedial surface. A triangular wedge of the articular is visible between the prearticular and surangular in ventral view. Posterodorsally the prearticular is deeply concave, where it forms the posteromedial part of the glenoid for articulation with the quadrate. Anteriorly the cup-like articular surface gives way to a transversely compressed, dorsoventrally thin lamina that extends forwards and is broken on both sides.

The articular forms approximately half of the articular surface for the quadrate and is reniform in dorsal view ( Figs 14 View FIGURE 14 H, 15F). It is concave medially, where it overlies the prearticular, while laterally it becomes convex, overlying the surangular. The articular appears to extend ventrally between the surangular and prearticular, although it is excluded from the posterior surface of the lower jaw by contact of these elements. In ventral view the articular forms a triangular wedge between the surangular and prearticular, so that the apex of the triangle points posteriorly, and the base of the triangle forms the smooth, slightly concave surface of the adductor fossa.

Axial skeleton. Although the presacral axial skeleton is preserved in three specimens of Chasmosaurus (ROM 843, CMN 2245, CMN 2280), it has not previously been described in detail (although it was briefly described in Mallon & Holmes 2006). The cervicals and dorsals of NHMUK R4948 are the least distorted of any Chasmosaurus specimen and are uncrushed. Dimensions of the cervical and dorsal vertebrae are given in Table 2.

Cervical vertebrae. The first three cervical vertebrae are fused together to form the cervical bar or syncervical, a feature that constitutes a neoceratopsian synapomorphy ( Dodson et al. 2004; Campione & Holmes 2006; Tsuihiji and Makovicky 2007). In NHMUK R4948 the atlas, axis and cervical three are clearly differentiated from each other by rugose sutures ( Fig. 16 View FIGURE 16 ). The atlas is longer than wide and rounded in transverse section, with a deeply concave anterior facet for the occipital condyle, as in Triceratops ( Hatcher et al. 1907) and Centrosaurus (ROM 767). The anterior rim is smooth, although it is finished with plaster on the left side. The lateral sides of the atlas are smooth except where they rise to rugose tumescences about half way along the length of the element: these are rounded, situated dorsally on the intercentrum and represent the articular facet for the single-headed atlas rib, which is unpreserved. The suture with the axis is expanded transversely and rugose laterally; there is no evidence of the atlas centrum (odontoid process) and it probably fused to the axis early in ontogeny. The atlantal neural arches are broken on both sides.

The axis forms the second fused element in the cervical bar. It is similar in length anteroposteriorly to the atlas and the lateral sides of the centrum are smooth. Ventrally a very slight ridge is present extending anteroposteriorly, a feature also observed in ROM 843 ( Chasmosaurus belli ). In left lateral view a small parapophysis projects laterally from the anterior part of the lateral centrum, as in Triceratops ( Hatcher et al. 1907) ; on the right side this area is broken. Posteriorly the centrum expands transversely along the suture with cervical three and is rugose. As in all ceratopsians, the neural arch of the axis is very large ( Campione & Holmes 2006; Tsuihiji & Mackovicky 2007). It is broken along the anterior edge, and comprises two transversely thin sheets of bone that converge above the neural canal to form an elongate, posteriorly inclined neural spine that is fused to the anterior surface of the neural arch of cervical three, forming an elongate plate that presumably served as an anchorage point for some of the musculature necessary for the support of the large ceratopsid head ( Hatcher et al. 1907; Romer 1956; Campione & Holmes 2006; Tsuihiji 2010). The end of the neural spine is rounded dorsally and a cleft separates it from the dorsal end of the spine of cervical three. Posteroventrally a foramen separates the axis neural arch from the neural arch pedicle of cervical three; this is broken around its edges. A foramen in present in the same location in ROM 843, CMN 2280 ( Campione & Holmes 2006: fig. 3B, C), Triceratops ( Hatcher et al. 1907) , Styracosaurus (CMN 344) and Centrosaurus (CMN 344). In left lateral view a small diapophysis projects laterally on a sturdy transverse process from the side of the neural arch; this area is broken on the right side.

The centrum of cervical three, the final element in the ceratopsid cervical bar ( Tsuihiji & Mackovicky 2007), is approximately the same length as the atlas and axis. The posterior centrum facet is flat, the lateral surface is smooth, and ventrally a very slight ridge is present extending anteroposteriorly. A small, rounded parapophysis is located anterodorsally on the lateral surface of the centrum; it is supported by a bulge that continues onto the posterolateral surface of the axis, similar to Triceratops ( Hatcher et al. 1907) . The anterior surface of the neural arch and neural spine of cervical three is coalesced with those of the axis and prezygapophyses, and sutures have been largely obliterated, as in Triceratops ( Hatcher et al. 1907) ; in contrast, the postzygapophyses of cervical three extend posteriorly from just dorsal to the neural canal, face ventrolaterally, and are broadly separated, as in Centrosaurus (ROM 767). The diapophysis, which is smaller than the parapophysis, is located posterodorsal to the latter and situated on a short transverse process that projects ventrally and slightly posteriorly. A foramen pierces the neural arch ventrally at the base of the diapophysis and appears to extend dorsomedially in left lateral view; this feature is not present on the right side because the diapophysis has been covered with plaster in this area. The neural spine is short and stout, being broader transversely than it is anteroposteriorly.

A further five cervical vertebrae are preserved, bringing the total number of cervicals to eight, and based on their articulations with each other, the morphology and sizes of the anterior dorsals, and comparisons with CMN 2245 ( Chasmosaurus belli ), this probably represents the complete cervical column ( Fig. 17 View FIGURE 17 ). However, no quarry map exists and the sequence of the vertebrae assumed herein is based on parapophysis location, height of the neural spines and matching of similarly sized articular surfaces. Cervical vertebrae are distinguished from dorsal vertebrae on the basis of parapophyseal position: cervicals bear a parapophysis located on the centrum or neurocentral suture, whereas in dorsal vertebrae the parapophysis is located on the neural arch.

The centra of cervicals four to eight are amphiplatyan, transversely wider than they are anteroposteriorly long, and laterally bear several small foramina. Foramina are also observed piercing the sides of the centra in ‘ Anchiceratops ’ (CMN 8547). Brown (1917) noted that the cervical centra of Centrosaurus become progressively shorter more posteriorly along the column; in contrast, the lengths of the centra in NHMUK R4948 remain subequal. Cervicals four, five and six bear a flattened ridge ventrally that extends between the anterior and posterior centrum facets; in cervicals seven and eight the ridge is less well defined. The parapophysis of cervicals four to six is located on the anterior half of the lateral centrum and slightly closer to the neurocentral suture than the ventral margin, as in Triceratops ( Hatcher et al. 1907) . This process is rounded, concave, faces posterolaterally and is slightly raised from the lateral surface. Striations extending anterodorsally are present above the parapophysis in cervical four. In cervicals seven and eight the parapophyses are located more dorsally on the centrum than in the other cervicals, and in cervical eight they are oval in shape, with the long axis angled anteroventrally, and are larger than in the other cervicals.

The neural arches of cervicals four to eight exhibit variable preservation. Cervical seven is present as a centrum only; however an isolated partial neural arch and spine that may belong to this cervical is present. The height and angle of the neural spine are congruent with the interpretation that this is the neural arch of cervical seven.

In all cervicals the neural canal is circular or tear-drop shaped in outline and is approximately as wide as it is tall. The neural arch in cervicals five to eight leans slightly anteriorly, a feature also seen in other specimens of Chasmosaurus (CMN 2245, ROM 843). The prezygapophyses are located immediately dorsal to the neural canal and are widely separated, with ovate articular surfaces that face dorsomedially. A ridge extends anteriorly from the anterior margin of the neural spine, lying between the prezgyapophyses, and prominent depressions are present either side of this ridge, as is also the case in Triceratops ( Hatcher et al. 1907: fig. 51). The postzygapophyses are broadly separated in posterior view and have articular surfaces that are oriented ventrolaterally. These surfaces are reniform in outline in cervical four but oval in cervicals six and eight.

In cervical four, the diapophysis, which is preserved completely on the left side, is positioned on a short transverse process that extends laterally and slightly ventrally. In contrast, the fourth cervical of Triceratops has transverse processes that extend dorsolaterally ( Hatcher et al. 1907). Cervicals five and eight of NHMUK R4948, the only other cervicals in which they are preserved, have transverse processes that extend dorsolaterally. By contrast, in Centrosaurus (ROM 767) these processes extend more strongly laterally. In cervical eight of NHMUK R4948 the transverse processes are longer and more robust than they are in cervical five, and bear ovate diapophyses with flat to slightly concave articular surfaces. The base of the transverse process is pierced by a large foramen in cervicals four and six, but this feature is absent in cervicals five and eight; the area is not preserved in cervical seven. A similar foramen is present in cervical five of Chasmosaurus belli (CMN 2245), and in cervicals four and five of Styracosaurus (CMN 344), but no such features were observed in C. russelli (CMN 2280). The neural spine is completely preserved in cervicals four, seven and eight, and it projects dorsally, as it does in CMN 2245, CMN 2280, Triceratops ( Hatcher et al. 1907) , ‘ Anchiceratops ’ (CMN 8547) and Styracosaurus (CMN 344). In cervical four the spine broadens transversely towards its apex and the aforementioned ridge that extends forward between the prezygapophyses extends approximately halfway up the spine before splitting to give way to a flattened anterodorsal surface. In Triceratops , the top of the neural spine of cervical four is also broader transversely than it is anteroposteriorly ( Hatcher et al. 1907). The summit of the neural spine in cervicals seven and eight of NHMUK R4948 is not expanded transversely at the top, but is bifurcated by a groove posteriorly so that it is arrow-head shaped in cross-section with the apex pointing anteriorly.

Dorsal vertebrae. Seven dorsal vertebrae comprising centra and parts of the neural arch are present, along with two additional centra and five isolated partial neural spines ( Figs 18 View FIGURE 18 , 19 View FIGURE 19 ). Four of these appear to be anteriorly positioned and probably represent dorsals one to four based on the height of the neural arch, the angle of the transverse processes and the size of the neural spine relative to that on the last cervical vertebra ( Fig. 18 View FIGURE 18 ). Other dorsals (here termed dorsals five to seven) are more posterior dorsals based on neural spine and zygapophysis morphology, although it is not possible to determine which vertebrae they represent ( Fig. 19 View FIGURE 19 ). This suggests that the middle dorsals were not preserved: comparisons with other Chasmosaurus specimens suggest that around four dorsal vertebrae are missing (CMN 2245; ROM 843).

The centra of all dorsals are amphiplatyan, wider than long, and the lateral sides are smooth and generally lack foramina, except for dorsal four, which bears two small nutrient foramina on its right side. The lateral sides are not deeply excavated, in contrast to those of Centrosaurus (ROM 1426; Lull 1933). Ventrally, a flattened ridge extends from the anterior to the posterior articular surface in dorsal one. In dorsals two to seven the anterior articular surface is slightly larger than the posterior surface and both articular surfaces extend ventrally under the centrum as rugosities, producing strong, thickened rims to the centra, which increasingly approach each other in the more posterior regions of the dorsal column, a feature that also occurs in Agujaceratops ( Lehman 1989: fig.13), but is not apparent in Centrosaurus ( Lull 1933) . The neurocentral suture, obliterated on the cervicals and dorsals one and two, is clear on dorsals three to seven, suggesting fusion of the neural arches to the centra may have proceeded posteriorly during ontogeny, and that the animal was not fully osteologically mature at the time of death ( Irmis 2007).

The neural canal of all dorsals is rounded, except for that of dorsal one, which is oval, being taller than wide. The prezygapophyses of dorsals one to four face dorsomedially and their bases are separated from each other by a groove that extends to the top of the neural canal. This is similar to the condition in Triceratops ( Hatcher et al. 1907: fig. 52), but contrasts with that in Centrosaurus ( Lull 1933) and Styracosaurus (CMN 344), in which the bases of the prezygapophyses meet ventrally. This groove becomes more prominent in the posterior dorsals, in which the prezygapophyseal articular surfaces rotate to face entirely dorsally. Accordingly, the postzygapophyses face ventrolaterally in dorsals one to four, but entirely ventrally in dorsals five to seven, a feature also present in Agujaceratops ( Lehman 1989) , ‘ Anchiceratops’ ( Mallon and Holmes 2010) and Triceratops ( Hatcher et al. 1907: fig. 52). A ridge extends ventrally from the postzygapophyses, along the midline to the top of the neural canal. The zygapophyses of Chasmosaurus belli (ROM 843) are rather different from those of NHMUK R4948: they are fused to each other ventrally forming a continuous, concave (prezygapophyses) or convex (postzygapophyses) surface for articulation with the preceding or following vertebra, as in Centrosaurus ( Lull 1933) . It is possible that this feature is subject to ontogenetic variation, and this warrants further investigation. Additionally, in all of the dorsals of the latter two taxa, regardless of their position in the axial column, the prezygapophyses face dorsomedially, while the postzygapophyses face ventrolaterally.

In dorsal one the parapophysis, an oval, concave facet, is located below the base of the transverse process, level with the neural canal. In dorsals two to four, the parapophysis is located more dorsally, and is present as a rounded, deeply concave facet situated on a robust, laterally projecting stalk that arises from the transverse process approximately level with the prezygapophyses. In dorsal six, the only posterior dorsal in which the parapophysis is preserved, it is a smaller, circular facet located on the anteroventral transverse process below the diapophysis and dorsal to the prezygapophysis. In several dorsals of ROM 843, the capitulum of the rib is fused to the parapophysis; this may be a pathological or ontogenetic feature and is not seen in NHMUK R4948.

In NHMUK R4948 the diapophysis is located at the distal end of the dorsolaterally projecting transverse process; in dorsals one to three the cross section of the transverse process is shaped like an inverted ‘L’. The diapophysis is expanded anteroposteriorly, rounded, and angles ventrolaterally, as in Triceratops ( Hatcher et al. 1907) . In dorsal four and dorsal six, the only posterior dorsal in which the transverse process is preserved, the transverse process is ‘T’ shaped in cross section, an anterior ridge having been developed. The diapophysis is not as expanded in dorsal four as it is in dorsals one to three; in the posterior dorsals it is not preserved. The neural spine is preserved completely only in dorsal four; however, the base of the spine is preserved in dorsals three, five and six. In dorsal four the neural spine is blade-shaped, strongly transversely compressed, tapers dorsally and is oriented slightly posteriorly. In the posterior dorsals the neural spine is much broader anteroposteriorly (as in Triceratops [ Hatcher et al. 1907], ‘ Anchiceratops ’ [CMN 8547] and Centrosaurus [ROM 767]) although it remains compressed transversely, and appears to have projected more strongly dorsally. Mallon and Holmes (2006) suggested that the neural spines of the anterior dorsals of CMN 2245 ( C. belli ) projected more strongly posteriorly than the posterior dorsal neural spines, a feature not observed in CMN 2280 ( C. russelli ). Our observations of NHMUK R4948 support the suggestion that the anterior neural spines projected more posteriorly than the posterior neural spines in C. belli , and this may prove to be a specific difference, although more specimens are needed to confirm this.

Sacral vertebrae. The centra of six vertebrae are fused together to form the sacral rod, which appears to comprise two dorsosacrals and four sacrals ( Fig. 20 View FIGURE 20 ; dorsosacrals are those that bear ribs similar in morphology to dorsal ribs, while the four true sacrals, homologous to the four sacrals of basal ornithischians, bear sacral ribs: see below). In contrast, nine vertebrae are fused to form the sacral rod in CMN 2245 (one dorsosacral, four sacrals and four caudosacrals) and eleven are fused in ROM 843 (three dorsosacrals, four sacrals and four caudosacrals). Lehman (1989) reported ten vertebrae in the sacral rod in Agujaceratops (three dorsosacrals, four sacrals and three caudosacrals), and there are also ten in Triceratops ( Hatcher et al. 1907: two dorsosacrals, four sacrals and four caudosacrals). In NHMUK R4948, the anteriormost vertebra fused to the sacral rod, the second dorsosacral, bears prezygapophyses that face dorsally and are identical in morphology to those of the posterior dorsals. The dorsosacral vertebrae bear laterally-projecting transverse processes that are comparable in anteroposterior width to those of the dorsals. No evidence for parapophyses is visible on the lateral sides of the vertebrae, suggesting that they either bore single-headed ribs or more likely that the transverse processes contacted the preacetabular process of the ilium. Their neural spines are strongly inclined posteriorly, in contrast to those of Triceratops , which extend dorsally ( Hatcher et al. 1907: fig. 54), and are not fused to each other or to those of the other vertebrae in the sacral rod. The posterior four vertebrae in the sacral rod bear sacral ribs and are probably homologous to the four sacral vertebrae present in primitive ornithischians ( Butler et al. 2008). Foramina pierce the fused neural arches between the third and fourth vertebrae fused into the rod (sacrals one and two) and the fourth and fifth vertebrae fused into the rod (sacrals two and three). Ventrally, all of the vertebrae in the rod bear a deep groove that extends along the centrum anteroposteriorly, a feature present in chasmosaurines but developed to a much lesser extent in centrosaurines ( Lehman 1989). The neural spines of vertebrae three to six (sacrals one to four) are coalesced into a transversely compressed sheet and sutures between them have been obliterated. No ossified tendons are preserved. The posterior centrum facet of the final vertebra in the rod, sacral four, is rugose and unfinished and suggests an additional vertebra, a caudosacral, may have contributed to the sacral rod but was not fused at the time of death. Since other specimens of Chasmosaurus (CMN 2245; ROM 843) bear four caudosacrals, it is likely that several additional vertebrae would have fused into the sacral rod during ontogeny, and the unfused state of these vertebrae provides further evidence that the animal was not osteologically mature at the time of death. The sacral ribs of the third, fourth and fifth vertebrae in the sacral rod (sacrals one to three) are dorsoventrally deep and anteroposteriorly compressed, and slightly hour-glass shaped in cross section. There are two distinct articular surfaces, one dorsal and one ventral, homologous with the diapophysis and parapophysis of dorsal vertebrae, respectively. Both of these articular surfaces would have contacted the medial surface of the ilium. The ventral articular surfaces of sacrals one to three are coalesced into one anteroposteriorly elongate surface, while the dorsal articular surfaces remain distinct from each other and do not fuse, as in Triceratops ( Hatcher et al. 1907: fig. 54) and Pentaceratops ( Wiman 1930: pl.5). The rib of the posteriormost vertebra fused to the sacral rod, sacral four, does not bear distinct dorsal and ventral articular surfaces for the ilium; instead these coalesce and are joined to the ventral articular surfaces of the other sacral ribs, resulting in the rib angling anterodorsally, a feature also observed in Triceratops and Pentaceratops .

Ribs. Many dorsal rib fragments are preserved, but these are broken into small pieces and no complete rib remains. Anterior dorsal ribs are ‘Y’-shaped, with the tuberculum and capitulum held on elongate processes at right angles to each other. The tuberculum is anteroposteriorly expanded and oval in outline, while the capitulum, which sits on the longer of the two branches of the ‘Y’, is not expanded relative to the shaft and is also oval in outline. Proximally, the rib shaft is anteroposteriorly compressed and oval in cross-section; more distally it becomes triangular in cross-section with the apex pointing anteriorly. The process bearing the tuberculum in the more posterior dorsal ribs is much shorter proximodistally than it is in the anterior ribs, while the process for the capitulum is much longer. The tuberculum is not significantly expanded anteroposteriorly, whereas the tuberculum is larger than in anterior ribs. The proximal shaft is transversely compressed and its anterior surface is concave so that it is ‘C’ shaped in cross-section. More distally, the shaft is triangular in cross-section but the apex points posteriorly, in contrast to the situation in the anterior dorsal ribs ( Fig. 21 View FIGURE 21 ).

Appendicular skeleton. Elements of the pectoral girdle, forelimb, pelvic girdle and hind limb are well preserved in NHMUK R4948; unfortunately no manus or pes material was recovered. Dimensions of the appendicular skeleton are given in Table 3 View TABLE 3 .

Coracoid. Both coracoids are well preserved ( Fig. 22 View FIGURE 22 A–D). They were not fused to the scapula, in contrast with the situation in other specimens of Chasmosaurus (ROM 843; ROM 839; CMN 2245; CMN 2280) and most other ceratopsids ( Dodson et al. 2004), probably because the specimen was not osteologically mature at the time of death. The coracoid is gently convex laterally, gently concave medially, and the bone surface is smooth. Proximal to the sutural surface for the scapula the dorsal margin of the coracoid is rugose and thickened, representing the inferred origin of the m. supracoracoideus (Maidment and Barrett unpublished data); a similar roughened area was observed in Styracosaurus (CMN 344) and an unnamed pachyrhinosaur (TMP 2002.76.01). The coracoid foramen is fully enclosed on the lateral surface; medially it opens onto the sutural surface for the scapula. The glenoid is deeply concave and faces posteroventrally. A prominent sternal process projects from the anteroventral surface well below the level of the glenoid, as in ROM 843, CMN 2280 and other ceratopsids ( Dodson et al. 2004).

Scapula. Only a small portion of the proximal plate and the distal end of the left scapula are preserved ( Fig. 22 View FIGURE 22 E–G). The sutural surface for the coracoid is rugose, unfinished and transversely compressed dorsally; it thickens ventrally close to the glenoid. The dorsal margin of the scapula is drawn into a small, poorly defined acromial process that rises gently anteriorly as in all ceratopsids ( Triceratops , NHMUK R3886-8, Hatcher et al. 1907: fig. 64; ‘ Anchiceratops’, CMN 8547; Styracosaurus , CMN 344; Centrosaurus , ROM 767, 1426; pachyrhinosaur sp.

indet., TMP 2002.76.01; in the centrosaurines examined the acromial process appears to be shorter, occupies less of the scapula shaft, and is less well developed than in the chasmosaurines examined). The dorsal margin of the acromial process is folded over laterally to form a flattened apex in lateral view just posterior to the coracoid suture for inferred attachment of the m. deltoideus clavicularis (Maidment & Barrett unpublished data), as in other specimens of Chasmosaurus (CMN 2280, ROM 839), Styracosaurus (CMN 344), Triceratops ( Hatcher et al. 1907: fig. 64) and an unnamed pachyrhinosaur (TMP 2002.76.01). The glenoid is thickened transversely relative to the rest of the proximal plate. It is pentagonal in cross-section with the apex pointing ventrally, and it faces anteroventrally. Posterior to the glenoid on the ventral surface are rugosities and a shallow depression, probably indicating the origin of the m. triceps longus (Maidment and Barrett, unpublished data). Similar features are seen in the same location in other specimens of Chasmosaurus (ROM 839; ROM 843, CMN 2245; CMN 2280) and in other ceratopsids ( Triceratops NHMUK R3886-8, Hatcher et al. 1907; ‘Anchiceratops’, CMN 8547; Styracosaurus , CMN 344; Centrosaurus , ROM 767, 1426; pachyrhinosaur sp. indet., TMP 2002.76.01). In medial view a shallow canal extends from the coracoid suture posteriorly and would have joined the coracoid foramen when the elements were in articulation, as in CMN 2245. Posterior to this a small foramen pierces the medial surface. The small portion of the distal end of the scapula suggests that the scapula blade was not significantly flared posteriorly, as is the case in other specimens (CMN 2280; ROM 839) and other ceratopsids ( Triceratops , NHMUK R3886-8; ‘Anchiceratops’, CMN 8547; Styracosaurus , CMN 344; Centrosaurus ROM 767). It is rugose and unfinished, suggestive of a cartilaginous suprascapula.

Humerus. Both humeri are uncrushed and quite well preserved ( Fig. 23 View FIGURE 23 ). In anterior view, the proximal end of the humerus is rectangular in outline. The medial tuberosity is distinguished from the dorsal surface of the element by a notch, which is also present in other well preserved Chasmosaurus specimens (ROM 843; ROM 839) and developed to a greater degree in Triceratops ( Hatcher et al. 1907: fig. 65). The deltopectoral crest is small and projects anterolaterally. Its anterolateral surface is incomplete on both humeri, but appears to be straight in lateral view, as in other chasmosaurines (e.g. Chasmosaurus , CMN 2280; CMN 2245; ROM 839; ‘Anchiceratops’, CMN 8547, Triceratops, Hatcher et al. 1907 : fig. 66) but in contrast to the condition in centrosaurines (e.g. Styracosaurus , CMN 344; Centrosaurus , ROM 767; pachyrhinosaur sp. indet., TMP 2002.76.01) in which it is concave upwards in lateral view. The apex, which points anteriorly, is well preserved on both sides in NHMUK R4948 and it forms a flattened, dorsoventrally elongate surface that probably formed the insertion of the m. pectoralis (Maidment & Barrett unpublished data). It is similar in morphology to that of other specimens of Chasmosaurus (ROM 843; ROM 839), Styracosaurus (CMN 344), Centrosaurus (ROM 1426) and an unnamed pachyrhinosaur (TMP 2002.76.01).

The apex of the deltopectoral crest reaches approximately midlength, and below it the shaft constricts rapidly before flaring for the distal condyles, which are rugose and unfinished in appearance, suggestive of a thick cartilaginous cap. The humeral head is restricted to the posterior surface, as is the case in all quadrupedal ornithischians (SCRM pers. obs., 2009–2010). It is distinct, rounded, and rugose and projects posteriorly. The posterior surface of the medial tuberosity bears some vertical striations. The posterolateral surface of the deltopectoral crest is strongly striated for the inferred insertion of the deltoid musculature, a feature seen in many ceratopsids (e.g. ‘ Anchiceratops ’, CMN 8547; Styracosaurus , CMN 344; Centrosaurus , ROM 767, 1426; pachyrhinosaur sp. indet., TMP 2002.76.01; Vagaceratops, CMN 41357; Triceratops, Hatcher et al. 1907 ) and other specimens of Chasmosaurus (CMN 2280, CMN 2245; ROM 839; ROM 843); the area of insertion of the m. pectoralis on the apex of the deltopectoral crest is clearly distinguished from the insertion of the deltoids by a deep groove, also observed on ROM 839, CMN 2280 and Triceratops ( Hatcher et al. 1907: fig. 66) but not on the more poorly preserved humeri of ROM 843. The posterior surface of the humerus lateral to the head bears a deep fossa, a feature observed on other ceratopsid humeri ( Triceratops , NHMUK R3886-8, Hatcher et al. 1907: fig. 66; ‘Anchiceratops’, CMN 8547; Styracosaurus , CMN 344; Centrosaurus , ROM 767; pachyrhinosaur sp. indet., TMP 2002.76.01; Vagaceratops, CMN 41357). The fossa probably represents the origin of m. triceps brevis (Maidment & Barrett unpublished data; contra Lehman 1989) and is much better developed in chasmosaurines than in centrosaurines. In many ceratopsid humeri, a foramen pierces the posteromedial humerus on the proximal half of the element (e.g. CMN 2280; CMN 2245; ROM 843; Styracosaurus , CMN 344; Centrosaurus , ROM 767; pachyrhinosaur sp. indet., TMP 2002.76.01; Vagaceratops, CMN 41357, Triceratops, Hatcher et al. 1907 : fig. 66); this may correspond to the insertion of the m. latissimus dorsi. In NHMUK R4948 the posterior shaft is patched with plaster on both sides and the foramen cannot be seen. In chasmosaurines (e.g. Chasmosaurus , CMN 2280, 2245; ROM 843, Triceratops, Hatcher et al. 1907 : fig. 66) the foramen is dorsal to the apex of the deltopectoral crest, whereas in centrosaurines ( Styracosaurus , CMN 344; Centrosaurus , ROM 767; pachyrhinosaur sp. indet., TMP 2002.76.01) it is located approximately level with the apex and more medially.

Ulna. Both ulnae are present ( Fig. 24 View FIGURE 24 A–L), although the olecranon process is eroded on the right ulna, and part of the shaft is reconstructed. On the left ulna, the olecranon process is large, rounded in cross-section and tapers dorsally, as in ROM 843 and other ceratopsids (e.g. ‘Anchiceratops’, CMN 8547; Centrosaurus , ROM 1426; Styracosaurus , CMN 344; pachyrhinosaur sp. indet., TMP 2002.76.01; Vagaceratops, CMN 41357; Agujaceratops, Lehman, 1989 ; Triceratops, Hatcher et al. 1907 ). A well-developed anterolateral process is present, as in all quadrupedal ornithischians (SCRM pers. obs., 2009–2010). It forms a right angle to the medial process, which is larger and has a flattened anterior surface against which the posterior surface of the proximal radius articulated, comparing favourably with ROM 843 and ROM 839. The distal end is not expanded relative to the shaft, and it is ringed in vertical striations, which are also observed in ‘Anchiceratops’ (CMN 8547), Styracosaurus (CMN 344) and Centrosaurus (ROM 767, 1426). A large roughened scar on the anteromedial surface of the distal end marks the articular facet for the distal end of the radius, as in ROM 839 and other ceratopsids (e.g. ‘Anchiceratops’, CMN 8547; Centrosaurus , ROM 767, 1426; Styracosaurus , CMN 344; Triceratops, Hatcher et al. 1907 : fig. 67). The presence of a large anterolateral process of the ulna results in the radius being located anteromedial to the ulna rather than wholly anteriorly, as it is in basal, bipedal ornithischians, and this makes pronation of the manus possible ( Bonnan 2003).

Radius. Only the proximal and distal ends of the right radius are preserved ( Fig 24 View FIGURE 24 M–P). The proximal end, which was expanded transversely and anteroposteriorly relative to the shaft, is flat dorsally. It is oval in cross-section with the long axis oriented transversely, as in Chasmosaurus belli (ROM 843), Chasmosaurus sp. (ROM 839) and other ceratopsids ( Centrosaurus ; ROM 767, 1426; Styracosaurus , CMN 344; pachyrhinosaur sp. indet., TMP 2002.76.01; Agujaceratops, Lehman, 1989 ); the posterior surface is flattened for articulation with the medial process of the ulna. The distal end is also expanded relative to the shaft. It is expanded more medially than it is laterally in anterior view, and it extends further ventrally on the medial side than it does on the lateral side. A small boss is present on the posterolateral surface for articulation with the distal ulna, a feature also seen in ROM 843, ROM 839 and other ceratopsids (e.g. ‘Anchiceratops’, CMN 8547; Centrosaurus , ROM 1426; Styracosaurus , CMN 344; pachyrhinosaur sp. indet., TMP 2002.76.01); although the two elements articulated they were clearly not tightly associated with each other and rotation of the distal end of the radius around the ulna was probably possible, allowing some degree of pronatory and supinatory rotation at the wrist.

Pubis. A partial right pubis is the only pelvic element preserved ( Fig. 22 View FIGURE 22 H–K). The acetabular region of the pubis is broader transversely than it is deep dorsoventrally. The acetabulum is cup-shaped, concave and faces dorsolaterally. The anteromedial corner of the acetabular rim is triangular in shape, rugose and unfinished, and represents the articular surface for the pubic peduncle of the ilium. Posteriorly the acetabular rim angles posteriorly and is rugose for articulation with the ischium; these features are also observed in ROM 843, the only other Chasmosaurus specimen to preserve this portion of the pubis. The postpubis has almost completely broken off and is not preserved, except for a small portion proximal to the acetabular region of the pubis. It is flattened dorsoventrally and slightly concave dorsally, as in Vagaceratops (CMN 41357) and Triceratops ( Hatcher et al. 1907) . The prepubis is elongate, dorsoventrally flattened and transversely as broad as the acetabular part of the pubis, as in other ceratopsids ( Triceratops , NHMUK R3886-8, Hatcher et al. 1907; pachyrhinosaur sp. indet., TMP 2002.76.01; Vagaceratops, CMN 41357; Pentaceratops, Wiman 1930 : pl. 6). The anterior margin of the prepubis is broken. It is greatly expanded dorsoventrally relative to the rest of the prepubis, and is also expanded medially, so that the medial surface of the prepubis is concave upward when viewed from above at the anterior end, a feature observed in other specimens of Chasmosaurus (CMN 2245; ROM 843) and other ceratopsids ( Triceratops , NHMUK R3886- 8, Hatcher et al. 1907; Vagaceratops, CMN 41357, Pentaceratops, Wiman 1930 ; pl. 6). This medial expansion may have contacted the posterior dorsal ribs ( Sternberg 1927; Lehman 1989).

Femur. The right femur is preserved but it is anteroposteriorly crushed and the medullary cavity has collapsed ( Fig. 25 View FIGURE 25 ). The femoral head is separated from both the shaft and greater trochanter by a distinct neck, as in all neornithischians. It is slightly flattened posteriorly and bears several ventrally-trending striations. The head is separated from the greater trochanter by a saddle-shaped depression. The greater trochanter projects dorsally to approximately the same level as the head, a feature common to most neornithischians. Laterally, the greater trochanter is flattened, broad and slightly striated along its dorsal margin. The anterior trochanter appears to be fused to the anterior side of the greater trochanter, as in ROM 843 and all ceratopsids ( Dodson et al. 2004), although the cleft between the two in lateral view is filled with plaster. The anterior trochanter can be clearly distinguished from the greater trochanter because it projects further laterally and does not project as far dorsally, as in Triceratops ( Hatcher et al. 1907: pl. 14). The shaft is poorly preserved and crushed. The fourth trochanter is present as an obliquely-trending ridge in posterior view, as in ROM 843 and other ceratopsids ( Styracosaurus , CMN 344; pachyrhinosaur sp. indet., TMP 2002.76.01; Triceratops, Hatcher et al. 1907 : pl. 14); its surface is flat to concave. Medially it is fringed by a series of small horizontal striations, probably for the m. caudofemoralis longus (Maidment and Barrett unpublished data). In posterior view, the medial distal condyle is larger than the lateral one; the lateral condyle is divided into a posterior and a lateral surface, separated by a groove in lateral view, as in all neornithischians.

Tibia. The right tibia is present, but the shaft is crushed and the distal end is poorly preserved ( Fig. 26 View FIGURE 26 A–D). Proximally, the tibia is in good condition and broader anteroposteriorly than it is transversely. The cnemial crest is transversely compressed. It projects further dorsally than any other part of the proximal tibia, is slightly concave laterally and flat medially, and is similar in morphology to ROM 839 Chasmosaurus sp. (ROM 839), C. belli (ROM 843) and Styracosaurus (CMN 344). The proximodorsal part is covered in plaster. The rounded fibula condyle projects strongly laterally and is rugose: many grooves and canals cross this surface and are suggestive of a thick cartilaginous cap; it is similar in morphology to that of C. belli (ROM 843) and Triceratops ( Hatcher et al. 1907: pl. 16). Posterior to the fibula condyle the proximal end is transversely broad and flattened dorsally. The bone surface is rugose and unfinished in this area. The distal end is also broader anteroposteriorly than it is transversely, in contrast to the condition in other ornithischians. Since the distal end is very crushed and poorly preserved, it is likely that it has rotated around to lie in this orientation as a result of post-mortem deformation. The posterior surface of the distal end now faces laterally. It is crushed but gently convex. The anterior surface, which now faces medially, bears two distinct surfaces separated by a groove, as in Triceratops ( Hatcher et al. 1907: pl. 16). The fibula would have articulated with the lateralmost of these surfaces, which projects further ventrally than the medial surface, and is slightly striated (as in ROM 843). The calcaneum would have articulated ventrally, and the surface for it is convex. The facet for the astragalus, in contrast, is flattened.

Fibula. The right fibula is slender and elongate ( Fig. 26 View FIGURE 26 E–F). Its proximal end is transversely compressed while the distal end is anteroposteriorly compressed, as in Centrosaurus ( Brown 1917; Lull 1933). The distal end is expanded more than the proximal end, which is slightly broken anteriorly. The shaft is flattened on its medial surface proximally and it bears a ridge on its lateral surface about one third of the way from the proximal end that probably represents the insertion of m. iliofibularis ( Dilkes 2000). About half way down the shaft, the posterior surface bears a rugose, lumpy muscle scar, probably for the m. peroneus brevis ( Dilkes 2000).

TABLE 1. Skull, braincase and lower jaw measurements, NHMUK R 4948. Numbers in brackets in the second column correspond to measurements in Godfrey and Holmes (1995) for easy comparison.

Element Rostral Measurement taken Dorsoventral height Dimension (cm) 22
Rostral Left premaxilla Max. anteroposterior length Anteroposterior length, as preserved 15 32
Right premaxilla Dorsoventral height 28.5
Left maxilla Right maxilla Nasal horncore Anteroposterior length, as preserved Anteroposterior length, as preserved Height (5) 37 36.5 15.5
Nasal horncore Nasal horncore Right jugal Right jugal Right epijugal Anteroposterior length at base Circumference at base Orbital margin to jugal posteroventral tip (9) Orbital margin to lateral temporal fenestra margin (10) Height 14 35.5 28.5 12 5
Right epijugal Right supraorbital horncore Right supraorbital horncore Orbit Orbit Right quadratojugal Anteroposterior length at base Anteroposterior length at base (2) Height (1) Dorsoventral height (13) Anteroposterior length (11) Dorsoventral height 6.5 10 5.5 13.5 10 17.5
Left quadratojugal Dorsoventral height 17.5
Right quadrate Left quadrate Right squamosal Right squamosal Right squamosal Parietal Dorsoventral height Dorsoventral height Dorsoventral height to jugal notch (14) Max. dorsoventral height (15) Anteroposterior length, as preserved Median bar length, as preserved 29.5 29.5 22 36 90 86
Parietal Foramen magnum Parietal fenestra transverse width (25) Transverse width 49 3.1
Foramen magnum Occipital condyle Dorsoventral height Transverse width 3.4 6.5
Occipital condyle Right lower jaw Dorsoventral height Transverse width, posterior view 6.7 14
Left lower jaw Transverse width, posterior view 14

TABLE 3. Postcranial measurements, NHMUK R 4948.

Element Measurement taken Dimension (cm)
Left coracoid Dorsoventral height 32.0
Right coracoid Dorsoventral height 33.5
Left humerus Length 59.0
Left humerus Midshaft circumference 25.5
Right humerus Length 59.5
Right humerus Midshaft circumference 25.0
Left ulna Length 44.5
Right ulna Length 46.0
Right femur Length 78.0
Right femur Midshaft circumference 34.5
Right tibia Length 58.5
CMN

Canadian Museum of Nature

AMNH

American Museum of Natural History

TMP

Transvaal Museum

Kingdom

Animalia

Phylum

Chordata

Class

Reptilia

Order

Ornithischia

Family

Ceratopsidae

Genus

Chasmosaurus

Loc

Chasmosaurus russelli Sternberg, 1940

Maidment, Susannah C. R. & Barrett, Paul M. 2011
2011
Loc

Mojoceratops perifania

Longrich 2010: 683
2010
GBIF Dataset (for parent article) Darwin Core Archive (for parent article) View in SIBiLS Plain XML RDF