Redescription of Ankylosaurus magniventris Brown 1908 (Ankylosauridae) from the Upper Cretaceous of the Western Interior of North America

Abstract

The armor-plated dinosaur Ankylosaurus magniventris is redescribed based on specimens from the Hell Creek Formation of northeastern Montana, USA., Lance Formation of Wyoming, USA., and from the Scollard Formation of south-central Alberta, Canada. Except for brief descriptions, most of these specimens have not been described in detail. Ankylosaurus is one of the largest known ankylosaurids, having an estimated length of up to 6.25 m (20.5 ft). It is characterized by a long, low skull having very prominent cranial “horns” that project laterally or dorsolaterally. The body armor includes a large half-ring that sat across the base of the neck and shoulders and a large, low tail club.

Full text

Redescription of Ankylosaurus magniventris
Brown 1908 (Ankylosauridae) from the Upper
Cretaceous of the Western Interior of North
America
Kenneth Carpenter
Abstract: The armor-plated dinosaur Ankylosaurus magniventris is redescribed based on specimens from the Hell
Creek Formation of northeastern Montana, USA., Lance Formation of Wyoming, USA., and from the Scollard Formation
of south-central Alberta, Canada. Except for brief descriptions, most of these specimens have not been described in
detail. Ankylosaurus is one of the largest known ankylosaurids, having an estimated length of up to 6.25 m (20.5 ft). It
is characterized by a long, low skull having very prominent cranial “horns” that project laterally or dorsolaterally. The
body armor includes a large half-ring that sat across the base of the neck and shoulders and a large, low tail club.
Résumé : Le présent article fournit une nouvelle description du dinosaure cuirassé Ankylosaurus magniventris basée
sur des spécimens de la Formation de Hell Creek du nord-est du Montana, de la Formation de Lance, du Wyoming,
toutes deux aux États-Unis, ainsi que de la Formation de Scollard du centre-sud de l’Alberta, au Canada. Sauf pour de
brèves descriptions, la plupart de ces spécimens n’ont pas été décrits en détail. Ankylosaurus est l’un des plus gros an-
kylosauridés connus; il peut atteindre une longueur estimée de 6,25 m (20,5 pi). Il est caractérisé par un crâne long et
bas avec des « cornes » crâniennes très proéminentes qui se projettent latéralement ou dorsolatéralement. La cuirasse
du corps comprend une large demi-bague placée à la base du cou et des épaules ainsi qu’une une large et basse
massue caudale.
[Traduit par la Rédaction] Carpenter 986
Introduction
Armor-plated dinosaurs, the ankylosauromorphs, first
appeared in the Sinemurian (Carpenter 2001; Norman 2001)
and disappeared during the Maastrichtian, just before the
end of the Cretaceous (Carpenter and Breithaupt 1986). They
spread globally and, based on their sedimentological occurrence,
spread into a variety of environments from desert to coastal
plain. It may have been their adaptability to a host of envi-
ronments that led to their longevity as a group for 135 million
years.
The best known ankylosaur in popular culture is Ankylo-
saurus, probably because it was featured as a life-sized re-
construction at the 1964 World’s Fair in New York City.
Ankylosaurus, as A. magniventris, was named by Barnum
Brown in 1908 for a partial skull and skeleton from high in
the Hell Creek Formation (Maastrichian) exposed in badlands
near the head of Gilbert Creek in Garfield County, Montana
(Fig. 1). The specimen had been found in 1906 by Peter
Kaisen, one of Barnum Brown’s collectors, during their work
in the Hell Creek Formation between 1902 and 1910. Brown
had actually collected armor plates of Ankylosaurus several
years earlier, in 1900, while excavating the holotype of
Dynamosaurus imperiosus (= Tyrannosaurus rex) in the Lance
Formation, Niobrara County, Wyoming. However, Osborn
(1905, 1906) assumed that the armor belonged to the theropod
when he named and described the specimen, a point Brown
also maintained (1908) when he named Ankylosaurus several
years later. Osborn (1916) continued to believe that the plates
belonged to Tyrannosaurus years later. However, numerous
Tyrannosaurus skeletons are now known and none have armor
plates associated with them. Comparison of the plates with
other known ankylosaurs from North America shows that
they are most similar to those of Ankylosaurus, Osborn and
Brown not withstanding (their comparisons were made with
plates of Euoplocephalus, which was originally cataloged at
the American Museum of Natural History as Ankylosaurus).
A third specimen of Ankylosaurus was collected by Brown
and Kaisen in 1910 about 4.5 m above the base of the
Scollard Formation along the banks of the Red Deer River in
southern Alberta (Sternberg 1951). This specimen includes a
nearly uncrushed skull and tail club. Another skull and single
mandible were collected by Charles M. Sternberg and
T.P. Channey in 1947 a kilometre north of the Brown and
Kaisen specimen (Sternberg 1949) and about 6 m above the
base of the Scollard. This skull is the largest Ankylosaurus
Can. J. Earth Sci. 41: 961–986 (2004) doi: 10.1139/E04-043 © 2004 NRC Canada
961
Received 16 September 2003. Accepted 4 May 2004. Published on the NRC Research Press Web site at http://cjes.nrc.ca on
16 August 2004.
Paper handled by Associate Editor H.-D. Sues.
K. Carpenter. Department of Earth Sciences, Denver Museum of Nature and Science, 2001 Colorado Blvd., Denver, CO 80206
USA (e-mail: kcarpenter@dmns.org).

© 2004 NRC Canada
962 Can. J. Earth Sci. Vol. 41, 2004
specimen known, but is poorly preserved and badly distorted
by crushing. Finally, a small section of fused distal caudals
was found in the Powder River drainage in southeastern
Montana. This specimen was recovered during the 1960s
from high in the Hell Creek Formation.
Romer (1927) figured and described a pelvis he called
Ankylosaurus, but it is clear now that the specimen is actually
one of two pelves of Euoplocephalus collected by Barnum
Brown and crew from the Horseshoe Canyon of Alberta.
The confusion stems from the original identification of the
Euoplocephalus material at the American Museum of Natural
History as Ankylosaurus. Walter Coombs (1971) corrected
the identification and gave reasons for separating the two
taxa. He later gave a brief description of the American Museum
specimens of Ankylosaurus in a series of publications (Coombs
1978a, 1978b, 1979; Coombs and Maryanska 1990) that
were accompanied with illustrations prepared by Erwin S.
Christman apparently for a lengthier description of the taxon
planned by Brown (Romer 1927, p. 251). Coombs also planned
for a lengthy description but has since abandoned this project.
Owing to the importance of Ankylosaurus as the archetype
of the armored dinosaur, a detailed description is presented
below of the major specimens. Isolated osteoderms and teeth
have also been found in the same strata, but these are not
included.
Institution abbreviations
AMNH, American Museum of Natural History, New York,
USA.; CCM, Carter County Museum, Ekalaka, Mont., USA.;
DMNH, Denver Museum of Natural History (now the Denver
Museum of Nature and Science), Denver, Colo., USA.; GI
SPS, Geological Institute, Section of Palaeontology and
Stratigraphy, Ulan Bator, Mongolia; NMC, National Museums
of Canada, Ottawa, Ontario, Canada; PIN, Paleontological
Institute, Moscow, Russia.
Systematic paleontology
Thyreophora Huene, 1915
Ankylosauromorpha Carpenter, 2001
Order Ankylosauria Osborn, 1923
Family Ankylosauridae Brown, 1908
Genus Ankylosaurus Brown, 1908
Ankylosaurus magniventris Brown, 1908
HOLOTYPE: AMNH 5895, top portion of skull, two teeth, five
cervical vertebrae, 11 dorsal vertebrae, three caudal vertebrae,
right scapulacoracoid, ribs, and dermal armor.
REFERRED SPECIMENS: AMNH 5214, complete skull, left and
right mandibles, six ribs, seven caudal vertebrae with associated
tail club, left and right humeri, left ischium, left femur, right
fibula, and dermal armor. AMNH 5866, 77 plates and smaller
osteoderms; CCM V03, section of fused caudal vertebrae;
NMC 8880, skull and left mandible.
LOCALITIES: AMNH 5895, 61–67 m below the K–T boundary,
Hell Creek Formation, upper end of Gilbert Creek, probably
somewhere in Section (Sec.) 27 or 28, Township (T) 22N,
Range (R) 40E, Garfield County, Montana, USA.; AMNH
5214, 45.4 m below K–T boundary, Scollard Formation, left
bank of the Red Deer River, center of Sec. 26, T 33, R 22,
Alberta, Canada (Sternberg 1951); AMNH 5866 (part), Lance
Formation, somewhere in Sec. 14, 15 or 16 T 40N, R 63W,
Seven Mile Creek drainage, Niobrara County, Wyoming, USA.
CCM V03, high in the Hell Creek, east side along the Powder
River drainage, somewhere in R 1S, T 53E, -24 km south of
Powderville, Powder River County Montana, USA. NMC
8880, 43.9 m below K–T boundary, Scollard Formation, right
bank of the Red Deer River, SE ¼ of Sec. 35, T 33, R 22,
Alberta, Canada.
REVISED DIAGNOSIS: largest known ankylosaurid, up to 6.25 m;
premaxillae expanded laterally by internal sinuses, crowding
external nares to lateral sides; maximum width of maxillary
tooth rows same as width of premaxillary beak; external nares
opposite 1st maxillary tooth; large, triangular osteoderm fused
to postorbital and squamosal, directed posterodorsolaterally;
large triangular osteoderm fused to jugal and quadratojugal,
directed posteroventrolaterally; cranial ornamentation of large,
flat polygons, including a large, diamond-shaped internarial;
sharp supraorbital osteoderms continuous with squamosal
osteoderm; greatest number of cheek teeth of all ankylosaurids
(34–35/35–36); quadrate process of pterygoid directed laterally,
not posterolaterally; cervical half-ring of three keeled plates,
outermost has a laterally projecting keel; body armor includes
relatively smooth textured plates, with sharp edge or lower
keel along one margin.
Description
Skull (Figs. 2–8)
Two complete and one partial skull of Ankylosaurus are
known (Fig. 2). The holotype, AMNH 5895, consists of the
top half of the skull (Figs. 2A; 3A, 3B), which exposes the
Fig. 1. Distribution map of major Ankylosaurus magniventris
specimens (isolated teeth and armor not included) in order of
historical discovery. 1, 1900 associated with Tyrannosaurus rex
(holotype of Dynamosaurus imperiosus); 2, 1906 holotype
Ankylosaurus magniventris AMNH 5895; 3, 1910 AMNH 5214;
4, 1947 NMC 8880; 5, in 1960s CCM V03.

© 2004 NRC Canada
underside of the cranium (Fig. 3C, described in the following
text). The preserved portion measures 55 cm long and 63.5 cm
wide. Based on the more complete skulls, it is estimated to
have had a premaxillary – occipital condyle length of 61 cm.
The smallest skull, AMNH 5214 (Figs. 2B–2D; 4), has a
premaxillary – occipital condyle length of 55.5 cm and a
width of 64.5 cm across the quadratojugal horns, whereas
the largest skull, NMC 8880 (Figs. 2E, 2F; 5), has a
premaxillary-occipital condyle length of 64.5 cm and is
74.5 cm wide across the quadratojugal horns. In dorsal view,
the skulls are triangular and are wider than long; length of
both skulls is 0.86% of width. The nuchal margin overhangs
the occiput.
There is considerable difference among the three skulls,
which for the present, is best explained as taphonomic and
(or) individual variation. In profile, the snout of AMNH
5214 is arched dorsally above the level of the cranium, whereas
it appears not to have been nearly as much in AMNH 5895
(compare Figs. 2C, 2D with 2A); the condition in NMC
8880 is unknown because of crushing (Figs. 2E, 2F). The
rear portion of NMC 8880 is the least part crushed of that
skull because of reinforcement by the fused dermal plates on
the corners of the skull and the braincase. The snout is
considerably crushed because of the lack of reinforcement
in the nasal cavity.
The orbits are subcircular to slightly oval (Fig. 2). They
are visible in anterior view because of the anterior tapering
of the skull (Fig. 6A), suggesting some stereoscopic vision
was possible. The area around the orbit is flat, rather than
raised posteriorly, as in Tarchia, or completely raised, as in
Tsagantegia (Tumanova 1993). Above the middle of the orbit,
a crest is developed by the supraorbitals at the dorsolateral
margin of the skull. The crest merges into the posterior dorsal
cranial horn (Figs. 2B; 3B; 4B), thereby forming the
postorbital–squamosal horn (identified from its anatomical
position). Although there is a sulcus bisecting the crest, indi-
cating that at least two horny scales were originally present,
it is not known whether there were two separate ossifications
that fused to the skull, although this seems likely based what
is known for Pinacosaurus and Euoplocephalus (Carpenter,
Carpenter 963
Fig. 2. Skulls of Ankylosaurus magniventris in lateral views. (A) holotype AMNH 5895 reconstructed (faded portion). (B) interpreta-
tive sketch of AMNH 5214 with parts labelled. (C) left and (D) right sides of AMNH 5214. (E) left and (F) right of NMC 8880.
Scale in centimetres. en, external nares; jqjh, jugal–quadratojugal horn; o, orbit; pm, premaxilla; posqh, postorbital–squamosal horn; su,
sulcus separating a supraorbital scale from the postorbital–squamosal scale.

© 2004 NRC Canada
964 Can. J. Earth Sci. Vol. 41, 2004
unpublished notes). The postorbital–squamosal horn is directed
posterodorsolaterally, but project only slightly above the level
of the parietal region (see Figs. 4F; 6A). In Tarchia,Euo-
plocephalus, and Saichania, there is only a squamosal horn,
and this projects posterolaterally. A ventral counterpart to
the postorbital–squamosal horn is the well-developed jugal–
quadratojugal horn (named for its anatomical position). It
projects posteroventrolaterally (Figs. 2B; 4F); Tarchia,
Euoplocephalus, and Saichania have a similar horn, although
in Tarchia, the horn is attached by a constricted neck of
bone.
Brown (1908) originally suggested that armor plates were
coossified to the skull surface, but analysis of other ankylo-
saurs suggests that is not entirely the case. Instead, remodeling
of the skull surface by the overlying scales of the skin has
left the scale pattern in the skull surface (Vickaryous et al.
Fig. 3. holotype AMNH 5895 in (A) dorsal view compared with (B) interpretative sketch. (C) ventral surface of the holotype AMNH
5895 compared with that of (D) Talarurus plicatospineus PIN 3780/1. White arrows and lines show interconnecting passages. Black
curved arrow shows path of air into the nasal chamber. Straight black arrows in C denote approximate location for cross-section in
Fig. 8B. af, adductor fossa; aow, anterior orbital wall; av, arched vault (palatine); e, endocranial cavity; ino, internarial ornamentation;
ins, internarial septum; nc, nasal chamber; nc-a, nasal chamber, anterior opening; nc-p, nasal chamber, posterior opening; nm, nuchal
margin; o, orbit; ot, olfactory tract; pow, posterior orbital wall; posqh, postorbital–squamosal horn; ps, premaxillary sinus; qf, fossa for
quadrate head; s, sinuses; sm, sutural surface for maxilla. Scales in centimetres.

© 2004 NRC Canada
Carpenter 965
Fig. 4. Skull of Ankylosaurus magniventris, AMNH 5214 in dorsal (A), ventral (C) and posterior (E) views. Interpretative sketches in
dorsal (B), ventral (D), and posterior (F) views. bo, basioccipital; bs, basisphenoid; bt, basipterygoid tubera; ch, choana; cm, crest on
maxilla; ec, ectopterygoid; en, external nares; ex, exoccipital; fm, foramen magnum; ino, internarial ornamentation; jqjh, jugal–
quadratojugal horn; ltf, lateral temporal fenestra; ms, maxillary shelf; mx, maxillary; nm, nuchal margin; oc, occipital condyle; paf,
proatlas facet; pal - sp, palatine secondary palate; pm, premaxilla; pop, paroccipital process; posqh, postorbital–squamosal horn; ppf,
postpalatine foramen; pt, pterygoid; ptf, post-temporal fenestra; ptp, pterygoid process; q, quadrate; qj, quadratojugal; so,
supraoccipital; stf, subtemporal fenestra; su, sulcus between scales; tr, tooth row; v, vomer. Scale in centimetres.

© 2004 NRC Canada
966 Can. J. Earth Sci. Vol. 41, 2004
2001; Carpenter 2001; Carpenter et al. 2001). The jugal –
quadratojugal and postorbital–squamosal horns, however, are
probably separate osteoderms that have fused to the cranium,
as in Euoplocephalus (Vickaryous et al. 2001). The remodeling
of the skull surface has obliterated all of the sutures between
the cranial bones, as is common among adult ankylosaurs
(Carpenter 2001). The pattern of cranial scales is variable,
although certain patterns are present in a manner analogous
to lizards (see Vickaryous et al. 2001). At the front of the
snout is a large diamond-shaped pattern, the internarial scale,
between the external nares in both AMNH 5895 and AMNH
5214 (the condition in NMC 8880 cannot be determined).
Above the orbit are two supraorbital osteoderms that form a
continuous ridge with the postorbital – squamosal osteoderm
as mentioned earlier in the text (Figs. 2B; 3B; 4B). Along
the nuchal margin are a large and small pair of scale patterns
(Figs. 3B; 4B), although these may be asymmetrical on the
left and right sides.
In lateral view, the front of the snout is arched, as in
Euoplocephalus, rather than flat dorsally and abruptly truncated
anteriorly, as in Tarchia and Saichania. The external nares
are elliptical and face laterally, rather than anteriorly or antero-
laterally as in many ankylosaurids. Coombs (1971) suggested
that the lateral facing external nares was due to expansion of
the nasal over the premaxillae, whereas Maryanska (1977)
thought that dermal ossifications had displaced the nares.
However, the presence of sinuses on the ventral side of the
snout (Fig. 3C) demonstrate that, in actuality, the expansion
is due to laterally expanded sinuses developed within the
anterodorsal part of the premaxillae. Premaxillary sinuses
also occur in Saichania (Maryanska 1977) and Euoplocephalus
(Coombs 1978a, fig. 5A), although the premaxillae are not
as laterally expanded. The expansion has crowded the external
nares towards the lateral side of the snout in Ankylosaurus.
Recessed within the external nares is a subvertical intranarial
septum (Fig. 7). An intranarial septum is also seen in other
ankylosaurids, such as Pinacosaurus,Euoplocephalus, and
Saichania, although in these ankylosaurids the septum is
more horizontal and externally located. This septum separates
the nasal passage (opening A of Hill et al. 2003; paranasal
aperture of Vickaryous and Russell 2003), located dorsally,
from a sinus passage (opening C of Hill et al. 2003;
Ankylosaurus does not appear to have an opening B). There
are five sinuses on each side of the snout (Fig. 3C). The
anteriormost, as just noted, is located within the expanded
premaxilla. Brown (1908) originally identified the sinuses as
air chambers and compared them with sinuses in elephant
skulls. Mapping of the remaining four sinuses relative to an
overlay of the palatal region of Ankylosaurus demonstrates
that they are located within the maxillae (Fig. 8A with Fig. 3C)
Fig. 5. Skull of Ankylosaurus magniventris, NMC 8880, in stereoscopic ventral view, with right mandible in occlusion. Scale in centimetres.
Fig. 6. (A) skull of Ankylosaurus magniventris, AMNH 5214, in anterior. (B) skull of Ankylosaurus magniventris, NMC 8880, in pos-
terior. Scale in centimetres.

© 2004 NRC Canada
and may be part of the reason why the maxillae extend laterally
well past the tooth row as the maxillary shelf. In Talarurus
(Fig. 3D), there are only two sinuses on each side of the
snout; the number in other ankylosaurids is unknown.
The function of the sinuses in ankylosaur skulls is prob-
lematic. Maryanska (1977, p. 117) suggested the function
was to reduce skull weight, house a nasal gland, or act as a
resonating chamber. However, because sound in tetrapods is
made by the larynx (syrinx in birds), not by blowing air
through the nostrils, a resonating function in ankylosaurs
seems very doubtful. Weight reduction of the skull was probably
minimal because the amount of bone reduced is only a small
percentage of the total volume. In addition, there is no corre-
lation between skull size and the numbers or sizes of the
sinuses. As for housing a nasal gland, there is so much vari-
ation individually (Hill et al. 2003), as well as among the
different taxa, that such a hypothesis also seems unlikely. As
Witmer (1997) has noted, it may be that the sinuses have no
function, but are outgrowths of the paranasal sinus that occupy
the region of the former antorbital fenestra. It is equally pos-
sible that there is a function, but not one that we understand
at this time.
In Ankylosaurus, a pair of elongated midline chambers,
partially floored by expanded palatines, are the nasal cavities
or chambers (Fig. 3C with Fig. 8B); these chambers were
illustrated by Brown (1908, fig. 3) in top view. The chambers
are separated from each other by a midline septum formed in
part by the vomers and nasals. This septum divides the snout
sagittally into two mirrored halves, as it does in other
ankylosaurids (e.g., Euoplocephalus Vickaryous and Russell
2003). The nasal chambers are pierced by two openings, the
anteriormost of which connects to the internal nares or choana.
Although the air passage is looped, it does not extend poste-
riorly as reported for Euoplocephalus by Coombs (1978a;
see also Vickaryous and Russell 2003). Another chamber,
the posterior nasal chamber, apparently directed air past the
short olfactory lobes (Fig. 8B). In Talarurus, the olfactory
tracts are elongate and extend well anterior to the orbits
(Fig. 3D), unlike Ankylosaurus where they terminate opposite
the posterior part of the orbit (as interpreted from the ventral
side of AMNH 5895).
The ventral or palatal side of the skull is well preserved in
both AMNH 5214 (Figs. 4C, 4D) and NMC 8880 (Fig. 5),
although the sutures between adjacent bones are best seen in
AMNH 5214. The premaxillary scoop is broader than long
and it forms, with a small part of the maxillae and the
anteriormost part of the vomers, the anterior secondary palate.
The tomial ridge in both specimens is damaged, thus the
depth of the cutting edge is unknown. However, the ridge is
complete enough in AMNH 5214 to show that it did have a
narrow, inverted V-shaped notch at the midline and that the
ridge extended lateral to the anterior portion of the maxillary
tooth row, well posterior to the external nares. The ridge is
separated from the tooth row, as it is in all ankylosaurids, by
a shallow groove. An elongate incisive fenestra is present
along the midline of the premaxillary scoop in AMNH 5214,
but this is apparently absent in NMC 8880. The suture between
the two premaxillae in AMNH 5214 may be traced from the
palatal side to the dorsal side (Fig. 6A), suggesting that the
premaxillae are not completely fused together.
The maxillae are laterally expanded in Ankylosaurus,as
well as most ankylosaurs, giving the impression that the side
of the face is bulged out. This expansion in ankylosaurs may
be due to the development of maxillary sinuses in a manner
that prevents reduction in the volume of the nasal chamber.
The maxillae have a ridge along their lateral margin, which
is confluent with the tomial ridge of the premaxillae, that
Vickaryous and Russell (2003) refer to as the tomial crest of
the maxilla in Euoplocephalus. However, this ridge probably
did not form part of the cutting edge of the mouth, but rather
marks the insertion of a fleshy cheek. The presence of cheeks
in ornithischians is controversial (Galton 1973 vs. Papp and
Witmer 1998, Czerkas 1999), although indirect evidence does
support its presence in ankylosaurs. In both Panoplosaurus
and Edmontonia, there are ellipsoid armor plates (apparently
absent in ankylosaurids) covering the cheeks. These plates
are dermal in origin and must have been embedded in a
fleshy, not muscular, cheeks. Because both Panoplosaurus
and Edmontonia have maxillary ridges similar to Ankylosaurus
and Euoplocephalus, it is safe to assume that these, and
probably all ankylosaurids, had fleshy cheeks as well (contra
Papp and Witmer 1998).
As just noted, the maxilla contribute to a small portion of
the anterior secondary palate. The maxilla extends medially
at its anterior end to contact the vomers, unlike Pinacosaurus
where the two bones are separated by the premaxilla
(Maryanska 1977). The maxillary tooth row occupies a
ventrally projecting ridge that is visible in lateral profile
(Figs. 2B, 2C). The row is separated from the lateral margin
of the maxilla by a broad maxillary shelf (Fig. 4D). The
tooth rows diverge posteriorly: that of AMNH 5214 curve
posterolaterally, whereas those of NMC 8880 are straighter,
but angle posterolaterally (compare Figs. 4C, 4D and 5). The
posterior width between the tooth rows is about the same as
the maximum width of the premaxillary scoop, as in Tarchia
Carpenter 967
Fig. 7. Narial region of Ankylosaurus magniventris, AMNH
5214, showing the internal septum and passage into the
maxillary sinus beneath it (arrow).

© 2004 NRC Canada
968 Can. J. Earth Sci. Vol. 41, 2004
and Saichania; the beak is considerably wider in Euo-
plocephalus, Tsagantegia, and Pinacosaurus. There are 34
to 35 alveoli in the maxillae of AMNH 5214, which is the
highest number among ankylosaurid (see Table 1). The left
tooth row in AMNH 5214 is 21 cm long and the right 20 cm
(both are damaged in NMC 8880 and both are missing in
AMNH 5895). Along the lingual side of each alveoli is a
small foramen, as in Euoplocephalus (Vickaryous and Russell
2003) and dinosaurs in general, through which the replacement
tooth may be seen.
The vomers are fused along their midline and form a Y-
shaped structure in cross-section. The vomer keel extends
ventrally almost to the level of the tooth row. It definitely
extends ventral to the tooth row in some ankylosaurids, such
as Saichania. Dorsally, it extends to contact the palatal vault
(Figs. 4C, 4D, 5) thus dividing the nasal chamber sagittally.
A similar division of the nasal chamber also occurred in
Talarurus based on the midline fragment of bone (Fig. 3D).
Anteriorly, the vomers form a wedge between the posterior-
most portions of the premaxillae, where they form a small
portion of the anterior secondary palate. Posteriorly, the vomers
contact the pterygoids along the midline.
The palatines form most of the posterior secondary palate
(Figs. 4C, 4D; 5). These are angled anterodorsally. They are
sutured laterally to the maxillae, ectopterygoids, and pterygoids
and medially to the vomers and pterygoids. A small foramen
is present at the posteromedial suture with the pterygoids
that Vickaryous and Russell (2003) call the palatine aperture
in Euoplocephalus. As hypothesized earlier in the text, the
foramen opens into the posterior part of the nasal chamber
where the olfactory lobes are located.
The ectopterygolocephalus (Vickaryous and Russell 2003).
It is situated on the posteromedial side of the maxilla, lateral
to the palatine, and forms a small portion of the ventrally
projecting pterygoid flange. The jugal floors the orbit and
forms the lateral margin of the subtemporal fenestra (Figs. 4C,
4D). It contacts the maxilla anteriorly near its suture with
the ectopterygoid and the quadratojugal posteriorly. The quad-
ratojugal is mostly hidden by a large, triangular osteoderm,
although a narrow wedge of it is visible in posterior view,
where it forms the ventral margin of the lateral temporal
fenestra (Figs. 4E, 4F).
The pterygoids are anteroposteriorly short and broad. The
anterior process, which primitively in dinosaurs extends dorso-
anteriorly, extends dorsally to slightly posterodorsally. The
Fig. 8. (A) Overlay showing the relative position of the sinuses on the ventral surface of AMNH 5895 to the palatal region of
Ankylosaurus (AMNH 5214). (B) off-axis sagittal reconstruction of Ankylosaurus (straight black arrows in Fig. 3C) showing the air
flow direction into the nasal chamber and sinuses. Although air flow is shown as unidirectional, it is possible that the ventral opening
was plugged by tissue and the air flow was bidirectional in the posterior nasal chamber. Reconstruction is an interpretation based on A
and Fig. 3. Silhouette of braincase modified from Vickaryous and Russell (2003) for Euoplocephalus. Scale in centimetres.
Taxon Maxillae Dentary Reference
Ankylosaurus 34–35 35–36 This study
Euoplocephalus 19–24 21 Vickaryous and
Russell 2003
Pinacosaurus 14–17 16–19 Hill et al. 2003;
Tumanova 1987
Tarchia 18 18 Tumanova 1978
Tsagantegia 19 Tumanova 1993
Saichania 22 16–17 Maryanska 1977
Table 1. Tooth count in various ankylosaurids.

© 2004 NRC Canada
dorsal position is present in Stegosaurus (DMNH 2818), but
until the condition can be ascertained in the ankylosauromorph
Scelidosaurus, it is not known if the condition was acquired
independently in the two groups (absent in Scelidosaurus)ora
character unifying the thyreophorans (present in Scelidosaurus).
The pterygoids form the anterior and medial sides of the
subtemporal fenestrae (Figs. 4C, 4D). The suture between
the left and right pterygoids is not visible in any specimen,
instead they meet to form a midline keel that is continuous
with the vomerine keel. Posteriorly, there is a small V-shaped
notch between the two pterygoids, giving the pterygoid
complex a W-shape. In addition, the pterygoids extend pos-
teriorly under the basipterygoid process, but do not appear to
be fused to them. The pterygoid flange extends ventrally below
the middle of the orbit; its distal end is rounded. The quadrate
process extends directly laterally, rather than posterolaterally
as in most other ankylosaurids. In addition, it forms an over-
lapped joint with the pterygoid process of the quadrate, as it
does in all dinosaurs.
The quadrate is an anteroposteriorly short, broad bone
(Figs. 4C, 4D). It is almost vertical, in marked contrast to
the anterior slope seen in Saichania (Maryanska 1977) and
Tsagantegia (Tumanova 1993). As is typical for ankylosaurids,
the mandibular condyle of the quadrate is completely hidden
in lateral view by the jugal–quadratojugal “horn” and is located
below the posterior edge of the orbit. Because the fossa for
the quadrate head is visible on the underside of the cranium
AMNH 5895 (Fig. 3C), the quadrate could not have been
coossified to the squamosal; the same was also true of Talarurus
(Fig. 3D). The quadrate in Ankylosaurus also forms the pos-
terior margin of the subtemporal fenestra (Figs. 4B, 4D).
Both cranial fragments of Ankylosaurus (AMNH 5895) and
Talarurus (PIN 3780/1) demonstrate that the supratemporal
fenestrae were closed by the surrounding bones (Figs. 3C,
3D) and not simply the fusion of armor plates to the skull
roof, as suggested by Maryanska (1977). Such closure is also
noted in the skull of the Jurassic polacanthid Gargoyleosaurus,
implying that it occurred early in ankylosaur evolution post-
Scelidosaurus.
The basioccipital and basisphenoid are completely coossified
in all specimens (Figs. 4C, 4D; 5). The basisphenoid is
underlain by the pterygoids, thus most features cannot be
seen. There is an elongate, shallow fossa extending between
the basisphenoid tubera and basipterygoid process, as noted
in other ankylosaurids (e.g., Maryanska 1977; Vickaryous
and Russell 2003). The occipital condyle projects ventro-
posteriorly and is more prominent than that of Saichania
(Maryanska 1977). Several foramina pierce the lateral sides
of the basioccipital below the level of the foramen magnum
in a manner similar to Euoplocephalus (Vickaryous and
Russell 2003); these will be described elsewhere in a review
of ankylosaur braincases.
The rear of the skull is broad and low (Figs. 4E, 4F; 6B).
The supraoccipital is an inverted triangle that is broad and
low and has a vertical, low midline ridge. Like most ankylo-
saurs, it barely contacts the dorsal margin of the foramen
magnum. The occipital condyle is crescentic or reniform in
shape as it is in Scelidosaurus and all ankylosaurids and
polacanthids. The exoccipitals form the lateral and part of
the dorsal walls of the foramen magnum, as well as contribute
to almost one-quarter of the occipital condyle at its dorsolateral
margins. The exoccipitals and opisthotic form the paroccipital
processes, which medially contact the ventral margin of the
supraoccipital. The processes have small proatlas facets on
them dorsolaterally above the foramen magnum. Similar facets
are known for most ankylosaurs (missidentified as nodes
bearing muscle scars in Pinacosaurus and Saichania by
Maryanska 1977), but only in Gargoyleosaurus have the
small triangular bones been found in situ. The paroccipital
processes arc gently laterally, so that the distal ends are below
the level of the foramen magnum floor; they are not fused to
the quadrates as they are in Saichania (Maryanska 1977) and
Tsagantegia (Tumanova 1993). The processes are not visible
in dorsal view (Figs. 4A, 4B). The lateral temporal fenestra
are visible in the occipital view, just dorsal to the quadratojugal
(Figs. 4E, 4F; 6B).
Lower jaws (Fig. 9)
Both mandibles of AMNH 5214 and the left mandible of
NMC 8880 are known; as yet, the predentary has not been
found. The mandible is proportionally lower to its length
than in Saichania,Tarchia, and Euoplocephalus and is more
similar to that of Shamosaurus suggesting a retention of a
primitive character. The tooth row is almost straight in lateral
view, rather than arched dorsally as in Euoplocephalus,
Saichania,Tarchia, and Pinacosaurus. As in most ankylosaurids,
when articulated to the cranium, the mandibles cant or slope
lingually to accommodate the medial placement of the maxillary
teeth due to the maxillary shelf (the mandible retains its
primitive erect position in Shamosaurus, polacanthids, and
nodosaurids). This canted position also makes the ventral
margin visible in lateral view. The left mandible of AMNH
5214 is 41.5 cm long, with a 21 cm long tooth row; the right
is 42 cm, with a 22.5 cm tooth row. The mandible of NMC
8880 is incomplete anteriorly, and the preserved portion is
41 cm long. The adductor fossa, located on the medial side
below the coronoid process, is remarkably small for such a
large skull. For example, the largest skull, NMC 8880, has a
midline length 64.5 cm, yet the fossa is only 5.3 cm long.
This contrasts with Edmontonia (NMC 8531) with a skull
midline length of 49 cm and a fossa length of 7.9 cm. These
data suggest that the adductor muscles were smaller in
Ankylosaurus than in the nodosaurid Edmontonia, and this
is probably related to the much smaller teeth of Ankylosaurus.
The fossa is closed posteriorly, as it is in Euoplocephalus,
whereas it is open in Tarchia,Pinacosaurus and Shamosaurus.
Laterally, the mandible has an extensive rough surface
(Figs. 9A–9C) that projects slightly ventrally below the low
margin of the jaw (Figs. 9A′–9C′). In Pinacosaurus,Tarchia,
and Euoplocephalus, there is a dermal ossification that
contributes to a large part of this rough surface. The sharp,
ventrolateral projecting keel on the mandible of Ankylosaurus
suggests that it is a keeled plate, and thus is of a similar origin.
The dentary is proportionally longer relative to its height
than in Euoplocephalus. There are 35 alveoli in the left dentary
of AMNH 5214, and 36 in the right. This is far more teeth
than in any other ankylosaurid (see Table 1). In dorsal view,
the tooth row is curved medially (Figs. 9A′′ ′,9B
′′ ′), mirroring
that of the maxillae rows (Figs. 4C, 4D). The thin, medial
walls of the alveoli are missing in both mandibles of AMNH
5214, but are present, along with some poorly preserved
teeth, in NMC 8880. Posteriorly, the tooth row extends medial
Carpenter 969

© 2004 NRC Canada
970 Can. J. Earth Sci. Vol. 41, 2004
Fig. 9. Mandibles of Ankylosaurus magniventris. AMNH 5214: left in (A) lateral, (A′) medial, ( ′′
A) ventral, and (A′′ ′) occlussal; right
in (B) lateral, (B′) medial, ( ′′
B) lateral interpretative sketch, (B′′ ′) medial interpretative, (B*) ventral, and (B**) occlussal; Incomplete
NMC 8880 in (C) lateral, (C′) medial, and ( ′′
C) occlussal. Scale in centimetres. af, adductor fossa; al, alveoli; ar, articular; cor,
coronoid; d, dentary; fm, mandibular foramen; Mc, Meckelian groove; nf, nutrient foramen; pd, facet for predentary; pra, prearticular;
sp, splenial; sym, symphysis.

© 2004 NRC Canada
to the anterior base of the coronoid process. In lateral profile,
the dorsal or occlusal margin is slightly higher anteriorly
than posteriorly, but this height is proportionally less than in
Euoplocephalus (Vickaryous and Russell 2003, figs. 15C,
15D). The symphyseal or mandibular ramus is tapering and
is oriented horizontally; it does not curve medially as abruptly
as in Euoplocephalus. There is a short, triangular process
that projects laterally near where the symphyseal ramus
attaches to the main body of the dentary, best seen in dorsal
or ventral view (Figs. 9A′′,9A
′′ ′, 9B*, 9B**). This process
apparently extends the mandibular beak laterally to conform
to the broad premaxillary beak. The facet for the predentary
(not yet found) is a shallow groove along the front of the
symphyseal ramus. The symphysis surface is smooth, except
for a groove that splits the surface horizontally. A similar
groove is seen in most ankylosaur dentaries.
Medially, the Meckelian canal is visible just posterior to
the symphyseal ramus. Posteriorly, it is covered by the splenial,
which covers most of the medial side of the mandible. The
splenial extends ventrally and covers the ventral edge of the
mandible. Its posterior border is below the coronoid process
and extends to the small intermandibular fenestra, where
upon it extends posteriorly. This fenestra is not within the
splenial as it is in Euoplocephalus (Vickaryous and Russell
2003), but at the boundary between the splenial and the
prearticular (Fig. 9B′′ ′). The coronoid is a small bone located
at the end of the tooth row and it forms the anterior margin
of the adductor fossa. The coronoid process is a very low,
broad triangle above the adductor fossa that is even less
developed than in Euoplocephalus but more developed than
in Pinacosaurus.
The prearticular is a subrectangular plate forming the medial
wall of the adductor fossa and of the retroarticular process.
The intermandibular fenestra occupies a position at the antero-
ventral margin with the splenial. The articular is a rectangular-
to square-shaped bone best seen in dorsal view (Figs. 9A′′ ′,
9B**). It is very wide, consequently so is the retroarticular
process. The process angles posteromedially (i.e., postero-
lingually) as it does in many ankylosaurs. It has a wide
quadrate fossa. There is a tubercle or mound at the postero-
lateral corner of glenoid, a character less developed in Euoplo-
cephalus.
The angular is not well delineated because most of its lateral
surface has been extensively reworked by the overlying scales
and by the probable coossification of an armor plate to its
surface as in Euoplocephalus (Vickaryous et al. 2001).
Regardless, the ventral portion of the mandible in the vicinity
of the angular is thickened (Figs. 9 ′′
A, 9B*). The angular
does not wrap onto the medial surface as in Euoplocephalus
(Vickaryous and Russell 2003). The borders of the surangular
are not well delineated because of remodeling of the bone
surface. Sutures along the dorsal or occlusal surface are visible
and these show that the surangular forms the lateral wall to
the adductor fossa and the retroarticular, as well as forming
most of the coronoid process as in Euoplocephalus (Vickaryous
and Russell 2003).
Teeth (Fig. 10)
Brown (1908) reported that two teeth were recovered from
the matrix-filled sinus of the snout (Figs. 10A, 10B). Coombs
(1990) has expressed doubt that these teeth belong to
Ankylosaurus, claiming that they do not match well with
teeth in maxilla of skull AMNH 5214 (Figs. 10C–10E), nor
to those of NMC 8880. Nevertheless, he conceded that the
teeth were all found with skulls, which makes their identifi-
cation as Ankylosaurus teeth highly probable. Coombs further
raised the possibility that a second species of Ankylosaurus
was present as a means to explain the differences. However,
both he and I have independently concluded that this is
highly unlikely based on the skeletal remains. Considering
the great variation in the teeth of Edmontonia and Euoplo-
cephalus (personal observations), for which multiple skulls
are known, I have no difficulty accepting these teeth as
those of Ankylosaurus magniventris.
Most of the crowns of Ankylosaurus are significantly taller
than wide, whereas those of the contemporary Edmontonia
are typically as wide as they are tall. In addition, the teeth
are very small relative to their skulls, and wear is on the
crown face rather than apical across the crowns as in
nodosaurids. Some of the teeth are slightly curved posteri-
orly, and these are from the posterior portion of the tooth
row. One side of the crown is typically flatter than the other.
Denticle count along the margins is variable, ranging from 6
to 8 on the anterior side of the apical cusp, and 5 to 7 on the
posterior. The cingulum is best developed on one side. The
crown may have some faint fluting, but these are unlike the
meandering and anastomosing rugosities common in Euo-
plocephalus.
Isolated, shed teeth are known from microvertebrate localities
throughout the Lance, Hell Creek, and Scollard formations
(Carpenter 1982a; Carpenter and Breithaupt 1986). Although
they indicate a broad geographical distribution of Ankylosaurus,
their rarity mirrors the skeletal remains.
Cervical vertebrae (Figs. 11)
There is a nearly complete series of cervical vertebrae in
the holotype, AMNH 5895. These mostly probably represent
c3–c7, although Brown (1908) thought the first one was the
atlas. The centra are anteroposteriorly shorter than wide.
Their articular surfaces are slightly amphicoelous and circular
throughout the series. In Saichania the articular faces become
progressively dorsoventrally flattened ellipses towards c-6
(Martyanska 1977), and in Pinacosaurus, c-3 and c-4 are
parallelograms, with the anterior articular face higher than
the posterior face (Maleev 1954). The lateral sides of the
centra in Ankylosaurus are concave (i.e., deeply excavated).
The articular faces are mostly on the same level, except for
c-4, in which the anterior face is displaced higher than the
posterior (Figs. 11B′,11′′
B). Maryanska (1977) reports that
c-3 and c-4 have displaced articular faces, which is true of
Talarurus as well (Maleev 1956). Ventrally, there is a thick
medial ridge for the longitudinal ligament; this ridge is most
prominent on the last two cervical vertebrae. The parapophyses
are well developed and become progressively higher on
succeeding centra.
The neural arches are short and very widely separated so
that the neural canal is almost twice as wide as tall. The neural
spines are anteroposteriorly short and erect, becoming taller
posteriorly; this is in marked contrast to Pinacosaurus,in
which the spines decrease in size. The neural spines are also
very broad transversely, as in Saichania and unlike Euoplo-
cephalus. Entheses are very extensively developed along the
Carpenter 971

© 2004 NRC Canada
972 Can. J. Earth Sci. Vol. 41, 2004
front and rear edges of the neural spines from the ossification
of the insertions of the nuchal and interspinal ligaments.
Although entheses maybe pathological in humans, they are
also normal signs of aging (Resnick and Niwayama 1983).
Because entheses are common in adult dinosaurs, they probably
are an indication of a old individual. In Ankylosaurus, these
entheses indicate that large sheets of ligaments had been
present for support of the massive head and that the individual
is assumed to have been aged, but not necessarily very old.
The lateral edge of the neural spines curve posteriorly to
connect the postzygapophyses. All of the zygapophyses are
widely separated and are about at the same level as seen in
lateral view. This is in marked contrast with Euoplocephalus
and Talarurus, where the postzygapophyses are higher. The
transverse processes are short, circular in cross-section and
extend horizontally, where they end with large diapophyses.
Dorsal vertebrae, ribs and ossified tendons (Fig. 12)
As with the cervical vertebrae, there is a nearly complete
series of dorsal vertebrae (d1–d11) in the holotype AMNH
5895 (Fig. 12). The overall size of the centra is close to that
of a large Euoplocephalus (AMNH 5337); however, the centra
of Ankylosaurus are less concave laterally giving the centra
the appearance of being more massive. The first dorsal vertebra
shares features of both the cervicals and other dorsal vertebrae.
The centrum is anteroposteriorly short relative to its width,
as in the cervicals, and the neural spine anteroposteriorly
short and relatively narrow. Overall the centra of the succeeding
Fig. 10. Opposite crown views of Ankylosaurus magniventris teeth. (A, A′) AMNH 5895 maxillary? tooth figured by Brown 1908; artist
unknown; (B) AMNH 5895 maxillary? tooth; (C–F) AMNH 5214; (C) maxillary tooth, (D) maxillary tooth, (E) maxillary tooth,
(F) dentary tooth; (G) NMC 8880 maxillary? tooth. Note wear facets (arrows) on cingulum and crown faces that are characteristic of
ankylosaurids. Cross-hatch patterns on A and B are areas restored in plaster of Paris. Scale in millimetres.

© 2004 NRC Canada
Carpenter 973
Fig. 11. Cervical vertebrae of Ankylosaurus magniventris (AMNH 5895). Dorsal view and lateral view as currently preserved articu-
lated with plaster of Paris on a steel rod. Third cervical in anterior (A), lateral (A′,′′
A), and dorsal (A*) views. Fourth cervical in an-
terior (B), lateral (B′,′′
B), and dorsal (B*) views. Fifth cervical in anterior (C), lateral (C′,′′
C), and dorsal (C*) views. Sixth cervical
in anterior (D), lateral (D′,′′
D), and dorsal (D*) views. Seventh cervical in anterior (E), lateral (E′,′′
E), and dorsal (E*) views. Note
the rugosities on the neural spines indicative of entheses (mineralized fibrocartilage that mark the attachment sites of ligaments, ten-
dons, or articular capsule to bone). Scale in centimetres. c, centrum; di, diapophysis; na, neural arch; nc, neural canal; ns, neural spine;
pa, parapophysis; poz, postzygapophysis; prz, prezygapophysis; vmr, ventral midline ridge.

© 2004 NRC Canada
974 Can. J. Earth Sci. Vol. 41, 2004
Fig. 12. Dorsal vertebrae of Ankylosaurus magniventris (AMNH 5895). (a) (A) First dorsal, (B) second dorsal, (C) third dorsal,
(D) fourth dorsal, (E) fifth dorsal, (F) sixth dorsal vertebrae. (b) (G) seventh dorsal, (H) eighth dorsal, (I) ninth dorsal, (J) tenth
dorsal, (K) eleventh dorsal vertebrae. Scale in centimetres. c, centrum; di, diapophysis; na, neural arch; nc, neural canal; ns, neural
spine; pa, parapophysis; poz, postzygapophysis; prz, prezygapophysis; tp, transverse process.

© 2004 NRC Canada
dorsal vertebrae are spool-shaped: the centra are long and
cylindrical; their length increases to the seventh vertebra,
where length and width are subequal. The sides of the centra
are concave but less than in Euoplocephalus or Talarurus,
and their articular surfaces are almost circular and slightly
amphicoelous. The neural arches are massive, and the neural
canals small and laterally compressed; the canals are very
tall in Pinacosaurus because of tall neural arches (Maleev
1954; Maryanska 1977). The neural spines are very elongate,
narrow plates that overhang the back of the centra. Entheses
are present marking where the supraspinous and interspinous
ligaments attached. As Maleev (1956) has noted for Talarurus,
there is a gradual decrease in the size of the posterior dorsal
vertebrae, including the centrum, neural arch, and height of
the neural spine.
The prezygapophyses meet along the midline and form an
elongate trough, into which the co-joined, elongate postzy-
gapophyses articulate in a tongue and groove; similar joints
occur in other ankylosaurs, such as Saichania (Maryanska
1977), Pinacosaurus (Maleev 1954), and Talarurus (Maleev
Carpenter 975
Fig. 12 (concluded).

© 2004 NRC Canada
976 Can. J. Earth Sci. Vol. 41, 2004
1956). This tight union between sequential vertebrae limits
lateral movement as noted by Brown (1908) and by Maleev
(1956) for Talarurus. The position of the parapophysis changes
from being at the base of neural arch in the first dorsal vertebra,
to being completely on the arch by the third. Their triangular
shape elongates dorsally up the neural arch in successive
vertebrae and onto the transverse process, until they connect
with the diapophyses. The transverse processes are relatively
short and steeply angled upwards more than in Euoplocephalus;
they do not extend beyond the height of the neural spines.
The diapophyses are elongate, inverted tear-drop-shape by
expansion onto the ventral length of the transverse process,
and by the fifth dorsal vertebra the diapophysis meets the
parapophysis in an hourglass-shaped facet for the rib head, a
feature also seen in Talarurus. This union eliminates the
normal gap between the diapophysis and parapophysis. In
the last four dorsal vertebrae (d8–d11), their ribs are coossified
to them indicating that these ribs were not involved in the
bellows-like action of the more anterior ribs during respiration;
fusion in Saichania begins with d6 (Maryanska 1977). The
broadness of this part of the rib cage (see Brown 1908, fig.
12) suggests a large hindgut fermentation digestive system
analogous to that seen in herbivorous lizards (Iverson 1980);
in such lizards the colon diameter is greatly enlarged and
partitioned internally into chambers for the microflora (Iverson
1982; McBee and McBee 1982; King 1996).
Contrary to common statements (e.g., Molnar and Wiffen
1994), the ribs are not T-shaped throughout their length.
Rather, the proximal end is L-shaped, the middle T-shaped,
and the distal end oval; a similar pattern is seen in other
dinosaur ribs and is related to the development of the various
intercostal muscles. Near the proximal end of the ribs is a
slightly rugose area for the origins of the M. serratus. These
scars are common in dinosaurs and have been wrongly ascribed
for the attachment of the scapula. More distal scars are also
present on the anterior ribs (see Brown 1908, fig. 14) that
are probably associated with insertion of the M. pectoralis.
Fragments of ossified tendon are preserved near the tops
of the neural spines, one on each side. Brown (1908) reports
that four dorsal vertebrae of the holotype were found in
articulation and that these showed that the tendons over-
lapped two succeeding vertebrae. Unfortunately, the vertebrae
are now all disarticulated and this relationship lost. The shaft
of the tendons are rod-like and the ends expanded for attachment
to their individual epaxial muscle.
Caudal vertebrae (Figs. 13, 14)
Few caudal vertebrae are known, including four from AMNH
Fig. 13. Caudal vertebrae of Ankylosaurus magniventris (AMNH 5895). First(?) caudal vertebra in anterior (A), lateral (A′), and posterior
(′′
A) views. Anterior caudal vertebra in anterior (B), lateral (B′), and posterior ( ′′
B) views. Middle caudal vertebra in anterior (C), lat-
eral (C′), and posterior ( ′′
C) views. Posterior middle caudal vertebra in anterior (D), lateral (D′), and posterior ( ′′
D) views. Scale in
centimetres.

© 2004 NRC Canada
5895 and the posteriormost caudal vertebrae of AMNH 5214
and CCM V03. The anterior and mid caudal centra (AMNH
5895) are anteroposteriorly short and slightly amphicoelous,
with a raised notochordal prominence in the center. The
anteriormost caudal vertebra (first?) is complete (Figs. 13A–
13 ′′
A). It has short, laterally projecting caudal ribs, which
are coossified to the centrum. In Talarurus, the caudal ribs
of the anterior caudal vertebra project anteroventrally, with a
slight upsweep (see Maleev 1956, figs. 8, 9). The neural
arches are massive, whereas they are thin in Talarurus. The
neural canal in both taxa tapers dorsally. The neural spine is
straight and slender, although it widens laterally distally; the
short spine is slightly angled posteriorly, whereas it is more
erect in Talarurus. The pre- and postzygapophyses are widely
separate. The chevron encompasses a large, oval haemal canal
and is fused to the posterior half of the centrum. The ventral
spine is short and projects slightly forwards. In Talarurus,
the chevron is proportionally longer, encompasses a triangular
haemal canal, and projects posteroventrally. The centrum of
the third(?) caudal vertebra of Ankylosaurus is wedge shape
in lateral view (Fig. 13B′), being anteroposteriorly longer
dorsally than ventrally. In all the caudal vertebrae, except
the first, the chevron facets are prominent on both the anterior
and posterior sides.
The posteriormost seven caudal vertebrae form the handle
(terminology of Coombs 1995) of the tail club (Fig. 14; see
additional description in the section on “Armor”). The centra
are elongated, but the lengths are only slightly more than
their widths. The articular faces of AMNH 5214 are in contact
indicating complete absence of the intervertebral disc; in
CCM V03, a larger specimen, the centra are coossified to-
gether. The postzygaphyses of both specimens are reduced
and expanded along the sides of the broad, low neural spines
forming a wedge-shaped structure. The prezygapophyses mirror
these changes by becoming elongate on both sides of the
preceding postzygapophyses. In dorsal view, the post-
zygapophyses and prezygapophyses of sequential vertebrae
form interlocking V-shaped structures. Similar modifications
of distal caudal vertebrae are seen in other ankylosaurid tail
clubs, including Euoplocephalus,Pinacosaurus,Saichania,
and Talarurus. Ventrally, the chevrons in AMNH 5214 are
elongated and each has a notch on the anterior edge to receive
the posterior portion of the preceding chevron; the result is a
continuous tube for the caudal artery.
Ossified tendons are present on sides of the caudal vertebrae,
just in front of the tail knob (terminology of Coombs 1995;
Figs. 14A, 14B). These tendons do not immobilize the caudal
vertebrae of the handle, because these are already immobilized
by the structural changed noted earlier in the text. Coombs
(1979, 1995) has noted that functionally, these ossified tendons
reduce the stretch of tendon, where it is very long, and transmit
a great deal of force distally when the tail is used as a
weapon.
Scapulocoracoid (Fig. 15)
The left scapula and coracoid are coossified in AMNH
5895, with the coracoid angled about 40° medially relative
to the long axis of the scapula. The scapular blade curves
Carpenter 977
Fig. 14. Distal tail section of Ankylosaurus magniventris (AMNH 5214) showing development of the tail club in left lateral (A, B),
dorsal (C, D) and ventral (E, F) views. Segment of coossified distal caudals representing a significantly larger tail club (CCM V03) in
lateral (G) and dorsal (H) views. Scale in centimetres. c, centrum; ch, chevron; ot, ossified tendons; poz, postzygapophysis; pz,
prezygapophysis.

© 2004 NRC Canada
978 Can. J. Earth Sci. Vol. 41, 2004
slightly posteroventrally and is slightly expanded distally.
Medially, it is 61.5 cm long from the coracoid suture posteriorly
to the middle of the distal end; the height of the distal end of
the blade is 24.5 cm. The acromion is a overhanging lip on
the lateral side of the dorsal margin (Fig. 15A), as in Euoplo-
cephalus and Pinacosaurus. Although flange-like, it does not
form low on the lateral side of the scapula, as in polacanthids
(Carpenter 2001). Medially, there is a thick, horizontal ridge,
15 cm long, that bridges the scapular–coracoid suture, although
it is mostly developed on the scapula (Fig. 15C). A somewhat
similar medial ridge is present in the crocodile, where it is
associated with the M. supracoracoideus longus. The ridge is
present in Euoplocephalus, although it is not as prominent.
The ventral margin of the scapula has entheses which mark
the insertion of the M. serratus ventralis. In addition, there is
a prominent enthesis that marks the origin of the M. triceps
longus caudalis. The scapular portion of the glenoid has a
prominent, ventrally projecting, triangular postglenoid process.
Fig. 15. Left scapula of Ankylosaurus magniventris (AMNH 5895) in lateral (A), ventral (B), and (C) medial views. Scale in centi-
metres. ac, acromion; co, coracoid; cof, coracoid fenestra; gl, glenoid; mb, medial brace; Msvt, entheses marking the insertion for M.
serratus ventralis thoracis; Mtlc, enthesis of the M. triceps longus caudalis; sc, scapula.

© 2004 NRC Canada
This acts as a “stop” and brace to the humeral head, as well
as providing for the origin of the lateral head of the M. triceps
longus.
The coracoid is incomplete, but does retain the region
around the coracoid foramen (Fig. 15). The foramen pierces
the coracoid in a posteromedial direction, exiting very near
the scapular–coracoid suture, just below the coracoid portion
of the medial ridge. The glenoid portion of the coracoid is
large and faces ventrolaterally to accommodate the angling
of the coracoid relative to the scapula. The suture to the
scapula is partially open within the glenoid (Figs. 15B, 15C).
It is a relatively wide groove that extends medially but ends
well before the coracoid foramen. There is no evidence of
fusion of the coracoid with d1, as reported by Maryanska
(1977) for Saichania (I am unable to see this in her plate
34); it is most probable that this is pathological because it is
unknown in another specimen of Saichania (GI SPS 100/1305).
Humerus (Fig. 16)
Both humeri are present for AMNH 5214. The left has a
midline length of 54.2 cm from the top of the humeral head
to the ventral portion of the intercondylar notch; it has a distal
width across the condyles of 25 cm and a proximal maximum
with from internal tuberosity to deltopectoral crest of 29.5 cm.
The right humerus has a midline length of 53.6 cm, a distal
width of 27 cm, and a proximal width 28.5 cm. The humeri
are short and, in anterior view, are very broad due in large
part to the laterally directed deltopectoral crest (Figs. 16A,
16B). However, the breadth of the left humerus (often figured
in various publications, e.g., Coombs1978b, Fig. 6A) is
exaggerated from compaction as evidenced by the flattened
humeral head and distal condyles. The right humerus probably
retains the more normal appearance. In lateral view, the
humerus is more sigmoidal than in Euoplocephalus.
Proximally, the humeral head is developed mostly on the
posterior side; it does not extend onto the anterior surface, as
in Pinacosaurus (Maryanska 1977). The axis of the hemi-
spherical head is about 45° from the vertical, projecting
posterodorsally. In anterior view, the proximal end is rounded,
with the triangular internal tuberosity and top of the delto-
Carpenter 979
Fig. 16. Humeri of Ankylosaurus magniventris (AMNH 5214). Left humerus in anterior (A, A′), lateral ( ′′
A), posterior (A′′ ′, A*), and
medial (A**) views. Right humerus in anterior (B), lateral (B′), posterior (B ′′ ′B), and medial (B′′ ′) views. Note that the left humerus
is more flattened than the right (compare ′′
Aand B′), as evidenced by the shape of the humeral heads and distal condyles and the
amount the deltopectoral crest projects forwards. Scale in centimetres. dpc, deltopectoral crest; hd, head; in, intercondylar notch; it, in-
ternal tuberosity; ld, M. latissimus dorsali insertion scar; lsr, lateral supracondylar ridge; of, olecranon fossa; rc, radial condyle; tm, M.
teres major insertion scar; uc, ulnar condyle.

© 2004 NRC Canada
980 Can. J. Earth Sci. Vol. 41, 2004
pectoral crest well below the level of the head. This round-
ness is in sharp contrast with Saichania, where the proximal
end is almost straight across because the internal tuberosity
and crest are not much lower than the head. The crest projects
anterolaterally, rather than strictly laterally, as previously
reported. Consequently, the plane of the crest is not in the
same plane as the distal condyles. The humeral shaft below
the crest is short, and it is this that gives the impression that
the deltopectoral crest extends distally more than in nodosaurids
(e.g., Coombs and Maryanska 1990).
Distally, the two condyles are separated by a moderately
developed intercondylar notch. The development of this notch
is variable in ankylosaurs (compare Talarurus PIN 557-97 in
Maleev 1954, fig. 8 and PIN 557-3 (as Pinacosaurus)in
Maleev 1956, figs. 14, 15) and will probably be shown to
occur in Ankylosaurus as well. The laterally placed radial
condyle is more rounded in lateral view than the medially
placed ulnar condyle. The lateral epicondyle for the extensors
is not as well developed in Ankylosaurus, and ankylosaurids
in general, than in nodosaurids.
Ischium (Fig. 17)
The partial left ischium of AMNH 5214 lacks its distal
end. As preserved, it is 56.5 cm long and 23.5 cm long
anteroposteriorly across the proximal end. The proximal end
is cupped and forms most of the medial wall of the shallow
acetabulum. Medially, there are prominent sutural surfaces
for the second and third true sacral ribs (Figs. 17C, 17D).
The proximal end is anteroposteriorly expanded by the pubic
and iliac peduncles; these taper ventrally into the laterally
compressed shaft (Fig. 17E). In lateral view, the shaft is
very slight bowed anteriorly.
Femur (Figs. 18A–18A*′)
The left femur of AMNH 5214 is 67 cm long. It is robust
and is anteroposteriorly compressed, being oval in cross-section.
In general, the trochanters are more pronounced than Euoplo-
cephalus and the femur more massive. Proximally, the femur
is 28 cm wide. The head faces dorsomedially, with an axis
about 30° from vertical. The greater trochanter is separated
from the femoral head by a neck (Fig. 18A′′ ′ ). Laterally, the
greater trochanter extends down the shaft as a prominent
ridge (Figs. 18A–18A*; as discussed by Carpenter and
Kirkland (1998), it is not the horizontal dorsal surface of the
femur that is lateral to the femoral head as is frequently
assumed). The anterior trochanter is not separate from the
greater trochanter. The fourth trochanter is visible on the
posteromedial corner of the shaft as a rugose scar that is
slightly raised from the shaft (Figs. 18A′′ ′–18A***). The
anteroposterior narrowest point is between the greater and
fourth trochanters; the femur at this point has a minimum
circumference of 36.2 cm. On the distolateral surface of the
Fig. 17. Left ischium of Ankylosaurus magniventris (AMNH 5214) in lateral (A, B), medial (C, D), and posterior (E) views. Sale in
centimetres. a, acetabulum; ip, iliac peduncle; pp, pubic peduncle; ss2, sutural surface for sacral rib 2; ss3, sutural surface for sacral
rib 3.

© 2004 NRC Canada
Carpenter 981
shaft is a very large scar for the M. gastrocnemius (Figs. 18A′–
18A*); this scar is more prominent than in Euoplocephalus.
The distal width is 25.5 cm. The articular surface of the
condyles wraps around the femur, but are most prominent
ventroposteriorly. The intercondylar notch is not well developed
ventrally, but it is both anteriorly and posteriorly (Fig. 18A*′).
Part of the intercondylar notch extends dorsally onto the
shaft as the popliteal fossa.
Fibula (Figs. 18B–18B′′ ′)
The left fibula of AMNH 5214 is laterally compressed
and slightly bowed posteriorly; it is 44.8 cm long. It is
anteroposteriorly broad compared with most other ankylosaurs,
with the ends only slightly more expanded than the shaft;
proximal width is 12.8 cm, distal width 8.5 cm. Laterally, a
ridge develops from the anterior edge and crosses onto the
lateral surface as a brace just above the distal end. The medial
surface of the shaft is slightly concave. It is not known if a
calcaneum has fused to the distal end, as is assumed for
adult Euoplocephalus (Coombs 1986).
Metatarsal (Figs. 18C–18 ′′
C)
A fourth metatarsal, 15 cm long, is known for AMNH
5214. The pes of advanced ankylosaurids is reduced to three
digits (II–IV) as noted by Coombs (1986), and presumably
this includes Ankylosaurus. The proximal end of the meta-
tarsal is broadly triangular, almost a right triangle, thus iden-
tifying it as a metatarsal IV. The proximal surface is not flat,
Fig. 18. Hindlimb elements of Ankylosaurus magniventris (AMNH 5214). Left femur, including interpretative drawing, in anterior (A),
lateral (A′,′′
A), posterior (A′′ ′, A*), medial (A**, A***), and distal (A*′; arrow points anteriorly) views. Right fibula and interpretative
drawing in lateral (B, B′) and medial ( ′′
B,B
′′ ′) views. Right fourth(?) metatarsal in anterior (C), lateral (C′), posterior ( ′′
C), and proxi-
mal (C′′ ′) views. Scale in centimetres. 4th, fourth trochanter; fg, fibular groove; g, scar for the M. gastrocnemius; gt, greater
trochanter; hd, head; hn, head neck; it, M. ilio-tibialis scar; lc, lateral condyle; le, lateral epicondyle; mc, medial condyle; pf, popliteal
fossa.

© 2004 NRC Canada
982 Can. J. Earth Sci. Vol. 41, 2004
Fig. 19. Armor plates of Ankylosaurus magniventris (AMNH 5895). Possible midline osteoderm from the right half of the first cervical
half-ring (A). Left cervical half ring in dorsal (splayed out flat) (B) and posterior (B′) views. Based on articulated skeletons of other
ankylosaurids (e.g., Euoplocephalus, Saichania), the rest of the armor consists of dorsal osteoderms, although the smaller keeled
osteoderm (J) may be from a forelimb. Scale in centimetres.

© 2004 NRC Canada
Carpenter 983
but is sloped anteromedially (Figs. 18C–18 ′′
C). The metatarsal
is broader distally than proximally. The fossa for the collat-
eral ligament is present on the lateral side.
Armor (Figs. 19, 20)
Various armor plates (osteoderms) are known for the
holotype, AMNH 5895, and AMNH 5866. Between the two
sets of armor, it seems that a representative collection of armor
for Ankylosaurus is now known. Isolated plates from the
Hell Creek, Lance, Scollard, and other contemporary deposits
are known and some have been referred to Ankylosaurus.
However, the identity of these plates has not yet been confirmed
nor are they included here.
The armor ranges in size from a centimetre in diameter to
one 35.5 cm long. The body plates are remarkably flat,
although a weakly developed or low keel is present near one
of the margins (Figs. 19B–19E, Figs. 20G–20I) in both AMNH
5895 and AMNH 5866. As frequently noted for ankylosaurids,
the plates are relatively thin-walled and hollowed ventrally.
These large plates are readily differentiated from those of
the contemporaneous nodosaurid Edmontonia sp., in which
the keel is usually well developed, usually near the midline,
and which rises from one margin towards the other; an over-
hanging tip is common as well. There are some smaller sub-
ellipsoid osteoderms with a keel along the midline (e.g.,
Figs. 19F, 19G; 20O, 20P) that are similar to some of the
Fig. 20. Armor plates of Ankylosaurus magniventris associated with the holotype of Dynamosaurus imperiosus (= Tyrannosaurus rex)
(AMNH 5866). Based on articulated skeletons of other ankylosaurids (e.g., Euoplocephalus, Saichania), the osteoderms (A–E) are
probably lateral caudal spines; (F) a lateral body osteoderm; (G–I) the very low, asymmetrically located keeled plates are probably dor-
sal body and caudal osteoderms; (J–M) possibly sacral armor. The position of other osteoderms is less certain: (N–O) small, low-
keeled osteoderms; (P) small osteoderms, some which may have occupied the space between the larger plates. Scale in centimetres.

© 2004 NRC Canada
984 Can. J. Earth Sci. Vol. 41, 2004
keeled, oval osteoderms in Edmontonia. Finally, the armor
of Ankylosaurus has a remarkably smooth texture, as compared
with those of Euoplocephalus.
A left cervical–scapular half-ring is present in AMNH
5895 (Figs. 19A, 19A′). Although broken, it consists of
three plates fused along their margins, the smallest medially
and the largest laterally. Although all of the plates are keeled,
the most prominent keel projects laterally from the largest
plate. Similar laterally projecting keels occur on the cervical
armor of Pinacosaurus (Godefroit et al. 1999) and Saichania
(Maryanska 1977). There is no evidence of an underlying
band of bone, as is seen in Saichania (Maryanska 1977), and
all traces of sutures are obliterated by fusion. What is assumed
to be the anterior edge of the armor belt is relatively straight,
whereas the posterior side is angled. When paired, a wide,
inverted V-shaped gap is created dorsally on the posterior
side above the shoulders. The gap was probably filled with
small ossicles to allow movement of the neck. The armor
belt is wider than half the width of the skull, suggesting that
it is equivalent to the second cervical ring of other ankylo-
saurids, However, the width (Fig. 19B′) is considerably wider
than can be accommodated by the neck alone, suggesting
that it lay at the base of the neck and onto the shoulders
(Fig. 21B).
As noted earlier in the text, the large body plates are unusual
in that they are relatively flat, but have a sharp edge or low
Fig. 21. Skeletal reconstruction (A) and life restoration in dorsal (B) and lateral (C) views of Ankylosaurus magniventris.Forthe
larger NMC 8880 use scale bar a, for AMNH 5214 use scale bar b.

© 2004 NRC Canada
Carpenter 985
keel along one side (Figs. 19B–19E; 20G–20I). These unique
plates are unknown in the contemporaneous Edmontonia and
show that the collection of plates AMNH 5866 belong to
Ankylosaurus. These plates most probably occurred on the
back. Several unusual flat, keel-less plates are also known
(Figs. 19F; 20J, 20K) and, based on Euoplocephalus
(Carpenter 1982b), these may be from the sacral region.
The dorsoventrally compressed pointed plates (Figs. 20A–
20F) resemble those along the sides of the tail in Saichania
and probably had a similar distribution in Ankylosaurus. Oval
keeled plates (Figs. 19G–19L; 20N, 20O) may be from the
dorsal surface of the tail, as well as the sides of the limbs.
Finally, a variety of small irregular osteoderms and ossicles
(Fig. 20P) probably occurred around and among the larger
plates.
The tail club is only known for AMNH 5214, and this is a
very large structure (Fig. 14). Given the great variety in
knob shape in Euoplocephalus (Coombs 1995), there is no
reason to assume that the club is typical for Ankylosaurus.
The knob of Ankylosaurus is composed of a pair of major
plates laterally, a midline row of minor plates ventrally, and
a terminal pair of minor plates (Fig. 14). The knob falls into
the oval–round category of Coombs (1995) and most closely
resembles a specimen in the Royal Tyrrell Museum collec-
tions (Coombs 1995, fig. 1B). The great variety of knob
shape in Euoplocephalus, and presumably Ankylosaurus, makes
the head-mimicry hypothesis for the tail club by Thulborn
(1993) very doubtful.
Discussion
A skeletal reconstruction of Ankylosaurus is shown in
Fig. 21A and a life restoration in Fig. 21B, C. The largest
specimen, NMC 8880, is estimated to have been an individual
about 6.25 m long and about 1.7 m tall at the hips. In
contrast, the smallest specimen, AMNH 5214, is estimated
to have been about 5.4 m long and about 1.4 m tall at the
hips.
Ankylosaurus was a rare component of the Maastrichtian
dinosaur fauna. Preliminary indications, based on the strati-
graphic distribution of most material referable to this taxon
(including teeth), suggests that it is restricted to more “upland”
facies of the Hell Creek, Scollard, and Lance formations,
well away from the coastal “lowlands.” It was apparently
contemporaneous with the nodosaurid Edmontonia sp., which
is more common in the costal, “lowland” facies. But the two
apparently had limited overlapping ecological niches since
their remains have not yet been identified from the same
localities. This separation is probably due to the more selec-
tive feeding by the narrow-muzzled Edmontonia and more
low-browse cropping of the broad-muzzled Ankylosaurus (see
Carpenter 1982bfor further discussion).
A phylogenetic analysis of Ankylosaurus and all other
ankylosaurs is in preparation and will be presented elsewhere.
Suffice it to say that the characters of Ankylosaurus include
a mixture of plesiomorphic (e.g., long, low dentary) and
derived (e.g., complex nasal chamber, pneumatic premaxillae).
Acknowledgments
Over the years I have looked at the Ankylosaurus material
on several different occasions. The individuals in charge of
the collections have changed over that period, but I thank
them all for their assistance and for access to specimens:
Richard Day and Kevin Seymour, Canadian Museum of Nature;
Charlotte Holton, Eugene Gaffney, and Mark Norell, American
Museum of Natural History; board of directors, Carter County
Museum, Ekalaka, Montana. I thank Chris Ott for arranging
to bring the Carter County Museum specimen to Hill City,
South Dakota, for me to study it, as well as for its locality
information. I thank Walter Coombs for discussions about
Ankylosaurus and for his encouragement to pick up the project
he never finished. Thanks to Rob Welch for translating the
E. Maleev articles and Ruth Griffith for translating T.
Tumanova’s monograph. Pen and ink drawings of Ankylo-
saurus bones are mostly by Erwin S. Christman, AMNH.
They are reproduced here courtesy of Mark A. Norell, AMNH.
Thanks to Jolyon Parish, Matt Vickaryous, and Associate
Editor H.-D. Sues for their review comments. Also to Mary
Dawson for catching a grievous scaling error. Finally, a very
special thanks to Charlotte Holton, AMNH, who has often
gone underappreciated. Charlotte was always patient and
helpful to me as a young undergraduate in the 1970s when I
first looked at ankylosaurs, helping me to locate specimens
and providing me access to the illustrations of ankylosaurs
used here.
References
Brown, B. 1908. The Ankylosauridae, a new family of armored
dinosaurs from the Upper Cretaceous. American Museum of
Natural History Bulletin, 24: 187–201.
Carpenter, K. 1982a. Baby dinosaurs from the Lance and Hell
Creek formations and a description of a new theropod. Contri-
butions to Geology, 20: 123–134.
Carpenter, K. 1982b. Skeletal and dermal armor reconstructions of
Euoplocephalus tutus (Ornithischia: Ankylosauridae) from the
Late Cretaceous Oldman Formation of Alberta. Canadian Journal
of Earth Sciences, 19: 689–697.
Carpenter, K. 2001. Phylogenetic analysis of the Ankylosauria. In
The armored dinosaurs. Edited by K. Carpenter. Indiana University
Press, Bloomington, Ind., pp. 454–483.
Carpenter, K., and Breithaupt, B. 1986. Latest Cretaceous occur-
rence of nodosaurid ankylosaurs (Dinosauria, Ornithischia) in
western North America and the gradual extinction of the dinosaurs.
Journal of Vertebrate Paleontology, 6: 251–257.
Carpenter, K., and Kirkland, J.I. 1998. Review of Lower and Middle
Cretaceous ankylosaurs from North America. In Lower and
Middle Cretaceous terrestrial ecosystems. Edited by S.G. Lucas,
J.I. Kirkland, and J.W. Estep. New Mexico Museum of Natural
History and Science Bulletin, 14, pp. 249–270.
Carpenter, K., Kirkland, J.I., Burge, D., and Bird, D. 2001. Disarti-
culated skull of a new primitive ankylosaur from the Lower
Cretaceous of eastern Utah. In The armored dinosaurs. Edited by
K. Carpenter. Indiana University Press, Bloomington, Ind.,
pp. 211–238.
Coombs, W. 1971. The Ankylosauria. Ph.D. thesis, Columbia
University, New York, N.Y.
Coombs, W. 1978a. The families of the ornithischian dinosaur
order Ankylosauria: Journal of Paleontology, 21: 143–170.
Coombs, W. 1978b. Forelimb muscles of the Ankylosauria (Reptilia:
Ornithischia). Journal of Paleontology, 52: 642–658.
Coombs, W. 1979. Osteology and myology of the hindlimb in the

© 2004 NRC Canada
986 Can. J. Earth Sci. Vol. 41, 2004
Ankylosauria (Reptilia, Ornithischia). Journal of Paleontology,
53: 666–684.
Coombs, W.P. 1986. A juvenile ankylosaur referable to the genus
Euoplocephalus (Reptilia: Ornithischia). Journal of Vertebrate
Paleontology, 6: 162–173.
Coombs, W.P. 1990. Teeth and taxonomy in ankylosaurs. In Dinosaur
systematics: perspectives and approaches. Edited by K. Carpenter,
and P.Currie. Cambridge University Press, Cambridge, UK.,
pp. 269–279.
Coombs, W.P. 1995. Ankylosaurian tail clubs of middle Campanian
to early Maastrichtian age from western North America, with
description of a tiny club from Alberta and discussion of tail
orientation and tail club function. Canadian Journal of Earth
Sciences, 32: 902–912.
Coombs, W., and Maryanska, T. 1990, Ankylosauria. In The
Dinosauria. Edited by D. Weishampel, P. Dodson, and H.
Osmolska. University of California Press, Berkeley, Calif.,
pp. 456–483.
Czerkas, S.A. 1999. The beaked jaws of stegosaurs and their impli-
cations for other ornithischians. In Vertebrate paleontology in
Utah. Edited by D. Gillette. Utah Geological Survey Miscella-
neous Publication, 99-1, pp. 143–149.
Galton, P.M. 1973. The cheeks of ornithischian dinosaurs. Lethaia,
6: 67–89.
Godefroit, P., Pereda Suberbiola, X., Li, H., and Dong, Z. 1999. A
new species of the ankylosaurid dinosaur Pinacosaurus from the
Late Cretaceous of Inner Mongolia (P.R. China). Bulletin de
l’Institut des Sciences Naturelles de Belgique, Sciences de La
Terra, 69(supplement B): 17–36.
Hill, R.V., Witmer, L.M., and Norell, M.A. 2003. A new specimen
of Pinacosaurus grangeri (Dinosauria: Ornithischia) from the
Late Cretaceous of Mongolia: ontogeny and phylogeny of
ankylosaurs. American Museum Novitates, 3395, pp. 1–29.
Iverson, J.B. 1980. Anatomical modifications of the colon in lizards
of the subfamily Iguaninae. Journal of Morphology, 163: 79–93.
Iverson, J.B. 1982. Adaptations to herbivory in iguanine lizards. In
Iguanas of the world: their behavior, ecology and conservation.
Edited by G.M. Burghardt. Noyes Publications, New York,
pp. 60–76.
King, G. 1996. Reptiles and Herbivory. Chapman Hall, London, UK.
Maleev, E. 1954. Armored dinosaurs of Mongolia, Family Syrmo-
sauridae. Trudy Paleontologicheskogo Instituta Akademii Nauk
SSSR, 48: 142–170.
Maleev, E. 1956. Armored dinosaurs of Mongolia, Family Ankylo-
sauridae. Trudy Paleontologicheskogo Instituta Akademii Nauk
SSSR, 62: 51–91.
McBee, R.H., and McBee, V.H. 1982. The hindgut fermentation in
the green iguana, Iguana iguana. In Iguanas of the world: their
behavior, ecology and conservation. Edited by G.M. Burghardt.
Noyes Publications, New York, pp. 77–83.
Maryanska, T. 1977. Ankylosauridae (Dinosauria) from Mongolia.
Palaeontologia Polonica, 37: 85–151.
Molnar, R.E., and Wiffen, J. 1994. A Late Cretaceous polar dinosaur
fauna from New Zealand. Cretaceous Research, 15: 689–706.
Norman, D. 2001. The anatomy and systematic position of Scelido-
saurus harrisonii Owen, 1861. Journal of Vertebrate Paleontology,
21: 84A.
Osborn, H.F. 1905. Tyrannosaurus and other Cretaceous carnivorous
dinosaurs. American Museum of Natural History Bulletin, 21:
259–265.
Osborn, H.F. 1906. Tyrannosaurus, Upper Cretaceous carnivorous
dinosaur (second communication). American Museum of Natural
History Bulletin, 22: 281–296.
Osborn, H.F. 1916. Skeletal adaptations of Ornitholestes,Struthio-
mimus,Tyrannosaurus. American Museum of Natural History
Bulletin, 35: 733–771.
Papp, M.J., and Witmer, L.M. 1998. Cheeks, beaks, or freaks: a
critical appraisal of buccal soft-tissue anatomy in ornithischian
dinosaurs. Journal of Vertebrate Paleontology, 18: 69A.
Resnick, D., and Niwayama, G. 1983. Entheses and enthesopathy.
Radiology, 146: 1–9.
Romer, A.S. 1927. The pelvic musculature of ornithischian dinosaurs.
Acta Zoologica, 8: 225–275.
Sternberg, C.M. 1949. The Edmonton fauna and description of a
new Triceratops from the Upper Edmonton Member. National
Museum of Canada Bulletin, 113: 33–46.
Sternberg, C.M. 1951. Complete skeleton of Leptoceratops gracilis
Brown from the Upper Edmonton Member on the Red Deer
River, Alberta. Annual Report of the National Museum of Canada
Bulletin, 123: 225–255.
Thulborn, T. 1993. Mimicry in ankylosaurid dinosaurs. Record of
the South Australian Museum, 27: 151–158.
Tumanova, T. 1978. New data on the ankylosaur Tarchia gigantea.
Paleontological Journal, 1977: 92–100.
Tumanova, T. 1987. Armored dinosaurs of Mongolia. Joint Soviet–
Mongolian Paleontological Expedition Transaction, 32: 1–77.
Tumanova, T. 1993. A new armored dinosaur from southeastern
Gobi. Paleontological Journal, 27: 119–125.
Vickaryous, M.K., and Russell, A.P. 2003. A redescription of the
skull of Euoplocephalus tutus (Archosauria: Ornithischia): a foun-
dation for comparative and systematic studies of ankylosaurian
dinosaurs. Zoological Journal of the Linnean Society, 137: 157–
186.
Vickaryous, M.T., Russell, A.P., and Currie, P.J. 2001. The cranial
ornamentation of ankylosaurs (Ornithischia: Thyreophora): re-
appraisal of developmental hypothesis. In The armored dinosaurs.
Edited by K. Carpenter. Indiana University Press, Bloomington,
Ind., pp. 318–340.
Witmer, L.M. 1997. The evolution of the antorbital cavity of
archosaurs: a study in soft-tissue reconstruction in the fossil
record with an analysis of the function of pneumaticity. Journal
of Vertebrate Paleontology 17(1, supplement, Memoir 3): 1–73.