A DIAPSID REPTILE
FROM THE

PENNSYLVANIAN OF KANSAS

ROBERT R. REISZ

Department of Biology
University of Toronto

Erindale Campus

Mississauga, Ontario

Canada L5L 1C6

SPECIAL PUBLICATION
OF THE
MUSEUM OF NATURAL HISTORY, UNIVERSITY OF KANSAS

NUMBER 7

1981

1^\ EmstWIayrLfcrary

. ^ <) / UNIVERSIT^BI^'XANSAS PUBLICATIONS

I ' ( MUSEUM OF NATURAL HISTORY

Copies of publications may be obtained from the Publications Secretary, Museum
of Natural History, University of Kansas, Lawrence, Kansas 66045. A list of
available Special Publications is provided on the inside back cover.

University of Kansas
Museum of Natural History

Special Publication No. 7
February 18, 1981

A Diapsid Reptile From the
Pennsylvanian of Kansas

BY

Robert R. Reisz

Department of Biology
University of Toronto

Erindale Campus

Mississauga, Ontario

Canada L5L 1C6

University ok Kansas

Lawrence

1981

UNIVERSITY OF KANSAS PUBLICATIONS
MUSEUM OF NATURAL HISTORY

Editor: Joseph T. Collins
Editors for this issue: E. O. Wilev and Larry D. Martin

h/[r'7 Special Publication No. 7

LIBRARY pp. 1-74; 26 figures

1 table
II II o Q ?nnR Published February 18, 1981

- RD I

U SiTY

Copyrighted 1981

BY

Museum of Natural History

University of Kansas

Lawrence, Kansas 66045

U.S.A.

Printed by

University of Kansas Printing Service

Lawrence, Kansas

ISBN: 0-S933S-011-3

This is the first of a series of Special Publications honoring the contributions of
Dr. Theodore H. Eaton to the fields of vertebrate paleontology and zoology. The
series will center around the Pennsyl\anian fauna of eastern Kansas, and will
include papers on fishes, amphibians and reptiles, as befits the wide-ranging
interests Dr. Eaton has shown in his research.

E. O. Wiley
Labry D. Martin
15 June 1980
Lawrence, Kansas

CONTENTS

INTRODUCTION - 1

ACKNOWLEDGEMENTTS - 2

DESCRIPTn'E AXD SYSTEMATIC HiSTORY 2

SYSTEMATIC DESCRIPTION 4

Family Petrolacosauridae Peabody, 1952 4

Genus Petrolacosauriis Lane, 1945 4

Pefrolacosmirus kansensis Lane, 1945 4

OSTEOLOGY - - 5

Skull 5

Dermal Bones of the Skull Roof 9

Dermal Bones of the Palate 21

Ossifications of the Palatoquadrate Cartilage 23

Ossifications of the Braincase 24

Mandible 26

Dentitiox 27

Sclerotic Plates — 28

Vertebrate 28

Cervical Vertebrae 29

Dorsal Vertebrae 33

Sacral Vertebrae 34

Caudal Vertebrae 35

Ribs 36

Pectoral Girdle 38

Pelvic Girdle 40

Limbs .-. 43

PHYLOGENETIC REL.4TIONSHIPS OF PETROLACOSAURUS 54

Hypotheses of Relatio.vships 57

Basic Taxa 58

Shared Derived Characters Testing the Hypothesis of

Relationship Betv^^een Petrolacosaurus and Paleothyris 61

Shared Derived Char.acters Testing the Hypothesis of

Relationship Between Petrolacosaurus and Youngina 62

Petrolacosaurus Compared with Aracoscelis 64

HEARING IN PETROLACOSAURUS AND OTHER EARLY REPTILES .. 66

CONCLUSIONS 69

SUMMARY 71

LITERATURE CITED 72

INTRODUCTION

AH living reptiles can be grouped into
four orders: Chelonia, Rhynchocephalia,
Squamata and Crocodilia. The latter three
can be associated with the largest assemblage
of fossil reptiles, collectively called diapsid
reptiles. The diapsid condition refers to the
presence of two pairs of temporal openings
on the skull roof behind the orbits. This con-
dition has been retained in a primitive form
in crocodilians and in the sole living rhyncho-
cephalian Sphenodon ptinctatum but has been
modified in the squamates by the progressive
loss of dermal bones in the temporal region.

Diapsid reptiles have an evolutionary his-
tory long thought to extend only into the late
Permian. The early evolution of this assem-
blage is not well documented in the fossil
record. Consequently, the origins and phy-
letic relationships of member groups have
been subject to dispute.

On several occasions, Watson ( 1951, 1954 )
reiterated his belief in Goodrich's concept of
a basic dichotomy among reptiles. According
to this theory, one group called the Therop-
sida led to mammals and included the mam-
mal-like reptile orders Pelycosauria and The-
rapsida. The other group, the Sauropsida,
also called "true" reptiles, included the diap-
sid reptiles and the turtles (Chelonia). Wat-
son argued for independent origins of the
Theropsida and Sauropsida from anthraco-
saurian amphibians via some unknown inter-
mediate forms. He based this conclusion on
his theory of the evolution of the ear (Wat-
son, 1953). He discounted the possibility of
captorhinomorphs having given rise to the
Sauropsida and, hence, to the diapsid rep-
tiles. He actually placed the Captorhino-
morpha among the Theropsida. In 1957 Wat-
son suggested that the diapsids evolved from
seymoriamorph anthracosaurs via the Millero-
sauria.

Vaughn ( 1955 ) also accepted the saurop-
sid-theropsid dichotomy and felt that the

Sauropsida, and therefore diapsids, could not
be derived from captorhinomorphs.

Romer (1966) rejected Watson's theory of
the evolution of the ear and reiterated his
belief that all true reptiles can be derived
from primitive captorhinomorphs. In a re-
view of early reptilian evolution, Romer
(1967) suggested not only that the sauropsid-
theropsid dichotomy, insofar as it exists, has
little phylogenetic significance, but also that
the term Sauropsida has no systematic or
phylogenetic meaning. Furthermore, he em-
phasized that the diapsids may not belong to
a single phyletic unit. He argued that two
discrete diapsid groups should be recognized,
the Archosauria and the Lepidosauria. He
proposed that the Lepidosauria (the squa-
mates, rhynchocephalians and the ancestral
eosuchians) on the one hand, and the Archo-
sauria (the two dinosaur orders, pterosaurs,
crocodilians and the basic Triassic group, the
thecodonts) on the other, evolved separately
from the ancestral captorhinomoqjhs. It was
his suggestion that the miileretid skull pattern
represents an intennediate stage between the
ancestral captorhinomorphs and the lepido-
saurs. '

In 1967 Reig set forth, and later (1970)
elaborated, a radically different hypothesis for
the origin of archosaurs. He proposed that
the ancestors of archosaurs should be sought
among varanopsid pelycosaurs. Most of the
comparable features cited by Reig as indica-
tive of phylogenetic relationships are, how-
ever, primitive characteristics found in the
ancestral captorhinomorphs (Romer, 1971).

Gow (1972), in his reconsideration of the
milleretids, contended that this group gave
rise directly to the Squamata. He felt on the
other hand, that the Rhynchocephalia and the
Archosauria had an entirely independent ori-
gin.

The above workers have based their theo-
ries of the origins and early evolution of the

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

diapsid reptiles mainly on the meager evi-
dence provided by the known Permian coty-
losaurs and on the Upper Permian and Trias-
sic diapsids. Thanks mainly to the labors of
Robert L. Carroll and the late Frank E. Pea-
body, the problem of the origins and early
evolution of diapsid reptiles can be reconsid-
ered. Carroll, during a period of about ten
years, described and reconsidered all of the
available Pennsylvanian and Lower Permian
specimens of primitive captorhinomorphs. One
significant conclusion became evident: the
described members of this group conform to
a single conservative morphological pattern
and remained suflBciently generalized to indi-
cate that i^rimitixe captorhinomoqohs could
be ancestral to most subsequent reptilian line-
ages (Carroll and Baird, 1972; Clark and Car-
roll, 1973).

In 1952, Peabody published a partial de-
scription of Petrokicosaiinis kansensis and
suggested that this reptile from the LTpper
Pennsylvanian was a member of the most
primitive diapsid order, the "Eosuchia." The
fragmentary nature of the specimens available
for this study led others, however, to question
Peabody's contention. Subsequently Peabody
spent two grueling summers — 1953 and 1954
— collecting more specimens but died before
he could undertake a reconsideration of the
anatomy of Petrolacosatirus. This collection
fonus the basis of the present study. These
specimens show that Petrolacosaiirus is a
diapsid reptile with affinities to both primitive
captorhinomorphs and the earliest "eosuch-
iaiis."

My interest in Petrolacosaiirus was aroused
during a visit to the Kansas University Mu-
seum of Natural History in Lawrence, Kansas,
where T. H. Eaton invited me to examine this
interesting reptile. It became apparent that
the available specimens of this genus, much
superior to those described in 1952, warranted
a thorough reeonsideration. The purpose of
the present study, undertaken with Dr. Eaton's
kind permission, is to describe the earliest
known diapsid reptile, Pctrohicosaurus kansen-
sis, and to consider its relationship to both

primitive captorhinomorphs and to more ad-
vanced diapsid reptiles. The problem of hear-
ing in Petrolacosaurus and other early reptiles
forms an integral part of this study.

Acknowledgements

I am greatly indebted to T. H. Eaton of
the University of Kansas Natural History Mu-
seum, at whose suggestion this study was
undertaken and who kindly allowed me to
borrow the Petrolacosaurus specimens at his
disposal. I am particularly grateful to Dr. and
Mrs. Eaton for courtesies extended to mc
during the time spent studying the Kansas
collection.

Thanks are due also to R. L. Carroll whose
wise guidance, friendliness, and unrelenting
support were invaluable. I am very grateful
to P. L. Robinson of University College for
her many valuable and instructive sugges-
tions, for numerous stimulating conversations,
and for making the time spent studying in
London so pleasant. I ha\'e profited also from
several interesting discussions with Donald
Baird of Princeton University and Malcolm
Heaton of University of Toronto.

Thanks are due to Bobb Shaeffer and E. S.
Gaffney, both of the American Museum of
Natural History, New York, for extended
loans of specimens.

DESCRIPTIVE AND SYSTEMATIC
HISTORY

Lane first described Petrolacosaurus kan-
sensis in 1945 and referred the genus to the
family Si)henacodontidae of the order Pely-
cosauria ( 1946) on the basis of a tarsus found
with an isolated hindlimb and peKis. An
isolated iorelimb and some other skeletal re-
mains were- described as Podar'^osaurus hih-
hardi and referred to the family Araeoscelidae
of the order Protorosaiiria ( 1946) on the basis
of the slender humerus, radius, ulna, and the
ribs.

In 1952 Peabod\ publisiicd a detailed de-
scription of (he CJarnett reptile and placed
Podargosaurus in synon)iny with Petrolaco-

1980

PENNSYLVANIAN DIAPSID REPTILE

saurus. He concluded that "Petrolacosaurus
represents a new family of primitive reptiles
showing relationships which place it at the
base of the Eosuchia while also evidencing
strong relationship with primitive cotylo-
saurs." The fragmentary nature and imma-
turity of the known specimens led others to
reject his conclusions. Peabody's association
of an isolated edaphosaurian pelvis (KUVP
1425) with the Petrolacosaurus material add-
ed to the confusion.

Watson ( 1954 ) , after an examination of
the known specimens concluded that this
genus was a theropsid. He based his conclu-
sion on the nature of the quadrate in the
small, immature skull. He felt that this quad-
rate was similar to those in pelycosaurs and
captorhinomoiphs.

Although he reserved final judgment until
a more complete description was undertaken,
Vaughn (1955) was inclined to agree with
Watson's diagnosis. He also suggested that
Petrolacosaurus may be related to Araeoscelis

and that the former genus may become the
type of a new family of Araeoscelidia.

Tatarinov (1964) reassigned the Family
Petrolacosauridae to the Order Araeoscelidia;
he based his conclusions on the similarities
seen in the palate and postcranial skeleton of
Petrolacosaurus and Araeoscelis. Tatarinov
followed Romer (1956), however, in placing
the Araeoscelidia within the Subclass Eury-
apsida.

Romer (1956) tentatively placed Petrola-
cosaurus within the Family Ophiacodontidae
(Pelycosauria). In 1966 in a joint paper with
Stovall and Price, Romer maintained that
"Apart from the possible but unproven diap-
sid nature of the temporal region, there is no
reason to assign Petrolacosaurus to the Eo-
suchia." He felt, however, that certain diag-
nostic features in the post-cranial skeleton
". . . strongly indicate that this genus belongs
to a group of archaic edaphosaurians from
which both Edaphosaiinis and, at a much
later time, tlie caseids may have arisen."

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

SYSTEMATIC DESCRIPTION

Family PETROLACOSAURIDAE Peabody,
1952'

Genus Petrolacosauriis Lane, 1945

Type species. — Petrolacosaurus kansensis
Lane, 1945.

Si/noni/m. — Podarg^osaurus Lane, 1945.

Occurrence. — Upper Pennsylvanian of
Kansas.

Diagnosis. — Same as for species.

Petrolacosaurus kansensis Lane, 1945
Figs. 1-25

Synonym. — Podargosaurus hibbardi Lane,
1945.

Revised diagnosis. — Medium sized early
diapsid reptile (Fig. 1) with well developed
superior and inferior temporal fenestrae and
elongate, narrow suborbital fenestra. Parietal
without the ventrolateral flange that extends
into the superior temporal fenestra in younger
diapsids; posterior splenial bone present in
mandible; marginal dentition unusually thin-
walled. Six elongate cervical vertebrae; twen-
ty six presacral vertebrae; a pair of mam-
millary processes on the neural spine of first
sacral vertebra; large dorsal isciadic notch;
slender forelimbs equal in length to more
massive hindlimbs; propodials equal in length
to epipodials.

Distinguished from captorhinomorph rep-
tiles by relatively smaller skull, superior, in-
ferior temporal fenestrae and suborbital fe-
nestra, greater length of neck, partial rather
than complete coossification of atlantal cen-
trum and axial intercentrum, mammillary

' The family Petrolacosauridae has been formerly
included in the order Eosuchia (Pcal)ody, 1952; Rcisz,
1977). Study of the ahove order indicates tliat it is
an artificial assemblage of Late Paleozoic and Mcso-
zoic diapsid reptiles. The "Eosuchia" has no ta.xo-
nomic validity because there are no known diag-
nostic features that can distinf^i'sh it from other diap-
sid reptiles. The diagnostic features of the mono-
generic family Petrolacosauridae are the same as for
the genus.

processes on neural spines and excavation of
neural arches of dorsal vertebrae, longer tail,
and greater relative length and slenderness of
limbs.

Separated from Araeoscelis by the inferior
temporal fenestra, more lightly built skull and
dentition and straight ventral margin of
cheeks.

Differs from other diapsid reptiles by sim-
ple lateral cmbayment of parietal rather than
a ventrolateral flange in addition to the em-
bayment, larger lacrimal and squamosal,
larger quadratojugal, lack of retroartieular
process, denticulate parasphenoid and mas-
sive stapes.

HoloUjpe. — KUVP ( Kansas University' Mu-
seum of Natural History) 1424, adult right
hindlimb consisting of most of the femur, the
distal part of the tibia and fibula in articula-
tion with a complete pes.

Hypodigm. — KUVP 1423, complete right
forelimb, immature. KUVP 1426, right hind-
limb lacking femur, immature. KUVP 1427,
splenial, \ertebral column, scattered ribs, fore-
limbs, partial pectoral girdle, partial hind-
limb, immature. KUVP 142S, questionable
skull fragment, partial maxilla, partial verte-
bral cohnnn, scattered ribs, incomplete fore-
limbs and hindlimbs, partial pelvic girdle,
scattered gastralia, immature. KU\T 1429,
left forelimb, immature. KUVP 8351, partial
skull, five cervical vertebrae, immature. KUVP
8,355, right carpus, immature. KU\"P 9950,
isolated right parietal, mature. KUVP 9951,
partial skull, vert<>bral column, ribs; on sepa-
rate blocks femora and peKes, left lower hind-
limb, right lower hindlimb, slightly immature.
KU\'P 9952, partial .skull (mainly dermal
skull roof), mature. KUVP 9956, a series of
articulated \ertebrae from C3 to S2, with
some closely associated ribs, a short series of
articulated caudal vertebrae, mature. KUVP
9957, three separate blocks containing left
scapulacoracoid with hinnerns, radius and
ulna in articulation and partial carpus, slight-

Fig. 1. — Petwlacosaunis kansensis. Reconstruction of skeleton, X 0.75. Composite.

c

1980

PENNSYLVANIAN DIAPSID REPTILE

ly immature. KUVP 995S, left clavicle, slightly
immature. KUVP 9959, seven separate blocks
containing badly preserved subadult skull,
interclavicle, immature, lower jaws, scat-
tered \ertebrae and ribs, forelimbs, hindlimbs,
both ilia, very immature. KUVP 9961, left
pelvis, mature. KUVP 9962, ilium with first
sacral rib attached, immature. KU\T 10320,
badly preser\ed specimen containing skull
fragments, vertebrae, hindlimbs, immature.
KUVP 33602, isolated left maxilla, slightly
immature. KUVP 33603, fragments of skull
material, distal end of humerus, immature.
KUVP 33604, scattered skull material, atlas-
axis complex, cleithrum, distal end of hu-
merus, radius, ulna, mature. KUVP 33605,
seven posterior dorsal vertebrae and first sac-
ral vertebra, first sacral rib, all articulated,
slightly immature. KUVP 33606, partial skull,
posterior dorsal, sacral and anterior caudal
vertebrae, some associated ribs, clavicle, par-
tial scapulacoracoid, humerus, partial pelvis,
right hindlimb with scattered pes, mature.
KUVP 33607, slightly dissociated skull, first
three cervical vertebrae, cleithrum, right
scapulacoracoid, olecranon, partial carpus,
mature. KUVP 33608, poorly preserved skull
and cervical vertebrae. KUVP 33609, right
scapulacoracoid, mature.

OSTEOLOGY

Skull

By using the available material, it has
been possible to obtain a composite, fairly
accurate reconstruction of the skull, in which
most of the external details of the dermal
roof, occiput, palate, the lateral and medial
aspects of the mandible can be discerned. The
quality of preservation prevents any recon-
structions of the braincase beyond those seen
in Figs. 3b and 4.

The skull as reconstructed has the princi-
pal dimensions listed in Table 1. These meas-
urements indicate that Petrolacosaurus has
retained many of the primitive cranial pro-
portions seen in primitive captorhinomorphs

but also shows a number of significant de-
partures from this pattern.

In dorsal aspect (Fig. 3a) the skull is long
and relatively narrow. The width at the quad-
rate exceeds, only slightly, one-half the length
of the skull. Maximum width is reached just
behind the orbits in the region of the tem-
poral openings. In the antorbital region the
skull table extends far laterally, obscuring the
posterior part of the narial opening and the
maxilla when seen in dorsal view. In the
postorbital region both temporal fenestrae are
visible from above. The summit of the skull
(in cross-sectional view) in this area is at the
level of the medial suture of the postfrontal
wedge. Laterally the skull roof slopes gently
down to the level of the upper temporal
fenestra. From here the grade of the down-
ward slope increases greatly. The curvature
between the skull table and the cheek in the
temporal region is thus formed by the ventral
part of the postfrontal that lies beneath the
postorbital and the dorsal part of the post-
orbital and squamosal. The posterior margin
of the skull table is deeply emarginated where
the dorsal aspect of the parietal ends.

The antero-posterior slope of the skull
table is illustrated in Fig. 2a. The skull is
long and shallow in lateral \iew. The maxi-
mum height, reached in the orbital region
represents only about 35 per cent of the
length of the skull. The gently convex pos-
terior margin of the skull is nearly perpen-
dicular to the ventral margin. The jaw articu-
lation lies only slightly below the level of the
tooth row. The marginal tooth row occupies
58 per cent of the skull length.

The outline of the palate is defined by the
\entral edges of the cheeks and by the pos-
terior margin of the braincase (Fig. 3b). The
internal nares, bounded by the premaxilla,
maxilla, palatine and vomer, are long and
narrow. Both the vomer and palatine extend
ventrally at the margin of the opening form-
ing ridges. The interpterygoid vacuities are
of moderate size. The suborbital fenestrae
are well developed, elongate openings, albeit
smaller than in Youngina (Olson, 1936; Gow,

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Fig. 2. — Pctwlacosaurti.i kansensis Lane. Rpconstniction of slaiU and niandihle in lateral view (A), and of
mandible in medial view (B) based mainly on specimen.s KUVP 9952 and KUVP 33607, X 2. Abbreviations
u.sed in Figs. 2-13: a, angnlar; ae, atlantal centrum; an, atlantal neural arch; art, articular; a.\, ;i.\ial centrum;
a.\i, axial intercentrum; bo, basioccipital; bs, basisphenoid; bt, basipter\goid process; co, coronoid; ct, cultri-
form process; d, dentary; ec, ectopterygoid; ex, exoccipital; f, frontal; j, jugal; 1, lacrimal; m, maxilla; n, nasal;
o, opisthotic; p, parietal; pf, postfrontal; pi, palatine; pm, preniaxilla; po, postorbital; pop, paroceipital process;
pp, postparietal; pra, prearlicniar; prf, prefrontal; i^ro, proolic; ps, parasphenoid; psp, postsplenial; pt, pterygoid;
q, quadrate; qf, (juadrate foramen; qj, (juadratojugal; rl, atlantal riii; r2, axial rib; s, stapes; sa, surangular; sc,
sclerotic plates; sni, septomaxilla; so, supraoccipital; sq, scpiamosal; st, supratemporal; t, tabular; tr. fl. pt,
transverse flange of pterygoid; v, vomer.

1975). With the exception of the ectoptery-
goid, the paired palatal l)Oiies hear denticles;
in addition, three major rows of large teeth
radiat(> laterally, diagonally, and anteriorly
from the hasicranial artienlation. Small den-
ticles arc also present on the parasphenoid.
The palate is anchored directly to the dermal
sknll roof at three points: anteriorly to the
premaxilla and laterally to the maxilla and
to the jngal. Posteriorly, the palate is indi-
rectly anchored to the skidl roof by the (jnad-
rate and the braincase. The oceijiital condyle
is set only slightly anterior to the posterior

margin of the jaw articulation. The subtem-
poral fossae are quite large, occupying about
23 per cent of the palatal area.

The overall impression from the occipital
aspect (Fig. 4) is of a fairl\- low, wide skull.
Part of the skull roof is \isible in this view.
\ well developed nuchal ridge extends from
the dorsal occipital border to the foramen
uKignum, which has the same height as width.
The post-temporal fenestrae are large. The
occipital cond\le is located low on the occipi-
tal face, only 4 nun Irom the line joining the
\entral borders of the (juadratojugals and

1980

PENNSYLVANIAN DIAPSID REPTILE

B

Fig. 3. — Petrolacosaurus kansensis Lane. Reconstruction of skull in dorsal (A), and ventral (B) views,
based mainly on specimens KUVP 9952, 33606 and 33607, X 2. See Fig. 2 for key to abbreviations.

about 13 mm below the posterior edge of the
parietal.

The largest openings in tlie skulls are the
orbits. Each occupies 38 per cent of the total
skull length. The orbits incise the skull table
deeply.

The upper temporal fenestra faces mainly
dorsally, but is also visible in lateral view
because of the curvature of the circumtcm-
poral bones. The lower temporal fenestra, on
the other hand, faces mostly laterally, but is
also visible from the dorsal aspect. Its pos-
teroventral corner is not preserved. The post-
orbital bar, composed of parts of the post-

frontal, postorbital, and jugal is slender; the
orbit and temporal fenestra are, therefore, not
widely separated.

The external naris is relatively large in
relation to the skull. This opening faces an-
terolaterally and only slightly dorsally. The
external naris is floored laterally by a shelf
formed by the premaxilla and maxilla.

A large pineal foramen opens dorsally on
the midline, approximately at the level of the
anterior margin of the upper temporal fe-
nestra.

The lower jaw, as reconstructed (Fig. 2),
has the principal dimensions listed in Table 1.

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY NO. 7

Fig. 4. — Petrolacosaurus kansensis Lane. Reconstruction of skull in occipital \ie\v, mainly based on speci-
mens KUVP 9951 and 33607, x2. See Fig. 2 for key to abbreviations.

Table 1. — Measurements of the skull and lower jaw
of Petrolacosaurus kansensis, made primarily on the
basis of the restoration of KUVP 33607. Measure-
ments of the captorhinomorph Protorothyris archeri
are based mainly on the type specimen (Clark and
Carroll, 1973). Measurements in millimeters.

Protorothyris

Petrolacosaurus

archeri

karuensis

(MCZ 1532)

Length of skull

59

56

Width of skull in the

temporal region

33

31

Height of skull

at the orbit

20

20

Length of tooth row

33

37

Length of orbit

22

16

Antorbital length

21

22

Temporal length"

13

17

Length of

internal naris

17

15

Length of

sviborbital fenestra

11

Length of

subtemporal fossa

20

20

Length of lower jaw

57

54

Length of tooth

row on lower jaw

30

34

Height of lower jaw

10

10

Length of

adductor fossa

16

15

Height of

adductor fossa

5

7

° Temporal length is measured from the posterior end
of the supratemporal to the orbit.

It is a long, slenderly built structure, very nar-
row in height and width. The tooth row
occupies 53 per cent of the length of the

mandible. The coronoid area is gently convex,
expanding dorsally well above the level of the
alveolar shelf. Posteriorly the dorsal border
curves slightly ventrally to the level of the
articular. There is no retroarticular process.
The ventral border of the ramus is nearly
straight. The adductor fossa is long, occupy-
ing 30 per cent of the length of the mandible.
Between its anterior, strongly acute angle
(formed by the coronoid) and its wide, gently
concave posterior end (formed by the articu-
lar) the fossa expands posteriorly between the
prearticular and surangular. The foramen
intermandibularis caudalis is not e\ident, but
this may be due to partial disarticulation of
tiio adjacent bones in the mandible.

With the possible exception of the max-
illa, lacrimal and (juadratojugal, all the bones
of the dermal skull roof are sculptured. The
hones are marked by slight anastamosing
groo\es and ridges that radiate from the cen-
ter of ossification. On the cheek bones sculp-
tin-ing is restricted to the area immediately
surrounding the centers of ossification. The
roofing bones, on the other hand, are almost
complctly covered by sculpturing. In a num-
ber of specimens sculpturing appears to be
more extensive than on others. Preparation of
the inner surface of these skulls reveals that
some sculpturing is apparent even on the ven-
tral aspect of the bones. Possibly, in these
specimens, crushing was so severe that the

1980

PENNSYLVANIAN DIAPSID REPTILE

normal sculpturing was exaggerated by the
superposition of the internal bone structure
onto the surface. On the lower jaw, only the
anterior one-third of the dentar\' is sculptured.
Near the tip the lateral surface of the dentary
has delicate rounded pitting. More poste-
riorly the pitting takes on the shape of elon-
gate grooves.

Exposure of the internal surface of the
skull roof of Petrolacosaurus revealed the
presence of a series of ridges or thickened
areas along the major lines of cranial stresses.
These ridges (Fig. 25) form a framework of
strengthened areas that connect the orbital
margin with the snout, maxilla and suspen-
sorium. What makes the ridges seen in
Petrolacosaurus significant is their relative
thickness when compared to the much thinner
areas of the dermal skull roof and palate. The
use of the resin imbedding compound during
preparation of the skulls made complete re-
moval of the matrix possible. This revealed
that in the non-ridged areas the skull is made
of very thin bone. The anteromedial part of
the prefrontal, the medial half of the nasal,
the anteroventral portion of the squamosal
and most of the quadratojugal, are less than
one-tenth the thickness of the ridged areas
of the same skull. Such differences in the
thickness of the skull elements cannot be
attributed to some unusual type of preserva-
tion because other genera, found in the same
deposits do not show such variation in bone
thickness. In Captorhinus and Dimetrodon
(Heaton, 1979; Romer and Price, 1940),
where the external and internal surfaces of
the skull are known in great detail, the dif-
ferences in bone thickness do not approach
the range seen in Petrolacosaurus. The in-
ternal surfaces of the skull roof and palate in
Permian and Lower Triassic "eosuchians" are
inadequately known for comparative pur-
poses.

Dermal Bones of the Skull Roof

Premaxilla. — The paired premaxillae form
the anterior margin of the skull. Each ele-
ment has three slender processes that connect

the dorsal, lateral and palatal sides of the
skull to constitute the region of the snout.

The unusually slender nasal ramus of each
premaxilla, exposed in both external and in-
ternal views in KUVP 9952, join at the mid-
line to form the anterior border of the snout.
The dorsal tips appear as slight posterior
projections on the dorsal surface of the skull
and fit into the bifurcated anterior ends of
the nasals. The dorsal process reaches its
greatest width at the point where it becomes
visible on the superior surface of the skull.

The maxillary ramus contributes to the
ventral border of the external naris and meets
the maxilla in a long diagonal suture. In
dorsal view the maxillary ramus gives a
rounded outline to the snout.

The palatal ramus forms a continuous
shelf anteriorly that extends between the ex-
ternal nares; it also forms the rounded an-
terior border of the internal naris. At the
midline the premaxilla forms a wedge-like
posterior vomerine process best seen in KUVP
8351 (Fig. 5), that fits into a corresponding
slot in the vomer.

Anteriorly and laterally, the premaxilla is
sculptured and is pierced by small foramina.
In lateral view, the generally slender appear-
ance of the premaxilla is accentuated by the
height and extreme slenderness of the dorsal
process and by the emargination of the maxil-
lary process, where a medially directed flange
provides support for the nasal capsule. There
are from three (KUVP 9952) to five (KUVP
33607) slender, conical premaxillary teeth on
either side of the midline.

The premaxillae of the protorothyridids
Hijlonomus, Paleothijris, and Protorothyris are
similar to but slightly more massive than
those of Petrolacosaurus. Reconstructions of
these protorothyridids (Carroll and Baird,
1972; Clark and Carroll, 1973) show broad
butt joints between the premaxillae and vo-
mers. This is unlikely, especially since the
anterior end of the vomer in Paleothijris is
narrow and bifurcate. The vomerine process
of the premaxilla is poorly known in protoro-
thyridids.

10

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Fig. 5. — Pctrolacosaurus kanscnsis Lane. Immature skull and lower jaws, KUVP 8351, X 4. (A) Dor-
sal view of palate exposed by loss of skull roof, oblique view of lower jaws. (B) Ventral view of palate and
oblique view of lower jaws. For key to abbreviations, see Fig. 2.

Maxilla. — As in primitive captorhino-
morphs and "(-o.suchian.s," the maxilla is the
longest hone of the skull roof. It extt-nds from
the midpoint of the external naris past the
midpoint of the orbit, over .53 per ecnt of the
skull length. It forms the posteroxentral bor-
der of the external naris but is exeluded from
the orbit by the lacrimal and the jngal
(KUVP 9951, Fig. 6). Anteriorly, the max-

illa is .suturally attached to the premaxilla
o\er a large surface area, as indicated by the
rugose area on the anterior medial surface
of the maxilla in KU\P 33602 (Fig. 9). To-
gether with the premaxilla, the maxilla forms
not only the ventral border of the external
nares but also a medially directed shelf for
the nasal capsule.

Posteriorly, the bone expands dorsally to

1980

PENNSYLVANIAN DIAPSID REPTILE

11

Fig. 6. — Pctrolacosaurus kansensis Lane. Mature skull and caudal vertebrae, partially scattered, KUVP
9951, X 2. Braincase of same specimen shown in Fig. 7. See Fig. 2 for key to abbreviations.

reach its greatest vertical expansion immedi-
ately behind the level of the caniniform teeth,
one-third of the distance from the anterior
end of the maxilla. From this summit, the
dorsal border of the maxilla slopes gently
posteroventrally in an almost straight line.
The lateral surface of the bone is pierced by
small labial foramina, probably for cutaneous

branches of the superior alveolar nerve and
the maxillary artery. The lower margin is
nearly straight, as in primitive captorhino-
morphs and younginid "eosuchians."

The internal surface of the maxilla is more
important taxonomically than the lateral. As
in all primitive reptiles, the simple conical
subthecodont maxillary teeth are attached to

12

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Fic. 7. — PetTolacosaums kansensis Lane. Braincase of adult individual, KUVP 9951, X 2. Braincase ex-
posed in ventral (A) and dorsal views (B), with the lateral, occipital and ventral elements compressed into a
single plane. See Fig. 2 for key to abbreviations.

the alveolar shelf. In KUVP 9952, the entire
left maxilla is exposed. Thirty teeth are in
place, with room for five others. Laterally
the teeth arc covered by a thin sheet of bone
that extends up to one-third of the length of
the teeth below the alveolar shelf. The me-
dial surface of the alveolar shelf is smooth in
the region where it forms the lateral border
of the internal naris and the lateral border of
the suborbital fenestra. Between these two
openings, directly beneath the anterior mar-
gin of the orbit, the alveolar shelf bears
a small striated process, that is exposed in
both KUVP 9952 (Fig. 8b) and KUVP
33602 (Fig. 9a). This process which occupies
only 17 per cent of the length of this bone,
is sutured to the palatine. It is significant
that, as in all "eosuchians," the maxilla in
Petrolacosaiirus is attached directly to the
palate only through this jirocess. Posterior to
this contact with the palatine, the maxilla
forms the lateral border of the suborbital fe-
nestra. The maxilla is not attached to the
ectoptcrygoid. In primitive captorhinomorphs
and pelycosaurs, on the other hand, the max-
illa is sutured behind the internal naris (more
than 50 per cent of the total length of the

bone) to both the palatine and the ectoptcry-
goid.

There is no buttressing on the medial sur-
face of the maxilla, above the caniniform
teeth. This area is somewhat strengthened,
however, by the ventral margin of the lacri-
mal which fits into a long longitudinal groove
along the dorsal margin of the maxilla. The
cross-section of the maxilla (Fig. 9b) in the
region of the canines shows that the alveolar
shelf is hollowed out from above, giving the
maxilla the outline of an inverted letter "h."

Septomaxilla. — The septomaxilla, pre-
served only in KUVP 9952 (Fig. 8), is a thin
sheet of bone whose shape and relationships
cannot be determined because of crushing.
It was probably attached to the anterior edge
of the maxilla. In most primiti\'e reptiles the
septomaxilla, when preserved, appears as a
thin rectangular sheet of bone that has been
curved around a simple cone (Heaton 1979).
Only in pelycosaurs is the septomaxilla robust
with a thickened base and a large vertical
sheet placed transversely in the posterior por-
tion of the external naris (Romer and Price
1940).

LacrinuiL— The lacrimal in KUVP 33607

1980

PENNSYLVANIAN DIAPSID REPTILE

13

Fig. 8. — Petrolacosaiinis kanxctuis Lane. Mature skull, KUVP 9952. The e.\ternal (A) and internal (B)
surfaces of the skull talilc and left cheek are shown, but the preserved part of the right cheek is exposed only
in lateral view. See Fig. 2 for key to abbre\iations.

(Fig. 12 and 13) i.s a large shoet of bone tliat
extends from the naris to the orbit. From its
broad contribution to the narial border, the
laterally exposed surface of the lacrimal ex-
tends slightly upward, increasing its width
posteriorly, and forming the anteroventral
border of the orbit. Its suborbital ramus ex-
tends far posteriorly and makes contact with
the anterior end of the jugal. The ventral
border of the lacrimal is overlapped by the
maxilla from the region of the canines to the
posterior tip of the bone, as seen in the right

side of the skull in KUVP 33607. Conse-
quently, less than half of the height of the
suborbital ramus is seen in lateral view.

The right lacrimal, in KUVP 33607, shows
three foramina that open into the orbit. Two
large lacrimal puncti are situated in a deep,
conjunctival groove near the lateral border
of the lacrimal. The groove is partially ex-
posed in lateral view. A much smaller fora-
men is visible on the medial surface of the
suborbital ramus that may have carried the
anterior orbital artery, much as in the primi-

14

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY NO. 7

lb

C

Fig. 9. — Petrolacosaums kanscnsis Lane. (A) Isolated left maxilla, KUVP 33602, exposed in medial view,
X 2. (B) Section through maxilla at b-b, X 5. (C) Left parietal from a poorly preserved skull, KUVP
9959a, exposed in ventral view, X 2. (D) Quadrate and quadratojugal from poorly preserved, immature skull,
KUVP 33603, exposed in partial ventral view, X 2. See Fig. 2 for key to abbreviations.

tive captorhinid Eocaptorhinus laticeps (Hea-
ton, 1979). The lacrimal canal is completely
enclosed within the lacrimal bone.

The orbital margin of the lacrimal is
strongly buttressed. At the level of the an-
terior border of the orbit, the lacrimal reaches
the alveolar shelf of the maxilla and reinforces
the maxillary process where it is sutured with
the palatine. This is best seen in Fig. 8,
where the suborbital ramus of the right lacri-
mal is exposed in medial view.

Nasah. — The nasals, exposed both ven-
trally and dorsally in KUVP 9952 and 33607,
are paired bones lying along either side of
the midline of the skull, between the pre-
maxillae and the frontals; they form part of
the skull roof in the antorbital area. The pair
are sutured to each other along most of their
length. Each nasal has a posterior process
that is well preserved in KUVP 9959b, which
diverges from the midline and covers part of
the frontals dorsally. Anteriorly, the bone
curves strongly down to receive the dorsal

process of the premaxilla. The remaining part
of the anterior border is free and forms the
dorsal border of the external naris. Laterally,
the nasal is sutured to the lacrimal and the
prefrontal. The lacrimal fits over a depression
on the nasal, seen in KUVP 9952 (Fig. 8),
whereas the prefrontal is attached to the ven-
tral surface of the nasal.

The nasal is transversely convex (in cross-
section) in the region where it is attached
to the lacrimal, forming a dome over the nasal
capsule. On the ventral surface, a prominent
ridge runs anterolaterally from the posterior
border.

Prefrontal. — The prefrontal, seen in lateral
view in KUVP 9952 and 33607 and in dorsal
view in KUVP 33606, is a triangular bone
with an anterior plate that forms a wedge
between the nasal and the lacrimal, and a
thickened posterior ridge that forms the an-
terodorsal border of the orbit.

The anterior plate is a thin dorsoventrally
convex sheet of bone, partially overlapped by

1980

PENNSYLVANIAN DIAPSID REPTILE

15

Fig. 10. — Petrolacosaurus kansensis Lane. Skull and lower jaws of young adult, KUVP 33608, X 2. The
slightly disarticulated palate and braincase are e.xposed in dorsal view by loss of skull table. Part of the
cheeks and the left parietal are exposed in internal view. See Fig. 2 for key to abbreviations.

the lacrimal and nasal. A heavily .sculptured
ridge that runs from the anterior plate to the
orbital margin separates the lateral and dorsal
aspects of the skull in the antorbital region.
A heavy posterior thickening of the prefrontal
forms the anterodorsal angle of the orbit.

The thickened dorsal orbital rim portion
of the prefrontal extends far posteriorly and
is sutured medially to the frontal in a well
developed tongue and groove joint. The dor-
sal orbital margin is rugose and probably
ser\'ed as an attachment for part of the or-
bital fascia.

The prefrontal extends medially along its
anterior orbital rim portion to form the verti-
cal edge to which the orbitonasal membrane

was probably attached. A well developed
ventral process of the prefrontal orbital ridge
is strongly sutured to the medial surface of
the thickened orbital rim portion of the lacri-
mal. The prefrontal is heavily sculptured near
the orbital rim.

Frontal. — The frontals are the third in the
series of paired bones lying along the midline
of the skull. Each is a long prominent bone
that lies on the dorsal surface of the skull.
In KUVP 33606 (Fig. 11) the two frontals are
joined to each other along most of their
length in a slightly wavy tongue and groove
suture. Each is bounded anteriorly by the
nasal and laterally by the prefrontal and post-
frontal. Between the latter two elements the

16

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Fic. 11. — Pctrolacosaurus kanscnsis Lane. Mature .';!ai!l and left lower jaw, KU\'P 33606, X 2. Part of
the dermal skull roof is exposed in donsal view, in almost perfect association; partial left lower jaw is seen
in lateral view. See Fig. 2 for key to abbreviations.

frontal forms the orbital margin, and is slight-
ly rugose to provide better attachment for the
orbital fascia. Posteriorly, the frontal inter-
digitatcs with the anterior margin of the
parietal. The frontal forms a wedge, postero-
laterally, which partially separates the post-
frontal and jiarietal. The suture between the
postfrontal and the frontal is also of the
tongue and groove type, similar to the suture
between the prefrontal and frontal. In lateral
view, the frontal is convex in outline, follow-
ing the curve of the dorsal margin of the
orbit. In cross-section, the frontal is slightly
convex over the orbit. As seen in KUVP 9952
(Fig. 8b), the frontal has a strongly devel-
oped ridge on its ventral surface that runs the
length of the bone, roughly parallel to the
midline. It is to this ridge that both the pre-

frontal and the postfrontal are attached in
order to brace the anterodorsal and postero-
dorsal corners of the orbit. This ridge is most
pronounced over the orbit.

Parietal. — The parietal in Pctrolacosaurus
is readily distinguishable from that of capto-
rhiuomorphs and pelycosaurs because it forms
part of the border of a well developed su-
perior temporal fenestra. It resembles the
parietal of other "eosuchians" and of arae-
oscelids. The description given here is based
on well preserved parietals in KUVP 9951,
9952, 9959a, 33606, and 33607, and on isolated
single elements in KUVP 9950, 33603, and
33604. Along the anterior part of its lateral
border the parietal fonns an "L" shaped su-
ture with the postfrontal and a well developed
lateral process that joins the postorbital and

1980

PENNSYLVANIAN DIAPSID REPTILE

17

forms the antcnidorsal rim of the upper tem-
poral fenestra. The central portion of the
lateral border of the parietal is deeply con-
cave and forms the dorsal rim of the superior
temporal fenestra. The remaining portion of
the lateral border is formed by a narrow wing
that is directed posterolaterally and gently
ventrally beneath the supratemporal to reach
the squamosal. This wing is deeply excavated
on the dorsal surface in order to receive the
elongate supratemporal.

The posterior margin of each parietal is
deeply embayed between the posterolateral
wing and the midline. At the midline the two
parietals form a posteriorly directed nuchal
crest. At the occipital border the parietal ap-
pears to have a narrow sheet of bone that
extends ventrally onto the occipital surface
to support the postparietals and tabulars
(Fig. 8a). The nearly straight inteiparietal
suture is interrupted by the large, round
pineal foramen.

The parietal is gently convex anteropos-
teriorly; the downward slope is evident in
lateral view (Fig. 2).

In one specimen KUVP 9959a (Fig. 9c), a
parietal, badly worn at the edges, has been
preserved in ventral view. It reveals a slight
longitudinal crest about halfway from the
midline to the upper temporal fenestra. This
crest presumably marked the position of the
cartilaginous taenia marginalis of the chon-
drocranium and the point of attachment of
the supraoccipital. Lateral to this crest the
ventral surface of the parietal has a gently
convex ridge that follows the medial margin
of the upper temporal fenestra and extends
onto the posterolateral wing of the bone.
From this ridge to the temporal opening the
parietal becomes progressively thinner.

Postparietal. — The postparietal, preserved
only in KUVP 33606 (Fig. 11), is completely
restricted to the occipital surface. It is only
slightly visible in dorsal view. The element
retains its primitive reptilian relationships — a
paired bone in contact with the parietal, tabu-
lar, and supraoccipital.

Supratemporal. — A right supratemporal is

only partially preserved in KUVP 9952 (Fig.
8). Its configuration can, however, be estab-
lished on the basis of the deep, narrow groove
on the posterolateral process of the parietal
and on the posterodorsal aspect of the squa-
mosal, in which this element lay. As in all
captorhinomorphs were this element is pre-
served, the supratemporal in Petrolacosaurus
is a long, narrow, slightly curved bone that
lies on a process separating the superior tem-
poral fenestra from the occipital surface of
the skull. Posteriorly, it is anchored by a
depressed facet on the posterodorsal corner of
the squamosal. This indicates that the supra-
temporal probably serves to strengthen the
parietal-squamosal contact, bracing the pos-
terodorsal corner of the skull roof against the
pull of adductor muscles.

Tabular. — The tabulars, as seen in KUVP
9952, are relatively small, elongate bones,
completely restricted to the occipital sm-face.
This element, although much reduced, re-
tains its primitive relationships through con-
tacts with the postparietal, parietal, supratem-
poral, and probably with the supraoccipital
and squamosal. As reconstructed, the tabular
forms a small part of the border of the post-
temporal fenestra.

Postfrontal. — The postfrontal, seen in ex-
ternal and internal views in KUVP 9952 and
33607, is a relatively large, "L"-shaped bone
that foiTns the posterodorsal margin of the
orbit and has a long ramus lateroventrally
beneath the postorbital. The postfrontal,
as reconstructed, is a shaiply convex bone,
with most of the medial portion occupy-
ing a dorsal position while most of the
lateral ramus extends down on the cheek. Its
anterior apex fits into a small slot on the
frontal. The medial border of the postfrontal
extends far posteriorly, bound by the frontal
and the parietal. The shape of the area of
attachment of these two bones can be seen
best in KUVP 33606 (Fig. 11). The angle
between the medial and posterior edges of
the postfrontal is close to 90 degrees. This
corner of the bone is wedged deeply within
the parietal, and the articulation is reinforced

18

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Fig. 12. — Petrolacosaurus kansenxis Lane. Mature skull and lower jaws with articulated cervical vertebrae,
and an isolated right clethnim, KUVP 33607, X 2. Tlie component .skull elements are disarticulated, and com-
pressed into a single plane. The riglit mandible and the \ertel)rac are exposed in lateral view. See Fig. 13
for opposite view of same specimen. See Fig. 2 for key to abbreviations.

1980

PENNSYLVANIAN DIAPSID REPTILE

19

Fig. 13. — Petrolacosmirus kausemis Lane. Mature skull and lower jaws with articulated cervical vertebrae,
and an isolated right cleithrum, KU\T 33607, x 2. The component skull elements are disarticulated, and
compressed into a single plane. The left mandible lies over the palate. See Fig. 12 for opposite view of
same specimen. See Fig. 2 for key to abbreviations.

20

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

by a facet in the parietal that supports the
posterior edge of the postfrontal from under-
neath. The ventrolateral ramus of the post-
frontal fits imder and follows the curve of the
upper half of the postorbital.

The ventrolateral process of the post-
frontal appears to be massive and triangular
in cross-section. The postfrontal in KUVP
9952 ( Fig. 8 ) has a well exposed ventrolateral
ramus. In lateral view, a slight ridge is pres-
ent close to the orbital margin. The medially
inflected anterior margin of the postorbital
fits over this ridge and covers the ramus of
the postfrontal. In medial view, the post-
frontal shows a well developed ridge that
extends to the ventral tip of the bone. This
ridge on the postfrontal is equivalent to the
medially extending plate of bone seen on the
prefrontal. Both ridges reinforce the orbital
margin.

Postorbital. — The postorbital, exposed in
lateral view in KUVP 33607 and in both lat-
eral and medial views in KUVP 9952, has the
triradiate configuration typical of diapsid rep-
tiles. Although slightly disarticulated from
the rest of the skull elements, its relationship
to the surrounding bones can be established
with confidence. The ventral process of the
postorbital extends around the postfrontal
covering much of its surface and entering the
posterior orbital border. This process reaches
far ventrally to attach to the anterodorsal
facet of the jugal. The length and surface
detail of the ventral process of the right post-
orbital in KUVP 9952 indicates that almost
one-half of this process was applied to the
inside surface of the jugal, and must have
extended near to the jugal-ectopterygoid su-
ture. The concave anterior rim of the post-
orbital is thicker than the rest of the bone
and is medially inflected. The posterior mar-
gin of the ventral process of the postorbital
forms part of the anterior border of the lower
temporal fenestra. The dorsal process forms
the anterior border of the upper temporal
fenestra and extends as a narrow slip of bone
between the postfrontal and parietal. This
dorsal process was flattened in both the right

and left postorbitals of KUVP 9952. In KUVP
33607 the right postorbital was also flattened,
but the left retains much of its original shape
(Fig. 13). The posterior process of the post-
orbital is sutured to the squamosal and forms
the anterior half of the intertemporal bar.

Jugal. — The jugal, exposed in lateral and
medial views in KUVP 9951 (Fig. 6), 9952
(Fig. 8), 33608 (Fig. 10), and in KUVP
33607 (Figs. 12 and 13), forms part of the
cheek region below and behind the orbit.
Its triradiate configuration can be seen clearly
in all specimens and is typical of forms with
lower temporal fenestrae. The two ventral
rami form not (jnly the lower margins of the
orbit and the lower temporal fenestra, but
also the ventral border of the skull behind the
maxilla. The suborbital ramus extends far
forward to reach the posterior projection of
the lacrimal. In both KUVP 9952 (Fig. 8)
and KUVP 33607 (Fig. 12) the lateral surface
of the suborbital ramus is deeply recessed
close to the anteroventral border, indicating
that in this region the jugal was partially cov-
ered laterally by the maxilla. The lower tem-
poral ramus of the jugal, completely pre-
served only in KUVP 9951 (Fig. 6), is much
wider than the other rami. It extends below
the lower temporal fenestra, and probably
attaches to the quadrato-jugal. The dorsal
ramus of the jugal forms the lower part of
the postorbital bar. The anterior border of
this ramus bears a long groove that slants
diagonally upward and backward. This groove
receives the ventral process of the postorbital.

The orbital margin of the jugal is thick-
ened on the medial surface. Immediately
above the posterior end of the maxilla, this
thickened region of the jugal has a well devel-
oped medially projecting tubercle that is su-
tured to a stout lateral process on the ecto-
pterygoid. This tubercle, partially exposed
in KUVP 33607, but also seen in KUVP 9951
and 9959a, provides the only contact between
the palate and the jugal. The posterior ramus
of the jugal is smooth and ver>' thin.

Quadratojugal. — The quadratojugal, which
is only partially preserved in KUVP 33603,

1980

PENNSYLVANIAN DIAPSID REPTILE

21

33607, and 3360S is a thin, elongate sheet of
bone that forms the ventral border of the
skull behind the jugal, and meets the squa-
mosal behind the lower temporal fenestra.
Although this bone is poorly preserved in all
the above specimens, there is evidence to in-
dicate that the jugal and quadratojugal attach
to one another below the temporal fenestra.
In KUVP 33608 (Fig. 10), the badly pre-
served jugal and quadratojugal appear to be
in contact. In addition, the height and length
of the quadratojugal, relative to the quadrate
in the immature specimen KUVP 33603 (Fig.
9c), indicates that the quadratojugal extended
far anteriorly below the lower temporal open-
ing as a tall, thin sheet. The jugal also ex-
tends posteriorly as a relatively tall, thin
sheet; its posterior outline is indicative of an
overlapping suture with the quadratojugal.
The quadratojugal, therefore, not only forms
the ventral edge of the lower temporal fenes-
tra, but also contacts the jugal. The quadra-
tojugal wraps posteriorly around the dorsal
process of the quadrate as indicated in KUVP
33603.

Squamosal. — Although seen in many speci-
mens, the squamosal is incompletely pre-
served in all, and is therefore, difficult to
interpret. The size and shape of the squa-
mosals in KUVP 33607 and 33608 indicate
that most of the cheek region behind the tem-
poral fenestrae is covered by this large plate-
like bone. The anterior portion of the right
squamosal in KUVP 33607 (Fig. 12) shows
very clearly that, as in all primitive diapsids,
this bone forms the rounded posterior borders
of the upper and lower temporal fenestrae
and extends an anterior projection between
the fenestrae to form the posterior portion of
the intertemporal bar. Although incomplete,
the posterior border of the squamosals in
KUVP 9952 (Fig. Sb), 9959a and 33607
(Figs. 12 and 13) is sufficiently preserved to
indicate that this element forms the rounded
posterior border of the cheek region, as in
primitive captorhinomorphs. In KUVP 33607
the long posterior strip of the right squamosal
which is partly separated from the rest of the

bone, probably wrapped around the dorsal
process of the quadrate to present in occipital
view a narrow strip that forms the lateral
border of the posterior temporal fenestra and
may have also provided support for the tabu-
lars. The posterior region of the squamosal
is, therefore, similar in Petrolacosaurus and
captorhinomorphs and more primitive than in
all other diapsid reptiles, where the squa-
mosal does not cover the dorsal process of the
quadrate completely. In all known Permian
and Triassic "eosuchians" it is the quadrate
that forms the concave posterior margin of
the skull and eventually the otic notch. There
is no otic notch in Petrolacosaurus, Arae-
oscelis or primitive captorhinomorphs.

The shape and configuration of the pre-
served squamosals also indicates that, dor-
sally, this bone enters the superior surface of
the skull, contributes to the medial edge of
the upper temporal fenestra, supports the
supratemporal, and comes in contact with the
parietal. Ventrally the squamosal overlaps
the quadratojugal.

The squamosal in KUVP 33607 is thickest
at its center of ossification, which is located
at the level of the intertemporal bar. From
here the bone thins anteriorly and anteroven-
trally, but remains thick posteroventrally. The
posterior border of the squamosal in KUVP
33604 and 33608 is also thick. At margins of
the temporal fenestrae, the squamosal is very
thin. The anteroventral corner of the squa-
mosal is not preserved in any specimen.

Dermal Bones of the Palate

Vomer. — The vomers are best preserved
in KUVP 33608 (Fig. 10) and KUVP 33607
(Figs. 12 and 13) where they appear as
paired sheets of bone lying anteriorly along
the midline of the palate. In ventral aspect
the vomer is a long roughly triangular ele-
ment, occupying 28 per cent of the total
length of the skull. The entire lateral border
of the vomer forms the medial rim of the
internal naris. Anteriorly and posteriorly the
vomer is sutured to the premaxilla and pala-
tine, respectively. The two vomers are su-

22

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

tiired to each other at the midline along t\vo-
thirds of their length. Posteromedially the
vomers are separated along the midline by a
narrow wedge-shaped anterior extension of
the pterygoids.

Medially the vomer is strengthened by a
longitndinal ridge on the dorsal and ventral
surfaces (Fig. 10). The ventral ridge carries
a row of large teeth that approaches but does
not reach the premaxilla anteriorly. The
posterior three-fourths of the ventral surface
of the vomer is also covered by small denti-
cles. The ventral surface is concave in cross-
section. The lateral border of each vomer
curves ventrally but is not ridged.

The dorsal surface of the vomer is more
complex than the ventral, with a tall median
ridge, a flat central surface and a longitudinal
depression near the lateral rim ( KUVP 33607,
Fig. 13). The median ridge probably
strengthened the contact between the vomers
and supported the nasal septum. The flat dor-
sal surface of the vomer is narrow anteriorly
but widens posteriorly, where it is partially
covered by the large anterior process of the
palatine. The longitudinal depression is a
deeply concave surface, bound medially by a
ridge that probably represents the medial
boundary of the internal naris.

Palatine.— The palatine in KUVP 3360S
(Fig. 10) and KUVP 33607 (Fig. 13) is a
rectangular sheet of bone pressed against the
lateral edge of the pterygoid. The shorter
sides of the rectangle form the sutures with
the vomer anteriorly and the cctopterygoid
posteriorly. The long lateral side of the pala-
tine extends in a wide process to the maxilla
anteriorly as it does in Youngina and other
diapsids. This massive process occupies one-
third of the total length of the bone, its su-
tural surface is concave and scarred thus in-
dicating that the palatomaxillary suture was
heavily stressed. Between the maxillary proc-
ess and the anterior suture with the vomer,
the palatine forms the posterior margin of
the internal naris.

Behind the maxillary process, the lateral
edge of the palatine is free as in all diapsid

reptiles, forming the anteromedial border of
the suborbital fenestra. The palatine is not
ridged along the rim of the fenestra. Small
denticles, similar to those seen on the vomer,
cover the surface of the palatine ventrally. In
addition, a ridge with larger teeth runs di-
agonally across the palatine from its postero-
medial corner toward the maxillary process.

In dorsal view, the palatine overlaps the
lateral edge of the pterygoid and forms a
stout process anteriorly over the dorsal sur-
face of the vomer. The ectopterygoid appar-
ently overlaps the posterior part of the dorsal
surface of the palatine.

Ectopterygoid. — The left ectopterygoid is
preserved in KUVP 9952 (Fig. 8). It is
slightly out of position relative to the jugal
and is seen in partial lateral view. A well
developed process of the cctopterygoid pro-
jects laterally to articulate with a correspond-
ing tubercle in the jugal, as it does in Youn-
gina and other primitive diapsids. Anterior
to this the lateral border of the cctopterygoid
is free of sutural contact and forms the pos-
teromedial and posterior boundary of the
suborbital fenestra. Posteriorly, the ectoptery-
goid forms the anterior boundary of the sub-
temporal fossa; medially it is suturally at-
tached to the pterygoid. Posteromedially this
attachment is formed by a massive rounded
abutting suture. Anteromedially, the ecto-
pterygoid fomis a small sheet of bone that
extends onto the dorsal surface of the ptery-
goid. Anteriorly, the ectopterygoid overlaps
the dorsal surface of the palatine.

Ptenjgoid .—The pterygoid, seen in both
dorsal and \cntral \iews in KUVP 8351,
33604, 33607 and 3.360S, is the longest bone
of the skull, similar in most features to that
in protorothyridids (Carroll and Baird, 1972).
Although partially covered by other skull ele-
ments, its configuration and relationship to
the surrounding bones can be established with
confidence. The pterygoid has four processes
common to all primitive reptiles: the palatal
ranuis, the transverse flange, the quadrate
ramus, and the small process for the basi-
pterygoid articulation.

1980

PENNSYLVANIAN DIAPSID REPTILE

23

The palatine ramus extends anteriorly
from the region of the basiptcrygoid articula-
tion to its acuminate temiination between the
vomers; it is widest where it articulates with
the ectopterygoid. Along the posterior part
of its medial edge the palatine ramus bounds
the interpterygoid vacuity. A dorsally in-
flected ridge is present on the anterior one-
third of the rim of the interpyterygoid va-
cuity. This inflected ridge continues anterior-
ly and helps to form a greatly strengthened
suture between the pterygoids along the mid-
line. The palatine ramus carries two longi-
tudinal ridges on the ventral surface. These
bear irregularly arranged teeth, up to 0.5 mm
in diameter and not more than 1.0 mm in
height. The medial ridge runs to the anterior
tip of the pterygoid and is continuous with
the tooth bearing ridge on the vomer. The
shorter lateral ridge is continuous with a
similar ridge on the palate. The rest of the
palatal ramus of the pterygoid is covered
with numerous small denticles, about O.I mm
in diameter.

The transverse flange of the pterygoid
extends anterolaterally at about 60 degrees to
the sagittal plane. The flange slants ventro-
laterally and extends below the lower margin
of the cheek. The ventral surface of the
flange is covered with a large number of ir-
regularly arranged teeth of varying sizes. The
largest teeth, located on the posterior rim of
the flange, are up to 0.8 mm in diameter and
1.2 mm in height. The anterior portion of the
flange is covered by smaller teeth.

The quadrate ramus, as seem in KUVP
8351 (Fig. 5a and b) is essentially a large
vertical plate running posterolaterally from
the basiptcrygoid region to the quadrate. The
ventral edge of the quadrate ramus is thick-
ened to form a ridge, above which the inner
side is slightly concave. Posteriorly, the ra-
mus is attached to the inner surface of the
dorsal process of the quadrate. The contact
between the two bones is extensive both
anteroposteriorly and dorsoventrally. The
laterodorsal surface of the quadrate ramus in

Fig. 5a shows an extensive area of attach-
ment to the epipterygoid.

The basicranial process of the pterygoid is
not well preserved in most specimens. In
KUVP 33604 this process is exposed in dorsal
view. It shows that the anteromedial part of
the process has a small dorsally oriented ar-
ticulating surface for the attachment to the
basisphenoid. Most of the articulation, how-
ever, must have been provided by the epi-
pterygoid.

Ossifications of the Palatoquadhate
Cartilage

Epipterygoid. — A small triangular piece of
bone lying on the crumpled left quadrate
ramus of the pterygoid in KUVP 8351 has
been identified by Peabody (1952), with
some misgivings, as the epipterygoid. This
identification is of doubtful validity, how-
ever, and the element is not included in the
reconstruction. In the same specimen, the
dorsal expanse of the quadrate ramus of the
right pterygoid is well preserved. Dorsome-
dially. the quadrate ramus of the pterygoid is
strongly grooved at the level where the epi-
pterygoid should be located. This attachment
area is extensive, running posteriorly from the
region of the basiptcrygoid process to the
dorsal process of the quadrate. The conclu-
sion that the epipterygoid extended so far
anteriorly is supported by the fact that the
articulating surface of the basiphenoidal proc-
ess faces dorsolaterally and that the engaging
process on the pterygoid is too thin in all the
specimens, where this region is preserved, to
provide support on its own. In addition, the
epipterygoid in pelycosaurs and captorhino-
morphs fonued partly, or wholly, the articu-
lation with the basisphenoid (Heaton, 1979).

The groove on the pterygoid for the base
of the epipterygoid is not excessively long in
PetrolacosauTus when comparison is made
with other primitive reptiles. In Dimetrodon
(Romer and Price, 1940, PI. 7), for example,
the base of the epipterygoid extends from
above the dorsal process of the guadrate to

24

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

the interpterygoid vacuities, well past the
level of the basicranial articulation.

Quadrate. — In Petrolacosaurus, the quad-
rate is a tall, well developed element, similar
in configuration to that in captorhinomorphs
and pelycosaurs. It is composed of a rela-
tively flat sheet of bone that extends dorsally
and anteriorly from the articular condyle. The
articular area of the quadrate seen in ventral
view in KUVP 33603, 33607 and 33608, con-
sists of two subhemispherical surfaces, a lat-
eral and a medial, separated by a groove. The
lateral surface faces ventrally. The medial
surface, on the other hand, faces ventrolater-
ally and is located farther anteriorly and ven-
trally than the lateral surface. The tall dorsal
process of the quadrate preserved only in
KUVP 33608, joins the quadrate process of
the pterygoid anteriorly. The posterior mar-
gin of the dorsal process of the quadrate is
covered by the squamosal and quadratojugal
except in the region where the palatoquadrate
foramen emerges, precluding the possibility
of fonuing the margin of a squamatc type of
otic notch. The lateral surface of the quadrate
forms part of the posteromedial wall of the
adductor chamber.

Ossifications of the Braincase

Varasphenoid. — The parasphenoid is a me-
dian dermal element that provides a ventral
cover for the braincase. In ventral view, the
region of the parasphenoid ventral to the
basisphenoid has the general outline of an
isosceles triangle, typical of all primitive cap-
torhinomorphs and early "eosuchians." Tlie
cultriform process, seen in both dorsal and
ventral views in KUVP 9951 (Fig. 7) extends
forward between the interpterygoid vacuities.
It is "V"-shaped in section and bears a row
of tiny denticles on its posterovcntral aspect,
as in primitive captorhinomorphs. The cultri-
form process probably supported the sphe-
nethmoid.

The triangular portion of the parasphenoid
hides most of the basisphenoid from ventral
view. The ventral surface of the parasphe-

noid is covered with denticles at the level of
the basisphenoidal tubera. The flanks of this
rostrum are not well preserved. The location
of the anterior internal carotid foramina can-
not be established. The sides of the para-
sphenoid are formed by two rounded crests
(Figs. 5b, 7a, and 13) that converge an-
teriorly and widen posteriorly. These crests,
the cristae ventrolaterales, leave a wide de-
pression on the ventral surface between them.
The posterior end of these crests extend lat-
erally beyond the basioccipital. The ventral
surface of the parasphenoid between the two
lateral ridges is formed by a thin sheet of
bone underlying the anterior end of the
basioccipital.

Exposed in dorsal view, the parasphenoid
in KUVP 33608 has a thin sheet of bone on
each side that probably extended dorsally
(Fig. 10b) to attach to the prootie, contrib-
uting to the lateral wall of the braincase.

Basisphenoid. — In ventral view the basi-
sphenoidal tubera project anterolaterally be-
yond the dermal parasphenoid. Each process
has a slightly constricted waist and ends in a
gently convex articular head. The articular
area faces mainly antcrodorsally and is little
exposed in ventral view. Preservation of all
the known basisphenoid-parasphenoid com-
plexes either in dorsal or ventral views, makes
identification of the suture between these ele-
ments impossible.

The basisphenoid in Petrolacosaurus can
be seen best in dorsal view (Figs. 5a, 7b, and
10). The anterior portion of the basisphenoid
forms a short rostral process. Its anterior
termination is deeply incised between the
cristae trabeculares to carry a small branch of
the anterior internal carotid artery. The cris-
tae trabeculares form the point of attachment
of the trabecula communis to the basisphe-
noid.

Along the midline, just above the basi-
sphenoidal tubera, a small pit in KUVP 9951
and 33608 (Figs. 7b and 10) probabb- marks
the exits of the anterior internal carotid arte-
ries, and their subdivisions, the pituitary ar-
teries. A small trans\erse ridge separates this

1980

PENNSYLVANIAN DIAPSID REPTILE

25

region from the more posteriorly located sella
turcica.

The basisphenoid supports a well devel-
oped dorsum sella, comparable in height to
that in modern iguanid lizards and nearly as
tall as in Captorhitms. The dorsum sella in
Petrolacosaurus, exposed in KUVP 8351 and
33608 (Figs. 5a and 10) is a thin trans-
verse sheet of bone strengthened laterally
by the sellar processes. In both specimens
the dorsum sella and the sellar processes have
been crushed postmortally down onto the
dorsal surface of the parasphenoid obscuring
the cavum cranii, the largest concavity of the
basisphenoid, where the anterior body of the
medulla rested.

Basioccipital. — The short, stout basioccipi-
tal forms the ventral surface in the posterior
portion of the braincase and the major por-
tion of the occipital condyle. The ventral sur-
face of the basioccipital is fully exposed pos-
teriorly; its anterior portion, however, is cov-
ered by the posterior end of the basal plate
of the parasphenoid. The well developed
basioccipital tubercle, which forms most of
the exposed ventral surface of the basioccipi-
tal, has a transverse ridge anteriorly between
the cristae ventrolaterales of the parasphenoid,
and a longitudinal ridge that extends along
the midline to the occipital condyle. The
margins of this tubercle probably formed the
anterior and medial limits of the M. rectus
anterior insertion.

Unlike those of pelycosaurs and capto-
rhinomorphs, the basioccipital in Petrolaco-
saurus is not completely covered dorsally by
the exoccipitals. The latter do not join at the
midline to exclude the basioccipital from the
floor of the foramen magnum.

Exoccipital. — In posterior view the ex-
occipital has a wide base that forms the
dorsolateral part of the condyle. Above the
condyle, an ascending process runs up along
the side of the foramen magnum and is ap-
plied dorsally to the ventral edge of the su-
praoccipital. The posterior surface of the
ascending ramus bears a facet for the proat-
las. Between this and the main body, the

exoccipital forms the medial border of the
vagus foramen. Below this foramen, and
slightly medially, are two small hypoglossal
foramina. At the base, the pair of exoccipitals
approach each other but do not join at the
midline (KUVP 33608, Fig. 10).

Prootic. — The prootic is the least well
known element of the braincase in Petrolaco-
saurus. This element, preserved only in KUVP
9951, is similar in outline to that of Capto-
rhinus (Price, 1935).

Supraoccipital. — The supraoccipital, also
exposed in KUVP 9951 and 33607, forms the
broad occipital plate. It not only forms the
arched roof of the foramen magnum but also
extends down on both sides to the exoccipi-
tals.

Dorsomedially, the postparietals overlie
the process of the supraoccipital that extends
to the parietals. Dorsolaterally, the tabular
covers the slightly ridged process of the su-
praoccipital that extends to the squamosal-
parietal suture.

Posteriorly, the supraoccipital bears a pro-
nounced median ridge separating the prob-
able insertions of the M. recti capiti posterior.
This ridge runs from the postj^arietals to the
foramen magnum, where it fans out laterally.
In KUVP 9951, the supraoccipital has been
crushed flat, exaggerating the lateral extent of
this bone. Normally the two dorsolateral
"wings" of the supraoccipital, exposed in Fig.
7, face mostly laterally and only slightly pos-
teriorly; they formed the lateral wall of the
braincase behind the prootic.

Opisthotic. — The opisthotic, also found
only in KUVP 9951 and 33607, is partially pre-
served. The available material indicates that,
as in pelycosaurs, Araeoscelis and Youngina,
the opisthotic in Petrolacosaurus forms an os-
sified paroccipital process. Unlike Araeoscelis,
however, this process did not quite reach the
quadrate. In addition, a ventral process of
the opisthotic extended down along the lat-
eral margin of the exoccipital-basioccipital
complex. The level of its ventral termination
is, however, indeterminate.

Stapes. — The stapes preserved in KUVP

26

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

8351 and 33607 retain the pattern found in all
primitive Paleozoic reptiles. It is a relatively
massive bone consisting of a large proximal
head and a gently tapering columellar process
pierced by the stapedial foramen. The proxi-
mal head is poorly preserved, but it probably
consisted of a footplate and a dorsal process.
The distal tip of the columella in KUVP 33607
is slightly expanded and its surface is un-
finished.

Mandible

Dentanj. — This long, unusually slender ele-
ment, best preserved in KUVP 33606 and
33607, carries the mandibular tooth row and
occupies a considerable area of the outer
surface of the mandible. Anteriorly, it ex-
cludes the splenial from the lateral surface.
It is bounded posteroventrally by a narrow
slip of the splenial. Posteriorly, the dentary
covers the angular and surangular and nar-
rows dorsally. Its posterior tip lies near the
dorsal margin of the mandible, covering most,
but not all, of the coronoid. The single lat-
eral tooth row is supported by a medially
directed shelf on the internal surface of the
dentary, close to the dorsal edge of the bone.
The lateral margin of this shelf is, as in the
case of the maxilla, much higher than the
medial one. Anteriorly, the shelf forms most
of the symphysial surface. This area is ex-
posed only in KUVP 3.3607; it is quite small
and bears no marked ridges. Behind the sym-
physial area, the dentary is completely ex-
posed medially to the level of the eighth
tooth. Further posteriorly, the medially di-
rected shelf is strongly ridged, indicating the
area of attachment of the splenial and coro-
noid. Below the shelf, the dentary is strongly
excavated for the channel of the Meckelian
canal; the dentary forms the roof and lateral
wall of the canal in its anterior portion. In
KUVP 33607 there are 28 teeth attached to
the dentary with shallow alveoli for five more.

Anp,ular. — The angular, also preserved in
KUVP 33607, is a large "V"-shapcd element,
occupying part of the medial and lateral sur-
faces of the mandible. It forms the ventral

portion of the mandible posterior to the den-
tary and splenial, and continues back to the
end of the jaw as a deep keel below the sur-
angular and prearticular. This bone is over-
lapped aiiterodorsally by the dentary, but it
overlaps the surangular and articular on the
lateral surface and the prearticular on the in-
ternal surface of the mandible. Below the
prearticular fossa the angular forms the floor
of the Meckelian canal and part of the lateral
wall.

Surangular. — The surangular, also exposed
in KUVP 33607, is a simple sheet of bone that
occupies the posterior portion of the mandi-
ble. Its lateral exposure is relatively small;
it is overlapped by both the dentary and the
angular. Internally, it is overlapped by the
posterodorsal process of the coronoid. Pos-
teriorly, the surangular has a large exposure
along the lateral wall of the adductor fossa.
The upper edge of the bone is somewhat
thickened; posteriorly, it swings inward to
attach to the articular and form the posterior
wall of the fossa.

Splenial. — The splenial, as seen in KUVP
33607 and 33608, forms most of the anterior
portion of the medial mandibular surface. The
lateral exposure of this bone is limited to a
narrow strip below the posterior part of the
dentary. Anteriorly, the splenial is in contact
with the dentary above and below the anterior
opening of the Meckelian canal. This contact
is maintained for some distance posteriorly
with the splenial forming the medial wall and
floor of the canal. Posterodorsally, the splenial
covers the anterior projection of the coronoid.
A posteriorly directed thin sheet of the splen-
ial covers part of the prearticular and pos-
terior splenial and comes in contact with the
angular.

Posterior splenial. — A small elongate sheet
of bone, that tapers to a point both anteriorly
and posteriorly, lies on the medial surface of
the mandible in KUVP 33607. Anteriorly, a
large portion of this bone was probably cov-
ered by the splenial. It is sutured to the pre-
articular dorsally and the angular ventrally.
This bone may be comparable to the posterior

1980

PENNSYLVANIAN DIAPSID REPTILE

27

splenial seen in some amphibians; it presuma-
bly formed part of the medial wall of the
Meckelian canal. A posterior splenial has not
been reported in other Paleozoic reptiles.

Prearticuhr. — The prearticular, a major
element on the inner surface of the mandible,
is well preserved only in KUVP 33607. Pos-
teriorly it is applied to the inner surface of
the articular. The suture between these two
bones is not visible. The conjoined edges of
the prearticular and articular are drawn out
medially to form a small shelf at the postero-
ventral corner of the adductor fossa. From
this area, the prearticular extends anteriorly
as a sheet of bone along the inner surface of
the mandible, forming the medial margin of
the adductor fossa and most of its medial
wall. The prearticular is overlapped below
by the angular and the postsplenial. Anterior-
ly, the prearticular is covered by the posterior
end of the splenial. Anterior to the adductor
fossa, the prearticular is again drawn out
medially to form a shelf ventral to the coro-
noid. In this region, the prearticular forms
part of the medial wall of the Meckelian
canal.

Coronoid. — Only a single edentulous coro-
noid is present in Petrolacosaurus. In KUVP
33607 the anterior half of this bone is a thin
sheet, applied to the medial surface of the
dentary below the marginal tooth row, and
in contact with the splenial anteriorly and the
prearticular ventrally. A thicker area forms
the anterior margin of the adductor fossa and
part of its dorsal rim. The posterodorsal end
of the bone lies beneath the surangular, on
the outer part of the mandible.

Articular. — The articular joins the lateral
and medial walls of the adductor fossa and
forms the articulating surface with the quad-
rate. This surface consists of two parallel,
elongate concavities, the inner groove lying at
a lower level than the other. Externally, the
articular bone is almost completely covered
by the surangular and angular, but forms the
convex posterior end of the mandible. There
is no distinct retroarticular process. What
looks like a small retroarticular process in

KUVP 33607 is actually part of the rounded
posterior end of the articular that was pushed
posteriorly when the lower jaw was crushed.
A short angular process extends medially and
somewhat ventrally from the articular surface
of the articular, probably for the insertion of
the superficial pterygoideus musculature.

Dentition

The teeth in Petrolacosaurus are simple
conical structures. In both the upper and
lower jaws there is a marginal scries of sub-
thecodont teeth. There are also many palatal
teeth. The marginal teeth, arranged in a
single row, are slightly compressed and sharp-
ly pointed. Towards the tips, the teeth bend
slightly posteriorly and are marked by longi-
tudinal grooves. This longitudinal grooving, or
fluting, occurs only towards the tip and is
often seen in well-preserved Paleozoic rep-
tiles.

On the prcmaxilla, the anteromedial tooth
is the largest. Posterolaterally, the premaxillary
teeth show progressive diminution in size.
The first three teeth tend to be inclined slight-
ly anteroventrally. Two teeth in the anterior
portion of the maxilla are larger than the
others and hence may be designated as ca-
niniform teeth. This is a primitive amniote
condition that persists in many Permian rep-
tiles. In Petrolacosaurus, the caniniform teeth
are only slightly longer than the first pre-
maxillary teeth. These relatively large teeth,
the two to four smaller premaxillary teeth,
and the four pre-caniniform teeth create a
seemingly efiBcient food trap. The post-ca-
niniform teeth, although definitely smaUer
than the caniniform teeth, remain quite large
until well posteriorly in the series; only the
final eight to ten teeth show successive dimi-
nution in size. The upper tooth row continues
back to the posterior extremity of the maxilla.

The mandibular marginal dentition is ba-
sically homodont with reduction in tooth
height only in the last five to seven teeth
(KUVP 3.3606, Fig. 11). The lower tooth row
does not seem to extend back to the extreme

28

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

end of the dentary. The teeth <it the level of
the maxillary caniniforms are slightly longer
than the rest of the mandibular teeth. Al-
though the anterior three teeth on the dentary
point slightly anteriorly as on the premaxilla,
they show slight progressive diminution for-
ward.

In both the upper and lower jaws, the
teeth sit on a medially directed shelf and are
supported laterally by a thin sheet of bone. In
medial view, the teeth show slight vertical
ridging close to the base. Cross-sectional
views of the maxillary teeth (Fig. 9b) show
that they are essentially thin walled eones with
an open base. Large pulp cavities extend
close to the tip of the teeth. The teeth sit
in shallow pits on the jaw. Implantation has
been strengthened by a thin sheet of bone
that extends from the circular margin of the
pit onto the lateral surface of the tooth base
and is solidly fused to it. This fusion gives
the appearance of ridging both on the outside
and the inside of the teeth close to the base.

Palatal teeth occur on the pterygoids, pala-
tines, vomers and the parasphenoid. Rela-
tively large teeth (up to 1.2 mm in length),
irregularly distributed along ridges on the
palate, extend in three directions from the
basipterygoid area. A variable number of
large teeth extend laterally onto the trans-
verse flange of the pterygoid. Another group
of slightly smaller teeth extend diagonally
across the pterygoid and palatine. A third
group extends anteriorly along the medial
edge of the pterygoid and the vomer.

Numerous small teeth (about 0.3 mm in
length) are rather evenly distributed over
most of the palatal plate between the rows
of larger teeth. The teeth on the parasphe-
noid are also cjuite small. A niunber of broken
teeth indicate that both large and small pala-
tal teeth have well developed pulp cavities.
There are no teeth on the inner surface of the
mandibles.

Sclerotic Plates

Series of thin translucent sclerotic plates
lie in the orbit of the skulls in KUVP 9952

and 33606. The seven plates in the orbit of
the skull in KUVP 33606 are only partially
preserved, but they appear rectangular and
do not overlap one another. After the better
preserved plates in KUMNH 9952 (Fig. 8a)
were covered with a resin imbedding com-
pound, the outline of what appear to be five
small, triangular plates were found wedged
between five larger, nearly rectangular plates.
In both specimens the sclerotic plates were
flat, but this may be due to crushing.

Vertebrae

Notwithstanding certain specializations,
the vertebral column of Petrolacosaurus re-
tains most of the general characteristics seen
in the captorhinomorphs. Each segment of
the column includes a centrum, neural arch
and intercentrum. Each centrum has the
basic hourglass configuration, with expanded
hollow ends facing adjacent centra and a
narrow waist. A ventral longitudinal ridge
extends between the anterior and posterior
ends of the centrum to strengthen the hour-
glass structure. Anterodorsally, thickened
areas serve as points of attachment for the
neural arches. Thus, all the centra except the
first are amphicoelous and all are notochordal.
The notochordal canal is strongly constricted
at the middle of each centrum. The neural
arches are fused solidly to the centra and su-
tures have been obliterated. The articular
areas of the zygapophyses are simple plane
surfaces, somewhat tilted from the horizontal.
The intercentra are well developed.

A presacral vertebral count of 26 is defi-
nitely established on the basis of several ar-
ticulated specimens. There are six cervical
\ertebrae that form a distinct neck region.
The calculated length of the cervical series in
KUVP 9951 is about 72 mm; the calculated
length of tlie dorsal series in the same indi-
vidual is about 192 mm. There are two sacral
vertebrae as is typical in most primitive rep-
tiles. An approximate count of 60 to 65 for
the number of caudal vertebrae is based on
two partial specimens, KUVP 1427 and KUVP

1980

PENNSYLVANIAN DIAPSID REPTILE

29

1428 (Peabody, 1952, Figs. 2 and 4). The
distal part of the tail is composed of at least
twelve spool-shaped centra without neural
arches. The calculated length of the caudal
region based on these two immature speci-
mens is about 315 mm.

Cervical Vertebrae

Atlas-axis complex. — The proatlas (Fig.
14) is a relatively small but robust element.
The two elements of the pair do not appear
to be in contact. There is a rudimentary neu-
ral spine on each element and a small posterior
zygapophysis that articulates with the atlas.

The articulating surface of the posterior 2^'ga-
pophysis faces medially so as to be aligned
almost with the saggital plane. An anterior
zygapophysis is also present but faces for-
ward to articulate with the corresponding
facet on the exoccipital. The proatlas in
Petrolacosaurus is similar in outline to the
proatlas in the primitive captorhinomorphs
Hijlonomus and Paleothtjris; it is much short-
er than the same element in the pelycosaur
Ophiacodon.

The atlas in Petrolacosaurus includes four
elements: the intercentrum, two neural
arches and the centrum. The intercentrum

A

ii

c:::^

1)

Fic. 14. — Petrolacosaurus Jcansensis Lane. Cervical vertebrae, all x 1.5. (A) Lateral view of left proatlas,
KUVP 33607. (B) Ventral and anterior views of atlantal intercentrum, KUVP 33604. (C) Lateral and me-
dial views of left atlantal neural arch, KUVP 33604. (D) Anterior and lateral views of atlantal centrum and
fused axial intercentrum, KUVP 33604. (E) Outline of the atlas-axis complex, based on KUVP 33604 and
33607. (F) Lateral view of axis, KUVP 33604. (G) Lateral and dorsal view of third cervical vertebra, based
on KUVP 9951 and 9956. (H-I) Lateral view of fourth and fiftli cervical vertebrae, KUVP 9951. (J) Lateral
and dorsal views of the si.xth cervical vertebra, based on KUVP 9951 and 9956.

30

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

has the usual crescentic shape, but is slightly
larger than its countcqoarts in the remainder
of the column with the exception of the axial
intercentrum. A strongly concave anterior
surface receives the ventral portion of the
condyle. The posterior surface, which is less
deeply cupped, is apposed to the axial inter-
centrum. Laterally there is a small articular
surface on each side for the capitulum of the
first cervical rib. This is a rather primitive
feature seen in captorhinomorphs and pely-
cosaurs. Ventrally the intercentrum has the
longitudinal ridge seen in the other cervical
intercentra.

The atlantal neural arch (Figs. 12, 13 and
14 ) is a paired structure. There is a thin shelf
of bone above the neural canal, but the two
sides apparently are not in contact. The post-
zygapophysis extends far posteriorly to artic-
ulate, by a medially directed face, with the
prezygapophysis of the axis. A posteriorly
directed spine, external to this facet, extends
for a short distance beyond the point of artic-
ulation with the axis. At the anterodorsal an-
gle of the atlantal neural arch a roughly circu-
lar surface faces laterally and slightly dorsally
to connect with the proatlas. Anteriorly, the
neural arch articulates with the dorsolateral
portion of the condyle. A wide ventral proc-
ess descends over the lateral surface of the
atlantal centrum and articulates with it via a
medially directed pedicel. The base of the
neural arch has a small ventrally directed
process for articulation with the tuberculum
of the atlantal rib.

The atlantal centrum is well developed
although considerably smaller than other pre-
sacral centra. Dorsally, it is gently grooved
to form the floor of the neural canal. Antcro-
dorsally, there is on both sides a well devel-
oped articulating surface for the pedicel of
each neural arch. Behind this area the upper
and lateral surfaces of the centrum are free.
In immature specimens ( Peabody, 1952, Fig.
3), the centrum has, in anterior view, the
shape of an inverted "U" that sits loosely on
the axial intercentrum. In more mature speci-
mens (KUVP 33607, Fig. 12), however, the

notochordal canal is completely closed by a
more fully ossified centrum. In these mature
individuals the upper part of the anterior sur-
face surrounds the relatively small noto-
chordal pit and articulates with the central
portion of the condyle. The posterior surface
of the centrum articulates with the upper half
of the axial centrum and is of normal con-
struction.

Ventrally, the atlantal centrum sits on the
axial intercentnmi and does not reach the
ventral surface of the column. Even in ma-
ture specimens, although the above two ele-
ments are fused, the areas of fusion are
marked by a wide band of unfinished bone.
A somewhat similar pattern is seen in Heleo-
satirus (Carroll, 1976).

The axial intercentrum is relatively large,
exceeding in length all the other intercentra.
This intercentrum is more plate-like than cres-
centic and has the well developed longitudi-
nal ridge seen in the atlantal intercentrum.
In lateral view, it has the shape of a right
angle triangle, the sides of the 90 degree
angle facing posteriorly and ventrally. There
are no obvious facets for rib articulation at
the lateral margins of the element as seen on
the atlantal intercentrum.

The axial centrum-neural arch complex,
well preserved in KUVP 33607 and 9951, is a
robustly built stnicture in Petrolacosaurtis,
lying at the functional end of the vertebral
column and linking it to the skull. Its dual
function is clearly indicated by the form of
the axial neural arch (KUVP 33607 and
9951), which posteriorly is not greatly differ-
ent from that of the succeeding cervical verte-
brae. Anteriorly, it is highly modified to pro-
vide origins for muscles inserting on the occip-
ital region of the skull and to support the
skull through the remainder of the atlas-axis
complex. The neural spine is strongly built,
unusually thickened and elongated antero-
posteriorly to a length about equal to that of
the centnnn below. The posterior border is
heavy, and in one specimen small processes
near the top project posteriorly, possibly
marking the origin of the M. spinalis capitis.

1980

PENNSYLVANIAN DIAPSID REPTILE

31

The massiveness and the anteroposterior elon-
gation of the spine unquestionably provides
added support to such muscles as the M.
rectus capitis and the M. obliquus capitis.
The anterior portion of the neural spine proj-
ects anteriorly dorsal to the posterior part of
the atlantal centrum and neural arches, as in
Araeoscelis, and other primitive reptiles.

The postzygapophyses of the axial neural
arch are normally developed; the prezyga-
pophyses are weak, project little from the
surface of the arch, and face laterally as well
as slightly upward. A ridge runs posteriorly
from the prezygapophyses. The transverse
process or diapophysis is stoutly built, al-
though not greatly elongated, and originates
far ventrally on the neural arch. It projects
ventrolaterally at about 45 degrees to the
sagittal plane. The diapophysis is buttressed
by converging ridges from the neural arch
and centrum. The articular surface is small,
however, and keyhole-shaped.

The neural arch contributes significantly
to the anterior centrosphene, which has a
much greater vertical extent than the posterior
articulation. This greater anterior articulating
surface provides suflBcient contact for the
combined height of the atlantal centrum and
axial intercentrum. This basic pattern is seen
in such diverse forms as Ophiacodon. Ileleo-
saurus, Araeoscelis, Palcothyris. Platecarpus,
and Gavialis. The construction of the atlas-
axis complex indicates that movement of the
head was concentrated at the articulation be-
tween the condyle and the atlantal elements.
There appears to have been little movement
between the atlantal centrum-axial intercen-
trum and the axial centrum.

The cervical vertebrae in Petrolacosaurtis
(KUVP 9951, 9956) are conspicuous in their
degree of elongation. A number of later but
still relatively primitive reptiles, such as Arae-
oscelis, Prolacerta, Macrocnemits, and Tanij-
stropheus are also long-necked.

In the anterior part of the vertebral col-
umn, an abrupt elongation of centra starts
with vertebra 7 and continues to the axis. The
centrum in vertebra 7 is 1.25 times longer

than the average mid-dorsal centrum, and in
vertebra 6, it is 1.35 times longer; the centra
in vertebrae 2 to 5 are of equal length, being
1.55 longer than the average mid-dorsal cen-
trum. All these elongate centra have a slightly
smaller vertical diameter than the dorsal
vertebrae. In all cervical vertebrae the an-
terior centrosphenes are inclined anteriorly.
This inclination is most marked on the an-
terior cervical vertebrae, where the posterior
centrosphenes are also tilted forward. This
specialization in Petrolacosaurus would have
permitted raising, and possibly maintaining,
the head high above the ground.

There is no bevelling on the centra for
accommodating the large intercentra; the lat-
ter were probably continued dorsally in carti-
lage. In all cervical vertebrae, the longitudi-
nal ridge at the bottom of the centrum ex-
tends far ventrally, fomiing a strong keel,
with its lateral surfaces concave in section.
The keel of the anterior cervical vertebrae,
forms a nearly straight line between the ends
of the centrum, and is "V"-shaped in section.
On the posterior cervical vertebrae this keel
does not extend as far ventrally and tends
towards a more rounded ventral margin.

The cervical neural arches are attached by
a pair of pedicels to the anterior half of the
centra. Anteriorly, the margin of the pedicels
form the dorsal half of the centrosphene.
Above the centrosphene, the anterior edge of
the arch forms a slight notch before curving
up and anteriorly beneath the prezygapophy-
sis; this notch forms the posterior boundary of
the intervertebral opening. The dorsal margin
of the posterior end of the pedicel is far for-
ward on the upper surface of the centrum.
From this point the posterior margin of the
arch curves strongly to form the anterior
boundary of the intervertebral opening. The
elongate shape and anterior extent of this
opening is in contrast with the situation in
the dorsal region of the column where this
opening is rounder and considerably smaller.

The transverse jjrocesses on vertebrae 2
to 7 show the following gradual changes from
the pattern in vertebrae 2: (1) shift to a

32

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY NO. 7

higher position on the lateral surface of the
neural arch; (2) increased lateral elongation;
(3) change of angle of projection of the
process from 45 degrees to the sagittal plane
in vertebra 2 to almost 90 degrees in verte-
bra 7; (4) increase in the size of the articular
surface with the rib tuberculum and accom-
panied change in shape to an oval head plus
an attached thinner ridge than runs antero-
ventrally to the anterior centrosphene (there
is apparently no gap on the process to sepa-
rate the oval head from the thinner ridge for
the passage of the segmental artery ) ; ( 5 ) de-
velopment of a ridge connecting the prezyga-
pophysis with the dorsal margin of the dia-
pophysis. Similar changes are seen in Araeo-
sceJis, but the shift is not as gradual as in
PetrolacosauTus and occurs between vertebrae
7 and 9.

In response to the greatly elongated cen-
tra, the zygapophyses in the cervical region
are also unusually elongate and well-devel-
oped. Both pre- and postzygapophyses are
strengthened by longitudinal ridges. In addi-
tion, the prezygapophyses are supported by
buttresses directed dorsally and anteriorly.
The postzygapophyses are braced above by
paired supports descending and diverging
posteriorly. These supports, when viewed
from the rear, have a convex margin laterally
giving the neural arch its slightly swollen
appearance. Posteriorly from the third cervi-
cal vertebra, the neural arches show a gradual
increase in this convexity, which reaches its
maximum on the posterior dorsal region. This
convexity represents additional bone above
the articular surface to withstand the pressure
acting on it.

The prezygapophyses of the cervical ver-
tebrae extend far beyond the anterior end
of the centra. The postzygapophyses extend
only as far as the posterior end of the centra.
This is in contrast to the situation in the
dorsal region where both zygapophyses ex-
tend slightly beyond the ends of the centra.

Both zygapopliyses extend far laterally,
well beyond the lateral limit of the centrum.
The articular surfaces of the zygapophyses

are large and are curved transversely. The
plane of the articular surfaces is close to the
horizontal, but the exact angle cannot be
established. Because of the large articular
surfaces, the members of each pair of zyga-
pophyses are not widely separated.

The neural spines are situated well toward
the posterior end of the vertebrae, centered
between the postzygapophyses. Each spine is
bordered both anteriorly and posteriorly by
paired ridges curving dorsally from each zy-
gapophysis. These ridges continue dorsally
for some distance. Between these ridges, the
base of the spine shows signs of lateral exca-
vation, but the fossa thus fonncd does not
liecome really conspicuous until the seventh
vertebra. The neural spine of the third cervi-
cal vertebra is low in lateral aspect. In dorsal
view, the spine is unusually wide and bears
two deep longitudinal grooves along the dor-
sal surface. The neural spines change in
shape and gradually increase in height pos-
teriorly along the column. The spine on
vertebra 5 is roughly rectangular in lateral
aspect and bears a distinct mammillary proc-
ess on each side. Posterior to the fifth verte-
bra the neural spines have a pair of mammil-
lary processes; these persist until vertebra 11.
The shape of these processes and their posi-
tion on the neural spine changes along the
colunm: in vertebra 5, the pair of mammil-
lary processes occurs anteriorly, in vertebra
6 it lies about midway along the spine and
by vertebra S it is near the posterior end.
The mammillary processes, which are also
seen in Araeoscelis and Prohicerta probably
provided attachment for strong intervertebral
muscles or ligaments (Vaughn, 1955; Cow,
1975).

Intercentra in the cervical region are rela-
tively large, well-developed elements; they
are up to 1.5 mm long in KUVP 9951 and
33607. The exposed ventral surface bears a
well-developed longitudinal ridge; there are
no capitular facets at the lateral extremities
of the intercentra. Each intercentrum extends
far dorsally between the centra.

1980

PENNSYLVANIAN DIAPSID REPTILE

33

Fig. 15. — Petrolacosaunis kansensis Lane. Dorsal, sacral and caudal vertebrae with some associated ribs,
all X 1.5. (A-C) Lateral view of vertebrae 7, 8, 13 and 14, KUVP 9951. (D) Lateral view of vertebrae 17
and 18, KUVP 9951. (E) Lateral view of vertebrae 25 and 26, KUVP 33606b. (F) Dorsal and lateral views
of first sacral neural arch and ribs, KUVP 33605. (G) Partially preser\'ed second sacral vertebra and rib in
anterior view, KUVP 33606b. ( H-I ) Lateral and anterior views of anterior caudal vertebrae and associated
ribs, KUVP 33606c.

Dorsal Vertebrae

Dorsal vertebrae (Fig. 15) are consider-
ably shorter than the cervical vertebrae. The
centra in KUVP 9951 decrease gradually in
length from 9 mm in anterior dorsal verte-
brae, to 8 mm in typical mid-dorsals and
down to 7 mm in posterior dorsals. The
centra in the dorsal vertebrae are stoutly
built, with a length to posterior height ratio
of 1.3:1 in vertebra 14, in contrast to the
condition in the cervical region where the

ratio is 2.5:1 (vertebra 3). The absolute
iieight of the centra increases slightly in the
dorsal region, and there is bevelling of the
ventral surface anteriorly and especially pos-
teriorly to accommodate the intercentra. In-
terccntral gaps are, therefore, small. Between
the anterior and posterior ends of the centrum
the lateral surface is strongly concave when
viewed in cross-section. The keel, seen in
cervical vertebrae, is retained in dorsal verte-
brae, but the ventral border of the keel is

34

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

broadly rounded as the longitudinal ridge
does not descend far vcntrally.

The neural arches of the dorsal vertebrae
contribute much less to the anterior centro-
sphcne than in cervical vertebrae. The trans-
verse processes in the dorsal region are not
only (juite different from those in the neck
but also undergo considerable change. The
transverse processes in the anterior dorsal
region have undergone a number of changes
from the pattern seen in the posterior cervical
vertebrae: (1) the transverse processes have
migrated anteriorly onto the prezygapoph-
yses; (2) the lateral elongation of the trans-
verse process has decreased; (3) although
the articular surface changes in configuration
to correspond to the cross-sectional outline of
the head on the dorsal ribs, there is no gap
between two oval areas of articulation as
there is in Araeoscelis, but a single elongate
surface of articulation; (4) the transverse
process, because of its position on the pre-
zygapophysis, is braced by the same ridges
as the latter; (5) in addition, however, a
well-developed ridge runs posteromedially
from the top of the transverse process to the
body of the arch.

Caudad from vertebra 8 the transverse
processes gradually decrease both in their
lateral projection from the surface of the
arch and in the length of the articular sur-
face. By vertebra 20, the articular surface is
quite short but still retains both "heads" on
the neural arch. Posterior to this vertebra, the
articular surface is located further antcro-
ventrally and continues to decrease in size
so that by vertebra 23 only the diapophyseal
area of attachment is present; it lies close to
the centrum near the anterior centrosphene.
Presumably the rest of the articular surface
for the rib is now part of the centrosphene.
In vertebra 26, the tiny articular surface sits
on a small knoblike projection at the antero-
vcntral corner of the neural arch as in Araeo-
scelis.

The pre- and postzygapophy.ses of all
dorsal vertebrae extend laterally slightly be-
yond the centra. The zygapophyseal planes

are tilted medially at about 10 degrees to the
horizontal. The zygapophyseal surfaces are
smaller than in the cervical region.

The neural arches of all the dorsal verte-
brae are excavated by deep lateral fossae.
The fossae are found by the prezygapophyscal
buttress that forms the anterior margin of the
neural spine and by the ridge that extends
anteroventrally from the postzygapophysis.
The fossae are most pronounced on mid-dor-
sal and posterior dorsal vertebrae. The neural
arches arc slightly swollen above the post-
zygapophyses, but only in the dorsal verte-
brae.

The gradual increase in neural spine
height, noted in the cervical region, continues
throughout the anterior dorsal region until
vertebra 18. The spines on vertebra 7 to 11
bear short dorsolaterally directed mammillary
processes; these are located on the posterior
half of the spine, close to the tip. The mam-
millary processes disappear completely by
vertebra 12. In a number of articulated pos-
terior dorsal vertebrae (KUVP 9951, 33605
and 33606 ) , the spines vary in size; every sec-
ond spine is either somewhat wider antero-
posteriorly (KUVP 9951 and 33606) or taller
(KUVP 33605) than the intervening one. The
function of such alternations in the size of
the neural spines is not understood. Similar
]>atterns of neural .spine size alternation have
been observed by the author in Captorhinus,
Eocaptorhimts, Araeoscelis and Lahidosaurus.
Well-de\elopcd intercentra are present
throughout the dorsal region of the column.
They are about 1.5 mm long in KUVP 9951.

Sacral Vertebrae

Sacral \ertebrae are preserved in KUVT
9951, 33605 and 3.3606. The first sacral verte-
bra is distinguishable from the posterior dor-
sal vertebrae by the following features: the
two prezygapophyses extend as far laterally
as on the dorsal vertebrae, but the postzyga-
pophyscs extend only about half as far from
the midline. The extremely massive dia-
pojihyses extend anterolaterally from high on

1980

PENNSYLVANIAN DIAPSID REPTILE

35

the neural arch. The neural spine is inclined
posteriorly and has a maniniillary-like process
(KUVP 33605) on each side at its postero-
dorsal tip. Anterodorsally, the spine bears two
longitudinal grooves along the crest, marking
the probable site of the strong interspinous
ligament that runs to the last dorsal vertebra.

The centrum is slightly smaller than those
of the posterior dorsal vertebrae. Ventral to
the diapophysis, but separated from it by a
deep groove, is the small longitudinal para-
pophysis. It lies on the lateral surface of the
centrum at the anterior end. The intercen-
trum does not fuse with the centrum as re-
ported in Araeoscelis (Vaughn, 1955) and a
number of other Paleozoic reptiles (Romer
and Price, 1940).

Less information is available on the sec-
ond sacral vertebra. The pre- and postzyga-
pophyscs conform to the width set by the
postzygapophyses of the first sacral vertebra.
The anterodorsal edge of the neural spine is
also grooved on the second sacral vertebra.
This suggests that the first and second sacral
vertebrae were also connected by well-devel-
oped interspinous ligaments. The neural spine
on the second sacral vertebra is considerably
smaller than the one on the first. The dia-
pophysis is also smaller, although it is still
considerably larger than the processes on the
mid-dorsal vertebrae. The distinct diapoph-
ysis and parapophysis form the articulation
with the rib. No infonnation is available on
the second sacral intercentrum or centrum.

Caudal Vertebrae

The anterior caudal vertebrae in Petrola-
cosaurus are somewhat similar to the second
sacral vertebra and are considerably altered
from the pattern seen in the dorsal series.
They are roughly 25 per cent smaller in all
dimensions than the mid-dorsals and conse-
quently are less stoutly built. The width of
the centmm is slightly greater than the height
and is equal to the length. The longitudinal
ventral ridge on the centrum does not extend
as far ventrally as in the mid-dorsals and its

ventral edge is not as broadly rounded. The
lateral excavations above the ridge are not as
deep as in the dorsal vertebrae. The zyga-
pophyses are lightly built with considerably
smaller zygapophyseal surfaces than in the
dorsals. The articular surfaces are tilted at
about 25 degrees to the horizontal plane.

The lateral surface of the neural arch of
the anterior caudal vertebrae is less deeply
excavated than in the dorsal vertebrae. No
swelling is apparent on the arch, above the
postzygapophyses. The spines are also much
smaller and alternation of width is not evi-
dent. The transverse process, on the other
hand, is more massive than on the dorsal
vertebrae and only slightly less well-devel-
oped than in the second sacral vertebra. It is
difficult to establish whether the diapophysis
is distinct from the parapophysis. The in-
completely exposed articular surfaces suggest
that only one, large, roughly oval surface pro-
vides the area of attachment for the rib.

The most anterior caudal intercentra are
similar to those of the sacral and presacral
vertebrae anterior to them. Posterior to the
second caudal vertebra the centra gradually
decrease both in width and height, but the
length remains relatively constant well into
the mid-caudal region. Centra in this region
of the vertebral column appear relatively long
and laterally compressed. At the fourteenth
caudal vertebra, the length to height ratio is
1.8 to 1; by the 39th caudal vertebra, the
ratio is 2.2 to 1. Both the ventral longitudinal
ridge and the lateral excavation of the centra
diminish gradually; by the sixteenth caudal
the ventral border of the centrum is gently
rounded. The neural arches also diminish
gradually in size. The diapophyses disappear
by the thirteenth caudal vertebra. The neural
spines diminish both in height and lateral
extent, disappearing at about the twenty-
fourth caudal vertebra. The spines of the
anterior and mid-caudal vertebrae follow the
general tendency of other primitive reptiles
to a slight backward inclination. This postero-
dorsal inclination increases as the spines de-
crease both in height and lateral extent pos-

36

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

teriorly along the series. Zygapophyses persist
beyond the fortieth caudal as simple anterior
and posterior projections above the spool-like
centra. They are, however, definitely lost by
the fiftieth caudal.

Associated with the five anteriormost cau-
dal vertebrae but not preserved in position,
are two chevron bones. The first chevron, in
KUVP 336()6c, probably fits between the third
and fourth caudals. As seen in several speci-
mens, they continue posteriorly at least as far
as the fortieth caudal vertebra. They are
apparently outgrowths of the intercentra and
consist of a crescentic intercentral portion,
and two thin ventral arms that converge dis-

tally to enclose a vertically elongate foramen.
The ventral projection is "Y"-shaped in pos-
terior view. The first chevron appears to be
small and delicately built. The second is much
larger, with a long ventral projection beyond
the foramen. As expected, the chevrons de-
crease in length posteriorly, albeit gradually.

Ribs

.\s in all known primitive reptiles, ribs are
present on every vertebra from the atlas to,
and including, the proximal caudal vertebrae.
The head of most ribs includes capitular and
tubercular portions, but these vary consider-

FiG. 16. — Pctrolacosaunts kanxctisis Lane. Ribs, all X 1-5. (A-B) Atlanta! and axial ribs in lateral view,
KUVP 9951 and 33604. (C) Ventrolateral view of fifth rib, KUVP 9956. (D) Medial view of sixUi (cerv-
ical) rib, KUVP 9956. (E) Medial view of .seventh (first dorsal) rib, KUVP 9956. (F) Anterodorsal view of
eighth rib, slightly enished, KUVP 9956. (G) Anterior view of rib 15 slightly crushed, KUVP 9951. (H)
Proximal ends of anterior dorsal ribs 9-11, KUVP 9956. (I) Posterior dorsal ribs 23 and 25, KUVP 33606b.
( J ) Ventral view of first sacral rib and lateral view of first sacral vertebra, KUVP 33606b.

1980

PENNSYLVANIAN DIAPSID REPTILE

37

ably along the column both in the degree of
their development and in their relation to one
another. In the anterior cervical, posterior
dorsal and sacral regions, the ribs have dis-
tinct gaps between the tubercular and capitu-
lar surfaces of articulation for the passage of
the segmental artery. Along the rest of the
presacral column the rib heads have a single
continuous hourglass-shaped area of articula-
tion, with rounded dorsal and ventral margins
and a thinner spanning portion. The dorsal
and ventral ends of the area of ai+iculation
probably correspond to the capitulum and
tuberculum of dichocephalous ribs. None of
the ribs is immovably fused to the body of the
vertebrae.

In the cervical region (Fig. 16) the shafts
of the ribs typically run without cur\'ing dis-
tally from the expanded heads. The atlantal
rib (KUVP 3.3607) is the smallest of the cervi-
cal ribs. The tuberculum and capitulum are
joined by a thin connecting web. The tu-
berculum articulated with the atlantal neural
arch, while the capitulum reached vcntrally
to the atlantal intercentrum. Distally from
the expanded head, the rib narrows rapidly
to a point. The second (axial) and third
cervical ribs are each more than twice the
size of the first. Although they retain the
same structural pattern as the first rib, they
do not taper to a point distally. Unlike that
of the first three ccrvicals, the distal end of
the fourth, fifth and sixth ribs is flattened and
expanded to a paddle-like blade, probably to
provide additional support for the levator and
serratus musculature that runs to the scapula.
The seventh cervical rib is almost twice as
long as the sixth, but unlike that of the dorsal
ribs, its shaft is nearly straight.

The fourth and fifth cervical ribs have a
small anterolaterally directed process close to
the anteroventral end of the head. The sixth
and seventh ribs have a thin dorsally directed
flange, running along the external edge of the
shaft. On the sixth this flange runs along
most of the length of the shaft, whereas on
the seventh it is restricted to the proximal
half.

In a typical dorsal rib (Fig. 16), the shaft
extends dorsolaterally from the capitulum,
whose articular surface is oval in outline and
occupies the head of the rib. The tuberculum
does not project markedly from the shaft. Its
articular surface is also oval in outline and
broader distally. The two heads are con-
nected by a narrow web of bone whose edge
apposes the corresponding articular surface
on the neural arch. Distally from the tuber-
culum, the rib curves strongly laterally and
ventrally. This curvature indicates that the
trunk was relatively high and narrow.

Except for the rather flattened promixal end,
typical dorsal ribs are oval in section. A small
ridge is usually present along the postero-
dorsal margin. Proceeding posteriorly along
the dorsal series, the posterior dorsal ribs as
far as rib twenty-one appear to be essentially
similar to more anterior dorsal ribs but show
a gradual decrease in length and in the size
of the head. From the twenty-second rib
back, this decrease becomes very pronounced.
Rib twenty-three is about half as long as a
typical dorsal rib. The shaft is slender and
almost straight. The tubercular and capitular
heads are separated by a small depression for
the passage of the segmental artery. The tu-
berculum of this rib still attaches to the neural
arch. The capitulum, however, attaches to a
small facet on the centrum, which is confluent
with the anterior centrosphene. Rib twenty-
five is still shorter and more slender.

The first sacral rib preserved in KUVP
9951, 99.56, 33605 and 33606b consists of a
short, massive, cylindrical proximal portion
that expands laterally to reach out towards
the inner surface of the ilium. Its distal mar-
gin is applied to the inner surface of that
bone but not fused to it. The first sacral rib
articulates with its vertebra, in a large round-
ed tubercular articulation that begins on the
neural arch and passes ventrally onto the
centrum. The capitulum is separated from
the tuberculum by a small groove and, as in
Araeoscelis (Vaughn, 1955) extends antero-
ventrally to attach to the intercentrum. Be-
yond the articular area, the neck of the rib

38

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

narrows slightly, then aljriiptly flares latoralTy
as a cupped expansion that widens below, and
is boundt^d anteriorly and posteriorly by two
heavy ridges. Between these ridges the rib is
strongly concave with its lateral surface al-
most vertical. The rib terminates laterally in
a broad, roughly "U"-shapcd strongly curved
border, which articulates with a correspond-
ing groove on the ilium.

The second sacral rib is much more slen-
der than the first. The smaller tuberculum
and capitulum articulate with the neural arch
and centrum respectively. The shaft of this
rib extends as far laterally as the first, but its
distal expansion against the ilium is less pro-
nounced. Its distal end, as in Araeoscelis
(Vaughn, 1955), has a dual function: bracing
the posterior end of the first rib from below
and articulating with the ilium.

The caudal ribs are markedly different in
character from those preceding them. They
extend laterally, curve slightly forward, then
turn sharply posteriorly in a horizontal plane.
The proximal part of each rib in KUVP
33606c has a longitudinal groove on the ven-
tral surface, reflecting the probable dicho-
cephalous character of the head, yet examina-
tion of the proximal surface of the rib reveals
no distinct tubercular and capitular surfaces
for attachment to the vertebra. A single, oval
surface serves this function. The shaft of the
rib is circular in section. Posteriorly the rib
tapers to a pointed tip. The first and second
caudal ribs are unusual in having a laterally
directed portion of the rib. This flange bears
longitudinal ridges, probably for muscle at-
tachment.

The anterior caudal ribs are elongate. The
first of the series reaches far beyond the pos-
terior end of the corresponding vertebra. Suc-
ceeding ribs run posteriorly parallel to, and
medial to the first. They diminish rapidly in
length and disappear by the eleventh or
twelfth caudal vertebra.

Pectoral Giixdle

The first description of the shoulder girdle
of Petrolacosaurus was based on an immature

.specimen, KUVP 1427 (Peabody, 1952, Fig.
2). The study of much better material, be-
longing to mature individuals in KUVP 9957,
33606 and 33607 indicates that Peabody's left
scapulocoracoid is in fact an incompletely
preserved scapula with no trace of either
anterior or posterior coracoids. In addition,
the elements tentatively identified as the
clcithrum and clavicle are quite different from
the well preser\'ed, complete cleithrum in
KUVP 33607 and clavicles in KUVP 9958 and
33606. They may, in fact, be interpreted as
is(}lated lower jaw elements since a splenial
lies nearby.

Cleithrum.— The cleithrum in KUVP 33607
(Figs. 12 and 13) is a long, narrow bone
with a slightly expanded head, quite similar
in general proportions to that found in Penn-
sylvanian captorhinomorphs (Carroll and
Baird, 1972). The head consists of an oval
plate that was probably applied to the an-
terior end of the cartilaginous extrascapula.
Its outer surface is only slightly sculptured.
The slender shaft is flattened. The medial
surface of tlie bone is strongly grooved to re-
ceive the anterior margin of the scapula. The
lower end appears to ha\e o\erlapped the
clavicle. The M. levator scapulae inferioris
presumably attached to the well developed
ridge that runs along the anterior margin of
the cleithrum (Holmes, 1977).

Clavicle.— The clavicle in KUVP 33606,
found in close association with the .skull and
other postcranial elements, is exposed in me-
dial view. A second, isolated clavicle belong-
ing to a juvenile indi\'idual is exposed in
lateral ^■icw (KUVP 9958, Fig. 17c). The
clavicle consists of a narrow dorsal process
and an expanded plate. The outer .surface of
the dorsal process shows two groo\'ed areas.
Near the upper end, the anterior margin of
the clavicle is longitudinally striated to re-
ceive the overlapping lower end of the cleith-
rum. The whole of the posterior margin of
the shaft is strongly grooved to receive the
anterior edge of the scapula. At the end of
the shaft the external lip of the groove broad-
ens posteriorly and conceals the groove from

1980

PENNSYLVANIAN DIAPSID REPTILE

39

Fig. 17. — Petrolacosatirus kansensis Lane. Pectoral girdle. (A) Lateral view of left scapulacoracoid in ar-
ticulation with limb, KUVP 9957a, X 0.75. (B) Ventral view of .slightly immature interclavicle, KUVP 9959b,
X 1.5. (C) Lateral view of left clavicle, immature, KUVP 9958, X 1.5- (D) Lateral and medial views of right
scapulacoracoid, KUVP 33609, X 2. See Figs. 12 and 13 for cleithrum.

lateral view. The narrow dorsal process of
the clavicle probably provided excellent at-
tachment for the clavicular portion of the
deltoid and trapezius muscles. The plate-like
portion of the clavicle is relatively massive in
comparison to that in primitive captorhino-
morphs, but is comparable to the clavicle of
Araeoscelis (Vaughn, 1955). This plate is in-
flected medially on the shaft at an angle of
about 45 degrees. The expansion of this por-
tion of the clavicle is formed largely by a
strong anterior curvature of the thickened
anterior margin. The lower portion of the
clavicle has striations running to the ventro-

medial margin. The ventromedial margin is
gently convex. It appears that the clavicles
covered the anterior part of the head of the
interclavicle head but did not meet at the
midline.

Interclavicle. — The isolated, immature, in-
terclavicle found in KUVP 9959b has a dia-
mond-shaped head and a long stem (Fig.
17b). The exposed ventral surface of the
head is strongly convex. The greater part of
the head is depressed, bears no sculpturing,
and lies internal to the clavicles. This iso-
lated element has been associated with Petro-
lacosaurtis partly because the depressed areas

40

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

on the head correspond closely in outline to
the medial end of the clavicular blade. In
addition, the rai.sed portion of the head, which
forms the posterior quarter of the diamond
and extends anteriorly between the clavicles,
bears the distinctive type of sculpturing seen
on the clavicle and the dermal skull roof
elements. The stem of the interclavicle is a
flat straight blade that runs along the ventral
midline of the pectoral girdle. It lies on the
external surface of the bony scapulacoracoid
and the unossificd sternum. The ventral sur-
face of the blade is missing in KUVP 9959b,
except for the posterior third, where a series
of bilaterally arranged longitudinal grooves is
seen. The anterior and posterior tips of the
bone bifurcate.

Scaptdocoracoid. — The scapulocoracoid
( Fig. 17d ) is present in four adult specimens,
KUVP 9957a, 33606c, 33607 and 33609, and
appears to be ossified as a unit with no su-
tures separating the anterior and posterior
coracoids and scapula. The only indication
of a division between the anterior and pos-
terior coracoids is a delicate crack that runs
dorsoventrally through the glenoid cavity in
the right scapulacoracoid of KUVP 9957a. In
immature specimens, however, the component
elements show postmortem separation.

In general proportions the scapulocoracoid
resembles those of the primitive captorhino-
morph Protorathyris archeri (Clark and Car-
roll, 1973, Fig. 6) and Araeoscelis (Vaughn,
1955) except that the- scapular blade in Petro-
lacosaurus is much wider. The straight dorsal
margin of the scapula is unfinished, indicating
that the endochondral girdle continued dor-
sally in cartilage. The anterior portion of the
blade is very thin transversely. The posterior
margin of the; scapula curves posterodorsally
as in Araeoscelis. Ventrally this margin is
continued as two ridges. One forms the pos-
terior boundary of the supraglcnoid buttress
and extends posteriorly onto the posterior
coracoid. A second, more prominent ridge
extends straight down to the massive anterior
glenoid buttress and forms the anterior boun-
dary of the triangular supraglcnoid buttress.

As in Paleothyris and Protorothtjris the supra-
glcnoid foramen opens externally on this
second ridge, just above the tip of the supra-
glcnoid buttress.

The glenoid cavity has the usual screw-
shape seen in most early tctrapods (Romer,
1956; Holmes, 1977). The anterior part of
the glenoid projects far laterally, supported
by the massive anterior glenoid buttress. The
external opening of the coracoid foramen lies
within a deep pocket, beneath this buttress.
Behind the glenoid there is a prominent proc-
ess for the origin of the coracoid head of the
triceps muscle. This structure, common in
pelycosaurs, is also found in Protorathyris and
Araeoscelis. The large external surface of the
anterior coracoid area is smoothly confluent
with that of the scapular blade. Close to the
ventral margin, the external surface of the
coracoids is unfinished.

The medial surface of the right scapulo-
coracoid in KUVP 33609 closely resembles
that of Protorathyris (Clark and Carroll,
1973). A system of two ridges has the outline
of an inverted "T." One of these ridges
extends vertically onto the scapular area me-
dial to the supraglcnoid buttress. The second
diverges anteriorly and posteriorly from the
base of the first and supports the anteroven-
tral part of the anterior coracoid and the
glenoid respectively. In contrast to the con-
dition in Protorathyris, where there are two
internal openings of the supraglcnoid fora-
men, there is only a single internal opening in
Petrolacosaurus, in the upper part of the deep
subcoracoscapular fossa. The coracoid fora-
men probably opened into the lower part of
the subcoracoscapular fossa, as in other early
reptiles, but this area of the fossa is slightly
damaged in KU\'P 33609.

Pelvic Girdle

Both Lane (1945) and Peabody (1952)
based their description of the pelvic girdle on
the isolated left half of a pelvis (KUM' 1425)
tliat is definitely tliat of an edaphosaurian
pelycosaur. There is now conclusive evidence
that this element does not belong to Petrola-

1980

PENNSYLVANIAN DIAPSID REPTILE

41

Fig. 18. — Pctrolacosatirus kansensis Lane. Pelvic girdle, X 1.5. (A) Lateral \iew of slightly crushed left
pelvis, KUVP 9961. (B) Medial view of right iliac blade, uncrushed, KUVP 33606c. (C) Medial view of
shghtly crushed left pelvis, KUVP 9951.

cosaurus. In KUVP 33606 part of the left
side of a mature pelvis, exposed in medial
view, is closely associated with a series of
posterior dorsal, sacral and anterior caudal
vertebrae, on a small block which also con-
tains other remains of Petrolacosaurus. In
KUVP 9951c the complete pelvis, exposed in
lateral and medial views, is closely associated
with both hindlimbs and lies near a series of
articulated presacral vertebrae and a scat-
tered skull of the same genus. These pelves
are unique in several features and can be
readily distinguished from those of other
Paleozoic reptiles. The pelvis mistakenly as-
cribed by Lane and Peabody to Petrolaco-
saurus probably pertains to Edaphosaurus sp.

which has also been found in the Gamett
quarry.

In its general proportions and in the struc-
ture of the component elements, the pelvic
girdle (Fig. 18) resembles that of Araeoscelis
more closely than the girdle of any other rep-
tile, but also retains a number of primitive
features seen in primitive captorhinomoriDhs.
In contrast to the fectoral girdle, which in
Petrolacosaurus is lightly built and is ossified
as a unit in mature individuals, the pelvic
girdle is massive and the sutures between the
component elements are always easily dis-
cernible in medial view. The acetabular cav-
ity is primitive in its configuration, with the
three raised comers of the cavity supported
by prominent buttresses on each of the three

42

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

pelvic elements. In both KUVP 9951 and
9961 the unfinished articular surface of the
acetaijulum extends to the posterior margin of
the pelvis, as in Araeoscelis (Vaughn, 1955).

Ilium. — The ilium has a long blade which
extends far posterodorsally and only slightly
anteriorly. This condition, also seen in all
captorhinomorphs, Araeoscelis and ophiaco-
dont pelycosaurs, is considered primitive. The
posterior end of the blade in Petrolacosaurus
is thin in cross-section, and its lower corner
tends to have been lost during preservation.

The internal surface of the ilium, superbly
preserved in KUVP 33606c, is more significant
than the lateral. Near the dorsal edge a well
developed shelf of bone extends medially
along much of the length of the iliac blade.
This process, not found in captorhinomorphs,
is well developed in Araeoscelis (Vaughn,
1955), and in ophiacodont pelycosaurs. The
central region of the medial iliac surface is
slightly recessed and rugose below the me-
dially directed shelf to receive the sacral ribs.
Neither of these structural features of the in-
ternal surface of the ilium are found in capto-
rhinomorphs. The posterior half of the medial
surface of the blade shows a series of longi-
tudinal ridges, similar to those seen in the
primitive captorhinomorph Coelostegus (Car-
roll and Baird, 1972, Fig. 14) for the origin of
epaxial musculature.

Pubis. — The pubis, well preserved in both
KUVP 9951c and 9961 is massi\ely built. As
in Araeoscelis, the anterolateral margin is
formed by a thick ridge that is a continuation
of the anterior iliac ridg(\ A large tubercle
projects dorsolaterally from this ridge. Its
function, according to Vaughn (1955) was to
raise the anteroventral end of the iliopubic
ligament in the absence of a well developed
anterior expansion of the iliac blade. This
tubercle is also found in Palcothyris and in
primitive ophiacodonts, but in these genera
is much smaller than in either Petrolacosaurus
or Araeoscelis. A second, smaller process ex-
tends ventrolaterally from the ridge connect-
ing the lateral pubic tubercle to the acetabu-
lum. This process, also seen in Araeoscelis,

probably served as the origin for the ambiens
and pubotibialis muscles. The true pubic tu-
bercle, located at the distal end of the ridge is
also massive. As in other early reptiles the
pubic part of the puboischiadic plate faces
more or less vcntrally and is not fully seen
in lateral view. The external opening of the
obturator foramen, which is much larger than
in primitive captorhinomorphs, lies directly
beneath the anterior end of the acetabulum.
On the internal surface of the pubis, the un-
finished symphysial surface is relatively nar-
row, running along the ventromedial margin.
The internal opening of the obturator foramen
is obscured in all the preser\ed specimens,
but probably lies on the upper part of the
large pubic surface which faces not only
dorsomedially but also anteriorly.

Ischium. — The ischium in Petrolacosaurus
is the most distinctive element of the pelvis.
In captorhinomorphs and pelycosaurs the long
dorsal margin of the ischium is strengthened
by a rounded ridge. No such ischiadic ridge
is found here. Posterior to the acetebulum,
tlie thin dorsal margin of the ischium is
deeply notched, a feature unique to Petrola-
co.^aurus. The posterior margin of the notch
flares dorsolaterally. Posterior to the notch
the margin of tlie ischium remains thin and
cur\es medially to the posteromedial corner
of the bone. The M. ischiotrochantericus ap-
parently passed from the posteromedial sur-
face of the ischium laterally ventral to the
ilioiscliiadic ligament to insert onto the proxi-
mal end of the femur. The dorsolateral flare
of the ischiadic notch probably served as the
attachment area for a shortened and hence
stronger, ilio-ischiadic ligament. The notch
itself likely served as a lateral passage for M.
ischiotrochantericus. On the internal ischiadic
surface a ridge extending posteromedially
from the notch, roughly parallel to the pos-
terior margin, limits the area of origin of the
above nniscle. This posterior area of the
ischium is slightly offset from the general
interna] surface. No such specialization is
seen in any other early reptile.

1980

PENNSYLVANIAN DIAPSID REPTILE

43

i (111

Fic. 19. — Petrohcosanrus kamensis hane. Reconstruction of tlie right humerus, X 1.5. (A-D) Proximal dor-
sal, proximal ventral, distal dorsal and distal ventral surfaces, based mainly on KUVP 9957 and 33606a.

Limbs

A total of twenty-six limbs, belonging to
fourteen individuals have been recovered.
Most of these were originally complete as
they lay in the matrix. Only eight were com-
pletely isolated from the body. Of the limbs
associated with the girdles, only one was in
articulation (KUVP 9957a, Fig. 17).

Humerus. — The reconstruction of the hu-
merus (Fig. 19) is based mainly on two ma-
ture but crushed bones, found in KUVP 9957a
and 33606a. The humerus is of the tetrahe-
dral type common to primitive reptiles
(Holmes, 1977) but the shaft is better de-
veloped than in primitive captorhinomoiphs
and pelycosaurs. The width of the proximal
end of the humerus forms about 27 per cent
of the length of the bone, and the distal width
is about 29 per cent of the length. The shaft
is exceedingly slender, only about 7 per cent
of the length. These proportions are compara-
ble to those found in Araeoscelis (Vaughn,
1955).

The estimated twist of the distal upon the

proximal plane is 60 degrees. Following
Romer and Price's (1940) terminology, the
humeral surfaces in Petrolacosaurus can be
divided into proximal dorsal, distal dorsal,
proximal ventral and distal ventral surfaces.

The proximal dorsal surface was probably
gently convex along the long axis of the bone.
Severe crushing flattened the proximal head.
The surface shows a distinct ridge for the
M. latissinius dorsi. A long narrow ridge for
the M. deltoideus insertion runs distally from
the delto-pectoral crest.

The proximal ventral surface was probably
deeply concave below the articular surface.
Tlie M. coracobrachialis brcvis inserted in the
concave region. The anterior proximal area,
above the delto-pectoral crest, is rounded and
lacks any rugosity. In modern lizards, the M.
supracoracoideus inserts on this area ( Holmes,
1977). Distally, a distinct delto-pectoral crest
juts out anteroventrally from the proximal
head. At the summit of this crest is a rela-
tively small circular area with a rugose sur-
face to which attached the powerful pec-

44

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

toralis inusculature. Distally, the crest slopes
rapidly into the shaft; its internal border sends
a delicate ridge distally to the entepicondyle.

A well-developed entepicondylar foramen
and a large snpinator process distinguish the
distal head of the humerus of Petrolacosaurus
from that of most pelycosaurs (Romer and
Price, 1940) as well as from Araeoscelis. The
radial nerve was only partially surrounded by
bone as it traversed the humerus, whereas
in Araeoscelis it was fully enclosed in bone
( ectepicondylar foramen ) .

The entepicondyle does not extend far
laterally. Its rugose lateral and distal margin
furnished the areas of origin of the flexor
musculature of the lower arm and foot. The
inferior humeral nerve apparently ran along
the deep groove on the dorsal surface of the
entepicondyle, passed through the elongate
entepicondylar foramen, and entered the fore-
arm, as in S})Iien()don (Miner, 1925). On the
distal dorsal surface, the entepicondyle is
separated from the ectepicondyle by a long

groove that widens distally. The moderately
developed ectepicondyle is a long ridge that
widens as it lains distally. Its end projects
distally above and beyond the general con-
tours of the humerus. Its rugose distal head
furnished a small area for the origin of much
of the extensor musculatine of the lower arm
and foot. Anteriorly the ectepicondyle is
bounded by the long ectepicondylar groove.
In pelycosaurs the entepicondyle is usually
larger and the ectepicondyle projects further
dorsally than in Pctrolacosaiinis (Romer and
Price, 1940).

Anterior to the deep ectepicondylar
groove, which carried the radial nerve, the
supinator process lies below the level of the
ectepicondyle and the general dorsal surface.
Unlike that in pelycosaurs, the supinator proc-
ess in Petrolacosaurus does not project strong-
ly outward but extends far distally. Its rugose
tip lies close to the distal end of the bone
and probably furnished a small area for the
origin of the supinator muscles. The fully

I (III

Fic. 20. — Petrolacosaurus kamcnsis Lano. Left radius (A) and ulna (B) in dorsal (prea.xial) view, KUVP
9957, X 2.

1980

PENNSYLVANIAN DIAPSID REPTILE

45

enclosed ectcpicondylar foramen in Araeo-
scelis extended the possible area of origin of
the supinator musculature, as in modern liz-
ards (Romer, 1944). A ligament, running
from the supinator process to the ectepicon-
dyle in Petrolacosaurus, may have served the
same function.

The lateral edges of the distal ventral sur-
face are formed by the margins of the entepi-
eondyle and the supinator process. Between
these margins the ventral surface is relatively
flat but is pierced by the oval entepicondylar
foramen. Much of the distal end of the hu-
merus is occupied by the radial and ulnar
articulations. The radial articulation (capi-
tellum) is a slightly elongate swelling with
an unfinished surface. The capitellum not
only projects ventrally but also slightly dis-
tally, suggesting that the epipodials met the
longitudinal humeral plane at an obtuse angle.
The capitellum is contiguous with the ulnar
articulation posterior to it. The concave sur-
face of the latter faces essentially distally.
Proximally from the articular area, the ventral
surface is slightly depressed.

Radius. — The radius (Fig. 20) is a long,
slender element with slightly expanded ends.
In the articulated left forelimb of KUVP
9957a (Fig. 17) the length of the radius is
.92 per cent of the length of the humerus.
The width of the proximal end measures about
14 per cent of the length of the bone and the
distal width about 12 per cent of the length.
The proximal head is cup-shaped for the
humeral articulation. Because of crushing, its
outline cannot be determined. The lateral and
medial edges of the bone, seen in anterior
view, are partially formed by two longitudinal
ridges that probably separated the extensor
and flexor muscles. The medial ridge is quite
prominent proximally and is slightly nigose.
The distal head of the radius is turned slightly
laterally so that the autopodial surface can be
seen in lateral view.

Ulna. — The ulna (Fig. 20) is also quite
long and slender; it is slightly longer than the
humerus. The maximum width of the proxi-
mal end approaches 15 per cent of the length

of the element, whereas the distal end is
narrower — about 12 per cent of the length.
The shaft, as in the radius, narrows to 5 per
cent of the length of the bone. The ulna has
a well developed olecranon and a concave
sigmoid notch for articulation with the hu-
merus. The tip of the olecranon is capped
by a ridge that curves over the apex of the
bone. The main tendon of the triceps muscu-
lature probably attached here. The sigmoid
notch occupies the whole thickness of the
medial side of the head. The articular surface
is higlily curved, turning through an arc of
about 140 degrees, while the distal end of the
notch faces proximally, its proximal end faces
medially and slightly distally. The anterior
surface near the humeral articulation is ru-
gose. The anterior and posterior surfaces are
separated medially by a rounded ridge. The
distal end of the ulna is expanded and is
twisted slightly laterally. The autopodial sur-
face is convex.

Femur. — The reconstruction of the femur
( Fig. 22 ) is based on the holotype KUVP
1424, as well as on KUVP 995ld and 33606.
Dorsoventral crushing has flattened the fe-
mora and bent the internal trochanter. This
prevents consideration of the femur in an-
terior and posterior views. The femur is
closely comparable to that of primitive capto-
rhinomorphs and pelycosaurs. It is more
massive than the femur in Araeoscelis
(Vaughn, 1955).

The width of both the proximal and distal
ends of the bone is about 30 per cent of the
length. The relatively massive shaft narrows
to about 13 per cent of the length.

The proximal head of the femur is termi-
nal as in all primitive reptiles and not in-
flected medially. The articular surface is
crescentic. The anterior portion of the artic-
ular area is the broadest. Distal to the head,
a prominent anterior crest extends onto the
ventral surface of the femur. The crest's
proximal end probably forms the internal
trochanter for the tendinous insertion of the
M. puboischiofemioralis externus. The distal
extension of the crest along the shaft forms a

46

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Fio. 21. — Petrnlocasaurus kansrnsis Lane. Rctonstniction of the right manii.s in dorsal (A) and ventral
(B) views, based mainly in KUVP 1423, 8,355 and 9957b, x 2.0. Abbreviation.s used in this FiKure: i, inter-
medium; Ic, lateral centrale; mc, medial eentrale; pis, pisiform; r, radiale; ii, ulnare; 1-5, di.stal eaqials; I-V,
metacarpals.

linea aspera for the insertion of the- addnctor
musculature. Tlicrc is no distinct fourth tro-
chanter. The ventral surface of the proximal
head is occupied by the deeply hollowed
intertrochanteric fossa for the M. puboischio-
fcmioralis externus.

In dorsal view, the proximal head has a
smoothly curved, conical .surface broken pos-
teriorly by a well developed tuberosity ap-

parently for the probable insertion of the M.
ischiotrochantericus. The smooth, cur\ed, dor-
sal surface extends distally onto the broad
shaft. Two prominent ridges on the dorsal
surface of the distal head run [Moximally from
the articular area of the anterior and posterior
condyles respectively. The extensor musctda-
ture of the lower leg attached to these longi-
tudinal ridges. He\'ond these areas of muscle

1980

PENNSYLVANIAN DIAPSID REPTILE

47

Fig. 22. — Petrolacosaurus kansensis Lane. Reconstruction of the left femur in dorsal (A) and ventral (B)
views, based mainly on KUVP 9951 and 33606, x 1.5.

attachment, the dorsal condylar surfaces are
concave. Between these ridges the femur is
notched dorsally by a deep intercondylar
fossa for the tendon of the quadriceps muscle.
The surfaces for articulation with the tibia
and fibula extend onto the dorsal surface.

The distal ventral surface of the femur is
deeply concave and rugose. A prominent
posterior ridge extends distally along the pos-
terior condyle. Its continuation onto the ar-
ticular area forms the boundary between the
areas of articulation with the tibia and fibula.

Tibia. — The tibia and fibula are known
in at least two specimen;; — KUVP 9951b and
33606. The tibia (Fig. 23) is quite long, its
length equalling that of the femur. As with
other distal limb bones, the tibia is slender.
The width of the head in one specimen is
about 20 per cent of the length of the bone,
and the distal width is nearly 12 per cent of
the length. The shaft has a minimum diam-
eter of only 5.5 per cent of its length. These
proportions are comparable to those in Araeo-
scelis (Vaughn, 1955).

The tibia is immediately identifiable by
the conspicuous anterior concavity of the
shaft and by the large head. The articular
surface of the head is similar in outline to
that of primitive captorhinomorphs (Carroll,
1969) and pelycosaurs (Romer and Price,
1940). The proximal portion of the tibia is
roughly triangular in cross-section as is the
outline of the articular surface. Two well
developed ridges pass distally along the lat-
eral and anterior margins. The latter prob-
ably afforded attachment for an interosseous
ligament. As in Araeoscelis the distal articu-
lar surface of the tibia can be divided into
two parts that form a fimi, locked joint with
the astragalus (Vaughn, 1955).

Fibula. — The fibula (Fig. 23) is very long
and slender. The width of the head in a well
ossified specimen is about 10 per cent of the
length of the bone; the distal width about 15
per cent of the length. These proportions are
also comparable to those in Araeoscelis
(Vaughn, 1955). The fibula in Petrolaco-
saurus is rather blade-like, with prominent

48

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Fic. 23. — Petrolacosaurtis kansensis Lane. Tibia and fibula, X 1-5. (A-B) Dorsal and ventral views of
the left tibia in KUVP 9951. (C) Distal articular surface of (B). (D-E) Dorsal and ventral views of the left
fibula in KUVP 9951.

medial and lateral ridges. The medial ridge
for the interosseous ligament is blade-like.
The medial margin of the fibula, formed by
the above ridge, is slightly concave. The
proximal portion of the filiula is somewhat
twisted in relation to the distal one. The
proximal articulatory surface for the femur
is key-hole shaped, due to crushing. The
distal surface, on the other hand, is elongate
and divided into distinct astragalar and cal-
caneular areas.

Mantis. — The manus is well represented
by eight specimens. Nearly all features of
the manus are seen in both dorsal and ven-
tral views. The general number and arrange-
ment of bones resembles closely that of cap-
torhinomorphs. The preserved carpal bones
are cither in perfect articulation, or only
slightly disarticulated. All the elements are
well ossified with fully developed articulating
surfaces. They appear to have fitted com-

pactly together to form a well knit structure
with little cartilage between articular sur-
faces (Fig. 21).

In general proportions, the carpus in Pe-
frolacosaurus is similar to that in Araeoscelis
(Vaughn, 1955). It is slightly more elongate
than the carpus in primitive captorhinomoqihs
(Carroll and Baird, 1972). This elongation is
probably related to the elongation of the
radius and ulna. The ulnare is the largest
element of the carpus, is attached proximally
to the ulna and the small pisiform, and is
supported distally by the fourth and fifth
distal carpals. It varies in outline from a
discoid in KUVP 8355 to a highly elongate
structure in KUVP 33606, equal in length to
an anterior dorsal vertebra. It is possible to
account for this variation by the apparent im-
maturity of the former specimen. The ulnare
of tlie mature individual, seen in KUVP 33607
is a massive element, braced by a pair of

1980

PENNSYLVANIAN DIAPSID REPTILE

49

ridges near the medial edge. The more prom-
inent of the two ridges extends on the dorsal
surface between the proximal and distal artic-
ular surfaces. A similar, but smaller, ridge
exposed in KUVP 9957b, extends proximodis-
tally on the ventral surface. Both the proxi-
mal and distal articulatory surfaces are con-
tiguous with a pair of unusually massive
medially and vcntromedially directed facets
that articulate with the intermedium and
medial centrale respectively. Between these
facets, the medial margin of the bone is
deeply notched to form the lateral surface of
the perforating foramen.

The intermedium in Petrolacosaurus is
also an elongate element, although it is short-
er than the ulnare. In KUVP 33607 the inter-
medium is two-thirds the length of the ulnare.
The surface of the proximal articulation with
the ulna is continuous with a massive medial
articular surface with the ulnare. This sur-
face, exposed in both KUVP 9957b and 33607,
faces somewhat ventrally. The small, oval
distal articulation of the intermedium is re-
stricted to the lateral centrale, as in Araeo-
scelis (Vaughn, 1955) and pelycosaurs (Ro-
mer and Price, 1940), rather than being
supported by both the lateral centrale and
ulnare, as in Paleothyris (Carroll, 1969). A
well developed ridge extends proximodistally
along the lateral edge of the intermedium.
Between the articular surfaces with the lunare
and the lateral centrale, the intermedium
forms the rounded medial surface of the per-
forating foramen. The medial edge of the
intermedium is thin and is free of contact
with the radiale.

The radiale, best preserved in KUVP 8,3.55
is a short, massive element supported by both
the medial and lateral centrale. As in Atqco-
scelis and numerous other early reptiles the
radiale in Petrolacosaurus bears a longitudi-
nal ridge on its dorsal surface. On the ventral
surface, this element lias a pronounced proxi-
mal depression, which is interrupted by a
proximally directed ridge. The radial surface
is nearly transverse to the long axis, as in
Araeoscelis (Vaughn, 1955) rather than being

strongly tilted dorsally as in many pelycosaurs
(Romer and Price, 1940). The bulbous distal
articular surface fits into the angle formed by
the centralia.

The lateral centrale in Petrolacosaurus is a
relatively short element which has contacts
with most of the other carpal bones. In addi-
tion to supporting the three proximal carpal
elements, it also articulates with the medial
centrale and the third and fourth distal car-
pals. Between the ventrolaterally directed
process for articulation to the ulnare and the
proximal articulation to the intermedium, the
medial centrale contributes to the margin of
the perforating foramen. The dorsal surface
of the lateral centrale also bears a prominent
longitudinal ridge. Whereas in primitive cap-
torhinomorphs (Carroll, 1969) and pelyco-
saurs (Romer and Price, 1940) the lateral
centrale is elongate and the medial centrale
is short and broad, in Petrolacosaurus the two
centralia are similar in shape.

The medial centrale is similar in shape to
that in Araeoscelis (Vaughn, 1955). It sup-
ports proximally much of the weight of the
radiale and articulates distally not only with
the first and second distal carpals but also
with a portion of the third. Its ventral sur-
face, exposed in KUVP 9957b bears a promi-
nent finger-like process that extends towards
the first distal carpal.

The fourth distal carpal, as in other early
reptiles, is the largest of the series, and bears
most of the support of the ulnare. Medially
it does contribute to the support of the lateral
centrale. The ventral surface of the fourth
distal carpal is considerably larger than the
dorsal and overlaps much of the head of the
fourth metacarpal and a portion of the small
triangular fifth distal carpal. The third distal
carpal supports most of the weight of the
lateral centrale and part of the medial cen-
trale. It is only slightly smaller than the
third, at least in dorsal view. In KUVP 9957b
the third distal carpal is exposed in both dor-
sal and ventral views. The ventral surface of
this clement is considerably smaller than its
dorsal surface because the large surface of

50

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

articulation with tlii' fourth distal carpal faces
ventrolatcrally. In addition to this unusual
feature, the third distal carpal also has a
prominent finger-like process that extends dis-
tally beneath the head of the third meta-
carpal, well beyond the carpal-mctacarpal
articulation. The finger-like processes on the
ventral surface of the third distal carpal and
the medial centrale and the ridge on the
radiale may have provided attachment areas
for the flexor musculature of the forelimb.

The first and second distal carpals support
the medial centrale and are comparable in
size to the fifth.

A complete set of metacarpals occurs in
perfect articulation with the rest of the imma-
ture forelimb in KUVP 1423 ( Peabody, 1952,
Fig. 6c). These metacarpals show a range in
length from 4 mm (first) to 14.5 mm
(fourth). The proximal heads overlap one
another, as in Araeoscelis (Vaughn, 1955)
and many diapsids (Carroll, 1975), restricting
considerably independent metacarpal move-
ment. Such metacarpal overlap is not found
in captorhinids (Holmes, 1977). As indicated
by Peabody (1952, pp. 28 and 29) both meta-
carpals and phalanges are relatively long and
slender, as in other small early reptiles. The
phalangeal count is 2, 3, 4, 5, 3. The terminal
phalanges are developed into long, sharp
claws, similar to those found in primitive cap-
torhinomorphs and early pelycosaurs.

One of the most striking features of the
manus in Petrolacosaurus relates to the orien-
tation of the articular surfaces between the
ulnare and intermedium, between the ulnare
and lateral centrale, and between the third
and fourth distals. The ventral edges of these
surfaces are widely separated if the carpus
is reconstructed as a comiiletely flattened
structure (Fig. 21). This indicates that in
life Petrolacosaurus had a strongly curved
carpus, with a medio-laterally convex dorsal
surface. Most of this cur\'ature was formed
along the long axis of the carpus. The extent
of this curvature cannot be established with
certainty, ho\\(\er, because of crushing. The
type of cariiai cnr\ature seen in Petrolaco-

saurus has not been found in any other Paleo-
zoic reptile. This may, howe\er. be due to
the general lack of information about the
\entral aspect of the carpus in early reptiles.

A definite joint plane is lacking within the
carpus, and even between the carpus and the
paired distal limb elements. Movement be-
tween the carpus and the metacarpals appar-
ently was also slight because the proximal
heads of the metatarsals overlap one another,
the line between the carpals and metacarpals
is curved, not straight, and the distal ends
of the third and fourth distal carpals restricted
the downward movement of the correspond-
ing metacarpals considerably. It is probable,
therefore, that bending during locomotion
was achie\ed by discreet accommodation
throughout the manus.

Pes. — The pes is almost completely articu-
lated in the type specimen KUVP 1424 (Pea-
body, 1952, Fig. 9), but most of the bone is
gone, leaving an impression on the matrix.
The interpretation of the tarsus given here
( Fig. 24 ) is based on well preser\ed left and
right tarsi, exposed in both dorsal and ventral
views (KUVP 9951a and b) and a partial
right tarsus (KUVP 33606) from mature indi-
viduals. Peabody suggested that the first tar-
sal bone was lost in Petrolacosaurus. This
element is, however, present in both the left
and right tarsi on KUVP 9951a and b. The
general arrangement of the tarsal elements
resembles that of Araeoscelis (\'aughn, 1955)
and to a lesser extent that of the primitive
captorhinomorph Paleothyris (Carroll, 1969).
The shape and level of ossification of the
component elements indicates that the pes
was a compact structure, as was the mature
manus.

The astragalus is the most distinctive ele-
ment in the tarsus. It is a long, relatively
slender bone. As in other primitive reptiles
it is essentially "L"-shaped, with a distinct
neck region, the proximal end of which forms
part of the tarso-fibular articulation. The
dorsal surface of the astragalus, exposed in
KUVP 9951b, is gently concave. The lateral
border of the astragalus forms a massive

1980

PENNSYLVANIAN DIAPSID REPTILE

51

articulation with the calcaneiim along the
length of the bone, interrupted only near the
distal end by a small notch that forms the
medial border of the perforating foramen.
This longitudinal articular surface faces dor-
sally, in part, in KUVP 9951b suggesting that
the pes, like the manus, is curved transversely
but with a concave dorsal surface. The latcro-
distal corner of the astragalus bears a condy-
loid process, similar to that in Araeoscelis
(Vaughn, 1955), which articulates with the
calcaneum laterally and to the fourth tarsal
distally. The distal border of the astragalus
is notched between the condyloid corner and
the articulation with the centrale. In con-
trast to the condition in captorhinomorphs and
pelycosaurus, where the tibial surface on the
astragalus is convex, that of Petrolacosaurus
(as it is exposed in medial view in KUVP
33606) is strongly concave, with two massive
ridges, separated by a trough. One of these
ridges forms the thickened distal two-thirds
of the medial border. The second prominent
ridge located on the ventral surface runs
roughly parallel to the lateral border. This
arrangement of a pair of ridges separated by
a deep trough is also seen in Araeoscelis and
in the eosuchian Kenyasaitrus (Harris and
Carroll, 1977). The distal articular surface of
the tibia is divided into a lateral trough and
a medial ridge, which fits snugly onto the
astragalus to fomi a finn, locked joint. On
the \entral surface of the astragalus a strong
ridge connects the distal lip of a deep fossa,
which extends to the perforating foramen,
with the proximal end of the tibial surface.
A notch in the distal border, exposed in dor-
sal view, extends onto the ventral surface as
a small fossa which recei\es the latero-proxi-
mal end of the centrale.

The calcaneum in Petrolacosaurus is a
relatively simple oval stiiicture with a thick
medial portion that supports the proximal
fibular facet and the distal tarsal articulation.
A somewhat thinner plate extends laterally.
Both dorsal and ventral surfaces are gently
concave. The straight medial margin, which
articulates with the astragalus, is interrupted

by the small notch that forms the lateral
margin of the perforating foramen. The
broad, rounded distal border of the calca-
neum apposes the fourth and fifth distal tar-
sals. An unfinished area, marking the area of
articulation with the distal tarsals, extends
far onto the ventral surface of the calcaneum.
Except for this extended area of articulation
there is nothing to distinguish this calcaneum
from those of pelycosaurs.

In contrast to the condition in primitive
pelycosaurs, where there are both lateral and
medial centralia in the pes (Romer and Price,
1940), there is only a single large centrale in
Petrolacosaurus and most captorhinomorphs.
It supported the astragalus. This element,
completely preserved in KUVP 9951a and b,
is much compressed proximodistally. It artic-
ulates with the rounded distomedial surface
of the astragalus. The proximal articular sur-
face of the centrale is deeply concave, where-
as the distal surface for articulation with the
first three distal tarsals and the lateral surface
for articulation with the fourth distal tarsal
are strongly convex. Both dorsal and ventral
surfaces of the centrale show a central de-
pression bound proximally by the ridged mar-
gin of the bone.

All the distal tarsals are well preserved in
KUVP 9951a and b and only slightly disartic-
ulated from one another. As in other early
reptiles, the fourth distal tarsal is the largest
of the series and bears much of the weight
of both the astragalus and calcaneum. The
dorsal and ventral surfaces are gently concave.
The dorsal surface of this bone is contiguous
distally with the broadly convex articular sur-
face for the fourth metatarsal. All other artic-
ular surfaces of the fourth tarsal are sharply
offset from both the dorsal and ventral sur-
face's. Medially the fourth distal tarsal artic-
ulates with the centrale and the third distal
tarsal by a complicated system of concave
and convex surfaces (Fig. 24). The proximal
articular surface is strongly concave. Later-
ally the fourth distal tarsal articulates with
the fifth distal by an extensive dorsolaterally
inclined surface. The first distal tarsal is also

52

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

deeply concave pioximally in Petrohicosaurus
and matches the length of the fourth. The
lateral surface of the fifth distal tarsal is
smoothly continuous with the narrow distal
surface, and is unfinished. This indicates that
both these surfaces articulated with the large
head of the fifth metatarsal. The fifth meta-
tarsal was, therefore, widely divergent.

The third distal tarsal not only has rela-
tively small, simple, dorsal and ventral sur-
faces, hut also has extensive and complicated
articular surfaces with the centrale and the
second and fourth distal tarsals. The first and

second distal tarsals, like the third, have
larger lateral and medial surfaces than cither
the dorsally or ventrally exposed areas.

The articular surfaces between the proxi-
mal and distal elements of the tarsus extend
in a straight line roughly peipendicular to the
long axis of the pes; in addition, the proximal
surface, formed by the astragalus and calca-
neum has a rounded, strongly convex outline
and fits into the strongly concave distal artic-
ular surface fomied by the centrale and the
fourth and fifth distal tarsals. The configura-
tion of the mesotarsal articulation indicates,

Fit;. 21 — PetTolacosaunts kan.wmis Lane. Pes, all X l-S- (A) ReconstmcHon of the left pes, in dorsal view,
based mainly on KUVP 1424 and 9951b. (B) Ventral view of the left tarsus, restored with the mesotarsal
surfaces exposed, KUVP 9951a. (C) Medial view of the right astragalus, KUVP 33606. (D) Metatarsals in
dorsal view, KUVP 99511). .abbreviations used in this Figure: as, a.stragalus; caj, caleaneum; cp, centralia
pedis; fib, artieulation with fibula; tib, articulation witli tibia; 1-5, distal tarsals; I-V, metatarsals.

1980

PENNSYLVANIAN DIAPSID REPTILE

53

therefore, that hinge movement between the
proximal and distal elements is probable,
while the distal elements are solidly attached
to one another with little possible movement
between them. Similarly, in primitive capto-
rhinomorphs and pelycosaurs, a functional
mesotarsal joint may have been present, but
the corresponding articular surfaces are not
as well adapted to movement as in Petrolaco-
saurits.

The metatarsals in general resemble the
metacarpals, except that the fourth and fifth
metatarsals ha\e large proximal ends. The

heads of both of these metatarsals have
strongly ridged medial portions and slender
lateral portions which also extend proximally.
The proximal articular surfaces are not per-
pendicular to the long axis of the metatarsal.
The proximal ends engage each other in
characteristic overlap. This overlap is, how-
ever, more accentuated in the pes than in the
manus.

The phalangeal count, based upon the
type specimen is 2, 3, 4, 5, 4. The terminal
phalanges are modified into claws similar to
those in Paleothijris.

54

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO.

PHYLOGENETIC RELATIONSHIPS
OF PETROL ACQS AURUS

In the first detailed study of Petrolaco-
saurus Peabody (1952) tried to establish the
probable relationships of this animal by eoni-
paring it with reptiliomoiph "amphibians"
(diadectids and seymouriamorphs), capto-
rhinomorph reptiles, pelycosaurian reptiles
and diapsid reptiles. He separated those
morphological features of Petrolacosaiirus
that are also found in many amphibians
(.shared primitive characters, or symplcsio-
morphics of Hennig) from those he believed
were "progressive features" present in capto-
rhinomorphs or pelycosaurs or cosuchians
(shared derived characters, or synapomor-
phies of Hennig). His conclusion that Petro-
lacosaiirus is an eosuchian diapsid reptile
rather than a pelycosaur and that it is close
to the primary dichotomy of the reptilian
radiation was generally rejected (Stovall,
Price and Romer, 1966). The reasons for this
rejection are two fold. The fossil material
availalilc to Peabody for study was meager
and crucial portions of the skull and post-
crauial skeleton were unknown. In his analy-
sis of relationships Peabody included among
his "progressive features" many morphologi-
cal characters tliat later proved to be charac-
teristics of early reptiliomorph amphibians or
all early reptiles, rather than features restrict-
ed to particular reptilian taxa. For example
the suture pattern of the palate in Pcfrolaco-
saurus is close to that of Youngina, as indi-
cated by Peabody (1952) but is also similar
to the pattern seen in primitive captorhino-
morphs (Carroll and Baird, 1972), pelyco-
saurs, millerosanrs (Gow, 1972) and even
such reptiliomorph amphibians as Gephyro-
stegus (Carroll, 1970) and Limnoscelis
(Heaton, in press). The discovery and study
of nearly complete, well preserved specimens
of Petrolacosaurus and recent studies of other
Paleozoic reptiles (Carroll, 1964, 1969; Car-
roll and liaird, 1972; Clark and Carroll, 1973;
Gow, 1972, 1975; Heaton, 1979; Holmes, 1977;

Reisz, 1972; N'aughn, 1955) pro\ide a strong
basis for the critical evaluation of the phylo-
genetic relationships of this unique early rep-
tile.

For the purposes of this discussion strict
monophyly for the .\mmiota is assumed. It
is further assumed that the following grovips
of late Paleozoic and early Mesozoic fossils
can be included within the .\mmiota: pareia-
saurs, procolophonians, millerosanrs, meso-
saurs, pelycosaurs, captorhinomoq^hs and
"cosuchians." In order to test rigorously hy-
potheses of relationships of the Petrolaco-
sauridae some well-tested equivalent level
phylogenetic system is needed within the
Ammiota, but this is not available. No re-
stricted outgroups can be developed either;
only an unresolved multichotom\- containing
at least four of the above seven groups. It
was therefore felt that the following pro-
cedure should bi' pursued: Those characters
shared by most of the above early reptiles will
be considered shared primiti\-e characters
( symplesiomoqDhs ) for .\mmiota and re-
moved from discussion when testing hypothe-
s(\s of relationships of Petrolacosaurus. These
characters are listed below and separated
from later discussions because retention in
Petrolucosatirus of many of these primitive
characters does not in itself indicate close
lilnlogcnetic relationships within the Am-
miota (Hennig, 1966). Hypotheses of rela-
tionships of Petrolacosaurus will be then
presented, the basic taxa discussed and the
hypotheses tested using shared derived char-
acters.

Phimitive Cn.\RACTEns

(1) While pelycosaurs show exponential
increase in size during the Penns>l\anian, pro-
torothyridids, "cosuchians," millerosanrs, mes-
osaurs and procholophonids retain almost the
same body size throughout their history
(Reisz, 1972). Among the protorothyridids

1980

PENNSYLVANIAN DIAPSID REPTILE

55

the skull also increases in length only slightly
(Table I). The skull of the largest known
specimen of Petrolacosaurus is only slightly
longer than the skull of the largest known
protorothyridid, Protorothyris archcri.

The outline of the skull in dorsal view
resembles an isosceles triangle with a rounded
apex and with sides that bulge out slightly,
close to the base. In captorhinids, the skull
becomes bulkier and wider than in protoro-
thyridids. Among pelycosaurs a great variety
of skull outlines exists from the slender, elon-
gate skulls seen in ophiacodonts to the bulky,
wide configuration of edaphosaurs and case-
ids.

The skull profile remains relatively low.
The maximum skull width is greater than the
maximum lieight in any transverse section of
the skull roof. The ventral edge of the cheek
is nearly straight from the snout to the sus-
pensorium. The posterior edge of the skull
is tilted only slightly forward. In advanced
pelycosaurs, on the other hand, the skull
profile deepens in the temporal region and
the ventral edge of the cheek is often strongly
cur\'ed ventrally.

The outline of the skull in protorothyri-
dids, millerosaurs, "'eosuchians" and Petrola-
cosaurus, in occipital view, resembles a trape-
zoid with rounded upper angles. The nature
of preservation of most primitive reptiles
precludes precise determination of the angle
between the cheek and the skull table. It is
only slightly greater than 100 degrees in the
captorhinomorph Protorothyris, and similar
estimates have been computed tor Palcothyris
on the basis of the width of the palate and
skull table (Carroll and Baird, 1972). The
wealth of material available for the recon-
struction of Petrolacosaurus .shows an angle
of 100 ± 5 degrees. A more precise determi-
nation of this angle was not possible because
of the curvature of both tlie cheek and skull
table. In pelycosaurs other than caseids and
eothyridids the occiput is much higher than
in either protorothyridids, eosuchians, mil-
lerosaurs, or Petrolacosaurus.

(2) The interrelationship of the bones of

the cheeks, skull table, palate, occiput and
mandible remains relatively constant in mil-
lerosaurs, protorothyridids and primitive pely-
cosaurs (Gow, 1972; Romer and Price, 1940;
Reisz, in press). Most features of this pattern
are retained in the skull of Petrolacosaurus in
spite of the morphological advances associ-
ated with crania] fenestration.

The dorsal expansion of the maxilla re-
mains moderate in most early reptiles. The
long lacrimal extends from the external naris
to the orbit, and fonns a large portion of the
anterior orbital margin. Although Petrolaco-
saurus, most captorhinomorphs, millerosaurs
and primitive pelycosaurs retain this condi-
tion, in more advanced pelycosaurs, "eosu-
chians" and mesosaurs the lacrimal fails to
reach the external naris because of the dorsal
expansion of the maxilla. In most early rep-
tiles the jugal forms part of the ventral mar-
gin of the cheek. The squamosal and the
quadratojugal form the slightly convex pos-
terior margin of the cheek. The postorbital
extends to the supratemporal. The supratem-
poral bone is large in parieasaurs, procolo-
phonians, millerosaurs and primitive pelyco-
saurs. In ventral view, the occipital condyle
extends posteriorly to the level of the jaw
suspension. In protorothyridids, millerosaurs,
pelycosaurs, "eosuchians" and Petrolacosaurus,
the long pterygoids are wedged anteriorly be-
tween the vomers. Three rows of well-devel-
oped palatal teeth radiate from the basiptery-
goid area in most early reptiles; the largest
palatal teeth are present on the transverse
flange of the pterygoid. The long, relatively
narrow palatine abutts anteriorly against the
plate-like vomer that forms most of the medial
margin of the internal naris. The ectoptery-
goid is small. The parasphenoid is denticu-
late.

The paired postparietal and tabular bones
are restricted to the occiput. The occipital
condyle is located far \entrally on the occiput,
close to the level of the jaw articulation. A
large subvertical strip of the squamosal ex-
tends onto the occiput. The posterior edge
of the dorsal process of the quadrate is cov-

56

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Fic. 23. — Petrolacosaurus kansctuis Lane. Pattern of internal cranial ridges or thickenings in (A) dorsal,
(B) ventral and (C) lateral \ie\vs, x 2. Composite. See tevt for further explanation.

1980

PENNSYLVANIAN DIAPSID REPTILE

57

ered by it, except in the region of the quad-
rate foramen. The dorsal process of the quad-
rate is inchned shghtly anterodorsally. In
mesosaurs, early "eosuchians," captorhino-
niorphs, pelycosaurs, procolophonids there is
no otic notch. The stapes is massive and has
both a dorsal process and a footplate.

The mandible is long and slender. The
coronoid region is little expanded. There is
no retroarticular process.

(3) In most early reptiles the marginal
teeth are arranged in a single row. They are
simple conical structures, slightly recurved,
with subthecodont implantation. An anterior
food trap is partially associated with some
caniniform tooth development.

(4) Vertebrae are amphicoelous and noto-
chordal. Relatively large intercentra are con-
tinued in the tail as chevron bones. The
atlantal intercentrum is very large, has a well-
developed ventral median ridge, and a pair
of articular facets for the capitulum of the
atlantal rib. In most pelycosaurs, in millero-
saurs, "eosuchians" and captorhinomorphs
the atlantal centrum is excluded from the
ventral border of the column by the axis
intercentnmi. In Hylonomiis, the earliest rep-
resentative of the captorhinomoi"phs, the
atlantal centrum is crescentic in outline and
open ventrally. The axial intercentrum is
longer than those in the rest of the column.
In Petrolacosaurus the large axial intercen-

Pu/c'olhyri'

trum is ventral to the crescentic atlantal cen-
trum and is loosely attached to it. Although
these elements are disarticulated in Hylono-
miis [BM(NH)R. 4168], their relative posi-
tion was apparently similar. In all later
captorhinomorphs, however, the axial inter-
centrum and atlantal centrum are indistin-
guishably fused. The axial neural arch of
Petrolacosaurus is strongly built, and has a
large spine that extends anteriorly over the
atlas, as in the captorhinomorphs, "eosuch-
ians," millerosaurs and pelycosaurs.

The cervical ribs have flattened, slightly
expanded ends. The sacral and anterior cau-
dal ribs are firmly attached to the vertebrae.
There are two sacral ribs.

The primitive reptilian configuration of
the girdles in early captorhinomorphs, miller-
osaurs and some pelycosaurs is also seen in
Petrolacosaurus.

The limbs are well ossified. Both proximal
and distal limb elements have thick shafts
and broad ends. The humerus has a well
developed entepicondylar foramen and a su-
pinator process. The manus and pes are rela-
tively large and broad, with a long-necked
astragalus and a plate-like calcaneum.

Hypotheses of Relationships

Two of the hypotheses presented and
tested here are illustrated in Fig. 26. The first
hypothesis is that Petrolacosaurus shares a

PeltvL

'dcmdiniis

) 'oim^ina

Fig. 26. — Cladogram illustrating hypotlieses of relationships of Petrolacosaurus. See text for further expla-
nation.

58

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

more recent ancestor with the captorhino-
morph Paleothyris than witli the synapsid
Eothtjris. All synapsids could have been used
but Eothijris was selected as the representa-
tive of this group because it is the most
primitive pelycosaur and is most similar to
Paleothyris and protorothyridids. This hy-
pothesis is tested by characters 1-6 (see dis-
cussion below ) . The only morphological char-
acter that may falsify this hypothesis is the
presence of the lower temporal fenestrac in
both Eothijris and Petrolacosaurus. Other
significant differences in the architecture of
the skulls of these reptiles and the presence
of six shared derived characters in the un-
fenestrated Paleothyris and Petrolacosaurus
suggest that the lower temporal fenestrac in
the above taxa are not phylogenetically ho-
mologous.

The second hypothesis is that the primi-
tive diapsid Youngina shares a more recent
ancestor with Petrolacosaurus than with the
protorothyridid Paleothyris. All protorothyri-
dids could have been used as the outgroup
for the tested characters but Paleothyris was
selected because it is the most primitive, well
preserved protorothyridid, is mo.st similar to
Petrolacosaurus and is therefore best suited
for the search of characters unique to diap-
sids. Youngina was chosen for comparison
with Petrolacosaurus because it is the best
known early "cosuchian" diapsid reptile
(Cow, 1972; Carroll, 1977) and is closest
morphologically to the latter species. In addi-
tion, there is little evidence to indicate that
the "Eosuchia" are strictly monophyletic.
Although onl\' Youngina is shown in the clad-
ogram (Fig. 26), characters 7-14 (see di.scus-
sion below) also test the hypothesis of a
monophyletic Diajisida including Petrolaco-
saurus. No characters are known that falsify
this hypothesis.

Basic Taxa

Eothyris is a small, remarkably priiniti\e,
pelycosaur known from a single skull from
the Lower Permian of Texas. The familv

Eothyrididae. erected as a pro\isional group
by Romer and Price (1940), also included
the fragmentary remains of three pelycosaurs
of large size because they are "Ophiacodont
pelycosaurs, primitive in most known regards
but paralleling the higher sphenacodonts in
the development of much enlarged canines
and showing a tendency toward elongation of
the \ertebral column" (Romer and Price,
1940, pp. 246-247). Both the presence of
caniniform teeth and long vertebrae represent
primitive conditions for amniotes, and the
dental and vertebral patterns seen in the
advanced ophiacodont Ophiacodon arc de-
rived conditions. Both Stereophallodon and
Stereorhachis are ophiacodont pelycosaurs
that have retained the primitive dental and
vertebral pattern and ha\e been therefore
included into the family Ophiacodontidae
(Reisz, in press). Balduinonus trux appears
to present a more difficult taxonomic problem
because the type specimen includes fragments
of large ophiacodont \ertebrae and the maxilla
of a large sphenacodont pelycosaur (pers.
obs.). The study of tiiis material is now in
progress, but it is certain that none of the
materials associated with this species is an
eothyridid pelycosaur.

Langston, in his 196.5 description of
another small eoth\ridid Oedaleops campi,
indicated that this species is remarkably simi-
lar to Eothyris. Oedaleops has smaller eanini-
fomi teeth than Eothyris, .similar in size to
those found in early captorhinomorphs, but
Oedaleops is peculiar in that the supratcm-
poral projects slightly beyond the posterior
edge of the cheek. The exact configuration of
the posterior end of the supratemporal in
Eothyris is not known because of overprepa-
ration. This posterior projection of the supra-
temporal may represent a shared derived
character for the family, but further materials
are needed before this can be verified.

Eothyridids, represented only by the spe-
cies Eothyris parkeyi and Oedaleops campi,
are closely related to the caseid pelycosaurs
as indicated by a series of shared derived
characters (Reisz, in press). Significant

1980

PENNSYLVANIAN DIAPSID REPTILE

59

shared piimiti\e cliaractcis found in both
caseids and fiotliyridids that appear in a
derived form in all otiier pelycosaurs are
worth noting: The width of the skull is great-
er than the height, e\en in the region of the
snout: the frontal either does not extend to
the orbit (Eotlu/ris) or the orbital margin of
the frontal is very short (Oedaleops and all
caseids ) ; the supratemporal bone is large and
broad. In eaptorhinomorphs, millcrosaurs and
proeolophonids the skull retains the priniiti\e
cotylosaur pattern of a low-profile skull. This
primitive condition persists only in caseid and
eothyridid pelycosaurs. In other pelycosaurs,
the snout in the edaphosaurs, and the whole
skull in ophiacodonts, waranopsids and sphen-
acodonts becomes narrow and deep. Only in
cotylosaurs ( Limnoscelis, Diadectcs, Sey-
mouria) and the caseid and eothyridid pely-
cosaurs is the primitive pattern of the frontal
retained. In other pelycosaurs, in millcrosaurs,
and in early eaptorhinomorphs one-third of
the dorsal orbital margin is fomicd by the
frontal. The broad supratemporal, a primi-
tive reptilian and cotylosaurian character is
modified in eaptorhinomorps, diapsids and
advanced pelycosaurs, but persists as a broad
sheet in caseids and eothyridids.

Paleothyris, the best known protorothyri-
did captorhinomorph, was described by Car-
roll (1969). This small reptile is represented
by two nearly complete skeletons and numer-
ous disarticulated specimens, all recovered
from a single Middle Pennsylvanian Sigiillaria
stump from Florence, Nova Scotia. Paleothy-
ris exhibits a combination of morphological
characters that make it the least specialized
well preserved member of the family Protoro-
thyrididae. The group of small reptiles placed
within this family is characterized by the
possession of features usually considered
primiti\-e (plesiomorphic) for the Amniota.
It is in fact difficult to find any derived char-
acters that are unique to this group. The
reasons for this are that ( 1 ) no attempt has
been made to search for mor^^hological char-
acters that are unique to this group, (2) the
placement of the species within the family

was based on solely primitive ("ancestral")
characters, and (3) the osteology of most spe-
cies is based on single partially preserved
sp(>cimens, or several poorly preserved scat-
tered skeletons. Of the better known protoro-
thyridids, Cephalerpeton ventriarmatum
(Carroll and Baird, 1972) is specialized in
having only 16 large maxillary teeth while
Anthracodwmeus longipes is specialized in
having long limbs. The most significant spe-
cialization of Coelostegus prothales is seen in
the skull roof, where the posterior margin is
deeply embayed. The Lower Pennian Pro-
towthijris archeri (Clark and Carroll, 1973)
has an unusually large skull and tall neural
spines. These species probably represent sep-
arate monophyletic taxa rather than a single
monophyletic group, but better preserved
materials are needed before such an hypoth-
esis can be tested.

A series of diapsids from the later Permian
and early Triassic are placed on the basis of
primiti\'e characters, within the order "Eo-
suchia.' They arc small, and most are poorly
known. The assigned genera appear to con-
stitute a number of distinct lineages. The
main stock is represented by Youngina from
the Upper Permian Daptocephcdus zone of
South Africa. This species is known from sev-
eral skulls described by Broom (1914), Olson
(1936), Gow (1975) and Carroll (1977) and
from postcranial skeletons described by
Broom (1924), Watson (1957) and Gow
(1975). The skull of Youngina is in many
significant features similar to that of Petro-
lacosaunis, including large upper temporal,
lower temporal, post-temporal and suborbital
fenestrae. There is no otic notch, but the
posterior margin of the (juadrate is not cov-
ered by the squamosal. The \ertebrae are
((uite primitive and close to the pattern seen
in eaptorhinomorphs, but they also ha\e
slightly swollen neural arches and j-jowerful
zygapophyses, with nearly horizontal articu-
lating surfaces, as seen in Pctrolacosaunis.

Heleosaurm (Broom, 1907; Carroll, 1976a)
from the Cisteccphahts zone, is known from
a single partially preserved specimen appar-

60

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

ently combining primitive "eosuchian" fea-
tures with a number of highly speciahzed, and
possibly archosaurian, adaptations. In this
specimen only the palate, marginal bones of
the cheek region, lower jaws, an articulated
series of presacral vertebrae and ribs, dermal
armour-plates along the vertebral column, in-
complete girdles and femora are preserved.
The known parts of the skull of Heleosaurus
are similar to the corresponding parts in
Petrolacosaiirus and Younghui despite the the-
codont marginal dentition. Heleosaurus also
resembles Petrolacocaurus in a number of
postcranial features including elongated cer-
vical vertebrae with long prezygapophyses,
tilted centrosphenes, mammillary processes on
the neural spines and well-developed ventral
keels. According to Carroll (1976a) the fe-
mur in Heleosaurus is far advanced from the
primitive reptilian configuration, in correla-
tion with an upright posture. Careful exami-
nation of the specimen reveals, however, that
the exposed posterior surface of the femur
was mistakenly identified by Carroll as the
ventral surface. The exposed surfaces of both
the proximal and distal heads and the curva-
ture of the femur correspond closely to those
of primitive captorhinomoqihs and is not spe-
cialized in the manner suggested by Carroll.

Another genus from the Cistecephalus
zone, Galesphyrus, was originally described
by Broom (1915) as a therapsid, but its
"eosuchian" nature has since been recognized
by Romer (1966) and Carroll (1976b). Al-
though this small reptile is known only from
two incomplete postcranial skeletons it has
been associated with the "Eosuchia" on the
basis of general similarities in size and pro-
portions to Youngina. and the particular simi-
larities of the vertebral, carpal and tarsal
structures. Many of these morphological
characters are also seen in Petrohicosaurus.

The postcranial skeleton of a new "eosuch-
ian" reptile, Kcntjasaurus, from the Triassic
beds near Mombasa, Kenya has been de-
.scribed by Harris and Carroll (1977). Tlie
type specimen of Kenijasaurus has a series of
articulated vertebrae that extend from the

posterior end of the pectoral girdle to the tip
of a long tail. Crushing makes interpretation
of the trunk vertebrae difficult, but it appears
that, as in Petrolacosaiirus, the neural arches
are massively built, the zygapophyses extend
far laterally and the articular surfaces for the
ribs in the anterior dorsal region are elongate
with expanded dorsal and ventral ends. Al-
though the preserved appendicular skeleton
of Kenijasaurus is in some features similar to
that of Petrolacosaurus, the forelimb is spe-
cialized in having a wide paddle-like humerus
and the fifth tarsal appears to have fused to
the fourth tarsal.

A number of primitive, but apparently dis-
tinct aquatic eosuchians from the Lower
Triassic of Tanzania and Madagascar are cur-
rently being described by Mr. Philip J. Currie
at the Provincial Museum of Alberta, Edmon-
ton. The early descriptions by Haughton
(1924) and Piveteau (1926) reveal little de-
tail, but show general postcranial proportions
similar to those noted in Petrolacosaurus.

Prolacerta (Camp, 1945; Cow, 1975) from
the Lower Triassic Lystrosaurus zone of South
Africa represents yet another "eosuchian"
lineage. Although the skull of Prolacerta has
an unusual combination of lizard-like and
thecodont-like features, it also shares a large
number of primiti\x> characters with Young,ina
and Petrolacosaurus. There is a well devel-
oped upper temporal fenestra, but the lower
temporal bar is absent. Although there is an
otic notch formed by the quadrate, the squa-
mosal extends far ventrally, preventing strep-
tostyly. Well de\elopcd post-temporal and
suborbital fenestrae are also present. The
dentition is apparently thecodont (Cow,
1975), although the anterior and posterior
edges of the teeth are not serrated. The post-
cranial skeleton of Prolacerta is surprisingly
similar to that of Petrolacosaurus. The atlas-
axis complex and the other cervical vertebrae
are nearly identical, the configuration of cervi-
cal, dorsal and sacral ribs and their pattern
of articulation are similar. In contrast to the
condition in Petrolacosaurus, the caudal re-
gions of Prolacerta and Youngina are short

1980

PENNSYLVANIAN DIAPSID REPTILE

61

and the laterally directed ribs appear to be
firmly fused to the centra. The pelvic girdle
is similar to that of Petrolacosaurus. The limbs
are lightly built, with the distal paired limb
elements equal in length or longer than the
proximal elements. The tarsus of Prolacerta
is derived compared to Petrolacosaurus in
lacking a discrete fifth tarsal bone and in
having a hooked fifth metatarsal.

This group appears to have the equiva-
lent position among diapsid reptiles as the
eothyridids have had among pelycosaurs
(Romer and Price, 1940) — a provisional "gar-
bage bag" assemblage of primitive forms. As
it now stands the "Eosuchia" has no taxonomic
validity because it cannot be defined in terms
of shared derived characters not found in any
other group. Until this group is thoroughly
reviewed, and proper diagnosis provided, no
new forms should be placed within it.

Shared Derived Characters Testing

THE Hypothesis of Relationship

Between Petrolocosaurus and

Paleothyris.

The following characters correspond to
characters 1-6 in Fig. 26.

1. — Loss of contact ])etween the postorhi-
tal and supratemporal. In Eothijris, other
pelycosaurs, millerosaurs and other early am-
niotes the postorbital has a long posterior
process that extends to the supratemporal.
None of the captorhinomorphs, Petrolaco-
saurus or diapsids show this primitive condi-
tion, and the loss of contact between these
bones is considered derived for captorhino-
morphs and diapsids.

2. — Reduction in the size of the supra-
temporal bone. In Eothijris and most pelyco-
saurs, in millerosaurs and even in cotylo-
saurian "amphibians' the supratemporal is a
large sheet of bone wedged between elements
of the skull table. In Paleothyris, captorhino-
morphs, and in Petrolacosaurus and other
primitive diapsids this bone is modified into
a small, slender element generally located in
a shallow trough of the parietal bone. This

condition is considered derived for capto-
rhinomorphs and diapsids. In carnivorous
pelycosaurs the same advanced pattern is
seen, but is probably arrived to independently
within the synapsids.

3. — Reduction in size or loss of the tabular
bone. In Eothijris. and all other pelycosaurs
the tabular bone is a large sheet located on
the occiput. In Paleothyris, and other protoro-
thyridid captorhinomorphs, in Petrolacosaur-
us, Younp,ina and some other diapsids the
tabular bone is greatly reduced in size, a con-
dition considered derived for these forms. In
captorhinid captorhinomorphs and in many
diapsids the tabular bone is lost.

4. — Proximal and di.stal limb elements are
elongate and lightly built. In primitive pely-
cosaurs (Reisz, in press) the limb elements
are relatively thick and massively built. Be-
cause the postcranial skeleton of Eothijris is
unknown, no direct comparison is possible.
In both Paleothyris and Petrolacosaurus the
proximal and distal limb elements the shaft
is exceedingly slender, usually less than 10 per
cent of the length; the extremities of the bones
are also reduced in size, with the areas of in-
sertion and origin of muscles important in the
power stroke and recovery concentrated near
the ends. Although this condition is consid-
ered advanced for these fonns, the heavy
captorhinid caiJtorhinomorphs and the aquatic
diapsids appear to have secondarily acquired
massive limbs.

5. — The cms and pes are narroiv and long
uith overlapping metatarsals and metacar-
pals. In most pclycosaius ( Romer and Price,
1940) the pes and crus are broad and the
metatarsals and metacarpals do not overlap.
In both Paleothyris and Petrolacosaurus as
well as most protorothyridid captorhino-
morphs and primitive non-aquatic "eosuchian"
diapsids, the cms and pes are narrow and
thus the proximal heads of the metacarpals
and metatarsals overlap extensively. Although
this condition is considered derived for the
above forms, captorhinid captorhinomorphs
appear to have either retained or secondarily
de\cloped the primitive condition.

62

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

6. — Presence of a single centrale in the
pes. In most pelycosaurs (Romcr and Price,
1940) two ccMitralia pedis are present. In
Paleothtjris, as reconstructed by Carroll
(1969), there is a tiny lateral centrale, but
the medial centrale has the outline of a pea-
nut. This suggests that the tiny lateral ele-
ment may have been an artifact and the much
larger imusually shaped centrale represents
the fusion of the two centrale seen in pelyco-
saurs. In all other captorhinomorphs, in Petro-
lacosaiiriis and in diapsids a single large
centrale is present, considered to represent
the derived condition. In a few advanced
carnivorous pelycosaurs the same derived
condition has been observed, but this prob-
ably represents convergence.

Shared Derived Characters Testing

THE Hypothesis of Relationship
Between Petrolacosaurus and Youngina

The following characters correspond to
characters 7-14 in Fig. 26.

7. — Presence of superior temporal fenestra.
The most significant departures in Petrolaco-
saurus, Yotmgina and other "eosuchian" diap-
sids from the captorhinomoqDh skull pattern
involve the development of fenestrae in the
skull roof. The upper and lower temporal
fenestrae, typical of diapsid reptiles, have the
same position and orientation in Petrolaco-
saurus as in Youngina. The lower temporal
fenestra is, however, similar to that in Eothij-
ris. The derived characters listed in the pre-
vious section shared by the unfencstrated
PaJeothyris and the diapsid Petrolacosaurus
with Eothyris as the outgroup suggests that
the lower temporal fenestrae in the latter two
species developed independently.

8. — The .shape and interrelationship of the
circtnnfcnestral hones are modified as a rc.sidt
of diapsidy. In Petrolacosaurus the postorbital
is triradiate in outline as in Youngina, Prola-
certa (Camp, 1945; Cow, 1975), Euparkeria
(Ewer, 1965) and numerous other diapsids.
The dorsal ramus of the postorbital is in con-
tact with a small lateral projection of the

parietal. The ventral ramus extends down
onto the anterodorsal margin of the jugal and
the posterior ramus of the postorbital comes
in contact with the squamosal forming the
anterior half of the intertemporal bar. In
Prolacerta, and even in the "eosuchian" de-
rivatives, Ktdineosaurus (Robinson, 1962) and
Euparkeria, where the skulls are known in
great detail, .specific relationships of the post-
orbital to surrounding bones are the same
as in Petrolacosaurus, despite differences in
other cranial features. In all four genera, the
dorsal ramus of the postorbital sends a small
wedge between the parietal and the post-
frontal. The anteroventral ramus replaces
much of the postfrontal and jugal in forming
the posterior margin of the orbit, and extends
far down onto the inside surface of the latter
element, close to the jugal-ectopterygoid su-
ture.

The lateral portion of each parietal is also
modified as a result of fenestration. In capto-
rhinomorphs the contact between the parietal
and the postorbital and squamosal is exten-
sive. In Petrolacosaurus, as in Youngina and
all other diapsids, these contacts are greatly
reduced and the lateral margin of the parietal
is deeply emarginated to form much of the
edge of the superior temporal fenestra. In
addition, the parietal in Petrolacosaurus has
a narrow lateral process that forms the antero-
ventral corner of the upper temporal fenestra,
a condition also seen in Prolacerta, Kuhneo-
satirus and Euparkeria.

In Prolacerta, Euparkeria and Kuhneo-
saurus a well developed flange of the parietal
is directed ventrolaterally, roughly parallel to
the direction of the adductor muscle action.
The condition in Youngina, where the parietal
has a very small ventrolateral process, prob-
ably represents an intiMiiK^diatc stage between
Petrolacosaurus and the Triassic diapsid rep-
tiles. A well developed ventrolateral process
of the parietal is al-so seen in modern lizards
and Sphcnodon. It is to the dorsal surface of
tliis llangt' that much of the adductor muscu-
lature is attached. This adaptation represents
a secondary stage in the evolution of the

1980

PENNSYLVANIAN DIAPSID REPTILE

63

upper temporal fenestra and probably devel-
oped in response to the need for more secure
muscle attachment. The presence of the
temporal opening provided the opportunity
of attaching muscle fibers to the outside sur-
face of the parietal. A similar development is
seen in adxanced mammal-like reptiles, where
the adductor musculature also invades the
dorsal surface by passing through the lateral
temporal opening. This occurs long after
initial fenestration and extensive adaptive
radiation of early mammal-like reptiles — the
pelycosaurs.

The squamosal of Petrolacosaurus, al-
though retaining its plate-like configuration,
is modified anteriorly and dorsally to form
the margins of the temporal openings and the
intertemporal bar. As in Youngina and all
other diapsids, the contact between the squa-
mosal and the postorbital is much reduced
over the condition in captorhinomorphs. The
posterior part of the jugal is modified from a
plate-like sheet to two relatively narrow bands
that extend dorsally and posteriorly. The
large suture between the jugal and squamosal
of captorhinomorphs is replaced in Petrolaco-
saurus by the lower temporal opening.

9. — Presence of a well developed suborbi-
tal fenestra. Elongate, well developed sub-
orbital fenestrae are seen in Petrolacosaurus,
Youngina and all other diapsids. No suborbi-
tal fenestrae are present in Paleothtjris, capto-
rhinomorphs or pelycosaurs.

10. — The maxilla, palatine, ectoptenjgoid
and jugal bones and their interrelationships
are modified as a residt of the presence of tJie
suborbital fenestra. As in Youngina, Heleo-
saurus and other diapsids the suture between
the cheek (maxilla) and the palate (palatine
and ectopterygoid ) in Petrolacosaurus is in-
terrupted by the suborbital fenestra. The
suture between the palatine and maxilla is
reduced in length and part of the alveolar
shelf of the maxilla is smooth. The ectoptery-
goid in Pctrolaco.murus is smaller than in
Paleothijris and other protorothyridids (Car-
roll and Baird, 1972) and has a robust lateral
projection. The contact between the ecto-

pterygoid and the cheek is greatly reduced,
as the jugal develops a small oval medial
process for attachment to the lateral process
of the ectopterygoid. There is no contact
between the ectopterygoid and the maxilla.
The same arrangement is found in Youngina,
Heleosaurus, Euparkeria and even Kuhneo-
saurus. In Paleothijris and other protorothy-
ridids the palato-maxillary suture extends to
the posterior end of the palatine. In addition,
it is probable that the ectopterygoid also
attached suturally to the maxilla, since in
these forms the ectopterygoid was a posterior
extension of the plate-like palatine (Carroll,
1969). In captorhinids the ectopterygoid is
lost, and the maxillary contact with the ecto-
pterygoid is replaced by a strong contact
between the jugal and the pterygoid (Heaton,
1979). In advanced pelycosaurs such as Di-
metrodon, the jugal also comes to brace the
pterygoid, while the ectopterygoid-maxilla
suture is reduced.

11. — A pair of well developed post-tem-
poral fenestrae are bounded in Petrolaco-
saurus, by a narrow occipital flange of the
squamosal, the small tabular, the sitpraoccipi-
tal and the well developed paroccipital proc-
ess of the opisthotic. In Youngina these fenes-
trae are somewhat larger than in Petrolaco-
saurus, but are of fundamentally similar con-
figuration. In Paleothijris and primitive cap-
torhinomorphs the occiput is poorly known,
but the post-temporal fenestra appears to be
much smaller than in Petrolacosaurus (Car-
roll, 1969). In pelycosaurs the post-temporal
fenestra is greatly reduced as a result of the
plate-like configuration of the tabular and of
the supraoccipital-opisthotic complex (Romer
and Price, 1940, Fig. S).

12. — The relative size of the skull, as indi-
cated by the ratio betiveen the length of tlie
skidl and trunk, is considerably smaller in
Petrolacosaurus than in Paleothyris and cap-
torhinomorphs, and is similar to that in
Youngina and Heleosaurus. The postorbital
region of the skull is much reduced in Petro-
lacosaurus and Youngina. The orbits in
Petrolacosaurus, Youngina and other early

64

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

diapsids are, on the other hand, relatively
larger than in captorhinomorphs.

13. — The limhs are long. The humerus
and femur in Petrolacosaurus are equal in
length to about seven mid-dorsal vertebrae.
In Youngina and Kenijasaunis the femora re-
tain the same relative length, but the humeri
are relatively shorter, equal in length to about
five trimk vertebrae. In Paleothyris and most
primitive captorhinomorphs the femora and
humeri are equal in length to from 4.5 to 5
dorsal vertebrae. In "eosuchians" the radius
and ulna, tibia and fibula are nearly equal to
(Youngina Galesphijrus, Kenijasaunis) or
even longer than the himierus and femur re-
spectively (Prolacerta). The distal end of
the tibia of the above genera is relatively
much larger than that of the captorhinomoriDh
Paleothyris, suggesting a firmer joint with the
astragalus.

14. — Locked tibio-astragaJar joint. The
pes of Petrolacosaurus is generally similar to
that of Galesphijrus, Prolacerta, Kenyasaurus,
and Tangasaurus (the pes of Youngina is
poorly known) except for the relatively large
size of the proximal elements (Carroll, 1977).
The tibial facet of articulation on the astra-
galus of Petrolacosaurus, formed by two lon-
gitudinal ridges separated by a deep trough
fits snugly onto the distal articular surface of
the tibia to fonn a firm, locked joint. This
configuration can be detected only if the
tarsus is exposed in ventral or medial views.
Thus, only Petrolacosaurus and Kenyasaurus
are now known to have this character, but it
is probable that other "eosuchians" have the
same pattern. No pclycosaur, millerosaur or
captorhinomorph has this advanced condition.

Petrolacosmims Co.mpared With
Aracoscelis

The reptile Araeoscelis from the Lower
Permian of Texas has been often allied by
Romer (1956, 1966) and VViUi.ston (1925)
with a group of poorly known late Permian
and early Triassic genera, referred to as pro-
torosaurs. Roth Vaughn (1955) and Carroll

( 1969 ) felt, however, that Araeoscelis and a
closely related form, Kadaliosaurus from the
Lower Permian of Europe, should be placed
within the captorhinomorphs on the basis of
their similarities to the primitive captorhino-
morphs.

Vaughn ( 1955 ) has also recognized the
possibility that Araeoscelis and Petrolaco-
saurus are closely related and suggested that
they may belong within the same family. In
fact, these genera resemble each other so
closely that there is no doubt about their
close relationship in spite of some apparently
significant differences in their skulls. These
two reptiles differ only in a few osteological
features.

The dentition in Araeoscelis is robust and
shows some cusp development in contrast to
the delicate, piercing type of dentition seen
in Petrolacosaurus. In association with this
adaptation to a different type of diet, the
massive prefrontal extends down to the max-
illa and excludes the thin lacrimal from the
orbit. The postorbital has a long posterior
projection that reaches the supratemporal. A
recently uncovered specimen of Araeoscelis
confirms Vaughn's interpretation of only up-
per temporal fenestrae in the postorbital por-
tion of the skull roof. The postorbital region
of the skull is very deep with the articulation
between the jaws far below the level of the
maxilla. The occipital plate as restored by
Vaughn (1955), is somewhat more extensive
than in Petrolacosaurus and the post-temporal
fenestra arc therefore small. This inteq^reta-
tion is based on a single, poorly preserved
occiput in an immature specimen. The lower
jaw is also relatively massi\e. All these dif-
ferences in cranial morphology may be adap-
tations of Araeoscelis to a different diet from
the insectivorous Petrolacosaurus.

The atlantal centrum in Araeoscelis is in-
distinguishahly fused to the axial intercen-
trum, a condition which is somewhat ad-
vanced over that in Petrolacosaurus. There
are nine elongate cervical vertebrae in Araeo-
scelis, according to Vaughn's reconstruction
versus six in Petrolacosaurus. The neural

1980

PENNSYLVANIAN DIAPSID REPTILE

65

arches of the two genera differ in that those
of Araeoscelis have lower neural spines and
the diapophyses and parapophyses are widely
separated in the dorsal region of the column,
with no bony web to connect them. The
girdles are very similar; the only distinguish-
ing features are the wider scapular blade in
the pectoral girdle and the ischiadic notch of
the pelvis in Petrolacosaiirus. The limbs in
Araeoscelis are slightly more specialized than
in Petrolacosaiirus in the presence of a fully
enclosed ectepicondylar foramen on the hu-
merus versus a supinator process, in having
a more slender femur, a cnemial process on
the tibia and in occasional fusion of the fourth
and fifth tarsals. These few osteological dif-
ferences in the postcranial skeleton can be
readily accounted for by the fact that Petro-
lacosaurus is the larger (therefore requires
relatively more bony support) and geologi-
cally older animal (therefore more primitive
in structure).

Clearly Petrolacosaurus and Araeoscelis
are closely related, with the younger genus
specialized to a diet that necessitates a
stronger, more massive skull. There are three

possible ways of interpreting the morphologi-
cal characteristics of these two genera: (1)
they evolved separately from an unfenestrated
ancestor which already achieved some of the
postcaptorhinomorph skeletal features shared
by both genera; (2) they evolved from a rep-
tile with an upper temporal fenestra, by
strengthening the skull in the line leading to
Araeoscelis, and by lightening the skull and
also developing a lower temporal fenestra in
the line leading to Petrolacosaurus; (3) Arae-
oscelis evolved from a diapsid condition like
that of Petrolacosaurus by closure of the lower
temporal fenestra during adaptation to a dif-
ferent diet.

Similarities in the configuration of the
upper temporal fenestra and of the suborbital
fenestra suggests that one of the latter two
possibilities may be valid. A more precise
interpretation of the taxonomic position of
Araeoscelis and the evolutionary relations of
the two genera must await the results of a
study (now in progress) of two recently dis-
covered and well preserved skeletons of
Araeoscelis.

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

HEARING IN PETROLACOSAURUS AND
OTHER EARLY REPTILES

Advanced cranial osteological features of
late Permian and early Triassic "eosuchians"
include modifications of the posterior margin
of the cheek, the stapes and the posterior end
of the mandible. The posterior border of the
squamosal in Youngina does not completely
cover the dorsal process of the quadrate, so
that this bone is visible in lateral view. The
posterior border is slightly concave. The
shaft of the stapes is slimmer than in capto-
rhinomorphs and PetTolacosaurus. This ar-
rangement of a light stapes and a slightly con-
cave quadrate is somewhat similar to that
seen in modern lacertilians, where the strong-
ly notched quadrate supports a tympanic
membrane. It has therefore been assumed
that Youngina had a tympanum. The condi-
tion in Euparkeria and Prolacerta is even
closer to that seen in living lizards. The con-
cavity of the posterior edge of the quadrate
is much stronger than in Youngina, there is
a well developed retroarticular process and
the stapes is reduced to a slender rod.

In PetTolacosaurus, on the other hand, the
squamosal forms the slightly convex posterior
border of the skull and the dorsal process of
the quadrate is completely covered. The stapes
is massive (Figs. 5 and 13).

Consideration of the functional implica-
tions of structures observed in early reptiles
is impossible without comparisons witli living
forms. Altliough generalizations about the
middle ears in living reptiles are tenuous
because of tlie morphological differcTices seen,
even within families, certain overall charac-
teristics can be noted in forms which are
sensitive to airborne sound. Reptiles sensitive
to airborne sound have a well-developed tym-
panum held taut by extensive bony support.
The areal ratio between the tympanum and
stapedial footplate is large (at least 30 to 1)
and the mass of the stapes is low. The stapes
is tv^pically a slender osseous rod with a
slightly expanded footplate that is suspended

within the oval window (Werner and Wever,
1972).

In lizards the quadrate, freed from its
squamosal cover, forms the strongly notched
posterior border of the cheek. In addition,
the lower jaw extends backward, beyond the
articulation between the quadrate and articu-
lar. In most extant lizards the strongly
notched quadrate and the retroarticular proc-
ess of the lower jaw support a large t)'mpanic
membrane. The posterior border of the tym-
panum is in part supported by the M. de-
pressor mandibularis which originates from
the top of the occipital plate of the skull and
inserts on the posterior end of the retroarticu-
lar process. Because similar osteological fea-
tures are seen in such Triassic diapsids as
Prolacerta, Euparkeria, Kuhneosaurus, Paleo-
gama and Paliguana, it is possible that in
these forms the quadrate and the retroarticu-
lar process supported a relatively large tym-
panum. None of these structural features are,
however, seen in Petrolacosaurus, captorhino-
morphs or even pelycosaurs. There is neither
a notch at the posterior edge of the cheek,
nor a posteriorly directed retroarticular proc-
ess. Under these conditions, the M. depressor
mandibularis had to hug the posterior border
of the skull as it extended down to the lower
jaw. There was therefore, no place for the
tympanum. There is no evidence for any
bony support for a tympanum. This support
would be needed to maintain high membrane
elasticity in the hydraulic and curved-mem-
lirane level action of the middle ear mechan-
ism (Wever and Werner, 1970; Manley,
1972).

All these features indicate that there could
be no tympanum in Petrolacosaurus, capto-
rhinomorphs or pelycosaurs. Vaughn (1955),
Hotton (1959) and Parrington (1955) have
suggested that the t>'mpanum in primitive
reptiles were located more medially and were
connected to the outside by an external audi-

1980

PENNSYLVANIAN DIAPSID REPTILE

67

tory meatus. These reconstructions did not
take into account that the tympanum had to
be located at the distal end of the stapes
and that the suggested locations of the tym-
panum were occupied by extensive cranial
and neck musculature.

In captorhinomorphs, pelycosaurs and
Petrolacosaunis the stapes is very massive and
is firmly attached both to the quadrate and
the braincase. Its higher inertia makes initia-
tion of movement much more difficult than
in modern forms. The footi^late of the stapes
is also very large. This makes high areal ra-
tios between a t\'mpanic membrane and
stapedial footplate impossible. To achieve
comparable areal ratios to those seen in mod-
ern reptiles, the tympanum would have to be
too large to fit anywhere behind the skull.
It is even more significant that the large
footplate was fitted within a bony socket of
the braincase, the fenestra ovalis externa
(Hcaton, 1979), and that the distal end of
the stapes was probably continued in a short
cartilaginous process that fitted in the stape-
dial recess of the quadrate. All of this indi-
cates that the stapes could not fomi an ef-
fective link within a mechanical transducer
system sensitive to airborne sounds. Although
Hotton attempted to show that the massive
stapes of Dimctrodon can be activated by a
relatively small membrane, his poorly cali-
brated experiments (1959, Fig. IB) provide
no useful evidence in this regard. No attempt
was made during the experiments to establish
the intensity of airborne sound. In addition,
the model was not intended to approach the
structural pattern of the middle ear of any
reptile.

The middle ear in captorhinomorphs, pely-
cosaurs and Petrolacosaunis could not func-
tion as in most modern reptiles. In modern
reptiles a highly effecti\e mechanical trans-
ducer amplifies small sound pressures im-
pinging upon a large tympanic membrane
into strong pressures exerted at the fenestra
ovalis. Without an eflFectivc tiansducer mech-
anism the great difference in impedances of
air and inner ear could not be compensated,

and most of the aerial sound impinging upon
the inner ear through \arious pathways would
not be transmitted, but absorbed and re-
flected (Wever and Lawrence, 1954). We
would, therefore, expect the ear of these
primitive reptiles to be insensitive to airborne
soiuid.

In spite of the supposed relative in-
sensitivity of the primitive reptilian type of
ear to airborne sound, there is little doubt
that with reduction of stapes mass and the
development of bony support for a tympa-
num, the ability of the ear to perceive low
frequency airborne sounds would greatly in-
crease. The presence in Euparkeria and Pro-
lacerta of a light stapes and a large area for
a bone supported tympanum, suggests that
these genera have already achieved higher
sensitivity to airborne sounds than captorhino-
morphs, or Petrolacosaunis.

It is relatively easy to derive the structural
condition of the ear region of Prolacerta and
Euparkeria from that in captorhinomoqohs or
Petrolacosaunis. The lightening of the stapes,
the loss of the squamosal cover of the pos-
terior end of the quadrate and the associated
notching in the quadrate are related with
the middle ear's increasing efficiency as a
mechanical transducer. The development of
a retroarticular process may, however, have
the dual advantage of (a) increasing the
mechanical advantage of the M. depressor
mandibularis, and (b) increasing the effi-
ciency of the middle ear by both increasing
the available surface area between the quad-
rate and the M. depressor mandibularis, and
by providing better support for the ventral
part of the tympanum. What conditions
would lead to selection of a middle ear sensi-
tive to airborne sound? Vocal communica-
tion has been cited as a possible factor ( Man-
ley, 1973). The fossil evidence indicates that
middle ears with efficient transducer mechan-
isms developed from the captorhinomorph
condition at least three times (therapsids,
"eosuchians" and chelonians).

Studies now in progress ( Heaton, pers.
comm.) indicate that the stapes in primitive

68

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

reptiles braced the braincase against the quad-
rate in a metakinetic skull. In the lineage
giving rise to lizards anterior rotation of the
braincase relative to the skull roof (metakine-
sis) was limited by the development of ac-
cessory structures on the supraoccipital and
parietal and, to a lesser extent, by the full
ossification of the paroccipital process, there-
by freeing the stapes from its supportive
fimction. Once freed, the stapes become a
hearing ossicle. The condition seen in Petro-
lacosaurus may represent the first stage in the
sequence, where the stapes is still massive,
but in contrast to the condition seen in primi-
tive captorhinomorphs, the paroccipital proc-
ess is fully ossified and extends far laterally.
Youngina would represent an intermediate
stage, where several modifications of the ear

region are taking place. The type specimen
of Youngina capensis shows that the squa-
mosal no longer covers the posterior margin
of the quadrate, and the posterior margin of
the quadrate is slightly concave, but not
notched. The type of Youngopsis rubidgei
has a stapes with a relatively slender shaft
and the paroccipital process extends dorso-
laterally to a notch between the squamosal
and the quadrate. Both Youngina and Heleo-
saurus have very short retroarticular proc-
esses. It is unlikely, however, that a tympa-
num was present in these "eosuchians." The
condition seen in Prolacerta, Palaeagama and
Euparkeria represent the final stages in this
sequence, where the osteological modifica-
tions necessary for having a middle ear sensi-
tive to airborne sound have been completed.

1980

PENNSYLVANIAN DIAPSID REPTILE

CONCLUSIONS

The detailed study of the osteology of the
Pennsylvanian reptile Petrolacosaurus kan-
sensis adds to our understanding of the anat-
omy of early reptiles. A good knowledge of
the osteology of this form also permits de-
tailed comparisons with other Paleozoic rep-
tiles.

Several authors have suggested that Petro-
lacosaurus should be allied with pelycosaurs.
This association is invalid for the following
reasons: As already indicated, the morpho-
logical features in which this reptile is de-
rived (the synapomorphies of Hennig) over
the captorhinomorph pattern are not at all
similar to the derived features seen in pely-
cosaurs. The similarities between Petrolaco-
saurus and early pelycosaurs are mostly
primitive characters, also shared with the
captorhinomorphs. In addition it has been
argued (Reisz, 1972) that the lower temporal
opening seen in the pelycosaurs developed
in response to selection for more efficient use
of the jaw musculature in forms of increasing
body size. The skulls of the Middle and Up-
per Pennsylvanian pelycosaurs arc larger and
considerably more massive than the skulls
of cither Petrolacosaurus or early captorhino-
morphs. Early pelycosaurs were, therefore,
able to include into their food supply animals
of greater size, and were probably able to
feed upon captorhinomorphs and other small
contemporary tctrapods. In the postcranial
skeleton, pelycosaurs show advances over the
primitive captorhinomorph pattern that can
be readily attributed to accommodation for
the greater body size. Petrolacosaurus, on the
other hand, evolved in a completely different
direction. Its diapsid skull configuration prob-
ably developed in response to the selective
advantage of maintaining a relatively small,
light-weight skull at the end of a long neck.
The postcranial features indicate that Petro-
lacosaurus was more agile than the capto-
rhinomorphs or pelycosaurs. The combination
of a relatively small, lightly-built skull and

greater agility suggest that this reptile adapt-
ed to a somewhat different food supply than
captorhinomorphs and pelycosaurs. It is dif-
ficult to specify what food was selected, but
it is probable that Petrolacosaurus fed mainly
on terrestrial arthropod invertebrates.

Close phylogenetic relationships of Petro-
lacosaurus to early "eosuchians" are indicated
by derived characters in common with
Youngina, Galesphyrus, Heleosaurus, Kenya-
sauras, and Prolacerta. Most derived petrola-
cosaurid features, not seen in either capto-
rhinomorphs or pelycosaurs, are found in
Youngina. The similar pattern of fenestration
in the skulls of Petrolacosaurus and Youngina
is the most significant of these advanced fea-
tures. Youngina and other diapsids are more
advanced than Petrolacosaurus in some cra-
nial features, but the latter genus is geologi-
cally much older, and understandably more
primitive. The most significant of these spe-
cialized features, not found in Petrolaco-
saurus, involve osteological modifications to-
wards development of a middle ear system
sensitive to airborne sound. There is evidence
to indicate that such modifications in the re-
gion of the ear may have been initiated by
the adaptations already seen in Petrolaco-
saurus. Similarities in function, shape and
orientation of the temporal openings and the
interrelationship of the circumfenestral bones
are strong indications that the upper and
lower temporal fenestrae in Petrolacosaurus,
Youngina and other diapsids are phylogeneti-
cally related.

The erection of the family Petrolacosauri-
dae by Peabody (1952) is valid because there
are several derived characters that can be
used to define this taxon. These derived char-
acters are: an unusually slender, lightly built
premaxilla, unusually thin walled marginal
dentition, sLx elongate cervical vertebrae,
mammillary processes on the neural arch of
the first sacral vertebra, a well developed
ischiadic notch and forelimbs equal in length

70

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

to hindlimbs. Placement of the family Pctro-
lacosauridae at the base of the order "Eo-
suchia" (Peabody, 1952; Reisz, 1977) is how-
ever inappropriate because the latter taxon
is undefined. The inclusion of early diapsid
reptiles into the order "Eosuchia" solely on
the basis of characters that are primitive for
diapsids is invalid. Both the best known
"cosuchian" family, the Younginidae and the
order "Eosuchia" therefore need to be re-
viewed and testable hypotheses of relation-
ships constructed before their phylogenies can
be reevaluated.

Acceptance of Petrolacosaurus as the old-
est known diapsid reptile indicates a long
hiatus in the fossil record of the diapsid
radiation between its first evidence in the late
Pennsylvanian and the subsequent differentia-
tion in the late Permian. The extent of the
early Triassic diapsid radiation confirms that
the origin of the group should be sought
much earlier, probably well within the Paleo-
zoic, as Petrolacosaurus appears to substan-
tiate, but the intervening gap is puzzling.

The conditions of preservation and the

nature of the fossils at Gamett, Kansas sug-
gest that Petrolacosaurus and other terrestrial
tetrapods, invertebrates and plants were
washed into a quiet lagoon from dry ground
(Peabody, 19.52). The gap in the fossil rec-
ord may indicate that the early diapsids lived
in environmental conditions which were not,
under normal conditions, conductive to pres-
ervation— areas away from standing water,
or that areas where they may have been
buried have either been eroded away or not
exposed by erosion.

Recent studies of Upper Permian and
Lower Triassic diapsid reptiles from South
Africa ( Carroll, 1975, 1976, 1977; Cow, 1975;
Harris and Carroll, 1977) and a preliminary
examination of Permo-Triassic diapsids from
Madagascar and Tanzania, show that this
stage of the diapsid evolution is far more
complicated and extensive than formerly be-
lieved. A more complete picture of the early
diapsid adaptive radiation is required in order
to understand fully how Petrolacosaurus re-
lates to later diapsids.

1980

PENNSYLVANIAN DIAPSID REPTILE

71

SUMMARY

Fossil evidence indicates that Petrohco-
saurus kaiisensis, a primitive diapsid reptile
from the Upper Pennsyhanian of Garnett,
Kansas presents a combination of morpho-
logical features that place it near the base
of subsequent diapsid lineages, while also
evidencing close relationships to capto-
rhinomorph reptiles. Shared derived char-
acters (synapomorphies) of diapsids and
Petrolacosaurus include the presence of well
developed superior and inferior temporal,
occipital and suborbital fenestrac. Close
phylogenetic relations between Petrolaco-
saurus and protorothyridid captorhinomorph
reptiles is indicated by the following shared
derived characters: the supratemporal is
reduced to a small narrow sliver of bone
that lies in a shallow groove formed by the
parietal and squamosal, the postorbital does
not reach the supratemporal, the tabular is

small, and the limbs are lightly built. Com-
parison of this species with captorhinomorphs
and other early reptiles indicates that the
diapsid skull configuration may have devel-
oped in response to the selective advantage
of maintaining a relatively small, light weight
skull at the end of a long neck. The con-
figuration of the ear region in both primitive
captorhinomorphs and Petrolacosaurus indi-
cates that these animals were insensitive to
airborne sounds. There is, however, evidence
to suggest that the selective forces for modi-
fications towards development of a middle
ear system sensitive to airborne sounds may
have been initiated in Petrolacosaurus.

Petrolacosaurus is the only described
member of a distinct family of diapsid rep-
tiles whose exact position is difficult to ascer-
tain until the younger diapsids, commonly
called the "eosuchians," become better known.

72

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

LITERATURE CITED

Broom, R.

1907. On some new fossil reptiles from the
Karroo beds of Victoria West, South Africa:
Trans. S, Afr. Phil. Soc, 18:31-42, figs. 7-11.

Broom, R.

1914. A new thecodont reptile: Proc. Zool. Soc.
London, 1914: 1072-77.

Broom, R.

1915. On the origin of mammals: Philos. Trans.
Roy. Soc. Lond., ser. B, 206:1-48.

Broom, R.

1924. Further evidence on the structure of the

Eosuchia: Amer. Mas. Nat. Hist. Bull., 51:67-

76, figs. 1-4.
Camp, C. L.

1945. Prolacerta and the protorosaurian reptiles:

Am. Jour. Sc, 243:17-32, 84-101, figs. 1-6.
Carroll, R. L.

1964. The earliest reptiles: Jour. Linn. Soc.

(Zool.), 45:61-83, figs. 1-14.
Carroll, R. L.

1968. A Diapsid (Reptilia) parietal from the
Lower Permian of Oklahoma: Postilla, 117:1-7,
figs. 1-3.

Carroll, R, L.

1969. A Middle Pennsylvanian captorhinomorph,
and the interrelation.ships of primitive reptiles:
Jour. Paleont., 43:151-170, figs. 1-12.

Carroll, R. L.

1970. Quantitative aspects of the amphibian-
reptilian transition: Forma et Functio, 3:165-178,
figs. 1-5.

Carroll, R. L.

1975. Permo-Triassic "lizards" from the Karroo.
Paleont. afr., 18:71-87, figs. 1-8.

Carroll, R. L.

1976a. Eosuchians and the origin of Archosaurs.
In: Churcher, C. S. (ed.) Athlon: Essays on
Paleontology in Honour of Loris Shano Russell.
Misc. Publ. R. Out. Mus.:58-79, figs. 1-11.

Carboll, R. L.

1976b. Galesjjhtjrus caitemis, a younginid eo-
suchian from the Cistecephalus zone of South
Africa. Ann. S. Afr. Mus„ 72 {4):54-68, figs. 1-3.

Carroll, R. L.

1977. The origin of lizards. In: Andrews, S. M.,
Miles, R. S. and Walker, A. D., (eds.): Problems
in Vertebrate Evolution. Linn. Soc. Symp. Series.
No. 4:3.59-396, figs. 1-15.

Carroll, R. L. & BAmi), I).

1972. Carboniferous stem reptiles of the Family
Romeriidae; Harvard, Mus. Comp. Zool., Bull.,
143(5):321-364, figs. 1-14.

Clark, J. & Carroll, R. L.

1973. Romeriid reptiles from the Lower Permian.
Harvard, Mus. Comp. Zool., Bull., 144(5):353-
407, figs. 1-25.

Ewer, R. F.

1965. The anatomy of the thecodont reptile
Euparkeria capensis Broom. Philos. Trans. Roy.

Soc. Lond., ser. B, 248:379-435, figs. 1-19, pis.
31-34.

Cow, C. E.

1972. The osteology and relationships of the
Millerettidae (Reptilia: Cotylosauria ) . Jour.
Zool., Lond., 167:219-264, figs. 1-25.

Cow, C. E.

1975. The morphology and relationships of
Yutingina capensis Broom and Prolacerta broomi
Parrington. Paleont. afr., 18:89-131, figs. 1-37.

Hahrls, J. M. & Carroll, R. L.

1977. Kenyasaurus, a new eosuchian reptile from
the Early Triassic of Kenya. Jour. Paleont., 51:
139-149, figs. 1-6.

Haughton, S. H.

1924. On reptilian remains from the Karroo beds
of East Africa: Quart. Jour. Geol. Soc, 80:1-11.

Heaton, M. ].

1979. Cranial anatomy of primitive captorhinid
reptiles from the Late Pennsyhanian and Early
Permian, Oklahoma and Texas. Bull. Okla. Geol.
Surv., 127:1-84.

Heaton, M. J.

(in press). The Cotylosauria: .\ reconsideration
of a group of archaic tetrapods. In: Panchen,
A. L. (ed. ), The terrestrial environment and the
origin of land vertebrates. Systematics Assoc,
Spec. Symp.

Hennic, W.

1966. Phylogenetic Systematics. Transl. by D. D.
Da\is and R. Zangerl. Illinois Univ. Press, 263
pp. Urbana, 111.

Holmes, R.

1977. The osteology and musculature of the pec-
toral limb of small captorhinids. Jour. Morph.,
152:101-140, figs. 1-20.

HOTTO-V, N.

1959. The Pelycosaur tympanum and early evo-
lution of the middle ear. Evolution, 13:99-121,
figs. 1-6.

KUHN-SCIINYDER, E.

1963. I Sauri del Monte San Giorgio: Comm.
Instit. Paleont. Univ. Zurigo, no. 16:811-854, figs.
1-31.
Lane, H. H.

1945. New Mid-Pennsylvanian reptiles from Kan-
sas. Trans. Kansas Acad. Sci., 47:381-390, figs.
1-3.

Lane, H. H.

1946. A sur\'ey of fossil vertebrates Kansas: Part
3, the reptiles: Trans. Kansas Acad. Sci., 49:289-
332, figs. 1-7.

Lancston, W.

1965. Oedaleops campi (Reptilia: Pelycosauria ) .

A new genus and species from the Lower Permian

of New Mexico and tlie family Eothyrididae. Bull.

Texas mem. Mus. 9, 1-46.
Manley, G. a.

1972. The middle ear of the Tokay Gecko: Jour.

Comp. Physiol., 81:239-250, figs. 1-5.

1980

PENNSYLVANIAN DIAPSID REPTILE

73

Manley, G.A.

1973. A review of some current concepts of the
functional evolution of the ear in terrestrial
vertebrates. Evohition, 26(4):608-621, figs. 1-6.
Miner, R. W.

1925. The pectoral limb of Ertjops and other
primitive tetrapods. .\nier. Mus. Nat. Hist. Bull.,
51:145-312, figs. 1-104.

Oelrich, T. M.

1956. The anatomy of the head of Cteno.saura
pectinata (Iguanidae): Misc. Publ. Mus. Zool.
Univ. Michigan, no. 94:1-122, figs. 1-59.

Olso.v, E. C.

1936. Notes on the skull of Yomigiiia capensis
Broom. Jour. Geol., 44:523-533, fig. 1.

Olson, E. C.

1961. Jaw mechanisms: rhipidistians, amphib-
ians and reptiles. Amer. Zool., 1:205-215, figs.
1-7.

Parrington, F. R.

1949. Remarks on a theory of the evolution of
the tetrapod middle ear. Jour. Laryng. Otol.,
63:.589-595, figs. 1-4.

Parrington, F. R.

1955. On the cranial anatomy of some gorgonop-
sids and the synapsid middle ear. Proc. Zool. Soc.
Lond., 125:1-40, figs. 1-14.

Parrington, F. R.

1958. The problem of the classification of Rep-
tiles: Jour. Linn. Soc. (Zool.), 44(295) :99-115,
figs. 1-7.

Peabody, F. E.

1949. Mid-Pennsylvanian pelycosaurs from Kan-
sas: Geol. Soc. America, Bull., 60:1913 (abs.).

Peabody, F. E.

1952. Pctrolacosaurns kansensis Lane, a Pennsyl-
vanian reptile from Kansas. Paleont. Contr. Univ.
Kansas, Vertebrata: 1-41, figs. 1-11, pis. 1-3.

Peabody, F. E.

1957. Pennsylvanian reptiles of Gamett, Kansas:
Edaphosaurus. Jour. Paleont., 31:947-949, fig. 1.

PiVETEAU, J.

1926. Paleontologie de Madagascar, XIII. Am-
phibiens et reptiles permiens. Ann. Pal. (Paris),
XV: 1-128, pis. 12.

Price, L. I.

1935. Notes on the brain case of Captorhinus.

Boston Soc. Nat. Hist., Pr., 40:377-386.
Reic, C. a.

1967. Archosaurian reptiles: a new hypothesis on

their origins. Science, 157:565-568, fig. 1.
Reic, C. A.

1970, The Proterosuchia and the early evolution

of the archosaurs; an essay about the origin of a

major taxon. Bull Mus. Gomp. Zool., 139(5):

229-292, figs. 1-16.
Reisz, R. R.

1972. Pelycosaurian Reptiles from the Middle

Pennsylvanian of North America. Bull. Mus.

Comp. Zool., 144(2) :27-62, figs. 1-20.
Reisz, R. R.

1977. Petrolacosaunis, the oldest known diapsid

reptile. Science, 196:1091-1093.

Reisz, R. R.

(in press) The Pelycosauria: A review of phylo-
genetic relationships. In: Panchen, A. L. (ed.).
The terrestrial environment and the origin of land
vertebrates. Systematics Assoc, Spec. Symp.

Robinson, P. L.

1962. Gliding lizards from the Upper Keuper of
Great Britain. Proc. Geol. Soc, no. 1601:137-
146, figs. 1-2.

ROMEB, A. S.

1944. The development of tetrapod limb muscu-
lature— the shoulder region of Lacerta. Jour.
Morph., 74:1-41.

Romeb. a. S.

1947. The relationships of the Permian reptile
Protorosaunis. .i^mer. Jour. Sci., 235:19-30.

Romer, a. S.

1956. Osteology of the Reptiles. University of
Chicago Press, Chicago: i-.xxi, 1-772, figs. 1-248.

Romer, A. S.

1966. Vertebrate Paleontology. University of Chi-
cago Press, Chicago: i-v, 1-468, figs. 1-443.

Romeb, A. S.

1967. Early reptilian evolution reviewed. Evolu-
tion, 21(4 ) :831-833. figs. 1-3.

Romer, A. S.

1971. Unorthodo,\ies in reptilian phylogeny. Evo-
lution, 25( 1 ) :103-1 12, figs. 1-3.

Romer, A. S. & Price, L. I.

1940. Review of the Pelycosauria. Geol. Soc.

America, Special Paper no. 28:i-x, 1-588, figs.

1-71, pis. 1-46.
Stoval, J. W., Price, L. I. & Romer, A. S.

1966. The postcranial skeleton of the giant Per-
mian pelycosaur Cotiilorbynchus romeri. Bull.

Mus. Comp. Zool., 135:1-30. figs. 1-17.
Tatarinov, L. P.

1964. Order Placodontia: Subclass Lepidosauria.

In: Osnovy Paleontologii, 12:332-3.38, 439-493.
Vaughn, P. P.

1955. The Permian reptile Aracoscelis restudied.

Bull. Mus. Comp. Zool., 113:305, 467, figs. 1-15,

pis. 1-2.
Watson, D. M. S.

1951. Paleontology and modem biology. Yale

Univ. Press, New Haven: i-xii, 1-216, figs. 1-77.
Watson, D. M. S.

1953. Evolution of the mammalian ear. Evolu-
tion, 7(2): 159-177, figs. 1-7.

Watson, D. M. S.

1954. On Bolosaunis and the origin and classi-
fication of reptiles. Bull. Mus. Comp. Zool., Ill:
297-449, figs. 1-37, pi. 1.

Watson, D. M. S.

1957. On Millerosaurus and the early history of
the sauropsid reptiles. Philos. Trans. Roy. Soc.
London, ser. B, 240:325-400, figs. 1-23.

Werner, Y. L. & Weveb, E. G.

1972. The function of the middle ear in lizards:
Gekko gecko and Euhlcpharis macularius (Gek-
konoidea). Jour. Exp. Zool., 179:1-16, figs. 1-12.

Wever, E.G.

1973. The function of the middle ear in lizard's-

74

SPECIAL PUBLICATION MUSEUM OF NATURAL HISTORY

NO. 7

Divergent types. Jour. Exp. Zool., 184( 1 ) :97-125,
figs. 1-32.

Wever, E. G. & Lawrence, M.

1954. Physiological acoustics. Princeton Univer-
sity Press: i-xii, 1-354, figs. 1-142.

Wever, E. G. & Werner, Y. L.

1970. The function of the middle ear in lizards:

Crotapliylus collaria (Iguanidae). Jour. E.vp.
Zool., 175:327-342, figs. 1-9.

WiLLISTON, S. W.

1925. The osteology of the reptiles. Harvard
Univ. Press, Cambridge, Mass.: 1-300.

3 2044 072 228 8

AVAILABLE SPECIAL PUBLICATIONS
MUSEUM OF NATURAL HISTORY, UNIVERSITY OF KANSAS

1. Catalogue of publications in herpetology published by the University of
Kansas Museum of Natural History. By Linda Trueb. Pp. 1-15. December
1976. $0.25 postpaid.

2. Catalogue of publications in mammalogy published by the University of
Kansas Museum of Natural History. By Robert S. Hoffmann. Pp. 1-19.
15 February 1977. $0.25 postpaid.

3. Maintenance of rattlesnakes in captivity. By James B. Murphy and Barry
L. Annstrong. Pp. 1-40. 29 December 1978. $3.00 postpaid.

5. The natural history of Mexican rattlesnakes. By Barry L. Armstrong and
James B. Murphy. Pp. 1-88. 14 December 1979. $6.00 postpaid.

6. 1980 Catalog of publications of the University of Kansas, Museum of Natural
History. Pp. 1-28. September 1980. $1.00 postpaid or free with orders.