\^

PERMO-CARBONIFEROUS VERTEBRATES
FROM NEW MEXICO

By

E. C. CASE, S. W. WILLISTON,

AND

M. G. MEHL

WASHINGTON, D. C.
Published by the Carnegie Institution of Washington

1913

QE
C373

CARNEGIE INSTITUTION OF WASHINGTON
Publication No. 181

Copies of this Book
w«re tirst issued

St^' 22 1913

PRESS OF J. B. LIPPINCOTT COMPANY
PHILADELPHIA

PREFACE.

The present work includes a series of papers dealing with the Permo-
Carboniferous deposits of north-central New Mexico and their vertebrate
fauna, so far as known. By courtesy of the Carnegie Institution of Wash-
ington it is pubHshed as a part of the series (publications 55, 145, and 146)
by E. C. Case, dealing with the fauna of the Permian beds of North America.
It has seemed best to us to publish these results jointly in order, by mutual
study, to avoid as far as possible any errors, misunderstandings, or dis-
cussions in literature. The material in each case was studied by the author
whose name appears first in the caption and the description and figures were
then referred to the other author for careful criticism and discussion. Both
authors accept fully all responsibility for the contents of the different chap-
ters, except where dissenting or supplementary notes are attached, signed
by the individual author. In all bibliographical references to the contents
of this work the names of the authors should be given in the order in
which they appear over each chapter.

The chapter by Mr. M. G. Mehl was prepared under the direction of
Dr. Williston and both he and Dr. Case accept responsibility for the state-
ments and figures.

The material described was collected by a joint expedition headed by
Drs. Williston and Case in Rio Arriba County, New Mexico, in the summer
of 191 1. The beds had been previously explored by Mr. David Baldwin,
in 1878 and 1879, chiefly in the interest of Yale University. Mr. Paul C.
Miller, of the University of Chicago, was one of the party and Prof. F. v.
Huene, of the University of Tubingen, was our guest for a short time.

Several of the forms known from New Mexico have been described in
part or wholly elsewhere and such descriptions are not repeated. References
to these articles are given in the list and bibliography of the faima.

E. C. Case.
S. W. Williston.

CONTENTS.

Description of the vertcbratc-bcaring beds of North-central New Mexico. .S. W. Wilhston and E. C.

Case I

Description of Aspidosaurus novomexicanus WiUiston. E. C. Case and S. W. Williston 7

Description of Chenoprosopus milleri gen. et sp. nov. M. G. Mehl 11

Description of a nearly complete skeleton of Diasparactus zenos Case. E. C. Case and S. W. Williston 17

Description of a nearly complete skeleton of Ophiacodon Marsh. S. W. Williston and E. C. Case. ... 37

Description of Scoliomus puercensis, new genus and species. S. W. Williston and E. C. Case 60

Description of certain collections of lx)nes referred to Sphenacodon Marsh. E. C. Case and S. W.

Williston 61

Description of Edapkosaurus Cope. S. W. Williston and E. C. Case 71

Edaphosaurus notomtxicanus, sp. nov. S. W. Williston and E. C. Case . 74

iv

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO

By

E. C. CASE, S. W. WILLISTON, AND M. G. MEHL

Chapter I.

DESCRIPTION OF THE VERTEBRATE-BEARING BEDS OF NORTH-CENTRAL

NEW MEXICO.

S. W. WiLLISTON AND E. C. CaSE.

The following description of the Permo-Carboniferous beds of New Mexico
is a portion of a more extended article by the authors in the Journal of Geology,
only such portions being repeated as bear upon the character of the beds and indi-
cate the conditions under which the animals lived and were preserved.*

The regions explored were El Cobre Canon, north of Abiquiu, and the expos-
ures in the walls of the Puerco and Gallina Rivers, all in Rio Arriba County.

107 00'

106 30

107 '00

Fig. I. — Map of the Mesa Prieta and adjoining country. Fossiliferous Permo-Carboniferous areas shown
by shading. Comparison with the Gallina topographic sheet will make details clear.

El Cobre Canon or basin is formed by the erosion of an unsymmetrical dome-
shaped anticline, more or less faulted on the northeastern and southeastern sides,
the brim formed everywhere by the massive sandstones of basal Upper Triassic
age, the strata sloping in all directions, but chiefly east and west. The basin thus

• The Permo-Carboniferous of northern New Mexico. Journal of Geology, vol. xx, No. i, pp. 1-12.

I

2 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

formed is about 2.5 miles in its greatest extent, in a north-and-south direction.
Its very steep walls, for the most part about 700 feet in altitude, attain their
greatest height in the northwest part, where the altitude may exceed 800 feet and
where the Permian exposures are the greatest.

The erosion of the floor of this basin, acting on the beds of alternating sand-
stones and clays, has formed a series of steps or low cliffs, which for the most part
dip at a small angle toward the west. Toward the sides and upper end of the
cafion these ridges become more prominent, frequently forming high bluffs and
cliffs. The lowermost beds in the canon are deep chocolate-colored sandstones
and fine conglomerates; the latter weather into low, rounded hills, frequently
streaked with greenish layers. Bone fragments were found in these layers in vari-
ous places in the basin. Above these darker colored sandstones are more massive
sandstones, weathering more or less whitish, which ascend at the north end of the
canon to perhaps 350 feet above the stream bed. All vertebrate fossils that we
foimd, of Permian or Permo-Carboniferous age, were below these sandstones, which
form a fairly definite horizon about the basin and which may be taken as the lower
limits of the Trias.

It has been questioned by us elsewhere whether the vertebrate fossils found in
Texas, Oklahoma, southern Kansas, Illinois, and Pennsylvania are really of Permian
age. At the south side of the cafion Case found a perfect cast of a Spirijer,
identified by Professor Schuchert as S. rockymontanus Marcou, a form occurring in
Colorado in the Pennsylvanian. Though the specimen was found free, so that its
exact horizon could not be determined, its excellent preservation proves conclusively
that it had not been carried far from its original bed, and inasmuch as vertebrate
fossils are found in the deepest strata of the canon it seems quite certain that the
specimen came from an intercalated bed among those yielding so-called Permian
vertebrates. No other explanation seems possible. It is the conviction of both
the present authors that the lowermost at least of the strata yielding vertebrate
fossils are of Pennsylvanian age, and this conviction is strengthened by the known
position of the vertebrate horizons in Texas, Kansas, Illinois, and Pennsylvania,
that of the last-named region being definitely known to be Pennsylvanian.

Section I. EL COBRE CANON.

Feet.

Yellow sandstone and conglomerates 75 Upper Trias

Purplish and gray clays 40 ?

Purplish sandstones 20 Lower Trias? Barren

PurpUsh clays and nodular sandstone 25 " "

Red sandstone 5 " "

Bright red clay 35

Purple clay 12 " "

Bright red sandstones 22 " "

Co^-se purplish sandstones 12 " "

Bright red clay with greenish nodules and purplish bands . . 100 " "

Coarse, hard purplish sandstone 8 " "

Bright red clay and sandstone 65 " "

Hard red and purplish sandstone 6 " "

Bright red sandy clay, with purplish streaks 90 " "

Purplish and dark brown clay 22 " "

Red clay and hard red sandstone 30 " "

Hard purpUsh sandstones 35 " "

Red clay 7 ?

Purple sandstones 3 ?

Red clay 22 ?

Permo-Carboniferous red and brown sandstones and clays,
fossiliferous

A section of the west wall of El Cobre Cafion, as far down as the horizons
jrielding fossils of Paleozoic age, is given above. It must be especially remem-
bered, however, that this and Section II, on p. 4, will not apply in detail to any

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 3

Other place, since it has been our experience in the Red Beds that detailed sections
made in any given place can not be depended upon perhaps a quarter of a mile
away. The top of this section, as already stated, yields vertebrate fossils of Upper
Triassic age, and some of the Triassic vertebrates described by Cope from New
Mexico came from El Cobre Canon.

A survey of the surrounding country from the summit here, as also from the
Piedra Lmnbre, shows everywhere these basal Upper Triassic rocks as the lower
or lowermost exposures.

From El Cobre Canon the expedition turned westward on the Chama to the
mouth of Caiiones Creek, and then southwest across the Piedra Lumbre Mesa to El
Rito ("branch") of the Puerco. The top of this mesa is formed of rocks of Upper
Triassic age, and near the base of Mount Pedemal, which rises several hundred feet
above the mesa, appears for the first time the heavy layer of gypsum marking the
upper limits of the Red Beds, or so-called Trias. From the top of this mesa a good
view of the adjacent country is afforded. To the north is the Mesa de los Viejos,
with the Chama apparently occupying a fault line between, and the Arroyo Seco in

Fig. 2. — Map of region around Abiquiu, showing location of El Cobre Caflon.

a valley formed by the basal Upper Trias rocks sloping from the brim of El Cobre
Canon on the east and the superincumbent Triassic rocks on the west. To the
west lie the Mesa Prieta and the smaller Capulin Mesa, separated by the Puerco,
Chama, and Capulin, streams whose courses seem to have been influenced strongly
by the faulting and dipping of the Trias.

The Puerco, to the mouth of the Poleo, has cut down into Permian strata, which
attain their greatest exposure on the Poleo about a mile from its mouth. Our
first camp was made on the Poleo (Arroyo de Agua), about a mile above its con-
fluence with the Puerco.* Near the junction of the two creeks is a steep-walled
cliff of Permian rocks about loo feet in height, with a more or less fiat table-land
above it, a mile or so in extent, separating it from the Trias above. Farther west,
where the Permian rocks find their greatest exposure, and where the Baldwin
quarry lies, from which so many of the fossils in the Yale collection came, the very
steep bluffs, in many places so steep as to be unclimbable, are about 700 feet in
height. They are composed of alternating red sandstones and clays, with white
and purple sandstones, clays, and conglomerates at the upper part, corresponding

• This Puerco Creek is not the one which gave origin to the name of the Puerco formation,
formatiun name was derived from a stream by the same name farther to the southwest.

The

4 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

quite to the massive sandstones forming the brim of El Cobre Canon and the
top of the Mesa Poleo, dipping northward to the foot of the Mesa Prieta. Phy-
tosaur bones were found at the base of this white sandstone and in the pebbly
conglomerates immediately underlying them. Permian fossils were found only in
the lowermost 300 feet of these exposures, the intervening 300 feet of more or less
vertical red clays and sandstones here, as everywhere else in the Rocky Mountain
region, being quite barren. These rocks lie here, as elsewhere, apparently quite
conformable with the superincumbent and subjacent beds, and doubtless repre-
sent the Lower Trias and perhaps more or less of the Upper Permian. The sec-
tion of the bluff herewith given was made opposite our first camp on the Poleo,
about a mile from the mouth of the creek; as is the case with the section at El
Cobre, it can be depended upon only for a short distance on either side ; the strata
often change abruptly from sandstones to clays and vice versa.

On the north side of the adjacent valley of the Poleo the strata dip northward
to the walls of the Mesa Prieta; just south of the creek they dip abruptly south-
ward. About 2 miles above the mouth of the creek the beds bend down sharply
and disappear beneath the alluvial deposits of the creek bed, doubtless indicating
the line of a fault. Beyond this point the walls of the Mesa Prieta, formed exclu-
sively of Upper Triassic and superincumbent beds, descend to the adjacent valley
of the Poleo and Capulin creeks.

The Mesa Prieta rises about 1,500 feet above the beds of the Poleo and
Capulin. Near the middle of the bluffs, at about the 8,000-foot line, there is a
heavy bed of gypsum, which is taken to be the upper limits of the Trias, though,
as we have said, in the entire absence of all fossil remains through 400 feet of these
beds at least, everywhere, their age is assumed simply from their color — evidence
that is, to say the least, exceedingly dubious, the more so from the fact that there
is no petrological distinction between the Trias and Permian. Above this gypsum
layer are the brownish and purplish shales of the Jurassic* and the lighter-colored
sandstones of the Cretaceous, all lying quite conformably with the Red Beds below.

Section II. POLEO CREEK.

Gray sandstones, mostly even-grained with pebbly con- F««t-

glomerates and shales below. Phy tosaurs 30 Upper Trias

Softer gray sandstones, weathering into sand 30 ?

Sandy clay, with beds of thin black shale; plant remains,

fossil wood 40 Upper Trias?

Sandy clay, black and green 13 Lower Trias? Barren

Purplish sandy clay 6 " "

Coarse yellow sandstone 33 " "

Loose gray sand 6 " "

Green and purplish sandstones 3 " "

Gray and purplish sandstones 13 " "

Hard clay, variegated and jointed (clifls) 3 " "

Purplish sandstones and red clay 30 " "

Loose white sand with beds of red clay 40 " "

First red nodular layer 6 " "

Soft fine-grained, light red sandstones, cross-bedded, forming

tops of pyramids and clifls 6 " "

Soft fine-grained, light red sandstones, cross-bedded with

lighter bands of pebbles 3 " "

Second red nodular layer, with clay 3 " "

Red clay 2

Coarse cross-bedded sandstones 6 " "

Third red nodular layer 6 " "

Red cross-bedded sandstones 13 Lower Trias? Barren

• The beds immediately overlying the gypsum have been called Dakota by Darton {Bull. U. S. G. S.
No. 43S) and Shaler {Bull. U. S. G. S. No. 31$, p. 26*). We searched in these shales for fossils, but without
Buccew. Elsewhere the beds overlying the Trias are either the marine Sundance (Wyoming), the Hallopus
beds (Cafton City, Colo.), Morrison (southern Wyoming), or Lower Cretaceous (Kansas).

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.
Section II. POLEO CREEK— Continued.

Feet.

Dark red and green clay 3 Lower Trias? Barren

Coarse red and green sandstones 17 "

Hard red sandstones, cliffs 3 "

Red sandstone and clay with thin band of harder sandstone 18 "

Hard, red, coarse sandstone 3 "

Red clay 35

Red sandstone, bluffs 23 Base (?) of Trias

Red sandstone with thin seams of clay, reptile bones 18 Top (?) of Permian

Red clay with thin nodular layers 55

Dark red, coarse sandstones (cliffs) 6 Permian, fossiliferous

Red sandy clay 35

Dark red, coarse sandstone, jointed 18

Red clay, even texture, vertical rain erosion 50

Dark red clay 8

Red shaly clay 17

Heavy gray sandstones 25

Red sandstones and clays ?

About 1.5 miles beyond the little settlement called Capulin a stream flows
into Capulin Creek from the north along the line of a fault which divides the Mesa
Prieta from the Capulin Mesa. The strata of the Mesa Prieta at this point dip
slightly northwest, but those of the Capulin Mesa dip east and northeast. The
west face of the Capulin Mesa rises 1,000 feet or thereabouts above the valley of
the Gallina River. Just north of the Cerro Blanco is a high red wall similar to
that north of the Poleo Creek, the uppermost rocks bearing phytosaur remains.
It is confidently believed that the lowermost exposures here are of Permian age,
but no fossils were found.

There is a sharp break between the Capvilin Mesa and the Cerro Blanco.
The rocks of the latter dip sharply to the west and are overlain by the Jurassic
shales and the Cretaceous sandstones and shales. At the foot of these Upper
Triassic rocks, north of Cerro Blanco, and opposite the face of the Capulin Mesa
bluflf before referred to, were found various small fresh-water invertebrates, and
bone fragments referred provisionally to the genus Caslophysis Cope. The horizon
of these remains can hardly be less than 100 feet above the basal Upper Trias
sandstones, and in all probability the original types came from the immediate
locality where the fragments were fotmd by Case. The Cerro Blanco takes its
name from the massive beds of white gypsum which cap it, descending steeply
below the creek bed to the south and dipping to the west. From the top of
the Cerro Blanco one can look miles to the north and west, and the view there-
from is a revelation to the geologist. To the east lie the mesas of more or less
horizontal rocks of predominantly Upper Triassic age; to the west the strata are
deeply tilted and eroded into valleys; a few miles farther west the beds of the
Wasatch badlands lie horizontally upon the uptilted edges of the Mesozoic strata.

Upon the whole, the general features of the Red Beds in northern New Mexico,
as in many places elsewhere, may be summarized as follows:

The Upper Trias rocks, about 600 feet in thickness, perhaps more, are pre-
dominantly softer and lighter colored, often orange colored, yellowish, and whitish,
and more aeolian in character, with the upper or uppermost beds more or less
gypsiferous. These beds, as in the Lander region, have basal sandstones, reddish
or white, with conglomerate and clay layers below them yielding phytosaur and
labyrinthodont bones (both types were found at El Rito), corresponding well
with like vertebrates from the Keuper of Europe. Below these beds there are not
less than 350 feet (in the Lander and Kansas regions perhaps 900 feet) of more
uniform red sandstones and clay layers, usually weathering into more vertical
bluffs, that are utterly barren of all fossils and supposed to be of Lower Triassic

6 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

and Upper Permian age. Below these and conformable with them, in New Mexico
and probably elsewhere, are not less than 300 feet, probably more, of prevailing
coarser and darker- colored, often brownish sandstones, and dark-colored clay
beds, yielding vertebrate remains hitherto considered to be of Permian age, but
which in all probability are in part at least of upper Pennsylvanian age.

List of Known Vertebrates from the Permo-Carboniferous Beds of New Mexico.

Pisces. Shark, like Pleuracantkus.
Amphibia.

Eryops (?) reiiculaius. Cope, Am. Nat., vol. xv, p. 1020; Case, Pub. 146, Carnegie
Institution of Washington, p. 30.

* Eryops {?) grandis Marsh. {Ophiacodon grandis) Marsh. Am. Jnl. Sc, vol. xv, p.

211. Williston (Eryops grandis), American Permian Vertebrates, Chicago, 19 11,

p. 10.
Aspidosaurus novomexicanus Williston. American Permian Vertebrates, Chicago,

igii, p. 12.
Platyhysirix rugosus Case. Case (Ctenosaurus rugosus), Bull. Am. Mus. Nat. Hist.,

vol. xxviii, p. 176. Williston (Platyhystrix rugosus), American Permian Verte-
brates, Chicago, 191 1, p. 135.
Chenoprosopus milleri Mehl.
Gams indet., a small unidentified humerus.
Reptilia.

Diadectes {Nothodon\) lentus Marsh. Marsh (Nothodon), Am. Jnl. Sc, vol. xv, p. 410.

Case (Nothodon), Publication No. 145, Cam. Inst. Washington, p. 30. Williston

(Nothodon), American Permian Vertebrates, Chicago, p. 16. Case and Williston

(Diadectes lentus), Am. Jnl. Sc, vol. xxxiii, p. 339.
Animasaurus carinatus Case and Williston. Am. Jnl. Sc, vol. xxxiii, p. 339.
Diasparactus zenos Case. Bull Am. Mus. Nat. Hist., vol. xxviii, p. 174.
Elcobresaurus baldwini Case. Publication No. 55, Cam. Inst. Washington, pp. 28

and 89.
Arribasaurus navajovicus Case. (Dimetrodon navajovicus) Publication No. ss,^Cam.

Inst. Washington, pp. 56 and 137.
Ophiacodon mirus Marsh. Am. Jnl. Sc, vol. xv, p. 411. Baur and Case, Trans. Am.

Phil. Soc, vol. XX, p. 5. Williston, American Permian Vertebrates, Chicago, 1911,

p. 81.
Sphenacodon ferox Marsh. Am. Jnl. Sc, vol. xv, p. 410. Baur and Case, Trans. Am.

Phil. Soc, vol. XX, p. 4. Case, Publication No. 53, Cam. Inst. Washington,

p. 66. Williston, American Permian Vertebrates, Chicago, 191 1, p. 78.
Edaphosaurus novomexicanus Williston and Case, sp. nov.
Scoliotnus puercensis Williston and Case, gen. et sp. nov.
Limnoscelis pcdudis Williston. Am. Jnl. Sc, vol. xxxi, p. 378. Idem, vol. xxxiv,

p. 457. American Permian Vertebrates, Chicago, 191 1, p. 23.

The genera Dimetrodon and Clepsydrops previously determined from the New
Mexican locality have not been confirmed by later work.

* The reference of these two species to Eryops is in a measure provisional, as later studies indicate a
closely allied but distinct genus.

t The characters of Nothodon, so far as they have been determined, are identical with those of Dia-
decUs,but there is a very considerable possibility that the skeletal characters other than those of the skull,
the only part known, may justify the original generic name.

CHAPTER II.

A DESCRIPTION OF ASPIDOSAURUS NOVOMEXICANUS WILLISTON.
By E. C. Case and S. W. Williston.

Very few fossils were found on the west side of the Puerco River opposite
El Rito, but from a small lens of clay inclosed in a bed of sandstone Mr. Paul
Miller collected several associated bones which prove upon preparation to be the
skull, portions of the vertebral column and carapace, the scapulae, femora, humeri,
portions of the pelvis, and some parts of the smaller limb bones of an animal
indistinguishable from Aspidosaurus novontexicanus Williston. The type of this
species (No. 8io Yale University Museum) is only partially preserved and the
identification of the present specimen may not be accurate, but the resemblance
of the parts preserved, especially the skull, is so close that it seems best to use the
earlier name, provisionally, until more material can be obtained.

The most striking thing about this fossil is the strong resemblance of the
skull to that of Cacops aspidephorous Williston from Texas. Aside from the condi-
tion of the otic region, the two skulls are almost identical. The same can be said
for the pelvis, the limb bones, and the vertebral column, aside from the dorsal
armor ; this last is apparently composed of a single row of plates formed by expan-
sions of the neural spines, the typical condition for the family AspidosauridcB.
A fragment of detached armor accompanying the specimen seems to be composed
of two plates to each vertebra, but as the fragment consists of the plates alone,
this point is not absolutely certain. The plates do not lie in two vertical series,
as in the DissorophidcB, but on the same level. On two of a series of separate
plates, each being one of a pair, there is a descending process evidently for attach-
ment to a neural spine; the presence or absence of a descending process in the
others can not be made out because of the condition of the specimen. The mean-
ing of this peculiar arrangement is not evident. Perhaps, as suggested by Willis-
ton, the forms included in the families AspidosauridcB and DissorophidcB axe more
closely related than has been supposed and this form may be a connecting link.
However, the separation of the two families is very definite and easily determined
and it seems desirable to wait for more evidence before abandoning the distinction.
It is possible that the plates may come from the caudal region or (less probable
but still possible) they may belong to another animal; this last suggestion is
strengthened by the extremely slender descending process, which is far too deli-
cate to have been a part of any of the neural spines preserved.

The skull: There is a strong resemblance between this skull and that of Cacops,
and this extends to all minor details if we except the condition of the otic region
and the ridges which run posteriorly from the orbit. In Cacops, Williston described
two ridges running back from the posterior inner comer of the orbit, inclosing a
triangular space; in this specimen there is but one ridge, the upper. The pos-
terior portion of the skull has been nearly perfectly preserved on the left side, but
that of the right has been largely restored in plaster. Fortunately the ends of the
tabulare bones and the quadrates are preserved on both sides and this renders it
very certain that in this specimen the otic opening was a deep notch, not closed
by a posterior bar of bone as in Cacops and Dissorophus. This determines the

7

8

PBRMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

position of the quadrate bone in the lower bar of the otic opening and shows that
it did not form part of the posterior bar of the opening in Cacops.

A careful examination of the bones of the temporal region of the left side shows
a series of lines, very certainly sutures, as they are different in character from
cracks which accompany them, and as some of them show, for a short distance,
the peculiar wavy form of sutures The course of these sutures and the relations
of the bones thus outlined are shown in fig. 3 a. Similarly the outlines of the
maxillaries, the lachrymals, and the jugals can be made out. From the strong
resemblance of the skull to that of Cacops, in other regards, it is probable that the
general course of the sutures was the same as in that genus. The otic notch is
widely open posteriorly and the surfaces of the bones bordering it are smooth;
but the adjacent portions of the skull, both on the sides and the top, are marked
by a fine sculpture and pitting.

On the lower surface of the skull (fig. 3 b) the bones have the same general
form and relationships as described by Williston in Cacops. The occipital condyles
are well formed and inclined sharply backwards and inwards. The basisphenoid

Fig. 3. — Aspidosaurus novomexicanus, No. 673, University of Chicago. X Jj.

A, upper surface of skull, shaded parts restored; B, lower surface of skull;

C, three dorsal vertebrae with attached armor plates.

is well developed, with strong basipterygoid processes corresponding to the enlarge-
ment in the same position in Cacops. The basisphenoid was well ossified and
evidently as fully functional as in the Cotylosauria. The ophisthotics are pre-
served, but the inner ends, perhaps due to injury, do not have the broad and
fan-shaped form figured by Williston for Cacops. On the left side there is a
slender styliform bone near the anterior end of the otic notch, which is without
doubt the stapes, as it corresponds exactly in position with the same bone in
Cacops. Anteriorly the large rhinencephalic chamber lies detached, by injury,
from the basisphenoid, and on its lower surface lies the poorly developed para-
sphenoid. The lower jaws lie in position and most of the teeth are hidden, but a
few in the right jaw and more in the maxillary show them to have been slender
and conical; there is no evidence of any enlarged teeth, either in the upper or the
lower jaws. The matrix and the lower jaws conceal the anterior portion of the
palate, but so far as can be made out there is no trace of the large teeth which
appear on the palate of Cacops.

The femora, humeri, scapulce, and the pelvis (lacking only the crest of the
ilium, as preserved) are indistinguishable from the same bones in Cacops.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 9

The vertehrm are typically temnospondylous. There are three short series
of vertebrae, one with the armor plates attached (fig. 3 c). In this the neural
spines are short and heavy and the neural expansions are firmly attached to the
spines; a careful microscopic examination of the broken surface reveals no trace
of a suture or any interruption of the course of the bone fibers as preserved. The
diapophyses are simple broad plates directed outward, with the single articular
face directly slightly forward at the anterior end, as is common. There is no indi-
cation that the ribs were other than single-headed. In another short series of
vertebrae the intercentra and the pleurocentra are preserved ; the former are simple
without any process for the rib, as in Aspidosaurus glascocki Case, but the condi-
tion is such that it is impossible to say whether there was a facet or not. The
pleurocentra are small elements of indefinite form in the specimen.

There is a single detached fragment showing a series of armor plates, as men-
tioned above. This is bent upon itself so that at one end the ventral faces of the
plates are in contact. The descending process of the anterior of the first pair of
plates is very slender and could not have joined any of the vertebrae preserved.
There is no descending process on the second plate. In the posterior plate there
is also an indication of a descending plate, but this is less certain than in the first.
If these plates came from the same animal as the vertebrae they can only have
come from the caudal region, in which no plates occur in Cacops or in the Yale
specimen of Aspidosaurus novomexicanus ; if from another animal, they represent
a form as yet undescribed.

A few poorly preserved phalanges show that the foot was broad and short.

Chapter III.

A DESCRIPTION OF CHENOPROSOPUS MILLERI, GEN. ET SP. NOV.

By Maurice G. Mehl.

The specimen herein described was discovered by Mr. Paul C. Miller, of the
University of Chicago expedition, in 191 1, on Poleo Creek, in the vicinity of Arroyo
de Agua, northern New Mexico. It was found about 50 feet away from and ap-
proximately 2 feet below the skeleton of Ophiacodon Marsh described by Williston
and Case in the following pages. The horizon is almost identical with that of the
type specimens of Ophiacodon and Sphenacodon Marsh, collected by Mr. David
Baldwin in 1878. The illustrations in this chapter were drawn by the author.

Fig. 4.^Skull of Chenoprosopus milleri, X yi- A, lateral view; B, dorsal view.

The material on which this genus and species is based consists of an incom-
plete skull. The anterior third and the left side, to the median line, are complete,
with the exception of the loss of the upper posterior border of the skull, a little
of the quadrato-jugal region, and a small portion about and including the left
naris. Fragments from the right side were obtained, but they are so badly crushed
and so incomplete that they are of little use other than in determining the position
of the teeth. The specimen was inclosed in a hard sandstone matrix which, for
the most part, was freed from the bone quite readily, leaving many of the sutures
distinct and the pitting undisturbed. The skull is somewhat distorted and slightly
compressed laterally. This distortion has cracked the surface and this, with the
rather thorough ossification, has made the determination of some of the sutures
extremely difficult. Only those that can be determined with certainty are shown
in full lines, while the others are indicated by dots.

3 II

12 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

The skull is long and narrow, broadly rounded anteriorly, with the sides
diverging slightly and regularly in the anterior four-fifths of the length. From
this point the outline is slightly convergent. The greatest length of the specimen
is about 288 mm. The width at a point 30 mm. back of the tip of the rostrum,
the approximate length of the radius of the rounded anterior border, is about
54 mm. The greatest width of the skull was probably not over 90 mm. The
sides rise abruptly to the plane of the table of the skull. In a lateral view the
profile is triangular and low (fig. 4 a). The upper border of the anterior half is
nearly straight and rises gradually from a height of about 13 mm. near the tip to
about 33 mm. near the middle. From this point it rounds up gently in a concavo-
convex curve to the horizontal plane extending over the posterior third of the
skull. The greatest height is approximately 65 mm.

The orbit is nearly round and is about 32 mm. in diameter. The anterior
border is a little back of the middle of the skull. The plane of the orbit is parallel
with the lateral margin of the skull and is directed out and slightly upward. No
sclerotic plates are present. The naris, which has a rounded anterior border and
is straight behind, is about 10 mm. wide and 7 mm. long. It is placed near the
lateral margin of the skull, about 50 mm. behind the tip of the rostrum. Its plane
is directed out and upward. No pineal foramen has been observed. It is possible,
however, that one was present, but slightly to the right of what seems to be the
median line, and that it has been broken away in this specimen. From the middle
of the posterior border of the orbit a depression extends back and slightly down for
a short distance. A similar depression extends forward from the anterior border
of the orbit and a third, somewhat smaller, lies a little behind the naris and in a
line with the orbit and the naris. The dorsal surface (fig. 4 b) of the rostrum is
slightly concave laterally from a point near the tip back nearly to the middle of
the skull. The entire surface of the skull is marked with irregular pitting. In the
upper posterior portion the pits are large and deep and are separated by irregular
ridges. Anteriorly and below the pits become progressively smaller. Occasion-
ally they rim into each other to form antero-posterior grooves. This is especially
true on the posterior end of the jugal. Here several lie parallel and are crowded
close together. Extending back from the lower posterior comer of each naris is a
smooth, shallow groove, evidently a mucus canal. The specimen probably belongs
to an adult individual, for the ossification is so complete that no trace of suture
is seen between the parietals or the frontals.

The premaxillcB are large and broad. They form the lateral boundaries of
the nares as well as their anterior border. Their posterior border forms a con-
cavity into which the nasals extend. The sutures bounding the maxillcB are uncer-
tain, except for a short distance near the anterior end. From the narrow anterior
extremity, which enters but slightly into the posterior border of the nares, they
gradually broaden posteriorly to within a short distance of the orbit. From here
they rapidly narrow to the posterior end, which apparently takes part slightly in
the lower anterior border of the orbit. The extremity seems to be extended into
an upper and a lower process that embrace the jugal. The teeth extend back to
the middle of the orbit. Many of the teeth of the premaxillae and the maxillae are
broken or wanting ; still a fairly good idea of their size and number can be gained
from those present; they are placed rather irregularly and show some variation
in size. Apparently there is no arrangement according to size except that the pos-
terior ones are all small. The variation in length and diameter is due, perhaps,
to new teeth taking the place of those that have been lost. On each premaxilla
there are 8 or more teeth and on each maxilla about 18, or between 52 and 56 in

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 13

all. They are all conical, slender, and more or less recurved. At the posterior
end of the maxillae there are a few that reach a length of only 4 mm. From this
point they increase rapidly in length to an average of about 1 1 mm. At the base
the cross-section of all is probably 3.5 mm. or less.

The septomaxillcB are small, triangular elements, fully twice as long as wide,
and form nearly the entire posterior border of the nares. They extend back be-
tween the nasals and the maxillae, but are apparently separated from the lachrymals
by a considerable space. The nasals are large, rectangular in shape, about three
times as long as wide. While the lateral boundary is not certain, the anterior
half seems to be formed by the maxillae and the posterior half by the lachrymals.
Anteriorly, the nasals take part slightly in the inner posterior border of the nares,
posteriorly they have a broad connection with the frontals and connect slightly;
perhaps, with the prefrontals.

The frontals are long and narrow, being fully four times as long as wide. They
extend nearly to the posterior border of the orbits. The anterior three-fifths of
the lateral boundary is formed by the prefrontals and the postero-lateral border
seems to be formed by the post-frontals. Lying about 25 mm. or 30 mm. behind
the nares and between the nasals and the maxillae there seems to be a long, narrow
element, the lachrymal. To all appearances it takes part in neither the border of
the nares nor that of the orbit. The posterior boundary is formed by the pre-
frontals and the maxillae. The prefrontal is a triangular bone forming the entire
anterior border of the orbit. Anteriorly it connects slightly with the nasals. The
postfrontal forms the upper border of the orbit and apparently excludes the frontal
from that opening by uniting with the prefrontal anteriorly. It extends back of
the posterior border of the orbit some distance between the parietal and the post-
orbital. While the suture on the inner side is not certain the element seems to
have but little lateral extent. The parietal is sub-rectangular, long and narrow.
The posterior extremity is missing and its full extent can not be determined.
Laterally it is bordered from the front backward by the postfrontal, postorbital,
and supratemporal.

The postorbital is rather large, as wide as long, and forms the entire posterior
border of the orbit. Immediately behind this and in contact with the parietal
above lies a large rectangular element, probably the supratemporal. Bounding
this below is a small portion of the squamosal, the lateral extent of which can not
be determined, as the lower part is missing. At best, however, it must have been
a comparatively small element. A small rounded notch, apparently the ear notch,
separates the supratemporal and the squamosal for a short distance posteriorly.
The jugal is a large element extending from the anterior border of the orbit back
at least as far as the ear opening. It forms the lower posterior border of the orbit
and increases somewhat in width posteriorly.

Unfortunately a complete knowledge of the region of the ear opening is pro-
hibited by the condition of the specimen. When found the skull was broken through
at this point and some of the bone was weathered from the exposed surfaces. The
quadrate is present in the matrix, however, and serves to give some idea, at least,
of this part of the skull. From the upper side of the quadrate a thin, plate-like
bone extends up and forward in a direction as if to connect with the skull just
above the ear notch. The outer edge is rough and pitted like the surface of the
skull, while the posterior, incurved surface is smooth. The position of this process
is such as to suggest its homology with the downward exten.sion of the tabulare
in Cacops Williston.* In that genus the tabulare extends back and downward to

• Williston, Bull. Geol. Soc. Am., vol. 21, p. 354.

14

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

fuse with the quadrate and thus completely close the ear notch. A thin plate of
bone extends forward from the lower anterior inner edge of the tabulare and leaves
but a small slit-like opening at the bottom of the large depression back of the ear
notch. No such process is present in Chenoprosopus milleri, however, and the
opening, if an opening, must have been considerably larger in the latter genus.

Little can be said of the palate; the anterior part is still inclosed in an extremely
hard matrix, and although the left posterior part is exposed, the bones are some-
what crushed and their relations can not be told with any degree of certainty. The

Fig. 5. — Chenoprosopus milleri, X yi. Palate view.

palate is essentially fiat. The conjoined palatines and pterygoids are broad and
leave a long, comparatively narrow inter-pterygoid opening. This is especially
true in the posterior fourth of the palate, where the posterior processes of the
pterygoids converge in a gentle curve inward to join the quadrate. In fact, unless
the skull is compressed much more than the specimen would suggest, the opening

at this point is a mere slit. The posterior
pterygoid process extends upward in a thin
plate, inclosing a large infra-temporal vacuity.
This is approximately rectangular, about 65
mm. long and 35 mm. wide. The inner ante-
rior border is broken, but suggests a postero-
lateral process such as is seen in Cacops.
While the parasphenoid is not preserved, one
can say with some degfee of certainty that its
lateral extent was slight, because of the nar-
row inter-pterygoid space.

Although the entire palate dentition can
not be shown, some of the details are worth
considering. Extending forward for a distance
of about 40 mm. from the anterior border of
the infra-temporal vacuity is a low rounded
ridge, the posterior end of which is continuous
with the inner border of the posterior process
of the pterygoid. The anterior end curves outward slightly, away from the palate
opening. This ridge is about 5 mm. wide and 3 mm. high and is closely studded
with very fine tubercles. A similar ridge, but less pronounced, runs parallel to
this on the outer side, at a distance of about 6 mm. It extends forward from the
anterior border of the post-temporal vacuity for a distance of about 25 mm. This
is also studded with the fine tubercles, as is the inner edge of the pterygoid to some
extent. Along the lateral edge of the palate, running in a line parallel to the maxil-
lary teeth at a distance of about 11 mm., are exposed four large teeth. These are
spaced irregularly, the first about 60 mm. back from the tip of the rostriun and the

Fig. 6.^Diadectid vertebra, X >^. A, lateral
view; B, anterior view.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 1$

last near the middle of the skull or about 50 mm. in front of the anterior border
of the pdst-temporal vacuity. Two are placed close together in the posterior, one
of which is about 2 7 mm. in front of the last tooth in the series. Others are probably
present but hidden by the matrix. They are stout, conical teeth, more or less
recurved, averaging about 9 mm. at the base, and are about 19 mm. long. Cross-
sections show their labyrinthine structure.

A single vertebra was found crushed into the palate of the specimen (fig. 6).
The centrum is about 25 mm. in diameter and 24 mm. long. The general pro-
portions and even the details are very similar to those of Diadectes, with the
exception that the hyposphene of Diadectes is not present. From Diasparactus it
apparently differs only in that the latter has a rounded, somewhat longer spine,
while the present specimen has a quadrangular section with the diagonals directed
antero-laterally and transversely and with somewhat concave sides. It is pos-
sible that the vertebra belongs to Nothodon. In short, it is distinctly of the Dia-
dectid type.

To assign to Chenoprosopus a definite systematic position among the Permian
vertebrates would be mere speculation at best, till more is known of the skeleton ;
but one is justified in considering the genus as belonging to the Temnospondyli,
in spite of the fact that it has certain reptilian aspects. The vertebra found in
association with the skull has all the appearances of that of a cotylosaurian, and
its size and the amphibian nature of the skull show that the association must
have been accidental and that the two could not belong to the same type. The
narrow inter-pter>'goid opening and the small parasphenoid that must have been
present are not unknown among the Temnospondyli. In Cacops, while the inter-
pterygoid space is wide, the parasphenoid is exceptionally narrow and in Tre-
matops * it is vestigial or even wanting. Then, too, the irregular pitting of the
skull, the probable mucus canals, and the large, conical teeth of the palate show
its amphibian character.

Of the American Permian Amphibia, Chenoprosopus resembles Cacops most
nearly, perhaps. It differs from Cacops, however, in the smaller nares, not only
relatively but absolutely, and in their posterior position in Chenoprosopus; in the
much narrower inter-pterygoid space of the latter, and in many minor points. The
differences from Trematops are even greater: Trematops has an unpaired mucus
canal or opening in the anterior part of the rostrum, large anterior nares con-
fluent with the antorbital vacuities, and a wide palatal opening, none of which
are present in Chenoprosopus. In general appearances the skull very much resem-
bles that of the Eiu-opean Archegosaurus.\ Both are comparatively long and
tapering and in neither are the nares terminal. Even the arrangement of the
cranial elements in the two forms, so far as it has been made out in Chenoprosopus,
is similar. The posterior borders of the skulls, however, if the present interpreta-
tion of Chenoprosopus be correct, are very different. In Archegosaurus the border
is indented by an otic notch, while in Chenoprosopus the posterior border is evi-
dently formed by a bar that closes the ear notch, as described above. The wide
inter-pterygoid space and the narrow pterygo-palatines of Archegosaurus % are also
strikingly different from the condition found in Chenoprosopus. Besides these,
there are many minor differences, such as the antero-posteriorly elongate nares
and the more or less upward directed orbits of Archegosaurus, which are in con-
trast with the laterally directed orbits of Chenoprosopus and the nares that are

• Williston, Jour. Geol., vol. 17, No. 7, Oct.-Nov. 1909, p. 642.
t Jaekel, Zeitschr. d. d. Geol. Gesellsch., vol. 48, 1896.
X Credner.

i6

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

slightly wider than long. In all probability a more complete knowledge of the
form will show that the distinctive characteristics are of family value. (See fig. 7.)
The laterally directed orbits of Chenoprosopus are strongly suggestive of a
terrestrial life. The sharp, conical teeth of the maxillaries and the tubercle-
studded ridges on the posterior part of the palate would fit the animal admirably
for feeding on the worms and the larvae of the gigantic insects of its time.

Fig. 7. — Chenoprosopus miUeri, X %.

A, outline of skull, lateral view: pmx, premaxilla; smx, septomaxilla; na, nasal; mx, maxilla; /, lachry-
mal ;/r, frontal; pjr, prefrontal; 0, orbit; pt}, postfrontal; pto, postorbital; pa, parietal; ju, jugal; ptr, ptery-
goid; qu, quadrate; ta, tabulare; on, otic notch; st, supratemporal (?); sq, squamosal.

B, dorsal view; lettering as in A.

The skull described above is No. 670 of the University of Chicago collections.
The vertebra is No. 669. The writer is indebted to Dr. S. W. Williston for the
privilege of studying this material and other specimens that are used for compari-
son and also for his kind suggestions and corrections in the preparation of this
chapter.

Chapter IV.

DESCRIPTION OF A NEARLY COMPLETE SKELETON OF DIASPARACTUS

ZENOS CASE.

By E. C. Case and S. W. Williston.

The specimen here described was found by Case in a reddish, clayey sandstone
somewhat below the middle of the Permo-Carboniferous strata of El Cobre Canon,
in Rio Arriba County, New Mexico. When discovered, only a fragment of a tooth
was exposed, but excavation revealed nearly the entire skeleton. As the skull was,
unfortunately, near the surface, it is badly injured by weathering, and a portion
of the upper surface was destroyed before fossilization, but enough remains to
reveal by careful study most of the important characters.- The atlantal and axial
intercentra are preserved ; the atlas is lost ; the axis is preserved, but is separated
from the rest of the vertebral column. Beginning with the third cervical, there
is a series of five vertebras in connection with the scapulae, the clavicles, and the
interclavicle. The posterior portion of the scapula of the left side is destroyed,
but that of the right side, the clavicles of both sides, and the interclavicle, with
the exception of the tip of the distal end, are perfect. Following the break between
the sixth and the seventh vertebrae the series is complete to the twenty-third
caudal. Though no fit can be found between the sixth and the seventh vertebrae
the series was in position when excavated by Case and no reasonable doubt can
be entertained that the relations are as described. This is borne out by the
close relation which exists between the number of presacral vertebrae in this speci-
men and that found in other members of the same family, the Diadectidae. The
neural spines of many of the dorsal vertebrae and all but one of the caudals have
been destroyed. The pelvis and the limb bones of both sides have been preserved
in large part. The right ilium has lost the crest. All of the long bones of the limbs
are preserved, but the carpus and tarsus and the phalanges of both sides have been
more or less disturbed and many of the bones lost. Only a few of the ribs have
been preserved, and only the base of a single chevron. No trace of abdominal
ribs can be detected. This is in accord with the other specimens of this family
recovered, where no trace of such structures has been seen, though they would
certainly be expected.

In I QIC Case * described a short series of posterior dorsal and sacral vertebrae
(Am. Mus. Nat. Hist. Cope Coll. No. 4794) from New Mexico and proposed the
name Diasaparactus zenos to designate them. The posterior dorsal vertebrae of
this specimen agree so closely with the described type that it seems best to retain
the name for this specimen.

The weathered condition of portions of the specimen, the refractory nature of
the matrix, in places, and the softer substance of the bones has rendered the prep-
aration of the specimen very difficult, and it is to the painstaking care and skill
of the preparator, Mr. Paul Miller, of the University of Chicago, that we owe
our ability to describe the specimen so completely.

The skull: Lying close to the surface and being badly broken by weathering,
the skull is imperfectly preserved and its study has been exceptionally difficult,

•Bull. Am. Mus. Nat. Hist., vol. xxviii, art. xvii, p. 174, 1910.

17

1 8 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

but enough has been made out to render certain the points here described. The
top was destroyed previous to fossilization, so that it is impossible to give the
exact height, but the length and width are correct within a few millimeters. In
general the skull resembles that of the members of the family Diadectidae. It
may be best compared with the skulls of Diadectes lenius Marsh and Animasatirus
carinatus described by us in the American Journal of Science.* It resembles the
latter in the width of the posterior end, the narrowed occipital portion, and the
fact that the posterior ends of the quadrates and the tips of the squamosal or
tabulare bones are about on a line with the occipital condyle. It is also probable
that the quadrates had the same backward and inward inclination of the outer
faces as in Animasaurus rather than a position nearly parallel to the sides of the
skull as in Diadectes. Other comparisons with the two genera are given in the
description of the lower surface. The nares and orbits are similar in form and
position to those of the other members of the family ; the pineal foramen can not be
made out, owing to the destruction of the upper part of the skull, but certain indi-
cations point to the fact that it had the same relatively large size. The outlines
of the various bones of the skull can not be determined, owing to the condition
of the surface.

The maxillaries have the usual form, semi-crescentic, on the ventral surface,
with the concavity on the outer side. The alveolar surface is broad, accommo-
dating the laterally expanded roots of the cheek teeth. Fourteen teeth are indi-
cated in the maxillaries and premaxillaries, but as the last of the maxillary teeth
is still somewhat expanded laterally, though smaller than the others, it is probable
that there was a fifteenth tooth, small, conical, and almost rudimentary, as in
Diadectes. This is the usual number in the family. The teeth of the premaxil-
laries were elongated and probably had chisel-shaped cutting edges, as in Dia-
dectes, but as the incisor teeth of both the upper and lower jaws are badly weathered,
only the central portion is preserved and this point may be in doubt. Some of the
better-preserved ones seem to suggest a possibly conical form. The roots are very
elongate, at least twice as long as the crowns. There is a distinct radial arrange-
ment of the dentine around the pulp cavity, very clearly shown in several of the
teeth. In none of the cheek teeth is the crown preserved, so that it is impossible
to say more than that they were of the type common in the Diadectidae.

The posterior part of the skull is preserved, in part, on the right side, showing
the position and a part of the outer face of the quadrate. This bone was inclined
so strongly inward that it looks even more backward than outward, resembling
Animasaurus in this respect. The posterior edge of the quadrate is nearly flat
and there is no trace of a quadrate foramen, but this may be due to the condition
of the specimen. The prosquamosal reached down to the lower edge of the quad-
rate and then extended forward in a nearly horizontal line rather than rising some-
what abruptly, as in Diadectes phaseolinus.

The lower surface of the skull (fig. 8) shows the general form of the palate
common in the diadectids, but combines characters of both Diadectes and Ani-
masaurus.

Resemblances to A nimasaurus : The anterior end of the palate narrows rapidly,
due to the narrow muzzle; the prevomers and the anterior portion of the ptery-
goids are shown only by the broken edges, but it is apparent that they must have
met in the median line and in all probability formed a high median keel which
extended back to a small opening just anterior to the basisphenoid ; the posterior
end of the skull is broad across the quadrates, but narrowed in the occipital region ;

• Am. Jnl. Sc., vol. xxxiii, April 1912, p. 339.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

19

the quadrates are inclined backward and inward ; the lower ends of the quadrates
and the temporal or tabulare bones reach as far back, nearly, as the occipital condyle.

Resemblances to Diadectes lentus: The basisphenoid is broad and short
instead of narrow and elongate as in Animasaurus ; the articular faces of the quad-
rates are broad antero-posteriorly ; in Animasaurus they are narrow.

In this specimen no trace can be found of the palatine processes of the max-
illaries so characteristic of the family, but as the skull is poorly preserved and
has been badly crushed it seems to us more probable that the processes have been
injured beyond recognition than that they were absent.

The lower jaw: On the right side the articular portion is missing, but on the left
side it is the posterior portion of the jaw which is preserved, so that the length and
proportions of the jaws can be made out. It
was less deep than in Diadectes, but in other
respects very similar. There was a well-
defined coronoid process and the usual two
Meckelian openings on the inner side, the an-
terior one seemingly somewhat smaller than
in Diadectes. The anterior end of the two
jaws, in position, was narrower, correspond-
ing to the narrowed muzzle. The broken
condition of the specimen reveals the deep
sockets of the incisor teeth and the roots of
several teeth marked by the deep linear
grooves due to the folding of the dentine.
The articular surface of the articular bone
shows two deep concave faces, which, in op-
position to the corresponding prominences
on the lower face of the quadrate, limited
the motion of the jaws to a strictly verti-
cal plane. As in the skull, the outlines of
the individual bones can not be made out.

The shoulder-girdle: This, as is not uncommon in the specimens of the family,
is well preserved and almost undistorted ; this is due to its strength, the close artic-
idation of the bones, and probably also to the presence of a considerable quantity
of tough cartilage. The distal half of the right scapula, portions of the cleithra,
and the tip of the posterior end of the interclavicle are wanting. In comparison
with the scapula of Diadectes, the blade is shorter and the proximal portion wider.
The posterior edge is more sharply curved, forming a semicircle. The lower, or
inner, edge is straight in the middle, curving gently at the posterior end to join
the posterior edge of the bone and more sharply at the anterior end to join the
anterior edge. The anterior edge is nearly straight, with scarcely noticeable faces
for the clavicle and cleithrum. The posterior edge is crescentic in outline, becom-
ing more sharply curved toward the proximal end. On the outer face there is a
prominent preglenoid tuberosity anterior to and above the humeral cotylus; pos-
terior to it the edge of the bone is broad and bifurcates to include the supraglenoid
fossa with, apparently, a supraglenoid foramen at its bottom. On the other, ante-
rior, side of the tuberosity is a deep and wide opening, which terminates at the
bottom in the supracoracoid foramen. Just beneath the preglenoid tuberosity and
anterior to the articular surface is a very large, funnel-like foramen, the glenoid,
which is the largest foramen observed by us in this position in any of the cotylo-
saurs. The articular surface has the usual two facets, the upper, anterior, facing

F1G.8.-

Diasparactus zenos Case, X %. Restora-
tion of lower surface of skull.

30

PERMO-CARBONIPEROUS VERTEBRATES FROM NEW MEXICO.

obliquely downward and slightly backward and the lower, posterior, facing obliquely
upward and slightly forward. The inner surface of the scapula is hidden on both
sides in large part, rendering it impossible to describe the subscapular fossa. It is
impossible to trace any sutures between the bones, but both coracoids are surely
present.

The inter clavicle: This bone is very little expanded at the anterior end and is
strongly united by deep sutures with the clavicles, resembling very closely in this
respect the unnamed clavicles and interclavicle from Texas (No. 4390 Am. Mus.
Nat. Hist. Cope Coll.) described by Case in Publication 145 of the Carnegie Insti-
tution of Washington, page 79, figure 26. The posterior portion of the inter-
clavicle is nearly straight, gradually becoming narrower toward the posterior end,
which is lost.

Fig. 9. — Diasparaclus zenos Case, X >^. A, anterior view of clavicles and inter-
clavicle; B, upper view of right clavicle; C, outer surface of right scapula;
D, posterior view of right scapula.

The clavicles: These are very strong proximally and higher than the anterior
end of the interclavicle. They join the sides of the interclavicle but meet behind
it, so that there is a median depression or fossa between the proximal ends of the
bones anteriorly. Seen from in front the proximal ends of the clavicles are broader
than high, presenting a wide flat face to the anterior edge of the scapula. A few
centimeters back the clavicles become suddenly flattened, so that they are very
thin antero-posteriorly, in the natural position of the bones, but wide vertically;
they continue in contact with the scapula for their full length. The distal ends
of the bones become gradually contracted to narrow points.

The cleithrum: There is no trace of the cleithra attached to the scapulae on
either side, but there are two isolated fragments which appear to be portions of

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

21

the bones from the two sides. The posterior end of the bone was broad and thin
and evidently applied to the outer face of the posterior end of the scapula. Anteri-
orly it contracted to a narrow, relatively short process, which was applied to the
upper surface of the scapula or the distal end of the clavicle.

The humerus: The bone of the right side is perfect, but that of the left side
has lost a small fragment from the upper border of the entepicondylar foramen.
In general the form is the same as in Diadectes phaseolinus, but differs in some
particulars. The proximal and distal ends have about the same inclination to
each other as in D. phaseolinus and there is the same lack of any well-defined shaft.
The articular face for the scapula is narrow and extends nearly parallel to the
greatest length of the head of the humerus instead of obliquely across it, as in the
pelycosaurs and some Cotylosauria. Near the inner end of this face is a deep,
rather elongate pit, with a prominent lower border but open on the upper edge.
Beyond this pit the end of the bone falls away sharply to the inner, deltoid, pro-

Fic. 10. — Diasparactus tenos Case, X yi- A, anterior surface of left humerus;
B, posterior view of left humerus; C, view of inner side of left humerus.

cess, which extends almost directly outward and does not curve downward and
inward toward the anterior face of the bone, and does not send an extension down-
ward on the shaft, as in the pelycosaurs, nor even so much as in D. phaseolinus.
On the extremity of this process is a well-marked pit for the attachment of a strong
ligament. The process on the posterior side of the proximal end is fully as promi-
nent as that on the anterior end.

In D. phaseolinus there is a strong process on the anterior, radial, edge of
the shaft near its lower end, which extends almost directly outward, probably for
the attachment of the supinator muscle; a similar process occurs in the same
position in Seymouria, Limnoscelis, though small, and in Sphenacodon and Dimetro-
don, and has been called the ectepicondylar process. In the amphibians Eryops
megacephalus, Tremaiops, etc., there is a like process. In this genus there is
no such process; the shaft of the left humerus is perfectly smooth at the position
of the process, but on the right humerus there is a slightly elevated rugosity.
In D. phaseolinus the distal end is relatively broader, the entocondylar process

2 2 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

is longer and higher, and the foramen is smaller. The humerus of the specimen
of D. phaseolinus, with which this specimen was compared, is actually larger and
the foramen actually smaller, increasing the relative difference. In this specimen
the entocondylar process of the right humerus is slightly distorted, the upper edge
running obliquely downward from the shaft to the lower end; but on the left
humerus, which is undistorted, it passes out from the shaft more directly and there
was a long, somewhat oblique, inner edge to the process. On the lower comer
of this process there is a deep pit for the attachment of a ligament.

On the back of the bone directly opposite the articular surface for the radius
is the strong ectocondyle, similar to that on the humerus of D. phaseolinus. The
position and large size of the ectocondyle, directed prominently backward and
even inward from the plane of the distal end, seem to be correlated with a powerful,
thickset humerus, short fore-arms, short, flattened feet, and especially with non-
acuminate ungual phalanges. It is the condition found in Eryops and Trematops
among the amphibians, Ophiacodon among the pelycosaurs, Limnoscelis, Dia-
dectes, Diasparactus, and Seyntouria among the cotylosaurs. Williston found the
digits more slender in Seymouria, but no indications of acuminate claws have been
found in that genus. The great development of these processes for the attachment
of the extensor and supinator muscles on the outer side, of the flexor muscles on the
inner side, and the great mechanical advantage these muscles must have had, are
certainly correlated with a powerful manus; but the structure of the arms and
hands in the Captorhinida is like that of the more cursorial structure found in
other pelycosaurs and cotylosaurs, and indicate widely different habits for this
group of Cotylosauria.

The face for the radius is semicircular in form, with the lower, inner, border
truncated; the articular surface is flat or even concave, indicating the presence
of a large amount of cartilage and differing notably from the same face on the
humerus of the pelycosaurs, where it is strongly convex. As the face has the same
appearance on the bones from both sides it must be regarded as natural, but may
be merely ontogenetic in character. The face for the ulna is elongate and oblique
in position, passing from within outward and downward. It is largely confined to
the distal end of the bone, appearing only slightly on the lower, anterior, face, and
not at all on the upper, posterior.

The ulna resembles that of D. phaseolinus, but is proportionately longer and
more slender, without such heavy extremities and without the accompanying
rugosities. The shaft is wide but thin. The upper end has an oblique, nearly
fiat, slightly twisted, articular surface for the humerus; it was simply applied to
the corresponding face of that bone and did not embrace it, as in the pelycosaurs.
The lower end is a little expanded and presents two articular surfaces lying at
less than a right angle to each other and not separated by any non-articular space.
They are set at such an angle with the flat surface of the bone that they look more
inward than directly downward. The distal end of the shaft is curved inward
slightly at the lower part, increasing considerably the effect just mentioned. Viewed
from the posterior edge, the bone has a slight but very evident sigmoid outline.

The radius is shorter and heavier than the ulna; the shaft is triangular in
section, with the thicker portion on the ulnar side. The proximal end is expanded,
with a semicircular face truncated on the posterior side. The articular surface is
gently concave; this concave face opposed to the concave face on the humerus
indicates the presence of a large amount of cartilage in the joint. The distal end
is elongate laterally, but rather narrow antero-posteriorly. There are two faces;
a short inner face, which, in connection with the oblique face on the inner side of

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

23

the distal end of the ulna, forms a notch into which the intermedium fitted, and
a larger outer face, nearly horizontal and occupying most of the distal end.

The carpus: In no specimens of the Diadectidas previously collected have
more than a few of the bones of the carpus been preserved and in none are they
in position. In this specimen most, if not all, of the carpal elements of the right
side are in connection with the radius and ulna and some of the metacarpal bones,
though somewhat disturbed by the bending of the foot backward and inward.
The carpus of the left side is separated from the bones of the fore-arm and not all
are preserved. Seven bones are clearly preserved on the right side and there are
indications of two others. The bones of the proximal row and a centrale are large
and well formed, but those of the distal row are reduced to small nodules and were

Fig. II. — Diasparacltts zenos Case, X K- A, anterior view of left ulna; B, inner view of left ulna; C, ante-
rior view of left radius; D, upper end of left radius; E, lower end of left radius; F, lower ends of right
radius and ulna with carpals of proximal row, upper surface: G, carpals of distal row of right side with
attached metacarpals, upper surface. This is the distal portion of the carpus shown in F, which is bent
back upon itself; H, metacarpal and phalanges of third digit of right front foot.

evidently inclosed in a considerable amount of cartilage, resembling in this respect
the foot of Limnoscelis, described by Williston.* The ulnare is roughly oval in
shape, with broad articular faces for the intermedium, the ulna, and carpale IV.
The outer edge is narrowed above, but widens below to form a triangular articular
surface which probably afforded attachment to a pisiform, which is not preserved,
however, in either carpus. The intermedium is thick and oval in outline ; it articu-
lated fairly closely in the notch formed by the adjacent faces of the radius and
ulna, and touched the radiale, ulnare, and the centrale. The radiate is smaller,
rather longer than wide, and lay along the larger, inner, facet of the radius. As
in Limnoscelis, it was not large enough to articulate with this face for its whole
length. The centrale is elongate with a flat lower face; the upper face is flat on the
radial side, but near the ulnar end there is a sharp upward projection which fitted
between the ulnare and the intermedium. On the radial end of this bone there is
a small fragment partly inclosed in matrix which may indicate the second carpale.
The bones of the distal row are all small and of indefinite form. One, of larger
size, probably carpale IV, lies below the ulnare, a second below the ulnar end of

* Am. Jnl. Sc., vol. xxxi, May 191 1, p. 390.

24

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

/;

'"^(^CE?

■m^tK—J'

the centrale. None appears in the specimen directly beneath the centrale. Below
the radiale is an uncertain fragment which appears to be the proximal end of the
metacarpal with a first phalange attached to its lower end. On the inner side of
the proximal end of this fragment is a small nodule which is apparently carpale I.
The distal portion of the foot was very broad and strong, with heavy, short
phalanges. As neither of the fore feet is complete, it is impossible to give the
phalangeal formula, but it is very probable that it was the usual reptilian number,
2. 3. 4i Si 3- The metacarpals have very broad proximal and distal extremities,
which narrow rapidly to a shaft not over half so wide as the ends. Viewed from the
sides the extremities are wider than the shaft, but the difference is not nearly so
great as in the horizontal view. The articular surfaces are flat. The phalanges
are broad, short, and thin, with only a slight constriction of the sides to indicate a

shaft. The articulations are better formed than
in the metacarpals, as there are well-modeled
concave and convex surfaces indicating a consid-
erable freedom of motion. The shape of the ar-
ticulations between the phalanges indicates the
possibility of closing the hand, but also a con-
siderable resistance to any bending back of the
digits. The terminal phalanges are broad and
oval; the proximal articulation with the adjacent
phalange was close. The edges of the bone are
thin, but there is a distinct thickening on the
under side near the distal end and a distinct con-
cavity of the lateral edges, indicating the attach-
ment of a strong flattened claw or nail-like claw.
The interpretation of this type of limbs and
feet, so characteristic of some of the contemporary
amphibians and most of the contemporary coty-
losaurs, with the exception of the Captorhinidae,
has been in dispute. Abel accepts them as fosso-
rial. Matthew, in a review of Abel's Paleobiologie
(Science, March i, 191 2), expresses the opinion
that the animals were aquatic in habit. From
this conclusion we must enter our dissent. There is not a character shown in the
Limnoscelidae, Diadectidae, or Seymouriidae that could lend support to this hypothesis
unless it be the imperfect ossification of the mesopodium. The structure of the
body throughout is heavy; the tail is short and without propelling power, as indi-
cated by the short spines and chevrons. The fingers are short and stumpy, with
well-articulated phalanges and with the terminal ones clean-cut and provided with
heavy nails or blunt claws, all of which would be out of place if their main func-
tion was the support of a swimming web. In most, if not all, aquatic reptiles and
amphibians the foot is elongate with shapeless phalanges, not well articulated.
Furthermore, it is quite certain from the natural articulations that the elbow was
normally bent at an angle and it is very doubtful whether the fore-arm would
have been extended fully in line with the humerus. Nor can we now believe, in
opposition to views previously expressed by Case, that these animals were essen-
tially fossorial in habit, since the short, though powerful, legs (in some forms at
least) were not long enough to extend the hands in front of the head. It seems
more probable that these animals; as well as the Eryopidae, were marsh-dwelling
creatures, doubtless living among the rank vegetation in the vicinity of water.

Fig. 12. — Diasparactus zenos Case. Res-
toration of right front foot. X yi.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

25

which they occasionally entered, but using the feet in digging and excavating for
food. Invariably animals using the feet as propelling paddles have the propodials
more or less elongated, while in tail-propelling animals not only the propodials
but the epipodials are shortened and the digits elongated. In none of these groups
will the cotylosaurs enter. Upon the whole it seems to us that the habits of these
animals may have most nearly approximated the present habits of the marsh
turtles.

The pelvis is very nearly perfect. The crest of the ilium of the left side is
broken away and the median portion of the pubis is slightly injured, otherwise
the bones of both sides are complete. As in the rest of the skeleton the nature
of the bones makes it impossible to trace the sutures. The crest of the ilium is
elongate antero-posteriorly, with the posterior end somewhat higher than the
anterior, so that the upper line slants downward and forward. The upper edge is
thin, but 2 or 3 centimeters below the crest there is a sudden thickening, forming
a distinct shelf with a nearly horizontal upper surface. The posterior end is formed

Fig. 13. — Diasparaclus zenos Case, X>^. A, right side of pelvis; B, lower view of pelvis; C, anterior
view of pelvis; face of sacral vertebra somewhat restored.

by a thickened process, about a centimeter wide, which is the direct continuation
of the shelf mentioned. Below the shelf the ilium contracts rapidly until it is
less than half as wide as the crest. The acetabulum is of the form common in the
Cotylosauria, rather shallow and longer than high. The anterior edge is not well
defined, but posteriorly there is an elevation of the rim on the proximal portion
of the ischium forming a buttress for the head of the femur. Above, there is the
usual tuberosity on the iliimi at about the middle of the upper line of the cavity.
The pubis is shorter than the ischium. The anterior edge, opposite the ace-
tabulum, is convex, but toward the lower end the edge becomes wider and the bone
is turned so that it is nearly horizontal. The anterior end is thickened with an
elongate, flat, oval face indicative of the presence of a prepubic bone as suggested
by Broili. The lower portion of the bone is nearly flat (fig. 13 b), but towards the
median line it is turned sharply down and, in union with the bone of the opposite side,
forms a deep symphysis indicated on the outer surface by a prominent keel. The
pubic foramen lies just below the anterior end of the acetabulum, it penetrates
the bone directly, passing slightly upward and forward to open on the flat anterior
face described below. Viewed from the front, the cavity of the pelvis is slightly

26

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

oval in outline, a little higher than wide, and the pubes present broad faces tri-
angular in outline and slanting inward, downward, and forward, but the anterior
ends of the bones are horizontal (fig. 13, c), so that at the base of the broad faces
there is a flat shelf, perhaps slightly indented at the median line where the two
bones meet.

The ischium is longer than the pubis, extending posterior to the acetabulum
almost twice as far as that bone does anterior to it. The upper margin begins
near the upper posterior edge of the acetabulum and runs backward and down-
ward in a concave line. The posterior part of the margin turns downward
and joins the posterior border in a sharp angle; this border is nearly straight and
passes sharply inward, downward, and a little forward. The union of the bones
of the two sides leaves a notch at the posterior end of the pelvis. As in the pubis,
the main portion of the bones is horizontal below the acetabulum, but the inner

Fig. 14. — Diasparactus zenos Case, X yi. A, posterior view of right femur;
B, posterior view of left femur; C, anterior view of left femur.

edges turn down sharply and meet in a median symphysis, which finds expression
on the lower surface in a sharp keel, continuous with that formed by the pubes.
With the exception of the prepubic faces, never seen by us before but described
by Broili in a specimen of Diadectes from Texas, now in Munich, the whole pelvis
differs very little from those of the specimens of Diadectes already known and
closely resembles that of Limnoscelis and Seymouria.

We are unable to explain the meaning of the peculiar shelf on the outer side
of the crest of the ilium. In Diadectes the upper ends of the neural spines, in
the dorsal region, are broad and rugose and, in some cases, even expanded; this
has led Case to make the suggestion that there might have been an imperfect
dorsal armor, such as occurs in Pareiasaurus, and that the shelf may have been the
place of contact with the lower end of a broader plate, or plates, covering the pelvic
region. No trace of any dorsal armor has been discovered in any specimen of the
family. In Diasparactus the dorsal spines, as described below, are higher than
in any other known genus of the family and there is no distal expansion or rugosity
of the spines ; the shelf is fully as well developed, however, as in the genus Diadectes.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

27

The femur is very similar to that of D. phaseolinus, but rather more slender.
The head is convex dorsally, but with a deep trochanteric cavity on the ventral
face. The articular face is crescentic, confined to the end of the bone and somewhat
wider at the inner end. On the outer side of the proximal end is a strong process
incHned inward, on the bone of the right side, but this is probably due to crushing
and the process normally stood directly out from the body of the bone as in Dia-
dectes. The most prominent part of this process is just above the middle of the
shaft, and a short ridge extends downward to the middle of the ventral face of the
shaft ; this ridge can be traced as a low prominence obliquely across the face of the
shaft to the outer angle of the lower end. The shaft is proportionately longer than
in D. phaseolinus, and oval in section, but the outline is marked by the prominence
just described and the narrower radial border. The lower end has the form com-
mon in the pelycosaurs and cotylosaurs; on the inner side is a semicircular face
for the tibia and on the outer a more elongate face (antero-posteriorly) for the

Fig. is.—Diasparactus zenos Case, X yi. A, upper surface of right tibia;
B, anterior view of right tibia; C, inner view of right tibia; D, inner view
of right fibula.

fibula. The face for the tibia is semicircular in outline and is inclined downward
and backward; it is nearly cut off from the fibular face by deep grooves on both
the anterior and posterior sides of the bone. Just above the fibular face is a deep
pit on the outer side.

In a specimen of D. phaseolinus, No. 4684 American Museum, the articular
faces of the distal end of the femur are concave with raised edges, indicating the
presence of a considerable mass of cartilage; this condition, however, seems to be
peculiar to the particular specimen, and as it occurs on many of the articular sur-
faces may be due to exceptional conditions of fossilization or to the age of the
animal — especially as a similar condition does not occur in other specimens identi-
fied as belonging to the same species. In Diasparacius the articular faces are well
rounded and better formed, resembling the condition found in the Pelycosaurs
and indicating a more perfect joint. Compared with the femur associated with
the type skull of Chilonyx, the femur of Diasparactus is less rugose, shorter, and the
fibular face does not descend so far below the rest of the distal end of the bone.
The suggestions of similarity between Chilonyx and Diasparactus indicated by the
position of the quadrates is not borne out by the femora.

The tibia is short and heavy, but this is not carried to the extreme seen in
Diadectes or Diadectoides. The proximal face is crescentic, due to the deep incision

3

28

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

on the anterior face of the upper end. The outer half of the face is longer and
broader than the inner, which extends rather farther forward and is inclined
downward, forward, and slightly inward. The two halves of the face are much
more unequal than in Diadectes. The heavy proximal end contracts rapidly to a
relatively slender shaft, triangular in section. The distal end is expanded antero-
posteriorly, but is little wider than the shaft. The face is semilunar, due to a shal-
low concavity on the inner side of the distal portion of the shaft. The articular
surface is convex, presenting a posterior portion which looks almost directly down-
ward and an anterior, inclined upwardly and
inwardly; these two portions are not divided
into separate faces.

The fibula is decidedly different in form
from the same bone in D. phaseolinus. In that
animal the upper end of the bone is nearly cir-
cular in outline, joining the shaft below with a
sharp change in outline. The lower part of the
bone widens gradually and the distal articular
surface is nearly flat (see fig. 30, p. 82, Publica-
tion 145, Cam. Inst. Washington). In Diaspa-
ractus the head is formed by a gradual widening
of the shaft and is very little thickened. The
outer edge of the whole bone is straight, nearly
to the distal end, where it turns sHghtly out-
ward; the inner border is quite concave, the
curve becoming sharper toward the distal end.
The distal end is considerably wider than the
proximal. There are two articular faces; one
larger, nearly horizontal, occupying nearly the
whole of the distal surface, and a smaller, inner
The whole bone is very thin, with slight con-
cavities on the inner face just proximal to the edges.

The tarsus of the left side is nearly complete and there is very little disturb-
ance; that of the right is less perfectly preserved. On the right side the tibiale
and fibulare and two bones of the distal row are in connection, but the tibiale is
badly injured. The tarsus, like the carpus, must have contained a considerable
amount of cartilage, for tarsalia I, II, III are mere nodules, while the proximal
ends of the metatarsals are broad and heavy. The tibiale is much nearer in form
to that of Labidosaurus and the pelycosaurs than to that of Litnnoscelis. Williston
describes the tibiale of Limnoscelis as a nearly cuboid bone which in his figure
occupies only the inner half of the distal articular face of the tibia. He regarded
it as either a tibiale or a tibiale + intermedium, but is inclined to believe that only
two bones may appear in the tarsus of the reptiles. In Diasparactus the tibiale
is rather elongate and, assuming the position in which the bone was found to be
correct, is roughly right-angled when seen from in front. The inner face (fig.
17 A, a), which was in contact with the fibulare as the bones lay, is wide and tri-
angular in outline ; on the lower face of the horizontal portion there is a deep groove,
wider and deeper above and becoming shallower toward the lower edge. On the
lower part of the inner face there is another groove (fig. 17 a, b). The lower end
has a large, nearly square face, looking straight downward. The upper part of
the horizontal portion carries no articular face, but on the outer vertical edge
there are two rough faces, probably articular. The larger looks outward, upward,

Fig. 16. — Diasparactus zenos Case, X 1/2.
Upper surface of left hind foot.

one, inclined upward and inward.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

29

17. — A, Diasparactus zenos Case, left tibiale from the
front, X }^; B, left tibiale of same from the rear, X H;
C, carpus of an unidentified species of Diadectes, in the
collection of the University of Chicago: /, tibia; /, fibula;
a, astragalus (tibiale); c, calcaneum (fibulare).

and slightly forward; the posterior, smaller face looks directly outward. If the
bone is viewed from the inner side its outline is quite suggestive of the tibiale of
the Cotylosauria, but it seems that this is more of a coincidence than otherwi.se
and that the true position of the bone is different from that in the order mentioned
and that the faces articulate in a different manner. Allowing the bone to remain
in the horizontal position in which it was found, which seems quite correct, it
lies nearly across the tarsus, its inner end taking a small part in the articulation
with the fibula and its large, outer, oblique face accommodating the tibia.

The fibulare is oval, with a broad face on the upper side for the fibula and
a narrower one on the inner side
for the tibiale. The lower and
outer edges are thinner, but still
quite broad. The distal row con-
sists of four bones. Tarsalia I,
II, III are small, oval, flattened
nodules with no articular surfaces
and indicate the presence of a con-
siderable mass of cartilage in the
foot. Tarsale IV is larger, roughly
triangular, with broad edges and a
deep concavity on the lower sur-
face. There are no traces of cen-
trale bones. The attitude of the
tarsus in this and related fonns was far different from that in the Pelycosauria or
in the modem Lacertilia; the foot was not set obliquely to the axis of the foreleg,
but was continued directly forward.

The phalanges are similar to those described from the front foot. The meta-
tarsal of the first digit is short and very wide, that of the fourth digit is longer than
any in the front foot, and the whole digit was also, apparently, longer. The first
phalange of this digit is short, but much wider and heavier than the metatarsal.

The vertebral column: The condition of the vertebral column has been de-
scribed in the general account of the specimen. The atlas is wanting; in only one
specimen of the Diadectidae (No. 1075 Univ. Chicago), the nearest related group,
has this bone been observed. It is a fiat disk, perforated by the notochordal canal,
with detached neural arches. The arches were weak, with small and weak posterior
zygapophyses for attachment with the axis. It is not known whether they were
divided, as in the Stegocephalia (young stages of Eryops and Trimerorhachis) , but
in all probability they were. The atlantal intercentrum is preserved in the speci-
men of Diasparactus; it is thick and heavy, with the anterior articular face for the
occipital condyle wider than that for the axis.

The axis: This is not in connection with the rest of the vertebral column or
the skull, but is identified by the neural spine; this is thin and elongated antero-
posteriorly, with the posterior edge nearly straight and the anterior edge slanting
downward and forward to project over the atlas, quite as in Dimetrodon and Dia-
dectes. The sides of the neural arches are not expanded and convex as in the
dorsal vertebrae. There is a short transverse process for the attachment of a well-
formed rib. The centrum is small, with nearly circular faces; the sides are smooth,
somewhat contracted, and meeting in a sharp median keel below. The lower edge
of the anterior face is beveled by an articular face for the axial intercentrum, which
is nearly as large as the first intercentrum.

3© PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

The third vertebra * has lost the neural spine, but judging from the base the
spine was broadly diamond-shaped in section with the anterior and posterior edges
somewhat extended as in the succeeding vertebrae of the presacral series. The
neural arch is slightly convex, beginning to assume the form of the neural arch in
all the Cotylosauria, but is still much narrower than those of the dorsal series. The
anterior zygapophyses are injured, but were evidently of good size. The neural
canal is, as the specimen is prepared, of relatively enormous size, having a diameter
greater than that of the centrum ; this is apparently natural, but if so is a most sur-
prising feature. The sides and the bottom of the centrum are concealed by the
matrix, but it is apparent that the centrum retains the small size seen in the axis
and that there was a large intercentrum. It is a notable fact that throughout the
presacral series the accommodation for the intercentrum is largely made by the
beveling of the anterior face of each centrum and that there is a much smaller face
on the posterior end. The transverse process starts from the side of the anterior
zygapophysis near the anterior end and runs downward and forward nearly or quite
to the anterior face of the centrum at the lower edge; this is somewhat obscured in
the third vertebra, but is very apparent in those immediately following. The upper

end of the transverse process stands well out beyond
a line connecting the outer edges of the anterior and
posterior zygapophyses; in this and the rest of the
cervicals the characteristic shortening of the trans-
verse process does not appear; it begins at the first
of the dorsal vertebrae.

The fourth to the seventh vertebrce have lost the
neural spines, but the bases show that they were all
rather slender, with the diamond-shaped section de-
scribed above. The neural arches become gradually
Fig. i8.— Upper view of third, fourth, wider and rounder until on the seventh they have
fifth, and sixth cervical vertebra; assumed fully the form characteristic of the order.

ol Dmsparaclus zenos,y. }i. ^. i , . , . ., ^,

The centra of the sixth and seventh are rather more
elongate than the anterior ones, somewhat narrowed on the bottom line, but with-
out a distinct keel. The intercentrum between the sixth and the seventh is narrow
and shallow, not rising much above the lower edge of the centrum ; this is the form
observed in all of the succeeding portion of the column except the immediate pre-
sacrals. The transverse processes are still prominent, but decreasingly so, and in
each vertebra they originate a little farther back until in the seventh the upper
end rises from a point a little anterior to the middle of the neural arch. The
articular face for the rib is flat and runs downward and forward, touching the
anterior edge of the centrum near the bottom line, so that if the vertebra is viewed
from the lower surface the anterior end is considerably wider than the posterior.

There is a break following the seventh vertebra, but as the next vertebra is
of the correct size and form, and especially as the vertebrae were in position as they
lay in the ground, it may safely be assumed that the next in series is the eighth and
that the lack of a fit is due to the loss of some small amount of matrix, either in
the collection or preparation of the specimen. From the eighth dorsal to the twenty-
third caudal, the forty-seventh of the complete series, the contact is preserved or,
in the caudal series, recorded from the matrix in cleaning.

The eighth vertebra is imperfect, having lost the spine and a part of the left side.
The neural arch is low and convex, with considerable breadth across the zyga-

• Williston believes that this may be the fourth or fifth. Case believes, from comparison with Dia-
dectes, that it is the third.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 31

pophyses, laterally. The outer edge of the upper end of the transverse process is
just on a line with the outer edges of the anterior and posterior zygapophyses.
The articular face runs down to the anterior edge of the centrum.

The ninth vertebra is the first with a complete neural spine ; this is oval in sec-
tion throughout, with the anterior and posterior edges slightly contracted; the
upper end is not rugose nor expanded; it terminates in a flat, smooth face.

From the tenth to the sixteenth vertebra there is little or no change in form.
The maximum width of the neural arches is attained in the tenth and retained to
the fifteenth. The upper ends of the transverse processes do not reach beyond
the line of the zygapophyses, and beyond the tenth they shorten very rapidly,
until in the last of this series they form only very slight prominences on the sides
of the centra. The neural spine of the tenth is similar to that of the ninth, but
does not exhibit a slight anterior curvature visible in the anterior one; this curv-
ature may be due to accidents in fossilization or may be the last remnant of a

Fig. 19. — Diasparaclus zenos Case. Lateral view of the last eleven presacral vertebrx, X yi-

decided forward curvature of the spines in the cervical region. On the posterior
face of the thirteenth the broken surface shows the edges of distinct hyposphene
and hypantrum articulations. The posterior edges of the neural arches of the last
vertebra in this series are nearly straight vertically ; they descend almost directly
from the base of the neural spines and do not overlap the anterior portion of the
succeeding vertebrae, as in Diadectes. The centra are short, without keels on the
mid-line, but are somewhat narrowed below.

The seventeenth vertebra is slightly displaced from the sixteenth, but is held in
place by matrix. It shows a very decided change in form from the preceding one.
The neural spine is a little higher and the neural arch narrower, with steeply sloping
sides. The transverse process is here and in the succeeding vertebra2 identical in
form and condition with those of the type specimen of Diasparactus zenos. No. 4794
Am. Mus. Nat. Hist. Cope Coll. The portion of the original description applicable
is here repeated.*

" This new genus and species of the family Diadectidas is characterized by the small
size of the centra compared to the height and spread of the neural arches, and the short
transverse processes. The whole vertebra is relatively very thin antero-posteriorly, so that
while it has the general form of all members of the family, it looks much higher and wider
and the small centrum gives it. something of a high-shouldered kite-shape, when viewed from
the front or rear. The transverse processes are exceedingly short, never extending out
beyond the edges of the zygapophyses and in most cases not reaching so far."

* Case, Bull. Am. Mus. Nat. Hist. vol. xvii, p. 174, 1910.

32 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

The remaining presacrals are similar in form. The neural spines become a
little heavier and shorter, the neural arches are narrower, the transverse processes
become shorter by reduction of the lower end, not reaching to the anterior end of
the centrum, and the position of the neural arches changes so that the posterior
zygapophyses hardly reach beyond the posterior edge of the centrum, while the
anterior ones reach far forward, almost to the middle of the preceding centrum.
In the dorsal series the condition is almost exactly the reverse of this last-mentioned
point. Between the last two presacral vertebrae the intercentra are very broad
antero-posteriorly, but still quite shallow.

The sacral vertebrce are two in niunber; both have lost the neural spines. The
first has a very short and narrow neural arch with well-developed anterior but
weak posterior zygapophyses. The sacral rib of the first is very heavy; its proxi-
mal attachment is from the anterior end of the anterior zygapophysis to beyond
the middle of the neural arch. The body of the rib extends nearly straight down-
ward, but has a sUght posterior inclination. The lower end is concealed in large
part by the ilium, but it can be seen that it is a broad, thin plate with a wide pos-
terior extension, which was applied to the inner side of the crest of the ilium.
The whole vertebra is set rather high, so that the crest of the ilium does not rise
greatly above the middle of the neural arch.

The second sacral vertebra resembles the first in most regards, but the ribs are
very much smaller and more slender. Starting with a strong attachment to the
sides of the neural arch and the centra, they rapidly dwindle to a very slender
process without any distal expansion. The lower part extends downward and then
turns backward, just touching the inner side of the ilium. These ribs take very
little part in the support of the pelvis and resemble the ribs of the anterior caudal
vertebrae very closely. It might be said that there is but a single sacral vertebra,
so small is the share of the second sacral vertebra in the support of the ilium. The
sides and bottoms of the two vertebrae are hidden by matrix, but as they are slightly
separated in the specimen it may be said with certainty that they were not closely
united in hfe.

The anterior caudal vertebra closely resembles the second sacral. The spine
is the only one of the caudal series which is preserved. It is short and sharply
curved to the rear, so that the flat distal end looks directly to the rear. The ribs
have strong triangular heads firmly attached to the centra and occupying most
of the side; they are rather elongate and, like the rib of the second sacral, extend
downward and outward, but turn sharply to the rear at the distal end. The lower
face of the centrum is wide, slightly concave antero-posteriorly, and gently convex
from side to side. There is no trace of a keel.

The second to the sixth caudals have the same form as the first, becoming
gradually smaller. The ribs maintain the same form, but rapidly shorten. The
lower surface of the first five is concealed by matrix; the sixth and seventh are
poorly preserved and as yet not freed from the matrix. It is uncertain at what
point the chevrons began.

The eighth caudal has a relatively small neural arch set well forward on the
centrum. The spine was slender and inclined slightly forward, but, in common
with all the caudals, except the first, has been destroyed. The base of the centrum
is shortened by the development of a large face on the anterior end for the accom-
modation of the head of a chevron bone. The posterior end is beveled by a similar
but smaller face. Between the eighth and the ninth centra is the base of a badly
injured chevron; this is the only one of the series of chevrons that has been pre-
served and it is so badly injured that little more than its presence can be noted.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

33

Fig. 20.— Diasparactus zenos Case. Lateral
view of the ninth to the twenty-third cau-
dal vertebrae, X 'A.

From this point back the character of the vertebrae changes regularly and very
slightly. The last trace of a rib is seen on the eleventh or twelfth. The sixteenth
and seventeenth still have well-formed neural arches located on the anterior half
of the centrum; the posterior zygapophyses are elongated. From this point back
to the twenty-third, the last preserved, the neural spines are low, almost rudi-
mentary, and the zygapophyses are elongated, interlocking strongly at first, but
becoming reduced in the posterior part of the tail. In only the last few vertebra
is there any tendency for the centra of the vertebrae to become at all elongate,
and even here it is very slight. There is no indication of any great length of tail.
The ribs: Anterior to the first vertebra in series, reckoned as the third cervi-
cal, there is a well-developed rib, of moderate
length, lying in the matrix just within the edge
of the scapula. This is the rib of the axis; it
has a wide, undivided head and a slender shaft
terminating in a point. The ribs of the third,
fourth, and fifth vertebrae are in general form
similar to the ribs of the same vertebrae in
D. phaseolinus, but as they are injured on the
left side and covered by the scapula on the
right, the exact form can not be given. The
rib of the third vertebra (fig. 21 a) has an
elongate, undivided proximal end; beyond this it contracts to a narrow shaft; the
anterior edge is straight to the distal end, but the posterior is extended backward

in a sharp point almost at right angles to shaft for a centi-
meter or more and then obliquely forward to the distal end ;
this gives a narrow, triangular form to the rib, resembling
the same rib in D. phaseolinus, but longer. The shape of
the next two ribs is less certain, but they are larger than
the third and were evidently of the same triangular form.
The fifth is the largest of the series. The sixth rib is nor-
mal in form, with an undivided head. There is no trace of
the plates which overlie the sixth, seventh, and eighth ribs
in Diadectes. Beyond the sixth the ribs are not preserved
in place until the fifteenth and sixteenth are reached. On
the left side of the series from the eighth to the fifteenth
are attached the imperfect shafts of five ribs, but neither
the proximal nor the distal ends are preserved. The shafts, as preserved, are rather
more slender than in D. phaseolinus. On the left side of the fifteenth (fig. 2 1 b) and
sixteenth the ribs are retained in place, but are bent back parallel to the column.
The heads are rather short and are undivided; the shaft of the rib stood almost
directly out from the vertebrae, with little downward curvature.
There is no trace of any abdominal ribs.

THE RESTORATION OF DIASPARACTUS ZENOS.

In the restoration presented (fig. 22) we believe that there is very little
that can be questioned. The skull, though badly shattered, presented nearly all
the characters, and its resemblance to both Diadectes and Animasaurus is so close
that we feel justified in all the points which we have recorded. Neural spines are
present both in sufficient number and in sufficiently diverse positions to render
certain their general characters throughout the column. The atlas is restored
in toto from Diadectes, but the presence of the atlantal and axial intercentra gives

Fig. 21. — Diasparactus zenos
Case.XK. A, rib of third
cervical vertebra from
left side; B, head of rib
attached to right side of
fifteenth vertebra.

34 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

an idea of the character of the lower surface, and the similarity of the axis to
Diadectes renders it very probable that the atlas was of the form shown. Ten
vertebrae have been added to the extremity of the tail ; this may be wrong by one
or two either way, but scarcely more. . Williston thinks the tail to have been rather
longer than does Case. That chevrons were present is shown by the base of one,
and the form of these bones varies so little, in a general way, that it is safe to con-
clude that they were nearly as represented ; they may have been a little longer and
may have had a slightly different inclination. , The phalangeal formula was very
certainly the usual one, 2,3,4,5, 3-4 ; this is shown by the presence of digits on one
or another foot carrying 2,3,4, and 5 phalanges. We have assumed that the posi-
•tion of the tarsus, as it lay in the ground, was correct and little if any disturbed;
it seems very satisfactory as compared with other forms. The presence of two
centrale in the tarsus is not certain and is improbable. For all other points there
is full evidence from the .skeleton.

That Diasparactus was a member of the family Diadectidae there can be no
doubt. The general form of the body is strikingly similar and the individual

Fig. 22. — Restoration of Diasparactus zenos. About yi natural size.

bones show differences that can not be reckoned greater than generic. The skull
is closer to that of Animasaurus than Diadectes, but possesses characters of both.
The development of the three broad ribs in the cervical region is another close
resemblance to Diadectes, which we doubt not will be found in Animasaurus when
the skeleton of that genus is discovered. The absence of the plates over the sixth,
seventh, and eighth ribs may be natural or may be due to accident. The presence
of a well-developed rib on the axis is a new but not unexpected feature. The shoulder
and pelvic girdles and the limbs are strikingly similar to those of Diadectes; the
most apparent differences are in the vertebrae. The high and more slender neural
spines are distinctly indicative of habits somewhat different from those of Dia-
dectes; if the idea be retained that the Diadectidae were in process of developing
a dorsal armor similar to that of Pareiasaurus, which was perhaps realized to some
extent in Diadectes, then Diasparactus may be regarded as representing an early
stage in the process.

The animal had gone far toward attaining the strongly articulated vertebral
column forming the axis of what must have been a very stiff and rigid body, so
evident in the members of the family, but had perhaps not yet abandoned freedom
of motion in the dorsal series so extensively as had Diadectes: this is perhaps borne
out by the less powerful sacrum ; in Diadectes the two vertebras support the pelvis.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

35

in Diasparactus the second plays little part, as indicated above. For the reasons
cited in the description of the foot, we are of the opinion that the Diadectidae were
not aquatic animals, but marsh-dwellers; we believe these animals to have been
very sluggish, harmless eaters of vegetation or small invertebrates, and to have
been able to present no more than a passive defense against the attacks of the
larger carnivorous forms, pelycosaurs, or forms yet unknown. Such habits would
inevitably lead to concealment and to the persistence of any forms favored by the
development of some form of protection, as an armor. We know that the period
in which these animals lived was one of those, frequently repeated in the history
of the world, in which there was a keen contest between attack and defense, the
development of a peculiarly efficient type of predatory form (in this case the
pelycosaurs of the genus Dimetrodon) was paralleled by the development of armor
on the weaker forms (in this case both amphibians and reptiles) . It is very probable
that Diadectes followed the general trend. The lateral position of the orbits does
not indicate habits such as those of the Stegocephalia, in which the orbits are near
the surface of the skull and look largely upward. It is possible, but only offered
as a suggestion, that the large size of the orbit may indicate a large eye and so
nocturnal or crepuscular habits.

Measurements.

Total length of the animal, as restored 1.35

Length of 10 vertebrae added to complete the

tail 105

Length of skull, as restored 167

Width of skull, across quadrates, as restored . . .154
Length of arc of slightly flattened right clavicle .159

Length of arc of right scapula 180

Width of anterior end of same 128

Extreme length right humerus 151

Width proximal end of same 084

Width distal end of same, slightly distorted . . . .094

Width distal end left hiunerus 081

Length right ulna 1 18

Length right radius 084

Length complete third digit of manus, left side .073

Length metacarpal of same digit 026

Width proximal end of same 024

Height of ninth vertebra 117

Breadth across posterior zygapophyses of same .074
J^ength across anterior-posterior zygapophyses.

same 039

Length bottom line of centrum, same 0.023

Height of fifteenth vertebra 134

Breadth across posterior zygapophyses of same .084
Length across anterior-posterior zygapophyses

of same 038

Length of bottom line of centrum of same 029

Length of bottom line of pelvis 166

Height of pelvis to top of ilium, a little ex-
aggerated by pressure 016

Length of crest of right ilium 1 16

Breadth across center of acetabulum 092

Length of right femur 146

Length of right tibia 102

Length of right fibula 118

Approximate breadth of tarsus 082

Length of fourth digit of pes, right side, minus

terminal phalange 086

Length of metatarsal of same 037

Length of first digit of pes, left side 045

Breadth of proximal end of metatarsal of same .023

CHAPTER V.

DESCRIPTION OF A NEARLY COMPLETE SKELETON OF OPHIACODON

MARSH.

By S. W. Wn-LISTON and E. C. Case.

The genus Ophiacodon was proposed in 1878 by the late Professor Marsh, to
include two new species of reptiles based upon fragmentary and disconnected
material collected by the late David A. Baldwin from a bone-bed near Poleo Creek
in Rio Arriba County, New Mexico. No distinctive characters were given for
the genus, nor was it possible to give such, since the two species Professor Marsh
described, O. mirus and O. grandis, represent not only two distinct genera, but
two different orders and classes as well ; the second species belongs to the temno-
spondylous genus Eryops, or an allied form. The characters given in the brief
description of the genotype were so meager that it was impossible to identify the
form without comparison of the type. The genus therefore remained doubtful till
recently, when Williston examined the type specimens in the collections of Yale
University. The genotype, O. mirus, was described by Marsh as follows:

"A third genus of reptiles allied to the last described [Sphenacodon] is indicated by
various well-preserved remains from the same locality. The teeth are all carnivorous in
type, conical in form and all are similar. Those in the anterior part of the jaws are recurved,
and in general shape resemble those of serpents. The rami of the lower jaws are united by
cartilage. The vertebrae are veiy deeply biconcave, and even perforate, and the inter-
central bones are large. In the present species the teeth are nearly smooth, and somewhat
compressed.

"The following measurements indicate the size of the reptile:

"Extent of anterior sixteen teeth in dentary 75 mm.

Extent of five lower teeth 20

Height of crown of fourth lower tooth 10

Depth of lower jaw at symphysis 15

Extent of seven anterior maxillary teeth 33

Height of crown of first maxillary tooth 9

Antero-posterior diameter of crown 3 "

Of the original material, only the fragmentary mandible and maxilla could
be positively referred to this genus and species. A study of the associated material
in the type collections, by a process of elimination, indicated various other parts
of the skeleton which might, provisionally, be associated with the genotype. The
greater part of this material came from a bone-bed, hereinafter designated as the
Baldwin Bone-bed, in which the bones were widely distributed, and almost wholly
disconnected. Descriptions and figures of such bones of this bone-bed as seemed
to be conspecific were given by Williston in his American Permian Vertebrates,
p. 81, plates xxxiv-xxxvii. A comparison of these figures with those of the present
paper will show that most of the bones there provisionally referred to O. mirus
were correctly identified, as follows : clavicle, scapula, humerus, radius, ulna, ilium,
femur, and tibia. The vertebra figured and described is more doubtful, since it
differs materially in its more elongated spine. The sacrum, very doubtfully re-
ferred to this genus, belongs with Sphenacodon.

37

38 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

The material herein described, consisting mostly of a single skeleton (Cata-
logue No. 650 University of Chicago) of remarkable completeness, came from
another bone-bed, which for convenience is called the Miller Bone-bed, after its dis-
coverer, Mr. Paul C. Miller. This bone-bed is about half a mile east of the Baldwin
Bone-bed, and apparently in quite the same horizon. The skeleton, as fossilized, must
have been nearly or quite perfect. Perhaps a half or more of the tail, the distal half,
had been weathered out of the matrix and strewn on the surface below. Unfortu-
nately, before the skeleton, which lay about a foot above the chief bone-layer, was
recognized a stroke of the pick had mutilated the sacral region, which, with some of
the adjacent spines and the underlying left foot, was lost in the loose material thrown
lover the dump. Although diligent search for these lost parts was later made, not
many of them were recovered. Dr. Williston has, regretfully, to add that he alone
is responsible for the mutilation.

Upon the recognition of the skeleton lying in an orderly position, the speci-
men was carefully traced out and removed by Mr. Miller, with our aid, in a single
large block of matrix secured by bandages. The skeleton, as received in the
laboratory, was carefully cleaned of its investing matrix on the upper (right) side,
and then embedded in a plaster mold, after which it was turned over and worked
out from its better preserved lower (or left) side. As it lay in the matrix (pi. i,
fig. 2) it had what seemed to be a most remarkably natural position, with its
various parts all in articulation, or nearly so, as far as the pelvis. The right foot
and hand, falling on their inner sides, had their inner digits partly folded under
the others, but with little disturbance of their articulations. The shoulder-blade
of the upper side had slipped downward and forward a few inches. Apparently
after death, and before maceration had gone far, the cadaver had been gently
pushed cephalad and ventrad some 6 or 8 inches, dragging the under limbs dorsad
and caudad, bringing the left hand under the lumbar vertebrae and the left foot
under the pelvis. The curve of the dorsal vertebrae, as the skeleton lay, is per-
haps a trifle greater than was usual in life, and the skull evidently had suffered
more flexion than was commonly assumed in the living animal. These positions
and curves, however, have not been greatly modified in the mounted skeleton
(pi. I, fig. 3), the skull and vertebras, for the most part, being still united by their
original matrix.

Lying on the same level, and perhaps 2 or 3 feet distant from this skeleton,
there was another (No. 652) skeleton of the same species, almost identical in size
and characters, the larger part of which had been washed out from its matrix and
strewn down the hillside, intermingled with the distal caudal vertebrae of the more
complete skeleton. From the matrix of this second specimen the pelvis, posterior,
dorsal, and sacral vertebras, the nearly complete left hind leg and a part of the
right one, and various other bones were recovered, parts which, fortunately, supple-
ment the more complete specimen almost perfectly. Furthermore, Ophiacodon
mirus seems to be well represented in the underlying bone layer by isolated bones,
many of which were collected, but have not yet been worked out. The skeleton,
therefore, as herein described, lacks some distal caudal vertebrae and the vertebral
spines of the posterior dorsal and sacral regions only. From all of which it results
that, aside from the intimate structure of the skull, there is perhaps no genus of
Permian vertebrates now more completely known than Ophiacodon, unless it be
Limnoscelis.

The matrix of the Miller Bone-bed is a dark brown, indurated sandy clay,
weathering loose near the surface, but very hard, almost stony, farther in. It
was removed rather easily from the bones, leaving the original smooth surface.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 39

The skull of Ophiacodon * is remarkable among Permian vertebrates for its
relatively large size and delicacy of structure. In part the result, perhaps, of the
extreme thinness of most of its bones, the skull was more or less compressed from
side to side in fossilization, though without affecting its general shape and con-
tours. So frail is its structure that it was found inexpedient to remove it entirely
from its supporting matrix. It was, therefore, carefully cleaned of its investing
matrix on its upper or right side and embedded in a plaster mold, after which it
was turned over, with the rest of the skeleton, and cleaned on the other, the better-
preserved side. Both sides have been photographed and studied by us, but the
right side is no longer accessible, embedded as it is in plaster for the better preser-
vation of the specimen.

The skull is remarkably narrow, high, and long. The very small nares are at
the extreme front end; the small orbits are far posterior. The upper side is flat-
tened in the frontal and parietal region ; its greatest width is just back of the orbits.
The occipital region forms a declivity, as in Dimetrodon. The parietal foramen has
not been detected with certainty; it must have been small. Immediately in front

Fig. 23. — Ophiacodon mirus Marsh. Lateral view of skull, X ,'■>.
pa, parietal; />o, postorbilal; pf, prefrontal; /, lachrymal; y,
jugal; (ji, quatlrato-jugal; q, quadrate.

of the orbits the upper surface narrows; thence to the nares the border, formed
chiefly by the nasals, is gently convex in outline. Doubtless in life the very broad
sides of the face were gently convex, but, as preserved, the thin bones forming
them are nearly in contact, producing a light concavity between the thickened
alveolar border of the maxilla and the thickened nasal border; the bones of this
region are scarcely thicker than writing paper, for the most part, and are so cracked
as to render impossible the certain determination of their connecting sutures.

In fig. 23 the skull is shown as it lies in its matrix, except that the two posterior
premaxillary teeth have been brought into anatomical association with the first
one. The general shape of the skull, as shown, seems to be quite that of life; pos-
sibly the dentigerous margin of the maxilla is a trifle too convex in outline. The
nares, as has been said, are very small, and are situated near the extreme front
end of the skull. The orifice of each naris is nearly circular in outline, directed
laterally and perhaps a little forward in life. The orbits also are remarkably small
in comparison with the size of the skull. They are subtriangular in shape, broader
above, the lower end separated from the posterior end of the maxillae by the mod-

• Ophiacodon minis Marsh, American Journal of Science, xv, 411, May 1878; Baur and Case, Trans.
Amcr. Phil. See, n.s., 1899, p. 5, 1907; Williston, American Permian Vertebrates, p. 81, plates xxxiv-
XXXVII, Oct. 191 1.

40 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

erately broad jugals. Above, the orbits are roofed over by a projecting plate from
the frontals. The opening of each orbit, as preserved, looks laterally and not at
all upward, and this must have been essentially the condition in life. The pos-
terior border, formed by the postorbital above and the jugal below, is thickened,
as is also the antero-superior border. Below, in front, the margin is very thin,
and the precise edge here has been lost in preparation. Immediately back of the
upper, dilated, part of the orbit, and separated by a narrow but firm bar, is the
relatively small, oval, lateral temporal fenestra, an opening 25 mm. in its greater,
13 mm. in its conjugate diameter, the former at an angle of about 45 degrees with
the condylar-narial axis of the skull. Its postero-inferior margin is thin; elsewhere
the border is thicker and rounded. Above, this vacuity is separated by a thick,
subcylindric or oval bar from a smaller, oval vacuity or foramen, the supratem-
poral fenestra, which is about 12 mm. in its greater diameter at the bottom of a
fossa, which is partially over- roofed by the posterior prolongation of the parietal.
There seems to be no doubt that this opening is a normal one, not due to some
accidental separation of contiguous bones, since it is alike on the two sides of the
skull. On each side the clay matrix filling the fossa and vacuity was removed with
care, proving conclusively that the thick, rounded, smooth, and deep margins
were everywhere normal, without signs of sutural roughening or fractured surfaces.
This opening, bounded below by the thickened supratemporal arcade, which
evidently is composed of the united postorbital and squamosal, posteriorly by what
is certainly the squamoso-parietal bar, and superiorly by the parietal, can be inter-
preted as nothing else than a true supratemporal fenestra.

Below and back of these vacuities there is a broad expanse of thin bone,
doubtless composed of a squamosal element, the jugal and the quadratojugal. The
striations on the clean surface of the bones in this region, as also the position of
lines which seem to be sutural, indicate their relative extent and articulations
about as they are shown in the figure. With these interpretations, the jugal in
front forms only a small part of the free lower border of the maxillo-quadrate arch.
A sutural line seems to distinguish the jugal from the post-orbital above, as shown
in the figure between the orbit and lateral fenestra. Posteriorly the diverse stria-
tion and an intervening sutural line distinguish the jugal from an element that
is doubtless the true quadratojugal, which extends far forward on the lower, free
border of the arch. The position of this suture is very different from that bound-
ing the jugal behind, positively and definitely known in Dimetrodon. In this genus
there is but one bone articulating with the jugal posteriorly, the prosquamosal of
Case, the squamosal of authors; at least neither of us is convinced of the presence
of a lower bone articulating with the jugal, which if present must have joined its
extreme tip only, taking no part whatever in the margin of the temporal fenestra.

Case believes that there is no lateral quadratojugal bone in Dimetrodon, but
that the bone properly so-called is attached to the posterior side of the quadrate,
extending upward. A recent examination of the structure of Dimetrodon in this
region by Williston in material that is, perhaps, as good as any known, tends
to convince him of the general correctness of Case's observations, though he is not
at all assured of the correctness of his interpretations. The squamoso-jugal suture
in Dimetrodon passes backward obliquely from above downward to the extreme
tip of the jugal. Back of the jugal, covering the quadrate, there is an element
which seems to articulate with an upper bone lying posterior to that identified as
the prosquamosal by Case. Whether or not this bone touched the extreme tip of
the jugal Williston can not decide, but he thinks not. Between it and the quadrate
posteriorly there is a distinct foramen, identified as the quadrate foramen by Case.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 4I

These relations are proven incontestably by material in the collections of the
University of Chicago, and Williston vouches for them. The best evidence we
have seen of the presence of a lateral quadratojugal bone in any pelycosaurian
reptile is that furnished by the present specimen.

The posterior thickened border of the quadrate slopes obliquely backward
from the long axis of the skull. How much of the quadrate is actually visible
from without, how much is covered by the roof bones, can not be said. There
is certainly no quadrate foramen.

The face in front of the orbits, as has been said, is of great extent. Sutural
lines can not be detected, though the longitudinal striation of the broad, long,
and somewhat thickened nasals indicates their width, and probably their length,
reaching far back toward the orbits. The prefrontal seems to be large, and the
extent of the jugal seems to be indicated by striation and sutural lines. Whether
the lachrymal (postnarial) extends the whole distance to the margin of the naris,
as in the cotylosaurs, is doubtful ; certainly if it does it must be a very slender bone.

The maxilla is a very long bone, with its dentigerous border somewhat thick-
ened and convex in outUne. Twenty-nine maxillary teeth are preserved, with
vacant spaces for six or seven more, which, if present in life, must have been lost
before fossilization. Possibly the whole number of teeth in this bone may have
been 36. The teeth are slender, conical, recurved, and pointed, and they do not
vary much in length as far back as the middle of the series. At about the junction
of the first and middle thirds of the series there is a distinctly stouter tooth, corre-
sponding to the large maxillary tooth of Varanosaurus. The maxilla is thickened
above it, and the dentigerous border here is more convex or subangular in outline.
The teeth extend posteriorly to opposite the lower angle of the orbit. In the
mandible, so far as can be observed, the teeth are all uniformly slender, especially
the more anterior ones, with the exception of the first one or two, which seem to
be stouter. In the premaxillae there are three teeth on each side, all larger than
the ones immediately following in the maxilla; the foremost is perhaps as stout
as the enlarged maxillary tooth.

The matidible is long and slender, with a marked concavity of its alveolar
border corresponding to the convex dentigerous border of the maxilla. The man-
dible is a little thickened in front, where the two sides meet in a short symphysis
about an inch in length. The coronoid angle is not high ; its border is gently convex
above, where it slightly underlaps the free border of the quadratojugal. In the
space left free back of the teeth, between the jugal and the mandible, the rather
broad surface of the pterygoid, or ectopterygoid, is visible; the bone clearly abutted
against the mandible in life. As regards the structure of the mandible little can
be said, save that the very long splenial meets its mate in a symphysis, as is usual
in the early reptiles.

The present specimen seems to demonstrate, for the first time in a Paleozoic
reptile, the normal presence of both supratemporal and lateral temporal fenestras.*

• Neither of us is quite convinced of the presence of both vacuities in the Proganosauria and Protero-
sauria. The presence of the upper vacuity in Paleohatteria has been, we believe, assumed because of the
resemblances otherwise of that genus to Sphenodon. Credner, in his original description of the genus,
alludes very briefly to its presence as " wahrscheinlich " ; apparently no certain evidence of its existence
was found in any specimens which he studied. In a later paper he restores the skull after Sphenodon and
says the vacuity is present, but gives no reasons for this statement. We have seen that other Permian
reptiles having like " rhynchocephalian " characters do not have two vacuities on each side, and the pres-
ence of an upper one in Paleohatteria must remain a subject of legitimate doubt until additional evidence
incontestably demonstrates its presence. A very recent examination of the type specimens of Paleohatteria
in Leipsic has convinced Williston that only one temporal vacuity is present.

42 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

In Dimetrodon Case has asserted the presence of an upper vacuity, but the evi-
dence is now apparently conclusive, and is so accepted by him, that, in some speci-
mens at least, there is no upper vacuity; and Williston, not having studied the
matter thoroughly, is not in a position to express a positive opinion as to its normal
absence, though he thinks such is the case. It will, perhaps, require additional
specimens of Ophiacodon to prove its constant occurrence in this genus, but it is
our firm opinion that the opening in the present specimen is a normal character
of the genus; it certainly is of this specimen. And this is said with a full appre-
ciation of the reputed fenestrae in the skulls of Procolopkon and Dicynodon, later
shown to be errors.

As a bit of philosophy we may say that observers are often led into the recog-
nition of characters that accord with preconceived opinions. If one finds many
characters agreeing in different forms, he is naturally disposed to assume that
other characters do also. For a long while Procolopkon was supposed to be closely
related to the rhynchocephalian reptiles — a supposition now summarily disposed
of, we think — to such an extent that it was located with the "Diapsida" and
" Diaptosauria " even. But Sphenodon and the Rhynchocephalia have been respon-
sible in the past for many taxonomic sins, and we have not yet quite escaped from
the shadow of their misleading influence.

In the present reptile the presence of a supratemporal fenestra was wholly
unexpected, and its discovery was rather a shock, since it seems almost hopelessly
to confuse the taxonomy of the Reptilia. That there is such a fenestra in the
present genus is, we think, beyond reasonable doubt. What is its significance ? It
will be seen that not only the upper vacuity, but the lower one also, is remarkably
small, smaller proportionately than is known in any other double-arched reptile.
Are they rudimentary ? Or vestigial ? Is their small size due simply to the remark-
able development of the facial part of the skull without a corresponding develop-
ment of the posterior part ? Against this explanation is the fact that the temporal
region of Ophiacodon is, for the most part, covered by a broad expanse of bone,
necessitating the assumption that in the evolution of this particular type of skull
from some smaller one having proportionately larger vacuities everything has
developed except the vacuities. This hypothesis we think may be rejected. There
seems to be but one conclusion — -that the openings are rudimentary.

The presence of one or two temporal vacuities has hitherto been regarded as
a sort of fetich in reptilian taxonomy and phylogeny — -as a crucial test of relation-
ships, not only of ordinal but even of subclass value in classification. For some
years past they have been considered as a sort of noli me tangere character, in that,
whatsoever of iconoclasm is permissible in regard to the feet, limbs, girdles, and
vertebrae, hands must be kept off the sacred temporal region of reptiles. But
if we do not invade the sanctity here we shall be led into a dangerous morass of
speculation. In all other respects Ophiacodon is not only a "synapsidan" reptile,
but a primitive one at that. To separate the form ordinally and classify it with
the "Diapsida" would be taxonomic speculation run wild.

Could we know that Ophiacodon is the beginning of a morphological phylum
that ended in the Archosauria or Rhynchocephalia we would be quite justified,
notwithstanding all of its other and intimate relationships with the one-arched
or synapsidan forms, in separating it as the representative of a distinct order
of reptiles. In any true phylogenetic classification the lines of descent must be
followed back to where they diverge, to where the branches ultimately are sepa-
rated by nothing more than specific characters, since in the ideal classification of
animals and plants the truest is that in which orders, families, and genera are

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

43

ultimately distinguished by specific characters only, where the chain from ancestor
to descendants nowhere lacks a single link. No character is ever more than varietal
until it is fixed and isolated physiologically by heredity and time.

One may be speculative and assume that Ophiacodon is really the beginning
of the double-arched phylum, the beginning of a subclass, but there is not the
slightest proof that such is the case. It will be many years before such an hypoth-
esis can be proven or disproven, if it were ever possible, and classifications based
upon hypotheses are unscientific. If the present form is merely a "mutation" or
sport, then certainly the presence of a supratemporal vacuity has no taxonomic
value, for we have no faith in the theory that species ever arise in such sudden
ways. In other words, the most profound taxonomic characters must have had
beginnings when they were merely varietal or specific ; not till time and long descent
have fixed them do they achieve a higher rank. In the present case we are con-
vinced that, without further evidence to the contrary, the presence of two temporal
vacmties and two temporal arches in Ophiacodon can not be assumed to have
more than generic value.

Vertebra and ribs: As has been stated, the vertebral column as collected was
without break or even noticeable disturbance as far back as the sacrum, lying

Fig. 24. — Ophiacodon mirus Marsh. Cervical vertebra; from the side, the sixth also from in front, X ^2.
ax, axis; an, atlantal neuropophysis; 0, odontoid; r, r, ribs; q, quadrate; pa, proatlas.

with the skull and all or nearly all the ribs in natural articulation. Six or seven
of the vertebrae may properly be called cervical, and twenty or twenty-one dorsal ;
the entire number, as in Dimetrodon, Varanosaurus, Theropleura,* and probably
all other pelycosaurs in the narrower sense of the word, is twenty-seven in all
certainty.

Proatlas (fig. 24, pa). A proatlas has never before been recognized in an Ameri-
can Paleozoic reptile. Its existence in the present genus was not unexpected, since
Williston has already suggested that the facets on the sides of the foramen magnum
in Dimetrodon are for the articulation of this bone.f In the present specimen
the proatlas lies quite in position, articulating posteriorly by a well-formed zyga-
pophysis on each side with the arch of the atlas, and in front on each side with a
facet on the margin of the foramen magnum. Because of a slight injury on the
right side, it can not be said positively that the element is single, though in much

* Twenty-six presacral vertebrae back of the atlas are figured by Case in Theropleura retroversa (Publi-
cation 55, Carnegie Institution of Washington, pi. xill, fig. i).
t American Permian Vertebrates, p. 93.
4

44 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

probability it is. It forms a sort of bridge between the atlas and the occiput, above
the spinal canal, with an obtuse protuberance or rudimentary spine above on
each side, like that on each atlantal arch.

Atlas (fig. 24): The atlas is preserved perfectly on the left side. Each neuro-
centrum is slender, curving back to articulate, by a well-formed zygapophysial
facet, with the axis at the base of the spine in front. On the front of each arch
above there is a zygapophysial facet, slightly constricted from the arch itself, for
articulation with the postzygapophysis of the proatlas. Anteriorly a process
descends downward and backward, articulating at its extremity with the atlantal
rib, which lies closely in position in the specimen. The neurocentrum, as a whole,
differs only a little from that figured by Williston in his American Permian Verte-
brates, plate xxxiv, figs. 10, 11, from the Baldwin quarry, the chief difference
consisting in the more slender process which is represented by a mere protuberance
in the atlas of Ophiacodon. In much probability the figured neurocentrum belongs
with Sphenacodon. The arch is supported chiefly by the odontoid, resting only
sUghtly upon the atlantal intercentrum below. The intercentrum is of somewhat
larger size than the intercentrum between the odontoid and the axis.

The atlas, it is seen, is of a primitive type, almost identical in structure with
such vertebrae as are known in the tails of certain temnospondyls, with the omission
of the chevron. In fact, the union of the neurocentra with each other and with
the intercentrum would result in an almost typical embolomerous vertebra.

This condition of the atlas in these primitive reptiles suggests certain conclu-
sions. The reptilian atlas could not possibly have been evolved from the atlas of
any known amphibian. In the temnospondyls the atlas is composed exclusively
of the paired neurocentra fused with the intercentrum; the pleurocentra have,
apparently, disappeared entirely. The proatlas also is quite unknown in any
amphibian. The proatlas is generally regarded as the remnant of a preatlantal
vertebra which has disappeared in the reptiles, and the presence of zygapophysial
articulations in these early reptiles would substantiate that theory. Gadow,
however (Evolution of Vertebral Column, etc.), insists that the proatlas is merely
the spine of the atlas. It has been known for some time that the amphibian occi-
put, whether of living or ancient forms, docs not correspond to the reptilian; and
we also know that the atlas in the two classes is not homologous. If, then, the cor-
roborative proof furnished by these reptiles supports the theory that the reptilian
atlas is in reality at least the second, perhaps the third vertebra of the amphibians,
it would seem probable that the amniote atlas was derived from a true holospondy-
lous vertebra, but one in which the primitively large intercentrum was persistent ;
perhaps such a vertebra as is found in Seymouria.

Among modem reptiles a proatlas is known in the Crocodilia and Sphenodon;
also in certain dinosaurs and the Therapsida. In Procolophon, as Broom informs
us, the atlas and proatlas are almost identical with those of the Pelycosauria, except
that the intercentra in front and behind the axis are parial, a condition also found
in some modem turtles. In the Dinocephalia, according to the same authority,
the atlas and proatlas seem to be typically pelycosaurian in structure. Indeed,
it may be assumed that this condition is that of all primitive reptiles. Brown
(Osteology of Champsosaurus) describes the arch of the atlas in the very archaic
genus Champsosaurus as articulating directly with the exoccipital by an anterior
process. This condition would be remarkable, and one wonders whether there
may not have been an intervening proatlas which has not been recognized.

The axis (fig. 24) differs from that of Dimeirodon especially in the much greater
antero-posterior expansion of its spine, which is even greater than in other known

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

45

poliosaurids. The spine is thickened behind, thin in front, projecting much an-
teriorly, and is helmet-shaped in outline. The atlantal zygapophyses in front are
relatively small. The centrum differs very little from that immediately following.
Postaxial vertebrce (figs. 24-26) : The centra of the first four vertebrae back of
the atlas are especially characterized by the obliquity of their articular faces; this
obliquity decreases on the fifth, and the centra of the following vertebrae all have
their terminal articular faces nearly at right angles to their long diameters. On
the first four postaxial centra the cartilaginous surface of the anterior rims is carried
far back. The capitula of the ribs articulated in part on this surface, but its great
extent is doubtless due to the increased mobility of the vertebrae in this region.
For this reason, as also because of the position of the scapulae in the specimen,
the first six, if not the first seven, vertebrae may be considered as true cervicals.

Fig. 25. — Ophiacodon mirus Marsh. Seventh to twentieth vertebrae from the side, X yi-

The centra of the cervical vertebrae are a trifle longer than those of the dorsal
region, which are all nearly uniform in length. The under side is gently concave
in outline. The anterior vertebras have a sharp, thin keel below; in the posterior
vertebra; the keel is thicker, with angular or parallel sides. The articular faces of
the zygapophyses are directed at a considerable angle inward and outward. The
zygapophyses are strongest in the cervical region. The spines of the vertebras are
of nearly imiform height throughout the presacral series, ascending above the
centra about 2.5 times their vertical diameter, or a little more. Anteriorly, as far
as the tenth or eleventh vertebra, they are stout, with a truncate, cartilaginous
surface at the extremity, of an oval or ovate outline. Back of the eleventh or
twelfth vertebra the spines decrease in thickness and increase in width, becoming
quite thin and flat in the posterior region, as shown by the outline figures of the
extremities. Very remarkably the fourth and the ninth spines, especially the fourth,

46

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

are conspicuously different from those immediately adjacent, as shown in t]\c
figures; so different, indeed, are these spines that had they been found isolated
they would have been unhesitatingly referred to a different genus from that of the
other vertebrae. Doubtless they are anomalies that will not be found in other
specimens of the same species.

The diapophyses are very much as they are in Varanosaurus * and Dimetro-
don. Those of the cervical region are stouter, and are directed outward, downward,
and backward, becoming almost transverse in position in the sixth, quite as in Vara-
nosaurus. Beginning with the seventh vertebra the processes project transversely,
gradually becoming shorter and less stout, until in the last, that is, the first presacral,
they are mere tubercles for the sutural attachment of very short ribs. In all the
presacral vertebrae, unless the first three are exceptions, the articular surface for the
tuberculum is continuous on a thin prolongation extending downward and forward
to the intercentral space. The diapophyses are located exclusively upon the arch,
as shown by distinct sutural Hnes distinguishing the arch from the centrum.

<«f:23

Fig. 26. — Ophiofodon mirus Marsh, X 'A. A, twenty-first to the twenty-sixth vertebrae
from the side; B, first presacral, sacral, and two proximal caudal vertebra; from the side.

The intercentra were probably preserved throughout the presacral series, but
a few have been lost in the preparation of the specimen. They measure about
12 mm. from side to side and 4 mm. in width.

Sacrum: Unfortunately, as already explained, the sacrum of the more perfect
specimen (No. 650) was lost in collection. Of the second specimen (No. 651) the
posterior presacral vertebrae, the sacnim, and numerous vertebrae of the tail were
found loosely united by matrix, though with their processes more or less broken
away and lost. The sacral vertebras, so far as they are preserved, scarcely differ
from those immediately preceding, except in their ribs and rib attachments. The
sutural surfaces for the ribs, shown fully on the right side, are very extensive,
especially in the first of the two vertebrae. The diapophysial surface is broad and
overhanging; the parapophysial surface, of less extent, reaches quite to the lower
margin of the centrum at the anterior end. On the second vertebra the surfaces
are less extensive, about equally divided between arch and centrum, and the ribs
of both vertebrae projected at their capitular end into the intercentral space. The
distal extremities of the sacral ribs of both vertebrae are lost, but a nearly complete
rib found among the loose bones in the wash shows a structure not unlike that
of Varanosaurus. There were but two pairs of sacral ribs, as shown in the figure.

Caudal vertebroB: With the more complete specimen thirteen vertebrae were
found associated, connected by matrix in two series, as shown in fig. 27. From

' Williston, American Permian Vertebrates, plates i, 11.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

47

the wash of the two specimens numerous other caudal vertebrae were obtained,
but it is impossible to say to which skeleton they belong; in all probability they
belong with both, though there seem to be few or no duplicates. As they differ so
little in their characters, only a few have been figured — the four distal ones shown
in fig. 27 B. The first one shown in fig. 27 a may be either the third or the fourth
of the series. The tail as a whole must have been very much like that of Vara-
nosaurus, long and slender, and probably included, as in that genus, fifty or more
vertebrae. Conspicuously different, however, are the numbers of pygal vertebrae,
that is, those without chevrons, in the two genera. In Varanosaurus there are not
more than three, while in Ophiacodon there are at least six, and possibly seven.
The spines decrease rapidly in size and height, the first complete ones preserved,
those of the eighth and ninth caudals, being even shorter than the corresponding
ones of Varanosaurus. Between the two series shown in figs. 27 a and B there is,
in all probability, but one vertebra missing, the perfect spine shown being that of
the fourteenth or fifteenth vertebra.
Between the last of this series and
the next one shown in the figures,
probably the thirtieth, a dozen ver-
tebrae are preserved. The smallest
vertebra figured is probably the
fortieth or thereabouts, possibly the
forty-fifth.

From all of which it is quite cer-
tain that the tail was long and slen-
der, perhaps more slender than that
of Varanosaurus, of which forty-
seven caudal vertebrae have been
found in a continuous series, with
at least a half dozen terminal ones
missing. There are ten or more

pairs of sutiu-ally united ribs, the proximal ends of many of which are yet connected
with their vertebrae, but whose lengths can not be determined. This is about the
number occurring in Varanosaurus and most other contemporary reptiles. The ante-
rior caudal centra are rather sharply keeled below, but the tmder side soon becomes
fiat, or has a pair of grooves separated by a low keel. Evidently well-developed
chevrons were present throughout the series back of the sixth or seventh, though
only a few fragments of them have been recovered. The tail very clearly was not
natatorial in structure.

Ribs: All the vertebrae as far back as the eleventh or twelfth caudal are costif-
erous. The ribs have the usual articulations of the primitive reptiles, the head
resting more or less in the intercentral space, the tubercle articulating with the
arch. In no presacral vertebra is there a distinct parapophysial surface on the
centrum, but the tubercle articulates with a more or less prominent diapophysis.
In Ophiacodon there is a more or less continuous articular surface between the
capitulum and tuberculum, as in most of the contemporary cotylosaurs. In most
of the ribs this intervening surface is narrow, and in some it is obsolete, but in none
is there a distinct emargination as in other known pelycosaurs.

In many of the primitive reptiles, such as Diadectes, Lahidosaurus, Capto-
rhinus, Pareiasaurus, Paleohatteria, and Sphenodon, as also the "Microsauria,"
the ribs are commonly called single-headed, but incorrectly, since all have the two
articular surfaces, capitulum and tuberculum, the former articulating in or near

Fig. 27. — Ophiacodon mirus Marsh. A, proximal caudal
vertebrae from side; B, proximal and distal caudal ver-
tebra from side, X K.

48

PERMO-CARBONIFF.ROUS VERTEBRATES FROM NEW MEXICO.

the intercentral space, the latter with the arch at the extremity of a more or less
prominent diapophysis. And this is the primitive rib attachment in reptiles. In
the further evolution of the ribs these two surfaces, primitively continuous, become
separated by an emargination, leaving a distinct foramen between the articulated
rib and its vertebra. This modification of the articulation was, however, acquired
in different phyla independently, since we find the continuous articulation, the
"single-headed" rib, in such diverse reptiles as the "Microsauria," Diadectes, Labi-
dosaurus, Ophiacodon, Paleohatteria, and Sphenodon, while the emarginate form, or
"double-headed" rib, was acquired in such a generalized cotylosaur as Seymouria
(fig. 28 b). In all these reptiles, however, the ribs are distinctly double-headed in
the proper sense of the word, since they have both capitulum and tuberculum. Per-
haps some convenient single word to characterize each type, such as holocephalous
and dichocephalous , will be useful.

Ophiacodon not only has holocephalous ribs, but also dilated cervical ribs, a
character (so far as known) hitherto peculiar to the Cotylosauria. Beyond the

Fig. 28. — ^A, cervical ribs of Ophiacodon mtrus Marsh, X K- B, dorsal rib of Seymouria
baylorensis Broili, X i; C, right ilium of Ophiacodon mirus Marsh, inner side, X /^.

seventh the ribs are all cylindrical and slender to their extremities, becoming quite
short in front of the sacrum. The first three cervical ribs, those articulating with
the atlas, the axis, and the third vertebra, are short, and but little dilated distally,
attached exclusively, it seems, to the diapophysis.

Ventral ribs: The size and characters of the ventral ribs are best shown in the
photographic illustration (pi. i, fig. i). They formed a large sheet in the specimen
as found, covering the whole of the space back of the coracoids and between the
ends of the ribs, quite to the pelvis. They are very slender, and lie closely together,
each meeting its mate in the middle in an acute V. The length of no individual
rib can be determined, since they are all more or less broken into short, contiguous
pieces. It is possible that some of them, if not many, were continuous throughout
their whole lengths.

Pectoral girdle and extremity: The pectoral girdle and both anterior extremities
lay quite in natural arrangement as found in the specimen, except that the right
scapula had been pushed downward and a Uttle forward, an inch or two from its
natural place ; and the left hand was partly concealed beneath the posterior dorsal
vertebras. The right arm and hand, closely articulated throughout, were directed
downward and forward ; the inner fingers were doubled under the outer, ulnar ones.

The inter clavicle, preserved in place, has a broad, thin, flattened stem, some-
what convex below, and a moderately dilated anterior extremity. The stem
extends backward, as articulated, about an inch beyond the posterior angles of the
coracoids. In the present specimen (No. 650) it is somewhat flattened by pressure;

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

49

in life it doubtless had a considerable convexity below, the front end turned upward.
The dilated end is concave above, convex below, and grooved on each side for
the attachment of the clavicle.

Clavicles: Both clavicles are entire and firmly attached to the interclavicle,
but are compressed and somewhat distorted. Figures 29 b and c have been made
from an isolated specimen which agrees perfectly and is undistorted. As seen, the
mesial extremity is only a little dilated in comparison with the clavicles of both
Varanosaurus and Dimetrodon. Of this extremity, the upper side is moderately
concave, the lower convex, and both sides have conspicuous radiating ridges and
grooves, more pronounced near the margins. The slender stem turns upward and
outward, in the articulated position, at an angle of about 130 degrees. As articu-
lated, the upper ends are about 125 mm. apart. The front border is thickened
throughout; the posterior border of the dilated part is thin. Beyond the dilated

Pig. 29. — Ophiacodon mirus Marsh, X H. A, clavicular girdle from below; B, right clavicle from in
front; C, right clavicle from within; D, right scapula-coracoid, outer side.

part, on the posterior side, there is an everted, thin flange, which overlaps the lower
front border of the scapula. Distally the shaft for a short distance is oval in cross-
section, the posterior border only a little thinner and without a surface for attach-
ment to the scapula. Still more distally the outer anterior side is flattened and
grooved, and striate, as though for the articulation of a slender cleithrum, of which,
however, no definite evidence has been found in the specimen.

Scapula-coracoid: The scapula-coracoid of the right side is very little dis-
torted (fig. 29 d). The front border of the scapula, however, was so intimately
united to the clavide that it could not be recovered entire. On the left side the
scapula lay over the humerus in part, and was pressed outward in its middle por-
tion; its outline, however, was complete, and it has been used to complete the
figure as shown. On both sides the scapulae lay nearly or quite in position in the
skeleton; that of the right side had been pushed downward and forward an inch
or two beyond the position occupied by the left one, whose distal margin crossed
the base of the spine of the eighth vertebra.

so

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

Fig. 30. — Ophiacodon mirus
Marsh. Right front leg,
dorsal side, X yi-

The scapula is broad, resembling that of Dimetrodon more than that of Vara-
nosaurus. The posterior border is thickened and rounded, and curves upward
and backward. The supraglenoid fossa is broad and shallow ; it is pierced by the
usual supraglenoid canal. The front border is thinned throughout; it is straight
near the middle for about an inch, where the distal end of the clavicle is apposed.
Both above and below this straight border there is a shallow emargination of the
very thin bone. That a small cleithrum was present is possible, but not certain.
Lying just above the end of the left scapula is a small, thin bone, so closely applied
to the vertebrae that it could not be detached ; it may be a vestigial cleithrum. The
upper or distal border of the scapula, as usual, is truncated and convex in outline,
for the attachment of a suprascapular cartilage. The coracoid of Williston, the
procoracoid of authors, is narrow and, in its somewhat flattened condition as

preserved on both sides of this
specimen, is not directed in-
wardly as much as in other
forms. The broad stem of the
interclavicle filled a part of
the interval between the two
bones, but it seems probable
that in life there was a consid-
erable cartilaginous continua-
tion of the truncate margin of
each coracoid. The posterior
coracoid, or metacoracoid of
Lydekker and Williston, the
coracoid of authors,* is of
considerable size, its separating suture showing con-
spicuously on both sides ; the suture between the ante-
rior coracoid and the scapula is not distinguishable.
It is an interesting fact that, wherever this bone
is ossified, the suture uniting it with the scapula and
anterior coracoid remains distinct much longer than
that between the anterior coracoid bone and the
scapula ; it is nearly always visible in Dimetrodon, and
the posterior coracoid is often found isolated. Will-
iston believes that this is further evidence of its dis-
appearance in modem reptiles, as in Varanosaurus
and Seymouria among the ancient ones. The bone,
whatever it may be, coracoid or metacoracoid, agrees closely with that of Dimetro-
don. Its suture passes through the postglenoid facet a little in front of its middle,
and not far back of the preglenoid facet, and about half an inch posterior to the
supracoracoid foramen. Its posterior border is thickened, concave in outline, and
somewhat everted. At the upper part of this border and somewhat inward from
the posterior end of the glenoid facet, there is a prominent process, like that of
Dimetrodon, Casea, Champsosaurus, etc. On the inner side of the scapula is the
usual subscapular fossa, with the opening of the supraglenoid foramen at its upper,
that of the supracoracoid at its lower end.

Humerus: The humerus is an unusually short and stout bone for a zygocro-
taphic reptile, resembling somewhat in its general characters those of Diadectes and
Limnoscelis. It has a very broad entocondyle, a very large, hemispherical capitel-

*Poradiscussionof the homologies of coracoids, see Broom, Anatom. Anzeiger, Band 41, p. 625, 1912.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

SI

lum, and a stout ectocondyle directed obliquely dorsad. The plane of the upper
extremity is relatively but little divergent from that of the lower. The rather
broad, spiral, and concave glenoid articular surface winds about the bone from the
dorsal outer side to the ventral inner side (fig. 31 a), and fits very perfectly into
the glenoid cavity of the scapula. As articulated with the scapula, the humerus
has a somewhat obliquely upward, outward, and backward position, showing con-
clusively that the animal in life could not have raised itself very high above the
ground. Indeed, the humerus could not have been brought much below a hori-
zontal position in life without partial dislocation from the scapular socket. The
lateral process is stout and is situated at about the junction of the upper and middle
thirds of the bone ; in the articulated position of the humerus it was directed down-
ward and a little forward. There is a small median process or eminence. The
ectepicondylar process is stout and protuberant, directed downward and forward
in the articulated position of the bone, and is situated a little beyond the middle
of the distance between the lateral process and the distal extremity (fig. 31 b).
The capitellum is unusually large, nearly circular in outline, and comprises about
one-third of a sphere ; the plane of its base is directed forward at an angle of about
45 degrees. The very stout ectocondyle is directed dorsad at an angle of almost

Fig. 31. — Ophiacodon mirus Marsh. A, left humeras, ventral side,
X K; B, left humerus, distal end, X K; C, left ulna, radius, and
carpus, ventral side, X >i ; D, left carpus, dorsal side, X K-

90 degrees with the plane of the entocondylar expansion ;
like the entocondyle, it has a terminal cartilaginous sur-
face for the attachment of powerful extensor muscles. This unusual dorsad inclina-
tion of the condyle is very characteristic of short, stout legs, and especially of short
humeri, as seen in Diadectes, Diasparactus, and Limnoscelis. The entocondyle is
very broad, and has an extensive surface on the distal border for the attachment of
strong flexor muscles. The articular surface for the ulna winds obliquely about the
distal border from the margin of the capitellum inward.

Altogether the prominent and stout lateral processes, the protuberant ectepi-
condyle, probably for the attachment of the supinator muscle, and the dorsad
inclination of the ectocondyle indicate a powerful muscular control of the arm and
hand, very much greater than is seen in either Dimetrodon or Varanosaurus.

Radius: The radius (fig. 31c), about three-fourths the length of the humerus,
has stout extremities and a relatively slender shaft. The truncated upper end is
subtriangular in outline, with its dorsal angle rounded and its articular surface
gently concave. The posterior surface above is flattened from side to side obliquely
inward; the dorsal surface is strongly convex. The somewhat obliquely truncate

52 PERMn-CARBONlFEROUS VERTEBRATES FROM NEW MEXICO.

distal end is broadly subtriangiilar, with the dorsal angle rounded, the inner angle
more acute. The two sides of the bone are nearly symmetrically concave in outline.

Ulna: The ulna (fig. 31 c) is a longer and stouter bone than the radius, and
is almost or quite the length of the humerus. The olecranon is considerably pro-
duced, and has its distal cartilaginous surface continuous with that of the sig-
moid fossa. Here, again, it is evident that the triceps was unusually powerful and
that its action was exerted at much mechanical advantage. The inner border of
the bone is deeply concave and is thicker than the outer or radial border. The
distal extremity, somewhat less expanded than that of the radius, has its truncated
articular surface for union with the carpus oblique to the long axis of the bone.
Its inner angle is convex, evidently for articulation with a pisiform bone, which,
however, has not been found. The ventral side of the bone is gently concave
longitudinally above, convex distally.

Carpus: The carpus was found in almost perfect articulation on the right side;
the bones forming it on the left side were slightly displaced. On each side the
carpus has been restored so perfectly to its original condition that scarcely anything
more could be desired. The number and arrangement of the bones are quite as in
Dimetrodon and also as in Varanosaurus, except that the first centrale and the fifth
carpale are well ossified in Ophiacodon. The radiale is a short, thick bone, very
convex in front. The intermedium is elongate and helps to form a half of the
articular surface for the ulna. The ulnare is also elongate and rather narrow; its
inner margin curves ventrad. The perforating foramen is rather large, formed
by the ulnare, intermedium, and second centrale. Of the five carpalia the fourth
is the largest, as usual, the third is next in size, the fifth is small, and the second
is not much larger. The first centrale is intercalated between the radiale, second
centrale, and first three carpalia. The second centrale is larger, articulating with
all the bones of the carpus except the first and fifth carpalia. The carpus, as
articulated, is convex from side to side on the dorsal side, concave on the palmar.
The articulations of the radiale, intermedium, ulnare and second centrale are all
close, permitting little motion between them. The articulations of the first centrale
and all the carpalia are loose, not only with each other, but also with the bones of
the first row, especially so on the radial side of the wrist; the free surface of the
first centrale is broader behind than in front.

Hand: The bones of the right hand were all found in intimate articulation
with each other and with the carpus, save the distal two phalanges of the thumb,
which doubtless were lost in collection. Of the left hand the distal phalanges of
several fingers could not be found, but the proximal phalange of the thumb was in
position, giving the complete structure of the hand, with the exception of the ungual
phalange of the first digit. The thumb metacarpal is, as usual, small and broad;
the second, third, and fourth metacarpals increase markedly in length, but not
nearly so much so as do the corresponding ones of Varanosaurus. The fifth meta-
carpal is about two-thirds the length of the fourth ; it is slender and curved, and was
evidently divaricable in life; possibly, as in many lizards, it was freely separable
in life to the base of the metacarpal. The ungual phalanges differ markedly from
those of other known American Pelycosauria or Theromorpha in their less curved
and less jwinted form. They are more like nails than claws, almost flat on the
under side, somewhat thinned at the extremity, but not at all pointed. The animal
could have made no use of them in climbing, or as offensive or defensive weapons.

In figure 30 the right arm and hand are shown almost precisely in the position in
which they were found in the matrix, save that some of the digits have been straight-
ened out and a few of the loosened carpal bones have been more closely adjusted. A

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

53

plaster cast of the arm and hand was made before removal from the matrix. After
removal from the matrix the bones were placed in the mold and cemented together
in that position, a position shown to be quite nattiral by the artictdar surfaces. The
divarication and curves of the fingers have been given as demanded by the articular
surfaces. The elbow, it is seen, is strongly bent, and it seems certain, because of
the humeral and epipodial articular surfaces, that this flexure was the usual one in
life; complete extension of the forearm would have been impossible.

Pelvic girdle and extremity: The pelvis of specimen No. 650 was somewhat
injured in collection, but enough remains to show its absolute identity in char-
acters and size with that of specimen No. 651, of which only the ilium and ischium
of the right side are missing. The pelvis is distinctly of the true pelycosaur type,
but is especially characterized by the massive sutural union of the pubes.

The ilium (fig. 32 a, il) agrees quite with that figured by Williston in his American
Permian Vertebrates (plate xxxvii, figs. 4, 5) and there tentatively referred to Ophia-
codon. It has a long and slender posterior prolongation, in part wanting in the present

Fig. 32. — Ophiacodon mirus Marsh, X ^•

A, left pelvic bones, from without.

B, the pelvis, from above.
il, ilium; pu, pubis; is, ischium.

specimens. In the articulated skeleton the bone
lies somewhat obliquely, with the posterior ex-
tremity elevated. This posterior process is ver-
tical and thin on the posterior part ; anteriorly it
has a transverse thin shelf or plate, jutting inward. The articular surfaces for the
sacral ribs are at the anterior end of this shelf, partly in front, partly below, forming
an oval excavation and showing a strong union with the ribs. The sutural borders
for union with the pubis and ischium are strong and massive, meeting a little
below the middle of the acetabulimi in rather more than a right angle, that for
the ischium a little longer than that for the pubis. On the inner side the convex
thickening which joins the posterior end of the pubic symphysial thickening limits
the brim of the true pelvis, the false pelvis extending far in advance.

The pubis (figs. 32 and 34 b) projects strongly forward, with a heavy lateral
bar, curved somewhat inwardly. On the margin of these bars, on the outer side at
about the middle, there is a thin, everted pectineal process. The two pubes meet
in a very thick and strong median symphysis very unlike that of either Dimetrodon
or Varanosaurus, where the general shape of the pubis is nearly the same. The two
articulated bones extend far in front of the brim of the true pelvis, as a sort of
shallowly concave, nearly horizontal trough, supported on its sides by the thick-
ened bars. Back of the broad symphysis the two bones together form a concave,

S4

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

rather deep spout leading into the narrow and deep concavity formed by the ischia.
The perfect condition of the two pubes of specimen No. 651 permits an accurate
coaptation of the whole pelvis, an upper view of which is shown in fig. 32 b. The
true pelvis, it is seen, is narrow, the greatest diameter of its brim being scarcely
more than an inch, and its depth is less than 2 inches. And this is the outlet
through which the eggs of a creature nearly 6 feet in length must have passed.
The obturator foramen pierces the pubis in its usual place. Between the pubes
and the ischia, as articulated, there is a diamond-shaped pubo-ischiadic vacuity
of moderate size, as in all the true pelycosaurs.

The ischium (fig. 32) is of the usual shape, an elongated, hatchet-shaped bone
meeting its mate in an acute angle below. Its upper posterior border is thickened and
everted, and there is an angular excision between it and its articulated mate behind.

What the significance of the peculiarly elongated and horizontal expansion of
the pubes is we can not suggest; that it has something to do with the habits of the

Fig. 2,^.-0 phiacodon mirus Marsh, X 14- A, left femur, ventral surface; B, left femur, fibular side; C, right
femur, ventral side; D, right femur, dorsal side; E, right femur, tibial side; F, left fibula, ventral side;
G, left fibula, tibial side.

animal seems probable. In the present genus the massive symphysis between the
pubes indicates a firm and strong pelvis, one able to sustain a considerable weight.
In Limnoscelis the massive symphysis extends through both pubes and ischia; in
Seymouria the pubic symphysis is weak, while that of the ischia is deep ; in both
Dimeirodon and Varanosaurus the symphyses of both ischia and pubes are slight.
Femur: The femur is relatively stout, its extremities considerably expanded.
The digital fossa extends a full third of the length. The trochanter stands out
prominently, but ends abruptly, not continuing into the adductor ridge. Below
the trochanter, on the fibular side, there is a distinct roughening for the attach-
ment of muscles. The tibial condyle projects strongly inward from the longitudinal
axis of the bone, its articular surface looking inward at an angle of about 45 degrees.
As usual, the fibular condyle has much the larger articular surface, which is con-
nected by a narrow band on the ventral side with that for the tibia. Neither
surface can be seen broadly from the dorsal side, indicating a normally much flexed
knee in life. The left femur (fig. 33 a) has a remarkable anomaly : On its fibular
side (fig 33 b) the proximal articular surface is broad and has a distinct rim separat-
ing it by a considerable interval from an elongate oval cartilaginous surface at the

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

55

extremity of a thin, elongate process directed ventrad and laterad. This surface
doubtless corresponds to the outer end of the cartilaginous surface of the proximal
end, though separated by a thin, non-cartilaginous interval ; and it is very much
more protuberant. The trochanter on the tibial side is unusually produced ventrad ;
its lower border also is situated further down the shaft, and its extremity is smaller.
In other respects this femur does not differ from its mate.

Tibia (fig. 34 a) .• The tibia is rather more than four-fifths the length of the femur.
It is a rather stout bone, with its upper extremity much expanded. The articular
surface for the tibial condyle is elongate and crescentic in shape, separated by a
ridge and groove from the rather large surface of an oval shape which articulates
with the posterior side of the fibular condyle of the femur.

Fibula (fig. 33 p and g) : The fibula is a little longer than the tibia. It has a
rather slender shaft, and the lower end is much expanded. The upper end is mcd-

FlG. ,^4. — OphiarodoH mtrus Marsh, X H- A, left tibia, dorsal
surface; B, right pubis, inner surface; C, right foot, dorsal side:
D, right tarsus, ventral side.

erately expanded, with a subcrescentic articular surface
placed at an angle of about 60 degrees with the plane of
the lower end. Its articular surface fits well the elongate, curved articular border
of the femoral condyle, sliding, in articulation, forward and inward and backward
and outward; that is, in the extended condition of the leg, the plane of the distal
extremity is horizontal, while in the much-flexed condition this plane is turned
obliquely. The outer border of the bone is nearly straight, the inner border deeply
concave to the lower end. The truncate border is gently curved on the inner side
for articulation with the astragalus.

Foot: The right foot of specimen No. 650 (fig. 34) lay in the matrix with most
of its bones in close articulation. The first three digits were doubled under the
others, but had all their phalanges attached; one claw, only, was lost in the collec-
tion of the specimen. Of specimen No. 651, the left foot was nearly as perfect, the
calcaneum and centralia missing.

The tarsus has the same general structure as have the tarsi of other American
zygocrotaphic reptiles, except that there are two free centralia instead of the single
one hitherto known. The calcaneum is broad and flat, subcircular in outline, with
its free borders thin. The astragalus is rather more elongate than usual, and its
tibial facet is separated rather widely from the fibular one. The perforating fora-
men between the two bones opens near the distal end of the articulating borders.
The two centralia were found quite in position, as were all the other bones of the

S6

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

tarsus save the first two tarsalia, which, with their respective toes, had been pulled a
little distance away from the others. They articulate between the astragalus and the
first two tarsalia and have their free surfaces in front much less extensive than those
behind, indicating the possibility of considerable over-extension of the inner toes and
corresponding tarsalia upon the proximal ones. The fourth tarsale is, as usual, con-
siderably the largest, and the fifth is the smallest. In fig. 34 c the tarsus is shown
in the obhque position of life; in fig. 34 d it is shown from behind, lying horizontally,
the surface somewhat convex. This convexity is that in which the bones were found
in the matrix, and the position is that assumed in articulation; evidently it was the

normal articulation of the foot in life, and it is doubtful
whether the foot could have been straightened out in the
same plane with the bones of the leg.

The curvatures of the toes, as shown in fig. 34 c, are
quite those which the metatarsals and phalanges assume
in close articulation with each other. The first meta-
tarsal is, as usual, short and broad; its proximal articu-
lar surface is of considerable extent and is convex, both
dorsoventrally and transversely, showing conclusively
that the toe was capable of much over-extension as well
as of much free lateral movement. On the other hand,
the third and fourth metatarsals, of much greater length,
fit closely at their proximal ends with their respective
tarsalia, and were not capable of much movement, either
laterally or dorsoventrally. The second metatarsal, of
considerably greater length than the first, had also much
freedom of movement in both directions, as had also
the fifth metatarsal. The distal end of the astragalus
has a much greater extent of surface than the sum of
the opposing surfaces of the centralia, indicating a con-
siderable arthrodial action between them. The ungual
phalanges are more nail-like than claw -like; they are
thin and flattened, and even somewhat dilated at their
tips. They are scarcely curved and are not pointed.

The foot was broad and flat, one well adapted for
soft or uneven ground, but quite unfitted for climbing
or running. The posture of the leg shown in the restor-
ation is clearly a natttfal one. The articular surfaces are
sharply and clearly defined.

This tarsus of Ophiacodon, the most generalized
known among reptiles, may be compared with that of Treniatops, a contemporary
temnospondylous amphibian, as shown in the accompanying figure, alter Williston
(fig- 35)- The astragalus, it is seen, corresponds quite in its relations with the com-
bined tibiale, intermedium, and fourth centrale of the amphibian, with the perfo-
rating foramen at the junction of the intermedium and centrale with the fibulare,
precisely as in the carpus. The first and second centralia of the reptile likewise cor-
respond with the first two centralia of the amphibian, each supporting its respective
tarsale. The third centrale plus fourth tarsale of the amphibian tarsus also occupies
the place of the enlarged fourth tarsale in the reptile.

Note by Williston. — The tibia in figure 35 is shown precisely as it lies in the matrix
in articulation with femiu" and tarsus. In the original figure the fibular side of the tibia
is shown, in the belief that the bone had suffered rotation. Such rotation, however, did

Fig. iS-^Tremalops milleri Will-
iston. Left hind leg, ven-
tral side. Reduced.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 57

not occur, since the side shown in the present figure corresponds quite to the ventral side
as determined since in various other forms. The foot lies nearly in the position as regards
the leg that is shown in the present figure; possibly in life it was angulated more to the
tibial side, leaving more room for the fibula, which is unknown. The arrangement of the
bones of the tarsus agrees well with the observations of Baur on Archegosaurus, who, how-
ever, remained in doubt as to the presence of a fourth centrale.

Concerning the homologies of the tarsal bones in the reptiles, there has been not a little
dispute, and the subject is by no means yet at rest. Gegenbaur, in 1864, considered the
astragalus of mammals as a composite formed by the fusion of the tibiale and the inter-
medium, and this view is the one usually accepted at the present time. Baur, however,
(Morphologisches Jahrbuch, xi, 468), after a careful examination of the embryos of numer-
ous reptiles and mammals, reached the conclusion that the astragalus is a single element,
with no evidence of fusion; that, in the Prototheria and Eutheria at least, "ein Intermedium
tarsi niemals embryologisch nachweisbar ist." Among certain marsupials only, and as
an occasional anomaly in the human foot, a small bone is found wedged in between the
astragalus, tibia, and fibula at the back part of the ankle joint, which Bardeleben believed
to be the intermedium, but which Baur was disposed to consider a neomorph or sesamoid.
In the belief that the astragalus represents but a single bone, which he was inclined to
consider the intermedium, Baur recognized in the "tibial sesamoid" the real tibiale. This
"tibial sesamoid" is a bone not infrequently found in mammals, but unknown in reptiles,
on the tibial side of the tarsus, articulating with astragalus, navicular and internal cunei-
form. Baur gave four hypotheses: First, that the tibiale is represented by the tibial
sesamoid, the intermedium by the marsupial sesamoid of Bardeleben; second, that the
tibial sesamoid is the tibiale, and the astragalus wholly the intermedium; third, that the
tibial sesamoid is the first tarsale, the marsupial bone the intermedium, and the astragalus
is the tibiale; fourth, that the tibial sesamoid and the marsupial bones are real sesamoids,
and the astragalus is the combined tibiale and intermedium. Although Baur recognized
the possibility of four free centralia in the amphibian foot, he did not recognize the possi-
bility of more than one in the reptilian tarsus, since none is known in any recent reptile,
nor more than one in any reptile hitherto. It seems to me that the presence of two free
bones between the astragalus and the inner tarsalia in the present genus offers an explana-
tion of the "tibial sesamoid," if it be necessary to derive the bone from a true tarsal element.
The first of these centralia occupies the position of the sesamoid bone in the mammals,
nearly, and it is not at all impossible that its presence may have been continuous in that
phylum from which the mammals arose.

In a previous paper (Amer. Permian Vertebrates, p. 44) I have expressed the opinion
that a free intermedium tarsi is never pre.sent in adult reptiles as a distinct bone, as based
upon the fact that in all the earliest-known reptiles, from the Middle Pennsylvanian to
the Trias, there are but two bones in the proximal row of the tarsus, bones corresponding
quite to the astragalus and calcaneum. Nor can I find in any later reptile, either extinct
or living, any certain evidence of its presence. We may safely say that if the intermedium
once had disappeared in the ancestral forms it never reappeared as a normal functional
element in their descendants. Any other interpretation of the known facts would require
the existence of two distinct phyla of vertebrates, beginning nearly as early as the Missis-
sippian, and continuing without break to the present time — one in which the intermedium
had disappeared, the other in which it was persistent, but of which we have yet found no
certain representatives.

Schmalhausen (Anat. Anzeiger, 1908, p. 378), from the study of pig embryos, reached
the conclusion: "dassbei den Vorfahren der Saugethiere mindestens eine Dreizahl der
Centralia Tarsi vorhanden war; das eine proximale ist jetzt im Astragalus mit dem Inter-
medium zusammen eingeschlossen, das zweite ist im Naviculare erhalten, und das dritte
kann, wie bei Schweine, mit dem Cuboid verschmelzen." The tibiale, according to him,
has disappeared, or is represented by the vestigial sesamoid. On the other hand. Broom
declares that the astragalus is composed exclusively of the tibiale, because he finds in
Oudenodon a small bone occupying the position of a vestigial intermedium, possibly corre-
sponding to the bone found in the marsupials.

58

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

It is generally assumed that the fourth tarsalc of modem mammals is a compound
bone formed by the fusion of the fourth and fifth tarsalia. There is no embryological
evidence of such fusion, and with SewartsefT (Anat. Anz., 1904, p. 483) and Baur I believe
that the fifth tarsale has entirely disappeared in all modern vertebrates. It is an actual fact
that in some Permian reptiles the fifth tarsale is gone, not fused, but disappeared as a bone.
I am not one of those who must find in the human skeleton distinct ossificatory or even
embryological evidence of every lost bone that has existed in ancestral forms. In some cases
it is quite certain that bones do fuse; the temporal and frontal bones of mammals, to say
nothing of the opisthotic, etc., are sufficient evidences of the fusion not only of cartilage
bones, but membrane bones as well. But, in many cases, it is equally evident that bones
actually disappear. Possibly they remain as potential or latent elements, as mere germs,
which under stimulation in later forms may develop sporadically into a semblance of their
original forms, such as some of the accessory or anomalous bones found in the human
carpus and tarsus, the centrale carpi or the cuboides secundarium for instance. And yet.
it would seem tliat we are going a long way back, to the Trias or Pennian perhaps, to find
the normal functional prototyp)es of some of these germs. Indeed, if we must find the
functional antecedents of all the anomalous bones found in the human carpus and tarsus
we shall have to go back quite to the Stegocephalia, if not the fishes. It seems to me more
probable that many of such anomalies are to be explained by duplication, by the splitting
or division of normal germs, as we know is quite possible.

RESTORATION.

The accompanying restoration, made with painstaking care (fig. 36), and the
mounted skeletoji as shown in the photographic illustration (pi. i, fig. 3) have very
little that is conjectural or even doubtful. The spines of the posterior lumbar,

^^^IS

Fig. 36. — Skeleton of Ophiacodon mirus Marsh as restored. About one-tenth natural size.

sacral, and the basal caudal vertebrae are unknown ; they are shown in the drawing by
barred outlines ; in the mounted skeleton they are restored in light-colored plaster-of-
paris. Enough caudal vertebras from different parts of the tail are known to render
it certain that it was very much like that of Varanosaurus, that is, slender and nearly
as long as the presacral region. It was doubtless composed of about fifty-five ver-
tebrae, of which about twenty are wanting. The chevrons are represented by frag-
ments only. The scapula is placed very nearly as the left one was found lying in the
specimen. The right scapula lay immediately over the left one, save that it had been
pushed downward a short distance. The curvature of the vertebral column in the
cervical and dorsal region as found, and as shown in pi. i , fig. 2, which is a photograph
of a cast of the specimen as it lay in the matrix, is perhaps a little exaggerated, though
doubtless normal, since all the vertebrae lay in immediate contact and without notice-
able displacements. The curves, however, are probably not those often assumed in
life ; they have been reduced a little in the figure. The skull in the mounted skeleton

i'^'

I . Oplu,icodon mirui Marsli. Ventral ribs, natural size.

z. Same. Specimen 650 University of Chicago, as lying in original matrix,

3. Same. Skeleton as mounted. About one-tenth natural size.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 59

has been left in matrical association with the cervical vertebrae. The ventral ribs are
figured as nearly as possible as they were found on the left side, the under side as the
skeleton lay in its matrix, filling up most of the open space below the ribs in front
of the pelvis and extending forward quite to the hind end of the interclavicle. It is
evident that there was no cartilaginous sternum in life back of the coracoids.

HABITS AND RELATIONSHIPS.

The relatively large size of the skull of Ophiacodon is very conspicuous in the
restoration. It is quite a third of the length of the body to the base of the tail,
and doubtless more than a sixth of the entire length of the creature. It was very
narrow and high, and had an inconsiderable weight in life, notwithstanding its
length, composed as it is of very thin and delicate bones. The nostrils and eyes
were relatively small, though quite large enough for the body. The legs are short
and stout, with broad feet and flattened ungual phalanges ; and the tail was slender
throughout. Without taking into account the character of the claws, it is evident
that Ophiacodon was neither a swimming nor a burrowing animal; its tail would
have been of no use in the water in propulsion ; and the front legs were not nearly
long enough to excavate a hole for the head to enter. And it is also apparent
that Ophiacodon was not a swift-moving reptile. Doubtless, like so many of its
congeners, it spent its life about the flat marshes and low plains, feeding upon
such small reptiles and amphibians as it could capture — which did not require
much speed, since nearly all were sluggish creatures— possibly varying its diet
with soft-bodied invertebrates. Its long, slender and recurved teeth were well
adapted for the capture of slippery creatures, but, with the jaws, they were too weak
to withstand much struggling of strong-bodied prey.

Aside from Theropleura, the relationships of Ophiacodon are not very intimate
with any known reptiles. That it should be located in the same suborder with the
true Pelycosauria seems altogether reasonable, notwithstanding the paired tem-
poral vacuities and the holocephalous ribs. The pelvis, especially, has the very
characteristic elongation of the pubes, projecting far in advance of the true pelvic
brim ; and the puboischiadic vacuity is as in Dimetrodon and Varanosaurus. The two
sacral vertebrae, a not very profound taxonomic character, allies the form with
the Poliosauridse. The relatively short spines, of uniform length, are like those of
Varanosaurus and its allies, though, unlike them, the anterior ones are thickened,
doubtless for the better support of the elongated head. That the extraordinarily
elongate spines of Dimetrodon is a very important taxonomic character does not
seem as probable as formerly, since there is now but little doubt that the spines
of Sphenacodon are not unlike those of Ophiacodon, though the skull and dentition
seem to be quite like those of Dimetrodon.

It is very probable, if not certain, that the genus Theropleura, from the Texas
beds, is a closely allied, if not identical genus. The vertebrae, Hmb bones, and such
parts of the skull as are known, as figured by Case (Cam. Inst. Wash. Pub. No. 55,
pis. 3 and 13) all resemble the corresponding parts of Ophiacodon closely. Mandi-
bles and maxillae from the Craddock bone-bed in Texas, mentioned by Williston
as probably belonging in a new genus,* are also very like the same parts in Ophia-
codon, save that the teeth are more compressed.

If Ophiacodon and Theropleura are eventually found to be closely allied but
distinct genera, it may become advisable to separate them into a distinct family,
the Ophiacodontidae. For the present they may both find a place in the family
Poliosauridae.

• American Pennian Vertebrates, p. 75.

6o

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

A DESCRIPTION OF SCOUOMUS PUERCENSIS, NEW GENUS AND SPECIES.

By S. W. WiLLisTON AND E. C. Case.

A new genus of pelycosaurian reptiles is represented by some remains (No. 656
University of Chicago) found by Dr. v. Huene in the vicinity of Arroyo de Agua on
the Puerco in New Mexico. The specimen consists of a complete left humerus, the
two ends of the right humerus, the proximal end of an ulna, a nearly complete left
femur, the proximal end of its corresponding tibia, a centrum, and a few fragments.

The genus is especially characterized by the unusual divergence of the planes
of the proximal and distal ends of the humerus. Usually, among Paleozoic reptiles,
the plane of the proximal end diverges from that of the distal at an angle of from
45 to 60 degrees, though in some cases the two planes are at nearly right angles
with each other. In the present specimen the proximal end is turned outward
much more than a right angle, perhaps 125 degrees, to such an extent, indeed, that

G

Fig. ij.—Scoliomus puercensis, X K- A, left humerus, radial side; B, left humerus, ventral side; C, left
humerus, dorsal side; D, left ulna, proximal end; E, left femur, ventral side; F, left tibia, proximal
articular surface; G, left tibia, proximal end from in front; H, a dorsal centrum, left side.

the inner side of the proximal end is broadly visible from the dorsal side when the
distal end is lying horizontal, whereas in all other humeri known to us from the
Permo-Carboniferous the inner side is always more or less visible in this position.
The entepicondylar foramen is large; it is situated rather low down on the bone,
and is separated by a heavy bar from the inner margin. The femur is unusually
slender. It differs from the femora of other known Permo-Carboniferous reptiles
in the broad separation of the articular faces for tibia and fibula. The specimen is
doubtless somewhat depressed, notwithstanding which the projection of the dorsal
part beyond the articular surface for the fibula is unusual, and seems to indicate
that, in natural articulation, the leg must have been constantly and considerably
flexed upon the thigh. The upper end of the tibia shows the same broad separa-
tion of the articular facets for the leg bones. The trochanter is, for the most part,
unfortunately wanting in the specimen. The linea aspera is distinguishable as a
slightly rugose line, and is not at all prominent.

The single centrum preserved has an unusual character in the distinct but
small parapophysial facet near its front margin. The diapophysis is not prominent.
The centnmi is sharply keeled below.

CHAPTER VI. ^

A DESCRIPTION OF CERTAIN COLLECTIONS OF BONES REFERRED TO

SPHENACODON MARSH.

By E. C. Case and S. W. Williston.
Sphenacodon feroz Marsh.

Marsh, Am. Jnl. Sc, vol. xv, 1878, p. 410.
Baur and Case, Trans. Am. Phil. Soc.. vol. .\.\, 1894, p. 4.
Case, Publication No. 55, Cam. Inst. Wash., 1907, p. 67.
Williston, American Permian Vertebrates, 191 1, p. 78.

Marsh's original description of this genus is as follows :

" In the present genus the anterior teeth are somewhat like those of the reptile described
above, but the p)osterior, or more characteristic ones, are totally different. The crowns
are much compressed, and have very sharp cutting edges without crenulations. In the
present species the carnivorous teeth are crowded together, and the crowns placed slightly
oblique, and twisted. The jaws were comparatively short and massive. The rami of the
lower jaws were apparently united by cartilage only, and the symphysis was short. The
vertebrae are deeply biconcave.

" Meastxrements from the type of this species are as follows:

" Length of the dentary bone 150 mm.

Space occupied by the teeth '. 130

Extent of four anterior caniniform teeth 25

Extent of twenty compressed teeth .' 105

Height above jaw of second lower tooth 15

Dcprh of dentary bone at symphysis 26

Height of crown of corapres-sed tooth 8

Transverse diameter 4

"This reptile was about 6 feet in length, and carnivorous in habit. Its remains are from
the same locality in New Mexico that yielded those of Nothodon."

Marsh made no attempt to classify this animal further than to remark on its
carnivorous habits. Baur and Case regarded it as belonging to the Clepsydrop-
sidae, and Case compared it directly with Dimeirodon. These conclusions were
drawn from the imperfect lower jaw which forms the holotype of the genus and
species. Williston assembled much more material in the Yale collection, and
associated several bones which he regarded as doubtfully belonging with the genus ;
he also was unable to distinguish the genus positively from Dimetrodon. The
collections made by the expedition to New Mexico in the summer of 191 1 added
much new material, but more decisive information as to the characters of the
genus has been obtained from the Yale collections, which have been further worked
out since the time of Williston's studies in 19 10, and which have been attentively
studied by him the present season. The material obtained by the Chicago-Michigan
expedition may be described as follows:

This material consists of three lots of bones found in close association, together
with numerous isolated bones, all from the banks of Poleo Creek in Rio Arriba
County, New Mexico. The associated lots are :

First, ten dorsal vertebrae, eight of which are matrically connected, five of
them with spines; and a separate coracoid. From the Miller quarry.

61

62 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

Second, an imperfect skull, including a nearly complete lower right ramus,
lacking only the articular region ; the anterior part of the left ramus, almost identi-
cal with the fragmentary holotype ; the imperfect right maxilla with part of the
nasals ; a less perfect left maxilla ; the external process of both pterygoids ; both post-
orbitals ; the lower border of the left orbit ; a complete left quadrate ; and a fragment
of the right quadrate. With this skull were closely associated many other bones,
including two complete vertebrae, three less complete, two clavicles, and a femur.
From the Miller quarry.

Third, a poorly preserved skeleton, of which two dorsal vertebras, part of the
right maxilla, and a fragment of the scapula have been cleaned. This specimen
was found isolated, embedded in a soft gray sandstone.

Other bones regarded by us as belonging to this genus are two scapula-cora-
coids, an imperfect pelvis, an axis, atlantal intercentrum, and three imperfect
maxillae, all from the Miller quarry.

Skull: It is not surprising that Case, in his first description of this genus,
referred it to Dimetrodon, since, with all the fragments of the skull before us,
representing a goodly portion of its structure, we are unable to discover a single
character that would separate the form generically from that genus. It is prac-
tically only in the vertebrae that a distinction is possible. Williston has pointed
out that the number of teeth in Sphenacodon is less than in Dimetrodon; he counted
twenty-four in the holotype mandible; in the right mandible before us there are
twenty-two, and, allowing for the presence of one anterior to the enlarged teeth
and one at the posterior end, there would be exactly twenty-four. In the left
mandible, making allowance for four which had been broken away from the upper
margin, and for one in front of the enlarged tooth, and one at the posterior end,
there are also twenty-four. The maxillae are all imperfect, but they certainly had
less than twenty teeth each. Williston counted sixteen in a maxilla associated with
the holotype mandible. In the specimens before us the largest number is fourteen,
with a possible maximum of three more, making seventeen. Case, from a study
of numerous specimens of Dimetrodon, concluded that the number could not have
been far from twenty-eight in the mandible and twenty in the maxilla. None of
the teeth, as mentioned by Williston, are crenulate. In one specimen the bases of
the large teeth are quadrangular in outline, as described by Case for Theropleura.
The upper margin of the diastemal notch, anterior to the enlarged maxillary tooth,
is nearly on a line with the alveolar border of the maxilla ; that is, it does not ascend
as in the more specialized members of Dimetrodon (e.g., D. gigas). In other char-
acters, so far as the material we have studied goes, the skull of Sphenacodon can
not be distinguished from that of Dimetrodon, and certainly none of these characters
can be considered of generic value.

The premaxillcB in two specimens show but two large teeth in each, the anterior
one much larger than the second. Williston has figured a specimen from the Yale
collection with a third, posterior tooth, much smaller than the others.

The maxilla (fig. 38) has two enlarged teeth, which seemingly alternated in func-
tion, as in one of them the anterior tooth is the larger and the following one .shows
evidence of incomplete development. In another specimen the second tooth is the
larger. This seems to bear out the conclusion expressed by Case that the large teeth
in Dimetrodon alternated in function — that a single one performed the main work
while the other was maturing, although in many specimens the two teeth are of
nearly equal size. In the skull under description there are no teeth in the notch
anterior to the enlarged tooth, but in other specimens in the collection, and also in a
specimen in the Yale collection, as figured by Williston, there are two small teeth.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

63

The nasals are too incomplete to describe fully. They were narrow and seem
to closely resemble the same bones in Dimetrodon, indicating that, as in that genus,
the skull was narrow and high in the facial region. The cast of the top of a skull
observed in a large boulder in the field shows the same character.

The mandible (fig. 39 a) of the right side is complete, with the exception of
the articular region. It resembles very closely the jaw of Dimetrodon. The only
difference noticeable is the sharp upward bend of the anterior end, causing the
anterior teeth to point backward at a considerable angle with the axis of the jaw.
This is,_ in part at least, due to distortion of the right mandible, since the ramus

Fig. 38. — SphenacodoH ferox Marsh, X yi- A, right maxillary, imperfect; B, inner surface of

fragment of second right maxillary.

of the left side does not show this character to anywhere near the same extent.
The sutures, so far as they can be made out, are identical with those of the Dime-
trodon mandible.

The pterygoids are represented by the external processes only; they are indis-
tinguishable from the same processes in Dimetrodon. A single row of medium-
sized, conical teeth is found on the lower margin.

The lower rim of the orbit, formed by the maxilla and jugal, the postorbital
bones of the two sides, and the quadrate are indistinguishable from the same
bones in Dimetrodon.

Fig. 39. — Sphenacodon ferox Marsh,

A, inner surface of a right ramus; B, quadrate.

There can be no doubt of the great resemblance of the skull of Sphenacodon
and Dimetrodon — so great a resemblance that, on the strength of this part only,
we would not be justified in considering the genera distinct.

Three imperfect scapula are preserved in the collection. One, associated with
a considerable part of the skeleton in a bad state of preservation, is known only
from a fragment of the blade. Another, an isolated specimen, lacks the anterior
edge and has the two proximal bones attached (fig. 40 a). The third, also an iso-
lated specimen, lacks the anterior border and the border of the coracoids. As a
whole the shoulder blade is quite similar to that of Dimetrodon. The sutures between

64

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

the scapula and coracoid (metacoracoid Williston) and procoracoid (coracoid Willis-
ton) are distinct, and the posterior bone is easily separable — there are two isolated
posterior coracoid bones in the collection. The anterior border of the bone is
destroyed in both specimens, so that it is impossible to determine the presence or
absence of facets for the attachment of a clavicle or cleithrum. The proximal end
carries a large preglenoid tuberosity, with an articular glenoid face looking back-
ward and downward. There is no foramen on the posterior face of the preglenoid
tuberosity, such as occurs in Dimetrodon. Below the posterior part of the edge of
the tuberosity there is an indication of a glenoid foramen, and below the anterior
part of the same border is the usual supracoracoid foramen. Both foramina open
into a subscapular fossa on the inner side of the bone, as in Dimetrodon. The posterior
of the two coracoid bones has the same form as in Dimetrodon. The glenoid face looks

Fig. 40. — Sphenacodon ferox Marsh, X }4. A, right scapula; B, left clavicle; C, fragment of left
side of pelvis; D, left side of axis; E, anterior view of axis.

forward, upward, and outward, completing the glenoid cavity identical in form and
position with that of Dimetrodon and Ophiacodon. The anterior coracoid bone, the
procoracoid of authors, is complete in one specimen, but the scapula is broken
away along the line of imion between the two (fig. 40 a). This edge is perfect,
extending almost directly outward from the lower part of the scapula. As a whole,
the proximal margin formed by the two coracoid bones is nearly straight, except
at the front end.

Clavicles (fig. 40 b) : Associated with the skull are two clavicles. That of the
right side is more perfect than that of the left; its proximal end is wide and flat.
and evidently articulated with the interclavicle, as in Dimetrodon. The distal end
is narrower, and there is a deep groove on the lower edge which received the border
of the scapula. The anterior border is deeply concave longitudinally. The whole
bone is diflferent from that of Dimetrodon; in that genus the proximal end is wider
and the distal end narrower, without a groove on the lower edge for the scapula.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 65

Pelvis (fig. 40 c) : The pelvis is represented by a portion of the left half, show-
ing the cotylus and the outer part of the pubis. A much better specimen of the
Yale collections is shown in fig. 43 d.

The cotylus is rather narrow and deep, with prominent edges on all of the bones;
in this it differs from Dimetrodon, in which the edge of the cotylus on the pubis is
not elevated. This permitted the femur of Dimetrodon to be extended much more
directly forward than in Sphenacodon, in which it must have stood almost directly
out from the side of the pelvis. The ilium is too incomplete for any detailed descrip-
tion ; the crest was broad, differing notably from that of Theropleura, which is narrov/,
resembling the ilium of Varanosaurus. That it was firmly attached to the sacrum
is shown by the prominent rugosities on the inner side. In Dimetrodon there is a
narrow process on the proximal edge of the ilium, which forms a buttress against
which the end of the femur fitted; in Sphenacodon the whole edge is elevated,
though there is a narrow portion on the posterior part which carries the face for
the femur. The outer edge of the pubis is thickened ; it runs directly forward from
the cotylus and is slightly deflected at the distal end. The opening of the pelvis
was rounder than in Dimetrodon, in which it was distinctly angular at the lower
part. The pubic foramen perforates the bone at a point just below and anterior
to the cotylus.

The vertebral column is represented by an atlantal intercentrum, an axis,
several vertebrae associated with the skull, and ten others closely associated to-
gether. The intercentrum is broad below and of good size. Its anterior face is
deeply concave, both vertically and horizontally, indicating a well-rounded con-
dyle. Its posterior face is smaller and more shallow; the posterior edge of the
lower surface has a shallow notch in the middle. The presence of atlantal ribs is
indicated by small facets on the upper part of the posterior margin. The axis
resembles that of Dimetrodon much more closely than that of Ophiacodon or Thero-
pleura. In the latter the spine is short and the distal end is fully as long, antero-
posteriorly, as the rest of the spine ; the anterior and posterior margins are nearly
vertical, with no projection of the lower part of the anterior end forward over the
atlas. The axis of Sphenacodon (fig. 40 d, e) has an elevated neural spine, the
lower part of which extends forward, terminating in a slight rugosity, beyond which
the border slopes upward and backward slightly concave in outline to the rather
wide and triangular apex.

The posterior border of the spine is nearly vertical and for its lower two-
thirds deeply excavated by a narrow groove, which increases in depth downward
until it reaches its maximum at the level of the posterior zygapophyses. The
edges of this groove are sharp, and there is a small but prominent knob or process
just over each posterior zygapophysis. We have never seen such a process on the
axis of Dimetrodon or other pelycosaur. The anterior zygapophyses are small,
but well formed; they look almost directly upward and but slightly outward.
The posterior zygapophyses are relatively large for the size of the axis. The
anterior face of the centrum is elongate and heart-shaped, due to the development of
a sharp keel on the lower surface. On the upper edge of the articular face there are
two large, elliptical, concave faces, centantra, looking slightly inward and down-
ward; they are relatively much larger than any observed in Dimetrodon. The
posterior face of the centrum is circular, with no faces on the upper margin corre-
sponding to those of the anterior face.

Centantra and centrosphenes do not seem to occur on the post-axial vertebrae.
The sides of the centra are deeply concave and terminate below in a sharp keel,
which has a very sharp line on its lower surface. The keel and concavities of the

66

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

sides terminate before the anterior and posterior borders are reached, due to the
wide flaring out of the edges; this flaring is especially prominent in the posterior
face. The transverse process rises from the neural arch and the upper part of the
centrum, and extends outward, backward, and downward, as in Dimetrodon. The
axis is decidedly like that of Dimetrodon, differing only in the tuberosities over
the posterior zygapophyses, the large size of the latter, and the size of the centantra
above the anterior articular face. (Compare figs. 32-35 and 58, 59, Publication
55, Cam. Inst. Washington.)

The ten dorsal vertebrae, eight in connection (fig. 41 a), are from the middle or
posterior part of the series; they all have short, smooth spines, somewhat flattened,
and with the anterior-posterior diameter the longer. The base is narrowed and
somewhat recurved, above which the spine expands gradually to the upper end and
terminates in a flat face. They are easily distinguishable from the vertebrae of
Ophiacodon, in which the spines are less dilated and thickened above. We observe
no rugosity on the sides of the spines just above the base, such as occurs in

Pig. 41. — Sphenacodon ferox Marsh, X yi. A, dorsal vertebrae; B, anterior view of a dorsal vertebra;
C, lateral view of same vertebra shown in fig. B.

Dimetrodon, and which doubtless marks the attachment of the dorsal muscles
and the beginning of the thin cutaneous covering of the spines. The spines are
relatively much higher than in Ophiacodon or Theropleura. That there was
some variation in the height of the spines is indicated by the fact that of the
five vertebrae with spines attached, one, the last, has its spine notably shorter
than the others.

The short transverse processes rise from the base of the neural arch and extend
directly outward ; the upper side of the process is thickened and convex ; the lower
is very narrow and does not pass obliquely downward to join the anterior border
of the centrum, but is set off from it sharply, indicating dichocephalic ribs. As
there is no face for the rib on the anterior edge of the centrum in these specimens,
nor any on the ends of the single dorsal intercentrum preserved, it seems probable
that the capitulum articulated, as usual, with the intercentral cartilage. The
sides of the centra are concave and there is a sharp median keel. The articular
faces are elongate, and, due to the presence of the keel, narrowed below.

Two vertebrae (fig. 41 b, c), probably from the anterior dorsal region, have
much higher spines and even more prominent transverse processes than those
of the series of ten. These stand out almost directly from the sides of the neural

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

67

arch, with the broad upper surface on a level with the anterior zygapophyses. It
is evident that these vertebrae closely resemble those of Dimetrodon, especially the
form described as D. giganhomogenes by Case, but the faces of the centra are rela-
tively more elongate and the longer transverse processes do not dip so far down-
ward, as is common in all species of Dimetrodon. As compared with Theropleura,
which Sphenacodon resembles in some respects, the centra are much narrower
below and the sides are concave, the faces of the anterior dorsal vertebrae are
more elongate, and the spines are higher and narrower.

It is apparent that Sphenacodon was closely related to Dimetrodon. The skull
is, so far as can be determined from known material, almost identical; the few
resemblances to Theropleura which have been pointed out are all of a primitive
character. We regard Sphenacodon as a less specialized member of the group of
dimetrodonts, which had not yet acquired those habits, whatever they were, which
led to the excessive development of the spines and the more elongate and slender
limbs. The position of the femur, directed nearly straight outward, indicates a
more sprawling habit of body. Perhaps the animal had not yet so thoroughly
adapted itself to the dry-land habitat which seems to be characteristic of the
specialized, long-spined forms.

Fig. 42. — Sphenacodon ferox Marsh, No. 818, Yale University, X %■ A, basioccipital, intercentra, and
odontoid seen from the side; B, left side of an axis; C, anterior view of a posterior cervical vertebra;
D, lateral view of a dorsal vertebra; E, anterior view of a posterior cervical vertebra.

Note by Williston. — The foregoing material collected by us in New Mexico is of great
interest, but leaves in doubt several important problems as to the relationships and struc-
ture of this genus; this doubt is partly, not entiiely, dissipated by the better collections of
the Yale Museum, made by Baldwin so long ago. Williston had the privilege of study-
ing this material, though not as ftilly as he desired, the past summer. Several specimens,
which must be referred to the genus Sphenacodon, have been thoroughly prepared, and
settle beyond dispute the general characters of the genus, though it is by no means
certain that they all belong to the same species as that originally named by Marsh from
material from the Baldwin quarry. The most important and best preserved of these speci-
mens is No. 818 in the Yale collections, coming from the " Rito Puerco." It was embedded
in a red sandstone, and the bones are somewhat crushed, and many fragments are missing.
It includes the two innominata, scapula, humerus, part of skull, axis, about twenty verte-

68

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

brae, apjiarently in a connected series, a few caudal vertebrae, and an incomplete femur.
Figures of many of these bones, carefully made by Williston, are given in figs. .^2-45 and will
render an extendctl desbription unnecessary. The vertebral spines, it will be seen, are all
of nearly uniform length, though they shorten somewhat toward the axis; they are dilated
and thickened above, rounded and stout below. On a number of the posterior ones
(fig. 43 A, B, c) there is a facet, it will be observed, for the capitulum, an unusual thing in
the Pelycosauria. The axis agrees with the specimen previously described. The nearly
complete innominate bone (fig. 43 d) is indistinguishable from that of Dimetrodon.
The scapula (fig. 44 a) shows only minor differences in comparison with an unusually
well-preserved scapula of Ditneirodon, collected by Mr. Hatcher on Coffee Creek, Texas, and
preserved among the Yale collections (fig. 44 b). The sutiiral lines between the scapula
and coracoids are shown very clearly in both specimens. In the specimen of Sphenacodon
the preglenoid facet is imperfect. The humerus (fig. 45 a) also shows an amazing resem-

FiG. 43. — Sphenacodon ferox Marsh, No. 818, Yale University, X %■ A, B, C, lateral view of three
posterior dorsal vertebrae; D, right innominate bone.

blance to that of Ditneirodon* The femur of this specimen is imperfect; a perfect femur
from another specimen, agreeing well with this incomplete one, except that it is slightly
smaller, is shown in fig. 45 b.

The sacrum is quite that of Dimetrodon, so far as the incomplete specimens show; it
has three vertebras. The parts of the skull of this and other specimens in the Yale col-
lection indistinguishable from it agree very well with the ones referred to the type speci-
men, except in size.

As has been stated, there seem to be minor differences in the length and characters of
the spines among the material in the Yale and Chicago collections that might indicate
specific variations. Possibly, when the complete anatomy of the genus is known, as in
Ophiacodon, these differences may justify its separation into two or more closely allied
genera. The size, typically, is smaller, and no such broad-ended spines as those figured
herewith as belonging with specimen No. 8i8 of the Yale collection have been discovered
in the Baldwin quarry, whence came the holotype specimen of Marsh ; though on the other
hand, this collection does contain spines too long for Ophiacodon, though thinned above.
But these details are unimportant at the present time. The one thing now absolutely

• It was chiefly on this hurtierus, together with the incompletely prepared spines of this specimen,
that I reported in an earlier paper the occurrence of Dimetrodon in the New Mexican fauna. As a fact,
there is not a single specimen in either collection which can now be referred with any probability to the
genus Dimetrodon. I think it may be definitely said that the genus does not occur in New Mexico.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

69

determined is that Sphenacodon is a typical pelycosaur, very closely allied to Dimetrodon,
with only minor differences in skull, girdles, extremities, and vertebral centra, but with
very much shorter spines, which are not pointed at their extremities, but, on the other hand,
are dilated antero-posteriorly. The spines are long, extraordinarily long for a reptile, but
they could not have formed any such frill as that of Dimetrodon or the basilisk lizard.

And this structure of the vertebra has an irnportant bearing in any disaxssion as to the
meaning of the spines in Dimetrodon. That Dimetrodon could have developed such extra-
ordinary* spines without affecting to a greater degree the characters of the skeleton proves
conclusively their relative physiological unimportance.* Certainly, had the enormous
dorsal expansion of Dimetrodon been of profound importance in the life economy of these

Fic. 44. — Sphenacodon ferox Marsh. A, right scapula, No. 818, Yale University, X h;
B, right scapula of a large Dimetrodon from Coffee Creek, Wilbarger County, Texas,
No. 661, Yale University, X Vi.

creatures, it must have materially afTected the structure of the skeleton elsewhere. That
Sphenacodon is more primitive than Dimetrodon must be admitted to be perhaps another
bit of evidence of the greater antiquity of the New Mexico deposits than the upper ones,
at least, of Texas.

* Case would hesitate to indorse the statement that these elongated spines were physiologically unim-
portant. He has long considered (and frequently stated his belief) that the enormous development of the
spineii in Dimetrodon and probably also in Edaphosaurus (Naosatirus), imposed upon the creatures a ph\s-
iological burden so great in its demands upon the energies of the individual, both for their original
production and the repair of frequent injuries, that it was an important, if not the chief, cause of their
extinction.

7°

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

With differences that can hardly be accounted of generic value elsewhere in the skeleton,
it would be idle to claim for the greater spinous expansion of Diinetrodon a family value in
their classification. Whence it follows that in the strict application of the laws of priority,
the family name Sphenacodontida (Marsh, May 3, 1878) must take precedence over Clepsy-
dropid4B (Cope, May 8, 1878), provided no question of scientific rectitude on the part of
its author is raised.

Fig. 45. — Sphenacodon ferox Marsh, X %. A, ventral suru^c ui ^ right humerus.
No. 818, Yale University; B, ventral surface of right femur, X H-

CHAPTER VII.
A DESCRIPTION OF EDAPHOSAURUS COPE.

By S. W. Williston and E. C. Case.

The detailed history of the genus Edaphosaurus has been given by Case *
and need not be repeated here. Briefly, the genus was described by Cope in 1882,
based upon a fairly complete but crushed skull and a single vertebra, the axis.
The characters presented by this specimen were deemed of sufficient importance
by Cope to warrant the erection of the family Edaphosauridae, in which, however,
he erroneously associated the cotylosaurian genus Pantylus Cope, a genus having
somewhat similar palatal teeth, but differing widely in other respects. In 1884 he
referred a second species, E. microdus Cope, to the same genus, based upon some
fragments of the plates bearing the teeth and a series of vertebrae. Two years
later, however. Cope transferred the latter species to his new genus Naosaurus,
where it has since remained.

Until recently no additional specimens have been referred with certainty to
Edaphosaurus, and the identity of the two genera, Edaphosaurus and Naosaurus, has
been in doubt. Six years ago, however, Case discovered in the Texas deposits
typical Naosaurus vertebras associated with a mandibular dental plate, which
pretty nearly convinced him of the identity of the two genera, as shown by the
following quotation from his cited work, page 45:

" If this association is a true one, as seems certain, the name Naosaurus must be given
up, as it is preoccupied by Edaphosaurus, and the subfamily Naosaurinae will disappear,
and the members will be placed in the family Edaphosaurida;. It may seem that there is
undue hesitancy in uniting the two genera on the evidence cited, but to any one familiar
with the occurrence of bones in the Texas beds the possibility of accidental association is
so evident that the greatest conservatism seems the best course. * * * To me it seems
extremely probable that the two genera must be united."

As will be seen. Case was quite right in his opinion; the evidence is now posi-
tive that the two genera are closely aUied if not identical, and the name Naosaurus
must be abandoned, unless it should be found that the typical species N. claviger
Cope is generically different from Edaphosaurus in ways that are yet unknown.
That the genus belongs in the family Edaphosauridae is, in any event, definitely
proven. In some ways it is unfortunate that the identity or close relationship
of the two genera was not more strongly insisted upon by Case; it might, perhaps,
have prevented the incorrect restorations of Naosaurus which have gained currency
in textbooks and popular works. The following, to the best of our present knowl-
edge, is the correct synonymy of the genus and species:

EDAPHOSAURIDA.

Cope, Proc. Amer. Phil. Soc., xx, 1882, p. 450; ibid, 1883, p. 631; Pal. Bulls. 35, 36. Case, Revision
of the Pelycosauria, 1907, 68.

Bdaphosaubus.

Cope, Proc. Amer. Phil. Soc., xx, 1882. p. 448. Case, Revision of the Pelycosauria, 1907, p. 69.
Naosaurus Cope, Amer. Nat. xx, 1886, p. 544; ibid, xxi, 1878, p. 319; Proc. Amer. Phil. Soc.
' XIV, 1878, 44 (Dimetrodon). Case, Revision of the Pelycosauria, 1907, p. 58.

• Revision of the Pelycosauria, Cam. Inst. Washington, Publication 55, 1907, p. 34.

71

72 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

BoAPiiosAURUS — Continued.

Edaphosaurus pogonias Cope, Proc. Amer. Phil. Soc., x.x, i88j, p. 448; Trans. Amer. Phil. Soc, xvii,

1892. p. 15, pi. xi, ff. 5, 5a. C;vsc, op. cit., 1907, pp. 69, 151, pi. xxxiv.
Edaphosaurus microdus Cope, Proc. Amer. Phil. Soc., X-\ll, 1884, p. 37.

Saosaurus microdus Cope, Amer. Nat., .xx, 1886, p. 544; Trans. Amer. Phil. Soc, xvi, pp. 287,
295. Case, op. cit., 1907, p. 61.
Edaphosaurus claviger Cope.

Naosaurus claviger Cope, Amer. Nat., xx. 1886, p. 544; Trans. Amer. Phil. Soc, xvi, 1888, pp.
287, 293, pis. ii, iii. Case, op. cit. pp. 59, 139, pis. xxviii, xxix, xxx, 1907.
Edaphosaurus cructger Cope.

Dimelrodon cruciger Cope, Amer. Nat., xxi, 1878, p. 829; Proc. Amer. Phil. .Soc, xtx, 1880, p. 44.

Naosaurus cruciger Cope, Amer. Nat., xx, 1886, p. 544; Case, op. cit., 1907, pp. 60, 146.
Edaphosaurus novomexicanus Williston and Case, infra.

Until the material described in the present paper was discovered, the type
specimen of Edaphosaurus pogonias preserved in the American Museum was the
only one of the genus confidently known. This specimen, unfortunately, was so
crushed in fossilization that its true characters were very difficult to determine;
attempts to restore it by Case and Broom were only in part successful, as will
be seen.

In the summer of 1910, Mr. Paul Miller was so fortunate as to discover, on
Moonshine Creek, in Baylor County, Texas, probably from the lower Clear Fork

Fig. 46. — Skull of Edaphosaurus ? pogonias (Cope) Williston and Case. No. 625. University of Chicago,
X Ji. A, from the left side, a, section of the jaw; B, from above; C, from behind, ang, angular; /r,
frontal; /, lachrymal; m, maxilla; n. nasal; pa, parietal; pfr, prefrontal; pof, postfrontal; pi. pterygoid;
sq, squamosal; sur, surangular; po, postorbital; oc, occipital condyle.

beds, an excellent skull referable with certainty to Edaphosaurus, but whose pre-
cise specific determination is impossible at the present time. From E. pogonias it
seems specifically dififerent in the proportions and shape of the bones of the upper
side, and, as the skull is quite unknown in the other described species, it may be a
long while before its specific identity is certainly proven. In much probability
.some one or another of the three species, E. claviger, E. cruciger, and E. microdus,
is identical with E. pogonias, and the giving of a new specific name to the present
specimen, while temporarily convenient, would in the end probably encumber the
synonymy to a still greater extent. This species, until more is known of the allied
ones, may be known as E. ? pogonias Williston and Case. Comparisons of the
specimen will be made in the discussion of the following species, E. noz'omexicanus,
the skull of which seems more nearly like that of the type.

The skull (fig. 46) as a whole is very high, narrow, and short, subquadrilateral
in shape, less deep in front, with the nasal region convex. Its greatest width above

PERMO-CARBONIFEROUS VBRTEBRATES FROM NEW MEXICO. 73

is between the orbits, due to the projecting, overhanging, and somewhat upwardly
curved roof of the orbits, which is convex on its margin and thinned. The parietal
region is narrower, but certainly less narrow than in the genotype specimen, and
is concave transversely. The parietal foramen is situated rather far back. In
front of the orbits the skull rapidly narrows triangularly, nose-like, and is convex
in outline, terminating in a thin, median, prominent ridge, which descends con-
vexly between the closely approximated nares. The side of the skull is conspicuous
for the large, oval, single temporal vacuity, lying immediately back of the orbit,
but higher posteriorly, and narrowed in front. It has the thickened lateral margin
of the parietal above, the thickened and short postorbital bar in front, the convex
border of the parietal arch behind, and a rather slender temporal bar below. The
quadrate, covered by a temporal bone, descends far below the arcade posteriorly,
leaving a large open space in the articulated skull between the temporal bar and
the upper margin of the mandible. If any quadratojugal bone is present in the
arcade it must occupy an anomalous position. There is but a single temporal
arch and a single temporal opening, in the formation of which it is quite evident
the squamosal bone must participate, and into which it seems quite as evident
that the quadratojugal does not enter. If this vacuity be, as is urged by Huene
and Broom, the lateral temporal vacuity, and not the upper one, the relations
of the bones have become sadly mixed. In all probability the temporal bar is
formed exclusively by the squamosal and jugal, and it is also not at all improbable
that the squamosal reaches toward, if it does not actually join with, the postorbital.
By this we do not mean to say that we believe the opening is the upper vacuity,
but simply that the attempt at present to differentiate the upper and lateral vacui-
ties in the reptilian single-arched skull is premature and unreliable.

The orbit is subrotund in shape, a little longer in the oblique diameter from
above downward anteriorly. Its roof is very prominent, projecting horizontally
and a little upward, its border gently convex. Below, it is separated by a short,
slender bar from the vacuity above the mandible in the articulated skull. The
face in front of the orbit is somewhat cpncave on its upper part. The external
narcs are small, round in shape, situated close to the front margin at a considerable
distance above the alveolar margin.

Posteriorly the skull is deeply concave between the parietal projections. On
either side there is a concavity above the stout suspensorium, which articulates
with the quadrate a little above the level of the temporal arch. Doubtless there is a
post-temporal vacuity between the parietal bar and the paroccipital, but it can not
be distinguished in the specimen. On the right side, lying back of the downwardly
curved parietal process, and reaching inwardly to the middle line, there is a long,
rather slender, and curved bone, which seems to be distinct.. If so, it must corre-
spond to that figured and described by Case as the " ? epiotic" in the type specimen
of the genus. Broom, however, doubted its presence. If the bone be really distinct,
and it seems to be, it can hardly be the tabulare, inasmuch as we know of no verte-
brate skull in which the tabularia meet in the middle line. Its relations seem to be
more those of the postparietal ; but even this interpretation is open to doubt.

The mandibles, as seen from the side, are very broad posteriorly. The tooth-
border is nearly straight. From the rounded and not at all prominent coronoid
angle the thin upper border descends obliquely to the articulation for the quadrate.
In front, the two sides meet in a very firm symphysis, the front border of which
is thinned and convex in outline, descending below the mandibular border back
of it, where the mandible is the narrowest, its width not half that oppo.site the
coronoid elevation.

74 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

The skull, while in most excellent preservation so far as its shape and absence
of distortion or loss of parts are concerned, will not permit the determination of
the sutures with any degree of certainty. Certain lines, by their symmetry on the
two sides, seem in much probability to indicate sutures, but it is quite impossible
for any one to say with certainty that they really do so without the corroboration of
additional material. These lines are indicated by dots; those of the face and upper
side seem very probable; the others are more doubtful (fig. 46).

Dentition: On the right side of the specimen the maxillary teeth are all, or
nearly all, shown clearly; those of the mandibles are more obscure. There are
four premaxillary teeth, if the division between the maxilla and premaxilla is
where it is assumed to be, below the back part of the naris. Back of these four
teeth there are definitely nine or ten on the maxilla in front of the orbit. These
teeth are not nearly so broad at the base as they were figured by Case ; they do not
differ much in size, though the median ones are the longest and largest. Back of
these teeth there are four or five, possibly more, extending at least as far as the
middle of the orbit; they are of smaller size. Broom criticizes Case's interpreta-
tion of the maxillary teeth, but his own is no better; there are not nearly so many
as he figures, and they extend much farther backward.

In the closely articulated position in which the mandibles are preserved in
this specimen it was impossible to ascertain the most essential characteristic of the
genus, the palatal and mandibular teeth. In order to determine their extent and
position a section nearly 2 inches in length was smoothly sawed from the right
mandible; and this section enables one to determine with certainty, not only the
presence of such teeth as were described in the genotype, but also their position.
In this section of the mandible as separated from the pterygopalatine, numerous
round teeth are seen, as also those of the opposing surface, the two sets attached
to their respective bones in a plane divergent from the vertical by about 45 degrees,
as shown in the figure of the front end of the section (fig. 46 a, a). The palatine
and splenial teeth have hitherto been supposed to be placed more nearly horizontal.
On the outer side, the stout and rounded upper border of the dentigcrous ptery-
goid plate is seen just above the coronoid, where it joins the jugal. From this
junction the border is directed backward and downward for more than two-thirds
the distance to the articulation of the quadrate with the mandible. The apparent
fusion of the pterygoid with the jugal would suggest the presence of a transverse
bone. So far as the teeth are visible in the section broken away from the ptery-
gopalatine plate, they seem to be quite like those described in the following species.
The narrow space between the mandibles below does not permit the preparation of
the palatine region, especially so as posteriorly the region is more or less occluded
by a fragment of the vertebral column.

Edaphosaurus novomexicanus, n. sp.

A somewhat crushed specimen, discovered by Dr. v. Huene in the vicinity of
Arroyo de Agua, New Mexico, may be referred provisionally to a new species.
Although somewhat fragmentary, the specimen is of importance as furnishing for
the first time indubitable evidence of the relationship of Edaphosaurus and Naosau-
rus. The specimen consists of the skull and a connected series of fifteen vertebrae,
together with the two incomplete scapulae, the right clavicle and a part of the left
one, the left cleithrum in articulation with the scapula, a number of ribs, a humerus,
and a radius. Some of the vertebral spines had been weathered out of the matrix,
and the pieces into which they were broken can not be reunited, because of the
absence of many.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

75

The skull is about as complete as the genotype, with the exception of the man-
dible, of which nothing has been detected. Like that specimen, it is crushed nearly
flat. By careful labor the broken parts have all been reunited, and the whole has
been carefully cleaned of its investing matrix ; unfortunately some fragments were
not recovered. The skull is shown in fig. 47, as carefully drawn by Williston.
The upper side resembles in its proportions the original type better than it does
that of the skull described on the preceding pages, especially so in the relative
widths of the interorbital and interparietal regions.

On the palatal side the two large plates bearing the teeth are nearly complete,
that of the right side quite so. About seventy-five teeth are visible on this side;
possibly there are a few more. The teeth vary not a little in size. They are
hemi-ellipsoidal in shape. The more newly erupted, smaller ones have the apex

Fig. 47. — Edaphosaurus navomexicanus Williston and Case, X }4- A, skull
from above; B, skull from below; bo, basioccipita! ; bs, basisphenoid;
ep, epipterygoid; po, paroccipital; pi, pterygoid; q, quadrate.

somewhat pointed, but most of them are worn a little at the top. They are inserted
in pits, thecodont, or protothecodont. The anterior part of the plate is angulated
somewhat from the larger posterior part, about in the place where Broom thought
the palato-pterygoid suture occurs; possibly the angulation indicates a suture,
but that can not be determined. The maxillary teeth extend back to about the
middle of the pterygo-palatine teeth, and beyond the middle of the orbit. There
is no evidence of a posterior palatine vacuity ; there is certainly none in the place
figured by Broom in his restoration. The region in front of the middle is injured
in the specimen, and its structure can not be made out; there are no vomerine teeth
visible.

Posteriorly the basisphenoid has a deep median fossa, on either side of which,
behind, there is a descending basisphenoid process; the distal end of the stapes is
visible. The quadrates are visible in large part They are shaped much like those
of the true pelycosaurs, consisting of a thin, expanded plate directed obliquely
forward and inward, articulating on the outer side with the lower part of the squa-
mosal or quadratojugal, on the inner side by a large and firm surface with the

6

76 PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

posterior end of the pterygoid, the posterior lower edge of which extends to within
8 mm. of the condylar margin. The condyle is transverse, the inner side descend-
ing lower than the outer, with a trochlear groove in the middle. The posterior
part of the pterygoid, beyond the dentigerous plate, is a thin, flat, or gently con-
cave process, articulating at its extremity with the quadrate, as described. On
the right side, lying on the outer side of the quadrate expansion, which is here
exposed, there is a smooth rod, expanded above, where it lies in juxtaposition with
the parietal; more slender below, its lower end doubtless dislodged from its original
position above the dentigerous plate. It must be the epipterygoid.

The occipital condyle is gently convex, subcordate in outline, with the usual
notochordal pit in the middle. On either side of the foramen magnum there is a
projection having, apparently, an articular facet, doubtless for articulation with a
proatlas. On the side of the skull there was a slender temporal bar like that of
the skull described in the preceding pages, the most of which was broken away
and the pieces lost. Its free, thin border curves downward posteriorly to very
near the condylar margin of the quadrate, with which the bone articulates. Broom's
conjectural restoration here is incorrect ; the free border of what he supposed was a
foramen in the arch is merely the lower border of the temporal arch ; and there are
no indications on the smooth surface here of a suture distinguishing a quadratojugal.

From the two specimens of skull figured herewith it will be seen that the
conjectural restorations of Edaphosaurus were both incorrect ; in some details that
of Case is the better, in others that of Broom. The orbit is not as large as Broom
figures it; the quadrate descends much lower; the single temporal vacuity is of
a different shape; there is no foramen in the temporal arch; the whole skull is
narrower and higher, and there is no palatine vacuity. Broom was probably more
nearly correct in the sutures of the face; Case in those of the frontal region. Case
was also apparently correct in the detection of the bone in the occipital region
called by him " ? epiotic." The sutures everywhere need corroboration, and
otherwise should not be accepted as certain.

VertebrcB (fig. 48) : Fifteen vertebrae were found in a continuous series articulated
with the skull, and their matrical attachments have been preserved in preparation.
The posterior three are more or less fragmentary, and the spines of the posterior
eight, which were exposed when discovered, are more or less wanting. Only a
fragment of the atlas has been detected. As in most of the Pelycosauria, the first
six of the vertebrae, counting the altas, may be considered as cervicals. With the
seventh a much heavier rib was articulated, and it doubtless was the first thoracic
vertebra. As in the different species known as Naosaurus, the cervical vertebrae
are remarkable for their decrease in size, as also for their much shorter ventral
lengths. To such an extent, indeed, are these vertebrae shortened on the under
side that closely articulated the skull would be directed obliquely downward and
backward. The twelfth vertebra, of much larger size, has its articular ends nearly
parallel with each other, while those of the first two or three vertebras are inclined
in such an angle that the lower border is a third shorter than the longitudinal
diameter on the floor of the neural canal, and is but little more than half the length
of the lower border of the twelfth vertebra. Throughout the cervicals the lower,
concave border is sharply keeled ; further back, the lower side is more obtusely
keeled. The anterior diapophyses, beginning with the axis, are more slender, but
as long as those further back. In the cervical region they are directed downward
and backward; posteriorly they are more transverse in position and are situated
near the front end of the vertebra. The zygapophysial surfaces of the cervicals
are extensive and nearly horizontal in position. The zygapophyses of this region

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

77

project nearly as much as do those of the dorsal region, notwithstanding the
smaller size of the vertebrae. Like the centra, they evidently permitted not only
extensive vertical flexion, but a considerable lateral one as well.

The axis has a rather narrow spine, not expanded anteroposteriorly above,
and it is but little longer than the height of the centrum. It is thin in front, much
thickened and rugose behind, and is directed obliquely forward. The precise
length of the spine of the next following vertebra can not be determined. It is
flattened at its base, cylindrical, slender, and evidently pointed above; its extremity
is directed forward at an angle of about 45 degrees with the long axis of the centrum.
The spine of the fourth vertebra could not be connected with its centrum, and
its length is, consequently, unknown. That of the fifth vertebra is complete, or

Fig. 48. — Edaphosaurus navomexicanus, X yi. A, vertebrae as numbered, No. 674, University of
Chicago; B, rib of eighth or ninth vertebra.

practically so. It is directed upward and forward at a little less angle than the
preceding ones; it is gently curved below, nearly straight above, and is obtusely
pointed at its extremity. It is oval in cross-section above its flattened and rugose
base, and has no lateral tubercles. The next spine is stouter, but is preserved only
as far as the extremity of the one preceding it, with which it lay in close matrical
contact above; it must have been at least 3 inches longer than that spine. It has
two or three tubercles on its sides, in the form of rounded protuberances, pro-
jecting about a fourth of an inch. The sixth spine is still stouter, but only about
5 inches of it can be connected with the centrum. It is directed less obliquely
forward, and the single pair of tubercles that this portion bears are larger. On the
tenth, the base of the spine, about 3 inches in length, was recovered; it is directed

78

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

vertically. Among the many pieces of spine recovered of the posterior vertebrae,
but few of which could be connected with their vertebrae, the lateral tubercles
are more numerous and rather more prominent, but none could have been more
than three-quarters of an inch in length.* A few of the intercentra are preserved
with the vertebrae, one of which is figured, as seen from below.

Large quantities of broken ribs were secured, but only one rib could be restored
completely or nearly completely ; it belongs with either the eighth or ninth vertebra.
Its shape and size will be sufficiently well understood from the figure (fig. 48 e).
The rib seems remarkably heavy for a skull of the size of the present one.

Pectoral girdle: Both scapulae are preserved, imperfectly. That of the right
side is more complete on the lower part, that of the left on the upper. The latter
has a nearly complete cleithrum in close articulation with it. The right clavicle

Fig. 49 — Edaphosaurus novomexicanus. Pectoral girdle, X %.
c, cleithrum; cl, clavicle; sc, scapula.

was found lying across the left scapula, and is complete. No indications of an
interclavicle have been observed in the material. The right scapula is figured,
with some slight additions, as also the cleithrum, reversed from the left side (fig. 49).
The cleithrum is larger than has been observed in other pelycosaurs. It is a rather
stout cylindrical bone, lying in close articulation along the front edge of the scapula
above, thinned below for union with the clavicle, but not dilated at either extrem-
ity. The scapula is peculiar in its short, narrow, rather stout blade, and its great
antero-posterior extent below. Both scapulae are more or less crushed, and were
doubtless less flat in life than shown in the figure. The large supraglenoid foramen
pierces the bone back of the ridge leading to the preglenoid facet, not the outer
face, as in Dimetrodon. A comparison of these scapulae with that figured and de-
scribed by Case f leaves little or no doubt of their generic affinity (fig. 50). Case
was especially struck by the short blade of his specimen, and the great antero-

• Among the collections made by Mr. Baldwin from Poleo Creek, New Mexico, now preserved in the
museum of Vale University, is a considerable fragment of a large spine, closely resembling those of " Nao-
taurus," that has elongated lateral processes.

t Journal of Geology for May 1903, page 397.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 79

posterior extent below, characters unlike those of other scapulae that have been
observed from Texas. It undoubtedly belongs with some one of the species of
Edaphosaurus described from Texas.

Humerus, radius: The humerus and radius of the right side, found lying closely
related to the scapula, are so poorly preserved that only their chief characters can
be made out. The humerus is more slender than usual; the entepicondylar foramen
appears to be larger. It measures 130 mm. in length, with the least diameter of
the shaft 14 mm. The radius is 80 mm. in length, or less than two-thirds the length
of the humerus. In Dimeirodon the radius is more than four-fifths the length of
the humerus.

Fig. 50. — Edaphosaurus sp. Scapula-coracoid of a specimen found
on Coffee Creek, Texas. No. 186, University of Chicago, X }^-

CONCLUSIONS.

It will be observed that the cervical spines of the present species differ mate-
rially from those figured and described as belonging with "A^ao^oMrMS," in their
greater anterior inclination, more slender and pointed shapes, and in the smaller
development of lateral tubercles so characteristic of the genus. It is quite possible
that these differences are of generic value, and that Naosaurus will eventually be
found to be a valid genus. But this can not be determined until the type species of
Edaphosaurus has been definitely correlated with its proper skeleton. It is more
probable that the New Mexico species will prove to be the aberrant form, requiring
a new generic name. If these differences really exist in the Texas species, it will be
quite time enough to validate the name Naosaurus when the facts shall have been
determined. For the present, at least, Naosaurus has no claims for existence.

In E. novomexicanus, with the neck extended horizontally, the pointed spines
of the anterior vertebrae would project over the occiput. If the neck were fully

8o

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO.

flexed, with the articular surfaces of the centra in contact or parallel, the head
would be folded under the breast, the pointed cervical spines projecting directly
forward. That such a posture was possible seems certain; that it was an habitual
posture of the animal in life, either of offense or defense, it would be hazardous to
say ; and yet one can not resist the belief that it was.

The function of the greatly elongated spines in both this genus and Dimetrodon
has been the subject of no little speculation and some very fanciful theories. There
is little evidence that the spines were covered in life by a homy epidermis. They
of course could not have projected freely without a nutrient covering of some kind.
If, on the other hand, they were enveloped in a continuous covering, there could
have been little or no motion of the vertebral column in a vertical plane, since even
the slightest angle of divergence of the centra would have increased the distance
between the spines enormously at the periphery. The looseness of the cervical
vertebrae is almost certain evidence of the possibility of considerable flexion and
that the spines throughout most of their extent could not have been connected

Fig. 51. — Edaphosaurus sp., X K- No. 3333, University of Michigan. A, frontal;
B, basioccipital, upper surface; C, basioccipital, lower surface.

by ligaments, even elastic hgaments. They must have been freely movable in the
cervical region at least, probably invested by a firm skin.*

It is a curious fact that no genus of vertebrates is more commonly met with in
the Texas Permian deposits than Edaphosaurus; isolated and fragmentary bones are
found everywhere ; a day seldom passes that the collector does not see one or more
Edaphosaurus vertebrae. It is also very rarely that he finds two or more verte-
brae in anatomical relationship, and hitherto there has been no record of the posi-
tive association of any other parts of the skeleton, except the pelvis. A legitimate
conclusion from these facts would seem to be that Edaphosaurus was an inhabitant
of the river plains and not of the more quiet bays and mud fiats. The skeletons
were broken up and brought down to the lower lands by streams and freshets.
Williston has figured and described f a leg of a reptile that he suspected belongs
with Edaphosaurus. It is peculiar in having a relatively long and slender femur and
short tibia and fibula, characters invariably associated with aquatic habits in
extinct reptiles. If this leg really belongs with Edaphosaurus, and there seems to
be a similar relation between the humerus and radius, it would add to the prob-
ability that the species of Edaphosaurus were more or less swimming, fluviatile
animals, living along the river plains and subsisting upon fresh- water moUusks and

• Case has shown that in Chameleo cristatus the elongated spines of the dorsal region are united by a
flexible skin, with a strong thread of connective tissue running along the apices of the spines. A few scat-
tered fibers of muscular tissue were found between the spines. This suggests the condition that existed in
the long-spined pelycosaurs. (Case, Science, xxix, p. 979.)

t American Permian Vertebrates, p. 75.

PERMO-CARBONIFEROUS VERTEBRATES FROM NEW MEXICO. 8 1

other invertebrates. Whatever may have been the use of the cervical spines in
offense or defense, a Hke explanation will hardly account for the greatly elongated
dorsal ones.

The relationships of Edaphosaurus and Dimetrodon, notwithstanding the simi-
larity of their dorsal spines, can not be very intimate. The conclusion is irresistible
that these resemblances were, for the most part at least, merely the result of con-
vergent evolution. We know now that there were other widely different animals
in the New Mexican Permian and the European Trias with greatly elongated dorsal
spines. All of which render the meaning of their remarkable development still
more of a problem.

Note by Case. — Further confirmation, if such be necessary, has been added to the
identity of Edaphosaurus and Naosaurus by the discovery, during the summer of 1912, by
Case of a nearly complete vertebral column of Edaphosaurus (Naosaurus) associated with
scapula, pelvis, limb bones, and a broken and fragmentary skull indistinguishable from the
type specimen of Edaphosaurus pogonias. Figtue 51, a, e, c, shows the occipital region and
frontal plate of this skull. It is of interest to add that the humerus possesses an ectepi-
condylar foramen as well as an entepicondylar foramen.

'liHShW

QE Case, Ermine Cowles

8iil Penno-carboniferous

C373 vertebrates

P&ASci

PLEASE DO NOT REMOVE
CARDS OR SUPS FROM THIS POCKET

UNIVERSITY OF TORONTO LIBRARY

llKit.-:

''■'^L

<, ^