LIBRARY OF THE

UNIVERSITY OF ILLINOIS

AT URBANA-CHAMPAIGN

560.5
FI
V.I6

y

UNIVEHSITY OF

IlllNOIS LIBRARY

AT URBANA-CHAMPAIGN

GEaOGY

The person charging this /"^ .e^al is re^
sponsible for its return to the hbrary from
Xch it was withdrawn on or before the
Latest Date stamped below.

Theft, mutilation, and underlining of books

are reasons for disciplinary action and may

result in dismissal from the University.

c ,M.Kini<; LIBRARY AT URBANA-CHAMPAIGN
UNIVERSITY OF lUINOIS LIBRAKT

OCT f 3 976
own 9 11977

L161 — O-1096

vrr t, «-'

CI

FIELDIANA • GEOLOGY

Published by
FIELD MUSEUM OF NATURAL HISTORY

Volume 16 March 31, 1966 No. 3

The Phylogenetic and
Functional Implications of

The Armor of
The Dissorophidae^

Robert E. DeMar

Assistant Profeksok of Geology, University of Illinois

Chicago Circle Campus

Since the work of Williston (1914), there has been no general work
on the Dissorophidae, except for brief synoptic treatments by Wil-
liston (1916), Case (1946), and Romer (1947). A number of other
authors have considered scattered dissorophid specimens, but have
not greatly altered the general concept of the family.

This paper suggests new concepts of the phylogeny of the Dis-
sorophidae and of the function of their armor. The work has been
made possible by the discovery of a number of new materials col-
lected since 1930 by the Museum of Comparative Zoology at Har-
vard College, and by the University of Chicago.

The writer wishes to express thanks to Dr. Herbert Barghusen
and Dr. James Hopson, who were especially helpful at various times
during the writing of the paper. Dr. Everett C. Olson originally sug-
gested the project and gave repeated useful advice during all stages
of the work. The drawings were made by Mrs. Jane Hubby.

A number of institutions were materially helpful in the loan of
valuable specimens. These institutions include the American Mu-
seum of Natural History, The National Museum, The University of
Texas, Field Museum of Natural History, and the Museum of Com-
parative Zoology.

1 While this paper was in press, Chicago Natural History Museum adopted the
name Field Museum of Natural History. In this paper the catalog designations
remain CNHM.

Library of Congress Catalog Card Number: 66-19172

No. 1007 55 Vil liW"^ '3^" '^'-^

GEOLOGY LIBRARY

SEP 27 1357
OMiVERSlTV Or ii-i-^-^

56 FIELDIANA: GEOLOGY, VOLUME 16

Abbreviations

All abbreviations are explained when used, except for those of
museums. They are :

CNHM — Chicago Natural History Museum

AMNH — American Museum of Natural History

UT — University of Texas

NM^ — National Museum

MCZ — Museum of Comparative Zoology

YPM — Yale Peabody Museum

The Dissorophidae is now one of the largest families of Paleozoic
amphibians in number of genera. It is known from the Westphalian
and Stephanian stages of the Pennsylvanian of Europe and North
America, and from the Lower Permian of North America, and the
Middle Permian of Europe. Because of its early discovery and
the large number of people involved in its description, the history
of the family's taxonomy and phylogeny is long and complicated,
with a number of disputes as to the proper taxonomic value of cer-
tain morphological characteristics. This paper considers the phy-
logeny and taxonomy of the family as developed primarily through
the agency of the morphology of the dorsal dermal ossifications, usu-
ally called armor.

In order to properly relate the present description of the dissoro-
phid armor to the taxonomy and phylogeny of the group, a review of
the history of the classification of the Dissorophidae follows, with spe-
cial emphasis on early ideas on the origin and evolution of the armor.

Original Descriptions and Relationships

The family name, Dissorophidae, was first used by Boulenger
(1902) in an article on amphibians in the tenth edition of the Ency-
clopedia Britannica. He did not define what he meant by the family
or list the genera, except for Dissorophus, which he notes (p. 383),
"is remarkable for an extraordinary exo- and endo-skeletal carapace."

The first described dissorophid was Dissorophus multicinctus,
which Cope (1895, p. 998) assigned to the Amphibia. He commented
that it was "a veritable batrachian armadillo." Cope (1896b) de-
scribed the genus Otocoelus, and for this genus and Conodectes {Sey-
mouria) he erected the family Otocoelidae, which he believed to be
ancestral to the turtles.

Continuing with the same basic framework, Case (1905, 1905a)
reviewed the characters of the order, Chelydosauria (erected by Cope

DEM AR : DISSOROPHIDAE ARMOR 57

in 1898 in a syllabus not available to the writer), and included the
genera, Otocoelus, Conodectes and Diadedes, primarily on the basis of
the enlarged otic notch. This group was supposed to be related to the
Cotylosauria, and to be directly ancestral to the turtles.

Williston (1908, 1909) agreed in principle to this arrangement,
though he dropped the term, Chelydosauria. A measure of the con-
fusion involved in this work previous to 1910 is that all later authors,
including Case (1911, 1946), WilHston (1910, 1914), Watson (1920),
and Romer (1947), consider all described species of Otocoelus and
Dissorophus to be members of a single species, D. multicinctus.

Williston (1910a) described a second dissorophid, Cacops, and
erected the family Dissorophidae for Cacops and Dissorophus, failing
to notice that Boulenger had already used the family name. The
characteristics of the family were not well chosen, however, since he
did not mention the skull shape, and considered such characters as
reduced parasphenoid, miscellaneous limb characters, and number of
teeth to be important at the family level of taxonomy. At least,
he did recognize a family of armored amphibians with a deep otic
notch.

In a later paper, Williston (1911) uncertainly expanded the family
without comment (except for a question mark on a list), by including
Aspidosaurus (Broili) . In the same paper, Williston remarked on the
close relationship of Dissorophus to Cacops, suggesting that they are
just barely generically distinct.

Case (1911) did a very thorough job of discussing previous work
on the Dissorophidae. He expanded the family, naming a new genus,
Alegeinosauru^, and referring specimens previously described as Oto-
coelus or D. articulatus to D. multicinctus. At this time Case also
erected a separate family for Aspidosaurus, the Aspidosauridae. His
description of the Dissorophidae, which is representative of his and
Williston's diagnoses of the time, follows:

Small, not exceeding 500 mm. in length. . . . Occipital condyles separate.
. . . Otic notch represented by a large fenestra. . . . Parasphenoid reduced to a
slender rod. . . . Two functional sacral ribs. . . . Dorsal armor composed of the
laterally expanded neural spines, smooth above and overlain by narrow dermal
plates, as wide as the body and alternating with the neural spines in position
. . . Clavicles and interclavical small, without sculpture. . . . Cleithrum very
large. . . . The two halves of neural arch united. . . . Intercentra thick, con-
stricting the notochordal space. No processes on the intercentra for the ribs.

This description is good, considering the small amount of avail-
able material, though it has some factual errors, i.e., "parasphenoid

58 FIELDIANA: GEOLOGY, VOLUME 16

... a slender rod," and frequently has extrapolations based on a
single genus.

Williston (1914) redefined the family on somewhat better grounds.
He noted the shape of the skull, the bones entering the orbit, and the
nature of the sacrum. The family was not well defined, but this clas-
sification was superior to earlier ones. In this paper (1914) for the
first time he divided the family into two subfamilies, the Dissorophi-
nae and the Aspidosaurinae. This division was based on differences
in the otic notch, which was closed behind in the Dissorophinae, and
open in the Aspidosaurinae; and on the dorsal armor, which he be-
lieved to be double (two armor segments per vertebral segment) in
the Dissorophinae, and single (one armor segment per vertebral seg-
ment) in the Aspidosaurinae.

Williston (1916, p. 202) described the Dissorophidae as follows:

Terrestrial armored amphibians of about one-half meter in length. Skull
short, subtriangular, moderately depressed, without lateral line canals or inter-
frontal or intertemporal bones. Ear notch deep, open behind or closed. Two
occipital condyles. A median row of bony shields over cervical and dorsal
vertebrae. Scapula with large cleithrum. Clavicular girdle small, not rugose.
Mesopodials and pelvis fully ossified. Two sacral vertebrae.

This definition can be considered to be a final statement by earlier
workers on the Dissorophidae. At that time, Williston continued the
division of the Dissorophidae into two subfamilies. One, the Dis-
sorophinae, contained the genera, Dissorophus, Cacops, and Alegeino-
saurus, and was characterized by two armored segments per vertebral
segment and a closed otic notch. The other, the Aspidosaurinae, con-
tained the genera Aspidosaurus and Broiliellus, and was characterized
by one armored segment per vertebral segment and an open otic notch.

Thirty years later Case (1946, pp. 378, 379) followed the same
classification, except that he agreed with Watson (1920) that Zygo-
saurus (Eichwald) from the Russian Permian, should be referred to
the subfamily, Dissorophinae.

Romer (1947, p. 314) discussed the family with particular atten-
tion to the shape of the skull, the depth of the otic notch, and the
terrestrial nature of the life of the animals. He did not mention the
subfamilies used previously, perhaps because it is not his habit to use
subfamilies for amphibians. He suggested that the Dissorophidae
are closely related to the Trematopsidae, as did Olson (1941).

Romer (1947) began the practice of placing certain Pennsylvanian
genera in the Dissorophidae. His family included lArkanserpeton,

DEMAR : DISSOROPHIDAE ARMOR 59

IPlatyrhinops, Alegeinosaurus, Aspidosaurus, Broiliellus, Cacops, Dis-
sorophus, Tersomius, and Zygosaurus.

Gregory (1950, p. 854) placed the Pennsylvanian genus Amphi-
bamus in the family, and following a discussion of that assignment,
he concluded that many other Pennsylvanian genera should also be
included. Gregory conceived the Dissorophidae to be a group of
amphibians with a standard skull roof pattern, with large tabulars,
a large otic notch, relatively large orbits, and a short vertebral col-
umn. He considered the armor to be a specialization of the later
members of the family. In addition to Amphibamus, he suggested
the inclusion of the Pennsylvanian genera Mordex, Limnerpeton lati-
ceps, and Ricnodon limnophyses.

Presumably, he would retain Romer's dissorophid genera, with
the consequence that his family would include twelve to fourteen
genera.

The present concept of the dissorophid armor is therefore rather
ambivalent. There is the original concept of Williston and Case,
which does not include the recently described Pennsylvanian forms,
and which divides the armor into two different types. And there is
also the concept of recent authors who perceive an earlier, unarmored
part of the family, which later evolved into the armored forms. In
this later concept the details of the arrangement of the armor are
not considered.

Generally, the terms used in this paper are those currently in use,
and should not cause confusion, but the vocabulary used for the ar-
mor needs some clarification. The armor consists of segmental ossifi-
cations dorsal to the neural spines and along the antero-posterior
dorsal midline. It has various lateral extents, being sometimes quite
extensive, and sometimes narrowly restricted to the midline. In order
to discuss the armor properly and to relate the armor of one animal
to that of another, the terms "internal" series and "external" series
are used. The internal series refers to armor closest to the neural
spines, whereas external series refers to armor alternating with the
internal series, but just above it. Some animals, such as Broiliellus,
"Aspidosaurus" novomexicanus, and Aspidosaurus have only one
series, the internal series, whereas Dissorophus, Cacops, and Alegeino-
saurus have both sets of armor. Therefore, according to this ter-
minology, all animals with only one set of armor have the internal
series, and only those with a second more dorsal set, have the exter-
nal series. These terms have geometric meaning only, are intended
for convenience in discussion, and have no meaning in homology.

60

FIELDIANA: GEOLOGY, VOLUME 16

one inch
I I

Fig. 1. Diagrams of the armor of Dissorophus and Broiliellus. A, Anterior
portion of the shield of Dissorophus, from a dorsal aspect. B, Anterior portion of
the shield of Broiliellus, from a dorsal aspect. C, Lateral cross-section of Dissoro-
phus, as seen from the left. D, Single segment of Dissorophus, as seen anteriorly.
E, Lateral cross-section of Broiliellus, as seen from the left. Abbreviations: e, ex-
ternal segments of armor; i, internal segments of armor; f, flange on internal seg-
ments of the armor of Broiliellus and Dissorophus; n, neural spine.

In fact, the armor of the Dissorophinae and of the Aspidosaurinae
are not considered to be homologous in this paper. The usage of
these terms is illustrated in figures 1 and 2,

In addition, despite the use of the term "armor," it is not the in-
tention to imply, without discussion, that the function of armor is
wholly, primarily, or even partly, protection, though the possible
function of the armor will be discussed in a later section. The rea-
son for using the term "armor" here, is that previous authors have
repeatedly used it, so that it is nearly a part of the general scientific
vocabulary for dermal ossifications of the sort to be discussed.

DEMAR: DISSOROPHIDAE ARMOR

61

B

one inch
1 I

Fig. 2. Diagrams of the armor of Aspidosaurus and Cacops. A, Anterior por-
tion of the shield of Cacops, as viewed dorsally. B, Lateral cross-section of the
armor of Cacops, as viewed from the left. C, Anterior view of a single segment of
Cacops. D, Anterior view of a single segment of Aspidosaurus. Abbreviations:
e, external segment of armor; i, internal segment of armor; n, neural spine.

Dissorophus multicinctus Cope (1895)

Specimens Examined:

The holotype, AMNH no. 4593, consisting of a part of the verte-
bral column, ribs and dorsal armor, all articulated, from the Permian;
AMNH no. 4376, a skull and anterior part of the skeleton, from the
Permian; CNHM no. 1204, an articulated portion of the dorsal armor
and vertebral column, from Hog Creek; CNHM no. UR 813, articu-
lated part of the dorsal armor and vertebral column from Olson's
Broiliellus pocket'; CNHM no. UC 648, a partial skeleton and skull
from Coffee Creek; NM no. 21904, a partial skeleton, from the Per-
mian; NM no. ?1555, a skull, several articulated partial armor cara-
paces and vertebral columns, and a foot from W. Coffee Creek;
MCZ no. 2122, consisting of four fairly complete skulls, seven fairly
complete armor carapaces and vertebral columns, and numerous frag-

' Olson's Broiliellus pocket has never been mentioned in the literature. It con-
sists of a pocket of vertebrate fossils including Dissorophus, Trematops, and a new
species of Broiliellus, and is located east of Coffee Creek, Baylor County, Texas in
the Arroyo Formation of the Permian. The name "Broiliellus pocket" is used
informally on the specimen labels at Field Museum of Natural History.

62 FIELDIANA : GEOLOGY, VOLUME 16

ments from all parts of the body, from Indian Creek; MCZ no. 1163, a
skull and partial skeleton, and numerous fragments, from Hog Creek.

Distribution:

All of the specified localities are in the Arroyo Formation of the
Lower Permian of Willbarger and Baylor Counties, Texas. Those
without a specific locality designation are probably from the same
horizon but this cannot be verified.

Discussion:

The dorsal dermal armor is the most conspicuous feature of Dis-
sorophus, and was described by Cope (1895, 1896, 1896a). Williston
(1910) and Case (1911) added materially to the knowledge of dissoro-
phid armor. However, more numerous and better preserved materials
afford an opportunity to make superior descriptions and conclusions
at this time.

The general concept of the carapace developed by Williston and
Case is that it is very much like that of Cacops, except that it is much
more extensive. In the most complete carapace at their disposal,
there were "twelve or thirteen vertebrae participating in the shield,"
and they felt that additional elements must have been lost. The
number of participating vertebrae was believed to be fifteen, as in
Cacops, or more.

Except for some details of description, I cannot agree with their
concept of the dermal armor. Because the similarity of the armor of
Cacops to that of Dissorophus is considered to be superficial in this
paper, the dermal armor is completely redescribed. Figures 1 and 3
illustrate the armor of Dissorophus.

The number of vertebrae beneath the shield is uncertain, because
none of the specimens is so well preserved that the vertebrae and ar-
mor plate elements can be counted and numerically related. In order
to arrive at a figure on the number of vertebral segments beneath the
carapace, one has to admit the general assumption that there are two
shield, or carapace, elements per vertebral segment. This assumption
is clearly correct in most instances, but the most anterior segment
may have several, or perhaps only one vertebra beneath it. Also, the
posterior part of even the best preserved carapace may not be com-
plete. The most probable number of segments beneath the shield is
21, although it may be one less or one more, and there may be some
variation among individuals.

Fig, 3. Dissorophus muUicinctus, MCZ 2122. Dorsal view of two armored
carapaces.

63

64 FIELDIANA: GEOLOGY, VOLUME 16

All of the armor segments, except for the first and last, have a
standard shape. They are transversely elongate, with a transverse
dimension of about 7 cm. and an antero-posterior length of about
1 cm. They are about 3 mm. thick, though this measurement is quite
variable. On one specimen of the carapace (MCZ no. 2122), the
transverse width of the armor segment related to the first vertebral
segment is 7 cm.; the seventh (counting two to a vertebral segment)
is 7.5 cm.; the 17th is about 5.5 cm.; and the last or 20th is about
4.8 cm. The total length of the shield is about 29 cm.

The shield, as a whole, reaches its greatest transverse diameter
approximately in the center, and tapers gradually, both anteriorly
and posteriorly. At the lateral margins the shield curves ventrally
so that it is convex upward, along an antero-posterior axis.

As has just been stated, Dissorophus has two pieces of armor per
vertebral segment, with the possible exception of the anteriormost
and posteriormost segments. Members of the external series overlie
members of the internal series so that, from a dorsal aspect, one sees
alternately, from front to back, members of each series. This arrange-
ment is diagrammatically represented in fig. 1.

A typical segment from each series of the armor may be described
briefly as follows: The dorsal surface of each piece of armor in the
external series is deeply pitted, like the surface of the skull. The ven-
tral surface is smooth, with no processes. A typical segment of the
internal series is very much like one of the external series, but on the
dorsal surface, where the internal plates are overlapped by external
plates, the surface pitting is subdued, or absent. More important,
each piece of the internal series has a flange along its mid-ventral
line, that slants ventrally and anteriorly from the armor into the
space between the neural spines, where it comes to a point. No doubt,
the armor was held in place by ligamentous attachments to the neural
spines. There are no co-ossifications of the internal segments of ar-
mor and the neural spines, so the armor segments cannot be thought
to be ossified extensions of the neural spine. (See also, the discussion
of "Aspidosaurus" novomexicanus.)

Before going on to other aspects of the armor, it might be pointed
out that there are several miscellaneous, but interesting, anomalies in
the armor plate. In general, the armor is a single mid-dorsal group,
but on one specimen (MCZ no. 2122) (fig. 3), the first external piece
of armor is present, but only on the right side of the animal. It ter-
minates normally on the right side, but in a blunt point in the midline
of the shield on the left. On the same specimen, a more posterior

DEMAR : DISSOROPHIDAE ARMOR 65

position for external armor is occupied by two, left and right, other-
wise normal pieces of armor.

The anteriormost armor segment is much longer antero-posteri-
orly than any other armor segment, and is peculiar in its relationship
to the neural spines. In most specimens this relationship cannot be
seen, for various reasons. But in Williston's figured specimen of D.
multicinctus, it is quite clear that the neural spines of a few (two or
three) vertebrae are split, and the ventral flange is inserted partly in-
to the split ends and partly between the adjoining neural spines. It is
not possible to give any more details because of the poor preservation
of the specimen. It would be necessary to section or otherwise injure
the holotype to make a complete study. An important matter here
is the insertion of the ventral flange into the split ends of the neural
spine, but it is equally important that only the first anterior segment
is so involved. Also, the anteriormost observed segment of other in-
dividuals is not so inserted, which suggests that this is an exceptional
condition in Williston's figured specimen. If the specimen could be
more satisfactorily studied, possibly the confusion could be cleared up.

The anteriormost piece of armor is difficult to relate to the geom-
etry (serial homology) of the rest of the carapace. Case (1911) and
Williston (1910) thought that it was an enlarged single piece of armor
(presumably of the external series), not having much to do with the
vertebral column underlying it. However, they had little material,
and there is a great deal of variation in the armor segment occupying
the first position.

Apparently, the first segment of armor is the result of fusion of a
number of adjacent armor segments (fig. 1). There is a great deal
more observed variation in the antero-posterior length of the first seg-
ment than in the length of other armor segments. This suggests that
the first segment does not grow gradually, but by fusion with already
present segments.

In further support of this contention, the posterior edge of the
first segment sometimes is swollen dorsally, and rests above the lead-
ing edge of the next posterior segment, which is a member of the in-
ternal series. On the other hand, sometimes the posterior edge is not
swollen, and instead, inserts below the leading edge of a member of
the external series. This is caused by the progressive addition of seg-
ments to the posterior margin of the anteriormost segment. The
situation is well illustrated in a number of specimens (fig. 3) .

On one of the specimens, numbered MCZ no. 2122, there are two
nearly complete shields preserved (fig. 3). In the specimen with the

66 FIELDIAN A : GEOLOGY, VOLUME 16

shorter first segment, the next posterior segment is a member of
the external series and appropriately overlaps the posterior edge of the
first segment. From this it would be interpreted that the posterior
part of the first segment is a member of the internal series. The first
segment is about twice as long, antero-posteriorly, as the armor seg-
ments posterior to it, so it can be concluded that it is composed of
two fused segments. The anterior edge would be a member of the
external series.

The other shield shown in figure 3 (MCZ no. 2122) has a larger
first segment and, unlike the above mentioned specimen, has a slightly
swollen anterior dorsal margin, overlapping the next posterior seg-
ment. From this situation it would be interpreted that the first ar-
mor segment is composed of three fused segments. Starting anteriorly,
these would be a member of the external series, one of the internal
series, and finally one of the external series.

In specimen CNHM no. UC 648, the first segment is even longer
antero-posteriorly and overlaps the next posterior segment. Cross-
ing this enlarged segment are three transverse swellings that appear
to mark the positions of fused members of the external series. It
would therefore appear that this first segment is composed of five
fused segments, three of the external series and two of the internal
series.

In summary, these observations lead consistently to the conclu-
sion that the enlarged first segment of armor of Dissorophus multi-
cinctus is formed by the posteriorward fusion of alternating members
of the external and internal armor series. No specimen is known with
an unenlarged first segment consisting only of the external series, but
specimens interpreted as having 2, 3, and 5 fused segments have been
described.

The only other pecularity of the armor is seen in the posteriormost
armored segment of the vertebral column. This vertebral segment
may not be typical, as it appears to lack the segment of the external
series. However, none of the available specimens is well preserved
in the last segment, and no definite statement can be made.

The armor of Dissorophus is very similar to that of Broiliellus,
except that Broiliellus lacks the external series. One of the most
cogent reasons for believing that the two genera are closely related
lies in the morphology of the armor. On the other hand, the armor
of Dissorophus is not at all similar to that of Cacops, Aspidosaurus or
Alegeinosaurus.

DEMAR : DISSOROPHIDAE ARMOR 67

Broiliellus Williston (1914)

Specimens Examined:

Broiliellus texensis, holotype, CNHM nos. UC 684 and UC 685,
each consisting of a skull, dorsal armor, and miscellaneous limb bones,
from Mitchell Creek. Williston (1914) did not designate which of
these two specimens was to be the holotype. Since UC 684 is a some-
what better specimen, I hereby designate it the holotype (fig. 4) . A
new species of Broiliellus, CNHM no. UR 810, consists of a carapace
from Olson's Broiliellus pocket^ of East Coffee Creek.

Distribution:

Broiliellus texensis is known only from one locality in the Clyde
Formation, Lower Permian of Archer County, Texas. The new spe-
cies is known only from Olson's Broiliellus pocket of the Arroyo For-
mation, Baylor County, Texas.

Discussion:

Broiliellus texensis was originally described by Williston (1914) on
the basis of two specimens, both of which had well preserved dorsal
shields of segmented armor. His description of the armor is incorrect
on a number of points. He estimated the number of segments to be
higher (15) than exist in fact. He did not properly understand the
relationship of the shield to the vertebral column, and as a result of
his inaccurate count of the number of segmental elements, he mis-
judged their antero-posterior dimension.

Williston's concept of the carapace was that it consisted of trans-
versely elongate segments, one to each vertebral segment, "with no
connection whatever to the underlying [neural] spines." Presum-
ably, he meant by this that the armor segments were not ossified from
the tips of the neural spines, as they are in Aspidosaurus. At the
same time, he believed Broiliellus to be closely related to Aspido-
saurus, and rationalized this conflict by considering the co-ossification
in the latter genus to be secondary.

He is correct in stating that there is one vertebral segment per
armor segment. In no specimen can one observe the connection be-
tween armor and the underlying vertebrae, but in specimen CNHM
no. UC 685, it is possible to count nearly undisturbed armor and in-
tercentra on opposite sides, and the counts are the same. It can be

' See the discussion of the Dissorophus specimen from that locality (p. 61).

68 FIELDIANA: GEOLOGY, VOLUME 16

concluded that only members of the internal series of armor segments
exist in Broiliellus.

However, it is a different story in regard to the total number of
armored vertebral segments. The armor segments somewhat overlap
one another with the trailing edge of each segment overlapping the
leading edge of the next succeeding segment. The line of demarca-
tion is often unclear, and sometimes the imbricated arrangement has
caused breakage across the segments during preservation. These fac-
tors cause difficulty in counting the number of segments, but it ap-
pears that each of the two specimens described by Williston has
thirteen vertebral segments covered with armor, rather than fifteen,
as described by Williston. It is important to take note of this matter
because the armored vertebral count in Broiliellus and in Cacops has
been used as evidence of the relationship of the two genera (fig. 4).

On specimen CNHM no. UC 684, the shield as a whole is 14 cm.
long. In the anterior segments, the maximum transverse dimension
is 4 cm. Posteriorly, the shield narrows to 2.2 cm. The shield is
nearly flat, but is rounded along the median axis.

The most anterior armor segment is about 13.5 mm. in antero-
posterior length, which makes it 1.5 mm. longer than the second seg-
ment. Williston (1914) indicated that the first segment is consider-
ably larger than the other segments. He probably thought that the
posterior border of the second segment, which is vaguely differenti-
ated from the anterior portion, was the third segment. The inter-
pretation of the second segment as two segments would result in an
appearance of great disparity in antero-posterior dimension between
the first segment and the segments following it.

The remainder of the carapace, except for the last armor segment,
is made of elements whose typical dimensions are about 12 mm. an-
tero-posteriorly, 40 mm. in width, and 2 or 3 mm. thick. The borders
of each of the segments, from the dorsal aspect, are slightly curving,
convex anteriorly and concave posteriorly.

The most posterior segment appears to be slightly larger, in an-
tero-posterior dimension, than the others, and to have convex anterior
and posterior boundaries, from a dorsal aspect.

It has not been possible to study the ventral aspect of the armor
segments of B. texensis, because of the nature of the matrix. How-
ever, in a new, closely-related species from the Arroyo Formation,
with armor in every way like that of B. texensis, preparation of the
ventral surface reveals a ventrally directed point, similar to the con-
dition in Dissorophus. It can be concluded that the armor of Broili-

DEMAR: DISSOROPHIDAE ARMOR

69

Fig. 4. Broiliellus texensis, CNHM UC 684. Dorsal view of a skeleton, with
special reference to the armor.

ellus is homologous with the internal series of Dissorophus, and that
the relationship of the armor to the neural spines is like that of Dis-
sorophus. Figs. 1 and 4 illustrate the armor of Broiliellus.

Cacops aspidephorus

Specimens Examined:

Holotype, CNHM no. UC 647, consisting of a nearly complete,
articulated skull and skeleton; CNHM no. UR 806, consisting of ar-
ticulated fragments of the carapace; CNHM no. UC 885 (fig. 5), con-
sisting of several articulated vertebrae and armor segments; CNHM
UC 900, consisting of several articulated neural spines and armor
segments (fig. 6) ; and CNHM no. UC 977, consisting of several artic-
ulated neural spines and armor segments (fig. 7) .

Distribution:

C. aspidephorus is known only from the Cacops bone bed of the
Vale Formation of Willbarger County, Texas, except for two humeri
from the Choza Formation (Olson, 1956).

Discussion:

Williston (1910) described a complete skeleton of Cacops aspi-
dephorus, CNHM no. UC 647, which includes a complete dorsal ar-

Fig. 5. A, Cacops aspidephorus, CNHM UC 885. Dorsal view of the armor
B, Cacops aspidephorus, CNHM UC 977. Dorsal view of a posterior portion of
the shield. C, Lateral view of the same from the right.

70

DEMAR: DISSOROPHIDAE ARMOR

71

one inch
I I

Fig. 6. Armor of Cacops aspidephorus, CNHM UC 900. A, dorsal view,
B, Lateral view from the left. C, Posterior view.

mored shield. This specimen is now mounted in Field Museum of
Natural History, and is therefore difficult to study. Because my
interpretations of the dorsal armored shield of Cacops are signifi-
cantly different from Williston's, I have described additional mate-
rials from the Cacops bone bed that Williston did not mention. Most
of the following concepts are based on these specimens. Figs. 2, 5
and 6 illustrate the armor of Cacops.

Williston's (1910a, pp. 258, 259) concept of the dermal armor was
accepted by all later authors. He states, "In the structure of the
dorsal shield or carapace the genus (Cacops) is identical with Disso-
rophus, save the carapace of the latter genus is very much more
extensive."

As Williston states, there are two pieces of armor per vertebra,
with 15 vertebrae covered with armor. The pieces are arranged alter-
nately in two groups as follows: There is an internal group in which
each segment is co-ossified with an expanded neural spine. Alternat-
ing with these segments are the segments of the external group, which
are arranged so that each segment lies partly on the posterior border.

72 FIELDIANA: GEOLOGY, VOLUME 16

and partly on the anterior border of adjoining members of the inter-
nal group. This situation is diagrammatically illustrated in fig. 2.

This carapace involves 15 vertebrae, and begins with the anterior-
most neural spine, the axis. The first and last armor segments are
each members of the internal series.

Typically, a segment, either internal or external, is about 2.5 cm.
in transverse dimension, and 1 cm. long. Both the first and last seg-
ments are about 1 cm. in transverse dimension, and the last segment
(one of the internal series) is about 2 cm. long.

The total length of the carapace of the mounted specimen, CNHM
UC 647, is 15.5 cm. The individual segments are rounded over the
tops of the neural spines, so that from either anterior or posterior
aspect, they look like inverted, rather open Vs.

Williston apparently did not consider the armor to be fundamen-
tally different from one dissorophid to another, therefore he did not
stress the nature of the dorsal part of the neural spine and its con-
nection with the dermal armor segments.

The neural spines are indistinguishably fused to the internal seg-
ments of dermal armor. From an anterior aspect, the neural spines
are swollen dorsally, and thus adjoin the armor with an expanded
end. This swelling is rather variable. In some instances, the dorsal
expansion of the neural spine is marked; in others it is slight. Appar-
ently, the specimens of older individuals have more greatly expanded
neural spines.

In neither Broiliellus nor Dissorophus is the armor fused to the
dorsal end of its neural spine, whereas Cacops, Alegeinosaurus, and
Aspidosaurus all have such fusion in all armor segments of the inter-
nal series. This suggests that Cacops is not closely related to Dis-
sorophus or Broiliellus, but is closely related to Alegeinosaurus and
Aspidosaurus. However, see the discussion of "Aspidosaurus" novo-
mexicanus.

Aspidosaurus

Specimens Examined:

A. chiton. — The only specimens are at the Museum of the Alte
Akademie, Munich, nos. 84 and 85. One consists of most of a skull
and the other of a portion of the vertebral column and dermal armor.
They were not examined. A. glascocki, the holotype AMNH no. 4864,
a portion of the vertebral column, ribs and attached dermal armor,
with fragments of the skull; A. apicalis, the holotype AMNH no.

d^r.^

one inch
I I

Fig. 7. Aspidosaurus crucifer, MCZ 1258. A, Lateral view of two articulated
neural spines with armor. B, Anterior view. C, Lateral view from the right of the
posteriormost armored segment. D, Anterior view of the posterior most armored
segment.

73

74 FIELDIANA: GEOLOGY, VOLUME 16

4785, consisting of the apices of several neural spines, now lost and
therefore not examined; A. crucifer, the holotype CNHM no. UC 1205
(fig. 8), consisting of a neural spine [this is probably UC 171 of Case
(1903 and 1911)], UT no. 40030-5 (fig. 8), consisting of two associ-
ated neural spines, and MCZ no. 1258, consisting of several articu-
lated neural spines (fig. 7) . In addition, specimens NM 22075 (fig. 9)
and UT no. 40030-8 (fig. 9), each consisting of a neural spine, are
probably specimens of Aspidosaurus.

Distribution:

All specimens are from the Permian, as follows: A. chiton was col-
lected early, and its locality is not surely known, but Romer (1928)
says that it is from the Arroyo Formation of Texas. A. glascocki is
from the Belle Plains Formation of Texas (Romer, 1928) . A. apicalis
is from the Abo Formation of New Mexico, which suggests a Putnam
Formation equivalent.

Discussion:

There is much confusion concerning the genus Aspidosaurus for
a variety of reasons. The genotype, A. chiton is not available for
study, and all other species are represented by fragmentary materials.
"Aspidosaurus" novomexicanus is represented by good materials, in-
cluding a skull, but clearly does not belong in the genus Aspidosaurus
(Langston, 1953) . There is no question that the genus Aspidosaurus is
a valid one, and each of the species also appears to be valid except for
"A." novomexicanus which clearly is Broiliellus or even another
genus. However, the lack of comparable and complete materials
makes it impossible to know if the species belong together in one
genus. Romer (1947, pp. 161, 162) states that several genera, or even
families, may be involved.

However, even with this confusion, it is possible to compare the
specimens of Aspidosaurus armor with that of other dissorophid gen-
era. First of all, the specimens of armor assigned to the genus Aspi-
dosaurus all have the same general structure, in that they have dermal
pitting on the dorsal and lateral margins of the neural spines. The
important point here is that the armor is fused to the neural spine,
and not free as in Dissorophus and Broiliellus. Furthermore, it lacks
the mid-ventral flange of the internal series of each of these genera.
Figures 2, 7 and 9 illustrate the armor of Aspidosaurus.

The genotype, Aspidosaurus chiton (Broili, 1904) is represented
by a considerable portion of the skull and a portion of the vertebral

one inch,
I 1

Fig. 8, Neural spine of Aspidosaurus crucifer. A, CNHM UC 1205, Lateral
view. B, Posterior view of same. C, Neural spine of Aspidosaurus cruciferl,
UT 40030-5, lateral view. D, Anterior view of same.

75

76 FIELDIANA: GEOLOGY, VOLUME 16

column and armor. From the descriptions, one can surmise that the
armor consists of small, roof-shaped pieces fused to the dorsal surface
of the expanded neural spines. These armored segments overlap onto
the next posterior segment; as a result the anterior dorsal surface of
each segment of armor is differentiated from the remainder of the
segment by having more subdued dermal pitting.

Aspidosaurus glascocki (Case, 1910) is represented only by the
holotype, which consists of a portion of the vertebral column, ribs
and attached dermal armor plates, and portions of the skull. The
dermal plates are narrow and very coarsely sculptured externally.
They are firmly fused to the expanded dorsal ends of the neural spines,
and are slanted to either side in the usual roof -like manner. The der-
mal plates overlap anteriorly and posteriorly, but in no ordered man-
ner, and appear to be partly fused to their nearest neighbors.

Aspidosaurus apicalis (Cope, 1878) was represented only by the
holotype, AMNH no. 4785, consisting of the apices of several neural
spines. This specimen has apparently been lost. Case (1911) de-
scribed it as "only the apices of the spines known. Flat, oval in out-
line with a peg-like apex." Little more can be said, except that the
figures show the dorsal surface of the neural spine to be flat, with a
median dorsal keel.

Aspidosaurus crucifer Case is one of the best represented species,
and consists of three specimens. The holotype CNHM no. UC 1205
(fig. 8) is a neural spine. UT no. 40030-3 consists of two neural
spines, and MCZ no. 1258 (fig. 7) consists of several articulated
neural spines.

In 1911, the holotype and only known specimen of A. crucifer at
that time was succinctly described by Case as ". . . only a single
neural spine known. The apex with a prominent median spine and
narrow lateral process. Adjacent process touching, but overlapping
very slightly."

A very significant morphological feature had been described by
Case (1903). He said:

The anterior and posterior edges of the apex are free from the deep pits and
are marked by a narrow space of striations showing that the spine was over-
lapped by the edges of some other element; it is evident from the shape of the
spine that this could have been an extra element intercalated between the
spines . . . probably one of the dermal ossifications such as occur along the spines
of Pareiasaurus.

Case (1911) apparently interpreted these smooth, striated sur-
faces as the area of a slight overlap of adjacent armor segments, but

DEMAR: DISSOROPHIDAE ARMOR

77

one inch

Fig. 9. A, Neural spine of an unidentified dissorophid, NM 22075, lateral
view. B, Anterior view of same. C, Neural spine of an unidentified dissorophid,
UT 40030-80, dorsal view. D, Anterior view of same.

his 1903 statement appears to be the correct one. Another specimen,
MCZ no. 1258, consists of several articulated neural spines and der-
mal armor. These spines clearly do not overlap, yet they have a band
of unpitted bone on their anterior and posterior borders. The most
reasonable explanation is that, external to those neural spines and
alternating with them, was another dermal plate closely applied to
the swollen tip of the neural spines, and overlapping slightly on adja-
cent surfaces. It may be pointed out that this condition is reminis-
cent of the internal series in Cacops and Alegeinosaurus, and is
an important reason for the assignment of Cacops, Alegeinosaurus
and Aspidosaurus to one family. These facts are shown in the dia-
grams of fig. 2 and in the illustrations of Aspidosaurus and Cacops
armor. See especially figs. 8 and 9.

78 FIELDIANA: GEOLOGY, VOLUME 16

Another interesting thing about the same specimen (MCZ no.
1258) is its anterior-posterior differentiation of the armor (fig. 7).
Two of the spines are fairly typical of the genus Aspidosaurus, but
the two posterior spines are a great dorsal extension of the neural
spines, with little lateral expansion. On the posteriormost spine
the dermal pitting is posterior to the spine, as well as dorsal and lat-
eral. The anterior spines are not known for this specimen, but they
were probably very variable. The main point is that this single speci-
men shows wide variation, with some overlap into the morphological
type represented by Platyhystrix.

In connection with this differentiation, the genus Platyhystrix can
be appropriately mentioned here. A large number of disarticulated,
undescribed specimens of that genus exist at the Museum of Com-
parative Zoology at Harvard. They emphasize the importance of the
proper evaluation of differentiation along the spine, when one is ana-
lyzing the status and relationships of Aspidosaurus specimens. A
brief study of these specimens reveals some specimens with relatively
small unexpanded spines, others with greatly expanded spines, and
some in which the spine is similar to the holotype of Platyhystrix, in
which there are side spurs and an enormous bladelike dorsal exten-
sion. These extreme variations in specimens found in the same de-
posit seem a contradiction to the interpretation that several species
are involved. The net result of the observation of this variation
from Aspidosaurus-Wke spines to those of utterly unfamiliar aspect,
is that one should treat the status of the various Aspidosaurus spe-
cies and that of Platyhystrix with considerable caution.

Alegeinosaurus aphthitos

Specimen Examined:

The holotype, AMNH no. 4556, consisting of a partial anterior
axial and appendicular skeleton, including several articulated seg-
ments of dorsal dermal armor.

Distribution:

Coffee Creek of the Arroyo Formation, Lower Permian of Will-
barger County, Texas.

Discussion:

Alegeinosaurus aphthitos is another species represented by a single
fragmental specimen that appears to be sufficiently unique to warrant

DEMAR : DISSOROPHIDAE ARMOR 79

calling it a separate genus. Here again the armor is very important.
Case (1911) described the armor thus:

Dorsal vertebrae with neural spines expanded, but not wider than the ver-
tebrae. Small dermal plates overlying the neural spines directly, and each
overlapping the preceding plate, not wider than the vertebrae. No enlarged
anterior plate. . . .

Williston (1910, p. 260) originally stated that Alegeinosaurus aph-
thitos was certainly a species of Aspidosaurus. By 1914, he had de-
cided that it was a separate genus on the basis that the dermal plates
were not co-ossified with the expanded neural spines. Apparently,
he envisioned an evolutionary series that began with an Alegeinosau-
r2^s-like animal with separate dorsal dermal ossifications above the
neural spines. This animal evolved into an Aspidosaurus-Mke animal
by the fusion of the nerual spines to the dermal plates. In connection
with this concept, Williston (1914) maintained the idea that Alegeino-
saurus was not closely related to Cacops, and, in fact, placed them in
separate subfamilies. However, Romer (1947, p. 159) stated that
Alegeinosaurus closely resembles Cacops, except for uncinate proc-
esses on the ribs.

These conflicting ideas need clarification, and my examination of
the holotype, the only specimen, indicates the following regarding the
armor. The neural spines rise about 2 cm. above the zygapophyses
and measure about 2 cm. transversely, and 0.8 cm. longitudinally.
From an anterior aspect, the head of each spine is swollen, forming a
triangle with the apex upward. These apices, from a dorsal aspect,
are dermally pitted, and are divided into two parts along a trans-
verse midline, one of which slopes anteriorly, and the other posteri-
orly. Just dorsal to these spines, and between pairs of them, are a
series of dermal plates about 1.5 cm. long and 2.5 cm. broad, with
dermal pitting as in typical armor. Along an antero-posterior mid-
line, they are sharply folded downward, so that they look like in-
verted Vs.

One thing that stands out on this specimen is the excellent preser-
vation of the bone. There is no weathering, and delicate structures
are excellently revealed. There can be no doubt that Williston was
incorrect in assuming only a single piece of armor per segment, as
in Aspidosaurus. Clearly, there are two pieces; one part, the in-
ternal series, which displays dermal pitting on the dorsal extremi-
ties of the neural spines; the other, the external series, consisting of
the separate inverted V-shaped pieces.

80 FIELDIANA: GEOLOGY, VOLUME 16

As Romer (1947) pointed out, this species is clearly very closely
related to Cacops. The armor, however, is distinct from any known
specimen of Cacops, and as a result it must be considered a separate
genus, at least until more materials are located.

Especially pertinent about this armor, is that single specimens of
the swollen neural spines would be identified as from the genus Aspi-
dosaurus, but articulated materials can be separated from Cacops only
with some uncertainty. This, fundamentally, is the basis for suggest-
ing that these three genera belong together in one subfamily. How-
ever, before that step is taken, it is necessary to discuss some other
materials.

"Aspidosaurus" novomexicanus

Specimens Examined:

The holotype, YPM no. 810, consisting of a skull with articulated
vertebral column, dermal armor, and miscellaneous skeletal elements;
CNHM no. 673, consisting of a partial skull and parts of the dermal
armor and skeleton.

Distribution:

Cutler and Abo Formations of the Permian of New Mexico.

Discussion:

This species is represented by only three specimens, all from New
Mexico. Two of them are mentioned above, and, in addition, there
is a skull, CNHM no. UC 40103.

Case and Williston (1913) assigned the holotype to Aspidosaurus
as a new species, A. novomexicanus, but it is clear from the work of
Romer (1947) and Langston (1953) that, in reahty, it is more closely
related to Broiliellus. More detailed assignment will have to await
more complete comparisons.

The holotype is not well preserved, and is obscure because of this
poor preservation and incomplete preparation. Also, I have had no
opportunity to examine the California specimen, which, however, has
no armor. The information given here is based mostly on the Chi-
cago specimen, and as of now it appears that the Chicago specimen
is not the same species as the holotype. More work is necessary to
be sure of this.

The reason for considering this species, despite its inadequate tax-
onomic assignment, is that the dermal armor of Field Museum of

DEMAR : DISSOROPHIDAE ARMOR 81

Natural History specimen and its relationship to the vertebral column
are unique.

Case and Williston (1913) stated that there are two dermal plates
per vertebral segment, but examination indicates that this is false,
and that there is actually only one dermal plate per segment. As in
the case of the armor of Broiliellus texensis, there tends to be trans-
verse breakage of the pieces of armor under the stress of compaction,
with the result that more pieces appear to exist than actually do.

The armor segments are about 1 cm. in antero-posterior measure-
ment, and, unique for Broiliellus-like animals, are only about 2 cm.
in transverse measurement. Of special interest is the relationship of
these armor segments to the neural spines. Unfortunately, the entire
specimen is not articulated, so that details of regional variation can-
not be evaluated properly. However, a number of things stand out.
The anterior plates are fused to the neural spines, and near this junc-
ture the neural spines are thicker than they are ventrally. On the
other hand, the more posterior plates are not attached to the neural
spines, but, like the internal plates in Dissorophus and Broiliellus,
they have a thin, ventrally directed flange along the antero-posterior
midline that, presumably, articulated between the neural spines.

The posterior armor and its relationship to the neural spines is
then much as one would expect in an ancestor of Broiliellus. It sup-
ports the idea that the armor of Broiliellus and Dissorophus belongs
in a distinct morphological plan that is different from that of Aspido-
saurus, Alegeinosaurus, and Cacops.

However, the anterior armor appears to be very much like that
of Aspidosaurus, or the internal series of Cacops.

It is not possible to consider "Aspidosaurus" novomexicanus a spe-
cies of Aspidosaurus, because its skull is nearly identical to that of
Broiliellus. In fact, Langston (1953) preferred to rename it as Broili-
ellus novomexicanus because of its marked similarity to that genus.

It appears that we are on the horns of a dilemma. It is possible
that the solution to this dilemma is that the armor of Cacops and
Aspidosaurus may be derived from Broiliellus-like armor. However,
this would seem to be unlikely, because (as will be discussed) there are
other important morphological distinctions between the two groups.

There is another possibility. It was noted that the anterior seg-
ments of the armor of Dissorophus are intimately associated with the
neural spines by a ventral flange inserted into the split ends of the
neural spines. Apparently, a little mobility is retained by this ar-

82 FIELDIANA: GEOLOGY, VOLUME 16

rangement. At any rate, it would take very little for the neural
spines and the ventral flanges to fuse together, with this intimate
arrangement. The mobility provided by the unfused condition in
Dissorophus can be attained in a form such as "Aspidosaurus" novo-
mexicanus, which has small pieces of armor, by having the armor seg-
ments move slightly over one another.

Another possibility (suggested by Case and Williston, 1913) is
that all of the pieces found together are not properly associated to-
gether as one species, but this seems unlikely. In any case, it appears
that the posterior dermal armor fits into the concept of Broiliellus
and Dissorophus-Wke armor, and that something atypical has occurred
in the anterior vertebral and dermal armor segments.

Summary and Conclusions

All of the well known Permian dissorophids are armored, whereas
the Pennsylvanian dissorophids are not (Gregory, 1950) . Therefore,
it appears that the armor evolved sometime during the very Early
Permian or Late Pennsylvanian. Moreover, the armor falls into two
morphological groups. One group includes Dissorophus and Broiliel-
lus; the other includes Cacops, Aspidosaurus and Alegeinosaurus. In
each group genera with both one and two pieces of dermal armor per
segment are known. In addition, each group possesses genera with
transversely narrow pieces of armor, though only Dissorophus and
some species of Broiliellus have transversely broad pieces of armor.
The two groups can be called Dissorophinae, for Dissorophus and
Broiliellus, and Aspidosaurinae, for Aspidosaurus, Cacops, and Ale-
geinosaurus. These names are merely the old subfamilies of Williston
(1914), but with new members.

The key to differences in the armor of these two groups is con-
tained in the segments of armor which touch the neural spines, here
called the internal series. In the Aspidosaurinae, the neural spines
are expanded dorsally, and the dermal armor is firmly fused to this
expanded terminus. The external series is immediately above the in-
ternal series, and each segment of the external series lies between, and
overlaps slightly on the border of the adjoining pair of segments of
the internal series.

In the Dissorophinae, the internal series is not fused to the neural
spine, and furthermore does not lie directly above it. Instead, each
segment of the internal series has on its ventral midline an anteriorly
and ventrally directed fiange which extends into the space between

DEMAR : DISSOROPHIDAE ARMOR 83

two adjacent neural spines. Very often the neural spines are grooved
on their anterior and posterior surfaces, making a closer articulation
with these flanges. Somewhat exceptionally, the anterior armor seg-
ment of Dissorophus apparently has a large flange which extends into
the split ends of the neural spine, but still retains some mobility. The
external series, which exists only in Dissorophus, lies just external to,
and alternates with, the internal series. Each segment of the exter-
nal series of armor overlaps the segments of the internal series ante-
rior and posterior to it.

In order to evaluate the homology of these armor segments prop-
erly, we must consider the stratigraphic distribution of the various
species, in addition to the morphological data just mentioned. The
early species of the Dissorophinae is Broiliellus novomexicanus; that
of the Aspidosaurinae is Aspidosaurus sp. In these species, only the
internal armor series is preserved. Each subfamily later developed
at least one species which has the external armor series, too. These
species are Dissorophus multicinctus, of the Dissorophinae, and Cacops
aspidephorus and Alegeinosaurus aphthitos, of the Aspidosaurinae.
From this information, it can be concluded that the internal series
developed first, and later the external series developed.

From the above stratigraphic and morphologic information we
can reason about homology as follows: Because of structural differ-
ences in the internal series of armor in each group, the two series can-
not be homologous (have a common phylogenetic origin) . And even
though the external armor plates of both groups look similar, they
also cannot be homologous because of the non-homologous nature of
the internal series. We have then two separate origins for the armor
of the Dissorophidae.

Longiscitula houghae presents a third possible origin of the armor.
The only armor preserved resembles that of the first segment of Dis-
sorophus, but the structure of the skull of Longiscitula is distinct from
that of Dissorophus. In fact, it resembles Trematops, except for the
dissorophid otic notch (DeMar, in press).

Of course, the characteristics of the armor should not be the only
criteria used to separate the Dissorophidae into two or more groups.
As would be expected, the skulls and the sacral ribs exhibit a number
of characteristics which support the idea that the Aspidosaurinae and
the Dissorophinae evolved separately, early in the evolution of the
dissorophids.

Skulls are known in Cacops and Aspidosaurus chiton, of the Aspi-
dosaurinae, but only that of Cacops was available to me for study.

84 FIELDIANA: GEOLOGY, VOLUME 16

In the Dissorophinae, skulls are known in " Aspidosaurus" novomexi-
canus, B. texensis, in a new species of Broiliellus from the Arroyo For-
mation, and in Dissorophus. The skull of Cacops stands apart from
the skulls of the Dissosophinae in being relatively deeper, in having
especially large, laterally directed orbits, and a different pattern of
exostoses on the skull roof.

Of even more significance is the nature of the sacral region in the
Dissorophidae. Williston (1910) described the remarkable fact that
Cacops has two sacral ribs. This unusual condition (among amphib-
ians) was thought to be the standard condition in the family, though
no other dissorophid was known which had the sacral region pre-
served. In retrospect, this extrapolation seems very reasonable be-
cause the family is armored, which is itself a remarkable condition.

Therefore, it was a surprise to find, among the specimens of MCZ
no. 2122, a well preserved specimen of Dissorophus multicinctus that
demonstrated that Dissorophus had a single sacral rib. The rib is
what one would expect in a rhachitome, though it is somewhat special-
ized in that it has a great ventral expansion which wraps around and
articulates closely with the base of the ilium. It is difficult to see how
a sacral rib like this could evolve into the two sacral ribs of Cacops.

This clear distinction between Cacops and Dissorophus in the sa-
cral region, which is consistent with the distinctions between them
in the armor and in the skull, would seem to indicate that much of
their other general similarities is a result of parallel evolution. This,
then, is further reason for concluding that there are two, and possibly
three separate developments of armor in this one family.

The nearly simultaneous separate evolution of armor several times
over would suggest some moulding evolutionary influence. Presum-
ably, this influence was environmental. In a later paper, a general
framework of the evolution of Early Permian faunas will be devel-
oped. In this paper, the idea will be developed in relation to the ar-
mor of the dissorophids, and will otherwise be only briefly discussed.

It is, of course, somewhat difficult to obtain reliable information
on the distribution of Permian vertebrates, particularly the older
specimens, because many collections were made without specific lo-
cality data and with little stratigraphic reference. As noted earlier,
fair information is available for most specimens and information on
some others can be inferred.

It would appear that most dissorophids were collected from places
very near where they actually lived. The common genera are pre-

DEMAR : DISSOROPHIDAE ARMOR 85

served in an articulated condition, with several individuals of the
same species associated. Presumably, this is also the case for the
animals that are found with the dissorophids. The preserved assem-
blages may thus be considered to represent life assemblages.

Dissorophids are not found with obviously aquatic or semiaquatic
animals. These would include xenacanth sharks and lung-fishes, em-
bolomeres, and other tetrapods with weak limbs and /or lateral line
organs (Trimerorhachis, Lysorophus and the like). Instead, they are
found with fairly terrestrial appearing reptiles, such as Varanops,
Casea hroili and Seymouria, and with amphibians like Trematops, or
in association with other members of the family Dissorophidae. This
provides some indication that the Dissorophidae lived in more ter-
restrial associations than many of the Permian tetrapods.

The osteology is consistent with the above interpretation. Most
of the Dissorophidae are well ossified. In connection with this ossi-
fication, the limb bones have well terminated articulations, the manus
and pes (in the few known cases) are well formed, the vertebrae are
closely interlocked with well formed central elements. The ribs ar-
ticulate closely with the vertebrae, the skulls are strongly formed,
and the separate bony elements tend to be fused together. The large
otic notch is reinforced by a post-otic bar, and the pelvis is strongly
attached to the vertebral column by means of one or two specialized
sacral ribs.

In addition, the armor fits into this general model, since it is not
merely in the skin above the vertebral column, but is closely inter-
woven into the texture of the column by means of direct fusion to the
neural spines and articulation with adjoining plates, or by close ar-
ticulation with the neural spines with a ventrally directed flange and
by articulation with adjoining plates. One action of the armor on the
vertebral column, regardless of any other function it may have had,
was to strengthen and restrict the movements of the vertebral col-
umn. In Dissorophus, the anteriormost armor segment is enlarged
and its anterior end articulates closely with the posterior edge of the
skull, forming an integrated system. In this system, the skull articu-
lates closely with the vertebral column and the carapace, which also
articulate closely with one another. Thus the armor of Dissorophus
apparently decreases the flexibility of the body, as the pieces of armor
are closely pressed together, allowing little motion. In connection
with this inhibition of movement, the armor is integrated into a well-
ossified vertebral column and powerful rib series. The armor could
thus have served to strengthen the back by decreasing its flexibility.

86 FIELDIANA: GEOLOGY, VOLUME 16

The development of this condition can be traced back to Broili-
ellus, an earlier genus. In that animal, the general structure of the
armor is similar, but there is apparently less reduction of flexibility
because there is a space of about 1 cm. between the carapace and the
skull, and the individual armor segments form a less tightly inte-
grated unit with the vertebral column.

These features of the dissorophids are entirely consistent with the
kind of adaptive features that would be expected if any tetrapod
evolved from adaptation to an aquatic environment to adaptation to
a terrestrial environment. In fact, under a broad interpretation,
these are the types of features that would be expected if any organism
evolved in that manner. A terrestrial environment places certain
demands on organisms that live in it which are different from those
imposed by aquatic environments. These relate to reproduction,
drying, locomotion and strength. Only the last two can concern us
directly, because no information can be gathered on the other two
by examination of the fossils. The problem of drying will be con-
sidered briefly.

From the structure of the animals and from the distribution of
the specimens, it appears that the later dissorophids wei e terrestrial,
and the earlier forms from which they evolved were aquatic or semi-
aquatic.

Consistent with this conclusion is the concept that, in the Disso-
rophinae line, the armor serves as a mechanism to prevent drying.
The armor is, of course, dermal bone, which would serve to thicken
the dermis, and, thus to allow less transfer of moisture through the
skin. Terrestrial animals must handle the problem of evaporation
through the skin and this is one possible solution.

It is curious, and a bit ironic, that one possible function for dis-
sorophid armor that does not fit into this general plan is that of pro-
tection. It would seem that aquatic and semiaquatic existence would
place animals in as much danger from carnivores as would terrestrial
existence. However, regarding protection, if some such function can
be presumed, it is interesting to note that in Dissorophus the ribs are,
frequently preserved lying along the boundary between adjoining
segments of armor of the internal series. One could, therefore, de-
velop the concept that the armor and ribs were integrated into a
protective shield with three layers.

Finally, it can be noted that the presence of dermal ossifications,
or "armor," in dissorophids cannot be considered unusual since such

DEMAR : DISSOROPHIDAE ARMOR 87

ossifications have been reported in Eryops as small round or oval ossi-
fications (Romer and Witter, 1941), similarly in Trimerorhachis (Case,
1935), in Archegosaurus and Sderocephalus (Broili, 1927) and in Mi-
crerpeton (Gregory, 1950). In fact, Eryops has dermal pitting on the
dorsal surface of its neural spines which is very similar to the situa-
tion in Aspidosanrus.

Of special importance is Kotlassia, described in detail by Bystrow
(1944), which has dermal ossifications (armor) as prominent as that
of the Dissorophidae. In detail it differs in a number of ways, the
most obvious of which is that there are three rows of armor so ori-
ented that there is a mid-dorsal row with one row on either side. The
armor is transversely elongate with one transverse set per segment.

If there is any single cause for the development of armor, such as
occurs in the above animals, it cannot be simply terrestrial adapta-
tion, since the above animals are largely aquatic. Therefore, in order
to arrive at a conclusion on the function of armor, we must consider
both the animal's environmental context, and its other morphological
features. This problem will be discussed in a later paper.

REFERENCES

BOULENGER, G. A.

1902. Amphibia. Encyclopaedia Britannica, 10th ed., pp. 381-384.

Broili, F.

1904. Permische Stegocephalen und Reptilien aus Texas. Paleontographica,
51, pp. 1-16.

1927. tiber die Hautbedeckung der Archegosauridae und Actinodontidae.
Zeitschr. Deutsch. Geol. Ges., 79, pp. 375-384.

Bystrow, A. P.

1944. Kotlassia prima Amalitzky. Bull. Geol. Soc. Am., 55, pp. 379-416.

Case, E. C.

1903. New or little known vertebrates from the Permian of Texas. Jour. Geol.,
11, pp. 394-402.

1905. The osteology of the Diadectidae and their relations to the Chelydosauria.
Jour. Geol., 13, pp. 126-159.

1905a. Characters of the Chelydosauria. Science, n.s., 21, p. 298.

1910. New or little known reptiles and amphibians from the Permian(?) of
Texas. Bull. Am. Mus. Nat. Hist., 28, pp. 163-181.

1911. Revision of the Amphibia and Pisces of the Permian of North America.
Carnegie Inst. Pub. 146, pp. 1-179.

1935. Description of a collection of associated skeletons of Trimerorhachis.

Michigan University, Mus. of Paleont. Contributions, 4, pp. 227-274.
1946. A census of the determinable genera of the Stegocephalia. Trans. Amer.

Phil. Soc, n.s., 35, pp. 325-420.

88 FIELDIANA : GEOLOGY, VOLUME 16

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

1913. A description of Aspidosaurus novomexicanus. Carnegie Inst. Pub. 181,
pp. 7-9.

Cope, E. D.

1878. Descriptions of extinct Batrachia and Reptilia from the Permian forma-
tions of Texas. Proc. Amer. Phil. Soc, 17, pp. 505-530.

1895. A batrachian armadillo. Am. Naturalist, 29, p. 998.

1896. Permian land Vertebrata with carapaces. Am. Naturalist, 30, pp. 936-
937.

1896a. Second contribution to the history of the Cotylosauria. Proc. Amer.

Phil. Soc, 35, pp. 122-139.
1896b. The ancestry of the Testudinata. Am. Naturalist, 30, pp. 398-400.

DeMar, R. E.

In press. Longiscitula houghae, a new genus of dissorophid amphibian from the
Permian of Texas. Fieldiana: Geol., 16, no. 2.

Gregory, J. T.

1950. Tetrapods of the Pennsylvanian nodules from Mazon Creek, Illinois.
Am. Jour. Sci., 5th ser., 248, pp. 833-873.

Langston, W.

1953. Permian amphibians from New Mexico. Bull. Calif. Univ., Dept. Geol.
Sci., 29, pp. 349-416.

Olson, E. C.

1941. The family Trematopsidae. Jour. Geol., 49, pp. 149-176,

1956. Fauna of the Vale and Choza: II. Fieldiana, Geol., 10, pp. 313-322.

ROMER, A. S.

1928. Vertebrate faunal horizons in the Texas Permocarboniferous red beds.

Bull. No. 2801, Texas Univ., pp. 67-108.
1947. Review of the Labyrinthodontia. Bull. Mus. Comp. Zool., 99, pp. 1-368.

RoMER, A. S. and R. V. Witter

1941. The skin of the rhachitomous amphibian Eryops. Am. Jour. Sci., 239,
pp. 822-824.

1942. Edops, a primitive rhachitomous amphibian from the Texas red beds.
Jour. Geol., 50, pp. 925-960.

Watson, D. M. S.

1917. The evolution of the tetrapod shoulder girdle and fore limb. Jour. Anat.,

52, pp. 1-63.
1920. The structure, evolution, and origin of the Amphibia, the orders Rhachi-

tomi and Stereospondyli. Trans. Royal Soc. London Philos., B, 209, pp. 1-73.

Williston, S. W.

1908. The Cotylosauria. Jour. Geol., 16, pp. 139-148.

1909. The faunal relationships of the early vertebrates. Jour. Geol., 17,
pp. 636-658.

1910. Dissorophiis Cope. Jour. Geol., 18. pp. 526-536.

1910a. Cacops, Desmospondylous, new genera of Permian vertebrates. Bull.
Geol. Soc. Amer., 21, pp. 249-284.

1911. American Permian vertebrates. Univ. of Chicago Press, pp. 1-145.

1914. Broiliellus, a new genus of amphibians from the Permian of Texas. Jour.
Geol., 22, pp. 49-56.

1916. Synopsis of the American Permocarboniferous Tetrapoda. Contrib.
Walker Mus., 1, pp. 202-206.

r