UNlVEHSn Y OF
ILLINOIS LIBRARY
AT URBANA-CHW/IPAIGN
GEaOGY -

The person charging this /"^ .^^^^^^^^^^
sponsil3le for its return to the hbra Y ^rom
which it was withdrawn on or before the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.iKinK; LIBRARY AT URBANA-CHAMPAIGN
UNIVERSITY OF lUINOIS LIBRAKT

OCT f 3 |976
OO,'"! 9 1977

L161 — O-1096

o--^ t /

/

FIELDIANA • GEOLOGY

Published by
FIELD MUSEUM OF NATURAL HISTORY

Volume 16 August 31, 1966 No. 4

Cardipeltis

An Early Devonian Agnathan

of the Order Heterostraci

Robert H. Denison

Curator, Fossil Fishes

This study of Cardipeltis was prompted by the acquisition of new
and well preserved material from Cottonwood Canyon in the w^estern
flank of the Bighorn Mountains, east of Lovell, Wyoming. Two
articulated but imperfect specimens have been discovered and for the
first time give a reliable, albeit incomplete, picture of the structure
of this strange heterostracan. The systematic determination of the
new material has required comparisons with Cardipeltis from the
Water Canyon formation of Utah and the Beartooth Butte formation
of Beartooth Butte, Wyoming. For the loan of material from the
former I am much indebted to the late Dr. M. G. Mehl of the Mis-
souri Geological Survey, and from the latter to Dr. Donald Baird of
Princeton University.

The Cottonwood Canyon locality was first described by Black-
stone and McGrew (1954) who published a measured stratigraphic
section and reported the presence of a vertebrate fauna of Early
Devonian age. More recently Sandberg (1961) has referred this
deposit to the Beartooth Butte formation and described other oc-
currences that he believes to be similar at localities in northwestern
Wyoming and southern Montana. As at Beartooth Butte itself,
the formation in Cottonwood Canyon is a channel fill, local in occur-
rence and variable in thickness and lithology. It lies on the eroded
upper member of the Bighorn Dolomite, and is overlain by later
Devonian limestones and dolomites.

Parties from Field Museum of Natural History first visited this
locality in 1959 and have since worked there in 1960, 1962, and 1963.

Library of Congress Catalog Card Number: 66-25If93

No. 1011 89 lUl'-:'- "•< '-

_P 27 1367

90 FIELDIANA: GEOLOGY, VOLUME 16

All of the collections for this Museum have been obtained in an ex-
cavation well up on the side of the south wall of the canyon, just
west of the first tributary canyon from the south. The locality lies
in SE Yx, NE %, Sec. 4, T56N, R93W, Bighorn County, Wyoming.
A section, measured in 1962 by B. G. Woodland and E. S. Richard-
son, Jr., shows the nature of the rocks at the excavation site and the
stratigraphic position of the fossils collected. Their section, with
some modifications and additions, is as follows:
Dolomite, forming a cliff above the excavation (presumably Middle
or Late Devonian).
Beartooth Butte formation

A. Covered interval, in part at least a very friable shale, buff to
drab, and weathering light yellow, about 20 ft.

B. Red shale, high in carbonate, with green and mottled angular
clay pebbles, 1 ft.

C. Purple shale, high in carbonate, with fossil plants and fishes,
7 in.

D. Red shale, high in carbonate, with fossil fishes, Yi in.

E. Mudstone, high in carbonate, fissile at top, becoming sandy
beneath, base irregular, 1 ft., 8 in.

F. Limestone, greenish-gray or reddish, or mottled reddish and
greenish; with a few fossil fishes, some articulated, 33/^-4 ft.

G. Red shale, high in carbonate, irregularly bedded, 3 in.
H. Hard, gray, silty shale, high in carbonate, 6 in.

L Soft, greenish shale, high in carbonate, with conglomeratic
lenses and with fossil fishes, 6 in.

J. Soft, friable, gray shale, high in carbonate, with local red
bands and numerous fossil plants, grading laterally into hard-
er greenish-gray siltstone, 1 ft.

K. Greenish mudstone, high in carbonate, not bedded; base not
seen, a few feet.

L. Gray and reddish dolomitic sandstone, forming ledge, 17 ft.

M. Light gray or reddish dolomitic conglomerate (at the base of
the formation), 5 ft.

Field Museum of Natural History collections were obtained from
beds E to J. Bed F, which is relatively hard, massive and difficult
to work, has yielded relatively few specimens, but among them are
articulated pteraspids and the two articulated Cardipeltis. The
great majority of the specimens were collected in beds H, I, and J.

DENISON: CARDIPELTIS 91

The specimens described are in Field Museum of Natural History
unless otherwise noted.

SYSTEMATICS

The new material of CardipelHs from the Bighorn Mountains has
made it possible to extend the diagnosis of the genus. Previous
diagnoses have been given by Branson and Mehl (1931), Bryant
(1933), Denison (1953), and Stensio (1958).

Cardipeltis Branson and Mehl

Type species — C. wallacii Branson and Mehl

Revised diagnosis: Cardipeltis includes large, rather flat-bodied
Heterostraci in which the dorsal shield consists of a large dorsal disc,
a rostral region of small plates, and probably a pair of orbital plates.
Laterally there are several paired marginal plates with dorsal and
ventral laminae. The ventral shield is a mosaic of small plates boun-
ed anteriorlj'- by two or more larger postoral plates. There is a single
pair of external branchial openings that notch the dorsal disc deeply.
The tail is narrow, slightly longer than the shield, and covered with
relatively small scales, except ventrally where they are large and
subrectangular. There are no paired or median fins. The super-
ficial ornamentation consists of coarse dentine ridges dorsally, and
of large, rounded dentine areas surrounded by very broad ridges
ventrally. The lateral line pattern is distinguished by five dorsal
transverse commissures, and a much reduced ventral system. The
dorsal exoskeleton is 2-3 millimeters or more thick and has a moder-
ately thick spongiosa with rounded chambers arranged somewhat
irregularly.

Cardipeltis wallacii Branson and Mehl

Cardipeltis wallacii Branson and Mehl, 1931, Jour. Geol., 39, pp. 517-523, fig. 2,
pi. 2, figs. 1-3; Bryant, 1932, Proc. Amer. Philos. Soc, 71, pp. 240-241, pi. 5,
fig. 1; 1933, Proc. Amer. Philos. Soc. 72, p. 310; Tanner, 1936, Proc. Utah
Acad. Sci., 13, p. 81; Denison, 1953, Fieldiana: Geol., 11, pp. 305-318, figs.
66-71; Stensio, 1958, Traite de Zool., 13, fasc. 1, fig. 184; 1964, Traite de
Paleont., 4, vol. 1, fig. 87; Tarlo, 1962, Acta Palaeont. Polonica, 7, fig, 9.

Cardipeltis sinclairi Bryant, 1933, Proc. Amer. Philos. Soc, 72, pp. 308-310,
fig. 6, pi. 18, fig. 1; Obruchev, 1941, Trav. Inst. Pal., Acad. Sci. U.R.S.S., 8,
no. 4, p. 16; Denison, 1953, Fieldiana: Geol., 11, p. 306.

Cardipeltis oblongus Bryant, 1933, Proc. Amer. Philos. Soc. 72, pp. 310-311,
fig. 7, pi. 19; Denison, 1953, Fieldiana: Geol., 11, p. 306.

92 FIELDIANA: GEOLOGY, VOLUME 16

Type. — In their original description Branson and Mehl listed
two specimens in the University of Missouri collection under the
heading "Type," but they did not designate a holotype. The better
specimen, 601VP, a natural mold of both dorsal and ventral surfaces
of a dorsal disc, is here designated the lectotype (Branson and Mehl,
1931, pi. 2, figs. 1-2). The other specimen, 602VP, a natural mold
of the ventral surface of a dorsal disc and an associated natural
mold of the outer surface of part of another dorsal disc, becomes a
paratype.

Occurrence. — Early Devonian; Water Canyon formation. Black-
smith Fork and Cottonwood Canyons, Cache County, Utah; Bear-
tooth Butte formation, Beartooth Butte, Park County, Wyoming.

Diagnosis. — The dorsal disc has a median length in known speci-
mens of 132-159 mm. The ratio of its maximum width^ to its med-
ian length is 1.05 to 1.15. The ratio of the greatest width across the
anterior part of the disc to that across its postbranchial lobes is .73
to .78. The branchial notches are of moderate depth, and separated
from each other by 95 to 105 mm. The ornament of the dorsal disc
consists of rather coarse ridges, generally radiating laterally and
antero-laterally from the midline, but with many irregularities; the
ridges are subdivided into tubercles near the margins, and occasion-
ally over much of the disc.

Discussion. — Cardipeltis ivallacii, the type species, comes typi-
cally from the Water Canyon formation of northeastern Utah. When
Cardipeltis was discovered at Beartooth Butte, Wyoming, it was
first referred to C. wallacii, but later was removed by Bryant to two
new species. C. sinclairi Bryant was distinguished from C. wallacii
by the greater width of the anterior part of its dorsal disc, its gen-
eral proportions, and its deeper branchial notches. However, the
width and proportions are greatly influenced by the amount of flat-
tening, which is usually more at Beartooth Butte than in the Water
Canyon formation. If the dimensions are measured along the curv-
ature of each disc, the proportions computed from them are similar
in wallacii and sinclairi. As will be shown below, the depth of the
branchial notches increases as the disc grows, so this character must
be related to size. Moreover, in a dorsal view of an uncrushed disc
the branchial notches appear more shallow because of the curvature
of the disc. Actually the branchial notches are very similar in
wallacii and sinclairi. The other differences which I considered

' All dimensions have been measured along the surface curvature, not in a
straight line.

DENISON: CARDIPELTIS 93

valid in 1953 (p. 306), I now find either to be slight or variable. In
C. sinclairi the postbranchial lobes are not significantly more convex,
and the posterior margin is not more deeply emarginate. The lateral
margins anterior to the branchial notches and the anterior margin
are more convex in known C. sinclairi, but the differences are slight
and the shapes apparently variable. The ornamentation is extremely
variable. I conclude, therefore, that there is no valid reason for
distinguishing C. sinclairi.

The second species from Beartooth Butte, C. oblongus Bryant,
was distinguished from C. sinclairi largely by its narrower dorsal
disc. However, a considerable part of the right postbranchial lobe
of the type was not exposed when Bryant described this species, and
now that this specimen has been prepared, the maximum width
(following the curvature of the disc) is 165 mm., not much less than
in C sinclairi. The ratio of maximum width to the median length
is 1.05 (as compared to 1.10 in the type of sinclairi), and the ratio of
the greatest width across the anterior part of the disc to that across
the postbranchial lobes is .78 (as compared to .73 in the type of sin-
clairi). If allowance were made for the crushing of the left post-
branchial lobe of the type of oblongus, maximum width and the width
ratios would be even closer to those in sinclairi. Bryant also stated
that the posterior border of the dorsal disc was less emarginate in
oblongus, but this is no longer so since the type has been completely
exposed. Bryant's two species also agree closely in size and ornamen-
tation, so there appears to be no valid reason for separating them.
C. oblongus thus becomes a junior synonym of C wallacii.

Cardipeltis richardsoni, new species^

Type. — FMNH, PF3902, a dorsal disc, exposed on the outer side
(fig. 1).

Referred specimens. — PF3897, a nearly complete, articulated spec-
imen (figs. 3, 9), prepared on dorsal side; PF3893, an incomplete
dorsal disc, external side; PF3790, 3886, 3890-2, 3896, 3899, (fig. 4),
3903-4, dorsal discs, inner side.

Occurrence. —Early Devonian, Beartooth Butte formation, south
side of Cottonwood Canyon, SE }4, NE M, Sec. 4, T56N, R93W, on
the west fiank of the Bighorn Mountains, Bighorn County, Wyoming.

1 Named after Dr. Maurice L. Richardson of Lansing, Michigan, in grateful
recognition of his generous assistance to the Division of Paleontology at the
Field Museum of Natural History.

94

FIELDIANA: GEOLOGY, VOLUME 16

Diagnosis. — The median length of the dorsal disc in known speci-
mens ranges from 104-133 mm. The ratio of its maximum width to
its medium length is 1.03-1.15. The ratio of the greatest width

Fig. 1. Cardipeltis richardsoni, type, outer surface of an incomplete dorsal
disc, PF 3902 (X2/3). bro, branchial opening; brp, plates roofing branchial duct ;
pbl, postbranchial lobe.

across the anterior part of the disc to that across the postbranchial
lobes is .75-.83. The branchial openings are situated about half-way
between the midline and the lateral margins of the dorsal disc, and
are separated from each other by about 45-60 mm. The ornament
of the dorsal disc consists of rather coarse ridges, with many irregu-
larities in the details of their arrangement and a tendency to be
arranged parallel to the anterior and lateral margins.
Discussion. — See under C. bryanti.

DENISON: CARDIPELTIS

95

Fig. 2. CardipelHs bryanti, type, inner surface of dorsal disc, PF 3905 (X 2/3).
brd, branchial duct; gl, growth line; pbl, postbranchial lobe.

Cardipeltis bryanti, new species^

Type. — FMNH, PF3905, a dorsal disc, exposed on the inner sur-
face (fig. 2).

Referred specimens. — PF3895, an articulated specimen, prepared
on ventral surface (fig. 7) ; PF3883-4, dorsal discs showing the inner
surface.

Occurrence. — As for C. richardsoni.

• Named in honor of the late Dr. William L. Bryant, a student of Devonian
fishes.

96 FIELDIANA: GEOLOGY, VOLUME 16

Diagnosis. — The median length of the dorsal disc is 121-137 mm.
in known specimens. The ratio of its maximum width to its median
length is .92. The ratio of the greatest width across the anterior
part of the disc to that across the postbranchial lobes is .86-.87. The
branchial openings are situated about half-way between the midline
and the lateral margins of the dorsal disc, and are separated from
each other by about 45 mm. The ornament of the dorsal disc con-
sists of rather coarse ridges, but the pattern is not well known.

Discussion of C. richardsoni and C. hryanti. — The known speci-
mens of Cardipeltis from Cottonwood Canyon in Bighorn County,
Wyoming, all differ from C. wallacii in one striking characteristic,
the more central position of the gill openings, approximately half-
way between the midline and the lateral margins of the dorsal disc.
In C. ivallacii, on the other hand, the gill openings are near the mar-
gins of the disc, and are widely separated from each other. This is
the primary basis for specific distinction. The majority of speci-
mens from Cottonwood Canyon have a broad dorsal disc with the
width across the posterior part much greater than that across the
anterior part. These are referred to C. richardsoni. However, four
specimens have a distinctly narrower shield, in which the dorsal
disc is longer than wide (measured as always along the curvature),
and the anterior and posterior parts of the dorsal disc are more
nearly the same width. This difference in shape and proportions is
surely not the result of post-mortem distortion. Specimens from
this locality are sometimes distorted but this is due to compaction
and there is no evidence of stretching or compression in a horizontal
plane. The differences between C. richardsoni and C. hryanti are
thus believed to be valid biological distinctions, and they are assumed
to be specific, though it is not impossible that they may be sexual
or have some other explanation.

MORPHOLOGY

Previous descriptions of Cardipeltis have been based on dis-
articulated dorsal discs, marginal plates, ventral plates and scales.
The two newly discovered articulated specimens, PF3895 (fig.7) and
3897 (fig. 3) permit a more complete description and restoration.
The dorsal disc is bounded laterally by a series of marginal plates
which are sharply folded under to form well developed ventral
laminae. Anterior to the dorsal disc there is a short rostral region,
but this is poorly preserved. The ventral shield consists of the ven-
tral laminae of the marginal plates laterally, two or more large pos-

DENISON: CARDIPELTIS

97

Fig. 3. Cardipeltis richardsoni, dorsal side of a nearly complete, articulated
specimen, PF 3897 (X 2/3). bro, branchial opening; brp, plates roofing branchial
duct; dd, dorsal disc; ds, dorsal scales of tail; mg^, mg ,, first and fourth marginal
plates; pm, platelets formed around margin of dorsal disc; ro, rostral region.

toral plates anteriorly, and a large number of small polygonal tesserae
centrally. The tail is narrow and slightly longer than the dorsal
shield.

Dorsal disc: This has been described quite completely before by
Branson and Mehl (1931, pp. 520-521), Bryant (1933, pp. 309-310;
1935, p. 120) and Denison (1953, pp. 306-314). Its most prominent
external features are the branchial openings, which are much closer
to the midline in C. richardsoni and C. bryanti than in C. wallacii.
Two specimens of the former (figs. 1 and 3, bro) show that the

98 FIELDIANA: GEOLOGY, VOLUME 16

"branchial notches" of previous descriptions are roofed over laterally
by small dermal plates to enclose a pair of oval external branchial
openings. From these external openings, the branchial ducts extend
inward under the dermal covering in a direction slightly anterior of
lateral. The basal layer of the exoskeleton dips down to line the
branchial ducts internally, and the spongiosa thickens to form the
sides of the ducts.

The articulated specimen of C. richardsoni (fig. 3) throws some
light on the growth of the dorsal disc. The central part of the disc
has the dentine ridges arranged in an irregular pattern, and pre-
sumably it calcified as a unit. Stensio (1958, pp. 330-331) inter-
preted a similar ridge pattern of C. wallacii as indicating three growth
centers, but there is evidence against this, as will be shown below.
Growth of the dorsal disc, at least its superficial part, proceeded by
apposition around the margins, particularly the lateral and anterior
margins. The growth increments were not more or less continuous
growth rings, as in pteraspid discs, but a number of small platelets
that formed and grew around its margins. A narrow ring of small
platelets surrounds the disc of PF3897 (fig. 3, pm) and other platelets
that have been incorporated into the shield are recognizable by the
pattern of their dentine ridges. This type of growth is similar to
that described in Psammolepis by Heintz (1957, pp. 158-159), but
Cardipeltis differs in having only a very narrow band of platelets
instead of the wide one that occurs between the disc and marginal
plates of Drepanaspis and other psammosteids. The external branch-
ial openings seem to have retained the same position that they had
when the disc was first calcified. However, as the disc grew, the
branchial duct apparently also grew laterally and became roofed over
by the fusion of small platelets to the shield ; these platelets are still
recognizable by their pattern in PF3897 (fig. 3, brp).

The most prominent feature of the inner surface of the dorsal disc
(figs. 2 and 4, brd) is the pair of branchial ducts, which are partly
lined by the inwardly bulging basal layer. Centrally, the midline
of the disc is a shallow groove, becoming a distinct ridge in the pos-
terior part. No specimen shows internally any indication of the
platelets that were added to the shield during growth, so presumably
the platelets were superficial and were undergrown by the basal
layer when they became attached, as in Psammolepis. Parallel to
the disc margin many specimens show a series of gentle ridges that
presumably represent growth lines (figs. 2, 4, 5, gl). These growth
lines and the platelets, fused and unfused, that are visible externally
are the main evidence for the growth of the disc, as no juvenile

DENISON: CARDIPELTIS

99

Fig. 4. Cardipeltis richardsoni, inner side of dorsal disc, PF 3899 (X 2/3).
brd, branchial duct; brp, plates roofing branchial duct; gl, growth line

Cardipeltis have been discovered. The center of growth is indicated
on the inner side of the shield of some C. richardsoni (fig. 5, cr) by
a radiation of interrupted ridges. The center Hes in the midline
between the external branchial openings, and is evidence that growth
proceeded from a single center, not from three as claimed by Stensio
(1958, p. 331) in C wallacii. Most of the growth, judging from the
internal growth lines, took place laterally and anteriorly, but narrow-
er increments were also added to the posterior edge.

There is no indication on the inner side of the dorsal disc of gill
pouches^ or the other internal structures that are sometimes seen
in small cyathaspids and pteraspids. However, the position and
direction of the branchial ducts, and particularly their lateral and
internal terminations, indicate that the gills must have been situated

* These have recently been interpreted as indications of segmental muscles by
Tarlo and Whiting (1965).

100

FIELDIANA: GEOLOGY, VOLUME 16

Fig. 5. Cardipeltis richardsoni, restoration of inner side of dorsal disc showing
growth lines and aspidine radiation, based mainly on FF 3891-3892; and position
of lateral-line canals, based mainly on PF 3898 (X about 2/3). brp, inner surface
of plates roofing the branchial ducts; cdj^_^, first to fifth dorsal transverse sensory
commissures; cr, center of aspidine radiation; cso, commissure between supra-
orbital and medial dorsal canals; gl, growth lines; ig, inner rim of branchial duct;
Idl, mdl, lateral and medial dorsal sensory canals; soc, supraorbital sensory canal.

under the anterior half of the dorsal shield. The posterior part of
the shield must have enclosed the whole abdominal region, since
the tail was narrow and presumably entirely locomotor in function.
Marginal plates: These plates form the entire lateral margins of
the shield, and are sharply folded to form dorsal and ventral laminae
that enter into both the dorsal and ventral shields. They are com-
parable to the branchial plates of cyathaspids and pteraspids, and
are variable in number and shape. In the articulated C. bryanti (fig. 7,
mg, 1-6) there are six on each side. In the articulated C. richardsoni
(fig. 3, mgi, mg^) there are only four on each side, and the extent of
the posterior two of opposite sidles is not the same. PF3900, a frag-
ment of an articulated Cardipeltis sp., appears to have eight or nine
marginal plates on one side.

DENISON: CARDIPELTIS 101

The most anterior marginal plate is the longest. It is 52 mm.
long in the articulated C. richardsoni (fig. 3, mgi), but isolated plates
PF3950) are known as long as 73 mm. As in all the marginal
plates, the ventral lamina is wider than the dorsal. The ventral

Fig. 6. ICardipeiiis sp., inner suriace of orbital plate, PF 3932 (X 3).

lamina of the first marginal (fig. 7, mgi) is widest where it abuts
against the small polygonal tesserae of the ventral shield, but it
tapers rapidly anteriorly in a slightly concave edge that lies against
a postoral plate. The dorsal lamina (fig. 3, mgi) is wider posteriorly
where it lies opposite the dorsal disc, and tapers slightly anteriorly
opposite the rostral region. One complete plate (PF3947) resembles
the posterior two-thirds of the first marginal, so apparently this
plate can occasionally be subdivided. Most of the ventral lamina
is covered with very broad (to 5 mm.), irregular, fiat-topped, closely-
spaced dentine ridges. Along its medial, anterior and posterior edges
are rows of narrower ridges (about 1 mm. wide) that are more
numerous on larger plates; these ridges presumably represent growth
increments. On the dorsal lamina the ridges are narrower (about
1 mm. wide) and similar to those of the dorsal disc; these ridges are
more or less parallel to the anterior, medial, and posterior edges, and
include near those edges, ridges or tubercles that presumably re-
present growth increments. Near the lateral margin of certain
specimens, on both dorsal and ventral laminae, the dentine ridges
become higher and often sharp-crested, more widely spaced, and
sometimes strongly scalloped on their margins. The crests of the
ridges near the margin, especially on the ventral lamina, are common-
ly, or even usually, dull, due to the absence of the durodentine and
the superficial part of the dentine (fig. 8C). This looks like the re-
sult of abrasion, but as shown below (p. 112), this explanation is
not completely satisfactory.

102

FIELDIANA: GEOLOGY, VOLUME 16

vt^

Fig. 7. Cardipeltis bryanti, ventral side of nearly complete articulated speci-
men, PF 3895 (X 2/3). Ivl, lateral ventral sensory canal; mgj^_Q, first to sixth

marginal plates; pol, postoral sensory canal; pop, postoral plate; vs, ventral scales;
vt, ventral tesserae.

The second marginal plate is somewhat shorter and of more uni-
form width. Its length is 43 mm. in the articulated C richardsoni
(fig. 3), 40 mm. in the articulated C bryanti (fig. 7), while the largest
disassociated plate (C sp., ^F3957) is 55 mm. long. The ridge
pattern is much as in the first marginal. The third marginal is very
similar, but still shorter. PF3900, which is a fragment of a small,
presumably young adult, articulated Cardipeltis sp., has two short

DENISON: CARDIPELTIS 103

plates occupying the position of the second marginal of the other
articulated Cardipeltis. It has two plates, though not so clearly-
distinct, occupying the position of the third marginal. Isolated
marginals, as well as the second and third marginals of the articulated
specimens (fig. 7, mga), may show in their ornamentation a partial
subdivision into two plates. It is probable that such marginals arose
as two or more smaller plates that later fused to form the adult
plates. The most posterior marginals are variable. In PF3895
marginals 4 to 6, and in PF3900 the posterior marginals, are of de-
creasing size, the last being as small as adjacent scales. In PF3897
(fig. 3, mQi) this region is occupied by a single long, narrow marginal.
Along the ventral laminae of the anterior marginals are occasional
pores of a lateral line canal.

The marginal plates of Cardipeltis sp. from Utah previously
figured by me (1953, fig. 69) may be provisionally identified as
follows: first marginals. A, B, G; second marginal, H; marginals
from mid-length, C-F.

Rostral region: The articulated Cardipeltis richardsoni (fig. 3)
is the only specimen with a rostral region, but it is so badly crushed
that little can be made of its structure. Only two deductions are
possible. First, this area appears to be covered by many small
plates, rather than a large rostral plate such as occurs in pteraspids
and psammosteids. Second, the rostral region is short, extending
probably no more than 2 or 3 cm. in front of the dorsal disc. Clearly
there was no extensive anterior shield as restored by Stensio (1958,
fig. 184 A). If my identification of the lateral line canals is correct
(see p. 109), the posterior ends of the supraorbital canals lie on the
most anterior part of the dorsal disc, and the pineal organ must have
underlain the most anterior part of this disc, or possibly a small
pineal plate immediately in front of it.

Orbital plates: The articulated specimens of Cardipeltis do not
preserve any orbits, and if they were once present, as is probable,
they must have been far anterior (presumably just in front of the
first marginal plate in the badly crushed area of PF3897 (fig. 3).
The collection from Cottonwood Canyon in the Bighorn Moun-
tains contains an isolated orbital plate (PF3932, fig. 6) originally
exposed to show its inner surface. This has been imbedded in plastic
and the matrix removed from the opposite side, but unfortunately
none of the superficial layer was preserved, so it was impossible to
demonstrate its affinity to Cardipeltis. If it belongs to a group other-
wise represented in the large collections from this locality, it could

104 FIELDIANA: GEOLOGY, VOLUME 16

only be an osteostracan or a heterostracan. The Osteostraci are
unlikely because only a single small fragment of this order has been
found. If it is a heterostracan, it belongs most probably to Cardipeltis
or to a pteraspid. The shape of PF3932 is distinctly different from
that characteristic of pteraspids; it is flat, has a nearly square body
with the small orbit in its presumed antero-lateral corner, and has
a narrow, presumably posterior process. This orbital plate probably
belongs to Cardipeltis.

Ventral shield: This is known largely from the articulated speci-
men of Cardipeltis bryan.ti (PF3895, fig. 7). Its lateral margins are
formed by the series of six marginal plates (4-8 in other specimens)
described above. Between the marginals the shield is formed by
two or more large postoral plates anteriorly, and a large number of
small plates or tesserae (fig. 7, vt) over the rest of the area. The
latter are polygonal (usually five or six-sided) and flat -topped. With
the exception of two elongate plates, they range in maximum length
in PF3895 from about 14 mm. anteriorly and centrally, to about 4
mm. posteriorly. Similar isolated tesserae in the collection are as
long as 19 mm. The ornament typically consists of an irregularly
oval, flat, central dentine area separated from a surrounding dentine
ring by a narrow intercostal groove. The ring of dentine is often
subdivided into two or more parts, and may represent the first growth
increment around the original oval plate. A few plates on PF3895
have a partial second ring of dentine, much narrower than the first
ring. One isolated ventral plate (PF4016) has five interrupted
growth rings, although only two of them are complete. Occasionally
two adjacent plates with originally distinct central areas may fuse
and continue their growth as a single elongate plate (fig. 8B). How-
ever, two elongate plates on PF3895 (fig. 7), approximately 28 mm.
in maximum dimension, have a single elongate central dentine area;
these plates are equivalent in extent to three or four of the adjacent
plates. A fragment of a ventral shield (PF3934) consists of nine
ventral plates that have apparently fused, and another fragment
(PF3907) has three or four fused plates. Fusion of ventral tesserae
probably was unusual, however. It is possible in places, but un-
likely, on the articulated C. bryanti (fig. 7). It clearly has not
happened in PF3791 and 3900, two partial articulated specimens
that show the ventral shield. Qn the posterior part of the ventral
shield the plates become smaller (to about 4 mm.) and more scale-
like. Instead of a flat, central dentine area, they have a ridge that
may have a sharp crest and scalloped edges. This is surrounded by
convex-topped ridges separated from each other and from the central

DENISON: CARDIPELTIS

105

iR^'1

^A*

Fig. 8. Cardipeltis sp. (X 4/3). A, postoral plate, PF 3936; B, fused ventral
tesserae, PF 3933; C, ventral lamina of marginal plate showing removal of super-
ficial part of dentine, PF 3948.

ridge by wide intercostal grooves. Very rarely a ventral plate has
a pore that may represent the opening of a sensory canal. A plate
of the ventral shield of Cardipeltis sp. from Utah was figured by me
in 1953 (fig. 70A).

Postoral plates: At the front end of the ventral shield between the
anterior parts of the first marginal plates lie two or more large plates
(fig. 7, pop) . These may be compared to the postoral plates of most
pteraspids and of certain cyathaspids, and to the "Komplexplatten"
of Drepanaspis (Gross, 1963, p. 140). They appear to be variable
in size, shape and number. Based on an interpretation of their
dentine pattern, they grow from irregularly rounded centers by the
addition of broad bands of dentine. Two or three centers may fuse
to form a large postoral plate, or two may fuse and one remain
separate so that there are two plates on one side, one behind the
other. In the articulated Cardipeltis bryanti (fig. 7) there are three
plates in a lateral row, one approximately median, and one on either
side of it, but they are not developed quite symmetrically. It is
possible that a median plate was usually present and such does
sometimes occur in Drepanaspis. Growth of a very large postoral

106 FIELDIANA: GEOLOGY, VOLUME 16

plate (fig. 8A) may be interpreted from its dentine pattern as follows:
three irregular, roundish dentine areas formed close together, one
behind the other; these were fused at the next growth stage when they
were surrounded by a broad band of dentine. Several additional
dentine bands were added during further growth of the plate as a
single unit, but these were not of uniform width; they were very wide
on the lateral side, were less developed anteriorly and posteriorly,
and were very narrow or absent on the medial side. The last two
increments were narrow ridges on the lateral side only.

A prominent lateral line, identified as the lateral ventral canal
(fig. 7, Ivl), extends antero-posteriorly across the lateral half of the
postoral plate, well laterad of the growth centers. It is usually de-
veloped as a more or less continuous canal with pores on either side
of it, often connected to the main canal by open grooves. A few
pores of the postoral canal extend obliquely from the antero-lateral
corner towards the median line (fig. 7, pol).

Postoral plates of Cardipeltis have been figured previously by
Bryant: in 1933 (pi. 21, fig. 1) as "Gen. and sp. indet."; and in 1935
(pi. 17, figs. 1-2) as "Innominate plate."

Because of the poor preservation of the anterior ends of both
articulated specimens of Cardipeltis, the mouth is not known. Pre-
sumably it lay in front of the postoral plates, was terminal or nearly
terminal, and was provided with oral plates, much as in Drepanaspis.

Tail: The tail is preserved only in the articulated specimens of
C. richardsoni (figs. 3 and 9) and C. bryanti (fig. 7). In the former,
its length is about 190 mm., somewhat longer than the dorsal shield,
which is estimated to be about 155 mm. The tail is narrow: in the
articulated C. bryanti its maximum width at the base is 27 mm.
compared to the maximum width of the shield of 122 mm.; in the
articulated C. richardsoni it was at least 30 mm. wide, but the shield
of this specimen has a maximum width of about 180 mm. Its shape
in cross-section anteriorly was probably subtriangular, with a flat
base and a rounded dorsal edge. The size of the tail relative to the
shield was small, so presumably the abdomen did not extend into it
and it was entirely locomotor in function.

Except for ventral scales, isolated scales are very rare in the
collection, presumably because their outer surface adheres to the
matrix and only their base is exposed. The ventral scales (fig. 7, vs)
are very large, paired, and arranged in a double row along the lower
surface of the tail. Their exposed part is subrectangular, wider
than long, with a maximum width of about 12 mm. in PF3895, but

DENISON: CARDIPELTIS

107

Fig. 9. Cardipeltis richardsoni, ventral view of articulated specimen, PF 3897,
showing ventral and lateral sides of tail ( X 2/3). ds, dorsal scales; vs, ventral scales.

as wide as 19 mm. in one isolated scale. Anteriorly on each scale
there is a wide unornamented area that is overlapped by the scale
in front; this extends forward into a long process laterally. Between
the large paired scales is an occasional small median scale. The tail
is incompletely preserved in PF3895 (fig. 7), so the posterior develop-
ment of the ventral scales is unknown. However, the ventral scales
in the part preserved show a gradual decrease in size posteriorly,
and at the end of the preserved part one larger crested median scale
appears. The dentine ridges of the ventral scales in PF3895 are
short, well spaced, and directed obliquely postero-laterally, except
that laterally they are parallel to the margin. On C. richardsoni
(PF3897, fig. 9, vs) the ridges are similarly arranged, but spaced more

108 FIELDIANA: GEOLOGY, VOLUME 16

closely. Isolated scales show differences such as development of
tubercles on lateral or anterior margins, or an antero-posterior arrang-
ment of ridges.

The rest of the tail is known only from the articulated C. richard-
soni (PF3897, figs 3 and 9) . It is covered with relatively small scales
that decrease in size posteriorly and near the tip may be less than
2 mm. in length. No strongly developed ridge scales are apparent,
so it is probable that the dorsal edge of the tail was rounded, especial-
ly in its anterior half. The shape of these scales is somewhat rhom-
bic, and they are covered with short, crenulated dentine ridges
arranged longitudinally, and becoming finer on the small posterior
scales. The shape of the posterior part of the tail is not precisely
determinable. A long dorsal lobe (fig. 9, ds) is separated by a deep
split, 40 mm. long, from a median lobe that is probably slightly
longer. This in turn, is separated from the shorter ventral lobe
by a deeper split, about 70 mm. long. I know of no comparable
subdivision of a tail in a modern fish, so it is tempting to assume that
this is the result of crushing during preservation. However, the
splits are bordered by evenly aligned scale edges which make them
look like natural edges.

Lateral Sensory Canal System: In 1953 (pp. 309-310) I described
the pattern of the lateral line system of Cardipeltis wallacii. The
pattern was indicated by linear arrangements of large pores that
lay between or notched the dentine ridges. Some of these pores are
visible in specimens of C. richardsoni (PF3893, 3897). One dorsal
disc of C. sp. from Cottonwood Canyon in the Bighorn Mountains
(PF3898) is preserved in such a manner that the lateral line pattern
is quite completely shown. This disc was crushed in such a way that
the chambers of the spongiosa are depressed and the lateral line
canals stand up in relief on the inner surface of the disc. The pat-
tern agrees with that figured for C. wallacii (Denison, 1953, fig. 68)
except that additional canals are apparent. There is another pair
of transverse commissures (fig. 5, cdi) lying between the last two
previously identified in C. tvallacii. This commissure has now been
found in C. ivallacii also; a few pores from it are doubtfully identified
in PF804 (Denison, 1953, fig. 66), and its pores are quite completely
visible at the left posterior part of another incomplete dorsal_disc
(PF895; Denison, 1953, fig. ^7). This means that there were quite
certainly five dorsal transverse commissures behind the supraorbital
canals of Cardipeltis, as compared to four in the Cyathaspididae.
Another pair of canals previously unknown in Cardipeltis can be seen

DENISON: CARDIPELTIS 109

between the anterior ends of the median dorsal lines of PF3898 (fig. 5,
soc). These can only be the supraorbital canals, or supraorbital and
pineal canals of Stensio's terminology (1958, p. 401), and demonstrate
that there was no important additional part of the lateral line system
anterior to the dorsal disc, as restored by Stensio (1958, fig. 184A).
The supraorbital canals are connected with the anterior parts of the
median dorsal lines by a short commissure (fig. 5, cso) which has not
been identified in other Heterostraci. A few pores of the supraorbital
canal are possibly identifiable in C. wallacii (PF804).

The ventral sensory canals are poorly developed and are known
only in the articulated C. bryanti (fig. 7) and in isolated postoral
(fig. 8A) and marginal plates. A row of pores extending obliquely
toward the midline from the anterolateral corner of each postoral
plate is presumably the postoral canal of cyathaspids and pteraspids
(fig. 7, pol). A prominent canal extending antero-posteriorly in the
outer half of each lateral postoral plate may be the lateral ventral
line (fig. 7, Ivl). Scattered pores along the medial margins of the
ventral laminae of the marginal plates may be related to the canals
near the lateral margins of the ventral shield of Amphiaspis, con-
sidered by Stensio (1958, p. 404) to be a hypotrematic series not
known in other vertebrates. Only a single pore has been identified
on the small tesserae of the ventral shield, so presumably the ventral
transverse commissures and medial ventral canals were reduced and
possibly largely absent.

Histology. The relatively abundant and well preserved material
of Cardipeltis from Cottonwood Canyon in the Bighorn Mountains
permits a much more detailed description of its histology than was
possible in my earlier paper (1953, pp. 307-308). In that paper I
stated that the total thickness of the shield was 1.3-1.5 mm. in
C. wallacii, but it is now apparent that this was based on crushed
sections. In sections of the dorsal disc of Cardipeltis sp. from the
Bighorn Mountains the thickness ranges from 2.0-3.2 mm., but even
here there has been some crushing of the spongiosa. A section across
the median internal ridge of the dorsal disc (slide 4670) measures
4.5 mm. in thickness, and this has been thinned considerably by
collapse of the spongiosa. So the Cardipeltis shield really is relatively
thick for one of the Heterostraci, much thicker than those of Cyatha-
spididae and Pteraspididae, though exceeded by some Psammosteidae.

The superficial layer (figs. 10, 11 A) consists of ridges of dentine
separated by intercostal grooves. On the surface of the ridges is a
thin, transparent, presumably densely calcified layer, penetrated at

no

FIELDIANA: GEOLOGY, VOLUME 16

its base by the tips of the dentine tubules. It is assumed to be duro-
dentine. The dentine has a thickness of .19-.24 mm. above the pulp
canals, and often shows a distinct lamination that continues into the

■Wa^-' ^

i m A

'^. -<«^

1

'*"*l8S

^

im ,. mafw%'^

^

^

.' »^

IfnWr'f^iiiW

ifgH

' -' -3

■ft.' .^^^IflHHHil

Fig. 10. Cardipeltis sp., transverse sections through dorsal shield. A, outer
part of middle layer, and tubercle showing superficial removal of dentine at its
center, slide 4669 (X 62); B, section through entire thickness of dorsal disc,
slide 4684 (X 20).

aspidine surrounding the intercostal grooves. The ridges of the
dorsal disc are 1 mm., more or less, in width, and usually have a
slightly convex outer surface, but they may be sharply crested near
the edges of the marginal plates. Their margins are usually gently
scalloped on the dorsal disc (fig. 11 A) and more strongly so on the
marginal plates. On the ventral surface of the marginal plates, the
ridges become wide and flat, and on the plates of the ventral shield
there are wide areas of dentine separated by narrow grooves. On the
dorsal disc, the intercostal grooves are widely open, and typically
are pear-shaped in section.

In the dentine ridges, the tubules extend from the pulp chambers
towards the surface and sides of the ridges. Near the surface they

DENISON: CARDIPELTIS

111

branch into tufts of small twigs, but otherwise there is little or no
branching of tubules. The pulp chambers are scattered along each
ridge, two or three in the width of one ridge (fig. 11 A). They do not
unite or connect with a continuous pulp canal, at least in adults, but
are connected by vertical or oblique canals to an irregular system of
canals at the base of the ridge. The latter connect by occasional
horizontal canals to the intercostal grooves, and by vertical canals
to the canal system of the underlying spongiosa. The material of
the lower part of the ridges is aspidine, arranged mostly in lamellae
around the various canals.

On some ridges, and occasionally over small areas of the shield,
the crests of the dentine ridges are dull due to the removal of their
surface (fig. 8C). Thin sections reveal that only the durodentine
and the most superficial part of the dentine are gone (fig. lOA).
This removal is common and even usual along the edges of the
marginal plates, especially on their ventral laminae. It maj'' extend
over much of the ventral lamina (PF3948, fig. 8C), or be scattered
over the dorsal lamina (PF4004), and also occurs here and there on
some dorsal discs (PF3893, 3902) and on ventral tesserae and scales.
The common surface removal on the edge of the marginal plates
resembles the condition in some psammosteids, where it is explained

Fig. 11. Cardipeltis sp., tangential sections through dorsal disc (X 20). A,
section through ridges of superficial layer, showing dentine on left and pulp canals
on right, slide 4667; B, section through spongiosa, showing probable branched
lateral-line canal in upper right, slide 4685.

112 FIELDIANA: GEOLOGY, VOLUME 16

as the result of abrasion (Heintz, 1957, pp. 159-162). There are
some difficulties with this explanation in Cardipeltis. There are
places where the removal has formed small distinct concavities on
the crests of otherwise convex ridges (PF3955, 4004) ; or the duro-
dentine may retain quite sharply elevated edges around areas of
dentine removal (PF3948). These particular examples of removal
cannot be explained as the result of wear. Resorption is a possibility,
but only if the dentine ridges had retained a covering of soft tissue.
A more likely explanation is that the covering skin was abraded,
permitting an infection by bacteria, protozoans, or algae which
caused the removal of the underlying durodentine and dentine.
This type of damage is known to occur to bones of modern fishes and
other aquatic vertebrates (Schaperclaus, 1954, p. 552.)

The middle layer of the Cardipeltis shield may be divided into an
outer part with relatively small chambers or canals, and an inner spon-
giosa with relatively large chambers. The outer part (figs. lOA, B)
is seen in some sections to be a horizontally laminated layer of aspi-
dine about 0.3 mm. thick, immediately underlying the laminated lin-
ing of the intercostal grooves and the base of the ridges. Its canals
range in diameter from 0.06 to 0.22 mm., except for one large canal,
0.72 mm. in diameter, that may be a lateral line canal. Some of the
canals pass up into the canal system at the base of the ridges, while
others extend horizontally or parallel to the lamination.

The large-chambered spongiosa, or inner part of the middle layer,
is usually more or less crushed in the material that I have examined.
In the better preserved specimens, the chambers generally are round-
ed or subrectangular, usually 0.6 to 1.4 mm. in maximum diameter,
and surrounded by walls of aspidine that are laminated parallel to
the surface of the chambers (fig. IIB). In one section (slide 4684,
fig. lOB) the chambers are unusually large, as much as 1.7 mm.
horizontally and 1.6 mm. vertically; in this specimen the spongiosa
resembles that of Cyathaspididae and Pteraspididae, though the
chambers are less regular. The walls of the chambers are typically
0.13-0.22 mm. thick, but in one specimen (slide 4086) the walls are
thicker and the chambers smaller. Two thin sections (slides 4082,
4086) show structures radiatmg from the surface of the chambers and
at right angles to the lamination; these are not tubules, and may be
calcified fiber bundles. A few canals penetrate the walls and connect
adjacent chambers, and a few enter through the basal layer. In cer-
tain areas where the shield is thick, as in marginal plates, in the

DENISON: CARDIPELTIS 113

median inner ridge of the dorsal disc, and at the sides of the branchial
openings, the increased thickness is due almost entirely to an expan-
sion of the spongiosa.

The basal layer of the dorsal shield (fig. lOB) ranges in thickness
from about 0.5 to 0.8 mm. It is strongly laminated parallel to its
lower surface, and often shows more or less vertical structures that
may be Sharpey's fibers. Relatively few ascending canals penetrate
the basal layer to the chambers of the spongiosa. In the ventral
plates the basal layer may be thinner (0.3 mm.) or even absent.

The character of the spongiosa of Cardipeltis usually distinguishes
it from pteraspids, cyathaspids and probably traquairaspids, in which
the middle layer mainly consists of large, prismatic chambers. It
also differs from psammosteids, at least the later ones, in which the
spongiosa is very thick and filled with many small, rounded chambers.
Its histology is closest to that of certain Late Silurian and Early
Devonian Heterostraci with broad dentine ridges, particularly Cor-
vaspis, Oniscolepis, and Kallostrakon. Kallostrakon has narrower
ridges with a simpler nutrient canal system, but its dentine tubules
are similar to those of Cardipeltis; the middle layer has a more typ-
ical spongiosa with many irregularly shaped chambers, and the basal
layer, and in fact, the whole shield is thinner. Corvaspis has a large-
chambered middle layer in the center of its plates, and a fine-cham-
bered spongiosa near the margins; in places it shows great similarity
to that of Cardipeltis. The dentine ridges of Corvapis differ in having
the tubules arise in clusters, and in the absence of any marked hor-
izontal arrangement of the nutrient canal system. Oniscolepis differs
in its rather thin spongiosa with small rounded chambers. All of
these genera have typically very fibrous aspidine, more so than in
Cardipeltis, but this is possibly a result of more favorable preservation.

ADAPTATION

Cardipeltis was a very broad-shielded fish, with a relatively flat
body. Its dorsal shield may have been moderately convex, but the
ventral surface of its shield and tail were surely rather flat. The
mouth must have been nearly or quite terminal in position, and the
gill openings dorsal. All of these characters suggest a bottom-
living fish. The relatively small tail was not a powerful swimming
organ, and its large ventral scales must have reduced its flexibility.
In addition to this, the thick, heavy external skeleton indicates that
Cardipeltis was a relatively sluggish swimmer. Adaptively it is most
closely similar to the Psammosteidae.

114 FIELDIANA: GEOLOGY, VOLUME 16

The Beartooth Butte formation at Cottonwood Canyon is
thought to be a stream deposit, and the abundance of Cardipeltis
here and its occasional articulation indicates that it was a stream
dweller. The same is probably true at Beartooth Butte in north-
western Wyoming, but in Utah Cardipeltis occurs in the Water
Canyon formation which is at least in part marine. One cannot
conclude, therefore, that Cardipeltis was restricted to fresh waters.

RELATIONSHIPS

There is no question about the relationsip of Cardipeltis to the
Heterostraci; it is confirmed by the arrangement of the exoskeletal
plates, the single pair of branchial openings, and the histology.
Cardipeltis has, however, been assigned various relations within the
Heterostraci: to the Cyathaspididae, Pteraspididae, Psammosteidae,
Corvaspididae, or to a family or higher category of its own. Its
greatest resemblance is perhaps to the Psammosteidae, but this is
largely a result of a similar adaptation, and is believed to be due to
convergence. One of the most important differences between Cardi-
peltis and psammosteids is the position of the paired branchial
openings: in Cardipeltis they notch the dorsal disc and may become
enclosed by the disc; in psammosteids the position is like that of
pteraspids, — at the posterior end of the branchial plates, bounded
medially by the cornual plates. This alone means that Cardipeltis
is not related to either psammosteids or pteraspids, except perhaps
remotely through a common ancestor. Traquairaspids are also
distinct from Cardipeltis in that their branchial openings penetrate
the branchial plates.

There is a distinct possibility that Cardipeltis is related to certain
late Silurian and early Devonian Heterostraci with broad dentine
ridges, such as Kallostrakon, Corvaspis, and Oniscolepis. Kallo-
strakon is perhaps closest in its type of ornamentation and in its
histology, but this is of little help because it is known only from
fragments and is of uncertain affinities itself. This is true also of
Oniscolepis. Corvaspis is better known, but still only from fragments,
so that it is not yet possible to attempt a reconstruction of its shield,
or to determine its precise relationships.

There remain two theories of the origin of Cardipeltis: 1) that it
is derived from cyathaspids by a subdivision of the shield ; or 2) that
it evolved from primitive Heterostraci in which the shield was formed
of small tesserae. Derivation from Cyathaspididae would involve
a number of changes in addition to those related to adaptation to

DENISON: CARDIPELTIS 115

bottom living. The rostral area would have to separate from the
rest of the dorsal shield and become divided into small plates. Prob-
ably the lateral epitega of cyathaspids would have to separate from
the dorsal shield and surround the orbits. The branchial plates
would have to subdivide into a number of marginal plates. The
ventral shield would have to subdivide into tesserae. Anterior
ventral plates would have to form the postoral plates. The plates
would have to be separated by growth zones, permitting the attain-
ment of the relatively large size of CardipelHs. The tail would have
to lose its dorsal and ventral ridge scales and evolve large, paired
ventral scales. The shield would have to thicken and develop wider
dentine ridges and a less regularly arranged spongiosa. None of
these changes are impossible nor even unlikely, but, when taken all
together, derivation from cyathaspids seems most improbable.

Derivation from a heterostracan with a shield composed of tes-
serae would appear more probable. There is evidence that such a
condition was primitive in Heterostraci : it occurs in Ordovician
Heterostraci ; fused tesserae are evident in some early cyathaspids;
and other genera with fused or unfused tesserae are not unusual in
late Silurian and early Devonian strata. Derivation of CardipelHs
from such an ancestor would involve the formation of larger plates
dorsally, and at the same time retention of tesserae on the ventral
shield, on the rostrum, and between the growing plates of the dorsal
shield. This appears to be the more likely hypothesis and fully
justifies the retention of Cardipeltis in a family of its own, the
Cardipeltidae.

SUMMARY

Bryant's species of Cardipeltis, C. sinclairi and C. oblongus, are
referred to C ivallacii Branson and Mehl. New material of Cardi-
peltis, including articulated specimens, from the Early Devonian
Beartooth Butte formation, exposed in the flanks of the Bighorn
Mountains east of Lovell, Wyoming, is referred to two new species,
C. richardsoni and C. bryanti. The dorsal shield consists of a large
dorsal disc notched or pierced by the pair of external branchial
openings, a series of paired marginal plates, a rostral region composed
apparently of small plates, and probably of paired orbital plates.
The ventral shield consists, between the marginal plates, of a large
number of isolated tesserae, and two or more large postoral plates
anteriorly. The mouth was probably nearly terminal. The tail was
narrow, slightly longer than the shield, and flat-based anteriorly.

116 FIELDIANA: GEOLOGY, VOLUME 16

Additional canals are described in the dorsal lateral-line system, but
ventrally this system was reduced. The exo-skeleton was thick and
had a rather thick spongiosa of somewhat irregular, rounded cham-
bers. Cardipeltis is believed to have been a bottom dweller in
streams and perhaps in sea margins. It was probably derived from
primitive Heterostraci with a shield composed of isolated tesserae,
and is best retained in a family of its own, the Cardipeltidae.

REFERENCES

Blackstone, D. L., Jr. and P. O. McGrew

1954. New occurrence of Devonian rocks in north central Wyoming. Billings
Geol. Soc, Guidebook, 5th Ann. Field Conf., pp. 38-43, 2 figs.

Branson, E. B. and M. G. Mehl

1931. Fishes of the Jefferson formation of Utah. Jour. Geol., 39, pp. 509-531,
figs. 1-2, pis. 1-3.

Bryant, W. L.

1933. The fish fauna of Beartooth Butte, Wyoming. Part 1.- The Heter-
ostraci and Osteostraci. Proc. Amer. Phil. Soc, 72, pp. 285-314,
figs. 1-8, pis. 1-21.

1935. Cryptaspis and other Lower Devonian fossil fishes from Beartooth
Butte, Wyoming. Proc. Amer. Phil. Soc, 75, pp. 111-128, figs. 1-2,
pis. 1-18.

Denison, Robert H.

1953, Early Devonian fishes from Utah. Part II. Heterostraci. Fieldiana:

Geol., 11, pp. 291-355, figs. 61-85.

Gross, Walter

1963. Drepanaspis gemuendenensis Schliiter. Neuuntersuchung. Palaeontogr.,
121, Abt. A, pp. 133-155, figs. 1-11, pis. 6-9.

Heintz, Anatol

1957. The dorsal shield of Psammolepis paradoxa Agassis. Jour. Paleont. Soc.

India, 2, pp. 153-162, figs. 1-3, pis. 17-19.

Sandberg, Charles A.

1961. Widespread Beartooth Butte formation of early Devonian age in
Montana and Wyoming and its paleogeographic significance. Bull.
Amer. Assoc. Petrol. Geol., 45, pp. 1301-1309, figs. 1-3.

SCHAPERCLAUS, WiLHELM

1954. Fischkrankheiten. 3rd ed. Akademie-Verlag, Berlin, pp. i-xii, 1-708,

389 figs.

Stensio, Erik )

1958. Les cyclostomes fossiles ou ostracodermes. In Grass^, P.-P., Traits de

Zool., 13, fasc. 1, pp. 173-425, figs. 107-219.

Tarlo, L. B. and H. P. Whiting

1965. A new interpretation of the internal anatomy of the Heterostraci
(Agnatha). Nature, 206, pp. 148-150, fig. 1.

Y