The osteology of Protostega

I
,1

MEMOIRS

OF THE

CAKNEGIE MUSEUM.

VOL. II. NO. 7.

THE OSTEOLOGY OF PROTOSTEGA.
By G. R. Wieland.

The first mention of ancient gigantic marine turtles from America was made by
Cope in 1871 in a letter to Professor J. P. Lesley containing an account of a journey
in the valley of the Smoky Hill River in Western Kansas. This letter, as subse-
quently published in the Proceedings of the American Philosophical Society,1
includes a preliminary notice of the huge Niobrara Cretaceous turtle, Protostega
gigas. The type specimen was collected by Cope himself on a bluff near Butte
Creek in the vicinity of Fort Wallace. It was secured in a more or less fragmentary
condition after the manner of the vertebrate collecting of the earlier days of western
exploration, for, in the more extended description given in the Cretaceous Vertebrata
of the West, Cope says this fossil is made up of more than eight hundred separate
fragments.

The various parts of Protostega gigas (type), although mostly in situ, were con-
siderably removed from their natural position. The original specimen is now in the
collection of the American Museum of Natural History. It includes much crushed
cranial elements with portions of the lower jaw, ten nearly free ribs, several verte-
bra?, various. plastral elements, the shoulder girdle, a humerus, radius, and ulna, and
several metacarpals and marginals, as well as parts of uncertain position. Owing,
however, to the manner in which the ribs lay athwart the rather imperfect plastral
plates, the latter were supposed to be dorsal, and to represent a very primitive con-
dition of carapacial development with large fenestra. The radius and ulna were

1 Vol. XII., p. 175.

279

280 MEMOIRS OF THE CARNEGIE MUSEUM

not distinctly recognized as such although clearly figured. Owing also to a palpable
numerical error in the measurements of the cranium (" .50 M. = 24^ in.") the total
length of the original animal was estimated at thirteen feet, and is so referred to in
text-books ! The actual length is far less, as will clearly appear in the present
description of a much completer cotype almost exactly the size of the original Cope
specimen. This error was, however, in a sense prophetic, as some of the turtles of the
related Dakotan genus Arckelon, discovered by the writer twenty-four years after the
first Protostega, did actually reach, or possibly exceed, the enormous size of thirteen
feet in length.

Protostega remained a very vaguely known turtle until Baur 2 pointed out that
it must in its main characteristics agree with the ( !heloniidae, and that the plates
Cope supposed were dorsal must be plastral. That such was the fact was later more
definitely shown by Hay,3 who figured the nuchal and hyo- and hypoplastron of
another Niobrara specimen.

The next contributions to our knowledge of the osteology of the Protosteginre
were made by the writer, after his discovery of gigantic turtles east of the Black
Hills in the Fort Pierre Cretaceous in the summer of 1895. In the communica-
tions,4 which soon followed, the structure of the greater part of the carapace and
plastron was made known from remarkably preserved specimens. All the larger
limb bones were also determined and figured ; for the first time, indeed, in the case
of any extinct sea-turtles of America. There immediately followed these papers the
important contribution of Case,5 which, in addition to a careful discussion of the
systematic position of Protostega, added more particularly to a knowledge of the
cranial characters, as based on the description of various crushed, but otherwise well
defined disarticulated elements. The pelvis was also made known.

Further facts concerning the general cranial type in the Protosteginse were next
given by the writer6 in a paper describing the splendid skull of Arckelon, now on
exhibition in the Yale Museum. Later an attempt was made by Williston7 to
restore tbe tarsus and give the organization of the bind Hipper of Protostega, and by

2 (a) "Die Systematische Stellung von Dermochelys Blainville," Biolog. Centratblatt, IX., 1889. (J) "On the
Classification of the Testudinata," American Naturalist, XXIV., 1891).

3 " On Certain Portions of the Skeletion of Protostega gigas," Field Columbian Museum Publication, No. 7, 1895.

'(«) " Arclulon ischyros, a New Gigantic Cryptodiran Testudinate from the Fort Pierre Cretaceous of South
Dakota," Am. Jour. Science, December, 1895. (h) "The Protostegan Plastron," Ibid., January, 1898.

5 "On the Osteology and Relationships of Protostega,' ' Journal of Morphology, Vol. XIV. (This publication bears
the date 1897, but did not appear until some time in June, 1898, its presumptively true date. )

b " The Skull, Pelvis, and Probable Relationships of the Huge Turtles of the Genus Arckelon from the Fort Pierre
Cretaceous of South Dakota," Am. Jour. Sci , Vol. IX., April, 1900.

T "On the Hind Limb of Protostega," Ibid., Vol. XIII., Apiil, 1902.

WIELAND : THE OSTEOLOGY OK PROTOSTEGA 281

Wieland8 to show the carpal organization of Archelon ; but, as will be shown below,
both these efforts are almost entirely in error. From the foregoing review of the
slow progress of our knowledge of the Protostegimo it is seen that the third of a
century which has elapsed since Cope's discovery of Protostega gigas has not sufficed
to bring forth an entirely complete restoration of any single individual of these great
sea-turtles. How welcome then has been the discovery during the past two years
by Mr. Charles Sternberg in the Niobrara ( Vetaceous of Western Kansas of the
nearly complete specimens of Protostega gigas which permit the present descrip-
tion of the organization of the limbs, the most important of the parts yet unde-
scribed, as well as the least likely to be recovered in complete form. For happily
the elements of the first-secured and completer of these exceptional specimens,
though somewhat crushed, were found altogether, or nearly, in their naturally
articulated position, a condition imperatively necessary to a satisfactory description
of the flippers.

This rare fossil turtle was first briefly mentioned in Science by Professor Osborn9
as " * * * a complete skeleton of Protostega which lay on its dorsal surface with the
fore limbs stretched out at right angles to the median line of the carapace, measur-
ing six feet between the ungual phalanges." Afterwards it was secured for the
Carnegie Museum by Mr. J. B. Hatcher, who, though he crowded the brilliant
work, which might well have crowned the efforts of a long life, into a short one.
leaves this ripe fruition mixed with a sorrow surely not lessened by the fact that
the hardships of the plains of the Northwest and the Patagonian deserts had all too
plainly left their mark upon him.

Having expressed in conversation with Mr. Hatcher much interest in these more
recent discoveries of Protostega, I was invited to make a study of the newly acquired
material, this arrangement being concurred in by Dr. W. J. Holland, Director of the
Carnegie Museum. But early in July, 1904, when I visited the Carnegie Museum for
the purpose of doing this work, to my extreme sorrow I found, that, although 1 had a
brief word from Mr. Hatcher shortly before, he was so seriously ill that there could
be but little or no hope of his recovery. And indeed, as everyone feared, it was
but a few days before he passed away. However, it was under such circumstances
a relief to be busied, and Dr. Holland very kindly arranged for and furthered the
initial study of the material on hand.

Furthermore, during the past summer Dr. Holland has added to the collections
first obtained much additional material, also collected by Mr. Sternberg, including

8 "Notes on the Cretaceous Turtles Toxochelyz and Archelon, with a Classification of the Marine Testudinates,"
Ibid., Vol. XIV., August, 1902.

9N. S., Vol. XIX., No. 470, p. 35, January, 1904.

282

MEMOIRS OF THE CARNEGIE MUSEUM

one very important specimen, a large Protostega, represented by a quite complete
cranium and lower jaw, accompanied by a humerus, radius, ulna, wrist, and palmar
bones — all in place on the same slab. The acquisition of this specimen has been
as timely as fortunate ; for it has removed all doubt as to the carpal and tarsal
organization of Protostega, and made possible the avoidance of errors in the descrip-
tion of the completer sjjecimen, most of the elements of which had been dissociated
from their matrix, and the position of some of them rendered doubtful through the
mistaken zeal of their collector. It will hence greatly simplify our description of
the limb organization of Protostega as based on the free elements of the original and
completer specimen, No. 1420, Carnegie Museum Catalogue of Vertebrate Fossils, if
the more recently acquired and less complete Protostega with its parts in approxi-
mately natural position, be considered first.

Protostega gigas Cope (Cotype).

Specimen No. 14.21 {Carnegie Museum Catalogue of Vertebrate Fossils). — This fine
fossil is from the Niobrara Cretaceous of Hackberry Creek, Gove County, Kansas.
The ex situ portions of the original skeleton, which had weathered out and were
secured in more or less complete condition, include the left humerus, radius, ulna,

Fig. 1. Protostega gigas. Carnegie Museum Specimen, No. 1421, from Niobrara chalk of Hackberry Creek, Gove
County, Kansas. Superior view of skull with the lower jaw and a hyoid, the right anterior border of the carapace and
most of the right fore flipper, all in nearly natural position and still partially imbedded as collected on a single slab of
the chalk matrix. X s-

Skull. — pm.% premaxillary ; in., maxillary ; n.o., external nares ; p.f., prefrontal ; Jr., frontal ; o., orbital border ;
p., parietal ; s.o., snpraoceipital ; j., jugal ; q.j., quadratojugal ; sq., squamosal ; q., right quadrate.

Lower Jaw. — rf., dentalium ; a., angular; or., surangular ; sp., splenial ; (h., hyoid).

Carapace. — A'., portion of right ala of nuchal ; Cr., second rib ; lm., Sin., first and second marginalia.

Flipper. — H., humerus; S., radius; U., ulna; r., radiale(?); i., intermedium; «/., uluare ; p., pisiform; 1-5,
first to fifth carpalia respectively ; I.-V., first to fifth metacarpalia.

WIELAND : THE OSTEOLOGY OF PROTOSTEGA 283

ulnare, radiale, and pisiform, with fragmentary phalanges, and many broken pieces
of the plastron. The in situ portion consists of the right anterior part of the skele-
ton, and was secured on a single slab of the chalk matrix, in which it still remains
intact, as shown in the accompanying drawing, Fig. 1, by Mr. Prentice. It includes
the following : the lower jaw in oblique inferior view, with the two larger hyoidal
elements ; the skull in superior view, less the left squamosal and with the right quad-
rate disarticulated and its anterior surface up ; the right humerus, radius, ulna, wrist,
and palmar elements in superior view, and all in place, except the radiale which
has slipped across to the posterior border of the wrist ; the right ala of the T-shaped
nuchal with the first and second marginals, and the distal half of the second rib, all
in place and resting over the proximal extremity of the humerus.

Although the various bones are more or less crushed, after the manner of most
fossils from the Kansas chalk, the contours are exceedingly good. In the skull, as
simply crushed down in the vertical direction with little distortion of parts, there
appear in clear preservation and united by distinct sutures the premaxillaries,
maxillaries, prefrontals, frontals, postorbitals, and parietals, with the right squa-
mosal, jugal, and quadrato-jugal. The cervicals have in some way been dissociated,
but the skull lies in a normal position with respect to the anterior border of the
carapace, from under which the right fore-flipper projects in the normal position,
the hand being bent back pronately over the position of the third marginal, part of
which appears in view.

It will at once be seen what exceedingly satisfactory information is furnished by
the present specimen, as compared with all other examples of Protostega hitherto
found. The more important measurements afforded are the following :

em.

Lower jaw : Extreme length of ramus 37

Extreme length of median symphysis 16

Cranium : Extreme length (as crushed flat) measured from end of heak to posterior ex-
tremity of occipital crest 58

Median length of narial opening 7.5

Greatest width of narial opening 5.5

Orbital border of the prefrontal 6

Antero-posterior length of orbit 12

Humerus: Length across head 31?

Width of distal end 14

Radius : Length 19

UlDa: Length : 18

Carpalia : Width across the live carpalia in place 18

Metacarpalia : (Length of 1-5 resp.) ; 7( + or — ), 10.5, 12, 12.5, 10 ( + or — ) cm.

The above cranial and hand measurements have not hitherto been obtained in
Protostega, whence it is of interest to note, that, allowance being made for the crush-

284 MEMOIRS OF THE CARNEGIE MUSEUM

ing of the specimens of the Niobrara chalk, no wide differences in proportion are
evident on comparison with the much larger turtles of the genus Archelon from the
Fort Pierre Cretaceous. The lack of coossification of the mandibular rami and
especially the very low radial crest of the latter form will, however, serve to dis-
tinguish it from the fossil before us. But of far greater interest than any question
of generic values that may arise is the fact that the bones of the hand are, with
the sole exception of the radiale, indisputably in place, the most striking fea-
ture being the immense ulnare, and the sti'ong contact of the intermedium with
the first carpale. With these remarks we may pass on, basing our main de-
scriptions on the earlier secured and in some respects better specimen with its several
elements freed from their matrix.

Specimen No. 1420 (Carnegie Museum Catalogue of Vertebrate Fossih). — The
present specimen of Protostega gigas is more nearly complete than any other as
yet discovered. As originally imbedded in its matrix of chalk, nearly every ele-
ment was present in an exactly or approximately natural position, and recovered
as follows :

1. Portions of the skull, with the crushed lower jaw and two large and well
preserved paired hyoids.

2. The 1st, 4th, 5th, 6th, 7th and 8th cervical, the dorsal, sacral, and eight anterior
caudal vertebrae.

3. The nearly complete series of ribs, including the anterior pair of sacral ribs,
also the neurals, and various marginals.

4. The left shoulder-girdle and proximal half of the humerus, with the right
shoulder-girdle and complete front flipper in place, except the ungual phalanx of
the fourth and the second phalanx of the fifth digit.

5. The pelvic girdle and hind limbs complete and in place, less only the fibu-
lare (if ossified), the ungual phalanges of the right, fourth, and fifth digits and of
the left fifth digit.

6. A nearly complete though very much crushed plastron.

Unfortunately the collector of this surprisingly complete fossil, in an attempt to
remove and separate the bones from their matrix of chalk, mis-marked some of them,
and also made it virtually impossible to either replace more than a very few of the
marginals, or to determine the outlines of any of the plastral elements with exact-
ness. Hence it is not worth while, in view of the marked crushing undergone, to
give figures of any of these parts now, although they will aid in the restoration.
As will be evident to any student of the fossil vertebrates the removal of the fossil
from its matrix in the absence of the necessary knowledge, training, and equipment,

WI ELAND : THE OSTEOLOGY OF PROTOSTEGA 285

was ill advised. Such work is difficult enough in the best equipped labora-
tories. However none of the hones of the limbs are broken, and Mr. Sternberg
redeemed himself by discovering and securing in such excellent condition specimen
No. 1421, as just related. The position of the following parts is independently
determinable :

1. The fourth to the eighth cervical, the dorsal and sacral centra, with the
proximal caudal vertebrae and most of the ribs as shown in part in Fig. 2.

2. The shoulder-girdles and elements of the fore-arm ; in particular (as can be
determined by actual close articulation, or by such intervening matrix as the col-
lector permitted in much too small part to remain), the intermedium, first carpale,
metacarpale and its phalanges, and also carpalia 3-5.

3. The pelvic girdle and elements of the hind Hipper, in particular the inter-
medio-centrale ami tarsaiia 1-3, ami ', and 5 (fused), as well as several metatarsals
and phalanges.

It should be stated here that it is claimed that, as figured, no element of either-
Hipper is actually misplaced, although the extent of the crushing undergone makes
it quite possible that in one or two instances dorsal surfaces have been mistaken
for ventral ones, or vice versa, and that the ends of several phalanges may have
been reversed. Withal it is deemed important to be thus explicit in giving the
manner in which the accompanying text-figures and photographs have been
obtained, before taking up more detailed description. (Because of the compression
undergone it was not thought needful to figure either the lower jaw, cervicals, or
caudals. The dorsal view of the carapace can better be given after the restoration
now being made.

I. The Carapace. (Fig. 2.)
With the exception of the T-shaped nuchal and the marginals with dactylate
interior borders so peculiar to the Protostegimc in distinction from all other known
marine turtles, the main features of the carapace are very well represented in Fig.
2. The very thin, almost paper-like series of neuralia is present, though much
crushed down on the underlying neural arches. The ribs are free through at least
the distal two-thirds of their length. The first pair is comparatively slender, it not
being clear as to whether, or not. their distal extremities rested on the lateral ala>-
like expansions of the nuchal. The shield contour is, instead of elongate as once
supposed, relatively broader than in either Thalassochelys or Ghelone, the ratio of the
length of the dorso-sacral series to the breadth of the carapace being as three to four.

286

MEMOIRS OF THE CARNEGIE MUSEUM

Fro. 2. Prutoslegit gigas. Carapace less the nuchal and the marginal series. Inner view. 1-10, the first to tenth
ribs respectively ; s, the first sacral. Ten dorsal, the two sacral, and the first of the caudal centra are shown. (A trial
drawing preparatory to mounting of specimen. About one-eighth natural size.)

This greater breadth is quite characteristic of Cretaceous turtles, and a distinct ap-
proach to a quite orbicular form like that of Li/toloma.w It is a form, moreover, that
well accords with the powerful and widely expanded flippers now to be described.

II. The Front Flipper.

(Plates XXXI. and XXXII., with Figs. 3 and 4.)

The shoulder-girdle of Protostega is robust but presents no strongly marked

peculiarities, save the elongate coracoid which extended back to the pubis as

in Eretmoehelys, but not either Chelone or the long-bodied Dermochelys. The

10 Wieland, "Structure of the Upper Cretaceous Turtles of New Jersey : Lytoloma," Amer. Jour. Sci., Vol. XVII.,'
September, 1904.

WIELAND : THE OSTEOLOGY OF PKOTOSTEGA

287

features of the humerus are, however, highly characteristic, as is shown more
particularly in Figs. 3 and 4, which disclose certain approximations to the
humerus of Dermochelys not so readily discernible in the figures hitherto pub-

Fig. 3. Piolnslega gigas. Ental view of a large humerus. Actual length, 34 cm. <t, head ; h, radial crest ; c, ulnar
condyle ; d, ectepicondyle ; e, ectepicondylar foramen ; /, ectocondyle ; g, entocoudyle.

lished. It must be noted, however, that crushing has been such as to greatly dimin-
ish the distal breadth and render it uncertain as to whether there is an ectepicon-
dylar groove. Case speaks of a foramen in describing his specimen,11 though his
figure suggests a groove. I suspect that there may be present a true enclosed fora-
men rather than a deep groove ;is in Archelon. In the latter the groove is much
further back from the anterior border and the ectepicondylar process correspond-
ingly larger, while the radial crest is not nearly so prominent as in Protostega. (See
Fig. 3.)

11 See foot-note 5.

288

MEMOIRS OF THE CARNEGIE MUSEUM

The carpal and finger organization of Protostega has been hitherto wholly un-
known, and, in fact, the only fossil marine turtle from America in which these
parts have been described is Toxochelys.12 In discussing the carpal organization it is
of foremost importance to recall that the intermedium,, carpale 1, and metacarpale 1,
as well as carpalia 3-5, are still naturally articulated as in life, and in all particulars
agree precisely with these same elements in specimen No. 1421. Nor is there any

Fig. 4. Protostega gigas. Right shoulder girdle and flipper. X I- Niobrara Cretaceous. S, scapular; PC, pro-
coraco-scapular ; C, coracoid ; H, humerus ; R, radius ; U, ulna ; r, position of the radiale ; ;', intermedium ; «, ulnare ;
e, centrale ; 1-5 first to fifth carpalia ; p, pisiform ; m, metacarpal 1 ; p. phalanx ] ; I.-V., first to fifth fingers and un-
gual phalanges. The view is dorsal with the procoracoscapular rotated into the plane of the paper.

further doubt concerning the identification of the other elements and agreement in
toto with that specimen. As so clearly shown in the figures and photographs, the
triangular-shaped centrale articulates strongly with carpale 1. I have supposed
this was not the condition in Archelon,1* but must have been in error. No radiale
was recovered. Bearing in mind that the several carpal elements are somewhat
crushed, no further detailed description of them appears necessary, excejDt that it
should be remarked that the ulnare is of sub-hexagonal outline and relatively very
large, being of the general form seen in Thalassochelys, and especially Colpochelys,
rather than Dermochelys. The ulnare of Archelon is of nearly the same relative size,

12 Wieland, " Notes on the Cretaceous Turtles Toxochelys and Archelon, with a classification of the Marine Testudi-
nata," Am. Jour. Set., Vol. XIV., August, 1902.

13 Loc. cit., Wieland, " Notes on Toxochelys and Archelon," etc.

WIELAND : THE OSTEOLOGY OF PROTOSTEGA 289

but of rounder and more regular contour. The nearly ovate pisiform is distinctly
intermediate in development between that of the existing Dermochelys and the
Cretaceous congener Toxochelys in which the pisiform is smaller than in any other
distinctly marine turtle.

The first metacarpal is broad, and the first finger short and robust as in Toxochelys
(see Fig. 8) and other members of the Cheloniidse. Finger disparity is pronounced,
the second finger being little elongate as in Toxochelys, with the third and fourth
fingers of medium and nearly equal elongation, and the fifth fully as elongate as the
second. The first to third fingers bore free claws, but not the fourth and fifth.

In its general features the front flipper of Protostega agrees much more closely with
that of Toxochelys (Fig. 8) than with that of other forms, as one might well expect.
The existing Cheloninse do not present so close a likeness, because of the peculiar
elongation of the radius and dependent carpal variations, although the boundaries
are much the same, the centrale in particular being in contact with carpale 1 in both
cases. With Dermochelys, in which carpale 1 is small and excluded from contact
with the centrale, the points of likeness are more obscure, although there is no dis-
tinct suggestion in the carpal organization, that the former belongs to an utterly dif-
ferent race. Pisiform development is also more like that of the primitive forms
than in Dermochelys.

The relative size of the front flippers as compared with the carapace is great,
since they are not only robust, but have a spread equal to about 3^ times the length
of the dorso-sacral series of vertebras. In the existing carnivore Thalassochelys. this
ratio is nearly as great, being equal to about three, but falls to two to one in the
algaphagous Chelone. In the carnivore Dermochelys the ratio is about 2^ to 1, a
result of comparison rather unexpected in view of the very great length of fingers in
the latter, and accounted for by the great length of the clipper-built Dermochelan
body. There is in the comparison just made the very strongest suggestion that Pro-
tostega, more distinctly than any marine turtle thus far known, hunted prey, which
swam actively, and, bearing in mind other features, was perhaps even powerful.

In my description of the front flippers of Toxochelys14 I presented an interesting
tabular comparison of the relative percentages of length of the several elements of
the front flipper and humerus of various Testudinates, which exhibited the general
trend of change since the Jurassic in the development and variation of flippers from
a more generalized limb-type. I now include Protostega in this comparison. The
humerus is in each case considered as having a length of 100, and the finger lengths
as reduced to the same ratio, but including arbitrarily tor convenience the meta-
carpals, as follows :

"See Footnote 8.

290

MEMOIRS OF THE CARNEGIE MUSEUM

Dermochelys 100

Erelmochelys 100

Oolpoehelya 100

Protostega 1 00

Archelon 100

Toxochelys 100

Chelydra , 100

Acichelys = (Euryxternum) ' 100

43

53
57
60
54
58
52
57

Is

44
51
50
51
50
53
51

127
49
56
50

51
50

40

180
89
92

73
72
54

_•

•

&£•

be fi-

aj d,

£ a

fe §

•Sal
CO

«S

fc. B

W B

*

209

173

86

128

105

44

131

107

44

128

110

80

100

104

70

73

55

50

63

66

51

23
12
13

17

11

small

17

Inspection of the above table shows :

1. Strongly marked radial and ulnar decrease in length.

2. Marked tendency to radial elongation as compared with the ulna, with con-
siderable variation in the length of the radius and ulna as compared with the
humerus.

3. Nearly static length of tbe first finger in tbe Cheloniidse, with a sharp increase
in D&rmckhelys.

4. That elongation of the second finger appeared slowly.

5. Early and persistent increase in the length of fingers three and four.

6. More or less variable tendency to elongation of tbe fifth finger, with a sharp
increase in Dermochelys, and a suggestion that this finger may have first elongated
in some forms, and then undergone decrease in length.

7. Great and persistent pisiform increase, which began relatively early.

8. Increased finger disparity is mainly coordinated with depression of the radial

crest.

III. The Hind Flipper. (Fig. 5.)

The presence of an entire pelvis and the complete hind flippers, save only the
ungual phalanges of the right second digit, and of both fifth digits is quite all that
the most exacting anatomist could wish. Although as the result of compression in
the matrix the bones are of slightly broader outline than in life, their proportions
are, as in the front flippers, fairly well retained. Likewise, as is so often and so
fortunately the fact in the fossils from the Kansas chalk, most of the articular
facets and surface characters are, as in the other parts of this excellent specimen,
clearly indicated.

Regarding the pelvis it is only necessary to note that the obturator foramen
must have been completely enclosed by the close contact of the ischium and ento-
pubis on the median line as in Archelon.™ The figures given herewith would not of

15Wieland, lac. cit., "The Skull and Pelvis, etc., of Archelon," p. 247.

WIKLAND : THK OSTEOLOOY OF I'KOTOSTEGA

291

themselves be quite clear upon this point were it left unnoticed. The ilia are of
course shown in the lateral view; this, as the result of compression, being the only
view remaining in fairly exact proportion.

Similarly the left fibula, as shown in Plate II., is turned partly around with a
portion of the plastron adhering, the facet for contact with the distal end of the
tibia being thus directed upwards. There can be but the barest doubt that the posi-
tion of all the elements of the hind flippers as assigned in the accompanying Fig.
5 is correct. The presence of paired parts afforded a means of so exactly checking

Fig. 5. Protoslega yigas. Right pelvic girdle and flipper. X'- The dorsal or ectal view of flipper with the corre-
sponding ental view of the pelvic elements show in the plane of the paper. P, pubis ; F, fibula ; /, tihiale intermedium
and centrale fused, or caleano astragalar element ; 1-3, first to third tarsalia ; 4, fourth and fifth tarsalia fused ; m, meta-
tarsal 1 ; ;), phalanx I ; I -V., first to fifth toes and ungual phalanges.

the determinations as to exclude all possibility of error. Moreover, nearly all the
parts, save perhaps the two ulnares, were imbedded in their matrix in a closely
articulated natural position. And it should bt noted that tin tibiale, all tin tarsalia,
and the first metatarsal and some of the phalanges of both flippers remain so articulated,
the evidence as to their position hence being incontestable. Also metatarsal V.,
though not articulated, or otherwise joined by remaining matrix, is of such char-
acteristic form as would testify amply to its position, even though isolated. The
same may be said of metatarsal 1. The development is in both cases much the
same as in all the marine turtles. Unfortunately, however, no element referable to
a fibulare, which appears to have been present, is now determinable. Nor is a

292

MEMOIRS OF THE CARNEGIE MUSEUM

fibulare represented in the drawing, it being difficult to determine what may have
been its outline.

The first toe is short, the second, third, and fourth of nearly similar moderate
elongation, and the fifth somewhat elongate, as set on the strongly bowed and en-
larged metatarsal V., which is similar to that common to all sea turtles.

The first toe bore a heavy and free claw, and I believe the claws of toes II.— IV.
were also free and clawed, as indicated by their form and the curvature of the an-
tero-lateral or ungual ridges. Reference may be made to Fig. 6, showing the
lateral view of all the ungual phalanges which were all dissociated.

Allusion has been made to Professor Williston's restoration of the hind limb of

Prolostega,16 and, before pass-
ing on to a general comparison
with other forms, it is neces-
sary to deal with this restora-
tion more specifically. Willis-
ton's figure is that of a form
so different from Protostega and
all other turtles, that, if cor-
rect, it would indicate the ex-
istence of a new and hitherto
unknown genus, or even fami-
ly, of marine turtles. But this
cannot be ; for the form and
proportions of the individual
elements is throughout eesen-

Fio. 6. I'rofoslega gigas. Ungual phalanges. /, fingers ; h, toes.
Lateral view. X f. (Trial series not found to necessarily be the exact tially the Same as ill Protostega.
finger and toe series.) The g^ ^ fl^ the arrange_

ment of elements given by Professor Williston affords no approximation, and no
direct hint as to the tarsal organization in Protostega. Metatarsal V. is placed in the
position of metatarsal I., and the bone supposed to be metatarsal V. is probably from
another animal, or else, the metatarsal of the other flipper is much altered by the
compression to which nearly all the turtles from the Kansas chalk have been sub-
jected. The other tarsals are difficult to adjust, although it is in particular likely
that the element placed in the fibular position is a pisiform ; for in both form and
proportions it agrees with the pisiform of both of the present specimens, and differs
very markedly in these respects from the true tarsal elements. The phalangeal

16 See foot-note 7.

V \

\VI ELAND : THE OSTEOLOGY OF PROTOSTEGA

293

formula of 3, 3, 3, 3, 0, as proposed by Williston, is not that of Protostega, in which
the true formula is 2, 3, 3, 3, 3.

The strongly marked angular tubercle of the proximal end of the fibula, as
figured by Williston, reminds one of the similar and similarly situated process on the
femur of the Dinosaur, Camptosaurus. It is not present in any of the specimens of
Protostega known to me, and must at least indicate a new specific form. Should
further examination confirm the presence of this feature, the specimen in the collec-
tion of the University of Kansas, which shows it, should be known as the type of a
new species. And surely its most appropriate name would be one honoring the dis-
tinguished paleontologist who first studied it, and who rendered a very distinct ser-

Fig. 7. Toxochelys latiremis. Left frout flipper. Dorsal view. «, head ; b, radial condyle ; c. ulnar condyle ; e,
ectepicondylar groove; g, entocondyle ; R, radius; U, ulna; In, intermedium; Ul, ulnare ; C, centrale ; 1-5, first
to fifth carpalia ; P, pisiform ; I.-V., first to fifth metacarpals and fingers.

vice by figuring it as best he could in the absence of further material or means of
checking results.

Structurally, aside from the more primitive clawed condition, there is no hiatus
between the tarsal and other features of the hind flipper of Protostega and the exist-
ing sea-turtles, the comparison with Dermochelys being quite as close as any other, in
some respects closer in fact than with a somewhat aberrant although primitive form
like that of Colpochelys (Fig. 8). In Dermochelys and Protostega the relative size and
development of all the tarsal elements is suggestively alike. It is indeed increasingly
difficult to believe that the former genus belongs to an utterly remote and primi-
tively separated line. But unfortunately there is not yet known amongst the fossil
marine turtles a single hind flipper which is sufficient!)' well preserved to afford

294

MEMOIRS OF THE CARNEGIE MUSEUM

further detailed comparison with Protostega. The writer has, however, pointed out
the fact that the littoral, or semi-marine, turtles of the Cretaceous of New Jersey,
especially Osteopygis and Lytoloma, present many interesting primitive features,
amongst these being relatively longer hind limbs, the inference being a quite clear
one, that in the evolution of the limbs of the marine turtles, in correlation with cara-
pacial and other changes, the hind limbs underwent shortening, and the front limbs
a compensatory elongation. In fact in Osteopygis, a form now known to be very
closely related to the ancestors of the Chelonime, the femur is distinctly longer than

Fig. 8. Colpochelgs Kempii Garmau. ■ J Fore and hind flipper of a specimen in the United States National
Museum (ratio of fore to hind flipper, 1.25).

Fore Flipper. H, humerus with ectepicondylar perforation set well in from the anterior border : ft, radius in
natural position revolved somewhat beneath the ulna ( U) ; r, radiale, bounded by the radius, the intermedium and first
carpale ; i, intermedium, u, ulnare ; p, pisiform ; e, centrale bounded by the intermedium, ulnare, and first to fourth
carpalia ; I., V., first and fifth fingers.

Hind Flipper. 7\ tibia, F, fibula ; I., V., first and fifth toes. The tibia and fibula support a single heel or astrag-
alo-calcaueal element, («) theoretically formed by the fusion of four elements corresponding to the radiale, intermedium,
ulnare, and centrale. The first toe is borne on tarsale I., the second on tarsale II., and the third to fifth on tarsalia III.—
V., fused. Metatarsal 1 (m) is the exact equivalent of metacarpal 1, and metatarsal V. has a pisiformoid development.

the humerus, as is also the case in Chelydra, a tortoise which is in many respects sug-
gestive of specialization in some direction similar to that doubtless followed by the
direct ancestors of the marine turtles. In Lytoloma, an upper Cretaceous and lower
Eocene genus closely related to Chelone, the femur appears to have undergone some

WIELAND : THE OSTEOLOGY OF PROTOSTEGA 295

relative shortening. But though a nearly complete carapace has recently been de-
scribed,17 in the absence of further description of the Belgian material, now in process
of elaboration by M. Dollo, we know very little of the limb development in this
form, except by inference. In the case of Toxochelys, which is near to Lytoloma, we
fortunately know the organization of the front flipper within close limits (Vf. Fig.
7), while I have but recently described the carapace of a most interesting new
species, T. BauH.w In Toxochelys we find a markedly primitive form of flipper, in-
deed the most primitive known, although the humeral contour and finger disparity
indicate an animal capable of navigating the open seas. Also, the carapace of all
the Toxochelyds known has very large fontanelles, suggesting a marine life quite as
decidedly as the flipper; for though the laws of carapacial reduction are as yet only
surmised, it is only in the marine members of tbe Cryptodira that prominent
pi euro-marginal fontanelles occur.

Because of the facts given it does, however, prove decidedly interesting to find
that even in Protostega of the Upper Cretaceous, the bind limbs yet remain rela-
tively long, though fully developed as flippers. For in Eretmochelys and Ghelone as
well as in Dermochelys, tbe fore flippers vary from about 1.55 to 1.60 times the length
of the hind flippers; whilst in the orbicular bodied Protostega, with a tremendous
spread of front flippers as already pointed out, this ratio falls to 1.30, which is, bow-
ever, a little in excess of Colpochelys. In the latter genus which has in some respects
the most primitive flipper organization of any existing Cryptodiran the fore flippers
are but 1.25 times the length of the hind flippers. Doubtless there are many
further interesting numerical relationships not yet discovered. Though the rela-
tively strong pattern and great size of both fore and hind flippers in Protdstega and
the related Archelon, taken with the broad body and shortness of the dorso-sacral
series, may not indicate extreme and sustained swiftness, in conjunction with the
cranial features these proportions do go far to confirm the opinion already expressed
that these turtles, so powerfully equipped for both swimming and attack, may well
have hunted actively swimming prey. The Protosteginse plainly included the larg-
est and the most rapacious turtles which ever existed.

"Wieland, "Structure of the Upper Cretaceous Turtles of New Jersey ; Lytoloma," Am. Jour. Set. September,
1904.

18 Wieland, "A New Niobrara Toxochelys," Am. Jottr. Sc.i., November, 1905.

296 MEMOIRS OF THE CARNEGIE MUSEUM

IV. Systematic Position of Protostega.

CHELONIODEA (Baur).

(Superfamily of the Cryptodira.)

A parieto-squamosal arch ; palatine foramen and free nasals sometimes present,
(Desmatochelydinse) ; fourth cervical biconvex, with the centra of the sixth to
eighth much more modified in recent than in most Cretaceous forms.

[A) Dermochelydid.e.
No descending parietal processes ; no palatine foramen ; other cranial and limb
characters not remote from those of the Cbeloniidse ; carapace represented by the
nuchal only, and body enveloped in a leathery hide with an osteodermal mosaic ;
no claws. Genera: Dermochelys, Psephophorus, Eosphargis.

(B) Cheloniid.e.
Skull with descending processes of parietals, so far as known ; palatine foramen
sometimes present; vomero-premaxillar union usual but not constant; a normal,
though often much reduced carapace and plastron ; nuchal with or without articular
process on under side ; claws one, two, or more.

1. Protosteginx (Wieland) : No free nasals, and no palatine foramina; obturator
foramen small and enclosed by ischio-pubic contact on the median line, as in many
land forms ; nuchal T-shaped ; neuralia and pleuralia thin and investing the ribs
but slightly ; marginalia usually spiniferous on interior borders ; plastral elements
of medium development with numerous digitations on both outer and inner border
of the hyo- and hypoplastron ; epiplastron of Trionychoid and Acichelysoid form ;
body enveloped in a leathery hide (?) ; claws three or more.

(a) Protostega Cope. — Cranial elements nearly as in Arckelon ; mandibular rami
eoossified ; radial process of humerus strong; front flippers very broad and of a
more distinctly marine type than those of the Toxochelydinse (Toxochehjs); centrale
in contact with carpale 1 ; hind flippers relatively very large, and tarsal region most
like that of Dermochelys. Species : P. gigas Cope from the Niobrara Cretaceous of
western Kansas :

Ratio of spread of front flippers to length of the dorso-sacral series 3.2

Ratio of front to hind flipper 1.3

First to third fingers free, clawed.

First to fourth toes free, clawed.

WIELAND : THE OSTEOLOGY OF RROTOSTEGA 'I'M

V. Measurements of Protostega qigas.
(All based on the single individual illustrated in Text-figures 2-7 and Plates I.
and II. As most of the elements are much flattened lengths only are given, these
doubtless remaining much the same as in life.)

( A ) The Lower Jaw.

Length on median line (estimated, the tip being broken away) 26 cm.

Length along ramus 30 "

(11) The Carapace, Ribs and Durso-saeral Cenlia.

Total length of the ten dorsal centra.. 68 "

Total length of the three succeeding centra 10 "

Length of the dorso-eacral series 4.5 "

Extreme width of the carapace (exclusive of marginals) as crushed out flat and meas-
u led from tip to tip of the fourth ribs, both of which are complete and are the

longest of the rib series 106 "

Length of the first to tenth dorsal centra respectively, 5.5, 7.5, 8, 8, 9, 8, 7.5, 6.5, 4, 4 "

Lengths of the first and second sacral and first caudal centra respectively, 3, 3.5, 3.5 "

Lengths of the first rib 10+ "

" " second rib 49 "

" " fourth rib.. .. 51 "

" fifthrib 51

" " sixth rib 46 (?) "

" " seventh rib 43 (?) '

Length of liist sacral rib (in place) 3.2

( P) The Sliovldl r Girdle.

Distance apart of the two distal extremities (the scapular and the precoracoidal) of

the procoraco scapular 30 "

Length from interior bow of procoraco-scapular to the glenoid cavity 13 "

Length of the coracoid 40 "

, D) lite Pelvic Girdle.

Length of pubis (outer exterior border, on straight line) 18 "

Length of ischia 8.5 "

Length of ilium 18 "

Length of pelvis on median line 43.5 "

Width across pubis at broadest point 48 "

(E) The Limbs.

The greatest spread of the front flippers (as based on measurement from the approxi-
mate median line of the shoulder girdles to tip of longest finger) 250 "

Greatest spread of hind flippers 190 "

[V The Front Flipper.

Length from glenoid cavity to tip of linger III 106

Length of h uruerus 34

Length of radius 20

Length of ulna 17

Length of intermedium 7.5

Width of intermedium 4.5

Length of carpale 1 • 6

Length laterally across carpalia 3-5, which remains as recovered naturally articulated. 10 6

298 MEMOIRS OF THE CARNEGIE MUSEUM

Length of ulnare 10

Greatest width of ulnare 7

Length of pisiform 6.1

Length, inclusive of metacarpals, of

First finger 18

Second ringer 29.5

Third fi nger 43

Fourth finger 39

Fifth finger 30

(2) The Hind Flipper.

Length from acetabulum (or head of femur) to tip of linger III 85

Length of femur 27

" " tibia 20

" fibula 20 5

" " ti bial e 7

Greatest length of the first to fourth tarsalia as they yet remain naturally articulated

(on line across ankle) 11

Length, inclusive of metatarsals, of

First toe 18

Second toe 27

Third toe 30.5

Fourth toe 31

Fifth toe 20

EXPLANATION OF PLATE XXXI.

Protostega gigas. (Carnegie Museum. Cat. Vert. Fossils, No. 1420.)

(a) Right and left shoulder girdles, with right front flipper and the proximal half of the left
humerus.

(b) Pelvic girdle with both hind flippers.
(/i) The accompanying hyoids.

The elements are shown as photographed in superior view in approximately the position in
which they were imbedded and crushed out flat in the matrix of chalk. All the elements belong
to a single individual from the Niobrara Cretaceous of western Kansas.

300

X
X
X

EXPLANATION OF PLATE XXXII.

Protostega gigas. (Carnegie Museum. Cat. Vert. Fossils, No. 1420.)
Enlarged views of forearm and hand (upper photograph), and the foot (lower photograph).
Observe in the upper photograph that the intermedium, carpale 1, and metacarpale 1, as well
as carpalia 2-4 are still articulated as in the matrix ; and in the lower photograph that the tibiale
with tarsalia 1—5 and metatarsal 1 are likewise articulated as found in their normal position in the
matrix.

302

X
X
X

EXPLANATION OF PLATE XXXIII.

Dermoeh elys coriacece.

(d) Shoulder girdle with front flipper articulated, and the procoracoscapular thus shown
foreshortened.

(b) Hind flipper. Superior view.

I., V., the first and fifth fingers and toes respectively.

304

X

X