[ 1019 ] j Researches on the Structure , Organization , and Classification of the Fossil Reptilia.\#151 ; Part IX .
, Section 2 .
The reputed Mammals from the Karroo formation of Cape Colony .
By H. G. Seeley , F.R.S. Received January 4 , \#151 ; Read March 1 , 1894 .
[ Plate 89 .
] The reputed Mammals from the Karroo .
1 .
On Theriodesmus phylarclius .
887 , 1 described some associated bones of an animal from Klip font ein , Fraser berg , ir the name Theriodesmus phylarchus ( 'Phil .
Trans./ 1888 , B. , 24 ) .
The fossil since received some attention from anatomists , but I am not aware that the place nally assigned to it among the mammalia has been questioned .
Professor Carl deleben ( 'Proc .
Zool. Soc./ 1889 , p. 259 ) , who critically examined some points y interpretation , concurred that it cannot be said to belong to any known order of mals ; and affirms that it is not a reptile in the modern sense of the word .
He is that its characters are such that it might be placed between reptiles and imals , but does not offer evidence or argument in support of that determination , is not a substantial difference from my own conclusion , which regarded Themo-iws as illustrating the reptilian inheritance in the mammalian hand .
These lusions may be re-examined in the light of new evidence which I obtained in Africa .
Before stating that evidence I propose to examine the interpretation ie carpus and phalanges which Professor Bardeleben offers in place of that given *y paper , published in the 4 Philosophical Transactions .
' rofessor Bardeleben places four bones in the proximal row of the carpus instead he three bones shown in my restoration .
The result of this is to make the carpus er than the distal ends of the ulna and radius , so that in his restoration the radius 7 covers a small portion of the smooth proximal articular surface of the bone named me scapho-lunar , and by him named scaphoid .
That proximal surface shows no lence of division into two parts , and , therefore , the hypothetical position assigned t is not supported by evidence .
The proposal to name the bone scaphoid is based * m the hypothesis that I have mistaken the lunar bone for the central bone ; and discovery of this supposed error is the ground for introducing four bones into the 13.2.95 .
proximal row of the carpus .
No argument is offered in support of that interpretatior and no South African fossil is known with the undisturbed carpus composed in tba way .
In the fossil the carpus is manifestly somewhat displaced , though there is n question that the three bones which I have regarded as proximal are substantially i their natural positions ; and that the bones of the distal row of the carpus are also i their natural positions , in contact with the metacarpal bones , dhe transverse widt over these four distal bones is one-tenth of an inch less than the transverse widi over the three bones which I placed in the proximal row .
The difference of interpret tion is therefore with the bones which I regarded as central bones .
These bones a entirely distal in position to the proximal carpal bone , which is in contact with t !
distal end of the radius ; so that , as preserved , the bone which seems to me to be i principal central bone is between the magnum and the outer side of the bone whh articulates with the radius j and that central bone which Professoi Bardeleben idem fies as the lunar bone , is in close lateral contact on its inner side with the bones whit he regards as central .
There is , I submit , no evidence for the suggested transposith of the bone from the middle row of the carpus to the proximal row ; and no arguinet is offered in support of the change of interpretation .
The distal end of the rads gives no indication of having supported two bones ; while on the outer distal surfe of the scapho-lunar is a distinct facet which seems to correspond in size and posit i with the central bone which is below it .
\#166 ; .
It is , I think , unnecessary to discuss Professor Bardeleben 's views further upi this character of the carpus , because I have obtained new evidence of the struct* of the carpus in allied animals ( p. 36 ; Plate 2 , fig. 17 ) , with the bones united by mate , in other genera , demonstrating three bones in the proximal row .
^ And this seems* me confirmatory of the interpretation which I originally offered , since I am award no fact tending to a different conclusion .
Professor Bardeleben , however , proposed to obtain an additiona one in \#187 ; central row of the carpus by dividing into two the bone which I have letters ( 'Phil Trans. , ' B. , 1888 , Plate 26 , fig. 1 ) , as is shown in his drawing ( Bros. A. Soc. , ' 1889 , plate 30 , fig. 4 , ce1 , For.this division I believe there is no justifies to either in the specimen , or in the osteology of allied animals , and I have no evidi of it in the example of the carpus which I collected at Lady Frere ( Plate l , fag .
1 .
The bone which I have regarded as the innermost central bone in Babdeleben 's nomenclature is the pre-pollex .
I have no new evidence on \#166 ; = interpretation , and see no reason to vary my reading ; wwVlescribed \#171 ; question to those who examine the original specimen Whom 1 " ^nbe fossil , the possibility of there being a digit interior in position to the fi st oc ur ^ me but the evidence was too slender to justify me in doing moie th ' Tm\#151 ; r - \#171 ; 2 millim in diameter !
'andTu'b-qufdrate in form .
It may be connected dista#p AND CLASSIFICATION OF THE FOSSIL REPTILIA .
1021 nute ossification external to the trapezium .
" Professor Bardeleben does not 3 that passage , though he suggests names for the hypothetical bones of the digit described .
The small bone which is drawn in my figure , external to the trape-may be homologous with that named pre-pollex by him in *'timus .
But it has not been shown that this identification would bring the lo-lunar bone into contact with the trapezium .
\#166 ; ofessor Bardeleben has suggested the existence of epiphyses in the phalanges te third and fourth digits of Theriodesmus .
During the removal of the bone e from the stone I became convinced that the appearance of epiphyses was \gt ; ry .
It rests entirely upon the shortness of certain phalanges .
A.nd the original pretation was confirmed when I obtained a part of the hand of a larger animal , d near the Oude Kloof , in which the articular surfaces of these short phalanges veil preserved .
i renewed examination of the remains of I fail to find the same onderance of mammalian affinity in the type as formerly .
It is not that there is change in the value of interpretations of character in the several bones , for the imalian approximations remain what they were .
The whole basis of interpreta has been changed by the discovery of the limb-bones of Pareiasaurus .
And the lion which I now make shows how closely the reptile Theriodesmus may resemble immal , instead of demonstrating reptilian characters in a Mammal as had been .red .
be humerus of Theriodesmus differs in type from every known South African ile , but is too imperfect to admit of detailed examination of either the proximal iistal end .
Its distal end showed a form not unlike that seen in the lion and 9 marsupials , both in the rounded trochlear reflection of the condyle downward , in the slight development of the distal crest .
These characters make some L'oximation to those seen in Pareiasaurus , which , if not very close , is inteiesting , e one of the animals compared has slender limbs , and the other has limbs of smarkably heavy type .
The approximation leads me to believe that the imperfect osure of the proximal end of the humerus of Theriodesmus indicates part of idial crest which I have not seen developed in the same way at the proximal end the bone in mammals , in relation to its articular head .
It also seems evident that articular head of the bone has not the characteristic curve which is associated in higher mammals with a more or less hemispherical head .
There may be some \gt ; roach to an ovate form of the head in certain marsupials , but it is only among the .notremata that this region has a decided horizontal or transverse compression .
The ference between the mammalian form of the proximal end , which might be expected m the distal end of the bone in Theriodesmus , and the form actually preserved , is 9wn by shading in a figure already published ( 'Phil .
Trans. , 1888 , vol. 1/ 9 , p. 143 ) .
The two ends are not in harmony with known mammalia .
They seem to me to be better compared with the South African reptilia , because the characters in which the humerus referred to Cyuodr for instance , differs from that of a marsupial Mammal , are most marked in the transverse width of the shallovy proximal articulation , and in the proximal development of the radial crest .
There being apparently this resemblance to Theriodonts , all of which are distinguished from Mammals by the shape of the articular head of the humerus ( in the nev examples found by myself ) , it seems that the humerus of ei may h regarded as that of a Theriodont , exposed on its external aspect only , and not a that of a Mammal .
There is no ground to affirm that the bone is that of a Mamma because it cannot be paralleled among mammalia .
And although the Theriodon limbs are at present imperfectly known , there does not appear to be in the humern of Theriodesmus ( fig. 1 ) any point of structure which may not be paralleled amon Theriodonts , and their near allies , and especially in Pareiasaurus ( fig. 2 ) .
The ulna and radius are much more mammal-like in form , but when those bom were described very little was known of the limb bones of the South African After studying the ulna and radius of ( fig. 2 ) , at Tamboer , it seemt manifest that there is practically no difference of plan between those bones and tl corresponding bones of Theriodesmus ( fig. 1 ) .
The difference is essentially one f proportion ; part for part and curve for curve they are identical , but the heavy bom of the one are drawn out to a slender type in the other .
The most remarkable character of the ulna in is its prol\lt ; Myti\lt ; ^ proximally into an olecranon process , which helped to define the articulation for t humerus , and extended beyond it .
This process is compressed from side to Bid the proximal surface of the bone is concave in length laterally .
Its posterior contour-convex as it extends over the olecranon ; but the contour becomes concave externa in the lower half of the shaft .
These characters are identical in Pareiasaurus .
both genera the positions and forms of both proximal and distal articulations \#187 ; \#166 ; almost identical .
In the radius the contours of the bones , which have concave si# and truncated ends , are similar ; and in both the articular ends gradually contract , the middle length of the bone , without any approximation to the comparative pa e ft. of the sides of the bone seen in Mammals , so that the sudden expansion o -articular extremities of the radius , which is commonly found in Mammals , is wanting .
Therefore , notwithstanding the heavier build of , I doubt that Theriodesmus makes a nearer approach to that type m the radius than it does to the corresponding bones of Mammals which approxima ^ to it in form and relative stoutness of the bones .
SHI There are no new facts which illustrate the carpus .
Pareiasaurus proves proximal carpal beneath the radius , which I have regarded as formed y t e \#171 ; P and lunar bones , as in Theriodesmus , I have already indicated ways m diverges from Mammals towards some Heptilian types .
I have not met AND CLASSIFICATION OF THE FOSSIL REPTIL1A .
r xample of a carpus in which the central bones are similar ; but the blending ie3capho-lunar appears to be a character found in the South African Therosuchian ils which are known .
It is not possible to compare the carpus closely with that of Fig. 1 .
Humerus , radius , and ulna of the right fore limb of Theriodesmus phylarchus .
Fig. 2 .
Humerus , radius , and ulna of the right fore limb of Pareiasaurus Baini .
\#166 ; ' asaurus , because the carpus in that type was not extracted in an entire state 11 the rock .
The presence of a second central bone in Theriodesmus is a nearer i^ach to Halteria than to any Mammal ; while the possible presence of a third uhl bone would be a resemblance of an amphibian type .
^cccxciv .
\#151 ; b. 6 P It has already been observed that the metacarpus resembles that of Protorosaub ( ' Phil. Trans./ B. , 24 ) and Scapbosaurians , as well as carnivorous Mammals lfeeL Otter and Thylacynus .
The evidence no longer appears to me to warrant the view that the extremity the fourth digit of Theriodesmus was lost and replaced by frbro-cartilage .
In \#166 ; impression of the digit preserved , I believe the fourth phalange to be indicated ^ displaced laterally , so that I mistook it for indications of ligaments .
The ternn^j claw appears to be transverse , beyond the third and fourth phalanges .
Hence ^ number of phalangeal bones in the digits may have been I. , 2 , II , , 3 III , | .
IV .
, 5 ; V. , 3 .
If the pre-pollex really exists as a digit , the evidence is not inclusive that it includes more than a metacarpal rudiment .
I have no doubt thathe hand is more Reptilian than Mammalian , though the phalangeal bones approximate those of Monotremes in form .
Pareiasaurus shows no resemblance in having simiHy short phalanges between the longer hones .
Its digital formula appears to |K 1.3 * II .
, 4 ; III .
, 4 ; IV .
, 3 ; V. , 3 .
On this evidence I conclude that Theriodesmus is related to Pareiasaurus in \ is which were unknown when the genus was described .
And the resemblances see to me to be such that there is now ground for placing those types in subdivisions otts same group of animals .
It having been proved that Pareiasaurus is essenUl ; reptilian in the characters which are of value in classification , it follows that Tho-desmus cannot be a mammal .
The mammalian resemblances , however , which we been demonstrated in its bones show that the border limit between Reptilesmd Mammals approximates so closely in osteological character that the distinction those groups of animals has not hitherto been easily made on the evidence of them-limb .
Practically that distinction may he said to have disappeared , for the charars which divide the two groups as now stated are of small value in classification .
Fig. 3 .
Carpus of EerpetoeUrus , showing the bones which articulate with the radius and ulna , with part of the metacarpus and phalangeal bones of the digits , in a specimen from Klipfontein .
:11b There is every reason to believe that belongs to that group of ai *b 'PP\#187 ; ii .
10 ~1\#187 ; in i\#171 ; *""\#187 ; \#187 ; .
\amp ; .rlL dontia ; but os the dentition ie unknown there re no mesns of defining J AND CLASSIFICATION OF THE FOSSIL REPTILIA .
n an within that group , or of establishing that conclusion , by other than inductive dnce of Pareiasaurian affinities .
' II .
On Tritylodon longcevus ( Owen ) .
Plate 89 , fig. 15 .
[ i 1884 , Sir R. Owen described the anterior part of a skull from Thaba-chou , svoland , which was associated with remains of Kistecephalus and ^urt .
Journ. Geol .
Soc./ vol. 40 , p. 146 , plate 6 ) .
In its dentition the anterior t are placed almost as in rodents ; there are no teeth in the usual position of ues ; the molar teeth approximate to those of , and appear to have 3(oots divided as among most Mammals .
The author concluded that the osteology fce skull showed mammalian conformity with the characters of the teeth .
Xe evidence which I obtained at Lady Frere of the dentition in new fossil animals tb approximate to Theriodont types , leads me to re-examine the osteology of the Li of Tritylodon .
In its general form the contour of the skull of the reputed imal is similar to that seen in reputed Reptiles ( Plate 89 , figs. 15 , 16 ) .
This is Lsnt in the mode of development of the parietal crest which bifurcates at its trior termination in both types , in the way seen in carnivorous Mammalia .
The sion of the orbit is the same in both types , though in Tritylodon only its trior and inner border is preserved .
Sir R. Owen has assumed that the orbit is incomplete on its posterior border , as among carnivorous Mammals , but there 3S to me ground for inferring that it was completed posteriorly , as among liodonts .
The transverse expansion of the sub-orbital and supra-alveolar bar of enaxillary bone ( Plate 89 , fig. 15 mx ) is a character common to Theriodonts and amals ; but its lateral direction is singularly Theriodont ( fig. 16 mx ) , for many he skulls are characterized by a sudden transverse widening at the orbits , a character is associated with a transverse narrowing of the skull in front of the Its , which gives a somewhat bulbous aspect to the extremity of the snout , due to Brior convexity of the maxillary bones .
This character is found in Theriodonts 17hich , however , it is associated with the extension of the roots of the canine teeth r the maxillary bones , while in Tritylodon it results from a similar upward and akward extension of the roots of teeth which have been termed incisors .
The i-umstance that those teeth do not meet in the median line , but are separated by interspace wide enough to have contained small incisor teeth , coupled with the 1 that indications of successional teeth which have been partly absorbed , but are betimes not absorbed , are found behind the canines in Theriodonts , makes it ibtful whether those teeth may not be regarded as canines comparable to the ines of Theriodonts , rather than as incisors comparable to the incisor teeth of ^mmals like Rodents .
The nares are terminal in Tritylodon and in Theriodonts .
f^re is a median anterior process of the premaxillary like that which in Theriodonts forms the inter-narial septum in advance of the alveolar border , and this character 6 p 2 is at least suggestive that the anterior nares may have been conditioned as in Then donts rather than as in Mammals , though the small size of the premaxillary bom and the development of the nasal bones , is similar in both .
The posterior nares \#163 ; .
conditioned as in Theriodonts , opening between the hinder molar teeth .
This characv is not mammalian , though ^Marsupials make an approximation m having the pal ; , more or less largely perforated between the hinder molars .
And since the palatei not completely preserved , there is no proof that Tvitylodovb had no posterior transve^ bar in the palate such as characterizes Marsupials , though the absence of a bonylete on the inner border of the hinder molar teeth makes the existence of that mareu^ character improbable .
As far as the evidence goes the posterior nares are Theriodct .
In view of these resemblances , it may be mentioned that Sir R Owen thoughit possible that the parietal foramen , which is unknown in Mammals , might be indict at the anterior bifurcation of the parietal crest .
This would have placed it nine frontal bone .
No foramen exists in the position indicated , and the depression is as is seen in some Theriodont skulls .
If a parietal foramen existed it could only We been situate in the part of the parietal crest which is lost .
The most interesting region of the skull is that which borders the orbit .
Thfifal shows no obvious post-frontal bone .
Therefore , the orbit of the eye appears to \lt ; ffer from that of a Theriodont in not being completed posteriorly by bone .
It lsjunii this position that the skull is fractured .
There is no character which distinguishes the Theriodont skull from that of a Mammal than the presence a post-frontal bone , which , in all known specimens , forms a transverse bar at the u of the orbit , and it unites with the frontal bone by a squamous suture .
It deveps posteriorly a plate which extends into the temporal vacuity ( Plate 89 , pobcenoixy \#171 ; \#187 ; paW \#151 ; , _ ^_____^____ " +.La eidft\#171 ; of the frdal Marsupials , or other Mammals , which have the and classification of the fossil beptilia .
1027 , d ttan ; n Sir R , Owen 's figure .
The broken lateral surfaces which I have " d"d indicate a structure now lost which apparently extended outward from the the orbit .
It is , therefore , in the position of the transverse bar of the post-J bone but appears to have been more slender than in any Theriodont known present , which approximates to Tritylodon in form .
Thus , although the post- ' ital bar'is unknown , the evidence seems to me to warrant the conclusion as highly Abable that it was present in the complete skull and defined the back of the orbit *iu Theriodonts . .
The pre-frontal bone is distinctly shown in Sir R. Owen 's figure 1 , forming the itero-superior border of the orbit .
Its posterior border is impressed by the missing irt of what I have regarded as the post-frontal bone , so that the pre-frontal and Lt-frontal bones united above the orbit , and excluded the frontal from the orbital [ rder , as among Theviodonts .
The presence of a pre-frontal bone is a reptilian .aracter unknown among Mammals , and its identification has some bearing on the due of the recognition of the post-frontal .
The pre-frontal bone ( Plate 89 , fig. 15 pr.f ) is an external superior surface above the front of the orbit , which is -jf inch long , eing a long triangle -\#163 ; ginch wide in front where widest .
It is margined externally a sharp angular ridge , which separates the superior area from the internal orbital -ea which is nearly vertical , flattened , a little concave , looks outward and backward , rd appears to be * inch deep .
Anteriorly the bone joins the lachrymal bone by a ell-marked suture .
^ . .
The lachrymal bone shows the usual external area in front of the orbit , which is t of an inch long and ^ inch deep .
It meets the maxillary bone below and in ?
ont , and joins the frontal and pre-frontal above .
The frontal bone is M inch wide t the widest point , where the lachrymal joins the pre-frontal .
A sharp marginal idge , continuing that of the pre-frontal , separates the external surface from the iternal posterior orbital surface of the lachrymal bone , on which the lachryma .amen is seen .
# , The identification of the pre-frontal and post-frontal bones is in harmony wit t e her resemblances to Theriodonts , which have been enumerated .
Hence I be 1\#174 ; ve Lat what remains of the skull goes to show that Tritylodon was a Heptile , an t at le skull might be restored upon the Theriodont plan .
If the teeth had been aknown it would have been difficult to find grounds upon which it could have even Terred to the Mammalia , since the only character in which it differs from Theno-onts is the vertical elevation and median union of what I have regarded as posterior lates of the post-frontal bones .
The other parts of the skull at present unknown iay have been as reptilian as those of Theriodonts .
It follows that if is lassed as a Reptile on the evidence of skull structure now stated , it has molar teeth , nth complicated cuspidate crowns , and long , straight , divided roots apparent y , ^ are only known in Mammals* It is the only animal at present known in w nc ^liese characters are combined with reptilian skull structure .
So fax t e genus are 1028 PROFESSOR H. G. SEELEY OH THE FOSSIL REPTILIA .
may be said to be more Mammalian than any known Theriodont .
But just as the development of divided roots to molar teeth is not a general character among Mammals , so the presence of double roots to the molars of Tritylodon , if established may only show that the character is found in Reptiles which approximate to Mammals in many other characters .
Some of the lower Mammalia have two roots to the canines as is well known ; and others appear to have two roots to the incisors .
erctwieles uasutd ( R. Coll. Surg .
, 3876 ) has a double-rooted incisor seen in the translucent premaxillan bone in advance of the functional canine , which has a single root .
In Petrodromus tetra dactylus the last tooth in the premaxillary appears to have the root divided ( R. Coll Surg .
, 3368 ) .
There is no evidence of this kind of variation in the roots of teetl among Reptiles , recent or fossil , so that the character may be important enough to place Tritylodon in a group of animals intermediate between IMtUiinials an ; Theriodonts .
One feature in the arrangement of the molar teeth is an obviou resemblance to Mammals , for those of the two sides of the jaw are straight and near !
parallel ; while in Theriodonts they are frequently , though not invariably , curved , an diverge as they extend backward .
Fig. 4 .
Anterior extremity of the right ramus of a mandible .
Other animals may become known hereafter which are allied to Tntylodon .
i 1889 I examined in the South African Museum at Cape Town a small fragm\#187 ; i the right ramus of the lower jaw , obtained by Dr. Kannemeyer from near Burghe:-dorp in 1884 .
It is fractured ( / ) behind the first pre-molar tooth .
It shows tto ( or possibly four ) small , cylindrical , close-set incisor teeth ( i ) , with the crowns bro \lt ; .
The canine ( c ) is in close contact with the incisors .
It is ovate , with flattened act .
and has its long axis in the length of the jaw .
It measures half-an-inch from** to back .
Behind , the canine there appears to be a toothless diastema ( m ) ooP with matrix , which is concave in length , and more than one and a ha mcfijml between the canine and the small first pre-molar ( pm ) , which appears to be sr quadrate .
The animal was larger than Tntylodon , and has incisors and canines the usual type .
The general form of the jaw is such as might be expected in Tnty $ede/ Phil. Trans. 1894 .
B. 89 .
G.M " Woodman ?
3 .
del.et.11th .

