[ 267 ] IX .
An Account of the Devonian Fish , Paheospondylus Gunni , Traquair .
By W. J. Sollas , F.R.S. , Professor of Geology and the University of Oxford , and Igerna B. J. Sollas , B.Sc. , Lecturer on Zoology in , Cambridge .
Received May 19 , \#151 ; Read June 11 , 1903 .
[ Plates 16 and 17 .
] The small and obscure Devonian fossil , first described by Dr. Traquair in 1890 ( 1 ) , # under the name of PaUeospondylus Gunni , has excited a keen interest among Zoologists , owing to its asserted resemblance to the Lampreys ; so that although but a short period has elapsed since its discovery , it already fills an important place in our literature .
Its position in the animal kingdom has been made the subject of controversy , and is still open to discussion .
Dr. Traquair , whose extensive knowledge of fossil fishes gives great weight to his opinion , concludes that " it not only appears to be , but actually is a Marsipobranch " ( 7 ) ; Professor Howes is no less positive and speaks of it as " an unmistakable Marsipobranch " ( 26 ) ; Dr. Smith Woodward , while expressing himself more cautiously , definitely includes it among Cyclostomes ( 4 ) , as the sole representative of a new order , the Cyclise , a term which had been previously " proposed by Professor Gill ( 14 ) , to designate a group of uncertain rank , of which Palseospondylus is the only known member .
In face of this agreement , an equally imposing array of authorities may be cited as bolding different views .
Professors Gegenbaur ( 21 ) , Wiedersheim ( 17 ) , and Zittel ( 12 ) , regard the claims of Palseospondylus to a close alliance with Marsipobranchs as at least doubtful ; Dr. Bashford-Dean , who has made a special study of recent Marsipobranchs , is even more sceptical ( 23 ) , Huxley is said to have styled the fossil a " baby Coccosteus , " the late Sir J. W. Dawson ( 8 ) suggested a comparison with larval Amphibians , and Professor Graham-Kerr claims it for a Dipnoan ( 24 ) .
That competent observers should thus be divided in opinion will occasion but little surprise , when we recall how imperfectly the remains of the organism appear to be preserved and how incompletely they are exposed to view .
On the other hand , the large amount of genuine information , which has been extracted from this obscure material , may well be a matter for astonishment ; it affords a striking testimony to the powers of skilled investigators , provided with abundant specimens for comparison .
If on the present occasion , we are able to throw any fresh light on the subject , this * These numbers in parentheses refer to the entries in the Bibliography at p. 291 , 292 .
( 222 .
) 2 M 2 18.11.03 .
is entirely owing to our employment of a new method of research , that of studying the fossils in a series of sections taken at minute and regular intervals apart .
This \#151 ; though applied to comparatively few specimens , none of them , as originally exposed to view , so perfect as some which have passed through the hands of Dr. Traquair , or as some examples displayed in the collections of the British Museum\#151 ; has given definiteness and precision to much that was imperfectly known or only surmised before , and has exposed the general anatomy of the organism with unexpected completeness .
Dr. Bashford-Dean ( 23 ) has remarked that " even in fairly good specimens , the structures are not only difficult to determine , hut individuals will be found to vary in important details .
" This is perfectly true when the fossil is studied in the usual manner , but not when more searchingly investigated , indeed , that which has inspired us with the greatest confidence in our results has been their singular consistency .
Having once obtained a correct representation of the complete skeleton , it becomes possible to identify without difficulty all the parts of even the most fragmentary remains .
Still , much remains obscure , as must necessarily he the case when sutures are absent , articulations vague , and a possibility exists that soft cartilaginous parts may have vanished in the natural course of petrifaction .
Mode of Occurrence.\#151 ; The fossil occurs , apparently in not inconsiderable numbers , in beds of lower Old Bed Sandstone age , which are worked in a shallow pit at Achanarras , Caithness ; it is most abundant in bands 2 or 3 feet in thickness , met with at a depth of about 6 feet below the surface ( 6 ) .
The rock is a hard flaggy sandstone , composed of minute angular grains of quartz , and a few scales of muscovite mica , cemented by carbonates of lime and iron ; fragments of coaly matter , and occasional granules of pyrites are also present : when fresh it is bluish-grey in colour , or when much carboniferous matter is present , almost black ; it weathers usually to a yellowish-grey , sometimes to an ochreous tint .
The fossil consists of a black lustrous substance , to be described later , which is very resistant to the weather , so that fairly good specimens , with their black tint unimpaired , are sometimes to he seen as conspicuous though minute objects on the rotten surface of flagstones , which must have lain for a long time exposed to the air , as is shown by their having weathered inwards for a depth of a quarter of an inch or more .
The best specimens , however , are always comparatively fresh .
The fossil is often fairly complete , presenting sometimes the ventral , sometimes the dorsal surface to view ; as hitherto some doubt has been felt on this point , it may he as well to add that serial sections place the matter beyond dispute , on grinding down examples in which the dorsal surface is exposed , the ventral is revealed and versed , in those exhibiting the ventral surface the dorsal becomes revealed .
Not infrequently the fossil is incomplete , the dorsal or ventral moiety alone being present .
This difference as regards completeness , is in many cases connected with the manner of splitting of the rock ; under favourable circumstances , as when a thin ACCOUNT OF THE DEVONIAN FISH , PAL^EOSPONDYLUS GUNNI , TEAQUAIP .
269 slialey parting determines the splitting , a plane of separation may be produced , which will expose the fossil and yet leave it intact , but when such partings are absent or of comparatively insignificant thickness , then the plane of fracture must pass through the fossil itself , if this is to be exposed to view , and the nearer the fracture passes to the middle of the fossil , the better it will be displayed .
In this way slabs of rock may be obtained , one bearing the ventral , and the other the dorsal moiety of the fossil .
The splitting of the rock takes place naturally during the winter frosts , and the specimens thus set free are those which most readily catch the eye of the collector .
Most of the specimens which we received from the collector , Mr. Donald Calder of Thurso , were exposed on a more or less weathered surface , but at our request he sent us some obtained by splitting the rock with a hammer ; on taking serial sections of one of these slabs and its counterpart , we found , as we had expected , that the dorsal moiety lay in one slab and the ventral in the other .
Nothing , however , was gained beyond the knowledge of this fact ; our best specimens were those naturally exposed .
Good specimens are said to be rare , a collector who in a gathering of 20 finds a single good one may be regarded as fortunate .
The excellence of a specimen is usually judged , however , not only by its completeness , but also by the perfection of its exposure at the surface ; a factor of no account when the fossil is studied by serial sections .
Hence many a specimen which to superficial observation would appear far from good , will yield excellent results on grinding down .
As regards the outer part of the fossil which has lain exposed to the air , considerable imperfection might be expected , since although very resistant , the fossil is not absolutely weather-proof !
Greater interest attaches to the deeper part , included in the unaltered substance of the rock , for this has remained hermetically sealed up , and thus protected from all destructive action , since the time when its petrifaction was first completed .
Judging from these parts alone , we find that the organism was preserved in very various stages of decay , in some cases decomposition was far advanced and only fragmentary remains of the skeleton are left ; in others it was arrested at an early stage and the general anatomy of the hard parts is almost as well displayed as it would be in a recent specimen .
Still even in the best examples , parts are missing here and there ; the auditory capsules , probably on account of the thinness of their walls , are never quite complete , and the same is true of the rostral processes .
The outlines also are rarely even and flowing , but usually present an eroded and jagged edge .
The specimen in which this defect is least marked is also the smallest we have encountered , it measures nearly 14 millims. in length and the length of the head is only 2'5 millims. ; the contours of this , particularly in the region of the skull , are remarkably smooth and even .
The parts of the skeleton are , as a rule , singularly little displaced , not more than can be accounted for by gentle movement in the water , which surrounded them at the time of their death , or by the increasing pressure of accumulating sediment .
The delicate fin-rays still retain their association with the vertebral column , and even movable parts , such as the so-called " post-occipital plates , ' ' suffer disarticulation only in the rarest cases ; a neural arch may be detached from its centrum , but is never carried far away .
The effects of pressure are evident , but extremely slight ; the hollow cylindrical vertebrae have been squeezed into a flattened band , and the skull has suffered more or less distortion , parts originally separate have been fused together , and a general compression has taken place from above downwards ; but these changes , although greatly increasing the difficulties of interpretation , are the result of very trifling movements , so trifling as to suggest that the consolidation of the rock must have followed speedily upon deposition ; a suggestion which is strengthened by the extraordinary state of preservation of such specimens as are represented by fig. 1 , Plate 16 , fig. 7 , Plate 17 , in which the thin cranial walls rise vertically upwards , with but little indication of having been deformed by pressure .
As a consequence of the absence of disturbance during interment and petrifaction , the organisms retain the attitude in which they were left by death .
The head always lies with its dorsal and ventral surface parallel to the lamination of the flagstone , a rule to which we have found no exception , notwithstanding the large number of specimens we have studied ; this is obviously the result of its depressed tadpole-like or skate-like form .
The vertebral column as invariably lies on its side , except in the immediate neighbourhood of the head : this position having evidently been determined by the dorso-ventral extension of its median fins .
The different disposition of the skull and vertebral column naturally involves a twist somewhere , and the effects of this are manifest in the case of the first few vertebrae , which lie next the skull ; these are overturned so as to lie on their articular face , and are sometimes imbricated , or even in rare cases piled on one another ; this latter derangement can scarcely be accounted for by spasmodic muscular contractions , but would seem rather to have resulted from the position of the cranial base , which was raised slightly above the sea-floor by the underlying visceral arches .
Judged by their various size , specimens of nearly all ages are associated in the flags of Achanarras ; the smallest specimen we have examined measures 14'0 millims. in length , the smallest described by Dr. Basiiford-Dean only 6 millims. , but since the head of this is said to measure as much as 3 millims. while that of ours measures only 2*5 millims. in length , the specimen is probably incomplete .
The largest specimen is said by Dr. Bashford-Dean to attain a length of 52 millims. and may have slightly exceeded this , since the head is imperfect .
The average length is about 25 or 30 millims. Numerous species of other Lower Devonian fishes are found with Palaeospondylus , of which a list is given by Dr. Traquair ( 6 ) , but fish and some ill-preserved plants are its only associates .
The substance of which the fossil consists closely resembles ordinary coal in appearance ; it is black , lustrous , of considerable hardness , just below 4 of Mohr 's ACCOUNT OF THE DEVONIAN FISH , PALHSOSPONDYLUS GUNNI , TKAQUAIK .
271 scale , brittle , breaks with a concboidal fracture , has a density of from 1*35 to 1*7 , and burns with a smoky flame , leaving a considerable quantity of ash .
It does not dissolve in xylol , benzol , .chloroform , or carbon tetrachloride .
Its mineral characters thus closely correspond with those of coal .
It is structureless in thin slices , which thus afford no help in an attempt to determine its original histological character .
Associated fish remains , such as those of are formed of a similar material , having all the properties just enumerated , but with a slightly lower average density .
The mass of a Coccosteus is very considerable , so that it is possible to obtain enough of its substance for a complete chemical analysis .
Accordingly the black coaly matter of a specimen of Coccosteus decipiens from Achanarras was chiselled out of its matrix , crushed in a steel mortar , passed through the meshes of a fine sieve , and then placed in a separating apparatus filled with chloroform to remove associated shale and other impurities .
The lighter part , of density below 1*5 , was removed , dried , and again placed in the separator , this time filled with a mixture of chloroform and xylol , having a density of 1*37 ; the heavier portion , which thus had a density of between 1*37 and 1*5 , was removed , and dried in a water-oven .
It amounted in weight to about 2 grammes .
Mr. J. E. Marsh , M.A. , very kindly undertook its analysis , with the following results:\#151 ; Composition per Cent. I. Hydrogen ... 4*7 Carbon . . . .
68*1 Oxygen , \amp ; c. . .
11*4 Ash ... ... ... ... ..15*8 100*0 II .
III .
Mean .
4*5 4*6 4*5 68*7 68*3 68*4 11*1 11.1 11*3 15*7 15*0 15*8 100*0 100*0 100*0 An examination for nitrogen showed that this element was not present in any considerable amount , certainly not more than 0*5 per cent. Sulphur was not estimated .
The " oxygen , \amp ; c. , " in the tables given above is obtained by difference .
An analysis of the ash , made by one of us , showed the presence of 31*73 per cent , of phosphoric anhydride , equivalent to 68*77 per cent , of calcium phosphate .
The composition approaches very closely that of an impure cannel coal , as will be seen from the following table :\#151 ; Coal from Coccosteus .
Carbon . . . .
68*4 Hydrogen . . .
4*5 Oxygen , \amp ; c. . .
11*3 Ash . . . .
\#187 ; 15*8 Cannel coal .
66*44 7*54 12*2 13*82 100*0 100*00 If the ash be disregarded it even more closely resembles an ordinary non-caking coal , such as that of the South Staffordshire coal-fields ; as is shown in the next table :\#151 ; Coal from Coccostous .
Non-caking coal .
Carbon . .
81*1 79-39 Hydrogen 5*3 5-36 Oxygen , \amp ; c. . . .
13-6 15-25 100-0 100-00 Thus both in physical characters and chemical composition the substance of the Coccosteus fossil , and consequently of Pabeospondylus , is a true coal .
That coal might in some instances result from the decomposition of fish remains has often been conjectured , but never yet conclusively proved .
These results gain greatly in interest by the fact that occurs in another and very different state of fossilisation , as for instance in the calcareous nodules of the Moray Firth .
Within these the skeleton has been protected from the effects of pressure , so that the bones are not squeezed together as in the Achanarras specimens , but retain a separate individuality ; at the same time their organic basis has disappeared , leaving only the mineral matter , so that instead of being compacted into a black structureless coal , they are preserved as white clearly defined plates , in which histological characters are still plainly decipherable .
Thin slices under the microscope are seen to be crowded with the characteristic cavities left by bone-corpuscles , and to be traversed by numerous minute canals .
The corpuscular cavities are best displayed at the instant when the preparation of the thin slice is completed by the addition of balsam and a cover-glass ; they are then filled with air and consequently obvious , but as this is driven out by the balsam they rapidly disappear , only a single one here and there remaining visible .
Owing to the fortunate occurrence of species of Coccosteus in two different states of fossilisation , we are furnished with precise information as to the nature of the material which has been converted into coal ; it was evidently bone .
The organic basis of this may be presumed to have closely approached chondrin in composition , and this as regards the relative quantities of carbon and hydrogen , which enter into it , is so similar to cellulose that its transformation into coal need not occasion surprise .
The composition of the two substances is given below :\#151 ; Carbon .
Hydrogen .
Oxygen .
Nitrogen .
Rest .
Chondrin . .
4774 676 3P04 13-87 0'59 Cellulose ... 44'2 6 '3 49*5 \#151 ; \#151 ; To convert chondrin into coal it must lose all its nitrogen , a little more than one half its hydrogen , and about three-quarters of its oxygen , the total loss of substance amounting to about 41 per cent. This is a minimum estimate and may have been ACCOUNT OF THE DEVONIAN FISH , PALJEOSPONDYLUS GUNNI , TRAQUAIR .
273 much exceeded .
To what extent the mineral constituents of the skeleton would be added to or diminished we have no means of judging .
Evidently the chances are that a considerable contraction in bulk accompanied the process of carbonisation ; but the matrix about Palreospondylus adheres to it closely , no interspace exists between the rock and the fossil , either empty or filled with a mineral deposit , the sand-grains of which the rock chiefly consists are in immediate contact with the fossil .
Hence carbonisation may be inferred to have been practically completed before the rock was consolidated .
Taking all the facts into consideration we are inclined to think that the transformation into coal was not effected very slowly .
An important question next arises as to what was the original character of the skeletal substance of Pakeospondylus .
As already pointed out it now consists of coal , which , so far as can be seen , differs from that of Coccosteus solely by a slightly greater density , due probably to the presence of a somewhat larger proportion of mineral salts .
That it has been derived from a substance not greatly differing in composition from bone is highly probable , but there is no evidence to show that this substance actually possessed the histological structure of bone , nor , on the other hand , is there any direct evidence to prove that it did not .
The assumption has been made , that it consisted , not of cartilage bone , but of calcified cartilage ( 5 , 7 ) ; this may very well have been the case , but the suggestion arises solely from zoological analogies .
That the skeleton consisted of fairly hard substance is suggested by the remarkable manner in which thin walls and delicate processes are sometimes preserved .
In concluding this account of the mode of occurrence of Pakeospondylus , we may endeavour to reach some conclusion as to the physical conditions under which the sediments of its " gisement " were deposited .
The rarity of any fossils in the Old Red Sandstone , except fishes and Crustacea , and of these except sporadically or at special horizons , where they suddenly become abundant , is a very remarkable fact , and becomes all the more so when the rich and diversified nature of the contemporaneous marine faunas is recalled to mind .
Had the Old Red fish inhabited a fresh-water lake , it would have been strange if fresh-water molluscs had not lived along with them , and this argument is strengthened rather than weakened by the isolated occurrence of the so-called Anodonta Jukesii of Kiltorcan , in Ireland ; since , if this fossil has been correctly referred to the genus , it proves that fresh-water molluscs were in existence , and flourished in considerable numbers where conditions were favourable .
That these conditions were not general over the Old Red area seems fairly certain .
In Russia there is much evidence to suggest that some parts of the system were deposited in salt lakes , pseudomorphs of rock-salt occurring in some localities , and the generally unfossiliferous character of the Old Red Sandstone of the British Isles may be most readily explained by regarding it as the deposit of " dead " seas .
Perhaps no precise analogy can be pointed out amongst existing basins ; in some respects we are reminded of the Black Sea , in others of the Caspian or the Great Salt Lake of Utah .
VOL. cxcvr.\#151 ; b. 2 n The hypothesis of a lifeless sea leads naturally to the suggestion that the Old Red Sandstone fish were inhabitants of the freshwater rivers of the adjacent land , and were washed into the lakes by sudden freshets , just as happens at the present day to the fish which live in the rivers tributary to the Great Salt Lake of Utah .
The fish carried into this lake are said to perish instantly ; an equally sudden death , followed by a rapid burial , seems to have overtaken the assemblage of fishes now preserved in the flagstones of Achanarras .
Description of the Anatomy .
The Head : Dorsal Aspect ( Plate 16 , figs. 1 , 3 ; Plate 17 , figs. 7 , 9).\#151 ; Viewed from above the skull presents a median longitudinal trough , bordered on each side by paired lateral structures , and produced in front into numerous spine-like processes .
The trough , which extends throughout the length of the skull and occupies the middle third of its breadth , clearly represents the cranial cavity .
The most posterior of the lateral structures are a pair of cavernous bodies , which , as Dr. Traqtjair ( 5 ) rightly conjectured , represent the auditory capsules .
The most anterior are a pair of arcades , applied to the cranial walls , and partially divided by a transverse partition ; these we regard as the nasal capsules .
An important element , having somewhat the form of an inverted " L , " lies on each side , immediately behind , the nasal capsules , and was probably related to the eye , which was situated above it .
This element is indicated in Dr. Traquair 's figures by the letter a , and is spoken of by him as " the additional lobe .
" As it will be frequently referred to , it requires a distinctive name , and in order to avoid as far as possible the introduction of hypothesis into description , we propose to call it the " gammation , " from its fancied resemblance to the Greek letter " V. " Another element , lying immediately in front of the gammation and nearer the middle line , is almost entirely concealed by the nasal capsules , but is partly visible in the posterior compartment of the nasal capsule of the left side in Plate 16 , fig. 3 .
This may be termed the pre-gammation .
A transverse division of the skull into a somewhat larger anterior and smaller posterior portion may be made , owing to the fact that the gammation is more closely connected with the nasal than with the auditory capsule , with which , indeed , it never appears to come actually in contact .
In one specimen\#151 ; the smallest\#151 ; an obvious deep groove divides the coherent anterior region from the posterior ( Plate 17 , fig. 9 ) .
The cranial cavity is open in front , behind and above .
In the least distorted specimens of large size its walls are steep , for the most part vertical , and thin , about OT millim. in thickness , but in the smallest specimen at least twice as great .
Their height in proportion to the length of the skull is by no means inconsiderable ; ACCOUNT OF THE DEVONIAN FISH , PAL^EOSPONDYLUS GUNNI , TRAQUAIR .
275 in the case of one skull , which measures 5'3 millims. in length , their height is 1'0 millim. ; in another , 3*75 millims. in length , the height is 0'4 millim. They pursue an almost parallel course for the greater part of their length , that is for the posterior four-fifths , and are then bent slightly outwards ( Plate 16 , fig. 1 ) at about the anterior termination of the nasal capsules , after which they resume a parallel course , but at a greater distance apart , and with diminished height .
These low anterior walls are not visible in all specimens , they appear to be absent in the one shown by Plate 16 , fig. 3 , probably as a result of displacement .
At their posterior extremity the cranial walls are confused with the walls of the auditory capsules , but they retain their individuality in the anterior region of the capsules .
The upper edge of the posterior cranial walls is frequently more or less curved over towards the middle line ; this is evidently in part a result of distortion by pressure , but also may also indicate the presence of an incomplete cranial roof of which only the margins are preserved ( Plate 17 , fig. 7 ) .
The cranial walls are perforated by large vacuities , which are constant in position in all our specimens ; most posterior is a wide gap on each side by which the cavity of the auditory capsule is put into communication with that of the skull ; this , indeed , is a complete solution of continuity , for it is not roofed over .
A communication between the cavity of the ear capsules and the brain case is characteristic of some Teleosteans , but it might easily be produced in Palseospondylus as the result of post-mortem changes .
The next opening lies in front of the auditory capsule , opposite the concavity of the gammation ; as this is roofed over , it is not visible from above , a strip of paper has , therefore , been passed through to indicate it in the figures ( Plate 16 , figs. 1 , 3 ) .
The aperture from its constancy in occurrence and position may be regarded as indicating some feature original to the skull , most probably a foramen , which has been enlarged by subsequent decomposition .
We consider that it gave exit to the optic nerve , possibly to the fifth and seventh as well .
In front of the enlarged optic foramen no constant vacuity is met with in the side walls of the skull , we look in vain for such an aperture as might have afforded egress to the olfactory nerve .
But precisely at that point , where the sudden flexure of the cranial walls occur , there is an easy way into the anterior compartment of the nasal capsule , through which the nerve may have passed .
The posterior opening of the cranial cavity , corresponding to the occipital foramen , is not completed by a roof , a fact rendered all the more remarkable by the presence , immediately behind it , of the neural arch of the first vertebra , still retaining its original position with regard to the skull or only slightly displaced to a lower level , although the body of the vertebra has been separated from it , and turned over so as to lie on its articular face .
The arch is far from well preserved , however , and is usually in a very fragmentary condition .
The absence of an occipital arch in the PROFESSOR W. J. SOLLAS AND MISS IGERNA P\gt ; .
J. SOLLAS ON AN fossil does not .
necessarily prove that it was not present in the organism ; the apparently capricious way in which parts are preserved in one specimen and not in another , offers a warning against putting too much faith in merely negative evidence ; the number of specimens we have examined in which the dorsal surface was embedded in the rock , and , therefore , most likely to he well preserved , is not great , ( some six or seven in all ) , and with regard to these it must be borne in mind that if the organism came finally to rest on its ventral surface , the upper part of the skull would be the last to be covered up with sediment , and would thus be the longest exposed to oxidising and destructive action .
The floor of the cranium , as rightly surmised by Dr. Thaquatr , is continuous , save for a slight interruption in front , and there are no signs of a basi-cranial fontanelle .
It is fairly level in the posterior region , between the auditory capsules , such irregularities as do occur being obviously due to decomposition ; but just on a line with the anterior border of the auditory capsules or between this and a line joining the optic foramina of opposite sides , it makes a sudden step-like descent to a lower level , recalling in a striking manner the " saddle-ledge " which occurs in a corresponding position in many Elasmobranchs , and which marks in them the position of the pituitary fossa .
Beyond this region the floor descends very gently , and becomes slightly narrower ; the cranial walls ceasing to be quite vertical near their base , curve slightly inwards as they pass into the floor , so that the cranial cavity in this region , i.c. , between the nasal capsules , is narrower , as well as deeper than elsewhere .
The floor then gently rises and widens forwards , but again deepens near its termination between the low anterior walls ; in this region it is crossed by a fissure which extends transversely across it and separates two distinct elements of the skull from each other .
Beyond this it makes an upward rise and continues at this level to its termination .
The two elements just mentioned will be referred to again in treating of the ventral aspect of the skull , but the most anterior of them may conveniently be noticed here .
So far as it contributes to the floor of the skull it may be described as a transverse bar or fillet , somewhat higher in front than behind , providing a support on each side for the terminal half of the low anterior cranial Avails .
The thickness of the bar , which for brevity may be called the " ampyx , " is considerable , and it extends downwards to the ventral face of the skull , where it is seen as a very conspicuous ridge ( Plate 1G , fig. 5 ; Plate 17 , fig. 10 ) .
Further description of this element is rendered difficult , owing to its extension into other regions not sharply marked off from it , it is confused laterally with the walls of the nasal capsules and may be produced beyond them as a thin broad lamella which fills the angle between the capsules and the lateral rostral process .
The rostral processes or rostralia are forward extensions of the ampyx , arising from it at different levels , and imperfectly distinguished into a dorsal and ventral series .
ACCOUNT OF THE DEVONIAN FISH , PALHEOSPONDYLUS GUNNI , TRAQUAIP .
277 They are most completely preserved and displayed in a unique specimen described and figured by Dr. Traquair ( 7 , Plate 9 , fig. 1 ) .
In this they number altogether 11 , of which 2 are lateral , and by far the longest and strongest , 5 are dorsal , and continuously united by a ' thin membranous-like expansion , and 4 are ventral ; 1 of the dorsal series is median in position , the others are symmetrically disposed on each side of it , the ventral processes are symmetrically arranged 2 on each side of the middle line .
No signs of dorsal rostralia were discernible in any of our specimens as naturally exposed , hut clear evidence of their existence was obtained in sections .
The ventral and lateral , on the other hand , are generally visible in hand specimens .
The ventral rostralia are evidently extensions of the ampyx , as may be learnt from Dr. Traquair 's figures .
They proceed from its front margin near the ventral surface ; thus in the specimen shown in Plate 17 , fig. 11 , they make their first appearance in the third section upwards from the base , or 0'2 millim. from the ventral surface .
This specimen , however , being one of the first experimented on , was cut at intervals of OT millim. ; had the intervals been smaller the processes might have appeared sooner , but not by more than 0'025 millim. Lateral processes appear in the same specimen a little above the ventral , and are followed by dorsal ones , of which only the basal portions are present .
The left lateral process is bent near its termination towards the right side .
A specimen in which the ventral surface is unusually well preserved ( Plate 16 , fig. 5 ) shows four processes in the first section by which the ventral face of the ampyx is revealed ( counting from below upwards ) ; these are both unusually large for ventral rostralia , and are not exceeded in length or breadth by the lateral processes , which appear in the succeeding sections , and are followed later by dorsal processes .
These , like the ventral , appear to he extensions of the ampyx , hut are given off at a higher level .
An interesting series of sections is afforded by the specimen shown in Plate 16 , fig. 1 , the ventral surface is not particularly well preserved , but traces of the ventral rostralia appear in the second section , and continue through to the eighth , when others , apparently both dorsal and ventral appear .
The dorsal are the three anterior , two of which are conjoined by their expanded base , they are continued no farther than the next section , where the outer one on the left joins the adjacent outer process of the series behind .
Beyond this ragged fragments of broadly extended thin processes are seen in the succeeding four sections ; these are not connected with the ampyx , which is much reduced in breadth ( antero-posterior extension ) in the first of these sections , is interrupted in the middle in the second and third , and has wholly disappeared in the fourth .
The basal or proximal ends of the rostralia\#151 ; whether ventral or dorsal is difficult to say\#151 ; are well exposed on the upper surface of the ampyx in the fine specimen shown in Plate 16 , fig. 3 ; judging from their number and position they would seem to be dorsal .
They occur as sharply marked ridges symmetrically arranged on each side of the middle line .
The innermost are the smallest , scarcely extending beyond the front margin of the ampyx ; the second pair are larger and broader , extending for more than one-third of their length forwards beyond the ampyx , as well as slightly backwards , the ampyx accompanying them , so that its posterior margin is rendered sinuous , a median sinus corresponding to the space included by the innermost rostral pair , and a lobe on each side of this to the second pair .
The third or outermost pair are by far the largest , extending 1 millim. in front of the ampyx .
A median dorsal rostralium is not clearly expressed , hut a minute rounded projection in the middle line , just below the groove which separates the first pair , may possibly represent it .
The lateral rostralia do not differ greatly in size from the outermost dorsal , they appear to arise just below the latter , and that on the right side is bent sharply in the middle , so that its distal half crosses underneath the adjacent outermost dorsal process , of which it deceptively appears to be a branch .
The occurrence of this flexure in other specimens affecting the corresponding rostralia leads to the suggestion that the lateral rostralia pointed not only forwards and outwards but also downwards , bending would then naturally result from the action of pressure , such as might be produced by the mere weight of the organism .
A general shifting of the rostral apparatus backwards seems to have taken place in this specimen , leading to a concealment of the anterior cranial walls and carrying the base of the outermost dorsal process into the anterior compartment of the left nasal capsule .
In another specimen ( Plate 16 , fig. 5 ) two of the rostralia are seen to extend backwards into the cranial cavity for a distance of at least half a millimetre , while the base of another is found within the aperture of a nasal capsule .
Dr. Traqtjair seems to have observed something similar in his specimens , since he speaks of the lateral processes as passing " within the rim of the opening .
" From the facility with which they undergo displacement one can only infer that their attachment must have been very loose .
The rostralia seem to have retained their original disposition most completely in our smallest specimen ( Plate 17 , figs. 8 and 9 ) .
In this the ventral are very diminutive , the lateral well developed , but several others are confused with them ; two of the dorsal processes are seen on the upper surface of the ampyx , bordering a deep narrow groove .
The general appearance in this case strongly suggests that the rostralia formed a terminal fringe around the anterior aperture of the cranial cavity .
It is worthy of note that in this specimen the cranial cavity is slightly constricted just in front of the anterior walls of the nasal capsules , so as to be divided into a larger posterior portion , the true cerebral cavity , and a much smaller anterior one , with which the rostralia seem to be directly associated .
The Nasal Capsules.\#151 ; The structures , which we think may have supported the olfactory organs , occupy the greater part of the anterior half of the sides of the head ; ACCOUNT OF THE DEVONIAN FISH , PAL^EOSPONDYLUS GUNNI , TRAQUAIK .
279 they have the form of a half boat , longitudinally divided and inverted , the parts corresponding to the cut surfaces being applied to the walls of the skull .
A transverse vertical wall or bar crosses the cavity of each half-dome , almost in the middle , thus dividing it into nearly equal compartments , one anterior , the other posterior .
One end of the transverse bar is closely applied to the inner wall of the dome , with which it seems to be confluent , the other meets the wall of the skull , and in some well-preserved and scarcely distorted specimens seems to be continuous with it ; in others , which have yielded to pressure , it retains a separate existence , even when closely adpressed to the skull walls .
Usually both compartments of the capsules are open above , but in some cases the anterior is roofed completely over .
Both are freely open below .
Several small apertures in the cranial walls place its cavity in communication with that of the capsules , but these are not so constant in position as we should expect if they were original .
It does not seem likely that foramina , large enough to give exit to the olfactory nerves , would become obliterated by a secondary deposit of coal during the process of fossilisation , nor can we suppose that they have been closed up by pressure , at least in the case of such specimens as still present thin vertical cranial walls , which have certainly not been exposed to any adequate downward compression .
There is nothing , however , to interrupt the free communication of the cranial cavity with the anterior compartment at the point where the cranial walls are suddenly diminished in height ( Plate 16 , fig. 1 ) , and the exit of the olfactory nerves at this comparatively high level would not be inconsistent with analogy .
The use of the term nasal capsules may be objected to as a petitio ; if a purely descriptive name should be thought preferable , they might be spoken of as the " anterior hemidomes .
" The Auditory Capsules.\#151 ; As will be seen from the figures , these are of considerable size ; in a young Skate , of which the cranial cavity measures 24 mi dims , in length , the auditory capsules have a length of 6 millims. , giving a ratio of 1:4 ; in a specimen of Palaeospondylus the length of the cranial cavity was found to be 4'2 millims. , that of the auditory capsules 1*5 millims. , giving a ratio of 1 : 2'8 .
The capsules extend very nearly the whole length of the postero-lateral mass , but not over its outermost lateral portion , this is covered by a thin plate , which overlies the gills .
The walls of the capsules were evidently very thin , as is shown by the generally fragmentary character of their remains , yet in one instance they are well enough preserved to still show traces of the course of the anterior and posterior semicircular canals .
The free communication which exists between the cavity of the capsules and that of the cranium has already been mentioned .
The Ventral Aspect of the Head .
( Plate 16 , figs. 2 , 5 ; Plate 17 , figs. 8 , 10.)\#151 ; The ampyx with its rostralia forms the most anterior portion of the skull , and is succeeded immediately behind on the ventral surface by a somewhat " T ''-shaped element , which may be termed the " tauidion .
" The front margin of the " T " fits against the posterior margin of the ampyx , from which it is clearly defined , and from which it was readily separable , as may be seen in specimens where the parts have been slightly displaced by pressure ( Plate 16 , fig. 5 ) .
The slender shaft of the " T " is directed backwards in the middle line .
The tauidion forms the anterior part of the floor of the skull , and there is nothing to show that it does not pass insensibly into the cranial walls .
In specimens showing the ventral surface naturally exposed , a separation appears to exist , but in sections of these specimens it can no longer be traced .
The tauidion may consequently prove to be a region rather than a separate element of the skull .
Two small foramina perforate it on each side in the angle formed by the junction of the arms with the shaft , they are about 0T milliin .
in diameter , and occur 0T5 millim. above its most ventral surface .
They are indicated in Plate 16 , fig. 5 , by inserted slips of paper .
The Visceral Arches.\#151 ; Branchial arches are present , more or less completely preserved in all the specimens we have examined ; they are best displayed by the specimen shown in Plate 16 , fig. 5 .
Their position is invariably below the posterior region of the skull , concealed almost completely by the auditory capsules and the epibranchial plates .
They are four in number , and appear as a paired series of stout well-developed rods , diverging from the middle line outwards and backwards , those of each side remaining in general parallelism with each other .
The second and third are the longest and strongest , the fourth the shortest of the series ; thus , in the specimen just mentioned , the second measures 2*8 millims. in length , 0*5 millim. in height , and 0*3 millim. in breadth ; the fourth nearly 1*3 millims. in length ; in its other dimensions it resembles the second , but is slightly stouter .
Notwithstanding their difference in length , all four of the branchial bars terminate not far from the same straight line , and their anterior extremities in consequence make very different degrees of approach towards the base of the skull .
The second pair meet each other over the base just behind the tauidion , the third likewise , a little behind the second , but the distal ends of the first and fourth do not meet , and lie somewhat remote from the middle line .
The several bars are united dorsally near their posterior extremities , in the case of the second and third bars an apparent coalescence takes place , the two , in the specimen under description , becoming one for the last 0*1 millim. of their length ; the precise nature of the union is not always easy to make out , in some cases an evident bridge crosses from one to the other ; in others , the extremities pass into each other , the two bars forming a loop .
The third and fourth bars are united by a bridge near their extremity , and in addition by a broad plate , which lies transversely over their dorsal surface and looks like a separate structure .
The bars of the first pair are free at their anterior extremity , but attached posteriorly to the second , and are thus liable to a displacement , the anterior ends moving outwards , and the whole bar coming nearly into parallelism with the long axis of the skull .
The parts just ACCOUNT OF THE DEVONIAN FISH , PALJ50SP0NDYLUS GUNNI , TRAQUAIR .
281 described are probably cerato-branchial , but traces of epi-branchials are sometimes met with ; it is difficult to interpret otherwise the processes seen in connection with the back of the skull in Plate 17 , fig. 10 .
It has sometimes appeared as though basi-branchials might be present , but , if so , they are for the most part confused with the base of the skull .
The cerato-branchial bars are interrupted by perforations , revealed by sections , but whether these are original or not there is nothing to show .
In front of the branchial arches lies a pair of large and important structures , one on each " side of the middle line , separated from each other by the base of the skull or the backward extension of the tauidion .
Looked at each has somewhat the form of a trapezium , and bears two bars , one over its anterior , and the other over its posterior edge .
The anterior is much the smaller of the two ; it is separated by an interval from the overlying mass , and in one instance is brought into connection with this by its outer end alone .
A strip of paper has been inserted between the two parts in Plate 16 , fig. 5 .
The posterior , which is situated so as to continue the direction of the first branchial bar towards the middle line , is not separable in its present state from the skeletal mass immediately above it , but it has the appearance of having been originally a separate element .
In the identification of these structures lies in all probability the key to the interpretation of the affinities of Palseospondylus ; for facility of reference we propose to term them provisionally the anterior and posterior trapezial bars .
From a careful comparison of sections it would seem to result that the trapezium-like mass consists of the pre-gammation and the gammation ; the posterior trapezial bar is supported above by the gammation , while the anterior trapezial bar is borne by the pre-gammation .
Conflicting explanations will probably be found by different thinkers for these relationships ; that which we are inclined to regard as the most probable would identify the anterior trapezial with the mandible , the posterior with the hyoid , the gammation with the hyo-mandibular , and the pre-gammation with the palato-quadrate .
In this case the opening of the mouth must have been transverse .
The right and left trapezial bars are always separated by the base of the skull , perhaps as a consequence of the pressure of this upon them .
The Post-branchial Plates.\#151 ; These are the problematical " post-occipital " plates of Dr. Traquair , marked x in his figures .
They are comparatively large and important structures , found in the least distorted specimens lying immediately behind the skull , parallel with the vertebral column , one on each side of it .
Anteriorly , they extend for a short distance beneath the posterior region of the skull , so that their relation to other underlying structures can only be studied in specimens showing the ventral surface , or in sections .
These show , in the clearest manner , that the problematical plates have no connection with the cranium , but are directly related to the last pair of branchial bars , the rounded end or head of the plate resting in contact with the thick rounded anterior extremity of the bar , as though making articulation with it .
2 o VOL. CXCVI.\#151 ; B. That the post-branchial plates were freely movable in a horizontal plane is shown by the positions in which they lie ; sometimes one , sometimes the other crosses obliquely over the middle line , and in one beautifully preserved example both alike take a parallel oblique course , while their capitular ends remain in contact with the posterior branchial bars .
In this case the vertebral column is seen to overlie the plate that crosses its course .
Whether , they were also capable of movement in a vertical plane we are unable to say , there is no evidence to show that they were , and equally none to suggest that they were not .
It is difficult to be sure of the precise form of the plates , in some cases ( Plate 16 , fig. 6 ) they have the appearance of somewhat flattened cylinders , with a cavernous interior .
In the specimen referred to they measure 2'8 millims. in length , 0'5 millim. in breadth , and 0'3 millim. in thickness , but may have been somewhat thicker , since their dorsal surface is somewhat corroded ; in other cases ( Plate 16 , fig. 1 ) they resemble a plate curved from side to side into the form of a hollow half-cylinder , as if in conformity to the vertebral column which lies between them .
It seems not unlikely that the post-branchial plates may represent a modified branchial arch : in modern Elasmobranchs , as Gegenbaur has pointed out , the last branchial arch , whatever its number , is usually both stronger than those which precede it , and turned backwards to a greater extent , so as in some cases to assume a nearly parallel position to the vertebral column .
It is true that at the same time it becomes less movable , acting as a strut , or as a protection to the pericardium , but probably modern Selachians do not exhaust all the possibilities of modification which the last arch may undergo .
If the plates were capable of movement in a vertical plane , then they may have served as limbs , and the fourth branchial bar might be regarded as a potential coracoid , while the transverse plate overlying it and the third bar might represent a potential scapula .
In this case the homology with a fifth branchial arch would fail .
On the other hand , if their movement were limited to a horizontal plane , it might be maintained , and the function of the plates would then appear to have been to assist in securing the head to the vertebral column .
Specimens show that the neck was very liable to dislocation .
The Vertebral Column.\#151 ; The vertebrae , as displayed in reconstructions , confirm in all essential respects the description of Dr. Traquair .
The large centra have the form of hollow cylinders , with rounded ends , which are complete throughout the vertebral column , except in the region of the caudal fin , where only the ventral half appears to be well developed .
The neural arch roofs over a large canal , and seems not to have been very firmly attached to its centrum , since it is frequently found detached .
The backwardly directed processes which are given off from the ventral aspect of the caudal vertebrae are single and median .
They appear to be directly continuous with the centra .
We have not found any evidence to suggest that the notochord was more constricted in the middle region of the vertebral column than ACCOUNT OF THE DEVONIAN FISH , PALMOS POND YLUS GUNNI , TRAQIJAIR .
283 elsewhere ; as Dr. Bashford-Dean has asserted , the centra are somewhat longer and not so high as in the neighbourhood of the head , hut are not distinguished by other differences .
In the cervical region , a vertebra favourably situated for measurement was found to have a length of 0'35 millim. and a breadth of 0'7 millim. , giving a ratio of 1 : 2 .
In an overturned vertebra , showing the ring-like form , the height and breadth were nearly equal , and measured 0'7 millim. The length in this case was 0'3 millim. ; further back , five or six vertebrae behind the post-branchial plates , the length and height of the centra had become equal , and measured 0'55 millim. Judged by the lumen of the ring-like centra , the notochord must have been of great relative thickness , thus in one case the outer diameter of a ring was found to measure 0-9 millim. the inner 0'7 millim. , the walls being only 0*2 millim. in thickness ; so that , making no allowance for the enlargement which the central cavity may have undergone as a result of post-mortem changes , the notochord would have formed more than three-quarters of the diameter of the centrum .
There is nothing very remarkable in this , however , and ring-like remains of cyclo-spondylous vertebrae are not uncommon among fossil Elasmobranchs .
The first vertebra is distinguished from the rest by its very large and strong neural arch , which is intimately associated with the posterior region of the skull , so that in several cases it retains its original position more or less perfectly , although the centrum corresponding to it has been displaced and overturned .
The fact is interesting in connexion with the apparent absence of an occipital arch , and the suggestion naturally arises that this element may be represented by the neural arch of the first vertebra , which though not actually incorporated with the skull seems to be on the point of becoming so .
In the specimen shown on Plate 16 , fig. 3 , the arch of the first vertebra has been displaced as well as the centrum , and lies on its face , it has a span of 1*3 millims. and a height of 0*6 millim. , while the centrum measures 0- 9 millim. in -height and breadth .
The total height of this vertebra is now 1- 5 millims. , while the length of the head , measured from the front margin of the ampyx to a line joining the posterior margin of the posterior lateral masses , is 5'5 millims. The Absence of Ribs.\#151 ; The presence of ribs has been both asserted ( 15 and 19 ) and denied ( 16 ) , we have , therefore , made a very careful search for them , but without success .
Twice structures were found in connexion with the fossil which might have easily been taken for ribs , but when their true nature was ascertained by sections , they proved to be the remains of other parts slightly shifted out of their proper position ; in one case they were displaced post-branchial plates ( Plate 17 , fig. 10 ) , in the other detached neural arches ( Plate 16 , fig. 1 ) .
It is evidently almost as dangerous to assert the presence as the absence of parts from an examination of the specimens which are only superficially exposed .
The Absence of Limbs.\#151 ; Structures which might possibly represent the fragmentary and confused remains of a pectoral and pelvic girdle with their associated fins have 2 o 2 been met with in one specimen .
The specimen was studied in sections , and reconstructed , but without adding anything definite to our knowledge ; the vertebral column passes continuously through the confused problematical structures , but these cannot be resolved into definite parts .
The Affinities of Paleeospondylus.\#151 ; Suggestions have been made that Palseospondylus was a larval form , possibly of an Arthrodiran Fish , possibly of an Amphibian .
Dr. Bashford-Dean is much inclined to take this view , and has supported it by the assertion that the relative dimensions of the parts of the organism undergo great changes with growth , and , indeed , as represented by Dr. Dean in the following table , these changes are by no means inconsiderable .
\gt ; Table illustrating the Changes in the Approximate Proportions of Palaeospondylus at various Stages of Growth .
( Dr. Bashford-Dean ( 23 ) .
) Length of Length \lt ; 4 cranium .
Percentage of total length .
specimen in millimetres .
Breadth of head , Length of post-occipital plates .
Length of column .
Thickness of column .
6 50 30 50 4 13 30 23 15 70 17-5 21 12 7 79 2*0 22 17-7 11 6-8 83 1-8 2-1 16-5 11 66 83-5 1-7 52 15-4 10 \#151 ; 8P6 22 These results are of great interest , and if trustworthy of great importance .
Dr. Bashford-Dean furnishes us , however , with very little material for a critical examination of the process by which they were obtained ; no information is given as to how the length of the cranium was determined , nor as to the characters by which the completeness of the specimens was judged .
An examination of the figures illustrating Dr. Bashford-Dean 's paper does not inspire us with confidence as regards the latter point .
The actual results of measurement are not stated , but they may be elicited from the table , and are given from our calculations in the table below:\#151 ; Length of specimen .
Length of cranium .
Breadth of head .
1 Length of post-occipital plates .
Length of column .
Thickness of column .
G 3 1-8 3 2-4 13 3-9 2-99 1-95 9-1 2-86 17-5 3-675 2-1 1-23 13-825 3-5 22 3-74 2-42 1-496 18-26 3-96 24 3-90 2-64 1-584 19-04 4-08 52 7-908 5-2 \#151 ; 43-992 13-00 The measurements arc given in millimetres .
ACCOUNT OF THE DEVONIAN FISH , PAL/ EOSPONDYLUS GUNNI , TRAQUAIR .
285 A glance at these data is sufficient to show the laws of growth as represented by Dr. Bashford-Dean 's table of percentages are the result of an illegitimate method of comparison .
The fact that one specimen with a cranium measuring 3'9 millims. in length is provided with a vertebral column only 9T millims. long , while another with a head 3'9G millims. in length possesses a vertebral column 19'04 millims. long , does not illustrate a law of growth , but rather the danger of trusting to measurements made on specimens only superficially exposed to view .
The smallest specimen in the table is represented by a figure ( 23 , Plate 1 , fig. 9 ) in Dr. Bashford-Dean 's Memoir , which shows nothing but the vertebral column or a part of it , without any indication of a head ; to draw important conclusions from so obscure and insignificant a fragment appears to us hazardous in the extreme ; still more to make it the subject of measurements for comparison .
Our own smallest specimen has a total length of IPO millims. measured from the front margin of the ampyx to the extremity of the vertebral column , which may not be complete ; the length of the head between the front margin of the ampyx and a line joining the posterior margin of the postero-lateral masses is 2'5 millims. It does not differ in any structural character , so far as can be discovered , from our largest specimen , in which the length of the head is 5*5 millims. , except in the close apposition of the posterior border of the gammation to the cranial wall .
To arrange the specimens in the order of their total length , as Dr. Bashford-Dean has done , is not a proceeding which commends itself to our judgment ; any want of completeness in the length of the vertebral column cannot fail by this method to vitiate the whole of the results ; a safer method would be to select some one organ , of which the completeness can be more successfully estimated , as a basis of reference , and this would appear to be the head .
In the following series the specimens are arranged in the order of magnitude of the length of the head ( cranium ) , Dr. Bashford-Dean 's smallest specimen being omitted and measurements made from ours substituted ; measurements have also been added taken from Dr. Traquair 's most perfect specimen , they appear in the second line of the table :\#151 ; Length of cranium .
Length of column .
1 Ratio of length of vertebral column to length of cranium* .
2-5 11-5 4-6 3-5 20 5-7 3-675 13-825 3-76 3-74 18-26 4-86 3-9 9-1 2-33 3-96 19-04 4-30 4-67 27-1 5-8 7-908 43-992 5-56 * Taken as unity , The absence of regularity in the series of ratios shown in the last column is inconsistent with Dr. Bashford-Dean 's conclusions , and we can only suppose that some concealed imperfections in his specimens have led him to over-estimate their completeness .
On the other hand , the results suggest that the average ratio between the length of the head and " tail " is as one to five or six .
We are far from denying , however , that some change in the proportional size of the organs of Palseospondylus has taken place with growth ; indeed , there seems to be some evidence to show that it has .
Thus in our smallest specimen the length of the head anterior to a line joining the posterior border of the gammation ( which corresponds to a line joining the anterior margins of the postero-lateral masses ) measures 1*35 millims. , the part behind this line , IT millims. , giving a ratio of 1T8 : 1 ; in another specimen , chosen for its apparent freedom from distortion , the length of the anterior part is 2*465 millims. , of the posterior 1*785 millims. , giving a ratio of 1*38 : 1 , thus indicating a more rapid growth on the part of the anterior region , a result precisely corresponding to that found by Sewertzoff in his study of the early development of the Skate ( 27 ) , but further , the same observer has shown that this more rapid growth is confined to the rostral region , and does not affect the rest of the skull behind this region ; precisely the same fact is indicated by our two specimens : thus the low anterior walls of the skull measure 0*25 millim. in the small specimen and 0*765 millim. in the large one ; deducting these quantities from the length of the anterior division first mentioned , we have in the small specimen 1*35 \#151 ; 0*25 = 1*1 , and the ratio of this to 1*15 , the length of the posterior region , is 1 : 1*05 ; in the case of the large specimen 2*465 \#151 ; 0*765 = 1*7 ; the ratio of this to 1*785 is 1 : 1*05 , or identical with that found for the smaller specimen .
This suggests that the low anterior skull walls are closely associated with the rostral apparatus .
The ratio of 1 : 1*05 is so nearly unity as to show that the rate of growth of the parts to which it relates has been practically uniform .
It by no means follows , however , that Palseospondylus is a larval form ; it would have been surprising if no change in the relative dimensions of parts had taken place during growth , from a length of 14 millims. up to a length of 52 millims. , that of Dr. Dean 's largest specimen .
On the other hand , the absence of any important structural changes , which might have been expected to distinguish our smallest from our largest specimens , is a remarkable fact , and suggests that the characters of the adult were attained at a very early age .
Regarding Palseospondylus as an adult form , we next pass to a discussion of its relations to other Vertebrates .
Its claims to alliance with the Marsipobranchs have been ably urged by Dr. Traquair , who regards the presence of rostralia about a single anterior aperture as indicating a narial character for that opening ; if so , of course , the fossil must be regarded as a Monorhine ; the presumed absence of jaws has naturally led to the ACCOUNT OF THE DEVONIAN FISH , PAL^EOSPONDYLUS GUNNI , TKAQUAIR .
287 conclusion that it is a Cyclostome .
But the position of the rostralia at the anterior extremity of the cranial cavity is also that which they occupy in other fishes such as the Elasmobranchs , where they primarily serve as supports for the mucous glands .
These fill the anterior end of the cranial cavity , and are only partitioned off from the brain by a membranous septum , as may be readily demonstrated in a young Skate or Dog-fish.# The position of the rostralia is thus as much consistent with Elasmobranch as with Marsipobranch affinities .
As regards the presumed absence of jaws , we have already pointed out structures that are probably of this nature , and with this the last of the two most convincing arguments in favour of Marsipobranch relationship disappears .
There remain then features of indicative but not of fundamental importance , such as the asserted nature of the caudal fin-rays , which in their dichotomy do certainly recall a Marsipobranch character in a most striking manner .
We can offer no observations of our own on the nature of these structures ; our specimens were not well enough preserved , but Dr. Bashford-Dean sees no reason why they should not be dermal , and he considers the fact that they twice dichotomise supports him in this view .
The absence of ribs and limbs cannot by itself be regarded as furnishing an argument for a Marsipobranch alliance .
The arguments against such an alliance , however , have become much strengthened in the light of additional knowledge ; the characters of the cranium are foreign to the Marsipobranch , the highly developed vertebral column equally so , while the visceral skeleton is absolutely unlike anything known amoiig existing Marsipobranchs .
Professor Graham Kerr finds a remarkable agreement between the general features of Pakeospondylus and the Dipnoi as illustrated by Ceratodus , he compares the post-branchial plates with the cranial ribs of a young Dipnoan , and remarks on the presence on the ventral surface of the fossil of a pair of plates , which recall the two moieties of the hyoid arch of Lepidosiren .
These last are evidently what we should regard as a pair of branchialia , belonging to the second or third arch .
The post-branchial plates cannot be interpreted as cranial ribs , since they are not attached to the skull , but to the last branchial arch ; a glance at Plate 17 , figs. 8 and 9 , will be sufficient to suggest their true nature .
That a certain similarity may be recognised between the head of Palseospondylus and that of a young Dipnoan , such as Ceratodus , can scarcely be doubted .
If we turn to the figure by Sewertzoff ( 28 ) of a lateral view of the skull of an advanced embryo of Ceratodus , we shall find similar large auditory capsules , a cranial cavity not completed by a roof , with its walls higher behind than in front , and a large * In a young skate , 8 inches in length , the brain entirely fills the cranial cavity , with the exception of that part occupied by the mucous glands , and the wide space in front of the brain met with in the adult is non-existent .
foramen immediately in front of the ear , giving passage in this case to branches of the fifth and seventh nerves .
There is as yet nothing to correspond to the anterior hemidomes , but they might readily arise as developments about the eye and nasal capsules ; in such a case the eye would be situate in the posterior , and the nasal capsule in the anterior compartment of the hemidomes .
The branchialia , though very much smaller , correspond in number and position with those of Palseospondylus , but the hyoid and lower jaw are immensely larger than the parts we have identified with these elements in the fossil ; their suspension also is altogether different .
If Palaeospondylus is to be interpreted as a Dipnoan , we must either suppose that the lower jaw has been lost , or we must find it articulating with the gammation , and biting against the quasi-maxillary cartilage .
The first is not an impossible assumption , \#151 ; it might be as logical to conclude that the Mammalian lower jaws of the Stonesfield slates had no heads , as that Pala30spondylus had no lower jaw\#151 ; but it would be made use of as a valuable admission by those who might still prefer to regard Palseo-spondylus as a Marsipobranch .
As regards the other horn of the dilemma , it may be urged that we are mistaken in the identification of the lower jaw , and that it may occur in the requisite position .
We have done our best to find it there , and , in one instance , thought we had succeeded .
In a particularly fine specimen , of which the ventral surface is shown by Plate 17 , fig. 13 , a more or less distinct ridge may be traced running forwards from the gammation , conformable with the quasi-maxilla on the left side , close to it , and , as longitudinal sections show , below it .
The ridge terminates near the extremity of the transverse arm of the tauidion .
If this could be shown to be the lower jaw the resemblance to a young Ceratodus would be indeed remarkable .
Unfortunately none of our other specimens have afforded evidence that would confirm this identification , and we are compelled somewhat reluctantly to abandon it in favour of that we have already proposed .
Again , as a matter of fact , structures corresponding to the hemidomes are not present in Ceratodus , nor if they were , is it easy to see how the eye could have been placed in the posterior compartment , which is unprovided with a foramen , through which the optic nerve could find issue to it .
Tempting as the Dipnoan interpretation certainly is , we are unable to adopt it , chiefly on account of the great apparent difference in the nature of the suspensorium and manducatory apparatus .
Sir J. W. Dawson , on what appear to us insufficient grounds , has suggested that Pakeospondylus may prove to be a forerunner of the Batrachia or a primitive tadpole .
There may be other arguments than those he urged in favour of such a view , but the remarks just offered with regard to the Dipnoi will , for the most part , apply equally to the Amphibia .
No one seems yet to have proposed to associate the fossil with the Elasmobranchs , and yet they have much in common .
ACCOUNT OF THE DEVONIAN FISH , PALiEOSPONDYLUS C4UNNI , TRAQ .
UAIE .
289 Except for the presence of a roof* in that of the latter , the crania of the two are not very dissimilar , and even as regards the roof , this is not complete in the Selachians .
If we take the similar aural capsules as a starting point in the comparison , we find the eye of the Elasmobranch situated immediately in front of them , and the optic nerve passes out , frequently through a large foramen , opposite to it .
In Palseo-spondylus a large aperture exists in a similar position , and if it represents an optic foramen , then the eye must also have occupied a corresponding position .
The saddle-ledge , which marks the place of the pituitary body in the floor of the Selachian skull , is met with similarly placed in Palseospondylus .
If the hemidomes of Palseospondylus are really nasal , then the nasal capsules have a similar position to that which they occupy in the Elasmobranchs .
In the latter , however , there is no such marked transverse bar or median septum as occurs in the hemidomes .
Professor Bridge suggests that this may be represented by the partition between the nares , as seems very possible ; it would also seem possible that the posterior compartment might represent the ethmoidal incisure of some Elasmobranchs , such as Prionodon or Pristiurus : its posterior wall would then correspond with a pre-orbital process , and its lateral wall to the lateral ethmoidal process .
The pre-frontal gap of the Elasmobranch skull occurs in a region corresponding to that which lies between the low anterior walls of the fossil skull .
The position of the rostralia is similar in both .
There is nothing in the Elasmobranch skull to correspond with the transverse line of separation between the ampyx and the tauidion of Palaeospondylus .
On reviewing the resemblances and differences between the two skulls , it would appear that the former outweigh the latter ; on the whole , the general topography of the two is strikingly concordant .
In strong contrast with this conclusion is the evidence of the visceral skeleton .
The branchial arches in their number , limited to four , and in their position beneath the skull afford a character , which finds no parallel among the Elasmobranchs , and it is not till we reach the higher groups of Dipnoi and Amphibia , and then only in larval forms , that we again encounter it , but in these it is precisely repeated .
With an agreement so close in this respect , we might reasonably expect to find an ecpially marked correspondence between the jaws and suspensory apparatus , but , as we have already seen , this expectation is disappointed .
These structures in Palseospondylus remain , we believe , without any precise analogy .
Reasoning upon the observed facts alone the position of the gammation and pre-gammation is as anomalous on a Selachian as upon a Dipnoan interpretation ; they suggest the idea of a divided sub-orbital bar , but this , though ending close to the middle line of the skull on the one hand , and to the auditory capsules on the other , is united to neither .
The primitive connection of the quadrate bar with the trabecula in the Dipnoi and VOL. CXCVI.-B .
2 P Amphibia is behind the eye , and this is the connection of the gammation with the crania ] walls in our youngest specimen , but in larger specimens , though apparently freed from this connection , the gammation has not entered into union with the auditory capsules ; it may have done so , but there is no absolute proof that it did .
Still , our friend , Professor Bridge , who has carefully studied our reconstructions , remarks that " it looks as though such a connection must have existed " ; in tliis we fully concur , and if we may assume that the appearance represents the fact , we may identify , as we have done , the gammation with the hyo-mandibular , the pre-gammation with the palato-quadrate , the posterior trapezial with the hyoid , and the anterior trapezial bars with the mandible .
The lower jaw would , in this case , have been singularly diminutive ; it could have bitten against the palato-quadrate , and its position would have been transverse .
Its two halves are now separated in the middle line , and it is unprovided with teeth .
Teeth might , however , have been originally present ; they would have been very minute , and highly calcified .
Under the conditions which were favourable to the conversion of cartilage into coal , much carbonic acid would be evolved , in presence of which calcareous structures , such as teeth , might easily pass into solution .
The interpretation of the anterior arches which seems most probable is thus frankly Selachian , but no known Elasmobranch is provided with jaw7s so insignificant in size , or with a hyoid which , as in this case , greatly exceeds them in magnitude .
Dr. Goodrich , commenting on the relative dimensions of the trapezial bars , would regard the anterior as labial , and the posterior as mandibular .
On this view , of the four branchial arches the anterior would be the hyoid .
There is evidently room for a difference of opinion as regards the precise interpretation of the anterior arches , but that their facies is Selachian seems to us evident .
The position of the branchial arches is apparently a primitive character , and when persistent , correlated with comparatively small jaws .
In Dipnoi , although the jaws are large , they are placed far forward ; in Amphibia they are small , so long as the branchial arches are present beneath the skull ; their backward position in Selachians seems to be a direct consequence of the enormous development of the jaw apparatus , with its large posteriorly placed suspensorium .
The number of the branchial arches may or may not be primitive ; in the advanced embryo of Ceratodus , as described by Sewertzoff , there are only four , but in the adult Dipnoi there are five , an increase which suggests the possibility that the larger number of the Selachians may not have been an original character .
If the branchial arches present us with merely primitive , indifferent characters , generally possessed by ancestral fishes , then the remaining characters of Palaeo-spondylus which can be looked upon as distinctive are those connected with the jaws , and these seem to point rather in a Selachian direction than any other .
If Palseospondylus took its origin near the base of the Elasmobranch stem , it proceeded in its subsequent development along an independent course , losing its limbs , if it ever ACCOUNT OP THE DEVONIAN PISH , PAL .
EOSPONDYLUS GUNNI , TBAQUAIK .
291 possessed them , and acquiring a highly organised vertebral column , homoplastic in character with that of cyclo-spondylous Selachians .
It thus does not seem possible to place Palseospondylus in any of the accepted sub-classes of fishes ; it probably belongs to that legion of primitive forms which must have peopled the Devonian waters , but of which the vast majority have disappeared without leaving any trace of their existence .
We desire in conclusion to express our warm thanks to our numerous friends who have interested themselves in the progress of this investigation , to Dr. Smith Woodward and Dr. Traquair , who made special journeys to Oxford to examine our specimens , to Professor Weldon , to Dr. Goodrich , and particularly to Professor T. W. Bridge for much useful information and suggestive criticism .
Bibliography .
1 .
Traquatr , IT .
H. " On the Fishes found at Achanarras Quarry , Caithness , " 4 Ann. and Mag. Not .
Hist. ' ( 6 ) , vol. 6 , pp. 485 , fig. 4 .
1890 .
2 .
Woodward , A. Smith .
4 Catalogue of the Fossil Fishes in the British Museum , ' part 2 , p. 553 .
1891 .
3 .
Howes , G. B. 44 On the Affinities , Inter-Belationships , and Systematic Position of the Marsipobranchii .
" 4 Trans. Biol. Soc. Liverpool , ' vol. G. 1891 .
4 .
Woodward , A. Smith .
44 The Forerunners of the Backboned Animals , " 'Nat .
Science , ' pp. 597-599 .
1892 .
5 .
Traquair , It .
H. 44 A Further Description of ( Traquair ) , " 4 Proc. Boy .
Phys. Soc. Edin .
, ' vol. 12 , pp. 87-94 .
Plate 1 .
1893 .
G.----------------------44 Achanarras Revisited .
" lbicl .
, p. 279 .
1893 .
7 .
-------------------\#151 ; 44 A Still Further Contribution to Our Knowledge of Palwospondylua Gunni !
'Ibid .
, pp. 312-321 , Plate 9 .
1893 .
8 .
Dawson , J. W. 4 Salient Points in the Science of the Earth .
' London , pp. 285-286 .
1893 .
9 .
Traquair , IT .
H. 44 PalceospondylusGunni ( Teaquatr ) from the Caithness Flagstones , " 4 Ann. Scot .
Not .
Hist. , ' pp. 94-99 , Plate 3 .
1894 .
10 .
Neumayr , M. 4 Erdgeschichte , ' vol. 2 , pp. 12 , 112 .
1895 .
11 .
Dean , Basheord- .
4 Fishes Living and Fossil .
' New York , pp. 65 , 70-71 .
1895 .
12 .
Zittel , K. A. von .
4 Grundziige d. Palseontologie .
' Munich and Leipzig , p. 527 .
1895 .
13 .
Gill , Theodore .
44 Note on the Systematic Position of Palseospondylus , " in a Review of Dean 's 4 Fishes , ' 4 Science , ' N.S. , vol 3 , pp. 914 , 915 .
1896 .
2 r 2 292 PROFESSOK W. J. 80LLA8 AND M188 IGERNA B. J. 80LLA8 ON AN 14 .
Gill , Theodore .
" Note on the Devonian Palaeospondylus .
" Ibid. , vol. 4 , pp. 10 , 11 .
1896 .
15 .
Dean , Bashford- .
" Is Palaeospondylus a Cyclostome ?
" 'Trans .
N.Y. Acad. Sciences , ' vol. 16 , pp. 101-104 , Plate 5 .
1896 .
16 .
Traquair , B. H. " Note on the Affinities of ( Traquair ) , In Reply to Dr. Bashford-Dean of New York , " ' Proc. Zool. Soc. Lond. , ' pp. 314-317 .
1897 .
17 .
Wiedersheim .
' Elements of Comparative Anatomy/ p. 74 .
1897 .
18 .
Parker and Haswell .
'A Text-book of Zoology , ' London , vol. 2 , pp. 133-134 .
1898 .
19 .
Dean , Bashford- .
" Remarks on the Affinities of Palceospondylus In Reply to ( Dr. Traquair ) , ' ' ' Proc. Zool. Soc. Lond. , ' pp. 343-347 .
1898 .
20 .
Woodward , A. Smith .
' Outlines of Vertebrate Palaeontology , ' Cambridge , pp. 1-3 .
1898 .
21 .
Gegenbaitr , C. ' Vergl .
Anat .
d. Wirbelthiere , ' Leipzig , vol. 1 , pp. 65 , 363-365 .
1898 .
22 .
Dean , Bashford- .
'On the Embryology of Bdellostoma Stouti'(Festschrift Carl von Kupfer ) , Jena , p. 221 .
1899 .
23 .
------------------- " The Devonian ' Lamprey , ' Gunni ( Traquatr ) , " 4 New York Acad. Sciences Memoirs/ vol. 2 , pp. 1-32 , Plate 1 .
1900 .
24 .
Kerr , Graham .
" The Zoological Position of Palaeospondylus ( Traquair ) , " 4 Cam .
Phil. Soc. Proc./ vol. 10 , p. 298 .
1900 .
25 .
Sollas , W. J. " The Investigation of Fossil Remains by Serial Sections , " ' Rep. Brit. Assoc. ' 1901 , p. 643 .
26 .
Howes , G. B. " Presidential Address to Sec. D. Brit. Assoc. " 'Rep .
Brit , Assoc. ' 1902 , p. 626 .
27 .
Sewertzoff , A. N. 4 Die Entwickelung des Selachier-Schiidels ' ( Festschrift Carl von Kupfer ) , Jena .
1899 .
28 .
------------------- 44 Sir Entwickelungsgeschichte des Ceratodus Forsteri , " 4 Anat .
Anzeiger , ' vol. 21 , pp. 593-608 .
1902 .
ACCOUNT OF THE DEVONIAN FISH , PALH50SP0NDYLUS GUNNT , TRAQUAIR .
21)3 EXPLANATION OF PLATES .
Plate 16 .
Kg- Fig. 2 .
Fig. 3 .
Fig. 4 .
Fig. 5 .
Fig. G. Dorsal aspect of a reconstruction .
In the specimen from which the reconstruction was obtained the ventral surface was exposed to view , the dorsal was embedded in the rock .
Ventral surface of the same reconstruction .
Dorsal aspect of a reconstruction .
The specimen was embedded in the same way as that of Fig. 1 .
The ventral surface was consequently imperfect .
A slight distortion has led to a shortening of the anterior region of the head , and to the concealment of the anterior cranial walls .
The same reconstruction with the dorsal moiety removed .
The left hemidome , with the gammation , pregammation , and a part of the cranial wall , is shown in the upper left-hand corner .
Ventral aspect of a reconstruction .
The specimen from which the reconstruction was obtained was embedded by the ventral surface , the dorsal side is therefore the less perfect .
Dorsal aspect of the same reconstruction .
( The magnification of the figures on this plate is 10 diameters .
) Plate 17 .
Fig. 7 .
Dorsal aspect of a reconstruction .
The specimen was embedded on the dorsal side , but it had suffered much disturbance and decay before petrifaction .
The reconstruction shows the cranial wall of the right side curving roof-like over the cranial .cavity at the point marked Fig. 8 .
Ventral aspect of a reconstruction .
The specimen , which was the smallest in our collection , was embedded by the ventral surface .
Fig. 9 .
Dorsal aspect of the head of Fig. 8 .
Fig. 10 .
Dorsal aspect of another reconstruction .
The specimen was embedded on the dorsal side , and had been much compressed from above downwards .
( Figures 7 to 10 are magnified 15 diameters .
) Note.\#151 ; The specimens from which the reconstructions shown in figs. 1 to 10 were obtained were cut at intervals of 0'025 millim. Fig. 11 .
Ventral aspect of a reconstruction .
The specimen was embedded on the ventral side .
( Cut at intervals of 0T millim. X Fin .
12 .
Both aspects of another reconstruction .
( Cut at intervals of 0T millim. X 15 .
) Fig. 13 .
Ventral surface of an actual specimen ( X 7^- ) .
Explanation of Abbreviations used Figures .
am .
Ampyx .
br.c.y^ .
First to fourth cerato-branchials .
hr .
Epibranch ials .
br.b .
Basi-branchial .
br.p .
Post-branchial plates .
c.w.Cranial wall .
c.w.a , Anterior cranial wall .
e.br.pEpibranchial plate .
for .
Foramen in front of auditory capsule .
g. Gammation .
g. p. Pre-garamation .
h.Hemidome .
h.a.Anterior wall of hemidone .
h. p. Posterioi* wall of hemidome .
h. a.e.Anterior compartment of liemidome .
h.m.s. Median septum of hemidome .
h.p.c.Posterior compartment of hemi-dome .
h.p.w. Posterior wall of hemidone .
mx. Quasi-maxillary cartilage .
r. Rostralia .
r.l.Lateral rostralia .
r.d.Dorsal rostralia .
s.c.a.Anterior semi-circular canal .
s.e.p. Posterior semi-circular canal .
t. Tauidion .
tf Foramen in tauidion .
tp.b.a .
Anterior trapezia !
bar .
tp.b.p .
Posterior trapezial bar .
v. Vertebra .
v.b.Centrum of vertebra .
v.n.a.Neural arch .
Sottas and Sottas .
Phil. Trans. , B , vo 196 , Plate 16 .
s.c.a. e.brp .
\amp ; c.p. v. n v.b. br.p .
rfi- r.L. Fig. 2 .
v.n.a. Fig.l .
Fig. 3 .
Fig. 4 .
Fig. 5 .
Cp.b.a .
Fig. 6 .
C , f. h.m.s. h.a.c. h.p. \lt ; )\#166 ; p.o. cum .
g\gt ; br.cz .
br.c5 .
br .
Cq .
v.n.a. Phil. Trans. , B , ml .
196 , 17 .
Fig. 12 .
j\#151 ; / * !
I\#151 ; CMd .
\#151 ; h. \#151 ; c.w. bp.b.a .
bp.b.p .
Fig. 9 .
v -r , *S ip fS / ?\#163 ; .\#151 ; br.p \#151 ; am .
b. mx.

