FURTHER STUDIES 
ON 
Grammicolepis Brachiusciilus, 
Poey. 
BY 
R. W. SHUFELDT, M.D., C.M.Z.S. 
[Reprinted from Journal of Morphology, Vol. II., No. 2, November, 1SS8.] 
BOSTON: 
GINN AND COMPANY. 
1888. FURTHER STUDIES 
ON 
Grammicolepis Brachiusculus, 
Poey. 
BY 
R. W. SHUFELDT, M.D., C.M.Z.S. 
[Reprinted from Journal of Morphology, Vol. II., No. 2, November, iSSS.J 
BOSTON: 
GINN AND COMPANY. 
1888. FURTHER STUDIES ON GRAMMICOLEPIS BRACHI- 
USCULUS, Poey. 
By R. W. SHUFELDT, M.D., C.M.Z.S. 
During the spring of 1872, Professor Poey of Havana, Cuba, 
came into possession of a very remarkable form of fish, which 
presumably was taken in Cuban waters. 
Fortunate it was for science that it fell into such excellent 
hands, as that eminent ichthyologist promptly presented us 
{Anal, de la Soc. Esp. de Hist. Nat., Tom. 11, 1873) with a very 
excellent account of this more than rare type, the duplicate of 
which, so far as I am aware, has yet to be found by natural- 
ists. 
This account of Professor Poey's, as will be seen, was pub- 
lished in the Spanish language, and it has given me much 
pleasure to make a translation of it, and present it here as an 
introduction to some subsequent examinations which I had the 
rare opportunity of undertaking upon the skeleton of the speci- 
men in question. The osteology of this fish is very interesting, 
not only from the fact that it is the only specimen in the 
hands of science, but from its extraordinary peculiarities, and 
from the fact that it may some day be found upon our own 
coasts. 
My translation of the original description just referred to 
reads as follows : - 
"GRAMMICOLEPIS BRACHIUSCULUS. 
TYPE OF A NEW FAMILY IN THE CLASS PICES. 
By Don Felipe Poey, 
Professor in the University of Havana, and Corresponding Member of the Academy 
of Exact, Physical, and Natural Sciences of Madrid. 
The length of this extraordinary fish is 470 millimetres. The 
head enters five times into the total length of the body, and two 272 
SHUFELDT. 
[Vol. II. 
and two-thirds times into its greatest depth. The body is much 
compressed, and quite deep. The very large eye is contained 
two and a third times in the length of the head, and lacks the 
membrana adiposa. 
The branchial apertures are deeply cleft, but I fail to find 
more than four branchiostegal rays, without being able to assert 
that there may be a greater number of them. The snout is 
short. The prefrontal, the turbinal, and the anterior suborbital, 
are extremely hard, and covered with spiny rugosities. The 
preoperculum and interoperculum have rugose borders, while the 
remaining opercular bones are entirely so. The mouth is 
small, subvertically cleft; the premaxillary process is large, 
and is lodged in a fossa of the cranium. The maxillary is com- 
plicated. The teeth are simply a narrow row of minute pric- 
kles ; they do not occur upon the vomer, nor the palatines. 
D. 6-34; A. 2-33; V. 1, 6 ; P. 15 ; C. 1-13-1. The leading 
spine of the first dorsal series is rugose, as is the first ventral, 
the two post-anals, and the external ones of the tail, which latter 
show the condition equally well in either one. 
The rays of the pectoral, second dorsal, and the anal fins are 
compressed, and do not ramify at their extremities. The pecto- 
rals are very short and rounded. On the other hand, the vertical 
fins, the dorsal, and anal are well developed. 
The tail was injured, and apparently cut; the membrane 
which unites its rays had disappeared ; the peduncle which 
supports it is large, and capable of communicating a powerful 
impulse to the act of progression. The thoracic pectorals un- 
questionably possess a rugose spine and six flexible ones that 
are branched. 
Aside from the frontal bones and the suborbitals where the 
skin abruptly terminates, and the nasal portion of the snout, all 
the trunk and the head is covered with scales, including the 
inferior mandible. 
These scales in no way resemble those found among the 
acanthopterygean fishes. Their length greatly exceeds their 
width ; they have the appearance of parchment, - transparent, 
brittle when dry, - overlap each other, and are strengthened 
longitudinally by a raised lineal ridge. • 
Their contact with each other is so extremely intimate that it 
lends to the skin of either side a very smooth appearance - so No. 2.] 
GRAMMICOLEPIS BRACH1USCULUS. 
273 
much so, that the rough borders of the scales would not be 
suspected without the aid of the fingers. 
Thanks to the length of these scales, four, five, or six of them 
are sufficient to span the height of the trunk, one of such a 
series being crossed by the lateral line, where its presence is 
denoted by a raised ridge. 
The leading scales on the body, above as well as below, are 
shorter, and where carried on to the head, are doubly as firm as 
those found at the base of the fin rays. 
Without having done more than counted the scales in a longi- 
tudinal line, I calculate that the number is considerably above 
two hundred ; those of the head, although shorter, have the 
same form as those of the trunk. There are no scales upon the 
fins. 
The caudal peduncle develops neither a cartilaginous nor an 
osseous plate at its sides. Posterior to the arms the ventral 
keel is rough. 
The cranium is more cartilaginous in structure than it is 
osseous, except the frontals, which are rugose in lines in the 
supraorbital region, and bristly in front, as are the turbinals and 
suborbitals; these latter are four in number, the last three 
being very slender. There are two supratemporals. 
The inferior mandible is characterized by several rows of 
minute spines upon the dentary and articular elements. The 
vertebrae number 10+36. 
The anterior neural spine is not excavated, being lofty and 
smooth ; the five that follow are short and inclined backwards. 
The remaining ones are slender, which applies also to their 
haemapophyses. The last vertebra is without lateral spines. 
The pleurapophyses are inconspicuous, feebly developed, and 
have much the same size and shape as the epipleurals. I dis- 
cover but one pseudo-interneural spine in front of the one that 
supports the first dorsal fin ray. 
The specimen I described, when received by me was without 
gills and without abdominal viscera. Preserved as it was three 
days upon ice, its general color appeared to be white; but we 
have reason to suspect that in the fresh condition this fish can 
easily assume a violet tint. The hard parts of the head were of 
an intense violet shade. The ascending border of the preoper- 
culum, violet. The fins were white, changing to violet in cer- 274 
SHUFELDT. 
[VOL. II. 
tain lights; the caudal fin rays were of a reddish tint. Eye 
inclining to white. 
Family : The characters which are presented us in this fish 
are of such an extraordinary nature that they will not permit 
us to place their possessor in any of the recognized families of 
fishes. Its nearest affinities are with the Bcrycidce and with the 
Carangidce, two families widely separated from each other; I 
am inclined to believe that its better place is along-side of the last- 
named one. Its resemblance to the Bcrycidce is seen in the 
large eye ; the asperity of the cranium ; the rugosities upon the 
fin rays; the ventrals composed of more than five soft rays, 
over and above the spiny one ; its resemblance to the Carangidce 
is seen in the two free spines which precede the anal fin, and 
especially to Scriolas for lacking the bony plate of the lat- 
eral line; but in the number of its vertebrae it approaches the 
Scombridce, as the shape of its ventral fins are in pattern analo- 
gous to those of the Acanthuridce, and its unramified fin rays 
agree with the Balistidce. 
The character of the scale, to which ichthyologists have 
attached so much importance, separates it from all other forms 
known to me. 
My examination, then, authorizes me in establishing the fam- 
ily Grainniicolcpidce, based upon the following characters : Lat- 
eral line unarmed with bony plates; ventral fins composed of 
more than five soft rays; two free postanal spines; caudal ver- 
tebrae numerous; scales very long and narrow, without fan-like 
expansions or denticulations. 
Genus : The genus Grammicolcpis has for its etymology ypap,- 
[ugos, line ; Xem?, scale. 
The characters, in addition to those I have already pointed 
out for the family, are : Body deep, compressed ; eye large; 
mouth small; head, in part, rugose, which also applies to the 
interoperculum and the preoperculum; to all appearances a lim- 
ited number of branchiostegal rays; teeth mere asperities, the 
palatine arch without them; two dorsals, the first short, the 
second very extensive, its height insensibly increasing; pectoral 
short and rounded; the dorsal, anal, and pectoral fin rays do 
not ramify at their extremities. 
History : I saw this fish for the first time in Havana, on the 
5th of April, 1872, and I have not observed it since; neither No. 2.J 
GRAMMICOLEPIS BRACHIUSCULUS. 
275 
fishermen nor students of the class have been able to give me 
its name, because neither one nor the other have seen it to 
know it. It is, then, one of the rarest forms in existence. The 
skeleton I have sent to the eminent Professor Gill, who has it 
in his possession, though I do not know but that he has pre- 
ferred placing it in the collection of the Smithsonian Institution 
in Washington." 
This account is completed by a plate and its accompanying 
description, showing the fish one-third the size of nature, and 
various illustrations of its scales and other parts. 
Now about a year ago, Professor Gill did me the great honor to 
place in my hands, for a little more extended illustration, not only 
the skeleton of this rare type, but a life-size outline drawing of 
the fish made by Professor Poey himself. In addition to these 
treasures, this eminent zoologist also placed at my disposal sev- 
eral crania of fishes, representing the genus Caranx and others, 
to be used in the present connection. Situated as I now am, 
at an outpost in New Mexico, notwithstanding the great value 
of these crania for comparison, I can only regret that the material 
at my hand is not still more extensive, as it might be, were I 
more favorably situated to undertake this kind of a paper. Es- 
pecially would I like to examine specimens of Brama Raii,x 
which, if I have recalled the proper form, possesses vertical 
linear scales something like those in Grammicolepis, though, I 
believe, very much smaller. 
In order to give an idea of the external appearance of the 
subject of this article, I brought to my aid the two drawings of 
1 Since writing the above, a very valuable work upon ichthyology has appeared, 
viz.: The Fisheries and Fishery Industries of the United States. By G. Brown Goode, 
Asst. Direc. U. S. Nat. Mus. and a staff of Associates. Washington, 1884. On 
page 335 of the text of this book, we read of the Bramid/e that " The only member 
of this family of interest to us is the Brama Raii, called " Pomfret " in Bermuda, 
where a few individuals were observed by the writer in 1876. In 1880, an individual 
was obtained on the Grand Bank of Newfoundland, and more recently the species 
has been found to be somewhat abundant on the coast of Washington Territory and 
Vancouver's Island. This species was described from the coast of South America, 
under the name Brama Chilensis." In the second volume of this work, we find an 
excellent figure of Brama Raii, Plate 112, which shows the fish possesses vertical 
linear scales, although they are much shorter than they are in the subject of this 
article. 
The Pomfret also has its tail more deeply forked, and the dorsal fin is seen to be 
continuous. The eye is very much smaller, though otherwise there are some general 
external resemblances between the two forms. (R. W. S., 7 Aug., '86.) 276 
SHUFELDT. 
[Vol. II. 
Figure i. - Right lateral view of Grammicolepis brachiusculus. One-third size of nature. By the author, 
assisted by Poey's outline drawings, and existing material. 
Professor Poey, neither of which profess to be anything more 
than the merest outline of Grammicolepis, and the scales, fins, 
and other parts that accompanied the skeleton. These, taken in 
connection with the lucid description of the fish, and all care- No. 2.'] 
GRAMMICOLEPIS BRACHIUSCULUS. 
277 
fully compared, have resulted in my drawing presented in Fig. i 
of this memoir. 
Owing to the fact that many parts of the skeleton, from long 
keeping and their delicate structure, have warped considerably 
out of shape, I propose to devote myself on the present occasion 
only to such as seem most important of them, and chief among 
these stands the cranium. 
As I say, so far as I know, the specimen of the cranium of 
Grummicolepis before me is the only one in the hands of 
science, and a most extraordinary object it is. Three features 
strike us most forcibly when we first came to examine it: the 
enormous orbits, the truncate appearance of its anterior part, 
and the semi-transparency of its gelatinous-locking bones. 
Figure 2. - Left lateral aspect of the cranium of Grammicolepis brachiusculus; 
life size, drawn by the author from the specimen. Fr, frontal; Pa, palatine; S.O., 
supraoccipital; Sq, squamosal; Ep.O., epiotic; Pt.O., pterotic; E.O., exoccipital; 
Op. O., opisthotic; B. O., basioccipital; Bs, basisphenoid; Pr.O., prootic; Ptf, post- 
frontal; As, alisphenoid; Os, orbitosphenoid; Eth, ethmoid; Prf, prefrontal; Pr.S., 
parasphenoid; Vo, vomer. 
The peculiar rugose condition of a frontal bone, referred to by 
Professor Poey, is well shown in Figs. 2 and 3, Fr. It will be seen 
that these rugosities of the frontal radiate from a common cen- 
tre on its superior aspect, this centre being found at about the 
middle of the bone, or what would be the middle of its oblong 
figure were its anterior internal notch completed, and we do not 
regard its postero-lateral prolongation. This latter part of the 
bone forms the superior periphery of the orbit, and is produced 
backwards as far as the squamosal (Sq). To the inner side of 278 
SHUFELDT. 
[Vol. II. 
this process, the posterior border of the frontal shows at least 
one conspicuous notch, while its free margin overlaps the supra- 
occipital, and is in turn overlapped by the parietal (Pa) more ex- 
ternally (Fig. 2). Mesially, its surface turns upwards, more 
particularly behind, where with the fellow of the opposite side 
it grasps in the middle line the anterior portion of the supraoc- 
cipital crest. Below this point the two frontals have their 
straight, free, mesial edges roughly in contact with each other, 
and slope gradually downwards to the margin of that concavity 
which is found in front (Figs. 2 and 3). 
This extraordinary fossa on the anterior aspect of the cranium 
of Grammicolepis is entirely open above; its rugose and subcir- 
cular margin being formed by the frontals; while below it 
becomes conical with its apex in the middle line, and in the eth- 
moid. Above, where it is most capacious, it has its posterior 
wall formed by descending plates developed on the part of the 
frontals, the left one considerably overlapping the right. Below 
this, in the middle line, there is an opening of some size, which 
leads into a commodious chamber lying between the frontals 
above and the mesethmoid below. 
A frontal is truncate in front, where it overhangs the corre- 
sponding prefrontal, and internally articulates with the curiously 
shaped ethmoid. Behind this, and on its under side, it forms the 
major share of the roof of the orbit. Then occurs a longitudinal 
keel, which separates this from that other part of its under 
surface which forms the roof of the mid-chamber described in 
the last paragraph. Viewed together from above (Fig. 3), it 
will be observed that the rugosities of the frontals are limited 
behind by a subparabolic curve with its arc anteriorly directed. 
In this dried cranium a parietal (Pa) is represented by a 
thin, flake-like, semi-transparent piece of bone, of a form shown 
in Fig. 3. To the outer side of its mid-longitudinal line it de- 
velops for its entire length quite a prominent, though thin, crest, 
which is rugose all along its superior margin. 
The anterior three-fifths of the under surface of this element 
simply rests upon the frontal and supraoccipital, while the re- 
maining portion behind is more firmly attached, and really holds 
the bone in its position. Its outer free margin articulates prin- 
cipally with the inner border of the posterior prolongation of 
the corresponding frontal, though still more posteriorly it meets No. 2.] 
GRAMMICOLEPIS BRACHIUSCULUS. 
279 
to some extent the squamosal (Sq). With the epiotic (Ep.O.) 
it is connected simply by a feeble and thin band of bone. The 
parietals, then, seem to play the part here of binders, rather 
than their presence is at all essential to covering over any siza- 
ble vacuity in the cranial vault, that might exist were either of 
them removed. 
The supraoccipital is a very extensive ossification, and is char- 
acterized by a fairly prominent crest. 
Figure 3.- Cranium of Grammicolepis brachiusculus, seen from above; life size, 
by the author. Lettering as before. 
This crest is triangular in outline, with its apex above, and its 
base attached in the middle line, to the horizontal portion of 
the bone. From the middle point of this base, a narrow, fan- 
shaped development springs upon either side, which is incor- 
porated with the crest, to strengthen it, being carried nearly as 
high as its apex (Fig. 2). 
The broad and spreading horizontal portion of the bone forms 
the roof of the brain-case, while posteriorly, just before the ter- 
mination of the crest, it is bent abruptly down to meet the 
exoccipitals. The flexure, being sharply defined by a transverse 
line, the outer end of which, either way, terminates in the apex SHUFELDT. 
[Vol. II. 
280 
of a pyramid, the lateral and upper sides of each being also 
formed by the supraoccipital. The lateral aspect of this pyra- 
mid is overlapped by the epiotic (Ep.O.), while outwardly its 
free margin articulates with the squamosal (Fig. 3). 
Regarding one of these epiotics, we find that its fan-like 
portion is finished off behind by a semicircular piece, which is 
thickened below, where it becomes firmly attached to the neigh- 
boring bones. The blade portion is longitudinally fluted, but 
no rugosities are found upon it. This does not apply to the 
element at its outer side, the squamosal (Sq), which element 
develops very conspicuous rugosities upon its upper aspect in 
direct continuation with the longitudinal ones on the long, back- 
ward-extending process of the corresponding frontal (Fig. 3). 
At the distal extremity of the squamosal I detect a small, 
flake-like piece of bone, thoroughly attached, though individu- 
alized by sutural traces, which I take to be the representatives 
of the pterotic {Pt. O.). Beneath and beyond, the squamosal 
seems to make the usual ichthyic articulations with the post- 
frontal (Ptf) and the prodtic (Pr.O.). At its under side we 
find a small hyomandibular facet (Fig. 4, hf). 
A postfrontal of considerable size (Ptf) develops at its outer 
side, a sharp, descending, spicula-form process of bone, which is 
transversely pierced at its base by a small foramen. The infe- 
rior articular sutural trace of the postfrontal, as I make it out, 
is subcircular in outline, and closely meets corresponding mar- 
ginal concavities offered by the prodtic and alisphenoid. 
Each alisphenoid is necessarily a very extensive ossification, 
forming, as they conjointly do, the major part of the bony wall 
of the posterior aspect of the immense orbit (As). 
In front they articulate with far smaller orbitosphenoids, which 
in their turn meet in the median line anteriorly (Os). Now 
above the basisphenoid (Es), the alisphenoids and orbitosphe- 
noids are separated from each other, mesially, by a vertical 
vacuity, broadest below, gradually tapering to a blunt apex 
above, which constitutes a great fenestra for the anterior wall 
of the cranial casket (Fig. 4). 
As already stated, the hinder portion of the ethmoid (Eth) 
forms the mid-roof of the orbital space. This division of the 
bone is of an oblong outline, being encroached upon by the 
common, circular, anterior margin of the obitosphenoids behind, No. 2.] 
GRAMMICOLEPIS BRACHIUSCULUS. 
281 
the two elements being completely united. Below, it is convex 
from side to side, correspondingly concave above, where it 
forms the floor of the interfronto-ethmoidal chamber, already 
alluded to above. I have previously described how now the 
ethmoid is deflexed, and becomes concaved in front to form 
the lower limits of that excavation on the anterior aspect of the 
cranium. 
This, as we have already seen, terminates in a conical point, 
and even beyond this the bone is carried forwards as a median 
Figure 4.-Under side of the cranium of Grammicolepis brachiusculus: life size; 
reference letters the same as in former figures, with hf hyomandibular facet, and gp, 
foramen for the exit of the glossopharyngeal nerve in the opisthotic (Op.OP). 
triangular process, the apex of which rests upon the parasphe- 
noid (Fig. 3, Eth. Pr.S.\ 
Upon either side of the ethmoid, the flat anterior aspect of 
this cranium is completed by a broad prefrontal (Prf). The 
form assumed by one of these elements can best be appreciated 
by referring to Fig. 5, which represents the cranium of Gram- 
micolepis seen from behind, while the anterior face is in the 
horizontal plane. In this position the posterior aspects of the 
prefrontals come into view. 
Each one essentially consists of a thicker and vertical outer 282 
SHUFELDT. 
[Vol. II. 
column of bone, antero-posteriorly compressed, and an expanded 
inner portion, which latter is reenforced by radiating projections 
that converge to meet at a point at the lower part of the inner 
margin of the columnar portion. It is hardly necessary to state 
that these prefrontals form the externo-lateral parts of the ante- 
rior orbital wall, the ethmoid completing it mesially. 
Coming now to the vomer (Fh), we find it to be a thin scale- 
like bone, of a form best shown in Fig. 4. It rests in the 
longitudinal excavation of the anterior and lower side of the 
parasphenoid, while its firmest attachment to that bone seems 
to be by the periphery of its anterior margin. 
The parasphenoid ^Pr.Si) is gently arched as it spans the 
orbital space below, having its convex arc downwards. The 
lower side of this part of the bone, as I have already intimated, 
is longitudinally scooped out, while the upper side presents lat- 
eral surfaces, which are inclined so as to meet in a median line. 
Posteriorly, the parasphenoid makes the usual teleostean con- 
nections with the basioccipital, basisphenoid, and prodtic, being 
deeply cleft as it passes to cover the under side of the first- 
named element (Fig. 4). 
Occupying its usual position, the basisphenoid (Bs) not only 
develops the median process (Fig. 2) seen in so many true 
teleosts, but furnishes a firm horizontal roof for the three-sided 
pyramidal eye-muscle canal, the lateral walls of which are com- 
pleted by the prodtic and parasphenoid. 
As in the majority of osseous fishes, the prodtic is a well- 
defined and important element at the lateral aspect of the brain- 
case (Pr.O.). Its anterior margin is pierced by the foramen for 
the trigeminal nerve, from which point faint lines in the tissue 
of the bone are seen to radiate. 
The basioccipital (Bs} has its thickened and longitudinal por- 
tion underlying the brain-case, as in most fishes, being com- 
pleted behind the facet for the leading vertebra of the spinal 
column. This facet is comparatively rather small, with its con- 
ical depression very deep. 
At either of its sides the basioccipital develops an upturned 
and semicircular plate of bone, similar in structure to the other 
flat bones of the lateral cranial walls, which articulates with the 
lower margin of the opisthotic and the posterior margin of the 
prodtic (Fig. 2). No. 2.] 
GRAMM7C0LFFLS BRACHIUSCULUS. 
283 
The opisthotic (Op.O.) is large and occupies its usual position, 
as generally found, in the cranium of the teleosts. Its posterior 
margin is pierced by a conspicuous foramen for the exit of the 
glossopharyngeal nerve from the brain-case. The intersutural 
traces defining its borders are easily made out in the specimen, 
and this element contributes not a little to the lateral wall of 
the cranial cavity, - a large vacuity existing after its disarticu- 
lation. 
Each exoccipital (E.O^ develops at the outer sides of the ver- 
tically oval foramen magnum, a fan-shaped, bony thickening 
(Fig. 5), which nearly meets at the middle point above. 
Figure 5.- Posterior aspect of the cranium of Grammicolepis brachiusculus ; life 
size. Letters have the same significance as in former figures. The cranium is repre- 
sented as resting upon its anterior face on the horizontal plane, the line of sight 
being perpendicular to the latter, and passing through the imaginary centre of the 
foramen magnum. 
Further, these bones spread out so as to complete the hinder 
cranial wall, where the supraoccipital and osseous elements at 
the lateral angles have failed to do so. 
Now a number of the bones required to complete the skull of 
this fish have been lost, and, as I said before, the others in my 
possession are too much out of shape, from their fragile nature, 
for me to decide, with any degree of certainty, as to their several 
proper positions. This is much to be regretted, as I expect a 
complete skull of Grammicolepis would prove a very interesting 
and instructive object. 284 
SHUFELDT. 
[Vol. IL 
To return to the cranium for the purpose of taking a general 
glance at it: we are to note especially the almost entire absence 
of those parial and lateral crests, developed on the part of the 
parietals and squamosals, so manifest in some fishes, as for in- 
stance the genus Caranx; we are to note, also, the very peculiar 
texture of the bone that composes this cranium, being more like 
the material that is found in ordinary fish scales rather than 
bone ; particularly are we to observe the relation between the 
anterior portion of the supraoccipital crest, and the upturned 
portions of the frontals. 
There are but few striking features within the cranial case of 
this strange form of fish. For the most part, surfaces, convex- 
ities, and concavities on the outside give rise to similar surfaces 
on the inside, the last two being, of course, reversed. The fos- 
sae for the otoliths are ample and well defined, but the elements 
themselves have been lost. 
I have already expressed my regrets at not having at hand 
more extensive material wherewith I might compare this extra- 
ordinary fish; they only increased as my investigations pro- 
ceeded, while the remaining consolation left me, is, that I feel I 
have added at least my mite to the labors of Professor Poey; so 
should another specimen of Grammicolcpis fall into the hands of 
naturalists, we can, at least, meet it with drawings of its cranium 
and other skeletal parts, as well as with similar drawings of 
some of the forms to which it is supposed to be related. 
Through the courtesy of the Smithsonian Institution, and the 
kindness of Dr. Gill for selection, I find before me the cranium 
of a specimen of Caranx hippos, with the spinal column of the 
same fish (No. 13,561 S. I. Coll.). There is also the cranium 
(No. 11,385) of another and still larger Caranx, the species 
being unknown. This last specimen presents some points of 
peculiar interest not so well shown in the first. I have also the 
cranium of a specimen of Tcuthis coeruleus, which will be intro- 
duced to show certain points ; and finally, the cranium of Poma- 
canthus pane (No. 12,770 S. I. Coll.) brought forward to illus- 
trate still other points. 
Professor Poey's investigations evidently led him to believe 
that Grammicolcpis was more nearly related to the Carangidce 
than any other family of fishes known to him. And in this 
opinion, so far as I can see or am able to judge, I must concur. No. 2.] 
GRAMMICOLEPIS BRACHIUSCULUS. 
285 
The cranium of the Caranx No. 11,385, which bears a very 
close resemblance to C. hippos, shown in Fig. 6, although, be it 
known, it possesses marked differences, is composed of a bone 
tissue much more like that seen in the cranium of Grammico- 
lepis than any of the other specimens before me. As much as 
it is unlike it, it evidently approaches the semi-transparent and 
brittle condition found in our subject. The next thing that 
our attention is directed to, is the strikingly large orbit of this 
Caranx, and the evident, though distant, similarity of the ele- 
ments that go to form its walls. The chief difference we meet 
with here is the absence in the Caranx of the backward-extend- 
ing plate of the ethmoid seen in Grammicolepis, while there is 
much to support a probable relationship of the forms, in the 
parasphenoid, the basisphenoid, and less so in the prefrontals, of 
the two. 
Figure 6. - Left lateral view of cranium of Caranx hippos. Spec. No. 13,561, Col- 
lection of the Smithsonian Institution; life size, by the author. Letters signify the 
same as in the other figures. 
Again, in the Caranx, the ethmoidal mass, and parts, which of 
a consequence associate with it, are produced forwards, and we 
fail to find anything upon this aspect that in any way reminds 
us of the curiously truncate appearance of the front part of the 
cranium in Grammicolepis. 
Another marked difference is seen in the vomerine element; 
this bone, as we have observed, is in our subject merely a kite- 
shaped scale, in no way incorporated with the parasphenoid, 
being merely attached around its anterior rim. Now our sped- 286 
SHUFELDT. 
fvol. ii. 
men of Caranx, No. 11,385, foreshadows in its vomer what even- 
tually comes to pass, between this condition in Grammicolepis 
and what we find in C. hippos. In this latter fish, as will be 
observed by referring to the figures Fig. 6, et seq., the vomer 
is quite a solid bone, and is moulded upon the anterior end of 
the parasphenoid, forming a more or less massive termination of 
this end of the cranium. 
In our unspecified specimen of Caranx, this general appear- 
ance is likewise maintained; but upon a lateral view, we are 
enabled to look in between the vomer and parasphenoid, and 
the less solid formation of either can at once be appreciated as 
well as their less intimate relation to each other. It is a shame 
that this species is not known, nor was ever diagnosed when this 
specimen of cranium was taken, as this condition is very inter- 
esting in the present connection, as are several others, as we 
shall presently see. 
As representatives of the Carangidce, neither of these speci- 
mens develop a spine-like process descending from the post- 
frontal, which is a very marked feature of that bone in Grammi- 
colepis. It is, however, present in other teleosteans, as seen in 
one crania of Pomacanthus and Tetithis, Figs. 10 and 11. 
Before leaving this region of the cranium, I would like to 
invite attention to the anterior aspect of it, in this very speci- 
men of Pomacanthus (Fig. 10). It approaches to some degree 
the truncate appearance, so often alluded to in Grammicolepis; 
a closer resemblance, however, is vitiated by the extraordinary 
forward and upward projection of the vomer in Pomacanthus. 
Posterior to this bone, in the individual in question, an extraor- 
dinary concavity is seen, the sides of which are formed by the 
prefrontals and parasphenoid, being perforated on either side by 
a group of foramina. Its bottom is completed entirely by the 
latter bone. 
Teuthis coerulens offers us in its ethmoid and vomer just the 
very reverse of this condition, as may seen by a reference to 
Fig- li- 
lt may be as well to note in passing that in Pomacanthusparu 
the parasphenoid is very deep in the vertical direction, being 
longitudinally excavated above and continuous with the capa- 
cious eye-muscle canal, while anteriorly and below it is sharply 
carinated. Posterior to this carination the bone develops a No. 2.] 
GRAMMICOLEPIS BRACHIUSCULUS. 
287 
rounded and descending prominence, which is bifid, the two 
lamina being directed backwards and outwards. The behavior 
of the anterior end of the parasphenoid of Pomacanthus has 
already been described above. 
This bone is also wonderfully developed in our cranium of 
Teuthis (Fig. 11). Here the carination in front is exceedingly 
deep, while behind it, a distinct descending process is also seen. 
Figure 7. - Superior view of the cra- 
nium of Caranx hippos, same specimen 
as in Fig. 6; life size. 
Figure 8. - The same seen from be- 
low. Letters as in the former figures. 
Further, this elegantly developed element makes another curi- 
ous departure from all the specimens thus far examined. It is 
this : where it forks over the basioccipital behind, a large fora- 
men is found between the diverging limbs; this opening is in 
reality the apex of the eye-muscle canal in this fish, and conse- 
quently leads through that fossa to its continuation, which again 
is the longitudinal excavation on the upper side of the parasphe- 
noid or the floor common to the orbits. 
But to return to our cranium of the Caranx No. 11,385 : we 
find that the anterior and horizontal portions of the frontal bone 288 
SHUFELDT. 
[Vol. II. 
are quite transparent at their centres, while raised flutings radiate 
from their hinder points, forwards and outwards. The trans- 
parent areas are found to be even perforated in my specimen of 
Caranx hippos, so thin do they become. Now it will be remem- 
bered that in Grammicolepis, from the horizontal portion of either 
frontal, was developed an upturned, scroll-like projection, the 
free edges of the two bones meeting in the median line. There 
was thus formed sort of a conical prominence, the lower part 
and base of which was anterior, being terminated by the trans- 
verse rugosity in front, while the apex, or highest part, seized 
the free front margin of the supraoccipital crest. In Caranx 
hippos these vertical portions of the frontal bones are in close 
approximation, so that they appear to be the continuation for- 
wards of the supraoccipital crest; the sutural traces, however, 
have entirely disappeared; while in the cranium of our other 
Caranx, the method of formation is very evident from the fact 
that the vertical frontal plates are not thus coossified, but plainly 
show their individual origin as well as their relation and connec- 
tion with the anterior free margin of the supraoccipital crest, 
which is wedged in between them. (Compare Figs. 2 and 6, as 
well as 7 and 3.) 
Pomacanthus pane has very extensive rugosities upon its 
frontal bones, but these latter elements are exceedingly dense 
and thick, as is the anterior border of the supraoccipital crest in 
this fish, which measures at its widest part nearly five millime- 
tres across. 
Such forms as Pomacanthuspane do not develop conspicuous 
parietal and squamosal crests; they are still less manifest in 
Teuthis. On the other hand, in the Carangidce these crests 
constitute the most striking feature of the cranium. As already 
stated above, they are but feebly produced in Grammicolepis, 
though they are plainly indicated. 
All the lateral parts of the cranium, made up of the hinder 
portion of the parasphenoid, theprodtic, opisthotic, exoccipital, 
basioccipital, and below the squamosal line, are very much alike 
in Caranx and our unique subject, more particularly in our un- 
diagnosed specimen of a Caranx. But in the latter the basioc- 
cipital enters far more extensively into the formation of the eye- 
muscle canal than it does in Grammicolepis, as in the Caranx 
we find a condition existing, as regards the opening between No. 2.] 
GRAMMICOLEPIS BRACHIUSCULUS. 
289 
the hinder forks of the parasphenoid, very much the same as 
described for Teuthis coeruleus. 
Upon comparing the posterior views of the crania of Caranx 
hippos and Grammicolcpis (Figs. 5 and 9), we find, indeed, but 
few points of resemblance between them. The occipital crest 
in the former comes almost down, as it does in Pomacanthus, to 
the supero-median point of the foramen magnum. It is far 
above it in the latter fish. There the absence of the spreading 
lateral crests, seen in the Caranx, constitute a marked difference. 
Professor Poey's fish also has bony pillars developed by the ex- 
occipitals, one being on either side of the foramen magnum. 
These are absent in the Carangidce. 
Figure g.- Posterior view of the cranium of Caranx hippos, Spec. 13,561, Smithso- 
nian Institution Collection. Life size, by the author from the specimen. Letters as 
before. 
In C. hippos the facets on the exoccipitals for the first verte- 
bra of the column meet in the middle line ; these parts, how- 
ever, in G. brachiusculus have been injured, probably during the 
first dissection, so that I am unable to say positively upon this 
point in regard to them. In P. pant the first vertebra of the 
column coossifies with the basioccipital, but this condition does 
not obtain in Teuthis. In this latter form the supraoccipital 
crest also fails to reach the upper periphery of the foramen mag- 
num at its middle point. 
We find that both P. pane and Teuthis have the squamosal 
curled downwards and forwards in the most extraordinary 
manner, best marked in the latter type. This is well shown in 
lateral view in Fig. n, though I am not sure but that the piece 
there indicated by sq may not be a separate ossification, in 290 
SHUFELDT. 
[Vol. II. 
which case it would be a pterotic. I would have to dissect a 
young fish to decide this point. These processes are very 
conspicuous upon posterior view, and of course Grammicolepis 
can show nothing like them. 
Pomacanthus paru has another condition present, not seen in 
any of the other forms alluded to above. Just posterior to the 
probtic, and above the basioccipital and parasphenoid in the 
cranium of this fish, on either side, we find a subelliptical fora- 
men, with its major axis placed longitudinally, of no inconsider- 
able size, through which we can easily observe the movable 
otolith (Fig. io, otl}. 
Figure io. - Left lateral view of the cranium of Pomacanthus paru ; life size, by 
the author, from specimen 12,770 of the Smithsonian Institute. Lettering as before, 
with otl, otolith, and c.v., first vertebra of the spinal column, which is here coossified 
with the basioccipital. 
Before concluding our comparison of these crania, we must 
note another point in the cranium of Pomacanthus^ and this is, 
the ossified orbitosphenoids and the ethmoid meet in the middle 
and interorbital line, immediately beneath the frontals. 
We have already fully described above, the relation of these 
several elements in the subject of this paper, and how any such 
condition is entirely absent in Caranx. This latter form, how- 
ever, may have the ethmoid extended backwards in cartilage, 
which material may be missing in these dried preparations. No. 2/ 
GRAMMICOLEPIS BRACHIUSCULUS. 
291 
Now my material on this occasion will not admit of such a 
thing as an analysis of characters for comparison. In my opin- 
ion, without a thorough examination of the entire organization 
of not only the forms at my command at the present writing, 
but several others, such a tabulated synopsis, made up at the 
best from such fragmentary material, would be of but little ser- 
vice to us. The structure of Seriola taken in the present con- 
nection would come handsomely into play. Naucrates ductor 
would be another good form to examine. 
From a comparison of the crania alone, I should say that the 
relation between Grammicolepis and such a fish as P omacanthus 
parti was very distant, while its affinity with Teuthis cceruleus 
is still more remote. I should have liked, however, to have ex- 
amined some of the Salistidce, and perhaps glanced at one or 
two more of the Choctodonts. 
Figure 11.- Left lateral view of the cranium of a specimen of Teuthis cceruleus; 
life size. Kindly loaned the author by Professor Gill. Letters have the same signifi- 
cance as in the foregoing figures. 
Its relationship with the Carangida, as Professor Poey pre- 
dicted, is far more evident, though this, too, is extremely indi- 
rect, and many forms still unknown to us are required to 
demonstrate the connection. 
These forms must especially show an increased density in the 
cranial bones ; a decrease in the size of the eye and orbits; a 
gradual disappearance of the rugose condition of some of the 
flat bones of the cranium, particularly the frontals; a gradual 
protrusion of the snout; and finally the development of the 
parial cranial crests. 
Of the shoulder girdle. - Although a number of the bones are 292 
SHUFELDT. 
[Vol. II. 
lost, I am enabled at least to present a very good idea of the 
shoulder girdle in this fish. This is shown in Fig. 12, illus- 
trating this paper, and if the reader happen to have at hand a 
copy of my osteology of Amia calva, published in the Annual 
Report of the Commissioner of Fish and Fisheries for 1883, it 
will be well to compare it with the figure I there gave of the 
shoulder girdle of Micropterus salmoides (Pl. 14, Fig. 35). It 
represents these parts as they appear in a typical teleostean 
fish. A glance is sufficient to satisfy us that the general form 
of the proscapula {Ps} of Grammicolepis is very much like that 
element in the Bass, differing principally in being slenderer and 
more sloping, and in its relations with some of the other bones. 
I am very sorry that I have not at hand the shoulder girdle of a 
Caranx, as it would be interesting to compare it in the present 
connection. 
As is most usual in teleosteans, the hypercoracoid and hypo- 
coracoid (Fig. 12, Hyp.c. and Hyo.c?) are fused together, and in the 
present instance, to the proscapula also. The hypercoracoid 
{Hyp.c.} is pierced by the usual foramen seen in this element 
among typical teleosts. The anterior projection of the hypocor- 
acoid (Hyo.c)) is long and slender, almost reaching to the extrem- 
ity of the proscapula {Ps}. 
It will be noted in Fig. 12 that each of these elements 
develop a backward, extending process, and the letters Hyp.c. 
stand between them. This recess harbors the actinosts of the 
pectoral fin, when these parts are in situ. These pectoral fins 
have been carefully wrapped up by Professor Poey in a separate 
little package, and I find three of these actinosts attached to 
each fin. It does not appear as though any of them had been 
lost, and I am led to expect that that is the correct number in 
life. They are composed of rather elementary bone, as is so 
much of the rest of the skeleton in this curious fish. Now the 
bone marked T in Fig. 12 I take to be the teleotemporal, and 
designated by the same letter in my drawing of the shoulder 
girdle of the Black Bass. On this latter form, however, as it is 
also in Amia, the teleotemporal is very loosely attached to the 
rest of the girdle by ligament, while here in Grammicolepis, it is 
represented {T} by an exceedingly long and slender bone, which 
has its superior extremity moulded upon the side of the pro- 
scapula s, and firmly attached thereto. I fail to find in any of No. 2.] 
GRAMMICOLEPIS BRACHIUSCULUS. 
293 
the little packages put up by Professor Poey anything that 
might answer for a lower teleotemporal, T, of my Bass figure. 
We would, however, hardly expect to find such an addition, 
where the superior element proves to be so very much elongated. 
Figure 12. - Outer view of the principal elements of the left side of the shoulder 
girdle of Grammicolepis brachiusculus ; life size, from the specimen. Ps, proscapula; 
Hyp.c. hypercoracoid; Hyo.c. hypocoracoid; T, teleotemporal. The actinosts are 
attached to the vertical border of the recess in which the letters Hyp.c. are contained. 
Whatever we may see in the cranium of Grammicolepis to 
remind us of like parts in any of the Carangidce, such resem- 
blances are certainly not borne out when we come to compare 
the vertebral column of our subject with the column of a speci- 
men of Caranx hippos. Figs. 13 and 14 are presented to show 
the extraordinary departures that take place in this part of the 
skeleton. 294 
SHUFELDT. 
;vol. ii. 
Of all the bizarre structures that pertain to the organization 
of fishes which it has been my pleasure to examine, I cannot 
recall at this moment one that presents quite so supremely a 
fantastic arrangement as the eleven or twelve leading vertebrae 
in the column of Grammicolepis. These are represented in 
Fig. 13, but I have omitted to include the first vertebra, or that 
one which is found between the basioccipital and the one shown 
in the figure with the enormous neural spine. In it the neural 
spine is not developed, and its connections with the skull are 
very intimate. 
Taken in connection with Professor Poey's account of these 
parts, this figure obviates the necessity of my presenting a 
verbal description of any great length, as all the details can be 
plainly studied without any such additional assistance. 
I am inclined to think that the bony pillars, which I described 
in a previous paragraph, found on either side of the foramen 
magnum, and completely fused with the exoccipitals, are the 
halves of the neural spines of this first vertebra of the spinal 
column. To support this view, we find by placing this vertebra 
in position, that their pedicles spring from the centrum as in 
other vertebrae, and that, moreover, the sculpturing on the 
external surfaces of these pillars is precisely like that upon the 
sides of the neural spine of the second segment of the column. 
This last process is very strong, and quite firmly attached to 
its centrum : it curves gracefully first backwards, then upwards, 
in a gentle curve, as shown in Fig. 13. 
The succeeding four neurapophyses are inclined well back- 
wards, each one, as we advance in that direction, becoming 
shorter, more slender, and with a gradual disposition to assume 
the vertical attitude. This is nearly accomplished by the neural 
spine of the next segment following, or the seventh vertebra. 
Fig. 13 shows, also, the eighth, ninth, tenth, and eleventh ver- 
tebrae, and, as will be seen, the neurapophyses of these segments 
actually lean forwards. The one on the twelfth, not here 
shown, is nearly vertical again, while after that, they gradually 
incline backwards. The broken spines on the last two vertebrae 
of the figure I have restored by dotted lines. 
Now a glance at Fig. 14 is enough to convey to us that the 
arrangement of these neurapophyses are entirely different in 
Caranx hippos. In this latter drawing the first vertebra is No. 2.] 
GRAMMICOLEPIS BRA CHIUSCULUS. 
295 
shown, and it has a vertical neural spine movably articulated 
with its centrum. The succeeding spines, firmly fused with 
their centra, gradually become slenderer, longer, and more 
inclined backwards, to again become nearly vertical in the mid- 
series of the column, to incline once more as we approach the 
caudal end. 
In Gravimicolepis the ventral parial apophyses at first support 
the freely articulated ribs with their attached epipleural appen- 
Figure 13. - Left lateral view of the anterior end of the vertebral column of 
Grammicolepis brachiusculus, the first vertebra removed; life size from the specimen. 
dages, but in the seventh vertebra the capitulum of the rib 
completely fuses with the apophysis, and as the latter lengthens, 
the two become indistinguishably blended. The first and sec- 
ond vertebrae do not support ribs, and in the third pair only, 
they articulate high up on the sides of the centrum, at the base 
of the neural arch. 296 
SHUFELDT. 
[Vol. II. 
The appendages have been lost in my specimen of Caranx 
hippos, so I am unable to say anything about them. 
When we come to compare the existing differences in the 
latter halves of the spinal columns of these two fish, we find 
that they are quite as great as those shown in Figs. 13 and 14. 
Indeed, I can see there nothing to indicate that the forms in 
question have any relationship whatever and were they in the 
Figure 14. - Left lateral view of the first eleven vertebrae of the spinal column 
of Caranx hippos ; specimen 13,561, of the Smithsonian Institution collection; life 
size, from the specimen. 
class Aves, I should feel justified in placing one at one end of 
the system and one at the other. 
The absence of the hypural spine in Grammicolepis has already 
been commented upon by Professor Poey in his memoir at the 
head of this article; this apophysis is quite manifest in the 
genus Caranx, as in most true teleosteans.