mo

HARVARD UNIVERSITY

LIBRARY

OF THE

Museum of Comparative Zoology

BULLETIN

OF THE

MUSEUM OF COMPARATIVE ZOOLOGY

AT

HARVARD COLLEGE, IN CAMBRIDGE.

VOL. XVII.

CAMBRIDGE, MASS., U.S.A.

1888-1889.

Reprinted with the permission of the original publisher

KRAUS REPRINT CORPORATION

New York

1967

Printed in U.S.A.

C O N T E NTS.

Page

Xo. 1. — Studios from the Newport Marine Laboratory. —
XX. On the Development of the Cah'areous Phites of
Asterias. By J. W. Fewkes. (5 Thates.) July, 1888 . 1

X'o. 2. — Ou the Lateral Canal System of the Selachia and
Holocephala. By S. Garman. (5.3 Plates.) September,
1888 57

Xo. 3. — The Coral Reefs of the PLawaiian Tshuids. By A.
Agassiz. (13 Plates.) April. 1881) 121

X'o. 4. — Studies on tlie Primitive Axial Segmentation of the
Chick. By Julia B. Platt. (2 Plates.) July, 1889 . . 171

Xo. 5. — The Morphology of the Carotids, based on a Study
of the Blood-vessels of Chlamydoselachus anguineus Gar-
man. By n. Ayers. (1 Plate.) October, 1889 .... 191

Xo. 6. — Cave Animals from Southwestern Missouri. By S.
Garman. (2 Plates.) December, 1889 225

Xo. 1. — Studies from the Neivport Marine Zoological Laboratory.
Communicated hy Alexander Agassiz.

XX.

Oil the Beveloxunent of the Calcareous Plates of Asterias.
By J. Walter Fewkes.

1. General Observations.

2. General Changes in External Form brought about by the Growth of the

Calcareous Plates.

3. Development of Individual Plates, Eods, Pedicellariae, Spines, and Stone

Canal.

4. Comparisons with other Asteroidea.

5. Comparison of the Plates of Asterias and Amphiura.

6. Summary.

7. Explanation of the Plates.

1. General Observations.

Asterias, the most common genus of Asteroidea at Newport, in its
development passes tliroiigh a braehiolarian stage before it assumes a
stellate form. This brachiolaria is one of the most abundant larvse
found in our nets in surface fishing at certain times of the yeai'.
Although the development of the brachioharia from the e^^ of the
starfish has been accurately worked out, and the changes in the exter-
nal form of the young Asterias, after it begins to assume a stellate
form, have been well described by several naturalists, we are still igno-
rant of the mode and place of formation, and the sequence in the devel-
opment, of some of the calcareous plates which help to give the stellate
form to the young starfish after the absorption of tho brachiolaria. ^Ve
need more information as to how the ambulacral plates form, and when
they appear, as compared with the dorsoccntral and terminals. We
do not know how or when certain plates of the arm appear, and it is
desirable to study the character of certain so-called embryonic plates
reported to exist on the median line of the actinal side of the arm
in the larva.

VOL. XVTI. — XO. 1. 1

2 BULLETIX OF THE

Before we can arrive at any trustworthy conclusions as to the mor-
phulo^y of the Echiuoderms, animals as varied in external foi'm as the
Crinoids and Holothurians, it is necessary for us to Ivuow the character
of the early ditfcrcnces in the calcareous plates, and tlieir sequence and
mode of growth in the different groups. Tliese plates are the struc-
tures which, more than anv others, jrive the variety in external form to
the diftei'ent members of the Echinoderniata. It may be confidently
said that we know tlie general outlines of the growtli of the primary
plates of a representative Comatulid, Ophiuran, and Holothurian. We
know next to nothing of the early formed plates of the Echinoids,
and there is no subject which oifurs more interesting possibilities of
discovery than this. Little is known of the mode of growth of certain
of the plates of the body and arms in tliose Asterids which have a
nomadic brachiolaria.*

The following paper, therefore, is offered as a contrilmtion to the
recorded observations on the growth of the plates in the starfish.

The common species of Asterias found at Newport resembles closely
Asteracanthio)i beri/liniis of A. Agassiz, and has close alfinities with
Asterias viih/aris, tSL, and A. Forbes'd, Desoi'.f Although I suppose it
to be the same as herylhms, there are some peculiarities of coloration X
which would lead one to regard them as diti'erent. While the species
of starfishes found by me at Newport, in the adult condition, have fea-
tures of both ,1. vttlyaris and A. Forbesil, it is not possible for me to

* Our knowledge of tlie growth of tlie plates whicli form the mouth parts of
the staitish is fragineiitary and unsatisfactory.

t Tlie genus Leptasterias is thought to be sufBcicntly well separated from
Asterias by the character of its development to merit a new name, as shown by
Prof. Verrill.

t The fact that all females of both Astrracauthion palluhis, Agass., and Aatcra-
cant/lion hcrtjlinus, Agass., have a bluish tint, while the males have a reddish
color, according to A. Agassiz. indicates that there is a difference in color in the
female starfishes which we studied. The color of the females of the species of
starfish which I tried to fertilize was different from those of the species of Astera-
canthion used by A. Agassiz in tlie artificial imjjregnation of the starfish. 3Iany
specimens of female starfishes, which had ripe ova, have a chocolate-brown color,
and a bright orange madreporic body. Starfishes of this color were the only
ones which cast their eggs, although I had in the aquaria bright red and bluish
colored starfishes of all sizes. In A. Agassiz's specimens those with a bluish tint
are invariably females, while the reddish brown or reddish are males. I do not
know the color of our male Newport Asterias, but several specimens of the reddish
brown s[)ecimens laid eggs in great numbers on .«everal occasions. Ova nearly
mature were also cut out of specimens of tiiis color.

MUSEUM OF COMPARATIVE ZOOLOGY. 3

determine of which of these my brachiolariae are the young. As most
of the younger stellate forms were raised from brachiolarias captured
by surface tishmg, it is almost impossible to say definitely to which
species of Asterias they belong. I was not able to fertilize artificially
Asterias, although plenty of ripe ova were repeatedly found. The
difficulty seemed to be in all cases in procuring the males.

The following mention of their time of ovulation may be of assistance
to those who have in mind a visit to the New England coast for the
study of Echinoderm embryology.

1. The eggs of Ophiopholis were fertilized at Eastport, Maine, on
July 17th. The young of -4. squamata w^ere found at Newport, R. L, in
July, August, and September.

2. Echinarachnius can be artificially fertilized at Newport in August
and September. The probable time of ovulation is the end of August
and the first weeks of September. Plutei are abundant iu September.
A specimen of Arbacia laid eggs at Newport iu August. I have found
the majority of the plutei of Arbacia in July.

3. Large numbers of Leptasterias with attached young were taken
in Massachusetts Bay in April. Multitudes of a red pupa of some
Holothurian were collected at Provincetown in April.

4. The pupee of Synapta are found sporadic at Newport in August
and September by surface fishing. The auricularice of Synapta are
found iu July.

The material which has served for the following observations on
the starfish young was collected iu two ways. The younger forms in
some instances were raised from the brachiolari?e, collected by surface
fishing with the Miiller net. This material includes all stages from
the first appearance of the plates, or calcai'eous skeleton, up to the
young starfish with three pairs of ambulacral rafters. The remaining
specimens, from the young Asterias with three pairs of ambulacrals
into the oldest stages figured, were found on the under side of stones
near low-tide mark. The large stones near the outer landing-place at
the Laboratory were turned over, and the young starfishes were found
clinging to them. This method of collecting involves continued search,
as Asterias is not common in the immediate neighborhood of the
Laboratory.

The method by which the preparations of starfishes described in this
paper were made is as follows. The young starfishes were killed in alco-
hol (.35%). They were then rapidly passed through different grades
(50%, 70%, 90%) to absolute alcohol. They were then clarified in

4 BULLETIN OF THE

clove oil, and mounted in balsam. Those which were stained were
carried from 70% alcohol into Grenacher's alcoholic borax-carmine,
washed, afterwards placed in from 90% to 100% alcohol, then removed
to clove oil and balsam. The preparations mounted without staining
show very well the relation of the plates to each other, but it is
necessary to use a staining fluid to bring out the tissues of the organs
in the immediate vicinity of the calcareous skeleton.

In the study of the plates on the abactinal side of the disk of older
specimens, it was necessary to separate the arms from the disk proper.
No dissection was resorted to in this separation, for the arms are easily
broken from the disk along the suture between the first dorsal plate
and the second dorsal radial, leaving the former, as well as the genitals
and all intermediate plates between them, on the disk with the dorso-
central. In older stages staining fluid was used, but the best results,
as far as the plates are concerned, were obtained in specimens where no
artificial staining was resorted to.

The use of chloroform, which gave good results in Amphiura,* was
not resorted to in Asterias.

2. General Changes in External Form brought about by the
Growth of the Calcareous Plates.

By the growth of the calcifications in the growing Asterias the animal
assumes a stellate outline, passing into this form from a spherical or dis-
coid larva. These changes are almost wholly the result of change in form
or modification in the an-angement of the plates, but the peripheral
appendages, spines, pedicellarise, and spicules also play an important part
in this growth. When the growth of the primary plates begins, the
young starfish is not stellate in form, and all the early plates are con-
fined to the body. The elongation of the arms are the most prominent
results of the modification in the shape of plates, of addition to those
already existing, and of enlargement of the same. In the growth of
the arm no marked symmetry in the formation of plates on the actinal
and abactinal regions of the arm was noticed. There is also no sym-
metry observed in the gi'owth of the calcifications in the actinal and
abactinal regions of the body.

It is not in the province of this paper to give more of the develop-
ment of Asterias than is necessary to understand the relation of the

* I tried a few specimens of tlie young Amphiura with clove oil, and find that
this reagent clarifies them better than chloroform.

MUSEUM OF COMPARATIVE ZOOLOGY. 5

plates to one another, and to aid in their identifications and homologies.
A consideration of the internal organs is a most interesting and neces-
sary chapter in a study of the growth of the stellate form of the starfish,
but it is one of which little is written in the present paper. Some idea
of the origin of organs in immediate connection with the plates is neces-
sary, however, to understand the homologies of the calcareous formations
with which this paper is specially concerned.

The development of the brachiolaria of our common Asterias is well
known through the researches of A. Agassiz,* and is not here considered.
My account opens with a late stage of the brachiolaria, in which certain
calcareous nodules, described in the paper mentioned, have already ap-
peared, and in which the form of a stellate animal is obscurely marked
out. It is intended first to follow the general course of growth of these
plates collectively, and later in the paper, the development of individual
plates will be taken up one after the other.

In the starfish body, as is well known, there are two regions, called
the actinal and abactinal, the lower and upper, ventral and dorsal, which
may be studied. The primary plates in these two hemisomes differ
from the very first in number, arrangement, and distribution. No plate
is ever formed in the centre of the actinal hemisome comparable with
that in the middle of the abactinal, and it would be a task which the
author is not called upon to undertake to compare the ten ambulacrals
formed on the lower hemisome with the five terminals and five genitals
of the abactinal region of the body.

In the early condition of the plates there is an indication of the disk-
like form which the young Asterias has, but it is somewhat masked.
If we look at the lower or anal pole of the brachiolaria (PI. I. fig. 1)
laterally, and in such a way that the forming plates are on the side
turned to the observer, we can see ten small calcifications, an'anged in
two U-shaped lines, one within the other. If we so place the brachio-
laria that the anal pole is below, or pointing to the lower side of the
figure, the madreporic body on the left hand of the observer and the
anus of the brachiolaria on his right, we notice the five plates, now in

* On the Embryology of Asteracanthion heryUrais, Ag., and a Species allied to
A. rubens, M. T., Asteracanthion pallidus, Ag. Proc. Aiuei'. Acad. Arts and Sci.,
VI., 1863. Also separate, 1863.

Embryology of the Starfisli, published in December, 1864, advance Pt. I., Vol.
v., Contrib. Nat. Hist, of U. S., of L. Agrassiz. — The same, reprinted with descrip-
tions of the hard parts (calcareous skeleton) of several genera and species of
Asteroidea, under the title, " North American Starfishes," Mem. Museum Conip.
Zoology, v., No. 9, 1877.

6 BULLETIN OF THE

the form of calcareous spicules, t^-t^, of the larger U, begiuning with
one, t^, just south of the madi-eporic opening; followed by a second, ^^,
a little east of south of the first ; a third, t^, north of east of the second ;
a fourth, t\ east of north of the third ; and a fifth, t^, about due east
of the first. With these alternate the rods of the smaller U, the first,
g\ being placed about east of the madreporic opening, the second, r^-,
third, g^, fourth, g*, and fifth, g^, alternating respectively with the
1st -2d, 2d -3d, 3d - 4th, 4th - 5th, of the larger U. The members
of the larger U are the terminals ; those of the smaller U the genitals.*
Between the first genital and the fifth terminal lies a broader space than
between other consecutive plates, which is the open part of the larger
U. It is an unclosed region which forms the brachiolarian notch. As
the brachiolaria is slowly absorbed, this notch is more and more reduced
in extent, until it is almost wholly lost, when by this reduction the two
Us become rings forming the abactinal calcai'eous surface of the young
starfish.

If now we rotate the brachiolaria on its axis, through a right angle,
so that the madreporic body faces the observer, the anal pole being
still below, we have the following perspective of the two Us. It
will then be seen that the larger and the smaller Us do not lie in one
and the same plane, but that the U formed by the terminals is situated
on a greater circle than that of the genitals. This fact explains why it
is that the figure formed by the line of the latter is smaller than that
of the former. It is as if the U of the terminals was placed on the
great circle of a hemisphere, while that of the genitals follows a
smaller. The difference in size of the two letters (U) is due to the
spherical form of the walls of the stomach of the brachiolaria.

It is somewhat difficult to understand the exact relationship between
the dorsal and ventral or abactinal and actinal f surfaces of the young
starfishes, and the relation of the plates which form in these two regions.
These two surfaces are separated by the stomach of the brachiolaria,
and are not at first parallel, but form an acute angle with each other ;
and if the plane in which the plates of the abactinal hemisome were
continued to meet that of the primitive extensions of the water tubes,
they would cut each other at a small angle. A. Agassiz described them
as two "warped spirals," and if in early stages lines be drawn, connect-
ing the terminal and genital plates, the planes in which they lie will be

* The term " <Tcnital " is used to denote the same plates as " basal " by Sladen.
t " Ambuhicralcu " and " Antianibtilacralen Anhigen " of Ludwig (Entwick-
lungs-geschidite der AsUrina gibbosa, Forbes).

MUSEUM OF COMPARATIVE ZOOLOGY. 7

found not to be parallel with tlic plane in uhich a line drawn through
the tips of the radial water tubes, or the first-formed ambulacral jjlates
lie. As absorption of tlie brachiolaria goes on, however, these planes
get more and more nearly parallel, so that the two surfaces equidistant
from the axis are equidistant bom each other at all points.

Each of the five small cuU-de-sac, rir, from the water tube on the
ambulacral side of the young starfish forms a radial water tuije of
the starfish; and if a line be drawn from the tip through its middle to
the centre of the mouth, it miglit be thought to indicate the line of the
ray. In the same way, if a line be drawn from each of the notches
in the margin of the young starfish (I'l. 1. fig. 3) through the centre
of tiie mouth, it might be thought to indicate an interradius of the ab-
actinal side. The radius and interradius thus formed have in the adult
a definite relationship to each other. They do not coincide, as tliey
indicate entirely different regions of the voung starfish. If such lines
be projected in Plate I. fig. 3, it will be seen that there is a very great
variation in their relative distances from each other. This difference is
in part due to the oblii^uity of the two planes of actinal and abactinal
regions of the starfish.

The first addition, on the abactinal side, to the ten plates which form
the two Us of the early stages, is a small calcareous nodule, situated
within the smaller U near the fifth genital (I'l. I. fig. '!). Tliis nodule
is the beginning of the dorsoccntral, dc, and in the subsequent growth
of the fifth terminal towards the first genital by the absorption of the
brachiolaria, and the consequent reduction in size of the brachiolarian
notch, it is brought to occupy the centre of the abactinal region of the
starfish. The anus of the brachiolaria, which is the blastopore of the
gastrula, is situated quite a di.-^tnnce from this plate, and not near it as
recorded in Astcri^ua.

The growth of the fifth termmal, t^, towards the first genital, g^, which
from its vicinity to the madreporic opening is called the madreporic
body, is brought about by an absorption of the brachiolaria, and the
reduction in width of the notch as stated above. Before the complete
closure of the brachiolarian notch takes place, however, the terminals
have grown so large that marginal notches corresponding to interradii
have formed between them (PL I. fig. 2) on the rim of the body, and
an ajiproach to the stellate form begins to be visible. The increasing
growth of the rods of the forming star adds so much weight to the
brachiolaria that it sinks to the bottom of the aquarium in which the
animal is confined. The eleven plates of the abactinal region of the

8 BULLETIN OF THE

starfish antedate all plates on the actinal surface. With the formation
of the eleven plates of the abactinal hemisome, how fares it witli the
actinal ? "What plates have been added tu this portion of the body 1

If we so place the brachiolaria that the side opposite that already
described (abactinal) is made to face the observer, it will be seen that
the future circular water tube on this side (actinal) has the form of an
elongated U-shaped tube, with five blind extensions (culs-de-sac, fig. 3,
rw). If we focus the microscope down to the plates below, which lie on
the actinal surface, on the opposite side of the stomach, it will be seen
that each of these blind extensions of the tube corresponds roughly with
a radius of the starfish. Like the disk, the circular tube has likewise
a brachiolarian notch, but it is unclosed as yet. This tube and its
extensions are the water-vascular tubes of the future starfish ; the un-
closed U-shaped portion, the circular tube ; tlio prolongations, or culs-
de-sac, are the beginnings of the median actinal water vessel * of the
starfish rays.

One remarkable fact in regard to the relation of the system of vessels
to the abactinal plates may be mentioned. The middle of the region
on the periphery of which the circular tube of the actinal vessels lies,
or the mouth, does not coincide with and is not opposite to the dorso-
central, or the central plate of the abactinal hemisome, but is more
nearly opposite the point of origin from the circular vessel of the madre-
poric tube. This dislocation has been brought about by the fact that
the fifth terminal has grown to be much larger than the first terminal.
jNIoreover, the indentation which separates the first terminal from the
second, is in early stages quite obscure, for some unknown reason. The
same is true of the circular tube which near the fifth terminal, t^, is
well formed, while near the first terminal, t^, it is as yet unclosed. The
first radial median projection is half formed. This condition is due in
part to the fiict that the plane of the water ring is not yet quite parallel
wnth that of the terminals, as explained above, on the abactinal hemi-
some. The unequal development would seem to indicate that the fifth
terminal is the first to form, and that the first terminal is the latest

* The primary form of the circular bloodvessel of Asterias can be seen ia
Plate I. fig. 3, cbr. It occupies the same position as the homologous structure in
Asterina, and skirts the walls of the opening of the cesophagus into the stomach
anil the primitive water vessel. Extensions or radial vessels extending from this
into the arms were not noticed, and the central tube, ncr, is as ytt not closed in
or united at the brachiolarian notch. The early-formed nervous rinsr lies on this
region of the starfish, on the bloodvessel; but its tissue is with difficulty distin-
guished from that of the bloodvessel upon which it lies.

MUSEUM OF COMFAItATIVE ZOOLOGY. 9

formed terminal. It would seem to indicate that the madreporic open-
ing was a fixed point of departure, as far as the age of plates is con-
cerned. Unfortunately for this conclusion, it is only conjectural that
such is the case, and the terminals seem to arise simultaneously. On
the other hand, my observations on Asterias do not agi-ee with Lud-
wig's on Asterina, that the madrepoi'ic plate, or the genital near the
madreporic opening, is larger than the others, or has a predominance in
size in early stages.*

On each side of the radial water tubes, near the circular tube, in their
early condition, are formed tlie ambulacral oral plates, am, which are
placed at first in parallel pairs and are ten in number. The pro-
gi-ess of the growth of these plates and the addition of new ambu-
lacral rafters will be treated of in a special account of the gTowth of
these plates. It is to be noticed that the oral plates are in parallel
pairs at first, like the spoon-shaped plates, and other ambulacrals of
Amphiura.

Prominent among the calcareous formations which mark this stage
(PI. I. fig 3) in the development of the starfish are the abactinal
spines, s/9, which appear first on the terminals, and later on the genitals.
Their early appearance introduces a morphological question in the dis-
cussion of the homologies of the Echinoderms.

In a disk-shaped starfish (PL II. fig. 1) which follows hard on the
last, the brachiolaria has been almost wholly absorbed, and the brachio-
larian notch is very much diminished in width. The relationship of
the madreporic tube and the madreporic opening to the genital plate,
g^, is indicated by an arrangement of the branches of the calcareous
spicules on the outer border. There is no true madreporic body or
superficial calcification until after the stone canal has begun to form,

* In this way I interpret what Ludwig says (pp. 49, 50), in speakinsr of the
eleven first-formed abactinal plates : " 1st doch eines, welclies den iibrigen zehn
fast iniraer etwas voraus ist in Bezug auf die Zeit seines ersten Auftretens. . . . Aus
diesem Skelettstiick wird die Madreporenplatte des Seesternes."

In one or two specimens the predominance in size of the interradial [pace. Car-
penter) plate near the madreporic opening was thought to exist, but in others no
such prominence was noticed. No preparation which was made shows the " mad-
reporic plate " without one or two other genitals, and I have none with this plate
without terminals.

As Ludwig has shown in Asterina, all the terminals and other genitals do not
appear always simultaneously in Asterias. The dorsocentral is generally' belated,
and one or two genitals may be developed after the others. The same may also
be true of the terminals. (Ludwig, Entwicklungs-geschichte ilev Asterina ^ibbosa,
Forbes, Zeit. f. Wiss. Zool , Vol. XXXVII., 1882.)

10 BULLETIN OF THE

but a notch where it is latei* situated appears in a genital long before
tiie calcification of the stone canal has appeared. The niadrepoiic
opening, as shown by Ludwig's figures of Asterina, lies to tlie left of
tlie first genital.

The stellate form of the starfish is brought about by the growth of
the plates of the arms, by which the terminals or first jdates to ap|)ear
in the body are pushed out to the extremity of the rays. Among the
most important of these plates, on the aljactinal side of the body, are
the dorsals, d, and the marginals or laterals, m. On the actinal side,
the ambulacrals, am, and the interambulacrals, ad, accomplish a similar
result. All of these plates follow in their growth a general law, viz.
that the new plates are formed between those which have already
appeared and the terminals. A somewhat similar law holds in regard
to the formation of the plates on the disk between the dorsocentral and
the first radial or median dorsal. In tlie case of the latter plates (dor-
sals, (/), however, those nearest the dorsocentral are the last to form.
The first radial, d, or first dorsal, is therefore a point of departure, on
either side of which calcifications appear. On the arms the first formed
jDlates are nearest the first dorsal. In the plates formed in Amphiura,
the primary radials, or j-adialia, have a similar relationship.

In the general development of the water-vascular system, it may be
noticed that the extensions, /, from the medial water tube, which form
the am})ullfe, are formed before the ambuhicral rafters which ultimately
separate them. Tlie teet, /, are at first destitute of suckers, and are
arranged in two rows (PI. III. fig. 3) in each arm, one row on each side
of the middle line. A terminal tentacle (PI. III. fig. 4, tta) was ob-
served in a young starfish, in which tliere are but two pairs of lateral
feet, and no ambidacral rafters. Even in this early stage the eye spot
is well developed on the terminal tentacle, tta.

The madreporic tube forms a consi)icuous object in early starfish
larvfe, and passes by successive change into tlie madreporic canal of the
older specimens. The calcification of the canal was observed in early
specimens before the formation of the madreporic sieve, which is found
superficially on the abactinal surface of the adult. The cribriform mad-
reporic plate is of comparatively late formation in the growth of the
starfish. None of the imitortant primary plates of the young starfish
form by a constriction from otliers previously formed, but each kind of
plate originates from its own calcification. Plates which have originated
from two centres rarely consolidate, although their connection may be
of a very intimate nature.

MUSEUM OF COMPARATIVE ZOOLOGY. 11

In all the stages which have been studied no embryonic plate or
plates were found which are at one time well formed and later lost by
absorption. I have looked for these plates especially along the me-
dian actinal line of the arm, but was not able to discover tliem. The
large size of the spines would seem to stamp them as embryonic organs.
Certainly the size of the spine on the dorsocentral or on the terminals
is relatively much larger than in the adult and older stages. This pre-
dominance in magnitude of these appendages does not necessarily mean
that they are later lost, but it simply indicates a relationship of Asterias
in its youth to spiniferous forms. The later grosvth, when new spines
form, simply brings the spine to lose its relatively great size on account
of a corresponding growth of the plate and other spines.

3. Development of Plates, Rods, Pedicellariae, Spines, and

Stone Canal.

The following calcifications are considered in this place : —

Body.

Ahactinal hemisome.

1. Dorsocentral (Centrale, Luclwig), dc.

2. Genitals (Basals, Sladen, Interradialla, Ludwig), rp- -g^.

3. Plates on the radial line, dd^.

4. Abaxial interradials, gg, gg^.
6. Connectives,* c.

Actinal henmome.

1. Ambulacral orals, t am.

2. Interambulacrals J (Adambulacrals, aurt.), ad-ad^.

3. (Odontophores §) ; First Interbracliials, auct.

4. Interbracliials, auct.

Arms.
Ahactinal.

1. Terminals, t'^-i^.

2. Dorsals il (Intermediare, Lndwiq), d-d'^.

3. Marginals (Interambulacrals, Ludwig), m-m7.

4. Dorsolateral, dt.

5. Connectives, c.

* By the connectives the smaller plates connecting the larger are meant.

t I should prefer the term Oral, if the name had not been applied to certain
other plates.

J It may be as well to retain the old term, especially as they arise between
ends of successive ambulacrals.

§ An unfortunate designation. I prefer Interbracliinl. First bomologized by
Ludwig with oral, later {op. cit.) with unpaired marginal (Unparige Interanit)ulacral).

II The first of the dorsals is bomologized with radial of crinoid by Sladen.

12 BULLETIN OF THE

Actinal.

1. Anibulacrals, Ambulacral rafters.

2. Interambulacrals (Adambulacrals of late authors).

In addition, the following calcifications, appendages to the above
plates, are considered : —

1. Spines.

2. Pedicellariae.

3. Stone canaL

The last is an internal calcification, which is morphologically distinct
from the above, and arises in the walls of the water tube.

It is thought that almost all the larger calcifications of the mature
Asterias can be referred to some of the above-mentioned structures.
There are, however, spicule-like calcifications, as in the legs, which are
not considered in this discussion.

The plates first appear as a small calcareous formation in the midst
of cells, which color with reagents more deeply than those of the remain-
ing parts of the body in the immediate vicinity. A common form fol-
lowing the simplest is the extension of arms — commonly three —
which impart to the spicule a trifid shape. The extremity of each
branch subdivides, and continues division, anastomosing and joining
with other bifurcations from other branches. The thickness of the cal-
cification is at first small, being simply that of a spicule-like rod. By
the growth and anastomosis of the spicules the plate later assumes the
form of an open network. This network, in large plates like the termi-
nals and genitals, is open. In other plates, as the median dorsal, the
calcification has from the first the form of a disk in which are small
perforations.

The interambulacral plates are never thin, with loose open work, but
become compact at the very beginning of their development. It is
probable that the open character of the early formed terminals is cor-
related with the fact that these plates are formed in the brachiolaria.
It is questionable whether such slight rods would be strong enough to
preserve the shape of the stellate animal if they occupied the position of
ambulacrals, laterals, or dorsals, and were formed at the same time.
These latter plates are more compact from the beginning, for obvious
reasons. When the brachiolaria is absorbed, and the terminal oomes to
be pushed out in an exposed position, its form and compact calcification
is such as successfully to resist any injury which its exposure might bring.
It not only is strong enough for its own safety, but it serves as an effect-
ual shield for the newly formed ambulacral, interambulacral, dorsal, and

MUSEUM OF COMPAEATIVE ZOOLOGY. 13

other plates. It is possible, in this connection, that the large spines of
the terminals may also serve as structures for protection of the tip of
the ray. Animals small enough to take the young starfish for food
might well hesitate before eating an animal bristling with the sharp
needles of the rays of the young Asterias. While such an explanation
may be probable, it also seems not unreasonable that these enormously
large spines point to features of the ancestors of the starfish, and have a
morphological significance.

It is likewise to be noted, that in many genera of Asteroidea the
primarily formed spines are relatively of great size. This is also true
of many Ophiuran genera, as Ophiothrix and Ophiocoma. Among
Echinoids, Echinarachnius and several others have large primary-
spines, and in the former genus, as I have already pointed out, these
spines appear very early in the embryonic history. Their existence in
these groups would seem to indicate a morphological meaning.

It seems to me a significant fact, however, that while in Ophiothi'ix
the spines of the terminals bear the form of large hooks, as I have
noticed in an Ophiothrix from Santa Barbara, Amphiura in the pouch of
its mother is destitute of terminal primary spines. The spines of the
larval Asterias are larger than those of the larval Asterina.

The general character of the above plates are as follows : —

Flat, discoid. — The interradials, connectives, dorsocentral, dorsals,
and dorsolaterals are flat, discoid, or cylindrical in shape.

Massive, quadrangidar. — The interambulacrals are quadrangular and
massive.

Curved, crescentic. — The marginals and genitals are crescentic or
curved, or more or less bent out of a plane surface.

Elongated in the plane of first calcification. — The ambulacra! rafters
are elongated in the plane of the first calcification, and have the form
of bars or beams, rather than flat plates.

Elongated at right angles to pAane of first calcification, — The spines
are elongated in the plane opposite that in which the first calcification
occurs.

Cap-shaped. — The terminals are cap-shaped.

Tubular. — The calcification of the stone-canal is tubular and mul-
tiple, or originating from several centres.

Double calcification in same organ. — The calcification of the pedicel-
laria; is double from the first.

The above classification is not intended as a hard and fast division,
but only as a means of roughly separating the plates from each other.

14 BULLETIN OF THE

The different divisions grade into each other, and in early conditions
are not distinguishable.

Dorsocentral. — The dorsocentral plate [dc) is one of the earliest to
form, and is one of the least modified in its growth, of all the abactinal
plates of the body. It is believed to be homologous with the dorso-
central of Amphiura. Especial attention was paid to the time when
the dorsocentral forms in Asterias, and the development or stage of
growth of the terminals and basals when it first appears. This is
believed to be an important fact in comparisons both with Ophiurans
and with Echinoids.* The younger the larva is, the greater is the
distance of this plate from the plutean anus, or blastopore.!

In the youngest starfish (PL I. fig. 2) in which the dorsocentral was
observed, there were five terminals, t^-t'', and five genitals, ff^-g^.
On the periphery of each of the terminals there were two trifid spi-
cules, sp, which later developed into the terminal spines. The stel-
late form, or the position of the inten-adii, was mapped out by slight
indentations in the rim of the disk between the terminals. The termi-
nals have the form of simph* bifurcated and Y-shaped calcareous rods.
The position of origin of the dorsocentral is in a small space of the
dorsal or abactinal region enclosed by the second, third, fourth, and
fifth genitals. It is hemmed in by these plates except at one place,
the brachiolarian notch, which is an unclosed interval separating in
the present stage the first genital from the fifth terminal.

The dorsocentral originates as a simple calcareous rod, or nodule,

* I have already elsewhere (Bull. '^Ins. Comp. Zoul., XIIL, No. 4, pp. 122, 123)
devoted some space to a discussion of the time of appearance of this plate.

t The dorsocentral of Asterias is never as near the blastopore as in Asterina.
This fact is mentioned as its neigliborhood in the latter has been used by Carpenter
in comparisons of the centrodorsal and dorsocentral. Facts in Asterias do not sup-
port the supposition, that, when the dorsocentral first appears in this genus, it is closer
to the blastopore than in older stages, as Carpenter says is the case in Asterina.
Tliis does not deny that it may not be nearer in Asterina, but it is not in Asterias.
In Asterias no anus is formed, but the blastopore, which never opens among the
limestone plates, is simply closed after absorption of the brachiohiria. I am in-
clined to the opinion that tlie blastopore does not become the permanent anus,
but have made no observation on this point except that in my species of Asterias
the blastopore (brachiolarian anus) is simply closed, and never migrates around
the rim of the young starfish to the abactinal side. A. Agassiz, however, con-
siders that there is a close approximation of the calcareous plates and the anus,
and says {op. at., p. 46) that the " anus undoubtedly discharges at tins time through
one of the many limestone cells." But later he says, " I am not able to state this
positively, never having seen from any point discharges of fecal matter."

MUSEUM OF COMPARATIVE ZOOLOGY. 15

which immediately grows into a trifid spicule. As the starfish matures
the dorsocentral becomes branched (PI. I. fig. 3), forming a flat pen-
tagonal plate occupying the central region of the abactinal part of the
body of the starfish.

As the starfish grows older, the dorsocentral is fomid to carry a large,
well marked, centrally placed spine (PI. II. fig. i) ; later, two and three
other spines of large size form. These spines are relatively to the size
of the starfish much larger than those of the adult.

In the oldest larval starfish (PL IV. fig. 4) which was studied, the
dorsocentral still preserves its pentagonal form, and, on a line passing
through the dorsal region of the starfish arm, from its angles arise the
radials of the disk. The dorsocentral always preserves its central posi-
tion even into the adult starfish, and never undergoes any considerable
modification in outline or in size, although of course its size relative to
that of the starfish is smaller as the starfish matures.

The time of development of the dorsocentral in Amphiura is offer the
primary radials and basals.* As there are at first no plates which can
be compared with the radials in Asterias, we can simply say that the
dorsocentral is formed in the starfish after the basals ; but if we com-
pare the first median dorsal arm-plate of the starfish with the radial of
Amphiura, we must say that the dorsocentral originates before the first
radial in the starfish. This seems to me an additional argument,
although I confess not a strong one, against considering the first median
dorsal arm-plate as homologous with the radialia, or primary radial, of
Amphiura. Still it would seem that the relative time when plates
appear in the Echinoderms is unimpol'taut, as far as a determination of
their morphology is concerned. My observations on the time the dorso-
central plate appears, as compared with the ten primary plates of Aste-
rias, support the statements of A. Agassiz on this point.

Terminals. — The terminals, t^-t^, are the most conspicuous and
largest of all the primary plates in the embryonic life of the starfish.
While the starfish is yet in tho brachiolarian stage (PI. I. fig. 1) tlie
terminals appear, and in the oldest form considered they are still promi-
nent. From first to last, then, these plates are important calcifications
in the growing Asterias and in the modifications of its form. There are
five terminals. These will be designated in the following way. Begin-
ning with the madreporic opening, t^, and ending near the same body,
t^, passing around the anal pole of the brachiolaria from dorsal to
ventral side, they extend through a complete semicircle. The termi-
* Bull. Mus. Comp. Zool., XIII,, No. 4, p. 121.

16 BULLETIN OF THE

nals seem to arise almost simultaneously, although brachiolarise have
been found with two, three, or four terminals. The five terminals are,
however, all believed to be formed before the genitals appear. The ring
of terminals, beginning with t^ near the madreporic body, are placed
in their U-shaped figure at about equal distances apart, with the excep-
tion of t^ and t^. The space between these two last is the whole diam-
eter of the stomach of the brachiolaria. This space, which in older
stages appears as a notch separating the fii'st genital, g^, from the fifth
terminal, t^, is the brachiolarian notch. It is the notch which marks
the position of the madreporic body, and renders it a point of departure
in all morphological comparisons of the different groups of Echinoderms.
The notch and madreporic opening are separated by the first genital, g\

The stage of the young starfish directly following the one in which
the dorsocentral is first seen shows a condition in which the terminals
have elongated and extended Y-shaped appendages at their extremities.
These extensions have formed a rod perpendicular to the radius connect-
ing the dorsocentral with the margin of the formiug starfish. The law
of the first growth of the terminal seems to be that they elongate, form-
ing an extension across perpendicular to the radial lines, not parallel with
them. The spines of the terminals appear directly after the dorsocentral,
while yet the terminals are simple spicular bodies. Notches now begin
to deepen in the interradii of the forming starfish on the border of the
disk, separating the terminals. The terminals never coalesce with their
neighbors. Although the terminals originate in the body of the starfish
while yet a swimming brachiolaria, and form the most conspicuous
plates in the Asterias before the stellate form is marked out, they are
ultimately pushed to the extremities of the rays by the growth of the
plates of the arms. The first appearance of the stellate form of the
young Asterias results from the enlargement of these plates. The ter-
minals originate on the abactinal side of the body, and grow down on
the sides of the water-vascular portion of the extensions from this sys-
tem. By this growth they eiiclose the tube above and on two sides,
and come to have a cap-shape, an opening being left on the terminal
border for the passage of the tentacle. There is no growth downward
at the tip of the radii in which they lie, but a groove is left at that
point through which later the extremity of the medial vessel extends,
and in which point the eye-spot is situated.

The cap-shaped form of the terminals affords ample protection for the
immature ambulacrals, am, intei'ambulaci'als, ad, and marginals, m,
which first form under cover of the sides and dorsal portions of the

MUSEUM OF COMPAEATIVE ZOOLOGY. 17

terminals. This protection to the delicate forming calcareous plates is
afforded in later stages in the growth of the starfish arm, even in speci-
mens an inch or more in diameter. As the starfish grows older, the
terminals lose their prominence. The large spines and pedicellariae
which first form on the terminals are specially treated of elsewhere. It
may here be said that the spines of the terminals are the first spinfes to
form in the Asterid body. The same sequence and predominance are
true likewise of the pedicellarise.

The general form of a terminal of an older starfish (PI. III. fig. 3) is
such that it completely covers the tip and a part of the dorsal region of
the end of the arm. That part which covers the dorsal tip of the arm
is thinner than that upon the sides, and the groove through which the
end of the i-adial tube, or the tentacle, passes, is well marked.

Medial Dorsals of the Arms. — A row of plates along the crest of the
abaxial region of the arm of the starfish may be called dorsals or medial
dorsals. The median dorsal row of plates does not begin until after the
formation of three pairs of ambulacrals, and likewise subsequent to the
odontophores, mouth plates, genitals, terminals, and dorsocentral. The
starfish has begun to have a pentagonal or stellate form before the first
of this series develops. The first (cZ) of these plates to form appears in
the medial dorsal line of the radius, in the triangular space between two
genitals and the adoral edges of the terminals (PI. IV. fig. J). It ante-
dates the adambulacrals and the laterals. The new dorsals (PI. IV.
figs. 3, 4) form distally to those which have already appeared.

The question of what plates in Amphiura the first of the dorsal plates
corresponds to will be spoken of later. There are no radials formed be-
fore the terminals inside the ring of genitals ; but in other ways, as far
as position goes, the oldest of the series of median dorsal plates of the
arm corresponds with the first radial of Amphiura. When the arm of
the young starfish is broken from its disk, the line of fracture commonly
leaves the first dorsal with the arm, not with the disk. The median row
(PI. IV. fig. 4) of dorsal plates form in a continuous series on the middle
line of the dorsal (abactinal) region of the arm. The newest formed
plates, cP, (T, are those outside the plates already formed. They begin
as a simple branched calcareous spicule, and broaden into a flat plate.
Each median plate bears at first a single spine. The second median
dorsal plate forms after the first pair of marginals and the first pair of
interambulacrals. The oldest radial has a quadrate form ; the others,
when well developed, are triangular, with re-entrant angles, by which
are developed lateral rings and a median adaxial extension.

VOL. XVII. — NO. 1. 2

18 BULLETIN OF THE

The median row of plates is well marked, even into stages of the star-
fish of some size, and each plate bears at first a single spine.*

Lateral Dorsal Plates. — • The larger members of the network of plates
which connect the median dorsals with the marginals may be known as
the lateral dorsal plates. They were first detected in a young starfish
in which were four median dorsal plates ; and the first pair to ap-
pear is situated on one side of the third median dorsal. In a specimen
older than the last these plates were found on the second, third, and
fourth median dorsal plates. It will thus be seen that the first lateral
dorsal to form is not the pair which belongs to the first, but to the third,
median dorsal.

The lateral dorsals (dl) are semicircular or circular plates, with their
longer axes at right angles to the line of the radius. They are in young
stages destitute of spines. There seems to be little regularity in the
formation of additional lateral dorsals, and in older conditions they form
a dsedalus of plates very difficult to trace. They are intimately connected
by smaller calcifications, which will be spoken of as the connectives.

Genitals.f — The genitals {g^—g^) are the first interradial plates to form.
These plates are among the earliest plates of the starfish brachiolaria,
and in early stages in the growth of the body they are very conspicu-
ous. They probably originate after the terminals, and appear at first as
small calcareous nodules, alternating with the terminals. All of the
genitals, in young stages, are smaller than the terminals. The fact that
thev originate after the terminals is not an imimportant one. as it shows
that in this particular the starfish resembles Amphiura. Moreover, it
has been stated that one (^g^) of the genitals — namely, that near the
madreporic body — arises before the terminals. I find this statement,
as well as another that the size of the " basal " (genital) near the madre-
poric opening is larger than the remainder, and preserves its preponder-
ance in size in all younger stages of the growth of the starfish, not to
hold in the specimens of Asterias which were examined.

The genitals when first formed are simple nodules, which later form
branched spicules, as shown in Plate I. fig. 1. They lie in the interval
of the interradii between the terminals, while their centre of calcification
always begins in an interradius.

* The median row of dorsals and their spines correspond witli the " median
line of spines supported by a long narrow limestone plate extending from the basal
plate almost to the terminal radial," mentioned by A. Agassiz (p. 51, op. cit.).

t The same plates as those called in several late writings the basals, from
their supposed homology with the basals of Crinoids.

MUSEUM OF COMPAEATIVE ZOOLOGY. 19

The genitals never leave, or are pushed from, the body of the star-
fish, but as the complexity of their reticulation increases they fill al-
most the whole space of the interradius between the radially situated
terminals.

The genital which occupies the interradius in which the brachiolarian
notch lies, differs from the others in possessing an indentation on one
side which is perforated by the madreporic opening. This failure of cal-
cification is brought about by the growth of spicular extensions from
the edge of the genital which lies contiguous to the madreporic open-
ing, but is marked in no other conspicuous manner.

It will be seen, on a comparison of my description of the way the ac-
tinal plates of the starfish form with that given by A. Agassiz,* that
there is a difference in our accounts of the growth of these structures.
It would seem doubtful that so great a difference could be the result
of our studying different species or genera. According to Agassiz,*
the large clusters of calcareous deposits which I suppose to be the ter-
minals " unite along the edge of the rays, forming a continuous net-
work," and they are figured with such a union in Plate VI. fig. 10.*
It does not appear that he considers those plates which lie in the angle
of the rays as joining with the ray plates, or terminals, although re-
maining distinct from each other ; for, later, he says that the limestone
deposits in the angles of the rays do not unite laterally.

It would seem exceptional to suppose that the terminals do join or
unite at their edges, and that the first interradial plates, or genitals,
unite with them. Such a consolidation would prevent, for a time at
least, any subsequent growth of the arms, unless we suppose a resorp-
tion to take place. The plates simply interdigitate with each other in
Asterias, and there is no union, temporary or permanent, between termi-
nals and interradially situated plates or genitals. It is extremely diffi-
cult to distinguish the boundaries of the terminals and genitals in live
specimens, and it was only by the use of alcohol and some clarifying re-
agent that I w'as able to make cut the separation of the two.

Interamhulacral Plates. — The interarabulaci'al plates, ad, originate
after the corresponding ambulacral rafters, as separate calcifications
between the lateral t ends of successive ambulacrals. In their early

* "North American Starfishes," pp. 46, 48. In my references to A. Agassiz's
observations on the embryology of the starfish I have quoted from this paper
(Mus. Comp. Zool., Vol. VI.). This reprint contains valuable references to the
work of other observers made since the paper was first published.

t Most distantly removed from the median radius.

20 BULLETIN OF THE

condition they are protected by the cap formed by the sides and dorsal
region of the terminals, and, as the terminals are pushed out by growth
of dorsals and marginals, new interambulacrals continually form under
the shelter of the terminal. Tlie law of development of ambulacrals
and marginals, in the arm holds in the interambulacrals. The oldest
formed are those nearest the mouth ; the youngest are the nearest to
the terminals. The beginnings of at least three pairs of ambulacrals
are formed before the first true interambulacral appears. The marginals
antedate the interambulacrals.

Ther^ is little variety in the progress of the growth of the interam-
bulacrals, from the time they first appear as small calcareous nodules
(PI. III. fig. 2) until they form the compact blocks of older stages in the
growth. They differ from most of the other plates of the starfish in their
massive growth, and they never have the flat perforated plate form of the
dorsals or genitals. As they mature, they fit closely together, forming
square bloclis with re-entering angles, and are closely articulated. They
also carry a single long spine in early stages.

In younger forms of the starfish, before the ambulacrals have be-
come so crowded that there are four rows of feet, the interambulacrals
alternate with the ambulacrals. At that time the number of interam-
bulacrals is the same as that of the ambulacrals, with the exception,
however, of the newly forming ambulacrals at the extremity of the
ray. In all young starfishes the single row of interambulacrals stands
out clear and distinct from the other plates, while their number always
has a constant relation to that of the ambulacral rafters.

Marginal Plates. — Large and important plates of the arms, origi-
nating early in the development of the larva, may be known as the
marginal plates, m. These plates are formed at the extreme end of
the ambulacral rafters, between the adoral rim of the lateral extensions
of the terminal plates and the interbrachial region of the body. They
follow the same law in sequence of formation as the adambulacral, but
do not have the protection of the terminals in their early condition.

The first marginals to form appear in a stage between one with a
single median dorsal and one with two median dorsals (PI. IV. figs. 1, 2).
It is a curved plate, extending on the actinal side to the interbrachial
region and on the abactinal to the vicinity of the first median dorsal, d.
On its dorsal region it bears a single spine (PI. IV. fig. 2).

The first pair of marginals is firmly jammed in between the lateral
extensions of the terminals and the interradial portion of the circumoral
plates, and by its subsequent growth helps to push out the terminals in

MUSEUM OF COMPARATIVE ZOOLOGY. 21

the increase in length of the arms, A second pair of marginals is not
formed until after the second pair of iuterambulacrals has appeared.

The forming marginals bridge the intervals between adjacent inter-
ambulacrals. They are larger than the interambulacrals, but not so
massive, and form curved plates making the curve in the margin of the
arms. Their number is not so constant as compared with that of tlic
arabulacrals as with the adambulacrals, and in this respect even in very
young stages they recall the marginal plates of the adult starfish.

Oral Ambidacral Plates. — The plates, or calcareous framework which
surrounds the mouth of Asterias, date back to very early stages in the
growth of the starfish (PI. I. fig. 3). Rudiments of these structures
appear while yet the starfish has a disk-like form, and before the com-
plete absorption of the arms of the brachiolaria.

In the earliest condition in which these plates were seen there were
but eleven other plates in the starfish body, and these were all found
on the abactinal surface. These eleven plates are, of course, the single
dorsocentral, the five terminals, and the fi-e genitals.

The oral ambulacral plates appear on each side of the primitive ex-
tensions from the right water vessel, rw, which later form the five radial
water-vascular tubes of the arms. They appear in pairs and are tea in
number, a pair to each pair of legs.

In their earliest stages they are spiculate and elongated, their length
running parallel with the walls of the water tube (PI. I. fig. 3). This
fact is an important one, for it recalls the condition which we have in
the ossicles, or ambulacral plates, of Amphiura. I shall speak of this
condition later. The elongated rods or spicules have later (PL II.
fig. 1) small lateral branches, and a beginning of a network is to be
seen. The two members of the pair never grow together laterally in
the position that they are at first placed.

The condition of the watei tube, when the first pairs of circumoral
rods form, is briefly as follows. The tube has not joined about the
mouth, as the brachiolaria is not yet fully absorbed. The five median
water-vascular tubes, rw, are simple protuberances, without lateral
appendages. The pre-divided water system is asymmetrically placed
as regards the disk of the future starfish, and the five extensions do
not project beyond the stomach of the brachiolaifia. In the next form
in which the circumoral calcareous rods appear, we find that the brachi-
olaria has been wholly absorbed, and the starfish has assumed a stellate
form, brought about by an enlargement and growth of the termi-
nals (PL II. fig. 2). While the oral plates were placed with their

22 BULLETIN OF THE

lengths parallel with the median water tube^ they are now at right
angles to its course. The median tube has not yet extended to the
extremity of the edge of the terminal, but has formed two and three
pairs of lateral branches, — the first formation of the legs of the star-
fish. The ten oral ambulacral plates or rods, am, form a pentagonal
network, not yet united, but already in the approximate position which
it occupies in the adult (PI. II. fig. 3).

The plates are crescentic, with convexity pointing outward, perforated,
closely approximating near the middle line of the radius, and more dis-
tantly separated in the interbrachial regions of the starfish. As far as
their general appearance is concerned, they resemble incipient ambula-
cral plates of later stages of Asterias, as in former conditions they
resembled those of Amphiura.

In an older condition (I'l. II. fig. 3) of the oral ambulacral plates,
the interbrachial ends grow together, and at the same time become very
much more thickened in the interbrachial region. The ten oral ambula-
cral plates, am, now form a pentagonal ring about the mouth opening.
In this stage the rudiments of two pairs of ambulacral rafters, ani^,
have likewise appeared.

The outlines of the single member of the oral ambulacral ring of
plates at present are as follows. Each oral ambulacral plate has the
form of an elongated bar, enlarged at either end. The length of the
bar is at right angles to the line of the radius of the arm. On the ab-
oral side it is deeply concave, while on the adoral it is straight, slightly
curved. The radial extremity is bifid, divided into an upper and lower
branch (PI. III. fig. 1). The interradial extremity is enlarged into a
massive thickening, forming a club-shaped body whose aboral broad end
abuts the lateral wall of the terminal. The mass of the thickened part
of the oral interbrachial plate is on its actinal side, while on the abactinal
side it is concave, in which concavity fits a heart-shaped plate, ib, later
described as the odontophore. This thickening in the interbrachial
region of the oral ambulacrals corresponds with the interambulacrals,
and these plates represent interambulacrals of the oral ambulacral
plates, although they do not seem to be formed as separate calcifica-
tions. In early conditions no spines are found on the oral plates or
bars. No spines were ever detected in the ambulacral region of these
bars, although in older conditions of that part of the oral plates which
lies in the interradii spines were found, as in the other interambulacrals.

The subsequent growth of the interbrachial ends of the oral ambula-
crals is as follows. They grow at the expense of the ambulacral orals,

MUSEUM OF rOMPARATIVE Z^MiLOOY. 23

forminiz elongated bodies whose pfreatcst lengtli lies in the direction of
the interradius, or parallel to it. As their size increases, four spines
form on each of these plates, two on the aboral and two on the adoral
ends The existence of these spines would indicate that they are con-
solidated interambulacrals, and that interambulacrals as well as ambu-
lacrals enter into the formation of the oral ring of calcareous bars.

First InterbrcichiaL* — In very eai'ly conditions in the growth of the
oral rino- of bars, before the increase in size of the interambulacral ends
of the circuniorals, and before spines appear, there form in the inter-
brachial radii certain round or heart-shaped plates, which are thought
to be the first sign of the odontophore.

These plates, ib. lie on the abactuial side of the adambulacral circuni-
orals, in a space between them and the genitals, and on the adoral side
of the cii'cumorals. Their first form is round, or heart-shaped. As the
growth goes on they are pushed moi'e to the aboral region of the inter-
radius, but never extend beyond the cover of the interambulacral ends
of the circumorals, by whicli, in the increased compactness in growth
of the calcareous network, they are almost wholly concealed when the
starfish is seen from the actinal side.t This is the first time that the
odontophore has been described in starfishes as young as Plate II. fig. 4.
From their position of formation they seem to be homologous with
hiterbrachials, which will bo subsequently described.

The Ambulao'al Rafters. % — Under the name of ambulacral rafters
all actinal jjlates of the rays, with the exception of the circumorals and
the adambulacrals, will be included. The following plates are present
when the first pair of ainbulucrrds begins to form : dorsocentral, dc,
five genitals, ^7^-//, five terminals, t'^-t^, and ten circumoral ambula-
ci'als, am. In addition to these the interambulacral circumoral and the
odontophore, z6, are formed before the second pair of ambulacrals, am.
Tiie first of the median doi-sal row of plates, d, appear just after the
second pair of ambulacrals. The marginals, intei-ambulacrals, and second
median dorsal appear after the third pair of ambulacrals.

* Tlie name odontophore, with which this is liomologous, as pointed out by Lud-
wig. is ill}' suited for tlie first interbracliial plates on the actinal liemisome. The
true name of tliese plates can hardly be known until there is some uniformity of
opinion as to their homologies. Tlie term interbracliial does not commit us to
the tlicory that they are homologous with orals or with unpaired marginals.

t In certain deep water Asterids, according to Sladen, a part at least of tiie
odontophore is visible on the actinal surface of the adult. In very young speci-
mens of Asterias the same tiling is true. The feature in the deep-water starfishes
would seem to be embryonic.

} The plates referred to are commonly called ambulacrals.

24 BULLETIN OF THE

The ambulacra] s follow the law of formation of the other arm-plates,
■with the exception of the terminals. The first ambulacral to form is
nearest the circumoral, and new plates are added aborally to the first
formed. The new ambulacrals are protected by the cap-shaped termi-
nals. The ambulacrals originate as elongated rods, with axes at right
angles to the line of the radius. The two members of a pair do not
necessarily arise simultaneously. The position of origin is nearer the
centre of the arm, or nearer its median line than the peripherj'.

In the earliest condition in which the ambulacrals were seen, they
had the form of small calcareous nodules, one on each side of tlie
median line of the arm (PI. II. fig. .3). In older stages these nodules
elongate into bars, growing from the middle line towards the side of the
arm on its actinal region (PI. II. fig. 4). By an increase in the length
of the ambulacral bars, they bridge the interval between the middle
line of the arm and the lateral extensions of the terminals, although
they never join the last mentioned structures (PI. III. figs. 1, 2).

Each ambulacral bar has the following form. Near the middle line it
is enlarged, while on the aboral and adoral borders it is concave, in
order to leave an interval or space for the passage of the legs to the
ampullee.

As the growth of the arm of the starfish goes on, and new pairs of
ambulacrals are formed, the terminals are pushed out more and more
from the disk. At the same time the ends of the ambulacrals approach
one another on the median actinal line of the arm, and ultimately
become articulated together. Before, however, they join, they bifur-
cate on the median line, and form an iipper and a lower spur, as in the
circumoral calcareous rint;. As in earlv stages of the starfish, there are
only two rows of feet, one on each side of the median line ; the rows of
openings for the passage of the feet are also in two lines. It is only in
young starfishes of considerable size that we find four rows of openings
between the ambulacral rafters. In all the specimens figured there are
but two rows of feet.

The voung stages of Asterias studied bv me were never found to have
spmes on the ambulacrals, and neither in the oldest nor in the youngest
was there any median row of plates or spines of an embiyonic nature on
the actinal side of the arm.

Second Interhracliial. — "When the growing starfish, in which tlie
arms ha\e pushed themselves out to a considerable size, is looked at
from the actinal side, there will be seen in the interradii, in the space
left between the marginals and the abaxial end of the interbrachial ex-

MUSEUM OF COMPARATIVE ZOOLOGY. 25

tremities of the oral ambulacrals, a single interbrachial resembling the
first (odontophore), but abaxial to it (PI. V. fig. 8, ib"). This is called
the second interbrachial, ih"^. Other interbrachials outside (abaxially
to) this were found ; but in the genus Asterias the number and devel-
opment of these interbrachials is not as great as in some other genera.
The first of these interbrachials, " odontophore," might be regarded as
homologous with the orals of the Amphiura. The homology of the
others in Ophiurans is not clear to me. They are of course represented
in other starfishes, where they are sometimes very greatly developed,
imparting a characteristic form to the body, filling in the whole inter-
brachial region.

The remaining interbrachials may be numbered among early plates to
form in the young starfish. The}' are, however, the last plates to form
of all those which we have mentioned in our account of the early or
primary plates of the body.

"Wiien the starfish of a stage like that shown in Plate Y. fig. 7 is seen
from the actinal region, an irregular triangular interval is seen in the
iuterradius just outside the two interarabulacral circumoral plates, amd.
This interval is bounded by the adambulacral circumoral, amd, the first
interambulacrals, ad'^, one on each side, and the laterals, also one on
each side. In the centre of this space, on a line opposite the middle
of the interambulacrals, the first interradial or interbrachial takes its
rise. As the starfish matures, other interbrachials also form outside,
aborally from that which has already appeared.

Connectives. — Under this name are included certain plates of the
bod}' and disk of the starfish, which bridge the intervals between the
others, but which have a secondary place as compared with primary
plates. There are connectives on the abactinal region of the arms, and
others on the disk, but in either case they do not differ greatly from
each other. In the connectives we have a multiplicity of calcareous
plates, imparting a compactness to the abactinal hemisome. Their
form, size, and number are variable, and their morphological importance
of a subordinate character.

Spines. — The study of the primary spines of the young Echinoderm
is one which in most accounts of the development is not given very great
prominence, yet these bodies are in many genera among the first calci-
fications to appear, antedating in formation many plates which play a
most important part in the determination of the external form of the
animal. When the first spines appear in the starfish, there are only
eleven plates present, five terminals, five genitals, and a dorsocentral.

26 BULLETIN OF THE

The dorsocentral is hardly larger than a small calcareous nodule, and
the genitals and termiuals are but simple branched spicules. Not a
single plate of the actinal region of the body has yet appeared. It will
be seen that plates which have been regarded of importance in a discus-
sion of the affinities of the starfish with other Echinoderms are not even
present as a simple rudiment, when spines which few have yet consid-
ered of any great importance morphologically have begun to form, and
are well developed in many cases. Are we justified in thus neglecting
the spines, or have they no morphological meaning outside of a simple
classificatory interest 1 Tlie discussion of the meaning of the large size
of the first-formed spines will be taken up later. It is necessary now
for us to consider the size, arrangement, and distribution of these struc-
tures, their embryonic form, and their general mode of growth.

The spines originate as trifid spicules, and in their early stages are not
to be distinguished from calcareous plates. The earliest spines (PI. I.
fig. 3, sp) to appear are those at the outer rim of the terminals of
the young starfish, peripherally to these plates. They are at lirst ten in
number, or two to each terminal, and by the time the notclies which
indicate the interbrachial regions of the future starfish arc incised,
the number of immature primary spines has increased to nine. Later,
when the number of ambulacral rafters lias grown to six (tliree pairs),
thei'e are sis very prominent spines on the outer l)order of the termi-
nals. According to A. Agassiz, these spines are more or less fan-shaped,
and recall those of certain Echinoids. The dorsocenti-al, for a long
time after its first apijearance, bears a single long and prominent spine.
This calcification is jointed to the centre of the dorsocentral on the
aboral side, and later other primary spines are added to the dorsocentral.
The genitals (basals) have at first three long, slender spines, which
originate while yet the starfisli is borne by the brachiolaria. The
spines of the medial dorsal plates of the arms are prominent and single
at first, each situated in the middle of the plate upon which it is carried.
The lateral or marginal plates, m, of the arms bear long, stout, single
spines. Each interambulacral plate, rtr7, has at first a single spine. No
spines were ever observed in the anibulaci'als, but the extremities of
the oral aiubulacrals in the interbrachii bear four spines, two of which
lie on the edge adjacent the mouth. Tlie so-called odontophorcs * were

* From the position wiiicli they early occupy, it is not to be woiulered at tliat
spines are not developed on the first interbrachials or the odontopliorcs. They
are covered on the actinal side by the interbrachial ends of the oral ambulacrals,
so that spines could not be formed.

MUSEUM OF COMPAKATIVE ZOOLOGY. 27

not observed to carry spines, and the intertrachials, in their younger
staires at least, are destitute of these structures.

With the advancing growth of the starfish, the nuniber of spines on
the primary plates increases, and new spines are formed on new plates
as they appear. The later formed spines, however, never have that
prominence so marked in the younger and primary plates, but appear
more compact, and more like the spines of the adult Asterias.

Pedicellarice. — The pedicellarise, ^;ic/, were first observed on the
terminal plates in a stage of the starfish in which there were four
median dorsals (PI. IV. fig. 3). They were then confined to that
plate, being absent on all others. In an older starfish, or one with
seven median dorsals (PI. IV. fig. 4), they were likewise found in clus-
ters on the second marginal plates, m^, and one or two were likewise
seen on the plates, m^, between the second marginals and the termi-
nals. Although in both these stages large spines exist on the median
dorsals, d, there are no pedicellarise as yet formed upon or near them.
Like the spines, the calcifications of the pedicellariae are at first wholly
separate from the plates from which they rise. Unlike the spines,
however, their calcification is from the first double, or split longitudi-
nally into two separate parts.

Stone Canal. — The calcifications in the wall of the madreporic canal
•were observed in a larval stage before the external modifications of the
plate through which it opens were evident or had appeared. It con-
sists of a delicate tubular network of calcifications, formed by a lacework
of calcareous spicules, which appear to arise from many centres of for-
mation. They appear to form in the wall of the tube itself I am un-
aware that any one has described the stone canal in a young starfish in
which there were but seven median dorsal plates, yet it is well marked
at that age.

4. Comparison with other Asteroidea.

It is here intended to consider certain relations between the plates
of Asterias and the observations and comparisons which have been
made by others on the plates of young starfishes. The study of the cal-
careous formations of the adult and their history from the time when they
fi.rst appear has engaged the attention of several naturalists, and many
different conclusions have been arrived at in this study. With these
recoi'ded observations and interpretations I have been able to compare
my own on Asterias, and their concordance has strengthened my belief

28 BULLETIN OF THE

in them. In one or two instances, liowever, there are differences, either
of observation or of interpretation. It will perhaps be profitable, before
■we can discuss the relationship of Asterias with Ophiurans, that these
differences and concordances among Asteroidea be considered. The
subject deals with calcareous plates only.

The most important observations of the way in which the plates of
Asteroids develop are those of Krohn, Thomson, A. Agassiz on Asteracan-
thion, Ludwig on Asterina, and Lovdn on Asterias glacialis. The way in
■which the plates of Asterina develop is as well known as that of any
other Asteroid, if not better. As this development of Asterina pertains to
a starfish without a nomadic brachiolaria, and as Asterias has an indirect
development with nomadic brachiolaria, it is interesting to compare the
formation of the plates in the two types, and to note the differences
which occur. "Whatever the character of the metamorphosis of a star-
fish may be, — whether it has a nomadic brachiolaria, as Asterias, or
carries the young in brood-sacs, as in Pteraster, — it would appear that
the sequence of the growth of calcareous plates is little afi'ected by it.
How much the abbreviation in early development afi"ects the sequence
in the growth of plates is yet to be proved, and a complete series of
the young Asterias to compare with Asterina may give us valuable
information on this point. The figures of Asterina by Ludwig, and
those of Asterias by Agassiz, Krohn, Thomson,* and Loven,t in a way
supplement each other, yet much still remains to be done on late
stages of both genera.;}:

For a comparison of the way in which the plates of the abactinal
hemisome of the body of Asterias develop with those of other Asteroids,
I have little to add to what is known as far as the dorsocentral is con-
cerned. The various authors who have written on this subject do not
emphasize the fact that it is formed after the terminals and genitals, or

* Krohn and Thomson figure and describe isolated stages of growth. Agassiz
considers the whole subject of the development.

t Loven figures only later stages with stellate form.

X It would appear from the relative time and sequence of the appearance of
plates in related genera of Ophiurans being very diSerent, that it is not safe to
rely upon a similarity in time when calcifications appear in the comparisons of
homologous plates. Other naturalists have already commented on this fact. A
diversity in the time of the appearance of homologous plates in related species
seems to me paralleled in the fact that m two Asterids once thought to be generi-
eally the same, and even now, if their adult features alone are examined, regarded
as generically identical, one, A. tenera, has no nomadic brachiolaria, and the other,
A. herylinus, has such an elaborate 'metamorphosis with this stage.

MUSEUM OF COMPARATIVE ZOOLOGY. 29

that it antedates all the plates of the actiual hemisome. Agassiz sup-
posed it to be formed after the other ten abactinal plates, and Ludwig *
does not say that it is or is not formed at that time.

The time of formation and mode of growth of the terminals, at the tips
of the arms, seems to be the same as already described by A. Agassiz
and Ludwig. Agassiz found them to form before the genitals, wdiich
is true also in my larvte. I cannot verify the statement that the ter-
minals ever fuse with each other, as described by Agassiz. The
arrangement of spines on tlie terminals differs somewhat from those of
Asterina.

According to Ludwig (p. 50, op. cit.) one of the interradials (geni-
tals) precedes in time of formation and size the other genitals and
the terminals. This is the genital wliich later forms or fuses with
the madreporic plate. This predominance of the genital contiguous
to the madreporic opening was not noticed in Asterias, although the
relative distance and general situation of this plate as compared with
the madreporic opening are about the same as Ludwig describes for
Astei'ina.

My observations on the growth of the plates of the abactinal region
of the arms resemble those recorded by A. Agassiz, Loven, Ludwig, and
Viguier. The calcifications of the body in the abactinal hemisome also
resemble, with some exceptions, those already described. According to
the first author (p. 37), in an eai'ly condition after the eleven abactinal
plates were formed " the whole of the abactinal surface has become
coated with a very fine granular deposit of limestone." This formation
was not seen in the specimens of Asterias which were studied.

The observations on the mode of formation of the oral ambulacrals

* His youngest stages show eleven plates, and in the text he speaks of them
as if the " Centrale," dorsocentral, was synchronous in formation. (See Entwick-
lungsgeschichte der Asterina gibhosa, Forbes, Zeit. f. Wiss. ZooL, Vol. XXX VII.)

The homology of the calcifications of the pluteus of Ophiurans and Echinoids
witli calcifications in the stem of the Crinoids would seem far-fetched. The spines
of the pluteus are secondary developed structures, and it is believed by some that
they have no phylogenetic significance. The fact that they are wanting in the
brachiolaria of Asterias would look that way, but in Amphiura they are repre-
sented before the larva leaves the mother. It is possible in this instance to be-
lieve either that Amphiura is descended from a genus wliich had a pluteus with
spines, and in its abbreviated metamorphosis the rudiment of the spines only re-
mains, or that the plutean spines show relations witli otlier groups outside tlie
Opliiurans. Tiie latter conclusion does not appear absurd, and it may be possible
later to show that there is an homology between the stem of a Crinoid and the
plutean spines of an Ophiothrix.

30 BULLETIN OF THE

resemble those of Ludwig on Asterina,* and Loven's figures of .-1. gla-
ciahs,j but difler somewhat from Agassiz's. A. Krohn J was one
of the first correctly to figure the situation and early form of the
oral anibulacrals of the starfish in young stages in its growth. His
figure of the ten first formed oral ambulacral plates in the starfish
found in Bipennaria correspond closely with those of Asterias which I
have represented. He also figures the spines of the terminals, but does
not represent the terminals as they exist in Asterias.

Sir Wyville Thomson's figure and description§ of the early forms of
the ambulacral rafters of Asteracanthion violaceus closely resembles what
I have seen in Asterias. His figure of the oral anibulacrals and the
ambulacral rafters corresponds with mine. There is not as close a
likeness in the plates of the abactinal side which he has figured and
my own. The first dorsal seems more prominent in one than in the
other. Both of his figures, represented from the actinal and abactinal
hemisomes, are regarded as important contributions to our knowledge
of the early form of the calcareous plates of Asteroids.

Metschnikolf il has published very instructive figures of the young
and stellate forms of a starfish. In Plate XL fig. 8, he represents the
earliest form of the ambulacral orals about as they appear in Asterias.
They have here the form of simple calcareous spicules. The spines of
the terminals in this stage are also well showi. Another figure of a
young starfish by the same author (PL XII. fig, 1, A) represents five
terminals, an inner row of six genitals, and a dorsocentral. The mad-

* It may be borne in mind that the mouth of Asterias is "ambulacral," i. e.
formed for the most part of modified ambulacrals, while Asterina is classified as
"adambulacral," — mouth formed of both ambulacral and adambulacral. It is
consequently necessary that the j'oung stages of Asterina have adambulacral
calcifications distinct from ambulacral in the formation of oral plates. It would
seem from Ludwig's figure (fig. 1)8) and his lettering as if it was formed in this way.
In Asterias, however, similar plates are formed from the interradial ends of tiie
ambulacrals, and not as separate calcifications. If I am right in my observations,
it would seem that Asterias has an ambulacral mouth from the early stages of
growth.

t Etudes sur les Echinoide'es. K. Svensk. Vetensk. Akad. Handl., Stockholm,
Vol. XT. Pt. II.

t Krohn, August. Ueber die Entwickelung der Seesterne und Holothurien.
Arch. f. Anat. Piiysiol. u. Wiss. Med., 1853.

§ Thomson, C. Wyville. On the Embryology of Asteracanthion violaceus.
Quart. Jour. Mic. Sci., I., 18G1.

I! MetschnikofF, E. Studien liber die Entwickelung der Echinodermen und
Nemertinen. Mem. Acad. Imper. Sci. St. Pe'tersbourg, Vol. XIV. No. 8.

MUSEUM OF COMPARATIVE ZOOLOGY. 31

reoorite is represented in the same ring as the terminals, and in posi-
tion is a separate plate from the genitals. The genital which occurs in
the same interradius is double.

If this condition or arrangement is found on more extended observa-
tions to be the exact relation, it may lead to much light on the whole
question of starfish morphology; for if the madreporite is not a genital,
but a distinct plate, the fact adds strength to the belief that the
oral of Amphiura is the homologue of the odontophore, while the
basal of Amphiura is the homologue of the genital of Asterias. It may
render it necessary for us to regard the madreporite in starfishes as
ordinarily described as a consolidation of genital and madreporite,
which would somewhat affect accepted homologies. The separate
calcification of the stone canal, and the eccentric position of the madre-
poric opening as compared with the genital, point to to a compound
character of the madreporite.

A. Agassiz's account of the way in which the ambulacral and inter-
ambulacral plates of the arms of starfishes are formed, differs from what
I find in Asterias. He says (pp. 91, 92), "In the case of the young star-
fish, the radial plates of the abactinal system which form the dorsal part
of the arms gradually extend towards the edge of, and down on to the
actinal side, enclosing the water system little by little, and finally, as has
been described, covering the ambulacral tube, leavmg only openings for
the passage of the tentacles. ... In the starfishes, the actinal plates
formed by the bridges separating successive pairs of tentacles become
resorbed along the central line, the edges forming inwardly by spurs the
true ambulacral plates, and the plates which little by little develop so
as to form the edge of the arms are likewise formed from the plates
originally a part of the abactinal system. Those which are on the out-
side of the tentacles become the interambulacral plates, but differ in no
way from the plates forming the sides of the arms." *

If I rightly understand his account, there is considerable difference
between the w^ay in which ambulacral and interambulacral jilates form
in the starfish which I studied, and those which he describes. The

* It is hard to reconcile this view of the way these plates (ambulacral) form
with the figures of them by Krohn and Thomson. Agassiz's figures (Pi. VI.
fig. 12, PI. YII. fig. 1, Embryology of the Starfish) of the plates of the actinal side
of the arm differ from those of Krohn of a starfish from Bipcnnaria, and of Thom-
son of A. violoceus. They also differ from mine. Tlie separate ambulacral rafters
and oral ambulacrals are not represented, but the actinal calcareous plates are
represented as joined together.

32 BULLETIN OF THE

main difference is, that in the Asterias which I studied the plates on
the actinal side of the arm originate on that side, and there is no
growth downward from the abactinal surface enclosing the water sys-
tem. The ambulacral rafters and interanibulacrals originate as separate
calcifications on the actinal side of the arras, while no absorption of
plates previously formed was observed.

There is a close similarity in the early formation of the actinal plates
in x\sterias and of those of Asterina, followed b}' Ludsvig, and my in-
terpretation of some of the plates of the mouth is in most cases the
same as his.

Ludwig says (p. 49, op. cit.) that he is the first to make known the
primary position of the ambulacrals.* The difference in the early form
of the first and second pairs of ambulacrals, or those which form the
oral ring, is not especially considered by him, and his account does not
extend to the growth of the ambulacrals formed subsequently to the
oral or first pair.

It was not possible for me to observe any relationship in the time
when the members of the five pairs of ambulacrals form, or their se-
quence, as he has done in Asterina, although I have repeatedly found
young stai'fishes in which one pair of ambulacrals (oral) smaller than
the remaining, or in which one or more members of the five sets were
missing.

The five pairs of plates which Ludwig {op. cit.) letters JA in his ac-
count of Asterina are called bv him interambulacrals. Bv this term
it is understood that he means what are here called marginals. In the
development of these plates Asterias closely resembles Asterina.

Adoral to these plates lie five other plates, a single plate in each
interradius. These arc the first interbrachials, and are regarded as the
odontophores of authors. They are called the heart-shaped plates from
the shape which they have in the young Asterias. In Asterias the in-
terbrachial ends of the oi'al ambulacrals arch over the heart-shaped
bodies before the "lateral plates " are developed.t

* Ludwiii was the first to show how the ambulacrals originate in Asterina.
Tlie form and early condition of the oral ambulacrals of an Asterias-like star-
fisli with a brachiolaria was given by Krohn in 1853. Thomson [op. cit.) in
ISlil figures correctly the first form of the ambulacrals in A. violaceus. It would
therefore seem that in genera besides Asterina the subject had already attracted
observers.

t Ludwig was at first of the opinion that these odontophore plates are " Inter-
mediiire Skelettplatte," which I interpret to be the same as " orals"; lie later sup-

MUSEUM OF COMPARATIVE ZOOLOGY. 33

The position of the newly formed interambulacrals as regards the
terminal is similar in Asterina and Asterias.* In Asterina, from Liid-
wig's diagram, I should judge that these plates arise near the outer
ends of each pair of ambulacral rafters, so that a line through them
■would pass through the length of the rafter. In Asterias, however,
the interambulacrals arise in the interval between the outer ends of the
early formed ambulacral rafters.

Plates homologous to the under basals of Crinoids are recognized by
Sladen f in several species and genera of Asteroids, including Asterias
rubens and A. glacialis. I have been struck in the examination of
figures of deep-sea starfishes to see how close, in some instances, the
likeness between the test of certain of them and that of the yoimg Aste-
rias is. The young and adult of Zoroaster seem closely to resemble cer-
tain young stages of Asterias. Sladen comments on the " unmistakable
crinoidal facies " which the young Zoroaster has, and regards a young
stage of this Asterid as highly suggestive of the Ophiuroid genus Ophio-
pyrgus. Whether the resemblance between the young Asterias and the
young Zoroaster can also be interpreted as a crinoidal facies, I leave to
those more familiar with the Crinoids to consider. The simple an-ange-
rnent of plates in Zoroaster is an embryonic feature.

There is some danger in affixing to the plates of starfishes names of
plates which are current among students of Crinoids. While it may be
held to be proper to do so, if the Crinoids represent the ancestral con-
dition of Echinoderms, it might lead to error if they are simply special-
ized or degenerate descendants of other and older groups.

posed them " Unpaare Interambulacralplatte." It is thought that he means by
the latter the marginal plates.

* Fig. 97, Plate VII., of Ludwig's paper on Asterina would seem, from the po-
sition of the terminal as regards the feeler, F, to be a view from the abactinal side.
Such a conclusion is likewise supported by the relative position of the plates, A^,
or second pair of ambulacral rafters. If, however, the view from which the figure
is seen is from the abactinal side, it would seem as if other abactinal plates would
be represented. If they were figures of Asterias of the same age, such plates cer-
tainly might be expected to be visible. Perhaps the term " bei tieferer Einstel-
lung " explains the peculiar arrangement of plates and the loss of the abactinals.
I am unable to understand the figure, since the view is said to be from a prepara-
tion (fig. 96) which is shown from the actinal side.

t On the Homologies of the Primarj' Larval Plates in the Test of Brachiate
Echinoderms. Quart. Journ. Micros. Science, Vol. XXIV., new ser., 1884.

Judging from Sladen's figure of Zoroaster fnlgens (fig. 16), it seems that what he
calls the " underbasal " is also represented in the abactinal hemisome of a young
Asterias.

VOL. XVII. — NO. 1. 3

34 BULLETIN OF THE

It is a most important thing to know more of the early formed plates
of the genera Caulastei', Perr.,* and Ilyaster, D. & K., in which we have
appended to the middle of the abactinal zone a short peduncle. It
would seem that this peduncle is comparable with the stem of a Cri-
noid. In a young Ctenodiscus a protuberance in the same place as the
peduncle of Ilyaster has been noticed, but I have never seen it as long
as figured in Ilyaster. As Ctenodiscus is a common .starfish off the
New England coast, it would present a most instructive genus for the
study of the homology of the early formed plates of a starfish with an
abactinal prominence.

If it should be shown that this appendage to the abactinal region
of the genus Ilyaster is a remnant of the ancestral Crinoid stem, it
might be supposed that the Asteroidea have descended from criuoid-
like genera. It may likewise be true that the Crinoids are highly
specialized and descended from certain starfishes or Ophiurans. In
this case, perhaps, the arrangement of apical plates in the larval
starfishes is the most primitive, and may determine the nomenclature
of the Echinoids.

The pedicellariae of Asterias are relatively somewhat larger in the
young than in the adult. In their early condition they are short and
stunted, clavate, with at least two centres of calcification, which later
form the two jaws. Unlike the primary spines, the calcifications in
each pedicellaria are not consolidated, but double from the very first.
The theory that the pedicellarise are homologous to spines, renders it
necessary to compare calcifications which diff'er in shape from the very
first, and also to compare a primarily single with a double calcification.
Neither of these difficulties is necessarily fatal to the theory, nor does
the mode of development of spines and pef^' . .idria^ give w^holly satis-
factory proof of the theory.

The growth of the spines on the calcareous plates of Asterias re-
sembles that of the same stinictures in Asterina, as described by Lud-
wig. In the case of the spines of the primary plates they arise as
separate calcifications, and are not extensions from plates already
existing.!

* Memoire sur les Etoiles de Mer recueilles dans la ^ler des Antilles et le
Golfe du Mexique durant les Expeditions de Dragage faites sous la Direction de
M. Alexandre Agassiz. Nouvelles Archives du Museum, 2 ser., Vol. VI.

t The " ambulacral spines " observed by A. Agassiz on the outer edge of the
ainbulacral plates were not observed in Asterias. Is it not possible that tiiese
spines belong to the " interambulacrals " '?

MUSEUM OF COMPARATIVE ZOOLOGY. 35

The calcification of the stone canal is not treated by other naturalists
■who have studied the early torm of these organs in younger stages of
Asteroidea ; so that it is not possible to compare my observation with
others as far as this structure is concerned.

The growth of the calcifications of the pedicellarite in Asterias is the
same, or nearly the same, as has already been recorded by A. Agassiz.
No histological studies were made of these organs, and my attention
was not turned to their minute anatomy,

A. Agassiz found that the madreporic opening is placed on the
" actinal side in the angle between two rays," * and that it is protected
by a thick funnel-shaped projection. There seems to be a ditlerence in
the position of this opening in some of my larvae from those which he
studied, for in the younger larvaj of Asterias the madreporic opening
does not lie on the actinal side of the larva, even in considerably ad-
vanced stages. The position of the madreporic opening is of course a
most interesting thing in morphological studies of the young starfishes,
and more observations as to its position are needed.

In the figure already referred to which Metschnikoff gives of the
abactinal side of an unknown starfish, the madreporic body appears on
the very margin of the disk of the starfish, and would seem intermedi-
ate in position between a madreporic body on the actinal and on the
abactinal surface. In Ludwig's figure (fig. 94) of the young Asterina
in which a stellate form has been taken on, the madreporite (P) is
abactinal. In some of my older stages I was unable to discover the
madreporic plate, but believe it in all cases abactinal.

5. Homology of the Plates of Asterias with those of

Amphiura.

Of all the Ophiurans the growth of the plates is best known in a
viviparous genus, Avijjhiura squamata, Sars. While it is desirable to
know more of the sequence and method of formation of the plates in
an Ophiuran which has a development through a pluteus, the genera
which present this condition have as yet not been much studied as far
as the growth of plates is concerned, and our knowledge of the Ophi-
uran plate development wholly relates to Ampbiura.f In comparisons

* Op. cit., p. 45.

t The author is not of the opinion tliat any very great exception to the law of
the growth of new plates is brought about by what is called abbreviated develop-
ment in Ecliinoderms. In essential points the growth of the calcareous plates in
Asterias and Leptasterias, genera representing two types of development, is the

36 BULLETIN OF THE

between stellate forms of the two groups, Ophiuraus and Starfishes, I
have chosen, therefore, Amphiura on the one side and Asterias on the
other as representatives. It would seem as if it were necessary care-
fully to compare the stellate forms of Echinoderms before we can pass
to others, when the external foi'ms are so varied. It may seem as if
the difficulties in a comparison of Ophiurids and Asterids would be
small, but even here we find very great differences in opinion as to the
homology of certain plates, and a variety of interpretations upon struc-
tures of primary origin. While it is not proposed in this paper to go
beyond a comparison between stellate Echinoderms, it is believed that
a more accurate conception of the relationship of plates is possible than
that ordinarily accepted. There is range enough in the modification of
plates in Ophiurans and Asteroids to call for the best possible state-
ments of their relationship in the two groups.

The following theses may be stated in a tabular form, to indicate the
line of discussion which is to follow. Plates of Asterias and correspond-
ing plates of Amphiura are placed side by side.

Asterias. Amphiura.

Dorsocentral. DorsocentraL

First dorsal ? Radialia.

Genital. Basal.

Oral arabulacrals* Spoon-shaped plates.

Interbrachial ends of oral ambulacrals. Adambulacrals (1 and 2).

same. It is not intended to compare Asterias with any Ophiuran except Amphi-
ura. The many problematical questions which have arisen in a comparison of
Amphiura and Crinoids are also passed over in silence. 'My object in this paper
is to see if it is possible to arrive at a better idea of the homologies of the stellate
Echinoderms. A discussion of tlie homologies of the plates of either with those
of Echinoids, Crinoids, or Holothurians is not proposed. Ampliiura is chosen
for a comparison, for the reason that the development of its plates is better known
than those of other Ophiurans. From the statements of those who liave written
on the development of an Ophiuran from the pluteus, it would seem that there is
some difference in the sequence of the plates in Ophiurans with and those without
a pluteus, but the amount of difference is yet to be made out.

In this connection, it seems to me that published statements about Ophiophragmus
by Mr. Xaclitrieb have an interest. Naclitrieb (Studies, Johns Hopkins University,
Vol. IV. No. 2) finds that Ophiophragmus has a " development without pluteus."
He says he was able artificially to fertilize Ophiophragmus. This is the first time,
I believe, that an Ophiuran without pluteus has been artificially fertilized ; and,
judging from the statements which he makes, the development of Ophiophragmus
must be very peculiar.

* Lender this designation I refer to the plates which resemble ambulacral rafters,
surrounding the mouth.

MUSEUM OF COMPARATIVE ZOOLOGY.

37

(Odontophore) First Interbrachials.

Other Interbrachials.

Connectives.

Terminals.

Dorsals.

Dorsolateral.

IVIarginal*

Ambulacrals.

Interarabulacrals.t

Under-basals 1

Oral.

Wanting on arms.

Terminals.

Dorsals.

Wanting.

Wanting.

Ambulacrals.

Laterals (Adambulacrals).

Under-basals.

The ventrals of Amphiura and the plates of the stone canal and pedi-
cellariae of Asterias are not common to the two genera.

Spines (embryonic) are present in the laterals (adambulacrals) of
Amphiura. They are not found in the ambulacral rafters of Asterias.

It will be noticed in the above list, that two adambulacrals of Am-
phiura (1 and 2) are designated as the same as interbrachial ends of the
oral ambulacrals of Asterias. In Asterias they resemble other ambula-
crals, except that on their interradial ends they bear spines, while in
Amphiura they more closely resemble adambulacrals, and so they were
called by Ludwig. It will probably be said, that they are either ambu-
lacral or interambulacral (adambulacral), and abler persons than myself
may be able to show that the oral ambulacrals of Asterias are different
from the oral adambulacrals of Amphiura. I confess, however, that I
am unable to see that they may not be the same plates, now modified
iu one way, now in another. They are the most difficult plates to com-
pare of any in Asterias and Amphiura. Without being able to make up
my mind whether or not there are plates in the young starfish homolo-
gous with the torus and teeth of Amphiura, I think it not improbable
that these structures may be homologous with the "Aristotle's Lantern"
of lu-chius, but I am not willing yet to venture that statement. There
are in certain genera of Asteroids, of course, teeth which may be com-
pared with the teeth of the Amphiura ; but whether tliey are homolo-
gous is doubtful.

r. Herbert Carpenter, in his paper on the growth of the calcareous
plates of Amphiura, has pointed out differences in the time of appear-
ance of the same plates in the American and European varieties of
A. squamata. I regard this as a very fruitful field for morphological
study. T'he recognition of a great diflference iu the time of the appear-
ance of homologov? plates among Echinoderms seems to me an impor-
tant one. There is considerable variation in the time and sequence of
* Interambulacral of some authors. t Adambulacrals of recent authors.

38 BULLETIN OF THE

appearance of homologous calcareous plates in genera closely allied to
each other, which certainly looks as if the difference in the time of
appearance of plates is no valid objection to a strict homology of those
plates. How much a recognition of this principle may change or modify
existing ideas of Echinoderm morphology, as far as the plates are con-
cerned, remains to be seen. It is possible that some of the differences
in the sequence of the plates of Asterias as compai-ed with other genera
of Asteroids may be explained in this way. If we recognize so much
difference in the time of the appearance of homologous plates in genera
closely related, does it not call for great caution in this particular in the
comparison of genera of different groups ] It does not seem too much
to say, that an acceleration or retardation in the time of appearance
of primary plates may have led to essential differences in the exter-
nal forms of Echinoderms. On the other hand, it is very strange if
geographical distribution has brought about sucli a great difference in
the sequence of plates as that which Carpenter finds between the Ameri-
can and European forms of A. squamata. It seems as if there must be
some mistake in the identification either of the European or American
specimens. As far as external form goes, my Newport specimens closely
resemble A. squamata, and specialists in the study of Ophiuraus have
so identified them for me. Shall we call the American and European
representatives different species or different genera, or does A. squamata
in Narragansett Bay depart so widely from the same in European waters
as far as development goes ]

Dorsocentral. — There seems to be a uniformity of belief that the mid-
dle plate of the abactinal region of the body is homologous in Asterias
and Amphiura. The only essential point of difference is the presence of
a large spine in the young Asterias and its absence in Amphiura. This,
however, is not thouglit to be of importance enough to have any mor-
phological meaning. The author has no doubt that the dorsocentral
forms in the same relative position in both genera, as shown by the ob-
servations already recorded. Whatever objections, therefore, might be
urged on the ground that the sequence * is different, are not regarded
as fatal. The name dorsocenti-al is well chosen, but should not be
confounded with the centrodorsal of Crinoids.

Genitals. — The author homologizes the first ring of five plates, which
form in the interradii of both Asterias and Amphiura, with each other.

* The only plates of Asterias which can be homologized with the radialia of
Amphiura develop in Asterias ajler the dorsocentral. This fact probably has no
morphological meaning.

MUSEUM OF COMPARATIVE ZOOLOGY. 39

There would seem to be no expressed variety of opinion on this point
among those who have written on the subject ; the only important
question which has been raised being whether the term "genital" be an
appropriate one to designate them, and whether their fate is the same.
It is important, however, for us to call attention to this fact in re-
gard to the so-called genitals in the two genera. While they are both
primary plates on interradii, the absence of primary radials in Asterias
has brought them to occupy a difterent position as regards the first
formed plates in Asterias and Amphiura. For illustration, in Asterias
they form an inside ring in comparison with the primary plates (termi-
nals); in Amphiura they form an outside ring compared Avith primary
plates (radials) ; they form an inside ring as compared with terminals
in both. The relationship of one of the genitals to the madreporic
opening would seem to show that the ring of plates of which this is a
member is the same in both cases. Ludwig regards the first plate in
the dorsal hemisome in the interradius of Amphiura as a madreporite.
If he is right, there is no doubt that the first ring of plates in the inter-
radii in Amphiura and in Asterias are the same.

In considering these plates as genitals, too much importance cannot
be attached to the fact that no one has yet satisfactorily traced them to
plates with the genital openings in starfishes. It is certainly a form of
a priori reasoning to characterize them as genitals from the fact that
they are the first interradials and occupy a similar position to the geni-
tals of Echinoids. The one thing which we really know is that in As-
terias one of these first interradial plates bears a definite connection with
the madreporic body, and that it later occupies a similar position in one
of the interradii that the genitals do in the other. Is it not a jump at a
conclusion to suppose from this that they are necessarily genitals ? If
one should say that other interradials form the genitals, there are no
observations to show the error. That the first (orals) plates of the
interradii in Amphiura are genitals, we have even less to support our as-
sumption than in Asterias, One of these plates, according to Ludwig,
is perforated by the madreporic opening, and it is therefore supposed to
be a madreporite. All five are later consolidated in the system of plates
about the mouth, and bear no relationship to the genital openings.
Obviously these cannot be the same as the genitals of Asterias, if geni-
tals are homologous in Asterias and Amphiura. We are consequently
driven to this position : the first formed interradial plates on the abac-
tinal hemisome of Asterias do not enter into the formation of the mouth;
they occupy a position which would indicate that they are genitals, and

40 BULLETIN OF THE

one of them is early brought into connection with the madreporic open-
ing. Tlie plates of Amphiura which are in an homologous position ap-
pear not to have any relation to the genitals of the adult. One of them
is a madreporite. The ultimate fate of both in Asterias and Amphiura
is conjectural, but probably ditferent.

In other words, while the genital of a starfish may be brought by its
growth into intimate union with the end of the stone canal or v/ith the
opening of the madreporic tube, it may readily be seen that in Amphi-
ura the oral, a plate of totally difi^erent homology, might have the same
relation without the same homology. It would not make these plate?
homologous, i. e. genital and oral. Even if in Brisinga, with its ophiu-
ran and starfish affinities, the madreporic opening is found on the odon-
tophore, it is not necessary to regard the odontophore as a genital.
The whole thing of course hinges on the acceptance or denial of the dic-
tum that the presence of tlie madreporic orifice means homology of the
plate tlirough which it opens. While many naturalists in whose opin-
ion we have the most confidence hold that it does, that the existence of
the madreporic opening in a plate settles its homology once for all, to
me it leads in some cases to insurmountable difticulties. It seems to
me that the objection to the homology of the genitals of Asterids with
the orals of Amphiura is well taken, if its defenders adopt the theory
that the odontophore is an oral ; for the genitals of starfishes are cer*-
tainly not odontophores, even if in Brisinga one of the odontophores
bears the madreporic body, or is a madreporite. The mistake seems to
me just here : the madreporite may not be one of the genitals, or ho-
mologous with them, even in Echinoids, but is rather to be considered a
separate plate, which may have connection with one of the heart-shaped
bodies which form the odontophore. Ludwig was, I think, right in a
comparison of the odontophores of Asterids with the orals of the Ophiu-
rids. A. Agassiz has expressed the thought that the oral, or the plate
we now call the odontophore, is an interbrachial, an homology which is
sound, in the light of the development. The fact that in its early for-
mation that genital plate which is nearest the madreporic opening forms
wholly independently of the sti'ucture in question, has a meaning. In its
early condition this plate is not penetrated by the tube in Asterias; but
it is only later, after the brachiolarian notch has been considerably re-
duced in size, that the spicules begin to grow around the opening to en-
close it. It is not until the stellate form has been assumed, and the
arms have reached a considerable development, that the madreporic
plate appears. It may be said that all the primary plates of the body

MUSEUM OF CO]\IPAKATIYE ZOOLOGY. 41

— viz. terminals, dorsocentral, genitals, spines, and even ten of the oral
ambulacrals — have appeared before there is any calcification about
the terminal opening of the madreporic tube. Does not this fact call
for extreme caution in regard to statements that would lead us to ho-
mologize the madreporic plate with a genital rather than any other
Asterid plate ? Another thing, the outer edge of the genital grows
around the end of the madreporic body to enclose it, and the stone
canal forms before there is any cribriform plate over the opening.
There is yet one fact which ought to be taken into consideration with
those mentioned ; viz. four interbrachials (odontophores) form at first,
and the lower end of the madreporic canal occupies the homologous
position of the fifth. AVhile this fact does not demonstrate the ho-
mology of the madreporite, it is thought to have a bearing on the
subject. The madreporite is not in intimate union with the odonto-
phore after the odontophore forms, but it lies in the same inter-
radius.* I venture to say that, if it is not in some kind of connection
with the odontopliore in Asterias, as Perrier says it is in Brisinga, it cer-
tainly is very close to it.

Sladen {oj). cit.) finds no morphological relation between the odonto-
phore and madreporite in Brisinga. While we may or may not subscribe
to Sladen's criticisms of Perrier, he does not in either case seem to me
to prove that there is " no morphological relation" between madreporite
and odontophore, for when we go to very young stages in the growth
of Asterias, we find a very intimate connection between the madreporite
and odontophore.

As all the other odontophores of Asterias form on the actinal hemi-
some, and the plate which occujiies a similar position in the interradius
has a like position of origin, it would seem that Brisinga differs very
considerably from Asterias, as for as the formation of the madreporite is
concerned. In my specimens I certainly detected the odontophore of
the interradial in which the madreporic body is found on the actinal
hemisome, in the same relative position as the otlier odontophores. At
the same time the genital (g^) had not grown around the opening of
the madreporic tube. Here, then, were two separate plates on opposite
ends of an axis of the body. One could not be the other morphologi-

* Obviously, from tlie fact that one of the genitals grows around the madre-
poric opening, we are not obliged to regard this genital itself as homologous with
a madreporite. We migtit easily suppose the primary separation of the madreporic
opening and this genital much greater, and the stone canal so reduced that the
madreporic body occupies a position which it is said to have in Brisinga.

42 BULLETIN OF THE

callj. If the former is a madreporite, the latter is not ; it the latter is
a madreporite, the former cannot be. What additions now take place 1

(1) The genital grows around the opening of the madreporic tube.

(2) The calcifications of the stone canal form. Neither of these events
makes (g^) the first genital a homologue of the odontophore, the ho-
mology of which preserves its distinctness whatever occurs.

In the case of the orals of Aiaphiura, it is found that one of them
grows around the madreporic opening in the same way that one of
the genitals grows in Asterias. It seems to me that this fact alone
does not make genitals and orals homologous, and does not prevent
the homology of orals of Amphiura and odontophores of Asterias.*

A comparison of the odontophore with the oral of Amphiura was made
by Ludwig. According to Carpenter, he no longer holds that view,
although Carpenter does not say what plate in Ophiurans Ludwig now
regards as the homologue of the odontophore. As " interambulacrals
(marginals)"' are wanting in Ophiurans (Amphiura), it is difiicult to
interpret what plates here correspond with the unpaired marginals of
the starfishi Perhaps Ludwig might still consistently hold that while
the odontophore is an " Unpaare Interambulacralplatte," it is still a
homologue of the orals of Amphiura where no marginals like those of
the starfish are found. I believe that the first interbrachial (odonto-
phore) is homologous to the oral, but do not say that Ludwig now holds
such a view.

Sladen holds that the presence of the madreporite in a plate does not
mean homology of that plate among Echinoderms.t It does not lie in
the genital in certain starfishes, and it is disconnected with the orals in

* "Waiving the difference of opinion of Sladen and Perrier (Compt. Rend., Vol.
XCV., July 10, 1882), — the latter of whom holds that the madreporiform body in
Brisinga is always formed on one of the odontophores, and the former, that no
connection whatever exists between these two bodies, — is it not possible to con-
clude that we may have, in this ophiurid-like starfish, a genus in which the
odontophore, like the oral of an Ophiuran, has been modified by its proximity
to the madreporic opening, even if no connection has resulted 1

t In this connection it may be well to call attention to the migration of the
madreporic opening in Echinarachnius along an interradius from the edge of the
disk towards the centre, as shown in my paper on the development of this Clype-
astroid (Bull. Mus. Comp. Zool., Vol. XII. No. 4). A. Agassiz had already shown
the migration of the periproct in the same genus (Bull. Mus. Comp. Zool., Vol. IH.
No. 9, p. 295). The madreporite moves from the margin of the disk to the centre;
the periproct of Echinarachnius moves from the centre to the margin. It might
better be said in regard to the madreporite, that by the growth of plates about it
the madreporic body is pressed to the centre from the margin.

MUSEUM OF COMPARATIVE ZOOLOGY. 43

Astropbytidte. His reasoning seems to me conclusive, and I do not
regard the madreporite as a fixed point of reference in Echinoderm
morphology.

2'ermiiials. — The terminals in Amphiura and Asterias have so many-
points in common that I do not hesitate to regard them as morphologi-
cally identical. That they do not appear at the same time as compared
with the other primary plates does not appear of importance enough to
destroy the argument for their identity built upon their many resem-
blances of mode of growth, position, and relation to the median water
tube of the ray.

Sladen considers that the occurrence of plates which he calls " under-
basals" in the Asterid nullifies Studer's argument that the arrangement
of primitive plates in Asterids corresponds with the monocyclic Crinoids.
I do not subscribe to Studer's idea, but on the other hand I cannot but
ask if Sladen has not overestimated the morphological value of his sup-
posed discovery of the " under-basals " of Asterids in this connection.
It may be well to remember that in Asterias the plates which Sladen
likens to the Crinoid " under-basals " do not appear until after at least
the Asterias may have had a monocyclic stage. At tlie time the " un-
der-basals form," we might suppose that the starfish had passed out of,
or become more developed than, the " monocyclic stage."

From their place of origin and their subsequent growth the terminals
of Asterias and Amphiura are probably homologous. Moreover, it is
probable that these plates are not to be compai-ed with the other plates
of the arm, which originate between them and the axis of the body. It
had seemed to me possible to find a serial homology between this plate
and those of the arm joints ; to compare it, for instance, with a consoli-
dated dorsal, two laterals,* and possibly a ventral ; to find, in other
words, that the portion of the ray in which they lie is a true arm divis-
ion or "joint." My attempts, however, have not been rewarded with
great success, and it seems more probable that they are not comparable
with other arm plates. Students of Crinoid morphology do not find the
homologues of terminals in this group, and there is evidence that they

* Sladen says (op. cit., p. oO) : " The comparatively large size of the terminal
plate at an early stage of the young Asterid is due, in my opinion, to a coalescence
of primitive lateral plates with the primitive, or first formed, rudiment of the ter-
minal,— a circumstance wliich further strengthens my view of the secondary
character of the terminal plate." It would, however, seem from the growth of tlie
plate that such a coalescence does not take place, or at ail events I was unable to
observe it. It is suggested that the absence of lateral plates (marginals 1) allows
the sides of the terminals to grow into their places.

44 BULLETIN OF THE

are not homologous with the oculars of the Urchin. Whether there are
plates in the young urchins which are homologous with the terminals
of the starfishes is a question upon which more knowledge of the devel-
opment of the calcareous plates of the Echinoids may throw some light.

According to Ludwig (p. 188),* A. Agassiz " halt die jungen Termi-
nalia fiir die Dorsalschilder des Erstgebildeten Armgliedes, eine Bezel-
hung die ebenfalls nicht korrekt genau ist." Agassiz says, "The
only calcareous deposits we have (y y\ fig. 32) are evidently parts of
the first arm joints, the dorsal (y, fig. 32), and the side arm shields
(y, fig. 32) of that joint, which consist at present of but a few rods
indicating their future position." (Embryology of Echinoderms, p. 20.)
This was written of an Ophiuran with a pluteus, which is probably
Ophiopholis. Agassiz uses the term "arm joint" elsewhere in his pa-
per to designate one portion or section of the arm. I am unable to
understand exactly what Ludwig means b}' the above criticism of Agas-
siz, unless it is that the terminals (dorsal shields, Ag.) do not belong
to the first formed arm joint, or possibly that the terminals do not
indicate an arm joint. One of these interpretations is the best I can
make of his meaning. If we consider an arm joint among the Ophi-
urans to mean a portion of the arm with a dorsal, two laterals, and
a ventral with enclosed organs, the portion of the ray in which the
terminal lies might not be called an arm joint, since separate calci-
fications for lateral and ventral plates do not exist. However the
terminal may be homologized, it originates like a dorsal, and grows
around the terminal tentacle forming laterals and ventral. In the
structure of the plates, as in position and time of origin, it differs from
all other dorsals, and consequently morphologically may he held by
some not to be an arm joint, or to belong to the first formed arm
joint. Possibly that may be the meaning of Ludwig's criticism above
quoted. If not, I have been unable to see the force of his criticism.

In speaking of an early stage of Asterina in whicli the first dorsals,
or " radials," are formed, Sladenf says : " The first formed plates in
the viviparous or abbreviated larva are the primitive elements of the
terminal plates, the basal plates, and the dorsocentral plate. These
become well developed before any traces of the radial plates make their
appearance. Concurrent with the radial plates the lateral plates (inter-
ambulacral plates) are developed." He might have added also certain
mouth plates, ambulacrals, and intcrambulacrals (adambulacrals).

* Entwicklungrsgeschichte der Asleiina gibbosa, Forbes,
t Sladen, op. cit., p. 31, fig. ]5.

MUSEUM OF COMPARATIVE ZOOLOGY. 45

Primary Radials. — The first radials of a Crinoid, according to Sladen,
are represented by the luterniediare in Asterina.* There is but one set
of plates which can be compared with the Amphiura primary radials,
and these are the first of the dorsal plates of the arms. Notwithstand-
ing the fact that these plates have a different time of originf as regards
the genitals, Sladen regards it possible that they may be called the
primary radials.

If we accept these plates as homologties of the primary radials, they
must be regarded as very much belated in time of formation. Inside
the radials of the disk of Asterias are representatives of the under-basala
and similar radial plates.

Connectives. — The homologies of the numerous small plates and bars
by which the primary plates on the abactiual hemisome of the body are
joined together were not identified with definite plates in the Am-

* Although it can readily be granted that the first dorsals occupy the same
relative position as the radials in Amphiura, it is a question whetlier we are justi-
fied in carrying our comparisons so far as to horaologize them. I grant that their
position on the radius is the same, and that retardation in the time of appearance
of plates has very little importance morphologically ; but it must be remembered
that we are attempting to homologize plates in stages where we know there are
unrepresented plates inter se in stages where there are plates in the starfish which
cannot be referred to those of Crinoids, and plates in the Crinoid which have, as
far as we know, no representatives in the starfish.

t While it is generally true that in Ophiurans the radialia appear before the
"basals," in one case, according to P. Herbert Carpenter (On the Apical System
of Ophiurids, p. 8), basals are formed on the disk where no radialia are repre-
sented [Ophiomitra exigua, Lyman). It would seem improbable that in this in-
stance radialia had formed and been absorbed. Carpenter (Notes on Echinoderm
Jlorphology, Xo. XI.) says of the radialia, " They appear before the basals in the
Ophiurids, but after them in other groups." I confess surprise at this statement,
especially as three years before he had liimself pointed out that Ophiomitra exigua
"has no radials at all, nothing but the five interradial basals (3) intervening be-
tween the dorsocentral (1) and the radial shields," and I can only explain it on the
supposition that he believed that radials had formed and disappeared, that they
are formed later, or that he had changed his mind on the subject.

Carpenter's criticism of my use of ' abaxial basals " I will not consider here ;
but as I nowhere in my paper use the combination "adaxial interradials," I fail
to see why he should speak of any plates as ray adaxial interradials (" his adaxial
interradials ").

Carpenter's statement (op. cit., p. 313, lines 12-17) that every previous writer
regards basals and interradials as fundamentally distinct, seems to have been
written without remembering the fact that Sladen in considering certain star-
fishes {op. cit., p. 33, lines 27, 29) uses interradial for basal, and to explain what
he means by interradials uses the following combination : " interradials (i. e.
basals)."

46 BULLETIN OF THE

phiura. Undoubtedly a few of the larger might be compared, but, as
the connectives are not primary plates, they vary very greatly in num-
ber among different genera of starfishes. Moreover, they appear late
in development, even after generic features are acquired. A notice
of their modification is made under my comparison of Asterias with
other starfishes. Xot enough is at present known of the smaller plates
of Amphiura to attempt any comparison of them with the connectives
of Asterias.

Oral Amhulacrah. — The oral ambulacrals early take the form of the
"spoon-shaped" plates of Amphiura. In tlieir eai'ly condition they
approach more nearly the form of the ambulacrals of Amphiura than
any of the arabulacral rafters of Asterias, which leads me to think that
they are really ambulacra] in their nature. As growth goes on, the oral
ambulacrals lose their resemblance to the spoon-shaped plates, and re-
semble more closely those interambulacral plates which later appear in
the starfish arm. The early resemblances which have led me to regard
the oral ambulacrals as the same as the " spoon-shaped plates " are :
1. Their position and number. 2. Their elongation in early stages
parallel with the water tube. 3. Their time of formation. The only
one of these three reasons which lacks observation is the third. "While
■we know that in Asterias the first plates of the actinal region to form
are the circumoi'al ambulacrals, we do not yet know this for the spoon-
shaped plates of Amphiura. I believe that they are the first, but
cannot as yet definitely state this to be a fact.

The later resemblance to interambulacral plates is caused by their
growth to the inteiTadial regions and the presence of spines. Notwith-
standing this resemblance to adambulacral plates of the oral ambula-
crals in the interradii, Asterias is classified as a starfish with an ambula-
cral mouth. In starfishes in which the iuterambulacrals enter into the
formation of the mouth as well as the ambulacral, it may be supposed
that the growth of the adambulacrals of the mouth has prevented the
ambulacral from pushing into the interradii. These are possibly inter-
mediate, as far as the mouth goes, between Asterias and Amphiura. In
Amphiura the oral ambulacrals, even in late stages, are kept in the con-
dition of spoon-shaped plates similar in relative position to the water
tube to embryonic ambulacral oi-als of Asterias. The retention of a bi-
serial arrangement of the feet in Asteroids with adambulacral moiiths
may be correlated with the growth of the adambulacrals in early stages.
Like the structure of the mouth parts, the rows of legs are biserial,
as in Ophiuroids.

MUSEUM OF COMPARATIVE ZOOLOGY. 47

Interhrachial Ends of the Oral Amhulacrals. — The interbrachial
ends of the oral ambulacrals in the starhsh are represented in Amphi-
ura by two separate plates, known as the first and second pair of adam-
bulacrals. While in Anaphiura these plates arise from two separate
centres of calcification on each side of the arm (twenty centres in all),
and in Asterias they appear to arise directly from the ambulacrals of the
oral region as a single interhrachial (ten in all), their position and their
relation to the mouth lead me to suppose that they are really the same.
From the existence of a pair of spines on each oral adambulacral it is
necessary to suppose that two plates are consolidated in early stages in
Asterias. The greatest difficulty which has been encountered in sup-
posing the homology which I have indicated as a correct one is the fact
of origin from diff"erent centres of calcification in the one case, and from
the ambulaci-als in the other (Asterias). I am forced to admit from my
study of my preparations that it looks as if the account which is given
above of the origin of the oral adambulacrals from the ambulacrals
(oral) is correct, although an error may have crept in in this observation.
I have no doubt that the adambulacral is single (ten in all rays) from
the beginning. The existence of the spines in twos leads me to look
upon each oral adambulacral as a compound structure.

First Interhrachial. — The homologue of this plate of Asterias in Am-
phiura is one of the most difficult problems connected with the whole
subject of the morphology of the plates of Echinoderms. It is a most
difficult problem to determine what plates in Amphiura correspond with
these plates in Asterias. The term odontophore, as others have already
shown, is poorly chosen to designate these plates, but as the term has
received a signification which is difficult to denote in any other way
at present, it is here retained.

The plates which I have identified as odontophores bear the same
relation to the oral plates as the so called orals in Amphiura. They do
not, however, have a similar origin in the two.

In Amphiura the orals originate on the border of the abactinal hemi-
some, and gradually grow down on the actinal side until they come to
occupy a position relatively the same to the adambulacral as the odonto-
phores to the oral adambulacrals. One of these, according to Ludwig,
is a madreporite.

In Brisinga, according to Perrier, the madreporite is situated on one
of the odontophores. It would seem, therefore, a just conclusion, that
the odontophores and orals are homologous.

If, however, we accept the theory that the orals are homologous with

48 BULLETIN OF THE

the odontophores, we certainly cannot also believe that the genitals of
the startish, interradials of which the madreporic body may be one, are
also homologous with orals. There is a manifest impossibility that orals
can be the same as genitals, and odontophores at the same time. There
must be some mistake somewhere if they are compared to both.

If we examine the observations in regai'd to the fate of the primary
interradial plates (orals) in Amphiura as recorded, proof seems to be
wanting in observation that they do form the orals. We may readily
concede that the madreporic plate may form interradially, and that it
may grow down and form an oral, but is it not a leap at a conclusion
that the other plates in other interradii do the same thing 1 Can we
not suppose, then, that the madreporic plate is morphologically differ-
ent from so called genital plates ] Are we forced to place it in the
same category as other genitals 1 It seems to me that at present we
may say that it is possible that the madreporic plate of Asterias is a
modified homologue of an odontophore which has become consolidated
through the stone canal with a genital, and that it is the same as that
of the oral Amphiura. The orals of Amphiura are, then, the same
as the odontophores of starfishes.*

Dorsals. — The dorsals of Asterias are thought to be homologous
with the dorsals of Amphiura.f They originate in the same relative
position, have the same sequence in development, and to all appearances
are identical. While in Asterias they bear spines and in Amphiura
are destitute of these structures, this fact does not seem of great impor-
tance in showing a want of homology of the two. As the Asterias
matures, the relative predominance in size as compared with other
plates is diminished, while in Amphiura it is increased. It is thus
brought about that in the older stages of Asterias it is more difficult
to recognize the dorsal plates. This results both from the relatively
small development of the dorsals and the appearance of dorsolaterals
and connectives, neither of which are thought to be represented in
Amphiura.

* Tlie mode of growth of the odontophore of Brisinga, as recorded by Perrier,
seems to differ from that of Asterias, — a fact which does not seem surprising con-
sidering the otlier important differences in the two genera. My observations on
Asterias support Sladen and otliers, tliat the odontophores are formed on the
actinal heniisome.

t The dorsals in the young Amphiura were first figured in my paper on the
development of the calcareous plates of Amphiura (PI. IIL fig. 19). They are
rot represented in any of Ludwig's figures, although I believe that they will later
be found in stages younger than his fig. 21, as already pointed out.

MUSEUM OF COMPARATIVE ZOOLOGY. 49

Ventrals. — No plates corresponding with the ventrals of the Amphi-
ura were found in the young Asterias which were studied. In my
account of the development of the plates of Amphiura some difficulty
was found in a comparison of the way in which the ventrals develop
with the way the " embryonic median row of plates " corresponding with
these were formed in Asteracanthion according to A. Agassiz. I was at
that time anxious to study the embryonic ventral plates of Asterias,
and when opportunity occurred took up the subject for this purpose.
I was disappointed, however, for if embryonic ventral plates do exist in
some starfishes other genera must be studied. I found no trace of
them in any of my young larvae of Asterias.*

The ventrals of Amphiura are believed not to be represented in
Asterias. I have already elsewhere adduced evidence which is thought
to be conclusive, as far as mode of formation goes, that the ventrals of
Amphiura are not homologous with the embryonic plate of the middle
actinal line of the starfish ray as described by Agassiz. My argu-
ment then was that the ventrals in Amphiura are unpaired median]
calcareous deposits, while the theory would imply that the so called em-
bryonic plates of the starfish were formed by a coalescence from two cal-
cifications, one on each side (see Agassiz, op. cit., p. 91). As I am unable
to recognize in Asterias the middle actinal row of embryonic spines, it
is not possible for me to find in Asterias homologues of the ventral
plates of Amphiura.

Interambulacrals. — The interambulacrals of the starfish (Amphiura)
are recognized by some authors in the laterals, while others consider that

* It is, of course, possible that the species of Asterias which I studied may not
be the same as tliat in which tlie median actinal "embryonic plates " have been
described. The difference in tlie colors of the females, already mentioned, would
seem another fact pointing to such a conclusion.

t This fact is pointed out by Ludwig, and in the light of his studies it is prob-
able that the figure of Schultze (fig. 6) represents a ventral in the median line of
the under side of each arm. According to Ludwig, the erroneous idea that the
ventrals are originally paired structures has been reproduced by Carpenter (Oral
and Apical Systems of the Echinoderms, Part II., Journ. Micros. Science, Vol.
XIX. p. 21) and Semper (Reisen ini Archipel der Philippen, II. Holothurien,
1868, p. 1G2). Ludwig says that, to show tlie homology of the ventral plate of
Ophiurans with the adambulacral of the starfish Semper instances the fossil Pro-
t'lster Sedgirickii, Forbes, as an Ophiuran with paired ventrals. Semper saw the
difficulty of comparison of the unpaired ventral of most Ophiurans with paired
adambulacrals of starfishes. It seems to me that Ludwig meets the case of Pro-
taster exactly when he says, "Leider ist nun aber Protaster ein nocli so ungenii-
jend bekanntes Fossil dass man dasselbe iiberhaupt als Beweismittel in dieser
Sache nicht gel ten lassen kann."

VOL. XVII. — NO. 1. 4

50 BULLETIN OF THE

the Ophiurans have no iuterambulacrals (adambulacrals). In my paper
on Amphiura the laterals were regarded as adambnlacrals, following
Ludwig and P. Herbert Carpenter. I am not, however, sure that they
are not rather the marginals, and that Agassiz is right in considering
that there are no interambulacrals in Ophiurans. The early form and
growth of the marginals on the border of the arm in both Asterias and
Amphiura is similar. The interambulacrals in starfishes (x\sterias) never
approach in form the laterals of Amphiura. In genera allied to the
Ophiurans, Astropecten, etc., the marginal plates get ophiuroid in na-
ture, and resemble the lateral plates of the young Amphiura. In certain
deep-sea genera of starfishes there is a resemblance between the mar-
ginal plates and the lateral plates of some Ophiurans when dorsals are
not developed. In the light of these facts I must register my doubt
whether I am right in following those who regard the lateral plates of
Amphiura as adambulacral, homologous with the interambulacral (adam-
bulacral) of the starfish. It looks as if true adambulacrals were yet to
be made out in Amphiura. The resemblance of the marginal plates of
the young starfish to the marginal plates of Zoroaster is still another
embryonic feature of deep-water Asteroids.

The following summary may be made of the preceding obser-
vations : —

1. The first plates to originate in the young starfish are the termi-
nals. These plates are simple (not formed by a coalescence of more
than one calcification). They form a protecting cap, shielding the
newly formed ambulacrals, interambulacrals, and possibly margirials.

2. The genitals originate after the terminals. The genital which
lies contiguous to the madreporic opening does not always antedate in
time or excel in size other genitals. It grows around the madreporic
body, or end of the madreporic tube.

3. The madreporic body, madreporite, is a late formation (after the
rudiments of the stone canal),

4. The dorsocentral originates after the terminals and genitals, hefore
any plates on the actinal hemisome.

5. The first plates to form in the body after the genitals have a
radial position in a circle inside the genitals. The second circle is also
radial, and lies inside the circle of the first body, or somatic radials. A
third and inner circle appears before the interradial somatic plate.

6. The first plate in the circle outside that of the genitals is the
first dorsal of the arms. This plate (" Radial," Sladen), when the

MUSEUM OF COMPARATIVE ZOOLOGY. 51

arm of the young starfish is broken from the body, always remains on
the arm.

7. The dorsals, or median roi" of plates on the dorsal surface of the
arm, originate peripherally to the fii'st dorsal ("radial"), and are at
first relatively very large.

8. The dorsolaterals do not aj^pear iu the same sequence as the dor-
sal. (The oldest dorsolaterals may not be the nearest to the bod}-.)

9. Marginals appear after ambulacrals and before (of this there is
some doubt) interambulacrals (adambulacrals).

9. Oral ambulaci'als are the first plates to form on the actinal hemi-
some. When they first appear there are the following calcifications on
the abactinal hemisome : 1st. 5 terminals; 2d. 5 genitals ; 3d. 1 dorso-
central ; 4th. 30 spines on terminals.

10. The oral ambulacrals are at first parallel with the radial culs-de-
sac of the water system. By subsequent growth tliey are placed at
right angles to the same. They are at first ten in number.

11. The interbrachial ends of the oral ambulacrals of adjacent radii
(arms) gi*ow towards each other, forming two parallel ends in each
interradius, of which each bears two spines. The median end of each
oral ambulaci'al bifurcates into a dorsal and ventral part.

12. All other ambulacrals, with the exception of the oral, originate
with axes at right angles to line of radii. They form near the middle
line of the under side of the ray, and grow towards the peripheral.
The adoral are the first formed. They bifurcate in the neighborhood
of the median line.

13. The first interambulacrals (adambulacrals) form after the cor-
responding ambulacrals in the interval between the marginal ends of
successive pairs of the same. Their centre of ossification is from the
first different from that of the ambulacrals.

14. Marginals form before (?) interambulacrals (adambulacrals).

15. The first interbrachials (odontophores) originate as heart-shaped,
interradially placed calcifications, five in number, each abactinally
placed to the interbrachial ends of the oral ambulacrals. In one speci-
men four of these were first observed ; that which lies in the same
interradius as the madreporic tube was retarded in growth. The first
formed interbrachials are not wholly concealed from view, as in deep-sea
Asteroids. When the first interbrachials (odontophores) form, the
following plates have begun to appear:' 1st. 5 terminals; 2d. 5 geni-
tals; 3d. 1 dorsocentral ; 4th. 10 oral ambulacrals; 5th. 20 ambu-
lacrals.

52 BULLETIN OF THE

16. No ventral embryonic row of spines or plates was observed in
the genus and species studied.

17. Genitals of Asterias are thought to be homologous with "basals "
of Amphiura.

18. First interbrachial (odontophore) is homologous with the orals
of Amphiura.

19. Madreporic opening is placed on two homologically different
plates in Asterias and Amphiura.

20. 1st. Genitals; 2d. Dorsocentral ; 3d. Dorsals; 4th. Interarabu-
lacrals ; 5th. Terminals; — represented in Amphiura by, 1st. Basals;
2d. Dorsocentral; 3d. Dorsals; 4th. Laterals; 5th. Terminals. (Ho-
mologous plates numbered the same.)

21. Oral ambulacrals of Asterias are represented by the "spoon-
shaped plates" of Amphiura.

22. " First and second adambulacrals " of Amphiura have no homo-
logue in the mouth parts of Asterias.

23. Ventrals of Amphiura are not represented in Asterias.

24. Dorsolaterals and connectives of the arms were unrecognized in
Amphiura. The homology of the marginals is in doubt, and it is prob-
ably not represented in Amphiura. Possibly the marginals are homol-
ogous with the (adambulacrals) laterals.

Cambridge, January, 1888.

MUSEUM OF COMPARxiTIVE ZOOLOGY.

EXPLANATION OF THE PLATES.

a. Blastopore.

ad. Interambulacrals.

ad^, ad '2, ad^, ad*, ad^. Interambulacrals in sequence.

am. Oral ambulacrais.

am^, am", ain^, am*, am^. Anibulacral rafters in sequence.

amd. Interambulacral ends of the oral ambulacrais.

bn. Brachiolarian notch.

c. Connectives.

cbw. Circular blood-vessel.

cw. Circular water tube.

d^. First dorsal (radial).

d\d^,d*,d'=,d^,d-. Dorsals.

dc. Dorsocentral.

c?f/i, rf(/2, <f(f3. Early formed body plates in radial lines.

dl. Dorsolateral, or lateral dorsal.

/. Feet.

9' 9\ 9~' f> 5^. 9^- Genitals (basals).

9^- Genital near the madreporic opening, through whicli the

madreporic tube opens.

j'. Intestine.

^b- First interbrachial, or odontophore.

ib"^ Second interbrachial.

ibn. Interbrachial notch.

''• Lateral region of the terminal.

m, rn}, m-, m^. Marginals.

mb. Madreporic body.

mt. Madreporic tube

0. Mouth.

0^- CEsophagus.

o™- Oral ambulacrais.

I'd- Pedicellaria.

sp- Spines.

St. Stomach.

s'c. Stone canal.

t, 0, f-, t\ t*, t\ Terminals.

'- *^- Terminal tentacle, or extremity of the medial water tube.

"^- Unabsorbed region of the brachiolaria.

64 BULLETIN OF THE

In Plate I. Fig. 1, t*, t^, g-, g, and g* are seen through the stomacli of tlie
bracliiolaria; ^<i and ^-, </i andjr-', lie on the same side of the brachiolaria as the
observer ; g" and g^ are seen in profile, and lie on the lower surface.

In the figures where the ambulacrals are represented, they are made very much
darker than in nature, in order to illustrate diagramatically their relationship to the
other plates. All the figures were drawn with the camera lucida, reduced in size
in lithography. Where a single arm only is represented, it is to be remembered
that the unrepresented arms are in the same condition of development.

PLATE I.

Fig. 1. View of the posterior (anal) extremity of the body of a brachiolaria of
Asterias, showing the first appearance of the calcifications which ultimately form
the terminals and genitals. The greater part of the brachiolaria is not shown in
this figure.

Fig. 2. A somewhat older brachiolaria, in which the brachiolarian arms are be-
ginning to be absorbed. Tlie dorsocentral (dc) has just appeared. The specimen
from which this was drawn was no longer free-swimming, but was taken witii a
pipette from the bottom of the jar in which the starfishes were raised. (Abactinal
view.)

Fig. 3. A somewhat older starfish in which the brachiolarian arms are still more
reduced in size by absorption. (Abactinal view.) In this stage the radial canals
are seen through the body of the starfish, and the ambulacrals have just begun to
appear.

Fig. 4. The same, slightly younger than the last. This figure shows the plates
only ; the soft parts, with the exception of the brachiolarian arms, are not
represented.

Fig. 5. Lateral view of a brachiolarian younger than Fig. 4, showing position of
the anus, a, and the raadreporic body, mh. The lower part only of the brachiola-
rian is shown.

Fig. 6. Side view of a somewhat older starfish separated from its brachiolaria,
showing the relation of the madreporic body to the first genital, g.

PLATE II.

Fig. 1. View of a young starfish, from the abactinal side, in which the brachio-
larian arms are almost wholly absorbed, and the interbrachial notches have ap-
peared. The body of the starfish and of the brachiolaria is seen through the walls
of the animal.

Fig. 1«. Two rudimentary spines of the terminals in their primary form. The
part represented is one of the interbrachial notches.

Fig. 2. A somewhat older starfish, seen from the actinal side, showing the
mouth, stomach, radial tubes, and plates with spines.

Fig. 3. Young starfish, seen from the actinal surface, with the first ambula-
crals, oral ambulacral, am, well formed. The second pair of ambulacral rafters,
awi^, are just beginning to appear.

MUSEUM OF COMPARATIVE ZOOLOGY. 55

Fig. 4. An older starfish, seen from the actinal side, showing the arrangement
of the plates. The soft parts in this and the former figure have been removed,
bipines are removed from most of the terminals, and appear on one arm only.

PLATE III.

Fig. 1. Single arm of a starfish somewhat older than PI. II. Fig. 4. View from
the actinal side. Three ambulacral rafters have appeared. The youngest, am-, is
very small.

Fig. 2. Somewhat older starfish, in which four ambulacral rafters have formed,
and in which the interambulacrals have begun to develop. View of one arm
from the actinal region. The specimen is sliglitly older than PI. III. Fig. 1.

Fig. 3. View of tlie tip of the arm of a somewhat older starfish, showing the
terminals and the ambulacrals just forming under the protection of the terminal.
The ends of the spines are not represented.

Fig. 4. View of the arm and the plates of the body of a somewhat older star-
fish than Fig. 2. Seen from the actinal side. The outlines only of the plates of
the body are represented.

PLATE IV.

Fig. 1. Abactinal view of an arm of a starfish a little older than PI. III. Fig. 1,
showing dorsocentral, two genitals, the first of the medial dorsal line of plates,
and the terminal with its spines. Ambulacral plates in this and in Fig. 2 shown
through the abactinal system.

Fig. 2. Abactinal view of an arm of a starfish older than Fig. 1. Portions of
two genitals, the marginals, dorsals, and a terminal, are represented.

Fig. 3. View of a starfish from the abactinal region. From a specimen some-
what older than that shown in Fig. 2. Four of the arms liave the plates shown
in outline only. The plates of a single arm are shaded, and the spines are
represented.

Fig. 4. View of a single arm and the plates of the body of a starfish much
older than the last. Seen from tlie abactinal side. Soft parts have been removed.
Spines not represented on genitals and first dorsals.

PLATE V.

Fig. 1. View of the genital and first dorsal (radial) plates of the body of a
young starfisli, of about the same age as PI. IV. Fig. I. All the plates of the arms
removed.

Fig. 2. The same somewhat older, showing the first formation of the plates,
dd^, between the ring of genitals and the dorsocentral.

Fig. 3. Still oldtT starfish, showing the body plates between the ring of geni-
tals and the dorsocentral. View from the abactinal region. Tlie radial lines are
from d^ to dc.

Fig. 4. Much older starfish, showing the network of plates which compose the
abactinal surface of the body. These plates are from a starfish almost an inch in
diameter (from tip of one arm to another).

5G BULLETIN OF THE MUSEUM OF COMPAEATIVE ZOOLOGY.-

Fig. 5. Enlarged view of tlie interbrachial region of a young starfish. The
odontophore, U<, appears as a lieart-shaped plate in the angle of the ray. Actinal
view. Same age as Plate II. Fig. \.

Fig. 6. The same, somewhat older. Actinal view.

Fig. 7. Still older view of the same.

Fig. 8. View of an older stage, in which a second interbrachial, ih"^, has formed.

The shaded plates of Figs. 5, 6, 7, and 8 are oral ambulacrals and ambulacral
rafters. The two crescentic plates, unshaded, in the upper part of Fig. 5, are
tlie edges of the terminals, not marginals, m, of Figs. 6, 7, and 8.

FliWKES Ari'^rJ^IAS

---r^:-

-^fc-

v/ :

-il^i'S^

Pl.L

V

Sfl.

" 'Af'.z--:

Fewkes, Asterias

PL. il.

1»

^

1.

lib.

ant.

Tnh.

dc.

^^

iA.

efr.

J w ;-• del

Fewkes, Asterias

Pi. IiT.

1.

<ut.

-tut.

ad

di

2.

msp-

id'

tta,.

J.V/ F del.

B.Meisel. liOi

FK'vVKES, ASTFRiAS

Pf.

-/■^Pt;-:

'/r.

,„/.

Ul

^

Fewkes Asterias

d-' \

No. 2. — On the Lateral Canal System of the Selachia and
Holocephala. By Samuel Garman.

Soon after his return from the Hassler Expedition, in 1872, Professor
L. Agassiz placed before me, his pupil at the time, a specimen of one
of the Batoidei, with the remark, " See what you can find out about it."
A preparation of the lateral line system was one of the results. This
was followed, under Professor Agassiz's directions, by other preparations
of the same system, which, one after another, were handed over to the
artist of the Museum to be figured. The work was continued thus for
more than a year, without my knowledge of the fact that Dr. B. G.
Wilder had previously been engaged on similar work while assistant of
Professor Agassiz. And it was not until about 1883 that the Director
of the Museum desired me to prepare for publication my own material,
and that accumtilated under Professor Agassiz's directions. It was
found that, in order to use the manuscript left by Dr. Wilder, many
changes would be necessary, and it was thought better on the whole not
to attempt to incorporate it with my own. Dr. Wilder's dissections have
been used as far as possible in the descriptions. They were figured by
Mr. Roetter, and included a representative of each of the following
genera : Scoliodon, Prionodon, Mustelus, Triads, Isurus, Odontaspis,
Alopias, Ginglymostoma, Scylliorhinus, Heterodontus, Acauthias, Rhina,
Pristiophorus, Pristis, Raia {R. Icevis), Dasybatus (Z). tuherculatus) ,
Pteroplatea (^P . valencieiuiii) , Myliobatis (31. fremiyivillei and M. aquila),
Aetobatus, and Rhinoptera {JR. hrnsiliensis and R. jussieui). In addition
to these there were preparations of several types of which no use has
been made, as they had been duplicated in my own work. Chimsera
and Callorhynchus required no dissection ; a drawing of the latter had
been made by Mr. Roetter. The figures by this artist were made to
be lithographed, and were not at all suited to the engraver's process,
by which this publication was to be illustrated. Consequently outlines
have been used instead of his drawings.

This leaves me responsible for all the text, and for the dissections
and sketches of Dicerobatus, Pteroplatea (P. liirundo and P. marmorata),
Dasybatus (Z). nudxis and D. dipterurus), Teeniura, Urolophus, Disceus,

VOL. XVII. — NO. 2.

58 BULLETIN OF THE

Potamotrygon, Narcine, Torpedo (T. calif arnica and T. marmorata) ,
Eaia {R. ocellata), Uraptera, Syrrhina, Ehinobatus, Somniosus, Chlamy-
doselachus, Heptabranchias, and Cestracion (Zygsena) ; also for sketches
of Chimaera. Callorhynchus, and Pristiophorus ; and for the outlines
and sketches of such as had been drawn by Mr. Roetter.

With a few exceptions the names of the canals are those adopted by
Professor Agassiz.

The structures to which attention is here directed are those on the
Selachia and Holocephala, which correspond to the lateral lines of
the Fishes. On the individual they form a system of branching
canals, or tubes, tubules, and branchlets, which has received a vari-
ety of names at the hands of different writers : slime canal, mucous
tube, water canal, lateral line, etc. To avoid confusing with the
unbranched hyaline mucous ducts of the ampullae of Lorenzini, the
term canal will in this paper be applied exclusively to the organs com-
prised in the system under discussion. The manner in which the tubes
branch and connect, and the fact that they are sometimes represented by
mere furrows in the skin, make this designation the more appropriate.
In addition, since it has been pretty well established that their func-
tion is that of very delicate tactile organs, receiving and carrying the
slighter vibrations of the water, noting changes of density, currents, etc.,
a special name, Tremognosters, to be applied to these particular canals,
distinguishing them from the many other canals of the body, is intro-
duced as likely to prove still more convenient.

These canals, or tubes, lie in or under the skin on both body and
head. They open externally either as furrows or by means of pores,
that in some cases enter the tubes directly and in others are approached
by tubules. The inner layers of their walls are furnished with series of
nerve-endings, which, as also the external openings of tubular canals,
are segmentally arranged. In difierent pai-ts, the structure of the walls
varies from fibro-cartilaginous, on the top and sides of the skull, to
very delicate transparent tissue, under the snout. Granulation, appar-
ently resulting from calcification, occurs in the cephalic tubes of cer-
tain genera. There are no glandular attachments ; the vessels are
simply canals, open at each end, and more or less so along their sides.
On the head they are innervated mainly from the fifth pair, and on the
bodv bv a branch of the vagus, the nervus lateralis.

The development of the system, as worked out by Balfour in Scyllium,
coincides closely with that of the Teleostei. According to Beard the
lateral line in the embryo salmon first appears opposite the hyoid arch,

MUSEUM OF COMPARATIVE ZOOLOGY. 59

a little behind the ear capsule, on the level of the notochord ; and it is
formed by the splitting off of some of the cells of the inner layer of the
epiblast. From its point of origin, where it is broadest, it grows back-
wards along the body. This cord of cells, as Beard calls it, is no doubt
what Balfour describes, in Scyllium, as a linear thickening of the mucous
layer of the epidermis, or as a linear streak of modified epidermis. This
linear sense streak is in Balfour's opinion the primitive structure from
which the various forms of the line have originated. He says, further,
that the thickened streak becomes a canal in Scyllium, not by the fold-
ing over of the sides, as in Teleostei, but by the formation of a cavity
between the epidermic and the mucous layers of the epiblast, and the
subsequent enclosure of this cavity by the modified cells of the mucous
layer of the epiblast constituting the lateral line. The cavity appeared
first at the hinder end of the organ, and thence extended forwards.
After formation the canal gradually recedes from the surface, retaining
its connection, however, at a series of points corresponding to the seg-
ments, points at which the segmental openings are afterward formed.
As compared with the open canal on Chimcera, the tubular canal is a
secondary form. In regard to the innervation on the trunk by the
lateral branch of the vagus, the nerve was found to originate as the
other nerves, and, pushing its way backward, to follow the course of
the lateral line. Originally the line is supposed to have been restricted
to the anterior part of the body, and thence, extending farther and
farther backward, it carried with it the lateral branch of the vagus,
until ultimately the latter was prolonged far beyond the position it
originally occupied. Beard says of this nerve, in Salmo, that it origi-
nates far from the epiblast and growing backward approaches the skin
so as to lie between the two muscle plates just below the epiblast, never
fusing with the lateral line, but always separated from it by the cuticular
basement membrane of the epidermis.

Balfour found the canals of the head and the ducts of the ampullce
to be formed from the mucous layer of the epidermis, very much as
the lateral line ; but their innervation is effected by simple branches
of the fifth and seventh pairs, which reach them in various places
without following their courses, unlike the nervus lateralis and the
lateral line.

Primarily the openings at the ends and along the sides of the tubes
appear to have been in close relationship with the segments of the body,
both in regard to position and number. The relations are still apparent
in the numbers and in the positions of the tubules at their points of

60 BULLETIN OF THE

junction with the main tubes. But in many species, through the
descent of the canals below the skin, the consequent elongation of the
tubules, and the multitudes of branchlets by which they communicate
with the surface, the primary arrangement has come to be greatly ob-
scured. Alopias illustrates this to some extent in the Galei ; and in the
Batoidei instances are numerous in the Trygonidse, the Myliobatidae,
the Zygobatidee, and the Ceratopteridae.

The tubes contain a thin mucilaginous liquid. This is probably for
the most part an excretion, and not an absolute necessity in connection
with the function of the system, except, it may be, in so far as it serves
the purpose of lubrication. Its retention is hardly possible in the open
grooves of various genera, on which the office of the organs is undoubt-
edly the same. In discussing the purpose of the liquid, one must bear
in mind those Teleosts in which the sense bulbs open directly on the
epiderm, without either groove or tube, and the likelihood that they
represent the primitive condition of the system from which the furrows
and the tubular canals have been developed.

Absence of the mucous secretion on the skin of species well provided
with canals precludes consideration of the opinion that the object of the
latter is to cover the surface with slime.

Series of the follicles in immediate connection with the subrostral
canals of certain species lead to the conclusion that the nerve follicles
of Savi are really obsolescent tubes of the canal system, in which the
section that forms the enclosui*e or follicle owes its persistence to the
presence of the contained nerve. In other words, the follicles repre-
sent vanishing and rudimentary tubes. From this it would seem as
if Potamotrygon, Disceus, and Urolophus, among others, may be on
the way to lose the canals of their ventral surfaces, as has already
happened in the cases of Torpedo and Narcine.

The hyaline mucous ducts of the ampullae are unbranched, have but
a single aperture, are closed at the inner end, where entered by the
nerve, and are filled by a jelly-like mucous. Plate XXVIII. fig. 1,
represents a portion of those of Eaia Icevis, and their distribution as
compared with that of the principal canals, fig. 2.

GALEI.

On the Sharks the canal system consists of a vessel on each side of
the vertebral axis, extending from the snout to the tail, connected with
a similar vessel on the opposite side by a transverse branch near the

MUSEUM OF COMPARATIVE ZOOLOGY. 61

occiput, and sending another branch oetween the eye and the spiracle
toward the mouth. Between the lip and the eye, on the lower aspect,
the latter branch sends another backward past the angle of the mouth,
and, farther on its way forward, sends still another behind each nostril,
itself usually joining the main canal at the end of the snout. Com-
monly a branch from that passing the angle of the jaws extends under
the jaw, behind the mouth, toward the middle of the chin ; in some cases
this branch is disconnected ; in others it does not appear ; occasionally
it is continuous across the symphysis. The branch behind the nostril
passes toward a similar one from the opposite side of the head, either
uniting with it or approaching it closely for a short distance under the
base of the rostral cartilage, after which the two diverge slightly and
continue forward to the end of the rostrum. On the top of the skull
the tubes are more or less strongly attached to the cartilage, in troughs
or depressions in which they are often deeply seated. Under the base
of the rostrum there is also a firm attachment to the cartilage. Else-
where the canals lie at varying depths in the skin or below it. Gener-
ally they are tubes with openings to the surface through simple to
many-branched tubules. On species of Heptabranchias and on Chlamy-
doselachus they appear, in great part of their extent, as open dermal
grooves. On one of the genera of the Holocephala they are open
furrows, on the other they are tubes.

NAMES OF THE CANALS.
Plate 1.

The most convenient designations for the different canals, or parts of
canals, are those derived from the names of the portions of the body
traversed by them, or from those of the organs near which they pass.
The propriety of the names cephalic for all the canals of the head, and
corporal for all those of the body, is at once apparent. Their position
along the flanks makes the name laterals (J), by which they are c:ener-
ally known, a very appropriate one for the two main corporals. In the
Batoidei there is a canal peculiar to them extending out upon the pec-
toral fin; this may be called the pleural (p). On the lower surface
it becomes a subpleural. The areas enclosed by the pleural tubes are
the pleural areas ; those sometimes enclosed by scapular tubules on
the shoulder are called the scapular areas.

Running longitudinally on the top of the head are the two principal
cephalic tubes, the cranials (cr) ; anteriorlv, on the rostrum they be-

62 BULLETIN OF THE

come the rostrals (r), and after passing below the snout they are known
as suhrostrals (sr). At the end of each cranial, on the crown of the head,
an orbital (orb) canal runs outward behind the orbit ; below the eye,
and below the disk in the flat-bodied Selachians, they become the sub-
orbitals (so). A transverse tube from one lateral to the other, close to
the external openings of the aqueducts, ear openings as commonly
named, is the aural (au). A short occipital (oc) reaches from the aural
to the orbital ; or, in other words, from the lateral to the cranial.

On the ventral surface the canal passing lengthwise near each angle
of the mouth is the angular (ang). The portion to which this name
is applied is not a long one ; farther back on the same tube, the name is
supplanted by that o^ jugular (j). In some cases these canals are
definitely limited by a branch, the oral (o), putting out transversely
behind the mouth; but very often the oral is found to have lost its con-
nection with the other tubes. Earely the oral is continuous across the
symphysis. In front the angular meets either the suborbital or a canal,
the nasal (71), extending behind the nostril, between the latter and the
mouth ; sometimes the one is met, sometimes the other ; whichever it
may be, it marks the anterior limit of the angular. The two nasals
meet in front of the middle of the mouth, in most cases, and, unit-
ing, form a short median (m), from which two other tubes, the pre-
nasals (p«), diverge and run forward to the end of the snout. On a few
forms the nasals do not meet. In some instances there is a junction of
subrostral and nasal ; in others, the subrostral joins tJie suborbital ; in
one species the angular and the nasal join, in another it is the angular
and the suborbital ; but however the junctions may be arranged, a tube
of less or more extent lies between the end of the orbital and that of
the nasal. Its position is indicated in its name, orhito-nasal {on).

BATOIDEI.

If one of the round-bodied sharks were to be greatly depressed and
flattened, extension taking place on both dorsal and ventral surfaces,
the pectoral fin at the same time being expanded and applied to the side
of the trunk, the arrangement of the main tubes of the system would
be similar to that obtaining on the Batoidei. The subrosti-al, nasal,
prenasal, angular, jugular, and suborbital would appear on the lower
surface, as in the Skates and Rays. An important addition to what has
been recorded in the Galei occurs in the Batoidei : a pleural canal {jil),
which meets the lateral at the shoulder, runs outward on the pecto-

MUSEUM OF COMPAEATIVE ZOOLOGY. 63

ral fin, then forward, descending near the head, and, after a backward
course of varying extent on the lower surface, unites with the jugular.
Or, reversing the direction and starting below, from the jugular, the
canal goes out and forward under the pectoral, ascends at the side of
the head, then turns out and backward, describing a circuit toward the
margins on the top of the fin, and unites with the lateral at the shoul-
der. Most frequently it is the case that the pleural and the orbital
are connected by orbito-pleural tubules ; exceptionally these tubes meet
directly without the intervention of the tubule. No doubt the pleural
originated as a branch of the orbital. Besides this pleural canal on
the pectoral, there are usually present several others, post-pleurals,
from the scapular curve of the lateral toward the hinder part of the
fin, which also are not represented in the Galei. Ordinarily the upper
pleurals are abundantly supplied with tubules ; sometimes on the lower
surface tubules are entirely absent ; and on the Torpedoes the ventral
portion of the entire system is obsolete. Branches of tubules are gen-
erally in pairs ; a tubule forks to form a pair ; each of this pair forms
another pair in similar manner, and so on. This dichotomous branch-
ing of the branchlets may be kept up, as in the higher Rays, until on
reaching the outer layers of the skin a considerable space is occupied
by the mat or rosette formed of the very small tubes and their pore-like
openings on each of the tubules. Among the Torpedoes and the Skates
the simple unbranched tubule is the common form.

The origin of the pleurals of the Batoids, or the manner in which the
group became possessed of these canals, in addition to those possessed
by the Galei generally, is a question of considerable interest. Our only
clue to the solution of the problem is to be seen in Chlamydoselachus.
If the head and body of this shark were depressed, and the pectorals
expanded and applied, so as to produce the skate-like form, the spiracu-
lar canal would then extend back along the basal cartilages of these
fins, they being attached above the gill openings, thus forming the
pleural canal, the union of which with the scapular branch of the lat-
eral is only a secondary matter, as shown by the variety in modes of
junction, in the Rhinobatidse, the Raiidse, and such genera of Trygoni-
dse as Urolophus and allies. On the lower surface the subpleurals
would be supplied by the gular and the canal lying between it and the
lip, the oral being limited to the part anterior to their point of meet-
ing. The fact that the additional canals would be acquired in this way,
as a necessary consequence of the change of the form, leads, at the
least, to a strong presumption that the Batoidei are indebted for their

64 BULLETIN OF THE

pleurals and subpleurals to a Galeoid ancestor resembling Cblamy-
doselacbus as far as the possession of spiracular and gular canals is
concerned, if not further. And indirectly the tendency of such consid-
erations is to confirm the claim elsewhere advanced that that genus is in
great measure to be regarded as a persistent type.

HOLOCEPHALA.

The great differences between Chimaera and Callorhynchus in regard
to rostral appendages and claspers, are iu reality no greater than those
obtaining in their canal systems. Greater divergence than occurs in
these genera is not to be seen in the most dissimilar forms of the Sharks.
On Chimaera the canals are furrows, as on the body of Chlamydoselachus,
and the oral meets the angular ; on Callorhynchus the canals are tubes,
and angular, oral, and jugular meet the suborbital independently. At
the first glance, the differences in the distribution of the cephalic canals
in the two genera appear greater than they really are. On comparison,
the positions of laterals, aural, occipitals, cranials, and orbitals are found
to be similar. In both cases the oral and the jugular cross the median
line as series of pores or short grooves, the suborbital extends to the
end of the snout, the subrostrals unite under the rostrum to form a
median, then separate to meet the nasals, and the nasals are in front of
the nostrils, meeting across the middle without forming a median canal
or prenasals. It may be added, that in both types the lateral descends,
above the lower lobe of the caudal fin, to the lower edge of the muscles,
as in certain of the lower Galei.

Affinities with the Sharks, through ancestry, are indicated by the
correspondence in laterals, aurals, cranials, orbitals, angulars, and orals.
Special points of disagreement are seen in the union of the jugulars, the
prenarial location of the nasals, the absence of prenasals, the presence
of a median in the subrostrals, and in the connections of the occipitals.

COMPARISONS.

Whether the canal system is a suitable basis for homology and classi-
fication, either alone or in connection with other parts of the anatomy,
and its importance as such a basis, are to be determined by consid-
eration of the extent of its development and the amount of its va-
riability in the difi'erent types included in the class. An exhaustive
investigation of the subject would naturally demand a study of the

MUSEUM OF COMPARATIVE ZOOLOGY. 65

system in its relations to the general structure and to the habits of
the species, and, through the latter, to the surroundings and to their
influence upon its evolution and variation. Direct opportunity for
much of this is not within reach ; but from the material at hand
it may be possible to make approximatious that at least will be
tolerable.

Possession of the organ is quite general ; no exception has yet been
discovered either in the Selachia or in the Holocephala. A stage of
development of the system that is comparatively simple exists on those
forms usuallv called the lower, and on the course from them to the
highest the amount of complexity is found to correspond well with the
rank as indicated by the brain or other parts of the organization.
Between Heptabranchias and Alopias of the Galei, and between Pristis
and Dicerobatus of the Batoidei, each step is marked by variation in
contour and in the extent and complication of the system of the canals.
In the Holocephala a blow is apparently given to the idea that the
groove is the lowest, the primitive form of canal, by the fact that Callo-
rhynchus possesses tubes and not furrows like those of Chimsera. This
may be no more than an indication that the former is the most differ-
entiated type, the higher in rank. Bony Fishes, also, possess tubes.
The fact remains that it is among the lower forms of the Sharks and in
Chimaera that the grooves obtain. Furrows are unknown in the canals
of the Batoidei ; and it is in this order that the greatest degree of devel-
opment is attained by the system. Dichotomization of the tubules ap-
pears in the higher, and becomes excessive in the highest, forms of either
order. Types known to be sluggish in their habits are less abundantly
supplied with tubules, and the system is not so complex as on the more
active. It needs but a contrast of the Raiidae and the Myliobatidae
to make this obvious. Forms which have changed their habits and
become more addicted to resting on the bottom give evidence of the fact
in the gi'adual deterioration and disruption of certain canals on the
under portions of the body. That the canals are rather less subject
to variation, that is, that they respond less quickly to its causes, than
certain other organs, is intimated by the results of a comparison of
the species of a single genus. Close genetic relationship is asserted
by the canals of such species as Dasybatus walga and D. nudus, or by
Uraptera agassizii and Raia Icevis. Young specimens, or embryos, often
serve better as guides to descent and affinity than old ones, since canals
are present on the former which in some instances can hardly be found
on the older ones. In the embryo the canal system takes on its generic

VOL. XVII. — NO. 2. 5

66 BULLETIN OF THE

and specific peculiarities long before the characteristics of the outer skin
are acquired, even before the shapes of body and fins, and the peculiar
dentition, or the other features commonly relied upon to separate the
forms and groups, have become available. Consequently, the various
embryos may be recognized by means of the canals at j .iods when
identification by the specific and generic characters ordmarily .n-
ployed for the purpose of distinguishing them would be quite out of the
question.

Dichotomized tubules do not appear to any great extent on the lower
surfaces of such as habitually lie on the bottom. The tubules which
occur on the ventral canals of such species most often have their ex-
ternal apertures at the border of the disk. On the Torpedinidae it is
altogether likely that disuse has led to the loss of the entire ventral
portion of the system. Types addicted to flights through the water at
a distance from the bottom have the tubules and their branchlets more
alike on the upper and lower surfaces, as may be seen in such as the
Zygobatidoe and the Dicerobatidae. These and similar Rays have a
greater aggregation of the tubules and branchlets toward the hinder
portion of the disk, a distribution which suggests liability to danger
from behind, possibly while the creatures are feeding. The modifica-
tions in position and outline which the cephalic canals have under-
gone on these families, through changes in shape of head, snout, and
pectoral fins, and through change in the position of the mouth, become
very prominent when directly contrasted with the same canals on such
as the Trygonidae or the Raiidse.

The extent to which the canals may be used in classification is best
illustrated by comparing the systems on the various species or groups.
Necessarily the comparisons instituted below have been made very
much as if the genera were composed of the species under examination.
Further investigation of other species will, no doubt, bring to light dif-
ferences in matter of detail, serviceable in specific diagnoses, and possibly
such as may compel modification of our ideas of the generic charac-
ters ; but the attempt has been made here to use only such features as
are least liable to the minor variations. As the vessels have not yet
been studied on one sixth of the whole number of species in the class,
and as those on which the system has been worked out do not include
representatives of all of the genera, it follows that a synopsis con-
structed on the material here gathered could only be a temporary affair.
For this reason a short summary of differences is to be preferred to a
synopsis giving only a few of the more prominent ones.

MUSEUM OF COMPARATIVE ZOOLOGY. 67

A connection of the cranials, instead of the laterals, by the aural,
and the passage of the jugulars across the chest, at once separate the
Holocephala from both the Galei aud the Batoidei. The Batoidei are
separated from the Galei by the possession of the pleural canals.

Chimsera, in the Holocephala, is marked by the grooves, instead of
tubes, and Callorhynchus by the tubes, instead of grooves.

Among the Galei, on the base of the tail the lateral canals descend
to the lower edge of the muscles in Chlamydoselachus, Heptabranchias,
Heterodontus, Pristiophorus, Acanthias, and Somniosus, as in the Holo-
cephala. Open corporal canals resembling those of Chimsera appear on
Chlamydoselachus, Heptabranchias maculatus, and, in part, on Acanthias.
On other genera the laterals maintain their position near the vertebrae
of the tail, and the canals are tubular. On Scoliodon, Mustelus, and
the Hammerheads the lateral makes a decided bend below the second
dorsal fin ; and it ends at or near the end of the vertebral column in
Scoliodon, Triacis, Mustelus, Odontaspis, Scylliorhinus, Chlamydose-
lachus, Ginglymostoma, Cestracion, and Somniosus, not reaching so
far back in others. Disregarding the course of the column in Isurus, it
passes directly backward, ending at the edge of the muscles just above
the lower lobe of the caudal fin.

The aural is behind the "ear openings," and more or less curved back
in the majority of the Sharks ; it is in front of the openings in Chlamy-
doselachus, bisected in Heptabranchias, curved forward in the middle
in Acanthias and Chlamydoselachus, and nearly straight in Mustelus,
ScylUorhinus, Heterodontus, and Somniosus. Sometimes, as in Pristio-
phorus, it is deeply curved backward, much as in the Holocephala, or
in Dicerobatus.

The occipitals appear like continuations of the laterals, so slight is
their change in direction, in Acanthias, Rhina, Heptabranchias, and
Chlamydoselachus ; others have the tubes directed more or less obliquely
outward.

Somniosus is peculiar in that cranials, orbitals, and occipitals do not
meet on the crown.

On the frontal region the cranial curves are shallow in Prionodon,
Alopias, Isurus, Heterodontus, Acanthias, Somniosus, and Pristiophorus;
decided in Scylliorhinus, Mustelus, Triacis, Ginglymostoma, and Rhina;
more decided in Chlamydoselachus and Scoliodon ; and excessive in
Cestracion (ZygEsna).

A majority have the orbital bent forward in its lower portion ; in
Cestracion, Heptabranchias, and Chlamydoselachus it bends backward.

68 BULLETIN OF THE

In the suborbital, Scoliodon and Prionodou have a curve that reaches
upward in front of the orbit. This curve goes farther forward than the
eye in the greater number of the Sharks ; it lies under the orbit in
Isurus, Alopias, Cestracion, and Acanthias ; it is absent in Hepta-
branchias, Somniosus, and Chlamydoselachus ; and it goes forward of
the nostril in Ginglymostoma.

Between the nostril and the median, in much the greater number of
cases, the nasal canal is bent forward ; this bend is either absent or
faint in Isurus, Odontaspis, Alopias, Heptabranchias, Chlamydoselachus,
Scylliorhinus, and Ginglymostoma. On Isurus the nasals meet the
angulars ; on others, as on certain Batoids, they meet the subrostrals.

A union of the nasals in front of the mouth on most of the sharks
forms the median ; no such union takes place on Acanthias, Hepta-
branchias, Chlamydoselachus, or Pristiophorus.

Xo connection between prenasals and rostrals appears in Heptabran-
chias and Somniosus.

Most often the jugular is directed toward the middle of the first
branchial aperture ; Alopias differs in having this tube pass below the
gill opening.

The oral is continuous behind the mouth in Ginglymostoma ; it does
not meet the angular in Scoliodon, Prionodon, Triacis, Mustelus, Scyl-
liorhinus, Acanthias, and Rhina ; and it was not found in Heptabran-
chias, Somniosus, and Pristiophorus.

A general characteristic of the Batoids is seen in the pleural canals.
At once on passing from the Pristiophoridae of the Sharks to the Pris-
tidse of the Skates these tubes become prominent features.

The Pristidae are affected by an excessive elongation of the rostrals
and prenasals. Their pleurals are comparatively short, extending but
little on the pectorals. The scapular branches are few, but one, a post-
pleural, being present in the species sketched.

On the Rhinobatidge post-pleural branches are more numerous. In
general there is considerable resemblance between this family and the
preceding. The anterior cephalic canals are shorter, and there is a
sternal canal below the coraco-scapular arch.

All the Raiidfe are marked by greater extension of the upper pleu-
rals on the. pectorals. A strong branch extending back on the middle
of these fins is apparently common. The prominent narrow fold in the
subrostral varies in the different species : in Uraptera the fold has
been so much narrowed as to bring its sides together. On the ventral
surface of Raia ocellata the pleurals are obsolete.

MUSEUM OF COMPARATIVE ZOOLOGY. 69

The Torpedinidae have lost the canals of the lower surface. Rem-
nants of the missing vessels are found in the follicles of Savi, present
ou some. The pleurals and the orbitals unite directly, without the
intervention of tubules. In Narcine the entire extent of the system is
much less than in Torpedo, owing to the reduction in number and length
of the tubules. Yet in these respects there is not a little variation in
the species of Torpedo, as is seen by comparison of T.' calif ornica and
T. inarmorata.

In the Trygonidae, as, further along, in the Myliobatidse and their
allies, we see a disposition to enlarge the system by means of curves,
tubules, and dichotomous branchings much beyond what has been no-
ticed in the Skates. Through the increase in length of the main tubes
the courses of orbitals and suborbitals have come to be crossed by the
pleurals on both upper and lower surfaces. The presence of a small
enclosure, or more than one, on each shoulder, formed by scapular
branches, pre-pleural or post-pleural, or both, is somewhat general in this
section of the Batoids.

Potamotrygones as well as Thalassotrygones have the tubules of the
pleurals on the lower surface massed anteriorly, comparatively few
appearing under the posterior half of the disk. An obsolescent condi-
tion of the subrostrals obtains in Disceus and Potamotrygon ; where
parallel with the prenasals these tubes are merely lines of follicles,
without apparent connection by their cavities, marking the paths of the
canals. On the lower portion of the pleural of Potamotrygon there are
rather few tubules ; the sections of the oral are elongate and sinuous ;
the nasal meets the angular, and apparently there is a short sternal
canal. On Disceus the tubules are very numerous on the lower pleu-
rals, the parts of the oral are short and separated by some distance,
the nasal and the subrostral meet, and there is an orbito-pleural plexus
containing a large number of small areas. Differences similar in char-
acter, but less pronounced, exist on the upper surfaces of these genera.

Urolophus has no orbito-pleural plexus on the lower surface, its pleu-
ral tubules are not massed in front, and the suborbital is not provided
with a long loop pointing forward as in Potamotrygon. It has a short
sternal tube.

Tseniura resembles Urolophus more than it does the Potamotrygons.
Like the former it has the pleural tubules distributed along the tube,
and it has neither orbito-pleural plexus nor suborbital loop. It differs
from Urolophus in the multitude of its branchlets on the upper aspect,
in its pleural areas, and in the union of subrostral and nasal.

70 BULLETIN OF THE

Dasybatus has a considerable part of tlie anterior portion of the lower
pleural close to and parallel with the anterior edge of the pectoral.
Along this section the tubules are numerous. Backward they are dis-
tributed sparingly, if present. The features possessed in common by
the various species are best seen on the dorsal surface, since the ten-
dency toward variation has been much more active beneath the disk.
The suborbital alone is sufficiently diflferent to distiuguish the three
species figured : in D. nudus it is excessively elongate, it encloses a
peculiar series of spaces, and twice, in a couple of long reaches, it
comes abruptly to an end ; in D. dipterurus it is rather short and
somewhat sinuous ; and in D. tuberculatus it is greatly lengthened by a
complex series of zigzag folds or convolutions.

Pteroplatea differs greatly from the other genera in the distribution
of tubes and tubules on the dorsal surface. They are arranged to
reach the margins around the entire pectoral, and, though numerous
posteriorly, the branchlets are matted in multitudes in front. On
P. valenciennii the tubules are most plentiful, on P. marmorata less
abundant, and on P. hirmido still more scattered. These species dif-.
fer in the branchings and areas on the scapular region, as also in the
general arrangement and abundance.

The Myliobatidse, through narrowing the pectorals at the side of the
head, have had pleurals, orbitals, and rostrals brought close together
under the orbit. Those types which have the fins most reduced, or
absent, in this location, have these tubes arranged almost vertically one
above another on the side of the face. This family and the Zygobati-
dse agree in this respect ; they also agree in restricting the dorsal
canals to about half the distance from the vertebrae to the outer angle
of the pectoral, in the arrangement of the lower pleurals in pairs of
lines along the anterior margins or along the branchial and the abdomi-
nal areas, and in having a large portion of the oral longitudinally ex-
tended, as if folded with compression of the head, among the more
noticeable peculiarities. We might also have included the Dicerobatidse
in the majority, if not in all, of these agreements. These families are
readily separated by means of the cephalic canals. Forward from the
fontanelle in the Myliobatidae the orbitals cross the rostrals, a position
they do not attain in any of the lower families. Myliohatis freminvillei
has longer canals on the rostrum and a less number of pleural tubules
than M. aquila, and it has the subrostral separate from the prenasal.
M. aquila has the shorter anterior cephalic tubes, the greater number
of tubules on either surface, and has the subrostral joining the prenasal

MUSEUM OF COMPARATIVE ZOOLOGY. 71

near the median. Aetobatus agrees with M. aquila in regard to junc-
tion of subrostral and prenasal ; it diflfers from both of the sketched
species of Myliobatis in such characters as would be more likely to be
induced by greater compression of the head, as is seen in the foldings
of the oral, deeper curvature of the aural, etc.

Zygobatidse. — In this family the mouth, as compared with that of
the Myliobatidae, has been drawn backward closer to the gill openings
or the coraco-scapular, and the snout, through reduction, and retraction
to follow the mouth, has arrived at a position beneath the skull. The
process of the change is well written in the foldings, sinuosities, and
contortions of the cephalic tubes. Besides these particulars of charac-
terization there are others, apparently resultant from a shortening of
the longitudinal axis of the body, equally peculiar in this group. Of
these are the scapular and posterior pleural foldings. The species
figured differ in the number of cephalic tubules, in the scapular areas,
in the post-pleural folds, and in the oral, which is divided in one
species, united in the other.

Between the Dicerobatidae and the Zygobatidae there are more
points of resemblance than between the former and the Myliobatidae.
There is more of a disposition to mass the pleural tubules posteriorly
than in either; and the connection of the laterals across the vertebrae
has not before been noticed in any of the Batoids. A further peculi-
arity occurs on the inner anterior section of the lower pleural, on which
the tubules turn backward, as in Myliobatis freminvillei. The latter
is no doubt only a specific character. Of the cephalic canals, it is
hardly necessary to say anything, the distinctions arising from the
peculiar shape of the head are so excessively marked. Yet, as is
noted in the description of D. olfersii, the affinities existing between
Dicerobatus and Ehinoptera are shown by the canals of both body
and head.

The subjoined synontic list furnishes a condense! illustration of the
availability of the canal system in classification. Being based exclu-
sively on the species here described and figured, some of them repre-
sented by single specimens, it is to be expected that study of new forms
will necessitate modification and rearrangement.

72

BULLETIN OF THE

CHONDRICHTHYOIDEA.

With the nasal canals in front of the
nostrils ;
cranial meeting the aural

HOLOCEPHALA..

canals sulcate Chimcera monstrosa.
canals tubular Callorhynchus antarc-
ticus.
With the nasal behind the nostrils ;
cranial not meeting the aural

Selachia.
without pleural canals Galei.

with pleural canals Batoidei.

Galei.

Oral not connected with angular ;
median longitudinal ;
cranial curves abrupt, deep

Scoltodon terrce novce.
cranial curves shallow ;
an upward anal curve ;
jugular reaching toward upper
edge of gill opening;
anal curve prominent

Mustelus canis.
jugular reaching toward middle
of gill opening ;
anal curve low

Triads semifasdatus.
jugular curved downward

Scylliorhinus caniculus.
anal curve indistinct;
nasals not meeting

Acanthias americanns.
suborbitals and subrostrals on
top of snout Rhina squatina.
median transverse ;

anal curve absent Prionodon milbertii.
Oral connected with angular ,
divided at the symphysis ;
cranials folded over themselves

Cestracion tiburo.
cranials not folded ;

suborbital and angular meeting;
median transverse ;
jugular passing below gill
opening Alopias vulpes.

median longitudinal ;

jugular passing toward middle

of gill opening

Odontaspis americanns.

jugular short, passing toward

upper edge of gill opening

Heterodontus philippii.

spiraculars and gulars present

Chlamydoselachus anguineus.

not divided at the symphysis

Ginglymostoma cirratum.

suborbital and subrostral meeting

Isurus punctatus.
Oral absent ;

aural divided ;

nasals not meeting

Heptabranchias maculatus.
aural entire ;

nasals not meeting

Pristiophorus cirratus.

nasals meeting Somniostis carcharias.

Batoidei.

Subpleurals in front of gill openings ;
pleural descending on edge of disk

Pristis perfinatus.
pleural passing through disk ;
sternal divided ;
rostrals and prenasals long

Rhinobatus planiceps.
sternal entire ;
rostrals and prenasals short

Syrrhina brevirostris.
Subpleural and other ventral canals ab-
sent;
tubules very short Narcine brasiliensis
tubules medium ;

aural tubules very short

Torpedo marmorata.
tubules long ;
aural tubules long

Torpedo cali/ornica.

Subpleurals at side of gill openings ;

tuDules simple ; a long post-scapular

tubule ;

orbito-nasal at right angles with

pleural Uraptera ayassizii.

\

MUSEUM OF COMPARATIVE ZOOLOGY.

73

orbito-nasal parallel with pleural

Raja Icevis.
subpleurals partly obsolete

Raja ocellata.
tubules with dicliotoraous branchlets ;
pleural tubules distant from lateral
and hinder margins ;
subpleural tubules massed in front
of head ;
subpleurals and suborbitals form-
ing a network ;
subrostral and nasal meeting
Disceus strongylopterus:
no subpleural network ;
no lateral tubules on sub-
pleural ;
subrostral and suborbital
meeting

Potamotrygon tnotoro.
lateral tubules on subpleural ;
subrostral and suborbital
meeting

Urolophus halleri.
subpleural and suborbital
meeting, very tortuous
Dasybatus tuberculatus.
subpleural and suborbital'
not tortuous ;
orbitonasal a mere point
Dasybatus dipterurus.
no lateral tubules on subpleu-
ral ; tubules massed in
front;
suborbital forming a series
of four incomplete areas'
Dasybatus nudus.
pleural tubules reaching outer pec-
toral angles ;

areas included by pleurals wider
than long ;
lower pleurals emerge two sev-
enths way from the median
to tip of snout

Pleroplatea valenciennii.
lower pleurals emerge two fifths
way from median to tip

Pteroplatea marrnorata.
lower pleurals emerge half-way
from the median to the tip

Pteroplatea hirundo.
pleural tubules reaching little, if any,
more than half-way to the outer
angle ;
laterals not united across the ver-
tebrae ;
anterior cephalic tubes reaching
forward from skull ;
subrostral and prenasalunited;
median transverse

Myliobatis aquila.
median vertical

Aetobatus narinari.
subrostral and prenasal not
united

Myliobatis freminvillei.
anterior cephalic tubes not reach-
ing in advance of skull;
oral not crossing symphysis ;
no long tubules on the oc-
cipital

Rhinoptera brasiliensis.

oral crossing, symphysis ;

long tubules on the occipital

Rhinoptera jussieui.

laterals united by tubes across the

vertebrae Dicerobatus ol/eraii.

DESCRIPTIONS.
Chimsera.

Chimcera monstrosa (Plate II.) has open groovea throughout both corporal
and cephalic portions of the canal system. The delicate membranes of the
inner part of the furrows are protected bv closely set scales which overhang
from each edge. There are no tubules. On the snout the canals present a

74 BULLETIN OF THE

peculiarly scalloped appearance, caused by the rising of short sections of the
edges as prominent rounded flanges supported by ribbed cartilaginous expan-
BTons. At the end of each pair of these flanges the walls are lower, which
gives an appearance of rounded holes into the grooves. Between the holes the
edges are somewhat zigzag.

On the scapular region the lateral (l) makes a moderate curve upward ; on
the flank it is slightly sinuous; and on the anterior portion of the tail, near
the forward end of the lower lobe of the caudal fin, it descends to the lower
edge of the muscles, a position retained to the extremity. In its middle the
aural (au) is bent back, forming an angle from which a short post-aural branch
reaches toward the dorsal spine ; the canal crosses behind the aqueducts.
Aural and cranial (cr) are joined ; they are connected with lateral and orbital
(or6) by an occipital (oc) of moderate length passing downward and backward.
The cranials converge to some extent on the forehead ; on each side of the
frontal holder they turn out a little, but approach again on the snout. Below
the posterior border of the orbit the orbital meets the jugular (j) and the sub-
orbital {so). From this point the suborbital passes forward to join the ros-
tral (r) at the end of the snout, rising well up in front of the eye in an open
loop, somewhat inclined forward. An individual variation appears in each of
two specimens at hand : in one the angular (ang) unites with the jugular, in
the other with the suborbital. At first the angular passes downward to the
oral (o); thence it goes forward to the nasal (n) and the subrostral (sr). The
jugular runs obliquely backward and continues across the throat in a broken
line. Beneath the back part of the eye the oral leaves the angular, and may
be traced across the chin in a series of dashes or dots. The nasal lies in front
of the nostril ; it bends forward and meets its fellow in advance of the nares,
but forms neither median nor prenasals. From the suborbitals the subrostrals
extend toward each other and unite in a median ; a short distance posteriorly
they diverge to meet the angulars. The second specimen differs from that just
noted in having the angular united with the suborbital, and the jugular less
noticeable on the throat. The fii-st of these features is an approach to the
condition in CaUorhynchus, where angular, oral, and jugular connect with the
suborbital, but not with each other.

CaUorhynchus.

CaUorhynchus antarcticm (Plates III., IV.) differs from Chimtera, and agrees
with the majority of the Sharks, in possessing canals that are tubes, instead of
furrows. On the flank the lateral rises a little in the scapular region ; thence
it is sinuous to the end of the dorsal; and thence straight to a point above the
lower Ipbe of the caudal fin, where it makes an abrupt downward bend to the
lower edge of the muscles, which position it retains to the end.

In the middle the aural is mijch turned back; but it forms no angle and
sends oflF no branch. Forward from the aural the cranials are gradually con-

MUSEUM OF COMPARATIVE ZOOLOGY. 75

vergent, more decidedly so in advance of the frontal holder ((J), and they
approach each other closely on the thin portion of the proboscis. As in Chi-
maera, the occipitals connect the laterals and the orbitals with the aural and
the cranials, instead of connecting the aural and the laterals with the orbitals
and the cranials as in the Selachia. Orbitals and jugulars meet below the
pupil of the eye. The suborbitals are very long; they pass quite to the end
of the snout, and there meet the rostrals as the latter pass to the lower surface
of the rostrum. A row of short pieces of canals across the throat serves to
unite the jugulars. Under the fore part of the pupil the oral leaves the sub-
orbital; it curves forward on the cheek and the chin, and backward behina
the corners of the mouth. On the cheek it has what appears to be a more
slender tube just in front of itself. Not far in front of the oral the angular
descends toward the mouth from the suborbital. As it nears the lip it takes
more of a forward course, and, following near the border of the rostral flap,
finds its way down and backward to the edge of the lower surface, where it
turns under and inward to cross the wing and meet the angular of the oppo-
site side. After meeting the suborbital each rostral in its backward route ap-
proaches very close to its fellow, under the end of the rostrum, if it does not
unite with it. Farther back they diverge, and each turns up a side of the
snout, curving back as if to unite with the nasal. The union of subrostral
and nasal has not been traced. In the adult specimens the oral and the jugu-
lar show tendencies to obsolescence.

Although there are great differences in the shape of the head in this genus
and in Chimsera, in the arrangement of the canals in the two cases there is a
great deal of similarity. One has only to suppose the snout of Callorhynchus
shortened, so that the flap may be applied against the head, and the union
of subrostral and nasal, if not already existing, to produce an arrangement
essentially the same as that of Chimsera.

Scoliodon.

Scoliodon terrcB novas (Plate V.) represents one of the subdivisions of the
genus Carcharias, as arranged by Miillor and Henle. Comparison of this spe-
cies with Prionodon milherti, a representative of another of these subdivisions,
will give an approximate idea of the range of variation within that genus.

A small amount of curvature only is to be noticed in the thoracic portions
of the laterals. Below the second dorsal they make a slight descent, then
they ri.se rather higher than before, after which the canals retain the same rela-
tive height as regards the vertebral axis till they reach their terminations at
its end.

In its middle the aural has a shallow backward bend ; and it has one oi
similar depth in the opposite direction near- each end. The occipitals are
comparatively long; tbey are extended obliquely out toward the eye. A short
distance from these canals, each cranial makes a short but decided outward

76 BULLETIN OF THE

bend; in front of this, and between the orbits, it runs abruptly toward the
middle of the crown, before reaching which it makes a broad deep curve, car-
rying the tube outward to a point opposite the hinder margin of the fontanelle,
whence it passes in a nearly straight direction toward the tip of the snout.
Before it reaches the latter, it descends to the lower surface. Soon after leav-
ing the cranials, the orbitals sink deeply into the tissues of the side of the
head; approaching the skin again, each makes a broad curve around the orbit,
rising in front above the middle, after which it goes downward to meet the
angular, nearly half-way from the eye to the nostril. The angular is rather
elongate; its continuation, the jugular, ends in front of the middle of the first
branchial aperture. The sections of the oral are disconnected and detached;
the space separating them from the angular is about equal to that separating
them from each other; their length is about two thirds of that of each mandi-
ble. At its forward extremity the short orbito-nasal meets the subrostral and
the nasal; posteriorly it meets the suborbital and the angular. The nasals are
nearly transverse, and would be quite so if not for a decided curve forward
at the inner edge of each nostril. A short longitudinal median connects the
nasals and the prenasals; the latter have a moderate degree of divergence, and
unite with the rostrals.

Very prominent cranial curves, long occipitals, a shorter suborbital fold
in front of the eye, and a more pronounced nasal curve, are among the most
patent diflferences to be noted on this species as compared with Prionodon mil-
berti. An approach toward the conditions existing on Cestracion is to be seen
in the cranial, rostral, subrostral, and nasal curves, and in the great depth to
which the orbitals have sunk in the tissues behind the eye at the side of the
head.

Prionodon.

Along the trunk, the laterals of Prionodon milberti (Plate VI.) deviate but
little from a right line. There is a small degree of curvature behind the occi-
put. Opposite the anal the downward bend is hardly perceptible. On the
tail, above the anterior portion of the fin, the canal descends hardly half-way
to the lower edge of the muscles; it keeps the same relative position as far
back as to the hindermost of the vertebrae.

The aural is transverse, turned back a very little at each end. The occipi-
tals reach toward the side, behind the eye ; they are rather short. In the
coronal region each cranial makes a long shallow curve inward. From the
fontanelle they are nearly direct, converging somewhat ; and they descend
some distance behind the end of the snout. Depression of the head has
brought orbitals and angulars close together, on the cheek. In front of the
orbit, the suborbitals rise higher than the middle of the eye ; the loop formed
by them extends more than a diameter in front, and it is about half as far from
the orbit to the junction with the angular. The subrostral is long, with a
shallow curve around the nostril; the angular is long; the jugular is me-

MUSEUM OF COMPARATIVE ZOOLOGY. 77

dium; the orbito-nasal is short; the nasal is transverse, with a broad curve;
the median is very short; and the prenasals are long and connected with the
rostrals. There is a short disconnected oral behind each side of the mouth.
In their principal features, the canals represented in this form are interme-
diate between those of Mustelus and Scoliodon.

Cestracion (Klein).

Forward, the laterals of Cestracion tiburo {Plate VII.) are nearly straight;
backward, they have a pronounced curve, between anal and second dorsal, but
do not descend much below the middle of the tail, and they stop near the end
of the vertebral column.

The flatness of the head, the expansion of the snout, and the positions of
nostrils and eyes at such great distances from the occiput, have caused some
very peculiar contortions of the cephalic tubes. A small amount of backward
convexity is presented by the aural. The occipital is elongate and turned
toward the side. Between the eyes the cranial is very tortuous, turning upon
itself several times before taking its way toward the rostrum, where it again
makes a backward run before passing through. From the occipital the orbital
goes outward and backward a short distance, then passes through, behind the
lateral cartilages of the skull, to the ventral surface. Below the head the sub-
rostrals tend laterally near the edges until about half-way from the snout to
the eyes, where they ascend and run for a similar distance on the upper surface
before descending again just in front of the eyes, thus passing around the nos-
tril, and finally going with much directness to meet the nasal about balf-way
from the eye to the median. A great bend out in the direction of the eye is
made by the suborbital on the way forward to its junction with the angular,
the entire length of the tube being about three times the distance between the
point of appearance on this surface and that of the meeting. The orbito-nasal
is very short; it lies at right angles with the angular. Angular and jugular
are about equal in length. Near the comer of the mouth the oral is bent
toward the thorax ; it is connected with the angular, but does not cross the
symphysis. Were it not for a long narrow loop putting out toward the nostril
the nasal would be described as nearly transverse. This loop has parallel
sides, is slightly bent back, and has a tubule from its extremity. A median ot
moderate length gives rise to a pair of prenasals, Avhich are somewhat curved,
and which meet the subrostrals a little toward the side from the points at
which the latter make their appearance.

An arrangement of canals such as that here described might be developed
from a form like Scoliodon terrce novce by crowding the snout back toward the
skull and expanding the head at the sides. To push the long cranials back
toward the occiput would bring about the identical curves appearing on the
Hammerheads ; in fact, the curves on Scoliodon are just what would naturally-
lead to such a result. Expanding the head would necessitate the appearance

78 BULLETIN OF THE

of the subrostral on the upper surface to retain its position outside of the nos-
tril, the latter being on the edge. And the orbitals, being behind the expan-
sion, would sink deeper into the tissues; this also is simply going farther in
the direction already partially traversed by the orbitals of Scoliodon.

Mtistelus.

Anteriorly on Mustelus canis (Plate VIII.) there is hardly any departure
from a right line in the lateral. Over the anal fin the canal rises ; farther back,
it descends to its former level, and, not going below the middle of the caudal
muscles, it stops at the last of the vertebrae.

Comparatively little curvature is apparent in either aural or occipitals.
The latter are short and diverge toward the cranials. In front the cranials
converge ; opposite the fontanelle they turn directly toward the side of the
head for a short distance, then they run forward, almost straight, somewhat
convergent, and pass through the snout before reaching the tip. At first the
orbitals are transverse, but with a gradual curve they sweep below and about
half its diameter in front of the eye, where they turn back and downward to
meet the angular beneath the anterior third of the orbit. The angular is of
medium length ; the jugular is short, ending near midway from spiracle to gill
opening. A short horizontal orbito-nasal connects with a long subrostral, in
which there is but a slight bend at the side of the nostril. A prominent curve
appears in each nasal, between the nostrils. The median is long and longi-
tudinal. The prenasals are rather long; they connect with the rostrals.
Behind each ang;e of the mouth there is a short detached oral.

The canal system of Mustelus closely resembles that of Triacis.

Triads.

On Triacis semifasciatum (Plate IX.) there is hardly any curve in the scapu-
lar portions of the laterals. As in Mustelus, the anal cui've is a broad one;
the canal does not descend to the edge of the fin, and it stops at the end of the
column.

There is a slight forward bend in the middle of the aural, otherwise it is
almost straight. The occipitals are of medium length, and are divergent. In
their coronal portions the cranials are nearly straight. At the sides of the fon-
tanelle the bend is abrupt, but not deep. The rostral sections of these canals
vary in outline, converge, and descend before reaching the end of the rostrum.
Passing outward, the orbitals bend back slightly; they sink deeply into the tis-
sues behind the eye, and extend in front of the orbit more than its diameter.
A much more open loop is made by these tubes as they turn to join the angu-
lar than in Mustelus. The angular is of moderate length; the jugular is short
and turned up toward the superior edge of the gill opening, as in Ginglymo-
stoma, Scylliorhinus, and Mustelus. The orbito-nasals are short, convergent

MUSEUM OF COMPARATIVE ZOOLOGY. 79

forward. The subrostrals are moderate in length and slight in curvature.
The nasals are transverse, broadly curved forward between the nasal valves.
Approaching the symphysis, behind the comers of the mouth there is a pair
of detached orals. A short median and moderate prenasals, the latter con-
nected with the rostrals, complete a system closely resembling that of Mustelus
and with remote likenesses to that of Scylliorhinus.

An embryo of two and a half inches agrees so well with the adult that it
is readily identified by means of the canals. Without the aid of the system
identification would hardly have been possible.

Isurus.

Isurus punctatris (Plates I., X.) by the fusiform shape of its body compels
the laterals to diverge considerably along the middle of the flanks. With this
exception they are tolerably straight, there being hardly any deflection above
the anal, and but a scarcely noticeable upward turn on the tail. In reality the
laterals cross the muscular portion of the tail, not following the course of the
vertebrae, and they end at the lower (hinder) edge of the muscles close behind
(above) the lower caudal lobe in front of the concavity in the posterior margin
of the fin.

The aural is long, without prominent curves. The occipital is short, and
nearly transverse. Both coronal and rostral curves of the cranials are long
and shallow. The rostrals are short. In its downward course the orbital is
waved a little; as a suborbital it joins the subrostral below the forward margin
of the orbit. By its connections the short orbito-nasal would appear to be
reversed in direction. In one specimen the angular bends downward behind
the angle of the mouth to meet the oral, and the jugular makes an upward
turn, then goes half-way to the gill, to bend up still farther at the end; in an-
other case the angular and jugular form a single nearly straight line. The
oral is connected with the angular, and runs but little beyond the comer or
the mouth. At less than a quarter of the distance from the eye to the rostral
tip, the subrostral appears on the lower surface; from this point it is longitudi-
nal, faintly curving above the nostril. No nasal curve appears in the nasals.
The short median is nearly transverse. About one third of the prenasal is
bent abruptly to the side to meet the rostral.

Prominent among distinguishing characters are the caudal portions of the
laterals, the reversed orbito-nasal, the transverse median, the curveless nasal,
and the attached oral.

Odontaspis.

Odontaspis americanus (Plate XI.) exhibits neither scapular nor anal curves
in the laterals. The canal remains near the middle of the tail, and stops a
little forward of the last of the vertebne.

The aural bends back in the middle a very little. The occipital is of mod-

80 BULLETIN OF THE

erate length, and puts out directly toward the eye. On the crown the craniala
are straight until opposite the fontanelle, where they make a shallow outward
curve. The rostrals descend near the end of the rostrum. Starting down
and backward, the orbitals make a broad curve around the eye until beneath
it, where they become longitudinal and run more than a diameter forward
from the orbit, before turning down and back, parallel with themselves, to
join the angular. Both angular and jugular are long. From the angular the
oral bends back around' the mouth; it is not continuous across the symphysis.
The long orbito-nasal bends down imder the fore part of the eye before becom-
ing longitudinal. Above the nostril the subrostral turns abruptly toward the
nasal, in which there is no perceptible curve. The median is elongate and
longitudinal. At the median the rather short prenasals bend outward, then
turn forward to join the rostrals.

The type is characterized by absence of anal or caudal bends, by an elon-
gate occipital, a slight curvature in the cranials, a longitudinal loop in the
suborbitals, a prominent curve in the subrostrals, absence of a nasal curve,
and by the junction of oral and angular.

Alopias.

Alopias vulpes (Plates XII., XIII.). A very great development of the
canal system obtains in this Shark. There is no great difference in the main
tubes from what may be seen in allied genera ; it is in the enormous number,
the length, and the amount of branching of the tubules, that unusual features
are most patent.

Forward the lateral bends upward a little ; at the base of the tail it fol-
lows the vertebral- axis, keeping its position near the middle of the muscular
portion, and ends a little in advance of the notch in the hinder extremity.
Throughout the entire length the tubules are closely placed on the sides.
Anteriorly, on the thorax, they are directed toward the back. Nearly all of
those on the abdominal region are extended toward the belly. From the base
of the ventrals to the end of the anal the tubules have numerous branches,
some of which J)ass upward and others downward. On the tail the tubules
are sent toward the lower edge of the fin.

The aural is long; in prominent curves it bends back in the middle and
forward at each. end. The occipitals are short. In the cranials the curves are
shallow. At the crown the orbital starts back and down ; as a post-orbital it
is vertical; and in the suborbital it sinks below the eye. Slightly in advance
of the eye the suborbital turns back, and not far from the centre of the orbit,
over the front edge of the mouth, is the union with the angular. The latter
is of medium length. An uncommon arrangement of the jugular is seen here:
the tube is long and passes below the gill apertures. The oral is elongate and
connected with the angular; it makes a sharp bend around the corners, and' is
divided by a narrow interspace in the middle, behind the symphysis. At the

MUSEUM OF COMPARATIVE ZOOLOGY. 81

angular the orbito-nasals make an abrupt drop, beyond which they are longi-
tudinal. The nasals are long, and have but a small amount of curvature. The
median is short, and placed longitudinally. The prenasals aie long, and ab-
ruptly bent to the sides to meet the rostrals. Opposite thft nostrils there is a
decided outward bend in the subrostrals.

A great number of long tubules exist on the laterals, and on some of the
cephalic tubes. Those from the aural reach directly back. Those from the sub-
orbital extend backward or downward; some of them connect with the angular
or its tubules ; all are more or less branched. Above the mouth the branches
of the angular turn upward ; behind the oral a few of them go downward,
where, by meeting others from the oral, and by uniting among themselves,
they form a network. On the upper side of the jugular the branches are
much more numerous, but have not so many branchlets. Behind the angle
of the mouth the tubules of the oral are longer and more branched.

There is a striking similarity in the canals of Alopias and those of Odontas-
pis. This may be seen in cranials, aurals, orbito-nasals, suborbitals, subros-
trals, prenasals, nasals, angulars, and orals; and it appears fully to warrant
placing these genera side by side in a systematic arrangement of the Galei.

Heptabranchias.

- Heptabranchias maculatus (Plate XIV.). On the flanks of this species the
canals are shallow furrows, protected by enlarged overhanging scales of the
shagreen on the edges. In front the grooves commence above the forward
portions of the bases of the pectorals; all the canals farther in front are tubes.
Another specimen shows alternation in the lateral, between the aural and the
continuous lateral furrow, of irregular lengths of tube and groove. Over the
anterior lower lobe of the caudal the furrow bends downward in the direction
of the fibrous portion, which it approaches more gradually backward, and it
ends at the notch between the lower and the hinder sections of the fin.

The aural is divided by an interspace, behind the openings of the aqueducts,
as in H. pectorosus. A small amount of outward curvature marks the elongate
occipitals. At each side of the fontanelle the cranials bend out in a broad
curve toward the side of the head. In front of the nostrils the rostrals turn
back toward the subrostrals, but apparently without meeting them. These
tubes seem to be separated, just above the nostril, by a short interspace. On
the top of the head, again, the orbital is directed outward and a little forward-;
on the side it goes down and backward, without forming a suborbital, to join
the angular and orbito-nasal. The latter is very long, and takes the place
of the suborbital. Jugular, angular, and orbito-nasal form a single longitudi-
nal line; the first is short, ending in front of the middle of the first gill cleft,
the second is of moderate length, and the third is as long as both of the others.
The nasal is of moderate length, curves strongly toward the median line, but
does not meet its fellow, from the other side, to form a median. The pre-
voi. XVII. — Ko. 2. 6

82 BULLETIN OF THE

nasals bend out toward the rostral, without approaching closely, then pass for-
ward and end blindly near the tip, at a considerable distance apart. Above
the front edge of the mouth the subrostral meets the nasal in a sharp angle.
The nasal curve is comparatively slight. Traces of an oral were not detected.

Excessive thinness of the skin, by bringing the canals so close to the sur-
face, favors the presence of furrows rather than tubes, or, to go still further,
leads to the disappearance of the canals altogether, as in case of the orals of
this and other species.

Characteristic features of the system on this shark are the isolation of the
prenasal, the length of the orbito-nasal, the suppression of the suborbital, the
direction of the orbital, the bisection of the rostral, the division of the aural,
and the open lateral tubes. Several points, in occipitals, cranials, orbitals, and
orbito-nasals, recall similar ones in Chlamydoselachus; the latter, however, is
widely withdrawn by consideration of its lack of division in aural and rostrals,
the position of its prenasal, and its possession of oral, gular, and spiracular
canals.

H. pectorosiis is, in most particulars, similar to H. maculatus. Its laterals
end about two fifths of the length of the tail in advance of the extrem-
ity, making a decided and broad curve downward to the fibrous part of the
caudal.

A specimen of H. cinereus has closed corporal canals, or tubes, of similar
position and outline as the two species of this genus already noticed, but
reaching a little farther toward the caudal notch than in H. pectorosiis.

Chlamydoselachus.

Chlamydoselachus anguineus (Plate XV.) has the laterals open throughout
their whole extent, with the exception of less than an inch immediately be-
hind the aural. From each edge enlarged scales overhang the groove, enclos-
ing it in a measure and protecting it. Along the flanks the canals are nearly
straight. The caudal curve is very gradual in one specimen, more abrupt in
another, and on one side of the second descends, then rises to repeat the curve.
On the body, the canal lies a little above the crease between the muscles of
the back and those of the flank. On the tail, its track lies a little below the
middle of the muscular portion ; it continues thus, with a few slight breaks
posteriorly, to within an inch of the end of the vertebral column, where it
stops.

In the sketch the courses of the closed cephalic tubes are indicated by lines
of dots, each of the larger of which marks the opening of one of the short
tubules. The aural is closed. It has no tubules. Contrary to what obtains
in other Galei, it lies in front of the so-called ear openings. These openings,
however, are at the ends of tubes the inner extremities of which are in front
of the canal. The canal is nearly straight, bending slightly forward in the
middle and a little backward near each end. The occipitals are long and

MUSEUM OF COMPARATIVE ZOOLOGY, 83

extend forward with a very slight trend outward. On the crown the cranials
are parallel. At the sides of the fontanelle they bend abruptly outward, and,
as rostrals, run near the edge of the snout for some distance before going to the
lower surface. From the cranials the orbitals run outward and somewhat for-
ward ; near the side they turn backward and downward towai-d the corner of
the mouth. They end some distance behind the eyes. A long angular joins
the short jugular and the very long oral, which reaches almost to the sym-
physis. At the end of the jugular near the middle of the first branchial aper-
ture, there are two branches not found in any other of the Sharks examined : a
spiracular (sp), turning upward and forward toward the spiracle, and a gular (g),
turning down and forward near the median line, and finally uniting with the
oral a short distance from the inner end. Below the eye, in the position usu-
ally occupied by the suborbital, lies a very long orbito-nasal. The nasal is of
moderate length, and curves broadly in its posterior half. The subrostral is a
little shorter than the nasal; it bends upward over the nostril. Apparently
the prenasal is reversed in direction, meeting the nasal in front and running
backward to join the subrostral. Like the corporals, oral, gular, and spiracu-
lar are open grooves. lu the spiraculars and gulars of this Shark are found the
nearest approaches to the pleurals of the Batoidei.

Distinguishing peculiarities of the system on this type are seen in the pos-
session of spiracular and gular canals, in the position of the prenasals, and in
that of the aural, with regard to the ear openings. Similarity in the orbito-
nasals occurs in Heptabranchias. Somniosus by the same canals is interme-
diate between these genera and others of the order.

Ginglymostoma.

GinglymoKtoina cirratun (Plate XVI.). Over the shoulders the laterals have
little outward curvature ; in the anterior part of the tail they drop somewhat
abruptly from the middle to the lower portion of the muscular band, near the
edge of the fin, where they continue, ending with the vertebral column.

This form has a short broad head, and a very short snout. If compared
with one of the long-snouted species, it will be seen that there is a tendency
toward the transverse in the cephalic canals, which in those forms are nearly or
quite longitudinal. The aural is long, bending backward a little in the middle,
and as much forward toward each end. The occipital is of medium length; it
runs obliquely outward, with a slight curve toward the spiracle in the middle.
From the end of this canal the cranial turns rather sharply toward the crown;
it then passes forward, diverging a little from its fellow until opposite the fonta-
nelle, where it turns outward with less curvature than in Scylliorhinus. Ap-
proaching the edge, the rostrals run parallel with it until near the tip, where
they descend. The orbital is rather short. The suborbital is much longer and
passes forward more than three times the diameter of the orbit; above the nos-
tril it turns back, forming an angle, and meets the subrostral a short distance
forward from the eye. Angular and jugular are short; they are directed up-

84 BULLETIN OF THE

ward some, toward the top of the first branchial aperture. Behind the comer
of the taouth the oral makes a strong backward curve ; the tube is a long one ;
it crosses the symphysis and meets with the angular. Posteriorly the orbito-
nasal curves upward to meet the angular; the tube is elongate and nearly hori-
zontal. The nasal is long, sinuous, and almost transverse. Contrary to what
might be expected on a short snout, the median is long. As if reduction in
the length of the snout had proceeded faster than in that of the tubes, the
prenasals appear as if pushed back and folded on themselves ; each is turned
abruptly toward the side and bent into two folds. They unite with the
rostrals.

Prominent distinguishing features in this shark are the caudal canals, scapu-
lar curves, complete oral, long suborbitals and orbito-nasals, and the folded
prenasals. Of the genera studied it approaches Scylliorhinus most closely.

Scylliorhinus.

Scylliorhinus caniculus (Plate XVII.) has scarcely any curvature in the later-
als, and they end with the column, not going down to the fibrous portion of
the caudal fin.

A slight sinuosity affects the aural. The very short occipital is directed
toward the eye. Behind the fontanelle the cranials approach the median line
in a broad curve: in front of this, they turn abruptly out toward the edges of
the snout. Just before it joins the prenasal, there are several curves in the
rostral: in descending it runs forward, downward, inward, backward, and out-
ward. At the side of the nostril there is a prominent curve in the subrostral.
The suborbitals are longitudinal below the orbit ; at its forward edge they
pass down and backward to meet the short orbito-nasals. The nasals are
almost straight and transverse; the median is short; and the prenasals, sinu-
ous and moderately long, unite with the rostrals. From the angular the jugu-
lar curves up toward the upper edges of the gill opening, which it does not
reach. Behind each angle of the mouth there is a short disconnected oral.

Heterodontus.

On Heterodontus philippi (Plate XVIII.) the laterals diverge a little, behind
the occiput; farther back they are straight, without a curve over the anal fin,
until they reach the tail, above the lower lobe of which they descend to the
lower edge of the muscles. As it nears the end of the column, the canal
becomes a furrow.

Lateral and aural form a continuous curve, and are connected with the cra-
nial and orbital, which form a similar curve, by a very short occipital. The
cranial bends are broad, but not at all deep. This is true also of the suborbital,
which reaches nearly a diameter in front of the orbit, then drops vertically on
the subrostral. Angular and jugular are both very short. The oral joins the

MUSEUM OF COMPARATIVE ZOOLOGY. 85

angular, but does not cross the symphysis. The orbito-nasal is long. In pass-
ing the nostril, the short subrostral makes a decided curve. Behind the nostril
the nasal is bent toward the mouth. The median is very short. From the
median the short prenasal goes directly to the side, joining the rostral.

By the canals either Odontaspis or Ginglymostoma shows more affinities with
Heterodontus than does Acanthias. The differences in dentition between the
latter genera are scarcely greater than those apparent in the canal systems.

Acanthias.

Acanthias amcricanus (Plate XIX.). Backward from the shallow scapular
curve, the laterals of this species are nearly straight. Above the widest part of
the lower lobe of the tail, the tube makes a slight bend upward ; it does not
follow the vertebral column, but gradually approaches the lower edge of the
muscles, and stops in front of the last vertebra.

In the mi<ldle, towards the ear openings, the aural is bent forward. A con-
tinuous longitudinal line is formed by the elongate occipital and the lateral.
At the orbitals, the cranials make a rather sharp curve; opposite the fonta-
nelle, they make a broad and shallow bend. The upper portion of the orbital
is sinuous; behind the eye it is thrown backward; and beneath the orbit it
goes but half-way before turning back in a sharp angle to join the angular.
Jugular and anyular together are short. The orbito-nasal is long, and is bent
downward from the suborbital. By the side of the nostril there is a decided
bend in the subrostral. The nasals are long and bent so that the curves in
each approach the outlines of a Z; they do not meet to form a median, but
run close together as in Pj'istiophorus. Near the end of the snout the pre-
nasals converge, without seeming to join; they are located some distance from
the rostrals. The tubes are of lar^e calibre, and the tubules are numerous
and short. A short oral lies close to each angle of the mouth, entirely dis-
connected. On the tail, for a short distance from the end, the canal is open.
Figure 6 shows the arrangement of the scales and the form of the portion of
the canal included between the dotted lines.

An embryo of two and a quarter inches has tubes similar to those of the
adult, but the tubules are shorter or absent.

Somniosus.

Somniosus carcharias (Plate XX.) has tolerably straight tubular laterals.
They extend on the middle of the muscular portion of the tail, running as far
back as the hinder edge of the anterior lower lobe of the caudal fin; thence
they descend to tlie lower edge of the muscles, above the fibrous porticm, where
they continue to the end of the column.

Among the cephalic canals a very peculiar arrangement occurs on the occi-
put; the aural is transverse, and has its ordinary position; from its ends the
occipitals curve forward and inward, and end anteriorly without connecting with

86 BULLETIN OF THE

other tubes ; a short distance in front of their ends are those of the orbitals,
also disconnected; and still farther in front are the posterior extremities of the
cranials, like the others, making no connections. The orbitals pass directly
outward, then downward and forward, meeting the angular below the hinder
portion of the eye. At the start the cranials are transverse, they soon bend
forward, and, making very open curves around the fontanelle, becoming ros-
trals, converge toward the end of the snout, before reaching which they pass
through to the lower aspect. As subrostrals they go back and outward , mak-
ing a loop on the side and top around the nasal chamber, and pushing farther
back to meet the nasals. In comparison with that of other genera, the orbito-
nasal is rather long; it extends below the greater portion of the orbit. A
prominent curve toward the nostril marks the middle of each of the elongate
transverse nasals. From a median of more than ordinary length the prenasals
diverge and run forward, ending abruptly, under the tip of the snout, without
joining the rostrals. The angulars are of moderate length. The jugular is
short. An oral could not be found. The tubules are numerous, short, and
provided with large apertures.

Especially noticeable among the pecv;liar features met with in this species are
the separation of the orbitals and the cranials from the occipitals, the isolation
of tlie prenasals, the supranarial curve of the subrostrals, the length of the
orbito-nasal, the lack of tlie oral, and the caudal curve of the lateral. The
coronal arrangement of the ceplialic canals, and the subrostral curve, distin-
guish the genus from any of the other genera noted here. The orbito-nasal and
the disposition of the lateral on the tail are intermediate between sharks like
Heptabranchias and the majority of those of higher rank. In Cestracion (Zy-
gsena) only of the other Galei have we seen the subrostral return to the top
of the snout.

Rhina.

Ehina squatina (Plate XXL). With the great depression of the body of
this Shark, the lateral has to some extent been carried outward on the thoracic
and the scapular regions. Excepting slight waves in the outline, there is
hardly a deviation from a straight line in the tubes on the sides of the tail.
The canal does not reach quite to the hindmost vertebra. Above the thoracic
region near the aural a few of the tubules reach toward the median line.
Elsewhere along the whole length of the laterals the tubules are short and
directed out.

In consequence of the anterior position of the mouth and the shortness
of the snout, the canals of the front part of the head are greatly reduced in
length. Subrostrals, prenasals, and otlier tubes that in the balance of the
Sharks are confined to the lower surface, have T)een brought to the top by de-
pression of the head. The aural is long, curves backward slightly, and has
a few tubules extending toward the shoulders. The occipitals are long, and
<livergent forward. On the frontal region, the cranials curve toward each other;

MUSEUM OF COMPARATIVE ZOOLOGY. 87

their tubules, as those of the occipitals, start out, but turn and cross the tubes
toward the middle. Opposite the foutanelle the cranials make a broad open
bend, from which the short rostrals pass about half-way to the middle of the
mouth. From the front end of the rostrals the subrostrals turn back, around
and behind the nostril, to meet the nasals on the sides of the face. The pre-
nasals lie on the upper surface; they are nearly transverse, and turn back at
the ends as if to join the rostrals, but without making a junction.. Apparently
the very short median is vertical. The orbitals are entirely on the upper sur-
face. From the cranials, they go obliquely outward until past the orbit, then,
turning forward at a right angle, the suborbitals run a short distance beyond
the eye, where they turn out and backward, making a deep loop, convex in
front. They meet the angular opposite the eye. Both angular and jugular
are on the top of the disk. The orbito-nasal is on the side of the face; it is
comparatively elongate, running from opposite the middle of the orbit to
within a short distance of the nostril. The orals and the nasals belong to
the lower view. The former are elongate, disconnected, and do not reach the
symphysis; the latter are moderate and transverse, with a shallow curve for-
ward toward the middle. The tubules from the suborbitals are rather long
and pass outward; they, like the others, are unbranched.

The appearance of all the tubes, except orals and nasals, on the top, looks
as if resultant from depression that had caused great expansion of the ven-
tral portions of the body, and but little of the dorsal. This peculiarity alone
would serve to distinguish the genus from the other genera. There is nothing
in the canal system that favors the idea of close affinity with the Batoidei.

Pristiophorus.

Considerable uncertainty exists in connection with several points on the
sketch of Pristiophorus cirratus (Plate XXII.), because of the bad condition
of the specimen, a dried skin.

Back of the head the laterals turn outward somewhat; on the tail they
appear to lie near the middle of the muscular portion, stopping at the end of
the column.

The halves of the aural meet in a sharp angle at the middle; behind the
openings of the aqueducts they form a V, from the apex of which a short
tube extends directly back. A low inward sweep is made by the cranials, on
the crown. Beneath the eye the orbital does not quite reach the front edge
of the orbit; turning backward, it descends to join the angular on a vertical
from the centre of the pupil. The angular is longer than the jugular. Ap-
parently there is no oral. The orbito-nasal is rather long; and. Avith the
angular and the subrostral, it forms a longitudinal line. In front of the mouth
the na.sal is turned back; it has a moderate nasal curve, and does not connect
with its fellow to form a median. Prenasals and subrostrals are very long. In
the specimen they cannot be followed near the end of the rostrum.

There are marked resemblances between this Shark and Acanthias, which

88 BULLETIN OF THE

are at least suggestive of closer affinities in the distant past. These are brought
prominently forward in comparisons of such tubes as the medians and sub-
orbitals of the two forms.

Pristis.

Posteriorly the laterals of Pristis pectinatus (Plate XXIII.) are straight.
On the tail there is a slight downward tendency, and the canals end near its
extremity, at the lower edge of the muscles. Anteriorly, above the thoracic
region, they are drawn toward each other; at the shoulder each makes an out-
ward bend, irom which the scapular and the post-scapular branches extend.
A comparatively small area is enclosed by the pleural ; from the shoulder the
tube runs out and backward, then it turns forward, along the inner edge of
the pectoral fin, to meet some of the tubules from the orbital, near the hinder
part of the orbit, after which it makes a sharp bend and goes back a short dis-
tance parallel with its former course before passing down tlie side to the lower
aspect, about opposite the aural. One or two post-scapular branches, together
with the posterior pleural tubules, form a network of branchlets on the pectoral.
Lateral, pleural, and suborbital possess slender tubules. Similar ones on the
rostral have delicate branchlets.

On the ventral surface the pleurals run toward the gill openings, in front
of which, about one third of the distance to the mouth, they meet the jugulars.
The space included by these tubes is small.

The aural is deeply bent backward. A short occipital coimects it with the
orbital. The latter goes close below the e^'e and in front of it, al)0ut half a
diameter, passes to the lower surface around the edge. Near the fontanelle
the cranials diverge slightly, making a shallow bend; near the end of the ros-
trum thev converge, but diverge again at the tiji; in general, their course is
tolerably direct. These, as the other tubes of this surface, are beset with a
great many very fine short tubules.

Angular and iu'nilar are moderate. The orbito-nasal is short; in front it
meets the suborbital and the subrostral. Only a small portion ot the sub-
orbital is to be seen from below. The sulirostral is much elongated, has a
■waved course, and is bent prominently forward in front of the nostril. The
nasal is transverse, and waved in outline. The median is longitudinal and
short. From it the prenasals turn abruptly outward, toward the nostrils, be-
fore takiuL; a course of tolerable directness toward the rostral extremity. Close
to the latter they appear as if crowded back, so as to make a fold directed
toward the median line. Behind each side of the mouth there is a discon-
nected oral ; toward the middle the tube bends forward, at the outer end it is
turned back in a hook.

Although there is much resemblance between the majority of the canals
of Pristis and those of Fristiophorus, the presence of the pleural and the
scapular branches fixes the position of the former in the Batoidei.

MUSEUM OF COMPARATIVE ZOOLOGY. 89

Rhmobatus.

Rhinobatus planiceps (Plate XXIV.). From the shoulder girdle to their
ends on the tail the laterals of this species are nearly straight. At the pecto-
ral arch the scapular curves carry them outward, and back of the head they
approach each other. Posteriorly they send numerous tubules outward; ante-
riorly others are sent inward and backward. Behind the scapular there are four
post-scapulars, each of which has two or more tubules near the end. There is
more lateral curvature in the pleurals in this genus than was seen in Pristis ;
there is also a more intimate connection between them and the suborbitals, by
means of four or more of the tubules. Opposite the forward part of the orbit,
about half-way between it and the margin, the pleurals pass through the disk,
after sending numerous tubules on the pectoral. These pleural tubules are
of two kinds, one stronger, longer, and straighter; another finer, shorter, and
crookeder, distributed among the first. The course of the pleurals on the
lower surface is short; they meet the jugulars in front of the first branchial
aperture.

The aural curves back in the middle. A slight divergence obtains in the
short occipitals. In front of the eye the crauials curve outward sharply; they
approach each other nearest along the middle of the rostral cartilage. At the
end of the suborbital a branch is sent backward; in front of the eye its tubules
extend both inward and outward ; and still farther in front they are sent to-
ward the margin. Instead of going around its edge, the suborbital passes
through the di.sk very near the border. On reaching the lower surface this
tube makes a broad curve back, and joins the subrostral opposite the nostril.
A short orbito-nasal connects it with angular and jugular, both of which are
short. The nasal is bent back behind the nasal valves. The median is very
short. Near the middle of their length the elongate prenasals are curved
toward each other. The oral crosses the median line behind the mouth, but is
disconnected from the angular. Around the anterior border of the abdominal
chamber, beneath the coraco-scapular arch, there is a sternal canal, which
differs from the others in being more open; it does not cross the middle. This
canal was not seen in Pristis.

Syrrhina.

Syrrhina brevirostris (Plate XXV.). A description of the canal system in
this species would duplicate that of the preceding, excepting that rostrals, sub-
rostrals, and prenasals would be found to be greatly shortened. Other points
of difference, less important, are seen in the smaller number of branches of the
tubules and the undivided condition of the sternal. A study of the canals
of this species discloses little that favors separation from Ehinobatus, since it
differs less from species of that genus than some of them do from each other.

90 BULLETIN OF THE

Uraptera.

On Uraptera agassizii (Plate XXVI.) the upper surface is pretty well covered
witli tubes and tubules. On the ventral aspect the main tubes are all present,
but the tubules are few and short. There is nothing in the canals that will
distinguish this genus from Eaia.

On the shoulders, the laterals are thrown decidedly outward; farther for-
ward, they make a broad curve toward the vertebrae; and they converge toward
the base of the tail. Their tubules are of medium length and are most numer-
ous above the abdominal region. The pleurals reach far out on the pectorals,
enclosing an area, convex forward, somewhat lenticular in shape. At the angle
opposite the shoulder they send back a strong branch with many tubules on
its outer side. The most of the pleural tubules pass forward ; a few, espe-
cially of those near the anterior border, turn back. Opposite the eye the
pleural is connected with the suborbital by several tubules; thence it bends
toward the margin and descends about half-way from the orbit. A strong post-
scapular goes to the hinder angle of the pectoral ; the greater number of its
tubules are directed outward.

The occipitals are short and greatly divergent. Between the eyes the cra-
nials curve toward each other; in front of the orbits they bend apart; and on
the rostrum they converge gradually to their points of descent, near the end.
Outward from the curve in front of the orbit a tuft of tubules extends from
each. The suborbitals take a tolerably direct course to the end of the snout,
but pass down some distance before reaching it.

On the lower surface, the pleurals bend out opposite the first gill cleft and
inward opposite the shoulder girdle, in both the backward and the return
courses. They neither reach back behind the middle of the abdomen, nor out
to the middle of the pectorals.

The orbito-nasals are elongate, converging in front. The lower suborbitals
are only of moderate length, diverging backward. On the greater part of the
length of the snout the subrostrals are parallel with the prenasals; leaving
the latter finally, they pass outward and then forward to make a close fold on
themselves before taking a transverse direction in which they meet the subor-
bitals. At the inner edge of the nostril the nasals make a sharp bend, then,
converging forward, they unite to form a short longitudinal median. From
the median the prenasals at first bend outward rapidly, then converge gradually
toward the tip, near which they end without connections.

The following peculiarities are among the more likely to prove characteri.s-
tic: the shape of the pleural area, the short occipital, the length of the tubules,
the closeness of the fold in the subrostrals, and the longitudinal median.

Raia.

Rata Icevis (Plates XXVIT.-XXIX.). Some resemblance is seen in the shape
of this species and that of Uraptera, and there is still more in the arrangement

MUSEUM OF COMPARATIVE ZOOLOGY. 91

of the main canals. If these vessels alone were taken'into consideration, more
than specific distinction -would not be accorded the two types.

In comparison with the preceding the laterals on the smooth Skate approach
each other more gradually behind the shoulders and more abruptly in front
of them. Over the gills the branchial area is wider, and in general it is more
irref^ular in outline. The majority of the tubules on the hinder branch of
the pleurals run forward or outward, and on the post-scapular, toward the
hinder margin, a number of them turn backward.

The occipital is rather long. At the side of the eye a branch, from the
suborbital, turns back in the direction of the branchial area. Three or four
tubules connect the orbitals with the pleurals; the latter go down near the
margin, more than half the length of the snout in front of the skull.

Beneath the disk the subrostral is parallel with the prenasal more than half
the distance to the mouth; it then turns outward, and returning makes a sharp
curve and fold, not quite as close and complete as that of Uraptera, after
which it goes back obliquely, instead of transversely as in that genus, thus
bringing about a shortening of the orbito-nasal. The pleural lies close by the
side of the angular and orbito-nasal; opposite the mouth it bends outward
a short distance, then stretches back almost directly toward the posterior angle
of the pectoral; and about as far back as the middle of the abdomen it turns
to the coraco-scapular arch to meet the jugular. The space included is narrow
in front, and much broadened at the shoulder girdle. There is a moderate
amount of curvature in the nasal. The median is transverse. At the median
the prenasals make a broad bend ; they are not connected in front. The oral
is disconnected, and is in two sections.

Plate XXVII. shows the tubes and tubules of the npper surface; Plate
XXVIII. gives (fig. 1) the hyaline mucous ducts of the " ampullae of Loren-
zini,'' and (fig. 2) the main tubes of the canal system of the same surface ;
and Plate XXIX. contains a view of the lower side of the head in fig. 1, and
a sketch of the upper surface in fig. 2.

Eaia oodlata. (Plate XXX.) is one of the species with shorter snouts. In
consequence of the ro.stral shortening, the lengths of the prenasals and of the
rostrals have been decreased so much as to bring their forward extremities
almost back to a transverse line from the end of one suborbital to that of the
other.

Above the thoracic region the curves of tTie laterals are shallower, and the
scapular bends are less prominent, than in R. Icevis. The pleural areas are sub-
triangular, broader posteriorly. The posterior branch of the pleural is the
longer. A strong post-scapular extends from the shoulder obliquely out to
the posterior margin. This tube is provided with tubules on its outer half.
Laterals, pleurals, and orbitals also have tubules, which are more or less
irregular in regard to length.

The occipitals are of moderate length. The cranials have a prominent
curve opposite the fontanelle, and another near the orbitals; their tubules are
short, with the exception of several in front of the orbit.

92 BULLET1^^ OF THE .

On the ventral surface the pleurals are entirely absent, from the posterior
jugular extension. A moderate length obtains in the orbito-nasal. The curve
in the subrostral is very prominent, and a trifle more open than that of R. Icevis.
A decided curve appears in the nasal. It is difficult to determine whether the
median is longitudinal or transverse, it is so very short. Near the mouth, the
prenasals separate widely; forward, they are convergent but not connected.
An oral appears behind each half of the lower jaw.

Torpedo.

Torpedo californica (Plate XXXI.) goes much beyond the following species
in respect to the amount of surface covered by the tubes and tubules. On
the shoulders, the curve, or, better, the scapular angle of the laterals, extends
farther out, and, the batteries occupying a larger portion of the disk, the pleu-
rals are carried nearer to the margins. As in that species, pleurals and sub-
orbitals seem to form a continuous tube.

The aural is longer and straighter and in front of it the cranials converge
more. The rostrals extend farther toward the border, and are better provided
with tubules than in T. mannorata.

At the sides of the suborbitals, and the thoracic portions oF the laterals,
long tubules pass out upon the batteries, nearly across them. A marked con-
trast is presented by this distribution when compared with that of the follow-
ing, or of Narcine, in both of which the tubules venture little if any over the
surface of the batteries. A number of long tubules put out from the aural
toward the shoulders. Behind the angles on the pectorals formed by the
pleurals, there is a strong tubule with several branches; elsewhere the tubule-3
are simple.

There is much irregularity in the cephalic tulies, and it is probable that
there is considerable variation between individuals of the species in regard
to suborbitals and rostrals. In fact, there is great variance in the tubes of
opposite sides of the head of a single specimen. This is well illustrated by
dissections of the head of T. occidcntalis.

Torpedo marmorata (Plate XXXII.) accords substantially in the arrange-
ment of the canals with Narcine hrasiliensis, but the tubes are longer and more
crooked, and the tubules are of much gi-eater length. On the back over the
branchial region the laterals are considerably curved. Surrounding the large
batteries the pleurals approach very near to the front margins of the disk.
These tubes unite directly with the suborbitals. The occipitals are long.
Tlie cranials make a rather sharp curve in front of the eye, and they di.sap-
pear before reaching the end of the rostrum. Among the longest tubules are
those situated posteriorly on the pleurals and the t)rbitals, and anteriorly on
the laterals.

This genus agrees with Narcine in the absence of the canals on the lower
surface.

Absence of post-scapulars, or, better, the backward position of the pleurals

MUSEUM OF COMPAKATIVE ZOOLOGY. 9

o

where joined to the laterals, is an approach toward the Trigonidse rather than
toward the Raia3.

The lack of canals on the lower surface and the junction of pleurals and
orbitals sufliciently distinguish the Torpedoes from other families.

Narcine.

Narcine brasiliensis (Plate XXXIII.). Apparently there are no traces of
the canals on the ventral surface. On the back there is a very simple arrange-
ment of the system. The lateral passes directly to the end of the vertebral
column. Only a moderate degree of prominence is given the scapular curve.
Rather widely separated at the shoulders, the latends converge toward the
back of the head. The tubules are short and not many. At the outer edges
of the batteries the pleurals encroach but little on the pectorals. Half-way
from the eye to the margin they unite directly with the suborbitals.

The occipitals are long. About midway from the eye to the end of the
snout the cranials dwindle and disappear, after sending off a group of short
tubules in front of the eye. The curve around the eye described by the orbi-
tal is somewhat regular, and the canal ends near the margin. It sends out a
couple of short branches near the spiracle, and some shorter ones in front of
the junction with the pleural. A strong tubule reaches backward from an
angle in the pleural, opposite the scapular bends. From the lack of branch-
lets, the small size of the tubules, etc., the total length of the system is much
below the average of the order.

•o^

Potamotrygon.

Potamotrygon motoro (Plate XXXIV.). Upon the shoulders of this species
the laterals bend outward in a variety of curves. The anterior of these, the
greater ones, are concave, the posterior convex. A very little behind the girdle
the pleurals are met. There are two pre-scapular tubules, which do not en-
close an area. Behind the pleural there are several post-scapulars more or less
disposed to unite soon after leaving the main tube. The pleural starts from
several branches which form scapular enclosures. Backward from the aural the
laterals describe the outlines of a goblet, with the bowl e.xtended forward, and
closed by the aural. The occipital tubules are not far behind the ends of the
latter. At the fontanelle the cranial curves are but moderate. Tubules are
numerous on the head, and backward; in general they branch two or more
times. From the laterals the pleurals pass backward and outward , rather more
than in the sketch, until well out on the fin, where they form a somewhat
sharp angle and turn forward in a liroad curve: in front of the eye they turn
inward, and, passing under the orbital, they descend at the fore part of the
skull. Long tubules, with small groups of branchlets at their ends, extend
laterally toward the margins. Two or more tubules join the pleural and the
orbital. A branch goes back from the orbital at the eye, and, in front of the

94 BULLETIN OF THE

orljit, five or six long tubules reach toward the anterior border. The orbital
goes down at the side of the rostral, not half-way from the skull to the end of
the snout. Nearly or quite all of the tubules on this surface are branched.

On the lower aspect the pleurals pass toward the front margin, and, running
parallel with it, send out a number of simple tubules; afterward, along the
middle of the fin, they take a course of some directness to the posterior exten-
sion of the jugular, a little distance forvvard from the pelvis. A single tubule
marks the turning point. Bending around and back, as far as tlie mouth, in
a waved course, the suborbitals make a long loop forward. Behind this loop
they extend toward the gill apertures; in front of the latter they turn toward
the nostrils and meet the subrostrals opposite the mouth. The orbito-nasals
are of medium length, the nasals are long and moderately curved, tlie median
is short, and the prenasals are elongate and close together. Between the nos-
trils each subrostral makes a deep bend, on the nasal valve; they end, at the
side of the prenasals, in a series of rings or capsules connected with each other
by thin transparent tissue, which only near the mouth presents the semblance
of a tube. These rings are closed follicles, which do not appear to be con-
nected with the surface; they seem in most resjjects identical with the folli-
cles of Savi, and trace their origin to obsolescent canals, of which portions
surrounding certain nerve endings have persisted and become closed sacs.
Rings and enlargements also are seen ia the front portions of the prenasals.
On each side of the symphysis, near the teeth, a crooked oral reaches about
half-waj' to the first gill cleft. A short sternal crosses the middle in front of
the pelvic spine.

Distinguishing peculiarities appear in the presence of both pre- and post-
scapulars, in the isolation of the subrostrals, in the groups of tubules on the
front sections of the ventro-pleurals, and in the oddly shaped loop in the
suborbitals.

Disceus.

Disceus strongylopterus (Plate XXXV.). One of the most peculiar canal
arrangements to be found in the order occurs in this genus. Pre- and post-
scapulars are both present, and, outside of the prominent scapular curve in the
laterals, there is a pre- scapular area included by the pre-scapulars. The post-
scapulars are short ; by uniting among themselves or their branches they form
an irregular plexus. First passing back from the laterals, the pleurals then
turn forward at a sharp angle, and, in their course through the middle of the
pectorals, send toward the margin a large number of long tubules, each of
which bears a small group of branchlets at its end. Connecting with tlie orbi-
tals by means of^a couple of tubules, the ]deurals bend back toward the fore-
head, whence they run forward a little more than half the distance to the tip
before descending.

The aural is long and transverse. Starting outward from the aural, the
elongate occipitals turn forward, after sending out the occipital branches. In

MUSEUM OF COMPARATIVE ZOOLOGY. 95

front of the eye, the short cranials have a very sharp and prominent bend; on
the rostrum they are close together and nearly parallel. From the branch
sent back of the spiracles the orbitals incline a little outward, and proceed
thus until more than half-way to the edge, when they turn inward ; close to
the rostrals they again take a longitudinal course for a short distance, and find
their way down, in front of the pleurals, after sending out upon the fin ten or
a dozen long tubules.

On making their appearance on the lower surface the pleurals pass directly
forward; nearing the margin they turn and follow it around, gradually reced-
ing from it, to a point opposite the mouth, where they turn toward the abdomi-
nal chamber. Back of the shoulder girdle they turn slightly outward; and
when opposite the pelvis they turn toward it abruptly, meeting the extension
from the jugular at the edge of the abdomen. On the transverse posterior por-
tion, near the pelvis, there are a few tubules of medium length ; on the
portions anterior to the gills there is a multitude of tubules that reach to
the edge of the disk. The suborbitals emerge a little in front of the pleurals,
■which they cross, to run obliquely back until not far in front of the gills,
where they take an inward and forward course to meet the angulars a little
back of the mouth. As they pass backward they send off nine or ten branches
which by repeated forkings and fusions form networks, the outer limits of
which are the pleurals, and in which the inner areas are large and elongate
polygons and the outer small and short ones. The orbito-nasals are of moder-
ate length, connecting, as in the majority of the Sharks, with the angular and
suborbital posteriorly, and with the subrostral and nasal anteriorly. Each sub-
rostral makes a very prominent bend in front of the nostril; it does not return
far enough to reach the nasal valve; and it ends at the side of the prenasal,
near the skull, in a series of four or five swellings or follicles. No great
amount of curvature appears in the nasals. The median is very short. The
prenasals are close together and nearly parallel; they have several irregularly
placed rings or bunches along their sides. Not far from each angle of the
mouth there is a short disconnected oral.

The sketches were made entirely from the left side of the specimen.

This genus is well distinguished from its allies, the Potamotrygons, on the
one hand, and the Thalassotrygons, on the other, by the disposition of the tubes
on the shoulders, and the orbito-pleural plexus beneath the pectorals.

It is quite possible that the appearance of the follicles on certain of the
canals of the ventral surface, attended by deterioration and disappearance of
the tubes themselves, in this genus and Urolophus, and in species of other
genera, points toward a change made from habits similar to those of the typi-
cal Thalassotrygons, in which the lower canals possessed great utility, to others
leading the individual to remain habitually on the bottom, where the lower
vessels may be comparatively useless, which if persisted in lead to disuse
and ultimate obsolescence of the tubes, as in the Torpedinidee. It is not far to
the conclusion that, through their ancestors. Torpedoes, as well as Potamotry-
gons, were more closely related to the Thalassotrygons.

96 BULLETIN OF THE

Urolophus.

Urolophus hallcri (Plate XXXVI.). A striking feature in this ray is the
absence of the post-scapulars. Their position is occupied by the pleurals and
by the pre-scapular branches. In the genera Raia and Rhinobatus the pleural
met the anterior of the scapular branches ; in tliis genus it is the posterior.
From the anterior part of the scapular curve there is a pre-scapular branch
which connects with the scapular, enclosing a small pentagonal area. Behind
the aural the laterals converge in a gradual curve until rather close together.
The aural is transverse. The occipitals are elongate and diverge forward. At
each end of the aural, on the laterals, there is a small occipital tubule with a
number of brauchlets. The pleurals run forward a little way outside of the
basipterygium of the fin; they pass under the suborbitals and go through close
in front of the skull. One or more tubules connect these tubes with the orbi-
tals. Laterally long tubules extend more than half the distance to the border.
At the fontanelle, the cranials make deep curves outward ; beyond this they
approach each other until nearly in contact at the tip of the snout. The orbi-
tal .sinks deeply into the tissue; at the outer edge of the spiracle a tubule is
sent backward, farther forward others pass out laterally, one or more uniting
with the pleural, and in front five or si.x long ones reach toward the front edge
of the disk. The tube passes down about midway from the fontanelle to the
end of the rostrum.

Beneath the disk, on their appearance the pleurals run forward as far as
to the middle of the snout; thence they turn laterally and describe an arc of a
circle having a radius of about the distance between the first pair of gill open-
ings. This carries them back to a point opposite and near the coraco-scapular,
a point from which they pass directly to join the jugular e.xtension. Radiating
from the outside of the circle there are two- or four-branched tubules of me-
dium lenr'th. Emerrrins on the lower surface the suborbitals make a broad
sweep laterally, then turn back until behind the mouth, and then forward
toward the nostril till they meet the subrostrals. The connections of the short
orbito-nasal are similar to those of Isurus and the Holocephala, and not to those
of the majority of the Sharks. The angular is short, and reaches toward the
inner edge of the gill cleft. The jugular bends outward before running back
along the branchial apertures. No union is apparent between the rostrals and
the subrostrals. From the orbito-nasal the latter are transverse in general
course; they make a prominent bend forward in front of the nostril, and an-
other back upon the nasal valve, thence they pass forward at the side of the
prenasals, growing more and more delicate and transparent, and vanish before
reaching the middle of the snout. Individuals show the peculiar enlargements
or swellings in the tubes, in front of the median, that are seen in Potamotry-
gon. These rings or swollen portions closely resemble the follicles of Savi
They seem to be connected with the prenasals by the tissue of the walls, but
communication with the chambers of the tubes could not be discovered In
the nasals the curves are not very pronounced. The short median is hidden

MUSEUM OF comparatrt; zoology. 97

by cartilage. The prenasals diverge little from the parallel; they are discon-
nected forward. Behind the mouth, on each side, is a disconnected oral.

The tubules of the back are more or less dissected into fine branchlets,
which form small groups about the ends. On the head the branchlets are
close together. Only the principal ones were sketched, but, when possible,
they were followed to their terminations.

Urolophus torpedinus differs from U. hallcri mainly in matters of detail, in
tubules, etc. The specimen at hand has five of the enlargements, at the side
of the prenasal, in the subrostral; the tube ends with the fifth, seemingly
without other anterior connections, excepting by tissue from the walls.

The junction of the pleurals to the anterior scapular branches in Ehinoba-
tus, and to the posterior in Urolophus, indicates that both may have been
secondary attachments ; in other words, that the attachment of origin in the
pleural is that with the orbital

Taeniura.

Tceniura lymma (Plate XXXVIII.) possesses both pre-scapular and post-
scapular branches. What is called the pre-scapular area in Urolophus and in
Dasybatus becomes, by the junction of the pleural tube to the middle of the
outer boundary, a scapular area in this form. Three of the long tubules on
the hinder part of the pectoral diverge from a single short stem, which con-
nects them with the pleural. After connecting with one or more of the orbi-
tal tubules, the pleurals go downward near the forehead. Immediately in
front of the prominent scapular curve the laterals approach each other closely.
They send out the occipital branch near the end of the aural. Along the
greater portion of their lengths they are studded with short tubules, the ma-
jority of which are branched two or more times, making four or more of the
branchlets. On the skull the extremities of the dissected tubules interfere with
each other so much, and become so confused, that it is not possible to present
more than an approximation in the sketch.

An arrangement of the pleurals on the under surface intermediate between
that of Urolophus and Dasybatus is presented by this specimen. The lateral
curve of the tubes is not so round or so regular as in the former genus, and
the tubules are more massed along the anterior edge. Compared with the lat-
ter, the lateral sweep is more regular, and the tubules are much less crowded
along the anterior margin. From their point of appearance near the median
the pleurals describe curves which are entitled to rank as intermediates be-
tween those traced by the same tubes in the genera cited. At each end the
suborbital is bent so as to form a hook ; it meets the angular. Forward from
the nostril there is a deep fold in the subrostral; the tube does not return
quite to the nasal valve, and it cannot be traced beyond the base of the snout.
The nasal is moderately curved. There is a very short median, apparently
transverse. Behind each side of the mouth there is a short oral, which has
the appearance of being affected by the swellings elsewhere seen in subrostrals
VOL. XVII. — vo 2. 7

98 BULLETIN OF THE

or prenasals, possible precursors of the follicles of Savi, evidences of the
action of causes tending toward disruption and destruction of the canals.
Below the disk the tubules are rather short and are somewhat separated, but
not so much so as in Urolophus.

The great ditlerence of the canal distribution as compared with that ob-
taining on the Potaniotrygons is evidence to be added to that advanced by
the writer in 1877, of the necessity of separating the river Trygons from the
species properly belonging to the genus Tsniura.

Dasybatus.

Forward on the thoracic region of Dasybatus nudus (Plate XXXIX.) the
laterals are nearer to each other than they are above the abdomen. On the
shoulder the curve ic moderately prominent. There is a pre-scapular branch,
and also an area. There are no post-scapulars. The pleurals do not extend
much farther out than the basipterygia of the pectorals; their tubules reach
more than half-way to the margin of the disk, and end in small groups of
branchlets. The pleurals descend rather close to the forehead. The aural
is moderate; the occipitals are longer; and the occipital branches may spring
either from the occipitals themselves, or from the laterals, or from both, as
may happen, though most often they appear on the first. Posteriorly the
cranials converge; the orbital curve is pronounced; and the canals seem to end
on the snout. The orbitals cross the pleurals twice; they then go through to
the lower face of the disk, more than half the length of the rostrum from the
forehead. Their tubules are long and are branched at the ends. One only
joins the pleural.

Under the disk the figures outlined by the tubes are still more character-
istic. Within the subrostral loop, in front of the nostril, the pleural makes its
appearance. From this point it sweeps out and forward toward the tip of the
snout, crossed by the subrostral once and by the suborbital three times. Be-
fore reaching the extremity it turns, and, running close along the anterior mar-
gin, sends forward a great number of fine short tubules. Near the outer angle
of the pectoral it bends across the fin toward the pelvis, in front of Avhich it
meets the post-jugular extension. All of the tubules on this surface are along
the pleural in front. Four areas are outlined by the suborlntal. Only one of
them is completely circumscribed by it, from the fact that two of its branches
end without connections. With the aid of the pleural, in front, the otherwise
open areas one, two, and four are enclosed. The first goes back as far as the
nostrils, the second as far as the mouth, the third as far as the middle of the
space between the mouth and the gills, and the fourth area ends opposite
the second gill opening. Suborbital and subrostral meet at the short orbito-
nasal. Angular and jugular are both crooked in irregular flexures. The sub-
rostral is very much bent and folded; a prominent loop extends forward in
front of the nostril, and another upon the nasal valve. This canal disappears,
without visible connections, at the base of the snout The nasal has not a

MUSEUM OF COMPARATIVE ZOOLOGY. 99

great deal of curvature. The median is short. The prenasals are long. On
each side of the middle behind the lip is an oral that extends but little farther
than the corner of the mouth.

Marked vaiiatiou in the branches of the suborbital occurs on the specimen
sketched. Of species similar in shape, Dasybafus walga resembles this one
very much in the patterns described by the orbitals, but D. zugei is even
more simple than D. dipterurus in respect to the same tubes. In the first case
cited the similarity is so close as to raise doubts of the specific distinction of
the two.

Dasyhatus dipterurus (Plate XL.). Compared with the preceding, this species
shows less prominent scapular curves, sharper bends in the cranials, more
connections between orbitals and pleurals and less distance between their
points of descent, and a larger number of tubules. On the lower surface
the differences are a great deal more pronounced. The pleurals do not
reach so far laterally, and they bear tubules toward the sides and posteri-
orly as well as in front. The suborbitals traverse a comparatively direct
course, though affected by many small flexures, till they reach a point op-
posite the mouth, where they turn toward the nostrils, parallel with their
former route, and meet the subrostrals directly in advance of the first gill cleft.
The subrostrals also are sinuous; they form a prominent loop in front of the
nostril, and, apparently, vanish near the base of the rostrum after advancing
very little on the nasal valve. There is little doubt that the subrostrals and
rostrals join ; the latter pass to the lower surface, and may be traced back half
the length of the snout before the walls of the tubes become so thin and deli-
cate as to be undistinguishable from the surrounding tissue. The condition of
the orbito-uasal in this specimen is one of uncertainty: on one side the subros-
tral and the orbital meet, on the other side the subrostral and the nasal join.

Dasyhatus tuberculatus (Plates XLI,, XLll). Differences between this spe-
cies and the preceding are numerous, and very noticeable. But a moderate
degree of prominence is to be seen in the scapular bends of the laterals. The
scapulars and the pre-scapular area are separated. The occipital branch is situated
at the end of the aural. An intricate orbito-pleural plexus is formed, in which
the spiracular branch of the orbital is concerned, with the usual anteorbital
tubules. The tuljules are abundant, elongate and branched ; the posterior one
on each pleural is forked. Toward the forehead the cranials diverge gradually ;
opposite the fontanelle the curves are strong and sharp. Half-way from the
eyes to the end of the snout the orbitals pass to the lower surface. On the
same transverse line, below the snout, are the points of emergence of both
the orbitals and the pleurals, the latter being a tritie farther from the prenasals,
although on the top they started down close to the skull.

Considerable resemblances are seen in the outlines traced by the pleurals on
the ventral aspects of the three species sketched, from this genus. There is
the same outward prenarial curve, the same course along the anterior border
with the great number of short tubules, and a similarly crooked loute across
the pectorals in the direction of the pelvis. The suborbitals connect with the

100 BULLETIN OF THE

subrostrals, as in both of the preceding, but do not reach them by a compara-
tively simple course, as in I). dijAerurus, or through complex areas, as in
D. nudus. At the start they run forward to turn sharply back and to the
side, before reaching the pleurals; then they commence a series of perplexing
and seemingly erratic turns, doublings, and zigzags, that ultimately bring them
opposite the second gill clefts and thence forward to the short orbito-nasals.
Equally crooked is the course of the subrostral ; it makes two prominent loops
at the side of the nostril, one in front of it, and another at its inner side upon
the nasal flap, before making its way directly to meet the rostral at the tip of
the rostrum. The nasals partake slighth' of the tendency toward sinuosities,
as also the prenasals, orals, angulars, and jugulars. The median is short.

No attempt has been made in the drawing to follow the tubules of the head
or back. The laterals continue along the sides of the tail throughout the whole
of its great length. Under the snout the subrostral is difficult to trace, so
much so that the connection with the rostral may yet be considered an open
question.

Pteroplatea.

Plates XLIII.-XLV.

Nowhere else in the order, so far as it has come under our notice, does the
development of the canal system attain such a degree as in this genus. So
great is the number of tubules and branchlets that the larger portion of the
upper surface is a tangle of minute vessels. They are most closely grouped in
a broad band along the anterior margin, and on the head ; posteriorly they are
not nearly so much crowded. A space entirely unoccupied by them is found
on the middle of each pectoral, whence it extends upon the Ijranchial area.
Smaller spaces, (juite as free from them, appear in the scapular areas, and above
the abdomen behind the scapulars. On the ventral surface the extent of tlie
canals is not so remarkable; it is not much greater on this side of the disk than
in species of Dasybatus. The only cause that suggests itself for such an ex-
traordinary development of the system on forms that seem so poorly provided
with means of progression, of defence, or of procuring subsistence, is, a greater
dependence on vibrations in the water for knowledge of the presence of ene-
mies or of prey.

Pteroplatea hirundo (Plate XLIII. fig. 1). Resemblances in shape of disk
between this species and P. valenciennii are accuinpanied by similarities in the
figures outlined by the main canals. The pre-scapular area is large, and lies in
front of a scapular network, in the formation of which the post-scapuiar is also
con „d. Anteriorly the smaller canals are less numerous and much more
loosely arranged than in either of the following species. Posteriorly, also, the
vessels are less abundant, and their general appearance is more straggling and
scattered than ' the case on the same locality in those forms. The pleural
appears on the lower surface about half-way from the median to the tip of
the snout.

MUSEUM OF COMPARATIVE ZOOLOGY. 101

Pteroplatea marmorata (Plate XLIII. fig. 2). Greater length oi disk and
less lateral extension naturally bring about differences between this species
and the following, in regard to shapes and outlines of areas, etc. For instance,
the curve in the pleural behind the shoulder is comparatively deeper and
shorter, and the finger-like area, projected toward the outer angle, stretches
obliquely backward instead of nearly straight outward.

A pair of strong pre-scapular branches are situated close together on the for-
ward portion of the scapular curve. These are not connected with the scapular
area, which is at some distance from them and close to the pleural canal. An
elongate post-scapular branch lies near to, and for some distance parallel with,
the pleural. The masses of tubules and branchlets are more compact than in
the preceding, and less dense than in the following. On the lower surface the
pleural emerges two fifths of the distance from the median to the tip.

PterojAatea valenciennii (Plates XLIV., XLV.). Neither of the other spe-
cies figured possesses so great a number of tubules as this one. Forward, the
tubes are completely hidden. The laterals may be traced without removing
the smaller vessels, except close to the back of the head. In the scapular sec-
tion the curves are extensive, though not very prominent toward the pectoral.
Near the aural the tubes are rather close together; from their bends in this
vicinity long tubules, with two or more branches, extend back beyond the
shoulders. Long tu,bules, also, put out from the sides of the laterals over
the abdominal chamber, beyond which the main canals continue to the end
of the tail. A couple of pre-scapular areas lie in front of the scapulars ; the latter
originate in a plexus of scapular and post-scapular branches, in which it is diffi-
cult to trace the main line. A broad shallow bend, toward the hinder margin,
brings the pleural behind and outside of the middle of the pectoral; there it
turns forward and slightly inward till in front of the middle, where it turns
directly toward the outer angle, making a deep notch, open backward. Some
distance from the angle of the fin the tube turns toward the eye, nearly paral-
lel with the margin, and, after meeting four or five of the suborbital tubules,
descends near the forehead between the suborbital and the cranial. Anteri-
orly the tubules are numerous, branching into a thicket ; posteriorly they are
not so many or so short, and do not present such a confused mass of branch-
lets. Occipital branches occur on both laterals and occipitals. Opposite the
orbits the orbitals make a prominent bend outward. On this bend are the
tubules connecting with the pleurals; behind it are several long branches; and
in front of it are a number of long tubules reaching forward. The orbitals
descend in advance of the pleurals, and much nearer the rostrals. On the
cranial the ante-orbitul curve is sharp and produced.

Ventrally the pleurals extend near the front edges of the pectorals, for more
than two thirds of the length of the latter, before turning toward the abdomen
and meeting the jugular extension a little backward from the scapular arch.
They eniv^rge in the posterior third of the distance from the median to the tip
of the SI )ut; not as shown in the sketch, where the median u too far back and
the pleural too far forward. The orbito-uasal is a mere point, as if the tubes

102 BULLETIN OF THE

crossed each other at right angles. For about half the width of the pectoral,
the suborbital passes directly outward, parallel with the pleural. From its
outermost point it goes to the orbito-nasal. The subrostral lies close to the
side of the prenasal, connects with the rostral, and does not reach the nasal
valve. The median is short. The prenasals are moderate, disconnected in
front. The pleural tubules of this surface, the lower, are most numerous on
the anterior section of the tube; they are not long, and have no branches

Myliobatis.

Plates XLVI.-XLVIII.

Above the disk the canals and their branches extend only about half-way
from the vertebral line to the outer angles of the pectorals. Within these
limits the surface is closely occupied. There is a great tendency to form ro-
settes or mats of branchlets, at the ends of the tubules. From the forehead
back to the base of the tail on each side of the vertebral column, the groups
are dense, and so large as to be nearly continuous as a single one. The pleu-
rals range close to the borders of the branchial areas.

On the lower surface the pleurals run outward and return near the anterior
border of the pectoral, and they then pass backward very near the basal carti-
lages, thus merely skirting the fin. Elongate tubules pass outward on the base
of the tin, hardly covering a fourth of the length, and others pass from the jugu-
lar extension inward upon the abdomen. In great part the orals are longituJinal.
A continuous tube has, in each of the species drawn, taken the place of the
separated sections of the oral, as apparent in the majority of the Batoidei.

Myliobatis aquila (Platea XLYL, XLVIL). Tubules and branchlets are
numerous, above the abdomen, on both inner and outer sides of the laterals,
in this species. The scapular curves are not very prominent; the scapular
angle is sharp. A pre-scapular enclosure, of moderate size, lies in front of the
scapular, and, by union of pleural and the elongate post-scapular branch, a small
scapular area is enclosed. The branchial area is somewhat well covered by mats
of branchlets, from the occipital and from the pre-scapular tubules. The occipi-
tal is elongate, and in some cases it bears the occipital branch ; in others, this
branch rises behind the aural. Tubules are very plentiful on the cranials.
The ante-orbital bend in these tubes is moderate, and their rostral portions are
short. Near the spiracle the orbital crosses the pleural, and it traverses two
thirds or more of the length of the snout before going to the lower surface.
In front of the eye the pleural rises upon the forehead; it makes its appear-
ance on the under side near the nostril. Pleurals, orbitals, and cranials are
thickly beset with tubules on or about the skull.

Beneath the disk the outward course of the pleural lies near the anterior
margin, for two thirds of the length of the latter; the tube then turns back,
making a sharp angle, to take a backward course close to the jugular. It does
not extend as far back as to the pelvis, and the spaces enclosed by it, with the
jugular and suborbital, are very narrow and elongate.

MUSEUM OF COxMPARATIVE ZOOLOGY. 108

Behind its junction with the orbito-nasal the elongate suborbital makes
a prominent loop. A similar loop occurs in the subrostral in advance of
the nostril, and a second appears between the nostril and the median. The
nasal is not greatly curved; it joins the subrostral, which in turn unites with
the prenasal. Median and prenasals are short. The oral is crooked and
branched; it extends back between the branchial clefts of the first pair. Tu-
bules of moderate length, more or less branched, reach out upon the abdomen
Irom the jugular, and upon the posterior areas of the pectoral fin from the
pleural.

Myliobatis freminvillci (Plate XLVIIL). Compared with the preceding this
Eay has a greater number of the pleural tubules massed together opposite the
spiracles, and fewer of them reaching out upon the body of the fin ; and it has
longer rostrals, orbitals, and prenasals. The scapular enclosure appears incom-
plete on its outer boundary. The occipital branch is connected with the long
occipital, and also with the lateral, possibly an individual peculiarity. Be-
neath the disk the areas enclosed by the pleural are more irregular, but the
oral is curved much more regularly. The nasal joins the subrostral, which
unites with the rostral. The orbito-nasal is short, being little more than a
crossing of nasal and angular. Subrostrals and prenasals are not united.

Aetobatus.

Aetobatus narinari (Plate XLIX.) has but slight scapular bends in the lat-
erals, and, apparentlj', has neither pre-scapular areas nor pre-scapular branches.
A post-scapular branch or two enclose a very small space. On the pleurals there
are few branches ; those that exist are long, reaching beyond the middle of the
fin. The branching.=5 of the tubules are similar to those of Myliobatis and its
allies. Opposite the end of the aural on each side is an occipital branch.
The occipital is long. A spiracular branch was not discovered on the orbital.
For a short distance this tube unites directly with the pleural, without the
intervention of tubules, as in most Batoids; it crosses the track of the crasial
twice, in front of the skull, and it descends not far from the tip of the rostrum.
The pleurals descend much nearer the fontanelle.

The arrangement of the lower pleurals is similar to that of Myliobatis,
though the canals extend farther outward or backward ; they are hardly so
close together in front, but are closer to the jugulars along the branchial clefts.
From both of the transverse lines of the pleurals, near the forward edges ot
each pectoral, the tubules run toward the front; from the longitudinal portions
of the same tubes they pass outward, and from the binder part of the extension
from the jugular they reach inward. The suborbital and the angular meet
below the posterior edge of the orbit, whence a long orbito-nasal connects them
with the nasal and the subrostral. Both of the curves in the subrostral, that
in front of the nostril and that on the nasal flap, are sharp and prominent ; the
tube joins directly with the prenasal a little way in front of the median. As
is generally the case in the group, the median is rather short; the point at

104 BULLETIN OF THE

which the nasals unite is vertically above, or a little in front of the mesial
forward bend, formed by the junction of the prenasals with the median. Con-
sequently the median may be described as nearly or quite vertical. The pre-
nasals are elongate ; they unite directly with the subrostrals, forming with them
a single tube on each side of the rostral cartilage, as in Myliobatis aquila. In
M. freminvillei, which more closely resembles Aetobatus in shape, these tubes
are closely applied, but remain separate. At each side of the median line the
oral of Aetobatus sends forward a sharp curve, and on the outside of each of
these a similar loop is sent outward; the tube goes some distance backward
from this second bend before turning outward and forward. It ends without
joining the angular.

The closeness of the relationships existing between this genus and the pre-
ceding are asserted in the characteristics of the canal system with as great
emphasis as in any other portion of the anatomy.

Rhinoptera.

Plates T. , lil.

So far as the general features of the canal system are concerned, this genus
resembles both of the preceding. At the same time there are respects in which
it differs decidedly from either of them. The majority of these are due to
difference in the structure of the head, yet the divergences are not wholly con-
fined to this portion. Again, on comparison with Dicerobatus the indications
of close affinities are very conspicuous on the trunk, but on the head the rela-
tionship becomes apparent only on closer study, being masked by the dissimi-
larity in shape.

Rhinoptera brasiuensis (Plate L.). Abrupt bends give the scapular fold in
this type more prominence than it would attain by a gradual curve, as it departs
but little from the main course of the lateral. This fold bears a pre-scapular
and also a post-scapular branch, and between them an elongate pre-scapular
and a much smaller scapular inclosure. Behind the end of the aural, on the
lateral, there is a strong forked occipital branch with a multitude of branchlets.
Leaving the scapular area the pleural goes back and outward a short distance,
where it has the appearance of being crowded back upon itself in a number of
folds; from these it extends with tolerable directness to the side of the head.
Its branches are few and long ; their ends are much dissected. The two tu-
bules in front of the posterior one are forked near the middle of their length ;
the hinder one branches a greater number of times.

All of the anterior cephalic canals are affected by many flexures, as if in
compensation for the short distances between the extremities of the tubes.
Bringing the mouth so far back toward the gill-openings, and ending the snout
below the forehead, gives the rostral canal a vertical direction, and carries
orbital and pleural under the anterior part of the skull. The tubules of the
orbital pass forward on the inclined portion of the forehead. No distribution
of the canals occurs on the upper siirface of the rostral fins ; the tubes seek the

MUSEUM OF COMPARATIVE ZOOLOGY. 105

lower face, going between these fins, and there become more sinuous and make
broad and sweeping bends.

At the side of the head, opposite the angle of the mouth, angular, pleural,
and suborbital are close together and parallel. Below, the pleural emerges
farther back than its point of disappearance on the top. It passes to the side
of the face, thence to the pectoral, where, in its outward and its inward course,
it traces a pair of lines along the greater part of the anterior border of the fin.
Returned from this it runs back nearly parallel with and not far from the
jugular extension toward the pelvic region. The tubules of both the anterior
lines are directed forward ; of the two posterior lines those of the outer line
are extended outward, and those of the inner toward the abdomen, inward.
Passing around on the rostral fin, near' its border, the suborbital reaches a
point on the side of the head, near the corner of the mouth, where it accompa-
nies the pleural while making a long loop outward; coming back from this, it
unites at once with the angular. The orbito-nasal is long, curving toward the
(jral. The nasal itself is neither long nor greatly curved. Two great loops
occupy the whole of the subrostral : one turning forward in front of the nostril,
and the other backward upon the nasal valve. Both the median and the pre-
nasals are short. The latter are not connected with the rostrals. Behind each
side of the mouth there is an oral of moderate length, in which the ends extend
transversely in opposite directions from a median longitudinal section.

Rhinoptera (Zijgobates) jussieui (Plate LI.). Prominent among the features
in which this species differs from the preceding are the increase in the number
of tubules on the cranials, the presence of a group of tubules immediately be-
hind the orbital on the occipital, the extension of the prelateral branch between
the spiracle and the cranial, the shapes of the scapular and the pre-scapular
areas, the augmented number of branches on the posterior scapular tubule, the
more regular curves in the suborbital and the subrostral, and in the union of the
oral across the median line. Besides these there are other particulars of vari-
ance, more or less important, as a smaller amount of curvature in the pleural
tubules of the ventral series, and greater parallelism in the prenasals, seen on
comparison of the drawings. A close relationship of these species is indicated
by the many points common to both.

Dicerobatus.

Dicerobatus olfersii (Plates LII., LIII.) presents a distribution of the corpo-
ral canals that, in the main features, bears much resemblance to that of Mylio-
batis, Aetobatus, or Rhinoptera. There is a similar nearly parallel arrangement
of two sections of the lower pleural near the front margin of each pectoral, and
of two others, closer together, along the basipterygia of the same fin. On the
dorsal surface the likeness to Rhinoptera is the greater. At the shoulder there
is a single large pre-scapular area. Near the scapular arch the sinuous folds of
the pleural are less prominent than in the preceding, but the branchlets of the
tubules nre even more massed toward the posterior angle of the fin. A post-

106 BULLETIN OF THE

aural branch on the anterior extremity of the lateral recalls the same feature
in Rhinoptera. Apparently the number of branchlets and openings is greater
in Dicerobatus than in either of the other genera cited, and they form closer
acroretrutions along the laterals or over the head. Connection between the
laterals, across the vertebral line in the vicinity of the shoulder girdle, has
not hitherto been observed. Still greater differences exist in the cephalic
canals. If a specimen of one of the species of Rhinoptera were to have the
pre-oral fins separated along the median line, and their inner edges carried
upward and outward so as to be united to the skull along the edge below the
eye, the mouth being at the same time much widened, an arrangement of the
canals might be brought about that would present a somewhat near approach
to that obtaining in Dicerobatus, so far as the distribution of the main vessels
is concerned. The affinities between these genera are well indicated in the
canals.

Laterals. — From the aural each lateral passes obliquely outward to the
post-aural branch; thence it takes its way toward the point of junction of
shoulder girdle and vertebral column. Nearing the latter, it sends a couple
of tubes across it to the lateral of the opposite side, and immediatel}' behind
them turns inward and around, under itself, so as to make a rounded loop just
in front of the pre-scapular enclosure. This may be an individual peculiarity.
Behind the area the pleural is met, and farther back numerous tubules are sent
out toward the median line. Half-way to the tail, or farther, some of the
tubules pass to the outer side of the canal. In front of the shoulder seven or
ei"ht tubules are sent inward toward the vertebrae. The greatest branches are
the post-aural and the scapular branch, by which the scapular area is enclosed.

Pleurals. — Each pleural encloses abranchial area of moderate size, that is
widest near the middle of its length and pointed toward each end. Twenty-
four branches pass outward from the canal, in the specimen at hand; the
median reach little more than half-way from the middle of the back to the
tip of the pectoral. The posterior of these tubules are more branched than
the anterior, the latter being short, confused, and irregular. To make its
descent to the lower surface the pleural passes through the edge of the disk, a
short distance behind the spiracle, and drops downward, meeting on the way
several tubes connecting with the orbital, until below the level of the eye,
where it turns forward nearly parallel with, and a short distance below, the
suborbital. With the latter it is connected at narrow intervals by short tubes,
a half-dozen or more in number. Below the pleural, in the suborbital region,
there are about a dozen short tubules with numerous fine branches, the open-
ings of which appear as thickly strewn dots on the surface. Some of these
tubules originate in the pleural, the majority, however, belong to the sub-
orbital. A little distance in front of the eye the pleural passes obliquely back-
ward and inward to the lower surface, making its appearance a very little in
front of the nostril. From this its course is somewhat irregular backward and
outward to a point below the spiracle, whence it turns still more outward and
upward toward the lower side of the pectoral near the anterior border. Oppo-

MUSEUM OF COMPARATIVE ZOOLOGY. 107

site the angle of the mouth three or four tubules are sent downward and for-
ward toward the lower border of the cephalic fin; a couple are sent inward
behind the mouth; and several short ones are extended inward just before
the tube enters its outward course along the border of the pectoral. Toward
the anterior margin of this fin the canal puts forth a large number of short
tubules, and near the outer angle, at the point of turning back on itself, two
elongate branches are extended toward the tip. Each of these branches bears
tubules. Retuniiiig toward the mouth the course of the pleiu'al is not much
farther from the edge. At first after making the bend it sends several tubules
forward, then, near the middle of the fin, a number turn backward; nearing the
first gill opening, some start forward, but turn and cross the canal, and still
nearer, before turning back by the side of the extension of the angular, a few
irregular branches are pushed forward. In their backward track the two inner
sections of these sub-pleurals lie close together. The outer of the two has by
fiir the greater number of tubules; from the shoulder girdle, anteriorly it sends
these outward; near the girdle they are turned inward to cross the inner tube
and reach the abdominal region. Opposite the middle of the abdomen, near the
end of the course, the majority of the branches turn out and backward; a few
only turn in to cross the tube; and at the turning-point of the tube an elon-
gate tubule goes back upon the ventral fin. As the canal goes forward toward
the jugular, it bears. several branches turning toward the ventral region, then
a few that cross the other section of the tube outward; but after leaving a point
opposite the middle of the belly, it bears no others.

Aural. — This tube is elongate and strongly bent back in its middle, behind
the openings of the aqueducts. Near the median line it sends back three or
four tubules.

Occipitals. — From the aural, each of these tubes extends toward the eye, at
the same time making a broad curve toward the branchial area, and sending
several irregular branches in the same direction.

Cranials. — A cranial goes forward from the end of each occipital directly
toward the tip of the outstretched cephalic fin, without passing beyond the
skull. Posteriorly each bears a number of tubules reaching toward the median
line, but beyond a third of the distance forward they all reach outward to the
supraciliary prominence, where the}' end in a band of thickly set punctures.
Anteriorly the tubules are more numerous, more slender, and more crowded.

Bostrals. — The rostral turns rather abruptly back and inward from the end
of the cranial. It runs near the front edge of the snout until about half-way
to the median line, where it passes to the lower surface. On the under side
of the snout it curves broadly in the direction of the prenasals, then, taking a
lateral direction before reaching the mouth, crossing and recrossing the nasal,
and making a bend forward in front of the nostril, it crosses the pleural before
meeting the orbito-nasal, which it joins opposite the corner of the mouth. One
of the rostrals (sr), that on the right side of the specimen dissected, appears
to be abnormal; it crosses the pleural, and a short distance behind it stops
abruptly, making no connection whatever.

108 BULLETIN OF THE

Orbitals. — The junctions of orbitals and cranials are deeply buried in the
tissues of the top of the head. The orbital is directed obliquely out toward
the eye. In front of the spiracle it passes through the cartilage to the side
of the head ; there it makes a shallow backward curve and meets three or four
tubes connecting it with the pleural. Eight or ten similar connections are
made from the suborbital, behind the point at which it is crossed by the pleu-
ral, as the latter passes to the lower (inner) surface between the fin and the
skull. The majority of the branches uniting with the pleural below the sub-
orbital in reality originated in the suborbital, but in crossing the other tube have
become joined to it. After being crossed by the pleural, the suborbital, en its
way forward, makes a number of sinuous windings, and sends forth a number
of strong many-mouthed tubules, which are nearly parallel as they reach ahead.
Anteriorly the tube divides. One section of it passes the edge of the fin to
take a course on the inner side along the margin toward the tip; near the latter
it turns back, in a slightly sinuous track along the middle of the inner surface
of the fin, crossing the pleural, and meets the angular some distance behind its
junction with the subrostral. The other section of the suborbital turns toward
the rostral, running between it and the edge of the snout. Apparently it con-
nects with the rostral, not far from the cranial, and descends, without going as
far toward the median line as the former, to meet the extremities of the pre-
nasals. The connections and extent of this portion of the suborbital are sub-
ject to a little uncertainty on account of the number, excessive delicacy, and
confused condition of the tubules, and the preclusion of injections by the
preservation of the specimen. The openings of the branchlets of the sub-
orbital form an elongate band of pores extending below the eye forward to tlie
upper edge of the fin.

Nasals. — These are strong transverse tubes ; they become calcified as they
approach the median.

Median. — This tube is elongate, transverse, and, like the nasals, enclosed
in a calcified envelope.

Prenasals. — These are of moderate length, calcified posteriorly, delicate and
slender anteriorly, and, apparently, connected in their front extremities by a
very slender vessel from the suborbital and rostral.

Orbito-nasals. -;- The orbito-nasals are of considerable length, and turn out-
ward posteriorly.

Jngulars. — Each angular makes a broad outward curve toward the front
margin of the pectoral.

Orals. — Excessive fineness and delicacy in these vessels makes it ver}"- diffi-
cult to work them out. They were first sketched from the low ridges formed
by the canals on the outer skin. The terminations and finer branchlets, of
course, could not be marked in this manner. On removing the skin, however,
some of the tubules were lost, and it was found better to give the sketch as
taken from the surface.

MUSEUM OF COMPARATIVE ZOOLOGY. 109

HISTORY.

The mucous ducts and the canals weie more or less confused by the
earlier writers. Usually both systems were treated as apparatus for the
secretion of mucus, and for distributing it over the skin. It was a
long time after the structural differences were pointed out before the
difference in function was recognized. On account of the confusion, the
list of authors treating of the canals is made to contain also those ti'eat-
ing of the ducts, as there are in most instances contrasts with the
canals, or references to them, even in such writings as are most ex-
clusively devoted to the ampullse of Lorenzini. And, further, to make
the literature approximately complete on the embryogeny, the innerva-
tion, and the general homologies of the system, it is found necessary to
include studies of similar organs on the Fishes, the Batrachia, and the
Insects. Consequently a few works are cited which have indirect con-
nection only with the subject of this paper.

As early as 1664 the outward openings of the ducts on the skin of
the Skate were noted by Stenonis. Those on one of the Sharks were
described by him in 1669. The information given by Blasius, in 1681,
was drawn from the publication of Stenonis.

Lorenzini, 1678, in observations on the Torpedoes, recognized the ex-
istence of the two classes of vessels, and distinguished them by their
distribution and by their branchings. Following the ducts he dis-
covered their swollen inner terminations, now called the " ampulke
of Lorenzini."

Monro, 1785, in his book on the "Structure and Physiology of
Fishes," figured both ducts and canals. Plate V. of his work traces the
canals on the head and shoulders of a Cod. Plate VI. exposes the ven-
tral ducts and the canals of a species of the genus Raia ; and Plate VII.
shows the ducts of the upper surface of the same Skate. According
to this author each system formed part of "a very elegant structure for
the preparation of the mucus."

Geoffroy, 1802, published his opinion that the mucous ducts of
the Skate were the analogues of the electric apparatus of the Tor-
pedo. His conclusions did not meet with ready acceptance among his
contemporaries.

Jacobson, 181-3, put out a short paper, entitled " Extrait d'un Mc-
moire sur un organe particulier des sens dans les raies et les squales,"
in the " Xouveuu Bulletin des Sciences, par la Societe Philomatique de

110 BULLETIN OF THE

Paris," VI., p. 332, in which he announces the discovery that the ducts
are organs of sense, carrying vibrations from the surrounding water to
tlie nerves. He also pointed out that these vessels could hardly be the
analogues of the batteries, both being found in the Torpedoes. Trevi-
ranus, Knox, and others followed, agreeing more or less completely with
his conclusions. Delle Chiaje, Savi, and other observers, still claimed
that the ducts were to be regarded as " organ i mucipari," distributing
the slime over the surface.

Blainville, 1822, and others of his time and later, among them Miiller,
looked upon the canals as apparatus for the secretion of mucus.

Savi, in 1840, announced his discovery of the " appareil folliculaire
nerveux " to the Scientific Congress at Florence, and a year later it was
published in the " Atti della tei'za Riuuione degli Scienziati Italiani in
Firenze."

Mayer, 1843, arrived at conclusions similar to those of Geoffroy, 1802,
and held that the mucous ducts of the Raife were the analogues of the
electric batteries of the Torpedinidse.

Savi, 1844, sent out his " Etudes anatomiques sur le Systems Xerveux
et sur rOrgane electrique de la Torpille," in Matteucci's work, " Traite
des Phenomenes Electro-physiologiques des Animaux," of which it forms
an appendix. Here he gives a detailed description and figures of series
of follicles on the Torpedo, which are apparently of the same character
as those sketched in the present work, on Disceus, Potamotrygon, and
Urolophus, and which are here proved to be part of the canal system.

AVithout mentioning all the writers who may have touched upon, or
referred to either canals, follicles, or ducts, we may simply call attention
to Wagner, 1847, to whom is to be credited the hypothesis that the
function of the follicles of Savi is to excite the activity of the electric
organs, and then proceed to several of the more important contributions
toward an understanding of one or another of the organs.

H. Miiller, 1851, makes three groups of the vessels, the greater part
of which are to him organs of sensation instead of secretion. To quote
his words, '• Unter der Rubrik ' Schleimkaniile ' sind bei den Knorpel-
fischen verschiedene gebilde zusammengefiisst, von denen nur ein Theil
den Schleimkanalen der Knochenfisch analog ist. Ein grosser Theil der
Kanale bei Knorpel- wie bei Knochenfischen hat bestimmt nicht Secre-
tion sondern Sensation zum Zweck."

Leydig, 1852, also makes three classes of the vessels, one class includ-
ing the ducts, another the canals, and another the follicles of Savi. He
characterizes them thus : —

MUSEUM OF COMPARATIVE ZOOLOGY. Ill

''1, als verzweigte Eohren, die in oder unter der Haut liegen. Sie setzen
zusanimen das System der Seitenlinie also die Seitenlinie selbst und ihre
Auslaufer ;

"2, als nicht verzweigte Rohren, welche iiiit einer erweiterung — Ampulle
— blind geschlossen beginnen und sich auf der ausseren Haut ofFnen ;

" 3, als geschlossene Blasen, die also nioht iu der Haut ausmiinden.

"Mit der ersten und zweiten Classe sind siimmtliche Rochen und Haie ver-
sehen, rait der ersten, zweiten und dritten zusammen bloss die Zitterrochen."

KoUiker, 1856 and 1858, and Max Schultze, 1862, showed the exist-
ence of a sensitive epithelium within the follicles.

Leydig, 1868, brought forward one of the most important contribu-
tions to knowledge of the organs under consideration. It was entitled
" Ueber Organe eines sechsten Sinnes," and it deals with the matter in
the most compi'ehensive way. The three classes of vessels are accepted
as organs of a sixth sense.

Boll's monograph, " Die Lorenzinischen Ampulleu der Selachier,"
appeared in the same year, 1868. As its name indicates, it was devoted
to the ducts, but references to the canals are included.

A valuable addition to the literature, and very exhaustive so far as
the follicles themselves are concerned, is the monograph, " Le vesicole di
Savi della Torpedine," 1875, by the same author. He is able in this
work to give no additional light on the physiological function of the
vesicles. The idea that they are a form of the canals has little in it
that is seductive to him, since it involves, as he says, asci'ibing one
office to two organs of very different structure in the Selachia generally,
or to three diverse organs in the Torpedinidte alone. The hypothesis of
R. Wagner, that the follicles provoked, in reflex manner, the activity of
the electric organ, he claims to have shown in 1873 to be without foun-
dation ; and he maintains that the opinion that the follicles of Savi
represent an organ of electric sense may only be discussed when the
presence or absence of analogous organs is established in the other
electric fishes.

To Balfour, 1878 and 1881, more perhaps than to any other one, we
are indebted for knowledge of the origin and innervation of the canals.
He first found the lateral nerve to originate as the other nerves, and to
push backward, following the lead of the canal and sending branches to
connect with it in the successive segments that were traversed. His
conclusions disagreed with those of Semper and Goette, who claim that
the lateral nerve originates directly from the epiblast of the lateral
line, but the results of more recent study favor his opinions rather
than theirs.

112 BULLETIN OF THE

In connection with the embryogeny, segmental distribution of the
nerve-endings, etc., it is found necessary to refer to a number of publi-
cations relating mainly to other organs of the Selachia, or to similar
organs in other classes of animals. Semper, Goette, Eisig, Dercum, Van
Wijhe, Hoffmann, Wright, and others, have all put forward contribu-
tions which may not be overlooked, though not in most cases directly
connected with the subject of this paper.

Solger, 1878-80, is the author of a number of papers relating to the
microscopical anatomy in Selachia, Holocephala, and Fishes.

Sappey, 1880, in his " Etudes sur I'appareil mucipare et sur le sys-
teme lymphatique des Poissons," did some work on the Selachia, the re-
sults of which are indicated on several plates illustrating the courses and
connections of the canals, as well as of the mucous ducts, of a Skate,
probalily Raia clavata, and of a Shark, probaVjly Galeus. This is the
nearest approach to a delineation of the canal system since the attempt
of Monro, nearly a centui'y previous. Some peculiarities are to be seen
on the plates in Sappey's publication, which apparently make the species
dissected for the drawings to differ greatly from others of their genera.
A number of the items of greatest variance are evidently the conse-
quences of incomplete observations. The most questionable points on his
Skate are these : (1) the connection of prenasal and subrostral ; (2) the
absence of connection between subrostral and rostral ; (3) absence of
junction of suborbital and orbital ; (4) the disunited condition of upper
and lower sections of the pleurals ; (5) the ending of the upper pleural
near the orbit ; (G) the presence of a transverse canal between the cra-
nials in front of the orbital ; and (7) the absence of the aural. On his
Shark neither aural, orbitals, nor orals would appear to have been
discovered.

De Sede, 1884, in his " Recherches sur la ligne laterale des Poissons
osseux," details the results of a number of essays toward a determination
of the uses of the organ. In this work he also instituted a number of
comparisons for the purpose of ascertaining its value in classification.
He occupies the position of a pioneer in the directions of his study.
From his experiments he decides that the line is a tactile organ of ex-
treme delicacy. In the Selachia the canals demand higher rank as aids
in classification than he accords them in the Teleostei, and his conclu-
sion that the apparatus is more necessary to the least migratory fishes
is dii'ectly opposed to what is seen on such Sharks as Alopias, or such
Rays as Dicerobatus.

P.card. 188.0. has made one of the most recent and important contri-

MUSEUM OF COMPARATIVE ZOOLOGY. 113

butions to the literature of the subject. In the main, his conclusions
agree with those of Balfour.

Though the distinctions between the canals, the ducts, and the folli-
cles had by a number of writers been kept prominently in sight for many
years, Professor Agassiz was the first to attempt the use of the canal
system as a basis for homologies, as an aid in classification, or as a means
of tracing affinities, purposes for which it is admirably adapted.

LITERATURE.

N. Stenonis.

Be Musculis et Glaudulis observationum specimen cum duabus epistolis

quarum una ad Guil. Pisonum de anatome Rajac^ etc. Amst., 1664.
Elementorura Myologiae specimen ; accedit Canis Carcharise caput dissectum,
et dissectus piscis e Cauum genere. Amst., 1G69.
S. Lorenzini.

Osservazioni intomo alle Torpediui. Pirenze, 1G7S; Loud., 1705, Angl.
Miscellanea curiosa s. Epliemeridum medico-pliysicorum, Ann. IX. et X.,

p. 3S9; in Miscall. Acad. Nat. Cur., Dec. Ann. IX. et X., obs. 17.
Schneider's Sammlung von anatom. Aufsatzen und Bemerkuugen zur Auf-
klarung der Pisclikuude, Th. I. p. 93, fig.
G. Blasius.

Anatome Animalium. 1G81. De Cane Carcharia, p. 264, from Stenonis.
M. B. Valentini.

Ampliitlieatrum Zootomicum. 1720.
Perrault.

CEuvres diverses de Physique et Mechanique. 1721. Vol. II.
A. Monro.

The Structure and Physiology of Fishes. 1785.

Vergleichung des Baues und der Physiologie der Pische. 1787. Schneider's
Translation.
P. Camper.

In Schneider's Translation of Monro's Tvork, pp. 16-19. 1787.
E. Geoffroy St. Hilaire.

" Sur ranatomie comparee dcs organes e!ectriq\ies de la raie torpille, du
gymnote engourdissaut, et du silure trembleur." In Ann. du Mus., I.
p. 392. 1801.
L. Jacobson.

"Extrait d'un Memoire sur un organe particulier des sens dans les raies et
les squales." In Nouv. Bull, des Sciences, par la Societe Philomatique de
Paris, 1813, VI. p. 332.

VOL. XVII. — NO. 2. 8

114 BULLETIN OF THE

G. R. Treviranus.

Vermisclite Schriften anat. uud pliysiol. Inlialts. 1820, "Ueber die Nerven

des fiinften Paars als Siunesnerven."
Untersuchungen liber die Natur des Meii'^clien, der Tbiere und der Pflanzen,

1832, IV.

H. D. de Blainville.

Principes d'anatomie comparee, 1822, I.

C. Huschke.

Eeitrage zur Phvsiologie und Naturgescbichte, 1824, I.
Isis, 1825, Taf. XI. f. 1.

R. Knox.

"On the Theory of a sixth Sense in Fishes," in Edinb. Jour. Sci., II. 1825.
Ferussac.

Analysis of Knox's paper on a Sixth Sense, etc., in Bull. Sc. Nat., II. 1827,
p. 12.
J. Muller

De glandularum secementium structura, 1830.

S Delia Chiaje.

Instituzione di anatomia comparata, 1836.

Anatoniiche disamine sulle Torpedini, in Atti del Reale Istituto d' Incorragia-
mento alle Scienze Naturali di NapoH, VI. 1840, p. 291.
J. Davy.

Researches Physiological and Anatomical, I. 1839, p. 85, PI. X., XI
P. Savi.

Atti della terza Riunione degli Scienziati Italiaui in Firenze, 1841, p. 334 ;

noted in Isis, 1843, Heft VL
Etudes anatomiques sur la Torpille, 1844; in Matteucci, Traite des Phe-
nomenes Electro-physiologiques des Animaux, p. 332, Xo. 10, PI. III.

F. Magendie.

Phenomenes physiques de la Vie, 1842.
C. Mayer.

Specilegium observationum anatomicarum de Organo electrico in Raiis aua-
lectricis, 1843, p. 9.

G. Carus.

Lehrbuch der vergleichenden Zootomie, 1843, I. p. 329
G. Retzius.

Oefversigt af Kongl. Vetenskaps Akad. Forhandlingar. 1845. Noticed in
Froriep's Notizen, 1848, V. p. 53.
C. Robin.

Bull. Soc. Philomatique, 1846.

Recherches sur uu appareil electrique des Raies. In Ann Sc. Nat., (3* ser.)
VII. pp. 193-204.

MUSEUM OF COMPARATIVE ZOOLOGY. 115

R. Wagner.

Ueber deii feiuereu Bau des elektnscheu Organs im Zitterrochen. In Ab-
hand. der Kon. Gesellsch. der Wissensrb. zu Gottingen, III., 1847.

H. Stannius.

Das peripberische Nervensysteme der Fische, 1849, p. 45.

F. Leydig.

Ueber die Scbleim Kanale der Knochen Fisclie, 1850, in Miiller's Arebiv

Anat., p. 170.
Zur Anatomie und Histologie der Chimsera monstrosa, in Miiller's Archiv,

1851, p. 249.
Beitrage zur mikroskopiscben Anatomie und Entwicklungsgeschichte der

Rocben und Haie, 1852.
Lehrbucb d. Histologie des Menscbeu u. d. Tliiere, 1857.
Ueber Organe eines secbsten Sinnes, 1868, in Nova Acta Acad. Caes. Leop.

Nat. Curios., XXXIV. 93.
Neue Beitrage zur anatomiscben Kenntniss der Hautdecke und Hautsinnes-

organe der Fiscbe. 1879.
H. Muller.

Verbandlungen der phys.-med. Gesellschaft zu Wiirzburg, 1851, pp. 134-144.

E. J. Bonsdorff.

Anat. Beskrifn. af Cerebral Nerverna hos Raja clavata, 1853, in Acta Soc.
Sc. Fenn., V.
Siebold and Stannius.

Zootomie der Fische, 1854, I. p. 105.
C. Eckhard.

Beitrage zur Anat. und Pbysiol., 1858 (1855-81).
A. Kolliker.

Untersuchungen zur vergleicbenden Gewebelehre, in Verhandl. der phys.-
med. Gesellsch. zu Wiirzburg, 1856.

Ueber .... Savi's appareil foUiculaire nerveux, in Verhandl. phys.-med.

Gesellschaft zu Wiirzburg, 1858.
Jobert.

Des Appareils electriques des Poissons electriques, 1858.

Etudes d'anat. comp. sur les organes du toucher chez diverses mammiferes,

oiseaux, poissons, et iusectes, 1872.
R. McDonnell.

Electric Organs of the Skate, 1861, in Nat. Hist. Review, p. 59.

On the System of the Lateral Line in Fishes, 1862, in Trans. Roy. Irish

Acad., XXIV.

F. E. Schulze.

Ueber die Nervenendigung in den sogenannten Scbleimkanalen der Fische
und iiber entsprechende Organe der durcb Kiemen athmenden Amphi-
bien, 1861, in Reichert und Du Bois-Reymond's Archiv fiir Anat. und
Physiol., p. 759.

116 BULLETIN OF THE

Ueber die Sinnesorgane der Seitenliuie bei Fisclien und Amphibien, 1870, in

Schultze's Archiv fiir mikroskop. Anat., VI. p. 62.
On the Organs ot" Sense in the Lateral Line in Eishes and Amphibians, 1872,

Abstr. in Arch. Zool. Exper. et Geuer., I. pp. i-iv.

Max Schultze.

Untersuchungen liber den Ban der Nasenschleimhaut, 1802, pp. 11-13.

A. Dum^ril.

Hist. Nat. des Poissons, 1865, I. 80-86.

F. Boll.

Die Lorenzinischen AmpuUen der Selachier, 1868, in Schultze's Archiv fiir

mikr. Anat., IV. 375.
Beitiage zur Pliysiologie von Torpedo, 1873, in Archiv fiir Anat. und

Physiol., p. 92.
Le vesicole di Savi della Torpediue, 1875, in R. Acad, de' Lincei, II. (2),

p. 385.
Die Savi'sclien Blaschen von Torpedo, 1875, in Reichert's und Du Bois-

Reymond's Archiv, p. 456.
Ueber die Savi'schen Blaschen von Torpedo, 1875, in Monatsberichte Bed.

Akad., p. 238.
Todaro.

Contribuzione alia Anatomia e alia Pisiologia dei tubi di senso dei plagio-

stomi, 1870.
Sulla Struttura dei plessi nervosi, 1872.

A. Goette.

Entwicklungsgeschichte der Unke, 1875.

C. Semper.

Das Urogenitalsystem der Plagiostomen und seine Bedeutung fiir das der
Uebrigeu Wirbeltliiere, 1875.

B. Solger.

Ueber die Seitenorgane der Fische, 1878, in Leopoldina, Heft XIV. p. 74.
Neue Untersuchungen zur Anatomie der Seitenorgane der Pische, 1879.
I. Der Seitenorgane von Chimaera, in Arch, fiir mikr. Anat., XVII. p. 95.

II. Die Seitenorgane der Selachier, 1880, 1. c. p. 450.

Ueber den feineren Bau der Seitenorgane der Fische, 1880, in Sitzungsb.

nat. Ges. Halle, p. 105.
Neue Untersuchungen zur Anatomie der Seitenorgane der Fische.

III. Die Seitenorgane der Knochenfische, 1880, in Arch, fiir mikr. Anat.,
XVIII. p. 364.

Bemerkung iiber die Seitenorganketten der Fische, 1882, in Zool. Anz., V.
p. 660.

A. A. W. Hubrecht.

Beit rage zur Kenntniss des Kopfskeletes der Holocephalen, 1876-77, in
Nicderlaudisches Archiv fiir Zool., III. p. 155.

MUSEUM OF COMPARATIVE ZOOLOGY. 117

F. M. Balfour.

A Monograpli on the Development of Elasmobrancli "Fishes, 1878, p. 141.

A Treatise on Comparative Embryology, 1881, p. 413.
F. Dercum.

The Lateral Sensory Apparatus of Fishes, 1S79, in Proc. Philadelphia Acad-
emy, p. 152.

H. Eisig.

Die Segmentalorgane d. Capitelliden, 1879, in Mitth. a. d. Zool. Station zu
Neapel, I.

J. Chatin.

Les organes des sens dans la serie aniniale, le9ons d'auatomie et de phy-
siologie comparee, faites a la Sorbonue, 1880.
P. C. Sappey.

Etude sur I'appareil mucipare et sur le systeme lymphatique des Poissons,

1879.
Extracts from preceding, in Guide du Naturaliste, II. pp. 29, 54.

J. W. van Wijhe.

Ueber die Mesodermsegmente und die Eiitwicklungsgeschichte der Nerven
des Solachierkopfes, 1882, in Natuurk. Yerhandl. Akad. Amst., XXII.
P. de Sede de Lieoux.

lleclierclies sur la ligne laterale des Poissons osseux, 1884.
La ligne laterale des Poissons osseux (Extr. de la These). Rev. Scientif.,
XXXIV. p. 467.
S. Garman.

An Extraordinary Shark, 1S84, in Bull. Essex Institute, XVI.
Clilamydoselachus auguineus, 1885, in Bull. Mus. Comp. Zool., XII.
J. Beard.

On the Segmental Sense Organs of the Lateral Line, etc., 1885, in Zool.
Auzeiger, VII. p. 220.

118

BULLETIN OF THE

LIST OF PLATES.

Abbreviations.

ang.

au.

cr.

9-
J-
I.

VI.

Angular.

Aural.

Cranial.

Gular.

Jugular.

Lateral.

Median.

n.

Nasal.

0.

Oral.

oc.

Occipital.

on.

Orbito-nasal.

orb.

Orbital.

P-

Pleural.

pn.

Prenasal.

pp.

Post-pleural,

r.

Rostral.

sc.

Scapular.

so.

Suborbital.

sp.

Spiracular.

sr.

Sub-rostral.

St.

Sternal.

Plate
I.

IL

IIL

IV.
V.

VI.

VII.

VIII.

IX.

X.

XI.

XII.

XIII.

XIV.

XV.

XVI.

XVII.

XVIII.

Isitrus punctatus sp. DeKay. With the names and abbreviations. Fig. 1,

top, 2, front, 3, lower, and 4, side view of head.
Chimcera monstrosa Linne. Fig. 1, side, full length ; 2, top, 3, front, 4, side,

and 5, lower view of head.
Callorliynchus antarcticus La C. ; Cuv. Fig. 1, side, and 2, back, full length ;

3, lower view of head.
CaUorhynchus antarcticus. Fig. 1, side, 2, top, and 3, lower view of head.
ScoUodon terrcR novce Rich. ; Gill. Fig. 1, side, full length; 2, half of lower,

3, half of upper, and 4. front view ol head.
Prionodon milherti (Val ) M. & H. Fig. 1, side, full length; 2, lower,

3, front, and 4, upper view of head.

Cestracion tihitro L. ; Duni. (Zijgczna auct.) Fig. 1, lower, and 2, upper

view of head.
Mustelus canis Mitch. ; DeKay. Fig. 1, side, entire ; 2, lower, 3, front, and

4, top view of head.

Triads semifasciatum Girard. Fig. 1, side, entire ; 2, lower, 3, front, and

4, top view of head.
Isurus punctatus sp. DeKay. Fig. 1, side, entire ; 2, lower, 3, front, and

4, top view of head.
Odontaspis americanus Mitch. ; Abb. Fig. 1, entire side; 2, top, 3, front,

and 4, lower view of liead.
Alopias vulpes Gmel. ; Bonap. Fig. 1, front, 2, top, and 3, lower view

of head.
Alopias vulpes. Entire view of side.

Hepiabranchias maculatus Girard. Fig. 1, side, and 2, top view of head.
Cldamydoselachus anguineiis Garman. Fig. 1, side, and 2, top view of head.
Ginglymosloma cirralum Gmel.; M. & H. Fig. 1, entire side; 2, lower,

3, front, and 4, top view of head.

Scylliorhinus caniculus sp. Linn. Fig. 1, entire side ; 2, lower, 3, front, and

4, top view of head.

Helerodontus philippi La C. ; Blainv. Fig. 1, entire side ; 2, lower, 3, front,
and 4, top view of head.

MUSEUM OF COMPARATIVE ZOOLOGY. 119

XIX. Acantkias americanus Storer. Fig. 1, entire side ; 2, half of lower, 3, half

of upper, and 4, front view of head ; 5, side of tail ; G, section of canal
at side of tail.

XX. Sonuiiosus carcharias Miiller; Garm. Fig. 1, top, and 2, lower view of

head.
XXI. Rhiiia squatina L. ; Dum. Fig. 1, top of head, with tubules ; 2, full
length, main tubes ; 3, front, and 4, lower view of head.
XXII. Pristiophorus cirratus Lath. ; M. & H. Fig. 1, top, and 2, lower view
of head.

XXIII. Pristis pectinatus Latham. Fig. 1, top, and 2, lower view of anterior

half of total length.

XXIV. Hhinobatus planiceps Garman. Fig. 1, top, and 2, lower view.
XXV. Syirhina brevirostris Jliiller & Henle. Fig. 1, top, and 2, lower view.

XXVI. Uraptera agassisli M. & II. Fig. 1, lower, and 2, upper view.

XXVII. Raia /cevis Mitchill. Upper surface, showing tubes and tubules.

XXVIII. Raia Icevis. Upper view, showing (1) the hyaline mucous ducts of the
" ampullse of Lorenzini," and (2) the main tubes of the canal system.

XXIX. Raia Icevis. Fig. 1, lower, and 2, upper view of head.

XXX. Raia ocellata Mitch. Fig. 1, top, and 2, lower view of disk.

XXXI. Torpedo californica Ayres. Upper view.

XXXII. Torpedo marmoruta Risso. Upper view.

XXXIII. Narcine brasiliensis Oif ; Henle. Upper view.

XXXIV. Potamotrijcjon motoro M. & H. ; Garm. Fig. 1, upper, 2, lower surface.
XXXV. Disceus strongi/lopterus Schomh. ; Garm. Fig. 1, upper, 2, lower surface.

XXXVI. Urolophus halltri Cooper. Upper view.
XXXVII. Urolophus halleri. Lower view.
XXXVIII. TcEniura lijinma Guv. ; M. & H. Fig. 1, upper, and 2, lower view.
XXXIX. Dasijbatus niidus Gthr. ; Garm. Fig. 1, upper, and 2, lower view.
XL. Dasi/batus dipterurus Jordan. Fig. 1, lower, and 2, upper view.
XLI. Dasybatus tuberculatus La C. : Garm. Lower view.
XLII. Dasi/batus tuberxidatus. L^pper view.
XLIII. Fig. 1. Pteroplatea hirundo Lowe. Upper view. Fig. 2. Pteroplatea

viarmorata Cooper. Upper view.
XLIV. Pteroplatea valenciennii Dumeril. Upper surface.
XLV. Pteroplatea valenciennii. Fig. 1, upper, and 2, lower view.
XL VI. Mi/liobatis aquila L. ; Cuv. Upper surface.
XLVII. Myliobatis aquila. Fig. 1, lower surface, and 2, side of head.
XLVIII. Myliobatis freminvdlii Lesueur. Fig. 1, upper, and 2, lower view.
XLIX. Aetobatis narinari ^wp\\. ; M. & II. Fig. 1, upper, and 2, lower surface ;
3, lower view of end of pectoral; 4, side and lower view of head;
5, tail and hinder part of body.
L. Rhinoptera brasiliensis Mtiller. Fig. 1, upper, and 2, lower view of disk ;
3, lower side of end of pectoral fin ; 4, side and lower view of head.
LI. Rhinoptera jussieui Cuv. ; Gthr. Fig. 1, upper, 2, lower, and 3, side and

lower view of head.
LII. Dicerobatus olfersii ^luller ; Gthr. Fig. 1, upper surface ; 2, side of head.
LIlL Dicerobatus oljtrsii. Lower surtace.

GARMAN, UTERAL SYSTEM

Lith of L S Pund»rMn & Son. Now Hmo. Cor

ISURUS PUNCTATUS.

GARMAN. LATERAL SYSTEM,

Photo Lith of L S Pundonon & Son, New Hovon, Conn

CHIMAERA MONSTROSA.

e.RMAN, LATERAL SVSTEM

Lilh, o( L S Poftdtriofi 6. So". Nm Hmtn, Conn.

ALOPIAS VULPES.

G»BM««,

LATERAL SYSTEM

PUTE XXI,

RHINA SQUATINA.

PtiolD, Lilh of L S. Pnndinon & Son, Now Hortn, Conn,

GIRMAN. UTERAL SVSTEM.

URAPTERA AGASSIZII.

», 0( L & Pun(k<loi> & So-V N«- H*»n Co^r.

GAHMAN. LATERAL S»STEM.

Photo bih o4 L S Pundonon & Son, Now Hoi«n. Conn.

POTAMOTRYGON MOTORO.

GARMAN. UTERAL SYSTEM

Phow Lilh. of L S Pund.non A Son. N«w Hmi^ Conn.

DISCEUS STRONGYLOPTERUS.

GARMiN, LATERAL SYSTEM

DASYBATUS NUDUS.

PhoW L.lh, ol L S Pund«i*)n & &»-. Ni- Hi't". Conn

GARMAN. LATERAL SV:»TEM

Uth e( i S. P«nd«iw" 4 S«n, New Hwn, Conn

OASYBATUS DIPTtRURUS.

GARMAN, LATERAL SYSTEM.

PTEROPLATEA VALENCIENNII.

Pholo Lith o) L S PumtartO" & Son. Nn> HmPn Conn

GARMAN, UTERAL SVSTEM

Phota Lllh ol L

S. PnAd»rM« i Son, N«« Htn". Con

AETOBATUS NARINARI.

GARMAN, LATERAL SYSTEM

Photo UOi. of U i Purxtwion A So«. H^ H**". Conn.

MYUOBATIS FREMINVILLEI.

, L.lk- .1 L S. !>.»«..." i S»-. N- "■••"■ <^""-

SHINOPTERA BRASILIENSIS.

GARMAN. UTERAL SYSTEM

PhoW bit. ot L S PnBd«<»n 4 Son N«w Ha>«n Conn

RHINOPTERA JUSSIEUI.

GARMAN. LATERAL SYSTEM

DICEROBATUS OLFERSII.

Pholo L.lh. of L S Pundonoo & So", Now Hi«n, Conr

CARMAN LftTERAL SVSTEM

Photo Lnh. ol L S Pundatwn & Son New Hb.oh. Conn

DICEROBATUS OLFERSII

GARMAN. UTERAL SYSTEM.

PUTE III.

CO

=>
O

O

CO

X

o

z
>

o

<
o

°fcctc Li'h. 01 I ?. Pu.-de.son S S'-n N^v,

Hav-ifi. Conn

GARMAN, UTERAL SYSTEM

PLATE IV.

J

■^^sai.

-- :^)^ /■

. Jc\\ \

V (

\

Phvtc. L'til of L S. Fnr.derson 4. Son. New Hav«n. Corn,

CALLORHYNCHUS ANTARCTICUS.

CARMAN, LATERAL SYSTEM.

PLATE V.

UJ

<
>

O

z

UJ

<

QC
Itl

h-

O

o
o

_l

o
o

CO

photo Litti. ol L S Pundefson .1 Son. New Haven. Cont..

GARMAN. LATERAL SYSTEM.

PLATE VL

DC
Hi

ca

O

Q
O

Q-

Photo. Ulh of L S. Pnnderson & Son. New Haven, Conn.

GARMAN. UTERAL SYSTEM.

PLATE VII.

O
cc

CD

Z

g

o

<

\-

«5
ui

o

Pholo. L;tK. cf L S. Pnndericn & Son, New Haven. Cortti.

GARMAN, LATERAL SYSTEM.

PLATE VIII.

CO

z
<
o

«>

Z3
_l
tiJ

h-

co

Plioto Liln of L S Pundersod & So.i, New Haven, Conn.

GARMAN, LATERAL SYSTEM.

PLATE IX.

Z3

o

CO

<

ill
<n

m

o

<

Photo Lith. of L. S Punderson & Son, New Haven, Conn

GARMAN, LATERAL SYSTEM.

PUTE X.

CO

O

z

CO

q:

=)
CO

Photo Lith of L S Punderson & Son, New Haven, Conn

GARMAN, LATERAL SYSTEM.

PLATE XI

W

Z3
Z
<

o

UJ

CO

a.
m

<
I-
z
o

Q
O

Photo. Lith. of L. S Purderson &. Son. New Haven, Conn.

GARMAN. LATERAL SYSTEM

PLATE XII.

Photo Lith- of L S Punderson Sl Son, New Haven, Conn

ALOPIAS VULPES.

GARMAN, LATERAL SYSTEM.

PLATE XIV.

Photo Lith. of L S Punderson & Son, New Haven. Cvnn

HEPTABRANCHIAS VJACULATUS.

GARMAN. LATERAL SYSTEM.

PUTE XV.

Photo Lilh of L S Pundereon 4 Son, New Hsven, Conn.

CHLAMYDOSELACHUS ANGUINEUS.

GARMAN, LATERAL SYSTEM

PUTE XVI.

Photo Lith of L S Punderson & Son. New Haven Conn

GARMAN, LATERAL SYSTEM,

PUTE XVII.

CO

_J
O

z
<
o

w

Z
X

cr
O

>
o

CO

Photo Lith, of L S Pundefson & Son, New Haven, Conn,

GARMAN, LATERAL SYSTEM.

PLATE XVII I,

a.
a.

09

3

o

Q
O
CC
lU

Photo. Lith. of L. S Punderson & Son. New Haven Conn.

GARMAN, LATERAL SYSTEM.

PLATE XIX.

CO

3
Z
<

o

a:

LU

<
<

z
<
o

<

Photo. Lith of L S. Pnnderaon & Son, New Haven. Conn.

GARMAN, LATER.,.. SYSTEM.

PUTE XX.

Photo Lith of L. S Punderson & Son, New Haven Conn

SOMNIOSUS CARCHARIAS.

GARMAN. LATERAL SYSTEM.

PUTE XXII.

CO

I-
•<t

oc
O

CO

q:
O

3:

Q.

o

a.

Photo Lith. of L S Punoerson & Son, New Haven, Conn

GARWAN, LATERAL SYSTEM.

PUTE XXIII.

I-
<

O

LlJ
Q.

02

H
GO
DC

Phi3to Lrth. of L S Pundersof^ &. Son. New H-iven, Conn

GARMAN. LATERAL SYSTEM.

PLATE XXIV.

Photo Lith. of L. S Pundersoi & Son, New Haven, Conn

RHINOBATUS PLANICEPS.

GARMAN, LATERAL SYSTEM.

PLATE XXV.

Photo LtlK 0* L S Punderson & Son. New Haven, Conn

SYRRHINA BREVIROSTRIS.

GARMAN, LATERAL SYSTEM.

PLATE XXVII.

Photo Lilh of L S Pundetson & Son. New H^,er Conn.

RAIA LAEVIS.

GARMAN, LATERAL SYSTEM.

PLATE XXVIII.

Photr. u:*fl. ot L S roaJerwr. & Sof. New H-H^n. Codn.

KAiA LAEViS.

GARMAN, LATERAL SYSTEM

PLATE XXIX.

Photo Lith. of L S Punderson &. Son. New Haven. Conn.

RAIA LAEVIS.

GARMAN. LATERAL SYSTEM

PUTE XXX.

Photo. Lith jf L S. Pn.Tdprson & Son New Haven. Conn.

RAiA OCELLATA.

GARMAN, LATERAL SYSTEM.

PLATE XXXL

Photo Lrth of L S Pundofson & Son. New Haven, Conn

TOFPEDO CALIFORNiCA.

GARMAN. LATERAL SYSTEM.

PLATE XXXII.

Photo Litfi of L. S. Punderson &. Son, New Haven, Conn.

TORPEDO MARMORATA.

GARMAN. LATERAL SYSTEM

PUTE XXXIII.

Photo Lith. of L- S Pundonon & Son, New Haven. Conn

NARCINE BRASILIENSIS.

GARMAN, LATERAL SYSTEM.

PUTEXXXVL

Photo. Litti. of L. S Pundofion & Son. New Hwen. Conn.

UROLOPHUS HALLERI.

GARMAN, UTERAL SYSTEM.

PUTE XXXVll.

Photo Lptli. of L S Pundervjn A Son. New Hmen. Con

UROLOPHUS HALLERI.

GARMAN. UTERAL SYSTEM.

PUTE XXXVIII.

Photo Lith of L S Pundo'son & Son. New Haven, Conn.

TAENIURA LYMMA.

GARMAN. LATERAL SYSTEM.

PUTE XLI.

Photo. Lith. of L. S. Pnnderson &. Son. New Haven. Conn.

DASYBATUS TUBERCULATUS.

GARMAN, LATERAL SYSTEM.

PLATE XLII.

Photo U'H. of L S Puf^derscn & Son, Ne-v Haven. Conn

D.AS'-'SATUS TUBilRGULpT US.

GARMAN. LATERAL SYSTEM.

PLATE XLIII.

Plio'.o Lilh of L. S Punderso'i 6l Son, Nev/ Haven Conn

I. PTEROPLATEA HIRUNDO.
2. PTEROPLATEA MARMORATA.

GARMAN, LATERAL SYSTEM.

PLATE XLIV.

Z
Z
UJ

O

z

bJ

-i
<
>

a.
o

Ul

P^ioto Lilh of L S Punderson &. Son, New Haven, Conn

GARMAN, UTERAL SYSTEM.

PLATE XLVI.

Z3

<

<

o

_i
>

Photo Lrth. of L. S Pundenon & Son, N«w Haven. Conn

GARMAN, LATERAL SYSTEM.

PUTE XLVII.

3
O

<

CO

CD
O

_l
>•

Photo bth. of L S Pundenon i Son. New H«wn. Conn.

No. 3. — The Coral Reefs of the Hawaiian Islands.
By Alexander Agassiz.

Before giving the results of ray observations on the Coral Reefs of
the Sandwich Islands, it may be useful to recapitulate the salient points
of the older theories of the formation of the coral reefs, from Chamisso's
(1815-1818) to Darwin's (1842), as well as the modification of the lat-
ter by Dana (1838-1842, 1849), and to enumerate briefly the objections
which have been made to the general application of the theory of sub-
sidence to the special cases examined by later investigators, from Agassiz
(1851) to the present time.

I need only refer to the earlier views of Forster, who imagined coral
reefs to have been built up from the bottom of the ocean, a view which
was naturally untenable after the observations of Quoy and Gaymard on
the limits of depth at which corals apparently thrive, as well as the later
observations of Ehrenberg on the coral reefs of the Red Sea.

Darwin, it should be remembered, examined only the Great Chagos
Bank, and based his speculations on the observations he made on this
single group, supplementing the knowledge, however, by a most exhaust-
ive analysis of the observations and descriptions of others, and a
most thorough examination of the hydrographic charts which had any
bearing on the subject. But no naturalist has had opportunities to
make a personal examination of the conditions of growth of corals and
of coral islands such as have been enjoyed by Dana, as geologist of the
United States Exploring Expedition. His Report on Coral Reefs and
Islands, published in 1849, contains a full account of his own observa-
tions (1838-1842) on the Hawaiian Islands, the Society Islands, the
Samoa and Viti groups, and his theories are based upon his own experi-
ence, far wider than that of any other Vrriter on the subject. He has
therefore drawn but little either from the descriptions of the voyagers of
the early part of this century, or from the hydrographic charts, both of
which form so essential a part in the Darwinian theory of coral reefs.

An examination of the hydrographic charts of the coral reefs, while
interesting, can lead to no so\md conclusion. Well as I know the
Florida reefs and part of the Bahamas, as well as the majority of the
West India coral reefs, I should hesitate to base any general conclusions

VOL. XVII. — NO. 3.

122 BULLETIN OF THE

upon an examination of these charts, much less to attempt any but very
indefinite deductions for local phenomena or conditions.

It seems scarcely necessary to discuss the opinions of Wilkes,^ who
dismisses the whole basis of the theory of coral reefs, besides the special
theory of Darwin, and who calmly says : " After much inquiry and due
examination, I was unable to believe that these great formations are
or can possibly be the work of zoophytes ; . . . I cannot but view the
labors of these animals as wholly inadequate to produce the effects
which I observed." There seems to be only one critical observation
worthy of remark. When speaking of the formation of lagoons he sug-
gests the possibility of the washing influx and efflux of the sea to carry
off in the shape of mud or sand, or in solution, the strata underlying the
central part of a lagoon. Wilkes further says : " It seems almost
absurd to suppose that these immense reefs should have been raised by
the exertions of a minute animal, and positively so to explain the pecu-
liar mode of construction by which reefs of an annular shape are formed,
when in nine cases out of ten they are of other figures." The last part
of his statement seems to have been lost sight of in the discussions on
atolls.

In striking contrast to Wilkes's opinions are the observations of
Couthouy, one of the naturalists of the United States Exploring Expe-
dition, who published in 1842 his views on the Coral Islands of the
Pacific.^ He suggests that corals are limited in their range of growth
by temperature rather than depth, and that wherever this is not be-
low 76° Fahrenheit, then, cceteris. paribus, they will be found to flourish.
Couthouy admits the correctness of the theory advanced by Darwin,
and believes the great thickness of the reefs to have been produced by a
gradual and long continued subsidence of the original shelf of coral, while
the surface was maintained at the same level as at first by the unceas-
ing additions made by the polyps. He also believes that the whole of
Polynesia is at present slowly rising, and gives his reasons for believ-
ing in a former subsidence of a great continent, mainly based upon the
identity of the volcanic and coralline rocks composing the majority of the
islands of Polynesia. Couthouy describes the seaward side of the encir-
cling reef of the Paumotu group as representing a succession of terraces
or plateaus ; the lowest of variable breadth, seldom exceeding one hun-

1 Wilkes, Charles, Narrative of the United States Exploring Expedition, Lon-
don, 1845. Vol. IV. p. 268.

2 Joseph P. Couthouy, " On Coral Formations in the Pacific," Journ. Bost. Soc.
Nat. Hist., 1842, p. 66.

MUSEUxM OF COMPARATIVE ZOOLOGY. 123

dred and fifty feet, and declining rapidly seaward. These terraces as they
recede from the sea become shallower and shallower, having at their ex-
tremity a sort of steep talus extending to the one next below it. This
is probably due to the action of the surf, as the force of the sea becomes
less and less where it is broken towards the shore line, though Couthouy
is inclined to see in these terraces the eflfect of subsidence. But as he
distinctly says that the outer terraces have only twelve to fifteen fathoms
of water on them, about the limit of reef-growing corals, his explanation
does not appear satisfactory even in the case of the terrace of Belling-
hausen's Island, on which he found twenty-eight fathoms.

It seems much more natural to look upon the channels left, either in
the barrier reefs or in atolls, as due to the original inequalities in the
level of the foundation of the reef. The corals would naturally reach the
surface soonest at the highest point, thus leaving passages which might
at particular parts of the reef, and under certain local conditions, be
gradually closed by the active growth of the corals, or might, on the
other hand, remain open wherever the tides or currents rushed through
them with sufficient force to check their increase, or where the silt was
deposited in such quantities as to obstruct it.

It is interesting to go back to Kotzebue's Voyage, and to find in the
chapter on the Coral Islands, by Chamisso, the following : " The corals
have founded their buildings on shoals in the sea ; or, to speak more
correctly, on the top of mountains lying under water. On the one side,
as they increase, they continue to approach the surface of the sea, on the
other side they enlarge the extent of their work." ^ He noticed the more
rapid growth of the coral where exposed to the action of the surf, and the
obstacles to their growth in the middle of a broad reef, due to the amass-
ing of shells and accumulation of coral fragments, and also to the forma-
tion of coral land by the cementation of the calcareous sand, gradually
increasing in thickness till it is covered by the sea only during certain
seasons of the year. He also noticed the formation of reefs more or less
circular, with an interior sea, having a depth sometimes of thirty to
thirty-five fathoms, which he explains from the action of the natural
causes above enumerated.

Dr. Guppy ^ justly says in Nature : " The development of the new
theory should be kept in mind. Chamisso seventy years ago advanced
the view that an atoll owes its form to the growth of corals at the
margin and to the repressive influence of the reef debris in the in-

1 Kotzebuc, Vol. HI. p. 331, London, 1821.

2 H. B. Guppy, Nature, March 15, 1888, p. 462.

124 BULLETIN OF THE

terior ; but this view gave no satisfactory explanation of the foundation
of such a coral reef, and Darwin was driven to his theory of subsidence.
The great defect in the way of Chamisso was, however, removed by Mur-
ray, who supplied the foundation of an atoll without employing subsid-
ence, and investigation in the Florida Sea (Agassiz) and in the Western
Pacific (Guppy) have confirmed his conclusions. The forms of reefs he
attributed to well known physical causes. Both Semper and Agassiz
have dwelt upon the importance of other agencies, and in our present
state of knowledge it will be wisest to combine in one view the several
agencies enumerated by them as producing the different forms of coral
reefs. On the outer side of the reef we have the directing influence of
the currents, the increased food supply, the action of the breakers, etc.
In the interior of a reef we have the repressive influence of sand and
sediment, the boring of the numerous organisms that find a home on
each coral block, the growth of Nullipores, the solvent action of the
carbonic acid in the sea-water, and the tidal scour. These are all real
agencies, and we only diff'er as to the relative importance we attach to
each. Future investigations will probably add others to the list, besides
ascertaining the mode and degree of action of each case."

According to the theory of subsidence, there is no limit to the thick-
ness which a coral reef, rising out of deep water, may attain. It may
rest upon rocks of any material situated in a region of subsidence, while
naturally it would extend mainly vertically, its horizontal width being
comparatively slight. According to the other theory, which does not
call upon subsidence to explain the formation of barrier reefs or of atolls,
the base of coral reefs, whether atolls or barrier reefs, may be any pla-
teau or eminence, either volcanic or the product of accumulations of lime-
stone banks, which have reached the requisite height for the growth of
corals. The presence of such limestone banks and of eminences, either
volcanic or other, which have reached the level at which corals will
flourish is nowhere better exemplified than in the West Indian region,
where we find atolls, barrier reefs, and fringing reefs in a region which
is eminently one of elevation. In an area of elevation we may have
comparatively thin reefs, forming a mere casing to the core upon which
they have grown, as Guppy has shown in the case of one of the Solomon
Islands.^

Mr. J. J. Lister, the naturalist of H. M. S. Egeria, Commander Al-
drich, describes Christmas Island as being a succession of horizontal
terraces, marking the pauses of its gradual elevation during which a

1 Nature, December 29, 1887, p. 203.

MUSEUM OF COMPARATIVE ZOOLOGY. 125

fringing reef was formed.* There was a cap of coral limestone over
the whole island ; the very top of the island, twelve hundred feet high,
being a block of worn and undermined coralline limestone, then tiers of
cliffs intervening between the top and existing sea cliffs. " Christmas
Island thus appearing to be a remarkable instance of the complete casing
with coral of an island, which, from the time that its first nucleus came
within the reef-building zone, has been steadily subjected to a movement
of upheaval varied by pauses, during which the cliffs were eroded by the
sea."

That the volcanic nucleus has not been exposed is undoubtedly due,
as has been suggested by Guppy,^ to the fact that the upheaved island
has not been exposed to denuding agencies for a sufficiently long period
of time.

Murray,^ who had unusually good opportunities for examining nu-
merous coral reefs of the Pacific, published a remarkable paper on the
formation of reefs in the Proceedings of the Royal Society of Edin-
burgh, in which he gives an explanation of the formation of channels
between barrier reefs and the mainland band, and of the lagoons of
atolls, based upon the solvent action of sea-water saturated with car-
bonic acid upon coral limestone. That this solvent action is a powerful
factor in corroding the surface of coral reef, and carrying off surplus
limestone in solution is not to be denied, but to consider it the principal
cause of the formation of lagoons and of channels between barrier reefs
is perhaps pressing the theory too far. It undoubtedly has acted in
many cases powerfully enough to corrode the whole surface of reefs ex-
posed to action of water so saturated with carbonic acid. I would re-
fer specially to the surface of reefs like the fringing reefs of Honolulu,
the corroded breccia reef rock found at many points of the Keys of
Florida, and the evidence of the same action to be found on the shore
deposits of coral along the whole northern coast of Cuba, where the
shore reef exists parallel with the great Cuban barrier reef. Similar
action, of course, is taking place constantly in limestone districts, through
which waters saturated with carbonic acid percolate, forming caves and
other cavities so characteristic of these formations.

* A similar condition of things exists at Barbados, where tlie volcanic reef centre
crops out only on the summit of the island, and the sides of the cone are covered
with coral reef terraces, which are one of the most characteristic features of the
island as seen from the sea. A. Agassiz, Three Cruises of the Blake, figs. 39, 46.

2 Nature, January 5, 1887, p. 223.

8 Murray, John, Proc. Royal See. Edinb., 1879-80, p. 505.

126 BULLETIN OF THE

Wherever in that region the fringing reef has attained any consider-
able width, we find that upon the portion nearer shore, where the corals
once flourished, they have died because the extension of the reef towards
the shore has excluded them from contact with the fresher sea-water
outside. This part of the reef has been corroded and eaten off, or was
dissolved, aa Murray suggests, by the action of the carbonic acid held
in sea-water, which absorbs a large amount of carbonate of lime.

But it should not be forgotten that this solvent action of carbonic acid
in sea-water cannot be considered as the chief agent in the formation of
barrier reefs. Take, for example, the case of the Florida reefs, or that
of the great barrier reef of Australia. The former are so far from the
neighboring chain of keys, and the latter so distant from the adjacent
mainland, that such an explanation of the presence of the channel sepa-
rating the one from the keys, the other from the shore, would involve the
solution and disappearance of the reef itself.

We are inclined to look upon the depth and extension of the ridge or
plateau upon which a barrier reef first establishes itself as the chief
cause of its growth and final form. Such a plateau having reached the
level at which corals flourish, the reef begins to grow, and its distance
from the mainland or from the adjacent islands is thereafter determined
by the contour lines of the submerged extension of the land seaward or
landward. Nevertheless, the effect of the relatively clear or muddy
water on the sea and land faces of the incipient barrier reef cannot have
failed to exercise an important influence on its seaward or landward
development.

Murray ^ clearly shows that the solution of dead carbonate of lime
shells and skeletons by sea-water is as constant as its secretion by living
organisms. He considers it probable that on the whole there is more
secretion than solution, and that there is at the present moment a vast
accumulation of carbonate of lime going on in coral areas no one familiar
with the subject will deny. This secretion diminishes with the depths,
while the rate of solution perhaps increases under pressure. He com-
pares it to the action of aqueous vapor in the atmosphere over land sur-
faces. When precipitation is in excess of evaporation fresh-water lakes
are formed ; when evaporation, on the contrary, exceeds it, salt lakes
are formed in inland drainage areas.

The discussion on the theory of solution which has taken place in
" Nature " between Reade and other geologists, does not appear to cover
the ground. The objections are mainly made by investigators who know

1 Nature, March 1, 1888, p. 414.

MUSEUM OF COMPARATIVE ZOOLOGY. 127

little of coral reefs from their own observations, some of whom have
ignored or flatly denied facts which can hardly be dealt with in so sum-
mary a fashion.

According to Reade,^ it seems very evident that if we accept the dis-
solution theory for the origin of coral lagoons, it seems impossible to
believe in the building up of calcium carbonate, or volcanic platforms, or
other peaks, from varying and unknown depths to the levels necessary
for the growth of coral reefs. If, on the other hand, we believe that
platforms are so built up, it appears equally destructive of the dissolu-
tion theory of lagoons.

In "Nature " of September 21, 1880, Mr. Reade says: "I think the
theory Mr. Murray sets forth, — that the cones or peaks on which he con-
siders atolls have been formed have been levelled up by pelagic deposits,
and thus brought within the limits of reef-building coral growth, — a
very far-fetched idea."

In the same journal, Darwin says : " I am not a fair judge, but I agree
■with you exactly that Murray's view is far-fetched. It is astonishing
that there should be rapid dissolution of carbonate of lime at great
depths and near the surface, but not at intermediate depths, where he
places his mountain peaks."

It is surprising that Reade ^ should have attempted to throw doubt on
the existence of calcareous submarine banks. The submarine banks are
not, as Mr. Reade seems to think, due to the tests of the pelagic fauna
alone. A submarine peak is not built up by the pelagic fauna, but it is
built up by the carcasses of the Invertebrates that live upon it, and for
which the pelagic fauna serves in part as food. Certainly, the amount
of limestone and shells of pteropods alone in some regions is very much
larger than any estimate made by Mr. Reade. The large number of
well known limestone banks of great thickness and extent should make
Buch a discussion unnecessary.

The *' pelagic cemetery " is farther down, and not on the surface, and
I would refer Mr. Reade to my article on the Florida Reefs, in the
Memoirs of the American Academy of Arts and Sciences, 1883, as well as
to the " Three Cruises of the Blake," for such proof as has thus far been
obtained regarding the existence of these huge masses of limestone
banks, eminently fitted, as I think I have shown, to form the base of
such coral reefs as those of the "West Indies, of Florida, of the shores of
Cuba, and of the great Alacran and other reefs on the Yucatan Bank.

1 T. Mallard Reade, Nature, March 22, 1888, p. 489.

2 Nature, AprU 5, 1888, p. 535.

128 BULLETIN OF THE

The following are the principal experiments which have been recorded
regarding the solvent action of sea-water on corals. According to Mr.
Robert Irwine,^ dead or rotten coral of several species of Porites, exposed
to sea-water of 1.0265 specific gravity, and of a temperature of from
70® to 80° Fahrenheit, was found to be soluble to the extent of 5 to 20
ounces to the ton. We have no data to show how far this capacity of
solution is in excess of the deposition of limestone due to the corals
themselves, or to the sand and debris carried into the lagoons or the
inner part of the reefs. No observations have been made regarding the
amount of carbonate of lime existing in lagoons, and in the sea-water on
the sea face of a reef.

Mr. W. G. Reid, in a paper read before the Royal Society of Edin-
burgh, February 6, 1888, observed that the solubility of carbonate of
lime increased with pressure.

Mr. James G. Ross ^ detailed in " Nature " other experiments show-
ing a considerable amount of solution. In a species of Oculina 0.0748
gramme was lost, from a specimen weighing 16.3164 grammes, in twenty
days. In another case, 0.1497 gramme was lost in thirty days by a
mass of Madrepora weighing 15.334 grammes. The above experi-
ments would both indicate the possibility of a very material deepening
of a lagoon by the solution of the coral. At such a rate of solution,
a lagoon four miles in diameter might be deepened one fathom in a
century.

The rotten condition of the old shore reef of Havana,^ completely
honeycombed as it is, shows how rapidly limestone is acted upon by
sea-water. The rotten reef rock of the Everglades, soaked by brackish
water, which is often accumulated in large bodies behind the old reef
ranges, has been described by Professor Agassiz. This water, saturated
with carbonate of lime, often rushes out with considerable volume after
a storm, and produces great havoc with the shore fishes of the adjoining
reef. Their dead bodies often line the shores of the Florida reefs for
miles, when there has been such an outburst of water saturated with
carbonic acid. The existence off shore of bands of sea-water similarly
saturated with carbonic acid may explain the great destruction of fishes
which so often takes place in fishing vessels carrying their catch from
the Florida Reef to Havana. It will have been noticed by all who have
ever seen a coral sand beach, or a breccia beach, or a beach composed

1 Nature, March 15, 1888, p. 461.

2 Nature, March 15, 1888, p. 462.

8 See Three Cruises of the Blake, Fig. B, p. xii.

MUSEUM OF COMPARATIVE ZOOLOGY. 129

of larger fragments of coral and reef limestone, that the constituent
parts of the beaches were, as could easily be seen by the eye, invari-
ably corroded far beyond the condition to which the sand or breccia, or
larger fragments, could be reduced if merely subjected to the triturat-
ing agency of the rollers. An analysis made by Prof. F. W. Clark, the
chemist of the United States Geological Survey, of such fragments,
either as sand or in all intermediate stages up to fragments of coral, or
coral limestone, showed very clearly that the chemical composition of
the pieces was practically the same.

The only analysis known of the chemical constituents of the sea-water
of the lagoon of an atoll is given as determined by Messrs. Stillwell and
Gladding,* from which it would appear that the amount of chlorine was
considerably larger than the amount given in the latest analysis of
Dittmar, and that the water of the lagoon is fresher than that of ocean
water.

It may not be out of place to mention here, that there is a most excel-
lent figure and plan of an atoll in an account of Caroline Island,'^ by
Prof. E. S. Holden, the director of the American. Eclipse Expedition of
1883. Not only is the description of the atoll admirable, but the illus-
trations of the various parts of the island are most characteristic, in-
cluding one of the best figures perhaps of a bird's-eye view of an atoll.
A map also accompanies the description, but unfortunately no sound-
ings are given.

Lagoons without openings are perhaps older lagoons, in which the
openings have from local causes been gradually closing, and from the
porous nature of the surface coral rock there still remains a chance for
the exchange of waters from the interior to the exterior of a reef.

Jukes,* who in 1845 surveyed the Great Barrier Reef of Australia,
came to the conclusion that the " northeast coast of Australia has either
been slightly elevated, or that it has at lea^t not suffered any depression
during a long period of time." From this he has satisfied himself that,
wherever we find coral reefs rising abruptly from unfathomable depths,
they must necessarily have been produced by the depression of the sea
bottom, the corals building on upwards as the bottom was slowly sink-
ing, so as to keep the upper portion of the reef always within the
required depth. The depression of the bottom, according to this view,

1 Mem. Nat. Acad, of Science, Vol. III. p. 96.

* Report of the Eclipse Expedition to Caroline Island, May, 1883, Mem. Nat.
Acad, of Science.

8 Jukes, J. B., Narrative of the Surveying Voyage of H. M. S. Flv, Vol. I. p. 311.

VOL. XVII. — NO. 3. 9

130 BULLETIN OF THE

has occupied a far longer period than that during which the northeast
coast -was either stationary, or had been slightly elevated. He urges
the parallelism of the outline of the Great Barrier Reef with that of the
northeast coast as evidence that the circumstances which modified the
outline of the coast likewise determined the general outline of the reef,
while subsidence would most assuredly produce the results observed on
the northeast coast of Australia if the rate of growth of corals were
absolutely identical with that of the subsidence of the bottom of the
sea. With our present knowledge of the mode of coral reef formation,
it seems unnecessary to explain the existing state of things by a sub-
sidence coincident in rate with the growth of corals, when observation
plainly shows us that there has been only a slight elevation or a stationary
condition of the coast line. Starting from the conditions Jukes imagine?
to have existed before the subsidence took place, only leaving the coast
nearly at its present level, we can imagine a fringing reef to have been
formed slowly, and to have little by little extended seaward, advancing
more slowly as the depth increased, while the talus for the upper limits
of coral grew, it increased in thickness, and to have ended in a barrier
and inner reef with channels very much like the reef we find to-day.

Is it credible that, along the whole length of the northeast coast of
Australia, the subsidence should, for a length of over one thousand
miles, be so nearly identical in amount as to have ended in forming
parallel to it the Great Banner Reef] The same question is one which
must be answered not only for Australia, but for all the atolls and
barrier reefs in the Pacific and other regions where such reefs exist.

We have all over the world many positive proofs of the elevation of
the land, sometimes on a gigantic scale, as in South America, for instance,
up to nearly three thousand feet. Neither can we deny that there are
many points, especially in the Pacific Ocean, where are to be fouHd
areas of subsidence ; but it is by no means proved that this subsidence
has been the main cause of the formation of atolls or of barrier reefs.
In fact, all the later investigations of coral reefs have, without exception,
rejected Darwin's theory of subsidence as explaining the formation of
reefs, and they have looked to other causes, which seemed to them more
natural, as probably more efficient in the growth of reefs. The ques-
tion is not whether subsidence has taken place even in the areas where
atolls or barrier reefs occur, — this may be considered as proved, — but
whether this subsidence has absolutely kept pace with the rate of growth
of corals. It is remarkable that Darwin, who is so strongly opposed to
all cataclysmic explanations, should in the case of the coral reefs cling

MUSEUM OF COMPARATIVE ZOOLOGY. 181

to a theory which is based upon the disappearance of a Pacific con-
tinent/ and be apparently so unwiUing to recognize the agency of more
natural and far simpler causes.

Granting that during the secondary period the great East India islands
were connected with Asia, and that there had been in the early tertiary
period a great subsidence, which may have extended throughout some
parts of the Pacific to the time of the formation of modern coral reefs, —
granting even that the summits of the islands now existing indicate
plateaus upon which the various archipelagos of the Pacific are based,
and point to a former extent of land far greater than now projects above
the surface of the sea, and also that the islands of the Pacific mark a
general subsidence along a line extending from the southeast to the
northwest, as is urged by Dana, — yet there is nothing in all this to show
that the subsidence has been the main cause of the formation of atolls
and barrier reefs, while the existence of such a subsidence in its turn
derives its strongest proof, with many writers, from the existence of atolls
and barrier reefs. As long as we can in so many districts explain the
formation of atolls and of baiTier reefs by other causes, fully sufficient
to account for the numerous exceptions to the theory of Darwin, which
have been observed by so many investigators since the days of Darwin
and Dana, it seems unnecessary to account for their presence by a gigan-
tic subsidence, of which, although we may not deny it, we can yet have
but little positive proof.

Dana has been led to reconsider the earlier and later observations,
and has given his results in the American Journal of Science.^ He most
distinctly rejects Darwin's hypothesis, that the slow subsidence upon
which he counted to form atolls and barrier reefs from fringing reefs
involved the whole central Pacific, besides other large areas, a Pacific
continent having disappeared through subsidence.

Whether subsidence is going on now, or has ceased after the formation
of atolls, which he ascribes to it, seems immaterial. The point at issue
is, how far is it possible for atolls and barrier reefs to begin in an area
of limited extent without a constant alternation of elevation and sub-
sidence. It seems to me that the rocky islets dotting the interior of
Kaneohe Bay could as well be cited as proof of subsidence, as the rocky

^ This part of the theory of Darwin, which seems a natural corollary of his
explanation of coral reefs, is most empliatically rejected by Dana, Am. Journ' of
Science, Vol. XXX. p. 90, 1885, and previously also in his Geology of the Explor-
ing Expedition, in 1849.

2 Am. Journ. of Science, August, 1885, p. 89, and September, 1885, p. 169.

132 BULLETIN OF THE

islets which dot the great lagoon-like waters of the Gambler group,
"leaving scarcely any doubt in the mind that the islets were the
emerged points of sunken lands ; and if this is evidence of subsidence,
then the atoll [of Keeling] which he [Darwin] examined was proof of
further subsidence, that is, one that had continued to the disappearance
of the sinking peaks." This is the proof which Darwin believed to be
almost certain evidence of subsidence.

Dana adds, as an argument in favor of subsidence, the existence of
deep fiord-like indentations in the rocky coasts of islands, both those in-
side of barriers and those not bordered by reefs. Certainly this is a
most unsafe method of reasoning, unless accompanied by sounding in the
fiords to show the continuation of the slope of erosion. As to the non-
existence in the ocean now, and the extreme improbability of the exist-
ence at any time, of submarine volcanoes or chains of mountains having
their numerous summits within a hundred feet of the surface, which has
been a favorite argument against the possibility of a volcanic base for
reefs, the recent deep-sea soundings of the Atlantic in volcanic districts,
like that off the west coast of Africa by the Talisman, have shown the
existence of numerous peaks and submarine banks, which in the track of
oceanic currents would soon be built up to the level at which corals can
thrive, and produce the very conditions denied by Darwin, A similar
state of things has been developed by the soundings of the Blake in the
West Indies.

Dana mentions the great width of a reef as an indication of subsid-
ence. I am unable to see the force of that argument. It seems merely
to indicate the great length of time which has elapsed since it began to
build. We might take for granted the evidence of subsidence as de-
duced by Dana for the Tahitian group and for the Samoan group, for
instance, and yet we should not have the proof that this subsidence
was coexistent with the formation of the different kinds of reefs.

If, as is supposed, we can have submarine banks of limestone formed
upon volcanic mountains or other steep slopes, the steepness of the slope
off the coral reef, does not argue anything in favor of subsidence. I do
not see that the large debris offer positive proof of subsidence, if they
have, as Murray supposes, gradually rolled down the steep talus of the
sea face of the reef, and have, as is certainly the case in the Sandwich
Islands, formed the surface, which may be of great thickness. Dana
infers, from the statement I made in regard to the former connection of
the Windward Islands ^ with South America, that there has been a sub-

1 Bull. Mus. Comp. Zool., 1879 ; Am. Journ. of Science, XVIIL 230, 1880.

MUSEUM OF COMPARATIVE ZOOLOGY. 133

sidence. It may also be that erosion has been amply capable of washing
away the land connections, and forming the banks on which the islands
rest as it were.

As to there not being any mound now approaching the ocean sur-
face in the western border of the Gulf Stream, the past history of the
Gulf Stream itself, of the Florida Plateau, and of the formation of the
Keys of Florida and of the present reef, seem to me to furnish just such
a foundation for reef-building as is required by Dana. The Mosquito
Bank, the Yucatan Bank, and the smaller banks between Honduras and
Jamaica, are all proof that immense limestone banks are forming at any
depth in"th6 sea, or upon pre-existing telluric folds or peaks, constituting
banks upon which, when they have reached a certain depth, corals will
grow. Is it claiming too much for erosion to say that some of the vol-
canic peaks may have been washed away and swept into the sea? Cer-
tainly this is not the case in any region where there is a rainy season.
The Sandwich Islands themselves, greatly modified as they have been by
erosion, furnish the best evidence that isolated peaks may have com-
pletely disappeared. A careful perusal of Dana's own account of the
effect of erosion on their topography, and of Captain Dutton's later
examination, shows how powerful a factor they regard erosion to have
been in these islands. And if we go farther towards the equator, or to
the region of cyclones and tornadoes, the action of erosion will be found
to be far more powerful than in the Sandwich Islands, which are on the
very edge of the rainy season district.

It is somewhat surprising that, in the discussion which has lately
been carried on in the English reviews,^ by the Duke of Argyll, Huxley,
Judd, and others, regarding the new theory of coral reefs, no one should
have dwelt upon the fact, that, with the exception of Dana,^ Jukes,'
and others who published their results on coral reefs soon after Dar-
win's theory took the scientific world by storm,"* not a single recent
original investigator of coral reefs has been able to accept this explana-
tion as applicable to the special district which he himself examined.
It is interesting to note that, however widely Darwin's theory was
accepted and spread in all text-books of Geology, neither L. Agassiz,^

1 " Nature " and " Fortnightly Review."

2 Dana in 1838-1842 ; " Corals and Coral Reefs," in 1872.

* Jukes in 1845, Narrative of the Surveying Voyage of H. M. S. Fly, Vol. L
p. 31, 1847.

* Darwin in 1842, "The Structure and Distribution of Coral Keefs" ; Darwin's
Coral Reefs, 1874, 2d edition.

& Agassiz, L., U. S. Coast Survey Reports, 1851 and 1866; also Methods of Study
(popular sketch). See Vol. VII., Mem. Mus. Comp. Zool.

134 BULLETIN OF THE

who examined the Florida Reefs in 1851, nor Joseph Leconte,^ his as-
sistant, who published subsequently views somewhat different from
those of L. Agassiz, nor E. B. Hunt,^ who promulgated a theory of the
formation of the Florida Reefs, nor A. Agassiz, who spent several sea-
sons in parts of Florida, on the Florida Keys, and on the Tortugas,
was able to accept Darwin's theory as offering an explanation of
the formation of the great reef extending from the Tortugas to Cape
Florida.

Agassiz, while in general he accepted Darwin's theory as applicable to
atolls, yet gave, in 1851, (Report of United States Coast Survey, repub-
lished with additions in the Memoirs of the Museum of Comparative
Zoology,) an account of the Florida Reefs, showing the living reef out-
side of the lagoon, and its position with reference to the line of Keys.
I subsequently gave a number of sections of the same reef from the
Coast Survey maps,® showing the formation of a barrier reef actually
going on, where the reef foundation grows lower and lower, and where
we need not have recourse to the theory of solution for the formation
of. a lagoon. The lagoon we can actually trace from its broadest point
at the Rebecca shoal, where the reef is submerged, to its narrowest point
at the northern extremity of the reef.

Leconte accepts the theory of subsidence as a satisfactory explanation
of the formation of atolls in the Pacific Ocean ; but in Florida, which he
visited with Professor Agassiz in 1851, he agrees with the latter in his
account of the formation of a barrier reef where there has been no sub-
sidence, and then he points to the Gulf Stream running parallel with
the trend of Florida, as the agent which has deposited the great mass of
the Florida bank below the level at which corals can grow. But there
is no evidence that the Gulf Stream ever ran in the direction assumed
by Leconte. Agassiz also accepted the theory of subsidence as generally
explaining the formation of the different kinds of coral reefs, though in
his account of the formation of the Florida Reef he does not go beyond
the depths at which reefs grow, and says nothing of the substructure or
foundation rock. In P'ebruary, 1878,* I called attention to the exist-

1 Am. Jour. Science, XXIII., May, 1857, p. 46, "On the Agency of the Gulf
Stream in the Formation of the Peninsula and Keys of Florida," by Joseph
Leconte ; also Elements of Geology, New York. 1878.

a Hunt, E. B., Am. Jour. Sci., 1863, Vol XXXV. p. 197.

' In the Tortugas and Florida Reefs, Memoirs of the American Academy, Vol.
XI., 1883.

* Agassiz, A., Letter No. 1 to C. P. Patterson, Supt., on the Dredging Opera-
tions of the U. S. Coast Survey Steamer Blake, Bull. Mus. Comp. Zool., V., No. 1,
1878.

MUSEUM OF COMPARATIVE ZOOLOGY. 135

ence of a great atoll, Alacran Reef, on an area of elevation on huge
limestone banks such as those of Yucatan.

Leconte insists on the fact that the Florida Reef, a true barrier reef,
has been formed where there could not be any subsidence, as continuous
increase of land is inconsistent with subsidence. According to Darwin,
barriers and atolls always show a loss of land, only a small portion of
which is recovered by coral and wave agencies, while on the Florida coast,
according to Leconte and Agassiz, there has been a continuous growth
of the peninsular by coral accretion, until a very large area has been
added. ^ He attributed the formation of successive reefs to the suc-
cessive formation of the depth condition necessary for coral growth,
and this latter, in the 'absence of any evidence of elevation, to the
steady building up ,by sedimentary deposits and extension southward
of a submarine bank within the deep curve of the Gulf Stream as it
bent its way round the west coast of Florida.^ The formation of bar-
rier reefs instead of fringing reefs on a coast which has certainly not
subsided, he attributes to the shallowness and muddiness of the bottom
along this coast. Only at a distance of twenty to forty miles, when the
depth of twenty-five fathoms is reached, and when, therefore, the bottom
is no longer changed by the waves, the conditions necessary for coral
growth could be found, and here a line of reefs would be formed, limited
on one side by the depth, on the other by the muddiness, of the water.

According to Leconte the building up of Florida and of the Keys was
due to the co-operation of several agents : —

L The Gulf Stream building up and extending a submarine bank
within its loop, but not in the position assigned to it by Leconte.

2. Corals building successive barriers on the bank, as the latter was
pushed farther and farther southward.

3. Waves beating the reefs into islands.

1 See Smith, Hilgard, Heilprin, and Dall, for the structure of the peninsula of
Florida.

- I can hardly see how Leconte states (Nature, October 4, 1880, p. 558) that
there are barrier reefs in Florida with lagoons from ten to forty miles wide, though
he subsequently (Nature, November 25, 1880) modifies this statement by indicating
this to mean the space between the southern coast of Florida and the line of Keys
(Old Barrier Reef) which widens from a few miles at its eastern part to more than
forty miles in its western part. But this is also misleading, as it refers to the time
when the Keys formed the reef, while now the channel between the line of Keys
and the present reef gradually widens from a narrow lagoon near Key Biscayne to
from six to ten miles wide, opposite the Marquesas, and is about one hundred and
fifty miles long.

136 BULLETIN OF THE

4. Debris from the reef and Keys on the one side, and the mainland
already built (Keys) on the other, filling up the successive channels,
and converting them first into swamps and finally into dry land, in all
of which he agrees with Agassiz's explanation of the causes which have
built the Florida Reef.

Neither was I able, when visiting the Alacran Reef, the reefs of the
Windward Islands,^ the elevated reefs of Barbados, of San Domingo, and
of Cuba, the great barrier reef of Cuba, and becoming acquainted with
the immense limestone banks so characteristic of the Caribbean region, to
satisfy myself that Darwin's theory of subsidence gave an explanation of
the condition of things now existing in an area of elevation, and includ-
ing all the types of reefs which he considered as characteristic of an area
of subsidence. If we pass to the Bermudas, Rein,'^ who carefully ex-
plored the islands, came to the same conclusion, and took a most
decided stand against the theory of subsidence. Rein is of the opinion
that coral reefs may grow wherever the conditions of the bottom are
favorable for the development of the corals. In these he includes the
temperature, the purity of the water, the supply of food by the sea, as
well as a solid substructure, whether this substructure be due to the
subsidence of the coast, or to an elevation of the bottom, this eleva-
tion being caused either by volcanic, organic, or other agency.

Rein also calls attention to the fact, that both Darwin and Dana
have assumed a possibility as a fact, and, the theory once given, have
attempted to prove the subsidence, instead of bringing the subsidence of
coral reefs as a proof of the theory. Proofs of subsidence have nowhere
been given except as explanations of existing phenomena, while the
proofs of elevations within the regions of coral reefs are innumerable.
Darwin and Dana explain the existence of deep channels between barrier
reefs and the coast, as well as the formation of atolls by subsidence, and
hence conclude from the existence of numerous barrier reefs and atolls
that the coasts have sunk, and many islands have been buried in the sea
to form atolls. It naturally follows that they calculate the vertical
thickness of coral reefs as due to the same cause, and nothing but boring
•will settle this point.

Rein further mentions a number of coral reefs from the Tertiary to
the Jurassic, none of which were more than thirty meters thick. Rein,'

1 Agassiz, A., Three Cruises of the Blake, 1888, Vol. I., " Tlie Florida Reef."

2 Rein, J. J., Beitrage zur physikalischen Geographic der Bermuda Inseln,
Bericht iiber die Senkenb. Naturf. Gesell., Mai, 1870, p. 140.

2 Die Bermudas-Inseln und ihre Korallenriffe, nebst einem Nachtrage gegen die

MUSEUM OF COMPARATIVE ZOOLOGY, 137

in his excellent sketch of the Bermudas, calls attention to the discovery
by Pourtales of a conglomerate otf the Florida reef, (the Pourtales Pla-
teau,) formed by the remnants of the calcareous remains of numerous
invertebrates mixed with coral ooze and sand, which has little by little
been built up from great depths. He suggests that the foundation of
the Bermudas consists of a submarine bank of a similar nature, which
has gradually been built up to the level at which coral reefs can flourish,
the Bermuda limestone itself having had its origin upon a mountain or
a terrestrial fold, which may consist of rocks having a greater or less
geological age. He thus accounts in a most natural manner for the
existence of the same rock which forms the smface of the Bermudas at
the greatest depths which have been excavated in making the dock at
that station. The limestone bank once having been built up to the level
at which corals will thrive, the floating embryos canned north by the Gulf
Stream found a foothold on which they began to grow, and founded the
existing active coral reef. The action of the winds on the beach sand
very soon formed the elevated ^olian rocks, which rise to a height of
over two hundred and forty feet, and of which he, Thomson,^ and
Moseley ^ have given such excellent accounts. Jukes ' had already, in
1845, given a similar account of such an yEolian formation at Raines
Island, and Dana, in the Geology of the United States Kxploring Ex-
pedition, had carefully described the formation of the sand drifts solidi-
fied into dunes and encrusting layers along the shores of Oahu.

Thomson, on page 304 of " The Atlantic," has given a graphic ac-
ccount of the mode of origin of the Bermudas, when once the weather
edge of the reef was raised above the level of the sea, and of the manner
in which the Bermudas of the present day have been built up as a bank
of blown sand in various stages of consolidation, though Thomson adopts
Darwin's theory, that the atoll of the Bermudas is due to the entire
disappearance by subsidence of the island round which the reef was
originally formed.

Thomson also gives, on page 309, excellent figures of the stratified
Eolian rocks of the Bermudas, and of xEolian beds in process of forma-
tion, and on the following pages a figure of a so-called sand glacier, or a

Darwin'sche Senkungstheorie. Verhandl. d. ersten Deutschen Geographen Tage
zu Berlin im Jalire 1881.

1 Thomson, Voyage of the Challenger, " The Atlantic," 1877, Vol. I., Bermudas,
p. 420.

2 Moseley, N. H., Notes of a Naturalist.

3 Jukes, J. B., Voyage of the Fly.

138 BULLETIN OF THE

mass of coral sand some twenty-five feet thick, progressing inland. He
also describes the mode in which the free coral sand is rapidly cemented
into limestone by the action of rain-water containing carbonic acid,
which takes up a little of the lime and on evaporating fprms the succes-
sive crust lines of demarcation between various layers of sand, forming
the stratification and lamination of the ^olian rocks. The section given
by Thomson, as exposed by the cutting made for the floating dock in
1870, seems to prove a slight subsidence, as there was found a bed of a
kind of peat at a depth of forty-seven feet, containing stumps of cedar in
a vertical position lying upon the hard bare rock. But it does not prove
that this subsidence, or a greater one, which cannot be proved, has been
the cause of the atoll shape of the Bermudas, any more than the slight
elevations of from twenty to fifty feet, such as we so often meet with in
volcanic districts, prove that the special type of coral reefs existing there
have been due to their influence.

Lieutenant Nelson * has given an account of the geological details of
the appearance of the diff"erent islands composing the Bermudas, and of
the encroachments by the sea and sands, and it did not escape him that
the whole of the Bermudas " may be called organic formations, as they
present but one mass of animal remains in various stages of comminution
and disintegration," and he also called attention to the organic com-
position of what he calls Bermuda chall^, which corresponds evidently to
what has more recently been called coral ooze. He was among the first
to notice the important action of Serpulae in cementing together pieces
of coral, and in certain localities forming even small independent reef
patches. This has been fully confirmed by other obsei-vers in other
districts.

Nelson has also suggested the possibility of the formation of submarine
mountains by the growth of marine invertebrates round any base they
may meet, the decay of their calcareous remains adding stability and
bulk to the colony, while around their summits coral reefs would grow.
He also says, very truly, " Zoophytes affect a vertical growth, and in
this attitude have a tendency to add to the accumulations of the exterior
fence, to the prejudice of the space circumscribed."

When we pass to the very regions explored by Darwin, Mr. Henry
0. Forbes,*^ who in 1879 examined the Keeling Atoll, forty -three years
after Darwin's visit, — the very one which Darwin first examined, and
which suggested to him his whole theory, — could not satisfy himself

1 Trans. Geol. Soc. of London, V., Part I., 1840, p. 103.

2 A Naturalist's Wanderings in the Eastern Archipelago, London, 1885.

MUSEUM OF COMPARATIVE ZOOLOGY. 139

that there was any proof of subsidence, or that the causes cited by the
opponents of Darwin's theory were not amply sufl&cient to account for
all the phenomena he observed there. Mr. Forbes, who spent more
than a month in its study, felt inclined to believe that the Keeling
Eeef foundation has been formed as suggested by Murray, Agassiz, and
Semper, and that the islets have been the result of the combined action
of storms and the slow elevation of the volcanically upheaved ocean
floor on which the reef is built.

Semper,* who visited the Pelew Archipelago in 1863, was among the
first to come to the conclusion that the presence of barrier reefs, atolls,
and fringing reefs in one district could not be explained by the theory
of subsidence, and he looked to natural and simpler causes to explain
the reefs of the Pelew Islands. He was one of the first, after the general
adoption of Darwin's theory of the formation of coral reefs, to visit an
atoll district in the Pacific, and he was the first also to point out for
that region a condition of things which seemed to him incompatible
with the accepted view. He found at the Pelew Islands, within a com-
paratively restricted area, atolls, barrier reefs, and fringing reefs. He
speaks of the channels eaten away between the coast and the barrier
reef, distant three to six miles from shore, and forming a labyrinth of
channels, which he considers as due to the action of currents, and in
which the flow of brackish water prevents the ready growth of corals,
while in the case of the barrier reefs less than half a mile or so from the
shore the action of the currents is reduced to a minimum and the chan-
nels scarcely marked. He speaks of elevated coral reefs of two hundred
and fifty feet in height, and comes to the conclusion that the presence
of atolls, barrier reefs, and fringing reefs in an area where there had
been elevation, and which had remained stationary for a long period,
does not indicate that they have been formed during a period of subsid-
ence, while their simultaneous existence would seem to preclude such a
conclusion.

Semper is inclined to attribute to the action of currents mainly the
great irregularities existing in reefs, which may form even closed atolls,
and are in great degree dependent for their ultimate shape upon the
configuration of the underlying base. On steep shores barrier reefs,
according to him, could not flourish ; only fringing reefs closely hugging

1 Semper, Carl, Die Philippinen und ihre Bewohner, pp. 100-108, Wiirzburg,
1869. A reprint, with additions, of Semper's article in Zeits. f. Wiss. Zool., XIII.
p. 558, 1868. Also, Die Natiirlichen Existenzbedingungen der Thiere, Leipzig,
1880, Zweiter Theil, p. 39.

140 BULLETIN OF THE

the shores, would thrive in such a position, and he lays great stress also
on the difference to be traced in the conditions of the two sides of the
same islands, where the one side is exposed to the action of the open
sea, while on the other side the long periods of calms are most favor-
able to the growth of corals. Although Semper does not deny that
subsidence may have accompanied in some cases the formation of atolls
and of barrier reefs, yet the explanation of the existing conditions of the
reefs of the Pelew Islands seems to him more plausible by the theory of
currents than by that of subsidence.

Semper has more fully developed these views in his " Natiirliche Ex-
istenzbedingungen der Thiere." He calls attention to the difference be-
tween the theories of Darwin and Dana, while Dana agrees with Darwin
that atolls and barrier reefs can only be formed in regions of subsid-
ence, he differs from Darwin in claiming that fringing reefs indicate a
greater amount of subsidence than either of the other types of reefs.
He looks upon the steep coast line of many volcanic islands as a proof that
there has been great subsidence, and of course upon such steep shores,
often with a vertical cliff of more than one hundred and fifty feet, there
is no possibility of the formation of a fringing reef. We must admit,
with Dana, that in the volcanic regions of the Pacific, for which Darwin
claims a general subsidence, there have been local phenomena of eleva-
tion, and also that in regions of elevation a slight subsidence may also
have taken place. But if we have to depend upon either elevation or
subsidence to account for the structure of the reefs, there seems to be
no possible application of a general law regulating the shape of the reefs.
Darwin's map of districts of elevation and of subsidence shows that he
considered a region of elevation as one where fringing reefs alone could
be formed. '

The basis of the whole of Semper's objections lies in the presence
of barrier reefs, atolls, and fringing reefs in the same region, and he
has attempted to prove that he can explain their presence and peculiar
conformation by the action of currents upon growing reefs in a region
which has been assumed, according to Darwin's theory, to be one of sub-
sidence. Semper and Rein were among the first to see the importance
of the discovery by Pourtales of the formation of great limestone plateaus
at considerable depth, far below that at which corals can grow, and the
possibility of having thus many extensive plateaus gi-owing gradually up
to the depth at which corals can flourish. The close connection of ele-
vated and growing reefs are strong proofs against subsidence. To estab-
lish this view, we are obliged to prove that the peculiar shape of the

MUSEUM OF COMPAEATIVE ZOOLOGY. 141

different types of coral reefs can be explained by the action of known
forces. The moment corals have begun to grow, there is nothing to
show that they are not at once subjected practically, though in a less
degree, to the same conditions as exist at the surface, since a more or
less extensive talus is formed in the sea just as at the sea level. The
apparently simple m^ethod of continuing the slope of the land into the
sea, and thus figuring out the depth of the reef, seems to me a most
fallacious one. Let ns look at the various sections which are known
on our northern coast off the Bahamas, off the coast of Florida, off the
Windward Islands, and off the coast of Georgia. These are all of differ-
ent types, and in a region of coral growth would lead to very diflFerent
conclusions. The Florida section, which has been given with consider-
able detail,^ is perhaps one of the most interesting. The great mass of
observations since the pi'omalgation of Darwin's theory is on the side
of the more recent explanation of the formation of coral reefs, while the
older theory rests upon an hypothesis of which it is under most circum-
stances extremely difficult to obtain any proof whatever.

Doctor Guppy,^ who spent considerable time in studying the Solomon
Islands, and more particularly the geology and the formation of the
calcareous limestones and reefs of the group, altogether dissents from
Darwin's explanation of the formation of such a reef as he observed.

Guppy, in his memoir on the calcareous formations of the Solomon
Group, has plainly shown that in that group of islands upraised reef
masses, whether atoll, barrier reef, or fringing reef, have been formed
in a region of elevation, and such upraised reefs are of moderate thick-
ness, their vertical measurement not exceeding the limit of the depth of
the coral reef zone, — one hundred and fifty to two hundred feet at the
very outside. While this is undoubtedly the case where the reef masses
rest upon a foundation of volcanic or older submerged rocks, yet the
presence of coral reefs upon foundations of modern limestone, as in the
West Indies, made up of fragments of the calcareous remains of all kinds
of invertebrates, among which may be deep-sea corals, makes it difficult
to fix very accurately the limit of demarcation between the reef lime-
stone proper and other recent limestones when both have been modified
and changed in elevated areas into the hard ringing compact limestones
80 characteristic of all areas of elevation. At the Solomon Islands, the

1 A. Agassiz, The Tortugas and Florida Reefs, Mem. Am. Acad., 1883.

2 Guppy, H. B., Sucrfjestions as to the Mode of Formation of Barrier Reefs in
Bougainville Straits, Solomon Group, Proc. Lin. Soc. of New South Wales, IX.,
1884, p. 949.

142 BULLETIN OF THE

presence of foraminiferal limestones of concretions of manganese, up to
a height of nearly nine hundred feet, (the limit is usually, according to
Guppy, five to six hundred feet,) would indicate a total elevation of
more than twelve or fifteen hundred feet, and there appears to be no
reason, from what we know of the formation of barrier and of fringing
reefs, and of their extension seaward, why the thickness of the reef lime-
stone should be limited to one hundred and fifty or two hundred feet
even in an area of elevation.

Guppy infers that corals may begin to build at greater depths than
those usually assigned, as some of the elevated reefs in the Solomon
Islands " rest upon partially consolidated calcareous ooze, which is not
found in depths under fifty fathoms on the outer slope of the present
reef ; that in the case of reefs with a gradual slope, where the lower
margin of the band of detritus lies within the zone of reef-building corals,
a line of barrier reefs will be ultimately formed beyond this band, with a
deep channel inside ; but if the band is formed on a steep slope, and
reaches beyond the limit of reef-building corals, no such barrier reef will
be found on account of the silt."

It is not necessary, as is supposed by Guppy,^ in his account of the
Coral Reefs of the Solomon Islands, to have an upheaval to bring corals
within the constructive power of the breakers. Their natural growth is
quite sufficient to raise them beyond that point. He gives for the forma-
tion of barrier reefs, and as an explanation of the existence of a lagoon
inside of the reef, the same explanation as is given by Leconte, — that the
outer growth of the corals is in the direction of clear water, while it is
limited inland by the silt and muddy character of the water of the
barrier reef channel. He is also inclined to attribute the cause of con-
secutive barrier reefs to elevation. This certainly has not been the
cause in Florida. The reefs have grown up from the bottom wherever
the platform had attained the proper level for coral gi'owth.

Bourne, who examined the Diego Garcia atoll ^ and the coral forma-
tions of the Indian Ocean, came to the conclusion that the whole charac-
ter of the Chagos group is very much opposed to the theory that atolls
and barrier reefs are formed during subsidence. There are several atolls
rising above the waves, that of Peros Banhos being fifty-five miles in
circuit, and composed of numerous small islets placed upon a ring-shaped

1 Guppy, Solomon Islands, Calcareous Formation of the Solomon Group, Proc.
R. S. Edinb., XXXII. Part III., .1885.

2 The Atoll of Diego Garcia and the Coral Formations of the Indian Ocean, by
J. C. Bourne, Nature, March 1, 1888, p. 414; April 5, 1888, p 646.

MUSEUM OF COMPARATIVE ZOOLOGY. 143

reef, through which there are several large and deep channels. Egmont,
or Six Islands, is an instance of an atoll in which the encircling reef is
perfect, and unbroken by any channels. There are several submerged
banks, nearly all of which have an atoll form. The great Chagos Bank
is a huge submerged atoll ; so are the Pitts, Ganges, and Centurion banks.
Darwin considered that the Great Chagos Bank afforded particularly good
evidence of the truth of the subsidence theory, yet Mr. Bouriie considers
that a more intimate knowledge of the Great Chagos Bank, and of the
relations of it and other submerged banks of existing land, shows this
view to be untenable. For as the rim of the Great Chagos Bank is on
an average only six fathoms below the surface, and in the most favorable
depth for growth of corals, there are actually six islets on the north-
western edge rising above high water. Bourne has also noticed the
great and rapid destruction of parts of Diego Garcia, both inside and out-
side of the lagoon, and has called attention to the transfer of material
due to storms and tides, showing that the normal action of tides and
winds and waves is constantly tending to lower the sea level, and thus
lay bare dry land that may have been formed by elevation or otherwise.
It does not seem surprising, therefore, that the majority of atolls and
barrier reefs are under such circumstances only just able to maintain
their surfaces above the sea level. He gives an explanation of atollons
in the centre of large lagoons, based upon the production of oceanic
conditions in the interior of a large lagoon, as in Tilla-dou-Matte, where
he thinks the atollons have been formed before any land reached the
surface, in wliich the islets forming the large lagoon were few in number
and distant from one another, so that the atollon would practically have
an oceanic character, and be swept by currents, establishing all the con-
ditions for a new atoll. The corals thus flourishing on the circumferen-
tial parts of the reef surrounding the islet, new atolls with shallow
lagoons would be formed as long as the deep channels between the outer
distant islets were swept by strong currents, becoming wider and deeper
because corals could not thrive in them.

Bourne emphasizes the favorable conditions under which corals flour-
ish as occurring in localities where there is a moderate current flowing
over them, not so strong as to dash them to pieces, but strong enough to
prevent the deposition of sand, these conditions being found everywhere
in external slopes. He lays greater stress on currents than on food sup-
ply, as he considers that to be at variance with the existence of thriving
coral patches within a lagoon. "While we do not deny the fact, yet the
lagoon patches do not spread as vigorously as the corals growing on the

144 BULLETIN OF THE

exterior of the reef, or else they would soon obliterate all traces of
the lagoon. Yet I can hardly see that he has made out a case, that
the corals on the outside of a lagoon on the face of a reef, in full expos-
ure to oceanic currents, laden with food, are not infinitely better off, and
naturally grow more vigorously, than tliose which, as in a lagoon, are cut
oif from a great part of their food supply. They are able to grow in
lagoons in spite of this, because they grow in localities which are kept
clean. As I have plainly shown in the Tortugas, all corals grow i-emark-
ably well on the edge of channels, above the sand drifted by the waves
and currents inland.

The following observations on the Coral Reefs of the Sandwich Islands
were made in the winter of 1885, and formed the substance of a lecture
delivered at Honolulu during my stay there. I have to thank Prof. W.
D. Alexander for important assistance during my visit, and for the com-
munication of valuable information from the ai'chives of the Surveyor
General's Office. Prof James D. Dana has given an admirable account
of the elevated coral reefs of Oahu, and of the extent of the distribution
of reefs on the Hawaiian group. Brigham has also added many interest-
ing observations on the coral reefs of the Sandwich Islands, and Captain
Button in his exploration of the group noted incidentally some points
bearing on the subject. Couthouy has also given a description of the
elevated coral reefs of the vicinity of Honolulu, as well as of the ele-
vated beaches of Kauai. ^ My own observations supplement those of
Dana. I have gone over very much the same ground he covered in
1843, limiting myself, however, to the examination of the reef area
proper, as far as it includes the living and the elevated reefs of the
islands which I visited, — Oahu, Maui, and Hawaii. For my knowl-
ed^re of the reefs on the other islands I am indebted to the observations
of Couthouy, Dana, and of Brigham.

All investigators of coral reefs agree that corals grow in greatest per-
fection in the comparatively still waters of inner channels. Thus, in the
Tortugfas, the largest masses of Mseandrinas and Astrseans are found in
the old channels between formerly distinct reefs, while the great coi'al
heads, measui'ing no less than twelve to fifteen feet in diameter, reach
their maximum size in the so-called ship channel between the outer
reef of Florida and the line of the Keys. As in Florida, so in the
Sandwich Islands, the most luxurious growth of Madrepores occurs upon
the face of the inner channels. There are, for instance, huge masses of

1 Coutliouy, Joseph P., Remarks upon Coral Formations in tlie Pacific, Journal
Bost. Hoc. Nat. Hist., 1842, p. 146.

MUSEUM OF COMPAEATIVE ZOOLOGY. 145

a species of Porites on the inner channels opening to the sea on the
frinofinic reef of the south shore of Oahu. In the enclosed harbor of
Kaneohe there are numerous examples of hummocks, on the summits
of which the corals have died on reaching the surface, while the sides
are still covered with magnificent clusters of Pocillopores and Porites.
Other hummocks in the same locality, not yet above the surface of the
water, remain covered with this luxuriant growth, giving shelter also,
wherever sand has accumidated between the single masses, to the simpler
Fungire so characteristic of the Pacific reefs.

Dana has called attention to the manner in which parts of the surface
of the inner reef of Tongatabu has become solidified by the cementing
material, sand and small fragments, into a huge pavement exceeding in
compactness that of the corals themselves, so that coral rock formed
from the filling of the interstices of masses of branching corals may
become solid enough to be used for building purposes, as is the case at
Honolulu.

The entrance to the harbor of Honolulu (Plates IV., VI.) is nothing
but a channel kept open by the flow of the river, which empties to the
west of Honolulu from the Nuuanu valley, and has killed the corals in
its path, scouring at the same time in freshets the whole harbor and
the adjacent limestone walls forming the channel (Plate VII.). This
and another channel farther to the westward separate the Pearl River
Reef from the Honolulu Reef proper. The river forming the Honolulu
harbor brings down a large amount of volcanic mud in its short course,
and has deposited this in the harbor and channel, so that there appeal's
to be nothing but dark volcanic mud for a considerable distance out
towards the entrance of the channel, where the coral limestone reap-
pears.

A similar channel, but not so well defined, exists opposite the creek
forming the drainage of Mauoa valley, which empties on to the reef at
"VVaikiki (Plates IV., IX.) ; but this river does not bring down the
amount of volcanic silt and detritus carried by the Nuuanu drainage, as
it deposits a great part of its burden along the plain through which it
flows before reaching the shore, whereas the river emptying from the
Nuuanu has a very steep course until it reaches the harbor (Plate X.).

The Pearl River Lagoon outlet, in its turn, divides the reef again by a
deep channel (Plates IV., XI.). The amount of fresh water pouring into
the lagoon is much larger than that emptying into the harbor, and some
of the deep ravines which drain into it extend nearly half the length of
the island toward Waialua. A good part of the western slope of the

VOL. XVII. — NO. 3. 10

146 BULLETIN OF THE

East Eange also drains into the lagoon. Pearl River Lagoon is the
remnant of an old entrance like that of Honolulu, when the old shore
line was just inside the great plain of coral rock extending to the west-
ward of the lagoon as far as Kalaeloa, the shore line being then its inner
line.

The very characteristic bedded coral sand rock so common along the
shores of the Florida Keys and of the Tortugas is not common on the
southern reef of the island of Oahu. It is replaced by the formation
of the massive coral sandstone pavement described above. This, how-
ever, is, as with the finer-grained sandstone, often broken into large rec-
tangular slabs, which in their turn have been uplifted by seas unusually
heavy, and thrown back on the more exposed beaches.

A very characteristic foi'mation found only on the shores of volcanic
islands fringed with corals is the peculiar pudding-stone formed of
rounded and water-worn pebbles of volcanic oi'igin, derived from adjoin-
ing basaltic rocks dipping into the sea. These pebbles are cemented
together by coral limestone, sometimes only a single stone in a mass of
white coral or the cementing material merely filling the interstices and
barely holding the pebbles together. So that we have all possible grada-
tions between a compact coral sandstone, with here and there a pebble
enclosed, and loose friable and poorly cemented rock. There is no local-
ity on Oahu where the process of formation of this conglomerate can be
better seen than at the very eastern extremity of the Honolulu reef, at
the foot of Diamond Head. This pudding-stone has already been de-
scribed by Dana, who also called attention to the fact that some of the
pebbles are evenly covered with a very thin incrustation of lime, look-
ing as if they had been dipped in milk. The lime in solution is also fre-
quenth' deposited in the seams of the volcanic rocks, which then resemble
irregular dikes, and their cavities when filled with limestone change
cellular lava into a sort of amygdaloid. Perhaps no better evidence of
the amount of carbonate of lime taken up by sea-water can be given
than that furnished by this constant deposition of lime from evaporation
of apparently pui'e sea-water.

The Sandwich Islands are peculiarly placed in the track of the trade
winds, so that they all present a dry and a moist side. One side is
rediant with verdure, and its mountain slopes are furrowed by innumer-
able streams, cutting deep valleys on the weather face. The streams
become powerful torrents during the rainy season, and pour an immense
amount of fresh water into the sea, — so large a quantity as materially to
influence the growth of coral reefs on that side. An examination of the

MUSEUM OF COMPARATIVE ZOOLOGY. 147

distribution of coral reefs, and of the streams of the islands of Oahu,
Maui, and Hawaii, clearly shows this interdependence (Plate I.), the coral
reefs being most prominent on the lee side. Combining with this .the
eflFect of the prevailing currents in bringing pelagic food to the growing
reefs, we have a most natural explanation for the absence of coral reefs
to the leeward of Hawaii, while the influence of the flood of fresh water
readily explains their absence along those parts of the shores of Maui, of
Oahu, of Hawaii, of Molokai, where they are not indicated on the map.
With the exception of a few patches of Pocillopora near Hilo, to the south
of the harbor, and on the west face of the island near Kawaihae and
Upolu, there are no corals to be found. In fact, we can hardly conceive
of a less favorable shore for coral reefs than the east face of Hawaii, where
from Hilo north there are in a length of about ninety miles over a hun-
dred water-flills, many falling perpendicularly into the sea from great
height, or pouring in rapid torrents down the steep banks and canons of
the eastern shore.^ Just as little could we expect, and for the same
reason, coral reefs to thrive on western Maui. Except at the junction of
the two parts of this island, we find now^heie conditions favorable for the
growth of coral reefs (Plate III.).

The coral formation of Kauai, which extends as a narrow growing reef
on the eastern and windward shore, has been described by Dana. He
has also given an accoiuit of the solidified beach deposits ^ similar to the
drift sand rocks of Oahu, and has figured a solidified beach deposit
occtirring along the shores of the Koloa district, the remnant of a narrow
frino-inu- reef, which seems to run more or less continuously along the
whole eastern sea face of the island.

Brigham^ is of the opinion that the reef near Koolaii (Kauai) has
been elevated. He also says* that ten or twelve miles west of Waimea
the coral reef has been elevated on a long wide ridge transversely to
the pi'esent shore line. Xear Lapa he speaks of a very curious sand-
bank, nearly sixty feet in height, formed by the winds and currents.

1 Dana does not consider that fresh water has a great influence in the formation
of harbors in coral areas, but it undoubtedly at low stages of the tide increases
the volume of water which scours the harbors, while the detritus it carries must
prevent corals from growing along its course, even if the fresh water was not itself
a check to the growth of coral. Though corals in many instances are known to
live close to fresh water, yet the fact remains that they do not thrive along coasts
where large bodies of water empty into the sea.

2 Dana, U. S. Expl. Expedition, Geol. "Report, pp. 275-277.

3 Erigham, W. T., On the Volcanic Phenomena of the Hawaiian Islands, p. 344,
Mem. Bost. Soc. Nat. Hist., I., Part III., 1868.

* Ibid., p. 349.

148 BULLETIN OF THE

He further says/ that the plain laud of Niihau, ■which comprises two
thirds of its surface, is composed of coral reef sand, and the detritus
washed from the mountains in successive layers. He also says that the
coral reef has been elevated from fifty to one hundred feet, and at the
southeast end of the islaud is quite level. This level portion is bare
and hard ; the coral structure is not evident, its fracture is couchoidal,
and it has a metallic ring. Opposite Kaula the reef is covered with sand
in round hills, which have a thin crust of earth.

Brigham has noticed that the limits of the coral reefs could readily
be traced by the marked change in color of the water of the fringing
reefs, which extend to a considerable distance from shore, usually re-
maining quite level as far as the outer edge, when they drop into deeper
water.^ No detached coral reefs are known in any of the channels
between this island. This is very noticeable oflF Molokai, whei-e there is
a fringing reef on the lee side, which can be plainly seen while steaming
along its shore. But whether the coral said to have been obtained
there by Rev. Mr. Andrew at a height of three or four hundred feet above
the level of the sea is drift coral sand, or mdicates a corresponding ele-
vation of the island, I am unable to state.

Nowhere do the drift coral sands seem to play such an important
part as on the windward side of some of the Sandwich Islands. This
is due to their position in the belt of the trade winds, and to the im-
mediate proximity of the fringing reef to the shore. In some cases
the sands merely drift with the wind, forming irregular banks, which
become cemented together by the action of the rains into a more or less
friable sand rock. The sand rock consists of thin distinct layers, indi-
cating the successive duration of the winds which have di'iven the sand
in a given direction ; the successive layers are frequently separated by a
thin smooth crust, formed by the action of water on the exposed surface.
On the weather side of Oahu, all the way from Kahuku Point to Diamond
Head, we meet with such sand drifts (Plate II.). "Where the hillsides
are more exposed to the full force of the trade winds in the range of an
old elevated reef which is pounding to pieces, as at Laie, the sands are
carried far inland towards Kahuku Point (Plate II.), where they form
well weathered pointed pinnacles of disintegrated sand rock, and assume
most fantastic shapes, reaching to a height of over two hundred feet
above the level of the sea, the material having been furnished by the drift
from the disintegration of the old reef; the loose sand is first swept in-
land by the trades, and banked up in layers, which are subsequently
1 Ibid., p. 351. 2 Ibid., p. 352.

MUSEUM OF COMPARATIVE ZOOLOGY. 149

furrowed and torn by the raiu waters, and either cemented or disinte-
grated into the shape they now present.

Dana has given sf figure ^ of one of the best examples of such drift
sand rock, which is found near Kahuku Point, at an elevation of abont
seventy-five feet above the level of the sea. As he states, the island of
Oahu has undergone an elevation of somewhat more than twenty feet,
since these sandstone bluffs were formed, and this bluff before its elevation
undoubtedly occupied the same relation to the fringing reef which now
forms the elevated plain back of Kahuku Point as the sandstone rocky
bluff of Laie Point holds to the present edge of the shore. Organic
remains are very rarely found in these coral sandstones, although an
occasional shell left by a hermit crab, or a thin fragment of coral or of
a Lamellibranch may sometimes be rolled up into the sand drift and
cemented in it. A walk on the long steep coral sea-beach extending
from Kahuku Point to Laie shows at once where the material for these
coral sandstone bluffs must have come from.

On Maui we also have a long coral sand beach stretching from Kahu-
lui Bay to Paia (Plate III.), from which drifts have been blown, form-
ing extensive coral limestone deposits on the base of the eastern slope of
Western Maui, near Wailuku. The drifts have in some cases formed
large heaps of considerable height, which have accumulated on the
mountain sides for nearly the whole length of the line of separation be-
tween Eastern and Western Maui. These accumulations of limestone
vary in thickness from a few thin layers, scarcely concealing the undula-
tions of the ground beneath and forming a thin veneer, to drifts of con-
siderable magnitude, with rounded tops, more or less disintegrated, and
showing plainly in section the successive layers which have formed them.
Through the thinner layers frequently crop out the grasses and planta
•which have been partially covered by the drifting coral sand, while in
the thicker deposits the vegetable matter is found in all possible stages
of decomposition, finally leaving tubular spaces, which have been attrib-
uted to annelids by some observers, and supposed thus to prove a very
considerable elevation of certain parts of the Sandwich Islands, as at
Wailuku, where Captain Button ^ mentions this drift coral sandstone as
fragments of an elevated coral reef

The low plain which separates Eastern and Western Maui (Plate III.),
extending from the landing at Maalaea to Kahului Bay, the harbor of
Wailuku, on the north, is the top of an old coral reef, which flourished

1 Geolopry of U. S. Exploring Expedition" p. 46.

2 Report of Director of U. S. Geological Survey, 1883, p. 201.

150 BULLETIN OF THE

when the inlet still gave free entrance for the sea-water driven through
it by the trade winds. The reef finally choked up this passage, flourish-
ing thereafter only at the northern edge, where it is still active. Little
by little the old reef has been completely hidden by the mass of drift
sand derived from the beach of coral limestone sand to the east of
Spreckelsville, which at one time may have been much farther inland.
The coral sand on the beach is finely triturated, and the finei- fragments
form regular dunes of all possible sizes, from small horseshoe-shaped
heaps, driven slowly along by the trades and growing constantly, so that
we find some of these dunes of no less than twenty feet in height, which
have travelled two to three miles towards the foot of the Western Maui
slope, where they are comparatively sheltered and become cemented to-
gether by the rain. The Spreckelsville beach thus supplies the drifting
coral sand, afterward hardened into the rock mentioned above, as well as
the remarkable sand dunes which travel inland, obstructing the roads
and the trails. They resemble the huge travelling sand dunea found on
the desert back of Mollendo, which frequently cross the railroad tracks
leading to Arequipa, and impede the progress of trains as much as snow
drifts do in a northern region.

In estimating the thickness of coral reefs,^ it has been usual to take
the declivity of the laud, and to calculate from the estimated slope and
distance from shore the thickness at any given point. This must be a
very defective method, at least in volcanic countries, where the fringing
coral reefs have frequently been entirely covered over by volcanic out-
bursts, such as ashes, lava, or perhaps torrential rains, bringing down
from the mountain-sides an unusual amount of detrital matter. The
drilling for artesian wells near Honolulu has most plainly shown this
alternation of growth of reef corals and of either lava outflows or water-

1 Darwin and Dana both argue that the subsidence of the land is the only pos-
sible cause for the thickness of a fringing or barrier reef, which may be as much
as one or two thousand feet. The evidence brought forward by Mr. W. O. Crosby
(On the Elevated Coral Reefs of Cuba, Proc. Bost. Soc. Nat. Hist., 1882. p. 124) does
not throw any additional light on Darwin's theory of subsidence ; it is of the same
character as all the statements which prove the subsidence by the existence of coral
reefs, and while there may have been coral reefs formed during subsidence, it
does not prove that their growth is due to subsidence any more than the presence of
elevated reefs proves them to be due to elevation. They grow and must have
flourished continuously in periods of both elevation and subsidence, as long as
neither the elevation nor the subsidence was more rapid than the rate of growth
of corals, and as long as the area in which they were found as elevated reefs was
inside of the limits of depth within which we know corals to grow.

MUSEUM OF COMPARATIVE ZOOLOGY. 151

•washed material. On the other hand, we may have the coral reef
forming merely a shell of very moderate thickness, covering the underly-
ing lava rocks. Such is probably the case with the inner reef of the
harbor of Kaneohe (Plate V.), where it is easy in the inner harbor to trace
all tlie transitions from lava islets rising high above high-water mark
(Mokolii, Plate V.), and surrounded at the base with a thin layer of coral,
or to similar islands scarcely reaching above the water level (Ahuo Laka
Mokuo Loe, Plate Y.), where the lava rock can be seen in the centre
of the Pocilloporse surrounding it, and again from these to numerous
similar islands (Plate V.), which, judging from analogy, have a nucleus of
lava, but, not reaching to the water level, have become entirely coated
with coral. Finally, there are larger islets which are covered by dead
corals in the centre, and fringed only by a circle of living corals, while
outside of the harbor we have a reef of greater thickness, probably form-
ing a regular fringing reef on the outside of the entrance to Kaneohe
Harbor (Plate Y.). The flat plain underlying the northern edge of the
harbor, having been built up to reach the water level for nearly its whole
width, is covered only with occasional patches of living coral in the
deeper parts, and with a flourishing growth of corals on the edges adjoin-
ing the inner harbor. Near Kahuku Point there are several most inter-
esting cases, showing the thin veneer of coral which must in some
instances cover the underlying lava. It is not uncommon to find at a
few rods from the shore what may be called coral tables. They are parts
of the elevated coral reef, left as pinnacles on the top of a projecting
mass of lava, the coral table being at the same level as the adjoining dis-
connected elevated coral reefs. These coral tables can hardly have been
left cemented where they were unless the intervening coral reef has been
all washed away, and they should not be confounded with similar un-
attached blocks upthrown and not necessarily cemented in their natural
attitude, such as have been described by Dana. A very fine specimen
of such a large unattached coral rock block is seen lying on the reef
across the entrance of Kaneohe Harbor.

In estimating the thickness of a fringing coral reef, the following indi-
cations, taken from sections of artesian wells bored in the vicinity of
Honolulu, will be of interest.

AYith the exception of Mr. .James Campbell's well and the well near
Pearl Eiver Lagoon, the artesian wells are at Honolulu or to the east-
ward and near Diamond Head, To ^Messrs. Lewes and Cooke I am in-
debted for data regarding the character of many of the wells. Water
■was reached at depths ranging from three hundred to six hundred and

152 BULLETIN OF THE

twenty feet, and none of the wells were started at a greater height than
forty-two feet above high-water mark.

Palace yard artesian well : —
72 ft. of coral rock.
6 ft. of lava.

Then 2G0 ft. lava to coral, thickness not given.
Then clav.
Then lava to 706 ft.

A second well half a mile inland from the above : —
30 ft, of boulders.

Coral was reached at 200 ft., of a thickness of 30 ft.
Then 250 ft. of clay.

At Waimea, Oahu, 900 ft. was drilled through hard ringing coral
rock ; then sand and lava were encountered.

Near Pearl Ptiver Lagoon, close to the road running above the ele-
vated coral rock plateau to the southeast of the Pearl Lochs, a well
passed through 300 to 400 ft. of coral rock.

Another well passed through

100 ft. of soil and boidders. 30 ft. of coral.

100 ft. of coral. 90 ft. of clay.

12 ft. of clay. 28 ft. of sand and boulders.

A well in Thomas Square : —

6 ft. of soil. 60 ft. of clay.

10 ft. of sand. 50 ft. of coral.

200 ft. of coral. 80 ft. of clay.

44 ft. of clay. 50 ft. hard pan.
10 ft. of coral.

Another well, after a few feet of surface soil, came upon a bed of
38 ft. of coral. 5 ft. of clay.

Then 22 ft. of white sand. 45 ft. of coral.

43 ft. of yellow sand. 30 ft. of clay.

47 ft. of lava. 100 ft. of coral
110 ft. of coral. 78 ft. of clay and coral mixed.

100 ft. of lava. 28 ft. of clay.

70 ft. of coral. 1 20 ft. of lava.

The " coral " in these wells was so ground up that it could only be
recognized as such from the larger fragments, and the so-called clay was
mainly lava detritus finely pulverized.

MUSEUM OF COMPARATIVE ZOOLOGY. 153

The well of Mr. James Campbell is thirty feet above high-water mark.

50 ft. of gravel aud beach saud. 20 ft. of soapstone.

270 ft. of tufli. • 110 ft. of brown clay.

505 ft. of hard coral. 48 ft. blue lava.

75 ft. of brown clay. 10 ft. of black clay.

25 ft. of washed gravel. 18 ft. of clay.

95 ft. of red clay. 249 ft. hard brown rock (lava).

28 ft. of white coral.

The well of Mr. A. Marques is at the mouth of Manoa Valley, 36.67
feet above high-water mark : —

10 ft. of earth. 30 ft. of clay.

20 ft. of coral. 150 ft. of lava.

40 ft. of lava. 268 ft. of clay, rock, and lava.

That of Mr. Dillingham is 38.72 feet above high water: —
90 ft. of loam. 25 ft. of coral.

40 ft. of coral. 40 ft. of clay.
60 ft. of clay. 300 ft. of lava.

That of Mr. Ward is 13.36 feet above high-water mark : —

15 ft. of loam. 23 ft. of coral.

180 ft. of hard coral. 107 ft. of sand.

4 ft. of clay. 4 ft. of sand.

24 ft. of coral and shells. 4 ft. of lava.

41 ft. of clay. 18 ft. of lava rock.
10 ft. of hard coral. 35 ft. of rock.

109 ft. of clay.

There are in the Museum at Honolulu pieces of wood, charred or de-
composed, brought up from a depth of two hundred and fifty feet from
one of the artesian wells. This would merely indicate that the pieces
had been washed down from the mountain sides on the then existing
slope, aud would not necessarily indicate a subsidence. The alternat-
ing of so many successive layers of clay and coral and lava indicates,
in my opinion, merely the gradual extension seaward of the shore line as
fast as lava detritus was washed down or flowed over from successive
eruptions, while the growth of tlie layers of coral indicates the period
of rest during which the coral beds were deposited, each in its turn
being overwhelmed by a layer of lava or laval detritus, until we reach
the existing condition of things. That such a succession of coral
growth and lava beds actually took place in the past can safely be

154 BULLETIN OF Tn^

inferred from the borings of the artesian wells, as well as from what we
see going on in the harbor of Kaneohe along its shore line, along the
shore line at Diamond Head and other parts of the fringing reefs east
and west of Honolulu, as well as in certain portions of the islands where
coral sand rock is intercalated between beds of lava of greater or less
thickness.

The great thickness of the coral rock can be accounted for by the
extension seaward of a growing reef, active only within narrow limits near
the surface, which is constantly pushing its way seaward upon the talus
formed below the living edge. This talus may be of any thickness, and
the older the reef, the greater its height would be, as nothing indicates
that in the Hawaiian district there has been any subsidence to account
for such a thickness of coral rock in its fringing reef.

Dana thinks that the western coral islands beyond Bird Island, in the
Hawaiian range, indicate participation in the general subsidence, which
he traces over a large part of the Pacific Ocean, as indicated by atolls
and barrier reefs. Yet he himself describes the fringing reefs of Kaui
and Oahu, and mentions their width as being considerable.

There appears to be no evidence that there has been any considerable
elevation in the Hawaiian Islands, twenty to twenty-five feet being prob-
ably the extreme ; while the existence of cinder cones with their base
close to the present sea level would indicate also that there had been no
special subsidence. There is, however, some evidence of subsidence on
the southern shore of Hawaii. At Kalapanu is a sunken plain about
a mile wide and two miles long, where there has evidently been a sub-
sidence of about fifty feet ; and the raised coral reef extending along a
part of the shore would indicate another change of level in former times.
Brigham has given a sketch plan of the plain in Fig. 27, on page 373
of his " Notes on the Volcanoes of the Hawaiian Islands." ^ Plate XIII.
shows the position of the sunken beach at Kaimu.

From the recent examination of the islands of the Hawaiian group,
and the explanations given by Dutton of the causes determining their
present physiognomy, it would be more natural to suppose that the
gradual building up of the various islands by overflows and eruptions had
overwhelmed such reefs as existed (if any did exist) during the period of
great volcanic activity of the islands. There have been as yet no sunken
reefs discovered in any of the channels between the islands, and as far a3
we know the reefs are all littoral formations of the present shore lines.

The greatest depth in the channel between Oahu and Molokai is 317

1 Mem. Bost. Nat. Hist. Soc, Vol. I. Part III.

MUSEUM OF COMPARATIVE ZOOLOGY. f 155

fathoms; between Molokai and Maui, 137 fathoms; and between East
Maui and Kahoolawe, only 40 to 50 fathoms ; while between Maui and
Hawaii there is a dejtth of 1107 fathoms, and over 1890 fathoms close to
Kaui between Kaui and Oahu. These soundings were kindly given me
by Hon. H. A. Wiedemann, and were taken by a vessel sounding for a
submarine cable to communicate between the islands.

It is interesting to note the structure of the reef as we pass over it
at high tide from the shore to the sea face. The slope of the channel
forming the harbor entrance is made by a steep bank of muddy whitish
ooze. The reef-flat itself, varying from half a mile in width to less, is
also covered nearer shore with coral ooze, and interrupted by small
rounded knobs of decomposed algae, Nullipores, and stalks of dead Sar-
gassum. These cover irregular patches of greater or less size, separated
by bare spaces of ooze. A little farther out, in depths varying from five
to six or even ten feet of water, we come across numerous rounded
patches, covered by clusters of Millepora, with here and there a group of
Pocillopora, and in the intervening bare patches the coral limestone is fre-
quently pitted by numerous Echinometradse and Diadematidte. Some
of the rounded knolls rise close to the surface, and sometimes even are
bare, leaving deep pools between them, in which the characteristic reef
fauna flourishes. Such knolls, when farther out to sea, and arranged, as
they often are, in regular lines of considerable length parallel in a gen-
eral way to the trend of the shore, form successive concentric lines of
breakers, diminishing towards the shore. Upon these the sea beat.s,
breaking up, pounding to pieces, and triturating the corals growing
upon the sides of the knolls, until they are changed into the ooze which
gradually cements the shore portions of the reef into a solid limestone
mass (see Plate VIII.), and freely supplies the fine material for the coral
sand beaches close to the land. On Plate VI. the lines of breakers on the
sea face of the reef are faintly indicated. On smooth days I could fol-
low beyond the outer line of the breakers the occasional patches of large
Millepores, or of Pocillopora, or Porites, or Astrtea, together with the
few Gorgoniee which run out on the somewhat steeper outer slope of the
reef. These gradually diminish, and as far as could be seen with the sea
glass disappeared completely in about ten fathoms. It was very easy to
examine the Honolulu portion of the reef by accompanying the fishermen,
who are in the habit of going out daily in their canoes just outside of the
breakers, and whose skill in crossing the swell in their outriggered canoes
it is very interesting to watch. The small amount of animal life on the
Honolulu reef (on the lee side of the island of Oahu) is surprising, as

156

liUi-iLiIliliiN

compared with that on the weather side on the reefs of Kaneohe Bay.
This difference is due to the fact that much of the pelagic hfe brought
by the trade winds against the weather side of Oahu is swept past the
lee side without bringing any great quantity of food to the coral reef.
This is plainly shown by the comparative scarcity of pelagic life, even on
the most favorable days, on the lee side along the sea face of the Hon-
olulu reef, as contrasted with that of Kaneohe Bay. On such days little
could be seen off 'Honolulu beyond a few Salpte, a huge species of
Appendicularia in its house, a few Diphyes and Praya, and a few pelagic
Crustacea, even when the wind had been blowing from the south, and
was driving the pelagic fauna towards the lee shores again. The Hon-
olulu reef contrasts also with the Florida Reef in the scarcity of Sponges.
The very gradual sea slope of the Honolulu reef is one of its marked
characteristics.

The in-shore flat of the reef, left bare at low tide, as well as all the
low land extending to the base of the hill slopes to a height of nearly
twenty feet above the level of the sea, is made up of coral reef sand.
This is the character of the whole reef, whether west of Honolulu or
east, all the way from the outskirts of the city to Waikiki, and to the
base of Diamond Head. At Diamond Head the coral reef sand is mixed
with the volcanic material washed down from its slopes, and where
it is washed directly into the sea we find the lava sand as well as the
coral sand remodelled by the action of the water, forming either layers
of clear lava sand overlaid by coral sand, or all possible gradations be-
tween a mixture of fine sands of the two and a modern conoflomerate
or breccia of the larger fragments cemented together by the lime car-
bonate held in suspension, or by the finer or coarser sands. Pot-holes,
gullies, and corrugations, due to the wearing action of the sea-woi-n lava
gravel rolling up and down the lava beds, characterized them wherever
exposed to the action of the breakers or of the sea ; while, if subsequently
protected, these or similar holes and corrugated surfaces are gradually
tilled by a deposit of finer or coarser washed material, which, becoming
cemented, produces very striking effects. Some of the larger pot-holes
in the lava beds in the adjoining elevated portions of the reef often con-
tain masses of Porites and of Mseandria of considerable size, more or
less washed', as well as numerous fragments of mollusks, the whole
cemented together in a solid calcareous mass, as we find it on the exposed
part of the shore edge of the reef Towards Diamond Head the outer
slope of the reef approaches the shore. (Plate TV.) The reef there is
narrow, and, owing to the greater depth off" shore and the narrowness of

MUSEUM OF COMPARATIVE ZOOLOGY. 157

the reef, there are only one or two lines of breakers acting directly upon
the shore portion of the reef.

«

Dana has described the so-called modern chalk of Oahu, which is
found at a single locality near Diamond Head, in a part of the elevated
coral reef. It is at the foot of a tufa cone rising from the water's edge,
and, as Dana has akeady stated, coral must have been thriving on the
shores when the eruption took place, as there are fragments imbedded
in the tufa, although the chalk itself is of later origin. There is nothing
to be observed throwing any light on the causes which have produced
this chalk at this particular part of the elevated reef, except that it
must have been deposited in a confined area, subject to special condi-
tions. Yet this chalk is not more similar to the modern chalk than the
modern chalk dredged ofif Nuevitas, which was deposited under most dis-
similar circumstances. The Oahu chalk appears to differ very slightly
from such deposits of fine coral sand as are deposited in sheltered locali-
ties on the shores of coral beaches. It does not contain any organic
remains, but has in addition the peculiar fracture of chalk, and, as is
stated by Dana, is used on the blackboard in some of the schools of the
islands.

At Makapuu the reef has been raised about twenty feet, and farther
north the whole coast is fringed with a growing reef, extending in some
places over three quarters of a mile in width. There are extensive sand
dunes, also, mentioned by Dana and Brigham, a short distance back of
the shore reef at the foot of Konahuanui.

The elevated reefs of the Sandwich Islands, although not elevated
more than twenty to twenty-five feet, are extensively quarried as lime-
stone for building purposes, especially those parts of the reef which evi-
dently formed its inner portion, and in which the corals and mollusks
living on the surface of the reef have been admirably presei'ved.

On the southern edge of the Aliapaakai basin, six miles west of
Honolulu and three quarters of a mile from the sea, there is a raised
coral reef which has been much displaced ; it has been fully described by
Dana. Living corals are comparatively rare upon the reef-flat. Large
specimens of Pontes ^ flourish in pits and hollows of the reef, and a
scanty marine fauna, with occasional masses of NuUipores and Sar-
gassum. The top of the edge of the reef is ban-en, and is deeply fur-
rowed ; it is only somewhat farther down the slope that the reef fauna
flourishes actively again.

^ Dana speaks of the huge size of individual masses of Poritesin the rock of the
inner reef of Tongatabu, which were twenty-five feet in diameter. Geology of
U S. Exploring Expedition, 1849, p 39.

158 BULLETIN OF THE

One cannot fail to be struck with the hardiness of the large masses of
Porites still found living, half exposed to the air at low tide, in the im-
purer water of the reef near to the shore, which seem to die more from
the effect of sediment than from the effect of the exposure to the sun,
or from the impurity of the water. In fact, Porites both at the Sand-
wich Islands and at the Tortugas are among the hardiest of reef corals.
As Jukes, Guppy, and others have noticed, in many species of corals
exposure to air is not always fatal, although in Florida the Madre-
pores, which hold to the Atlantic reefs the same relation the Pocilloporse
hold to the Pacific, are frequently killed over extensive tracts when
exposed to air by low tides or winds. As far as I could judge from an
examination of the sea face of the Sandwich Island reefs, the Pocillo-
porse do not extend to a depth of more than fifteen fathoms, and then
gradually disappear, though the sea face of the reef was swept by a con-
stant current running westerly, due to the trade winds, and during the
season of trade winds but little sediment found its way there to prevent
their active growth.

Dana has called attention to the great extent of the elevated reef of
Oahu, which occurs at the foot of the mountain slopes along the whole
southern face, at heights ranging from five to twenty feet above the level
of the sea, forming the large fiats of the Pearl River Lagoon. It is
nearly continuous from Makapuu to Kahuku Point, extending from
there to a small river emptying at Waimea,~where it abruptly ceases,
but flourishes again on both sides of Waialua, and along the greater
part of the northwestern coast near Waianae. The elevated reef attains
its greatest width near Kahuku Point, where it is nearly a mile wide,
and we can trace this elevated reef as a fringing reef before the elevation
of Oahu just as plainly as we now trace the present fringing reef of the
south shore of Oahu, and that in Kaneohe Bay.

Near Kahuku the drift sand-hills are of great size and height, and
resemble an elevated beach. The elevated reef near Kahuku and that
along the northwest end of Oahu are quite distinct from the solidified
sand-dune deposits.

At Laie^ the drift sand has formed hills of sandstone hard enough
for building purposes. These hills are thirty to forty feet high, much
broken and worn by the action of rain and wind into grotesque honey-
combed masses and ragged pinnacles, which, as Brigham says, have often
been mistaken for elevated coral reef rock.

The mouth of the stream at Waimea is often completely closed by a

1 Brigham, Mem. Bost Soc. ISfat. Hist., Vol. I. p. 368.

MUSEUM OF COMPARATIVE ZOOLOGY. 159

dam of sand at low stages of water, which is in its turn broken through
acain whenever sufficient head of water has accumulated behind it.

Some extensive ancient dunes, from one hundred to one hundred and
twenty feet in height, indicate an effect of the trade winds now no longer
acting. It may be that some of these more ancient inland dunes, which
have become solidified near Diamond Head, were formed under the influ-
ence of the trades before their full force on the eastern edge of the reef
was destroyed by the elevation of the long hill forming Diamond Head.
It may be that we owe to the presence of these dunes most of the coral
sand and calcareous material which we find up to a considerable eleva-
tion between Diamond Head and Honolulu in the Manoa Valley.

Although the Honolulu reef has a far less rich fauna than the reef of
Kahului Bay, yet the limestone which it forms, not having been exposed
to the action of the trade-wind surf, presents the reef much as we see it,
flourishing and gradually dying out in proportion to its proximity to
the shore. The corals, Serpulce, NuUipores, echinoderms, mollusks, and
even Crustacea, are not ground to pieces, as in reefs open to the violent
action of the sea, where all traces of their identity are destroyed in the
process of formation of the coral limestones characteristic of such ex-
posed reef shores. On the contrary, the coral heads themselves, as well
as all the animals flourishing upon such a protected reef as that of Hon-
olulu, are rapidly fossilified and imbedded in limestone, being gradually
covered with the floating lime held in suspension or in solution in the
water ; so that, whenever we get a good section of the shore coral lime-
stone, we invariably find, either on the reef-flat itself, or on the imme-
diate shore line, or on those portions of the reef which have been slightly
elevated, a limestone representing the reef as it grew and flourished, in
which we can plainly distinguish the different species of coral, as well as
the invertebrates which once lived in their shelter.

In the section to be seen on King Street, near the Prison Point, be-
yond the bridge, a face of about six feet in height was being quarried,
the highest point perhaps three feet above high-water mark, the Prison
Knoll, which is the continuation of this limestone ridge, being about ten
feet above high-water mark. This and another limestone knoll facing
the opening of the Nuuanu Valley are formed of a close white lime-
stone mass of decomposed reef rock, in which the individual heads are
more or less distinct, and which contains a large number of shells and
echinoderms imbedded in the mass cementing the corals together. The
western and eastern extensions of this old shoi'e line of the reef, which
must have been elevated about twenty feet above the sea level, has been

160 BULLETIN OF THE

preserved ; but in the continuation of the dividing ridge between the
valleys of Nuuanu and of the Manoa, they have been denuded on each
side of that ridge and appear again on the eastward at a few points
on Diamond Head and towards it on the west. But the lower levels
of the same reef can be traced continuously along the present south-
em shore line all the way from Diamond Head. The salt ponds and
fiats, which extend inland in the Manoa Valley between Honolulu and
Waikiki seem to indicate an inner lagoon much like that of Pearl River,
•which has gradually been filled up by the silt swept down by the river.

The comparative poverty of the fiiuna of the Honolulu reef is undoubt-
edly due to its being on the lee side of the island of Oahu, the outer face
of the reef alone obtaining a fair supply of food, brought by the westerly
current due to the trades, which runs along the south coast of Oahu.

Mr. Kose was kind enough to take me out in a canoe to examine the
corals in Kaneohe Bay. We found the bottom of the bay covered in
many places by numerous more or less circular patches of living corals
(Plate V.) in all stages of growth, from domes a few feet below the level
of the sea covered by fiourishing corals to small fringing reefs round the
shores of the islets and rocks which occur in the bay, and to elliptical
reefs, awash when the corals were living only on the outer slopes. The
bases of all these islands are undoubtedly summits of volcanic rocks
projecting above the general level of the bay, which have been coated
or surrounded by corals. To the southward of Kaneohe Head, on the
plain of Kailua, are extensive dunes similar to those already described.

The edge of the bay is itself entirely surrounded by a fringing reef
(Plate V.) of corals, mainly of species of Porites, which have gradually
died out near the shore, and thrive only near the deeper water on the
edge of the channel. The opening of the bay is ban-ed between Kapapa
and Ahuo Laka by a barrier reef (Plate V.) of very moderate thickness,
extending towards Kekepa and Mokolii, resting upon a lava bottom,
which is exposed in places. There are two entrances into the inner lagoon
of the bay. The breakers pound heavily upon this barrier, and from
it huge coral blocks are constantly thrown up and ground to pieces,
the sand being carried in towards the bay and forming the bar of the
harbor. Dana looks upon these huge blocks, as well as the islands oft'
Kaneohe Bay, as having been elevated from six to eight feet above high
tide.^ Kaneohe Bay is to a certaifi extent sheltered from the full force
of the trade winds by the small peninsula of elevated coral reef which
stretches to the eastward of the bay. On the barrier reef, as well as on

* Dana, J. T>., Geology of U. S. Exploring Expedition, p 253.

MUSEUM OF COMPARATIVE ZOOLOGY. 161

the isolated patches in the bay, we find mainly Pocilloporse, and shel-
tered by them in the interior circle of the coral patches solitary Fungiaj
with a few Gorgoniie, which seem to be much less numerous than upon
the reef near Honolulu, and there are also comparatively few algse and
corallines. The large masses of Porites growing near the shore have
little by little been choked by the silt coming down from the Pali, and
a volcanic sand Hat, with a band of living corals on the outer edge, is
thus formed near the fringing reef, in marked contrast to the coral sand
flats formed by the action of the breakers.

The whole bay before corals began to flourish upon it must have
contained a number of small volcanic islands, and a large number of
sunken volcanic rocks and ledges, which have become capped with coral
or surrounded by diminutive fringing reefs. What is now going on in
the Bay of Kaneohe on a diminutive scale, we may apply to groups of
volcanic islands in the Pacific. If we add to this the powerful agency
of accumulations of limestone on the deeper summits or banks, until
these surfaces are built up to a height at which corals can begin to grow,
we have all the various elements needed for the formation of fringing
reefs, barrier reefs, or atolls within a comparatively limited area, as is
the case in those archipelagoes of the Pacific where these various kinds
of reefs have been observed to occur together. The base upon which
the barrier reef of Kaneohe has been built up has probably been formed
by the washing down and disintegrating of a lava crest of considerable
height, if we may judge of it from Mokolii.

According to Dana and Darwin the line of barrier reefs and of atolls
indicates the former extent of the area of land before the reefs began to
form, which in some cases was three or four times that now above the
level of the sea, and in the whole Pacific I'eef district the atolls and reefs
are the monuments of islands which have long ceased to exist.

The formation of a barrier reef upon a foundation denuded to the
depth at which corals can flourish has not been observed before. Cap-
tain Wharton ^ gives a very interesting account of "the preparation of a
suitable foundation for coral builders by a process directly the reverse of
that of building up by marine organisms on mounds that have failed to
reach the surface," from which it appears probable " that the cinders
and ashes which formed, and still form, the summit of the volcanic
mound originally thrown up, are being by wave action gradually swept
away, and will continue to be so removed until the top of the bank is
reduced below the limit of such action, or the solid rock is laid bare."

1 W. J. L. Wharton, "Foundation of Coral Reefs," Nature, October 11. 1888.

VOL. XVII. — NO. 3. 11

162 BULLETIN OF THE

Dana argues against the possibility of coral reefs being planted upon
submarine banks of the requisite depths for corals to thrive, yet this is
actually what we see going on in Florida, and we can there trace all
the steps from the barrier reef to the keys, and to the incipient reef or
coral bank making a beginning upon the limestone bank which has been
raised by accretion to the depth requisite for the growth of corals. It
seems to me that a wide flat reef cannot be formed by a slow subsid-
ence, but must have grown, during a period of rest or slow elevation,
simply by the dying out of the coral next to the land, as has been ob-
served on the shore edge of Kaneohe Bay.

Captain Wharton ^ has also given, in a recent number of "Kature," a
number of instances of the growth of corals on banks, or on the edge of
banks, illustrating the formation of barrier reefs and of atolls without
the introduction of subsidence. He instances many cases in which reefs
now growing will when awash form perfect atolls of large size, enclosing
deep lagoons, without any further deepening by solution. He also calls
attention to the great width of many of the existing fringing reefs,
which should show more signs of solution than they do if Murray's
theory is sufficient to account for the formation of the whole interven-
ing lagoon. The rotten state of the surface of all coral reefs, espe-
cially fringing reefs, shows that there is considerable solution as well as
removal of material going on ; but the very fact that the majority of
these reefs are of great width goes to show also that solution alone is
not active enough to remove great masses and form lagoons. The
case of Rodriguez is cited by Captain Wharton, where, although there is
a rise of tide of nearly six feet, with every facility for a scouring action
and rapid change of water, yet there is a fringing reef of a width of
nearly four miles and three quarters, intersected only by narrow shal-
low channels. In the case of the Florida Reef there is nothing to show
that the outside reef has not arisen on the southern edge of the Florida
Reef plateau, when it attained a depth at which corals can grow. The
lagoon between the reef and the keys was certainly never filled by
corals which have been carried away by solution, though it has been
occasionally obstructed by the growth of patches of corals, which are ol
the same date or of a later growth than the coral reef proper.

Dr. Coppinger ^ describes Amirante Bank as a submerged atoll, which,

1 " Coral Formations," Nature, February 23, 1888, p. 393.

2 Cruise of the Alert, by R. W. Coppinger, London, 1882, p. 225. In Florida
there is nothing to show that detached barrier reefs cannot grow up to reach the
constructive power of breakers, aa Guppy seems to argue from the existence of
sunken barrier reefs.

MUSEUM OF COMPARATIVE ZOOLOGY. 163

if raised fourteen feet, would become a true atoll. According to Guppy/
elevation is necessary to form an atoll and bring it within the destruc-
tive range of the sea, and a fringing reef, according to him, could not
be formed near shore or grow outwards. No such elevating action exists
along the whole line of the Florida Keys, yet the reef has grown up to
the action of the breakers. The same is the case with Kaneohe Bay.

Neither fringing reefs, nor barrier reefs, nor atolls, grow exclusively
outward. I am inclined to consider the islets on the inner side of a
lagoon as remnants of islands formed by coral heads and coral sand
flats, rather than to suppose that the intervening channel has been eaten
away solely by the solvent action of sea-water. Leconte, as I have
stated, ascribes the limitation of corals towards the shore, and their
growth seaward, on the one hand to the muddiness of the inner waters
of the lagoon, and on the other to the purity of the water due to its
depth. It really seems doubtful whether the islands in the lagoon chan-
nel at Tahiti, mentioned by Murray,^ are portions of the original reef
still left standing, and not, as is the case with the coral heads in the
ship channel of the Florida Reef, independent coral patches, which have
not been overwhelmed by the action of sediment from the outer reef,
or as in the Tortugas, where we find active coral growth in the inner
Ime of the open channels. He further says, when coral reefs are much
broken up the coi-al growths in the lagoon are relatively abundant, while
there are but few coral patches and heads in the lagoons and lagoon
channels when the reefs rise to the surHice, or are nearly continuous.
This ie certainly the case at Kaneohe Bay.

Geikie * also calls attention, in the examination of maps of coral
regions, to the difficulty of the theory of subsidence, as in the case of
the Fiji Islands, where fringing and barrier reefs and atolls all occur in
close proximity, and where all evidence seems to point to elevation, or at
least to a long period of rest.

The fringing reef of Kaneohe Bay forms a scalloped outline, with an
occasional white coral sand beach where the wind has a wider sweep
than in the eastern and more protected parts of the bay. But these

1 Guppy, H. B., Notes on the Characters and Mode of Formation of the Coral
Reefs of the Solomon Islands, being the Results of Observations made in 1882-84,
dunng the Surveying Cruise of H. M. S. Lark. Proc= Royal Soc. Edinb., 1885-86,
p. 857.

2 Murray, Proc. Royal Soc. Edinb., 1879-80, p. 515, April 5, 1880, Vol. X.
No. 107.

3 Geikie, The Origin of Coral Reefs. An Address read before the Royal Phys-
ical Society of Edmburgh. Proceedings, Vol. VIII. p. 1, 1884.

164 BULLETIN OF THE

coral beaches alternate with darker lava silt brought down from the
neighboring mountains, often forming very extensive shore flats, of
which the Chinamen, bv dammin": out the sea, have taken advantage
for cultivation ; along part of Kaneohe Bay there is quite an extensive
flat thofoughly cultivated, where cattle are turned out. This flat is
formed of coral sand, extending far out to Mokolii Island from the
adjoining headland, from which diverge coral patches. On rounding the
noi'theru end of the ba}' we see that the reef north of Mokolii comes
close to the shore, and assumes the character of a narrow fringing reef,
following the shore line more or less regularly. At Kahana a small
harbor has been formed by the extension northward of the reef from
Makuua. Westward the road along the edge of the island runs behind
the shore sand dunes or the beach shelf, or on its summit. There is no
trace of elevation from the point to the south of Kaneohe Bay to the
great Kohuku reef-flat, extending south to Laie.

At Kohuku there is a fine bluft' of consolidated drift sand, of which
Dana has given an excellent figure. Similar drift rocks extend all round
the base of the slope of the foot-hills, marking the old shore line of the
Kohuku reef, which extends from this point as a flat coral plain, slightly
elevated (twenty to twenty-five feet) above the level of the sea, from half
a mile to a mile in width, to the present coral sand beach of Kohuku,
which is exposed to the full action of the trade-wind breakers, and has
thrown np a high sand-hank built up from the elevated reef. This ele-
vated reef extends all the way from Laie to Kohuku, the small outlying
islands being the remnants of sand drift rocks which are gradually being
eroded by the sea, and which have in former times, when the reef was
active, supplied the material for the innumerable sand drifts of the foot-
hills. These sand drifts have become gradually eroded into the most
fantastic shapes, covering the hillsides with innumerable small points
resembling Gothic spires. This elevated reef does not seem to be active
on its sea face. No soundings are available for that part of the shore,
and the heavy rollers break directly on the present sand beach, so that
the sea face is probably, quite steep.

Near Waimea the fringing coral reef ci'ops out here and there behind
the high sand beach formed from the disintegration of the underlying ele-
vated reef. There, as at Kohuku, the reef seems to stop abruptly near
the line of beach breakers, and the slope appears steep, there being no
trace of recent reef corals beyond the line of shore breakers. The coral
sand at the back of the beach was thrown up to a height of from ten to
twenty feet. The breakers form a small lake across a gulch, of which

MUSEUM OF COMPARATIVE ZOOLOGY. 165

they dam the outfet by throwing up a high sand dam ; this breaks
through when the water has accumulated a sufficient head. Half a mile
beyond Waimea the coral caps the lava beds at a height of nearly ten
feet above the high-water mark.

At one other point two miles beyond Waimea, towards Waialua, we
find numerous coral tables capping lava bases. Many of these coral
tables are blocks ten by twenty feet, one of them as much as fifty by
one hundred feet. The height of the lava supports is usually five or
six feet. This elevated coral reef, all the way from Kohuku to Waialua,
is cut by lava spits, which project beyond its surface and extend sea-
ward. At Waialua there is a very fine patch of the elevated coral reef,
from five to six feet above the level of the sea. Traces of this reef can
be seen for five or six miles to the southward of Waialua, along the
beach, made up of coral sand mixed with more or less lava sand, which
I'eaches towards Kaena Point.

As will be seen on examining Plate I., the reefs of Hawaii consist only
of isolated patches of limited extent near Hilo. On the sides of Upolu
Point, both east and west, isolated patches have been observed, near
Kawaihae, and at the southernmost point of Hawaii somewhat larger
stretches of coral exist.

The so-called coral reef to the south of Hilo (Plate I.), near Keokea
Point, consists mainly of detached patches of corals (Pocillopora). They
are very much like the patches of corals to be found on the west coast
of Mexico, and do not constitute a regular reef, although a good deal
of coral grows in this way, judging from the amount of coral sand and
fragments thrown upon the small beaches to the south of Hilo. The
patches are mainly the same species of Porites and Pocillopora which
form the reef of Honolulu. There is comparatively little animal life on
these coral patches, and the lava rocks off" Hilo do not appear to support
a rich fauna. Although small algse grow thickly on the rocks, no Sargas-
sum was found attached, as near Waikiki on the Honolulu reef, and
on the coral conglomerate of the Spreckelsville beach on the north shore
of Maui.

Near the northern extremity of Hawaii, near Honoipu, there is a
patch of coral and another patch to the eastward of Mahukona, clearly
seen from the railroad running round the northern point of Hawaii. On
the lee side of Hawaii there is, near the village of Kawaihae, a stretch
of coral reef, which protects a ba}' once a great resort of whalers. The
village itself is prettily situated at the northern end of a long coral
sand beach. To the south extend, as far as the eye can reach, the

166 BULLETIN OF THE

various ancient flows of lava which have come down the slopes of Hua-
lalai and of Mauna Loa ; while to the eastward of Kawaihae are seen the
older flows of Mauna Kea and the deeply furrowed canons extending
from the shore nearly to the summits of the Kohala mountains. An
interesting patch of elevated coral is also found on the edge of the
sunken plain of Kalapana, similar to the restricted patches now growing
on some points of Hawaii.

The coral reefs of Maui see (Plate III.) are found on the long beach of
the windward side of the island, on Maalaea Bay, and along the lee side
of Western Maui from Maalaea Bay to a short distance north of Lahaina.
The evidence I have been able to collect on the coral plain between
Maalaea Bay and Kahului indicates that Eastern and Western Maui
have been united by a coral reef, which flourished in the shallow passage
once existing between these two parts of Maui. The great coral plain
lying at the foot of Western Maui, and extending to meet the slopes of
Haleakala on the opposite side, is only the surface of the old coral reefs
which once flourished there. The plain which divides the two parts of
Maui is in some places scarcely above the level of the sea ; it abuts some-
what abruptly on the steep slopes of West Maui, while it passes imper-
ceptibly into the slopes of East Maui at Haleakala.

The corals mentioned by the Eev. Mr. Andrews, as found at elevations
of from five hundred to eight hundred feet even on the eastern slopes of
West Maui, are, as I have satisfied myself, all ^olian formations such as
I have described.

The large reservoir for the Hawaiian Commercial Company, below
Wailuku, from one hundred and twenty-five to one hundred and seventy-
five feet above the level of the sea, is built in natural depressions left
between the sand dunes which have been formed in former times on the
old beaches extending all the way across to Maalaea Bay from Kahului
Bay. The highest of these sand dunes mast be from two hundred to
two hundred and fifty feet above the level of the sea, and they have
become solidified into sandstones by the action of the rain. Coral sand
dunes can now be seen travelling across the road leading from Spreck-
elsville to Wailuku, some of which ai'e from twenty-five to seventy feet
high. But a great many have become fixed at a distance from the
beach from which they originated, having become overgrown by a species
of Bermuda grass.

Brigham says that on the windward shore of Maui ^ the coral sand is
piled up in ridges nearly one hundred feet above the sea, shifting with
1 Mem. Bost. Soc. Nat. Hist., Vol. I. p 367.

MUSEUM OF COMPARATIVE ZOOLOGY. 167

the wind, which sometimes drives columns of sand miles along the beach.
This is the material which forms the fine coral sand beach reaching from
Wailuku to Paia.

The stratified coral sand rock seen by Captain Dutton ^ at Diamond
Head, and to the east of the village of Wailuku on Maui, which he takes
as evidence of a recent upheaval of two hundred feet, and perhaps more,
are only consolidated sand drifts, such as I have described above. There
certainly is nothing in the character of this aeolian coral limestone to
compare with the consolidated reef rock at the level of the sea. The
shells he mentioned as imbedded in it have either been blown up by the
violent winter gales, or are the shells of gastropods carried up by hermit
crabs, which I have often met with more than a mile from the coast
in their wanderings.

I have not seen on the shores of Maui coral ledges indicating any ele-
vation. The highest masses of coral rock are fully within the reach of
the action of moderate, or even very heavy seas. The observations of
Rev. Mr. Andrews, quoted by Dana, in regard to the possible elevation
of Molokai and Maui, do not appear to me to indicate anything beyond
coral sand dunes.

The existence of coral sandstone on the east slope of West IMaui at a
considerable height, over extensive tracts, does not indicate any eleva-
tion, but is due merely to the seolian deposits which have found their
way to certain favorably situated places under the action of the pre-
vailing trade winds. Nowhere in the district I have examined on Maui
have I succeeded in finding any trace of corals beyond the height to
which fragments might be carried by the action of the waves or wind
and tides of unusually severe storms. The bedding of the sandstones
at considerable heights was evidently entirely due, as has been shown
by Dana, to the successive deposits of sand cemented together by inter-
rupted rain fall, forming the delicate crusts which separate the various
thin layers of coral sandstone which have accumulated at certain points.

I was greatly interested, on visiting the long coral sand beach which

extends from Kahului to Hamakuapoko, to find very much the same

action going on in the formation of coral conglomerate, breccia, and oolite,

which I had so often watched at Loggerhead Key, and on the island of

Key West on the beach north of Fort Taylor. This action was, however,

modified by the fact that a much heavier sea, due to the trade winds,

was driving upon the surface of the reef off the beach, and was still pow-

1 Hawaiian Volcanoes, by Capt. Clarence Edward Dutton. Fourth Annual
Report of U. S. Geological Survey, 1882-83. Washington, 1884, p. 81.

168 BULLETIN OF THE

erful enough on the beach itself to throw up huge masses of Porites, of
Pocillopora, and of Astrajans, and with them a h^rge number of shells
living on the reef The whole is pounded by the process into a sort
of coquina, which is cemented on the beach, mucli like coral breccia.
Owing to the steady action of the trades, the finer sands accumulated
on the beach would be blown up the slope and carried off to form the
travelling dunes, or the masses of drifting coral sand carried inland to
form the coral sand drifts, while quite heavy fragments were also blown
up bodily to the upper level of the beach.

Kahului Bay is sheltered by a wide, flat, active coral reef, the harbor
being an inlet of the western and widest end of the reef The reef ex-
tends easterly, gradually becoming narrower toward Paia, where it ends.
Only occasional patches of corals are found to the eastward of this point.
It is this extensive coral flat, covered with huge masses of Porites and
Pocillopores, upon which the full force of the trade-wind sea is pounding,
which furnish many of the larger blocks of the Maui coral coquina
which were left as formed in sheltered places, and were covered by
a luxuriant growth of a species of Sargassum ; the surface of many
of these blocks was protected liy masses of Nullipores and other calca-
reous algae. The Kahului beach is broken by numerous spits extending
out on the reef These spits are remains of lava flows which have be-
come covered with huge roimded masses of lava, and in part by frag-
ments of broken coral and by coral sand, sometimes one to two feet in
thickness. In this breccia and conglomerate, as well as in the stratified
coral coquina formed in its proximity, numerous rounded and water-
worn pebbles of lava have become imbedded. The coral lava conglom-
erate thus formed has a most striking appearance. I had been greatly
puzzled by finding similar deposits inland near Maalaea Bay, on the low
plain extending towards Kahului, and on some of the sugar plantations
at a distance :to the east from the road connecting the above-named
places.

The recent British Admiralty Chart, No. 1520, shows very well the
distribution of the coral reefs of the Sandwich Islands. The only
islands which I examined for reefs myself arc Oahu, Maui, and Hawaii,
and in passing close to the snnth sliore of Molokai I could readily see
from the color of the water that there was an extensive fringing coral
reef.

CAMnuiiMiT;, November, 1888.

MUSEUM OF COMPARATIVE ZOOLOGY. 169

EXPLANATION OF THE PLATES.

PLATE L

The Hawaiian Islands. From British Admiralty Chart. Soundings in fathoms.
The shaded shore plateaus show the position of coral reefs.

PLATE II.

Oahu. From Hawaiian Government Survey, W. D. Alexander, Surveyor-General.
The shaded shore plateau indicates the recent and ancient coral reefs.

PLATE III.

Maui. From Hawaiian Government Survey, W. D. Alexander, Surveyor-General.
The shaded shore plateau indicates the position of coral reefs.

PLATE IV.

The south side of Oahu. The shaded portions indicate the position of the af'-
tive reefs, extending along the south side of Oahu from Barber's Point towards
Makapuu Point, and in Kaneolie Bay. From British Admiralty Chart.

PLATE V.

Kaneohe Bay. From Hawaiian Government Survey, W D. Alexander, Surveyor-
General. The lined and colored parts indicate the extent of the active reefa of
Kaneohe Bay. Soundings in fathoms at low-water mark.

PLATE VL

Honolulu and its Fringing Reef. The line of breakers indicates the position of
tlie outer slope of the reef. From a photograph,

PLATE VII

The eastern side of the entrance of Honolulu harbor. From a photograph.
Diamond Head in the distance.

170 BULLETIN OF THE MUSEUM OF COMPARATIVE ZOOLOGY.

PLATE VIII.

The shore edge of the Fringing Reef, extending to the east of Honolulu harbor,
towards Waikiki, with Diamond Head in the distance. From a photograph.

PLATE IX.

The rotten shore edge of the Fringing Reef east of Honolulu, near Waikiki.
From a photograph.

PLATE X.

Passage cut through the shore edge of the Fringing Reef by the river coming
down the Nuuanu Valley to the westward of Honolulu, seen from Smith Bridge.
From a photograph.

PLATE XL
Pearl River Lagoon, the inshore arms of the Pearl Lochs. From a photograph.

PLATE XIL

Coral Sand Beach at Kahana formed from the triturated blocks of an ancient
coral reef which once flourished between Kahana Bay and Kahuku Point. From a
photograph.

PLATE XIII.
Sunken Coral Sand Beach at Kaimu, Hawaii. From a photograph.

\5

CO
CO

ri

X 5

O n^ in

a: "■

,^^-"
%

^f'^^^

1 't'

X

»

0

X

q:

a

1^

?

fel

^

3

0

•-*

0 1

^

cc

—

-1

u.

^

r3

•

a

^

z

i,-*.

c

<

X

u

<

^

0

y

5

V-

5

e
E

y.

•■■'1 ' '■! ■' r r

1,1.!.'

i 1

! ' I

l\

^i

ini

,',n);;,v/;i:i|lfi

iH, i ill *■' II ■' J v'l

IJii'l!'

Ill';

■ I -■! 1 1

l\

i

i,

i
!

•^

I'v

1 ■! '

1 i'

i|

i

II I |lj 1:

' ' '■" i?

i f i^'-',;' I'll !

■lillllitf^ul'il'ilW"
'>■■■ II") ■ ',

^ \! h.

''i!

:''i|'!l

\'y^

l;

0.

X

wm

%':m

y.

'A
■J.

N'o. 4. — Studies on the Primitive Axial Segmentation of the Chick}

By Julia B. Platt.

A. — The Successiox of the Protovertebr.e.

While studying the general development of the chick in connection
with my special work upon the segmentation of the medulla and the
origin of the cranial nerves, I came upon the following sentence in
Balfour's Comparative Embryology : " The first somite arises close to
the foremost extremity of the primitive streak, but the next is stated
to arise in front of this, so that the first formed somite corresponds
to the second permanent vertebra." ^ A foot-note calls attention to
the fact, that " further investigations in confirmation of this widely
accepted statement are very desirable."

Prof. His is hardly more definite. He says : " Die Urwirbel, welche
zuerst entstehen, sind, wie dies v. Baer bereits richtig erkannt hat, nicht
die vordersteu Halswirbel, sondern es bilden sich vor den zuerst ent-
standenen Wirbeln im 5 und 6 Stadium noch einige fernere. . . . Ich
mochte mit v. Baer vermuthen, dass ihrer jederseits zvvei entstehen,
mdglicher Weise ist indess diese Schiitzung zu niedrig gegriflfen."^

I believe that the " Photogramme zur Ontogenie der Vogel, von C.
Kupffer und B. Benecke," is the only other publication in which there
is either confirmation or refutation of the opinion advanced by Balfour,
in regard to the order in which the protovertebra3 are formed. From
the explanatory remarks with which Kupffer and Benecke accompany
their beautiful photographs of chick embryos, one is led to think that
the first protovertebra to appear is at least the fifth in the series, count-
ing from the anterior. In determining the order in which the proto
vertebra arise, they have been principally guided by the difference in
size found to obtain among the protovertebrse of the same individual.
The largest is considered the first.

1 These studies were pursued in the " Annex " Laboratory at the Museum of
Comparative Zoology, Cambridge, Mass., under the direction of Dr. Howard
Ayers. Zool. Laboratory, No. 14.

2 Vol. II. p. 16L

* Untersuchungen iiber die Erste Anlage des Wirbelthierleibes, pp. 82, 83.
VOL. XVII. — NO. 4.

172 BULLETIN OF THE

As my own investigations lead me to a conclusion at variance with
that of Prof. Kupffer, I have examined the photographs with care,
and am pleaSed to find that even in point of size the protovertebrse
which I consider to appear first may well dispute the claims of Prof
Kupffer's first protovertebrse. In fact, the criterion of size among
objects of such varying outline admits of a wide range of interpreta-
tion. It is not enough to judge from the superficial area, we must also
take into account the depth. Moreover, it often happens that a large
protovertebra appears opposite a small one, in which case size evidently
does not determine seniority, since opposite protovertebrse are known
to appear at about the same time. If precedence were established by
size, then we should expect to find a gradual decrease in size from first
to last ; yet this is not what we do find, and Prof. Kupff'er does not
affirm that any protovertebra is second to be formed because second in
size. Perhaps I am laying too much stress upon the matter of size,
since what Prof. Kupffer says in this regard is quite conditional. He
affirms only that, if size determines age, then we must consider this or
that protovertebi'a first formed.

Moreover, it is not clear upon what Prof. Kupffer founds his opinion,
that the second cleft in the mesoderm appears anterior to the first,^
unless it be founded upon the relative distances of these two clefts from
the anterior end of the primitive streak, as compared with the space
which intervenes between the end of the primitive streak and the first
mesodermic cleft in an embryo where only one cleft has yet appeared.
This, at least, is the argument used in regard to Figure 25, loc. cit., which
represents a chick with four distinct pairs of protovertebrse, and a fifth
faintly marked off" in the mesoderm posterior to the others. Here the
author considers the third from the anterior to be the one first de-
veloped, but allows that one might hesitate to decide between the third
and fourth : " denn die Entfernung des hintersten Spalts von dem vorde-
ren Ende des Primitivstreifs, harmonirt mit der Entfernung des ersten
Spalts vou demselben Punkte in den Figg. 20 und 22." (In Fig. 20
only one cleft has appeared in the mesoderm ; in P'ig. 22, there are but
two.) "Mag man nun aber den einen oder anderen der beiden hinteren
Urwirbel als den znerst abgegliederten auffassen, jedenfalls geht aus
den Figg. 24 und 25 hervor, dass die Segmentirung zunachst rascher
nach vorn als nach hinten vorschreitet, und ist darnach die Angabe von
Foster und Balfour zu berichtigen, die behaupteten dass die nachst-
folgenden Urwirbel hinter dem ersten Paar entstiinden." ^

1 Kupffer, p. 172. 2 Hjjd.^ pp. 173^ 174.

MUSEUM OF COMPARATIVE ZOOLOGY. 173

Now, Figures 20, 22, 25, are said to be enlargements of the ante-
rior portion of Figures 19, 21, 24, respectively. In the last mentioned
figures, the distance from the posterior extremity of the primitive streak
to the head fold varies but little, while the primitive streak alone, in
that chick which has three or four more protovertebrse than the other
two chicks, is shorter by the width of those protovertebrse. Comparing
the pictures, and allowing for a difference in total length, we find that
the width of three protovertebrse has been added in Figure 24 from the
region of the primitive streak to that of the embiyo, and this added
space we find occupied by protovertebr*. 1'his being the case, does it
not seem quite reasonable to suppose that the posterior vertebrae are
developed later than the anterior, in the space once occupied by the
receding primitive streak, — especially since we know the important part
played by the primitive streak in the formation of the mesoderm 1 I do
not attach much value to this argument, nor do I know if it would hold
true for other chicks which might be photographed ; I would merely
say that these three photographs, which are compared by Prof. Kupffer
for the purpose of determining the first protovertebra, are capable of
leading to conclusions quite different from those at which the author
arrives.

Prof. Kupffer further remarks : " An den Figg. 33 und 34 sind einige
Eiuzelheiten hervorzuheben. INlan geht wohl nicht fehl, wenn Man
das in der Contourzeichnung mit il' markirte Urwirbel-paar [of five
pairs, the fourth] als das erst entstandene auffasst. Abgesehen von der
Grosse [it does not seem to me larger than either the second or fifth
pair] spricht der Umstand dafiir, dass die Axe der Urwirbel dieses Paares
nicht senkrecht auf der Medianlinie steht, sondern eine leichte Neigung
noch hinten zeigt. Es harmonirt das mit der Richtung der beiden
ersten Spalten, Figg. 19-22." ■^ Let me call attention to the fact, that
the fifth pair of protovertebrse in Figures 33 and 34 shows also a
" Neigung nach hinten," but is evidently not therefore considered " das
erst entstandene."

I allow, indeed, that the obliquity of the angle made by the fourth
and fifth pairs of protovertebrse with the main axis of the chick in
Figures 33 and 34 corresponds well with the ol)liquity of the angle in
Figures 19 and 22, where, as will be remembered, but one pair of pro-
tovertebrse is developed, and there is therefore no doubt about its being
the first ; but how in regard to Figure 23, where " der erste Urwirbel
ist . . . deutlich begrenzt, vor demselben ein zweiter und hinter dem

1 Kupffer, pp. 175, 176.

174 BULLETIN OF THE

ersten ein dritter in der Abgrenzung begriffen " 1^ Here the second pair
of protovertebrse, the only pair as yet distinctly formed, is quite at right
angles with the main axis of the chick, while the only one obliquely placed
— the posterior — -is not as yet fully formed, and therefore cannot be
the first formed. Thus we see that, where there is but one pair of proto-
vertebree, it makes an oblique angle with the middle line ; where there
are three pairs, the third as }et but half defined proto vertebra is
oblique ; and where there are live pairs, obliquity belongs to the last
two puirs;, of which tlie fifth is not as yet distinctly separated from the
surrounding mesoderm. It seems, therefore, reasonable to suppose that
the characteristic of obliquity to the middle line belongs rather to the
protovertebra just formed, or forming, than to the protovertebra which
of any given number was first formed.

Figure 41 represents a chick with seven pairs of protovertebrse, of
which the fifth is marked as first. Figure 50 represents a chick with
nine pairs, of which the fourth is designated first; and the author
naively remarks, that " das Bild (Fig. 50) ist aufgenommen worden, um
an einem zweiten Beispiele die Constanz der Erscheinungen an den
Urwirbeln z\i demonstriren." Four pairs of protovertebra^ developed
in front of the first in the less advanced chick ; three pairs developed in
front of the first in the more advanced chick. Shall we not call this
variation, rather than constancy] I think enough has been said to
show that neither the size of the protovertebi-re, their relative distance
from the primitive streak, nor yet their obliquity to the main axis, is
a sufficient ground to warrant a decisive answer to the question in
regard to the order of their development.

It may be that Prof Kupffcr has other reasons than those which he
mentions for thinking it highly probable that the fourth or fifth pair
of protovertebrae is the pair first developed ; but in so fiir as the argu-
ments ho advances are concerned, one may well hesitate to accept his
conclusions until they be supported by further evidence.

I have studied one hundred and fifty or more chicks, and from their
external appearance I have been able to make out little in regard to the
order in wliich the fir.st protovertebr;c are formed. The length of the
chick at this early stage varies ; the shape and relative length of
the primitive streak vary ; the outline of the area pellucida is seldom
exactly the same in any two chicks. Tn fact, tliere seems to be no fixed
point froni which to measiire. But I have been more fortunate in the
study of serial sections. From these I learn that the cleft which sepa-

1 Kupffer, p. 172.

MUSEUM OF COMPARATIVE ZOOLOGY. 175

rates two protovertebrse is not at the time of its first appearance a
continuous break extending throughout the width and depth of the
vertebral ph\te. As a first trace of segmentation, there appears, in a
line transverse to the axis of the chick, a succession of slight depressions
on the dorsal surface of the vertebral plate. In a series of sagittal
sections these depressions appear and disappear, a single depression
seldom occupying more than two or three sections. As the chick grows,
these depressions gradually deepen and become more continuous, while
immediately below them a sliglit upward curve appears in the ventral
boundary of the mesoderm. A little later, a distinct line appears, con-
necting these two indentations, and separating the cells of the undiffer-
entiated mesoderm from those which are about to form a protovei'tebra.
The latter now take on the spindle shape that characterizes the proto-
vertebral cell, their long axes becoming gradually radial as regards the
protovertebral body. Soon there appears here and there a break in the
line that bounds the forming protovertebra, but, as a rule, cell prolonga-
tions, in a horizontal plane somewhat below the middle of the protover-
tebra, bind the new protovertebra, on the one hand to the undifferentiated
mesoderm, and on the other to the protovertebra last formed.

The foi'mation of a distinct and separate protovertebra takes place so
gradually that a single long section shows the jirocess in all stages of
progression. Passing from the fully formed protovertebrse through four
or five in process of formation, one readies the undifferentiated meso-
derm, in which a faint dorsal depression gives the first indication of the
place where a protovertebra will ultimately be cut off.

This being the case, it is evident that, provided the order in which
the protovertebrai are formed is in every chick the same, and pro-
vided that a protovertebra which has once started to develop contin-
ues to develop regularly, then, to determine the order in which they
have developed requires but a sufficiently complete set of sectioned
embryos.

In development, as in adult life, individual chicks may be expected,
within a certain limited range, to vary from the typical chick. Still,
if we find certain conditions to prevail in a large number of chicks,
without exception, we are warranted in assuming these conditions to be
normal. The protovertebrte are not formed until the mesoderm in the
region of the somites has grown to be quite compact, and is as many as
four cells deep, so that when the cell layers first take on their concen-
tric arrangement they lie in two layers about the central axis of the
protovertebra. I have never seen a protovertebra whose walls were

176 BULLETIN OF THE

but one cell thick, hence 1 am warranted in assuming that a certain
thickness pertains to the mesoderm from which a protovertebra is im-
mediately to be formed. Since the breaks iu the mesoderm appear so
gradually, it is possible, at the time the first cleft has appeared, to see
where the next is to follow, and consequently to determine definitely
on which side of the first protovertebra the first cleft appeared.

Figure 1, Plate I., is a section through a chick at the time when one
decided parting of the mesoderm occurs, and the chick when examined
entire by transmitted light showed but one mesodermic cleft.

From this section it will be seen that the first cleft lies anterior to
the first protovertebra, not posterior, as Kupff"er and Benecke supposed.
A partial cut separates this protovertebra from the posterior mesoderm,
while the protovertebra anterior to the first cleft is still closely con-
nected with the mesodermic band which runs forward to the head. In
fact, the anterior protovertebra is not entirely separated from the
anterior mesoderm until as many as four or five protovertebrae have
been formed. A curve in the dorsal outline of the mesoderm indicates
from the first, however, where the anterior protovertebra is to be ulti-
mately cut off. The series from which Figure 1 is taken shows more
conclusively than any other series which I have, the relation of the first
mesodermic cleft to the first protovertebra. It is difficult to find the
exact stage when the first cleft is far more developed than the second,
although the second one is already distinctly marked. However, in
almost any series made at the time when the first protovertebra has
just been formed, the same conclusion with regard to the priority of
the anterior cleft may be reached by comparing the number of entire
and partial breaks which occur iu successive sections anterior to the
first protovertebra with those that occur posterior to it. It will then
be found that the anterior cleft is both deeper and more continu-
ous than the posterior. T'he sections which I have drawn have
been selected from chicks, the stages of whose development follow
so closely upon one another, that, if it be supposed that between
any two stages the protovertebra marked a has developed into that
marked 1, while the cells included between the letters / and g have
developed into the protovertebra a, then the protovertebra?, which
in the preceding section were already partly cut off, posterior to 1,
must have become attached to the mesoderm from which they were
thus partly separated, or the order of development must vary in dif-
ferent chicks ; for if the above supposition were true, in the chick with
the greater number of protovertebrae the posterior ones would be less

MUSEUM OF COMPARATIVE ZOOLOGY. 177

distinctly set ofiF than in the chick with a less number of protovertebrse.
The sections, however, do not support this supposition, since they show
that, while the undifferentiated mesoderm posterior to the first proto-
vertebra (1) is of the requisite thickness to permit of the formation of
protovertebrse, the mesoderm anterior to a is too thin. This I find
to be the case in all chicks which I have examined at this stage of
development. Aside from the thinness of the mesoderm anterior to
protovertebra a, the sections also show that even in the anterior part
of a the cells have not increased as rapidly as in the posterior part.
Consequently, this protovertebra is characterized by a peculiarly asym-
metrical form, which gives it, in the early stages of its growth, an
appearance quite different from that of the other protoyeHebrae.
(See Figs. 2, 3, 4, and 6.)

Figures 5 and 5 a are taken from a series which was not cut parallel
to the axis of the chick, and therefore the appearance of protovertebra
a in Figure 5 is dekisive. These facts force me to conclude that,
contrary to the opinion advanced by Kupffer and Benecke, the develop-
ment of protovertebrse is much more rapid caudad than cephalad. In-
deed, after the first cleft in the mesoderm appears, dividing two forming
protovertebrae, the posterior mesoderm goes on continually diflferentiat-
ing into protovertebrse until the chick acquires its complete number
of protovertebrse, while it is only after three or four protovertebrae
have been thus formed posterior to the tii*st, that protovertebra a
becomes finally separated from the anterior mesoderm, after which this
mesoderm also slowly acquires sufficient thickness to form another pro-
tovertebra anterior to a.

Figure 5 a was drawn for the purpose of showing a group of cells
lying between / and g. They come into section near the outer (i. e.
lateral) boundary of the line of protovertebrae, and extend through
only three sections, while it takes twenty sections to pass through a
protovertebra. They are, in fact, the first indication of the formation
of a protovertebra anterior to a. A like group of cells is not found
on the opposite side, nor do they occur on either side of the chick from
which Figure 6 is taken. But I find a similar group of cells in the
same position (between / and g) in the series from which Figure 7 is
drawn. The section represented is that which shows best the character-
istic radial arrangement of these cells. Here, too, but few sections pass
through the group. In the succeeding stages they are more marked, and
by the time five or six protovertebrae have been formed posterior to 1
these cells have become distinctly protovertebr^l in their arrangement.

VOL. XVII. — NO. 4. 12

178 BULLETIN OF THE

Still the new protovertebra is little more than a half-protovertebra, for
it is much smaller than those immediately behind, and opens anteriorly
to the mesoderm of the head, into which it passes with no distinct boun-
dary. These peculiarities of protovertebra b are evident even in the
entire chick, if examined at this stage.

For some time I thought it possible that this incomplete protover-
tebra might be formed from protovertebra a, but this cannot be the
case, for I have traced its growth from a few cells grouped together
anterior to a, up to the time when the ninth nerve, passing behind
the ear capsule, crosses so near to this protovei'tebra as to leave no
intervening tissue from which another protovertebra could be formed.
That another protovertebra is not formed anterior to b is also evinced
by the fact, that in no subsequent stage does the line of protovertebrae
end anteriorly in a complete protovertebra, or in a protovertebra less
complete than b. If b became complete, and the tissue anterior to h
developed into an incomplete protovertebra, it would be possible to find
the stage in which this change took place.

In Kolliker's " Eutwicklungsgeschichte des Meuschen nnd der ho-
heren Thiere," there is a figure ^ representing an " Urwirbel ahnlicher
Korper vor der Gehorgrube, der von einem Ganglion und zwei Nerven
gebildet wird." Another figure ^ represents an embryo with two large
cell masses at the beginning of the protovertebral line, which Kolliker
calls " Urwirbeln ahnliche Massen." He supposes them to represent
the " vereinigte Anlage der Ganglion des Glossopharyngeus und Vagus."
The shape of these last mentioned protovertebra-like structures, and
their relative distance from the ear capsule, lead me to think it highly
probable that their posterior wall is formed by the incomplete protover-
tebra (6) of which I speak above, while their anterior wall is formed by
the ninth nerve. However this may be, since Kolliker afiirms these
structures to be at least in part ectodermic, they offer no objection to the
limitation which I place upon the mesodermic protovertebrae, of which
alone I speak.

My conclusions are, therefore, that the first break in the mesoderm
occurs anterior to the first protovertebra, and that two protovertebrae
(or, more correctly, one and a half) are slowly formed anterior to the
first mesodermic cleft, in the time occupied by the formation of six or
seven protovertebrae posterior to that cleft. With regard to the num-
ber of protovertebrae occurring anterior to the first, my work confirms
the estimate of His and Von Baer.

1 Loc. cit., p. 430. 2 Loc_ cit., pp. 142, 143.

MUSEUM OF COMPARATIVE ZOOLOGY. 179

In respect to the further development of these protovertebrse, I quote
from Froriep,^ the results of whose investigations have been corroborated
by Van Wijhe. "In der Occipital region, d. h. in dem zwischen ersten
Cervical-nerven und Vagus eingeschlossenen Abschnitt der "Wirbelsaule,
finden sich bei viertagigen Hlihnerembryonen vier Muskelplalten, welche
von hinten nach vorn (craniahviirts) an Grosse abnehmen. Es waren
also hier vier Urwerbel angelegt, welche in den Aufbau des Kopfes
eingehen. Der fiinfte liegt auf der Grenze von Kopf und Halswirbelsaule,
die Muskelplatte, die dieser liefert, beriihrt mit ihrem caudalen Rand
die Anlage des ersten Halswirbelbogens, mit dem cranialen die hinterste
Bogenanlage der Occipitalgegend."

"We see, therefore, that the first mesodermic cleft divides evenly the
four pairs of protovertebrae which enter into the formation of the head,
and that this cleft does not coincide with the occipito-cervical cleft of
the adult.

B. — The Relation of the Cranial and the Spinal Nerves to

THE Neural Axis.

The development of the rudimentary protovertebra h brings me
to that stage in the growth of the chick with which my work ostensibly
began. Contemporaneously with the formation of this prctovertebra,
the successive constrictions of the medulla appear, the neural crest is
formed, and the first cranial nerves arise (V., VII., VIII.).

If a chick be examined by transmitted light during the second or
third days of incubation, it will be noticed that the medulla is marked
by a series of swellings and constrictions which are directly continuous
with a line of similar swellings and constrictions in the region of the
spinal cord. These divisions of the medulla are such noticeable char-
acteristics that they have frequently been mentioned before the com-
paratively recent attempts to determine the number of head-segments
from the distribution of the cranial nerves. His, Rabl, and Balfour
speak of them as more or less transitory structures ; but in so far as I
know, Beraneck and Orr are the only authors who have attempted to
elucidate by their means the problem of cranial segmentation.

In an article upon the cranial nerves of the Lizard,^ Beraneck
describes five successive enlargements of the central canal of the

1 Archiv Anatomic und Physiologie, 1883. Zur Entwickelungsgeschichte der
Wirbelsaule, p. 226.

2 Des Nerfs Craniens chez les Le'zards. Eecueil zoologique Suisse, 1884..

180 BULLETIN OF THE

medulla, corresponding to which the external surface of the neural
wall is marked by five encircling swellings. These he designates
" replis meduUaires," (designated by Dr. Orr " neuromeres,") and he
atteiiipts to assign to them a segmental value, from their constant re-
lation to certain of the cranial nerves. With the anterior neuromere
he finds the fifth nerv^e connected ; with the third, the united seventh
and eighth nerves, while the roots of the ninth nerve come from the
fifth neuromere.

Dr. Orr, in a recent publication upon the development of the Lizard,^
gives a similar description of the relation existing between the neuro-
meres (replis medullaires) and the fifth, seventh, and ninth cranial
nerves. Beraneck has also published a detailed account of the neuro-
meres in the chick,^ and has assigned to each fold a segmental value.
The two anterior folds correspond to the two head-segments supplied
by the fifth nerve. The third neuromere belongs to the united seventh
and eighth nerves, these nerves consequently representing but one primi-
tive segmental nerve. The fourth neuromere he assigns to the segment
of the auditory capsule and sixth nerve, believing the relation which
obtains between the auditory capsule and eighth nerve quite secondary.
The fifth neuromere is connected with the ninth nerve. The tenth
nerve represents a transitory' condition between a spinal and cranial
nerve, and is consequently not entitled to its neuromere, since the swell-
ings and constrictions in the region of the spinal cord ai'e not considered
by Beraneck structures homologous with the neuromeres.

Beraneck claims to have found similar folds in tritons and elasmo-
branchs, although he was deterred from establishing their relation to
the cranial nerves by lack of material. To the above list of vertebrates
possessing neuromeres, I can add the salmon. Between the ages of
fourteen and nineteen days the medulla of the salmon is divided into
five distinct lobes or neuromeres, with the anterior of which the fifth
nerve is connected ; with the third, the ganglion of the seventh and
eighth nerves, while the ninth nerve passes from behind the ear capsule
close to the fifth neuromere. I did not have sufficient material, at the
age when the nerves first appear, to decide whether this relation between
nerves and neuromeres is primitive or secondary. From the constancy
with which Jive neuromeres appear in classes so widely separated as fish,
reptile, and bird, they would seem to be structures inherited from a
common ancestral form. Moreover, the constancy in the number of

^ Embryology of the Lizard. Journal of Morphology, 1887.

2 Replis Me'dnllftires du Poulet. Recueil zoologlque Suisse, 1887

MUSEUM OF COMPARATIVE ZOOLOGY. 181

neuromeres is not less striking than the constant relation which the
first, third, and fifth neuromeres bear to the fifth, seventh, and ninth
cranial nerves.

The constrictions which in the chick divide the neural tract into
cerebral vesicles oi " neuromeres," appear even before the neural walls
have formed a closed canal. They arise successively from before back-
wards, dividing the medullary tube into vesicles, which decrease in size
in the order of their formation. The third neuromere of the medulla
(fifth vesicle formed) is the only exception to the gradual diminution of
the successive vesicles. This neuromere (Fig. 17, Plate II.) is smaller
than either the second or fourth.

Anterior to the first protovertebra the neural canal is divided by a
series of such constrictions into seven vesicles. With the appearance
of each successive protovertebra, another constriction occurs opposite
the protovertebra, and another neuromere (as we may also designate
these neural swellings) is added to the preceding. The constrictions of
the spinal cord often appear before the formation of the corresponding
protovertebrae ; consequently these ectodermic diflferentiations are quite
independent of any mechanical influences consequent upon the formation
of the protovertebra?.

Beraneck has already carefully described and figured the medullary
folds, and in as far as the folds themselves are concerned I can but
confirm the results of his investigations ; yet I have ventured to repeat
the description in brief, because I would call attention to the fact, that
in the chick, at least, the order in which the various parts of the brain
are formed does not correspond with what is usually given as the typical
development of the vertebrate brain.

Wiedersheim says : " At a very early stage three swellings may be
seen on the anterior enlarged part of the medullary tube, which are
spoken of as the primary anterior, middle, and posterior cerebral vesicles
(fore-, mid-, and hind-brain). . . . The primary fore-brain and hind-
brain each become differentiated into two parts, and thus five divisions
of the brain may be distinguished. Counted from before backwards,
these are Prosencephalon (secondary fore-brain), Thalamencephalon
(primary fore-brain), ^Mesencephalon (mid-brain), Metencephalon (sec-
ondary hind-brain), and Myelencephalon (primary hind-brain)." ^ I
quote from Wiedersheim, because he expresses so concisely the opinion
commonly advanced by zoologists in regard to the formation of the
vertebrate brain.

1 Comparative Anatomy of Vertebrates, pp. 131, 132.

182 BULLETIN OF THE

From Figure 17, Plate II., it will be seen that the primitive brain of
the chick does not correspond with the above description. As the result
of my studies, I find that primarily the chick's brain consists of a suc-
cession of neural vesicles, from the first of which the three anterior
divisions of the adult brain are developed, namely, Prosencephalon,
Thalamencephalon, and Mesencephalon. Soon after the formation of
the first cerebral vesicle, the optic lobes appear as lateral outgrowths
from it, after which, for some time, the growth of this vesicle is not
relatively greater than the growth of the following vesicles, or than that
of the entire embryo. But with the first appearance of the cranial
flexure, the primary fore-brain begins to develop very rapidly, soon a
constriction marks off the mid-brain (Fig. 15), after which the fore-
brain grows out as an anterior vesicle, so that by the time the head is
completely bent, the primary fore-brain is represented by three large
vesicles, which open widely into each other (Fig. 14). The develop-
ment of these three vesicles is wonderfully rapid, and since it is exactly
coincident with the cranial flexure one may well connect the two events
causally, affirming the cranial flexure to result from the very rapid
development (increase of surface) of the dorsal and lateral walls of
the first cerebral vesicle.

The second primary vesicle develops into the cerebellum, and the
successive vesicles, including the neural swellings between the first five
protovertebrte, take part in the formation of the adult medulla oblon-
gata. Thus, either the development of the brain in the chick forms a
marked exception to the usual development of the vertebrate brain, or
the nomenclature of the three primary cerebral vesicles is inaccurate.
The prevalent impression, that the brain consists primarily of but
three vesicles, may possibly arise from the fact that the ectoderm does
not exactly follow the neural conformation. The surface of the head
at an early stage presents three main divisions, marked by slight con-
strictions, but these are quite independent of the neural tract. The
difference between the contour of the head and that of the brain will
be manifest if the chick be first examined as an opaque object, and then
by transmitted light. In the former case, the superficial outlines of the
head are most apparent ; in the latter, the medullary walls come into
strong relief.

Although I agree with Orr and Beraneck in regard to the number
and appearance of the neuromeres, and although I also find the ulti-
mate relation of the cranial nerves to these folds in salmon and chick
to be the same as that which they describe for lizard and chick, I do not

MUSEUM OF COMFAEATIVE ZOOLOGY. 183

find that this relation is a primitive one in the chick. From lack of
material, I was unable to study the younger stages of development in
the salmon ; but with regard to the chick, I can say definitely that at
its first appearance the fifth nerve does not arise from the outward con-
vexity of the first neuromere of the medulla, as Beraneck has repre-
sented it. It arises from the concavity that lies between the first and
second neuromeres, and its position at the side of the anterior neuro-
mere is quite secondary. This fact supports Marshall's theory, that
segmental nerves '' at an early period shift downwards, and acquire new
or secondary roots of attachment to the side of the brain." ^ It has
seemed to me that the secondary attachment of the fifth nerve was not
due to a new outgrowth connecting the nerve root and brain, but rather
to a gradual shifting of the position of the original root. However this
may be, it certainly cannot be owing to the outward spreading of the
walls of the fourth ventricle that the root of the fifth nerve comes to
acquire an attachment anterior to its primitive attachment. Beraneck
says: "Par suite des changements survenus dans la route du cei"veau
posterieur, . . . les replis sont encore plus dejetes sur les cotes de la
region cephalique et les racines nerveuses paraissent ^tre descendues et
se rattacher plus pres de la region ventrale qu'auparavant. . . . Les
changements survenus dans la forme et la structure du cerveau poste-
rieur me semblent suffire a expliquer cette descente apparente des nerfs
craniens sans qu'il faille recourir a une hypothese ne reposant sur
aucune observation directe." ^ The mere divergence of the walls of the
fourth ventricle may suffice to explain the more ventral attachment of
the fifth nerve, but is hardly sufficient to account for the fact that its
secondary attachment is anterior to the first.

Since the walls of the fourth ventricle follow quite closely the curvet
on the outward surface of the medulla, the concavity from which the
fifth nerve arises is represented on the inside by a ridge pi'ojecting far
mto the fourth ventricle. A series of horizontal sections through the
root of the fifth nerve at the time of its attachment between the first
and second neuromeres shows this ridge to be composed of lines of cells
converging like the rays of a fan towards the point from which the
fifth nerve takes its origin (Fig. 10). This convergent arrangement of
lines of cells is characteristic of each of the ridges which, projecting- into
the fourth ventricle, separate successive neuromeres (Fig. 12 b). If

1 A. Milnes Marshall. Segmental Value of the Cranial Nerves. Journal of
Anatomy and Physiology, 1882.

2 Replis Me'dullaires du Poulet, Rec. zool. buiss'e, pp. 334, 335.

184 BULLETIN OF THE

Figure 10 be examined, it will be seen that the fifth nerve draws its fibres
from both the posterior part of the first neuromere and the anterior
part of the second. Succeeding sections show the connection between
the nerve and the neuromeres, both anteriorly and posteriorly, to be
more extensive than in the section drawn. Beraueck says, apropos of
the connection between the fifth nerve and second neuromere : " J'avais
toujours ete frappe de ce que la seconde paire des replis medullaires
etait la seule ue paraissaut avoir aucune relation avec les nerfs craniens.
En etudiant la question d'un pen plus pres je reconnus que cette deuxi-
eme paire emet uu trouc nerveux qui vient se fondre avec celui partant
de la premiere. Ainsi les elements qui constituent le trijumeau sont
fournis par les deux premieres paires des replis medullaires. . . . Le
trijumeau correspondant a deux paires de replis, ne serait done pas un
nerf simple, mais resulterait de la fusion de deux troncs nervaux primi-
tivement independants." '

From the fact that the second neuromere, as well as the first, con-
tributes fibres to the formation of the fifth nerve, I cannot argue, as
Beraneck does, that these two neuromeres correspond to the two primi-
tive segments to which, from its peripheral distribution, the fifth nerve
is supposed to be related.

Of the two segmental nerves united in the fifth, one is surely repre-
sented by the main branch of the fifth with its Gasserian ganglion, the
other segmental nerve is usually supposed to be represented, at least
in part, by the ramus ophthalmicus profundus, with its ciliary ganglion.
In fact. Van Wijhe claims to have traced in the elasmobranch the back-
ward growth of the ramus ophthalmicus, from its connection with the
brain in the immediate neighborhood of the ciliary ganglion to its final
fusion with the main body of the fifth nerve. This being the case, one
would hardly look for the neuromere of that anterior segmental nerve,
which has thus lost its independence, back of the neuromere to which
the fifth nerve is itself attached. It is possible, indeed, to suppose that
the whole fifth nerve has transferred its connection from the posterior
to the anterior neuromere, but such a supposition takes away any seg-
mental value which may attach to the neuromeres from their present
relation to the cranial nerves.

Aside from any theory of segmentation, the fact remains that the fifth
nerve, at the time of its origin, arises from the depression between the
first and second neuromeres, while the cells composing its root are inti-
mately connected with the r.ells forming the mass of the ridge which

1 Loc. cii., p. 337.

MUSEUM OF COMPARATIVE ZOOLOGY. 185

projects into the fourth ventricle, opposite the external attachment of
the nerve. It is therefoi-e with this inner swelling, and not with an
outer one, that the fifth nerve is originally connected. It may here be
noticed, that neither the third nor fourth cranial nerve arises from the
middle of the vesicle with which it is connected, but both are attached
to the brain near the line which separates this vesicle from the following.

Passing from the concavity between the first and second neuroraeres
to that which separates the second and third, or, in other words, pass-
ing from the first medullary ridge (using this term to designate the
internal ridges) to the second, we come to the origin of the seventh
nerve, anterior to the small neuromere beside which lies the ganglion
of the united seventh and eighth nerves. In a series of horizontal
sections made at the time when these nerves have just left the neural
crest, it will be found that from each side of the third neuromere nerve
fibres pass downward and towards the middle of that neuromere, where
they unite in a large ganglion. (See Fig. 12b.) This ganglion is ulti-
mately and secondarily connected with the convexity of the third neuro-
mere. As I remarked above, this neuromere is smaller than any other
anterior to the origin of the tenth nerve, consequently the space included
between the large roots of the seventh and eighth nerves is small, and
might easily be overlooked in transverse or sagittal sections. (See Fig.
12 a.) The long axes of the cells composing each nerve are, as usual,
parallel to each other, but since the two nerves meet the ganglion from
different directions, it follows that the long axes of the cells composing
the seventh nerve make an angle with the long axes of those composing
the eighth nerve, and the ganglion is consequently connected with the
brain by two bauds of cells whose long axes diverge (Fig. 12 a.)

Although the fibres of the ninth nerve lie in close proximity to the
fifth neuromere, they may be traced back of the ear capsule to the con-
cavity which separates the fourth and fifth neuromeres, and corresponds
to the fourth medullary ridge. (See Figs. 11 and 11a.) The first
figure shows the course of the nerve, and passes through the neural
crest. The second figure is drawn to show the relation of the nerve
to the entire ear capsule and the medullary ridge. They are from the
same series.

Posterior to the ninth nerve, the long commissure of the tenth nerve
extends beyond the fourth protovertebi-a. The peripheral distribution
of this nerve, and the extent of its commissure, show it to be com-
posed of the fused roots of several spinal nerves. All of the spinal
nerves arise oppo:>ite the muscle plates (Figs. 13, 13 a) from correspond-

186 BULLETIN OF THE

ing concavities in the spinal cord. The nerve fibres primarily connect-
ing the spinal ganglia with the cord, form an almost continuous sheet
extending along each side of the spinal cord. At intervals correspond-
ing to the interspaces between the protovertebrce, these fibres are appar-
ently pushed to either side by the thickened mesoderm which projects
upward from the region of the notochord, so that the fibres between
every two such mesodermic thickenings are drawn into one spinal
ganglion. The width, therefore, of the band of fibres connecting a
spinal ganglion with the spinal cord, corresponds to the width of the
protovertebra lateral to it (Figs. 13, 13 a). From a glance at the gen-
eral relation of the neural concavities to the nerves which arise from
them, it will be noticed that, where the ganglion is connected with the
neural axis by a mass of fibres bound closely together, the curve of the
concavity from which they arise is sharp, as in the region of the medulla.
Where the fibres are spread out, forming a wide ganglionic connection,
the curve of the concavity is gontle, us in the spinal cord. But whether
its curve be gentle or sharp, the concavity in both medulla and spinal
cord is the source from which the nerves originate, and the neural
swellings which correspond to the nerve roots are the inner ridges, not
the outer neuromeres.

Beraneck is unwilling to homologize the neuromeres of the medulla
with the swellings of the spinal cord, because of the difference in the
later differentiation of the nervous tissue in these two regions, and be-
cause, as he affirms, the swellings of the spinal cord are transient. But
so are also the neuromeres of the medulla. By the fifth day of incu-
bation they are fiist fading away, yet at this time the wave-like form of
the walls of the spinal cord is still plainly visible (Fig. 13). Here, as in
the medulla, the segmentation is more manifest in the ventral region
than in the dorsal. Moreover, at the time when the neuromeres of the
medulla and the swellings of the spinal cord first appear, the tissue
throughout the neural tract is quite indiff'erent. Further, since the
first four folds posterior to Beraneck's ^five medullary folds lie also in
the region of the medulla, and are also connected with a cranial nerve,
— viz. the tenth, — it is practically impossible to draw the line sepa-
rating these folds from the more anterior. It is true that they are
smaller, but the diflference between the fifth neuromere and the next
posterior fold is not as great as the difference between the second and
third neuromeres. I therefore see no reason why the successive swell-
ings which originally mark the neural tract should not be regarded as
homologous structures.

MUSEUM OF COMPARATIVE ZOOLOGY. 187

Orr says : " Balfour described certain internal swellings of the lateral
wall of the hind-brain of elasmobranch embryos : ' Swellings of the
brain towards the interior of the fourth ventricle are in connection with
the first five roots of the vagus and the glosso-pharyngeal root, and a
swelling is also intercalated between the first root of vagus and the
glosso-pharyngeal root.' In his figure (Fig. 5, PI. XVI., I. c.) there are
no external marks of these divisions, and the ' swellings ' lie opposite
the nerve roots while in the region between the nerve roots there are
internal depressions. In the lizard, on the contrary, in the region be-
tween the nerve roots are internal ridges. The two conditions are thug
very different ; but possibly younger elasmobranch embryos might show
a connection between these swellings and neuromeres." ^ My experience
with the chick leads me to believe that possibly younger stages in the
development of the lizard than those which either Orr or Beraneck was
able to study may show a connectioa between the " swellings " and the
nerve roots.

1 Loc. cit., pp. 336, 337.

July, 1889.

188 BULLETIN OF THE

LITERATURE.

Balfour, F. M.

Comparative Embryology. London, 1885.

Development of Elasmobranch Fishes. London, 1878.
Beard, J.

Branchial Sense Organs. Quarterly Journal of Microscopic Science, 1885.
B6raneck, E.

Replis Medullaires du Poulet. Recueil zoologique Suisse, 1887.

Des Nerfs Craniens chez les Lezard. Rec. zool. Suisse, 1884.
Froriep, August.

Zur Eutwickelungsgeschichte der Wirbelsaule. Arcbiv Auat. und Physiol.,
1883.

Ueber Anlagen von Sinnesorganeu am Facialis u. s. w. Archiv Anat. und
Physiol., 1885.
Gegenbaur, Carl.

Die Metamerie des Kopfes und die Wirbeltheorie des Kopfskeletes. Morpho-
logisches Jahrbuch, 1887.
His, WUhelm.

Untersuchungen iiber die Erste Anlage des Wirbelthierleibes.
KoUiker, Albert.

Eutwickelungsgeschichte des Menschen und der hoheren Thiere.
Kupffer, C, and Benecke, B.

Photogramme zur Ontogenie der Vogel. Halle, 1879.

Marshall. A. M.

Development of the Cranial Nerves in the Chick. Quart. Joum. of Micros.

Sci., Vol. XVIII.
Segmental Value of the Cranial Nerves. Joum. of Anat. and Physiol., 1882.

Orr, H.

Embryology of the Lizard. Journal of Morphology, 1887
Von Baer.

Eutwickelungsgeschichte der Thiere.
Van "Wijhe.

Mesodermsegmente u. Entwickelung der Nerven d. Selacliierkopfes. 1882.
Wiedersheim, R.

Comparative Anatomy of Vertebrates, Eng. ed., 1886.

MUSEUM OF COMPARATIVE ZOOLOGY. 189

EXPLANATION OF THE PLATES.

LETTERING.

d, first cleft in the mesoderm.

o, i,j> ^, h indicate respectively successive clefts in the mesoderm posterior to the

first cleft, d.
f, g, position of first and second clefts anterior to d.
1-6, successive protovertebrte posterior to the first mesodermic cleft, d.
a, b, first and second protovertebrae anterior to the first cleft, d.
III., v., VII., VIII., IX., X., third, fifth, seventh, &c. cranial nerves.

A. Prosencephalon.

B. Thalamencephalon.

C. Mesencephalon.

D. Metencephalon.

E. Myelencephalon.
and. auditory capsule.
ect. ectoderm.

h. heart.

inf. infundibulum.

m. p. muscle plate.

n. c. neural crest.

n. ch. notochord.

n. sp. spinal " neuromeres."

p. V. protovertebrae.

sp. g. spinal ganglion.

V. fourth ventricle.

X. section broken.

1', 2', 3', 4', 5', neuromeres of the medulla.

PLATE L

Sagittal sections, showing the development of the protovertebraB.

Fig. 1. Thefirst mesodermic cleft is seen anterior to the first protovertebrse. X60

Fig. 1 a. Same, enlarged. X 180.

Fig. 2. Section from an embryo with two protovertebrae, showing the peculiar
shape of protovertebra o, which is due to the less rapid proliferation of cells in its
anterior portion. X 330.

190 BULLETIN OF THE MUSEUM OF COMPARATIVE ZOOLOGY.

Fig. 3. Embryo with three protovertebrae. The shape of protovertebra a much
tlie same as in the embryo of two protovertebrae. X 330.
Fig. 4. Section from a chick with four protovertebrae. X 180.
Fig. 5. " " " five " X 160.

Fig. 5 a. " " " five " X 160.

Fig. 6. " " " six " X 160.

Fig. 7. " " " seven " X 165.

Between /and g is seen the beginning of the protovertebra anterior to a.
Fig. 8. Sections from a chick with eight protovertebrae.

PLATE II.

Fig. 10. Horizontal section showing the origin of tlie V., VII., and VIII. cranial
nerves in a chick with fifteen protovertebrae.

Figs. 11, 11a. Two sections from the same series. Fig. 11 shows the origin of
the IX. nerve from tiie neural crest back of the auditory capsule {aud.). Fig.
11 a is a more ventral section, showing the relation of the ear capsule to the
neural concavity dorsal to which the IX. nerve arises. Horizontal section^.

Fig. 12. Shows the neuromeres between which the nerves arise, as indicated.
Sagittal section.

Fig. 12 a. Sagittal section, showing the two roots of the VII. and VIII. cranial
nerves.

Fig. 12 b represents the right medullary wall from a horizontal section, show-
ing the cell arrangement of the neural ridges connected with tlie VII. and VIII.
cranial nerves.

Figs. 13, 13 a. Sections from two 4|-day chicks. They show the relation of
the spinal gangUa {sp. g.) to the muscle plates [in. p.), and to the neuromeres of
the spinal cord [n. sp.). 13 a is sagittal; 13, horizontal.

Fig. 14. Median sagittal section, showing the relation of the five main divisions
(A, B, C, &c.) of the brain, after the cranial flexure.

Fig. 15. Median sagittal section, showing their relations at the beginning of the
cranial flexure.

Fig. lb shows the neural ridges in the medulla of a five-day chick. The roof
of the fourth ventricle and mid-brain have been removed.

Fig. 17. Diagrammatic representation of the relation of the successive cranial
vesicles at the time of their first appearance.

Fig. 18. Sagittal section through the lateral wall of the medulla in a five-day
chick. In hardening, the mesoderm has broken away from the neural wall.

Fig. 19. Sagittal section near the middle hue of tlie brain of a four-day chick.

Platt- Chick

Plate. 1.

&

^>

:^

.-7

P 1^ -» I .

tf*

I

1.1 ^'°

' fel^ ^

'.?'

^f;

BMeiee'.liih

Piatt - Chick

mm,,

lyser

No. 5, — The Morphology of the Carotids, based on a Study of the
Blood-vessels of Chlamydoselachus anguinews, Garman. By H.

A.YEES,

Chlamydoselachus * holds uudisputed claim to being the most
lowly organized Elasraobrauch yet discovered, and it was to have been
expected that primitive conditions of organization would be retained
to a greater extent than in any other known member of the group, the
vascular system not excepted.

Almost nothing is known of the vascular system of the Notidanidae,
but it may be inferred from a comparison of their other structures with
the corresponding organs in Chlamydoselachus that their vascular sys-
tem will not be found to retain all the primitive characters present in
Chlamydoselachus. This primitiveness of structui'e is expressed, 1st,
in the retention of a large number of aortic arches ; 2dly, in the pres-
ence of the complete dorsal aorta, of which the precardiac portion among
the remaining vertebrates is almost without exception either extensively
or completely ati-ophied ; 3dly, in the extensive venous spaces, always
simple in character, developed in the course of the large venous trunks.

While at the Banyuls-sur-Mer zoological station in the spring of 1885
I reached the conclusions, (a) that the vascular system of existing
sharks had been extensively abbreviated in the course of descent in con-
nection with the development of the head ; (b) that formerly there must
have been a much larger number of aortic ai'ches than we now find in
any member of the Elasmobranch group ; and (c) that with the loss of
the aortic arches the dorsal aorta of the branchial region had disap-
peared either entirely or in part. As examples of the latter condition

* The two papers containing the original descriptions of the systematic position
and anatomical characters of Cldami/doselachus anguineus from the type in the
Museum of Comparative Zoology at Cambridge, Mass., include all that is known
of the creature : —

1 Garman, Samueh New Sharks, Chlamydoselachus anguineus, etc. Bull.
Essex Institute, Vol. XVI. p. 1.

2 The same, 1884. Chlamydoselachus anguineus, Garman. A Living Species
of Cladodont Shark. Bull. Mus. Comp. Zool., Vol. XII. No. 1. Cambridge, Mass.,
1885.

VOL. XVII. — yo. .5.

192 BULLETIN OF THE

we have several species of sharks whose anatomy has been described by
Hyrtl,^ Parker,* and others, such as Scyllium, Mustelus, Zygeena, etc.
Furthermore, it seemed to me that the accumulation of blood-vessels
about the hypophysis cerebri could be reasonably accounted for on the
assumption that the vessel which Hyrtl ® describes is in truth a remnant
of a much larger vessel of functional activity during embryonic life
only, which sustained such relations to the vascular structures about
the pituitary space as would lead us to search for the remains of
preoral aortic arches. His words are as follows [loc. cit., p. 5): " Bei
feinen Injectionen lasst sicht leicht erkennen, dass diese Aorta, welche
Kopfaorta genannt zu werden verdient, durch eiue in die Mittellinie
des Schadel-basalkuorpels nach vorn gehende Fortsetzung bis zur Ein-
trittsstelle der Carotis interna in die Schadelkapsel sich erstreckt.
Tab. I. Fig. 1, lit. i." Again the vessels called carotids by Hyrtl,^'^
Miiller,® and Parker,*' '' seemed to complete the aortic circulation in
front by bringing blood from the ventral to the dorsal side of the di-
gestive tract, and were in a special sense the homologues of the carotids
of the Mammalia, of which more will be said further on.

These conclusions were the outgrowth of my verification of Hyrtl's*
descriptions and figures of the arterial blood-vessels of the head in Scyl-
lium, Acanthias, and Mustelus. I did not at that time expect to be
able to verify my conclusions by the dissection of any living animal,
and consequently considered the views which form the essence of this
paper to be of little value because not demonstrable. The vascular sys-
tem of Chlamydoselachus proves the contrai-y, however, and under the
circumstances it is now a matter of no inconsiderable interest to find
out how the Notidanid sharks comport themselves with respect to the
cephalic arteries.

I am not certain that Hyrtl has ever seen the "Kopfaorta" as it
exists in Chlam;ydoselachus, (i. e. imbedded in the cartilage of the basis
cranii,) for from 'us description of it quoted above it is not clear
whether the injected vessel figured in his plate runs in, i. e. through, the
basis cranii, or only in the middle line and ventrad of that structure.

3 Hyrtl, Joseph. Die Kopfarterie der Haifische. Denkschrift. d. Wiener Akad.,
XXXII, 1872.

* Parker, T. J. On the Blood-vessels of Mustelus antarcticus. Phil. Trans.,
Vol. CLXXVII., 18SG.

5 Das Arterielle Gefasssystem der Rochen. Wiener Sitzungsberichte, 1857.

6 Miiller, Johannes. Verglcichende Anatomie der Myxinoiden. Berlin, 1839-41.
■^ Parker, T. J. Zootomy, 1884.

MUSEUM OF COMPARATIVE ZOOLOGY. 193

At present I doubt not that Hyrtl found in every case only an extra-
cranial vessel so far as he traced it, and that perhaps he did not follow it
quite to its end on account of the non-penetration of the injection
mass beyond the point of entrance into the cartilage.* While study-
inof the same species at the Laboratoire Arago it did not occur to me
to seek for the vessel within the substance of the cartilaginous brain
case, and I always found the vessel very much as Hyrtl figures it,
though usually I was not able to trace it so far anteriorly as it is drawn in
his plate. Since my attention has been called to the matter I have dis-
sected only one specimen of Scyllium stellare from the Museum alcohol
collections, but the histological condition of the tissues did not permit
a satisfactory determination of the relations of the vessel. Further
study of this form is very desirable. So far as the homology of the two
vessels is concerned, there can be no question that they have strictly the
same morphological value.

To aid in understanding the relations of the carotids, and to serve in
the comparison of other forms, as well as to give an idea of the funda-
mental simplicity of organization of the vascular system, I shall first
describe the aortic system of blood-vessels in Chlamydosolachus passing
thence to a consideration of the homologies in other vertebrates in so
far as lies within the scope of this paper.

The sinu-aui-icular (see Figure 2) valve is placed in the middle of
the transverse axis of the sinus venosus with its Ions? axis at ri<?ht ansles
to the axis of the latter. It is slit-like and guarded on either side by
two broad semilunar tendinous folds, — the sinu-auricular valves. The
remaining auricular wall of the sinus is smooth, muscular in some
de,L,'ree, but thin.

The aperture of the auriculo-ventricular valve is placed to the left,
below and in front of the sinu-auricular valve, and pierces the thick ven-
tricular wall. Its ventricular end is provided with two cup-shaped folds,
the auriculo-ventricular valves. Its auricular end presents a radiate
figure formed by the tendinous cords of the muscular plate, which cen-
tre here. The dorsal wall of the auricle is smooth without, but rug-ose
within. It is scarcely thicker than the wall of the sinus venosus. The
ventral wall of this division of the heai't is thick and muscular, and
forms a triangular plate which projects beyond the edges of the ventricle,

* Miiller certainly did not see the vessel in any species studied by him, for he
says (/oc. cit. : " Bei den Haien u. Rocheii felilt die vordere umpaare Fortsetzung
der Aorta schon ganz, wis bei den Knochenfisciien aber die Haien besitzen nocli
einen circulus cephalicus ini Sinne Hyitl's," etc.

VOL. XVII. — NO. 5. 10

194 BULLETIN OF THE

from which at about the middle it may be separated. This plate reaches
cephalad to the middle of the couus arteriosus, but is not bound to it in
any way.

The conus arteriosus (see Figure 2) forms a thick spmdle-shaped
trunk about an inch long and one fourth of an inch in diameter. It is
provided with six rows of valves, all of which are quite small, except the
anterior set of three, which are large, tridentate, and formed of a white
tough tissue of a cartilaginous consistency. In Odontaspis the conus is
bounded anteriorly by three large valves, each of which consists of two
thin membranes united by a thick median bar which ends in a pointed
projection beyond the anterior border of the valve.

The pylangium terminates anteriorly in a synangium or bulbus arte-
riosus, from which spring three vessels, one median and two lateral, the
ventral aorta, and the sixth pair of afierent branchial arteries, respect-
ively. The synangium is not so well developed in Chlamydoselachus as
in Raja or most Elasmobranchs, for, while in the former a single pair of
afferent branchial arteries arise from it, in the latter the common trunk
known as the posterior innominate artery of the skate represents at least
three primitive afferent branchial vessels, and in consequence the synan-
gium represents a very much greater portion of the primitive ventral
aorta in Eaja than it does in Chlamydoselachus.

The synangium of Chlamydoselachus includes the enlarged end of
the ventral aorta lying between the last pair of pylangial valves and
the point of origin of the ventral aorta (in the restricted sense), and
the sixth pair of afferent branchial arteries. It forms a truncated cone-
shaped body, the homologue of the bulbus arteriosus of bony fishes.
The ventral aorta continues the synangium forwards along the median
line, and lies in a distinct sheath formed by the connective tissue sepa-
rating the basibranchial cartilages from the muscles of the floor of the
branchial basket. This sheath forms the outer wall of a lymph space.

The ventral aorta (see Figure 2) ends blindly in front in an anchor-
shaped enlargement formed by the bifurcation of tlie median trunk and
the separation of the two resulting vessels, — tlie anterior innominate
arteries, — which curve outwards, upwards, and backwards, quickly
splitting into two pairs of vessels the first and second afierent branchial
arteries. From the anterior edges of the anterior innominate arteries,
equidistant from the median line, arise two small vessels, which, passing
forwards, supply the muscles in the ventral wall of the throat. Other
small vascular twigs arise from the ventral aurta as well as the afferent
V)ranchial arteries to carry blood to tlie muscles of the ventral portion of

MUSEUM OF COMPARATIVE ZOOLOGY. 195

the branchial apparatus and to the heart. In the saurian Varanus, ac-
cording to Wiedersheim,^ (p. 704, Fig. 540, B,) it appears that a por-
tion of the ventral aorta (or ventral commissures 1) remains as a single
median trunk, from which both the common carotids are given off just
ventrad of the hyoidean apparatus, to rise on either side of the throat.
The common carotid trunks thus occupy the position of the hyoidean
efferent branchial vessels of the Elasmobranch. This common trunk
divides above the level of the pharyngeal roof into an external and an
internal branch. We thus see that in difiFerent animals the carotids
have not the same value in so far as their proximal ends are concerned.
The afferent branchial arteries (see Figure 2) number six pairs, and are
arranged in sets of two pairs each. While the first two pairs of arteries
arise from a common trunk, the arteries of the other pairs arise inde-
pendently, with the members of each pair placed opposite each other.

The anterior innominate artery does not divide into first and second
afferent branchial arteries until after it has curved upwards about a
quarter of an inch, when the first afferent branchial springs from the
anterior edge of the innominate and continues its trunk dorsad, curving
gracefully forwards, outwards, and upwards, then considerably backwards
to where it enters and suppUes the anterior half of the first gill cleft, or
the hyoidean demibranch. The second afferent branchial passes back-
wards at an acute angle from its origin at the posterior border of the
root of the first, and in its outward and upward course nears the third
afferent branchial where the latter enters its arch.

The sets composed of the afferent branch ials three and four, and five
and six, respectively, are so disposed that while five and six leave the
synangial end of the ventral aorta, three and four arise from the middle
of this trunk. There are slight variations in the size of the vessels
forming the pairs two, four, and five, the arteries of the left side being
larger than those of the right. The afferent branchials three, four, and
five run very nearly parallel with one another, and while the efferent
branchials of the pairs three and four continue this relation above the
intestine, efferent branchial five bends suddenly backward and fuses
with the sixth before entering the dorsal aorta.

There are six pairs of efferent branchial arteries (see Figs. 1 and 2),
corresponding pair for pair with the afferent branchials just described.
Only four pairs reach the median dorsal line to form the dorsal aorta ;
these are the second, third, fourth, and fifth. The first efferent branchial

8 Wiedersheim, Robert. Lelirbuch der vergleichenden Anatomie der Wirbel-
thiere. Jena, 1885.

196 BULLETIN OF THE

is connected by an anastomosing branch with the second, just as the lat-
ter emerges from its arch to enter the roof of the mouth, and thus at
least a portion of the blood from the hyoidean demibranch reaches the
dorsal aorta ; but as the trunk of the first efferent branchial artery
passes out of the hyoid arch it curves forwards along the outer edge of
the basis cranii, and runs as far forwards as the middle of the orbital
region, where it suddenly curves inwards to a point at one side of the
median line, just below the pituitary space, the floor of which it perfo-
rates to enter the cranial cavity. This is the first impression produced
on laying bare the vessels in Chlamydoselachus, but, as we shall see later
on, all of this trunk lying beyond the upper end of the hyoid is foreign
to the first efferent artery, whose continuation we are to seek in the
branch uniting it with the second, and is simply the trunk of the
common carotid artery.

The commissural branch uniting the hyoidean or first efferent bran-
chial artery to tlie second is fully as large as the arteries themselves,
and from its manner of union with the second and of its separation
from the first efferent branchial makes the conclusion unavoidable that
it is in truth the continuation of the trunk of the first efferent branchial,
which however fails of an independent union with the dorsal aorta, but
in a manner similar to that in which the homologous arteries in Scyl-
lium, Acanthias, and Zygsena (Hyrtl, loc. cit., Plates I.-III.) are united,
i. e. by the fusion of the latter with the next succeeding branchial
artery (see Figs. 10 and 11). This method of fusion is carried to
an extreme in the bony fishes, where all the efferent branchials of each
side unite to form the single pair of aortic roots (Midler, loc. cit.,
Plate III. Fig. 13), and is also represented in Chlamydoselachus in the
posterior section of the effei'ent branchial system by the fusion of the
sixth and fifth efferent branchials to form a single aortic root. Under
primitive conditions of the hyoidean gill this anastomotic vessel would
take blood from the anterior demibranch of the hyoidean gill sac and
the posterior demibranch of the mandibular gill sac, — holobranch of
Parker. On account of the reduction which has taken place in the
spiracular gill among existing Elasmobranchs, this vessel serves simply
to convey the blood from the anterior demibranch of the first gill sac
into the efferent artery of the first holobranch.

As in the afferent branchial system, so iu the efferent branchial
system, the component arteries are arranged in pairs, and the pairs
correspond, though the paired condition is less marked in the eflFerent
than in the afferent system.

MUSEUM OF COMPARATIVE ZOOLOGY. 197

The first and second efferent branchials unite to form a trunk that
reaches the aorta as the second aortic root. The third and fourth pairs
find independent outlets as the third and fourth aortic roots, while the
fifth and sixth pairs fuse directly to produce short common trunks, the
last or fifth pair of functional aortic roots.

The distance between the points at which the first and second and
the second and third efferent branchials enter the aorta is nearly the
same, while the common trunks of the fifth and sixth enter the aorta at
a distance behind the fourth twice as great as that between the other
pairs.

Unlike all other gnathostomous vertebrates, Chlamydoselachus has
a dorsal aorta (dorsal vessel) running the entire length of the noto-
chord, to which it is intimately attached through the greater part of its
course. There is, however, no trace of a chondrification of its walls,
such as frequently occurs in cartilaginous fishes (e. g. Sturio). For con-
venience in describing as well as on morphological grounds, it is desira-
ble to designate two sections of the dorsal aorta by the terms precardiac
and postcardiac. The former receives the aortic roots and supplies the
head with arteries, the latter gives off" the arteries to the trunk and tail.

The head and all the precardiac portion (see Figs. 1 and 2) of the
trunk are supplied with blood by means of the very primitive mns-
culo-spinal arteries and branches of functional or rudimentary aortic
arches. This fact is of the greatest importance in any discussion of the
homologies of the head arteries, or more exactly precardiac arteries, of
the higher Vertebrata. These vessels where more or less rudimentary
as regards their main trunks have undergone secondary changes, during
which the course of the blood currents through them may have been
reversed, and they have usually acquired new outlets, or inlets, as the
case may be.

It is convenient to make a further division of the precardiac section
into cranial, vertebral, and branchial portions. The precardiac section is
marked off posteriorly by the junction of the fifth pair of aortic roots
with the sub-chordal vessel. It terminates anteriorly in the pituitary
plexus. The postcardiac section extends from the junction of the fifth
pair of aortic roots to the tip of the tail. (See Figs. 1 and 2.) With
the exception of a slight enlargement in the occipital region, the diam-
eter of the aorta is constant between the occiput and the origin of the
coeliac artery ; from this point backwards the aorta gradually tapers into
the small caudal artery. From the occiput forwards the cranial sec-
tion of the precardiac aorta lies imbedded in the cartilage of the basis

198 BXJLLETIN OF THE

cranii ; sinking into the cartilage just in front of the posterior border of
the basi-occipital cartilage, it runs forwards, gradually rising to the
inner surface of the cranial floor, remaining equidistant. from the chorda
until near the anterior end of the latter, when the aorta dips slightly
to make a bold curve upwards into the pituitary prominence within
which it gives ofi' two lateral branches which separate from the median
vessel only gradually. (See Figure 3.) These three vessels make their
way through the cartilage, and by freely anastomosing with one another
form a small but sharply defined plexus, crowning the pituitary promi-
nence, but separated from the cartilage by several well defined layers of
connective tissue, one of which bridges over the pituitary depression,
and thus excludes the internal carotids from the cerebral cavity. The
plexus lies external to the dura mater.

As we have seen, the aorta is made up by the confluence of six pairs
of efferent branchial arteries, which pour their blood into the aorta
through only four aortic roots, and in this condition we recognize the
process of reduction, transposition, or utter obliteration at work in get-
ting the creature out of a lower into a higher stage of organization.
But the four aortic roots which bring blood from the gills are not the
only trunks which from their relations to the aorta and the body make
good their claim as aortic arches or roots. As Hyrtl has pointed out,
the pair of vessels running from the internal carotid trunk to the aorta
in all sharks is most surely an aortic arch, and although it has lost its
direct connection with a gill, which we have every reason to believe it
formerly had, it still retains its connections with a trunk which has
resulted from the obliteration of a series of efferent branchial vessels,
and through this trunk an indirect connection with the first two func-
tional gills of Chlamydoselachus. Besides this pair, there is another
whose relations to the dorsal aorta are such as to entitle them to rank
as aortic arches. I refer to the two side branches which the dorsal
aorta gives off as it approaches the pituitary plexus. To these six aortic
arches I would add a pair represented by the anterior portion of the
internal carotid arteries, and another pair represented by the efferent
vessels of the spiracular pseudobranch, which pour their blood into the
dorsal aorta through the ophthalmic artery, internal carotids, carotid
plexus, and pituitary plexus, all of which vascular structures anastomose
among themselves from without inwards, in the order given. To these
still another pair may be added, recognizable in the last pair of efferent
branchial arteries (the sixth functional pair), which by means of their
fusion with the fifth pair have not been counted. Chlamydoselachus

MUSEUM OF COMPARATIVE ZOOLOGY. 199

has then, iu all, nine aortic arches and remnants of arches, persisting
from an earlier and more primitive condition of organization. This
count is based on the demonstrations and suppositious, (a) that the
anastomotic branch between the hyoidean demibranch and first efferent
branchial is the continuation of the hyoidean efferent artery ; (6) that
the internal carotids after turning inwards and entering the pituitary
space unite with the dorsal aorta ; (c-) that the anterior end of Ihe cranial
aorta divides, and that the lateral vessels curve outward each side of
the anterior end of the notochord ; (d) that the efferent artery of the
spiracular pseudobranch is connected with the aorta ; and (e) that the
sixth efferent branchial artery formed at one time (probably during
embiyonic life) an independent aortic root. The fact should not be
overlooked that we thus find remnants of two, and perhaps three, aortic
arches in the pituitary space! (See table on page 218, and Figs. 1
and 2.) That the above method of counting the aortic arches in
Chlamydoselachus is a correct one, within very narrow limits of error,
no one will question who recognizes the vast changes that have been
brought about in the vascular system of a higher vertebrate — mammal,
for instance — -during its phylogeuetic course from the fish type up-
wards, and who recognizes the general law that an organism makes use
of rudimentary or disused structures to build up other structures, of
different function perhaps, for use under changed conditions of environ-
ment, provided the rudiment or disused structure be suitably placed.
The only portions of the vascular apparatus of the branchial region that
are suitably placed for use in case the aorta is reduced, are evidently the
dorsal and ventral commissural systems. That the dorsal vessel should
be chosen of the two is farther evidence of tlie law, for the dorsal
commissure is both more directly connected with the territory to be
supplied, and lies deeper in the tissues in a direct line toward the
brain ; besides, it normally carries the purified blood from the gills,
-which the ventral commissures do not to so great an extent, lying as
they do on the side of the gills where the currents are forming and
setting towards the dorsal vessel. As a further illustration, we find in
some species of Myxine the remnant of a ductus Botalli. This remnant
was in early adult life hollow, and connected the gills of its half of the
segment with the dorsal aorta. MuUer found in some cases that each
end of the thread-like remnant was still hollow. These tlireads arise
from the afferent branchial artery of the anterior gill sac, jjassing
thence upwards and forwai'ds, and fuse with the carotid trunks where
the latter anastomose with the first efferent branchial arteries. In

200 BULLETIN OF THE

this instance the vessel is atrophied during the process of the reduction
of the anterior gill, and the blood which formerly passed through it
direct to the collecting vessel above its gill now passes backward,
and its segment of the collecting vessel is now either carotid or
carotid root.

Since the number of efferent branchial arteries uniting to form the
dorsal aorta varies in different species, the dorsal aorta cannot be an
equivalent structure throughout the vertebrate series.

We know that the cranium itself is not an equivalent structure within
the limits of the Elasmobranchii. Abundant proof of this has been
collected by Gegenbaur,^^ Froviep,-^^ Van Wijhe,^* Dohrn,^^ and others.
Evidence based on the relation of blood-vessels to the cranial floor
is worthy of note in this connection. In some Elasmobranchs the an-
terior pair of musculo-spinal arteries pierces the cranial floor, in other
forms this pair is intimately related to the atlas. "Where the former
condition is present, we can say definitely that at least one vertebra
has been added to the cranium, but in the latter case wo may have to
deal with a suppression of one or more pairs of musculo-spinal arte-
ries, in which case we cannot draw conclusions as to the constitution of
the cranium.*

Since there can be little doubt, if any, that the j^rimordial cranmm is
the same in all cases, it follows that, during the early stages of on-
togeny, segments (cranial vertebra)) must have been added to the occip-
ital region in a large number of cases, and we are at once confronted with
the difficulty of determining the number of such added segments, and

1- Gegenbanr, C. Die Metamerie des Kopfes u. die Wirbeltheorie des Kopf-
skcletcs. Morpli. Jalirbuch, XIII., 1887.

13 Froriep, A. Ueber ein Ganglion d. H3'poglossus u. Wirbelanlagen in d. Oc-
cipitalregion. Arch. f. Anat. u. Physiol., 18G2.

1^ Yaw Wijlie, J. W. Ueber die Mesodermsegmente u. d. Entwickl. d. Nerven
des Selachierkopfcs. Konigl. Arad. d. Wissensch., 1882.

15 Van Bemmelen, J. F. Ueber vermiithliehe rudimcntiire Kieraenspalten bei
Elasmobranchiern. Mitt. d. zoolog. Stat, zu Neapel, VI., 1«85.

^•^ Doiirn, Alex. Studien zur Urgeschichte der WirbelthierkiJrper. Mitt. d.
zoolog. Stat, zu Neapel, VII., 1887.

* Transsections of the basis cranii of Chlamydoselaehus taken from the verte-
bral junction fcirward sliow at intervals calcified tracts leaving the central peri-
cliordal crust, and extending on either side out into the liyaline cartilage. Tliey
corresi>ond in position to the neural arches, transverse processes, and hypapophyses
of tlie vertebrte. (See Figure 8.) It may be possible to determine the nund)er of
vertebra; entering into the basis cranii of Chlamydoselaehus by making a perfect
series of such sections and counting the number of tliese vertebral remains.

MUSEUM OF COMPARATIVE ZOOLOGY. 201

thereby the limits of the primordial cranium. It is apparent from
Miiller's studies, that the Myxinoids possess a typical and well devel-
oped system of aortic vessels, and that Petromyzon differs more in
degree than in kind from the Myxinoid type, for the structural plan is
undoubtedly the same in both. The views which have hitherto been
held by morphologists of the nature of the aorta in Craniates do not
permit us to establish a homology of parts between this group and Am-
phioxus, the only living representative of the Acraniates.

Examining first of all the vascular system of Amphioxus as the type
from which we may expect the simplest exposition of the fundamental
vertebrate plan of structure, we find that it resembles in many respects
the annelid type. For our purpose now, it will be sufficient to describe
its precardial and postcardial sections. We find that, while they are not
distinctly separated, the former corresponds to the branchial and pre-
branchial systems of vessels, the latter to the dorsal aorta, its branches,
and its complement, the ventral (subintestinal) vein. Of importance is
the fact that there are at first two aortic arches, forming the anterior
termination of the aortic vessels and in this case of the vascular system,
one of which disappears later in life quite completely as an arch, but
persists in part as an artery supplying the naso-facial region, and that
these arches do not project to the anterior end of the notochord, — a
condition that may or may not be a secondary one.

Langerhans^ {loc. cit., p. 337) describes the arches and dorsal aorta as
follows : " Von der Arteria branchialis gehen zwar Gefasse unter dem
Constrictor veli zum Mund. Dann aber setzt sich das Herz fort in einem
sehr weiten rechts verlaufenden Aortenbogen, wahrend es links keinen
ahnlichen entsendet, sondern geschlossen ist. Dieser rechte Aorten-
bogen zieht hiuter dem M. constrictor veli nach oben, liegt in seinem
oberen Theil mit dem Muskel, zum Theil in derselben Querebene und
verbindet sich mit der rechten Aorta, wahrend die linke anscheinend in
keine Beziehung zu ihm tritt. Der Theil der Aorta unmittelbar hinter
der Einmiindung des Aortenbogens in die rechte Aorta ist bei beiden
gleich weit. Nach vorn aber setzt sich die linke Aorta bis zur
Mundhohle als schmales Gefass fort. Eechts dagegen biegt der grosse
sinuose Aortenbogen, nachdem er sich mit der rechten Aorta verbunden,
nach vorn um und erstreckt sich etwas unterhalb und seitlich von der
Chorda gleichfalls bis zur Mitte der Mundhohle nach vorn, um hier
abgerundet zu enden."

- o^

9 Langerhans, P. Zur Anatomie der Amphioxus lanceolatus. Arch. f. mikr.
Anat, Bd. XII, IRTG.

202 BULLETIN OF THE

Hatschek's studies of the development of the organs of Amphioxus do
not seem to have been extended to the vascular system, for he does not
mention the blood-vessels in his paper.

Schneider ^° quotes Langerhans, and adds the following observations
of interest. He says {loc. cit., p. 26) : " Am oberen Ende entspringt
von jedem Kiemenstabe eine (Taf. XIV. Fig. 2 v. b.) Kiemenvene
■welche bogenformig ein wenig nach riickwarts verlaiift und sich mit der
Aorta verbindet welche jederseits unter der Chorda liegt. Die Kiemen-
venen sind sehr dunnhaiitig, man' kann sie nur sehen, wenn sie mit
Blut erfiillt sind. Die Aorten der Kiemengegend zerfallen in zwei
Theilen. Der untere Theil liegt in der Falte welche die obere Branchial-
rinne seitlich begranzt, ihr Querschnitt ist spitzwinkelig, der obere
Theil liegt in dem Bindegewebe welches der Chordascheide nach unten
aufliegt und welche zu dem Gallertgewebe gehcirt. Die Aorta ist in der
Kiemengegend doppelt, hinter derselben wird sie einfach bis in das
Schwanzende. . . . Von der Aorta gehen dreierlei Zweige ab. 1. Arterieu
nach oben fiir die Muskeln der Leibeswand ; 2. Arterien an der Innen-
flache derBauchhohle. 3. Capillaren fiir den Darm. Obgleich die Zweige
der ersten imd zweiten Gruppe in ihren Yerlauf den Arterieu hoherer
Thiere gleichen, lassen sich doch ]\[uskeln an ihnen nicbt wahrnehmen.
Von der ersten Gruppe entspringt je ein Zweig in eiuem Myocomma
ungefahr in der Mitte desselben. Man kann ihn nach oben verfolgen
bis liber die Mitte der Chorda am weitesten in Kopftheile (Taf. XIV.
Fig. I, Ao. rechts). Die zweite Gruppe, die Arterien der Bauchhohlen-
wand, entspringen an jedem Ligament und laufen auf der Innenkante
des Ligaments nach unten. . . . Die der Mitte sich immer naherende
Fortsetzung der rechten Aorta sich bis in die Spitze des Kopfes verfol-
gen lasst. Die linke Aorta verhalt sich unregelmiissig, sie geht in
verschiedenen Exemplaren verschieden weit. . . . Audi Queraste sind
vorhanden, welche beide Aorten verbinden. AVenn die linke Aorta friih
verschwindet, liefern diese Queraste der rechten Aorta alle die Zweige,
welche sonst der linken ausgehen wiirden."

According to Johannes Miiller, the aorta in Petromyzon bifurcates a
short distance in front of the anterior pair of gill sacs, anterior to the
point of origin of the common carotid trunks, the two diverging limbs
communicating on either side with the right and left common carotid
trunks respectively. But in Myxine the aorta is continued to the
anterior end of the notochord, and in fact extends by means of short

^'^ Schneider, A. Beitrage zur Vergl. Anatomie u. Entwick. d. Wir^elthiere.
Breslau, 1879.

MUSEUM OF COMPAKATIVE ZOOLOGY. 20

o

terminal bi'auches (e. g. palato-nasal) to the anterior end of the body,
and there is no distinct bifurcation, the vessel, on the other hand, being
gradually reduced by the numerous lateral branches given off after its
passage thi'ough the circulus cephalicus. The aorta iu Myxine is
a relatively large vessel until it unites with, or, better, receives the
two converging bi-anches of the circulus, when it suddenly contracts,
and from this point on is clearly much reduced. This portion of the
canal still serves to keep oj^en a direct passageway from the heart to
the head, inasmuch as the arterial blood collected from the gills is not
only forced upwards and backwards, but also forwards, into the continua
tion of the aorta, or A. vertebralis impar, and into the carotid arteries.

I think, from the evidence gathered in the foregoing paragraphs, that
we are now in position to say definitely that between Amphioxus and
the Myxinoids on the one hand, and Chlamydoselachus as a representa-
tive of the Elasrnobranchii on the other, it is easy to establish a homol-
ogy of parts sui'prising in its completeness. The entire doi'sal aorta
exists in Chlamydoselachus, imbedded in part, it is true, in the basis
cranii, while in Mj'xine it lies in the connective tissue underneath the
latter. The long ventral aorta in these two forms has been much short-
ened, but still in showing traces of its reduction claims a descent from
an Amphioxus-like type. Of course we should not expect to find the
ventral aorta persisting after the gills in front of it had cei".,sed to be
functional, and it might easily shorten before such reduction of the
branchial apparatus had taken place, provided means were at hand to
enable it to perform its function of pouring blood into the gills. The
dorsal aorta, on the other hand, being a distributing trunk in a large
sense, would be looked for so long as its territory existed and was
not entirely supplied by new vessels ; and as we know that its teri'itory
persists in all vertebrates, and gi'eatly increases in extent among the
higher forms, the latter alternative is the only one we need consider
farther. We find, on examination, that the recession of the heart is
accompanied by the usurpation of the precardiac aortic territory by
some of its lateral branches or their smaller offspring. While it is trvie
in general that a reduction of the ventral aorta is followed by a reduc-
tion of the dorsal vessel, it is also true that the latter process takes
place much more slowly, and for the reasons given above. The only
indication of a jDersistence of the dorsal aorta in groups above the
fishes, of which I have been able to find reliable account, is given by
Goette."

11 Goette, Alex. Entwickelungsgeschichte der Unke. Leipzig, 1875.

204 BULLETIN OF THE

His figures on Taf. XVII. Figs. 306, 317, and 316 a, c, of frontal sec-
tions from embryos of Bombmator igneus show that the aortic roots
anastomose by means of transverse vessels, and that at an early stage
of aevelopment there is an anterior (median) prolongation of the dorsal
aorta between and beyond the aortic roots ! How long this remnant
persists is not stated by the author.

A series of transsectious (see Figs. 5 a to 6 g) through the basis
cranii of Chlamydoselachus in the regions designated below as 1, 2,
and 3 (see Figure 1) gave the relations of the chorda and aorta to
each other shown in the following table. The distances refer to meas-
urements in the perpendicular to the long axis of the animal. The
sections were taken from three portions of the basis cranii containing
the structures cut from the vertebral junction, middle distance, and the
pituitary region respectively, the last piece containing the whole of
the '' Sattellehne " and the pituitary space, with the foramina of en-
hance of the carotid arteries and the transverse canal.

TABLE OF RELATIONS OF CHORDA AND AORTA.
(c = chorda, a = aorta.)

( 1 ) From the vertebral junction.

c in middle of section.

a below and entirely outside basis cranii.

2

Section No. 1 \

( c, the same as No. 1.
I a, "

^{:

in middle of section.

inside cartillage, but on lower boundary.
( c in upper lialf.
( a in middle of lower quarter,
c in upper half.

(a ir

' ~ in middle of lower half.

( c in upper half.

( a in upper quarter of lowei half.

From this point on, the rise of the aorta is very gradual until the section
( 2 ) from middle distance between occipital region and pituitary space is
reached, when in

Section No. 1' 5 ^ '" "PP" ^^'""^ ^^ '^"*^™-
( a in the middle of section.

( c in upper quarter of upper half.

■^ ( a in middle of section.

c, the same as No. 2'.

3
4'

, j c, th(
1 a, "
U,"
la. "

MUSEUM OF COMPARATIVE ZOOLOGY. 205

( c in upper quarter.

\a in the middle third but above the middle.
From tlie middle region the chorda and aorta both rise until each
breaks through the surface of the cartilage, the chorda to end in the cal-
careous incrustation of the cartilage, and the aorta to enter the pituitary
space from the summit of the saddleback. Two sections from region
( 3 ) gave in

(■ c on the upper surface of cartilage.

Section No. 1" -s • ^ - , ,r

la in upper quarter of upper half.

« « 0" 5 ^ section in front of chorda end.

( a emerges from top of saddleback.

It will be seen from this table that the chorda occupies in the occipi-
tal region the middle of the cartilaginous plate of ttie cranial floor,
and that from the end of the cone-shaped body, which was in the indi-
vidual dissected about 1 cm. long, the thread remnant rises gradually
but continuously until it reaches the inner surface of the floor of the
cranium. This point, as the sections show, was behind the apex of the
pituitary eminence (Sattellehne). Further, we find that the course of
the aorta remnant is very nearly parallel with that of the chorda, and
that it issues from the apex of the pituitary eminence in exactly the man-
ner described by Gegenbaur for the chorda of other Elasmobranchs.

Gegenbaur " has investigated the subject of the chorda termination
very thoroughly, and judging from the text and plates mostly by means
of longitudinal section. There are several figures of cross sections of
the basis cranii, showing the structure which he calls " Chorda," but
which resembles the more ventrally placed aorta as I find it in Chlamy-
doselachus. As the author says, it is almost always a very difficult
matter to determine the exact position and manner of the chorda ter-
mination in adult animals ; much easier with young animals or older
embryos. The author studied, among others, the genera Acanthias,
Heptanchus, and Centrophorus. He says (loc. cit., p. 121): "Die
Chorda tritt mit ihrer Aufwartskriimmung immer naher an die in-
uenflache der Schiidelbasis und steigt dabei in der Sattellehne em-
por, welche sie dicht unter deren hinterer Flache durchsetzt, um niihe
an der Kante dieses Yorsprunges unter das Perichondrium zu treten.
Wo die Sattellehne starke corticale Verkalkungen zeigt ist das zuc^e-
spitzte Ende der Chorda noch in diese eingebetet. Fig. 7, Taf. XIV.
gibt eine Darstellung dieses Verhaltens in einem 22 cm. langen
Acanthias Embrj-o auf dem Medianschnitte. Beim ausgewachsenen

1" Gegenbaur, C. Das Kopfskelet der Selachier. Leipzig, 1872.

206 BULLETIN 05 THE

Thiere ist derselbe Zustand vorhanden, doch ist die Chorda im Ver-
gleiche zum Basalknorpel bedeutend schwiicher und die erweiterte Stelle
ist nur angedeutet. Das hervortreten des Chorda-Endes aus dem Ba-
salknorpel und die Lagerung unter dem bezuglicheu Pei'ichoudrium hat
KoUiker bereits gesehen, jedoch nicht der sehr auffalligen Beziehung
zur Sattellehne sonderu nur der Gegend der vor der Sattellehne gelager-
ten Hypophysis Erwahnung gethau : In einigen Fallen sah ich das aus
dem Knorpel der Sattellehne hervortretende freie Ende der Chorda liber
die Kante der Sattellehne nach voru lamgebogenen aber immer noch
unter dem Perichondrium verlaufend. So einmal bei einem 24 era.
langen Embryo von Acanthias, aber auch bei einem grossen Exemplare
von Centrophorus granulosus. Obgleich ich noch vier Acanthias-Emhryo-
nen darauf untersuchte, gelang es mir nicht, ein jenem dhnliches Verhalten
verhreitert zu finden . ' '

In the genera Heptanchus, Hexanchus, Centrophorus, Acanthias,
Squatina, and Cestracion, the author traced the chorda dorsalis through
the basis ci'anii to the saddleback (Sattellehne) of the pituitary depression,
and found that the chorda remnant as regards shape, size, and position
was very much alike in Hexanchus and Heptanchus. This I'emnant was
in the form of an elongated conical body projecting into the occipital
region of the basis cranii, forming of course the anterior continuation of
the chorda in the vertebral column. From the apex of the cone was
given off a pale thread of considerable size, which ran forwards usually
parallel to the outer surface of the floor of the cranium. The arch
formed in approaching the pituitary space is much weaker in Heptan-
chus than in Hexanchus. In Cestracion behind the pituitary saddle-
back the chorda swells out into a spindle-shaped body, whose fibrous
sheath is filled with a cartilaginous tissue, containing numerous round
hyaline cells. The structure of this spindle-shaped body, so far as
Gegenbaur's description goes, agrees with that of the ventrally placed
aorta in Chlamydoselachus, and not with the chorda, though it does
agree with the chorda and its tubular enclosure in a very large (10 ft.)
individual of Heptabranchias.

AYith reference to the persistence of the chorda in the cranial floor,
Gegenl)aur says (loc. cit., p. 122) : " Mit dem Xachweise der Fortdauer
eines Theiles der Chorda dorsalis im Cranium mancher Selachier ist fiir
diese ein niederstehendes Verbal tniss aufgedeckt, namlich die For-
setzung: eines bei den moisten xVbtheilun^en der ubrigen Vertebraten
bekannten embryoiialen und damit verganglichen Zustandes, der von den
ihn daurend besitzenden Formen her sich ableiten lasst. Aus jenen Yer-

MUSEUM OF COMPAKATIVE ZOOLOGY. 207

halten ergibt sich jedoch noch eiu anderes bedeutungsvolleren Moment,
jeues, namlich, welche deu von der Chorda durchsetzten Abschnitt des
Crauiums in gleichen oder doch zunachst abnlicheu Beziehungen zeigt,
wie sie die Wirbelsaule zur Chorda besitzt, so das darauf eine Vergleich-
uug jenes Abschnittes des Craniums mit einem Abschnitte der Wirbel-
saule sich stUtzeu kaun." And further, " Aus der vollen Wiirdigung
dieser Beziehung von Gehirn und ISTerven des hinteren Abschnittes
ergibt sich der offene Gegensatz zum Vorderen Schadelraum, der von
dem hinteren sehr verschiedene Gehirntheile umschliesst und ebeuso in
den ihn vei'lassendeu Nerven keinerlei Gemeinsamkeit mit deu von
Spinaluerven ableitbaren hinteren Nerven wahi-nehmen liisst. Die
Resultate der Vergleichung der einzelneu Abschnitte des Binnenraumes
am ausgebildeten Cranium sind somit mit der Priifung der Sonderuugs-
vorgange bei der Entstehung des Knorpelcraniums im Einklange."

To which if we add the weight of evidence afforded by the study of
the vascular arrangements described in a previous paragraph, we have
increased reason for the separation of the prechordal from the chordal
section of the cranium. For with this addition there is not a single im-
portant structure entering into the composition of the head which does not
shoio traces of the originally distinct separation of these regions, 7iow so
closely united among cdl the higher vertebrates.

Important in this connection is the relation of the chorda in Bom-
binator igneus figured by Goette, loc. cit., Taf. IX. Figs. 164, 165, and
166, Taf. XV. Figs. 283 and 284, in which the author found the chorda
dorsalis lying below the cartilaginnous cranium, although in intimate
contact with it. After its degradation, which takes place in an early
stage of development, this portion of the chorda is converted into a
(keel-shaped 1) calcareous crust, projecting from the ventral surface of
the cranium. It is evident that in such a case the dorsal aorta could
not becotpe enclosed in the cartilaginous cranium, and in this fact we
may have an explanation why a remnant of the aorta is not more
uniformly found among the Elasmobrauchs in general. It is important
to bear in mind the condition of the head region before the cartilaginous
cranium has been formed. In such a primitive animal, or at a cor-
responding stage in a n\ore advanced form, the notochord, dorsal (and
possibly ventral) aorta, digestive tract, and nervous cord extend through
the head region, all parallel to the long axis of the body, and held in
place by the connective tissue lying between them and whatever mus-
cular and skeletal structures may be present. Branches from the ner-
vous axis or the aorta easily find their way through the yielding tissue

208 BULLETIN OF THE

to their destination.* So long as the intervening spaces remain filled
with loose connective tissue, all these spaces are eminently vascular and
h-inphatic. The need of greater strength and power of resistance in the
slceletal axis calls for a solidification in and about the notochord, and
upon the extent of the hardened area depends the nature of the enclos-
ures. Usually, of course, the notochord forms the centre of the solidi-
fied tract, but we have just seen that it may lie on the ventral border
of this tract. The structures transverse to tlie axis are partly enclosed
in the solidified tissue, — the proximal portions in the case of the nerves,
the proximal or distal in that of the blood and lymph vcssels.t

The e]uestion of the homology of the carotid arteries has been touched
upon by many morphologists, and although the subject has never at-
tracted any very great attention, several explanations have been proposed
at various times. The usual one found in our text-books on compara-
tive anatomy and embryology was the result of determinations made by
the earlier embryologists, Bischoff, Rathke, and othei'S, of the ontogeny
' of the vascular system in mannnalian and other embryos.

Ivolliker's " account contains the whole matter in clear and concise
form, and I shall quote his words as a statement of the generally accepted
views. He says (loc. cit., p. 915) : "Die erste Form derselben (i. e. die
Arterien) die gleich nach der Entstehung des Herzens und wahrend der
Dauer des Kreislaufes im Fruchthofe getroffen wird, ist die (Fig. 560. 1)
dass das Herz vorn einen Truncus arteriosus entsendet der nach kurzen
verlaufe in zwei Ai'cus Aortse sich spaltet, die in der Wand des Kopfdarm-
hohle bogenfbrmig nach der Gegend der spateren Schjidelbasis und dann
langs dieser convergirend nach hinten laufen, um anfjinglich getrennt
von einander als doppelte Aorta descendentes zu enden und spater unter
einander zu unpaaren Aorta zu verschmelzen. Sowie die Kiemenbogen
. . . hervortreten, zeigf sich, dass der Anfang der Aortenbogen in den
ersten Kiemenbogen liegt, sowie dass audi fiir die folgenden Kiemen-

* These are the cranial and spinal nerves, the afferent and efferent branchials,
and musculo-spinal arteries for the most part, all of which do not run parallel to
tlie Inng axis.

t The hlood-vessels are affected most by this process, and all except the im-
portant trunks rapidly atrophy, leaving as a last trace a fibrous cord imbedded in
the solid cartilage.

1** fvolliker, A. Entwick. des Menschen, etc. Leipzig, 1879.

1^ Eathke, H. Entwickelungsgeschiclite der Natter. Konigsberg, 1839.

2^ Rathke. Ueber die Entwickelung der Arterien welche bei Saugethieren von
den Bogen der Aorta ausgehen. Arch, fur Anat. imd Physiol., 1843.

21 Kathke. Entwickelungsgeschichte der Wirbelthiere. Leipzig, IBGl.

MUSEUM OF COMPAEATDTE ZOOLOGY. 209

boo-en neue Aortenbogen hervortreten. ... Die bleibenden grosse Ar-
terieu geliea im wesentlichen aus den drei letzten Aortenbogen hervor,
dock erhdlt sich audi ei/i Theil des ersten urod zweiten Bogens in der Carotis
interna und externa.

" Von den drei letzten bogen wird der vorderste (der dritte der gauzen
Eeihe) zum Anfaug der Carotis interna, wahrend der Carotis commu-
nis aus dem Anfange des urspriinglichen ersten . Arcus Aortse sich ent-
wickelt. Yon der Aorta thoracica und abdominalis hat Remak zuerst
gezeigt das dieselbe beim Hiihnersembryo anfanglich doppelt siud,
indem die ersten Aortenbogen nicht vereinen sondern als sogenante
' primitive Aorten ' von der Wirbelsaule einander parallel bis zum hm-
teren Leibesende fortgehen. . . . Erst am dritten Tage verschmelzen
diese 'primitiven Aorten ' in ihren vordersten an der Wirbelsaule gele-
genen Theile. . . . Hier sind die langstbekannten Arteriae vertobrales
posteriores nichts anders als die ' primitiven Aorten ' und stellen zahl-
reiche Figuren dieses werkes dieselben als paai'ige Gefasse am Ivopte und
am Rumpfe dar. Die Verschmelzung dieser Gefasse . . . schreitet nach
hinten fort."

Important points in this consideration are (a) the double condition of
the aorta in the developing mammal ; and (6) the persistence of portions
of the first and second aortic arches in the carotid arches.

More recently Macalister -- has oftered an explanation of the carotid
arteries which does not harmonize with the facts as I find them, and I
shall first quote his remarks, and then show wherein it appears that his
conclusions are not tenable in the light of the comparative anatomy of
the lower fish forms. He says {loc. cit., p. 193) : " The an-angemeut of
the blood-vessels in the adult forms of the lowest, and in the embiyos
of the higher vertebrates, indicates that the history of the complicated
vascular system of the higher forms has been one of a development from
a simple and regular ancestral condition of metameric and intermeta-
meric vessels, through easily defined stages, to the more confused and
in'egular condition of the arterial system in the adults of the higher
forms. ... In the whole organism the vessels would thus form a
double series, two longitudinal ventral trunks, two corresponding dorsal
trunks, and the lateral uniting trunks in each segment. There are two
primitive dorsal vessels in vertebi'ate embryos, and their fusion can be
traced in the chick beginning at the forty-second hour of incubation.
This union commences behind the head, and travels backwards rapidly,

22 Macalister. The Morphology of the Arterial System in Man. Journ. Anat.
and Physiol., XX., 1886.

VOL. XVII. — xo 5. 14

210 BULLETIN OF THE

SO that after the fifth day there is but a single dorsal vessel for the middle
and the hinder part of the body, — the dorsal aorta. In the region of
the head and neck of mammals, the foremost ends of the two vessels re-
main i^ernianently se-parate as the internal carotid arteries. [Italics mine.]
. . . There were also originally two ventral longitudinal vessels, but
their union probably occurred even earlier than that of the dorsal. . . .
The setting apart of one portion of the single ventral vessel to form the
heart differentiates the pre- from the post-cardiac portions of the ventral
vessel. As a consequence of the cardiac differentiation, the only places
where complete metameric arcades remain are the precardiac segments.
. . . Behind the heart in higher vertebrates a series of vessels extend from
the dorsal aorta through the mesogastric fold, and end in the splanch-
nopleure ; these correspond to the dorsal extremities of the postcardia
lateral metameric arcades. With the condensation of the anterior seg-
ments which takes place in the formation of the skull, all distinct vas-
cular metamerism is lost, and the anterior segmental arches become
displaced backward or lost. The common and external carotids are con-
tinuations of the ventral aorta, while the root of the internal carotid is
the altered relic of the third arch, and the ascending continuation of
that vessel is the upper part of the dorsal aorta. . . . The only carotid
branches which in any way represent rudimental arcades are the occipi-
tal and posterior auricular arteries. . . . The cervical dorsal aorta (in-
ternal carotid) has only rudimental branches in the neck, represented
by the intercarotid ramuli. Its intracranial continuation gives off three
lateral neural branches, the posterior, middle, and anterior cerebrals
(the first originally being a carotid branch, its root being the so called
posterior communicating, but its anastomotic internal branch, which
joins the median anastomosis, dilates so as to form its functional
root). The ventro-lateral branches are reduced and modified as tym-
panic, vidian, receptacular, and ophthalmic branches."

As the foregoing quotations show, Macalister tacitly assumes that the
double nature of the dorsal vessel in mammalian embryos is comparable
with that of some vermian type. He claims that the anterior ends of
the double dorsal vessel are transformed into the internal carotid arteries
of the adult, which is certainly true, but not in the sense our author
intends it, as appears from the context, for he says further on, that the
root of the internal carotid is the altered relic of the third arch. These
views were, I believe, first expressed by Allen Thompson ^®' ^'^ in 1831,

26 Thompson, Allen. The Development of the Vascular System. New Phil.
Journ., Edinburgh, 1831,

2' The same. Quain's Anat., 9th ed., II., 1882.

MUSEUM OF COMPAKATIVE ZOOLOGY. 211

who also claims to have discovered the facts on which they were based.
Surprising is the statement tliat the external carotids are the continu-
ations of the ends of the anterior bifurcation of the ventral aorta.

It is not possible at the present time to prove beyond doubt that the
ancestor of vertebrates possessed only a single dorsal vessel ; but the
best evidence at the present time (the anatomy and development of
the higher worms, and of those vertebrates retaining most of 'the ances-
tral features) points to an ancestral form having a single median dorsal
vessel.

The embryological evidence cited by Macalister in support of his
views, is interpreted by Balfour, Gegenbaur, KoUiker, Hertwig, and
others, to be the effect of the shortening of the period of development,
the suppression of some of the stages and the adaptation to peculiar
embryonic environment.

In passing over the intermediate stages between the fish and mammal,
our author has lost sight of the homologies of the vessels he deals with,
and, so fiir as his account runs, has not seen the precardiac aorta in any
vertebrate, but considers the two commou carotid trunks to represent
the pair of aortte which his theoiy calls for.

The evidence which I have presented in the preceding pages shows
beyond question that the carotid arteries, instead of being derived from
the aorta or any of its branches, are derived from the commissures
which serve to connect the efferent branchial arteries with one another.
The bifurcated end of the aorta in the bird and mammal is only a rem-
nant of a previous complicated vascular apparatus. It is likewise
obvious that the carotid vessels cannot strictly be said to arise from, or
constitute the remains of, any particular pair of aortic arches, but repre-
sent all that is left of the commissural trunk from the most anterior
arch of the ancestral form to the most anterior arch of any given exist-
ing form.

About the time Macalister's paper on the homology of the blood-ves-
sels of man appeared in England, T. J. Parker, working in Xew Zealand,
published a paper in the Pliilosophical Transactions on the vascular sys-
tem of Mustelus antai-ciiciis, in which he advances decidedlv interestinsr
views as to the homology of the carotids of this southern shark. In the
first place, Parker proposes to establish the terminology of these vessels
on a scientific basis, and as the result of his studies objects to the use
of tlie terms " internal" and '■ external " to designate the anatomical
relations of the carotids as really misleading. He would substitute and
use exclusively throughout vertebrates the terms "anterior " and " pos-

212 BULLETIN OF THE

terior," * as correctly describing the morphological relations of the ves-
sels. Now, in as far as the terms are applied to the carotid arteries of
the higher vertebrates as names simjjiy, it matters little which set is
nsed ; but when it is proposed to select terms that shall harmonize with
the development of the vessels f under consideration, superior and infe-
rior are undoubtedly the correct ones, or dorsal and ventral.

In the light of Chlamydoselachus we may reasonably choose to retain
the terms internal and external as describing at the same time the
primitive condition of the vessels in the lowest vertebrates, and, when
we consider the relations of the vessels to tlieir territory of distribution
(their only constant relation), and not alone their origin (in the anatomi-
cal sense) and source of supply (their constantly varying relation), also
their anatomical relations up to and including man. The vessels are
really never anterior and posterior so far as their points of origin are
concerned, and this, as I understand it, is the basis of Parker's deter-
mination. As shown above, in primitive forms (e. g. Amphioxus and
Myxine), the region supplied by the internal carotids iu Acraniates
and Craniates is provided for by branches from the superior portion of
the curve of the first pair of aoi'tic arches. The carotids arise from
the dorsal prolongations of the aortic roots ; i. e. from the tract ho-
mologous with the superior commissures of Elasmobranchs, and not, as
Macalister and others suppose, the anterior ends of the two lateral
aortce on the one hand, or the corresponding parts of the ventral aortee
on the other. The carotid arteries are, in a strict sense, separated from
the aorta by the vessel crossing the space between the dorsal end of the
gill and the aorta whicli lies in the middle line. This vessel is equal to
the dorsal portions of the efferent branchial arteries (or the entire epi-
branchials of Parker). It is because the inferior commissures are
merely passageways for the transmission of blood to the distributing
vessels in the dorsal region of demand, that they do not persist, since
their function is early assumed by other vessels. As the aorta is gradu-
ally reduced by the backward journeying of the heart consequent on the
reduction of the branchial vessels and organs, the brain and the enclos-
ing head are removed farther and farther from the aortic arches from
which they originally received their blood supply direct, by means of

* Parker's A. cnrot. post. = Art. carot. int.

" " ant. = eff. br. art. of mandibular gill, and both these ves-

sels supply the refjion of the external carotid in sharks.

t Cf. Eathke (20) or Balfour (24). Comp. Embryol., II., 188L I refer here to
the jlevelopmcnt in the liigher vertebrates.

MUSEUM OF COMPARATIVE ZOOLOGY. 213

numerous small twigs, and the aortic roots plus the superior commis-
sures increase in importance with the retreat of the heart. In the Mam-
malia, where the aortic roots are reduced to the greatest extent, there
is a very great variety in the manner of origin of the carotids from the
aorta, or, in other words, the primitive relation of the carotid trunks to
the median aorta has undergone a variety of transformations that for
the most part are characteristic of the groups in which th^ are found.
They remain as a pair of lateral longitudinal vessels, each of wliich
almost universally divides into two branches, a dorsal and a ventral.
The dorsal supplies the strvictures contained in the cranium, and gives
off vessels into the orbital space ; it is the internal carotid. The ven-
tral branch supplies the visceral portion of the head and the cranial pari-
etes, at least in part ; it is the external carotid artery. The external
carotid is then only a ventral branch of the dorsal commissural trunk,
while the internal carotid continues the main stem of the common
carotid. Of course, the relative sizes of the vessels iindergo ceaseless
variations as we ascend the series ; but the rule is, that the higher the
position of the animal in the series, the more important relatively the
territory supplied by the external carotid becomes. For example, in
man the two vessels, internal and external, are of about the same size ;
in the cat, the internal is small and the external correspondingly large.
The two vessels may be united into a common trunk, and always are
when the aortic arch from which they aiise is much reduced, or they
may arise independently of one another, as in Myxine. So long as the
dorsal aorta persists entire, the carotids have no existence ; but just in
proportion as the precardiac section of the aorta is reduced, the carotid
arteries become more and more important, until they ultimately entirely
replace it, as in the vertebrates above the lower fishes. In Myxine,
" Aus dem Zusammenflus der Kiemenvenstamme entstehen vier Haupt-
arterienstamme fiir den Korper, ein vorderer und hinterer unpaarer
mittelerer, welche unter der Wirbelsaule hingehen, und zwei seitliche
vordere. Die vorderen Theile des Korpers besitzen also zwei Carotiden
und eine unpaare Wirbelarterie [i. e. precardiac; Aorta] die hintern
Theile des Korpers einen einzigen Artcrianstamm, die Aorta descendens.
Die Kiemenvenen der zwei oder drei letzten Kiemen gehen direct in die
nach vorn und unten gleich sich verlangernde Aorta. Die Kiemenvenen
der ersten oder zwei ersten Kiemen gehen nicht rnchr in die Aorta iiber,
sondern vereinigen sich jederseits in eine der Aorta parallele vena bran-
chialis communis, welche sich nach vorn als Carotide fortsetzt. Die
directe Fortsetzung der Aorta nach vorn, verlauft als arteria vertebralis

214 BULLETIN OF THE

impar dicht unter der Chorda und versieht die Seitenmuskeln, das
Riickgrath und Eiickenmark mit zweigen. Die Carotiden begleiten
die Speiserdhre und geben, ihren Seiten angewachsen, Zweige an sie
ab. Hinter dem Kopf theilen sich die Carotiden in zwei Aeste welche
eine Carotis externa und interna auf jeder Seite entsprechen. Die
aussei'en Carotiden vertheilen sich in dem Kopfmuskehi und in der
Lunge. Die beiden inneren Carotiden verbinden sicli bogenfdrmig unter
dem Anfang des Eiickgraths. Aus diesem bogen ; der auch von hinten
das ganz diinn gewordene Ende der uupaaren Wirbelarterie aufnimmt,
eutsteht nach vorn ein unpaarer starker Stamm. Dieser stellt gleichsam
eine unpaare Wirbelarterie des Kopfes dar, er verlauft unter der Wir-
belsJiule tiber dem Schlund nach vorn, dann unter der Basis des Hinter-
schadels und senkt sich, da wo der Basis hiiutig wird, in der Mitte iu
die Tiefe, wahrscheinlich die Hirnarterien abgebend, indem er zugleich
an dieser Stelle gabelig zwei dvinnere Aeste ausschicht, welche divirgircnd
zur Seite des Nasengaumenganges neben den Knorpeligen seitlichen
Gaumenleisten nach vorn weitergehen und dadurch in den Stand gesetzt
werden wahrscheinlich den Nasensack mit Zweigen zu versehen."

The cephalic circle is complete in only a few forms (Myxinoids, Petro-
myzon, and the lowest Elasmobranchs). Among the Eays, Sturgeon
(and all cartilaginous Ganoids'?), and Chimtera, it is incomplete in front.
But in every vertebrate except Amphioxus the internal carotids have
their ends united by anastomosis within the pituitary space (Figures
14 and 4 h) usually, but always in this cranial region.

Referring to the anastomotic branch between the hyoidean efferent
ai'tery and that from the first branchial arch, Parker suggests that the
union thus brought about is entirely a secondary condition, and that
the true efferent trunk of the hyoidean gill is to be sought in the pos-
tei'ior (internal) carotid artery. He says (p. G90) : " From the above
considei'ations, one is led to look upon the connection of the first
(hyoidean) efferent artery with the first epibranchial artery as a
secondary one, and it then becomes a matter of considerable interest
to find in Mustelus antarcticus distinct remains of the dorsal portion of
the hyoidean aortic arch, and of its connection with the dorsal aorta.
From the dorsal end of the first efferent branchial arterv arises a large
vessel, the posterior carotid artery. This tnmk passes forwards and
inwards ventrad of the proximal end of the hyomandibular, to the ven-
tral surface of the auditory capsule, and through a foramen in the skull
floor to the orbit. Its further course will be described hereafter ; the
point of interest for tho present purpose is, tliat shortly before entering

MUSEUM OF COMPARATIVE ZOOLOGY. 215

the foramen just mentioned [see Fig. 12] at the point x in Fig. 12, it
gives oflf a very slender vessel, y, which passes backwards and inwards
along the ventral aspect of the skull and vertebral column, and joins
with its fellow of the opposite side to form a delicate longitudinal
median trunk, z, which is continued backwards to the junction of the
first pair of epibranchial arteries. I think that there can be no doubt
that the posterior carotid artery, from its origin to the point x, to-
gether with its backward continuation, y, represents the dorsal portion
of the hyoidean aortic arch, or hyoidean epibranchial artery, the altered
direction of the vessel being accounted for by the changed position of
the hyoid arch. The middle trunk, z, is as obviously the actual an-
terior portion of what may be called the interhyoidean section of the
dorsal aorta. It has clearly nothing to do with the arteria verte-
bralis impar of Myxinoids, which it resembles at first sight, since the
latter is a secondary forward prolongation of the aorta altogether cepha-
lad of the gills. As this anterior portion of the dorsal aorta undergoes
complete atrophy — if indeed it ever exists — in the Rays as well as in
the Holocephali, it is a matter of some interest to find it persisting in a
typical Selachian, and one is led to inquire whether it is actually absent
in those two forms the arteries of which have been described, or whether
it has hitherto been overlooked. I can only say that I have failed to find
any mention of it." Parker does not give the title of Hyrtl's ^ important
paper in his list of literature, and makes no referetice to it anywhere in
his text. Presumably, then, he had at least no knowledge of its con-
tents, or he would certainly have greatly modified the paragraph just
quoted. In Chlamydoselachus, the arteries described by Parker for Mus-
telus are present as a strong pair of vessels diverging from the anterior
end of the vertebral portion of the precardiac aorta, curving outward
until they reach the internal carotid trunk, into which they open, some
distance behind the internal carotid foramina. The fusion takes place
even before the internal carotids begin to curve inwards toward the
median line. Parker's conclusion, that the posterior carotid to the
point X and the small vessel y form the hyoidean epibranchial artery
is clearly untenable when applied to the more primitive Chlamydose-
lachus.

Parker's argument, that the unpaired aorta formed by the confluence
of the vessels y is not comparable with the arteria vertebralis impar
of Myxinoids as described by Miiller, is I think insufficient, since we
know nothing of its developmental history to enlighten us as to its
origin and manner of growth, and the adult condition of the vessel cer-

216 BULLETIN OF THE

tainly allows the inference that it is a reduced primitive dorsal aorta, —
the exact homologue of the dorsal aorta of any vertebrate possessing the
precardiac section.

In studying the course of the blood in the vessels of Chlamydosela-
chus, we find that the complete ellipse formed by each efferent branchial
artery in the majority of Elasmobranch* species is wanting, and a
single trunk collects the blood from all the gill leaflets borne by an arch,
and consequently from one side only of any given gill sack. This is the
primitive condition, and from Dohrn's researches we know that it is
entirely in agreement with the embryonic structure of most of the
Teleost and Elasmobranch embryos studied. It also agrees with the
adult condition of Amphioxus. Parker very justly takes exceptions
to the current use of the term " branchial vein," as applied to an
efferent branchial vessel, and I quite agree with him when he says
respecting the nature of these arteries (loc. cit., p. 688) : "These vessels
are usually, but very incorrectly, called branchial veins. It would be
quite as justifiable to speak of the portal artery &.% to call these obviously
arterial vessels veins ; a capillary system may be interposed in the
course of an arterv or of a vein, but this does not make the efferent
trunk in the one case a vein, nor in the other the aSerent trunk an
artery." The collecting trunk is continued uninterruptedly to the dor-
sa^ aorta, so that an epibranchial artery in Parker's sense is not present
in Chlamydoselachus. He says (loc. cit., p. 689) : " From the dorsal end
of each arterial loop an epibranchial artery is continued backwards
and inwards (Fig. 11) ; by uniting with one another successively in
pairs these four trunks form the dorsal aorta. ... In the embryo the
aortic arches are continued directly from the ventral to the dorsal aorta.
In the Holocephali and Teleostei there is only one efferent artery to
each gill, corresponding to the anterior of the two efferent arteries in
the Plagiostome holobranch. This is very evident in Callorhynchus, in
which the single efferent artery of each gill is always cephalad of the
corresponding afferent trunk. These facts tend to confirm the opinion
to which one is led by the simple inspection of the parts in the adult
]\Iustelus (compare loc. cit. Figs. 6 and 17) ; namely, that the anterior
efferent artery of each holobranch is to be looked upon as its primary
revehent trunk and as strictly continuous with the corresponding epi-
branchial artery, the posterior efferent artery being a secondary vessel
which debouches not into the primary trunk of its own, but into that of
the next following gill." Such are Parker's conclusions from the study

* Cf. Hyrtl, loc. cit., p. 4, and Parker (4) and (7).

MUSEUM OF COMPARATIVE ZOOLOGY. 217

of the anatomy of an adult Mustelus. As a result of his studies of
the embryological history of the efferent arteries in Pristiurus, Dohrn ^
(page 3) establishes Parker's conclusions in a very complete and inter-
esting manner.

Goette's account of the relations of the developing carotids in the em-
bryos of Bombinator is as follows : "Bevor jedoch die Aorta vollstandig
angelegt ist, entwickelt sich eine neue besondere Verbindungsbahn
zwischen den ersten Aortenbogen und der Aortenwurzel. Die Carotis
hat sich namlich schon wahrend der Entwickelung des zweiten Aorten-
bogens bis an das Wurzelstiick des ersten Wirbelbogens verlangert, un-
ter welchem sie in die Sattelgrube eintritt, um von dort aus sich in zwei
Aeste fortzusetzen. Der vorderen verlauft als ihre gerade Fortsetzung
jederseits an der anatomischen Hirnbasis nach vom wobei er durch
das Austrittsloch des Sehnerven eine A. ophthalmica abgibt ; der andere
Ast (R. communicans carotidis posterior) steigt aus der Sattelgrube ge-
rade auf und uragreift dem Yorderhirn dicht anliegend, dessen Basaltheil
oder den Hirntrichter bis an seine Oberseite, wo er in dem sogenann-
ten mittleren Schadelbalken Rathke's eingebettet ist. Von dort aus
geht unser R. communicans in die Basalarterie seine Seite iiber, welche
alsdann auch eine hintere Fortsetzung im Riickenmarkskanal besitzt
sowie diese ihre Fortsetzungen unter dem Him und Riickenmark all-
mahlich zusammenrlicken, und sich endlich zum unpaaren medianer
Stamme vereinigen, erscheint dieser alsZusammenfluss jener nach hinten
konvergirendere Kai'otidenzweige. Die beiden primitiven Wirbelarte-
rien und ihre noch getrennten vorderen Fortsetzungen, die Basalar-
terien, bilden also jederseits die hintere Halfte, die inneren Karotiden
mit ihren hinteren Verbindungszweigen die vordere Halfte eines cere-
bralen Gefassbogens welche dem extracraniellen Herz-Aortenbogen
gleichsam von oben aufgesetzt ist . . . so hurt es bald noch zu ende der
ersten Larvenperiode ganz auf, iudem die primitiven Wirbelarterien vei'-
schwinden und ihrer Gebiet ganz den Carotiden iiberlassen (Taf. XXI.
Fig. 377)."

The mandibular gill remains in a rudimentary condition, called in the
Elasmobranch group spiracular gill or pseudobranch ; in the Teleosts, on
the other hand, the choroid gland (Balfour,^^ Miiller ^). It is not always
present in Teleosts, according to the latter authority, but where it is de-
veloped its branches supply the choroid plexus of the eye. The artery of

21 Dolirn, A. Die Entwicklung und Differenzirung der Kiemenbogen der Sela-
chier. Mitt. Zool. Stat, in Neapel., V., 1884.

25 Balfour, F. M. Comparative Embryology, II. p. 261.

218

BULLETIN OF THE

H

O rt c
cc •— ■ -^

o 2 <"

t« rz en
-;:- Cu-O a;

;- — —

O ai rt
t- 1— I b

c*

s+'

■o .2 .2

~ ^ ■" 5

o* r*^ ri —

•- '^ I?

s <i> II .5

E^ a; 3
O

c

^- => ^
— sa

t« S o
P.fcl3

;;°^ ■

o •- -r) •—
K C ;:i t^
OJ t- — <D

a* r-;

.2 s^

"a "
c c

n «

o

=« g

+i +^

5= =

t-

tn -^ C vj

.2 S =* .-

t- C (4

il> cS <u

— — O --CO'

• ='£'2

CO tr ^ • —

^ -ci - c

n .. ■" -

55.2

c«

5 = C

Z'^^ ■£

a c :j

c =s c

ri t. c

^-1 ^

C3
S
u

"3 o
0 s

o

o

-. o
C O V,

- >-

i; ~ o

o o

■S o
3-S

So

c
o

-♦-»

GO

es

C

CI

s

a

H

+

<5

+

o

<5

CO

+

3

+

c

CS

?-]

^

0

Q^

c3

^

;j

0

ti

c

0

ci

rt

,—i

-t3

(M

CO

o

o

C5

(M

cS

•^

0

2S

+

o

lO

o

O

r~

00

a

+

+

+

H

H

H

o

O

O

1-%

u

li'

Oi

cS

ci

eS

OS

[3

"B

1>

0

S

2

0

B

CS

OS

-4-3

a

c

d

a>

9

<D

c^

;-c

^

^

ii>

Stl

aj

0)

p^

•^3

-4-*

^

?i

00

"<*<

lO

CO

09

C/3

CO

WJ

is

!2

"S

[2

CS

0

0

u

0

<J

C

c

c

C

s

CS

C3

eS

cS

f

f-l

;h

fr-i

t-i

^

.=

r-a

^

^

.a

<i

<i

<^

<

<i

,

^

•

^

'5

"rt

.2

.2

.2

f^

f— 3

ZJ

0

0

0

c

0

B

s

c

B

c

CS

c4

ci

cS

(.4

(^

^

^

^

B

r^

ja

(13

1-^

CO

•^

•kJ

r— *

(M

■*

lO

cS

es

cS

CS

eS

0

0

0

-4.J

MUSEUM OF COMPAKATIVE ZOOLOGY. 219

the choroid gland comes from the hyoidean demibranch (Nebenkieme).
The gland usually lies within the bony orbit, and with very few excep-
tions it is present in those species possessing a pseudobranch (i. e. in this
case of course the hyoidean demibranch or its rudiment). The man-
dibular pseudobranch of Elasmobranchs and Ganoids lies behind the
orbital territory, but there are cases in which an evident approach to
the orbit is recognizable. The vessels of the mandibular pseudobranch
consist of an afferent and an efferent artery, as in the perfect branchiae,
but usually they are shifted in position, so as to run more or less par-
allel to the long axis of the body, instead of transverse, as in the normal
condition. The afferent trunk leaves the hyoidean efferent branchial
just before the latter leaves the arch and passes forwards to end in
the rete mirabile of the spiracular gill, while the efifei-ent trunk arising
from the rete passes forwards and inwards across the hind portion of the
orbit into the cranial cavity, where it unites with the dorsal aorta by an
anastomosis with the internal carotid, near the origin of the ophthalmic
artery. The homology of the mandibular artery of Callorhynchus, as
given by Parker, involves a mistaken identity, as we readily perceive by
referring to the author's own works on the Skate and Mustehis antardi-
cus, as well as by reference to figures by Hyrtl (3) and Miiller (6). It
seems to me clear that the vessel designated j^osterior carotid by Parker
is the arteria vertebral is.

It is of course possible that the arterise vertebrales of the Skate are a
pair of musculo-spinal branches of a now vanished dorsal aorta, but from
their prominent connection with the first efferent branchials of the Skate
it is more probable that they are reduced efferent branchials — of the
nmndibular gill 1 The relations of the afferent and efferent vessels to
the spiracular gill in Chlamydoselachus — a few leaflets of which still
persist — I have not worked out satisfactorily as yet.

There are traces of other lateral branches to be found in the cartilage
at either side of the aorta, between the occiput and pituitary promi-
nence. In two sections I saw lateral unpaired vessels passing out from
the median line to fade out in the cartilaginous tissue which appeared
to be the cause of their suppression. Heptabranchias shows similar
vessels. (See Figure 9.) They are so short and indistinct that it is
with difficulty they can be traced without entirely destroying the carti-
laginous floor of the skull in shaving it down thin enough to see them.
The microscopic sections prepared from one of these transsections
showed only a fibrous cord entirely destitute of a cavity. Presumably
then the vessels were functional only during embryonic life. There is

220 BULLETIN OF THE

a bare possibility that these vessels may have been the musculo-spinal
brancheo of the segments taken into the cranium.

To the characters which Garman has selected as of value in properly
placing Chlamydoselachus in the zoological system may be added : —

1. The dorsal aorta persistent throughout the entire length of the
chorda, its precardiac portion of large size to the occipito-atlantal line,
where it is suddenly much reduced to enter the cartilaginous basis
cranii, through which it runs below and nearly parallel with the chorda,
until it reaches the pituitary region, when it rises abruptly and becomes
in part suprachordal, ending in a vascular plexus.

2. The absence of a complete vascular loop surrounding each gill slit,
ending above in two efferent branchial arteries. Chlamydoselachus has
but a single efferent branchial, placed in each instance cephalad of the
corresponding afferent vessel, agreeing in this with the usual type of
structure found in embryos of other Elasmobranchs.

There are several other characters belonging to other portions of the
vascular system, of equal importance with the foregoing, indicative of
simple organization, which we may take up at some subsequent date.

The character supposed by ]\lUller to be diagnostic of the Cyclo-
stomes, namely, that the dorsal aorta was continued beyond the union
of the first pair of persisting eflferent branchial arteries, and that it was
still further connected with the anterior portion of the cephalic circle
(in Myxine), is not alone peculiar to this group of fishes, but is also
found among several Elasmobranchs. It still remains to be seen
whether it is absent in all the bony fishes (including the Ganoids). If
so, it would serve to show that the Cyclostomes and lower Elasmo-
branchs have retained their vascular apparatus in a much more primi-
tive condition than the remaining groups. Miiller did not find any
trace of the precardiac aorta in Sturio, and from his description of the
efferent branchial system it is extremely improbable that it exists in
any form.

P. S. — The substance of this paper was written out in nearly its
present form in the fall and winter of 1887, at which time the dis-
sections were made, but circumstances have delayed the publication
till this date.

August 1, 1889.

MUSEUM OF COMPARATIVE ZOOLOGY. 221

EXPLANATION OF FIGURES.

REFERENCE LETTERS.

a. artery (in Fig. 2, also auricle).

a. c. arcus cephalicus.

a. i. anterior innominate artery.

an. anastomotic branch of first efferent branchial artery.

a. pi. artery coimecting pituitary plexus with hypophysis plexus.

b. a. bulbus arteriosus.
br. brachial vein.

c. cranial aorta (in Fig. 7, the cavity of this vessel),
c. a. conus arteriosus.

c. c. anterior carotid commissure (art. com. ant. D. S.).

cent. vertebral centrum.

ch. chorda dorsalis.

cce. mes. coeliaco-mesenteric artery.

cor. coronary artery (4- liypobranchial trunk).

c. p. art. profunda cerebri.

c. p. a. posterior carotid commissure (art. com. post. D. S.)

c. c. cardinal sinus.

c. sh. chorda slieath.

ct. cartilage of the basis cranii, c', c", c'", three layers of basis cranii.

ct. sh. cartilaginous sheath of chorda and cranial artery.

c. V. cardinal vein.

d. dorsal aorta (posterior to k).

d. m. dura mater.

e. c. external carotid artery.

e. ex. elastica externa of notochord.

g. median groove in ventral surface of basis cranii.

H. Hyoid arch.

h. V. hepatic vein.

ky. hypophysis.

i, c. internal carotid artery.

i. c. f. internal carotid foramen.

i. j. V. inferior jugular vein.

k. cephalic aorta. Kopfaorta, arteria spinalis impar Hyrtl, arteria ver-

tebralis impar INliiller.

kl. calcareous incrustation.

222 BULLETIN OF THE

m. muscle.

me. membraneous wall overarching c near its anterior end.

m. s. arteriae musculo-spinales.

VIS. arteriae musculo-spinales of the head.

n. nasal artery.

n. p. neural process.

0. ophthalmic artery.

p. palatine artery (= maxillary of Parker).

p. c. e. art. post, cerebri ext. D. S.

p. c. I. art. post, cerebri int. D. S.

p. c. s. precaval sinus.

p. pi. pituitary plexus.

pt. pituitary space.

r. c. subdural rete mirabile.

r. c. V. right cardinal vein.

r. p. rete mirabile profunda cereori D. S.

s. cl. subclavian artery.

s.j.v. superior jugular vein.

sp. anastomosing branch to spiracle.

s. V sinus venosus.

tr. tropeic vein = lateral abdominal vein.

tr. b. fibrous trabeculae crossing the channel of c".

tr. c. transverse canal of pituitary region.

tr. p. transverse process.

r. ventricle.

V. a. ventral aorta.

vase. vascular layer.

I.-IX. First to ninth pairs of aortic roots (arches).
1-6. First to sixth pairs efferent branchial vessels.
l'-6'. First to sixth pairs afferent branchial vessels.
l"-5". First to fifth visceral arches.

Fig. 1. A sketch of a dissection of the efferent branchial vessels and the result-
ing aorta of Chlamijdoselachus anguineus, natural size. On the right of the figure
the vessels are sketched in the outline of the roof of the mouth, to the point of
emergence from the tissue surrounding the proximal ends of the gill arches indi-
cated on the left oval outhnes. The left internal carotid artery is not shaded,
and is sketched for a short distance only to show its course in the chiasm, at
which point the cephalic aorta is broken for the sake of clearness. The end of the
cranial aorta, and its branches connecting it with the pituitary plexus, are drawn
somewliat enlarged.

Fig. 2. A semidiagrammatic figure of a portion of the vascular system of the
same fish, showing the relations of the arterial and venous vessels, as seen from
the left side. Approximately natural size. At * the coronary artery is cut off,
nor are many of its branches shown. The venous vessels, heart, and ventral aorta
are left unshaded. The common and internal carotids have been displaced up-
wards, and the anastomotic branch broken for sake of clearness.

MUSEUM OF COMPARATIVE ZOOLOGY. 223

Fig. 3. A sketch of the left half of the hemisected cranium of Chlamydoselachus
to show the relations of the notocliord and cranial aorta to the basis cranii and to
the pituitary prominence and space. Natural size.

Fig. 4. (a) A view of the inner surface of floor of the cerebral cavity in the
pituitary region before the removal of the dura mater and tissue wliich separates
the brain cavity from the pituitary excavation and its contents (pituitary plexus,
transverse canal, and carotid chiasm). Natural size. (6) A dissection of the
ventral surface of the same, to show the chiasm of the internal carotids.

Fig. 5. A series of seven transsections of the basis cranii to show the relations
of the notochord, cranial aorta, and the median ventral groove to each other and
to the cranial floor.

Figs. 5 a and 5 6 from 1 shown in Fig. 1. Figs. 5 c and 5rf from 2; Fig. 5e
from 3; Figs. 5/ and be/ from tlie pituitary prominence. All the figures enlarged.
The series progresses cephalad, and the geometrical outlines refer to portions of
the median line of Fig. 1.

Fig. 6. Part of the section of the notochord figured in 5 a more highly magni-
fied to show the sheaths and external calcified layer. X circa 180 diameters.

Fig. 7. A section of the cranial aorta from Fig. 5 d, more highly magnified.
The fibrous trabeculae cross the cavity of the vessel in all directions. X 180
diameters.

Fig. 8. A transsection of the basis cranii of Chlamydoselachus, near the verte-
bral articulation, to show the figure made by the calcareous sheath (and its pro-
cesses) of the notocliord, resembling a vertebra of the trunk region.

Fig. 9. A transversely cut piece from the basis cranii of Heptabranchias sp.
near the anterior third of the distance between the pituitary prominence and the
occipital region, to show the chorda (and aorta ?) and the blood-vessels enclosed in
the cartilage.

Fig. 10. The efferent branchial system and aorta of Zygcena malleus, after Hyrtl.

Fig. 11. Diagram of the above, with the " cranial aorta" inserted.

Fig. 12. The efferent branchial system of Mustelus antarcticus, after Parker.

Fig. 13. The efferent branchial system of Myxine, after Miiller.

Fig. 14. The cephalic circle of Cephaloptera modified after De Sanctis.

AYKP.S- CKLAMYDOKEi.ACHtS GaR.

,^jjasa^.

It

/

/■

.pi.

i^,.

ja^^"

5?

7

Gi- .A,

«K/

<£:s».

'"ifkB

Hi

roc '^'^^

i^-m

B Meiscl. Uth

No. 6. — Cave Animals from Southwestern Missouri. By Samuel

Gakman.

Though a kuowledge of their inhabitants would appear to be of the
greatest importance in connection with the study of the cave life of
Kentucky, Tennessee, Indiana, and elsewhere, up to tlie present time
the caverns of Missouri have received little or no attention from the
zoologist. The existence of numerous and extensive caves west of the
Mississippi has been well known to geologists for a long time. There is
frequent mention of them in the various Geological Eeports ; but among
the notices only a single one touches on their animal occupants. The
cavernous belt of Missouri is a hundred and fifty miles or more in width,
and extends diagonally quite across the State from northeast to south-
west. On the Mississippi, roughly estimated, it reaches from Clark to
St. Louis County, and at the opposite extremity it stretches at the least
from Vernon to Howell County. The geological positions of the caves
range from the St. Louis limestone of the Lower Carbohiferous to the
third Magnesian limestone of the Lower Silurian. To the northward
the formations lie in a plane that nearly coincides with that of the
horizon. In Clark County the Keokuk limestone is at the surface ; in
St. Louis, it is that known by the name St. Louis ; and between these
points it is mainly the Burlington group that appears at the top. In
the southwest a section across the belt cuts from the Carboniferous
to the Silurian, as if toward a centre of upheaval in the southeastern
portion of the State.

Caves have been reported from some twenty different counties, and
in a number of instances particular ones have been described at length.
Among the better known are those of Ralls, Boone, Phelps, Greene,
Christian, Ozark, and McDonald. Whatever the causes, vhether dif-
ferences in the strata, the inclinations, the amount of fall iri the water-
courses, or in the water itself, the caves appear to become more extensive
and more numerous toward the southwestern portion of the State.
Fisher's Cave, in Ealls County, has an opening of ninety feet in width
by twenty in height, and more than four hundred feet from the entrance

VOL. XVII. — NO. 6. 15

226 BULLETIN OF THE

it connects with the surface by means of a sink-hole. The statement is
made that Conner's Cave, in Uoone County, has been explored for a dis-
tance of eight miles. Friede's Cave, in Phelps County, according to
report, has been traced for a number of miles. There are several large
caves in Ozark County, in the third Magnesian limestone. The sink-
holes, \fith. which so many of the caverns are connected, prove the
manner of forming to have been the same as that giving rise to the
Mammoth and other caves of Kentucky ; the I'ock, dissolved and disin-
tegrated, has been gradually removed by the water from the sink-holes.
There seems to be no I'eason to suppose the history of the majority of
these caves goes further back than tiiat of the later Tertiary deposits,
if so far. Such a small amount of divergence as exists between the
species peculiar to the caves and their allies outside is proof that the
former have entered their subterranean dwellings at a comparatively
recent period.

In one of his Reports, State Geologist Broadhead remarks that in
Christian County there is a stream that disappears in a sink to come
out again three quarters of a mile away by an opening ninety-eight feet
wide by sixty feet high, from which "a very clear, cool stream passes
out, in which by careful search crawfish without eyes can be found."
This is the only notice our search has revealed of the animals inhabiting
these caves.

An opportunity of adding something to our knowledge was recently
afforded by the kindness of Miss Ruth Hoppin, of Jasper County. My
attention was first directed to the matter by a note from her, accom-
panied by a specimen of 7''yphl.ichthys subterraneus, Gir., that had been
taken from a well. She said that similar fishes had been taken from
other wells in the neighborhood, and that the owners of the wells spoke
of subterranean streams flowing through. Experience elsewhere satisfied
me that there should be caves in the vicinity from which these streams
escaped, and at once my correspondent was asked if she would kindly
engage some one to explore any caves there might be near by, and also,
if possible, to get more specimens from the wells. She took up the
matter, engaged help, and, at great personal risk and inconvenience,
herself made explorations of a number of the caves, which, as was sus-
pected, were not at all rare in the district, the southern part of Jasper
County. Numerous specimens of Batrachians, Fishes, Crustaceans,
Moliusks, and Insects were collected and forwarded from time to time,
among them several new to science. Abstracts from the letters accom-
panying them give a very fair idea of their surroundings.

MUSEUM OF COMPARATIVE ZOOLOGY. 227

In one letter Miss Hoppin says : —

" I took a boy with me and weut to Wilson's Cave. The catch was not
large, but I was mucli encouraged by what I learned while there. The cave is
about iifty feet long, nearly as wide, oven-shaped, and high enough to stand
erect except around the sides. The farmer had enlarged the entrance to use
the place as a creamery. A small very clear stream flowed along the left side,
having a width of two feet and a depth of three, with a temperature of -f-54° F.
About ten feet from the entrance the light struck the stream in such a manner
that we could see everything in the water without a lantern. The first things
that caught the eye were a lot of white crayfish, a dozen in all, like those I
took from the wells. It seemed as if I might take every one of them. But
though blind, they have one or more of the other senses very keenly developed.
I am very sure they, as well as the white fi.she?, have the tactile sense developed
in an unusual degree. At the least touch upon the water they dart away. As
the net cautiously follows, they escape adroitly, making no blunders as to the
direction of the appioaching enemy, and hide in crevices of the jutting rocks or
in the muddy bottom of the stream. The nmd was easily stirred so that noth-
ing could be seen. These creatures, fish and crayfish, are only to be secured
by patient waiting and skilful management. The people at the cave say the
fish never bite, and cannot be taken with hook and line. The crayfish were
all found near the entrance, where there is considerable light. I'ollowing the
stream back to a dark recess, reached by crawling on the slippery rocks, the
lisht of the laiitern revealed a school of little white fishes, such as I secured
from the wells. All were very small. I saw half a dozen or more, but
secured only one. I concluded the crayfish liked the light. Perhaps they
remain near the entrance because thej'' find there a supply of food. We found
a few snails floating about, but saw none in the dark pool where the fish were.
Miss Wilson, who was with me, thinks the crayfish devour the others. She
has never seen them together, and says the latter keep away from the former,
though she had not noticed the crayfish catching or eating them. There was
nothing to prevent the crayfish ascending the stream to where the others

WtPd.

" An insect, a ' water spider,' common outside, is found inside near the light.
I did not find it back where the little fishes stayed. By crawling back under
the rocks one could see where the stream issues from the crevice. The passage
is too low and too small to be followed, as the water occupies almost the whole of
the opening. Beyond it, I think, one would find the home of the fish. The low
opening is arched over by solid limestone, and could hardly be enlarged arti-
ficially, as the main entrance has been. Several feet to the right of the stream,
having no visible connection with it, is ' the lake,' about fifteen feet in
diameter. It is now a muddy bed. In no place could I see an inch of clear
water. Just at the centre it has most water. When the water is high, the
lake is full of fishes. What becomes of them when the water is low ? I am
ashamed that I did not look into that mud a little more carefully. There

228 BULLETIN OF THE

must be some connection with the creek, either directly or hack through the
rocks. The lake is the place to catch the fish in high water. It is accessible,
if one does not count the ' gumbo,' which makes an almost impassable sticky
covering over the entire floor of the cave. This ' gumbo ' forms the banks of the
stream. It is difficult to keep one's footing there. With all our care it was
constantly falling into the stream, roiling the water and scaring the crayfish.
There was a small, dark-colored salamander near them. I brought home only
one small lisli and four small crayfish, the largest about two inches long. But
I am much encouraged, for I feel sure that in time we can get all you want,
and I realize there is ever so much that I can learn."

Samples of the "gumbo," or red mud, of the floors of the caves
were sent with the collectiou. Experiments with it proved its excessive
fineness and stickiness. Stirred about in the water at a depth only
of two and a half inches, it was more than three hours before it had
settled so that objects could be distinguished on the bottom. Twenty-
four hours later, a cloudy substance, an inch and a half in depth, seemed
to hang over the bottom, and it was more than two days before it had
completely settled.

Miss Hoppiu's first work was done in August ; in September she
made further eftbrts. In answer to questions, she states that the wells
from which specimens have been taken are about half a mile from
Centre Creek, the water level in wells and creek being nearly the
same. The wells were nine or ten in number, from five to eighty rods
apart, from eleven to thirty feet in depth, deeper in the higher ground,
and having a depth of water varying from two to four feet. In some
wells the rock at the bottom had been excavated. The water is what
is commonly called hard, i. e. impregnated with lime. After rains, some
of the wells have softer water than others, and the water stands higher
in these wells, indicating closer connection with surface drainage. All
the wells soon regain the common level. They become low in times
of drouth, but never dry out entirely, as is the case with a cave spring
near by, about twelve feet above the level of the creek. The tempera-
tures taken in the wells at low water ranged from -|-52° to 54° Fah-
renheit. During a storm, in the well having the highest water the tem-
perature rose to -f-S""- When the mercury stood at 90° to 95° in the
shade outside, the temperature was only 54° in Wilson's cave.

" After several days of a cold wave, the night temperature of the outer air
being 45° and the noon reading at 60° to 70° in the shade, I found the tem-
perature of the water in the cave had gone down about two degrees. . . .
The level of this cave is ten or more feet above the creek, and it is not affected

MUSEUM OF COMPARATIVE ZOOLOGY. 229

by the rains until several days after the creek has begun its rise. This state-
ment is from the people at the cave, and was not verified by me.

" From one well thirteen blind crayfish were taken by means of a net
formed of mosquito bar spread on the bottom. The specimens became en-
tangled in it. From the same well a few snails were taken. The owner re-
ports, that at various times, a year or more ago, the surface of the water would
be covered with ' little white lice or something of the kind.' Minnows were
put in, after which the lice disappeared. A blind fish was also put into this
well, but nothing had been seen again either of it or the minnows. From
each of the other wells one crayfish was taken; from one of the three, a few
minute centipedes. Earlier in the season these centipedes had been so numer-
ous as to render the water unfit for use ; they seemed to be inhabitants of
the well. Some minnows were put in, and the centipedes vanished. I heard
of them too late to make a satisfactory collection. From the Armstrong well
two small blind fish were taken, and one from the Adams well. From the
latter some snails were secured, also some large centipedes, these latter under
circumsjLances indicative of accidental presence. Repoits come from the coun-
try for miles around where fish and crayfish are taken. One well, an Arte-
sian, went dry when a neighbor dug another farther down the hill. It was
then found that the first well opened at the side directly into a small cave.
All of these wells are in limestone ; only in this formation is good water to
be obtained hereabout. The larger caves in this vicinity are under the lime-
stone cliflfs and hills that skirt Centre Creek. The wells are usually walled
with stones that leave spaces, through which the fishes may pass. There are
probably many small subterranean springs and streams, not one large under-
lying lake or stream, as popular belief has it.

" Day's Cave, from which a small collection is sent, opens under the cliffs.
After much digging the mouth was enlarged so that a small boy squeezed
through. Wilson's Cave is not large ; it is spanned by one limestone, and
floored w-ith the sticky 'gumbo.' This mud is utterly without grit. It
forms a crumbling bank on the approachable side of the stream, and the
minute particles are seen through the very clear water to be suspended in a
thin swaying cloud at the bottom of the water. This mud-cloud is so light
as to form no obstacle to the movement of the creatures which find it a ready
hiding-place. It renders a study of the animals at the bottom very difficult,
the water is so easily roiled. It required great care to catch the specimens ;
the stirring of the water frightened them away to their hiding places on the
bottom, or among the dark nooks and crevices of the jutting rocks of the. op-
posite bank, their movements at the same time stirring up the mud so that
nothing could be seen. In the farthest corner of the cave, where the water
comes in, would seem to be the home of the fishes ; here they were most
numerous and most active. When the water is low, they are found only
here, though the stream below is equally cold and deep. Apparently, they
avoid the light.

230 BULLETIN OF THE

" On my first visit, the water being low, no crayfish were seen in the dark
nook, the place favored by the fish. After the storm which had flooded the
caves, a few were found there. Though I watched for some time, I never saw
them pursue the fishes, as they might easily have done, guided by the stir in
the water. Both creatures are very sensitive to the slightest ripple. During
high water, a pool, ' the lake,' is formed, a little way from the stream in an-
other dark part of this cave. In low water the pool is cut off from the creek.
I found both species in it, the fish in the darkest part, and .saw no signs of
enmity. Most of the crayfish were found in the lower part of the stream, in
the twilight; the fishes could not be found without the lantern. At the time
of the floods, the cave is full, and the water rushes out furiously. . . . An-
other proof that the crayfish are more fond of the light is seen in the shallower
wells. That from which most were taken was more exposed to the sun. At
noon, when the light was more favorable, we could see them swimming about.
No fishes have been taken from this well. They were taken in the narrower
more shaded wells, of which the deep ones on the hills report fishes only.

" As to the food of the fishes, I discovered nothing. The mud where they
were was not so deep as farther down. An examination of it the length of
the cave brought to light many snails ; the shells of the living ones are whiter
and more nearly transparent than the floating dead ones. The largest crayfish
are of a dirty rusty color, and very bristly, in caves and in wells. One large
one is very soft and very white; no doubt it is newly moulted.

" Both fish and crayfish were less numerous after the freshet, and apparently
less active. The disturbance of the flood may have caused them to retreat
into their hiding places, only the weaker being left behind, or some may
have been swept away by the torrent. The sensitive creatures would soon
die in the light and heat outside, where the water is full of frogs and eyed-
crayfishes. . . . The specimens become opaque when they are put into alco-
hol ; they are almost transparent when alive, so much so that the action of
their internal organs can be observed. Repeated tests assured me the animals
were blind, though very sensitive to the sunlight. They died soon after catch-
ing, even in water frequently changed.

" The insects of the collection were taken in the lower part of the stream,
near the mouth of the cave. They are similar to, if not identical with, others
found in all the spring streams of this vicinity. They are very lively on the
surface of the water, constantly rippling it. I think the crayfish eat them, but
have no positive proof. On my first visit, insects and crayfish were very nu-
merous at this place. The latter were darting up towards the former. We
thought we could detect a faint odor about the insects (water spiders) that
might help to guide their enemies, but the vibration of the water would be
sufficient.

" Two aquatic and two terrestrial salamanders taken in this cave are in the
collection ; they are not peculiar to the cave. Some nearly a foot long are in
the creek outside."

MUSEUM OF COMPARATIVE ZOOLOGY. 231

Some time near the middle of October Miss Hoppin visited "Wil-
son's cave again, after some cold weather ; the water was four degrees
colder, and no fishes were to be seen. A couple of weeks later, after a
week of warm days, the water had taken on its summer temperature of
54°, but there was nothing to collect. The water at this time was so
low that the connection with the water of the inner cavern was broken,
the water in the stream being below the fissure from which it poured
earlier in the season. It appeared as if the fishes could not get out
into the cave till the water rose again. The opening into the inner
cavern would admit a small dog. There were no insects on the walls.
Something was heard that was supposed to be a bat, but it could not
be found. The neighbors said that after the floods white crayfishes
were frequently found out of the cave in the creek.

Various caves were visited, at a considerable distance apart ; the collec-
tions were in the main made from the wells and the two caves mentioned.
It is evident from the notes that the caves are numerous, and similar to
those in the same formations in other States. It is also evident, from
what is found in the stomachs of the fishes, that there is more to be
done in the way of collecting. A few fossils from the walls were sent.
Though not peculiar to it, all are forms common in the Keokuk lime-
stones, which lie at or near the surface in this district, known as the
lead region of Southwestern Missouri. The greatest altitude is rather
more than eleven hundred feet above the sea level. At the point under
consideration the drainage goes to the westward. The waters after-
ward go south in the Grand River, then southeast in the Arkansas, and
reach the Mississippi a little below 34° north latitude, two hundred and
fifty miles or more below the mouth of the Ohio. Directly eastward,
a considerable distance, the water is carried toward the mouth of the
Arkansas, near which it meets the Mississippi. Northeastward it is
less than twenty miles to points from which the drainage is carried
through the Osage River to the Missouri, the mouth of which is about
a hundred and fifty miles above that of the Ohio. Whether approached
by the way of the Arkansas or by that of the Missouri, the caves of
Jasper County and the neighboring counties are pretty effectually iso-
lated from the caves east of the Mississippi, — a fact not to be lost
sight of in discussing the distribution of the animals.

The collections contain a large number of specimens pertaining to a
rather small number of species. Of these the fishes and the Crustacea
claim most of our attention, being the only ones we can with safety call
peculiar to the caverns. As their testimony concerning the acquisition

232 BULLETIJ^ OF THE

of the species by the caves appears to differ somewhat, it will be well to
consider them separately. The following is the entire list of species : —

Geotriton longicauda, from Wilson's Cave.

Plethodon sp., larvae, " "

Cambarus virilis Hagen, from the wells, "Wilson's Cave, and streams.

C. sctosus n. sp., from the wells, WiLsou's Cave.

Asellus Hoppiiice n. sp., from Day's Cave, ua mud under stones.

Physa heterostropha Say, from caves and wells.

Scolopocryptops sexspinosa, from wells.

Plathemis trimaculata DeGeer, from mouths of caves.

Hygrotrechus remigis Say, from near mouth of Wilson's Cave.

Dineutes assimilis Aube, " " "

Agabus sp., from Day's Cave, under rocks,

Ceuthophilus Sloanii Pack, from the water in Wilson's Cave.

FISHES.

Typhlichthys subterraneus Girard, the only blind fish in the
collection, is represented by a large number of examples, the majority of
them taken from the wells, the balance from the caves, with the exception
of a single one from the creek outside. Compared with specimens from
Kentucky and Tennessee, they agree so exactly as to raise the question
whether the species was not originated in one of the localities and thence
distributed to the others. The opinion generally held is, that the cave
species of Indiana, Kentucky, and Tennessee originated in their respect-
ive localities. It is no doubt true for some of them. The idea is well
supported by the insects and Crustacea, the species in one section being
unlike those of the others. It may be urged that the respect in which
the fishes difi"er from them is more apparent than real, since these Crus-
tacea and insects were derived from a number of distinct species, while in
all probability the same species of fish entered the caves in each dis-
trict, and, being under the same influences in each, suffered the same modi-
fication in each. Reduced to its lowest terms the question, so far as the
fishes are concerned, is this : Were the blind fishes distributed to the
scattered localities where now found before or after they became blind 1
In favor of independent origins at distant points, it can be said that a
species, distributed over the valley, possessed of habits such as would
lead it to place itself under the modifying conditions of the cave in one
place, would be most likely led to do so in the others. On the other
hand, we have the more hesitation in accepting the conclusion that one
and the same species originated independently in two or more different

MUSEUM OF COMPARATIVE ZOOLOGY. 233

localities, from knowing that exact parallels in the development of
animals in nature, if they exist, are excessively rare. If our caution
prevents ready acceptance of two apparently exact evolutionary parallels
as really coincident, we become much more sceptical when the number
of parallels or coinciding lines is increased. There is no doubt that the
representatives of Typhlichthys subterraneus in the various caves were
derived from a single common ancestral species. The doubts concern
only the probability of the existence of three or more lines of develop-
ment, in as many different locations, starting from the same species and
leading to such practical identity of result. Suc|i . identical results
would demand substantially similar modifying elements, — darkness,
temperature, food, enemies, etc., ^i— and the same length of time subjected
to their influence. The likelihood of the existence of so many like ele-
ments in distant regions is inversely to the number demanded, though
one cannot say it is impossible. To accept the conclusion favoring inde-
pendent developments of the same species would involve acceptance of
the idea that the caves in each of the districts had been occupied for
about the same period of time. This, of course, would not furnish us
with any clue to the time of formation of the caves.

As an alternative, the opinion is here advanced that these blind fishes
originated in a particular locality, and have been, and are being, dis-
tributed among the caves throughout the valley. We are in the habit
of looking upon great rivers like the Ohio or Mississippi as impassable
obstacles to passage from 'cave to cave, rather than as thoroughfares.
In this we have certainly assumed too much. Various instances are on
record of the discovery of blind fishes that have strayed into the
open streams from their caverns. If there were means of determin-
ing the frequency of the occurrence of such instances, it would un-
doubtedly much exceed what we are now inclined to credit. Persons
acquainted with the streams of the Mississippi basin will agree that
their undermined banks provide series of recesses or caverns, extend-
ing from the rills at the sources of the tributaries to the Gulf. The
currents do not prove insurmountable to multitudes of fishes, no better
provided with locomotive organs than the blind fishes, passing up the
streams every season. Swept from the caves by the torrents in the
flooded mouths, the blind species would find itself protected at once
from light or enemies by the turbid waters. The temperature of the
water at such times is low, and, should the light penetrate so as to prove
detrimental, retreats exist on every hand in the excavations of the banks
or the mud of the bottom. What migrations these fishes mav make in

234 BULLETIN OF THE

winter we can only imagine. Hiding places are so numerous and exten-
sive as to suggest the possibility of the evolution of blind forms without
the caves. The great essential would be the disposition to avoid the
light, opportunities existing everywhere ; the surroundings then would
bring the organization into harmony with their demands, sooner or later
as the creature was more or less plastic and yielding ; disuse of the sense
of sight being followed by its loss and atrophy of its special organ.
Development of sightless forms in the holes and burrows of the banks,
or in the mud of the bottom of the river, would here follow a similar
course to that gone through at great depths in lake or ocean.

Crooked streams are not so impassable as one might suppose, even
to floating objects, insects, mollusks, etc. A twig or leaf dropped into
the current on the inside of the upper arm of a horseshoe curve in
a stream is carried near to the opposite shore before it leaves the bend,
and, especially if favored by the wind, often is carried completely across.
The passage is much easier to animals that swim, however feebly.
Taking everything into the account, it does not appear to be at all
necessary to credit Ti/phlichthys suhterraneus from Kentucky, Tennessee,
and Missouri with more than a single point of origin. The same may
be said of Amblyopsis spelcBUS of Kentucky and Indiana, and of the blind
crayfish of the same States.

In an article entitled " Life in the Wyandot Cave," Ann. Mag. Nat.
Hist., Ser. 4, VIIL, 1871, p. 368, Professor Cope makes this statement
concerning Amblyopsis : " If these Amblyopses be not alarmed, they come
to the surface to feed, and swim in full sight, like white aquatic ghosts.
They are then easily taken by the hand or net, if perfect silence is pre-
served ; for they are unconscious of the presence of an enemy, except
through the medium of hearing. This sense, however, is evidently very
acute ; for at any noise they turn suddenly downward and hide beneath
stones, etc. on the bottom." The statement is repeated in Amer. Xat.,
1872, p. 409. Such a development of this sense, in recesses where we
are accustomed to think any sounds other than those made by the rip-
pling or dripping water are almost unknown, is not what one would have
expected. Having this in mind, I wrote to Miss Hoppin asking her to
make experiments on Typhlichthys, and to determine what she could in
regard to hearing, feeding habits, etc. The quotations here given are
from her replies.

" For about two weeks I have been watching a fish taken from a well. I
gave him considerable water, changed once a day, and kept in an uninhabited
place subject to as few changes of temperature as possible. He seems perfectly

iroSEDM OF COMPAKATIVE ZOOLOGY. 235

healthy, and as lively as when first taken from the well. If not capable of
long fasts, he must Uve on small organisms my eye cannot discern. He is
hardly ever still, but moves around the sides of the vessel constantly, down
and up, as if needing the air. He never swims through the body of the water
away from the sides, unless disturbed. Passing the finger over the sides of the
vessel under water, I find it slippery. I am careful not to disturb this slimy
coating when the water is changed. . . . Numerous tests convince me that it
is through the sense of touch, and not through hearing, that the fish is dis-
turbed : I may scream, or strike metal bodies together over him as near as
possible, yet he seems to take no notice whatever. If I strike the vessel so
that the water is set in motion, he darts away from that side through the mass
of the water, instead of around, in his usual way. If I stir the water, or touch
the fish, no matter how lightly, his actions are the same."

From the stomach of one specimen the remains of an Asellus were
taken; from that of another, a young Camharus ; from a third, frag-
ments of an insect resembling Ceuthophilus ; and from others, portions
of a crustacean, of which we have several specimens from Day's Cave,
with well developed eyes, resembling Crangonyx, and from appearance
the main food dependence.

The total length of the largest fish is two inches and a quarter. The
eggs in the ovaries, August to September, are large, but with uo traces
of embryos.

CRUSTACEA.

In part, at least, the problem of the origin of the cave Crustacea is
simplified by the fact that they are so distinct in various caves as to
leave no doubt that they are descended from ancestors already of dif-
ferent species at the time of entering the subterranean habitations. The
blind crayfish of the Missouri caves is very distinct from any previously
known ; it is described below under the name Cambarus setosus. The
common species of the neighborhood, C. virilis, is also found to enter
the underground retreats, but it is not, of the outside forms, the nearest
ally of the blind form. The latter bears so much affinity to C. Bartonii
as to suggest derivation from it. A somewhat parallel condition exists in
the caves of Missouri and those of Kentucky. In these last, with the blind
C. pellucidus we find C. Bartonii, the nearest ally of the blind crayfish
of Missouri, C. setosus ; and with the latter again, in the Missouri caves,
is found an eyed species, C. virilis, more nearly allied to the blind one
in the Mammoth Cave. The relationship existing between the species
C. setosus and C. Bartonii is much closer than that between C. pellucidus
and C virilis. A distribution of C. Bartonii covering so large a portion of

236 BULLETm OF THE

the Upper Mississippi valley to some extent favors the idea of a deriva-
tion from it of C. setosus. The gi'eater differences between C. pellacidus
and all the known eyed species point toward a longer subjection of that
form to the spekcan influences. For comparison we give diagrams of
details of structure, antennal lamina, epistoma, and the two forms of
the anterior pairs of abdominal appendages of several species. These
are taken from the specimens and from the drawings. The degrees of
affinity are well indicated by the shapes of the first pairs of abdominal
legs. The slighter approach of C. pellucidus toward C. virilis is shown
by Figures 12 to 14 of the former, as compared with Figures 8 to 10 of
the latter ; and the nearness of C. setosus to C. Bartonii is apparent in
Figures 1 and 2 of the first, and 4 and 5 of the second. Figures 11
and 15 represent C. haviulatus, from the Tennessee caves, a form which
stands between C. setosus and C. pellucidus, nearer the former. Distribut-
ing the mentioned species into the gi'oups arranged by Professor Hagen,
■we shall have the aberrant form C. pellucidus in the first group, nearest
to the second, in which C. virilis belongs ; wliile C. Bartonii, C. setosus,
and C. hamulatus fall into the third group. Such close affinities as exist
between C. Bartonii and C. setosris do not permit their separation into
different genera, and the retention of the latter in the genus Camharus
cannot but be followed by the disestablishment of the genus Orconectes
and the return to the older genus of the two species heretofore included
in the later. Very young specimens of C. setosus correspond better with
the adults of C. Bartonii ; their eyes are more prominent in these stages,
and appear to lack but the pigment ; the rostrum also is less acuminate,
and its blunt lateral angles are present. The gonopods of the very
small ones agree with those of form ii. of C. Bartonii, the adult shapes
approaching those of form i. According to Miss Hoppin, the young of
C. setosus when alive are not so white as the older ones.

"At first, I attributed it to greater trausparency, but now I am sure the
color is in the shell, not that the internal organs can be seen because of the
transparent shell. They are not so dark, however, as the brook species
[C virilis'] of the same size."

In similarity to the case of Amhlyopsis, the presence of the same
species of blind crayfish in the caves of Kentucky and those of Indiana
is an indication of distribution from a single point of origin.

The Crustacea were placed in the hands of Professor W. Faxon for
identification. He has kindly furnished the descriptions of the new
species, which are given as they come from his pen. I have added on

MUSEUM OF COMPARATIVE ZOOLOGY. 237

Plate 11. a hasty sketch of au adult female of Cambarus setosus, one
half larger than natural size, and another of a specimen of Asellus Hop-
pince, three times the size of the specimen. On Plate II. Fig. 1, the
outer two joints of each leg of the hinder two pairs are bent under,
so that they appear one third shorter. The remainder of the collection,
the insects, moUusks, and the like, was examined by Professor H. Garman,
to whom I am indebted for identifications and notes quoted below.

Cambarus setosus Faxon.

" Carapace granulate on the sides, with scattered hair-like setse ; cervical
groove sinuate ; a small lateral spine just behind the cervical groove ; rostrum
of moderate length, triangular, excavated, lateral margins convex, no lateral
teeth (except in smaller specimens, which have a small, acute tooth on each
side near the tip); post-orbital ridges slightly developed, without anterior
spines; region behind the cervical groove relatively long; areola very narrow,
punctate. Abdomen about the same length as the carapace, with scattered
hairs; telson bispinose (occasionally trispinose) on each side. Anterior pro-
cess of the epistoma broadly triangular, margins more or less notched or
dentate. Eye-stalks and eyes rudimentary. Basal segments of antennules
furnished with a sharp spine near the distal end. Antennae longer than the
body; scale very broad at th& distal end, external border slightly convex,
ending in a short, sharp spine. Third pair of maxillipeds bearded within.
Chelipeds of moderate length; chelae long, very hairy, toothed on the inner
margin, granulate on the outer margin; fingers long, compressed, costate ;
carpus toothed on the inner face, granulate on the outer side; upper surface
of merqs granulate, lower surface with two rows of sharp spines. Third pair
of legs of the male hooked. First pair of abdominal appendages terminating
in two recurved hooks (similar to those of G. Bartonii). AnniUus ventralis
of the female prominent, with a deep central fossa.

" Length of one of the largest specimens, 2^ inches ; carapace, 1^ in.; from
tip of rostrum to cervical groove, W in.; chela, l^g in.; fingers, ^\ in."

From the wells come also two very small specimens with well devel-
oped eyes, probably C. virilis Hagen. They are too young to determine
with certainty.

Asellus Hoppinae Faxon.

"Anterior margin of head with a median concavitv, from the bottom of
which projects a rostral tooth; external angles rounded; the head widens
posteriorly, so that the hind margin is nearly as broad as the anterior margin
of the first thoracic segment; eyes of moderate size, oval. Thoracic segments
subquadrate, lateral margins convex, giving to the body with the head and

238 BULLETIN OF THE

abdomen an even, long oval outline. Abdomen suborbicular, slightly exca-
vated on the margin at the base of the caudal stylets. Basal segment of an-
tennule subspherical, second segment cylindrical, forming with the fir.-;t a well
marked peduncle; flagellum composed of six or seven segments; the tip of the
antennule does not reach the distal end of the penultimate segment of the
antennal peduncle. Peduncle of antenna composed of three short, followed by
two long segments; flagellum long, reaching, when bent backward, as far as
to the abdomen. Mandible furnished with a tri-articulate palpus. First pair
of thoracic appendages of male provided with a thick claw; on the palmary
border are two long teeth and a small blunted tubercle ; dactylus armed with
a blunt tooth or tubercle near the middle. Caudal stylets with two subcylin-
drical branches, the inner of which is somewhat longer than the outer. Color,
slaty brown mottled with lighter yellowish spots.

"Length, without caudal stylets, | inch ; breadth, ^\ inch."

INSECTS, etc.

The following are Professor Garman's notes on the invertebrates sent
him for examination.

" The invertebrates sent me for identification pertain to common species,
for the most part aquatic, such as one would expect to find at the mouths of
caves from which emerge streams of water. With the exception of a myria-
pod and a small grub, all have well developed eyes. One or two may be
classed as shade-lovers, since in ordinary situations they commonly affect re-
treats from which direct sunlight is excluded. The myriapud is totally blind,
but is not, so far as I know, an inhabitant of caves. It is one of a number
of widely distributed species, which spend much of their time in moist earth.
The absence of eyes in a dipterous larva of the lot has also no necessary rela-
tion to a life in caves, since larvae probably identical with it as to species are
frequently taken^among rubbish in open ditches and rivulets. The value of the
collection is therefore to be looked for in the direction of its remote bearing on
the problem of the origin of cave life, — a problem which needs for its com-
plete elucidation all details obtainable which may by any possibility throw
light upon this subject.

"The single mollusk of the collection, represented by many specimens from
Wilson's cave, is Physa heterosfropha Say, a species which is extremely common
in the weedy shallows' of ponds, lakes, and streams of the Middle States. It
does not Ordinarily avoid light more than other small fresh-water snails, and
had perhaps penetrated the cave in following up its food supply. The ex-
amples are quite typical of the species.

" A myriapod, Scoloporryptops sexspinosa Say, is represented by one specimen
marked ' From Wells.' Though lacking ocelli, it can hardly be considered a
cave species, inasmuch as it is found everywhere throughout the eastern

MUSEUM OF COMPARATIVE ZOOLOGY. 239

United States under wood and stones. Its occurrence in wells is of course
accidental.

" Three dragon-flies, two males and one female, taken at the mouth of Wil-
son's Cave, represent the Plathemis tnmaculata De Geer, a swift-flying, light-
loving insect which is common about fresh water in most parts of the^'unked
States.

" Seven examples of Hijgrotrechus remigis Say were collected in Wilson's
Cave, probably at no great distance from the entrance. These bugs prefer
shaded waters, and are commonly seen on the surface of pools under bridges
and culverts. Their eyes are relatively large, and they probably do not vol-
untarily visit regions entirely destitute of light.

^ " From the mouth of Wilson's Cave are four examples of the common whir
Iigig hpetle, Dineutes assimills Aube, difl'ering in no respect from examples col-
lected in other localities on open water.

"A second beetle, also aquatic, is represented by one specimen labelled
' Day's Cave, under rocks and stones in the mud.' It is a fine black Agabus,
probably A. suturaUs Crotch, but without authentic examples of this species
for comparison it is hardly safe to make this determination final. From the
Califoruian A. lugeus Le Conte, to which it bears a close general resemblance,
it seems to diff^er chiefly in having the sides of the prothorax a little rounded,'
and in having the basal margin sinuate.

"The 'cricket' seems to be Ceuthophilus Sloanii Pack., of which its discov-
erer says in a recent paper : ' The species is at once kno^vn by the conspicuous
pale dorsal band which extends from between the eyes to the fourth seg-
ment behind, dilating slightly on the front edge of segments 2 to 4 ; the brown
portion has scattered pale dots on each side of the line,' etc. The specimens
are labelled ' From the water in Wilson's Cave.'

" The remaining specimen is a fleshy, wrinkled dipterous larva, 7 mm. long
and 3 mm. in diameter, which was taken from a well."

November 16, 1889.

240 BULLETIN OF THE MUSEUM OF COMl'AKATIVE ZOOLOGY.

LIST OF DIAGRAMS.

PLATE I.

Fig. 1-3, 7. Camharus setosus Fax.

Fig. 4-6. C. Bartonil Fabr. ; Gir.

Fig. 8-10. C. virilis Hagen.

Fig. 12-14. C. pellucidus Tellk. ; Gir.

Fig. 11,15. C. hamulatus Cope; Fax.

PLATE II.

Fig. 1. Camharus scfosuK, 1| times nat., ?
Fig. 2. Asellus Hoppince, 3 times nat.

GARMAN, CAVE ANIMALS,

Plate

I.

7.

o,

lo.

12. 13.

Photo Litti of L S Pundeiso" &. Son New Haven Conn

GARMAN. CAVE ANIMALS.

PLATE II

Di,«.- I ;ik „( 1 <; Pi.nHoitnn & Son New Haven, Conn.

Harvard MCZ Librar

3 2044 066 302 928

Date Due

^n