• Mi:'/,, Ji •

fill

ipil'

wax.. ■>'■:' -i-:''^'''

:m^'

THE CORAL SIDERASTREA RADIANS AND
ITS POSTLARVAL DEVELOPMENT.

BY

J. E. DUERDEN.

Washington, U. S. A.

Published by the Carnegie Institution,

December, 1904.

CARNF.niE Institution of Washington,
Publication No. 20.

53f ^

Press of Gibson Bros.,
Washington, D. C.

CONTENTS.

Preface ^

Introduction ■

Adult Colony.

External Characters 7

Column wall 8

Tentacles 9

Disc and mouth 14

Color 15

Reproduction.. 16

External characters on decalcification iS

Anatomy and Histology 21

Column wall and disc 21

Tentacles 23

Stomodaium 23

Mesenteries 24

Mesenterial filaments 29

Skeletotrophic tissues 30

Septal invaginations, interseptal loculi,

and gastro-ccelomic cavity 34

Gonads 37

Corallum 38

Histology 40

Wall or theca 45

Septa 46

Synapticula 52

Columella 53

Dissepiments 55

Epitheca and basal plate 56

POSTLARVAL DEVELOPMENT.

Larva 57

Young polyp 6t

Tentacles 65

First cycle of exotentacles 65

First c^cle of entotentacles 66

Secondary exotentacles 69

Second cycle of entotentacles and third

cycle of exotentacles 71

Third cycle of entotentacles and fourth

cycle of exotentacles 73

Mesenteries 76

First cycle of mesenteries (protocnemes) ... 76

Second cycle of mesenteries (metacnemes) 79

Third cycle of mesenteries S3

Corallum 86

First cycle of entosepta and second cycle

of exosepta 86

Second cycle of entosepta and third cycle

of exosepta 93

Third cycle of entosepta and fourth cycle

of exosepta 99

Basal plate 112

Epitheca 115

Columella 117

Anatomy and histology of larva and young

polyp "9

Young polyps 122

References 125

Explanation of plates 126

Hi

PREFACE.

The researches of the late Prof. H. de Lacaze-Duthiers (1873, 1897),
Prof. G. von Koch (1882, 1897), and Prof. H. V. Wilson (1888) have made
us acquainted with many of the early stages in the development of corals.
They have served to establish such fundamental facts as the ectodermal
origin of the madreporarian skeleton and the sequence of the primary mesen-
teries and septa, results which must ever possess an importance to the student
of the Anthozoa. But for an understanding of many of the problems of adult
coral morpholog}', especially those associated with the relationships of the
mesenteries and septa, it has long been desirable that developmental stages
later than those studied by the authors mentioned should be investigated.
While resident in the West Indies I have followed day by day the postlarval
growth of the coral Siderastrea radians (Pallas) for a third of a year, and
secured the development of the tentacles and septa as far as the third cycle,
and that of the mesenteries to the completion of the second cycle. The
results are herein set forth.

In mau}^ respects the mature polyps of S. radians are of peculiar mor-
phological interest, but have never been fully described. An account is
therefore first given of the external characters and internal anatomy of the
adult colon}', and afterwards of the development of the young polyp from
the free-swimming larva. The manner of appearance and the relationships
of the tentacles, mesenteries, and septa are considered at some length, their
establishment being the principal object of the investigation.

The work was commenced while Curator of the Museum of the Institute
of Jamaica, continued as Bruce Fellow at the Johns Hopkins University,
Baltimore, and concluded at the American Museum of Natural History, New
York. For facilities afforded in carrying out the investigations I am under
obligations to the Board of Governors of the Institute of Jamaica, Prof. W. K.
Brooks, of the Johns Hopkins University, and Prof. H. C. Bumpus, of the
American Museum of Natural History. The research has been assisted by
an appropriation from the Carnegie Institution.

J. E. D.

University of Michigan, Ann Arbor, Mich.. U. S. A.,

IITH NOVKMBKR, 1904.

THE CORAL SIDERASTREA RADIANS AND ITS POSTLARVAL

DEVELOPMENT.

By J. E. DUERDEN.

INTRODUCTION.

The following are the more important references to and synonyms of this
well-kiiown species of coral :

Madrepora radinns, Pallas, Elench. Zooph., 1766, 322.

Madiepora astro'ttes, Linnaeus, S^s. Nat., ed. xii, 1767, 1276.

Madrepora galaxea, Ellis & Solander, Nat. Hist. Zooph., 1786, 16S, pi. 47, fig. 7.

Astrea galaxea, Lamarck, Svst. Anini. s. Vert., iSoi, 371; Le Sueur, Mdm. Mus. Hist. Nat. Paris, t.

VI, 1S20, 2S5, pi. XVI, fig. 13; Lamouroux, Expos. Meth., 18:1, 60. pi. xLix, fig. 7.
Aslr<Ea radians, Oken, Lehrb. Naturgesch., 1815, bd. i, 65; Milne-Edwards & Haime, Hist. Nat. Cor.,

1857, t. II, 506; Gregory, Qiiart. Journ. Geol. Soc. Lond., vol. Li, 1895, 277.
Astrea {Si'derastrea) galaxea, de Blainville, Diet. Sci. Nat., 1830,. torn. LX, 335; Man. Actin., 1834,

370-

AstrcBu astro'ttes, Ehrenberg, Corall. roth Meer, Abhandl. kongl. Akad. Wiss. Berlin, 1832, 319.

Siderina galaxea {pars), Dana, Zooph. Wilkes Expl. Exped., 1846, 218, pi. X, figs. 12, 12A, 12c.

Siderastrcea galaxea, Milne-Edwards & Haime, Ann. Sci. Nat., 1850, torn. Xii. 139; PourtalSs, Deep-Sea
Corals, 111. Cat. Mus. Comp. Zool., no. iv, 1871, 81 ; Florida Reefs, M^m. Mus. Comp. Zool.,
1880, vol. VII, pt. I, pi. XI, figs. 14-21, pi. XV, figs. 1-12; Quelch, Reef Corals, Challenger Reports,
1886, vol. XVI, 113; Heilprin, Proc. Acad. Nat. Sci. Phila., 1S90, 305.

Siderastr<JE\e\a radians, Verrill, Bull. Mus. Comp. Zool., vol. I, 1864, 55 ; Vaughan, Bull. U. S. Fish Com-
mission igoo, vol. 2, 1901. 309. pi. XV, pi. XVI, fig 2; Samm. der Geol. R. — Museums in Leiden,
1901, Ser. II, bd. 11, 6t ; Verrill, Trans. Conn. Acad. Science, vol. xi, 1901, 153, pi. XXX, fig. i.

The generic term has been thoroughly discussed in the recent papers of
Gregory (1895, p. 278), Vaughan (1900, p. 154), and Verrill (1901, p. 88).
Vaughan, fortunately for coral taxonomy, shows that the name Astrea, first
used biuomially as a coral genus by Lamarck in 1801, can not be retained
in madreporarian terminology, having been employed by Bolten in 1798 for
a group of gastropod shells. In this conclusion he is supported b}' Verrill.
The next generic name available is the Sideraslrea of de Blainville, 1830.
The specific term radians of Pallas (1766) has priority of the galaxea of
Ellis & Solander (1786), and Vaughan holds that ih& Madrepora as/rdites
of the twelfth edition of Linnaeus is the same as M. radians of Pallas.

The species is very common throughout the West Indies, and is recorded

2 SIDERASTREA RADIANS.

from Vera Cruz and Colon on the mainland. It is equally plentiful on the
Florida Reefs and flats, and around the more northern Bermudas. Gregory
(1895, p. 277) records it as fossil from the Low-level Reefs of Barbados and
the Pleistocene Reefs of the Bahamas.

A glance at the references shows that Siderastrea radians has been fre-
quently described and figured, but mainly as regards the corallum ; Le Sueur
(1820, p. 285), Pourtales (1871, p. 81), and Verrill (1901, p. 153) have, in addi-
tion, contributed brief notes upon the mode of occurrence of the living colonies
and the characteristics of the polyps.

As Astrtza radians the species is thus systematically described by
Milne-Edwards & Haime in their " Histoire Naturelle des Coralliaires "
(vol. II, p. 506) :

Polypary often fixed upon the Valuta turbincUus of Linnaeus, or spherical and free.
Budding takes place at the point of union of several calices. Calicos subpolygonal, appearing
thickened at the borders as a result of the strong development of the septal system, although
the walls are indicated only by fine lines. Columella formed by one or two compact tubercles,
scarcely visible, more distinct in young individuals. Three cycles of septa complete, and, in
general, a variable number of a fourth cycle, unequal. Interseptal chambers extremely
narrow. Septa much serrated, strong, very regularly crenulated at the border, nearly equal,
the primaries and secondaries a little larger. The teeth serrated, obtuse, and sub-equal.
The septa of the second cycle are fused by their internal border to those of the preceding
cycle. When the septa are broken from above they are found to be united by strong granules,
the spaces between the granules resembling small foramina. A specimen in this condition
has been considered by Lamarck as a distinct species under the name Astrcea picnctifcra.
In a vertical section the columella is compact and strong, the septa are perfect lamelte, cov-
ered with a radiating series of large granules ; the dissepiments are rudimentary, horizontal,
simple, and 0.5 mm. distant from one another. The species sometimes forms large masses.
The large diameter of the calices is from 3 to 4 mm., their depth z mm., or a little more.

An excellent engraving of a colony is given in Ellis & Solander's
"Natural History of Zoophytes" (1786, plate 47, fig. 7), under the term
Madrepora galaxea. In his report on " The Stony Corals of Porto Rico," Mr.
Vaughan (1901) adds a photographic reproduction of a flat incrusting colony
and also one showing the enlarged calices (plates xv, xvi). In the plates
accompanying Prof. Louis Agassiz's " Report on the Florida Reefs " (1880)
are reproduced (plate xv) a dozen beautifully executed drawings represent-
ing the polyps and skeleton in some detail. Many of the actual details with
regard to the tentacles and septa are different from those observed on the
Jamaica 5'. radians, but there seems no doubt that the same species is
intended in each case. The drawings of the polyps are interesting as show-
ing for the first time the bilobed character of the tentacles and their irregular
arrangement at varying distances from the center of the disc.

INTRODUCTION. 3

The species occurs plentifully in Kingston Harbor, Jamaica, from the
shore down to a depth of 5 or 6 feet. Here the colonies are small, and sub-
spheroidal or irregular in shape, partly or wholly incrusting some dead
shell or stone. Occasionally nearly globular colonies are obtained with the
polyps equally developed all over the surface. Such colonies lie exposed on
the sea-floor, or partly buried in beds of the aquatic phanerogam Tlialassia
marina^ and are associated with other free corals, such as Ma?iicina areolata
(Linn.), Cladocora arbuscula (Le Sueur), and Pontes divaricata (Le Sueur) ;
sometimes they are found incrusting shells inhabited by the living mollusc
or by hermit crabs. A flat incrusting form, growing upon stones and blocks
of coral, occurs on the coral flats outside the harbor in the area of reef
formation. Here the species may give rise to large fixed masses, 50 or
60 cm. across, while in shallow water the free colonies rarely exceed a
diameter of 10 cm. On the reefs S. radians is associated with another
common West Indian species of the same genus, viz, S. siderea (Ell. & Sol.).
This latter may produce large heraispheroidal blocks, and in situ is readily
distinguished from 5. radians by the larger size of the polyps and their
reddish brown color.

S. radians seems peculiarly hardy as regards the conditions under which
it will thrive. The shore waters within Kingston Harbor are often muddy
from the action of the strong day breezes, and at times the living polyps are
covered by deposits of sand and mud. Around Jamaica, however, the colo-
nies are rarely exposed to the direct rays of the sun, the fall of the tides
being very restricted ; but further north, where the diflPerence between high
and low tides is greater, both Pourtales and Verrill have found specimens
so situated that they are frequently subjected to the direct rays of the sun.
With regard to the occurrence of 5. radians in Florida, Pourtales, in " Deep-
Sea Corals" (1871, p. 81), writes:

In altitude it ranges higher than any other coral of the Floridian fauna, as small masses are
found flourishing in pools left by the tide. I have even seen small clusters left partially dry in
a hot sun, keeping up a communication with the water only by a few of the lowest polyps of
the group. From their position they must frequently have been thus exposed without incon-
venience. According to Prof. Agassiz the polyp has bilobed short tentacles at different
distances from the center.

Prof. A. E. Verrill (1901, p. 154) gives the following notes as to the
conditions under which the species is found at the Bermudas :

This species, which is abundant at the Bermudas, is more hardy than most reef corals,
for it can live and grow well in shallow water on mud flats, where it is laid bare by nearly
every tide, and where most other corals would be smothered in the mud, though 5". siderea

4 SIDERASTREA RADIANS.

and some forms of Isophyllia fragllis are usually found with it in such places. It is often
partlv buried in the white calcareous mud of the flats, and yet seems healthy there. It is also
abundant in the sm;dl, shallow pools left on the flats by the tide. But it is equally common
on the reefs, where it often grows larger. It is also found well grown in Harrington Sound.
Exposure to the dry air, or even to the hot sun, for an hour or so, does not kill it, if it be wet
beneath. Probably its porosity enables it to absorb sufficient water to prevent drying up.

The natural occurrence of the living coral under such varying condi-
tions as to purity of water and exposure indicates that this species, at any
rate, is not so extremely sensitive to its environment as corals generally are
found to be.

The polyps are so small that their external characters can be fully made
out only with the aid of a lens or a low-power microscope. Further, they
differ so much in appearance, according to the amount of expansion and
retraction, that to obtain their complete characteristics it is desirable to keep
the colonies under observation for some time and subjected to various condi-
tions. As the stocks often lie free on the sea-floor, they can be collected
without any injury to the polyps, and are sufl&ciently small to be kept in the
laboratory in ordinary glass vessels. With a little attention no difficulty is
experienced in keeping the polyps alive. Indeed, the colonies continued to
increase in size during conflnement, and while kept under observation new
polyps began to arise from the calices of previously dead areas.

The boring bivalve Lithodomus appetidiculatus Phi.* nearly always
occurs within the corallum, often several in each colony. The siphon has
a dark purple, funnel-shaped incurrent aperture and a small, tube-like
excurrent aperture, and protrudes a short distance beyond the surface of the
colony, the activity of the mollusc keeping up a strong circulation of water ;
larvae and other small floating objects may pass in rapidly at one aperture
of the siphon, and after a brief interval may be shot out from the other.
Numbers of cirripedes [Pyrgojua] also frequently infest the corallum, and
assist in maintaining a current of water over the living colony. It seems
not unlikely that these infesting organisms may be of much importance in
clearing away the mud and sand which accumulate on the surface of the
polyps, and the currents produced by them may also bring food within reach.

During the greater part of the day the colonies in the laboratory were
kept in the shade, or even in darkness, by placing the glass receptacles
under some cover. Under these conditions the polyps remained partly
expanded, but retracted when exposed for any length of time to the direct
rays of the sun. In the early morning and towards evening the vessels were

• For the determination of this species I am indebted to Prof. L. P. Gratacap, of the American Museum
of Natural History.

INTRODUCTION. 5

removed for a short time into the full rays of the tropical sun, or kept for
hours in diffuse light, when all the polyps expanded to their fullest degree,
and seemed tingling with vitality ; but in general, like most other corals
and actinians, the polyps seem most vigorous in the shade and at night time.
Detailed experiments on the phototropism of coral polyps are much needed.

When the colonies were exposed to strong sunlight for a time, in a limited
quantity of water, bubbles of oxygen gas were seen to form within the cavity
of the polyps and then to escape through the mouth. The production of
oxygen is without doubt dependent upon the metabolic activity of the uni-
cellular algcC or Zooxanthellae present in great numbers throughout the
endodermal tissues (p. 22). The filamentous green algse found perforating
the skeleton (p. 38) would doubtless have a similar activity, but the intensity
of the light reaching their chloroplasts must be much less than that falling
upon the more superficial Zooxanthellae.

The oxygen produced from the combination of the physiological activities
of the plant and animal was ample to keep a limited quantit}^ of water in a
proper state of aeration ; and once the colonies were established the water in
which they were living remained perfectly clear and sparkling without the
addition of other plant life. Generally it was unnecessary to renew the sea-
water, but from time to time fresh water was added to replace that lost by
evaporation. To make certain that the water contained sufficient calcareous
salts in solution to enable the young polyps to build up their skeleton it
was partly changed every few days. Like the adult polyps, the developing
individuals contained many Zooxanthellae within the endoderm.

For food the polyps were given small living v/orms, pieces of molluscs,
crabs, or fish. These they took readily by first extending their tentacles
and later their mouths over them. Sometimes a hapless polj'chaete would
attempt to creep over a colony, when it was quickly killed and ingested ; other
small worms, wriggling with great vigor, were unable to free themselves when
dropped upon a colony. The tentacles of the polyps fastened upon the
annelid at one or more places, and its movements soon began to weaken,
as if the creature were paralyzed. Each polyp immediately around the
worm opened its mouth widely, while those more distant extended their per-
istomes towards it, as if all were moved by a common impulse. As many as
half a dozen polyps would thus seize hold of a small worm at one time.
Naturally a difficulty arose when two polyps commenced to engulf the same
worm from opposite ends. The unfortunate annelid soon reached a state
of tension between the polyps, and the peristome of each of the latter was
drawn outwards to an unusual degree. For over an hour a worm was

6 SIDERASTREA RADIANS.

seen thus pulled at from each extremity. A large piece of a mollusc might
also be seized upon and partly ingested by two polyps at the same time, and
in the end their lips would approach and come into actual contact. A larval
polyp only foiir weeks old was able to hold a wriggling fragment of a small
worm and attempted to swallow it.*

For examination of their anatomy and histology some of the polyps were
preserved in the partly expanded condition in 5 per cent, fomiol and others in
a solution of corrosive acetic, after narcotization with menthol or magnesium
sulphate. Shortly after preserv^ation they were transferred to alcohol. Where
decalcification was necessary this was carried out by means of weak solutions
of hydrochloric or nitric acid, a few drops being added from time to time to
maintain a slight effervescence. For such a dense corallum as that of Sider-
astrea the process of decalcification required three or four days. After the
process the space heretofore filled by the skeleton was invariably found to
be occupied by a fluffy mass, which on examination under the microscope
was found to consist of delicate algal filaments along with fine organic
particles. Microscopic sections of the corallum were made by embedding in
Canada balsam and then grinding to sufl&cient thinness, as in the preparation
of ordinary rock sections.

♦Experiments conducted since this account was sent to the printer demonstrate that the mucus exuded
bjr the ectoderm cells plays an important part in the feeding of coral polyps, particularly where the food
particles are small. The latter falling on the poly pal surface become embedded in mucus, and if not revers-
ing the ordinary direction of movement of the stomodaeal cilia they are slowly wafted away beyond the
tentacles ; nutritive particles and solutions bring about a reversal of the action of the stomodteal cilia, and,
along witli mucus in the form of strands or threads, are gradually drawn into the polypal cavity.

ADULT COLONY.

EXTERNAL CHARACTERS.

The polyps are small and subpolygonal in outline, rarely exceeding
4 mm. in diameter when retracted, and 5 or 6 mm. on expansion. Six unequal
sides constitute the basal periphery of large polyps, while there are only four
or five in smaller examples. In general, the individuals in any colony are
irregularly disposed. In some regions an arrangement in parallel rows can
be made out, but usually this is interfered with by the intercalation of new
polyps among those fully grown. In most branching corals, alcyonarians,
and compound actinians, it is usual for the principal oral axis of the polyp
to be symmetrically placed with regard to the axis of the colony, but no such
regularity of orientation is manifest in the polyps of Siderastrea.

The superficial polypal tissues are so thin that the white corallum
below shows through, and occasionally the internal arrangement of the
mesenteries around the stomodaeum can be followed. This is especially
the case with polyps which have been preserved in formalin. Retracted
pol^-ps usually exhibit radiating lighter areas, which correspond with the
cycles of septa below, and narrower alternating darker areas, which corre-
spond with the interseptal spaces and internal attachment of the mesenteries
(plate 6, fig. 32).

The living polyps making up a colony are separable from one another
along a narrow polygonal groove which is always lighter in color than the
rest of the polypal wall. Along this groove the column wall is connected
with the corallum, not directly, but by the intermediation of the mesenteries,
and to whatever degree the polyps become expanded the wall is here never
much elevated. The polyps, however, are at no time raised far above the
corallum. On full expansion they rarely assume the true cylindrical form
with a flattened disc, like most coral polyps, but exhibit merely a dome-like
elevation of the walls over the calice, 2 or 3 mm. high (plate 6, fig. 31).

At no place, except in simple larval polyps and at the margin of a
colony, is the boundary of the column wall of the individual polyp directly
connected with the skeleton; that is, it is nowhere continuous with the
skeletogenic basal wall. The margin of the column of any polyp passes
uninterruptedly into that of the surrounding polyps, and, as will be shown
later, the mesenteries pass from the column wall to the skeletogenic tissues
below, leaving narrow interspaces, which place the digestive cavities of all
the polyps of a colony in communication.

8 SIDERASTREA RADIANS.

On partial retraction the column wall comes to rest upon the corallum,*
but the central area of the disc and the tentacles remain a little elevated, the
peristome especially so. On fullest retraction these also come to lie closely
against the skeleton and follow it all the way, nearly to the bottom of the
calice. Unlike the polyps of most corals, the column wall in Siderastrea is
incapable of folding over the tentacles and disc upon retraction ; both the
column wall and disc are simply drawn down upon the skeletal tissues,
leaving the tentacles and mouth exposed. Where, in other coral polyps, an
overfolding of the polypal wall occurs, it is brought about by the action of
the circular endodermal musculature of the column wall, sometimes in the
form of a sphincter ; but histological examination reveals that the columnar
musculature in S. radians is of the weakest character.

The genus Siderastrea^ in many respects, is allied to the mushroom
coral Fungia, both being included within the section Madreporaria Fungida.
The genus Fungia is not found in West Indian shallow waters, but the
various figures by Dana (1S46), Bourne (1887, 1893), and Saville-Kent (1893),
of different species of Fungia found elsewhere, show that in this genus
also the column wall in retraction is not drawn over the disc. The West
Indian species of the allied genus Agaricia are also unable to overfold the
column wall.

Apparently the outer surface of the polyps of S. radians is not ciliated,
for light particles dropped upon it remain there, and no ciliary motion is
recognizable when the living polyps are examined under a moderate magni-
fication. Similar particles dropped upon colonies of Manicina areolata are
gently swept away and over the sides in such a manner as to plainly indicate
ciliary activity, so that probably different coral species vary in this respect.
The outer surface is always uniformly ciliated in larvse (plate i, fig. i).

COLUMN WALL.

The columnar and discal areas of the polyps are not sharply separated
from one another under any condition of expansion or retraction. The outer
cycle of tentacles, which in ordinary cases marks the peripheral boundary of
the disc, here forms but an irregular circle, and its members are comparatively
widely apart (plate 6, figs. 31, 32). Including as column all that part of the
polypal wall between the polj'gonal line of attachment of the polyps to one
another and the outermost cycle of tentacles, its extent is very limited, being

* The free superficial regions of the poljp (column and disc) never come into actual contact with the
skeleton, as the living parts of the latter are always covered with the basal skeletotrophic tissues. When
they are said to rest upon the corallum the tissues lining the corallum always intervene.

ADULT COLONY. 9

represented b}' a zone not exceeding 2 mm. across. When the polyps are
expanded the proximal region of the column remains somewhat polygonal in
outline, but distally it becomes circular in section, rarelj^ assuming the true
columnar form.

The column does not present the two topographical subdivisions, calicinal
and pericalicinal, like most corals. Instead it extends directly from the edge
of the calice, so that it is wholly within it; in other words, no "Randplatte"
or " edge-zone " occurs in Siderastrea.

Structurally, the column wall is very thin, partly transparent, and
smooth ; on expansion vertical lines are clearly visible, indicating the internal
attachment of the mesenteries.

TENTACLES.

The oral disc is divisible into an outer tentaculiferous region and an
inner naked area or peristome, with the mouth in the middle. The tentacles
are arranged widely apart, and thus occupy a comparatively broad zone of
the disc, leaving only a very restricted smooth central area. As can be seen
from plate 6, fig. 32, the tentacular region during retraction includes the
greater proportion of the exposed part of the polyp, while the columnar area
is very narrow, and little remains of the naked portion of the disc. This
exceptional relationship of the columnar, tentacular, and discal areas results
from the wide intervals which separate the different cycles of tentacles, a
similar condition being apparently characteristic of the Fungidse generally.
In most coral and actinian polyps the tentacular cycles are placed close
together, the bases of the tentacles of one cycle being partly embraced b}' the
bases of the next cycle, and thus the tentacular zone is very narrow compared
with the remainder of the disc.

The tentacles of S. radians are remarkable among all West Indian
coral polyps on account of their form and arrangement. Under certain con-
ditions of partial or even full expansion of the polyps the organs can be
discerned with a lens as small sessile tubercles, arranged only approximately
in cycles, some as pairs and others singly, and corresponding in position
with the septa. In most cases they stand out as small, delicate, simple or
bifurcated finger-like outgrowths of the disc. As seen externally each
appears to originate over the central termination of the septum with which
it corresponds, that is, before the margin of the septum commences to dip
downward.

When the tentacles are fully extended and much enlarged they appear
as shown on plate 7, fig. 37. The stem consists of a broad basal part which

lO SIDERASTREA RADIANS.

in most members of tlie inner cycles becomes bifurcated a little above midway,
each half terminating in a light-colored knob which represents a batter}^ of
nematocysts ; the outermost members, however, are simple, like those of the
majority of coral polyps.*

In studying the internal anatomy of the polyp, it is found that the bifur-
cated members communicate with the entocoelic chambers, and the simple
forms with the exocoelic chambers. Development shows that the double
tentacles are at first simple, and such may be the condition of some of the
outer entocoelic in nearly mature polyps. The number of tentacles always
corresponds with the number of septa, the bilobed tentacles being situated
over the entosepta, and the simple tentacles over the exosepta. Their order
of appearance is described later in connection with the development of the
larval polyps.

The tentacular walls are smooth, that is, devoid of the nematoblast
tubercles which so frequently characterize the tentacles of coral polyps.
The aggregation of large nematoblasts is restricted to the apical knobs.
The stems are hollow and brown in color, the latter character due to the
presence of Zooxanthellse in the endodermal tissues. The knobs, on the
other hand, are solid and colorless. The organs are adhesive and able to
take particles of food from the tips of forceps. There is little difference in
size among the tentacles of the various cycles, but the bifurcated members
are larger than the simple.

During retraction, and even sometimes on full expansion of the polyp,
the tentacles appear as unstalked, spheroidal tubercles, sessile on the disc.
In the case of the bilobed tentacles, a knob is situated ou each side of the
septum with which the tentacle corresponds, while in the simple tentacles the
knob is directly over the septum (plate 6, fig. 32). No stem can be recog-
nized under such circumstances, aud sections show that it is not invaginated,
but has become part of the discal wall (plate 7, fig. 40). Such a condition
of the tentacular stems is ofteu found in corals, but rarely among the larger
tentacles of actinians. When the tentacles are fully expanded the}^ stand
obliquely to the surface of the disc, and are bent outwardly. The organs
have never been found invaginated, a phenomenon which frequently occurs
in coral polyps and alcyonarians.

• Verrill (1901, p. 154) makes the following remarks with regard to the tentacles of 5. radians : " the
tentacles are small, short, cylindrical, orclavate; they form several circles, and'appear somewhat scattered
those of successive cycles being in different circles and decreasing in size. Bii/ tk<'y are not bilobed, nor
trilobed, as A^assiz and Pourtales suf posed. This appearance is due to a smaller one standing on one
or both sides of a larger one, and close to it." From the part italicized it would appear that the tentacles
in Verrill's Bermudan specimens must have been altogether different from those in Jamaica polyps,
which, moreover, agree with those observed by Agastiz and Pourtales.

ADULT COLONY. II

The hexameral cyclic arrangement of the tentacles is by no means easy
to establish, and would be almost impossible to ascertain without anatomical
aid or assistance from the septa below. According to the usually accepted
cyclic formula for hexactinian polyps employed in systematic works, S.
radians would be said to have three cycles complete and a fourth incomplete,
and the cyclic formula would be 6, 6, 12, x, where x may be any number
from I to 24.

On the living polyp six bifurcated tentacles can usually be distinguished
as constituting an inner or primary cycle, which is widely separated from
the other cycles ; the remaining tentacles are more closely disposed and their
arrangement is somewhat obscure, simple and double members being inter-
mingled. Among them a second cycle of six bilobed tentacles can sometimes
be recognized, alternating with the first and remote from it. Beyond this
there is no true hexameral plan determinable. In most polyps a few mem-
bers of a third cycle of bilobed tentacles are present, but the full complement of
twelve entotentacles, necessary by the laws of actinian symmetry to complete
the alternation with the two previous cycles of six each, appears never to be
developed. The outer simple exocoelic tentacles correspond in number with
the sum of the members of the primary, secondary, and tertiary entotentacles
with which they alternate, but some belong to the third cycle and some to
the fourth. Only the first and second cycles are completely hexameral ;
the third and fourth are incompletely so.

In transverse sections through the polyps (plate 6, fig. 34) it is found
that the first and second cycles of mesenteries consist of six pairs each, but the
third cj'cle rarel}? or never contains the next number in the hexameral sequence,
namely, twelve. The number of entocoelic chambers corresponds with the
number of exocoelic chambers, therefore the number of entotentacles will
correspond with the number of exotentacles, as a tentacle communicates with
each mesenterial chamber. Whatever number of entotentacles be lacking
to complete the third cycle of 12, a like number of exotentacles will be
wanting to complete the fourth cycle of 24. Thus, considered morpho-
logically, the tentacular formula will be 6, 6, x, 6 + 6 + x, where x will
be the same in the two cases, and may be any number from i to 12 ; the
series 6, 6, x will represent the number of entotentacles, and 6 + 6 + x the
number of exotentacles. A polyp with 38 tentacles will have the morpho-
logical formula 6, 6, 7, 19 ; a polyp with 44 tentacles the formula 6, 6, 10, 22.

Where, in mature polyps, the hexameral sequence is incomplete, it seems
preferable to employ the morphological formula as compared with the ordinary
cyclic formula. In the first we have the true nature of the tentacles indicated ;

12

SIDERASTREA RADIANS.

in the second the penultimate cycle is assumed to be complete, whereas it is
made up of two different series of tentacles — entotentacles and exotentacles —
and simply represents an arrested stage in the hexameral plan.

Fig. I is a diagrammatic representation of the arrangement of the tenta-
cles in a polyp of Siderastrea with the morphological enumeration added.
In the two inner cycles (i, ii) six bilobed tentacles are complete, while only
five tentacles of the third cycle (in) are yet developed ; alternating with the
members of the three inner cycles of bifurcated tentacles is a simple tentacle

Fig. I. — Diagram showing the relationships of the septa and tentacles in the same
polyp as that from which fig. 34, plate 6, was taken. All the entotentacles are
shown with a double npex and the exotentacles with a single apex. The exo-
tentacles are all arranged as if forming the fourth or outermost cycle, whereas
some will belong to the third and some to the fourth cycle, according as a third
cycle entolentacle is present or not.

(x) belonging to the outermost series, but not forming a true cycle. The septa,
indicated by the thick radiating lines, show a corresponding arrangement.
For purposes of comparison the same polyp is also shown in transverse section
on plate 6, fig. 34, while the septa are diagram matically represented in fig. 2,
p. 13. By comparing figs, i, 2, and plate 6, fig. 34, it is seen that the three
inner cj'cles of tentacles communicate with the entocceles and correspond with
the entosepta, while the members of the outermost cycle communicate with
the exocceles and correspond with the exosepta.

The tentacles of S. siderea are of the same character as those of S. radians,

ADULT COLONY.

13

only somewhat larger ; tte entotentacles are bifurcated, and the exotentacles
simple. In the genus Agan'a'a, also belonging to the Madreporaria Fungida,
the tentacles are all simple, but in their wide distance apart they resemble
those of Siderastrea.

The fully expanded tentacles of most species of Fungia seem to be small
and club-shaped, but in F. crassitentaculata they are much larger than is
usual for the organs in corals generally. This is well shown in the figure

KiG. 2. — Diagram showing the relationship of the septa to one another in the same
polyp .IS that of fig. I,

of /^ crassitentaculata in Saville-Kent's " Barrier Reef." In all Fungice the
organs are widely apart, and, as in Siderastrea and Agaricia^ occupy nearly
the whole discal area.

The wide separation of the tentacles in the genera mentioned usually
tends to obscure their hexameral cyclic plan. The two pol3'ps of Siderastrea
represented on plate xv in Agassiz's " Florida Reefs " are evidence of this,
and also the remarks of Verrill, given in the foot-note on p. 10. Such is also
the case with Dana's figure of Fungia lacerta, but Bourne found the tenta-
cles of F. clavata to be arranged in distinct cycles. The positions usually
assumed by the tentacles of Siderastrea enable one to understand how such

14 SIDERASTREA RADIANS.

confusion may arise ; indeed, the polyps were for a long time nnder observa-
tion before any regular arrangement could be established with certainty.
It was only after observing the Jamaica polyps under various phases of
expansion that the cyclic plan of the first tentacles could be determined,
while, as regards the two outer cycles, such an arrangement is more
theoretical than actual. The tentacles on plate 6, fig. 32, from a camera
drawing of a preserved polyp, do not readily suggest a hexameral cyclic
plan.

Siderastrea is apparently the only madreporarian genus with true
dimorphic tentacles, the peculiarity having been first recognized by Agassiz.
The independent origin of the two moieties of the bifurcation, as described
later, is also a remarkable feature in the tentacular development of the
Zoantharia. Dimorphic tentacles are occasionally met with among the
tropical Actiniaria, e. g., Rhodaclis^ Phymanihus ; but here they are the disc
tentacles as compared with the marginal tentacles which show the distinc-
tion, and the morphological value of the former is uncertain. At any rate,
the disc and marginal tentacles of actinians are not comparable with the
entotentacles and exotentacles of Siderastrea.

DISC AND MOUTH.

The naked portion of the disc is very limited in extent. It is smooth,
circular, and usually partly depressed within the calice. So transparent is
the wall that the cyclic arrangement of the internal mesenteries and septa
can be easily followed. The central part or peristome is not separable from
the rest of the disc, except that it is usually elevated in a conical manner, and,
owing to a greater concentration of the endodermal Zooxanthellae, it appears
darker than the rest of the polyp.

The mouth is small and slit-like when closed, oval or circular when
open. The lips are not specially marked off from the peristome by any
tumidit}^, such as occurs in many anemones. In living polyps the mouth
sometimes closes along the middle by the approximation of the lips, but
leaves a small opening at each end. The apertures, however, can not be
regarded as representing a siphonoglyph or gonidial groove, for microscopic
examination reveals no difference in the character of the stomodaeal wall
between the sides and extremities. On narcotization the circular muscles of
the disc relax, and the oral aperture remains widely open and circular; the
stomod^al wall is then seen to be smooth, without ridges and grooves.

ADULT COLONY. 1 5

COLOR.

The pol3'ps on the under, unexposed surface of colonies ma}' be color-
less, but vary elsewhere from light to dark brown. Owing to the partial
transparency of the tissues the white corallum always shows through, and
gives rise to an alternation of light and dark radiating areas and to a light
polygonal line separating one polyp from the other. Sometimes the disc
presents a velvety green color along the mesenterial radii, and the oral aspect
of the tentacles and the angles of the mouth also may be of the same color.
The stomodseal walls and knobs of the tentacles are white, while the stems
of the latter are usually brown. More rarely the elevated peristome may be
rose-colored, and in some cases purple.*

The different shades of brown exhibited by the polyps are entirely due
to the presence of Zooxanthellae or unicellular algse within the endodermal
la3'er. The chromoplasts within these are brown, yellow, or greenish 3'ellow,
and when the algae are but few in number they give a light brown color to
the walls of the polyps, whereas when present in great numbers, overlying
one another, the polyps appear a dark brown. For this reason the thin
tissues of fully expanded polyps are much lighter in color than the tissues
of retracted polyps.

Examination of the walls of colorless polyps on the under surface of a
colony proves that the Zooxanthellae are either absent or occur very sparingly
in the endoderm. When, however, the colorless areas are exposed to light,
the polypal tissues begin slowly to assume a brownish color, and in a few
days 3'ellow cells are to be found in abundance.

On examining a living tentacle under the microscope the outer ectoder-
mal layer is seen to be quite colorless, but the yellow Zooxanthellae are
revealed by focusing through to the internal endodermal layer. The white-
ness of the spheroidal tips of the tentacles, compared with the brown color of
the stems, is also due to the fact that the knobs are wholly formed of ectoderm ^
the endoderm lining the cavity of the stems not being prolonged into them.
The superficial green or purple color is somewhat evanescent, and wholly
ectodermal in origin.

Different shades of yellowish brown are the prevailing colors of the coral
areas in Jamaica, and whether in the palmate or branching growths of

* Le Sueur (1820, p. 286) describes the color of the specimens seen b_v him at Guadeloupe as " d'un
rouge mele de violet." The color of 5. siderea as given by the same author is also very different from
that of the Jamaica specimens. Verrill (1901, p. 154) writes thus of the Bermuda 5. radians: ''The
general color in life is dull gray, yellowish gray, ocher-yellow, or rusty brown, sometimes tinged with a
purplish rosy tint; the polyps are paler, with the lips and tips of the tentacles whitish."

1 6 SIDERASTREA RADIANS.

Acropora {Madrepora), the hydrozoan Mi/Icpora, or the zoanthid Palythoa,
all owe their origin to the presence of unicellular algae with yellow chromo-
plasts within the endodermal cells.*

Stanley Gardiner and Hickson have both drawn attention to the part
which commensal algse probably play in the nutrition of the polyp. The
organisms are actually inclosed within the endodermal cells, and there can
be scarcely any doubt that the polyp receives from them nutritive carbohy-
drates, produced in the course of the metabolic activity of the vegetable cells
under the action of sunlight ; further, the Zooxanthellae are constantly
increasing in number by fission into two and then four, and perhaps some
are utilized as food directly by the polyp. The liberation of free oxygen,
already shown to take place by the activity of the chlorophyll within the
algal cells, must also be of importance in the vital activities of the polyps.

In many corals and actinians, however, Zooxanthellae are altogether
absent, hence their presence can not be considered as necessary to the life
of the polyp ; also, corals grown in the shade are colorless from absence of
the algae. Further, both actinians and corals will ingest almost any variety
of animal food that is oflfered them.

REPRODUCTION.

New polyps arise asexually as intercalary buds, usually at the point of
junction of three or more polyps, so that it is impossible to say whether one
polyp more than another is to be regarded as the parent, or as to how far
the structures in the bud arise in organic association with those of the older
polyps. Over the general surface of any colony bud polyps in different
stages of development are usually present, and for some time the}' remain
much lighter in color than the others, owing to a less growth of Zooxan-
thellae ; also, around the margin of colonies the addition of new polyps
is generally in rapid progress. These latter serve to enlarge the lateral
boundaries of the colonies, while the intercalary polyps occupy the spaces
produced as the coral grows in height, and, of course, in superficial area.
Instances of reproduction by the process known as fissiparous gemmationf

*In his recent vice-presidential address before the Section of Zoology of the American Association
at St. Louis, Prof. C. VV. Hargitt comes to the conclusion that the colors in ccelenterates and many other
groups of invertebrates have probably no adaptive or protective significance, and are in main the result
of the metabolic activity of the animal. The rich profusion and beauty of color in coral pol^'ps certainly
seems to have no protective or warning significance, but those due to the presence of commensal algx
have of course an altogether different ph^-siological importance from the other colors.

t The term is applied to a method of asexual reproduction occ^isionally found in gemmiferous colonies.
It appears as if an enlaiged polyp simply divides into two. The method, however, is altogether different
from growth in fission colonies generally. ("The Morphology of the Madreporaria — IV. Fissiparous
Gemmation." Ann. Mag. Nat. Hist., ser. 7, vol. xi, 1903.)

ADULT COLONY. 1 7

are very rare, but occasionally much enlarged calices are seen undergoing
binary' fission. The detailed study of the development of young bud polyps
has not been carried out, as the species is unsuited for the purpose. It has
been established, however, that the manner of appearance of the tentacles,
mesenteries, aud septa agrees closely with that of larval polyps to be
described later.

In what seem to be dead parts of a colony, new polyps may arise
within the old calices. Outwardly these polyps are at first wholly sepa-
rated from one another and from the other living tissues of the colony, and
are usually transparent and colorless, due to an absence of Zooxanthellas.
Similar renewals of growth have been found in other corals, and give rise to
the successive deposits of skeletal matter on dead, corroded surfaces, some-
times met with when masses of coral are broken across. The continuity of
anj^ one corallite is frequently maintained throughout the vertical extent of
the mass, notwithstanding the corroded surfaces here and there.

The renewal of the polyps within old corroded calices of a coral is a
subject for further investigation. So far as could be made out from external
observation alone such polyps in Siderastrea are altogether independent of
one another, and also unconnected with other living tissues of the colony.
It may be that within the deepest parts of the old calices there still remained
living remnants of the original polyps, and that from these new polyps were
formed as buds ; on reaching the same size as the original polyps the buds
would fuse with one another by their peripheral borders and present a
continuous covering of soft tissues. New skeletal deposits would cover
over the old surface, and the presence of the latter would afterwards be deter-
minable only on breaking the colony across.

Renewals of this character, by the production of independent buds, are
to be distinguished from the growth at the living margin of colonies which
is frequently seen encroaching over dead areas. The latter is of the same
character as ordinary marginal gemmation from the parent stock, even
though it extends over old corroded surfaces. Such new growth is very
frequent in branching stocks of Acropora, but may occur in almost any
colonial species.

The death of the pol3'ps over any restricted area of a colony may be
brought about by many causes, such as the colony becoming partly covered
by or embedded in sand or mud, or by adherence or contact with other
foreign bodies, or by exposure. Local death is very frequent in such large
hemispheroidal colonies as those of Mceaiidrina and Orbicella ; the middle
portion of the blocks may be dead and disintegrating while the sides are in

l8 SIDERASTREA RADIANS.

full vigorous growth. In addition to this, the death of colonies as a whole,
without any obvious environmental cause, is not infrequent. Stanley Gardi-
ner has recently drawn attention to this subject. During his investiga-
tions on the coral reefs in the Maldives and Laccadives the impression
was gathered " that practically all the colonies of a species in any one area
died, or that there were only the isolated deaths of individual large colonies,"
apparently without environmental cause. Gardiner is impelled to believe
that " The ripening of the generative organs of a large number of polyp
colonies of the same species in a single locality or habitat, followed b}- the
subsequent death of all these colonies, is a regular phenomenon."

In the course of my experience in the coral regions of the West Indies
I have met with no instance of the regional death of all the colonies of any
species, but frequently individual stocks have been encountered of which all
the polyps seemed to be in a state of maceration. This was particularly the
case with Poriies astrcsoides. Isolated colonies were obtained seemingly alive
and normal in color, but upon examination with a lens no distinct polyps or
tentacles were recognizable. The whole of the soft tissues seemed to be a
gelatinous mass in process of decay, the coloration being due to the persist-
ence of the yellow pigment cells characteristic of this species. Such colonies
were under exactly similar conditions to others living around them, and
without further investigation any suggestion as to the cause of death of
the polyps can be only the merest conjecture.

My own experience leads me to suppose that coral polyps do not die
after the ripening of the generative products, but that, from the same indi-
vidual, one series of larvae may follow another, for in numerous instances
{Fiwia, Poriies) in which polyps charged with larvae, all at the same stage
of development, have been examined, there were still many nearly ripe ova
within the mesenteries, as if preparing to give rise to another batch of larvse.

Sexual reproduction in Siderastrea takes place by the formation of
planulae from fertilized ova. The planulse undergo partial development, as
far as the appearance of the first four pairs of mesenteries, while within the
body-cavity of the parent, and are then expelled. After swimming freely for
a shorter or longer time they settle and give rise to young polyps. Larvae
were extruded toward the end of June and during July.

EXTERNAL CHARACTERS ON DECALCIFICATION.

Certain of the external features of the polyps can be studied only after
the latter have been freed from the skeletal matrix. The corallum of Sider-
astrea is very dense, and the process of decalcification requires several days.

ADULT COLONY. 1 9

At the close of the process a delicate mass of algal filaments remains in the
spaces previously occupied by the corallum, but is readily removed. The
exposed polypal wall then reveals a smooth surface, very complicated in
form, but ever3'\vhere continuous.

After removal of the skeleton the individual pol37ps are seen to be
wholly free from one another, except at the surface of the colony. No
other connection or communication between one polyp and another exists.
Each polyp appears as if made up of a deeply lamellated column, attached
above to a flattened layer which represents the column wall and disc. A
middle tubular space, formerly occupied by the columella, extends for some
distance upwards, when it terminates blindly. The lamellae represent the
tissues which occupied the interseptal spaces. They are now wholly free
from one another below, but in the upper region are united centrally. They
are easily torn apart along their central and upper lines of attachment.
One of the separated lamellse, slightly enlarged, is represented on plate 6,

fig- 33-

The whole of the external wall liberated by decalcification is the mor-
phological base of the pol3fp. Though columnar in form it in no way
corresponds with the column of the skeletonless Actiniaria, but represents
the flattened basal disc which, everywhere continuous, has become much
infolded and subdivided in correspondence with the upward growth of the
septa and columella. Thus the column of a decalcified retracted polyp is
really the verticallj^ elongated, much subdivided basal disc. The true column
wall, corresponding with that of an actinian, is here represented by the
narrow periphery of the flattened or concave superficial disc which constitutes
the upper end of the polyp. To understand the form ultimately assumed
by the originally flat basal disc, the arrangement of the septa and columella
in the individual corallite must be borne in mind, as every part of these is
covered by the basal wall. An early stage in the basal complexity is
represented in the section of the 3'oung polyp on plate 9, fig. 53.

The upper part of the liberated tissues appears delicate and somewhat
clear and transparent, but as the aboral termination is approached the walls
become denser and more opaque white. Such an alteration in the external
character of the embedded tissues is met with in nearl3^ all corals, and is
usuall3^ more marked than in Siderasirca. It is found to be associated with
a corresponding thickening and histological modification of the endodermal
layer (p. 32).

The pol3'ps when set free are subpolygonal in outline, and, as already
mentioned, are wholl3' distinct from one another, except at their uppermost

20 SIDERASTREA RADIANS.

termination, where they are all in communication and joined along the poly-
gonal boundary lines of the column wall and skeletogenic tissues immediately
below. An interval of only about i mm. separates laterally the columns of
contiguous polyps. The length of the polyps liberated from the skeleton
varies someAvhat, but whatever the thickness of the corallum the polypal
tissues are alwaj'S superficial, never extending downward for more than
about 5 mm. The largest polyps vary from 4 mm. to 5 mm. in length after
decalcification, while young buds extend much less within the corallum.
The diameter also varies from 3 mm. to 5 mm., and in mature polyps is prac-
tically the same throughout the column. The soft tissues of young buds,
before any columella has appeared, are more obconical, that is, longer centrally
and shortening as they pass to the periphery. Mature polyps are more
cylindrical and terminate somewhat abruptly, all the lamellse being of prac-
tically the same vertical length. Owing to the presence of synapticula, the
actual base of the individual lamellae is not always as regular or even as in
corals resting upon a smooth dissepiment ; the lower surface of a lamella
may, in fact, be partly folded round a synapticular bar, as in plate 6, fig. 33.

The individual interseptal lamellse, like the septa themselves, belong
to diflferent cycles and are of different radial lengths. Some are simple,
while others are subdivided peripherally into two or three parts. For the
greater part of their vertical length, that is, as far as the columella extends,
the lamellae are free, either singly or in groups of two or three, but towards
the upper extremity they are all joined to one another along their inner
margins (plate 6, fig. 34).

Any simple lamella when seen in surface view appears as a flat, sub-
rectangular plate, nearly of the same thickness throughout, and correspond-
ing, of course, with an interseptal space (plate 6, fig. 33). It is perforated
by rows of circular or oval apertures, arranged roughly in one to four vertical
series, the number varying with the cycle to which the lamella belongs.
In lamellse extending all the way from the periphery to the columella three
or four rows occur, while in narrower lamellse only one or two rows are present.
The perforations represent the spaces occupied by the calcareous synapticula
which stretch across the interval from one septum to an adjacent septum.
Numerous granulations occur on the septal walls (plate 10, fig. 63), but only
the large more peripheral ones bridge the interspace between two adjacent
septa, and, in doing so, necessarily perforate the soft tissues lining the septal
walls and the mesenteries contained within. The smaller granulations merely
cause indentations in the lining tissues.

Transverse sections through the lamellae show that they consist of the

ADULT COLONY. 21

two lateral walls wliicli line the adjacent faces of two contiguous septa, and
inclose between them a narrow chamber, broken up by the synapticular
perforations in such a manner as to present a complex canal-like condition
(plate 6, fig. 34, and plate 7, figs. 38, 39). The chambers represent the inter,
septal locnli less the thickness of the lining wall, but will, however, be spoken
of as interseptal spaces. The upper half of each contains a single mesentery,
but below they are empty. Moreover, only at the uppermost extremity of
the polyp is the mesentery found to extend outwardly as far as the peripheral
boundary. In the upper region of the polyp the inner border of each lamella
is open so that the chamber, along with all the others, communicates with the
central cavity of the polyp, while below all the chambers are closed centrally,
or are only in communication toward their centripetal edges (plate 7, figs.
38, 39). Where the chambers are wholly cut oflF from one another the
calcareous septa extend throughout the radial length of the polyp, and
are centrally united with others or with the columella (plate 7, fig. 39) ;
when the chambers are only partly separated, communicating towards the
center, then the septa between them do not extend as far as the middle of
the calice.

ANATOMY AND HISTOLOGY.
COLUMN WALL AND DISC.

In radial sections of retracted polyps the column wall and disc appear
as a continuous layer, and, histologically, the two are nearly alike through-
out. The wall as a whole is very thin, except where the knob of a tentacle
is included, when the ectoderm becomes thickened (plate 7, fig. 40, and plate
9) fig- 53); to a less degree the layer is also thickened along the line of
attachment of the mesenteries. In sections the three ccelenterate layers,
ectoderm, mesogloea, and endoderm, together, measure only about 0.07
mm. in thickness.

The chief constituents of the ectoderm of the column wall are support-
ing cells and clear mucous gland cells, while small nematocysts occur some-
what sparsely. The unicellular gland cells are most conspicuous toward the
periphery of the layer, where they appear oval, with clear or vesicular con-
tents. The nuclei of the various cells are oval, and arranged as a whole
within the inner half of the ectoderm, a few, more circular in form, being
found near the mesogloea. There is no trace of any ectodermal musculature
on the column, but a system of weak fibrils is developed over the disc, their
direction being radial. No evidence of external ciliation can be detected
even in the best preserved examples.

The mesogloea in the column, as elsewhere throughout the polyp, is

22 SIDERASTREA RADIANS.

feebly developed. In the column and disc it appears only as a thin dividing
lamella between the ectoderm and endoderm, but in the mesenteries it
becomes a little broader. It is homogeneous in structure, except for the
presence of minute connective-tissue cells. The mesoglcea of corals gener-
ally appears perfectly homogeneous, but in several West Indian species the
layer is found to contain migrant connective-tissue cells, such as occur in
the larger actinians.

The endoderm of the column and disc is a little broader than the ectoderm,
and its cells contain numerous Zooxanthellse, which are altogether absent
from the latter. The endodermal cells are much vacuolated, and distinct
cell outlines can be made out. The commensal algas are more numerous in
some regions of the endoderm than in others, though there seems to be no
regularity in their distribution. Where they are absent, or nearly so, the
endoderm is somewhat thinner, its cells are less vacuolated, and the layer as
a whole stains more intensely, the nuclei forming a more regular band.

A delicate circular muscle layer can be discerned over the inner face of
the mesoglcea of the column and disc, but no concentration, such as can be
regarded as constituting a sphincter muscle, occurs at any region of the
column. When studying the living characters it was found that the polyps
are unable to fold the column over the disc, hence the presence of a special
circular or sphincter muscle, as met with in most anemones and some few
corals, would be scarcely expected.

In sections which have been passed through Delafield's haematoxylin
the deeper part of the endodermal layer throughout the polyp, but particu-
larly the epithelial lining of the mesenteries, contains some substance which
stains intensely, and has a very irregular distribution. The appearance, as
seen in sections of mesenteries, is represented on plate 8, fig. 50, and in the
lining of an interseptal loculus on plate 8, fig. 47. Small, irregular, deeply
staining patches lie next the mesoglcea on both sides, and prolongations
extend for varying distances among the endodermal cells, occasionally
reaching almost to the surface. In tangential or oblique sections through
the endoderm an appearance of somewhat irregular longitudinal canals is
presented. The substance stains and behaves altogether in the same manner
as the contents of the mucous cells. No evidence of structure is presented,
nor is there any cellular character suggested. I conceive that the appearance
is due to intercellular spaces filled in the living condition with mucus or
some similar substance, capable of staining strongly in hsematoxylin ; per-
haps a hint of the 13'mphatic spaces of the higher animals.

This is the first time that such a system of apparently intercellular

ADULT COLONY. 23

spaces has been observed in the Zoantharia. Its appearance in sections of
Siderastrca stained in hgematoxjdin is very characteristic, and its presence
should be looked for in other forms.

TENTACLES.

When the polyps are preserved in the retracted condition, with the
tentacles sessile, the columnar and discal walls come to lie upon the skele-
totrophic layer of the septal edges, and the two knobs of the bifurcated
tentacles are far apart, one on each side of a septal ridge (plate 7, fig. 40),
while the knob of the simple tentacles lies immediately over the septum with
which it corresponds. In this condition the stem of the tentacle has altogether
disappeared as such, having become part of the disc. As a result the tenta-
cles are recognizable in vertical sections of polj^ps as mere hemispheroidal
thickenings of the ectoderm. In the case of the entocoelic tentacles a thick-
ening occurs on each side of a septal elevation of the disc, while a single
elevation directly over the apex of the septum represents an exotentacle.
In the thickenings long nematocysts extend nearly across the ectoderm, and
smaller ones crowd the periphery. The two forms of nematocysts are
represented on plate 7, fig. 44. Histologically the portion of the wall join-
ing the two tentacular knobs on plate 7, fig. 40, and representing the stem of
the tentacle, differs in no way from that of the disc.

In sections of polyps preserved with the tentacles extended the organs
appear as tubular outgrowths of the disc, terminated by knoblike enlarge-
ments. The stem seems to differ in no respect histologically from the disc,
not even in the greater number of nematocysts; the apex alone constitutes a
batter}^ of stinging cells. The cut ends of a delicate ectodermal muscle
layer can be distinguished in transverse sections of the stems, and a feeble
circular endodermal layer in longitudinal sections. The tentacular muscu-
lature is continuous with that of the disc, and a very definite nerve layer
occurs at the apex of the tentacles (plate 8, fig. 49).

STOMOD^UM.

The stomodseum is a short, thin-walled, depending tube, strongly ciliated
all round, and without any permanent ridges or grooves. Usually it is
oval in transverse section, and along its endodermal surface six pairs of
mesenteries are attached at equal distances apart. Histologically it is of
the same structure all the way round, there being no modification opposite
the directive mesenteries such as can be regarded as a siphonoglyph or
gonidial groove. This is also the case in all madreporarian polyps yet
described.

24 SIDERASTREA RADIANS.

The absence of a siphonogl3'ph in Madreporaria affords an interesting con-
trast with the one or two grooves usually present in actinian and alcyonarian
polj^ps. A siphonoglyph is generally wanting only in the lowest actinians
and in alcyonarians, and its absence in coral polyps would suggest their
more primitive nature. This is further borne out by the absence of ciliated
bands or Flimmerstreifen from tlie mesenterial filaments (p. 29).

The stomodaeal ectoderm is thicker than that of the disc, and differs
much from it in structure. Large nematocysts are distributed throughout,
but the main constituents are long, narrow, ciliated supporting cells, the
nuclei of which are arranged in a broad zone. The ciliation, which is uni-
form all round, generally persists in preserved material. Clear mucous
gland cells are not numerous, while here and there a granular gland cell
stands out very prominently by reason of the deeply staining character of
its contents. A distinct nerve layer is recognizable next the mesoglcea, but
no muscular fibrils can be detected. The endoderm is similar to that lining
the upper part of the polypal cavity.

At its lower opening into the polypal cavity the ectoderm of the stonio-
dseum is reflected up the inner or endodermal surface, and then extends
horizontally for a short distance along the free edge and each face of the
perfect mesenteries, becoming continuous with their mesenterial filaments.

MESENTERIES.

The number and arrangement of the mesenteries in mature polyps are
as follows : (a) Six pairs are united with the stomodseum, including two
pairs of directives attached at the opposite extremities ; {5) an alternating
cycle of six pairs, the members of which are free from the stomodaeum
throughout their length ; {c) a variable number of third-cycle mesenteries,
alternating with the two previous cycles, but very rarely, if ever, with the
whole cycle of twelve pairs developed (plate 6, fig. 34). Thus two cycles
of mesenteries are always developed, and a variable number belonging to a
third cycle. They may be represented by the formula 6, 6, x, where x
represents any number from i to 12.

Within each intermesenterial space, whether an entocoele or an exocoele,
an invagination of the polypal wall occurs which separates adjacent mesen-
teries from one another. Except in the uppermost region these septal
invaginations extend centrall}^ further than the mesenteries, while in the
lower parts of the polyp they extend so far as to meet in the middle and fuse,
dividing the gastro-coelomic cavity into distinct chambers, each of which
contains a mesentery (plate 7, figs. 38, 39).

ADULT COLONY. 25

The mesenteries extend from above downward for about two-thirds of
the total length of the polyp, the 3'ounger outer pairs terminating in advance
of the older inner pairs. Below they become very short transversely, and
are only slightly folded at their free extremity. Except in the uppermost
region of the polyp, their peripheral portion is everywhere perforated by the
synapticular bars. Further, they are connected outwardly with the wall of
the polyp only in the upper part of their course ; the connection breaks down
below, and the organs are seen in transverse sections with both their inner
and outer ends free. Thus at an early stage the peripheral portion of the
mesentery undergoes degeneration or resorption much in advance of the more
central part. The mesenteries, in fact, occupy only the middle and upper
part of the gastro-vascular cavity ; the proximal and peripheral regions are
practically destitute of them.

The mesenterial mesogloea is usually narrow, but there is much varia-
tion in the different polyps as to its thickness and hence that of the mesentery
as a whole. In the upper region the mesogloea presents vertical folds for
the support of the retractor muscles (plate 7, fig. 41). Generally the fold-
ings are diffuse and somewhat complicated, and extend along the entire
face of the mesentery ; in other cases they are stronger and more restricted
to the middle area. The oblique musculature is very weak throughout, but
occasionally the fibrils can be seen cut obliquely in transverse sections. As
would be expected, no basilar or parieto-basilar muscles occur. The mesen-
terial endoderm is everywhere richly supplied with Zooxanthellae, and also
with clear gland cells and others containing large granules. As described
in connection with the endoderm of the column wall and disc (p. 22), what
seems to be an intercellular system of mucous spaces occurs in the deeper
parts of the epithelium on both mesenterial face.

The arrangement of the mesenteries may be studied in more detail. As
represented on plate 6, fig. 34, and also in the diagrammatic figure on page 26,
the number of pairs within each of the six primary systems or sextants
varies from one to three. The members of a pair can be easily distinguished
by their similarity of size and by the retractor muscles being on the faces
turned towards each other. In each of the two ventral systems only one
pair of mesenteries (ll) is developed, while in the middle and dorsal systems
two pairs (11, in) are present, with the exception of the right middle system,
which contains three pairs. Where only one pair occurs, it represents one of
the six pairs of first-cycle metacnemes (ll), and this cycle is complete in all the
mature polyps examined. Where two pairs are present, one pair (ill) is shorter
than the other and is a member of the third cycle. It will be observed that

26 SIDERASTREA RADIANS.

in each sextant the third-cycle pair (iii) is situated on the dorsal aspect
of the second-cycle pair [ii). In the right middle sextant, where two pairs of
third-cycle mesenteries occur, a pair is present on both the dorsal and
ventral aspects of the longer pair. It is shown later that the mesenteries
of the first and second cycles are developed in a very definite sequence,
and such would probably be the case for the members of the third cycle were
the polyps isolated and free to develop normally all around. The individual
polyps in a colonial coral like Siderastrea^ however, are so closely arranged

Illb

I

KiG. 3. — Diagram of mesenteries of fig. 34, plate 6 i^cf. figs. I and a, pp. 12, 13).

that their growth after the first two cycles becomes largely influenced by
spatial necessities. Hence, in the particular polyp from which plate 6, fig.
34, is taken, one region, the right middle sextant, has progressed farther
than the others. Throughout the studies little constancy has been found in
the order of development of the members of the third C3'cle. The amount of
variability in the number and disposition of the mesenterial pairs is repeated
in the septa (p. 48). In no two polyps, among a dozen or so studied in
transverse sections, was the mesenterial plan the same for the third cycle ;
also, in none was the full complement of twelve pairs present. The normal
sequence of the mesenteries is more fully described in connection with the
development of the polyp.

ADULT COLONY. 27

The radial length of the mesenteries varies greatly in passing trans-
verse sections in review from the oral to the aboral extremity of a polyp. In
an expanded polyp the complete mesenteries extend from the colnmn wall,
then across the disc, and finally unite with the stomodasum, down which
they pass ; the members of the incomplete cycles, on the other hand, stretch
from the column wall only partly across the disc, and terminate at a greater
or less distance from the middle of the polyp. When the level of the corallum
is reached, all the mesenteries begin to lose their peripheral connection
with the polypal wall, and in transverse sections hang freely, unconnected
either peripherall}- or centrally, except where pierced by a synapticulum.
Even at the level of the stomodasum in partly retracted polyps the mesen-
teries are greatly shortened transversely, so that the peripheral parts of
the septal loculi are empty (plate 6, fig. 34). Where the loculi are inter-
rupted transversely by a synapticulum, a mesentery will sometimes stretch
from an inner chamber of the loculus to the next chamber, but in no instance,
as shown on plate 6, fig. 34, is one found to extend right to the periphery
of an interseptal loculus. Instead of this, most of the peripheral chambers
are empty, the mesenteries having become resorbed. The organs have a
much less radial or peripheral extent on plate 7, fig. 38, resorption having
been continued further than at the level represented by plate 6, fig. 34.
Mesenteries are wholly absent on plate 7, fig. 39, except in one or two
isolated instances. The stages prove conclusively that the mesenteries
extend but a ver}^ short vertical distance within the region of the synapti-
cula, but are longer centrally. The same fact is also well shown on plate 6,
fig. 36, representing the upper part of a tangential section of a polyp.

The descriptions given by Bourne (1887, 1893) of the anatomy of the
synapticulate coral Fungia indicate similar degeneration phenomena of the
mesenteries within the region of the synapticula in this genus. At first both
the primary and secondary mesenteries extend to the very base of the coral-
lum, but afterwards they are confined to the upper moiety of the calice,
though in Fungia no dissepiments are formed which cut off the polyp
from the basal plate. Bourne's figs. 10, 13, and 15 in his first paper show
that the mesenteries very rarely extend along the interseptal loculi all the
way from one synapticular perforation to another, but in practically all
the canals long or short remnants of the mesenteries are adherent to the
synapticular wall, and serve for the attachment of separate bundles of the
longitudinal muscles. In these cases the middle part of the mesentery within
each chamber has been resorbed, leaving only its two extremities. In Sider-
asirea there is nothing corresponding to the mesenterial muscle bundles in

28 SIDERASTREA RADIANS.

the synapticular region whicli Bourne describes for Fungia and Fowler (1888,
p. 9) for Stcphanophyllia jormosissima ; muscular degeneration proceeds along
with that of the mesentery in the West Indian form.

The actual process of resorption of the mesenteries, from the periphery
inwards, can be readily traced. It commences in the peripheral region of the
polyp even as high as the level of the retracted stomodaeum. The part of
the mesenter}' undergoing degeneration narrows and breaks up into separate
fragments ; the mesoglcea is often very thin for some distance, and the mus-
cular fibrils lining it form only an irregular, discontinuous layer. Generally
the portion of the mesentery some little distance from the peripherally fixed
part is the first to pass away, thus leaving an interval in the actual continuity
of the organ. The endodemi may disappear first, and the naked mesogloea
then extends a little beyond and thins out (plate 7, fig. 43), but usually the
epithelium persists the longest. Free fragments of the degenerating tissue
are often found in the interseptal chambers, and other fragments are occa-
sionally seen absorbed by the endoderm lining the chamber.

Mesenterial resorption must of course take place proximally in all corals
which continue to grow upwards as the lower part of the polyp is cut off
below by dissepiments. But in Sidcrastrea the peripheral parts begin to
disappear first, and the disappearance as a whole extends upwards propor-
tionally much further than in most corals. Perhaps this is in some way
connected with the presence of the synapticula in the peripheral region. In
other coral polyps the lower edge of the mesentery appears to be resorbed
more uniformly.

The mesenteries are too small and too completely inclosed within the
interseptal chambers to permit of their being dissected out and examined as
a whole, but by means of serial sections it can easily be seen that the S3map-
ticula actually perforate the mesenteries. In the first place, the complete
loculus shown on plate 6, fig. 33, reveals that the synapticula really per-
forate the walls, and necessarily anything inclosed within the two lining
walls in the same areas. Then certain of the synapticular perforations
included in plate 6, fig. 34, show that a mesentery is continued from each
wall bounding the perforation, passing centrally on the one hand and
peripherally on the other. On following such a mesentery both upward
and downward in serial sections it is found that when the synapticular
perforation disappears, having been followed through its vertical extent, the
two moities of the mesentery again become continuous. Clearly such condi-
tions can result only from the presence of actual fenestrse in the mesentery.
Similar relationships can be also followed in the series of sections represented
on plate 9.

ADULT COLONY. 29

Owing to the early disappearance of the mesenteries aborally and periph-
erally the organs are, at any one time, necessarily perforated for a very
restricted part of their course. Fnrthermore, as a result of the presence
of these synapticular bars, any mesentery must be capable of less retraction
than in species where the whole length of the organ is free to respond to the
action of the retractor muscle.

Desmocytes, continuous with the mesogloea of the lining of the inter-
septal chambers and the attached mesentery, are usually numerous around
the perforations of the synapticula (plate 7, fig. 41, and plate 8, fig. 45).

MESENTERIAL FILAMENTS.

Filaments occur on the mesenteries of all three cycles, though usually
they are imperfectly developed on the members of the third. On the incom-
plete mesenteries the organs commence a short distance above the lower end
of the stomodaeum, and are continued throughout the vertical extent of the
mesentery ; on the complete mesenteries they start from the lower termina-
tion of the stomodaeum, the ectoderm of the latter being continuous with the
filament. As in all Madreporaria yet described the filaments consist of only
a median lobe, supported basally on an expanded mesoglceal axis (plate 7,
fig. 42). Trilobed mesenterial filaments having lateral lobes supported on
a mesoglceal axis and bearing ciliated bands (Flimmerstreifen), like those
found in most Actiniaria, are not known to occur in corals.

The fully developed filaments of 5. radians are sharply separated from
the mesenterial endoderm by a well marked constriction on each side. The
endoderm is usually slightly swollen immediately behind the filament, but
rarely assumes the form of a definite lobe as in many other coral species. In
addition, the outline of the filament in section varies somewhat in the case of
the first-cycle mesenteries. At first it is cordate, and, histologically, is quite
uniform all round; but soon it becomes nearly circular, and the cellular
constituents of the middle part differ from those of the lateral {cf. plate 7,
figs. 42, 43).

The cellular constituents of the filaments are mostly ciliated supporting
cells, but with these are mingled large, clear, and granular gland cells and
nematocyst bearing cells, especially in the middle region of the sections.
Towards the sides and posterior borders the cells diminish in height and
are nearly all supporting cells (plate 7, fig. 42). At least two kinds of
nematocysts occur — a long, narrow, thick-walled form, similar to that in the
tentacular knobs, and a large, oval, thin-walled form, with the spiral thread
strongly marked (plate 7, fig. 44).

30 SIDERASTREA RADIANS.

The free end of the mesenteries and, consequently, the filaments also are
somewhat exceptional among corals in the small degree to which they are
convoluted in the lower regions. In most cases they pass vertically in a
straight course from their upper to their lower extremities without any folding.
Upon irritation the living polyps of most corals are able to extrude, through
temporary apertures in the body-wall, masses of contorted filaments along
with the portion of the mesentery' to which they are attached, but this
phenomenon has been observed on only one occasion in Siderastrea.

The question of the ectodermal or endodermal origin of the filaments is
alluded to in the description of the larva.

SKELETOTROPHIC TISSUES.

The skeletotrophic or skeletogenic tissues constitute by far the greatest
proportion of the soft parts of the polyps, including as they do the lining of
the calicinal wall, septa, and columella throughout. They represent the
original flat basal disc of the pol3'p, which has become greatly infolded in
correspondence with the skeletal ingrowths. Both the endoderm and ecto-
derm present certain structural peculiarities compared with the same la37ers
in the column wall and disc, and in some respects the calicoblast layer of
S. radians differs from that in most other corals.

The polypal wall lining the uppermost parts of the skeleton is extremely
delicate, especially over the edges of the septa. In sections the combined
ectoderm and endoderm vary from 0.015 to 0.003 mm. in thickness, both layers
being about equal (plate 8, fig. 47). The endoderm is a syncytium showing
no signs of cellular divisions, and the C3'toplasm has at first few or no
vacuoles, though below they become somewhat numerous. The contained
nuclei are large, round or oval, and closely arranged in a very regular row.
Zooxanthellse also occur, their diameter being often equal to the thickness
of the layer. Irregular mucous spaces, similar to those described in the
endoderm (p. 22), are also seen, except where the layer is at its thinnest.

The mesoglcea, both here and elsewhere throughout the skeletogenic
tissues, is rarely distinguishable as a distinct layer, but constitutes a mere
line of division between the ectoderm and endoderm. Where the mesoglcea of
the mesenteries is united with the skeletal covering the calicoblast ectoderm
is nearly absent, and the mesoglcea may then become swollen and continued
into the well-known structures which have been tenned desmocytes by
Bourne (1899). '^^^ combined mesenterial and skeletotrophic mesoglcea
here broadens in a fan-like manner, is striated, and stains more deeply than
usual (plate 8, fig. 45). Sometimes the desmocytes appear as more distinct

ADULT COLONY. 3 1

triangular or wedge-shaped structures, with the narrow apex towards the
mesogloea.

The function of the desmocytes or desmoidal processes, so closely asso-
ciated with the mesogloea, is considered to be that of attaching the soft
polypal tissues to the hard corallum, and they are always most numerous
along the line of union of the mesenteries with the basal wall. As the
mesenteries in Siderastrea are attached to the skeletogenic tissues only
in the uppermost part of the polyp, desmocytes are developed somewhat
sparingly ; furthermore, they are rarely found over regions where the
mesenteries are not attached. In such areas a broad calicoblast layer
usually intervenes between the mesogloea and the skeleton. Where the
skeletogenic tissues, including the mesenteries, are perforated by the synap-
ticula, desmocytes are usually present along the line of attachment of the
mesentery.

Bourne (1899) has worked out in a very masterly manner the origin of
the desmoidal processes from the single calicoblast cells. These cells, which
are strictly the desmocytes, later become connected with the mesoglcea, and
the structure takes on a feebly striated character. Although a careful search
has been made in sections of Siderastrea prepared in various ways, I have
found no certain stages in the formation of the desmoidal processes. Cells
of various forms occur within the calicoblast layer in regions where desmoidal
processes are already present, or may be expected, but none which could be
determined with certainty as desmocytes. Anyone familiar with the details
of madreporarian histology will appreciate the credit due to Bourne for work-
ing out the development of these structures with such a degree of complete-
ness. In their fully formed condition they are undoubtedly continuations of
the mesoglcea, and it is remarkable that they should have originated inde-
pendently and later come into fusion with the middle layer. Gardiner states
(1902, p. 138) that in the polyps of Flabellum riibrnyti the first appearance of
any desmocyte could be seen in a granular mass of protoplasm against the
corallum, to which, from the first, it seemed to be attached. Subsequently,
by growth inwards, it joins the structureless lamella, which may be thickened
so as to meet it.

The actual calicoblast layer of S. radians^ like the endoderm, is at first
very narrow, and in the growing areas of the skeleton shows no evidence of
cell limitations. The c^-toplasm appears continuous, non-granular, and with
or without mucous spaces, the whole staining a bright yellow in picric acid.
The nuclei are nearly as numerous as in the endoderm, and are large and
finely granular. The margin towards the skeleton is usually irregular in

32 SIDERASTREA RADIANS.

outline, often with many adhering particles which stain differently from the
layer itself (plate 8, fig. 47).

Away from the actual growing apices of the skeleton, both downward
and peripherally, the skeleton-lining tissues at once begin to undergo a
marked alteration ; both ectoderm and endoderm become greatly thickened
until they are three or four times their former size (plate 8, figs. 45-48).
The thickened endoderm is reticular, with many oval nuclei arranged in a
regular zone towards the margin ; Zooxanthellae are also numerous, and dis-
tinct mucous accumulations appear.

Vacuoles extend nearly across the layer, and usually appear clear and
perfectly transparent, but when stained with iron hsematoxyliu they present
a grayish reticulum. They are evidently mucous iu character, and would
seem to represent individual cells. In sections stained with Delafield's
haematoxylin and picric acid the vacuolar spaces stand out conspicuously,
the boundaries and reticulum being stained a dense blue, as with mucous
cells generally, and appearing as if enclosed in a yellow granular matrix.
Usually the mucous spaces do not extend as far as the outer margin of the
layer, and not always to the mesogloeal boundary.

The general cytoplasm of the calicoblast ectoderm is finely granular
throughout and does not stain readily except in blue-de-Lyon, which colors
the particles a bright blue. The nuclei are comparatively small and less
numerous than in the upper part of the layer, or in the endoderm. In
addition to the granular, matrix-like cytoplasm, cells with coarse granules
occur here and there which stain differently from the other granules and
appear to be altogether distinct structural elements. In sections stained
with carmine and blue-de-Lyon they stand out a conspicuous red against the
ordinary blue granules. They rarel}^ if ever extend wholly across the layer
and seem comparable with the other coarsely granular cells found elsewhere
throughout the ectoderm and in the mesenterial filaments.

Scattered throughout the calicoblast ectoderm are also a few small oval
nematocysts with a close spiral thread. Sometimes several may occur
together, but usually they are found singly. Different stages in their devel-
opment also can be observed. In the earlier stages they are large, with
homogeneous contents, and stain rather deeply, whereas when mature they
stain feebly, if at all.

The presence of nematocysts within the calicoblast layer, where appar-
ently they can be of no service to the polyp, is somewhat remarkable. It is
to be borne in mind, however, that the calicoblast layer is but a modified part
of the ectoderm which, over the column wall and oral disc, always bears

ADULT COLONY. 33

stinging cells ; and further, the C3'sts are usually present in numbers at
the aboral extremity of the larva which becomes the basal disc when the
larva settles. So far as can be made out from sections, without isolation by
maceration, the nematocysts in the calicoblast layer are similar to those in
the ectoderm of the column wall. Bourne (1899), in his studies of the
calicoblast layer of the Madreporaria, also encountered oval bodies which
suggested to him degenerate nematocysts. The close spiral thread found
in those of Siderastrea^ and their general behavior at different stages towards
reagents, leave no doubt as to their true nature.

The structure of the calicoblast ectoderm, as above given, differs some-
what from that usually met with in corals. As described in the papers of
Bourne, Fowler, and Gardiner, the layer is thickest in the regions of active
growth, and becomes very thin or uearly disappears elsewhere. Such is also
its condition in most of the species of corals examined by me (1902, p. 482).
It may appear either as a columnar epithelium [Madrepora) or have the
character of a syncytium. A well-developed cellular tissue, with abundant
protoplasm, would be expected where active formation of the skeleton is in
progress. Here, in Siderasfrea, the layer is thickest in those regions where
the secretory activity would be considered to be at a minimum. The granular
character of the cytoplasm, the small size of its nuclei, and the strong
vacuolization would not, however, suggest that the deposition of skeletal
matter was actively proceeding in the lower regions, as compared with the
large nuclei and more hyaline c3'toplasm along the edges of the septa.

It seems doubtful as to how far the thinness of the endoderm in the upper
region is to be associated with a similar condition of the calicoblast endoderm.
Over the spines of the columella, which are situated in the lower regions of
the polypal cavity, and are probably in a growing state, the two layers pre-
sent a great difference in this respect. The ectoderm is as thin as in the
uppermost growing parts of the polyp, while the endoderm is greatly thick-
ened, measuring as much as 0.05 mm. across, which is nearly equal to the
combined thicknesses of both ectoderm and endoderm in the lower regions.

None of the writers on the anatomy of coral polyps have referred to the
marked alteration undergone by the skeletotrophic endoderm in passing from
the upper to the lower parts of the polyp. It is, however, characteristic of
practically all West Indian corals, the contrast being usually greater than
that shown hy Siderastrea.

In his study of the minute anatomy of the polyps of Flabellum rubrum^
Gardiner (1902) found the calicoblastic ectoderm to be in many respects
like that of .S". radians. The layer is everywhere complete and persistent.

34 SIDERASTREA RADIANS.

even in the most roughly decalcified specimens, and no definite cell limita-
tions could be distinguished ; but over the greater part of the corallum it is
an extremely thin, finely granular layer, and as the edges of the septa are
approached the layer thickens, nuclei become more frequent, and tend to
exhibit a definite network. Where secretion may be supposed to be going
on most actively the layer is more hyaline, elsewhere it is granular. The
protoplasm is much vacuolated, as in Siderastrea^ except where secretory
activity prevails.

Where decalcification has been carefully carried out, a layer of the
organic matrix within which the skeleton is deposited usually remains
behind, and is closely adherent to the calicoblast layer, as shown at the left
side of plate 8, fig. 45. The matrix usually appears perfectly homogeneous,
very variable in width, and recalls the mesoglcea in its behavior towards
stains. Bourne (1899) found such a layer in many of the forms examined
by him, and regards it as a fine membrane — " limiting membrane" — separat-
ing the calicoblasts from the corallum, and comparable with the sheath which
incloses the spicules of the Alcyonaria. It is evidently a secretion of the
ectodermal layer, and at the apical points of rapidly growing corals, such as
Madrepora, I have found the deposit to be continuous throughout the whole
thickness of the corallum, and presenting just beyond its border the fibrous
scale-like appearance of the early skeleton (1902, p. 484). It is manifest that
the mass should be regarded as a homogeneous, mesogloea-like matrix within
which the minute calcareous crystals forming the skeleton are laid down, to
be compared with the matrix which Bourne has demonstrated for the skeletal
spicules of Alcyonaria. As a continuous structure it early disappears, and
in the older parts of the skeleton is represented merely by the fine organic
particles remaining after decalcification. Only under the most favorable
circumstances is it present as a continuous mass in sections of the newest
formed parts. Usually all that remains of it after decalcification is the deli-
cate membrane bordering the calicoblast layer on its skeletal aspect. When
of sufficient thickness to have contained skeletal fibers, now dissolved away,
this membrane appears fibrous, but immediately bordering the calicoblast
layer it is homogeneous.

SEPTAL INVAGINATIONS, INTERSEPTAL LOCULI, AND GASTRO-CCELOMIC CAVITY.

The septal invaginations {refoulemenis septaux of Delage & Herouard)
are the vertical, somewhat wedge-shaped infoldings or upgrowths of the
basal wall which in the living condition cover both sides of the septa,
and have been produced pari passu with them. Their presence results in

ADULT COLONY. 35

much subdivision and complexity of the lower part of the polypal cavity, as
compared with that of an actiuiau, where the basal disc retains its primitive
flatness. In the coral polyp radial parts of the basal disc are pushed vertically
upwards into the polypal cavity, concurrently with the deposition of calca-
reous matter, so that some portions of the aboral wall are of the height of
the septa. The earliest stages in the formation of the basal infoldings over
the septa are shown on plate 9, fig. 53. In the later processes of growth
the lower edge of the column wall is also carried upwards to about the same
height as the septal invagination,* hence it results that the basal wall is
represented by so many vertical lamellar infoldings, the column wall and
disc appearing as a mere superficial covering to these.

In transverse sections the invaginations are found to occur between
every two mesenteries, that is, they are both entocoelic and exocoelic in
position, the two series being equal in number, the exocoelic extending less
centrally than the others (plate 6, fig. 34). In the upper region of retracted
polyps the invaginations do not extend as far centrall}' as the primary mesen-
teries, though in the stomodasal region they are continued beyond the secondary
and tertiary mesenteries. As the lower regions are approached the foldings
stretch centrally beyond all the mesenteries until, in the end, the walls of
adjacent invaginations unite with one another and thus completely inclose
each mesentery in a separate interseptal loculus, the skeletal deposit being
also continuous from the periphery to the middle (plate 7, fig. 39).

As shown in the various figures, the septal invaginations do not remain
truly radial. At their inner extremities the exocoelic members turn laterally
to unite with an entocoelic invagination, and the entocoelic of the third cycle
are united with the entocoelic of the second. In this way continuity of the
invaginated walls in transverse section is broken, and the skeletal matter
within one invagination is likewise continuous with that in another. On plate
7, fig. 39, it is seen that some of the septa are continued radially as far as
the central columella, while the others extend centrally only indirectly by
fusing with the first. As a result of these many fusions the cyclic plan of

* In many corals, e. g., Cladocore, Manicina, the proximal margin of the column wall is lower than
the upper edge of the septa, and the latter are generally united peripherally in the thecal wall; the septa
have grown upwards into the polypal cavity, leaving the lower boundary of the column wall behind.
The mesenteries still remain attached to the outer column wall, but internally they have been
pushed upwards, as it were, along with the upward growth of the thecal wall. Owing to this, parts of
the polypal cavity and mesenteries are peripheral or outside the calice (^extracalicular or pericalicular)
and the remainder is within the calice (intracalicular). The portion of the polj'p outside and around the
calice is what is known as the "edge-zone" or "Randplatte," but, as noticed on p. 9, there is no edge-
zone in Siderastrea, the polypal cavity and mesenteries being wholly intracalicular when the polyps
are retracted.

36 SIDERASTREA RADIANS.

the invaginations is not readily established ; in fact, it would be almost impos-
sible to determine it without the assistance of the mesenteries. In addition,
complications are produced by the fact that the invaginations are perforated
by the synapticula, and thus, in sections, their walls appear discontinuous.

The gastro-ccelomic cavity in Siderastrea is very complicated in character
compared with that of an actinian polyp, or, indeed, with that of many other
corals. A double series of divisions, mesenterial and septal, are to be con-
sidered, except when the polyp is fully expanded and the column wall and
disc are raised wholly beyond the corallum. In this latter condition the
polypal cavity is merely divided into entocoelic and exocoelic chambers by
the mesenteries, as in an ordinary actinian polyp. Each chamber is
prolonged into a tentacle, but there has been found no lateral commuuica-
tion between one mesenterial chamber and another, such as frequently
occurs in actinians by means of stomata.

The coelomic cavities of the many polyps making up a colony are only
partly independent of one another. At the periphery of each polyp aper-
tures are found between the united edges of the contiguous column walls
and the skeletotrophic tissues, but no apertures or canals elsewhere connect
the different polyps. The communicating spaces are intermesenterial in
position, the polyps being more or less cut oflf from one another mesenterially
(plate 6, fig. 35). The spaces divide the superficial column wall from the
basal disc. Hence in colonial polyps the column wall and base are nowhere
in direct continuity with one another, except at the free part of the marginal
polyps; elsewhere they are connected only indirectly by the mesenteries as
these pass from column wall to basal disc.

When the level of the corallum is reached the polypal cavity is encroached
upon radially by the septal invaginations, and divided peripherally into inter-
septal chambers or loculi, each of which contains a single mesentery. At
first all the chambers communicate with one another toward the middle of
the polyp, which is still free from any calcareous deposit (plate 7, fig. 38).
At the level of the columella the middle region of the polypal cavity is
encroached upon, the septal invaginations extend from the periphery to the
center, and the entire cavity is broken up into distinct loculi (plate 7,
fig. 39). These are continued as far as the aboral end of the polyp, where
each loculus terminates blindly.

In addition, the interseptal spaces of Siderastrea^ as of other fungids, are
encroached upon concentrically by the synapticula which stretch across from
septum to septum, and in sections give the discontinuous canal-like character
to the polypal cavity seen towards the periphery of figs. 38 and 39, on plate 7.

ADULT COLONY. 37

In many corals the endoderm in the lower regions becomes thickened to such
a degree as to occupy the entire interseptal loculi, but in 5. radians a narrow
space remains as far as the aboral termination.

GONADS.

Among all the polyps which have been examined anatomically ova only
have been found to occur. The polyps studied were taken from colonies
other than those which extruded larvse. The gonads are situated below the
stomodseal region, only one ripe ovum as a rule being present on each mesentery
(plate 7, fig. 43). They occur about the middle of the transverse length of
the mesentery, and are found on the members of both first and second orders.
The ova are usually longer in diameter along the radial axis, and are indented
where the septal granules have intruded upon the narrow interseptal chambers.
Along with the single large &gg may be two or three developing ova without
yolk, but the conditions are not favorable for studying their origin from the
ordinar}^ endodermal epithelial cells.

The absence of male sexual elements from the examples studied can by no
means be taken as indicating a dioecious character of the species. Wherever
thus far in West Indian corals sexual differentiation is suggested the gonads
have been found to be female, but where the ova are best developed spermaria
have been found to accompany them ; spermaria have never been found alone
in any polyp. Hence there is much which suggests that coral polyps are pro-
togynous. Mr. Stanley Gardiner, however, has recently discussed the subject
of protandry in corals (Proc. Camb. Phil. Soc, XI, 1902). From an investi-
gation of a large number of developmental stages in Flabellum riibrum he has
been able to show that in this form spermaria arise first on the mesenteries and
that ova appear later, when the production of sperm acini ceases. " The ova
grow enormously, with the final result that the mass becomes entirely female,
consisting of usually 2 or 3 large ova, flattened on their sides against one another
and occupying the whole area of the former testes." Protandry is thus clearly
established in Flabellum rubrum. The claim for protogyny in the West Indian
corals hitherto studied rests upon the fact that spermaria have never been
found alone, but always in association with large numbers of ova; on the
other hand, many polyps have been found with ova alone, often few in number,
as if sexual maturit)' were but beginning.

Were a general protandry to be assumed, we should have to suppose
that all the pol3^ps containing onl}' ova had passed beyond their male
period and extruded all their spermatozoa, becoming wholly female. Prob-
ably, as in other groups of animals and plants, no hard and fast rule is

38 SIDERASTREA RADIANS.

followed by the different species of corals ; some will be protandrous wbile
others will be protogynous.

It is clear from the evidence adduced that a dioecious character can not
be assumed from the presence of only ova or spermaria until the entire
developmental history of the species has been followed ; a polyp having at
one time only one kind of sexual cells may later show both kinds.

CORALLUM.

Freshly macerated coralla are generally white or yellowish in color, but
frequently they are somewhat green, either as a whole or in patches. On
breaking a green corallum across the coloration is found to extend some
little distance below the surface, and, after decalcification, there remains
behind a green mass of fluff}' texture, which, on examination under the
microscope, is found to consist of filamentovis algae. Even where the skeleton
presents no superficial green coloration, a filamentous mass remains behind
on decalcification, and chlorophyll may be present in small quantities, or the
filaments may be dead and altogether devoid of protoplasmic contents. Under
the microscope fragments of the skeleton frequently show fibers crossing the
interseptal spaces or meandering over the surface, and sections of the corallum
are sometimes perforated by similar filaments (plate ii, fig. 66). Thus
the algse actually bore their way into the skeleton as well as spread over the
surface. They are particularly abundant in decaying masses of coral.

The coral-boring algae liberated by decalcification are represented by
both a non-septate and a septate form, the filaments of both var3ang greatly
in size. Associated with the algal filaments are clusters of a Leptotlirix-\-\k&
bacterium ; but whether its threads actually perforate the skeletal matter, or
merely grow over its surface, living upon the traces of organic matter, seems
uncertain. In a former paper I have pointed out the almost universal
presence of boring algse in corals, and their importance in disintegration
(Bull. Amer. Mus. Nat. Hist., xvi, 1902).

In addition to the parasitic algae, organisms such as cirripedes, molluscs,
or tubiculous worms usually occur, either inclosed within the skeleton or
adherent to its surface. These associated organisms are particularly numer-
ous on specimens obtained from muddy shores as compared wth those from
the clearer waters of the reefs, and often interfere with the regularity of form
of the corallites of the colony. The cavities resulting from the activity of
the boring molluscs weaken the skeleton greatly, so that the colonies can be
readily broken into fragments.

Numbers of the parasitic cirripede Pyrgoma sometimes occur, exhibiting

ADULT COLONY. 39

all stages in the growth of its protective shell. The older specimens are
completely inclosed and fused within the corallum, only the mouth of the
shell being exposed at the surface of the colony. Its outer surface is covered
by an irregular growth of the coral skeleton, and, being raised a little above
the general surface of the corallum, gives a marked irregularity to the colony
as a whole. The external ridges on the shell of the parasite have a curious
resemblance to the septa of the coral, and in the later growths it is somewhat
difficult to distinguish one from the other.

On certain macerated colonies many early stages were obtained showing
the manner of attachment of the cirripede within the living calice. At first
the presence of the small shell of the crustacean has produced no modifica-
tion of the coral ; it is simply adherent to a few of the septa. But later the
lower ridges on the surface of the cirripede, which closely simulate septa,
became fused and partly overgrown with the septa, thus causing the oblitera-
tion of the middle of the calice. In every case the intruder has fixed itself
wthin a calicinal cavity, never on the ridge connecting two calices. Appar-
ently the cirripede larva enters the polypal cavity and then bores through
the living tissues to the corallum, and in the process of growth the skeletons
of the two become fused. Afterwards the two organisms continue growing
together, the coral skeleton almost wholly inclosing the crustacean ; the latter
has no boring action. In the later stages the presence of the Pyrgoma
results in the production around it of imperfectly developed polyps.

On breaking through a colony it is frequently found that the corallum
is made up of growths of different periods, arranged upon one another in
irregular laj'ers, instead of a continuous increase in thickness from the middle
to the circumference. An old corallar surface has become dead and partly
corroded, and then a later skeletal deposit has been formed upon this, and so
on for several successive stages, the vertical axis of the corallites of the two
or more periods of growth usually corresponding. The secondary origin of
polyps in old calices has been already described (p. 17). The growth of the
corallum seems to be more continuous in blocks of the larger S. siderea than
in S. radians.

The actual degree of calcification of the corallum, if one may thus express
it, varies much in different colonies, and also in different regions of the same
colony. In some examples the septa are comparatively thick, leaving but
narrow interseptal spaces, and the calices are shallower than usual, the
columella being solid and prominent. In others the septa are thinner, the
interspaces correspondingly wider, the calices deep, and the columella either
absent as a definite projection or represented by one, two, or three prominent

40 SIDERASTREA RADIANS.

granules on the floor of the calice. Probably these diflFerences are determined
by the varj-ing conditions of growth of the poh^ps, whether rapid or slow.
The growing marginal corallites are alwaj's of the less calcified variety.
Such possible variations should be borne in mind, for if extreme coralla only
were available for study they might be almost regarded as distinct species.

Examination of the surface of the corallum, and also of sections, shows
that the individual calices of a colony are separated from one another only by
fusion of the peripheral edges of the septa. There is no thecal formation
distinct from that of the septa. The fusion is complete and continuous, how-
ever, and results in the entire lateral separation of one calicinal cavity from
another. This is clearly shown in the various figures of the corallum on
plate lo ; and, as further proof, it may be recalled that on decalcification the
tissues of the different polyps so far as they are inclosed within the corallum
are wholly cut off from one another.

The septa of adjacent corallites correspond peripherally, end to end, or
they may alternate ; and between these two extremes are all intermediate
conditions (plate lo, figs. 62, 64). At the peripher}^ the thickness of the septa is
usually greater than the width of the interseptal spaces, so that where the
septa of adjacent corallites alternate it is clear that a distinct wall of separation
is produced between one calice and another ; but where adjacent septa are end
to end a shallow interval may remain on each side of the two continuous
septa, and the septa are then feebly exsert.

The exsert condition of the septa is most marked toward the margin of
colonies, that is, in the region of new growth. Elsewhere the intervals are
generally filled by calcareous deposits, so as to bring the calicinal edge to
the same level as the septa. The usual superficial appearance is as if each
septum became bifurcated at its outer edge and then each half united with
a similar half of two adjacent septa belonging to a contiguous corallite.
The result is a very narrow zig-zag partition wall between adjacent corallites.
Usually the calicinal boundary as a whole reaches the same height as
the septa, but occasionally it is depressed, and the septal edges are inclined
towards it.

HISTOLOGY.

The different parts of the corallite will be now described, but first the
microscopic structure of the skeleton must be briefly considered. In connec-
tion therewith the writer has to express his indebtedness to the valuable
work of Miss Maria M. Ogilvie (Mrs. Ogilvie-Gordon) who, in her paper,
Microscopic and Systematic Study of Madreporarian Types of Corals (1S97),
has done more than any other investigator to further the histological study

ADUI,T COI^ONY. 41

of the madreporarian skeleton, following along lines initiated by Pratz (Palse-
ontographica, vol. xxix). Mr. T. Wayland Vauglian has given a resiime
of the subject in the introduction to his paper on The Bocene and Lower
Oligocene Coral Faunas of the United States (1900).

When the larger septa of Siderastrea are examined sideways, under a
low magnification, they present the appearance of the three septa shown on
plate 10, fig. 63. The peripheral vertical boundary of each septum consists
of a narrow, continuous, nearly straight ridge, which is the broken surface
of the thecal wall, really a part of the adjacent septum. Next this is the
synapticular area, the synapticula being' arranged rather closely in several
vertical rows. The surface as a whole is striate, becoming distinctly ridged
and grooved as the margin is approached. The septum terminates centrally
and above in a strongly serrated margin, the teeth being almost conical in
shape, continuous with the ridges, and varying but little in size. Minute
granules, usually terminating in one or more sharp points, occur over the
whole surface, including the synapticular area and even the marginal spines.

On septa of the second and third orders may also be seen a vertical row
of synapticulum-like structures near their inner border. They are shown on
the right (central) margin of the left septum on plate 10, fig. 63, and represent
the line of fusion or coalescence of the next adjacent septum in the series —
the third with the second, and the fourth with the third. They do not occur
on septa of the first order, as none of the other septa fuse with these {cf. plate
10, fig. 64). The disconnected character of the bodies serves to demonstrate
that the central union of one septum with another by the internal margin is
not continuous, but interrupted in character. It is best regarded as taking
place along the spinous or toothed edge of the smaller septum, in the same
way as the complete septa of the first and second orders fuse in an interrupted
manner with the columella b}' their spinous margins. In Siderastrea Miss
Ogilvie (1897, p. 179) regards the direct lateral coalescence of the septal
surfaces belonging to different septal cycles as homologous with the synap-
ticular union of septa, but it may be pointed out that the fusion is altogether
on the part of one of the septa, not a simultaneous growth from two adjacent
septa, as in synapticula proper. The interrupted coalescence of the septa,
both with one another and with the columella, necessarily leads to the pro-
duction of foramina near their inner edge, some of which are shown to the
left in plate 10, fig. 63.

The septal striae are directed upward and inward in a half fan-shaped
manner, starting from the vertical thecal wall. Those on adjacent septa
belonging to contiguous calices radiate in an opposite manner, each towards

42 SIDERASTREA RADIANS.

the center of its own calice (plate lo, fig. 63). The thecal ridge is thus the
area of divergence of the striae of adjacent septa. It is further found that
the striae on opposite sides of the same septum correspond with one another,
as can be seen by looking edgewise at the inner margin of an individual
septum ; hence opposite striae terminate in one sen-a. The synapticula and
granules are set mainly along the ridges of the septal face, but the former
may be so broad as to extend over more than one stria.

In many corals lighter lines run transversely across the striae, parallel
to the dentate septal edge, and mark successive septal margins, corresponding
to distinct periods of growth in the life of the polyp. The part between two
successive curves represents the addition made to the septum during a single
period of polypal growth, and is called by Miss Ogilvie (p. no) "the growth-
segment of the septum " or " a septal segment." The septa of Siderastrea
very rarely exhibit such lighter cross lines ; by Miss Ogihae the synapticula
are considered an essential part of every growth segment in this genus.

A magnified transverse section through a portion of the corallum shows
that the septa, columella, and synapticula are built up of the same structural
elements or units (plate 10, fig. 65). These appear in sections as so many
distinctly round or polygonal bodies, fused together into a compact continuous
whole. Each unit of structure under low power reveals lighter and darker
parts arranged in a concentric manner, and under high power is found to be
made up of radiating calcareous fibers (plate 11, fig. 66). The fibers seem
to start some little distance from a center, which is either clear or dark, and
practically homogeneous in structure ; further, they are seen to be arranged
in close concentric lamellae, and the boundary between each two layers is
often indicated by dark, finely granular particles, similar to those at the
centers of calcification.

Employing the terminology of Miss Ogilvie, each structural unit is a
irabecula, here seen in transverse section. The radiating bundles of fibers
s.r& fascicles, the light or dark centers from which the fibers radiate are centers
of calcification, and the concentric layers &re grozcth lamell(s.

In transverse sections of the septa of Siderastrea the centers of calcifica-
tion are some distance from one another, and each bears radiating calcareous
fibers all round ; but in the majority of corals the centers are closer, and
often appear in sections of the septa as a more or less continuous dark line
or band along the septal axis — the dark line of calcification. In this case the
separate fascicles or bundles of fibers are more distinct, and arranged in a
feather-like manner along each side of the dark line or layer.

Much discussion has taken place with regard to the nature of the center

ADULT COLONY. 43

or dark line of calcification. Writers on fossil corals particularly have devoted
their attention to the subject. For a complete account the reader is referred
to various sections of Aliss Ogilvie's work. By some authors the median
dark laj'er has been interpreted as 2i primary septiivi^ upon the faces of which
the lighter layers, then termed stereoplasm^ have been deposited, the two
formations making up the septum proper. Miss Ogilvie considers that the
dark appearance of the centers results from the presence of the carbonized
residue of the originally unchanged parts of the calicoblasts within which
she considers the madreporarian skeleton to be formed. It must be stated,
however, that the dark appearance is only seen when sections are viewed by
transmitted light. With reflected light the middle region appears lighter
than the rest of the septum, and thus can scarcely be occupied by black
organic matter.

Undoubtedly the middle line of the septum is often of a different struc-
tural nature from the two sides, and varies much in character in different
species. In Paleozoic corals I have occasionally found the middle part to be
dissolved away, making the individual septum appear as if formed of two
distinct lamellae, more or less completely separated. Furthermore, in the
very young septa represented on plate 11, fig. 70, the center of each trabecula
is clearly seen, under high power, to be devoid of any inorganic matter. In
the rather thick section one can focus down each center for some distance,
which would certainly not be the case were it occupied by a calcareous
deposit. I conceive that the so-called center of calcification is really the
organic center or axis around which the skeletal matter is deposited in a
radiating or feather-like manner, and that, at an early stage in the living,
growing skeleton, the center is occupied by the mesoglcea-like matrix within
which it has been shown that the calcareous fibro-crystals are deposited (p. 34).
The organic matrix of the centers is probabl}' in all cases impregnated
later with calcareous matter, differing in character and crystallographic
orientation from the true skeleton (stereoplasm), as shown by its dii^erent
appearance in sections and different solubility. Bourne has remarked (1899,
p. 539) : " I suspect that the dark ' centers of calcification ' will be found to be
the expression of a core of organic filaments, just as the central dark line in
the Alc3'onarian spicule is the expression of the central core of threads."
Results published since Bourne's paper seem to confirm his conjecture, only
the core is probabl}' homogeneous, and later becomes impregnated with inor-
ganic matter, probably from solution, not from calicoblastic acti\'ity.

The centers of calcification in S. radians are distant from one another
on an average about 0.125 mm. and the trabeculse show a corresponding

44 SIDERASTREA RADIANS.

diameter. The latter are very variable in size in the columella (plate lo,
fig. 65). Sometimes two centers of calcification, diverging from one another
in the median plane, are present in one concentric system, and the trabeciila
then appears double and is elongated in outline. Fusion between adjacent
trabeculas is usually indicated bj' a narrow, dark line.

A synapticulum may have either a separate center of calcification (true
synapticulum) or be formed from the fascicles of trabeculse in adjacent septa
which extend beyond the septa until they meet in the middle of the inter-
septal space (false synapticulum, p. 52). Bundles of fibers or fascicles from
individual centers also pass into the granulations over each face of the septa.

The growth lamellae in the individual trabeculae are not always clearly
indicated, but sometimes they come out with remarkable distinctness, having
a black, finely granular margin (plate 11, figs. 66, 68). Around the centers
of calcification they are strictly concentric, but towards the edges of the septa
they more nearly follow the septal outline. In transverse sections they are
distant from one another about 0.0025 mm. The black, finely granular
margin is probably organic residue of the same character as that at the
centers of the trabeculae.

A thin vertical section of part of two contiguous septa is represented on
plate II, fig. 67, and displays a series of elongated bands. One of these, the
continuous median band, represents the thecal wall, and from it the con-
stituent parts of each septum diverge. The darker circular or oval areas also
seen in the section are the transversely cut surfaces of synapticula, and each
originates in connection with a band.

The broad radiating bands are the septal trabeculae as they appear in
longitudinal section. Where the section passes through the middle of a
trabecula a row of close centers of calcification is revealed, and fascicles of
fibers pass from them obliquely upwards on both sides. Peripherally the
fibers of adjacent trabeculse meet all the way in a distinct thin line, so that a
compact septum is the result. Fresh trabeculae are seen to be inserted at
intervals between the old. The trabeculse are continued upward to the septal
margin and there constitute the terminal teeth, each tooth corresponding with
a single trabecula, but the preparation from which plate 11, fig. 67, is taken
was not sufiiciently wide to show the teeth. They vary but slightly in width,
being about 0.125 mm. across towards the margin of the septum. From the
inclined arrangement of the trabeculae it is obvious that a transverse section
of the septum will cut only the more peripheral at right angles ; the others
will be represented in varying degrees of obliquity.

Each separate dark area along the middle of a trabecula represents a

ADULT COLONY. 45

center of calcification, and from it fascicles of fibers radiate on both sides, tbe
black, finely granular matter extending but a very short distance. Each
center of calcification, therefore, represents a single growth period in the
upward progress of the septum. They are distant from one another about
0.013 mm. A trabecula is thus a vertical series of fibrous groups, each group
of fibers being deposited around a distinct axis of calcification. A trabecular
part is the name given to the separate groups of fibers formed by the fasci-
cles at one center of calcification. Bach stria seen under low power on the
surface of a septum represents a single trabecula, and the tooth in which an
opposite pair of striae terminates at the edge of the septum is the growing
apex or organic center of the trabecula. Around this organic center new
calcareous matter is constantly being added by the activity of the calicoblasts.

WALL OR THECA.

The wall or theca of a simple coral or of a corallite is defined by Vaughan
(1900, p. 48) as "that part of the skeleton that cuts oflf more or less com-
pletely the interseptal loculus from peripheral communication with the
outside." In Siderastrea the thecal wall at first appears indistinct, as if
represented only by the united peripheral edges of the septa of adjacent
corallites. Some of the septa of contiguous calices are in the same straight
line, in which case it is difficult to distinguish just where the calicinal
boundary comes ; in other places a septum of one corallite corresponds with
an interseptal loculus of the other, when the distinction between the two
calices is quite clear. Usually it appears as if the septa were forked
peripherally, and the two limbs of one septum unite in a zigzag manner
with two adjacent septa in the next calice.

The nature of the wall is best seen when a corallum is fractured verti-
callj'', and adjacent corallites can be viewed lengthways. As indicated on
plate 10, fig. 63, the theca is then represented by a narrow, vertical, continuous
ridge, which completely separates the interseptal loculus of one calice from
that of the other. Further, in the section shown on plate 11, fig. 67, this
vertical ridge is found to consist of a distinct trabecula, and from it the
trabeculae of adjacent septa diverge.

Miss Ogilvie (1897, p. iSo) has given a figure of Siderastrea (sp. ?)
somewhat similar to that of plate 10, fig. 63, but the thecal wall (pseudotheca)
is there represented as a discontinuous ridge, as if comparable with a row of
synapticula, and in such a way as would permit of communication between
one calice and another. This, however, is not its condition in S. radians.
In all instances I have found the wall to be continuous, and the fact that

46 SIDERASTREA RADIANS.

when a colony is decalcified adj acent polyps are found altogether cut off from
one another would prove that their cavities are not in lateral communication,
as would be the case were the boundary wall interrupted.

The terms psc lido/ krca and eiitheca of von Heider (1886) are well known
to coral students, the former referring to a theca produced by the septa
becoming so much thickened peripherally that they fuse together, while the
latter is applied to a theca formed from separate trabeculse between the
peripheral ends of the septa. Vaughan (1900, pp. 48-52) has considered
very fully the value to be applied to these distinctions, and comes to the
conclusion that " From the great variation not only in the same species, but
in a section of a single corallite, no special systematic importance can be
attached to the theca being of the so-called true or false variety. True theca
marks the Anlagen of new septa, as von Koch and Bourne have shown, or
occurs in calices where the septa are distant from one another and their outer
ends are not sufiiciently thickened to effect peripheral fusion."

The theca of Siderastrea is described by Miss Ogihae as a pseudotheca,
and, as above shown, it is certainly formed at the peripheral extremities of the
septa. On the other hand one, or rarely more, distinct centers of calcification
occur laterall}' in the interval between one septum and another, though not
to be distinguished from those of the septa except by being inclined to them at
an angle. In her figs. 43(2, 43*5 (p. 181), Miss Ogilvie represents such centers
of calcification, but regards them as synapticular. Although presenting the
same appearance in transverse sections, plate 10, fig. 63, and plate 11, fig. 67,
show that this is not their true character ; rather, the thecal trabeculse are
continuous vertical partitions. According to the accepted definitions it would
seem that the theca of Siderastrea should be regarded as a true theca, having
a separate trabecula between the ends of adjacent septa from which the tra-
beculas of the septa radiate.

An independent thecal wall, however, is found to be wanting in the
corallum of larval polyps, the epitheca being in no way comparable with this.
In the simple corallum, as far as reared, the septa remain free and exposed
peripherally, except in so far as they may be imited by synapticular growths
or covered by the epitheca (see plates 4 and 5). Likewise the outer septal
edges of the marginal polyps in a colony extend for a considerable vertical
height, exposed all the waj^.

SEPTA.

The detailed characteristics of the septa have been already described, so
that it only remains to discuss their arrangement and relationships.

ADULT COLONY. 47

All the septa of a corallite are well developed, both as regards their
thickness and radial extent ; even the members of the outermost cycle extend
centrally for about half the radius of the calice, and are but little narrower
than the others. The individual septa are usually thicker than the width
of the interseptal loculi separating one from another, so that, compared with
most other corals, the septal system is very compact, and the calicinal cavity
as a whole is correspondingly diminished. The septa vary but slightly in
thickness according to the cycle in which they belong.

The outer septa are not strictly radial throughout their transverse
length, but incline along their inner edges and fuse at intervals with the
lateral faces of the septa of the second and third cycles, thus forming groups
of three, five, or rarely seven (fig. 2, p. 13). The grouping, however, is not
alwaj's obvious at the surface of a colon}-. Thus Verrill (1901, p. 154) states
that the septa of the Bahama representatives are nearly all straight and

ixiuxnxi .iXLiixuxiiLXt

Fig. 4. — Grouping of the septa within a sextant (i-i) according as one, three, five, or seven septa are present.
The Roman numerals l-lll indicate the order of the septa and x the exosepta.

seldom fused. It is, however, alwa3's a marked characteristic in sections of
the corallum from a little below the surface downward (plate 10, fig. 64).

When examined in detail the grouping is found to be as follows : The
twelve largest septa, constituting the primary and secondary orders, extend as
far as the middle of the calice and are united directly with the columella.
Of these the six primary septa pass uninteiTuptedly all the way from the
periphery to the columella without any connection with the others except, of
course, through union by synapticula. The six secondary septa, on the other
hand, are also radial, but with them are united the members of the third
order, which never reach the columella. The members of the fourth or outer-
most cycle fuse one on each side with a tertiary septum. Where only one
secondary septum occurs within a primary sy.stem the two outermost septa are
fused with it, and a grouping of three is prodi:ced. Where a tertiary septum
and a secondary septum are within a sextant the grouping is in fives, and
where two tertiary septa and a secondary occur the grouping is in sevens.
The relationships are clearly shown in the series of figures above and also in
fig. 2, p. 13.

48 SIDERASTREA RADIANS.

In the course of development of the septa it is shown that from the
beginning the exosepta always constitute the outermost cycle, and are fused
with the entosepta immediately preceding them. Thus, when only two cycles
of septa are present (plate 4, fig. 23), the outer exosepta fuse with the inner
entosepta. Where three cycles are developed (plate 5, fig. 28) the exosepta
unite with second-cycle entosepta ; where four cycles, they fuse with the third-
cycle entosepta. At the early stage of development, represented in plate 5,
fig. 28, the second-cycle entosepta are fused with the first-cycle entosepta
instead of with the columella as in the adult. The union here takes place
in such a manner as to give a decided bilateral symmetry to the corallum.

The hexameral cyclic plan of the septa is not so manifest in S. radians
as in many corals, partly owing to the incompletion of the third and fourth
cycles ; but with a little care it can be established by the differences in size
of the septa and their other relationships. In mature corallites two inner
alternating orders (primary and secondary) of six septa each are always
present, while the third and fourth orders are represented by a varjnng num-
ber of septa. Owing to this varying incompleteness of the last two orders a
study of the septal arrangement yields some interesting results in connection
with the manner in which the later additions of septa take place, and also in
the relationships of the entosepta and exosepta.

The hexameral plan is found to be strictly followed so far as concerns
the first and second orders, and for the third and fourth orders so far as these
are developed. Were the third and fourth orders to be completed the total
number would be 48, arranged in the formula 6, 6, 12, 24; but this number
appears to be rarely, if ever, reached in S. radians, though frequently
exceeded in S. sidcrea. In one of the largest corallites 44 septa were present,
but the numbers generally vary from 30 to 36.

The six members of the primarj^ cycle are readily separable from the
others on account of their greater radial extent and thickness ; they pass all
the way from the thecal wall to the columella without any of the other septa
coalescing with them. They serve to distinguish the six primary systems
within which are enclosed varying numbers of smaller septa. The outermost
exosepta are also recognizable by their smaller size, one occurring between
every two adjacent members of the first, second, and third orders. Less cer-
tainty prevails in distinguishing the members of the second and third cycles,
though frequently the second-cycle septa are larger and more prominent than
those of the third.

In mature corallites three septa at least are alwaj^s present within each
primary system, while there may be five or seven (fig. 2, p. 13). Where only

ADULT COLONY. 49

three septa occur in a S3'stem the middle member belongs to the second order,
and the two lateral ones are exoccelic; where five septa are present, one
belongs to the second order, another to the third, and the three alternating
with these to the exoccelic cycle ; in the rarer instances in which seven
septa are developed the middle member is a second-order septum, the two
next in size are tertiary septa, and the four alternating septa belong to the
fourth or exoccelic cycle. The members of the outermost cycle fuse in an
intennipted manner by their inner ends with the members of the second
and third cycles, but not with those of the primary cycle ; the tertiary septa
in their turn unite with the secondary.

The actual ordinal relationships of the septa can be best understood
when taken in connection with the mesenteries, as revealed in transverse
sections of the polyp.

Plate 6, fig. 34, represents a transverse section from the lower stomo-
daeal region of a partly retracted decalcified polyp. The drawing was made
with the help of a camera lucida, but in the process of decalcification and the
preparation of the section slight displacements of the lamellae have taken
place. Still the general relationships remain as in the living condition, and
for all practical purposes the invaginations of the polypal wall correspond
with the septa which have been dissolved away. Fig. 2, p. 13, is a diagram-
matic representation of the septa of the same polyp, the synapticula and
spinous projections being omitted for the sake of clearness.

In the transverse section (plate 6, fig. 34) a large septal invagination
(entosepta, I, 11, iii) occurs within the interspace inclosed by the two
members of each pair of mesenteries, and a smaller one in the interspace
between every two adjacent pairs (exosepta, x). The entosepta inclosed by
the two pairs of directive mesenteries are known as directive septa, but in
the corallite itself there is nothing to distinguish these from the other
primary entosepta. They indicate the dorso- ventral (postero-anterior, sulculo-
sulcar) axis of the corallite, and can only be determined with certainty when
in association with the mesenteries.

The six interspaces between the six primary mesenterial pairs contain
a variable number of pairs of mesenteries : the two ventral systems each
contain one mesenterial pair (il), the right middle interspace three pairs
(ill, II, III), and the remainder two pairs each (in, 11). The six pairs of
secondary mesenteries (11) are easily distinguished from the tertiary mesen-
teries (ill) by their greater size and more central position. Of the latter
there are only five pairs in this particular polyp instead of twelve, as
considerations of hexameral symmetry would suggest.

50 SIDERASTREA RADIANS.

The septa are therefore related to the mesenteries in the following
manner :

The six primar}' septa are inclosed within the entocceles of the primary
cycle of six pairs of complete mesenteries, the six secondary septa within the
entocceles of the second cycle of mesenteries, the five tertiary septa within the
entocceles of the third cycle, while the members of the outermost or quater-
nar}' c^'cle of septa are all exocoelic, one between each two pairs of mesenteries.
The primary, secondary, and tertiary septa are all entosepta ; the members
of the fourth cycle are exosepta. Further, the number of exosepta equals
the Slim of the entosepta, and the number of cycles of septa is one more than
the number of cycles of mesenteries. Considering the septa as entosepta and
exosepta, the true septal formula is 6, 6, x, 12 + x, where x may be any
number from one to twelve. The variations in the number of septa within
the different calices are in pairs — an entoseptum and an exoseptum — and affect
only the members of the third cycle and their corresponding exosepta. These
two cycles always show a corresponding variation which is not explicable on
the usual supposition that the cycles of septa are developed in the order of
their importance. The order of appearance of the septa and the relationships
of the entosepta and exosepta will be discussed more fully in connection with
the development of the septa iu larval pol3'ps.

It will be shown that morphologically the exosepta are to be regarded
not so much as a separate cycle, appearing after the others are established,
but rather as the bifurcated continuations of the original cycle of exosepta, or
perhaps as new formations arising along with the entosepta which they
inclose ; each entoseptum has an exoseptum corresponding with it, which is
formed nearly at the same time. Exosepta are found to arise along with
each cycle of entosepta or even with each individual entoseptum ; but as new
cycles of entosepta are formed they are, as it were, shifted outwardly, so as
always to constitute the outermost cycle. Until the adult condition is reached
the exosepta are but temporary predecessors of the permanent entosepta.
The same relationship is found to hold between the entotentacles and the
exotentacles.

The above relationship between the cycles or orders of septa and mesen-
teries holds for most species of corals which have been examined. In certain
forms the exosepta are absent, e. g., Pectinia, Maiiicina^ when the cycles of
septa correspond in number with the cycles of mesenteries.

Where the hexameral cyclic sequence of a corallite is not completed it
is preferable to speak of the exosepta merely as exosepta or as septa of the
outermost cycle, not as a third or a fourth cycle ; for ontogeneticall}' some

ADULT COLONY. 5 1

belong to the third cycle and some to the fourth. They have no ordinal value
comparable with the entosepta.

In descriptive works on corals, when giving the number of septa
characteristic of an}- calice, it is usual to consider all the inner cycles as
hexamerously complete and then to regard all the missing septa as wanting
from the last cycle. Thus Milne-Edwards, in describing the septa of the
present species, says : " Three cycles of septa complete, and, in general, a
variable number of a fourth cycle." Also Verrill (1901, p. 153): "They
[the septa] form three complete cycles, with part of the fourth cycle devel-
oped, so that the number is usually 36 to 40." From the relationships here
set forth, and more fully discussed in connection with the development of
the septa, it is clear that such cyclic plans do not express the true ordinal
or morphological relationships of the septa ; the last and penultimate cycles
vary in the same degree, and the latter contains both entosepta and
exosepta.

The septal invaginations on plate 6, fig. 34, indicate how the exosepta
in most instances fuse with the third septa, but in the two ventral systems,
where no members of the third cycle are developed, they unite instead with
the secondary septa. In the serial sections the invaginations also show that
the fusion of the exosepta with the entosepta is not continuous throughout
their vertical length, but is effected only at somewhat regular intervals.

The section of a fragment of a corallum represented on plate 10, fig. 64,
includes the whole of one corallite and portions of the six surrounding
corallites, and shows their relationships to one another. The thecal wall
separating one corallite from those adjacent to it is seen to be very limited
in thickness. At this level in the complete calice twelve septa, representing
the first and second cycles, are united directly with the columella. The six
primary septa are distinguished by the fact that they extend all the way
from the periphery to the columella without fusion with any of the smaller
septa. In three of the primary systems there are only three septa — a second-
cycle entoseptum and two complete exosepta fused with it ; in the remaining
three sextants are five septa — a second-cycle entoseptum with one exoseptum
and a third-cycle entoseptum fused with it, the latter having two exosepta
in union with it.

Comparison ma}' be here made with a somewhat similar horizontal sec-
tion of a corallite of S. radians, introduced on plate xv (fig. 12) of Agassiz's
" Florida Reefs." Nine distinct, complete septa are represented, and in the
interspace between each two occurs a group of three septa, the two lateral
united with the middle, which, in its turn, extends to the columella. The

52 SIDERASTREA RADIANS.

arrangement of the septa in groups of three is continued all the way round,
and the hexameral plan is altogether obscured. Such an interpretation of
the septal plan is at variance with what is established above for the Jamaica
representatives of the same species.

As seen at the surface of a colony the septal edges extend almost hori-
zontally for a short distance from the periphery, and are then inclined down-
ward and inward somewhat sharply to meet the columella which forms the
floor of the middle of the calice. The actual depth of the calice varies some-
what in different colonies, but is usually about 2 mm.

SYNAPTICULA.

Viewed with a lens from above, the septa are seen to be joined to one
another laterally by thick transverse bars — the synapticula. In any trans-
verse section of a calice one to three, rarely four, synapticula are seen crossing
the space between every two adjacent septa, but are limited in their distribu-
tion to the peripheral half of the calice (plate lo, fig. 64). In a view of the
lateral surface of the larger septa they are found to be arranged in two or
three, rarely four, somewhat irregular vertical rows, each synapticulum being
circular or oval in section, and projecting at right angles from the septal
surface (plate 10, fig. 63). The presence of so many connections between the
septa gives a porous or reticular character to the more peripheral part of the
calice.

The synapticula are formed by the ultimate fusion across an interseptal
loculus of two opposite granulations on the faces of adjacent septa, the calca-
reous matter being deposited in a manner similar to that of other parts of the
coral lite. In Siderastrea it is only the granules in restricted spots on the
septal faces which become thus enlarged. The synapticula have a limited
distribution, and the other granulations on the septal face remain compara-
tively small, there being no intermediate examples. It is clear that in the
coiirse of their growth the enlarging granules must first indent the skeleto-
trophic tissues which line them, then bring the two opposite layers together,
and finally perforate them, as shown on plate 6, fig. 34 ; also, as described on
p. 28, the mesentery inclosed within the loculus is perforated at the same time.

Sections through the synapticula show that they are formed as lateral
processes of the septa, either by an extension of the septal trabeculse or from
a center of calcification of their own. In the former case a boundary in the
middle of the synapticulum indicates where the bundles of fibro-crystals from
one septum have met those from the opposite septum. In the other case it

ADULT COLONY. 53

appears as if the fibers from the synapticulum itself were bracing the two
septa. The first are known as false synapticula and the second as true
synapticula, but, as all recent writers have pointed out, there is no morpho-
logical distinction between the two. Both kinds appear in any section of a
corallite. Whether one or the other form is present depends mainly upon the
inter\^al between the two septa. Where sufi&ciently close, as towards the
middle of the calice, the interseptal space can be bridged without the forma-
tion of a new center of calcification, while such is necessary when the interval
is wide, as towards the periphery.

Miss Ogilvie (1897, p. 168) considers that in FungiaXho. synapticula are
formed by special interseptal invaginations of the basal wall of the polyp
subsequent to the septal upgrowths, without ever perforating the mesenteries,
and assumes such to be the case with Siderastj-ea. Bourne (18S6, p. 47) has
already shown that in Ftingia the mesenteries are really pierced by the
synapticula, Fowler (1888, p. 8) has accomplished the same for Stephatio-
phyllia forinosissima, and I for Siderastrea siderea (1902, p. 4S7). The con-
ditions revealed by the liberated lamella on plate 6, fig. 33, prove that the
synapticula are the products of definite areas of the original skeletogenic layer
producing the septa, not of special secondary basal upgrowths. Delage &
Herouard (1901) have accepted Miss Ogilvie's interpretation of synapticular
formation, gi\dng diagrammatic figures to illustrate how the upgrowth is
supposed to take place.

COLUMELLA.

The columella is represented in the mature calice by a more or less
compact, column-like structure forming the floor of the middle of the calice.
In decalcified polyps it is found to have elevated the central part of the
skeletogenic tissues in a tubular manner considerably beyond the peripheral
parts. Its superficial appearance varies much in different calices ; it may
be either papillose or smooth, dependent upon the degree of calcification of
the colony. Sometimes only one large, tooth-like papilla occurs, or there
may be two close together ; in others, again, there are two or three large or
chief tubercles along with several minute projections, or it may be formed
wholly of small, scarcely perceptible granules, when, to the naked eye, it
appears compact and smooth. This latter condition is found in coralla of
which all the parts are strongly calcified. In these the columella is conspic-
uous, raised for a short distance above the septa, and its free surface nearly
smooth.

Where several prominent tubercles are present some seem as if con-
tinuous with the teeth of the septa, suggesting that the columella is formed

54 SIDERASTREA RADIANS.

from tte united edges of tlie primary and secondary septa. Others of the
projections seem, however, to be quite independent of the septa. Secondary
calcareous matter is deposited among the projections, and this gives the more
compact character to the structure as a whole. The section represented on
plate lo, fig. 65, shows the relationship clearly. The columella is here
quite solid and distinct, and is structurally the same as the septa united with
it, having its own centers of calcification. In sections a little higher it is
spongiform, as the trabeculse are free from one another.

With such structural details alone available it is practically impossible
to say whether the columella of Siderastrea is a true or a false columella
(pseudocolumella) as these terms are understood in coral literature. Miss
Ogilvie (1897, p. 179) describes it as a "paliform pseudocolumella." A true
columella is considered to arise as an independent structure from the middle
of the basal plate, though the septal edges may secondarily unite with it. A
false or pseudocolumella, on the other hand, is an irregular skeletal tissue
formed from union of the inner or central ends of the septa, sometimes bound
together by a secondary deposit. Recourse to the developing coralla shows
the true nature of the columella in S. radians. As described later, and
illustrated by the figures on plates 4 and 5, independent skeletal upgrowths
are first formed from the basal plate, but later come into intimate relation-
ship with similar formations at the edges of the septa, and afterwards the
two groups are united into a solid, compact column by a deposit of calcareous
matter.

Developmentally, therefore, the columella of Siderastrea is a true colu-
mella, compact all the way, or compact below and spongy above.

Histologically the columella usually shows two or three large circular
trabeculas arranged in a row and surrounded by a number of smaller tra-
beculse (plate 10, fig. 65). The larger probably represent the true elements
of the columella, which arise directly from the basal plate, while the smaller
are the septal teeth which also take part in its constitution.

Viewed from the surface the columella is usually nearly circular in
outline and affords no aid in determining the principal axis of the corallite ;
but in sections some distance below the surface it generally presents a longer
and a shorter axis, and the former evidently corresponds with the directive
or principal axis of the corallite. To each extremity of the longer axis a
septum of the first order is attached, and the two are, no doubt, the directive
septa, though, owing to the difference in the number of septa on each side,
they are not always in the same plane (plate 10, fig. 64). The oval columella
may thus afford an important aid in the orientation of the septa and of the
corallites in a colony.

ADULT COLONY. 55

DISSEPIMENTS.

The dissepiments are extremely thin and delicate transverse partitions
which serve to cut off the living polyps from the dead part of the skeleton
below. In section they are only 0.0 1 mm. thick. Along with the columella
they serve for the time being as a support for the basal part of the polyp, the
dissepiment being morphologicall}^ comparable with the basal plate of the
larval polyp. In some cases the partitions are horizontal, but in others they
are deeply convex upwardly. They are situated at somewhat different levels
in different interseptal loculi, and in the same loculus var}^ from 0.25 mm. to
0.5 mm. apart. The distance apart of two adjacent dissepiments in a loculus
represents a period of growth of the polyp in its march upward, the polj-ps
throughout their lifetime retaining approximately the same vertical length
and transverse diameter. Asa result of the ver}' narrow interseptal loculi in
Siderastrea the dissepiments are rather insignificant features of the corallum,
but as a basal support they are probably as important to the polyp as in other
species of corals in which they are more prominent. Occasionally a dissepi-
ment is found intersected by a synapticulum, when both may be considered as
constituting the basal support of the particular lamella. Such was probably
the case in the interseptal lamella represented in plate 6, fig. 33, Usually
the lamellae present a more nearly straight basal extremity than is here repre-
sented, showing that the polyp is cut off below in a regular manner b}^ the
formation of dissepiments which alternate with the synapticula. The dissepi-
ments may thus correspond or alternate with the synapticula, the latter being
formed independently and in advance of them.

The dissepiments are easily studied in tangential sections of corallites,
and it is found that their histological structure is different from that of
septa. They possess no centers of calcification, but are thin and laminated,
composed of fibro-cr}- stals standing at right angles to the surface. They thus
recall the microscopic structure of the basal plate and epitheca, and like these
the}?- are lined by the calicoblasts of the polyp only on one side.

Miss Ogilvie (1897, p. 178) assumes that the synapticula and not
the dissepiments constitute the principal basal support of the polyp in
Siderastrea. This would follow were the S3'napticiila formed, as she assumes,
within special interseptal invaginations of the aboral body wall ; but, as
sho^\•n on p. 53, such is not their origin. They are produced by special areas
of the septal invagination, ultimately resulting in perforation of the polj^pal
wall. As the synapticula actually perforate the polyp they may be conceived
as also affording it support ; but only incidentally, as happens to be the case

56 SIDERASTREA RADIANS.

in the lamella shown in plate 6, fig. 33, is this basal. The synapticula do
not take the place nor assume the function of dissepiments.

EPITHECA AND BASAL PLATE.

The most careful observation of the margin of adult coralla fails to
reveal the presence of an epitheca or covering of the peripheral septal edges.
This is somewhat remarkable considering that such a skeletal formation is
developed in yoimg pol3'ps reared from the lan-a, and this whether the
polyps are isolated or growing contiguous to others (plates 4, 5). In
these larval polyps, however, the extent of the upward growth of the epitheca
has been found to vary greatly. In some specimens, apparently stationary
as regards growth, it formed a comparatively high external wall to the polyp,
wrinkled and diminishing somewhat in diameter from the base upwards, but
overtopping all the septa (plate 5, fig. 27), while in others, the largest,
actively growing corals, it was either absent or represented only by a low,
narrow rim (plate 5, fig. 28). Evidently it is a structure of importance
only in the early stages of growth.

Where, as frequently happens, a part of the growing margin of a colony
is free from the incmsted object, a flat, plate-like deposit covers and unites
externally the basal edges of the septa. Such a formation by the bud- polyps
will correspond with the basal plate of the larval pol^'p, but its peripheral
edge is rarely, if ever, upturned so as to cover the free vertical edges of the
septa, as in the case of an epitheca proper.

POSTLARVAL DEVELOPMENT.
LARVA.

Five diflferent colonies of ^. radians were collected in Kingston Harbor,
Jamaica, on the 6th of July, many of the polyps of which contained free
planulse. Although similar colonies had been obtained from this locality
on former occasions, and examined with regard to their fertility, this was the
first time that larvae were secured. On a second visit a week later two or
three other ripe colonies were collected, and others which on being sectionized
were found to contain ova.

On every colony the fertile polyps were in somewhat restricted patches,
not all the individuals coming to maturity at the same time. One, two, or
three planulse were plainly visible through the nearly transparent tissues of
the parent, and were often observed to pass into the tentacles, which thereby
became greatly distended, remaining so even when the polyps were retracted.
At times the planulas would glide into the lower regions of the polypal cavity,
and were then lost to view. While within the parent cavity it was impossible
to determine whether the movement of the larvae was, like that of the food
particles, dependent upon the ciliary activity of the lining endoderm or was
a result of the larva's own activity. Immediately upon being set free, how-
ever, the larvae were able to swim about, the ectoderm being already provided
with a layer of cilia.

The larvae were shot out suddenly, but the actual place of extrusion was
not determined, although prolonged observation was made. From their per-
sistent entrance into the tentacles it would seem that they made their escape
through these organs, and not through the mouth, which was usually closed
and depressed. In other instances, e. g., Manicina areolata and Favia
fragum, the sexual products and larvae have been seen to be given out
through the oral aperture, the proceeding being accompanied by a peculiar
jerking motion of the adult polyp. Von Koch (1897), however, found the
lar\'3e of Caryophyllia cyathiis to be expelled through the tips of the tentacles,
and Lacaze-Duthiers (1873, p. 30S) occasionally observed the same in Astroides
calycularis. Sections show that the knobbed tip of the tentacles in Sideras-
trea is without any permanent terminal aperture.

Larvae were freely extruded at the time of collection of the corals, and
continued to be discharged from time to time for about a month. Under ordi-
nary conditions one or two would be set free at intervals, but upon disturbance

57

58 SIDERASTREA RADIANS.

of the colony a score or so would be shot out together. Often when the disc
had become retracted a larva would remain within the upper peripheral region
of the pol3'p, fixed between the edges of a septum and the wall of a tentacle.

The lar\^se were able to swim about immediately upon being discharged,
and gyrated through the water first in one direction and then in another.
There was evidence of a feeble negative geotropism. During the first day
the}^ kept near the surface of the water, or gathered around the sides of the
vessels ; afterwards they traversed the water as a whole, though in the main
keeping near the surface.

To the unaided eye some of the planulge appeared as minute, opaque-
white, spheroidal bodies ; others were oval ; but the majority were elongated
and pear-shaped, as shown in plate i , fig. i . Under the microscope the ciliation
was found to be uniform over the whole surface. The broader pole (oral)
was deeply pigmented and invariably posterior in s\vimming ; no oral aperture
could be detected at first, but extrusions of Zooxanthellse, yolk, and cell
debris took place later, showing that the mouth was alread}- functional.
Under the microscope the brownish 3'ellow of the broad end was seen to be
due to the presence of numerous Zooxanthellse within the ectodermal layer ;
otherwise the ectoderm was colorless, while the endoderm was dark and non-
transparent. The narrow anterior end (aboral) was perfectly colorless, its
ectoderm being free from commensal algae. In some larvje the colorless end
(aboral) was larger than the other. The elongated specimens were able to
retract and extend themselves, and upon irritation or preservation altered
their shape so as to become nearly spherical. The ordinary pear-shaped
individuals were about 2 mm. long.

Throughout their free existence the majority of the larvae remained as
wholly opaque objects, and no trace of mesenterial divisions could be seen
from the outside. Some specimens, however, were so far developed as to be
distended at birth, or became so shortly afterwards. The tissiies of these
latter were more transparent than the rest, and eight mesenterial divisions
were obvious, thoiigh not all of the same vertical length (plate i, figs. 3 and 4).
The latest extruded larvae, after the colonies had been kept under somewhat
unfavorable conditions for two or three weeks, Avere nearly devoid of external
Zooxanthellas and therefore colorless.

After the first day or two many of the larv-fe showed positive geotropism
and sank to the bottom of the vessels, lying there motionless ; later, some of
these resumed their activities. By the evening of the second day a few had
fixed themselves by the narrow, anterior, colorless end to the bottom or side
of the vessel, or to objects placed within it. At first the lan^as would adhere

POSTLARVAL DEVELOPMENT. 59

by means of the actual tip of the narrow end (plate i, fig. 5) ; this would then
flatten out, and the whole body become much shorter, a small round or oval
aperture being visible at the free extremity (plate i, fig. 6).

Whether or not any individual larva would settle seemed very uncertain,
for out of several hundreds set free comparativel}^ few became permanently
fixed. Those which settled appeared to be the larger, better developed
specimens. Smaller vessels, glass slides, and cover glasses floated by
means of cork were placed within the receptacles so as to afford additional
surfaces for adhesion ; in other cases larvae were distributed within vessels
coated with parafiin. If fixation were not accomplished within the first few
days, it seemed to be impossible afterwards, though the larvae might continue
active for several weeks. In one instance about a score of planulae were
isolated and kept in a glass dish, and although they appeared healthy and
swam freely for a period of twenty days, externally the}' underwent no change
whatever.

Sometimes a larva would fix itself in a position apart from others ; but in
general many would settle close together at one and the same time, their walls
in some cases actually pressing one against another. Plate i, fig. 5, represents
three larvae in the first stage of fixation, in this case to a small pebble. The
narrow extremities nearly toiich, and it is obvious that when the larvae shorten
and their bases flatten, the latter will exert a mutual pressure as a result
of their closeness. In plate i, fig. 6, seven larvae already settled and fully
expanded are represented, as seen from above by reflected light. All were
closely adherent to a fragment of stone, and formed a miniature colony. To
the under surface of a small pebble thirty-eight other specimens were aggre-
gated in groups of two, three, or more, the members of any group touching
along their margin. One of these groups contained a dozen or more 3'oung
polj'ps, all in contact with one another, the mutual pressure producing a
distortion of the normally circular base.

During a single night another group of thirtj^-two became adherent to the
surface of a small glass dish. In this case nearly all the members were touch-
ing to a greater or less degree. One of the aggregations, consisting of fifteen
polyps, is represented in fig. 5, p. 60, as seen from the under surface of a
fragment of glass to which the individuals were attached. The colony when
drawn was two or three months old, and the skeleton was already in process
of development, being represented by a variable number of septa and by basal
and epithecal formations.

The aggregated larvae, or young polyps, as they may be called after
settling, remained closely associated during their subsequent growth, and in

6o SIDERASTREA RADIANS.

all respects resembled a distinct colony. Siderastrea radians thus exhibits
the phenomenon of a coral reaching its colonial stage by direct union or
aggregation of a number of independent larvse, the larvse being at first free
swimming and wholly unconnected with one another. The close association
is probably to be understood as a thigmotactic phenomenon, and calls for a
more extended study with abundant material. The subject of " Aggregated
Colonies in Madreporarian Corals " has been more fully discussed in " The
American Naturalist," June, 1902. It is there shown that similar associa-
tions are at times met with in other corals.

Externally the opaque larvae seemed all alike, but sections reveal slightly
different stages of growth as regards the number of mesenteries and their
connection with the stomodaeum. The sections further demonstrate that the

Fig. 5. — Aggregation of fifteen polyps derived from the same number of originally free-swimming larva:.
Only the basal sl<eletal appearance is represented.

opacity is due to the thickness of the endoderm and the numerous Zooxan-
thellae within its cells. Furthermore, the gastro-coelomic cavity is yet
unformed, or very limited in extent, the larvse being a nearly solid mass of
cells. The distended larvae, on the other hand, already showed from the
outside different stages in the development of the mesenteries, and the
internal cavity was more fully established [cf. plate i, figs. 3, 4).

After fixation all the larvae became distended, the walls more trans-
parent, and mesenterial divisions, twelve in number, were rendered conspicuous.
From the beginning of fixed life some few were more advanced than others.
These were the individuals bom in an already transparent state. On set-
tling they were nearly as large again as the other larvae, and their develop-
ment throughout was more rapid.

A few abnormal larvae were extruded and swam about like the rest.
The oral extremity of these was double and the aboral single, the bifurcation
taking place about the middle of the length (plate i, fig. 2). Lacaze-Duthiers
(1873, p. 312) likewise found such double monsters to be somewhat frequent

POSTLARVAL DEVELOPMENT. 6l

in Astroides. In these the oral end was sometimes bifurcated and sometimes
the aboral.

No trace of any calcareous deposit or skeletal formation was apparent
in any of the larvae before fixation, but calcareous matter was laid down soon
afterwards, the distinction between basal disc and free polypal wall, or skeleto-
genic and non-skeletogenic tissues, being then established for the first time.

It is convenient to describe the general course of the development and
character of the young polyps before proceeding with the detailed growth of
the individual organs — tentacles, mesenteries, and septa.

YOUNG POLYP.

Some of the lan'se became adherent to small pebbles, others to pieces of
glass, while most were fixed to the sides and bottom of the glass vessels in
which the colonies were placed. In this last case the vessels were sacrificed
in order to obtain the young polyps in a condition favorable for examination
under the microscope. By carefully breaking the glass suitable fragments
were secured with the polyps attached, and these could be transferred from
one aquarixim to another, or placed in dishes small enough to rest on the
stage of the microscope. In this way the growing polyps could be examined
at any time from both their upper and under aspects, their partial trans-
parency allowing the development of the mesenteries and septa to be followed
day by day. Thus the growth of the polyp and the skeleton could be observed
together and their various relationships studied.

Once the larvae were settled they seemed vigorous and hardy, and scarcely
au}^ of the young polyps arising therefrom succumbed, although subjected
to the somewhat adverse conditions of small aquaria.

The different organs, polypal and skeletal, began to appear shortly
after fixation, the various stages in the development of the tentacles, mesen-
teries, and septa being represented by the drawings and photographs on plates
1-5. The young polyps were early capable of expansion and retraction.
For the most part they remained in the expanded condition, with the tentacles
stretching out horizontally, or even overhanging. Sometimes the column
would extend vertically upwards and remain equal in diameter throughout ;
at other times the lower part of the walls would shrink over the corallum
and the remainder appear as a narrow column upon a broad pedestal. Upon
retraction the tentacles still remained exposed, but restricted to the central
region (plate 2). Even in the adult the tentacles have been found to remain
visible upon the fullest retraction, the column being incapable of folding
over them. On one occasion only a larval polyp was found with the

62 SIDERASTREA RADIANS.

column wall completely overfolding the tentacles and covering nearly all
the disc (plate 3, fig. iS).

After the first few days the polj'ps were able to feed upon small pieces
of molluscs or chsetopods placed upon the disc by means of forceps or a
needle. The tentacles would close upon the fragment and hold it as the
forceps were withdrawn. If the pieces offered were small enough they were
taken bodily within the gastric cavity, but when too large to be engulfed a
portion only would be drawn within the stomodaeum, remain there for two or
three hours, and then be pushed away over the side of the disc.

For some time Zooxanthellas were present within the ectoderm of the
disc and upper part of the column, and rendered these regions nearly opaque ;
but later they wholly disappeared from the outer layer, and, as in the adult,
were restricted to the endoderm. As the polyps grew larger their walls
became thinner and more transparent. The internal yellow cells were
thickly distributed along the two sides of the mesenteries and within the
endoderm of the stalk of the tentacles, but were sparse towards the lower
part of the column. They formed an important aid in determining the course
of the mesenteries within the living polyps.

The young polyps displayed great power of recovery from injury. In
two or three instances the glass to which they were attached was broken in
such a way that a polyp was completely divided into two, the halves adherent
to dilTerent fragments ; 3'et the two parts continued to live, and each became
a distinct, though smaller, polyp, nearly circular on healing, and restoring
in every way the normal form.

On the polyps attaining partial transparency, the internal cavit}'' could
be seen to contain granular matter and Zooxanthellae in a regular rapid
movement, and, by focussing at different heights, it was possible to make out
the complete course of the circulation. The particles passed directly upward
along the column, t4ien transversely across the disc, downward along the
endodermal surface of the stomodaeum, and thus right into the central cavity
of the polyp ; they then continued radially across the base outwards towards
its periphery, and once more up the column. The regularity of the course
was a little disturbed at the opening of the tentacles into the gastric cavity.
Some of the particles would circulate around this region in an eddy-like
manner, or pass a little within the tentacular cavity, and afterwards be
wafted along in the general current. The course is indicated in fig. 6.

Occasionally the lips of the stomodaeum would meet in the middle and
remain open at each end, but no incurrent or excurrent streams were ever
detected through the apertures thus formed. A bolus of organic debris,

POSTLARVAL DEVELOPMENT. 63

containing Zooxanthellse, would at times be seen whirling round and round
within the stomodaeum, and would afterwards be jerked out and lost to the
polyp. Even after three months such extrusions would still take place.

The young polyps varied much in their rate of growth, especially in
the later stages ; indeed, in some instances, development seemed to be wholly
arrested. The isolated polyps progressed most rapidly, but among these
many differences were recognizable. The polyps forming colonies made
scarcely any progress after the first week or two. In the group represented
in plate i, fig. 6, no increase beyond the six tentacles and twelve simple
septa occurred up to six weeks, when the colony was preserved.

The development of the various organs was characterized by certain
well-marked intervals of rest and growth. A week or two might pass without
any conspicuous change taking place, and then
progress would be somewhat rapid. The order of
appearance of the organs varied somewhat in dif-
ferent individuals ; at first some organs of a system
were developed a cycle at a time, and the remainder
in successive pairs from one border of the polyp to
the other. Perhaps under the artificial conditions
the duration of some of the intervals was partly
determined by the state of the water and the amount ^ ^ . ^

-^ ^ Fig. o. — Diagram snowing the course of the

and character of the food supplied. Still, there circulation within the imemai cavity of a

^ - , . • • 1 1 larval polyp.

were certam definite resting stages appearing m all

the pol3'ps and seeming to possess a phylogenetic as well as an ontogenetic

significance.

The following were the most conspicuous phases of growth : Of the
mesenteries the protocnemes were established at the time of settling of
the larvae, and the six pairs are shown to develop in a regular consecutive
manner \\4thout any marked intervals (p. 76). Four of the six bilateral
pairs of protocnemes were connected with the stomodaeum from the commence-
ment of stationary life, but at the termination of seventeen weeks the fifth
and sixth protocnemic pairs had not become complete. No further change
in the number of mesenteries took place for nearly four weeks, when, in
the larger polyps, the metacnemes began to appear. The different pairs of
these followed somewhat quickly upon one another from the dorsal to the
ventral aspect of the polyp, until the complete cycle of six pairs was estab-
lished (plate 3, fig. 14). The polyps remained at this mesenterial stage
for the rest of the time they were under observation — about two months.
There was no hint of the appearance of the third cj'cle.

64 SIDERASTREA RADIANS.

As regards the tentacles, six primary exotentacles appeared simulta-
neousl}^, shortly after fixation (plate i, fig. 6), followed by an interruption
of two or three weeks before the eutotentacles began to arise. These were
at first simple outgrowths of the disc, and a long interv^al took place before
a second moiety appeared in connection with each and thus established the
mature form (plate 2, fig. 11). The longest interruption, however, was after
the completion of the primary twelve tentacles, for in the most forward polj^ps
about three months elapsed before additional members appeared, when those
of another cycle began to arise in a successive manner.

Of the septa the six entocoelic representatives appeared as a complete
cycle at a very earl}' stage (plate i, fig. 7), followed shortly by the six alternat-
ing exoccelic members ; these latter were developed in some instances as a
complete cycle, but usually in successive pairs (plate 2, figs. 8, 9). Without
any marked rest other fragments or nodules were added to each system, but
in such a way as to render it doubtful as to which cycle they belonged (plate
2, fig. 12). On the whole, the growth of the septa was more continuous
than that of either the mesenteries or the tentacles.

Undoubtedly the most conspicuous interruption for all the systems of
organs w^as that between the protocnemic and metacnemic stages, and the
great differences between the manner of appearance of the two series of
mesenteries, septa, and tentacles further establishes this as the most impor-
tant ontogenetic and phylogenetic interval in madreporarian development.
Its significance is discussed later.

The time of appearance of the organs varied somewhat in different polyps,
sometimes even to the extent of several weeks ; but with regard to the actual
order followed no important differences were observ^ed in the mesenteries and
septa, though less constancy prevailed amongst the tentacles. Where devel-
opment was not in complete cycles the succession was bilateral, proceeding
from the dorsal to the ventral borders of the polyp. Contrary to what is
generally assumed, bilaterality, not radiality, was a conspicuous feature of
the growth from beginning to end ; and in most of the S3'stems of organs
the bilateral symmetry prevailed for lengthened periods, even where the
radial condition was ultimately assumed.

Many of the young polyps were preserved at different stages — while still
adhering to their original pieces of glass — and were prepared for microscopic
examination as a whole. The process was as follows : After narcotization
by means of magnesium sulphate or menthol the pol3'ps were killed by
pouring a solution of formaldehyde over them, and, still adherent, they were
transferred to alcohol, afterwards stained, dehj^drated, cleared, and mounted

POSTLARVAL DEVELOPMENT. 65

in Canada balsam as permanent preparations. The young coralla from
which the photographs on plates 4-5 were taken were prepared by macer-
ating the polyps in caustic potash and afterwards washing or carefully
brushing away the soft, loosened tissues. After maceration the skeletons
could generally be freed from the glass to which they were attached by
gently tapping the latter, but in some cases it was necessary to insert the
edge of a razor between the basal plate and the surface of the glass.

The young polj'ps intended for sectioniziug were detached after preser-
vation by dissolving away the calcareous deposit with weak hydrochloric
acid. In one case a living polyp of fourteen weeks became detached along
with the whole skeleton, and, thus free, continued its growth for some time.

In the course of their growth the polyps increased from i to 2 mm. in
diameter, the oldest polyps reared being but 2 mm. across the base.

TENTACLES.

FIRST CYCLE OF EXOTENTACLES.

The tentacles began to make their appearance two or three days after
fixation of the larvse, as rounded upgrowths over alternate mesenterial
chambers. At this stage only the six primary pairs of mesenteries were
present, giving rise to six entoccelic and six exocoelic chambers (plate i,
fig. 7). Six tentacular prominences appeared simultaneously, equal in size
and distance apart. They early showed an opaque white, knob-like apex
distinct from a short, more transparent stem. The single cycle which they
constituted served to delimit the lan^a for the first time into oral disc and
column wall, no indication of the boundary between the two areas being
hitherto determinable.

The stems of the tentacles were hollow, broad below, narrow distally,
and perfectl}^ smooth, that is, without nematoblast tubercles. The endoderm
within could be seen to be richly supplied with Zooxanthellas, while the ecto-
derm was colorless ; the knob was solid and colorless, without Zooxanthellse,
and constituted wholly of ectoderm with numerous radiating uematoblasts.

At first the actual mesenteric chambers from which the tentacles
protruded could not be determined with certainty, owing to the opacity of the
polyps. Soon, however, as the polyps grew larger and the walls more trans-
parent, the outgrowths were seen to communicate with the exocoelic chambers.
As noticed below, this relationship is contrary to the general rule of tentac-
ular development hitherto observed among the Zoantharia, and therefore
the greatest care was exercised to establish it beyond a doubt Externally
the entocoelic and exocoelic mesenteric chambers could be readily determined

66 SIDERASTREA RADIANS.

from their relation to the longer oral axis and from the known relation of
the complete and incomplete mesenteries at the Edzuardsia stage of develop-
ment; the two terminal chambers which include the longer oral axis are
always the directive entocoeles, and the two alternating chambers on each side
between these are the lateral entocoeles, while all the intermediate divisions
are exocceles. Most of the young polyps were so situated as to permit of all
these details being clearly observed, and in no instance were the first six
tentacles developed as outgrowths of the entocoeles, but were always disposed
as in plates i and 2, figs. 7-9.

As shown in plate i, fig. 7, the six tentacles first to appear alternated
with the first septa, the latter being entoccelic in position ; the smaller septa
which arise later (plate 2, fig. 9) correspond with the first tentacles, that is,
are exosepta.

FIRST CYCLE OF ENTOTENTACLES.

A long interval elapsed between the appearance of the primary cycle of
six exotentacles and the development of any others. Toward the close of
the fourth week additional small protuberances of the disc began to appear
in the interspaces between the primary series and a little nearer the center
of the disc. Like the first they soon showed a distinction into apical knob
and stem. In most polyps the six new tentacles arose simultaneously,
constituting a second inner cycle and alternating with the first ; but in some
cases the members developed siiccessively, though no dorso-ventral or other
regular sequence could be established. Any intervals which occurred in
the appearance of the diiferent members were very brief, except in one
instance where a fortnight passed between the appearance of the dorso-lateral
and the ventro-lateral pairs.

There were now twelve tentacles present, arranged in two alternating
cycles of six each ; an outer cycle of six large exotentacles, the first to arise,
and an inner cycle of six smaller entotentacles appearing later (plate 2, fig. 11).

Close examination soon revealed that the individual members of the inner
cycle were not disposed symmetrically midway in the interspaces between
those of the older cycle, but, as represented in plate 3, fig. 13, were a little to
one side. Even the tentacles arising from the axial entocoeles were some-
what to the one or the other side of the principal axis of the polyp. In most
cases the four small lateral tentacles were all situated toward the dorsal
aspects of their corresponding interspaces, while the tentacles over the direc-
tive entocoeles might be to the right or to the left of the directive axis. This
peculiar arrangement of the young entotentacles was best observed when
the polyps were only partly extended ; during full expansion they appeared

POSTLARVAL DEVELOPMENT. 67

almost symmetrically disposed with regard to the outer cycle and the ento-
coelic chamber below (plate 2, fig. 11).

The simple character of the entocoelic tentacles on their first appearance
is in marked contrast with their bifurcated form in mature polj-ps. As
already mentioned in describing the adult colony, the genus Siderastrea is
unique among West Indian corals in that its polyps possess dimorphic
tentacles. The entoccelic members consist of a simple stalk, divided above
the middle into two smaller halves, each of which is terminated by a white,
swollen knob ; the exocoelic tentacles, on the other hand, consist of a simple
stem with an apical knob. It was also found that in nearly mature polyps
the newly formed entocoelic tentacles of the higher cycles are sometimes
simple, while the older members are double (plate 6, fig. 32).

A few days after the completion of the second (inner) cycle of tentacles
in the young polyps, the double character of its members began to assert
itself. From certain of the entocoelic interspaces another knob-like pro-
tuberance arose beside the first tentacle, which in time became a distinct
tentacle with stem and capitulum like the other member (plate 2, fig. 12)-
The two moieties thus formed were for some time different in size, but later
became equal. The appearance was as if two small tentacles had become
intercalated between, but a little central to, the interspaces of the first-
formed tentacular cycle.

The second moiety of the inner tentacles exhibited no constancy in its
order of appearance from the different entocoeles. In plate 3, fig. 13, the ven-
tral axial member has appeared and also the two dorso-laterals, while over the
dorsal and ventro-lateral entocoeles the simple tentacle alone is present. In
four other polyps the dorsal axial tentacle was the first to become doubled,
and all the others remained single for a few days longer. In another polyp
the additional outgrowth first appeared over the two axial entocoeles.

The new outgrowth was at first wholly independent of the old, and to all
appearances the two were to be regarded as distinct tentacles, side by side,
each with its own peduncle. Later, however, a single proximal stem was
found to arise, which, uniting the two as by a single stalk, raised them some
distance above the disc, and the organ thus assumed the form of the ento-
tentacles already described in connection with the mature colony (plate 3,
figs. 15, 17).

The adult bifurcated eutocoelic tentacle is thus seen to be primarily
represented by two distinct tentacles, which only later are united by the
growth of a common peduncle. It represents a unique method of tentacular
growth among the Anthozoa. The double character appeared on only a few

68 SIDERASTREA RADIANS.

of the most forward larval polyps, and, for the time they were under obser-
vation, was in no instance established for all the six members.

For another period of six weeks or two months no further tentacular
increase took place, though in the meantime the second cycle of mesenteries
had made its appearance.

So far, four distinct stages are recognizable in the development of the
twelve prototentacles of Siderastrea, each stage separated by marked intervals
of rest. They are represented diagrammatically in the figures a-c on p. 72
as they would appear when fully completed.

1. The stage with six exocoelic tentacles, forming a single cycle. The
members are equal and appear simultaneously.

2. The stage with twelve simple tentacles, constituted of an outer cycle
of six large exotentacles and an inner alternating cycle of six small ento-
tentacles. The latter appear either simultaneously or successively in an
irregular sequence. They are not situated midway over the mesenterial
chambers. The four lateral members are disposed a little towards the dorsal
or ventral half of the interspaces, while the directive members are somewhat
to one side or the other of the principal oral axis of the polyp.

3. Six additional members of the inner entoccelic cycle appear, so that
the cycle is constituted of twelve independent tentacles in six pairs. They
arise in no definite sequence.

4. A single outgro\\i:h arises over the middle of each entocoele and forms
a common peduncle for each pair of entotentacles. Thus the tentacular
outgrowths from each entocoele come to resemble a bilobed adult tentacle.

The appearance of the exoccelic tentacles in advance of the entoccelic in
Siderastrca radians is unique in the tentacular development of Zoantharian
polyps according to the researches of Lacaze-Duthiers, Faurot, von Koch,
Appellof and others, who have made us acquainted with the post-larval
development of many species of anemones and corals. In the corals Caryo-
phyllia cyathus^ Manicina areolata^ and Favia fragum the twelve proto-
tentacles, constituting an inner and outer cycle, appear simultaneously, or
at most there is only a short interval between one and the other ; but
wherever such an interval occurs, the inner entoccelic cycle appears in
advance of the outer exocoelic, and the entoccelic members are larger than
the exoccelic (entacmseous). In general it will be found that the production
of other entoccelic structures — tentacles and septa — in corals is in advance
of the exocoelic.

In actinian larvae it is more usual for eight tentacles to appear first,
arranged either as a single cycle or as an inner and outer cycle of four each.

POSTLARVAL DEVELOPMENT. 69

These correspond with the eight interspaces between the four pairs of
Edwardsian mesenteries ; later, as the fifth and sixth pairs of protocnemes
increase in size, other two pairs of tentacles arise, and the whole become
arranged in two cycles of six each. Of the primary eight tentacles in
actinians six are the entocoelic members, and they early assume predomi-
nance over the exocoelic series \e. g.^ Lebninia, 1S99).

Clearly the number of external tentacles first appearing is determined
by the number of internal mesenterial chambers already formed, seeing that
the cavities of the former are continuations of the latter. Where, as in the
actiuian Lebrimia (1S99), onlj^ four pairs of mesenteries are present when
the tentacles make their appearance, it is manifest that only eight tentacles
will be developed, and that the other four will be formed only when the
growth of the fifth and sixth mesenterial pairs has given rise to four
additional mesenterial chambers. In the species of coral larvae so far fully
investigated the six pairs of primary mesenteries — four pairs of macrocnemes
and two pairs of microcnemes — are full}' established at or shortly after the
time of fixation and before the tentacles begin to protrude; hence six or
twelve tentacles arise simultaneously, one from each mesenterial chamber.

The long interval which elapses before the more complex entoccelic
tentacles are fully formed may perhaps account for the appearance in Side-
rastrea of the exotentacles in advance of the entotentacles. The young
polyp after fixation has to provide for itself, and for this purpose its weapons
of offense and defense, represented by the tentacular nematoblasts, are
necessary from the beginning.

The origin of the bifurcated entotentacles from two independent moieties
which appear at different times and are later elevated upon a common
peduncle is a second unique feature in tentacular development. Obviously
the presence of a double battery of nematocysts on each tentacle, in place of
a single battery, will be of advantage to the polyp ; and the growth of one
moiety in advance of the other is probably to be taken as indicating an
ancestry in which only one was present.

SECONDARY EXOTENTACLES.

Much interest attaches to the manner of appearance of the tentacles
bej'^ond the prototentacles, as, from the exceptional sequence of the first two
cycles, it is impossible to predict from the known development of other forms
how or where the next members will arise. One of the earliest stages is
represented by the fully expanded disc in plate 3, fig. 15. The second-
cycle mesenteries are seen stretching along the disc, decreasing in size

70 SIDERASTREA RADIANS.

successively from the dorsal to the ventral aspect. The six new pairs of
mesenteries have given rise to six additional entoccelic and six additional
exoccelic chambers, from which further tentacles may arise ; so that mani-
festlv the mesenteries with their mesenterial chambers are formed in advance
of their tentacular prolongations. At the stage represented in plate 3, fig. 15,
all the primary entotentacles are yet simple except the two dorso-lateral
members, each of which consists of a broad stem uneqvially bifurcated at
the dorsal extremity. The dorsal and ventral exotentacles are simple, and
the broad base of each communicates with the second-cycle entoccele below
as well as with the exocoele on each side of it. An additional tentacle is
seen on the ventral aspect of the middle primary exotentacle on each side of
the polyp, and as vet they are smaller than the exotentacles first formed.
The new tentacle has pushed the older a little to one side, and both are
obviousl}- exocoelic in position, though in the figure appearing to extend
over into the entocoele. There are now eight exocoelic tentacles forming an
outermost cycle.

A somewhat older stage is represented by the next figure (plate 3, fig.
16). The two middle exotentacles are equal in size and quite free from one
another, and both are now seen to be wholly unconnected with the median
second-cycle entocoeles. Another exocoelic member has also appeared on the
ventral aspect of the dorso-lateral exotentacle, on both the right and left sides
of the polyp, but as yet they are a little smaller than the primary exoten-
tacles. The four new exotentacles are situated at the same distance from the
center of the disc as the primary cycle of six exotentacles, and therefore,
instead of forming a new cycle, they merely add to the number of members
in a cycle already established. The outer cycle of exotentacles thus consists
of ten practically equal tentacles ; the two wanting to complete the hexameral
plan belong to the ventral system, which region is usually found to lag
behind the dorsal and middle systems in the extent of its development.

The two stages (plate 3, figs. 15, 16) demonstrate the important fact
that the primary six exotentacles remain exotentacles, and always constitute
part of the outermost cycle ; further, any new exotentacles formed become
equal in size with the primary exotentacles and are so disposed as to be
members of the same cycle. The primarj^ exotentacles do not become
entotentacles when new pairs of mesenteries arise in the same radii, as from
their original position might have been expected. On the appearance of the
second cycle of mesenteries, they seem for a time as if communicating with
their entocoeles (plate 3, fig. 15), but this condition is onlj' temporary. When
additional exotentacles arise and the mesenteries are further developed, both

POSTLARVAL DEVELOPMENT. 7 1

the older and the newer tentacles are seen to communicate only with the
exocoeles, and the second-cycle entocoeles are for a time without any tentacular
outgrowths. The cycle of exotentacles when complete would at this stage
consist of twelve members, of which six are primary and six are later forma-
tions, the latter appearing after the second cycle of mesenteries (fig.
7, d, p. 72).

The tentacular development of S. radians thus conforms with the law
of substitution established by Lacaze-Duthiers (1872, '73) and Faurot (1895)
for actinians — exotentacles remain exotentacles, and throughout the growth
of the polyp they continually change their relationship to the entotentacles
in such a manner as always to constitute the outermost cycle, and all are
uniform in size.

x\s in the case of the two cycles of primary tentacles the secondary
exotentacles appear in advance of the corresponding secondary entotentacles,
so that now the cycle of exotentacles contains double the number of members
of the cycle of entotentacles. In the several species whose development was
carried thus far by Lacaze-Duthiers and by Faurot the exocoelic and entoccelic
members appeared together along with a new pair of mesenteries, or the ento-
tentacles were in advance of the exotentacles ; later the entotentacles became
larger than the exotentacles.

SECOND CYCLE OF ENTOTENTACLES AND THIRD CYCLE OF EXOTENTACLES.

Clearly the stage beyond that represented in plate 3, fig. 16, will be one
in which the tentacles begin to make their appearance over the second-cycle
entocoeles. Such a condition is represented in plate 3, fig. 17, taken from the
same polyp as plate 3, figs. 15 and 16, when about a fortnight older than the
stage of plate 3, fig. 16. It is also the oldest stage reared. From the middle
second-cycle entocoele on each side a simple tentacle has arisen, intercalated
a little beyond the primary cycle of entotentacles, but yet within the cycle
of exotentacles. The two tentacles are the first representatives of the second-
cycle tentacles of the adult polyps, and the greatest interest attaches to the
fact that they have displaced, as it were, some of the members of the second
cycle (exotentacles), so that they now form a third cycle.

If tbe development of the polyps had been continued, there is every
reason to suppose that a new tentacle would have appeared over each of the
dorsal second-cycle entocoeles. Also in the further growth of the two ventral
systems another exotentacle would have appeared on each side, and then a
tentacle over the second-C3'cle entoccele, just as in the middle systems; then
all the simple second-cycle entotentacles would in time have become bifurcated

72

SIDERASTREA RADIANS.

like the members of the first-cycle entotentacles. In general, the sequence
of the entotentacles would follow that of the mesenteries with which they are
associated, although in plate 3, fig. 17, the growth within the middle sextants
is in advance of that of the dorsal sextants.

The completed system of tentacles would consist of: (a) Six bifurcated
entotentacles constituting the first cycle ; {6) six alternating bifurcated ento-
tentacles forming the second cycle ; (c) twelve alternating exotentacles, which
have appeared at two wholly distinct periods, making up the third cycle.

The developmental relations between the six second-cycle entotentacles
and the twelve third-cycle exotentacles are somewhat different in Siderastrea

X I X

Fig. 7. — a-^, series of diagrams illustrating the order of appearance and relationships
of the lirst three cycles of tentacles. The numerals I and II indicate the orders
of entotentacles and X the exotentacles.

from those hitherto met with in actinians. Lacaze-
Duthiers (1872) first found that after the completion
of the first two cycles of tentacles (six entotentacles
and six exotentacles), the tentacles of the third cycle
do not arise peripherally, alternating with the mem-
bers of the first and second cycles, as would be
naturally supposed, but that six new pairs are insinuated between the two
primary cycles at six intervals only. One member of each of the six new
pairs increases in size more than the other member, and ultimately consti-
tutes the second cycle of tentacles (entotentacles), displacing the primary
second cycle of exotentacles ; the other member of each new tentacular pair
attains the same size as the displaced exotentacles, and along with them
forms the third cycle of twelve tentacles. The third cycle is thus formed of
six tentacles from the original second cycle and six tentacles which are later
formations ; the members of the permanent second cycle are all new forma-
tions which have displaced the older second-cycle members. Lacaze-Duthiers
expressed these relationships under the term " law of substitution."

POSTLARVAL DEVELOPMENT. 73

In Siderastrea the second cycle for a time consists of twelve exotentacles,
six of wliich are primary formations and six are later outgrowths. The latter
appear in advance of the corresponding eutotentacles, not along with them, as
seems to be the case in most actinians. Afterwards, however, a new cycle of
six entotentacles is intercalated between the original first and second cycles,
displacing the latter to the third cycle, and itself constituting the permanent
second C3fcle. The law of substitution thus holds for Siderastrea as for
actinians, though the relative time of appearance of the entotentacles and
exotentacles differs from that usually followed. Faurot (1895) shows that in
Ilyantlitis parthenopeus the entotentacle from a new mesenterial pair arises
in advance of the exotentacle, exactly the reverse of that in Siderastrea.

THIRD CYCLE OF ENTOTENTACLES AND FOURTH CYCLE OF EXOTENTACLES.

As the larval polyps were not reared beyond the beginning of the
three-C3-cle stage, it will be necessary, in order to complete the survey of the
tentacular development, to have recourse to bud polyps of a colony. From
what has been established above it is manifest that the problem is not to
determine the manner according to which the fourth cycle of the adult polyp
will arise, but how a third cycle of entotentacles originates and is inter-
calated between the previously formed second and third cycles. The exo-
tentacles as a cycle are shown to have no ordinal significance in studies of
tentacular sequence ; this importance belongs to the entotentacles. Regarded
cyclicall}- the exotentacles are but temporary predecessors of the entotentacles
of the adult polyp.

On any colony are numerous polyps in which the tentacles vary from
twenty-four to nearly forty-eight. In the polj^p represented on plate 6, fig.
32, there are three third-cycle entotentacles, each with a corresponding
exotentacle. The former are yet at an early stage of development, only one
moiety of the tentacle having appeared. In fig. i, p. 12, is represented
diagrammatically the tentacular plan of the pol3'p of which the mesenterial
S3'stem is given in plate 6, fig. 34. Along with each of the five pairs of
third-cj'cle mesenteries (iii) have appeared two corresponding tentacles, and,
comparing with fig. 3, p. 26, one of the tentacles is seen to be entoccelic (iii)
and the other exocoelic (x). The new entotentacles, iii, are situated beyond
the second cycle of entotentacles, 11, but within the outermost exocoelic
cycle ; that is, they represent the commencement of a third cj'cle of ento-
tentacles, and the exotentacles are now being relegated to a fourth cycle.
At this stage, therefore, some of the exotentacles may be considered as
belonging to the third cycle and some to the fourth cycle. Under the circum-

74 SIDERASTREA RADIANS.

stances it was not possible to say whether the new exotentacles appeared in
advance of the entotentacles, but from the order actually observed in the
first three cycles it may be assumed that such was the sequence. It will be
observed that the new tentacular pairs are situated on the dorsal aspect of
the four sextants, along with one pair on the ventral side of the right middle
sextant.

New exotentacles and entotentacles have been found to arise shortly after
the appearance of new mesenterial pairs. Therefore, if the third-cj'cle mesen-
teries develop normally with the regularity outlined on p. 83, a single ento-
tentacle and exotentacle would be expected to appear on the dorsal aspect of
each sextant, and then the corresponding ventral series would commence,
both starting with the dorso-lateral systems and proceeding to the ventro-
lateral. In a colony where the polyps are so closely arranged as in Sideras-
trea^ however, the later growth rarely proceeds according to such a regular
law, so that very little importance can be attached to the order of appearance
of the tentacles making up the outer cycles. All the studies on corals seem
to indicate that the mesenteries, tentacles, and septa develop conformably to
a regular plan for the first and second cycles, but that the regularity is
frequently departed from in the later stages.

If the cyclic hexameral plan of the tentacles in S. radians were com-
pleted the formula would be 6, 6, 12, 24. The first three cycles would be
entotentacles and the last exotentacles ; the members of any one cycle would
be equal in size, but those of the inner cycles somewhat larger than those of
the outer cycle. The entotentacles would have appeared in a fairly regular
sequence, the inner cycle first, the middle cycle next, and the third cycle
last ; but it is altogether otherwise with the fourth cycle, constituted only of
exotentacles. Six of its members appeared a little in advance of the first
cycle of entotentacles; other six a little in advance of the second cycle of
entotentacles ; and twelve more in advance of the third cycle of entotentacles.
The exotentacles at each stage form a complete cycle ; first a cycle of six,
then a second cycle of twelve, which afterwards becomes a third cycle, and,
lastly, a third cycle of twenty -four, which afterwards becomes a fourth cycle.
In this way they always constitute the last or outermost cycle for the time
being. At the termination of each stage the number of outer simple exoten-
tacles equals the sum of the bifurcated entotentacles making up the inner
cycles ; but, until the adult stage is reached, as a cycle they occupy the
position which later will be occupied by the entotentacles.

When, however, the hexameral cyclic plan is not complete in the adult,
that is, where the last cycle of entotentacles commenced is not completed

POSTLARVAL DEVELOPMENT. 75

according to the hexameral sequence, then the formula must be different from
that given above if it has to express the true morphological value of the tenta-
cles as entotentacles and as exotentacles. Cyclic incompletion has been
shown to be practicall}? ahva3-s the condition in adult poh'ps of S. radians ;
two c^'cles of entotentacles are hex anierousl}- complete, but the third c^xle of
entotentacles consists of a variable number, rarel}- reaching the required num-
ber twelve. Regarded cjxlicalh', some of the exotentacles belong to the third
and some to the fourth cycles ; but until thej- form a complete hexameral
c3-cle, it is preferable to consider the exotentacles simply as the outermost
tentacles, not as forming a cycle. The morphological formula will thus be
6, 6, X, 6 + 6 + X, where x may be any number from one to twelve ; the
series 6, 6, x, will represent the number of entotentacles, and 6 + 6 + x
the number of exotentacles. The number of exotentacles will vary in the
same degree as the number of third-cycle entotentacles. Later, a similar
formula is established for the septa.

The different tentacular cycles in the adult polyps of Siderastrea are
much more widely apart than in the larval polyps (cf. plate 6, fig. 31 ;
plate 3, fig. 17). In the latter the closeness of the cycles recalls the condition
characteristic of the majority of anemones and corals.

The principal facts concerning the development of the tentacles of S.
radians may be now summarized.

1. The first tentacles to appear consist of a cj'cle of six members, one
from each of the six primary exocceles. They arise simultaneously, are
equal in size, and remain simple throughout.

2. Shortl}^ afterwards six smaller tentacles arise, forming a second cycle,
situated internally to the first, and communicating with the six primary
entocoeles. They may develop either simultaneousl}' or in successive pairs
from the dorsal to the ventral aspect. They are not symmetrically situated
as concerns the mesenterial chambers. Later, another entoccelic tentacle
appears alongside each of the first members, and then a common peduncle is
formed which elevates the two moieties so as to form the single bifurcated
tentacle of the adult.

3. The second cycle of mesenteries having appeared, a new exotentacle
arises from each additional exocoelic chamber within the six primar}' systems
of the polyp. In general the succession is from the dorsal to the ventral
svstems, and the new tentacle is on the ventral aspect of the older exoten-
tacle belonging to the same system, but many variations occur. The six
new exotentacles arrange themselves in the outermost or second cycle, which
thus consists of twelve members, equal in size, and all simple.

76 SIDERASTREA RADIANS.

4. Additional tentacles are now intercalated between the inner and outer
cycles already established. They arise, apparently without an}^ regular
sequence, from the six entocceles of the second pairs of mesenteries, and, like
the members of the first cycle of entotentacles, are at first simple, but after-
wards bifurcated. When the cycle is completed it constitutes the second
tentacular cycle of the mature polyp ; the original second cycle of six
exotentacles, now consisting of twelve members, becomes at this stage the
outermost or third cycle (law of substitution).

5. As the pairs of third-cycle mesenteries develop, two tentacles, one
entocoelic and one exoccelic, arise in connection with each. The entocoelic
representatives appear between the second and outermost cycles of the previous
stage, and thus constitute a new third cycle ; the exoccelic representatives
arrange themselves among the members of the outermost cycle, which now
becomes the fourth of the tentacular cycles. Here, as in the two previous
cycles, the entotentacles are at first simple, then another moiety appears, and
finally a common stalk.

6. The hexameral cyclic plan of the third-cycle entotentacles in S.
radians being usually incomplete, the morphological tentacular formula for
the whole polyp with four cycles is 6, 6, x, 6 + 6 + x, where x may be any
number from one to twelve.

MESENTERIES.
FIRST CYCLE OF MESENTERIES (pROTOCNEMEs) .

The earliest larvae reveal, both externally and in sections, only eight
mesenteries, arranged in four bilateral pairs, two axial and two lateral. Of
these the two lateral pairs are already united with the stomodseum, but the
dorsal and ventral axial pairs are free, the ventral pair being a little larger
than the dorsal (plate i, fig. 3). In other larvae, only a day or two older, the
two ventral mesenteries have become united with the stomodseum, but the
dorsal are still free, and two new pairs, the fifth and sixth, are beginning
to make their appearance (plate i, fig. 4, and plate 8, fig. 51). At the stage at
which the larva settles four pairs of mesenteries, two lateral pairs and the
dorsal and ventral axial pairs, are complete, and the fifth and sixth pairs have
become a little larger, but remain free from the stomodaeum (plate i, fig. 6).
This is the stage represented by all the larvae at or shortly after fixation, and
is evidently one of importance in the ontogeny of the pol3?p, for no further
mesenterial development took place for a period of three or four weeks.

In transverse sections of the larvae the vertical muscle fibers of the
mesenteries are already sufficiently well developed to permit of the directive

POSTLARVAL DEVELOPMENT. 77

and unilateral pairs being established. On the dorsal and ventral axial
pairs the musculature is on the faces of the mesenteries turned away
from one another, which, by this means, are recognizable as directives.
The musculature on each complete lateral mesentery is on the ventral
face turned towards the incomplete mesentery, while this bears its muscle
fibers on the dorsally directed face, so that each unilateral pair of mesen-
teries at this stage is made up of a complete and an incomplete moiety
(anisocnemic).

On the reasonable assumption that the comparative sizes of the mesen-
teries indicate their order of appearance in the larvae, the two bilateral pairs
which first unite with the stomodaeum (plate i, fig. 3) are the first and second
pairs to arise. At the stage available little difference in size is represented
by these two pairs, so that it could not be readily determined which is the
first pair and which the second. The results of Yon Koch (1S97), Wilson
(1888), and others who have studied earlier stages of coral larvae, leave no
doubt, however, that the ventral of the two pairs is the first, and that the
dorsal pair is the second. Frequently in other corals the first pair is for a
time much better developed than the others, and bears well-defined filaments
(1902, p. 528). From the relationships shown on plate i, fig. 4, and
plate 8, fig. 51, there can be no uncertainty as to which are the third
and fourth mesenterial pairs. The ventral pair of directives is larger,
and unites with the stomodaeum in advance of the dorsal pair ; it is there-
fore the third pair in the mesenterial sequence, and the dorsal directives will
be the fourth.

The incomplete mesenteries on plate i, fig. 6, representing the fifth and
sixth bilateral pairs, arise practically at the same time, and for the most part
remain equally developed. The more dorsal of the two in some instances
outstrips the other, and in other corals and actinians is frequently found to
unite with the stomodaeum in advance of the more ventral. Whatever diflfer-
ences exist in size or time of union with the stomodaeum support the view
now generally held that the incomplete pair of mesenteries between the first
and second pairs is to be regarded as the fifth, and hence the pair between
the first and third pairs is the sixth.

The sequence of the protocnemes thus established for Siderastrea
radians is in strict agreement with that ascertained by H. V. Wilson (1888)
for Mafu'chia areo/ala, and by Von Koch (1S97) for Caryophyllia cyathus,
both of whom had a complete series of larval stages for investigation. It is
also supported by the various early stages in other corals which I have had
under observation (1902, p. 450). The same order is followed by nearl}' all

78 SIDERASTREA RADIANS.

actinians,* thougli the evidence for the Ceriantheee and Zoantheas is not yet
conclusive.

The length of time the young polyps of Siderastrea remained with
only twelve mesenteries — eight macrocnemes and four microcuemes — would
indicate that the stage is one of phylogenetic, as well as ontogenetic,
significance. No marked resting stage in the mesenterial development
occurred until this was reached, while none of the pol3'ps showed further
progress until the expiration of three or four weeks, and some of the smaller
polyps remained thus for three months. Further, the fact that four pairs
united with the stomodseum, while two pairs remained separate for the whole
seventeen weeks, the six pairs of second-cycle mesenteries arising in the
meantime, would suggest a different significance for the complete and incom-
plete groups (S + 4).

The relationship, 8 + 4, is characteristic of the larvae and young polyps
of Wadreporaria and Actiniaria generally, and persists in some species
throughout the life of the polyp. Only the first four pairs of mesenteries
are ever united with the stomodaeum in the Edwardsidae, Gonaciinia, and
some other Actiniaria, and in the mature polyps of all West Indian species
of Madrepora and Poriies the same eight protocnemes alone are complete,
and the other four remain incomplete.

The long retention of freedom of the fifth and sixth pairs of protocnemes
suggests to my mind an ancestry in which the mesenteries as a whole, includ-
ing the metacnemes, were alternately long and short, excluding, of course,
the axial directives. Among modem examples this is retained in the mesen-
terial system of the zoanthids, Porites^ and Madrepora^ and was perhaps
characteristic of the Rugosa. The arrangement of the musculature on the
mesenteries in the Ceriantheas, always on the face towards one aspect of the
polyp, can be also understood if one considers that the incomplete mesenteries
present in such forms as the zoanthids are never developed; the cerianthids
retain the simplest protocnemic stage of any of the forms here considered,
having only four bilateral pairs.

In the Ceriantheae and Zoanthese the mesenteries beyond the protocnemes
arise at only one or two restricted regions of the pol3'p (bands of proliferation),
not all round the circumference, as in most modern anemones and coral
polj'ps ; and four such zones or bands of proliferation would appear to have
been characteristic of the mesenterial growth in Palaeozoic coral polyps. The

* Appellor (1900) has discussed fully the value to be assigned the accounts of Lacaze-Duthiers, Haddon,
and others as to various other sequences of the primary mesenteries in actinians. Throughout their work
Delage and Hirouard (1902) transpose the order of the fifth and sixth pairs as given above.

POSTLARVAL DEVELOPMENT. 79

mesenteries and septa in all these are never more than dicyclic. With the
later introduction of a radial polycyclic arrangement of the mesenteries and
septa, by the addition of exoccelic pairs in all the sextants, the fifth and sixth
protocnemic pairs begin to unite with the stomodosum, and so produce a
more approximate radial symmetry in the adult. But that the alternately
macrocnemic and microcnemic arrangement is more primitive seems to be
still suggested in the long separation from the stomodseum of the fifth and
sixth protocnemes, found to be characteristic of recent actiniarian and
madreporarian pol3'ps.

The first twelve mesenteries in Siderastrea have been shown to arise in
bilateral pairs, a member on each side of the median axis, and the order is
such that each new pair appears alternately in the successively oldest
chamber. As the mesenteries become fully established another paired
arrangement is introduced. The directives throughout the life of the polyp
constitute bilateral pairs ; but now the second and fifth mesenteries on each
side form what I have termed unilateral pairs, and likewise the first and
sixth mesenteries on each side. Though varying so much in their order of
appearance and in their primary relations to one another, the six pairs in
the end constitute a regular cycle and appear all of the same value ; at first
they are anisocnemic pairs, but they become isocnemic when the fifth and
sixth bilateral pairs come into union with the stomodasum. Thus a truly
radial disposition of the parts results from a primary bilateral origin, a result
found to be continually recurring in the development of the different systems
of organs.

SECOND CYCLE OF MESENTERIES (METACNEMEs) .

For a period of several weeks after fixation no addition to the twelve
protocnemes took place ; the four incomplete members continued to increase
in size, though remaining free from the stomodseum. Mesenterial filaments
began to show as dense, more opaque tissues within the interior of the polyp,
but their relative development could not be followed in the living polyp.

Towards the end of the fourth week some of the polyps presented
rudiments of the second-cycle mesenteries. Their first appearance externally
was as two narrow lines along the column wall, towards its aboral termina-
tion, and within the dorsal exoccele of the right and left sides. A few days
afterwards these were followed by a similar pair of lines within each of the
two middle exocoeles, and still later by a pair within each ventral exoccele
(plate 3, fig. 14). The actual time of appearance of the dorsal pairs, and
also the intervals between the three sets, varied somewhat in different polyps ;

8o

SIDERASTREA RADIANS.

the periods here given are for the most forward specimens. The six pairs
together constituted the second cycle of mesenteries or first cycle of metac-
nemes, the different pairs diminishing in size from the dorsal to the ventral
aspect of the polyp. Their order of appearance and comparative sizes are
diagrammatically shown in fig. 8 [d-f).

The interval between the appearance of the dorsal and middle, and
between the middle and ventral pairs, was usually well marked, so that at

Fig. 8 ifl,b,c,d). — Series of diagrams illustrating the order of development of the
first three cycles of mesenteries, a-c represent the order for the first cycle (protoc-
nemes), d, the first stage of the second cycle (metacnemes).

any time the different phases could be easily ob-
served. About four weeks elapsed between the
appearance of the first or dorsal pairs and that of
the third or ventral pairs. Usually, the pairs on
opposite sides of the polyp would arise simulta-
neously, but in one instance the left pair in the
ventral exocoele was apparent for over a week in
advance of the right pair in the corresponding
exocoele. In every instance the two members of a
unilateral pair appeared simultaneously and developed equally.

The metacnemes increased slowly in length, and reached the aboral
termination of the column long before they extended to the distal extremity.
For many weeks the pairs differed in their vertical extent corresponding
with their dorso-ventral appearance ; even at the close of the observations the
distinctions were strongly indicated, giving to the polyp a bilateral symmetry
(plate 3, fig. 17). After the third month they began to extend across the
oral disc, the dorsal pairs being first recognizable, then the middle, and later
the ventral. They never, however, reached the stomodaeum, this condition
being also characteristic of the adult polyp.

POSTLARVAL DEVELOPMENT.

8l

A dorso-veutral sequence in the development of the six pairs of second-
cycle mesenteries, as compared with their simultaneous appearance, has long
been known in actinians ; in fact, from the time of the Dixons' account of

Illb

Fig. 8 (',/,/, h), continued.— Scrits of diagrams illustrating the order of development of the first three cycles of mesenteries.
e and/ represent the second and third stages for second cycle, andf and h are two early stages of third cycle.

the relative sizes of the pairs in certain late larvae of Bunodes. Apparently,
however, the present is the first occasion on which the actual appearance has
been followed stage by stage in the living polyp.

The developmental stage now reached is one which frequently recurs
in zoantharian studies. Of the protocnemes, the eight Edwardsian members

82

SIDERASTREA RADIANS.

are complete and the fifth and sixth pairs are incomplete; the six pairs
of second-cycle mesenteries are incomplete, and either eqnal or show by
differences in sizes their dorso-ventral succession. The stage may be found
among adult Edwardsias, though two or more pairs of second-cycle mesen-
teries may here be wanting; adult members of the Ilyanthidas often present
it, and probably all higher actinians in the course of their development. It
is repeated b}^ the many pelagic anthozoan larvse described by Van Beneden
(1897, p. 189), numbered viii to xv, and which the author regards as belong-
ing to the Hexactiniaria.

Illb

Illb

lib

ille

Fig. 8 iy,k), continued.— Series of diagrams illustrating the order of development of the first three cycles of mesenteries,
y is a late stage, and /t the complete stage in the development of the third cycle.

A long interval of time, and a fundamental difference in their manner
of appearance, are found to separate the second-cycle mesenteries from those
of the first cycle. The primary mesenteries appear in bilateral pairs first
towards one aspect of the polyp and then towards the other aspect, and so
on, and only later constitute unilateral pairs. The secondary mesenteries,
on the other hand, are in unilateral, isocnemic pairs from the beginning,
and arise within the six primary exocoelic chambers in a regular dorso-
ventral succession, alternating with the primary pairs. Though presenting
such a difference in development, the two series agree in each forming a
cycle or order of six isocnemic pairs when mature.

It was to emphasize these fundamental differences in origin between the
primary mesenteries and the later members that, in a former paper, I termed

POSTLARVAL DEVELOPMENT. 83

the first cycle of six pairs of mesenteries " protocnenies," and all subsequent
mesenteries " metacnemes," whether becoming arranged in cycles or not.
The protocnemic stage is practically the same for all the different groups of
Actiniaria and Madreporaria. It is in the metacnemic stage that distinguish-
ing characteristics are introduced, development in all cases proceeding in a
manner altogether different from that of the protocnemic stage. The interval
is thus one of the greatest significance in the pli3-logeuy of the Anthozoa, as
well as in the ontogeny of the individual polyp.

The metacnemic development here established for Siderastrea is that
most usual for ordinary anemones and corals ; new isocnemic pairs appear in
the six primary exocoeles all round the polypal wall. In the group of the
Zoantheae the metacnemes appear at only two restricted regions, one within
the exocoele on each side of the ventral directives, and each unilateral pair
consists of a large and a small member (anisocuemic). In the Ceriantheae
the metacnemes appear as bilateral pairs at only one region, within what
seems to be the entocoele of the ventral directives. Furthermore, the metac-
nemes in the Zoantheae and Ceriantheae do not form one or more cycles
distinct from the protocnemes (monoc3'clic), as in Siderastrea^ most other
modern corals, and ordinary anemones (pol3'cyclic).

These differences in the origin of the metacnemes separate in the clearest
manner the zoanthids and cerianthids from ordinary actinians and corals.
On the other hand, the metacnemic similarity in actinians and madreporar-
ians proves -that the two groups are much more closely allied to one another
than to a:iy other group ; the only important difference between the two
consists in the presence or absence of a calcareous skeleton.

In general, it will be found that in the Actiniaria and Madreporaria the
organs as a whole bej^ond the protocnemic stage develop successively from one
border of the polyp to the other. The adult cyclic arrangement is clearly a
later modification of a primar}^ dorso-ventral plan ; in some instances, however,
the cyclic tendency strikes back, as it were, to the first appearance of certain
of the organs, for the prototentacles and protosepta usually arise a cycle at
a time. The second-cycle septa in S. radians are also interesting in this
respect, for in some polyps the six members of the cycle appeared simulta-
neously, but in others in successive pairs (p. 87).

THIRD CYCLE OF MESENTERIES.

The polyps reared from the larva were not kept alive beyond the com-
pletion of the first and second cycles of mesenteries ; hence for what follows
as to the order of appearance of the third cycle recourse will be had to

84 SIDERASTREA RADIANS.

colonial polyps arising asexually as buds. Fortunately, in any colony,
individuals of various sizes occur, and thus all the stages in the growth of the
third cycle can be secured. As already traced, the mesenteries consist of
six equal pairs of protocnemes all united with the stomodseum, the fifth and
sixth pairs having become complete, and six smaller, incomplete pairs of
metacuemes, alternating with the former. Following the laws of hexactinian
cyclic symmetry the third cycle of mesenteries should consist of twelve equal
pairs situated within the exocoeles formed by the first and second cycles;
the problem is to determine their order of appearance.

The earliest stage obtained in the formation of the third cycle of mesen-
teries is diagrammatically represented in fig. 8, ^, p. 8i, taken from one
of the small bud polyps of a colony. In addition to the primary and second-
ary cycles an isocnemic pair of mesenteries has appeared on each side of the
median axis, in the exoccele between the dorsal directives and the dorsal pair
of second-cycle mesenteries. Such an early stage would be expected on the
dorso-ventral succession already established for the second cycle. The
chamber within which the next pair of mesenteries will arise is, however,
one of much significance. The succeeding exocoelic chamber on each side
is between the dorsal pair of second-cycle mesenteries and the dorso-lateral
pair of first-cycle mesenteries, and it might be supposed that the new mesen-
teries would occupy the exocoeles in regular succession, from one border of
the polyp to the other. Instead of this it is found that the pairs arise succes-
sively within only the dorsal of the two exocoeles of each system. This
condition is shown in fig. 8, //, p. 8i, the next stage available, where a third-
cycle pair (ill) is found within the dorsal exoccele of each of the six systems.

A later stage secured in the establishment of the twelve third-cycle
mesenteries is given in fig. 8,/, p. 82, where an additional pair (iii, d) has
appeared, this time within the ventral exoccele of the two dorsal systems.
Clearly, if the succession here indicated were followed with perfect regu-
larity, other pairs would appear within the ventral exocoeles of the middle
and ventral systems, and the cycle would then be completed according to
fig. 8, k, p. 82. No stage exactly corresponding with this figure, however,
has been obtained, as the polyps of S. radians very rarely, if ever, complete
the third cycle of mesenteries.

The stages above presented prove that in the establishment of the third
cycle of mesenteries the dorso-ventral succession is twofold : First, a series
of six pairs within the dorsal exocoeles, and then a similar series within the
ventral exocoeles of each system.

The regularity in the sequence, represented by fig. 8, g-k, was secured

POSTLARVAL DEVELOPMENT. 85

only after examination of a number of polyps. In a colony in whicli the
polyps are so closely arranged as in S. radians it is found that the individuals
rarely undergo their later development with perfect regularity all round —
some regions will be in advance of the normal sequence and others behind.
The polygonal form assumed by the adults is evidence that pressure is
exerted upon a form which otherwise would be circular, as in simple polyps
reared from larvae. Spatial difficulties may therefore be held sufficient to
account in a large degree for the many iri'egularities obtained in the estab-
lishment of the third mesenterial cycle.

The mesenterial plan of two other polyps is given in figs. 3 (p. 26) and
1 1 (p. 100), and illustrates the variability encountered. In fig. 3 the sequence
is normal except that a pair of mesenteries (iii<?) has appeared within the
ventral exoccelic chamber of the right middle system in advance of the pairs
in the dorsal exocoeles of the ventral systems. The polyp represented in
fig. II presents many departures from the normal regularity ; two third-cycle
pairs occur within three systems, one pair within another, and two systems
are without any third-cycle pairs.

In Astrangia solitaria and Phyllajigia atnericana, where the polyps
are practically separated from one another and retain their cylindrical form
throughout, the regularity of development all round is more pronounced,
and the general order of appearance of the mesenteries established in S.
radians is found to be maintained from beginning to end (1902, p. 459).

Hitherto, the development of the third-cycle mesenteries has not been
actually followed either in actinians or corals. Faurot's studies in 1S95 were
confined mostly to the tentacles. Carlgren* has described a coudition of the
mesenteries in the actinian Condylactis cnientata^ in which the twelve pairs
of third-cycle mesenteries as a whole, as well as the exoccelic chambers in
which they are situated, show a gradual decrease in size in passing from the
dorsal to the ventral border of the polyp. Pairs \\\a and \\\d in the enumera-
tion of fig. 8, k^ p. 82, were larger than pairs iwb and iii^, and these than
pairs \iic and iii/; hence, if the condition obtained by Carlgren really repre-
sents the sequence followed in the growth of the third-cycle mesenteries in
Condylactis^ it is altogether different from that of corals, being simple instead
of twofold.

I believe it will be found in corals generally that the sequence of the
later mesenteries is by no means so regular as that of the earlier C3'cles.
The order followed by the organs in the first and second cycles is fairly con-
stant, but this can not be asserted of the third cycle, and probably the regularity

*"Zur Mesenterienentwicklung der Aktinien," Ofvers afR. vet.-Akad. Forh., Stockholm, 1897.

86 SIDERASTREA RADIANS.

will be even less in still higher cycles. As the polyps increase in size the
forces of growth are less likely to act with the regularity and uniformity
which they do in the earlier stages when the polj'ps are small; even though
in the end the cycles obtain hexameral completion it will be brought about
with much individual variation.

No polyps of S. radians having mesenteries belonging to a fourth cycle
have been found, and nothing is 3-et known as to the normal seqiience
according to which the cycle is established in other species.

The chief facts concerning the mesenterial sequence in 5'. radians may
be now summarized :

1. The six pairs of first-cycle mesenteries (protocnemes) arise as bilat-
eral pairs in a regular alternation from one aspect of the polyp to the other.
The first four pairs early unite with the stomodaeum, but the two last pairs
(v, vi) remain free for a long period. Later, the second and fifth, and the
first and sixth mesenteries, on each side, form isocnemic pairs, and the third
and fourth pairs constitute the directives.

2. The six pairs of second-cycle mesenteries arise bilaterally as uni-
lateral isocnemic pairs on each side of the polyp, and appear successive^
in the primary exocoeles from the dorsal to the ventral border ; ultimately
they become equal and exhibit perfect radial symmetry.

3. The twelve pairs of third-cycle mesenteries also arise bilaterally as
unilateral isocnemic pairs on each side. Normally six pairs appear in a
successive manner from the dorsal to the ventral aspect of the polyp, a pair
within the dorsal exocoele of each sextant ; then other six pairs appear in
the same succession, a pair within the ventral exocoele of each sextant.
Generally some of the pairs of third-cycle mesenteries are wanting in mature
polyps.

CORALLUM.

FIRST CYCLE OF ENTOSEPTA AND SECOND CYCLE OF EXOSEPTA.

Three or four days after fixation of the larva the skeleton was first
observed through the transparent tissues of the living polyp in the form of
six small radiating septal upgrowths, practically equal in size. At the
same time a narrow peripheral calcareous ring was seen, its outer surface
uncovered by the polypal tissues (plate i, fig. 7). The six septa were per-
fectly free from one another and from the outer annulus, and arranged at
equal distances apart within the six entoccelic chambers, thus alternating
with the cycle of six exocoelic tentacles first to arise. Each septum appeared
as a somewhat spindle-shaped bar with the upper edge strongly spinous and
the lower edge flat and adherent to the glass to which the polyp was affixed.

POSTLARVAL DEVELOPMENT. 87

A da}' or two after the formation of the first cycle of entosepta, the six
exocoelic septa began to make their appearance, in some cases simultane-
ously, but in others in successive bilateral pairs from the dorsal to the
ventral border of the polyp (plate 2, figs. 8, 9). Fig. 8 shows that a pair
of septa has appeared withiu the dorso-lateral exocceles, mere rudiments of
septa are found in the middle exocoeles, and as yet there is no indication of
the ventro-lateral exoccelic pair. For a long time, as shown by the coralla
on plates 4 and 5, the dorso-lateral pair is better developed than the middle
pair, and the middle than the ventro-lateral pair. The ventral pair in nearly
all cases remained conspicuously smaller than the other pairs. Where, how-
ever, the second cycle is fully developed (fig. 9), its six members are disposed
midway in the six interspaces between the members of the primary cycle,
and remain shorter than the latter. As in the entosepta the surface is spinous
along the edge and over both lateral faces.

Thus, within the first week two complete cycles of septa (protosepta)
were developed — a primary C3'cle, consisting of six equal entosepta, and a
secondary cycle of six smaller exosepta, the latter having appeared later and
diminishing in size from the dorsal to the ventral border. A narrow periph-
eral calcareous ring, unconnected with the septa, was also formed at the
same time. This is shown by later observations to be a marginal continuation
or upgrowth of the basal plate (p. 115), and therefore to be regarded as an
epitheca. Only the Edwardsian mesenteries were united with the stomodasum,
and of the tentacles the six exocoelic members alone were developed.

The order of appearance of the twelve protosepta is thus in marked con-
trast with that of the twelve protocnemes with which they are associated.
The latter have been found to arise in bilateral pairs, first towards one aspect
of the polyp and then towards the other, and the six pairs (four macrocnemic
and two microcnemic) are fully established before any of the septa arise.
The six entosepta, on the other hand, appear simultaneously ; and such is
usually the case with the six exosepta in other corals, though not in
Siderastrea.

The symmetrical growth of the skeletal structures, represented on plate
2, figs. 9 and 10, generally took place only in completely isolated polyps,
free to develop equally all round, and even in these irregularities were some-
times introduced. Among the young polyps forming the aggregated minia-
ture colony in fig. 5, p. 60, the septal development was scarcely alike in
any two. Where, as at the two extremities of the colony, one polyp partly
overfolds another, only half the number of septa occurs, while in the others
the alternation of large and small septa is inconstant; further, the epithecal

88

SIDERASTREA RADIANS.

wall of two contiguous polyps is common along the lines of adherence, and
the outline, instead of being circular, becomes more or less polygonal as a
result of the mutual pressure.

The protoseptal stage was completed in nearly all cases within the first

xb -^'

Fig. 9 («-#). — Series of diagrammatic figures illustrating the manner of development of the septa from one to three cycles
{cf. plates 1-5). The epithcca is represented as a white circle distinct from the septa.

fortnight, and some of the smaller polyps never reached beyond ; indeed, in
a few examples, the exocoelic cycle was never completed. Much variation
was observable as to the rate at which the corallum was laid dowTi, growth
in larger polyps being always in advance of that in smaller individuals.

POSTLARVAL DEVELOPMENT.

89

Within the larger, more vigorous polyps, further calcareous upgrowths
began to be formed peripherally during the course of the third week, some
appearing as angulated extensions of the primary septa and others wholly
independent of them (plate 2, fig. 12). Some of tlie isolated skeletal elements

I
■^ nil X

/

h J

Fig. 9 (f-j), continued.— Serits of diagrammatic figures illustrating the manner of development of the septa from one to three
cycles (c/. plates 1-5). The epithcca is shown uniting the peripheral ends of the scpu.

seemed like short additional septa, situated on different radii from the septa
first formed. The new members either retained their independence for a
long period or early became fused with the original septum in whose inter-
space they had appeared. In this latter instance the original septum was

90 SIDERASTREA RADIANS.

distinctly angulated at its peripheral end, and where two additional elements
were introduced in each chamber the periphery of the septum was strongly
bifurcated. In some few cases no separate fragments whatever would arise
within an interspace. The septum would then retain its original bar-like
form, but become longer and usually more rugged in outline, showing that
skeletal matter was being added. The deposition of new matter was usually
more forward in the axial chambers than in the lateral, but scarcely anj-
two polyps were alike in the detailed appearance assumed b}'- the skeleton
at this period.

Figs. 12, 19, and 20, on plates 2 and 4, represent the actual polyp or coral-
lum during this stage, while fig. 9, d^ p. 88, is an attempt to represent diagram-
matically the septal conditions at its close. It is a well-defined phase, and
important as showing the diflferent methods by which the septa may increase
in length and complexity. Although, as above remarked, no two polyps were
exactly the same, a general plan was determinable throughout the many
examples studied. The septa in all but the pair of ventral exosepta con-
sist of a simple, more central portion and the peripheral additions. The
former represents the enlarged primary septa seen in fig. 9, r, p. 88, while
the latter are newer formations. In general the peripheral deposits consist
of two or more separate fragments, placed at an angle with the central bar,
the angle being greater in the exosepta than in the entosepta ; the entocoelic
additions are, in fact, nearly parallel with the primary entoseptnm, and thus
more strictly radial. The two axial or directive septa are usually the most
strongly developed of the entire series, and remained thus to the end ; in
addition, the dorsal and the ventral frequently differ much in form from
one another.

In all the figures it will be seen that the two ventro-lateral exosepta
remain simple, enlarging but little be3'ond their primary condition, and
the middle exosepta are likewise somewhat less developed than the dorso-
lateral. Thus a decided dorso-ventrality is indicated in the growth of the
peripheral elements of the skeleton, as was the case with the simple exosepta.

For many weeks afterwards the only alteration in the septa consisted
in increased growth along the plan thus laid down. As the peripheral frag-
ments enlarged, the members of any group became fused with one another.
After the second month the development of the skeleton within the living
polyp could with diificulty be followed, owing to the complexity of the
internal tissues. Therefore, from this stage onward the septa will be studied
mainly from macerated corallites.

Fig. 23, plate 4, is from a photograph of the skeleton of a polyp ten

POSTLARVAL DEVELOPMENT. 91

weeks old, in which oiil}' six pairs of primary mesenteries and two cycles of
tentacles were present, while fig. 9, c, p. 88, is its diagrammatic representa-
tion. Each septnm now bears a closer resemblance to the septa in the
mature corallite, being narrow centrally and broad peripherall}', with spinous
projections over the entire surface. The general relationships of the septa
to the mesenterial pairs could be made out before maceration ; hence there is
no uncertainty as to the orientation of the corallum. The dorsal and ventral
directive entosepta present a somewhat bifurcated peripheral extremity, the
two limbs being nearly parallel, but the four lateral entosepta are simple
bars, with thickened peripheral ends formed from the enlargement and
complete fusion of the originally separate elements. On the other hand,
the dorso-lateral and middle pairs of exosepta are strongly bifurcated periph-
erall}', while the members of the ventral pair retain their simple character, and
are by far the smallest of the six pairs. The peripheral limbs in the bifurcated
exosepta may be quite separate from the single radial piece, though usually
they are joined. It is readily seen how by the enlargement and fusion of the
manj' detached fragments within each mesenterial space of plate 2, fig. 12,
such a septal condition as that of plate 4, fig. 23, has been obtained.

With the increase in thickness of the septa the interseptal loculi are
correspondingly diminished, but no actual synapticular unions are yet formed
across the interspaces. The dorso-lateral exosepta bend laterally and nearly
fuse centrally with the dorsal directive septum ; the middle exoseptura on
each side approaches the corresponding dorso-lateral entoseptum and fuses
with it, while the small ventro-lateral exoseptum on each side is fused with
the ventro-lateral entoseptum (fig. 9, e^ p. 88).

Much older corallites, but still at nearly the same stage of development,
are shown on plate 4 (fig. 24) and on plate 5 (figs. 25-27), and diagrammati-
cally b}' fig. 9,/, p. 89. The dorso-lateral exosepta are now fused with the
dorsal directive septum, the middle exosepta with two dorso-lateral entosepta,
and the ventro-lateral exosepta with the ventro-lateral entosepta. The exo-
septa, generally, are not so strongly bifurcated as in the corallite on plate 4,
fig. 23 (fig. 9, ^, p. 88). Fig. 10, a-d,, on p. 96, shows the diagrammatic rela-
tionships of the septa to the mesenteries throughout these early stages.

The general impression produced by the corallum at this stage, as seen
under a simple lens or low pow'er of the microscope, is that of two alternating
cycles of septa — a larger and a smaller. The tendency to fusion of adjacent
septa, exosepta with entosepta, which was found to be so marked a feature
of the adult corallite, is already exhibited by the two cycles, and gives a
bilateral symmetry to the calice.

92 SIDERASTREA RADIANS.

Four distinct stages in the development of the protosepta of Siderastrea
radians are thus recognizable :

1. The simultaneous appearance of six equal entosepta, a few da3'S after
the larva settles. A very narrow epitheca, distinct from the septa, appears
about the same time.

2. The appearance several daj's later, either simultaneously or in a
dorso-ventral sequence, of six exosepta which are smaller than the entosepta
and alternate with them. The septa of both cycles are simple, spinous
wedge-shaped upgrowths from the basal plate.

3. The appearance towards the periphery of most of the septa of one or
more short septiim-like bars or skeletal nodules.

4. The fusion of these detached fragments with the main septa, so as
to give rise to a broad peripheral termination which may be either simple
or bifurcated ; also the fusion of the exosepta with the entosepta by their
inner extremity.

A distinct dorso-ventrality in the rate of growth is maintained from the
second stage onwards, particularly with regard to the exosepta, thereby giving
a bilateral symmetry to the calice.

The protoseptal development of Siderastrea presents a general agree-
ment -with that of other corals whose earl}' history has been followed.
Lacaze-Duthiers (1873, 1897), however, found that in Astroides calycularis,
Balanophyllia regia^ Leptopsammia^ and Cladopsamniia the six exosepta
appeared simultaneously with the six entosepta ; but in Caryophyllia cyathiis
and others there is an interval between the two cycles as in Siderastrea,
whereas in Manicina areolata, as I have shown (1902, p. 491), it appears
to be doubtful whether exosepta ever appear. Lacaze-Duthiers represents a
a decided bilateral condition in the early development of the skeleton in
Astroides (1S73, plate xiv, fig. 29), but in other known cases the septa of
each cycle appear simultaneously, and are equal from the beginning.

The simultaneous appearance of the members of one or both cycles of
protosepta and also of the prototentacles may be compared with the succes-
sive appearance of the pairs of protocnemes. The protosepta and prototen-
tacles resemble one another in that both appear a cj'cle at a time, and from
the beginning exhibit radial symmetry, whereas the protocnemes arise in
bilateral pairs according to a well-defined succession, and, for a time, display
a strong dorso-ventrality. The simultaneous appearance of all the members
of a cycle is maintained for the septa onl}' so far as the first cycle or, at most,
the second C3-cle of the protosepta ; in the later growth of the organs a dorso-
ventral sequence is followed, quite as conspicuous as that of the mesenteries.

I'OS ri.ARVAL DEVELOPMENT. 93

When it is recalled that the twelve primary mesenteries and their cham-
bers are full}^ established before the tentacles and septa begin to make their
appearance it can be nnderstood how these latter organs ma}' arise a C3'cle
at a time. In their development beyond the protocnemic stage tlie mesen-
teries, tentacles, and septa follow one another very closely, and such would
probably be the case were the different pairs of protocnemes closely succeeded
by their tentacles and septa.* The organs as a whole in the Zoantharia are
unquestionabl}- to be regarded as developing in a bilateral dorso-ventral
order, not in a cyclic manner. Hitherto the polyps in Siderastrea are alone
in the throwing back of the dorso-veutrality of the skeleton as far as the
exocoelic protosepta.

The forked or bifurcated continuation of the protosepta, produced either
by continued peripheral growth or by the production of independent nodules,
has been observed both in Astroides calycularis and in Caryophvllia cyaihus,
in addition to the present species. Both Lacaze-Duthiers (1873, 1897) and
Von Koch (1882) give figures of the developing corallum which show that
the septa are prolonged peripherally much in the same way as in Siderastrea.
Von Koch at first considered that the forkings were concerned in the for-
mation of the theca b}' the union of those from adjacent septa, but in
Caryophyllia he found (1897) the true theca (Mauer) to arise independently.
In Siderastrea the bifurcations are found to be merely peripheral extensions
of the septa, becoming fused with the simple septum in the case of the
entosepta, but constituting two new septa in the case of the exosepta. They
in no way assist in the formation of a theca.

SECOND CYCLE OF ENTOSEPTA AND THIRD CYCLE OF EXOSEPTA.

Before describing the further development of the septa it will be helpful
to consider what are the septo-mesenterial relationships involved in passing
from a polyp with only two cycles of septa to one with three cycles. In the
first polyp only six pairs of mesenteries are present, and within the ento-
coeles of these are the six primary entosepta ; the alternating six members
of the second cycle of septa are within the exocoeles, on radii midway
between the entosepta (plate 2, fig. 9). In the second polyp with three cycles
of septa two alternating cycles of mesenteries are present ; the first-cycle
septa are contained within the entocceles of the first mesenterial cycle, the
second-cycle septa are within the entocosles of the second mesenterial cycle,

•Where in larval actinians onlv four pairs of mesenteries are present when the tentacles begin to
develop, only eight of the latter appear, one from each of the eight mesenterial chatnbers, and the
other four necessary to complete the hexameral plan arise after the establishment of the fifth and sixth
pairs of protocnemes.

94

SIDERASTREA RADIANS.

while the third-cycle septa are within the twelve alternating exocoeles, and
are therefore exosepta. Thus at both stages the exosepta form the outer-
most cycle [cf. fig. lo, b^ /, p. 96). The questiou naturally arises as to
whether, on the appearance of a second cycle of mesenteries, the exosepta
of the first stage, which there constitute the second cycle, remain as the
second cycle of entosepta of the later stage ; were they to do so the twelve
exosepta of the latter would be the only new formations. As regards their
actual position the six exosepta of the early polyp correspond with the
six entosepta of the later polyp, and it would be natural to assume that on
the appearance of the second-cycle mesenteries the latter have simply included
within their entocoeles the exosepta of a former stage, and then new exo-
septa have arisen within the newly formed exocoeles.

The latter is the view commonly held by writers on coral development.
Thus Delage & Herouard, in their "Traite de Zoologie Concrete" (1901, p.
55S), remark : " Quand, dans les interloges occupees par les septes du dernier
cycle, nait un nouveau cycle de couples de cloisons, celles-ci se forment de
part et d'autre du septa interloculaire qui, de ce fait, devient loculaire, et
bientot un nouveau C3'cle de septes se forme dans les nouvelles interloges
qui viennent d'etre formees. Les cj'^cles naissent successivement et jamais
un cycle ne commence a se former avant que le precedent soit complet."
Similarly, J. Stanley Gardiner (1902), in his account of the anatomy of
Flabellum rubrum, says (p. 133, italicizing added): "As the growth of any
corallite proceeds, more and more septa up to six cycles appear. The former
exoccelic order of septa become entoccelic by the development of new pairs of
mesenteries. The increase of mesenteries takes place pari passu with the
formation of new septa."

Unfortunately, the relationships involved in the above assertions have
not been actually followed, though from the known conditions no other
arrangement at first sight seems possible. The problem is one of the most
important in the developmental history of corals.

Without doubt the members of the primary cycle of entosepta in adult
polyps are the direct representatives or continuations of the six septa first
to arise, just as the six pairs of complete mesenteries throughout the life of
the polyp are the representatives of the protocnemes. Throughout their
later growth the primary septa retain their individuality, and remain within
the entocoeles of the primary C3'cle of mesenteries. Other considerations,
however, are involved in the relationships between the septa developing later
and those of the adult corallite.

Most of the polyps of Siderastrea remained for some time at the stage

POSTLARVAr, DEVELOPMENT. 95

already described on p. 92, the second-cycle mesenteries appearing in the
meantime and growing towards both the basal and the oral disc. As indi-
cated diagrammaticall}- in fig. 10, </, p. 96, the new mesenterial pairs corre-
spond with the space inclosed by the bifurcations of the dorso-lateral and
median pairs of exosepta, but seem as if about to embrace the incompletely
developed ventro-lateral pair.

About this time other calcareous upgrowths began to appear peripherally,
midway within the exoccelic bifurcations, and necessarily inclosed within
the eutocceles of the second-cycle mesenteries. Plate 3, fig. 15, gives the
discal view of such a living polyp in which the six pairs of second-cycle
mesenteries had been developed for some time. The new mesenteries, still
varying in size in agreement with their order of appearance, have now
begun to extend along the periphery of the disc, and the latter is resting
upon the upper edges of the septa with the tentacles fully expanded. The
septa are clearly seen through the transparent disc. Within the bifurcation
of each dorso-lateral and middle exoseptum has appeared an additional free
septum, included within the entocoele of the second-cycle mesenteries, and
in the same radius as the original exoseptum. Moreover, the exoccelic
bifurcations are now seen to be wholly exoccelic in position, situated in the
chambers between the pairs of the first and the second-cycle mesenteries ;
in the dorso-lateral sextants the forkings are free from the original median
exoseptum, but in the middle sextants they are united. The primary ento-
septa remain simple straight bars, and such is yet the condition of the
ventro-lateral exosepta. Fig. 10, e, p. 96, is a diagrammatic representation
of the polyp and corallum at this stage.

The new formations within the second-cycle entocceles suggest an inde-
pendent series of septa, and subsequent stages show that they represent the
second-cycle entosepta. The polyp of plate 3, fig. 15, in fact, displays the
early stages in the development of the second cycle of entosepta and the
establishment of a third cycle of exosepta, and in such a manner that the
actual relationships between the new septa and the second-cycle mesenteries
admit of no misinterpretation. The entosepta of the second cycle appear
peripherall}' as separate formations, but in their later growth centrally, as
shown on plate 5, figs. 28, 29, and fig. 10, yj p. 96, they come into union with
the primar}' exosepta, and the two then appear as a single continuous structure.
Further, each bifurcation of a primary exoseptum forms a new exoseptum,
belonging to a third cycle, and may be either distinct or united by its inturned
edge with a second-cycle entoseptum.

96

SIDERASTREA RADIANS.

Three important results in the development of the Madreporarian
skeleton are thus gained : (i) The primary exosepta do not continue their

Fig. io (o-/"). — Series of diagrammatic figures illustrating the relationships of the mesenteries and septa in the establishment
of the first two cycles of mesenteries and three cycles of septa.

growth peripherally in a radial manner, and constitute entosepta by becoming
included \vilhin the eutocoeleof the new second-cycle mesenteries. (2) The

POSTLARVAL DEVELOPMENT. 97

secondary entosepta are new formations, arising within the second-cycle
entocoeles independently of other septa; the second-cycle mesenteries never
embrace the exoccelic protosepta as such, but only after their fusion with
the new entosepta. (3) The appearance of the second-cycle entosepta is
in a dorso-ventral manner, not simultaneous, a cycle at a time.

Other coralla exhibit different stages, all pointing to the same conclu-
sion. That shown diagrammatically in fig. 9, //, p. 89, is in somewhat the
same stage as fig. 10, e. In the upper right sextant the second-cycle
entoseptuni (11) is y&t very distinct from the exoccelic bifurcations, but an
irregularity in the rate of growth is exhibited by the corresponding left
sextant, in that no entoseptum has yet formed. The growth in the two
middle sextants is in advance of it ; each of the sextants contains a primary
exoseptum, a secondary entoseptum, and two third-cycle exosepta as distinct
formations. As before, the development within the two ventro-lateral sex-
tants lags behind that of the middle and dorso-lateral sextants. Fig. 9, ^,
from another corallum, also shows important intermediate stages.

The next stages available are the coralla of the oldest polyps reared, two
of which are represented on plate 5, figs. 28, 29, and diagrammatically in fig.
9, y, p. 89, and fig. 10,/, p. 96. Each corallite now consists of three com-
plete orders or cycles of septa, the development within the two ventral
sextants having reached the same stage as that within the dorsal and middle
sextants. The first order contains six large septa radially disposed, the
directive septa a little larger than the others. The second order also
includes six septa, turned inwardly in such a manner that the two dorsal
fuse with the dorsal directive septum, the two middle fuse with the dorso-
laterals of the first order, and the two ventral with the ventro-laterals of the
first order. The third cycle contains twelve septa, the two adjacent members
fusing with each of the six septa of the second order. The coalescence of
the septa in this wajr imparts a bilateral symmetry to the calice which other-
wise would be radial. Further, adjacent septa are now for the first time
joined by synapticula. The polyps (plate 3, fig. 17) forming the coralla bore
two cycles of mesenteries, so that the first and second orders of septa are
entosepta, while the members of the outermost cj'cle are exosepta.

Previous stages have demonstrated that the members of the second
order of septa, although now continuous bars, have really a twofold origin.
The peripheral part of each appeared as a separate septum within an entoccele
of the second-cycle mesenteries and, later continuing its growth centrally,
fused with an exoseptum which was originally a constituent of the second
cycle of septa, so tliat now the two appear as a single septum. In each

98 SIDERASTREA RADIANS.

corallum there are indications still remaining of the distinctness of the
peripheral and central parts of the entosepta, as in the ventral sextants of
fig. 28, plate 5, where the entoccelic moiety is still free from the exoccelic
portion. The fusion of the two parts is a mechanical necessity in the
process of growth, seeing that both are on the same radii. It is manifest
that it is only by securing such intermediate stages that the true character
of the septa at the mature stage can be understood. When the secondary
entosepta have come into fusion with the primary exosepta there remains
no means of distinguishing their compound origin, and the ontogenetic and
phylogenetic significance of the secondary entosepta is obscured.

The twelve exosepta now forming the tertiary cycle undoubtedly arise as
continuations of the bifurcations of the six primar}' exosepta, but in
many places, as on plate 5, figs. 27 and 28, they still show a considerable
amount of distinctness. Regarding them as continuations of the two forks
of each primary exoseptum, it is manifest that they are to be considered
as appearing in advance of the entosepta of the second cycle, a relationship
which need not be wondered at, considering that in this species the exo-
tentacles are also found to arise in advance of the entotentacles. Originally
forming the second cycle, the exosepta now constitute the third septal cycle,
their place in the sequence being taken by the new second cycle of entosepta.

It is thus manifest that in the course of development of a coral the
exosepta of a former stage do not become the entosepta of a later stage ;
the latter are new formations appearing after the establishment of the
mesenteries with which they correspond, and consequently the mesenteries
and their included entosepta have the same ordinal value. The exosepta,
on the other hand, have no ordinal value ; they appear at each cyclic stage,
always constituting the outermost cycle. They are, in a measure, temporary
structures, predecessors of the entosepta, until the limit of growth of the
polyp is reached ; the}- serve as integral parts of the septal system of the
coral during all its intermediate stages, and are then overgrown by later
permanent septa.*

•In some corals the exoseptal predecessors appear to continue their independent growth in situ
without losing their individuality as skeletal structures in the central extension of the entosepta. In
these cases the entosepta do not grow far enough centrally to fuse completely with the exosepta already
there. I believe it will be found that this is the true nature of fait which are found in some corals as
small septum-like plates in front of the larger septa. The fact that pali seem not to occur before the
primary cycle of six septa, but only before those of later origin, is what we should expect if this surmise
be correct. The primary entosepta have never had exocoelic predecessors, as is the case with the later
entosepta. The pali would thus represent the persistent exocoelic predecessors of the entosepta beyond
the primary cycle.

POSTLARVAL DEVELOPMENT. 99

The closest morphological parallel is proved to exist between tlie develop-
ment of the septa and the tentacles. As previously shown, exotentacles are
present at each cyclic stage, but a new cycle of entotentacles intercalates
itself between the last cycle of entotentacles and the exotentacles, hence the
latter always remain as the outermost cycle ; only the entotentacles, like the
entosepta, have ordinal value. Thus the law of substitution iirst discovered
by Lacaze-Duthiers for the tentacles of hexactiuians is found to hold for the
septa also.

Belonging to the soft, flesh}' parts of the polyps, it can be easil}' under-
stood how actual displacement of the tentacles may be carried out, but such
is not possible with the hard, rigid skeleton. Hence the process of substitu-
tion must be conducted in a different manner in the two sets of structures.
A new peripheral eutoseptum arising independently can not displace an inner
exoseptura already occupying the same radius. In its growth centrall}' the
eutoseptum simply fuses with the exoseptum, and thenceforward the whole
of the septum must be morphologically regarded as an eutoseptum.

The septa, like the tentacles, are thus shown to arise in such a manner
that it is impossible to determine their order of development from their rela-
tionships in the mature polyp.

Though the septo-mesenterial relationships will remain the same, I
conceive that a like adult condition of the septa may be reached in differ-
ent ways in different forms of corals, as is the case with the tentacles.
The actual method followed in Sidcrastrea can by no means be assumed to
be that characteristic of corals generally. Probably some of the stages in
Siderastrea might be better interpreted were results available from other
forms; such, for instance, as the significance of the forking of the exosepta.
The fact that in Siderastrea the septa of one c3-cle fuse centrally with the septa
of the next inner cycle probably obscures the problem somewhat, as com-
pared with forms where the septa remain altogether free from one another.

THIRD CYCLE OF ENTOSEPTA AND FOURTH CYCLE OF EXOSEPTA.

None of the polyps reared from larv^ae were kept alive beyond the for-
mation of the first three cycles of septa, which consist of two cycles of
entosepta and one cycle of exosepta. Therefore the development of the
fourth cycle of septa must be studied from the bud polyps of a colony.
Fortunately, there are many polyps available for such an investigation, as
in any stock most of the individuals are at one stage or another towards the
establishment of a complete fourth cycle of 24 septa.

lOO

SIDRRASTREA RADIANS.

III2

Uli

Ills

nii

When describing the septo-mesenterial relationships in the mature
polyp, it was found that the fourth-cycle septa occur within the exocoeles
between all the adjacent pairs of mesenteries. On leaving the last section,
however, the third cycle was constituted of exosepta ; only the first and
second cycles were entosepta ; therefore, as in the case of the secondary and

tertiary C3'cles, the problem is first
to determine the morphological re-
lationship of the tertiary exosepta
of the developing polyp to the
tertiary entosepta, and also to the
quaternary exosepta of the mature
polyp. Do the tertiary exosepta
at the stage where only second-
cycle mesenteries are present be-
come the tertiary entosepta when
the third-cycle mesenteries appear,
or are they continued as the
fourth-cycle exosepta, in which case
the tertiary entosepta arise d^
novo ?

The relationships to be determined have been studied by means of
serial sections of nearly mature decalcified polyps. A complete series of
stages, representing the development of a pair of third-cycle mesenteries
along with the associated septa in their relation with the older mesenteries
and septa, is given on plate 9, figs. 54-60. The figures are taken from a
series of transverse sections of a retracted polyp, and will be described as
seen in passing from the lower stomodseal region to the uppermost margin
of the calice and polyp. On account of the presence of synapticula and the
peripheral resorption of the mesenteries, certain complications are introduced
which render a clear conception of the stages somewhat more difiicult than
would otherwise be the case. The diagrammatic figures on p. 103 will assist
in following the description of the sections.

The polyp from which the drawings were made is that of which the
mesenterial plan is diagrammatically represented in fig. 11, p. 100. Within
the ventral system on the left side occurs a pair of small third-cycle mesenteries
(11I3), situated on the dorsal aspect of a pair of second-cycle mesenteries
(11). Plate 9, figs. 54-60, represents the members of this ventral system
at different levels, and the mesenteries and septa are indicated by the same

Fig. II. — Diagram of mesenteries in the polyp from which figs.
54-60, on plate 9, were t.tkcn.

POSTLARVAL DEVELOPMENT. lOI

lettering {a~j) throughout. In the figures mesentery a is the lower moiety
of the ventro-lateral pair of first-cycle mesenteries, mesentery /" is the left
moiety of the ventral directives, mesenteries d and e are the pair of second-
cycle mesenteries, while b and c are the two members of the rudimentarj'- pair
of third-cycle mesenteries. The incompletely represented eutoseptum to the
left of «■ is a member of the primary cycle of septa, as is also the partial
septum to the right of /; the eutoseptum d-e belongs to the secondary
cycle, while the septa a-d and e-f, in figs. 54-56 are exosepta. The upper
margin of the section in fig. 54, is formed b}' the stomodaeal wall, while that
in all the subsequent figures is either the depressed disc or column wall.
This, upon retraction, has come to rest upon the septal edges, and adapts
itself to them in outline, being thrown into ridges and furrows.

Fig. 54 represents a section of the sextant taken from the lower stomodaeal
region. The members of the mesenterial pair, b^ c, of fig. 11, have not j-et
reached this level, so that the sextant contains only the pair of incomplete
second-cycle mesenteries, d, e. The mesenteries a and/" are united with the
stomodaeal wall centrally, but their peripheral extremity has undergone
resorption and, therefore, is free. At this level the second-cycle mesenteries,
</, ^, are feebly represented, being much reduced peripherally as a result of
resorption. The eutoseptum d-e is broad, but the exosepta a-d and e-f on
each side of it are narrow and partly turned towards it. Exoseptum a-d is
bifurcated at its peripheral end, but before dividing is connected by synap-
ticula with the entosepta on each side. Within the angle of bifurcation
occurs a small, empty loculus.

Plate 9, fig. 55, is from a section immediately above the level of the
depressed stomodaaum, so that it is bounded above b}^ the disc resting upon
the septal edges. At this level mesenteries d and e are larger than in the
former section and are without any peripheral degeneration; exoseptum e-f is,
united by its inner edge with eutoseptum d-e, but the bifurcated exoseptum
a-d is free. Within the loculus, at the angle of bifurcation of the exoseptum,
the rudiments of the pair of mesenteries, b, c, have appeared, but at this stage
there is no septal formation whatever within their entocoele. Already, there-
fore, it is manifest that the new pair of mesenteries does not inclose a pre-
viously formed exoseptum ; the two members lie close together within the
peripheral bifurcation of an exoseptum, a-d.

Plate 9, fig. 56, is taken from a section at a somewhat higher level.
The second-cycle mesentery d is now united with the depressed disc, but
the other moiety {e) of the pair is still free ; in none is there any peripheral
resorption. Exoseptum ^-^ is once more distinct centrally from entoseptum

I02 SIDERASTREA RADIANS.

d-e^ as in fig. 54 ; a sj^napticular growth from the left limb of septum a-d
perforates the mesentery b^ and a slight depression of the wall between the
mesenteries b and c is the first indication of an entocoelic skeletal ingrowth
about to separate the two.

The section from which fig. 57 was taken reveals important altera-
tions taking place. The second-cycle mesenteries d and e are both united
with the discal walls, and along withy are at this level not perforated by
any S3mapticular bars. The two peripheral limbs of septum a-d are now
becoming free from one another at their inner angle, and also from the central
radial portion ; both may be regarded as distinct exosepta a-b and c-d^ though
formerly they appeared only as the bifurcations of the single exoseptum a-d.
For the first time an actual entoseptum now occurs in the entocoele of the
new pair of mesenteries b and <:, and at its free extremity is on the point of
uniting with the synapticular process from the exoseptum a-b. The septum
b-c is clearl}- a new formation, appearing within the now widely bifurcated
peripheral extremity of the original septum «-</, or, better, between the two
exosepta «-<?, c-d.

The septal modifications suggested by fig. 57 are completed in the next
stage, fig. 58. The exoseptum c-d is now distinct, like septum a-^, from the
central radial part, while exoseptum a-b and entoseptum b-c are united with
one another towards their free edge by a synapticulum perforating the
mesentery b.

The section represented by fig. 59 passes through the exsert edges
of the septa, so that the column wall forms the lower boundary while
the upper boundary passes through the tentacular region of the disc, four
tentacular thickenings being represented. The mesenteries extend uninter-
ruptedly from one wall to the other, and include the free septal edges within
their chambers. The mesentery c is still free at its inner end, while the
mesentery b is perforated by the synapticulum still joining septa a-b and b-c.
The central portion of the original exoseptum «-a?has now nearly disappeared.

The next stage, fig. 60, is from a section near the uppermost extremity
of the retracted polyp and its calice, so that now each septum and the body
wall surrounding it are becoming distinct. The mesenteries b and c both
extend across the two portions of the column wall, and the septa a-b^ b-c, and
c-d are distinct from one another, the entoseptum b-c being as yet smaller
than the exosepta on each side of it.

Omitting the synapticula as merely incidental structures the develop-
ment of the three septa a-b, b-c, and c-d ca.\\ be diagrammatically represented
as in fig. 12 {a-f) below, which shows also their relations to the mesenteries.

POSTLARVAL DEVELOPMENT.

103

The first two figures (a, b) correspond with plate 9, figs. 54-56, of the
sections. The exoseptum is here simple, but bifurcated towards its periph-
eral extremity. At first no mesenteries are included within the angle of
bifurcation, but a pair has appeared in fig. 12, b. In relation to the mesen-
teries the two limbs are clearly exocoelic, and may even at this stage be
regarded as two individual exosepta.

In fig. 12, <;, one exoseptum lias become distinct, and a short septum has
appeared midway between the two older septa {cf. plate 9, fig. 57). Clearly
this is a new entoseptum arising a little later than the new pair of mesen-
teries. In fig. 12, d, the right exoseptum has also separated from the more
central portion which is along the same radius as the entoseptum icf. plate
9, fig. 58). The next figure, e, shows the exsert septa [cf. plate 9, fig. 59);
the central part is becoming smaller, while it has disappeared in the last

k £

Fig. la. — Series of diagrammatic figures illustrating figs 54-60 on plate 9.

figure, / [cf. plate 9, fig. 60). The entoseptum throughout extends the
shortest vertical distance of the three, and so far as the development has pro-
ceeded remains the smallest radially. Clearlj^ when growth is completed
the three septa will be related as shown in the six groups of fused septa in
fig. 10, /, p. 96, which is the usual relationship of the exosepta and entoseptum ;
the entoseptum in its further growth will have extended more centrally and
fused with the central portion of the original exoseptum, and the two exosepta
will turn inwardly to unite with it.

The series of sections plainly demonstrates that a new entoseptum
arises shortly after a new pair of mesenteries, not in advance of it, and the
two peripheral limbs of a bifurcated exoseptum become distinct, so that each
constitutes a new exoseptum. The new entoseptum takes the place of the
inner simple part of an exoseptum situated in the same radius.

The members of the third-cycle entosepta of the older polyp are new
formations which take the place of the earlier third-cycle exosepta. The
dorsal and ventral moieties of the bifurcated periphery of the third-cycle exo-

I04 SIDERASTREA RADIANS.

septa become fourth-cycle exosepta; in other words, the fourth-cycle exosepta
in the mature polj^p are the peripheral continuations of the third-cycle exo-
septa, which in their turn have been shown to be the peripheral bifurcations
of the primary exosepta. Exosepta thus remain exosepta to the end, each
time constituting a later cycle as new entosepta arise to take their place.

The results from a study of a series of sections from a nearly mature
bud polj'p thus agree stage by stage with those obtained in the progressive
development of the three cycles of septa in pol3-ps reared from larva. The
figures given bear the closest comparison with the corresponding details in
fig. lo {d-/), p. 96, which represents three stages in the septal development
of a larval polyp.

Close examination with a lens of the surface of macerated coralla
often reveals one or more stages similar to the above. In practicall}' all
the corallites of a colony the alternation of large and small (entocoelic and
exoccelic) septa is strongly marked ; but amongst the youngest corallites,
especially those around the margin of colonies, the regularity of the alter-
nating large and small septa is not always so pronounced. Occasionally a
group of two or three septa is seen presenting quite different relationships,
which can be explained only upon the septal development here described.

An exoseptum is sometimes seen with its peripheral end conspicuously
bifurcated, as in fig. 12, a, p. 103 ; in several instances a stage in which a
septum is beginning to appear midway between the bifurcation has been
met with, recalling the conditions in fig. 12, c, d ; while very often in
a group of developing septa a marked interval occurs between the central
and the peripheral halves of an entoseptum, and the exosepta on each side
are wholly free. In these instances it would seem that the new entoseptum
has not yet fully united with the central part of the old exoseptum, being
in the same stage as fig. 12, </, e.

It still remains to be seen what is the order followed in the appearance of
the third-cycle entosepta, for these do not appear a cycle at a time any more
than the six members of the second cycle of entosepta. The septa alone in
the dried corallum are insufficient to enable their sequence to be made out, as
they afford no certain means by which the principal or directive axis can be
determined, and from this the dorsal and ventral borders of the calice. It
has been shown, however, that in the case of each of the three cycles of ento-
septa the mesenteries appear in pairs only a little in advance of the corre-
sponding entosepta within them ; therefore, if the sequence of the third-cycle
mesenteries be determined, it can be assumed that the third-cycle entosepta
follow the same order.

POSTLARVAL DEVELOPMENT.

105

The order of appearance of the twelve pairs of third-cycle mesenteries
has been fully described on p. 83. It is found that the members arise after
the formation of the six second-cycle pairs as unilateral pairs in a twofold
succession. Normally a pair appears within the dorsal exocoele of each
sextant, successively from the dorsal to the ventral border of the polyp ; then
another pair appears within the ventral exocoele of each sextant, following
the same succession as the first series.

Fig. 13. — Diagram illustrating ihe order of dovelupmmt of the tirit three cycles of septa (I-III).
The small letters accompanying the Roman numerals indicate the sequence of the indi-
vidual septa in the cycle.

The normal sequence of the third-cycle mesenteries in Siderastrea
being established, we are justified in assuming that a like succession will be
maintained by the third-cycle entosepta, seeing that they arise shortly after
the mesenteries with which they are associated. This is shown in fig. 13,
p. 105, by the numerals i\ia-\i\f. The figure also represents the normal
sequence for all the entosepta of the first three cycles. The six primary
septa appear together, a C3^cle at a time ; the six members of the second order
arise bilaterally in a simple dorso-ventral succession ; the twelve members of
the third also arise bilaterally in a dorso-ventral succession, but in two
series — first, a series of six within the dorsal of the two interspaces in each

Io6 SIDERASTREA RADIANS.

sextant, aud then the remaining six in a like order, but within the ventral
of the two interspaces. The exosepta constituting the last outermost cj'cle
have no corresponding ordinal significance.

Studies on the mesenterial sequence of other corals indicate that a similar
septal succession will in all probability be followed by forms in which the
adult calice shows a regular hexameral cyclic plan. Individual departures
from the order may be expected, but are to be looked upon as irregularities ;
regularity of growth of the higher cycles is by no means so pronounced as
in the first and second cycles, which are less likely to be influenced by
spatial considerations. The sequence given is altogether diiferent from any-
thing which has hitherto been surmised for any coral, and further studies are
desirable to determine how far it admits of general application in the group.

From what has been revealed it is manifest that the exosepta do not
possess any true ordinal sequence comparable with that of the entosepta.
Exosepta have been found to be present at each stage, always constituting
the outermost cycle, and equaling in number the sum of the inner entosepta.
We may consider them as the direct continuations of the six primary exo-
septa, or, less likel3% as arisi:ig anew with each cycle of entosepta. Regarded
as the persistent representatives of the primary exosepta, they more nearly
conform to the law of substitution of the exotentacles in actinians as estab-
lished bj' Lacaze-Duthiers and Faurot. In actinians generally it is found
that after the protocnemic stage the tentacles appear two at a time, one ento-
coelic and one exocoelic, corresponding with the two chambers formed upon
the appearance of a new pair of mesenteries ; sometimes the entotentacles
appear in advance of the exotentacles, the reverse of what happens in Side-
rastrea radians. The entotentacles are always larger than the exotentacles,
the length of the former being in accordance with the order of appearance of
the cycle to which the}' belong, the largest being the first to appear. The
exotentacles all attain an equal length and are all relegated to the outermost
cycle, whatever be the cycle of entotentacles with which they appeared. They
constitute a single cycle of which the members are always smaller than those
of the cycle of entotentacles last to appear, and the number of exotentacles
in the last cj-cle is always half the total number of tentacles, and, of course,
equal to the number of entotentacles.

As new entotentacles are added the exotentacles become pushed aside so
as to occupy different radii at different times. The calcareous septa being
hard, fixed structures, do not admit of such rearrangement ; the new septal
growth has to be adapted to the old, resulting in the fusion of the new ento-
septa with the old exosepta.

POSTLARVAL DEVELOPMENT. 107

Partial studies on other corals, as well as cousideratious on the tentacular
development in actinians, suggest that the exosepta may arise in diflferent ways
in different species, and that a more precise significance as to their relation-
ships at different stages maj' be forthcoming- than is possible in Sidenistrea.
There are indications that in some forms an entoseptum and an exoseptum
arise together, thus more closely recalling the method followed by the tenta-
cles. Regarded as arising anew in each cycle, the two exosepta in Siderastrea
appear somewhat in advance of the entoseptum which is included between them.

The relationships proved to exist between entosepta and exosepta involve
important considerations when the cyclic hexameral sequence is not completed
in the mature corallite, as almost invariably happens in 6". radians^ as well
as in many other species of corals. As regards both the septa and mesen-
teries it is found that the last cycle is rarely a multiple of 6, but some
irregular number from i to 12, resulting from the fact that at maturity the
polyp does not complete the last cycle begun. Exosepta have been showTi
to appear always in close association with entosepta, whatever be the number
making up a corallite ; and, as often remarked, the two series are equal in
number and the exosepta alwaj's outermost in position.

If we regard a septal cycle as made up only of entosepta, or of exosepta,
then in mature corallites of S. radians the third entocoelic cycle and fourth
exoccelic cycle of septa will vary in the same degree. Whatever number of
entosepta be lacking from the third cycle to form the complete cycle of 12,
a like number of exosepta will be wanting from the fourth C3'cle.

When describing the number of septal cycles within a calice, the cyclic
hexameral plan of which is incomplete, it is usual in systematic works on
corals to regard the hexameral multiples as completed as far as the number
of septa will permit, and then to relegate to the last cycle all the remaining
septa not included in the hexameral formula. The cycles are all supposed
to be hexamerously complete with the exception of the last. Thus, with
regard to S. radians, Milne-Edwards states : " Three cycles of septa com-
plete, and, in general, a variable number of a fourth cj'cle." Likewise Verrill
(1901, p. 153), describing the same species, says: " They [the septa] form
three complete cycles, with part of the fourth cycle developed, so that the
number is usually 36 to 40."

The relationships proved to exist between the entosepta and exosepta in-
dicate that the above formulae do not express the true morphological charac-
ter of the septa. Any hexameral incompletion in the number of septa mak-
ing up a corallite affects both the entosepta and the exosepta, that is, both
the penultimate and the last cycles. If any septa are wanting to complete

Io8 SIDERASTREA RADIANS.

the liexameral multiple of the last cycle of entosepta, the same number
will be lacking from the exosepta. The third complete cycle, as understood
by Milne-Edwards and Verrill, is really made up of both tertiary entosepta
and of tertiary exosepta. The two kinds of septa are of very different value
in their development and relations to the polyp, and, as a matter of fact,
will be scarcely of the same thickness and radial length to justify their
being regarded as a cycle.

The cyclic formula, as above understood, may be written 6, 6, 12, x,
where x will represent any number from i to 24. Formulated in this way
the number 12 conveys the impression that the third cycle is really com-
pleted, and that all the additions made will belong to the next or fourth
cycle, whereas the}' will belong to both the third and fourth cycles. Beyoud
the two first cycles the septa do not arise a cycle at a time, but the penulti-
mate and last cycles are formed concurrentl}', or almost so. Incomplete
cyclic hexamerism, as met with in S. radians, is an intermediate condition
in the establishment of two adult hexameral cycles, not of one alone, and
attention should be drawn to this in the septal formula.

According to the relationships now established the morphological septal
formula for S. radians should be written 6, 6, x, 6 + 6 + x. In this
formula, 6, 6, x will represent the number of septa in the two completed
entocycles, x being the number in the last entoseptal or penultimate cycle
which does not yet complete the hexameral sequence ; while 6 + 6 -f x will
represent the total number of exosepta, x being the same number as before ;
some of the exosepta will be tertiaries and some will be quaternaries, the
nvimber of the latter being always double the number of tertiary entosepta.
The formula for a corallite having 36 septa would, according to the ordinary
cyclic formula, be written 6, 6, 12, 12, whereas, considered as entosepta and
exosepta, the formula will be 6, 6, 6, 18, the three first numerals indicating
the entosepta and the last the exosepta ; the usual cyclic formula of a corallite
with 40 septa would be 6, 6, 12, 16, and the morphological formula 6, 6, 8, 20.
In the first case 12 of the exosepta will be quaternaries and 6 will be ter-
tiaries ; in the second 16 will be quaternaries and 4 tertiaries.

Where the relationships of the septa to the mesenteries are clearly known
the morphological formula will more nearly express the real value of the septa
than the ordinary cyclic formula ; the latter has little significance unless the
hexameral sequence is fully completed. One can not say that a cycle is
really complete unless its constituents all have the same morphological value,
which is not the case where some are entosepta and some are exosepta.

The bilaterality of the polyp during development may be looked upon
as associated in turn with each cycle individually. Any cycle tends to attain

POSTLARVAL DEVELOPMENT. 109

its radial plan before the next cycle commences to form, when the additions
take place in such a manner as to again confer bilaterality npon the polyp
as a whole. Thus the first two cycles of septa become perfectly radial
before an additional cycle commences, and the growth of this is then con-
tinued in a bilateral manner ; likewise, the new second and third cycles
assume their radial stage before the members of the fourth cycle make their
appearance, proceeding from one border to the other. In like manner the
first-cj'cle mesenteries are nearly radial before those of the second cycle arise
and introduce a conspicuous bilateral symmetry ; and on these assuming the
radial plan the third-cycle mesenteries begin to appear, again in a bilateral
manner.

The successive dorso-ventral growth followed by the constituent mesen-
teries and septa of each cycle may also be regarded as conferring a certain
individualit}' upon the cycle. The diflferent cycles, arising independently, seem
to represent so many distinct recurring phases of growth in the life of the
polyp, not a continuous addition from one aspect to the other, as is usual in
permanently bilateral animals, particularly segmented forms. The members
of a cycle appear in a dorso-ventral sequence, and may retain their diflferences
in size for a long time, but in the end they become equal and thereby confer
radial symmetry upon the polyp. Then another cycle commences to form
in somewhat the same bilateral dorso-ventral succession, displays for a time
its consecutive origin, and afterwards attains radiality.

The conception of recurring phases of growth in cyclic coral polyps is
best realized when comparison is made with the mesenterial increase char-
acteristic of the Cerianthese. Here the mesenteries beyond the protocnemes
always develop in a regular bilateral successive manner, from the dorsal
(anterior, sulcar) to the ventral (posterior, asulcar) aspect, the oldest being
dorsal or anterior and the youngest ventral or posterior, recalling more the
method of growth of segmented animals ; in cerianthids there is never a
reversal of growth to the anterior end, followed by a successive series to the
other, such as occurs in ordinar}' hexactinians. Employing the term "band
of proliferation," introduced by Van Beneden in 1897, we may say there is
only one median band of proliferation in cerianthids, while in hexactinians
there are many such bands, the number increasing with age — at first 6, then
12, 24, etc.

In the Zoantheas also mesenterial growth is always in the same succes-
sion after the protocnemic stage. The increase takes place within the two
exoccelic chambers on each side of the ventral directives ; there are only two
bands of proliferation or zones of growth. In this case, however, the order

no SIDERASTREA RADIANS.

followed by the new mesenteries differs from that in hexactinians and
cerianthids; it proceeds from the ventral (posterior, sulcar) to the dorsal
(anterior, asulcar) aspect of the polyp, not from the dorsal to the ventral.

The bilateral development of the organs, from one border of the polyp
to the other, in ordinary actinians and corals, would seem to have no phylo-
genetic significance beyond the group of the coelenterates, and as yet we
appear to have no definite understanding as to what even this may be. The
approximate radial symmetry of adult ccelenterates is assumed from very
diverse developmental conditions {cf. hexactinians, zoanthids, cerianthids,
and the tentacles and other cyclic organs in the Hydromedusaa and Scypho-
medusse). Whatever may be said in favor of Sedgwick's well-known view
that the mesenterial arrangement found in cerianthids suggests the metamer-
ism of higher animals, there is clearly no support for such a conception in
the development of the organs in hexactinians.

SUMMARY.

1. In Siderastrea radians the six members of the first cycle of septa
appear simultaneously, shortly after fixation of the larva, situated within
the entocceles of the first cycle of mesenteries.

2. Six members of a second cycle are developed within the primary
exocceles, shortly after the primary cycle of entosepta. They are the tempo-
rary predecessors of a later permanent cycle, and arise either simultane-
ously or in bilateral pairs in a dorso-ventral order. Later, each becomes
bifurcated peripherall}^, either by the direct extension of the original septum
or by the production of separate fragments which subsequently fuse. The
bifurcations also appear in a bilateral dorso-ventral order.

3. The six members of the permanent second cycle of entosepta arise
within the entocoeles of the second-cycle mesenteries soon after these make
their appearance. The two right and left dorsal septa appear first, then the
two middle members, and, at a much later period, the two ventral, the series
thus exhibiting a decided dorso-ventrality. In the end they become equal,
and each fuses with the central part of the corresponding second-cycle exo-
septum previously developed, which now lose their individuality.

4. Twelve members of a temporary third cycle are situated within the
exocoeles between the primary and secondary pairs of mesenteries, and repre-
sent the bifurcated extensions of the six primary exosepta. The original
second-cycle exosepta thus become the third exocoelic cj^cle, their place
having been taken by the permanent second cycle of entosepta.

5. The later development of the septa in buds proves that a new third

POSTLARVAL DEVELOPMENT. Ill

cycle of 12 or less septa arises on the appearance of the pairs of third-cycle
mesenteries, in a similar manner to that followed by the permanent second
cycle. New entosepta appear within the entocoeles of the third-cycle mesen-
teries, and the bifurcations of the third-cycle exosepta become the exosepta
of the fourth C5'cle.

6. The third-cycle entosepta, following the mesenteries, are developed
in a bilateral dorso-ventral order, but in two series — first a series within the
dorsal moiety of each sextant, and then a second series within the ventral
part of each sextant.

7. Exosepta are present at each cyclic stage in the growth of the coral-
Inm, alternating in position and corresponding in number with the sum of
the entosepta. They never become entosepta, but always constitute the
outermost cycle of shorter septa ; only the entosepta have an}^ ordinal sig-
nificance. Until the adult condition is reached the exosepta are the tempo-
rary predecessors of the entosepta. The developmental relationships between
the entosepta and exosepta -are closely comparable with those between the
entotentacles and exotentacles. The law of substitution, first discovered by
Lacaze-Duthiers for the tentacles of Hexactinise, is thus found to hold also
for the septa.

8. Where the cyclic hexamerism of a corallite is incomplete the ordinary
cyclic formula does not express the true relationship of the septa. The ento-
septa and exosepta vary in the same degree, so that the morphological
septal formula for a corallite with three entoseptal cycles and one exoseptal
cycle is 6, 6, X, 6 -f 6 + x, where x may be any number from i to 12.

9. The cycles of septa and mesenteries represent so many distinct recur-
ring phases of growth all around the polyp, not a continuous increase from
one extremity to the other as in metameric animals. With the exception of
the first the members of each cycle follow a dorso-ventral succession, display
a bilateral symmetry for some time, and ultimately assume an approximate
radial plan. The succession for the third cycle of entosepta is twofold.

BASAL PLATE.

G. von Koch, in the course of his embryological studies of corals, found
a deposit of calcareous matter to take place between the ectoderm of the base
and the surface of attachment of the polyps. It is the first part of the skele-
ton to be formed by the activity of the calicoblasts, and from its position is
known in coral literature as the basal plate. It is present in the photographic
reproductions on plates 4 and 5, and its relationship to the polyp, at a rather
late stage, is shown on plate 9, fig. 53.

112 SIDERASTREA RADIANS.

In Astroides calycularis Von Koch (18S2) found the basal plate to take
the form of a thin circular disc, composed of spheroidal or elliptical crystal-
line bodies, from 0.005 to 0.008 mm. in transverse section. At first the
calcareous elements were arranged in an interrupted manner, but later the
openings were filled by further deposit of skeletal matter, and the plate as a
whole became thickened. In Caryophyllia cyathns (1897) the first rudiments
of the basal plate consisted of a small central deposit surrounded by six thin,
nearly triangular plates, interseptal in position. For a time these were dis-
tinct from one another, but later united with the central circular patch, and,
by further additions, became joined along their edges, thus constituting a
complete flattened disc or plate. The presence of a basal plate has since
been recognized in many forms of corals, and it is extremely doubtful
whether it is really wanting in any species.

For a long time the developing polyps of S. radians gave no external
indication of any skeletal formation which could be regarded as the
basal plate. The first evidences of the corallum were the six entocoelic
septa, which appeared to rise directly from the surface of attachment.
Under transmitted light they stood out as nearly opaque objects, while the
interseptal spaces at this and later stages were quite clear and apparently
devoid of any calcareous crystals. For nearly two months the laboratory
notes contained the assertion that no basal plate was developed in Siderastrea^
for, excepting the presence of the septa and epitheca, there was no interrup-
tion in the ordinary light passing through the polyp. When, however, the
living or preserved polyps were examined by means of polarized light, the
bright colors of the basal region revealed the presence of crystalline matter.
This is well seen in the photograph reproduced on plate 4, fig. 22, taken
with an exposure of twenty minutes to polarized light. Between crossed
nicols the field was black except in the region of the polyp ; the septa also
appeared black owing to their thickness and the irregular disposition of their
crystalline constituents, while all the interseptal areas were brightly and
variously colored.

It is obvious, therefore, that a basal plate was already developed between
the polyp and its surface of attachment, but was too thin to offer any appre-
ciable obstruction to the transmission of ordinary light. The plate must have
begun to form a day or two after fixation of the larva, for the septal upgrowths
were observed on the third and fourth days, and the deposit of basal skeletal
matter probably preceded them. As stated below, the plate does not undergo
much thickness, even in later stages ; hence, at no time would it offer much
interference to the passage of light.

POSTLARVAL DEVELOPMENT. II3

When any of the jj'oung polyps, two or three weeks old, were macerated
later, it was found that a thin continuous plate remained adherent to the
incrusted surface ; but none of the earliest stages in its formation were
obtained, showing whether it originated as a continuous layer or in sepa-
rate parts. All the polypal tissues have been macerated from the young
coralla represented on plate 4, figs. 19 and 20 ; they now show a well-developed
basal plate, slightl}^ upturned at the periphery, and bearing the septa on the
upper surface.

A portion of the basal plate of a macerated corallum of two weeks, magni-
fied about 300 times, is represented in surface view on plate 11, fig. 69. At the
upper boundary of the figure the epitheca also is shown in section, and upon
the plate are the first thickenings which will form the septa. The entire
surface of the plate exhibits a number of very thin, flat scales, roughly polyg-
onal in form. Sometimes they appear as if joined edge to edge like the
cells of an endothelium, or at other places as if overlapping. The average
diameter of the individual scales is about 0.03 mm. In most of them a distinct
fibrous structure can be recognized, the fibro-crystals either l3dng parallel
or, less often, presenting a radiating appearance. There is, however, no
suggestion of the fibers being arranged around a center of calcification, as in
the trabeculae of the septa.

According to Miss Ogilvie (1897, pp. 114-117) similar scale-like elements
are present on the surface of the corallum of corals generally, and she has
succeeded in isolating them from the almost transparent dissepiments of
Galaxea. Their breadth in this genus varies on an average from r 01 to
0.015 mm., while their height is about 0.003 ™™- -^ number of fibers are
present in each, sometimes in the form of divergent groups, but often lying
loosely side by side. It was from a study of these that Miss Ogilvie came
to the conclusion that each isolated skeletal element represented a calcified
calicoblast cell, thus returning to the old view of von Heider as to the origin
of the calcareous skeleton of corals in contrast with the more accurate results
of von Koch, Bourne, and Fowler, which show that the calcareous matter is
secreted wholly external to the ectodermal cells. The structure of the calico-
blast la5'er in the polyps of Siderastrea is also very conclusive as to the
ectoplastic origin of the calcareous fibers. Ever3'where it has been found to
be a simple laj^er, never many cells deep, as would be the case were the
calicoblasts themselves calcified and shed from time to time to build up the
skeleton.

From the relationships of the basal plate to the polyp, increase in its thick-
ness can obviously take place only on the upper surface. In addition to the

114 SIDERASTREA RADIANS.

septal rudiments, small projections are present here and there on this surface,
especially towards the middle of the plate, and some continue growing until
they become raised much above the general level of the plate and constitute
columellar spines. In the basal plate we have the skeleton in its earliest
and simplest condition as a flat deposit, and the septal and columellar forma-
tions are vertical upgrowths from it. The latter represent areas at which
the calcareous matter is laid down with greater rapidity by the activity of the
calicoblasts. The basal plate, however, is covered bj^ polypal tissues on only
its upper surface, while the upgrowths from it — septa and columella — are
covered on both sides ; hence the difference in their microscopic structure
shown on plate ii, fig. 70. There are no axial centers of calcification in
the elements of the basal plate, while such are very distinct in the septa.

When the young corallum is detached from a smooth surface, the lower
surface of the basal plate is also smooth and even, but when growing over a
rough surface it adapts itself to the irregularities.

The basal plate increased considerably in diameter from the time of its
first formation, keeping pace with the general growth of the corallum. In
the earl 3'^ stage on plate 4, fig. 19, it is only 1.3 mm. across, while in the
corallum of plate 5, fig. 28, it is 2 mm. in diameter.

As shown in the next section an epitheca begins to form at the edge of
the basal plate whenever growth ceases, while when lateral growth is con-
tinued there is little or no possibility of any upturned marginal deposit.

Apparently the original basal plate of 6". radiatts never becomes much
thickened. Some of the oldest coralla reared rather suggest that the central
interseptal portions may be resorbed, or, at any rate, remain extremely deli-
cate. When coralla of four months were macerated and separated from their
surface of attachment it was found that interseptally the basal deposit had
disappeared from the central regions, though retained towards the periphery
and along the septa. It may be that on account of its thinness interseptally
it had broken away in the process of maceration, while it was supported
septally and peripherally.

In colonial corals the basal plate is generally represented only in the
corallum of the primary larval polyp, the later bud polyps not admitting of
its formation. But wherever in growing colonies of S. radians the marginal
corallites extend beyond the incrusted object a thin parchment-like deposit
is found basally. It constitutes the basal skeletal support of the young bud
polyps, in the same manner as the basal plate of larval polyps, or as the dis-
sepiments at a later stage.

POSTLARVAL DEVELOPMENT. 1 15

EPITHECA.

Along with tlie first formation of the radiating septal upgrowths
appeared a narrow, peripheral calcareous ring, somewhat less opaque than
the septa, and for a long time wholly unconnected with them (plate 2).
The most careful examination of the living polyps proved that the annulus
was altogether external to the soft tissues ; while in preserved and cleared
specimens the polypal wall was found to pass within the inner border of the
skeletal deposit, not to be folded over it. The formation is undoubtedly to
be regarded as an epitheca, according to the definitions of this structure
given below.

The epitheca increased in height along with the growth of the polyp,
at the same time often narrowing a little transversely. It remained
throughout uncovered by any pol3^pal tissues on its outer surface, though
lined by the polypal wall on its inner surface (plate 9, fig. 53). Where
most fully developed its outer surface exhibited distinct incomplete annula-
tions, wrinklings, or accretion lines, as if representing separate intervals in
the deposition of calcareous matter (plate 4). By the time the polyps were
six or seven weeks old the lower region became discolored by the adherence
of foreign matter, such as filamentous algse and diatoms, and only the actual
margin was fresh and white. Opposite the septa the epithecal margin was
sometimes a little indented, but otherwise was of the same height all round,
somewhat exceeding that of the septa. Whenever, owing to uncongenial
conditions, the polj^ps shrunk from their former size, the peripheral deposit
remained behind, distinct and complete ; the growth of a second annulus
then took place at the new margin of the narrowed polyp, in such a way that
the new ring was wholly within the old (plate 5, figs. 25-27).

Later, as the septa increased in radial length, some of them came in
contact with or were actually fused with the epithecal deposit, but the micro-
scopic structure of the two remained distinct. When the coralla were freed
from their surface of attachment the epitheca was seen to be a direct upward
continuation of the edge of the basal plate, and, like it, shows no centers of
calcification. It is made up of circular lamellae, the fibro-crystals of which
are arranged at right angles to the polypal wall (plate 11, figs. 69, 70) ; the
epithecal deposit, in fact, corresponds to but one-half of a septum.

The epitheca was found to vary greatly in the extent of its development
in the various polyps. In some of the most forwardly developed specimens
it was practically absent or represented only by a thickened marginal ring
(plate 4, figs. 19, 20) ; in others it appeared as a very distinct parapet,

Il6 SIDERASTREA RADIANS.

elevated beyond the level of the septa (plate 5). Where two or more polyps
developed contiguous to each other a common epitheca was formed along the
line of contact (plate 5, fig. 30), but each became distinct later. In one
instance two polj'ps, along with their septa, were embraced in a common
epitheca.

This surprising variation in the amount of epithecal deposit appears to
be determined by the rate of growth of the polyp. As already noticed, the
polyps varied greatly in this respect, and the epitheca was best developed in
individuals which increased but little in size. The transverse narrowing in
the later stages may be taken to indicate that the polyps had become some-
what less in their basal diameter. Where a polyp was growing rapidly,
enlarging the diameter of its basal disc and extending the septa peripherally,
it is manifest that an epitheca could not be formed, or, if formed, would need
to be resorbed. In the two coralla represented on plate 5, figs. 28, 29, there
is practically no epitheca, but the basal plate is much thicker in its peripheral
half than in the middle. Presumably, the polyp did not rest long enough at
any one stage to permit of the secretion of an epitheca. Until maturity,
therefore, it is possible that individual corallites of the same species of
coral may be provided with or be destitute of an epitheca, according to the
slow or rapid rate of growth of the polyp ; the morphological value of the
structure becomes somewhat lessened when its formation is shown to be
dependent to such an extent upon physiological conditions.

The peripheral skeletal formation in Siderastrea is of interest in con-
nection with the much discussed question as to the nature of the thecal and
epithecal wall of corals. In some respects the structure recalls that described
as theca (Mauer) by von Koch (1897) in his paper on the development of the
skeleton of CaryophyUia cyathus, but a comparison at once establishes their
different values. In both species the structure in question arises as an inde-
pendent peripheral part of the corallum, and narrows from below upwards.
But in the species investigated by von Koch the annulus is inclosed on both
its inner and outer sides by an upgrowth of the basal wall, while in Sider-
astrea it is external from the beginning, only covered by the polypal tissues
on its inner wall and growing margin (plate 9, fig. 53). Again, in the first
the deposit early unites the edges of the septa, but in the other it remains
entirely free from septal connection for a long time, and then only joins their
edges as an independent external covering. The different character of the
theca (pseudotheca and true theca) in various Madreporaria has already been
alluded to (p. 45), and the original peripheral structure arising in CaryophyUia
would certainly belong to the "true theca" type. Miss Ogilvie (1897, p.

POSTLARVAL DEVELOPMENT. II7

159) separates the theca from the epitheca, as follows : " The essential differ-
ence which may be said to distinguish " theca " from " epitheca" is that the
theca, or wall, must be structurally associated with the peripheral ends of
septa, whereas the epitheca is in no structural connection with the septa, but
is a continuous concentric deposit exterior to all the other skeletal structures
of a calyx." And again (p. 248) : " The epitheca is an external basal struc-
ture, laid down at the angle of the aboral wall, where it bends towards the oral
or peristomal region of the polj'p (figs. 22, 36). It is the continuation upwards
or outwards of the embr3^onic ' basal plate,' and may be well-developed or
remain rudimentary." G. von Koch (1896, p. 254) describes the structure
in much the same terms.

These two definitions serve to distinguish clearly between the periph-
eral anuulus found in Caryophyllia cyathiis and that in S. radians. In
the one case it is structurally associated with the outer ends of the septa,
while in the other it is quite independent of these ; in one it is developed
within a special invagination, " thecal refoulement," of the skeletotrophic
wall which lines both its inner and outer surfaces, in the other it is wholly
external to the polyp, covered on one side only by the skeletogenic tissues,
not by an upfolding. In Caryophyllia von Koch found no trace of an
epitheca in addition to the theca, while in Astroides and now in the early
corallum of Siderastrea there is found to be an epitheca but no true theca.
Whatever calicinal wall is found in the mature corallum of these two genera
is a later structure formed by the coalescence of the outer edges of the septa
(pseudotheca).

By Ogilvie (p. 248) and Vaughan (p, 48) the epitheca is regarded as a
primitive structure in Madreporaria, while the thecal structures are of
secondar}^ origin. The latter writes : " The oldest type is where the ends
of the septa did not fuse distally, but simply had their outer ends bound
together by an epithecal covering." This clearly describes the early epithe-
cate corallum of Siderastrea. In this respect, therefore, the genus must be
considered as representing an older type than the truly thecate corallum of
such a form as Caryophyllia. Its very rudimentary condition in some of
the young polyps and its almost complete absence from the adult colony
would seem to prove that in this genus it is an embryonic structure of
diminishing importance.

COLUMELLA.

For a time the central part of the basal plate was free from any calcare-
ous deposit which could be regarded as a columella (plate 4), but, as the septa
increased in size, spinous upgrowths began to form near their inner extremity,

Il8 SIDKRASTREA RADIANS.

extending as far as the center of the calice (plate 5). Some of these upgrowths
appeared as separate projections from the basal plate, as if produced by special
invaginations of the basal wall of the polyp — the refoulement columellaire
of Delage & Herouard (1901, p. 558) ; others seemed to be direct continua-
tions of the septa, not requiring a separate upfolding of the basal disc. The
former undoubtedly represent a true independent columella, while the latter
by their union might give rise to a so-called pseudocolumella. At no time,
however, could any sharp distinction be drawn between the septal and the
independent upgrowths, except as regards their position. In the coralla rep-
resented on plate 4, figs. 23 and 24, two or three of the middle granules are
obviously distinct basal formations, while exactly similar spinous projections
occur at the end of some of the septa. The middle of the corallum on plate 5,
fig. 28, is also occupied by distinct spinous upgrowths, both basal and septal in
character.

A further stage in the columellar growth is represented in the coralla of
figs. 28 and 29. Here the intervals between the spines are becoming partly
occupied by the deposition of secondary calcareous matter, so that their act-
ual origin is obscured. In the mature corallum it was frequently found
(p- 53) that the columella is minutely spinous or papillose as seen from the
surface, and that in sections for some distance below it remains spongy,
becoming compact in the deeper regions by the later deposition of calcareous
matter. Further, where the coralla are very strongly calcified the spinous
character of the columella disappears even superficially, the interspaces
being altogether occupied by the secondary deposit, which keeps pace with
the septal and columellar spinous formations. These differences in the
mature calice thus coincide closely with those represented in the larval cor-
alla of figs. 23 and 29.

Both from its origin and mature characters the columella of S. radians
is therefore formed from three independent sources: (i) Separate basal
upgrowths; (2) septal spines, continuous with the central ends of the septa ;
(3) a secondary deposit filling up the interstices between i and 2. Histolog-
ically the trabeculse of all three are alike (plate 10, fig. 65).

The columella of Siderastrea is thus a " true " columella, to be distin-
guished from a " false " or " pseudocolumella" where there is no direct basal
upgrowth, but the entire structure is formed from the inner septal edges. In
its development it agrees most closely with the account which von Koch
(1897, p. 769) gives of the formation of the columella in Caryophyllia,

POSTLARVAL DEVELOPMENT. II9

ANATOMY AND HISTOLOGY OF LARVA AND YOUNG POLYP.

A number of free-swimming larvae were preserved in corrosive acetic,
and later studied b}' means of transverse and longitudinal sections, when all
were found to be at about the same stage of development. Only the impor-
tant features in which the}' differ from mature polyps will be here noticed.

The larval ectoderm is somewhat broader than the same layer in the
adult polyp. In sections it measures about 0.04 mm. across while that of the
adult is 0.03 mm. The outer surface is stronglj' and uniformly ciliated, the
enlai'ged base of each cilium being well defined. Numerous clear and gran-
ular gland cells occur, and towards the margin a zone of small nematocysts
0.015 mm. in length. The layer further diflfers from that of the adult in
containing a few Zooxanthellas, mainly restricted to the oral extremity.
These have been already noticed among the external characters as giving a
brownish color to the oral pole of the larva, and are apparently liberated
from time to time.

The larval ectoderm is broader at the base, where it measures 0.06 mm.
across, and, in addition, has undergone certain histological modifications.
The cells as a whole seem more compactly arranged, gland cells are less
numerous, and a nerve layer is present ; but nematocysts do not seem more
numerous than elsewhere. The whole structure recalls the aboral sense
organ which has already been found by McMurrich, Appellof (1900), and
myself (1902, p. 524), to occur in certain actinian and coral larvae, and is
evidently widely, though not universally, present in these two groups. The
characteristics of the organ, however, are usually more conspicuous than in
the present species, especially as regards the degree of development of the
nervous elements.

Thestomodaeal communication between the exterior and the larval cavity
was already established in all the larvse studied. In most the lumen is circu-
lar, while in others it is slightly oval. The surface is more strongly ciliated
than that of the outer ectoderm, and fewer gland cells occur; the two agree
in the presence of Zooxanthellae, though these are never found in the ecto-
derm of the adult stomodaeum.

The mesenterial filaments on pairs i and 11 on plate S, fig. 52, stand
out conspicuously from the endoderm on account of the number and deeply-
staining character of their nuclei, and their histological structure recalls that
of the stomodseal ectoderm. In the larvse, however, the organs are not yet
rounded oflF from the mesenterial endoderm.

I20 SIDERASTREA RADIANS.

In the serial transverse sections from which figs. 51 and 52, on plate 8,
were taken, the stomodseal ectoderm appears to be continued uninterruptedly
as the mesenterial filaments down the edge of the two lateral pairs of com-
plete mesenteries on their becoming free; but the ventral and dorsal direct-
ives (hi, iv), which are only attached to the stomodasal wall for part of their
length, are without such modified tissue along their free edge.

The continuity of the stomodasal ectoderm with the mesenterial fila-
ments in the various larvae of Siderastrea, and the close histological simi-
larity of the two, seem at first sight undoubted evidence of the ectodermal
origin of the latter, especially when it is found that filaments are not present
on mesenteries which do not reach the stomodseum, or do not extend as far
as its inner termination. In the larvas of many other Actiniaria and Madre-
poraria, however, it has been found that filaments may appear on mesen-
teries before they reach the stomodaeum, and even in complete mesenteries
an interval of undiflferentiated endoderm often occurs between the ter-
mination of the stomodaeal ectoderm and the filaments. From observa-
tions on other corals I consider that the mesenterial filaments arise from the
larval endoderm independently of the stomodasal ectoderm, but that continuity
is early established in the case of those mesenteries which unite with the
stomodaeum (1902, p. 476).

The internal cavity of the larvas is very limited in extent, the endoderm
nearly filling the whole chamber. In the youngest examples mere slits
represent the lines along which the polypal cavity will be formed later. The
cells of the endoderm are much vacuolated, and contain numerous Zooxan-
thellae scattered throughout. The lining of the mesenteries and also of the
intervening portion of the column wall is not arranged as a simple epithelial
layer such as characterizes the adult. Within most of the intermesenterial
spaces the endoderm is greatly thickened, and in transverse sections stands
out as a very distinct triangular projection into the coelomic cavity, leaving
only a narrow slit between itself and the mesenterial lining. The vertical
projections are the " Vorsepten " or prosepta of von Koch (1897) ; they are
the persistent parts of the mass of endoderm which at an earlier stage occu-
pies the whole interior of the early larva. On plate 8, fig. 52, taken from a
rather late larva, the prosepta are still conspicuous, and some are associated
with rudiments of the mesenteries ; the mesenterial endoderm is still greatly
thickened, and the central cavity is beginning to enlarge.

In an earlier paper on the larva of the actinian Lebrunia coralligens
(1899), I have shown that most anthozoan larvae are for some time nearly
solid, owing to the enormous development of the endoderm. There are, how-

POSTLARVAL DEVELOPMENT. 121

ever, very narrow, caual-like slits, and from these the adult gastro-coelomic
cavity is derived by the disintegration of the more central endoderm and the
shrinkage of that lining the bod3^-wall and mesenteries.

Some of the larvae of Sidetastrea sectionized reveal a gastro-coelomic
cavity further developed than that represented in figs. 51 and 52. Below the
stomodseal region the parenchymatous endoderm has broken down, and the
middle of the cavity is occupied by organic debris, in which Zooxanthellae,
cell walls, and granules of various kinds are recognizable. There is no
doubt that this is the organic debris which is extruded from time to time by
the larvae soon after their liberation from the parent polyp (p. 58). At the
stage in the young polyp at which the septa are well advanced (plate 9, fig.
53), the endoderm has become a simple epithelial layer throughout, resem-
bling in all respects that of the mature polyps.

In the earliest larvae sectionized eight mesenteries were already developed,
arranged in four bilateral pairs, as shown diagrammatically in fig. 8, a, p. 80.
The two lateral pairs are united with the stomodaeum, while the dorsal and ven-
tral axial pairs, representing the directives, are free, and of the two directive
pairs the ventral are slightly larger than the dorsal. In larvas a day or so
older the ventral directives have united with the stomodaeum, while the dorsal
are still free (plate 8, fig. 5 1) ; also at the stage with three complete mesenteries
two other bilateral pairs of mesenteries, the fifth and sixth in the sequence, have
made their appearance, arranged as on plate 8, fig. 52. Afterwards the dor-
sal directive mesenteries unite with the stomodaeum, and the larva has reached
the Edwardsian stage of mesenterial development presented at the time of
fixation (fig. 8, a-c, p. 80).

There appears to be no resting stage in the appearance of the mesen-
teries between the tetrameral and hexameral condition, such as seems to char-
acterize certain actinians [Lebnima), nor in the successive union of the first
four pairs with the stomodaeum. As already shown, however, the fifth and
sixth pairs remain as microcnemes for a prolonged period.

As represented on plate 8, figs. 51 and 52, the mesenterial mesoglcea is
extremely thin. The cut ends of very delicate muscular fibrils can be recog-
nized in transverse sections, and, according to their disposition on one side or
other of the mesogloea, assist in the determination of the paired arrangement.

The further development of the mesenteries, after fixation has taken
place, can be easily followed through the transparent tissues of the living
polyp, and has been already described.

122 SIDERASTREA RADIANS.

YOUNG POLYPS.

Very few of the young polyps reared were available for anatomical and
histological study, owing to the greater importance of the corallum at this
stage ; in most cases the soft tissues were removed by maceration in order to
secure the skeleton. A vertical section through a decalcified retracted larval
polyp, of about two months, is represented on plate 9, fig. 53. The section
of the skeleton has also been added, the details being taken from various
coralla of this age. The section of the pol3'ps includes the oral aperture, and
on the right is trul}' radial, passing through a mesenterial space and cutting
a septal invagination obliquely, while on the left side it passes obliquely
through a mesentery and also through a septal invagination. A third invag-
ination is included in the middle region of the section and probably represents
an early columellar upgrowth. The upper wall on the left has come to rest
upon one of the invaginations.

The septal and columellar invaginations are here of the simplest charac-
ter. They are merely continuous upgrowths of the basal wall of the polyp,
and agree with it histologically. They line both sides of the calcareous
upgrowths from the basal plate, while the latter in its turn is laid down by
the flattened part of the disc, increase in thickness taking place only on
the upper side.

Sections of coral pol3^ps at such an early stage are the most favorable for
demonstrating conclusively that the madreporarian skeleton is laid down
wholly outside the polj^pal tissues, and that all the skeletal complications are
formed within extensions of the basal disc, the invaginations proceeding /ar/
passu with the deposition of calcareous matter. This truth was first estab-
lished by von Koch by means of sections of polyps of Asfroides, somewhat
similar to that of plate 9, fig. 53 ; further, von Koch was able to demonstrate
the calcareous spheroids of the basal plate and septa in situ. His polyps
for this purpose were adherent to pieces of cork, so that the polyp and its
attachment could be sectionized together. The polyps of Siderastrea being
adherent to glass could be sectionized only after being freed through decal-
cification.

Decalcification of the polyp from which fig. 53 was taken was carried out
with great care, but only fragments of a very narrow skeletogenic ectoderm
remained. No hints of any mesogloeal processes or desmocytes occurred,
but such would scarcely be expected considering their rarity in the adult
polyp oi Siderastrea. The mesogloea itself is a very thin lamella. The basal

POSTLARVAL DEVELOPMENT. 1 23

endoderm also has undergone a great alteration compared with the same
la5-er in the other regions of the polyp. It is much thinner, all traces of cell
limitations are lost, and Zooxanthellse are wholly absent, while elsewhere in
the endoderm the algal cells occur in some abundance.

The passage from the calicoblastic ectoderm to the ectoderm of the
column wall is gradual, and can be studied on both sides of the section. It
is at this point that the epitheca is laid down. Were there any doubt as to
the epithecal or thecal nature of the peripheral calcareous riug such sections
prove absolutely that the deposit is uncovered by the polypal tissues on its
outer sides. Were it otherwise an invagination of the base would occur
towards the periphery, but none of the sections show any such folding.

At the region of the epitheca there is no sharp distinction between the
column wall and the basal disc. The ectoderm of the former is a broad
columnar epithelium, having many clear unicellular mucous glands distrib-
uted among the supporting cells ; the nuclei are restricted mostly to the inner
half of the layer. The left half of the disc includes the section of a tentacle,
represented only by the knob which is crowded with long, narrow nematocysts.
As in adult retracted polyps, the stem of the tentacle is not distinct from
the disc. Histologically the walls of the stomodaeal invagination closely
resemble those of the adult pol3'ps, and the ectoderm is marked off from the
rest of the body wall by its numerous ciliated supporting cells.

On both sides of the section represented in fig. 53 the stomodasal wall
terminates freely, but in sections which include a mesentery the ectoderm is
seen to be continuous with the mesenterial filaments along the free edge of
the mesentery. The two, stomodaeal ectoderm and mesenterial filament, are
much alike histologically, and by their brightly-staining character are easily
recognizable among the other tissues.

REFERENCES.

(Only the more frequently recurring references are here given ; the others are inserted in the text.)

1900. Appellof, A. : " Stiidien iiber Actinien-Entwicklung." Bergens Museums Aarbog. 1900, no. I.
1897. VAN Beneden, E. : " Les Anthozoaires de la Plankton-Expedition." R^sultats de la " Plankton-
expedition der Humboldt-Stiftung." Vol. 11, Kiel et Leipsic.

1887. Bourne, G. C. : "The anatomy of the Madreporarian Coral Fungia." Qiiart. Journ. Micr. Sci.,

vol. XXVII.
1893. : " On the postembrjonic development of Fungia." Ti'ans. Roy. Dublin Soc, ser. ii,

vol. V.
1899. : "Studies on the structure and formation of the calcareous skeleton of the Antho-

zoa." Quart. Journ. Micr. Sci., vol. XLi.

1901. Dklage, Y., and Heroltard. E. : "Traill de Zoologie Concrfete. Les Coelenterds." Tom. 11, pt. 2.

1899. Duerdkn, J. E : " The Edwardsiastage of the actinian Lcbrunia, and the formation of the gastro-

cceloinic cavity." Journ. Linn. Soc, Zool., vol. xxvii.

1902. ■ : "West Indian Madreporarian Polyps." Mem. Nat. Acad. Sciences, vol. viii, yth Mem.

1S95. Faurot, L. : " iltudes sur I'anatomie, I'histologie et le developpement des Actinies." Arch, de

Zool. Exp. et Gdn., ser. 3, torn. iii.
1S8S. Fowler, G. H. : "The anatomy of the Madreporaria, IV'." Quart. Journ. Micr. Sci., vol. x.\vni.
1902. Gardiner, J. S. : " South African corals of the genus Flabelliim, with an account of their anatomy

and development." Marine Investigations in South Africa, vol. 11, Cape Town.
1882. VON Koch, G. : " Ueber die Entwicklung des Kalkskeletes von Asteroides calyciilaris und dessen

morphologischer Bedeutung." Mitt. a. d. Zool. Stat, zu Neapel, bd. in.

1896. : " Das Skelett der Steinkorallen." Festschrift fUr Carl Gegenbaur. Leipzig.

1897. — — : "Entwicklung von Caryopkyllia cyathiis." Mitt. a. d. Zool. Stat, zu Neapel, bd. xil.

1872. Lacaze-Duthiers, H. de: "Developpement des Coralliaires. Prem. M^m., Actiniaires sans

Polypier." Arch, de Zool. Exp. et Gen., tom. i.

1873. : "Developpement des Coralliaires. Deux. Mdm., Actiniaires a Polypier." Arch, de

Zool. Exp. et G^n., tom. 11.

1897. : " Faune du Golfe du Lion. Coralliaires. Zoanthaires Sclerodermas." Arch, de Zool.

Exp, et Gen., 3 ser., tom. v.
1S97. Ogilvie, M. : " Microscopic and systematic study of Madreporarian types of corals." Phil. Trans.,

vol. CLXXXVII.

1900. Vaughan, T. W. : " The Eocene and Lower Oligocene Coral Faunas of the United States." U. S.

Geol. Survey, Monogr. xxxix.
igoi. Vkrrill, a. E. : "Variations and nomenclature of Bermudian, West Indian and Brazilian Reef
corals, with notes on various Indo-Pacific corals." Trans. Conn. Acad. Science, vol. xi.

1888. Wilson, H. V. : "On the development oi Mavkina areolata." Journ. Morph., vol. 11 (1889).

EXPLANATION OF PLATES.

Plate i.

Fig. I. — Larva immediately on extrusion, viewed by reflected light. The uniform covering of cilia is
indicated, and also the distinction between the outer ectoderm and the solid internal endoderm. The
darker color of the broader oral pole is due to the presence of Zooxanthells in the ectoderm. The
narrower aboral pole is anterior in swimming; the mouth is not yet functional.

Fig. 2. — Abnormal larva with two oral poles and one aboral.

Fig. 3.— Lana shortly after extrusion. The larva is now more swollen and transparent, so that four
pairs of mesenterial lines are seen; the mouth is also functional. Two pairs of mesenteries reach the
stomodieum, and two other pairs are free.

Fig. 4. — A second-day larva just before settling. Six pairs of mesenteries are now present, three
pairs of which reach the stomodieum.

Fig. 5. — Three larvte settling close together by the narrow aboral pole.

Fig. 6.— a group of seven third-day larva; which have settled so close together that their walls partly
overlie. Four pairs of mesenteries now reach the stomodsEum, and on three of the individuals the
rudiments of the six exoccelic tentacles have appeared.

Fig. 7. — A living polyp two or three days after settling, viewed by transmitted light. The dark
outermost rim represents the epitheca; the next lighter zone is the flat margin of the polypal wall into
which the polypal cavity does not extend. The six entosepta are opaque, but in such a view there is no
evidence of the basal plate. The six tentacles are exoccelic and alternate with the septa. The internal
Zooxanthellas have accumulated mainly along the sides of the mesenteries; the ectodermal Zooxanthella;
have disappeared except immediately around the mouth. Diameter of original, 1.5 mm.

Plate 2.

Fig. S. — Living polyp with a dorso-lateral pair of exosepta and a rudimentary pair of median exosepta
in addition to the six entosepta.

Fig. 9 — Polyp with six entosepta and six exosepta, the latter still revealing their dorso-ventral order
of development by differences in maa;nitude.

Fig. 10. — A living polyp of about the same stage as fig. 9, fully expanded and viewed from the side
as a transparency. Of the mesenteries only the insertions are represented.

Fig. II. — Polyp with twelve fully expanded tentacles, six large outer exoccelic and six small inner
entoccelic. Fifth week.

Fig. 12. — Polyp showing the irregular manner in which peripheral additions are made to the primary
septa. The ventral directive tentacle is double. Diameter of original, 1.7 mm.

Plate 3.

Fig. 13.— Expanded disc showing the doubling of certain of the entotentacles and the asymmetrical
position of others.

Fig. 14 —Expanded' polyp viewed from the side so as to display the manner of appearance of the
second-cycle mesenteries on the column wall.

Fig. 15. — Expanded disc resting on the septa and exhibiting the relationship of the second cycle mesen-
teries to the new entosepta and exotentacles. An additional exotentacle has appeared within the middle
sextant on each side, and the dorso-lateral entotentacles have each a single stem bifurcated distally.

Fig. 16. — Expanded disc showing the manner of increase of the tentacles. An exotentacle protrudes
from each of the twoexocoeles in the dorso-lateral and median sextants, but as yet there is only one from the
ventro-lateral sextants. The tentacle from each dorso-lateral entococle is already bifurcated as in the adult.

Fig. 17.— The same polvp as in fig. 16 at a later stage. On each side a second-cycle tentacle (II) has
grown out over the entocoele of the median second-cycle mesenteries, and the four lateral first-cycle ento-
tentacles (I) are bifurcated. Diameter of original, 2 mm.

Fig. iS. — A larval polyp of three months in which the disc and tentacles are indrawn and the disc is
almost covered by the overfolding column wall. The epitheca is shown around the margin.

126

EXPLANATION OF PLATES. 1 2/

Plate 4.

Fig. 19. — Photograph of early corallutn with six entosepta and six exosepta. The basal plate (proto-
theca) is fully formeci, but somewhat irregular in outline ; it is a little thicker and upturned towards the
margin. The six entosepta are nearly equally developed, while the alternating exosepta vary much in
size; the dorso-Iateral pair are best developed and peripherally' seem double, the middle pair are a little
smaller, while the ventro-lateral are yet rudimentary. Diameter, 1.6 mm.

Fig. 20. — A somewhat later corallum. The directive entosepta are irregular in form and differ
from one another, the ventral appearing as if bifurcated peripherally; the two dorso-lateral and right
middle exosepta also appear bifwrcated; the ventro-lateral exosepta are still much smaller than the other
exosepta.

Fig. 21.— a young preserved polyp photographed by transmitted light. In such a view the basal
plate is not apparent owing to its transparency. The septa are non-transparent and stand out as black
objects. Indications of the six pairs of radiating mesenteries are seen and also some of the tentacles, but
the details around the mouth are obscured by the mesenterial filaments and stomodaium. The dotted
surface of the polyp is due to the presence of internal Zooxanthellce. The septa are a little further
developed than in fig. 20, the exosepta still showing a decided dorso-ventrality. There is a marked
difference in form between the dorsal and ventral directive septa. Diameter, 1.7 mm. (In the process of
engraving the principal axis of the polyp has been turned coiinteyclockiuise through an angle of about
20 degrees from the vertical.)

Fig. 22. — The same polyp as in fig. 21'photographed by an exposure of 20 minutes to polarized light
between crossed nicols. Only the calcareous crystals transmitted the variously colored light. The
mottled surface represents the basal plate which is not indicated in the previous figure. The crystalline
matter of the septa is so thick and irregularly arranged that no light was transmitted ; hence the structures
appear black.

Fig. 23. — A typical corallum at a later stage, growth having proceeded upon the septal plan of figs. 19
and 20. The polyp remained at this stage for some time, and a well developed wrinkled epitheca has
formed, its upper edge higher than the septa; the latter are fused peripherally with the epitheca, but there
is no theca. The directive septa are somewhat bifurcated peripherally and also the dorso-lateral and
median exoseptal pairs. The dorso-lateral pair are almost fused centrally with the dorsal directive, the
middle exosepta are fused with the dorso-lateral entosepta on each side, and the ventro-lateral exosepta
with the ventro-lateral entosepta. The ventro-lateral exosepta are the smallest of the whole series. Sepa-
rate columellar nodules are present in the middle. For diagrammatic plan see p. 8S. Diameter, 1,8
mm. (In the process of engraving the principal axis of the corallite has been turned clock-wise through
an angle of about 20 degrees from the vertical.)

Fig. 24. — Another corallum at about the same stage as fig. 23, but showing variation in detail.

Plate 5.

Fig. 25. — A corallum in which the epitheca has gradually receded from the margin towards the
middle of the calice as the polyp shrunk in size. The older part of the epithecal deposit covers the
peripheral parts of the septa. The newest portion of the epitheca is the inner annulus. The septa are
at about the same stage as in fig. 23.

Fig. 26. — Corallum in which the polyp has twice diminished in size, somewhat suddenly. Three
concentric epithecse have been formed, the peripheral ends of the septa being partly exposed between each.

Fig. 27. — The septa are a little further advanced than in figs. 23-26 In the left dorso-lateral sextant
there are three distinct septa, a large median and two lateral; the median septum consists of a larger
central (exoseptum) and a small peripheral part (entoseptum). In the right middle sextant are also three
septa, the median of which consists of a central and a small peripheral part [cf. fig. 15).

Fig. 28- — This and the next corallum are the two oldest reared. Each contains three complete cycles
of septa, and some of the sextants show important stages in the passage from the second to the permanent
third cycle. In the polyps which formed the coralla the first and second cycles of mesenteries were fully
developed. The diagrammatic representation of the stage is given on p. 89, and emphasizes the manner
effusion of the various septa with one another. The directive septa are stouter than any of the others.
The three septa within the ventro-lateral sextants are not so strongly developed as those within the median
and dorso-lateral sextants, and show very distinctly the twofold constitution of the median entoseptum

128 EXPLANATION OF PLATES.

and the independent character of the exoseptum on each side (c/. fig. 15). Synapticula now connect some
of the adjacent septa. The origin of part of the columella from nodular upgrowths, independent of the
inner septal edges, is very evident. The basal plate is thicker peripherally than centrallv, and the epitheca
is onlv feebly developed at the margin. Diameter of original, 2 mm.

Fig. 29.— The details presented are practically the same as in the previous figure, but the growth of
the three septa within the ventro-lateral sextants has reached the same stage as in the other sextants.
The epitheca is better developed than in fig. 28.

Fig. 30.— Double corallum produced by two polyps the larv* of which settled close together. Each
has an independent epithecal formation, the part common to boih bemg very irregular. The septa have
not preserved the regular hexameral plan.

Plate 6.

Fig. 31. — Three fully expanded adult polyps, showing the forms assumed by the polyps and the
arrangement of the bilobed and simple tentacles.

Fig. 32 — Nearly mature polyp preserved in formalin in a partly expanded condition; surface view.
The tentacles over the third-cycle entosepta 'III) are still simple and resemble the exotentacles (X).

Fig. 33.— An isolated interseptal lamella from a decalcified polyp. The lower margin represents the
part of the lamella which rested upon a dissepiment, and about the middle nearly surrounded a synap-
ticulum. The apertures (.y«) are the spaces formerly occupied by the synapticula. The column wall,
disc, a bifurcated tentacle, stomodjeum, and a mesenterial filament are shown.

Fig. 34.— Transverse section of a mature polyp, passing through the stomodsal region. The figure
shows the arrangement of the mesenteries, the septal invaginations, and synapticula. In this and other
figures the spaces occupied by the skeleton are represented by the dotted areas.

Fig. 35. — Transverse section through a portion of the calicinal ridge of two retracted polyps, showing
the continuity between the mesenterial chambers of adjacent polyps, and also the vertical free edge of the
mesenteries as they extend from the column wall to the skeletotrophic tissues. In the upper part the
septa are exsert, but below they are united with the calicinal wall {ex. sepi., e.xoseptum ; ert/. sept.,
entoseptum).

Fig. 36. — Tangential section towards the periphery of a polyp. The column wall rests upon the septal
edges, and the mesenteries have a short vertical extent.

Plate 7.

Fig. 37. — (7, fully expanded entotentacle ; b, fully expanded exotentacle.

Fig. 38.— Transverse section of the same polyp as in fig. 34, taken a little below the stomodaeal region.

Fig. 39. — Transverse section of a polyp towards its aboral region. The mesenterial loculi are alto-
gether isolated at this level, and appear further broken up by the synapticula stretching from one septum
to another. The mesenteries have almost disappeared and the middle of the section is occupied by the
columella. The skeletal ectoderm and endoderm are both represented.

Fig. 40.— Tangential section through the tentacular region. A single bilobed entocoelic tentacle is
almost resting upon a septal edge. In the septal invagination {sep. inv.) the calicoblast layer is repre-
sented.

Fig. 41. — Transverse section through a mesentery at the level represented in fig. 35. The meso-
glceal plaitings for the support of the retractor muscles are simple in character, the oblique musculture
is feebly seen, and also the desmoidal processes where the mesentery is attached to the skeletotrophic
layer {sk. I.).

Fig. 42.— Transverse section through a mesenterial filament immediately below the stomodieal region.

Fig. 43.— Transverse section of a mesentery some distance below the stomodreal region, showing the
form of the mesenterial filament, and also a single ovum; the peripheral end of the mesentery is under-
going resorption.

Fig. 44. — a, small nematocysts from the knob of the tentacle and column wall ; *, elongated form of
nematocysis from the knob of the tentacle and mesenterial filaments; c, large oval nematocyst from the
lower part of the mesenterial filaments.

EXPLANATION OF PLATES. 1 29

Plate 8.

Fig. 45. — Section through part of a mesentery in union with the skeletotrophic tissues where they
surround a synapticulum. The mesenterial mesogloea for some distance from its termination is finely
striated, forming a desmoidal process. The skeletotrophic ectoderm on the left side shows faintly a
narrow portion of the homogeneous skeletal matrix.

F:g. 46. — Transverse section through an isolated portion of the lining wall of an interseptal loculus.
Each lateral extremity was terminated by a synapticuhnn. The section was stained in iron h:i-matoxylin,
and shows very distinctly the mucous spaces in the inner endoderm and calicoblast layer. The latter also
contains coarsely granular cells and nematocysts.

Fig. 47. — Transverse section through the skeletotrophic layers in their uppermost part, showing the
very narrow endoderm and calicoblast layer. The layers are syncytial in character. Irregular canal-like
spaces are filled with a deeply-staining substance. Small irregular particles are adherent to the free
surface of the ectoderm.

Fig. 4S. — Transverse section through a part of the skeletotrophic l.iyers from the lower region. The
■section gives practically the same details as fig. 46, but is more highly m.ignified.

Fig. 49. — Longitudinal section through an expanded tentacle. The stem is simple in character, while
the swollen knob has a nerve and a muscle layer at the base and nematocysts and gland cells {i^r. g. c.)
towards the periphery.

Fig. 50. — Transverse section through a portion of a mesentery showing the system of irregular
spaces in the endoderm. The ZooxanthelUe and cut ends of the retractor muscle fibers arranged along
mesoglceal plaitings are also represented.

Fig. 51. — Transverse section through the stomodseal region of a larva preserved shortly after extru.
sion. The six primary pairs of mesenteries are present, but the dorsal directives {iv, iv) and pairs v and
VI are yet free from the stomodteum. The gastro-coelomic cavity is beginning to be established in the
nearly solid endoderm ; Zooxanthellaj are present in the thick stomodioal ectoderm. In this and the
next figure the histological details of the endoderm are represented, while the ectoderm is convention-
ally depicted.

Fig. 53. — Transverse section through the same larva a little below the stomodie.il region. Mesen-
terial filaments are forming at the free end of the first and second pairs of mesenteries. Endodermal
thickenings (Prosepten) are present between liie larger pairs of mesenteries.

Plate 9.

Fig. 53— Radial vertical section through a young polyp, showing its relationship to the early coral-
lum. In the latter are shown the basal plate {b. //.), epitheca (ep.'), two septa, and acolumellar upgrowth
(col.)

Figs. 54-60. — Serial transverse sections through a portion of a polyp, showing the relationships of
the mesenteries and septa in the development of a third-cycle entoseptum (4-c) and two fourth-cycle
exosepta (a-b, c-d). The full description of the series is given on p. 100.

Plate 10.

Fig. 61. — A free corallum. Natural size.

Fig. 62. — A portion of the same corallum magnified about four times, showing various characteristics
of the corallites and their relationships to one another.

Fig. 63. — Upper part of three septa as seen in a vertical surfac; view. The left and middle septa extend
across a calice, but are separated in the middle by the columella and the broken central edges of other
septa in union with it. The right septum belongs to an adjacent calice, and is separated from the middle
septum by a vertical ridge which represents the line of fusion of an adjacent septum not in the same radius.
The two or three vertical rows of large elevations on each septum represent the broken surfaces of synap-
ticula ; the smaller elevations are granules. The inner central margin of the left septum shows the denti-
culations as they unite with the columellar tangle, and near them the fractured surfaces indicating where
an adjacent septum was interruptedly united by its inner margin.

Fig. 64. — Ground surface of a portion of a corallum showing the relationship of a single calice to the
six adjacent calices. Surface view. The long axis of the columella represents the principal axis of the
calice, the directive septa being at opposite extremities, and not quite in the same plane.

Fig. 65. — Portion of a thin transverse section of a single corallite showing the microscopic structure.
The columella, septa, and synapticula seem as if constituted of spheroids fused together; the spheroids
represent transverse sections through so many individual trabecule.

130 EXPLANATION OF PLATES.

Plate ii.

Fig. 66. — Tran.sverse section tlirousjh a single trabeciila, penetrated by filamentous boring algaa, and
showing the growth lamellre and their fibrous character.

Fig. 67. — Vertical section through a part of two adjacent septa. The septa are made up of trabeculae
arranged in a radiating manner ; they diverge from a continuous median trabecula which represents the
boundary between the two adjacent calices. The different appearances presented by the various trabec-
ule are dependent upon the part included in the section. Where the section passes along the middle of
a trabecula the interrupted dark centers of calcification are seen ; elsewhere only the diverging bundles of
fibrocrystals appear. The growth lamellss are not shown (c/^ fig. 68). New trabeculie are intercalated
at intervals. The dark circular and oval patches are sections of synapticula.

Fig. 68. — Terminal portion of two trabeculae as seen in longitudinal radial section, that is, parallel to
the surface of the septum, more highly magnified than in the previous figure. The section passes along
the middle of the trabeculse so that all the dark centers of calcification are seen. In the corallite from
■which the section was taken the growth lamelloe were very distinct, the boundary of each being indicated
by a dark granular deposit similar to that at the centers of calcification. The growth lamelliB are arranged
parallel with the toothed edge, that is, with the calicoblast layer which secretes them, while the fibro-
crystals making up the lamella are arranged at right angles. The figure should be compared with fig.
66, which shows a trabecula in transverse section.

Fig 69. — Portion of basal plate and epithecal boundary of a very early corallum. The epitheca is seen
in section, and the basal plate is sufficiently thin to allow of the passage of light. The basal plate is made
up of more or less distinct granules or scales showing a fibrous structure; the aggregation of darker
granules represents the first formation of a septum.

Fig. 70. — Portion of basal plate, septa, and epitheca of a somewhat older eorallum than that of fig. 69.
The sepia and epitheca are ground down to nearly the level of the basal plate Centers of calcification
are not present in the basal plate and epithec^i but are very prominent in ihe septa.

3I0ERASTREA RADIANS

PLATE 1

SIDERASTREA RADIANS

PLATE 2

/ J"

■/

10

11

SIDERASTREA RADIANS

PLATE 3

SIDERASTREA RADIANS

CLATE 4

3IDERASTREA RADIAN =

DLATE 5

SiDERASTREA RADIANS

PLATE 6

I X

Ilia I

^ ^" iR

\^^

■mest

rol.w.

"\

sj/n..

0

J

\J

/oni .

\\ mes.fil.

33

SIDERASTREA RADIANS

PLATE 7

SIDERASTREA RADIANS

PLATE 8

i \

'tiTlTi'- .- ■"'•■'•''

^

fc

•A

. I'^f

/•'

,^^®'

'^t!^

.0

-^■i:=^

"iV'^v_,

*37S

@J

50

300 x.

SIDERASTREA RADIANS

PLATE 9

ntest

3IDERASTREA RADIANS

= LATE 10

SIDERASTREA RADIANS

PLATE 11

Mill WHO] I.IRHAKV

li)H IflDR c

:) D

riWv«?iK!ii5M

m:<