‘4
i
é
i
a
a
i
|

) / ]

HARVARD UNIVERSITY.

I se Ake Y

OF THE

MUSEUM OF COMPARATIVE ZOOLOGY.
9524
Q py
Ge, akl
U

| 7 4]
Np i Opi Ay, //
Sr fuk Af | Over Ke a S99

i

QUARTERLY JOURNAL

OF

MICROSCOPICAL SCIENCE.

EDITED BY

E. RAY LANKESTER, M.A., LL.D., F.R.S., .

NONORARY FELLOW OF EXETER COLLEGE, OXFORD} CORRESPONDENT OF THE INSTITUTE OF FRANCE,
AND OF THE IMPERIAL ACADEMY OF SCIENCES OF ST. PETERSBURG, AND OF THE ACADEMY
OF SCIENCES OF PHILADELPHIA; FOREIGN MEMBER OF THE ROYAL BOHEMIAN
SOCIETY OF SCIENCES, AND OF TH ACADEMY OF THE LINCEI OF ROME;
ASSOCIATE OF THE ROYAL ACADEMY OF BELGIUM; HONORARY MEMBER
OF THE NEW YORK ACADEMY OF SCIENCES, AND OF THE
CAMBRIDGE PHILOSOPHICAL SOCIETY, AND OF THE ROYAL
PHYSICAL SOCIETY OF EDINBURGH ; HONORARY
MEMBER OF THE BIOLOGICAL SOCIETY
OF PARIS;

DIRECTOR OF THE NATURAL HISTORY DEPARTMENTS OF THE BRITISH MUSEUM) FULLERIAN PROFESSOR

OF PHYSIOLOGY IN THE ROYAL INSTITUTION OF GREAT BRITAIN.

WITH THE CO-OPERATION OF

ADAM SEDGWICK, M.A., F.RS.,

FELLOW AND TUTOR OF TRINITY COLLEGE, CAMBRIDGE 3

AND

W. F. R. WELDON, M.A., F.R.S.,

LINACKE PROFESSOR OF COMPARATIVE ANATOMY AND FELLOW OF MERTON COLLEGE, OXFORD,
LATE FELLOW OF ST. JOHN’S COLLEGE, CAMBRIDGE.

VOLUME 42. —New Senrtgs.
With Aithographic Plates and Engrabings on dood

LONDON:
J. & A. CHURCHILL, 7, GREAT MARLBOROUGH STREET.
1899,

CON TEN Ts.

CONTENTS OF No. 165, N.S., FEBRUARY, 1899.

MEMOIRS:

Outlines of the Development of the Tuatara, Sphenodon
(Hatteria) punctatus. By Arruur Denny, D.Sc., Professor
of Biology in the Canterbury College, University of New
Zealand. (With Plates 1—10) .

Abstract and Review of the Memoir by G. Hieronymus “On
Chlamydomyxa labyrinthuloides, Archer.” By J. W.
JENKINSON, M.A., Exeter College, Oxford

CONTENTS OF No. 166, N.S., MAY, 1899.
MEMOIRS:

On the Development of the Parietal Eye and Adjacent Organs in
Sphenodon (Hatteria). By ArtHUR Dewpy, B.Sc., F.L.S., Pro-
fessor of Biology in the Canterbury College, University of New
Zealand. (With Plates 11—13)

The Molluses of the Great African Lakes.—III. Tanganyikia
rufofilosa, and the Genus Spekia. By J. E. S. Moors.
(With Plates 14—19) : :

The Molluses of the Great African Lakes. —IV. Nassopsis and
Bythoceras. By J. E.S. Moorz. (With Plates 20 and 21)

Further Study of Cytological Changes produced in Drosera.
By Lity H. Hurts, ss a ae Oxford. (With
Plate 22). : :

Remarks on some Recent. Work on the pele anced: with a
Condensed Account of some Fresh Observations on the Entero-
pneusta. By Arruur Wi.xEy, M.A., D.Sc.

PAGE

89

111

155

187

203

223

iv CONTENTS.

CONTENTS OF No. 167, N.S., AUGUST, 1899.

MEMOIRS: PAGE
The Structure of Xenia Hicksoni, nov. sp., with some Observa-
tions on Heteroxenia Elizabethe, Kolliker. By J. H.

AsHwortH, D.Sc. (Lond.), Demonstrator in Zoology, Owens
College, Manchester. (With Plates 23—27) ; ; . 245

Notes on the Batrachians of the Paraguayan Chaco, with Observa-
tions upon their Breeding Habits and Development, especially
with regard to Phyllomedusa hypochondrialis, Cope. Also
a Description of a New Genus. By J. S. Bupeert, Trinity
College, Cambridge. (With Plates 28—82) 3 : . 305

The Development of Echinoids. Part 1—The Larve of Echinus
miliaris and Echinus esculentus. By E. W. MacBrips,
M.A., Professor of Zoology in McGill University, Montreal.
(With Plate 33) . : : . : ; . 335

Hydroids from Wood’s Holl, Mass.: Hypolytus peregrinus, a
New Unattached Marine Hydroid; Corynitis Agassizii and
its Medusa. By L. Murpacu. (With Plate 34) ; . 341

Note.—Arhynchus hemignathi. By Arthur E. Shipley . 361

CONTENTS OF No. 168, N.S., SEPTEMBER, 1899.

MEMOIRS :

The Structure and Metamorphosis of the Larva of Spongilla
lacustris. By Ricuarp Evans, B.A., Jesus College, Oxford.
(With Plates 35—41) . : : i : . 363

On the Communication between the Ccelom and the Vascular
System in the Leech, Hirudo medicinalis. By Epwin S.
Goopricu, B.A., Aldrichian Demonstrator of Comparative
Anatomy, Oxford. (With Plates 42—44) : . » a)

Balanoglossus otagoensis, n. sp. By W. Buaxtanp BENHAM,
D.Se., M.A., Professor of Biology in the University of Otago,

New Zealand. (With Plate 45). : : - . 497
The Movements of Copepoda. By E. W. MacBrips, Professor
of Zoology, McGill University, Montreal 5 : . 505

TirLe, INDEX, anb CONTENTS.

Outlines of the Development of the Tuatara,
Sphenodon (Hatteria) punctatus.

By

Arthur Dendy, D.Sc.,
Professor of Biology in the Canterbury College, University of New Zealand.

With Plates 1—10.

TABLE OF CONTENTS.

E
1. PREFACE : 5 : 1
2. InrRopUcTORY REMARKS : : 3 : d on
(a) On the Habits of the Tuatara : oo:
(6) On the Time occupied in ie yelopatent. mh List of

Embryos classified and arranged in Chronological Order . 9
(c) On the Structure of the Egg . : rae Ut
3. SysTEMATIC ACCOUNT OF THE STAGES IN aera - . 14
4. SUMMARY AND Discussion OF PRINCIPAL EMBRYOLOGICAL OBSERVA-
TIONS . - : - . 60
(a) The Formation ice the Genre Layers : : . 60
(6) The Formation of the Foetal Membranes : . 65
(c) The Modelling of the Body, and the Foundation of the Bone
pal Systems : : : é . 69
(d) The Embryonic Colour Mankines ‘ : ‘ meri
5. List oF REFERENCES : ‘ : : ‘ on
6. DESCRIPTION OF FIGURES : : : : 3 . 80

1. PREFACE.

Suortty after my arrival in New Zealand my friend Pro-
fessor G. B. Howes, LL.D., F.R.S., urged me to undertake
the investigation of the development of the Tuatara (Sphe-
nodon punctatus), but for various reasons I decided at that
time not todo so. Some time afterwards, however, while examin-

von, 42, PART 1.—NEW SER. A

2 ARTHUR DENDY.

ing the parietal eye in some embryo Australian skinks, I
was struck with the existence of an optic cup, apparently
similar to that of the paired eyes. This seemed so startling
that I resolved to attempt to obtain Tuatara embryos for the
purpose primarily of studying the development of the parietal
eye.

Knowing that Stephens Island, in Cook Straits, on
the recommendation of the Australasian Association for the
Advancement of Science, kad been proclaimed a reserve for the
Tuatara, it occurred to me by a happy inspiration to address
a letter to the lighthouse keeper, then entirely unknown to me
even by name, asking his assistance in the matter. I wrote
to him first on July 3rd, 1896, and most fortunately I found
in Mr. P. Henaghan, the principal keeper on Stephens Island,
an ardent enthusiast, who threw himself heart and soul into
the work in the interests of science; and I cannot sufficiently
express my gratitude to him for the magnificent supply of
material which he obtained for me, and also for the valuable
information which he gave me from time to time in his in-
teresting letters regarding the habits of the Tuatara.

Stephens Island being now a reserve, it was of course
necessary to obtain permission from the New Zealand Gevern-
ment to collect, and this, through the kind assistance of Sir
James Hector, F.R.S., was successfully accomplished. To the
Hon. the Colonial Secretary and to Sir James Hector I also
wish to express my gratitude for their courtesy.

I had proposed visiting Stephens Island myself during the
breeding season, which was supposed to begin about January,
but pressure of other business made it very difficult for me to
do so, and I ultimately resolved to trust entirely to Mr.
Henaghan for the supply of eggs. There is regular com-
munication between Stephens Island and the mainland only
once in six weeks, so that considerable difficulty and delay
were experieuced in sending the eggs, in consequence of
which a considerable number perished. Several lets of the
eggs were sent packed in moss or lichen in tin cans, but
they were very liable to go bad in this packing if delayed

OUTLINES OF THE DEVELOPMENT OF THE TUATARA, 3

on the voyage, and subsequently we found that much the best
way was to pack them, only a few together, in tin cans filled
with the coarse brown sand which occurs on the island. Thus
packed they travel admirably. I found it quite possible to
keep the eggs developing, buried in damp sand after their
arrival, but this was attended with considerable risk, as they
are very subject to the attacks of mould if kept too damp and
without sufficient ventilation, while they readily shrivel up if
allowed to get too dry.

The first consignment of eggs was received about the end of
January, 1897. Unfortunately the embryos all died, appa-
rently from drying up of the eggs, before they reached Christ-
church. Nevertheless one advanced embryo, the only one as
yet obtained of Stage Q, was removed from the egg in suffi-
ciently good condition to be of considerable value.

I obtained no more eggs that summer, my arrangements
being interfered with by a German collector, who visited
Stephens Island about the new year, having obtained a recom-
mendation to the keeper from the authorities. The only other
eggs found that summer were forwarded to him, but I am
informed that they perished in transit.

The next eggs which I received arrived, greatly to my sur-
prise, in July, 1897, a season of the year at which we did not at
all expect them. They were only three in number, and con-
tained very advanced embryos (Stage R) in excellent condition.

In November, 1897, the work began in earnest, and from
that time to the present I have received consiguments of eggs
at more or less frequent intervals. Latterly I have requested
Mr. Henaghan, when he finds a nest, to keep the eggs on the
island, to continue their development, and send them on to me
at intervals in small quantities, a plan which we find to work
admirably, and which I regret did not occur to me before, as
all the embryos thus obtained have been in much the same
advanced stage in development, and some more of the earlier
stages would have been more useful.

The result of our operations, however, has been to secure a
magnificent series of embryos, the majority of which have heen

A ARTHUR DENDY.

hardened in Kleinenberg’s picric acid after removal from the
egg, and preserved in alcohol. Those which were required for
sections were stained with borax carmine, and cut in paraffin
with Jung’s sliding microtome. I found it desirable to use oil
of cloves for clearing the embryos previous to embedding in
paraffin, as this allowed of more advantageous examination of
them as transparent objects than the turpentine method which
I usually employ.

It would, of course, be impossible in the time which I have
had at my disposal to give anything like a complete account of
the development of the Tuatara; and in this communication I
propose only to give a general outline, with special reference
to the formation of the germinal layers, the foetal membranes,
the modelling of the body, the foundation of the principal
bystems, and the classification of the embryos into stages ;
which will, I hope, be useful in future investigations either by
others or by myself.

This general account may be followed by more special
memoirs dealing with the development of the particular organs.
Already, indeed, I have sent home for publication one such
memoir “ On the Development of the Parietal Eye and adjacent
organs.”

In concluding this preface I must express my sincere
gratitude to Professor G. B. Howes for his kindly interest and
encouragement, and for revising the proof sheets, and super-
intending the execution of the plates in my absence from
England.

At my request he has undertaken to investigate the deve-
lopment of the skeleton, and for this purpose I am forwarding
to him a.supply of material.

2. INrRopucTORY REMARKS.
(2) On the Habits of the Tuatara.

Since I have not myself visited Stephens Island, it will be
as well to give the following account of the habits of the
Tuatara as nearly as possible in Mr. Henaghan’s own words.

OUTLINES OF THE DEVILOPMENT OF THE TUATARA. 5

“Our island,’ writes Mr. Henaghan,! “is densely covered
with scrub of various kinds; the soil is good in most places,
especially on the ridges; in the gullies the soil is of a light
brown colour, largely composed of oxide of iron. The birds
and lizards burrow into this soft soil, and one can often find
both living peaceably in the same hole. ‘There are three or
four kinds of petrel frequenting the island, and if you were
here now, which is their breeding season, you would be sur-
prised at the numbers of them,—there is hardly a foot of soil
but is undermined with their holes. Insects of various kinds
are also well represented, and I think the lizards feed largely
on them, especially beetles. I believe they also eat young
birds,—in fact, I have seen them do it.?, They, however, live a
large part of the year without any food, keeping constantly to
their holes. There are three or four kinds of lizards here alto-
gether, the Tuatara being the largest; the others are very
small. There have never been many lizards’ eggs got here
yet, though the Tuataras are very numerous.”

The breeding season apparently commences in November on
Stephens Island, and each female lays about ten (10) eggs.?
On November Ist, 1897, Mr. Henaghan removed ten eggs
from an individual which had probably been accidentally
crushed by a cow. These eggs, which were forwarded to me,
were apparently quite ready to lay, but in the one which I
opened on their arrival I found no embryo, and the others
unfortunately went bad subsequently, instead of continuing
their development, as I hoped they might.

The ground, Mr. Henaghan informs me, is so full of holes
that one has to dig ever so many before finding a Tuatara, and
then the probability is that it will turn out to be a male, these
being far more numerous than the females. The females are
much smaller than the males, but otherwise they do not appear
to be distinguishable externally.*

1 Letter dated October 28th, 1896.

* Compare Thomas (1) for a similar observation.

> Probably up to fifteen sometimes (cf. infra).

* According to Thomas (1) the male has the crests on the neck and back
far more strongly developed.

6 ARTHUR DENDY.

On December 11th, 1897, Mr. Henaghan wrote me:

“TI was out to-day searching for eggs, and I discovered two
nests. One of them was probably laid late in the last season,
as I noticed the embryos were fully developed. This I dis-
covered through one of the eggs getting broken, and the other
three I have sent you. You will observe that the advanced
ones are much larger than the new ones,—in fact, double the
size. |

“There were a lot more eggs in the same nest, but they
were shrivelled up and consequently no use. I think it takes
them several months to develop to maturity. All the eggs
we found this season, with the above exception, were in the
earlier stages of development. I notice a good many of the eggs
are shrivelled up, owing to the dry state of the ground. We
are having a long spell of drought just now, and I am afraid if
it continues much longer a lot of the eggs will perish. It is
only fair to give the lizard credit for a large amount of saga-
city in the way she selects places for depositing her eggs, for
it must be observed that there are plenty of enemies to
contend with ; as I have before mentioned, the birds are very
numerous, and continually scratching out holes. The lizard,
as a rule, shares the same hole with the bird, bat never lays
her eggs there, so that when searching for lizards’ eggs one
has to select a place with a sunny aspect and free from birds’
holes.”

In reply to questions as to the holes in which the eggs are
laid, I was informed as follows :—“ First, the holes are small
cavities made chiefly in the surface soil; the entrance is about
one and a half or two inches vertical height, and about three
inches horizontal width. ‘The chamber is continued at about
these dimensions for five or six inches into the ground on a
level surface, when it is then slightly increased to receive the
eggs. The eggs are packed close together in layers of two or

2

three, and owing to expansion in developing,’ it is with dif-

1 It is doubtful if the size of the egg has anything to do with the state of
development of the embryo. This question is discussed subsequently.
* See previous note.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 7

ficulty the eggs can be got out of some of them without injury.
The eggs, in the majority of cases, must be laid outside of the
chamber and carried in by the lizard, either by its mouth or
claws, and placed in position; not a vacant space is left in the
cavity where the eggs are deposited. The most of the nests
found were only a few inches underneath the surface. The
eggs are covered up with the soil removed from the chamber
in excavating, and well pressed in on the eggs, then the
entrance to the hole is stuffed with grass or leaves, and left to
appear as like the surrounding locality as possible.”

“* A few nests have been found at the extreme end of tunnels
extending into the earth for a distance of two or three feet.
The main tunnel is used by the lizard as a place of abode, and
is left open at the entrance. At the extreme end a small
chamber is scratched out at right angles from the main tunnel,
and in this are the eggs packed and covered up with soil.
The eggs in this case were probably laid in the main chamber
and then conveyed to the nest, as described in the first kind of
nests. The soil where the eggs were deposited is chiefly com-
posed of a mixture of clay and sand, but a good many of the
nests were found in surface soil, and, as before stated, only a
few inches in the ground. I may, however, say that no nests
were found in loose soil, and a very favourite place for them is
underneath a footpath. The ground being hard on top would
no doubt cause the rain to run off in the winter. From ten to
fifteen eggs is the average number found in any nest. In the
deep holes some of the eggs were scratched outside the en-
trance, and it was owing to this that these were found. If
you will picture to yourself the slope of a hill, you will easily
understand the above description of their operations. I very
much regret that your suggestion about retaining some of the
eggs in the nests for further development did not reach me
sooner; some of the nests, at all events, could be dealt with in
the manner you describe.”

On February 4th, 1898, I received another letter from my
indefatigable correspondent, containing most valuable informa-
tion. He tells me that they had been searching all the cleared

8 ARTHUR DENDY.

places on the island nearly every day since the first of January,
and were beginning to give up all hope of obtaining any more
egos for this season, when it occurred to him to try and find
out where the Tuataras laid their eggs before any clearings
were made for lighthouse purposes. ‘‘ Remember,” writes
Mr. Henaghan, “ that all the eggs found previously were got
on the clearings made for tracks, &c. In order to solve this
problem I went down a steep cliff about 400 feet, or in other
words about 200 feet above the sea. There is no vegetation
in this place, and to-day I started grubbing out some loose
earth; to my joyful surprise I found some eggs, good ones.
There was plenty of evidence that they used to lay here in
years gone by, for there were a lot of old egg-shells in the soil.
This place was so steep that it was with difficulty I could keep
my footing while working. Now I intend to keep these eggs
as you proposed ina previous letter; I have put them in a hole
in the ground close to my house, and in as favourable a situa-
tion for their development as there is in the island. I will be
able to examine them occasionally, and at the same time send
you a few every time there is a chance.”

I may conclude these quotations from my correspondent’s
interesting letters with the following important note :—‘‘ On
the 12th November one of my assistants excavated a track on
the side of a slope leading down to a sheep-pen. In making
this track he evidently cut into a lizard’s nest, but did not
notice it at the time. One day about the middle of January,
when we were carrying a sheep up this track to be slaughtered,
one of my children noticed an egg sticking out at the side of
the cutting. On examination it was seen that a nest had been
there. Some of the eggs, those that were farthest in, were
empty, showing that the young lizards had escaped, while the
egos that were exposed to the sun had the skeletons in them.
It is therefore quite plain that the eggs take about twelve (12)
months to develop, and it is with the object of proving this
that I send you the samples, hoping they may be useful.”

One of the most important facts brought out by the valuable
observations of Mr. Henaghan, above quoted is that the

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 9

Tuatara lays its eggs in special holes carefully concealed, and
not as a rule in its ordinary dwelling-holes. This fact doubt-
less accounts for the failure of other collectors to find the eggs.
As far as I know Mr, Henaghan is the only collector who has
yet obtained the eggs in the natural breeding-places.

(6) On the Time occupied in Development.

I quite agree with Mr. Henaghan that the eggs take at least
twelve months to develop,—indeed, I think they take rather
more, being laid in November of one year and hatched about
December of the year following. With a view to establishing
this belief I subjoin the following table, in considering which
it should be noted that those embryos which have not yet been
stained and cleared are only referred approximately to their
respective stages, while of course the different stages are con-
nected by intermediate ones, so that it is sometimes a mere
matter of choice to which stage a particular embryo should be
referred.

It must also be noted that, in the earlier stages of develop-
ment at any rate, the sequence of events is not always quite
the same. ‘Thus a feature which appears comparatively late
in one embryo may appear comparatively early in another;
and this inconstancy greatly increases the difficulties of classi-
fication.

The classification is, however, quite sufficiently accurate to
establish a most remarkable coincidence between the sequence
of dates and that of stages in development. The eggs were
numbered consecutively as they came in, and the numbers
missing from the following list belong to eggs which, for
one reason or another, yielded no embryos. Except where
otherwise stated, the date given is that on which the egg was

opened.

10 ARTHUR

DENDY.

List of Embryos obtained, classified in Stages and
arranged in Chronological Order.

DatTEs. STAGES. NUMBERS.
End of January, 1897. 5 8} nal
» duly, 1897 ; ~ 2, 3, 4
(Removed from female
on November Ist, 1897) (Ready for laying) . (28 to 37)
November 22nd, 1897 ER p@axs 2 . 5, 6,795.9
ex : 58
E 56, 64
F ee!
December 9th, 1897 : 4 G 59
LJ 40, 42, 43, 44, 46, 60, 62, 63
Ker. , 39, 41
ie 72
December 10th, 1897 : J 14 ?, 19; 74, 75
K : » ¥A5
H 78
December 16th, 1897 F ioe
L 47
December 27th, 1897 5 | u 50; 82, oF
(Died between Nov. 12th, 1897,
and middle of January, 1898). S 149, 150
January 3rd, 1898 M : Prediul
N 93, 94, 95, 96, 97
January 6th, 1898 {3 89, 90
S 138, 139, 140 ay ‘arrived dead)
January 10th, 1898 “ 510) ;
J aoe O ‘ 103
anuary 25th, 1898 5 ; {> 87.106 ?
March 8th, 1898 5 Pkt 141, 142
March 9th, 1898 : ety 143
March 12th, 1898 : adit 144, 145, 146
March 14th, 1898 ‘ dey 147, 148
April 5th, 1898 . : o 1k, =) loletioplins
May 12th, 1898 . : 5 IR 159, 160, 161
June 14th, 1898. c > oR 169, 170
June 24th, 1898 . : 7 lw . 162

I think that a careful examination of the foregoing table,
taken in conjunction with the observations of Mr. Henaghan
already quoted, establishes satisfactorily the following conclu-
sions :

(1) On Stephens Island the Tuataras lay their eggs in
November, and probably only at about this time.

(2) The eggs take about thirteen months to develop, during

OUTLINES OF THE DEVELOPMENYT OF THE TUATARA. 11]

which period it appears that a considerable number perish
naturally from drought.

(3) The earlier stages in development are passed through
much more rapidly than the later, and, Stage R having been
reached in March, the embryo makes little further progress
during the winter months, and does not hatch until about the
middle of the following summer (January).

There is only one other animal known to me, the eggs of
which take such a long time to develop, from the time of
laying to the time of hatching, and that is the oviparous
Victorian species of Peripatus, described by me under the
name of Peripatus oviparus (2).! As I have noticed else-
where, an egg of this animal which was laid in my vivarium at
Melbourne took no less than seventeen months to develop
before hatching. Sphenodon and Peripatus are both
doubtless extremely ancient types, which have persisted with
but little modification for a very long period, and one is strongly
tempted to connect this fact with the extraordinarily long period
occupied in their development from the egg.

(c) On the Structure of the Egg.

The eggs of the Tuatara are oval in shape, usually about
equally rounded at the two ends, but sometimes rather narrower
at one end than at the other. They vary very considerably in
dimensions,” and the size of the egg seems to bear no relation
to the stage of development of the contained embryo. Thus,
taking the measurements of eight eggs numbered consecutively
as they came in, and all containing embryos belonging to the
same stage (R), I found them to be as follows :

' Compare the observations of Mitsukuri and Ishikawa (5) on the similar
variation in size of the eggs of Trionyx.

2 Cf., however, a note in ‘Nature’ (vol. lvili, p. 619) by Mr. G. A.
Boulenger, F.R.S., on the embryo of Emp. orbicularis, contributed while
this paper was in the printers’ hands. Although most probably due to a
different cause, the “suspended gestation ” of the roe deer may be cited as a
kindred phenomenon (Bischoff, ‘ Entwick. des Rehes. Giessen,’ 1884, and
the ‘ Zoologist,’ 1889, p. 86).

12 ARTHUR DENDY.

Number 141, size 81 x 23°5 mm. Number 145, size 24 x 21 mm.
Ry eas. ay eo ene somes 5 - 146; 5, °28 <q2iebeee
5 143, 5, 27 xX 22:5 055 3° «| CO LATS 4, 88" X22 :beee
3 «(44 5, 33 2K 22 ; 148, 1, 30'b. 5¢ Zayas

Eggs containing embryos of other stages vary within about
the same limits, and it will be seen that their measurements
agree closely with those of Thomas (1), who gives the length as
varying from 2°d to 3°30 cm.

The colour of the eggs is dirty white, stained and mottled with
brown by the brown sand or earth in which they are deposited.

The shell is flexible, tough, and elastic, rather rough ex-
ternally, and contains in its outer portion a considerable amount
of caleareous matter. This calcareous deposit occasionally gives
rise to beautiful dendritic or stellate crystallisation patterns
on the surface of the egg. The inner surface of the shell is
smooth and white, and faintly opalescent. There is no distinct
shell membrane, but the shell itself can be readily separated
into layers which can be peeled off from one another.

In the recently laid egg the yellow yolk almost completely
fills the shell. ‘Thus the blastoderm when formed is separated
from the shell by only a very small amount of watery material.
Indeed, it is very difficult to open the shell without at the same
time piercing the blostoderm and allowing some of the yolk to
run out. I have, however, succeeded in puncturing the shell
so that only the clear liquid, evidently corresponding to the
“white” of a hen’s egg, escaped. At a much later stage
in the development there is a very large quantity of clear semi-
gelatinous material present, which escapes when the egg is
opened. This, though closely resembling the “ white ”’ of a hen’s
egg, does not at all correspond to it, but lies inside the allantois,
the outer layer of which, being pressed close against the shell, is
almost unavoidably punctured with the latter. No air-chamber
is present even at a late stage in the development. This fact was
specially determined by opening eggs of Stage R under water.

I have not been able to detect any vitelline membrane ; at
the earliest stages observed (c) it appears to have already dis-
appeared, or become fused with the blastoderm.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 13

The yolk, lying within the blastoderm, is of the usual yellow
colour. In the earlier stages of development (c) it appears to
be pretty uniformly soft and liquid, except where it is adherent
to the under surface of the blastoderm. It is full of large
transparent spheres and small granular spheres, floating freely.
The small granular spheres nearly all contain bodies which
may be termed “ crystalloids ” (fig. 107) ; otherwise they closely
resemble the yellow yolk-spheres of the hen’s egg. The crys-
talloids are clear, transparent, rod-shaped or spindle-shaped
bodies with truncated ends. The sphere appears to be at first
simply granular, and about 0°029 mm. in diameter (fig. 107, a) ;
then the crystalloid appears within it (fig. 107, 6), and gradually
increases in size, stretching the sphere as it grows, while the
granules of the latter gradually disappear until finally the
erystalloid is left without any trace of the original sphere
(fig. 107, c). Occasionally two crystalloids appear in the same
sphere (fig. 107, d); they may then give rise to irregular four-
sided bodies (fig. 107, e). I am inclined to think that ulti-
mately the crystalloids lose their form, and run together to con-
stitute the large transparent spheres which I regard as oil
globules, Already at Stage C the crystalloids in some parts
are aggregated apparently around the large transparent spheres,
this aggregation appearing to take place first just beneath the
blastoderm, where the yolk is already to some extent coherent.

As development proceeds the absorbent blood-vessels dip into
the yolk from the yolk-sac, and the large transparent spheres,
each surrounded by a layer of crystalloids, become attached to
these vessels like onions on a string. This ropy or radially
columnar character of the yolk in the later stages of develop-
ment is very striking even to the naked eye (fig. 106).

The closeness of the blastoderm to the shell, and the elastic
inward curling of the latter as soon as it is cut through, render
the successful removal of the embryo a matter of considerable
difficulty in the earlier stages of development. The embryo is
sometimes found adhering to the inner surface of the part of
the shell removed, and its remarkable transparency in the
earlier stages makes it still more difficult to manipulate.

14, ARTHUR DENDY.

3. Systematic AccOoUNT OF THE STAGES IN DEVELOPMENT.
Stage ©! (figs. 1—5).

This is the earliest stage which I have yet obtained, and to
it I assign embryos numbered 5, 6, 7, 8,9. The eggs from
which these embryos were removed were collected on Stephens
Island on November 14th, 1897, and the embryos were re-
moved by me on November 22nd, the day on which I received
them. ‘They were therefore at least eight days old.

Already the blastoderm appears to have spread completely
around the yolk, the epiblast forming over the greater part of
its extent a thin transparent membrane, to the naked eye
resembling a vitelline membrane, but composed of a layer of
polygonal, nucleated, flattened, or in some parts columnar
cells. This blastoderm lay so near to the shell that I found it
impossible to open the egg without puncturing it and allowing
some of the very liquid yolk to escape.

Beneath the superficial epiblastic layer of cells, and lying
in the outermost part of the yolk, are numerous nucleated
stellate cells which form a loose network holding the outer
part of the yolk together, and causing it to adhere to the under
surface of the epiblast, whereby the blastoderm outside the
area pellucida acquires a thick, blanket-like character. These
cells beneath the superficial epiblast may be termed “ lower-
layer 77 cells’ (fie.c4, es, Y.)).

Over a certain area, corresponding approximately to the
area pellucida, a large cavity has made its appearance in the
lower layer; this is the segmentation cavity (fig. 4, S.C.),
whereby the true embryonic portion of the blastoderm is widely
separated from the underlying yolk. The floor of the segmen-
tation cavity is formed by a very thin and delicate membrane,
which I propose to call the sub-embryonal membrane (fig. 2,
S.#. M., and fig. 4, the dotted line). The sub-embryonal
membrane is composed of yolk-spheres held together by a

1 T commence with the letter C in order to make allowance for earlier
stages which may be forthcoming later on.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 15

meshwork of stellate cells; it forms part of the lower layer,
and is continuous round the margin of the segmentation cavity
with the similar but much thicker layer which elsewhere lies
immediately beneath the epiblast. The roof of the segmenta-
tion cavity is formed by the area pellucida, which, however, is
not at all sharply distinguished from the more opaque area
which surrounds it.

In the hinder part of the area pellucida the embryo appears
as a cap-shaped structure (fig. 1) projecting above the general
surface of the blastoderm, from which it is marked off by a
deep fold shaped like a horseshoe, very pronounced at the
broad anterior end, but gradually dying out to the general
level of the blastoderm behind. This is evidently the head-
foldi (figs. 1, 3,5, A. F.).

‘Towards the close of this stage a distinct blastopore (figs. 1,
3, 4, B. P.), with a well-defined crescentic anterior margin,
appears at the posterior narrow end of the embryo.

The embryo at this stage is very difficult to remove and
harden satisfactorily. Kleinenberg’s picric acid does not seem
to be altogether suitable for the purpose. Owing, doubtless,
to contraction, embryos 6 and 8 became cracked on top, while
No. 7 appears to have shrunk inwards, so that it is now
concave above and convex below, instead of just the opposite.
Nos. 5 and 9, however, were satisfactorily preserved. The
former was mounted whole in Canada balsam, and the latter
cut into longitudinal sections.

A median longitudinal section passing through the blasto-
pore has the appearance shown in fig. 4. It will be seen that
in the area pellucida around the embryo the epiblastic cells
gradually lose their flattened character and become elongated
radially, prismatic, with very conspicuous, very darkly staining
nuclei. Beneath them the lower-layer cells, here free or nearly
free from yolk, form a fairly compact multiple layer. The
lower limit of this inner layer is perfectly definite, but no very
distinct hypoblast can yet be said to exist, though it will be
formed later on from the deepest of the cells. (The cells of
the sub-embryonal membrane, in the floor of the segmentation

16 ARTHUR DENDY.

cavity, doubtless belong also to the lower layer, as already
observed, but they take no part in the formation of the embryo,
and may in future be disregarded.)

In the embryo itself the epiblast and a lower layer can still
be distinguished, but with a totally different histological strue-
ture. Immediately behind the blastopore both layers merge
into a thick, dense mass of cells of very characteristic appear-
ance. Hach cell contains a very large but faintly staining
oval nucleus, containing a few small, darkly staining granules.
Under a low power the nuclei look as if they were the actual
cells. This mass of cells doubtless represents the primitive
streak (figs. 38, 4,5, Pr. S.). The blastopore (B. P.) appears
as a pitting in of its upper surface, as yet extending down-
wards and forwards only for a short distance, but destined later
on to form a very distinct neurenteric passage opening below.
The cells immediately surrounding the blastopore show a slight
trace of radial arrangement.

In front of the blastopore the general cellular mass of the
primitive streak passes into two perfectly distinct layers—an
outer epiblast composed of several tiers of radially elongated
cells, and an inner layer of cells not differing in any way from
those of the primitive streak itself. This inner layer, which
I propose to consider as at any rate chiefly mesoblastic, appears
to be simply a forward prolongation of the primitive streak
beneath the epiblast.

The epiblast of the embryo itself differs much in appearance
from that of the surrounding blastoderm; the outlines of the
cells are much less distinct, while their nuclei stain with much
less intensity, and are coarsely granular instead of compara-
tively homogeneous; moreover they are arranged in several
layers. Just in front of the blastopore the epiblast, though
thick, is not nearly so thick as the layer beneath it; but as we
trace it forwards its thickness greatly increases, attaining its
maximum about the anterior margin of the head-fold, and then
somewhat suddenly diminishing again as it turns downwards
and backwards to join the epiblast of the area pellucida. In
the region of the head-fold the epiblast forms a darkly stain-

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 17

ing granular mass, in which it is impossible to make out the
outlines of the cells at all. Between this and the primitive
streak its cells show a prismatic arrangement, and may be
regarded as forming a medullary plate (fig. 4, WM. P.). Just
in front of the blastopore, especially a little to each side of the
middle line, it shows a strong tendency to separate from the
layer beneath it.

The inner layer (P. S. M.) of cells (mesoblast), continued
forwards from the primitive streak beneath the epiblast, gra-
dually thins out till it reaches a point a little in front of the
middle of the embryo, where it is actually thinner than the
overlying epiblast; it then thickens again considerably in the
neighbourhood of the head-fold, and finally contracts once
more to become continuous with the lower layer (Z. Z.) of the
area pellucida.

Behind the blastopore the primitive streak passes gradually
into a thick mass of precisely similar cells, which can be traced
backwards for some distance beneath the epiblast of the area
pellucida. This mass of mesoblastic cells (fig. 4, Sz.) forms an
ill-defined transverse thickening behind the blastopore, and
evidently represents the so-called “sickle” of other reptilian
embryos.

It. will be observed that no amnion is yet present. It is
true the section represented in fig. 4 shows an uprising of the
blastoderm (pro-amnion) in front of the head-fold, but nothing
was visible in the living embryo, and I have no doubt the
appearance shown is simply due to contraction.

Stage D (figs. 6—14).

I have obtained only a single example of this stage—viz.
embryo 58, removed on December 9th, from an egg found by
Mr. Henaghan about the end of November.

The external appearance of this embryo, when viewed from
above as an opaque object without staining, is shown in fig. 6.
No trace of the amnion is yet visible. The head-fold still
projects conspicuously above the blastoderm, and has grown
forwards, while its anterior end has become considerably nar-

voL. 42, part 1.—NEW SER. B

18 ARTHUR DENDY.

rowed, so that the embryo is now broadest in the middle
about where the lower limb of the head-fold joins the under-
lying blastoderm (fig. 7, #. Sp.). The hinder half of the
embryo is narrower than the broadest part of the head-fold,
and forms a well-defined elevation above the surrounding blas-
toderm, with the conspicuous blastopore (B. P.) lying in the
mid-dorsal line near the posterior end. The most posterior
portion of the embryo is formed by the primitive streak behind
the blastopore, which rises up, and is very sharply defined
from the surrounding blastoderm (fig. 7).

In the mid-dorsal line of the embryo, between the blasto-
pore and the anterior extremity, a faint indication of the
medullary groove was visible even in the opaque, unstained
embryo (fig. 6, @.G.). After staining and clearing the
medullary groove was more conspicuous (fig. 7, MZ. G.), but
owing to a certain amount of crumpling in the head-fold its
course could not be exactly followed except in sections.

A selection from the series of transverse sections into which
this embryo was cut is represented in figs. 8 to 14, the sections
being arranged in order from in front backwards.

Fig. 8 represents a section through the area pellucida, just
in front of the head-fold, and it will be seen at once that it
presents a very curious appearance. In and about the middle
line the three germinal layers are all clearly differentiated.
The epiblast (£p.) lies above, composed of a single layer of short
columnar cells, with very large, darkly staining nuclei. The hypo-
blast (Hyp.) lies below, and is formed of short columnar cells
clearly derived from the lowermost cells of the original lower
Jayer. These hypoblast cells present an appearance of cilia-
tion, but this may be due to post-mortem changes. Between
the epiblast and hypoblast the remainder of the original lower-
layer cells form the mesoblast (Mes.), which has already split
into two perfectly distinct layers, an outer somatopleuric, which
cleaves to the epiblast, and an inner splanchnopleuric, which
cleaves to the hypoblast. Between these two a wide cavity
(P. C.) has appeared, the development of which causes the
somatopleure to bulge slightly upwards, and the splanchno-

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 19

pleure strongly downwards. ‘Traced backwards this cavity
appears to divide into two irregular fissures (fig. 9), lying one
on either side beneath the head-fold.

The development of a large coelomic cavity at this stage, and
in this situation, seems very remarkable. That it is not in
any way due to accident seems to be proved by the marked
differentiation of the hypoblast cells beneath it, which acquire
their characteristic short columnar form earlier in this parti-
cular region than anywhere else. Comparison with later stages
seems to me to prove pretty conclusively that it is due toa
very early development of that part of the ceelom which will
later on form the pericardium, being carried backwards into
the body of the embryo with the lower limb of the head-fold
(compare figs. 24, 50, P. C.).

Fig. 9 represents a section passing through the head-fold,
which lies quite freely above the blastoderm. In the blasto-
derm beneath the head-fold are seen the backward prolonga-
tions of the celomic space above mentioned, and the epiblast
has become flattened. In the embryo itself the epiblast on
the dorsal surface forms a medullary plate, in the middle of
which lies the medullary groove (M.G.). The alimentary
canal (4/.C.) is completely enclosed in this region, forming
a wide space lined by slightly flattened hypoblast cells, not
at all sharply marked off, especially below, from the miesoblast
cells. The latter, for the most part, form a loose network of
irregular cells lying between epiblast and hypoblast, but denser
and apparently undergoing more rapid division on the ventral
surface.

Fig. 10 represents a section just behind the head-fold, show-
ing the alimentary canal still widely open below. The epi-
blast on the dorsal surface on each side of the medullary
groove is widely separated from the underlying mesoblast, but
the floor of the medullary groove itself remains in close con-
tact with it. The mesoblast and hypoblast of the embryo are
clearly seen to be here formed primarily by infolding of the
original lower layer of the blastoderm before it has become
clearly differentiated into two layers. It seems likely, how-

20 ARTHUR DENDY.

ever, that part of the mesoblast in this region is derived from
the forward growth of the primitive streak described in the
preceding stage.

Passing backwards from the last section the medullary
groove is seen to disappear, and give place to a continuous
flat medullary plate, composed of radially elongated, columnar
epiblasu cells, and extending for some distance. As it ap-
proaches the blastopore the epiblast thins out in the middle
(fig. 11), and a groove (P.G.) appears on its upper surface,
which can be traced back almost to the blastopore. This
groove has prominent lips in the hinder part of its course
(fig. 12, P.G.). It probably corresponds to the primitive
groove of the chick.

For some distance in front of the blastopore the mesoblast
forms a very dense mass of cells, which appear to be rapidly
proliferating, and this is continued as a thick layer right
across the middle line, and also outwards, on either side, as
two thin lateral sheets (figs. 11,12). This mass of cells is
clearly derived from the forward prolongation of the primitive
streak described at Stage C (compare fig. 4). Itis perfectly clear
that the embryonic mesoblast originates from two sources, viz.
(1) from the original lower-layer cells of the area pellucida after
separation of the definite hypoblast (compare fig. 8),and (2) from
the cells of the primitive streak. The limits between the two
different kinds of mesoblast I have not been able to determine.

Fig. 12 represents a section just in front of the blastopore,
through the primitive streak. It will be noticed that the
columnar or prismatic epiblast of the embryo passes into the
undifferentiated mass of the primitive streak a little way in
front of the blastopore, so that the section under consideration
shows no differentiated epiblast on the dorsal surface at all.
The mid-dorsal line is occupied by the primitive groove (P. G.)
with its prominent lips. Beneath this, deeply embedded in the
dense mass of the primitive streak, lies a small cavity (V. £n.).
This is the cross-section of the anterior ventral end of the
neurenteric canal, which passes downwards and forwards from
the blastopore, but does not yet open below.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 21

On either side the cells of the primitive streak divide into
two distinct layers, an upper layer which joins the columnar
epiblast of the area pellucida, and a lower layer which forms
part of the lateral sheet of mesoblast running out into the sur-
rounding blastoderm.

Between this lateral sheet and the columnar epiblast, espe-
cially adherent to the under surface of the latter, a few irre-
gular mesoblast cells are clearly recognisable. These cells are,
I think, undoubtedly derived from the original lower layer.
It will be seen by reference to figs. 12—14 that the epiblast,
with these few adherent mesoblast cells, is separated by a dis-
tinct interval from the deeper or primitive-streak mesoblast.
This separation appears to be the commencement of the split-
ting off of the serous envelope (S. Hn.) which will be described
in later stages (compare figs. 30, 31, 55).

The hypoblast (Hyp.) forms an ill-defined and possibly im-
perfect layer of flattened cells, closely adherent to the under
surface of the primitive streak, and probably also derived, as
in the front part of the embryo, from the original lower layer
of the blastoderm.

The section represented in fig. 13 passes actually through
the opening of the blastopore (B. P.), which is seen as a wide
funnel-shaped depression on the upper surface of the primitive
screak, lined by distinctly columnar cells.

Fig. 14 represents a section through the primitive streak
behind the blastopore, and embraces a somewhat wider area
than the preceding figures, so as to show a portion of the area
opaca and its junction with the area pellucida. The yolk-laden,
sub-embryonal membrane, which forms the floor of the seg-
mentation cavity beneath the area pellucida, is not represented.
It will be seen that the blastoderm of the area opaca (A. O.) is
very much thicker than that of the area pellucida (A. P.) owing
to the enormous development of the lower layer with its
contained yolk-spheres. In this region of this particular
embryo the thickened lower layer of the area opaca passes
quite suddenly into the thin lower layer of the area pellucida,
forming a kind of “ germinal wall” (G. W.), but the transition

22 ARTHUR DENDY.

from the one to the other is not always by any means so
abrupt. Just outside the margin of the area pellucida bleod
isiands (B. I.) are making their appearance. These seem to me
to be derived from the sheet of mesoblast which grows out
from the primitive streak, and which extends about as far as
the germinal wall.

Stage E (figs. 15—384).

I refer to this stage embryos numbered 56 and 64, found on
Stephens Island about the end of November, and removed
from the eggs on December 9th. The two agree very closely
with one another in all essential features, and differ very
conspicuously from the preceding stage. This difference is
due chiefly to the development of the pro-amnion and amnion,
which has taken place in such a manner that the anterior end
of the embryo completely enveloped in its pro-amniotic
covering, now lies freely beneath the general surface of the
blastoderm, instead of above it as in the preceding stages, while
a large crescentic aperture, situated on the surface of the
blastoderm in front of the primitive streak, leads into the
amniotic cavity behind (vide figs. 15—20).

Although all the stages in the process have not been actually
observed, it is evident that the pro-amnion and amnion
develop in the following way :—The anterior end of the embryo
sinks down, carrying with it the part of the blastoderm (pro-
amnion) which lies beneath. The blastoderm (pro-amnion)
thus pushed downward seems to become stretched at the same
time, and thus forms a very thin sac (figs. 22—27, Pro-
Am.) whose lips meet together and unite in the mid-dorsal
line above the head. The pro-amnion thus formed around
the anterior end of the embryo is composed of two layers,
which must be regarded as epiblastic and hypoblastic re-
spectively, with no distinct mesoblast between them. The
cells of both layers have become flattened out, and so closely
pressed together that it is difficult, if not impossible, to dis-
tinguish them except where they pass into the epiblast and

hypoblast of the embryo itself (figs. 24,20). The thin pro-

OULLINES OF THE DEVELOPMENT OF THE TUATARA. 23

amniotic membrane thus formed around the head becomes
completely separated from the overlying blastoderm (area
pellucida), which lies at some little distance above, and exhibits
no special peculiarities (fig. 22).

As we trace the pro-amnion backwards we find that at the
posterior limit of the anterior closure of the alimentary canal
its inner layer becomes continuous with the epiblast of the
embryo proper, while its outer layer passes into the hypo-
blastic lining of the alimentary canal (fig. 25). A little further
back the outer or hypoblastic layer of the pro-amnion becomes
continuous above with the ill-defined hypoblast of the over-
lying blastoderm, to which the embryo is thus attached in the
mid-dorsal line (fig. 27). Still further back the amnion is
formed by an uprising of the epiblast on each side of the mid-
dorsal line, forming two folds which meet and fuse above the
medullary groove, and thus arch it over, while mesoblast cells
extend in between the two layers of the epiblast (figs. 28, 29).
Further back again the arching over of the medullary groove by
the amnion is as yet incomplete, so that the narrow amniotic
cavity opens widely to the exterior (fig. 380), and behind this
the amnion has not yet begun to develop (figs. 31, 32).

In embryo 64 the formation of the amnion has extended
back as far as the region of the primitive streak, which, how-
ever, is still incompletely arched over by the uprising folds of
the somatopleure (fig. 34).

A comparison of transverse sections through the posterior
portions of embryos of Stages C and D shows that the formation
of the amnion is here also accompanied by a downsinking of
the body of the embryo, but not nearly so marked as in the
anterior part (compare figs. 9—14 with figs. 22—382).

The separation of the serous envelope, commenced already
in the preceding stage, is progressing in the hinder part of the
area pellucida. Figs. 29—382 show clearly that it consists of
the superficial epiblast of the area pellucida, with a few
adherent irregular mesoblast cells (derived from the original
lower layer), which splits off from the deeper mesoblast derived
partly from outgrowth of the primitive streak.

24 ARTHUR DENDY.

The rudiment of the pericardium, observed in the preceding
stage, appears to be almost cbliterated by stretching in the
formation of the pro-amnion. It seems to be represented by
the split, containing a few mesoblastic cells, which lies between
the epiblast and hyboblast, where they turn up to join the
pro-amnion just in front of the opening into the anterior closed-
in part of the alimentary canal (fig. 24).

Having thus described the condition of the fetal mem-
branes, we may consider next the external features of embryos
of this stage.

The appearance of embryo 56, when viewed from above as
an opaque object without staining, is represented in fig. 15.
The embryo lies close to the hinder end of the pear-shaped
area pellucida, the outline of the anterior portion being some-
what dimly seen through the overlying blastoderm, At the
hinder end, just in front of the primitive streak, lies the large
crescentic opening into the amniotic cavity (Am. 0.) On
each side the primitive streak runs out into a wing of meso-
blast (Z.W.), which at once turns sharply forward and runs
for a while almost parallel with the long axis of the embryo,
but dies out before reaching the middle of its length. Of
course the mesoblast also extends backwards from the primitive
streak, and the latter, with its posterior and lateral mesoblastic
outgrowths, forms a crescentic thickening around the hinder
end of the embryo which corresponds more or less closely to
the “sickle” of other reptilian embryos.1

Fig. 16 represents the same embryo viewed trom beneath
after removal of the sub-embryonal membrane, a fragment of
which, however, remains just behind the embryo (S. #.dZ).
The anterior half of the embryo is now, of course, very much
more distinct, and is seen as a finger-shaped body, curving
somewhat downwards from its junction with the posterior half,
and projecting freely forwards beneath the blastoderm enve-
loped in its thin transparent pro-amnion. The mesoblastic
outgrowth from the primitive streak appears now to form a
conspicuous thickening all round the posterior half of the

' Compare Mitsukuri and Ishikawa (5).

OUTLINES OF THE DEVELOPMENT OF THE TUATARA., 25

embryo. On the ventral surface of the embryo the alimentary
canal is outlined by two parallel grooves, united in front by a
deep crescentic transverse fold which marks the limit to which
the anterior enclosure of the alimentary canal has as yet
extended (fig. 16, F. Sp.).

Fig. 17 represents the same embryo seen from above as a
transparent object after staining and clearing.

Figs. 18—20 represent three corresponding views of embryo
64. The sub-embryonal membrane not having been so com-
pletely removed in this case, the hinder end of the embryo
could hardly be made out at all as an opaque object.

Fig. 20 is especially worthy of note, as it shows very clearly
the spreading of the lateral wings of mesoblast (L. W.) derived
from the primitive streak. It will be seen that they spread
outwards and forwards as two irregular sheets, following
mainly the line of junction between the area pellucida and area
opaca. Reference to transverse sections (figs. 30, 31) show
that these sheets (Z.W.) lie in the floor of the developing
pleuro- peritoneal space which separates the serous envelope
from the future yolk-sac. ‘They lie, in fact, in the yolk-sac
itself, where they undoubtedly give rise to a portion at any
rate of the vitelline vessels (compare figs. 20 and 58).

Turning now to the more minute structure of these embryos,
the following features seem worthy of note.

The enclosed portion of the alimentary canal extends further
back than in the preceding stage, and is considerably narrowed,
while the hypoblastic cells forming its wall have assumed their
characteristic short columnar shape (compare figs. 9 and 23).
The change from flattened to columnar in the shape of the hypo-
blast cells extends also for some distance behind the limit of
closure (except in the mid-dorsal line), but gradually ceases as
we approach the primitive streak (figs. 25—33).

The notochord has made its appearance in the middle line be-
neath the medullary groove. It appears to me quite clear that
it has been formed by separation of a solid rod of cells from the
sheet of mesoblast which has already been noticed as growing
forward from the primitive streak, and which, it will be re-

26 ARTHUR DENDY.

membered, was continuous in the preceding stages across the
middle line beneath the medullary plate. This mode of origin
is very clearly shown in the section represented in fig. 33, and
also in the sections represented in figs. 81 and 32. As we
trace it towards the anterior end the forward growth of primi-
tive-streak mesoblast gradually dies out, and so also does the
notochord, which gradually narrows and finally disappears
somewhere between the point where the head end of the embryo
becomes free from the overlying blastoderm, and the point to
which the anterior closure of the alimentary canal has ex-
tended (vide figs. 25—82),

Beneath the notochord the hypoblastic lining of the alimen-
tary canal has not yet assumed its characteristic short columnar
form, even anteriorly.

The medullary groove at this stage has extended backwards
as far as the primitive streak, and has become continuous with
the neurenteric canal or blastoporic passage, which, however,
has not yet acquired its ventral opening (see figs. 31, 32).
The columnar cells lining the neurenteric canal appear to be
ciliated, but this appearance may be due to post-mortem changes.
The blastopore is no longer recognisable as a distinct structure,
its lips having become confluent with those of the medullary
groove in front of it. Posteriorly, however, in embryo 64 a
longitudinal furrow, not lined by columnar cells, is continued
backwards from the end of the medullary groove for some
distance over the surface of the primitive streak (fig. 34). This
furrow, like the similar one noticed in front of the blastopore
at the preceding stage (fig. 12), probably represents a portion
of the primitive groove, which would thus seem to extend both
in front of and behind the blastopore.

Tracing the medullary groove forwards (figs. 21—22) we
find that although it has become very deep, and its lips are
approaching one another, yet it is still open for either the
whole or the greater part of its length. In both embryos it is
greatly dilated in front (figs. 17, 20, 21) to form the rudiment
of the brain, and in No. 64 this rudiment is already differen-
tiated into fore-, mid-, and hind brain, the latter being repre-

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 27

sented on the ventral surface by two distinct dilatations (fig. 21),
while about the region of the mid-brain the lips of the medullary
groove are already united by a great proliferation of cells.

In embryo 56 the medullary groove exhibits a sudden con-
tortion or curvature in the region of the hind brain (fig. 17),
but as this is more conspicuous in some later embryos I shall
postpone any further account of it, as well as of the closure of
the medullary groove, until I come to deal with them.

Before leaving this stage it may be worth while to note
certain differences exhibited by the blastoderm in the neigh-
bourhood of the primitive streak, especially in embryo 64, as
compared with the corresponding parts at the preceding stage
(D.), although these differences are probably due rather to
individual variation than to progressive development. The
differences in question are strikingly brought out by comparison
of figs. 14 and 34, representing transverse sections through
the primitive streak of embryos 58 and 64 respectively. In
the former, as already observed, there is a distinct ger-
minal wall, due to the sudden diminution in thickness of the
lower layer of the blastoderm at the juncture of the area opaca
with the area pellucida, so that only a very thin stratum of
lower-layer cells is continued across the area pellucida and
beneath the primitive streak. In the latter (fig. 34) the lower-
layer cells form a thick stratum both at the sides of and beneath
the primitive streak, though in the front part of the embryo
they thin out suddenly on meeting the margin of the area
pellucida as in embryo 58. Thus the primitive streak is sur-
rounded, except above, by a network of stellate cells with yolk-
spheres entangled in their meshes (compare also fig. 385, Stage F).
In transverse sections of embryo 64, taken a little in front of
the primitive streak, the twofold origin of the mesoblast is
very clearly recognisable (fig. 33). The two lateral sheets of
mesoblast (P. S. M.), derived from the proliferation of the
compact rounded cells of the primitive streak, are seen thrust-
ing themselves in between the loose stellate cells of the original
lower layer (Z. ZL. M.), and dividing the latter into two parts,—
an upper very thin layer which adheres to the epiblast, and a

28 ARTHUR DENDY.

lower, thicker layer, from the lowest cells of which the hypo-
blastic lining of the alimentary canal is differentiated.

Stage F (figs. 35—46).

To this stage I refer embryos 61 and 72, collected about the
end of November, and removed from the egg on December 9th
and 10th. Though on the whole closely agreeing with the
last stage, these embryos differ in certain noteworthy respects,
so that it seems desirable to deal with them separately. They
illustrate well the great difficulty which I have experienced in
classifying the embryos, owing to the fact that events do not
always take place in exactly the same order in different indi-
viduals.

The external characters of the embryos themselves agree so
closely with those of the preceding stage that it seems un-
necessary to do more than refer the reader to figs. 87—4l.
Both embryos, however, differ from those described in the last
stage, in the remarkable fact that the amniotic cavity is con-
tinued backwards for some distance as a narrow canal (the
posterior amniotic canal), which opens to the exterior through
a crescentic aperture on the surface of the blastoderm at some
distance behind the primitive streak (figs. 37, 39, 40, 41).
This tubular prolongation of the amniotic cavity may either lie
in a straight line with the long axis of the body (fig. 41), or
obliquely to it (fig. 39). It is formed by invagination of a strip
of modified epiblast continued backwards from the primitive
streak, combined with an uprising of the edges of the strip,
which unite and close it in above (figs. 45, 46). The modi-
fication of the epiblast in question consists in a loss of defini-
tion on the part of its cells, the nuclei of which stain less
intensely than those of the normal epiblast of the area pellu-
cida. These characters are shown in fig. 46, which represents
a portion of a transverse section passing through the posterior
amniotic opening, where the amnion is still in process of
formation by the uprising of the epiblast on either side of the
modified strip. When the posterior amniotic canal has be-
come closed in, it lies altogether below the superficial epiblast,

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 29

and a few mesoblast cells make their way in between the two
(fig. 45). It is very narrow, and the cells of its roof are much
more clearly differentiated from the surrounding mesoblast than
those of its floor.

In both embryos of this stage the medullary groove has
become closed in the region of the mid-brain by the union of
its two very prominent lips to form a thick mass of cells,
which projects backwards so as to overlap the hind brain.
The thickening of the lips of the medullary groove in this
region is already prominent at the preceding stage, as shown
in fig. 27. Fig. 42 shows the thickened lips united to form
the roof of the mid-brain (A. WZ). Fig. 43 shows this roof
overlapping the hind brain, and fig. 44 is through the hind
brain a little further back, showing the medullary groove still
widely open. Fig. 35, representing somewhat diagrammatic-
ally a median longitudinal section of embryo 61, shows that it
is only in the region of the mid-brain that the closure of the
medullary groove has become completed.

In both these embryos also, but especially conspicuous in
No. 72 (fig. 39), there is a sharp bend in the region of the
hind brain. This bend appears to be due to the restraining
influence of the pro-amnion, which attaches the head to the
under surface of the blastoderm in the region of the mid-brain,
and seems to prevent the embryo from elongating in a straight
line. The result is that the head appears to be pulled up-
wards and backwards, as a horse’s head is pulled back by the
reins (compare fig. 36, and Stage G, tig. 50). This pulling
back of the head probably accounts for the overlapping of the
hind brain by the roof of the mid-brain (compare figs. 39 and
43, R. M.).

In embryo 72 the neurenteric canal is still closed below, but
in No. 61 it opens widely into the future alimentary canal
just behind the notochord, and in front of the great mass of
the primitive streak. This opening is clearly shown in the
longitudinal section represented in fig. 85, which also shows
very clearly how the notochord appears as a forward growth
from the region of the primitive streak, or, to speak more

30 ARTHUR DENDY.

accurately, as the axial portion of this forward growth sepa-
rated from the lateral portion ; while the hypoblastic lining of
the alimentary canal assumes its characteristic columnar form
from in front backwards.

The rudiment of the pericardium, which should lie just in
front of the letters “ #. Sp.” in fig. 85, appears in this section
to be quite obliterated, but it is seen again in the correspond-
ing position at the next stage (compare fig. 50, P. C.).

Stage G (figs. 47—50).

I have only a single embryo of this stage, numbered 59,
which was removed from an egg found on Stephens Island
about the end of November, and opened on December 9th.

When examined as an opaque object this embryo was seen
to differ from those of the preceding stage chiefly in the
appearance of the ventral surface (fig. 47), which showed the
as yet unenclosed portion of the alimentary canal as a shallow,
elongated trough, distinctly bounded all round by a thick
bolster-like rim. Close to the hinder end of the oblong area
thus enclosed the aperture of the neurenteric canal was con-
spicuously visible (fig. 47, N. Kn.). The backward tilting of
the head (fig. 50) is more pronounced than in the preceding
stages, and at the same time the head is turned to one side
(figs. 48, 49).

The structure and relations of the amnion and pro-amnion
remain much as before. The serous envelope has progressed
greatly in development (figs. 47 and 50, S. En.), having been
formed almost entirely by the splitting off of the superficial
epiblast, accompanied by a thin layer of mesoblast, from the
underlying portion of the blastoderm, The latter may now
be termed the yolk-sac (figs. 47 and 50, Y.S.). In the re-
moval of the embryo from the egg the foetal membranes were
considerably disturbed, and the yolk-sac to a large extent torn
off around the embryo; the result of this was to bring out
very clearly its relations to the overlying serous envelope, as
shown in fig. 47. Dorsally, of course, where the pro-amnion
joins the superficial blastoderm, the yolk-sac is reflected for-

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 31

wards to form the outer layer of the pro-amnion, as shown in
fig. 50. Behind this point the serous envelope remains united
with the true amnion in the mid-dorsal line to form a single
membrane, composed of three layers, viz. cubical epiblast on
the outside, mesoblast in the middle, and flattened epiblast on
the inside (fig. 50, Z.A.S.). The space developed elsewhere
beneath the serous envelope is, of course, the pleuro-peritoneal
space, and beneath this space the yolk-sac is not yet, at any rate
in most parts, clearly differentiated into mesoblast and hypoblast.

The posterior amniotic canal is still clearly traceable in
sections ; though, owing perhaps to the crumpling of the
membranes in removal, I failed to observe it in the embryo
before cutting. It lies embedded in the thickness of the serous
envelope above the pleuro-peritoneal space, and it opens in
front into the hinder end of the much wider amniotic cavity
above the embryo, while behind it still opens to the exterior
as before. In this specimen it lies obliquely to the long axis
of the embryo, and its anterior opening lies to one side of the
middle line. It is, therefore, not shown at all in the median
longitudinal section represented in fig. 50.

The alimentary canal, except as regards the development of
the thickened rim already noticed, has scarcely progressed
further than at Stage F. The short columnar or cubical
epithelium which lines it is well marked throughout the
greater part of its extent, but gradually thins out and becomes
flattened behind (fig. 50).

The notochord also has progressed no further. On the con-
trary, it seems to be scarcely so far advanced as in embryo 61
(Stage F), appearing to end in a mass of undifferentiated meso-
blast cells beneath the hind brain (compare figs. 35 and 50).

The pericardium is now again represented by a distinct
cavity lying beneath the anterior enclosed part of the alimen-
tary canal, just where the epiblast and hypoblast are reflected
upwards to join the pro-amnion around the head (fig. 50, P.C.).
Just where the bend takes place the epiblast and hypoblast are
separated from one another by a considerable space. This
space I believe to be the pericardium, although in this embryo

$2 ARTHUR DENDY.

the mesoblast does not appear to extend so far as this, but
ceases somewhat abruptly beneath the fore-brain, It certainly
seems to correspond to the spaces which I have identified as
pericardial in earlier stages, In front of the pericardium the
epiblast and hypoblast unite together closely to form the very
thin pro-amnion of the head region (fig, 50, Pro, Am.).

The medullary canal is now closed in dorsally, except in the
region of the fore-brain, and for a considerable space at the
hinder end, above and in front of the neurenteric canal, where
it still opens widely into the amniotic cavity. Owing to the
peculiar backward tilting of the head, the mid- and hind brains
are bent in a vertical plane somewhat into the form of the
letter S, as shown in fig. 50. The fore-brain is slightly twisted
to one side, so that its long axis lies obliquely to the long axis
of the spinal cord, and the right and left halves of the roof of
the fore-brain overlap one another slightly in the middle line
(fig. 48).

Stage H (figs. 51 and 52),

‘lo this stage I refer embryo 78, removed on December 16th
from an egg found on Stephens Island about the end of
November, Unfortunately the hinder half of this embryo was
accidentally injured after the first drawing (fig, 51) had been
made, so that Lam unable to give as full a description of it as
L could wish.

When viewed as an opaque object the embryo was seen to
differ from the preceding stage chiefly in the much greater
extent to which the closure of the alimentary canal has ex-
tended backwards. Thus the ventral transverse fold (fig. 51,
F. Sp.) caused by the turning forward of the splanchnopleure
lies only a short way in front of the middle of the body, The
enclosed portion of the alimentary canal (fig. 52, 4/4. C.) is
filled with yolk particles, whereby it is rendered very con-
spicuous when the embryo is examined as a transparent object
after staining. Beneath the hinder part of this enclosed
portion lies the pericardium (fig. 2, P.@.) in the form of a
somewhat quadrangular sac, narrower in front than behind,
In the pericardium hes the heart (fig. O2, Z74.), in the form of

OUTLINES OF THE DEVELOPMENT OF TH TUATARA, 83

a short peareshaped sac, ‘Che posterior and narrower end of the

,

heart receives the two vitelline veins (fig, 62, VW. VA) running
in the folds of the splanchnopleare, irom its broad anterior
end a very short bulbus arteriosus (fig, 62, 2B. Av.) vans forward
beneath the throat,

The head has straightened out again, and there is no trace
of the upward and backward tilting noticed in the preceding
stages, ‘The fove-brain lies at the anterior extremity, and has
not yet begun to bend downwards, ‘The overlapping of the
two lateral halves of the roof of the fovesbrain, already slightly
indicated in the last stage, is now much more pronounced, and
gives vise to the very curious appearance shown in fig, 52,
from which it will be observed that it is the left half whieh
overlaps the right,! On either side of the fore-brain a promi.
hent optic vesicle has grown out, and projects somewhat back=
wards (fig, 62, O, V.),

The mesoblist on either side of the medullary canal has
become segmented into about a dozeu not very distinet meso-
blastic somites, Lt was impossible to count them exactly,
partly because of the injury tothe posterior half of the embryo,
and partly because, on their first appearance, the mesoblastic

somites are not nearly so distinct as they are in the chick,

Stage J (figs, 58—73),
This stage is represented by a fine series of embryos, nume-
beved as follows :

ld? Collected on Stephens Island Noy, 14, removed from eee Dee, 10,
40

42 ” ” ” dU, ” ” v,
Ae

dd ‘

AG } ” ” ” Jd, ” ” 10,
48 Pe “A ” 30, ” ” 16,
Ng)

We) Vs re end of Nov,, ” »
Ob

7%

7A vi " ” ” ” 10,
76

7 ” ” ” ad dd 16,

' Vig, 62 is taken from the ventral surface and veversed by the microscope,
60 that the left side of the figure corresponds to the left side of the embryo,

VoL, 42, pant L.—NuW BUR, 0

34 ARTHUR DENDY.

Of course these embryos exhibit a certain amount of variation
amongst themselves, but not in my opinion sufficient to make
it necessary to separate them into distinct stages. I have
accordingly selected No. 44 as typical, and propose to describe
it in some detail, referring afterwards to such deviations from
the type as seem to deserve special mention in the others.

General Relations of the Blastoderm and Embryo.
—The living embryo was still very transparent, but a red-
coloured sinus terminalis was conspicuous in the blastoderm
surrounding the hinder part of the body (fig. 58, S. 7.). The
sinus terminalis forms the margin of the posterior half of a
clear-looking area in which the embryo lies. This clear area
exhibits a strong constriction in the middle, where the vitelline
veins are developing (fig. 56). In some embryos (e.g. No. 46,
fig. 54, and No. 60, fig. 53) the clear area thus acquires a very
characteristic 8-shaped outline at this stage; in No. 44, how-
ever, the shape is less regular.

The part of the blastoderm which forms the front half of the
5, corresponding to part of the original area pellucida, lies
entirely above the anterior half of the embryo, which projects
freely beneath it, and in front of the vitelline veins. This
part of the blastoderm is thin, and contains but little yolk,
hence its transparency. It is composed of epiblast and lower-
layer cells. Some of the latter (mesoblastic) adhere to the epi-
blast to form with it the serous envelope, which, however, is as
yet only partially separated from the underlying yolk-sac com-
posed of the remainder of the lower-layer cells. At the margin
of this anterior half of the clear area the lower layer (yolk-sac)
thickens suddenly and becomes heavily laden with yolk, which
accounts for its greater opacity. At the level of the vitelline
veins the serous envelope and yolk-sac join the amnion and
pro-amuion, thus giving rise to the transverse crease across
the back uf the embryo (fig. 56, Zr. Z.; compare also fig. 50, .
Stage G).

In the hinder part of the 8-shaped clear area, bounded by
the sinus terminalis, the serous envelope is separated from the
vascular yolk-sac by a wide pleuro-peritoneal space on each

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 35

side of the body of the embryo, forming a kind of blister (com-
pare figs. 69—71). Along the mid-dorsal line, however, the
body of the embryo is attached by the true amnion to the
under surface of the serous envelope (compare figs. 68-70),
except just above the primitive streak, where the two have
completely separated (fig. 71).

The Fotal Membranes.—As we have just seen, the
serous envelope (fig. 69, S. Hn.) has completely split off from
the underlying portion of the blastoderm (yolk-sac) throughout
the posterior half of the 8-shaped clear area, while in front of
this it is not yet widely separated. It is composed as before
of the highly characteristic but now somewhat flattened cells
of the superficial epiblast, with their very darkly staining
nuclei, and of a number of adherent mesoblast cells derived
from the lower layer, some of which appear to be specialising
as a distinct epithelioid layer on the under surface of the thin
transparent membrane thus formed (figs. 72, 73, Mes. .).
In the mid-dorsal line of the embryo, from about the level of
the vitelline veins to the primitive streak, the serous envelope
remains continuous with the true amnion (figs. 68—70).
Above the primitive streak, however, it lies quite separate
above the true amnion, so that the two lateral halves of the
pleuro-peritoneal space communicate freely above the embryo
in this region (fig. 71). They also extend backwards behind
the embryo as a single wide cavity, roofed over by the serous
envelope and bounded by the sinus terminalis. On the floor
of this pleuro-peritoneal cavity, formed by the yolk-sac, the
vitelline vessels are rapidly developing in the form of blood
islands (figs. 69, 72, B. I.), which together with the sinus ter-
minalis (S. 7.) are, as already noticed, probably derived from
the sheet of mesoblast which spreads out from the primitive
streak.

The pro-amnion, completely enveloping the anterior end of
the embryo in front of the vitelline veins, has exactly the same
structure and relations as before, being a thin transparent mem-
brane composed of two layers,—an outer hypoblastic, which
is really a portion of the yolk-sac, and an inner epiblastic,

36 ARTHUR DENDY.

which is really part of the somatopleure. Both these layers
are clearly shown in figs. 63 and 64, from which it will also be
seen that there is no recognisable mesoblast between the two.
Behind the vitelline veins, however, the true amnion envelops
the dorsal portion of the embryo, and is composed of somato-
pleure alone, with distinct epiblast and mesoblast, the latter
becoming continuous with the mesoblast of the serous envelope
in the mid-dorsal line (figs. 68—70).

Although it was no longer conspicuous externally, the pos-
terior amniotic canal is still clearly recognisable in sections
behind the embryo. It appears as a very narrow canal (figs.
72, 73, P. A. C.), embedded in the mesoblast of the serous
envelope, and extending obliquely backwards from the point
where the latter becomes disconnected from the true amnion
to a short distance beyond the sinus terminalis, where it still
opens to the exterior on the surface of the blastoderm. The
lurfien of the canal is distinct in the hinder part of its course
(figs. 72, 73), but in front it is closed, so that it is no longer
possible to trace it into connection with the amniotic cavity
above the embryo, as in earlier stages. The epiblast cells
lining the posterior amniotic canal seem to become gradually
assimilated to the surrounding mesoblast cells, so that it is
difficult, if not impossible, to recognise it at its extreme an-
terior end after its lumen has disappeared.

External Characters of the Embryo.—In this par-
ticular specimen the head end of the embryo projects beyond
the anterior margin of the clear area of the overlying blasto-
derm, so that when viewed from above as an opaque object
part of the head is concealed from view by the yolk-laden
opaque area (fig. 56). When viewed from below as an opaque
object the embryo exhibits the appearance shown in fig. 57,
The fore-brain, with the optic vesicles, has been bent down-
wards at right angles to the long axis of the body, so that
what was formerly its anterior extremity now lies ventrally.
Immediately behind the fore-brain a dark-looking triangular
depression, with its apex pointing backwards, marks the
position of the stomodeum (fig. 57, Séom.), where the veniral

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 37

wall of the alimentary canal has already begun to thin out,
preparatory to perforation by the mouth. A short way behind
this is seen the heart (fig. 57, H¢.), lying in the pericardium,
a little in front of the middle of the body. Just behind the
heart lies the opening of the anterior enclosed part of the
alimentary canal. The remainder of the ventral surface is
occupied by the as yet unenclosed portion of the alimentary
canal in the form of a shallow trough with prominent margins,
and with the aperture of the neurenteric canal (fig. 57, N. En.)
lying near its posterior end.

The Alimentary Canal.—The extreme anterior end of
the alimentary canal lies in the angle between the mid-brain
and the downwardly turned fore-brain, its blind extremity
being in the same transverse plane as the now posterior limit
of the fore-brain beneath it. Fig. 62 represents a transverse
section taken just behind the anterior extremity of the
alimentary canal, which in this region occupies only about
one third of the total width of the head. Beneath the
hind brain (fig. 63) it widens out until its side walls meet and
fuse with the lateral epiblast, preparatory to the formation of
the first visceral cleft. Its floor also here becomes extremely
thin, the epiblast of the stomodzum (fig. 63, Stom.) lying in
close contact with the hypoblast, and both being attenuated
preparatory to perforation by the mouth. On approaching the
pericardium, however (figs. 64, 65), the lateral and ventral
walls of the alimentary canal retreat from the epiblast and
assume their normal aspect, which is maintained until, on
reaching the opening to the exterior behind the heart, the
folds of the splanchnopleure diverge, and leave the alimentary
canal widely open below (fig. 67). This open condition is
continued to the posterior extremity behind the conspicuous
aperture of the neurenteric canal (fig. 70). The anterior
enclosed part of the alimentary canal is densely filled with
yolk particles, conspicuous both in sections (figs. 62—66), and
when the whole embryo is viewed as a transparent object
(fig. 59). Its hypoblastic lining is now composed almost
throughout of the characteristic short columnar cells, flattening

38 ARTHUR DENDY.

more or less, however, in the mid-dorsal and mid-ventral lines
towards the anterior extremity.

The Primitive Streak and Neurenteric Canal.—
The primitive streak, or at any rate the posterior mass of
mesoblast derived therefrom, forms a dense mass of cells
behind the neurenteric opening, somewhat triangular in
transverse section, covered by flattened epiblast, and arched
over by the true amnion (fig. 71). The blastopore has dis-
appeared from view externally, being included in the closure of
the hinder part of the medullary groove; but the neurenteric
passage still leads downwards through the floor of the
medullary canal, and places the cavity of the latter in free
communication with the cavity of the unenclosed portion of
the alimentary canal below (fig. 70).

The Notochord.-—The notochord can now be _ traced
forwards to a point just in front of the anterior extremity
of the alimentary canal, where it lies above the latter in the
angle between the mid- and fore-brain. Posteriorly it can, of
course, be traced back to the neurenteric aperture, with the
margin of which it is continuous. In the trunk region it is now
widely separated from the underlying hypoblast (figs. 57—69).

Mesoblastic Somites, Celomic Cavities.—The meso-
blast on each side of the spinal cord has become segmented
into fourteen mesoblastic somites, the first of which lies just
behind the auditory pit, and the last some distance in front of the
neurenteric aperture. These somites are cubical blocks, whose
cells have the usual columnar form and radial arrangement
(figs. 65—68, MW. S:). Anteriorly they are connected with the
peritoneal epithelium by the intermediate cell mass (figs. 59,
§7, 68, I. C. M.), beyond which the somatopleure and the
splanchnopleure diverge, leaving between them only a narrow
coelomic space in the body of the embryo, which is continuous
with the very large pleuro-peritoneal space outside (fig. 68).
Posteriorly, about the region of the twelfth mesoblastic somite
(fig. 69), the coelomic split can still be traced into the somite
itself, the latter having uot yet completely separated off from
the lateral plates of mesoblast,

OUTLINES OF THE DEVELOPMEN'T OF THE TUATARA, 39

In the head region, just outside the aortic dilatation on
either side of the front end of the alimentary canal, lies the
commencement of the head cavity. This is shown in fig. 62
as a small space in the mesoblast (H. C.). Anteriorly it
extends forwards for a short distance, dilating somewhat and
becoming irregular, while posteriorly it soon disappears.

The pericardium is now very largely developed, forming a
great quadrangular sac on the ventral aspect of the embryo
(figs. 59, 64—66, P. C.). The cavity of the pericardium is
continuous above with the narrow ccelomic space appearing
between the somatopleure and splanchnopleure in the body of
the embryo on either side of the alimentary canal (fig. 66).

In the region of the heart (figs. 65, 66) the splanchnopleuric
and somatopleuric mesoblast on either side both turn inwards,
and unite ventrally with the corresponding layers of the
opposite side, the splanchnopleuric mesoblast thus enclosing
the heart, and the somatopleuric helping to enclose the peri-
cardium. Behind the pericardium the somatopleuric mesoblast
turns upwards with the epiblast to form the true amnion,
while the splanchnopleuric mesoblast runs out with the hypo-
blast to join the yolk-sac (figs. 55, 68—71).

The Wolffian Bodies and Ducts.—These organs are
only just beginning to appeaf, and are as yet very ill-defined.
From about the fifth to the eleventh mesoblastic somites the
intermediate cell mass is thickened on its outer aspect (fig. 68,
I. C. M.). Behind the eleventh somite this thickening seems
to be growing freely backwards between the peritoneal epi-
thelium and the epiblast as a solid rod of cells, the rudiment
of the Wolffian duct (fig. 69, W.D.), which can be traced
backwards about as far as the thirteenth somite. Anteriorly,
opposite somites 6 and 7, segmental vesicles appear to be de-
veloping in the thickened intermediate cell mass (fig. 68, S. V.).
It would be beyond the scope of the present paper to trace the
development of these organs in detail, but it probably takes
place very much in the same way as described by Weldon for
Lacerta (6).

Vascular System.—The heart (fig. 59, Ht.) has now the

40 ARTHUR DENDY.

form of a wide tube slightly bent on itself, made up behind by
the union of the two vitelline veins running across from the
yolk-sac in the folds of the splanchnopleure just where the
latter diverge, and ending in front in the short bulbus arte-
riosus (fig. 59, B.Ar.). In transverse sections (figs. 65, 66)
the heart appears as a saccular diverticulum of the splanchno-
pleuric mesoblast, here composed of short columnar cells,
beneath the front part of the alimentary canal, hanging down
into the pericardium, and containing two epithelioid tubules.
These tubules (figs. 65, 66, Ep. T.) are continuous behind with
the vitelline veins, while in front they unite to form the bulbus
arteriosus. The bulbus arteriosus very soon divides into two
primitive aorte (figs. 62 to 69, P.Ao.), which run forward
beneath the alimentary canal for some distance, and then,
curving upwards, turn back and run above the alimentary canal
towards the posterior end of the body, giving off the vitelline
arteries (fig. 69, V. A.) in the trunk at the level of the twelfth
mesoblastic somite. Where the two primitive aorte turn up-
wards in front to reach the dorsal aspect of the alimentary
canal they dilate to form a large blood-space (fig. 62, P. Ao.) on
each side of the extreme anterior end of the alimentary eanal;
like the rest of the blood-vessels this space is lined by a distinct
epithelium.

In the splanchnopleuric mesoblast of the yolk-sac, forming
the floor of the large pleuro-peritoneal cavity posteriorly, nu-
merous blood islands have made their appearance (figs. 58, 69,
70, B. I.), and the posterior half of the sinus terminalis is
distinctly visible (figs. 58, 69, 71, S. T.), marking the limit to
which the separation of the serous envelope from the yolk-sac
has extended. These vessels of the yolk-sac, as already noticed,
appear to be formed, at any rate in part, from the sheet of
mesoblast which grows out from the primitive streak.

Nervous System and Sense-organs.—The medullary
canal has completely closed in above, except on what is now
the ventral aspect of the fore-brain, where the margins of the
medullary groove overlap one another as in the preceding
stage, but are not yet fused. Reference to fig. 60, in which the

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 41

right and left sides are reversed, again shows that it is the left
half of the roof of the fore-brain which overlaps the right.
Posteriorly, the medullary canal still communicates freely with
the as yet unenclosed portion of the alimentary canal through
the conspicuous neurenteric passage (fig. 70, NV. En.).

The fore-brain (figs. 59—61, F. B.), as already observed, is
bent down at right angles to the long axis of the body, and
presents two conspicuous hollow lateral outgrowths, the optic
vesicles of the ordinary paired eyes (figs. 59—61, O. V.).

The mid-brain (figs. 58, 60, 61, M. B.), nearly spherical
when seen from above, occupies the extreme anterior end of
the body. It now lies in the same straight line with the hind
brain, from which it is separated by a very strongly marked
constriction, while the curvature which existed in this region
at preceding stages has disappeared.

The hind brain (figs. 58, 63, 64, H. B.) shows indications of
transverse segmentation into at least four neuromeres, and its
roof is thinning ont in the usual manner (fig. 63). It tapers
very gradually behind, and passes insensibly into the spinal
cord.

On either side of the posterior part of the hind brain the
auditory pits (figs. 58, 64, Aw.) have made their appearance as
shallow depressions of the superficial epiblast, widely open to
the exterior and lined by elongated columnar cells.

Other Embryos.—Amongst the other embryos which I
have referred to Stage J, the following seem worthy of special
mention.

In No. 40 the posterior amniotic canal is still distinctly
visible in the unstained embryo when viewed as an opaque
object. It runs from the hinder end of the embryo obliquely
backwards, inclining towards the left, as in embryo 72 (Stage F,
figs. 37, 39), and opens to the exterior on the edge of the clear
area of the blastoderm. I have uot detected this canal ex-
ternally in any other embryo of Stage J, though, as noted
above, it was easily demonstrable in sections of No. 44.

1 This overlap was not observed until the sections were cut, and is there-
fore not shown in Fig, 59.

42 ARTHUR DENDY.

No. 46 exhibits an extremely interesting peculiarity with
regard to the development of the amnion. ‘Towards the hinder
end of the body, but in front of the neurenteric canal, a longi-
tudinal slit-like opening in the serous envelope was clearly
visible in the unstained embryo viewed as an opaque object
(fig. 54, Am.S.). Transverse sections (fig. 55) demonstrate
that this opening leads into the true amniotic cavity above the
embryo, and is due to the two uprising folds of the somato-
pleure not having yet united in the middle line, although the
amnion is completely formed both in front of and behind the
slit, and the posterior amniotic canal is also present behind the
embryo, and no longer communicates with the true amniotic
cavity. A similar opening into the amniotic cavity in this
region is, of course, a feature very characteristic of the chick at
a certain stage in its development, but it appears to be abnormal
in Sphenodon, as I have seen it in no other embryo. It is in-
teresting to notice just within the lips of the slit the sudden
transition from the large epiblastic cells of the serous envelope,
which become markedly columnar at the lips of the slit, to
the flattened epiblast celJs lining the amniotic cavity (fig. 55).

The downward curvature of the fore-brain may be greater
than in No. 44, as in No. 60, where the fore-brain is inclined
downwards and backwards, so that it comes to lie very close
to the heart. In some embryos the cleft, where the two
halves of the fore-brain have not yet united in the middle
line, is still clearly visible externally.

The development of the mesoblastic somites does not appear
to be nearly so definite and regular as in the chick, so that
these are of comparatively little use for the purpose of deter-
mining stages inthe development. It appears that a number of
mesoblastic somites are differentiated almost or quite simulta-
neously, but very indistinctly. In No. 60 only about four pairs
were recognisable in the stained embryo viewed as a trans-
parent object, and in No. 73 they were very indistinct.

I have already stated that some embryos (e. g. 46 and 60,
figs. 54 and 53) show much more clearly than No. 44 the

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 43,

characteristic figure-of-8 form of the clear-looking area of the
blastoderm surrounding the embryo.

In No. 48 the heart was observed to be beating in the living
embryo, but very slowly.

From this stage onwards I propose to describe only those
features which seem to be characteristic of each stage, marking
a definite advance on the preceding stage, or which otherwise
seem to demand special attention, as the organisation has now
become so complex that to describe all the remaining stages in
full detail would be far beyond the scope of the present paper,
especially as the development of the various organs conforms
very closely to that of already well-known types.

Stage K (figs. 74—82).
To this stage I refer six embryos, as follows:

38
20} Collected on Stephens Island Nov. 30, removed from egg Dec. 9.
41

45 oP) 2” ” 30, 2» »” 10.
49 ” ” » 30, >” 3” 16.
80 a _ end of Noy., - ly

Embryo No. 39 may be taken as typical of the stage. The
general relations of the embryo and foetal membranes re-
main as before (vide figs. 74, 75), but the posterior amniotic
canal has almost, if not quite, disappeared, having been only
doubtfully recognised in a single section.

The body of the embryo is still straight as far forward as
the mid-brain, but the anterior part of the head, including the
fore-brain, is bent downwards at more than a right angle, so
as to point backwards, thus considerably reducing the interval
between the head and the pericardium on the ventral aspect
(figs. 74, 75). There is as yet no distinct tail, but the hinder
end of the embryo has become, to a certain extent, folded off
from the yolk-sac, and projects backwards for a short distance
in the pleuro-peritoneal space between the yolk-sac and the
serous envelope (fig. 82).

The alimentary canal has begun to close in behind (figs. 75,
82), and the neurenteric canal opens into the short enclosed

AA. ARTHUR DENDY.

portion behind the notochord (fig. 82), so that it 1s no longer
visible from below. The stomodzum has become widely per-
forated by the mouth (fig. 78). Externally one pair of visceral
clefts is visible (figs. 74, 75). Transverse sections (fig. 78)
show them as a pair of outgrowths of the pharynx which do
not, however, as yet perforate the epiblast in the specimen of
which sections were cut. Just behind the mouth, and lying
above the front part of the pericardium, a median ventral
groove in the floor of the alimentary canal is conspicuous in
transverse sections (fig. 79, Th.). It is lined by columnar
epithelium, and doubtless represents the commencing thyroid
gland.

The two halves of the fore-brain have now completely
united, but the line of union can still be recognised in trans-
verse sections as a shallow median furrow between the paired
eyes (figi77, 7. 0.)

The auditory pits (figs. 74,75, 79, Aw.) are now well defined
and deep, but still widely open to the exterior.

The optic vesicles of the paired eyes have begun to invagi-
nate to form the optic cups (fig. 77, Op. C.), and opposite the
mouth of each the superficial epiblast is thickening to form the
lens (fig. 77, Le.).

The parietal eye makes its first appearance at this stage as a
small rounded diverticulum (primary parietal vesicle) of the
roof of the fore-brain just to the left of the middle line
and, owing to the cranial flexure, on the apparent ventral
surface (vide fig. 76, P. V., in which right and left sides
are reversed). The number of mesoblastic somites is about
twenty-three.

The head-cavities are conspicuous behind the paired eyes
even in the opaque embryo (figs. 74, 75, H. C.), and sections
show that each has begun to divide into two parts (fig. 77,
ETC).

Wolffian ducts and tubules are both present in the anterior
part of the trunk (fig. 80, W. D., W. T.) ; passing backwards
the tubules first disappear, then the ducts also.

In the front part of the trunk a longitudinal blood-vessel is

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 40

present on each side in the mesoblast of the amnion, just above
the line where the latter joins the body of the embryo (fig. 80,
B. V.). These vessels probably join the vitelline veins in
frout, but of this I was unable to make certain, owing to im-
perfection of some of the sections. In two embryos of this
stage the heart was observed to be beating.

One histological feature at this stage seems to deserve
notice. In the hinder part of the body the epiblast (fig. 81,
Ep.) is very conspicuously thickened on the flanks of the
embryo, being divided into distinct inner and outer layers,
widely separated but connected by strands of protoplasm con-
taining a few nuclei.

Stage L (figs. 883—91).
To this stage I refer four embryos, viz. :

47 Collected on Stephens Island Nov. 380, removed from egg Dee. 16.

50 29 bP] > 30, 3) 9 27.
82 be - end of Nov., 55 iy Bie
84 29 23 29 ) ” 27.

The separation of the serous envelope has extended widely,
so that it can be readily torn off nearly all round the yolk-sac
as a thin transparent membrane, but it is still adherent to the
true amnion above the embryo about the region of the shoul-
ders, at the spot marked with an X in fig. 83.

The living embryo is still almost transparent except for the
presence of the red blood, and the beating of the heart was
observed in three cases.

The external characters of the embryo are shown in figs. 83
and 84. Not only is the down-bending of the fore-brain very
strongly marked at this stage, but there is also a strong ventral
flexure in the region of the neck and shoulders, so that the
head end of the embryo hangs down into the yolk, enveloped
in the pro-amnion. A well-marked flexure in the opposite
direction has also made its appearance in the lumbar region.

At the hinder extremity of the body the short tail (figs. 83,
85, T.) projects freely into the pleuro-peritoneal cavity between
the yolk-sac and the serous envelope, and in No. 50 (fig. 85)

46 ARTHUR DENDY.

it has already begun to turn downwards and acquire its cha-
racteristic spiral twist.

The Wolffian ridges (figs. 83, 85, W.R.) have made their
appearance, and are quite conspicuous at the hinder end of the
body.

The cerebral hemispheres (figs. 83, 90, 91, C. H.) have
begun to grow out from the tore-brain, causing a marked
elongation of the head in front of the eyes.

The primary parietal vesicle (fig. 83, P.V.) is prominent,
still in the form of a simple hollow outyrowth of the roof of
the thalamencephalon, causing a projection of the external
epiblast between the paired eyes. The cavity of this vesicle
still communicates freely with that of the brain; its wall is
composed of a layer of columnar cells, like those forming the
brain roof, and is not yet thickened to form a lens.

The auditory vesicles (figs. 83, 84, 91, Aw.) are almost if not
quite closed.

In the paired eyes, pigment is just beginning to be visible
in living specimens, but it is not yet noticeable externally in
the preserved embryo.

The nasal pits (fig. 83, Na.) are just beginning to appear.

The first three visceral clefts are visible externally (figs. 83,
84), and four are recognisable in sections (fig. 91).

Perhaps the most characteristic feature of this stage is the
first appearance of the allantois, which is visible externally as
a small rounded or finger-shaped outgrowth (figs. 883—86, Ad/.)
on the ventral aspect of the body beneath the root of the tail
and between the two Wolffian ridges, and projecting into the
pleuro-peritoneal space above the yolk-sac. It originates as a
very thick-walled outgrowth from the posterior enclosed part
of the alimentary canal, its walls being continuous behind with
the Wolffian ridges. The transverse sections figured (figs. 87
—89), of which that represented in fig. 89 is the most
anterior, show this mode of origin sufficiently without further
explanation.

The closure of the alimentary canal posteriorly has extended
forwards for some little distance, and the enclosed portion

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 47

opens to the exterior by a narrow aperture (fig. 86, O. P. A.)
a short way in front of the allantois. No proctodzeum is yet
recognisable, and the neurenteric canal no longer opens into
the alimentary canal below.

No. 47 has about thirty mesoblastic somites recognisable
in the stained embryo viewed as a transparent object, but they
are very ill-defined at the hinder end, so that it is impossible
to count them exactly.

Sagittal sections through the head at this stage are very
instructive. It would be out of place to describe them in
detail here, but the following features, shown in figs. 90 and
91, are perhaps worthy of mention. In the fore-brain the
cerebral hemispheres (C. H.) and infundibulum (Jnf.) are
clearly recognisable, the latter projecting from the thalamen-
cephalon towards the pituitary body (Pzt.), The hind brain
(ZH. B.) is partially divided transversely into at least six neuro-
meres. ‘The notochord extends about as far forward as the
pituitary body, its extremity lying above and just in front of
the stomodeeum and below the anterior part of the hind brain.
The pituitary body (P7¢.) is just commencing its development
as a thickening of the front wall of the pharynx, composed of
columnar cells and extending forwards between the infundi-
bulum and the front end of the notochord. The two head-
cavities (H.C.) on each side are very conspicuous. The
anterior pair are connected across the middle line by a short
transverse canal (fig. 90, H.C. C.) which hes just in front of
the end of the notochord, while each of the posterior pair
gives off a conspicuous branch (H. C. M.) into the mandibular
arch.

Stage M (figs. 92, 93).

To this stage I refer two embryos, viz. No. 51, col-
lected on Stephens Island on November 30th, 1897, and
removed from the egg on January 38rd, 1898; and No. 81,
collected about the end of November, and removed from the
ege on December 27th, 1897.

No. 51 (figs. 92, 93) may be taken as typical of the

48 ARTHUR DENDY.

stage. The appearance of the living embryo as seen from
above is represented in fig. 98. The embryo beneath the thin
transparent serous envelope lies lengthwise in a well-defined,
oblong, clear area of the yolk-sac, corresponding more or less
closely in extent to the original area pellucida. Outside this
clear area the yolk-sac is very heavily laden with yolk, which
disappears abruptly at its margin.

The sinus terminalis (S.Z.) now lies at a considerable
distance outside the clear area of the yolk-sac, enclosing an
oval space about 10 mm. long by 8°8 mm. broad. Anteriorly
its two halves meet to form a large vein which runs backwards
and somewhat to the left, turning inwards at the left-hand
end of the fold (Zr. Z.) along which the yolk-sac is invaginated
around the anterior end of the embryo to form the outer layer
of the pro-amnion. In this vessel the blood flowed in a steady
stream towards the heart, which was beating about thirty-eight
times per minute. The larger vitelline vessels on the right
and left sides can be traced inwards as far as the edge of the
clear area of the yolk-sac, where they become lost to view
from above, dipping down to follow the splanchnopleure in-
wards along the floor of the pleuro-peritoneal cavity.

Fig. 92 represents the same embryo seen from below as
an opaque object after hardening. The anterior half of the
embryo, up to a point just behind the fore-limbs, is pushed
into the yolk-sac, and in life hangs freely down into the yolk,
enveloped in a thin transparent membrane (the pro-amnion),
composed of the true amnion inside, and an outer layer which
really belongs to the yolk-sac but which is not vascular. (In
hardened specimens the head is turned either to the right or
to the left, according to the way in which the embryo
happened to lie in the hardening fluid.)

The following features in the embryo itself may be regarded
as characteristic of the stage:

The ventral flexure of the body in the region of the
shoulders is more strongly marked, so that the head almost
touches the allantois (.Ad/.), which has increased considerably
in size.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 49

There are (in No. 51) forty-one mesoblastic somites.

The limbs appear as outgrowths of the Wolffian ridges, the
fore-limb (F. L.) opposite to mesoblastic somites 9—13 inclusive,
the hind limb (ZZ Z.) opposite to somites 28—82.

Still only three visceral clefts are visible externally in the
opaque embryo, but the visceral arches between them are
much more prominent than in the last stage, and the superior
maxillary process (S. M.) has begun to grow out.

The nasal pits (Va.) are much larger and more distinct.

The liver (Ziv.) begins to form a conspicuous projection just
behind the heart.

The proctodzum has formed as an invagination of the
epiblast on the ventral surface immediately behind the
allantois.

The primary parietal vesicle still remains as a simple sac
opening into the cavity of the fore-brain, slightly to the left of
the middle line. Its position is again marked, P. V., fig. 92,
by a projection between the paired eyes.

Stage N (figs. 94, 95).

To this stage I refer four embryos, numbered 93—97, all
collected on Stephens Island about the end of December, 1897,
and removed from the eggs on January 6th, 1898.

No. 96 (fig. 94) may be taken as typical of this stage,
which is distinguished from the preceding by the following
features :

The curvature of the body has increased somewhat.

The limbs are much more conspicuous, and the distal
extremity of the fore-limb (#. Z.) has become flattened to form
the hand.

The tail is larger, and more distinctly twisted into a spiral.

The allantois is larger, and has become vascular.

All four visceral clefts are visible externally when the embryo
is examined as an opaque object.

The superior maxillary process (S. 1.) is much larger.

The pigment in the paired eyes is for the first time con-

VoL. 42, PART 1,—NEW SER. D

50 ARTHUR DENDY.

spicuous externally, forming a broad rim around the lens,
interrupted by the choroid fissure.

The parietal eye and its ‘‘stalk’’ are both distinct, the eye
lying upon the roof of the brain, in front of and to the left of the
forward-pointing “ stalk,” the cavity of which still communi-
cates widely with that of the brain. The eye is almost, if not
quite, separated from the “ stalk,” though still in close contact
with it, and its front wall has begun to thicken to form the lens.

The paraphysis appears as a simple backward-pointing diver-
ticulum of the roof of the fore-brain, some distance in front
of the parietal eye, the two being widely separated by the greater
part of the thin roof of the thalamencephalon.

Another noteworthy feature of this stage is the condition of
the anterior part of the notochord, which is twisted into a
spiral beneath the hind brain, while its narrowed tip bends
sharply down and ends just in front of the pituitary invagina-
tion, as shown in fig. 95.1

About this stage also a feature which becomes very con-
spicuous later on in the yolk-sac outside the embryo begins to
make its appearance. This consists in the formation of radial
corrugations of the yolk attached to the under surface of the
yolk-sac just outside the clear area (compare Stage O, fig. 96,
R.C.), due to the development of the absorbent vessels in
connection with the vitelline circulation. Around these
vessels the yolk particles are aggregated as described in the
Introduction.

Stage O (figs. 96—100).
To this stage I refer four embryos, viz. :
89} Collected on Stephens Island about the end of December, 1897, and
ot removed from the eggs on January 6th, 1898.
92. Collected about the same time, and removed from the egg on January
10th.
103. Collected about the same time, and removed from the egg on January 25th.
The following features may be taken as more or less charac-
teristic of this stage:
1 Cf. Parker, W. BR. on the notochord in the embryo of the green turtle
‘Challenger Reports,’ Zoology, vol. i,

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 51

The anterior half of the embryo begins to withdraw from
the invaginated yolk-sac, but the extent to which this takes
place is not at all constant. Thus in No. 108 (fig. 96) the
withdrawal seems hardly to have commenced, while in No. 92
(figs. 97, 100) it has progressed so far that the fore-limb has
become completely extricated.

In No. 92 the yolk attached to the under surface of the
yolk-sac appears to be spreading inwards, so as to reduce the
clear area surrounding the embryo (fig. 97); but in No. 103
(fig. 96) the clear area is as large as in the preceding stage.
The attached yolk is more or less strongly corrugated.

The embryo, as its anterior end withdraws from the yolk-sac,
comes to lie on its left side, and the allantois, now large and
with a well-developed circulation, passes up on the right side
and comes to lie above the embryo, between it and the serous
envelope (fig. 100).

The alimentary canal has now become completely closed in
ventrally, except for a small aperture (figs. 97, 98, O. Al.)
where the splanchnic stalk passes out to the yolk-sac.

The limbs have elongated, and both manus and pes have
become evident (fig. 98), but no digits are yet visible, at any
rate externally.

The tail has lengthened considerably, and is coiled inwards
in a close spiral on the ventral surface between the hind limbs
(fie. 98,. 7).

The visceral clefts have begun to close up, but the hyoman-
dibular (figs. 96, 98, H. M.) is still conspicuous externally.

On the under surface of the head a broad fronto-nasal
process (fig. 99, F. N. P.) is present, but it does not yet
meet the large superior maxillary process (S. WM.) on either
side, so that the external nares (Na.) are not yet closed in
behind.

The mid-brain is very prominent (figs. 98, 99, WM. B.), and
just in front of it the parietal eye and its “stalk” are clearly
visible in the unstained embryo examined as an opaque object,
though as yet without pigment. The “stalk ” appears in the
middle line at the hinder end of the roof of the thalamence-

52 ARTHUR DENDY.

phalon as a round white spot with a darker centre. The eye
itself lies just in front and to the left of this spot, and appears
as another round white spot, distinguished from the one behind
it by its double outline, the inner circle representing the out-
line of the lens. The eye still lies close upon the roof of the
thalamencephalon, and closely pressed against the “ stalk”
behind. Its position is indicated by the letters Pa. E. in
fig. 98.

The paraphysis is still a simple diverticulum of the front
part of the roof of the thalamencephalon, but it has begun to
be folded.

Stage P (fig. 101).

Of this stage I have only a single specimen, numbered 87,
collected on Stephens Island about the end of November, 1897,
and removed from the egg on January 25th, 1898. I suspect
that this embryo was already dead before the egg was opened,
though apparently still in good condition.

The foetal membranes were evidently somewhat injured, pre-
senting at the time when the drawing was made the arrange-
ment shown in fig. 101.

The embryo has completely extricated itself from the yolk-
sac, and lies entirely above it on its left side, so that the pro-
amnion has ceased to exist. It remains attached to the yolk-
sac, however, by the now greatly elongated splanchnic stalk
(Sta.) containing the main trunks of the vitelline vessels.
The clear area of the yolk-sac around the embryo is in this
specimen still very extensive. The amnion (Am.) has been
ruptured and shrunk away from the front part of the embryo,
still, however, partially enveloping the posterior part of the
body. Owing doubtless to the rupture and shrinkage of the
amnion, the junction between the latter and the serous enve-
lope (S. En.) has been pulled backwards instead of lying in its
normal position above the shoulders. Later on, at Stage R,
the true amnion was again found continuously enveloping the
entire embryo, so that its rupture and absence from the anterior
portion in this specimen is doubtless due to accident.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 53

The allantois (Ad/.) is also in an abnormal condition, being
collapsed and shrivelled.

In the body of the embryo itself little advance is visible, at
any rate as regards external characters; and, having only a
single specimen in a doubtful state of preservation, I have not
cut any sections. In the paired eyes the choroidal fissure is
still just recognisable. No pigment is yet visible externally in
the parietal eye. The fronto-nasal process is still widely sepa-
rated from the superior maxillary processes, but has developed
a small median projection which becomes much more prominent
at the next stage. ‘The external opening of the hyomandibular
cleft (H. M. C.) has become very small, but is stili plainly
visible. No trace of digits is yet present externally in manus
or pes.

In short, the only important difference between this stage
and the preceding lies in the complete extrication of the
embryo from the yolk-sac, and the accompanying elongation of
the splanchnic stalk.

Stage Q (figs. 102, 103).

This stage is again represented by only a single embryo,
No. 1, collected on Stephens Island on January 18th, 1897.
Unfortunately the eggs sent in this first consignment all dried
up in transit, the sand in which they were packed not having
been sufficiently moist, and the embryo represented in figs. 101
and 102 was the only one of any service. This embryo was
dead when received, but still in a sufficiently good state of
preservation as regards external characters.

It exhibits a marked advance upon the preceding stage in the
following features :

(1) The fronto-nasal and superior maxillary processes have
united so as to complete the upper margin of the mouth, in the
middle of which a prominent beak-like projection is visible
(fig. 103).

(2) In consequence of this the external nares (fig. 103, Na.)
have become delimited as a pair of small crescentic apertures.

54 ARTHUR DENDY.

(3) Five digits appear as blunt projections on the margins of
both hand and foot.

(4) The hyomandibular cleft has completely closed.

I am inclined to believe that pigment first appears in the
parietal eye at this stage, but the sections are not sufficiently
well preserved to be conclusive.

Stage R (figs. 104—107).

Owing to its long duration this stage is represented in my

collection by a large number of embryos, viz. :

9—4, collected on Stephens Island on July lst, 1897, and removed from the
eggs about a week later.

141—148, collected on Stephens Island on February 4th, and removed from
the eggs on March 8th, 9th, 12th, and 14th, 1898.

151—158, removed from the eggs on April 5th, 1898.

159—161, removed from the eggs on May 12th, 1898.

162, removed from the egg on June 24th, 1898.

169, 170, removed from the eggs on June 14th, 1898.

I have also a number of eggs, doubtless containing embryos
of this stage, as yet unopened.

In this stage the embryos pass the winter, entering upon it
in February or March, and developing very slowly through the
winter months. Embryos removed from the eggs in March
almost exactly resemble in external characters those removed
in July, the differences not being, in my opinion, sufficient to
necessitate a separation into distinct stages. The external
differences concern chiefly the appearance of the head as
viewed from above, the younger embryos showing distinctly
the optic lobes and cerebral hemispheres through the integu-
ment, while in the older embryos (fig. 105) these are no longer
visible. As regards internal differences I have not made a
sufficiently detailed study to say much, but it is noteworthy
that while in the younger embryos the parietal eye still lies
close up against its so-called “stalk,” in the older ones it is
separated by a considerable interval from the blind end of the
latter, and the nerve of the parietal eye has made its appear-
ance. It is also noteworthy that whereas at the commence-

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 55

ment of this stage little or no ossification has taken place, this
process has progressed very considerably before its close.

This stage is a very good one at which to describe the con-
dition of the foetal membranes at what is probably almost, if
not quite, the acme of their development. The following
account is based chiefly upon dissection of the membranes in
Nos. 143 and 149, verified by dissection of others..

The embryo lies on its left side and lengthwise in the shell,
and by opening the eggs carefully under water, in which they
sink, it was found that there is no air-chamber.

On piercing the shell carefully a small quantity of a thin
colourless liquid is squirted out with some force, showing that
there must be considerable tension within the egg. This liquid
corresponds to the ‘‘ white” of a hen’s egg, but in the Tuatara
egg, as already observed, it forms only an extremely thin layer
between the shell and the serous envelope.

The serous envelope extends all round inside the shell.
Above the embryo it is still connected with the true amnion
over a small area in the region of the shoulders, and opposite
to this, on the other side of the egg, it becomes continuous
with the yolk-sac over another small area.

The true amnion closely envelops the entire embryo as a
thin transparent investment.

The allantois is very largely developed, and extends almost
completely around the embryo inside the serous envelope. It
is filled with a viscid, semi-gelatinous, transparent, colourless
liquid, closely resembling the white of a hen’s egg. By the
presence of this liquid it is greatly distended, so that its outer
limb is pushed closely against the serous envelope, while its
inner limb is pushed close against the true amnion above and
the yolk-sac below. I found it impracticable to pierce the egg-
shell without piercing the serous envelope and the outer limb
of the allantois at the same time, so that the emission of the
thin liquid “white” is immediately followed by a copious
oozing out of the viscid contents of the allantois. The tension
being thus relieved, it is possible to remove the shell without
further injury to the foetal membranes.

56 ARTHUR DENDY.

The short stalk of the allantois passes out on the right side
of the embryo, and the allantoic vessels at once divide into two
sets. One set passes upwards just in front of the right arm in
that part of the allantois which adheres closely to the true
amnion, while the other passes downwards in that part which
adheres to the yolk-sac. The vessels of the upper set are
reflected back above the embryo into the outer limb of the
allantois, under the serous envelope.

As already stated, the allantois extends almost completely
round inside the serous envelope, but it is of course interrupted
by the attachment of the latter, on the one hand, to the true
amnion above, and on the other to the yolk-sac below.

The yolk-sac is now attached to the ventral surface of the
embryo by a slender stalk about 5 mm. long. The vitelline
circulation embraces nearly the whole of the yolk, but there is
a rounded area, about 6 mm. in diameter, on the side opposite
to the embryo, over which the vitelline circulation has not yet
extended. Although this non-vascular area is well defined,
there is no longer a distinct sinus terminalis. It is in this
region, of course, that the serous envelope is still united with
the yolk-sac.

The radially columnar structure of the yolk, due to the
aggregation of the yolk particles around the absorbent vessels,
as described in the Introduction, is now becoming well marked.

For external characters the embryos of July may be taken as
typical. They present the following strongly marked advances
on the preceding stage:

(1) The head has acquired a markedly Chelonian form,
which may be better realised by reference to figs. 104 and 105
than from any description which I can give.

(2) A very conspicuous, sharp-pointed “shell-breaker” (fig.
104, S..B,) has been developed on the snout by thickening and
cornification of the epidermic cells.

(3) The external nares are represented by a pair of small
round white spots (fig. 104, Na.), having been completely
blocked up by a dense growth of cells which fills the outer
part of each nostril, and which is already present at the com-

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 57

mencement of this stage. This growth forms a darkly stain-
ing mass of tissue, apparently derived from proliferation of the
epiblast lining the nostrils, into which it gradually merges.!
The epiblastic lining of the nostrils is still perfectly sharply
defined on its outer aspect from the surrounding mesoblast.

(4) Owing to development of the eyelids the paired eyes
have acquired their proper elongated appearance.

(5) Pigment is present in the parietal eye (figs. 104, 105,
Pa, E.), rendering it conspicuous externally as a small black
circle surrounding a white spot. In the younger embryos of
this stage, where the outlines of the optic lobes and cerebral
hemispheres are still visible externally, the parietal eye is now
seen to lie above the apparent centre of the diamond-shaped
roof of the thalamencephalon. As far as one can judge, it has
become median in position.

(6) Teeth have appeared in both upper and lower jaws, in-
cluding the palatine teeth, but I have not succeeded in detecting
any vomerine teeth.

(7) The tail is to a large extent uncoiled, and lies with its
apex pointing backwards against the animal’s nght side, as
shown in fig. 104. A distinct corrugation, indicating the
future crest, already appears on its dorsal aspect.

(8) The digits have become greatly elongated, and the limbs
have much the same proportions as in the adult.

(9) Pigment has made its appearance in the general integu-
ment, forming a very conspicuous and remarkable pattern, as
shown in figs. 104 and 105. On the back and sides of the body
and tail this pattern takes the form of narrow, discontinuous,
longitudinal stripes of white on a grey ground, combined with
a less strongly marked development of much broader, transverse
bands of white, especially distinct about the root of the tail.
The limbs also are marked with longitudinal streaks and dashes
of white. In the mid-dorsal line of the body there is a very
characteristic narrow moniliform stripe, composed of a row of
small white spots which seem to indicate the position of the
spines of the future crest. On‘the dorsal aspect of the head

1 Cf. Addendum at the end of this paper.

58 ARTHUR DENDY.

there is very little pigment except for a broad transverse band,
incomplete in the middle, which runs inwards from in front of
each eye. Around the eyes are well-marked radiating bands of
pigment, and on the sides and under surface of the throat and
chin are seen irregular longitudinal bands of grey on a white
ground. Numerous very minute round white dots are scattered
over the upper surface of the head.

In embryo No. 2 the total length measured along the curve
of the back from the tip of the snout to the tip of the tail was
about 55 mm., of which the tail occupied about 22 mm.

The only internal structures which I propose to speak of at
this stage are the parietal eye and its immediate surroundings,
and as these will be fully dealt with in a special memoir they
need not detain us long here.

By the approximation of the cerebral hemispheres and optic
lobes in the straightening out of the cranial flexure the roof of
the thalamencephalon has become thrown into folds, and at
the same time strongly arched upwards.

At the commencement of this stage the parietal eye lies close
over the end of its so-called “ stalk’’ and above the hinder part
of the roof of the thalamencephalon. In July embryos, how-
ever, it has shifted forwards till it is separated from the end of
its ‘ stalk”? by aconsiderable interval, and comes to lie slightly
in front of the middle of the thalamencephalon and above the
paraphysis, which is now represented by a mass of convoluted
tubules, intermingled with blood-vessels, lying between the roof
of the thalamencephalon and the parietal eye.

The ‘stalk ” of the parietal eye has elongated considerably,
and its lumen has become obliterated at its point of attachment
to the brain. Immediately in front of this point the superior
commissure! has developed.

In the parietal eye itself the “lens” has become very sharply
marked off from the rest of the wall, which latter now forms
an optic cup clearly differentiated into two layers. In the
inner layer, next to the cavity of the eye, pigment has been
deposited between the cells, while towards the close of the stage

>

1 Erroneously termed the ‘ posterior ” in my preliminary notes (4, p. 442),

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 59

the nerve of the parietal eye makes its appearance in connection
with the outer layer.

A small quantity of pigment also appears on the floor of the
hollow distal portion of the ‘ stalk.”

Stage S.

To this stage I refer embryos numbered 138, 139, and 140,
which reached me on January 6th, 1898, having unfortunately
died on the voyage, still enclosed in the shell ; and 149 and 150,
which are the ‘ skeletons” referred to by Mr. Henaghan! as
having been found dead on Stephens Island about the middle
of January. All these embryos were evidently very nearly ready
to hatch when they died; in fact, in the case of Nos. 149
and 150 the shell was already ruptured and the embryo par-
tially extricated. The following description is based upon
Nos. 138, 189, and 140, which seem to be a little less
advanced.

The animal now closely resembles the adult, except in point
of size. The total length from the tip of the snout to the tip
of the tail in the specimen measured was 92 mm.; from the
cloacal aperture to the tip of the tail50 mm.; length of head
from the tip of the snout tothe level of the jaw-angles 15 mm. ;
width of head between the angles of the jaw 12 mm.

A yolk-sac about 12 mm. in diameter, densely packed with
yolk, is still attached to the embryo by a narrow stalk a short
way in front of the cloacal aperture, which now forms a large
transverse slit.

Scales and claws are present as in the adult. The nuchal

and caudal crests of spines are also present, the former as yet
very small, the latter well developed; while the dorsal crest is
not yet recognisable.
_ The “ shell-breaker” is still present, but relatively small as
compared with the preceding stage. It appears to be repre-
sented in the adult by the especially large scale at the apex of
the snout.

The palatine, maxillary, and posterior mandibular teeth are

1 Vide Introduction.

60 ARTHUR DENDY.

present as in the adult. In the front of the mandible the two
large cutting teeth of the adult are represented each by three
distinct conical, pointed teeth, a larger one behind and two
smaller ones in front. These three teeth probably fuse to-
gether in later life. Similarly in the premaxille each of the
two large cutting teeth of the adult is represented by three
distinct conical, pointed teeth, of which the outermost is much
larger than the other two. These also probably fuse together
in later life. I could find no vomerine teeth in the specimen
specially investigated as to this particular.

The parietal eye is no longer visible externally, but its
position seems to be indicated by a small median tubercular
scale lying a short way in front of the first nuchal spine.

The colour of the dead animal is dirty white, irregularly
mottled, and banded with grey. The longitudinal striping
has disappeared except under the throat and chin, where it
still remains distinct, but with the white stripes ®? now narrower
than the grey. The transverse banding is still clearly recognis-
able on the back and tail. The whitish or yellowish spots, so
conspicuous in some adult specimens, have not yet appeared ;
they would seem to be formed by gradual encroachment of the
grey pigment over the paler parts of the integument.

4. SUMMARY AND Discussion oF PRiNcIPAL EMBRYOLOGICAL
OBSERVATIONS.

(2) The Formation of the Germinal Layers.

The question of the origin of the germinal layers, and
especially of the hypoblast, is one of great difficulty, and I
have slightly altered my views on the subject since writing my
summary of results (4) some time ago.

The blastoderm spreads around the yolk at an extremely early

1 As was surmised by the late G. Baur (‘ Anat. Anz.,’ Bd. xi, p. 486) ; ef.
also Giinther, ‘ Phil. Trans.,’ 1867, pt. ui, p. 8.

2 These stripes appear to be represented by longitudinal rows of light-
coloured scales on the under surface of the head of the adult animal.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 61

stage, so that already at Stage C, and probably before, the yolk
is completely enclosed. This rapid spreading of the blastoderm
appears also to be characteristic of lizards? and chelonia,? and
perhaps of reptiles generally as compared with birds.

At Stage C, the earliest examined, the blastoderm is some-
what vaguely divided into area pellucida and area opaca, with
the cap-shaped embryo lying in the former (figs. 1, 2, 4).

In the area opaca the blastoderm consists of two distinct
layers, an upper epiblast, consisting of a single well-defined
layer of flattened or columnar polygonal cells (fig. 4, Hp.), and
a lower multiple layer of irregular stellate cells with numerous
yolk particles entangled in their meshes (fig. 4, Z. Z. Y.).
This lower layer evidently corresponds to the lower layer of
the chick, though it seems to be thicker and less sharply
defined from the yolk. A similar layer in Chelonians is
spoken of by Mitsukuri and Ishikawa (5) as ‘ the yolk ” con-
taining nuclei; but in Sphenodon it separates readily from
the underlying yolk, which is distinguished from it by its
want of coherence.

In the area pellucida the epiblast cells gradually become
more columnar in character as they approach the embryo, but
outside the embryo itself they still remain as a single layer
(fig. 4).

The lower layer in the area pellucida becomes divided into
two, which are separated from one another by a large cavity
extending right across beneath the embryo (fig. 4, & C.). This
cavity evidently corresponds to the similar cavity in the chick,
termed by Foster and Balfour (9) the segmentation cavity,
and by Marshall (7) the subgerminal cavity. According to
Weldon (6) it appears to be doubtful whether or not such a
cavity occurs in Lacerta.

The floor of the segmentation cavity in Sphenodon is
formed by a thin membrane which I have termed the sub-
embryonal membrane, and which is represented by the dotted
line in fig. 4, In the earlier stages of development it is

1 Compare Balfour (8).
* Compare Mitsukuri and Ishikawa (5).

62 ARTHUR DENDY.

conspicuous beneath the embryo after the blastoderm has been
removed from the yolk (figs. 2, 16, 19, S. #. M.), but it takes
no part in the development of the embryo itself. As far as I
can judge from the scanty literature at my disposal, it corre-
sponds to the secondary endoderm or yolk layer of Will (13) in
the gecko and other vertebrates.

Above the segmentation cavity the lower layer outside the
embryo is a multiple layer of irregular cells with little or no
yolk in their meshes (fig. 4). At Stage C it appears to pass
gradually into the lower layer of the area opaca, but in some
sections of later stages the transition appears very sudden,
there being a distinct germinal wall at the edge of the area
opaca (fig. 14).

In the embryo itself the epiblast is much thickened, and its
cells are arranged in a multiple layer. It is especially thick
in the region of the head-fold (fig. 4, H. F.), where the
epiblastic cells appear to be rapidly proliferating. Further
back it is thinner, and its cells distinctly prismatic, forming a
medullary plate (fig. 4, JZ. P.), which passes behind into the
very thick, undifferentiated cellular mass of the primitive
streak. Beneath the epiblast of the embryo there is a thick
layer of small rounded cells which also passes into the primi-
tive streak, from which, indeed, it cannot be _ histologically
distinguished, and from which it appears to have been derived
by forward growth and proliferation (fig. 4). In the region of
the head-fold this layer passes into the lower layer of the area
pellucida, but whether the cells of the latter actually extend
into the embryo at this stage (C) is very doubtful.

Posteriorly the cells of the primitive streak pass into the
prismatic epiblast of the area pellucida above, while below
this they are continued backwards almost to the edge of the
area opaca as a thick compact layer of small rounded cells, in
no way distinguishable from the primitive streak itself. This
layer in turn becomes continuous with the original lower layer
of the blastoderm behind the embryo.

It appears to me that we must regard the lower layer of the
embryo itself at this stage (C) as the mesoblast derived from

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 63

the primitive streak. It agrees closely in histological character
with the primitive-streak mesoblast of the chick as described
by Marshall (7), but differs in its much greater forward
extension, and in the much larger share which it takes in the
formation of the embryo. In this connection it is interesting
to note that, according to Foster and Balfour (9), Kolliker
holds that the mesoblast of the region of the embryo is derived
from a forward growth of the primitive streak.

In Lacerta muralis Balfour states (8) that a layer of
mesoblast spreads out in all directions from the primitive
streak, and is stated by Kupffer and Benecke to be continuous
across the middle line in the region of the embryo. This
latter statement Balfour regards as highly improbable, but it
certainly agrees with what I have observed in Sphenodon
(compare fig. 11).

“The chorda-entoblast ” of Hertwig and Mitsukuri and Ishi-
kawa in Trionyx (5), the “axial strip of invaginated hypo-
blast” of Weldon in Lacerta (6), and the “ head-process” of
Haswell in the emu (10), all seem to correspond to some
extent to the layer under discussion.

Posteriorly the primitive streak passes insensibly into a
transverse thickening formed of precisely similar cells, and re-
presenting the so-called “sickle,” as described by Mitsukuri
and Ishikawa (5) in Trionyx. From the sides of the primi-
tive streak, or one might say from the ends of the transverse
“sickle,” two lateral wings of similar cells (mesoblastic) grow
forwards and slightly outwards (at Stage E), outside the
embryo proper, and give rise to the vitelline vessels in the
vascular area (fig. 20, L. W.).

In the region of the primitive streak the dorsal surface of the
embryo sinks to the general level of the blastoderm, and here
the blastopore is established at Stage C by invagination, some
little way behind the medullary plate (fig. 4, B. P.). This
invagination, in the form of a pit lined by columnar cells,
gradually extends downwards, and at Stage F comes to open
by a single well-defined aperture into the space (the so-called
segmentation cavity) beneath the embryo, which will presently

64, ARTHUR DENDY.

be enclosed, at any rate in part, in the alimentary canal (cf.
fig. 35).

A primitive groove may be recognised both in front of and
behind the blastopore (figs. 11, 34, P. G.).

The hypoblastic lining of the alimentary canal is probably
formed in Sphenodon from flattened cells of the original
lower layer of the blastoderm. ‘These cells, at first absent, at
any rate in the middle line of the embryo (fig. 4), appear to
gradually spread inwards from each side and meet beneath the
notochord (figs. 8—14). The layer thus formed corresponds
to the “ Darm-Entoblast”’ of Hertwig and Mitsukuri and
Ishikawa (5). The latter authors state that in Chelonia
(Trionyx) also it passes gradually under the so-called “chorda-
entoblast,”’ and fuses in the middle line to complete the upper
wall of the digestive cavity.

Mitsukuri and Ishikawa (5) make much of the fact that in
Chelonia the mesoblast in front of the blastopore arises as a
paired mass from the junction of the “ chorda-entoblast ” with
the “ Darm-Entoblast ” on each side. I have seen nothing of
such a mode of origin in Sphenodon, where, as already noted,
it seems to spread forward from the primitive streak as a broad
and perfectly continuous sheet (vide fig. 11), the axial por-
tion of which presently separates as the notochord from the
lateral portions (vide fig. 33).

As in other cases, however, the mesoblast in Sphenodon
certainly has two distinct origins; (1) from outgrowth of the
compact rounded cells of the primitive streak, and (2) from
loose stellate cells of the original lower layer left between
epiblast and hypoblast after the formation of the latter. The
first mode of origin is especially conspicuous in the posterior
part of the body, but the exact share taken by each in the
subsequent development of the embryo I have not determined.

The hypoblastic lining of the digestive cavity, at first con-
sisting of flattened cells, gradually takes on a columnar cha-
racter. It is noteworthy that this change commences at the
anterior end of the body,—in fact, in front of the embryo alto-
gether (fig. 8), and gradually extends backwards (compare

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 65

figs. 22—82 and fig. 35). It also takes place later in the
middle line, beneath the notochord, than it does laterally.

(6) The Formation of the Fetal Membranes.

The formation of the foetal membranes in the Tuatara presents
several very peculiar features, and appears to me to be by no
means of a primitive nature; at the same time it presents no
features which are not met with in a more or less highly
developed state in other Vertebrate types.

The Pro-amnion.—At Stage C, it will be remembered,
the blastoderm of the area pellucida, forming the roof of the
large segmentation cavity (or subgerminal cavity) some little
way in front of the embryo, is thin and almost free from yolk
particles (fig. 4, Pro. Am.). This part of the blastoderm is evi-
dently to be regarded as the rudiment of the pro-amnion, but
we can hardly say that it consists, as in the chick, of epiblast
and hypoblast alone, with no mesoblast, for the lower-layer
cells still form a multiple layer, the lowest of which are only
just beginning to be doubtfully distinguishable as hypoblastic,
while the cells between these and the epiblast might certainly
be regarded as mesoblastic, though perhaps it is best to con-
sider the lower layer as being still undifferentiated into meso-
blast and hypoblast at all. We may therefore say that the
pro-amnion consists at this stage of epiblast and lower layer,
the latter being several cells thick.

At Stage D this condition is still maintained, but behind
the pro-amnion, though still in front of the embryo, the hypo-
blast has become clearly differentiated, and a large celomic
space has appeared in the mesoblast (fig. 8); this, however,
does not concern us now.

Between Stages D and E there is somewhat of a gap in the
series as regards the formation of the pro-amnion, but it is
evident that the head end of the embryo sinks down into the
yolk, carrying the pro-amnion with it. It thus comes to pro-
ject freely beneath the blastoderm, enveloped in a very thin
transparent membrane, the pro-amnion (fig. 22). This mem-

vo. 42, parr 1.—NEW SER. E

66 ARTHUR DENDY.

branous investment of the head undoubtedly consists really of
two layers, epibiast and hypoblast, which have become so
stretched and flattened that they can be distinguished from
one another in parts only, especially behind (figs. 24—27, 64),
while there is no trace of mesoblast, at any rate in the more
anterior portion of the pro-amnion. The pro-amnion is,
perhaps, best shown in the diagrammatic longitudinal sections
of somewhat later stages represented in figs. 35 and 50.

Thus the pro-amnion in Sphenodon forms a far more con-
spicuous feature than it does in the chick, owing to the much
greater extent to which the front end of the embryo sinks down
into the yolk. It is much more nearly approached by the
condition of the rabbit, as described and figured by Marshall (7),
but even here the pro-amnion does not appear to form nearly
so prominent a feature as it does in the Tuatara.

According to Mitsukuri (14), as I learn from an abstract in
the ‘ Journal of the Royal Microscopical Society,’ in Chelonia
also the head-fold sinks in the yolk below, but as I have not
been able to see the original memoir I do not know how far
this process agrees with what takes place in Sphenodon.

In Sphenodon it is only at a comparatively late stage in
the development (Stage O) that the pro-amnion ceases to exist.
At this stage the front end of the embryo, enveloped in the
true amnion (inner part of the pro-amnion), withdraws from
the pocket in the yolk-sac formed by the outer part of the pro-
amnion, and thus comes to lie entirely above the yolk-sac.

The Amnion.—In the anterior part of the body the true
amnion is formed, as we have just seen, by separation and
withdrawal of the somatopleuric portion of the pro-amnion from
the splanchnopleuric portion. From about the region of the
shoulders backwards it is formed by the uprising of two folds of
somatopleure alone, accompanied by a downsinking of the body
of the embryo. These two somatopleuric folds (figs. 30 and 55,
Am.) meet and coalesce in the mid-dorsal line above theembryo
in a perfectly normal manner. This process of amnion forma-
tion, however, does not cease at the hinder extremity of the
embryo, but is continued backwards for some distance behind

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 67

the primitive streak so as to give rise to a long narrow canal or
tunnel, the posterior amniotic canal, which for some time opens
on the surface of the blastoderm behind, and thus maintains
a free communication between the amniotic cavity above the
embryo and the very small space beneath the shell containing
the “ white” of the egg (figs. 35, 37, 39, 40, 41, 45, 71—73,
P.A.C.). The posterior amniotic canal is formed partly by
invagination of a specially modified linear strip of epiblast,
and partly by uprising of two somatopleuric folds which meet
and unite above it (fig. 46). It makes its appearance at
Stage F, and disappears by obliteration of its lumen at about
Stage K. After the serous envelope has split off from the
underlying yolk-sac it lies entirely in the thickness of the
former, embedded in the mesoblast (figs. 71—73).

So far as I am aware, the only other type in which anything
comparable to the posterior amniotic canal of the Tuatara has
been found is the Chelonian type, in which, according to
Mitsukuri, the amniotic folds continue to grow backwards, and
thus produce a tube which extends back from the posterior end
of the embryo and connects the amniotic sac with the exterior.
Although my knowledge of Mitsukuri’s observations is derived
solely from the abstract in the ‘Journal of the Royal Micro-
scopical Society’ (15), I suppose there can be little doubt as to
their close agreement with what I have myself observed in the
Tuatara, and we thus have a striking embryological confirma-
tion of the view so strongly insisted upon by Boulenger (16)
as to the close relationship of Sphenodon with the Chelonians,
a view which has hitherto been based entirely upon the
structure of the adult animal.

As to the functions of the posterior amniotic canal I have no
suggestions to make, but I understand that Mehnert has pub-
lished a paper (18) on the subject, to which, unfortunately, I
have been unable to obtain access.

The Serous Envelope.—The serous envelope is formed
only to a very slight degree by the outer limb of the uprising
amnion folds. Throughout nearly its whole extent it is formed
by splitting off of the superficial epiblast, accompanied by a

68 ARTHUR DENDY.

thin layer of mesoblast derived directly from the original lower
layer of the blastoderm, from the underlying yolk-sac (cf. figs.
29—82, 47, 50, 55, 68—74). It is doubtful if the sepa-
ration of the serous envelope from the true amnion is ever
quite complete, for at the latest stage at which it was possible
to examine the foetal membranes (Stage R) the two are still
connected above the shoulders over a small area. A similar
connection has been described by Hirota (17) in the case of
the chick; while in Chelonians, according to Mitsukuri (15),
this ‘‘ sero-amniotic connection ” remains much more extensive,
and separates the “‘ extra-embryonic cavities of the two halves
of the amnion” over the dorsal region of the embryo to the
end of development.

The Yolk-sae.—The yolk-sac is formed primarily from
what is left of the original lower layer of the blastoderm after
the serous envelope has split off from it. It appears, however,
that a large portion of the mesoblast derived from the primitive
streak spreads into or over it, and gives rise to, at any rate, the
commencement of the vitelline vessels (cf. figs. 15—17, 20,
30, 31, L. W.). The vitelline circulation (figs. 58, 93, 100,
101) closely resembles that of the chick, though perhaps the
absorbent vessels which dip down into the yolk are more
strongly developed, and the arrangement of the yolk-spheres
around them, like onions on a string, is certainly very striking
(fig. 106).

The splanchnopleuric portion of the pro-amnion must also
doubtless be regarded as forming part of the yolk-sac, but this
has been sufficiently described already.

The Allantois.—The allantois originates at Stage L in a
perfectly normal manner as an outgrowth of the ventral wall
of the alimentary canal close to the posterior end (figs. 883—
89). As development goes on it extends upwards on the right
side of the embryo, and comes to lie beneath the serous en-
velope (fig. 100). its walls become vascular, and it is very
greatly distended by the accumulation within it of a clear,
semi-gelatinous liquid. Thus the outer wall of the allantois
becomes closely pressed against the serous envelope, and the

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 69

inner wall against the amuion above and the yolk-sac below,
the two being separated by a wide cavity containing the semi-
gelatinous liquid above mentioned. Before the close of de-
velopment the allantois spreads almost completely around both
embryo and yolk-sac, being interrupted at Stage R only by the
sero-amniotic connection above the embryo, and by a small
area at the opposite side of the egg where the serous envelope
has not yet split off from the yolk-sac.

Marshall (7) describes the amniotic cavity of the chick as
forming a water-bath in which the embryo can move freely in
any direction. In the Tuatara this description is far more
applicable to the allantoic cavity, which vastly exceeds that of
the amnion in extent.

(ec) The Modelling of the Body and the Foundation
of the Principal Systems.

The Modelling of the Body.—The body of the embryo
is first recognisable (Stage C) as a cap-shaped structure (figs.
1, 3,5), the broad anterior end of which is raised up by the
head-fold, while the narrow posterior extremity, formed by the
primitive streak, lies at the general level of the area pellucida.
The general form of the embryo at this stage thus closely
resembles that of Trionyx, as figured by Mitsukuri and
Ishikawa (5). The anterior end of the embryo then elongates
and becomes at the same time narrowed (Stage D, fig. 6), and
as it sinks into the yolk, enveloped in the pro-amnion, it comes
to project freely beneath the overlying area pellucida (Stage
K, figs. 15—20). It remains in this position until Stage O,
when the front end of the embryo is withdrawn from the yolk-
sac, and comes to lie on its left side above it,

A distinct tail-fold is not formed until much later than the
head-fold, but at Stage J (and perhaps even at Stage H) the
whole body of the embryo is clearly outlined by the head-,
tail-, and side-folds (figs. 56—58), and by the time Stage K
is reached the tail has begun to grow freely backwards between
the yolk-sac and the serous envelope (fig. 82). As the tail
elongates it becomes coiled inwards in a spiral between the

70 ARTHUR DENDY.

hind limbs (ef. fig. 98,), but at Stage R this spiral has become
to a great extent straightened out, and the hinder part of the
tail lies against the animal’s right side (fig. 104).

The cranial flexure takes place in the usual way, commenc-
ing at Stage J, when the front part of the head, including the
fore-brain, is seen to be bent down at right angles (fig. 57).
Later on the whole of the front half of the body becomes
curved ventrally in a spiral, best seen about Stages M and N
(figs. 92, 94), while in the lumbar region there is a slight
curvature in the opposite direction (figs. 83, 84, 92, 94). By
the time Stage Ris reached the curvatures of the body have
to a large extent disappeared, and the cranial flexure has
straightened out, the head acquiring a marked Chelonian
aspect, with a conspicuous “ shell-breaker ” on the snout (fig.
104). A quite temporary flexure of the head in an upward
and backward direction is conspicuous in some of the earlier
stages of development, especially Stage G (figs. 48—50) ; it is
caused apparently by the restraining influence of the pro-
amnion, and very soon disappears.

The limbs make their appearance as outgrowths of the
Wolffian ridges at Stage M (fig. 92), and develop in a perfectly
normal manner, five digits appearing on each at Stage Q (figs.
102, 103).

The development of the visceral arches, superior maxillary,
and fronto-nasal processes appears to be perfectly normal
(cf. figs. 92, 94, 99).

The Central Nervous System.—The development of the
central nervous system, so far as investigated, takes place quite
in the ordinary manner, except perhaps for the temporary
modifications in the form of the brain due to the restraining
influence of the pro-amnion.

Already at Stage C the medullary plate (fig. 9, MZ. P.) is
recognisable as a flattened area in front of the blastopore,
where the epiblast cells are prismatic and arranged in two or
three layers. At Stage D (figs. 6, 7, 9, 10) the medullary
eroove appears as an invagination of the cells of the medullary
plate in the mid-dorsal line, At Stage E the front end of the

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 71

medullary groove is widened out and constricted into fore-,
mid-, and hind brains (figs. 20, 21), the hind brain already
exhibiting traces of further segmentation. At Stage F the
medullary groove begins to close in the region of the mid-
brain (figs. 35, 42). At Stage G the closure has extended
backwards and forwards, leaving the medullary groove open
above only in two places, viz.—posteriorly, above and in front
of the now conspicuous neurenteric aperture ; and anteriorly,
in the region of the fore-brain (fig. 50). The mid and hind
brains are now bent somewhat into the form of an § by the
backward and upward tilting of the head due to the restraining
influence of the pro-amnion (fig. 50), and, owing doubtless to
the same cause, the two halves of the fore-brain are beginning
to overlap one another in the middle line (fig. 48). At Stage
H the mid- and hind brains are straightened out again, but
the overlapping of the left half of the roof of the fore-brain is
more conspicuous, while the optic vesicles commence to grow
out on either side (fig. 52). At Stage G the medullary canal
is completely closed in above, though the left half of the fore-
brain still overlaps the right (fig. 60), and the margins of the
two do not actually unite until the next stage (fig. 77). At
Stage J, also, the fore-brain bends down at right angles, so
that the mid-brain comes to occupy the anterior end of the
body (figs. 58, 59), and the segmentation of the hind brain is
much more strongly marked (fig. 58). The roof of the hind
brain also is thinning out in the usual manner (fig. 63). At
Stage K the primary parietal vesicle is formed as an outgrowth
of the roof of the fore-brain a little to the left of the middle
line (fig. 76). At Stage L, by the closure of the neurenteric
canal, the central nervous system becomes completely shut in;
the cerebral hemispheres begin to grow out from the fore-
brain, and the infundibulum and pituitary invagination almost
meet one another beneath the thalamencephalon, while the
hind brain has become segmented into about six neuromeres
(fig. 90). At Stage M the mid-brain has become extremely
prominent, and the cerebellum begins to be conspicuous just
behind it (fig. 92). At Stage N the paraphysis appears as a

72 ARTHUR DENDY.

simple, backward- pointing diverticulum of the roof of the fore-
brain some considerable distance in front of the parietal eye.
At Stage O the cerebral hemispheres and optic lobes are both
prominent externally, with the thin roof of the thalamence-
phalon stretched out between them. Just in front of the optic
lobes lie the parietal eye and “stalk,” and just behind the
cerebral hemispheres lies the pineal gland, with its wall just
beginning to be folded. At Stage R the cranial flexure has
become straightened out, and the cerebral hemispheres and
mid-brain much more closely approximated to one another, so
that the roof of the thalamencephalon has become shortened
and folded, giving rise to the choroid plexus; while the para-
physis and parietal eye, which develop quite independently of
one another and are never really connected, are brought close
together. The paraphysis has now given rise to a convoluted
mass of tubules, intermingled with blood-vessels, which comes
to lie beneath the parietal eye and in front of its so-called
“stalk.” Meanwhile the superior commissure has made its
appearance just in front of the point where the stalk of the
parietal eye joins the brain. ‘The details of the development
of these organs, however, are beyond the scope of the present
paper, and will be dealt with in a later memoir.

The Ordinary Paired Eyes.—The development of the
paired eyes, so far as investigated, is shown in fig. 77, from
which it will be seen that it takes place in the normal verte-
brate fashion. The optical vesicle is invaginated to form an
optic cup, and the lens is formed as a thickening of the super-
ficial epiblast opposite to the mouth of the cup.

The Parietal Eye.—The development of this organ will
form the principal subject of the special memoir already referred
to, but it may be worth while to summarise the principal facts
in this place. It is formed from the primary parietal vesicle,
the origin of which as an outgrowth from the roof of the fore-
brain slightly on the left of the middle line has been noted.
The front part of the wall of this vesicle thickens to form the
lens, while the deeper part gives rise to the retina. ‘The latter
forms a deep optic cup which holds the leus in its mouth, the

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 73

two being sharply distinguished from one another though re-
maining in close contact. The wall of the optic cup divides
into two primary layers, in the inner of which pigment is de-
posited, while the outer becomes connected with the brain by
the special nerve of the parietal eye, which is not formed from
the so-called stalk. The so-called ‘‘ stalk ” of the parietal eye
probably represents the right parietal eye retarded in develop-
ment by the overlapping of the left half of the roof of the
fore-brain, and in an extremely degenerate condition.

The Ears.—The development of the auditory organs, so far
as followed, is quite normal. ‘lhe auditory pits appear about
Stage J as shallow depressions of the superficial epiblast on
either side of the hind brain, the epiblast being here com-
posed of a single layer of elongated columnar cells (figs. 58,
64). These pits gradually deepen (fig. 79), and about Stage L
they entirely close up, forming two sacs, each still lined by a
single layer of columnar cells (fig. 91, Aw.). Further than
this their development has not been followed.

The Olfactory Organs.—The nasal pits first appear as
shallow depressions of the superficial epiblast at about Stage
L (fig. 83, Na.). They have the usual crescentic outline, open
towards the mouth (figs. 92, 94, Na.). As development goes
on they gradually deepen, and with the downgrowth of the
fronto-nasal process, commencing about Stage O (fig. 99), the
external nares become defined in the usual manner as two cres-
centic apertures the margins of which are completed at Stage Q
(fig. 103). At Stage R a very remarkable feature makes its
appearance, the nostrils being completely plugged up by a
dense cellular mass derived from proliferation of their epiblastic
lining. The external nares now appear from the outside as
two small tound white spots (Fig. 104, Na.), being filled up to
the level of the surrounding epidermis with the plug of cells
just mentioned. This remarkable plugging up of the nostrils
takes place about the commencement of the long winter rest
which the embryo passes through at Stage R, and probably
bears some relation to this process of hibernation. I have no
material sufficiently well preserved to enable me to state when

74 ARTHUR DENDY.

the plug of cells is again removed, but this probably takes
place at Stage S, shortly before hatching.

We may, perhaps, compare the plugging up of the nostrils in
the Tuatara with the remarkable solidification of the esophagus
which takes place at a certain stage in the development of the
tadpole and other Vertebrates (Marshall [7] ).

The Alimentary Canal.—The development of the ali-
mentary canal takes place in the usual way by gradual in-
folding of the splanchnopleure. The front part is enclosed
first (figs. 9, 10, 23, 24, 35), and for some time remains
filled with yolk particles (figs. 52, 59). The stomodzum is
established in the ordinary way by a shallow invagination of the
superficial epiblast at Stages J and K (figs. 57, 63, 78). The
proctodzal invagination is formed considerably later, com-
mencing at about Stage M.

Of the outgrowths to which the alimentary canal gives rise
I have investigated only the allantois (which has been already
dealt with), the visceral clefts, and the commencement of the
pituitary body and of the thyroid gland.

The visceral clefts are four in number, and appear to develop
very much in the usual way as outgrowths of the hypoblastic
lining of the front part of the alimentary canal (cf. figs. 63,
78, 91). The first becomes visible externally about Stage K
(figs. 74,75). At Stage L three are thus visible (figs. 83, 84).
At Stage N all four can be seen from the outside (fig. 94), and
at Stage O they begin to close up again (figs. 96, 98), though
the hyomandibular remains visible externally till Stage P
(fig. 101).

The pituitary body was first cbserved at Stage L as a
thickening of the lining epithelium (presumably derived from
the stomodzum) beneath the anterior extremity of the noto-
chord (fig. 90). At Stage N it forms a deep pit with a very
thick wall composed of columnar cells (fig. 95).

The thyroid gland is visible at Stage K as a median, ventral

1 Cf. de Menron, ‘ Compt. Rend.,’ tom. cil, p. 1401 ; and ‘ Rec. Zool. Suisse,’
tom. ill.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 75

groove in the floor of the alimentary canal (fig. 79, Th.) just
behind the mouth, lined by columnar cells.

The Notochord.—The notochord appears at Stage H,
being formed by the separation of an axial rod of cells from the
thick sheet of mesoblast which grows forwards from the primi-
tive streak (figs. 26—33, 35, Not.). Thus the notochord in
Sphenodon appears to be undoubtedly of mesoblastic origin,
an origin which, it will be remembered, is also claimed for
it by some writers (e.g. Duval [11] ) in the chick. It extends
forwards to within a short distance of the infundibulum, and
in the later stages of its development its anterior end becomes
curiously twisted in an irregular spiral (fig. 95, Stage N), which
is recognisable even after the cartilage has developed around it
(Stage R).

The Celomic Cavities.—The first indication of coelomic
cavities appears extremely early, at Stage D, in the form of a
wide space in the mesoblast of the area pellucida in front of
and entirely outside the embryo itself (fig. 8, P.C.). This
space appears to be the rudiment of the pericardial cavity
which is so conspicuous in later stages, but which is not for a
long time shut off from the general body-cavity of the adult.
As the head-fold deepens the cceelomic space in question is
carried back, and comes to lie just in front of the opening into
the anterior enclosed part of the alimentary canal (fig. 50,
P.C.). It is possible that it does not always appear so early
as Stage D, for in embryo 61 (Stage F, fig. 35) I observed no
trace of it, and at Stage G (fig. 50) it appears as a space
between epiblast and hypoblast, with no mesoblast around it.
Possibly it normally appears very early, and is then more or
less completely obliterated, perhaps by stretching in the forma-
tion of the pro-amnion, to open out again later on. Already
at Stage H, however, it forms a conspicuous quadrangular sac
containing the heart, lying beneath the anterior enclosed part
of the alimentary canal (fig. 52).

Sections of embryos of Stage J, in which the pericardial
cavity is even more conspicuous (fig. 59), clearly show that it
is in free communication with the general body-cavity above,

76 ARTHUR DENDY.

on either side of the alimentary canal (fig. 66). The coelom
within the embryo develops in the ordinary manner as a split
between the somatopleuric and splanchnopleuric layers of
mesoblast (cf. figs. 66—69) ; this is directly continuous with
the large pleuro-peritoneal space outside the embryo, which
develops chiefly as a split between the serous envelope and
the yolk-sac (cf. figs. 68—72).

The head cavities appear at Stage J as two small spaces in
the mesoblast just outside the aortic dilatations on either side
of the anterior extremity of the alimentary canal (fig. 62).
At Stage K they are visible from the outside as dark patches
behind the paired eyes (figs. 74, 75), and each has begun to
divide into anterior and posterior portions (fig. 77). At Stage
L they have become very large (fig. 91), and those of the an-
terior pair are connected across the middle line by a short trans-
verse canal (fig. 90, H. C. C.), while each of the posterior pair
gives off a conspicuous branch into the mandibular arch (fig. 91,
H.C.M.). The head cavities are lined by short columnar cells.

The Mesoblastic Somites.—The vertebral and lateral
plates of mesoblast appear to be derived, at any rate mainly,
from the great sheet of mesoblast which grows forwards
from the primitive streak (compare Stage D, figs. 11 and
12; Stage E, figs. 29—34; Stage J, figs. 68—71). Trans-
verse sections of embryos of Stage E already, perhaps, show
indications of the division of the mesoblast into vertebral and
lateral plates (figs. 29—381 and 38), but it is not until Stage H
that the vertebral plate begins to be segmented into meso-
blastic somites or protovertebre. ‘The mesoblastic somites do
not appear with the same regularity as in the chick, so that
they are of little use for the purpose of classifying the embryos
in stages. Generally speaking, they may be said to develop
from before backwards, though a number of them seem to
appear almost or quite simultaneously at first. The coelomic
split between somatopleure and splanchnopleure at first ex-
tends into them (fig. 69), but they soon separate completely
from the lateral plates, forming squarish blocks composed of
radially elongated columnar cells (fig. 68).

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 177

The Wolffian Ducts and Tubules.—These organs first
appear at Stage J. Their development has not been followed
in any detail, but appears to take place very much as described
by Weldon (6) in Lacerta (vide figs. 68, 69, 80).

The Vascular System.—The heart was first recognised
at Stage H (fig. 52) as a somewhat pear-shaped sac, lying in
the pericardium, receiving the two vitelline veins behind, and
giving off the short bulbus arteriosus in front. It probably
originates as a split in the splanchnopleuric mesoblast. At
Stage J, when it was first observed in sections, it looks like a
ventral diverticulum of the splanchnopleuric mesoblast, here
composed of short columnar cells, containing two thin-walled
epithelioid tubules (fig. 66), continuous with the vitelline veins
behind, and uniting in front to form the bulbus arteriosus,
while the whole heart has begun to be bent into the form of
an § (fig. 59). A full description of the circulation at Stage J
has already been given, and need not be repeated ; it shows no
features of special interest, except perhaps the enormous dila-
tation of the primitive aorte on either side of the anterior
extremity of the alimentary canal (fig. 62). Beyond Stage J
the development of the vascular system within the body has
not been worked out. The development of the vitelline
vessels has already been referred to in dealing with the yolk-
sac.

The Teeth—The only feature of special interest observed
in connection with the teeth concerns the development of the
two large cutting teeth which are so conspicuous in the front
part of each jaw of the adult. At Stage S, shortly before
hatching, each of these is represented by three distinct, pointed,
conical teeth. No vomerine teeth were observed.

(qd) The Embryonic Colour Markings.

One of the most remarkable features of the development is
the appearance at Stage R of a well-defined embryonic pattern
on the integument totally different from that of the adult.
This pattern consists mainly of two distinct series of markings :
(1) a series of narrow, discontinuous, longitudinal stripes of

78 ARTHUR DENDY.

white on a grey ground ; and (2) a series of less well-defined,
much broader transverse bands of white (figs. 104, 105). At
Stage S, shortly before hatching, the longitudinal striping has
disappeared except under the throat and chin, but the trans-
verse banding is still clearly recognisable on the back and tail.
The small whitish or yellowish spots, so characteristic of some
adults, have not yet appeared, but the whole body is dirty
white, irregularly mottled, and banded with grey. Thus the
order in which the markings appear seems to be (1) longitudinal
stripes, (2) transverse bands, and (3) spots. ‘The stage at
which longitudinal stripes are present without the transverse
bands was not actually observed, but as most of the longitudinal
stripes disappear before the transverse ones, and are much
better defined than the latter at Stage R, we may assume that
they also appear before them. These observations are to a
large extent in agreement with the conclusions of Eimer (19)
and others as to the colour-markings of animals in general,
and especially of mammals ; but in the latter group, according
to Eimer, the spots arise before the cross-stripes. Eimer also
observes that the old features linger longest on the fore-parts,
as is also the case with the longitudinal striping of the Tuatara,
indications of which remain visible on the under surface of the
head even in the adult animal sometimes, if not always.

ADDENDUM.

Since this manuscript was forwarded to England I have
found that a precisely similar plugging-up of the nostrils has
been described by Parker (20, pp. 61, 64, 65, and 111) in
the embryo of Apteryx. ‘This fact seems to show that the
plugging has no connection with the hibernation of the embryo.
It is a most singular coincidence that this strange condition
should have been observed in two animals so widely separated
zoologically, and yet both occurring in New Zealand.

19.

20.

OUTLINES OF THE DEVELOPMENYT OF THE TUATARA. (8)

5. List or REFERENCES.

. Tuomas, A. P. W.—“ Preliminary Note on the Development of the Tuatara

(Sphenodon punctatum),” ‘Proceedings of the Royal Society,’
vol. xlviii. (Reprinted in the ‘New Zealand Journal of Science,’
vol. i, January, 1891.)

. Denvy, A.—“ Description of Peripatus oviparus,” ‘ Proceedings of

the Linnzan Society of New South Wales,’ series 2, vol. x.

. Denpy, A.—‘‘ The Hatching of a Peripatus Egg” (and other papers

on the same subject) in the ‘Proceedings of the Royal Society of
Victoria’ for 1893, et seq.

. Denpy, A.—“ Summary of the Principal Results obtained in the Study of

the Development of the Tuatara,’’ ‘ Proc. Roy. Soc.,’ vol. Ixiii, p. 440.

. Mitsuxuri, K., and IsHtkawa, C.—‘ On the Formation of the Germinal

Layers in Chelonia,” ‘Quart. Journ. Micr. Sci.,’ vol. xxvii, 1886.

. We.pon, W. F.—“ Note on the Harly Development of Lacerta mura-

lis,” ‘Quart. Journ. Mier. Sci.,’ vol. xxiii, 1883.

. Marsuatt, A. M.—‘ Vertebrate Embryology.’

. Batrour, F. M.—‘ Comparative Embryology.’

. Foster and Batrour.—‘ Elements of Embryology’ (2nd edition).

. Haswett, W. A.—“‘ Observations on the Harly Stages in the Develop-

ment of the Emu,” ‘Proceedings of the Linnean Society of New
South Wales,’ series 2, vol. ii.

. Duvat, M.—‘ Atlas d’Embryologie.’
. SpenceR, W. B.—‘‘ On the Presence and Structure of the Pineal Hye in

Lacertilia,”’ ‘Quart. Journ. Micr. Sci.,’ vol. xxvii, 1887.

. Witt, L.—‘ Development of Reptiles”? (Abstract), ‘Journal of the

Royal Microscopical Society,’ April, 1896.

. Mirsuxuri, K.—‘“‘ Foetal Membranes in Chelonia”’ (Abstract), ‘ Journal

of the Royal Microscopical Society,’ August, 1891.

. Mivsuxunri, K.—‘“‘ Foetal Membranes of Chelonia” (Abstract), ‘ Journal

of the Royal Microscopical Society,’ February, 1891.

. Boutencer, G, A.—‘ Catalogue of the Chelonians, Rhynchocephalians

and Crocodiles in the British Museum,’ 1889.

. Hirota, 8.—‘“‘Sero-amniotic Connection and Foetal Membranes of Chick ”

(Abstract), ‘Journal of the Royal Microscopical Society,’ Dec., 1894.

. Mennert, E.—“ Ueber Entwickelung, Bau, und Function des Amnion

und Amnionganges nach Untersuchungen an Hmys lutaria-tauri ca”
(Marsilii), ‘ Morphol. Arbeit.,’ iv, pp. 207—274. (Quoted in ‘ Zoologi-
cal Record,’ vol. xxxii, 1895.)

Eimer, G. H. T.—“ Markings of Mammals” (Abstract), ‘Journal of the
Royal Microscopical Society,’ February, 1889.

PaRkER, T. J.—‘‘Observations on the Anatomy and Development. of
Apteryx,” ‘ Phil. Trans.,’ B., vol. clxxxii, 1891.

80 ARTHUR DENDY.

6. DESCRIPTION OF PLATES 1—10.

Illustrating Mr. Arthur Dendy’s paper on “ Outlines of the
Development of the Tuatara, Sphenodon (Hatteria)
punctatus.”

List of Reference Letters.

Al. @. Alimentary canal. Ad?. Allantois. dm. Amnion. Am. C. Amniotic
cavity. dm.O. Amniotic opening. Am. 8. Slit along which amnion is not
yet closed. 4.0. Area opaca. 4.P. Area pellucida. Av. Auditory pit:
(and vesicle). B. 4dr. Bulbus arteriosus. B. J. Bloodisland. &. P. Blasto-
pore. B.V. Blood-vessel. Cb. Cerebellum. C. H. Cerebral hemispheres.
D. A. Dorsal aorta. Hl. Elbow. Zp. Epiblast. Zp. 7. Epithelioid tubule
in heart. Hye. Ordinary paired eye. F.B. Fore-brain. #. Z. Fore-limb.
F. Sp. Fold where the hypoblast turns round to run forwards, marking the
posterior limit of the anterior enclosure of the alimentary canal. G. WV. Ger-
minal wall. #.B. Hind brain. H.C. Head cavity. H.C. C. Connection
between the two anterior head cavities. H.C. M. Branch of head cavity in
mandibular arch. H. 7. Head-fold. H.Z. Hindlimb. H.W and A. W.C.
Hyomandibular cleft. H¢. Heart. Hyp. Hypoblast. J. C. M. Intermediate
cell mass. Jnf. Infundibulum. JZ. 4. S. Line along which amnion and serous
envelope are united. Ze. Lens of paired eye. Z. Ff. B. Left half of roof of
fore-brain overlapping right half. Ziv. Liver. JZ. Jw. Line of junction of the
two uprising amniotic folds above the medullary groove. J. Z. Lower-layer
cells of blastoderm. JZ. Z.1/. Mesoblast derived directly from lower-layer
cells of blastoderm. JZ. Z. Y. Lower-layer cells with yolk. Z.U. Line along
which the two halves of the roof of the fore-brain have united. JZ. W. Lateral
wing or sheet of mesoblast growing out from the primitive streak. Mand.
Mandibular arch. J.B. Mid-brain. I.C. Medullary canal. des. Meso-
blast. Jes, #. Epithelioid mesoblast on under surface of serous envelope.
M.G. Medullary groove. Mo. Mouth. J.P. Medullary plate. 2.8.
Mesoblastic somite. Ma. Nasal pit. WV. Zz. Neurenteric canal. ot. Noto-
chord. 0. Al. Ventral opening of unenclosed part of alimentary canal. 0. L.
Optic lobe. O.P.4A. Opening into posterior enclosed part of alimentary
canal. Op.C. Optic cup of paired eye. Op.S. Optic stalk of paired eye.
O. V. Optic vesicle of paired eye. 2.4. C. Posterior amniotic canal. Pa. #.
Parietal eye. P. Ao. Primitive aorta. P.C. Pericardium (or coelomic space
which will give rise to pericardium). P.G. Primitive groove. PA. Pharynx.
Pit. Pituitary body. P.P.S. Pleuro-peritoneal space. Pro. dm. Pro-
amnion, Pr. 8. Primitive streak. P.S. I. Mesoblast derived from primitive
streak. P. V. Primary parietal vesicle. .C. Radial corrugations in the

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 81

yolk attached to the under surface of the yolk-sac. &. M. Roof of mid-brain.
S.B. Shell-breaker (patch of cornified epidermis on snout). S.C. Segmenta-
tion cavity (subgerminal cavity). WS. #. 2. Sub-embryonal membrane. 8. Zn.
Serous envelope. Sv. ‘Sickle,’ = mass of mesoblast growing out from the
primitive streak behind. 8. /. Superior maxillary process. Som. P. Somato-
pleure. Sp.C. Spinalcord. Spl. P. Splanchnopleure. S. Z. Sinus terminaiis.
Sta. Stalk with vitelline vessels attaching embryo to yolk-sac, Stom. Stomo-
deum, S. VY. Segmental vesicle. 7. Tail. Zh. Thyroid gland. Thad. Thala-
mencephalon. 77. Z. Transverse line along which the anterior end of the
embryo is attached to the overlying blastoderm, and where the yolk-sac turns
forwards to form the outer layer of the pro-amnion. V. 4. Vitelline artery.
1 V. A. First visceral arch. 2 V.A. Second visceral arch. 3 V. A. Third
visceral arch. 4 V.4. Fourth visceral arch. 5 V.A. Fifth visceral arch.
1 V.C. First visceral cleft (hyomandibular), 2 V.C. Second visceral cleft.
3 V.C. Third visceral cleft. 4 V.C. Fourth visceral cleft. V.V. Vitelline
veins. V.4. Fourth ventricle, W.D. Wolffian duct. W.R. Wolffian ridge.
W.T. Wolffian tubule. X. (Fig. 47). Junction of clear and yolk-laden areas
of yolk-sac. X. (Fig. 83). Spot where serous envelope is still united with true
amnion. FY. Yolk. ¥.8. Yolk-sac.

Note.—The microscopic drawings have in almost all cases been made with
the aid of Abbé’s camera lucida.

Fics. 1—5. Stage C.

Fig. 1.—Embryo 9, upper surface, with portion of surrounding blastoderm ;
drawn from spirit specimen as an opaque object. x 10.

Fig. 2.—Embryo 9, lower surface, with the sub-embryonal membrane partly
torn away ; drawn as before. x 10.

Fic. 3.—Embryo 9, seen from above as a transparent object after staining
with borax carmine and clearing in oil of cloves. Zeiss A (with the bottom
lens removed), ocular 1.

Fic. 4.—Embryo 9, longitudinal vertical section, passing through the head-
fold in front and the blastopore behind. The dotted line indicates the posi-
tion of the sub-embryonal membrane. Zeiss A, ocular 1.

Fie. 5.—Embryo 5 (slightly younger than 9, before the formation of the
blastopore), seen from above as a transparent object in Canada balsam, after
staining with borax carmine. Zeiss A (with the bottom lens removed),
ocular 1.

Fics. 6—14. Stage D.

Fic. 6.—Embryo 58, upper surface, with portion of surrounding blasto-
derm ; drawn from spirit specimen as an opaque object. x 10.

Fie. 7.—Embryo 58, seen from above as a transparent object after staining
with borax carmine and clearing in oil of cloves. Zeiss A (with the bottom
Jens removed), ocular 1.

VOL. 42, PART 1,—NEW SER. F

82 ARTHUR DENDY.

Fics. 8—14.—Transverse sections of embryo 58, arranged in order ; drawn
under Zeiss A, ocular 1.
Fig. 8—Through the blastoderm in front of the embryo, showing the
large coelomic space in the mesoblast (P.C.).
Fig. 9.—Through the anterior portion of the embryo lying freely above
the blastoderm.
Fig. 10.—Through about the middle of the embryo, showing the alimen-
tary canal still widely open below.
Fig. 11.—Through the medullary plate and primitive groove.
Fig. 12.—Through the primitive streak just in front of the blastopore.
Fig. 13.—Throngh the blastopore.
Fig. 14.—Through the primitive streak just behind the blastopore. (This
figure includes part of the area opaca on each side, which is omitted
from the others).

Fics. 15—34. Stage EH.

Fic. 15.—Embryo 56, upper surface, with portion of surrounding blasto-
derm; drawn from spirit specimen as an opaque object. x 10. (The an-
terior portion of the embryo now lies beneath the blastoderm of the area
pellucida, through which it is seen.)

Fic. 16.—Embryo 56, seen from below after removal of nearly the whole of
the sub-embryonal membrane, showing the anterior half of the embryo pro-
jecting freely beneath the blastoderm of the area pellucida; drawn as before.
x 10.

Fic. 17.—Embryo 56, seen from above as a transparent object, the front
half showing through the blastoderm of the area pellucida; after staining with
borax carmine and clearing in oil of cloves. Zeiss A (with the bottom lens
removed), ocular 1.

Fic. 18.—Hmbryo 64, seen from above, with portion of surrounding blasto-
derm ; drawn from spirit specimen as an opaque object. x 10.

Fic. 19.—Embryo 64, seen from below after removal of the sub-embryonal
membrane from beneath the anterior end of the embryo, while the posterior
part is concealed by this membrane and the adherent yolk; drawn as before.
x 10.

Fic. 20.—Embryo 64, seen from above as a transparent object, after staining
with borax carmine and clearing in oil of cloves. The front part of the embryo
is seen through the overlying blastoderm. Zeiss A (with the bottom lens re-
moved), ocular 1.

Fic. 21.—Embryo 64, anterior end seen from below, as before.

Fics. 22—32.—Transverse sections of embryo 56 arranged in order; drawn
under Zeiss A, ocular 1. (In Figs, 283—26 the overlying blastoderm is
omitted.)

Fig. 22.—Through the head in front of the alimentary canal. The over-
lying blastoderm (4. P.) is shown above.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 83

Fig. 23.—Through the anterior enclosed part of the alimentary canal,
in front of the coelomic space, which will give rise to the pericardium.
Fig. 24.—Through the ecelomic space (P.C.), which will give rise to the
pericardium.
Fig. 25.—Just behind the opening into the anterior enclosed part of the
alimentary canal.
Fig. 26.—Through the anterior end of the notochord.
Fig. 27.—Just behind the spot where the front end of the embryo,
enclosed in the pro-amnion, becomes free from the underlying blasto-
derm (compere Fig. 35, 77. L.).
Fig, 28.—Through the trunk region.
Fig. 29.—Through the trunk further back, showing the very small
amniotic cavity overlying the medullary groove.
Fig. 30.—Through the posterior amniotic opening, showing the two up-
rising folds of the amnion not yet united.
Fig. 31.—Just in front of the primitive streak.
Fig. 32.—Through the neurenteric canal (VV. Zz.).
Fic. 33.—Embryo 64. Transverse section just in front of the posterior
amniotic opening. Zeiss C, ocular 1.
Fic. 34.—Embryo 64. ‘Transverse section through the posterior amniotic
opening and primitive streak, showing the primitive groove (P.G.). Zeiss C,
ocular 1.

Fics, 35—46. Stage F.

Fic. 35.—Embryo 6]. Median longitudinal vertical section, slightly dia-
grammatic.

Fic. 36.—Embryo 61. Similar section a little to one side of the middle
line.

Fie. 37.—Embryo 72, seen from above, with portion of surrounding
blastoderm ; drawn from spirit specimen as an- opaque object, the front part
of the embryo being seen through the overlying blastoderm of the area pel-
lucida. x 10.

Fie. 38.—Embryo 72, seen from below after removal of the sub-embryonal
membrane and adherent yolk; drawn as before. x 10.

Fic. 39.—Embryo 72, seen from above as a transparent object, after staining
with borax carmine and clearing in oil of cloves. The front part of the embryo
is seen through the overlying blastoderm. Zeiss A (with bottom lens re-
moved), ocular 1.

Fie. 40.—Embryo 61, seen from above, with portion of the surrounding
blastoderm; drawn from spirit specimen as an opaque object; the front part
of the embryo seen through the overlying blastoderm of the area pellucida.
x 10.

Fic. 41.—Embryo 61, seen from above as a transparent object, after staining

84 ARTHUR DENDY.

with borax carmine and clearing in oil of cloves ; the front part of the embryo
seen through the overlying blastoderm of the area pellucida. Zeiss A (with
bottom lens removed), ocular 1.
Fics, 42—46.—Transverse sections of embryo 72, arranged in order.
Fig. 42.—Through the mid-brain. Zeiss A, ocular 1.
Fig. 43.—Through the hind brain, overlapped by the roof of the mid-
brain. Zeiss A, ocular 1.
Fig. 44.—Through the hind brain further back. Zeiss A, ocular 1.
Fig. 45.—Through the posterior amniotic canal (P. 4. C.). Zeiss C,
ocular 1.
Fig. 46.—Through the opening of the posterior amniotic canal on the
surface of the blastoderm. Zeiss C, ocular 1.

Fics. 47—50. Stage G.

Fie. 47.—Embryo 59, seen from below as an opaque object in spirit. The
yolk-sae (Y.S.) is partly torn away from beneath the serous envelope (S. Hz.).
*. 10:

Fic. 48.—Embryo 59, seen from above as a transparent object, after staining
with borax carmine and clearing in oil of cloves. (The broad dark line across
the embryo at Zr. Z. is due to tearing and crumpling of the yolk-sac.) Zeiss A
(with bottom lens removed), ocular 1.

Fic. 49.—Embryo 59, seen from below, as before.

Fie. 50.—Embryo 59. Median longitudinal vertical section, slightly dia-
grammatic. (The posterior amniotic canal lies to one side of the middle line
in this specimen, and is therefore not shown. The separation of the serous
envelope and yolk-sac in front is exaggerated for this stage.)

Fies. 51,52. Stage H.

Fic. 51.—Embryo 78, seen from below with portion of the surrounding
blastoderm as an opaque object in spirit. At the bottom right-hand corner
the thick, blanket-like yolk-sac (Y.S.) is turned up to show the overlying
serous envelope (S. Hx.). xX 10.

Fig. 52.—Embryo 78, anterior half, seen from below as a transparent object,
after staining with borax carmine and mounting in Canada balsam. Zeiss A,
ocular 1. .

Fies. 58—73. Stage J.

Fie. 53.—Embryo 60, seen from above as an opaque object in spirit, with
portion of the surrounding blastoderm; the body of the embryo is seen dimly
through the 8-shaped clear area and serous envelope. X 10.

Fic. 54.—Embryo 46, seen from above as an opaque object in spirit, with
portion of the surrounding blastoderm as before, but the serous envelope
(S. En.) is removed from the bottom left-hand corner, showing the underlying
yolk-sac (¥.S.). x 10. (This embryo is exceptional in having the amnion

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 85

incompletely formed dorsally, leaving a slit-like opening (dm.8.) into the
amniotic. cavity.)

Fie. 55.—Embryo 46. Transverse section through the dorsal amniotic
opening, a little in front of the neurenteric canal. Zeiss A, ocular 1.

Fic. 56.—Embryo 44, seen from above as an opaque object in spirit, with
portion of surrounding blastoderm. The embryo itself is seen dimly through
the foetal membranes. ~X 10.

Fig. 57.—Embryo 44, drawn from below, as before. x 10.

Fic. 58.—Embryo 44, seen from above through the foetal membranes as a
transparent object, after staining with borax carmine and clearing in oil of
cloves. Zeiss A (with bottom lens removed), ocular 1.

Fic. 59.—Embryo 44, anterior half, drawn from below as before. Zeiss A,
ocular 1.

Fies. 60—72.—Transverse sections of embryo 44, arranged in order; drawn
under Zeiss A, ocular 1.

Fig. 60.—Through the fore-brain, where the left half of its roof overlaps
the right.

Fig. 61.—Through the optic vesicles of the paired eyes.

Fig. 62.—Through the anterior enclosed part of the alimentary canal in
front of the stomodzum.

Fig. 63.—Through the stomodum and hind brain.

Fig. 64.—Through the hind brain and auditory pits.

Fig. 65.—Through the first pair of mesoblastic somites, heart and peri-
cardium.

Fig. 66.—Through the heart further back.

Fig. 67.—Through the fourth pair of mesoblastic somites, showing the ali-
mentary canal widely open below and the vitelline veins on each side.

Fig. 68.—Through the sixth pair of mesoblastic somites.

Fig. 69.—Through the twelfth pair of mesoblastic somites.

Fig. 70.—Through the ventral opening of the neurenteric canal.

Fig. 71.—Through the primitive streak.

Fig. 72.—Through the pleuro-peritoneal cavity behind the embryo, to show
the posterior amniotic canal (P. 4. C.) lying in the thickness of the
serous envelope.

Fic. 73.—Part of the last section more highly magnified, to show the pos-
terior amniotic canal lying in the serous envelope.

Fies. 74—82. Stage K.

Fic. 74.—Embryo 39, seen from above as an opaque object in spirit, with
the foetal membranes torn away from the front half. x 10.

Fic. 75.—Embryo 39, seen from below, as before. x 10.

Fics. 76—8$2.—Transverse sections of embryo 39, arranged in order ; drawn
under Zeiss A, ocular 1,

86 ARTHUR DENDY.

Fig. 76.—Through the mid-brain and fore-brain, showing the primary
parietal vesicle (P. V.) on what is really the left side of the middle line.

Fig. 77.—Through the hind brain and fore-brain, showing the development
of the ordinary paired eyes.

Fig. 78.—Through the hind brain and stomodeum.

Fig. 79.—Through the hind brain and auditory pits.

Fig. 80.—Through the trunk, showing the Wolffian ducts and tubules.

Fig. 81.—Through the trunk further back, showing the thickening of the
epiblast on the flanks of the embryo.

Fig. 82.—Through the neurenteric canal, placing the hinder part of the
medullary canal in communication with the posterior enclosed part of
the alimentary canal.

Fies. 883—91. Stage L.

Fic. 83.— Embryo 47, seen from above as an opaque object in spirit. The
yolk-sac and serous envelope have been removed from above the head and on
the right. x 10.

Fic. 84.—Embryo 47, from below, as before. x 10.

Fie. 85.—Embryo 50, hinder end, seen from above as a transparent object,
after staining with borax carmine and clearing in oil of cloves. To show the
origin of the allantois.

Fie. 86.—Embryo 50, hinder end, seen from below, as before.

Fics. 87—89.—Transverse sections of embryo 50, selected and arranged
in order to show the origin of the allantois; Fig. 87 being the most posterior,
and Fig. 89 the most anterior. Zeiss A, ocular 1.

Fie. 90.—Embryo 50. Longitudinal section through the head, nearly in
the median plane, but not quite truly vertical; showing the notochord and
pituitary body. Zeiss A, ocular 1. (The section passes a little on one side of

he communication between mid- and hind brains.)

Fic. 91.—Embryo 50. Similar section on one side of the median plane,
showing visceral clefts and head cavities. Zeiss A, ocular 1.

Fies. 92,93. Stage M.

Fie. 92.—Kmbryo 51, seen from below as an opaque object in spirit. (The
Wolffian ridge and hind limb were not really seen until after the removal of
the yolk-sac.) x 10. |

Fic. 93.—Embryo 51, sketched from above while alive to show the vitelline
circulation, seen through the serous envelope. x 10.

Fies. 94,95. Stage N.

Fie. 94.—Embryo 96, seen from below as an opaque object in spirit. The
yolk-sac and serous envelope have been removed on the left side of the figure.
x 10.

Fic. 95.—Embryo 96. Part of a median longitudinal vertical section
through the head, showing the notochord and pituitary body. Zeiss C, ocular 1.

OUTLINES OF THE DEVELOPMENT OF THE TUATARA. 87

Fics. 96—100. Stage O.

Fic. 96.—Embryo 103, seen from below, with the foetal membranes intact,
as an opaque object in spirit. x 5.

Fre. 97.—Embryo 92, seen from below as an opaque object in spirit, show-
ing the front end partially withdrawn from the invaginated yolk-sac. x 5.

Fie. 98.—Embryo 92, from the left side, after removal of the foetal mem-
branes, seen as an opaque object in spirit. X 5.

Fic. 99.—Embryo 92. Ventral view of head, seen as an opaque object in
spirit. x 10.

Fie. 100.—Embryo 92. Sketch of living embryo from above, to show
allantois, allantoic and vitelline circulations. (The arrows show the direction
in which the blood was flowing; only the veins were conspicuous: the
transparent serous envelope is not indicated.) x 5.

Fie. 101. Stage P.

Fre. 101.—Embryo 87, drawn from above (right side of embryo) as an
opaque object in spirit, to show the complete extrication of the front part of
the embryo from the yolk-sac. The amnion and serous envelope have been
partially removed, and the allantois has shrivelled. x 5.

Fies. 102, 108. Stage Q.
Fic. 102.—Embryo 1, drawn from the right side as an opaque object in
spirit, after removal of the foetal membranes. x 5.
Fic. 103.—Embryo 1, drawn as before, but from the left side. x 5.

Fies. 104—107. Stage R.

Fic. 104.—Embryo 2, drawn from the right side as an opaque object in
spirit, after removal of the foetal membranes. x 5.

Fig. 105.—Embryo 2, drawn from above, as before. x 5.

Fie. 106.—Embryo 8. Portion of yolk-sac, with yolk particles adhering to
the absorbent vessels so as to give rise to a radially columnar structure ; seen
as an opaque object in spirit on a dark background. xX 5.

Fic. 107.—Embryo 3. Yolk-spheres and crystalloids ; drawn from a pre-
paration mounted in glycerine, after treatment with Kleinenberg’s picric acid
and alcohol. Zeiss D, ocular 3.

ABSTRACT AND REVIEW. 89

Abstract and Review of the Memoir by G.
Hieronymus ‘On Chlamydomyxa laby-
rinthuloides, Archer.”

By

J. W. Jenkinson, M.A.,
Exeter College, Oxford.

PAGE

I. Previous work . : : : 5 gx)
II. Preliminary account : 5 : F oe
III. The life history ; : : . #, #496
a. The ameebee : ‘ F : es)

b. The cysts : : : : me)!

IV. The cell contents : : : : > Or
a. The nuclei : ‘ : P 7 10k

6. The chromatophores ; 5 : . 103

ec. The oil-drops : . : . 104

d. The physodes and torial : 5 . 105

e. The crystals ‘ , : ‘ : 107

J The cell wall : : ‘ «, LO7

V. The systematic position of Gilenndonen ; : . 108
VI. Remarks , : 5 : : + 109

I. Previous Work.

a. Archer.—Chlamydomyxa labyrinthuloides was
originally discovered and named by Archer (1) in 1875. He
supposed it to be related to Labyrinthula (4), and described it
as a protoplasmic body, rounded or irregularly lobed, invested
by a cellulose cyst, colourless, or straw-yellow, and generally
of several layers, and living on water-plants, or parasitically in
the cells of Sphagnum or Hriophorum leaves, or in the air-
spaces of roots of Eriocaulon. He further saw it streaming
out through a rupture in the cyst as a sort of labyrinth-like
meshwork, the nodes of which were ameeboid, ingested Algz,
and finally encysted themselves anew.

90 J. W. JENKINSON.

In the protoplasmic body were (1) a hyaline, vacuolated
ground substance; (2) red granules of different sizes, varying
in number, and not always present ; (3) yellowish-green, round
or irregular granules closely resembling the chloroplastids of
certain Algz except in colour; they were smaller but more
numerous than the red, which Archer supposed to be derived
from them; (4) there were numerous small, pale blue homo-
geneous bodies, rounded in the resting stage, fusiform during
the streaming, capable of division, and of gliding on or in the
threads of the meshwork with a movement, according to Archer,
of their own; and (5) in the streaming condition there was a
“ diffuse chlorophyll.”

Reproduction was effected in two ways: (1) by the separa-
tion of the nodes of the meshwork ; and (2) by the division,
inside the cysts, of the protoplasmic body into several portions,
each of which became surrounded with a proper cyst of its own.

6. Lankester (8) suggested that the spindles of Chlamy-
domyxa, as also those of Labyrinthula, were to be regarded as
nuclei.

c. Geddes (6) described the resting stages. He did not
observe the streaming out of the protoplasm, although he re-
peatedly saw hernia-like protrusions of the contents of the
cyst, which re-encysted themselves outside the Sphagnum cell.
He also observed division of the contents of a cyst and the for-
mation of proper cell-walls round these, as well as the formation
of a septum between separate portions of a cyst.

He describes, in addition, two other kinds of cysts, one of
which he calls the Protococcus form, and believes to have arisen
from separated portions of the labyrinthine meshwork; the
other sort being due, he supposed, to individuals which had
migrated out of their cysts, spent some time as naked ameebe,
and then re-encysted.

He mentions, without figures or further description, cell-
nuclei ; and, besides these, masses of “ protoplasm,” usually of
a red colour, and occasionally nucleus-like bodies containing a
nucleolus, to be regarded as secondary cysts exceptionally
formed inside the primary ones. A yellowish pigment, pro-

ABSTRACT AND REVIEW. 91

bably xanthophyll, is present in conjunction with the chloro-
phyll; the distribution of the latter is irregular, but occasionally
definite corpuscles, the primitive form of the chloroplastids of
the higher plants, occur. The red colouring matter he believes
to be derived from the green ; it is found sometimes in minute
drops between the consecutive layers of the cellulose cyst.

As processes of reproduction he mentions fission, budding,
free cell formation, and rejuvenescence. Lastly, he believes the
organism to occupy the same position relatively to the lower
Algee as do the Myxomycetes to the lower Fungi.

d. Askenasy (2) has criticised these statements, suggesting
that the organism described by Geddes was not Chlamy-
domyxa at all, the latter being, he supposes, really related to
the Rhizopoda.

e. Biitschli (8) places Labyrinthula among the colonial
Rhizopoda (Mikrogromia, ete.), but considers it doubtful
whether Chlamydomyxa is really related to it, the “spindles”
of the two not being homologous.

f. It is highly probable that Janisch (7) really observed
Chlamydomyxa in 1859, mistaking it for Pleurostaurum or
Cocconema.

[g. Hieronymus has apparently not seen Lankester’s (9)
account of Chlamydomyxa montana. This was seen in the
streaming condition, and later on encysted. No ingestion of
food particles was observed. ‘The figure given l. c. is quite
like that of Hieronymus copied here. Lankester expresses
the opinion that the threads are pre-formed and comparable
with the axis of the Heliozoan pseudopodium ; he believes they
are covered by a layer of invisible hyaloplasm, the streaming
of which was the cause of the motion of the “ oat-shaped cor-
puscles”’ (“ spindles ’’).

In the encysted condition crimson oil-drops were observed,
and the division of the protoplasmic body inside the cyst into
several portions.. He observed no nuclei.—J. W. J.]

92 J. W. JENKINSON.

II. Pretiminary Account.

The present author has found Chlamydomyxa in the Riesen-
gebirge. It occurs in cells and intercellular spaces of Sphagnum
and dead Cyperacee and Graminee leaves. The extrusion of the

me s-4,

Fie. B.—A cyst out of which
the amceba is emerging, and
at the same time dividing;
one half will remain behind.
s, an egested oil-drop; z,a

Fie. A.—An ameeba dividing. cell-wall thickening.

cell contents, the ingestion of food by these, the re-entrance into
Sphagnum cells, and the re-encystment have all been observed.

Other organisms are found in the Sphagnum cells in com-
pany with the Chlamydomyxa. One of these is Chloro-
chytrium Archerianum (mentioned by Archer). Another

ABSTRACT AND REVIEW. 93

Alga, Urococcus Hookerianus, also occurs. It is mentioned
by Geddes, but supposed by him to be a stage in the deve-
lopment of Chlamydomyxa; he figures it in his figs. 3, 4, 5 a,
and 56. There is, however, no genetic connection between the

Fic. C.—A labyrinthine amceba dividing into several parts simultaneously.

two forms. His fig. 2 is a true Protococcacean, also often
found with the other two.

In Chlamydomyxa the nuclei are of a different shape, smaller
size, and much greater number than they are in Urococcus ;
the latter is also much more easy to cultivate.

With regard to Chlamydomyxa itself, we have to attempt
an answer to the following questions :

94, J. W. JENKINSON.

1. Do nuclei exist ?

2. What are the “ spindles” P

3. Are the yellow-green bodies chromatophores or symbiotic
Algee ?

4. What are the red bodies ?

5. Is “ diffuse chlorophyll” present ? and—

6. What is the systematic position of the organism ?

The following is an enumeration of the cell-contents accord-
ing to Hieronymus:

1. The Nuclei.—Of these there is only one in the quite
young cysts, in larger and older cysts two or more, and in the
largest several ; and conversely in those amcebz which have just
left the cyst there are many nuclei, in the final products of
division only one.

2. The Chromatophores.—These contain a green pigment,
probably identical with chlorophyll, and a yellow-brown pig-
ment. During the period of ingestion they fade.

3. Oil Bodies.—These are of a red colour, sometimes olive-
green or blackish brown. They are formed from aggregations
of dead chromatophores, and can be artificially produced in
sunlight. They are egested by the amcebe, and often left
behind in the cyst.

4. Rod-shaped crystalline bodies of calcium oxalate ;
they are formed in the hyaloplasma or in the cell-sap, and
exhibit Brownian movements in the vacuoles.

5. A vacuolated hyaloplasma, containing—

6. Granular or drop-shaped bluish, highly refractile cor-
puscles, deeply staining in the living cell, and identical with
Archer’s spindles and Crato’s physodes.

ABSTRACT AND REVIEW. 95

Fie. D.—1. Two thick-walled many-layered cysts. x 620. 2—4. Cysts
showing the distribution of the nuclei. Chrom-aceto-osmic and hematein-
ammonia preparations. xX 620. 5, The end of a living cyst treated with
methylene blue and pure water. x 5000. 6. Nuclei. Chrom-aceto-osmic
and hematein-ammonia preparation. x 10,000. 7and8. Chromatophores,
living. x 10,000. 9. Chromatophore. Chrom-aceto-osmie and fuchsin
preparation. x 10,000. 10. Living physodes. x 10,000.

96 J. W. JENKINSON.

III. Tue Lire History.

a. The Ameba Stage.—The amcebe commonly arise by
the extrusion from the cyst of the whole of the multinucleate
contents, which then divide into two with the appearance of
pseudopodia. The process is repeated until the whole is divided
into uninucleate amcebe, which then encyst (Fig. A).

There are often, however, modifications of this process. For
instance, the first division may take place during extrusion
from the cyst, in which single products of division may remain
behind and re-encyst (Fig. B); or the original amceba may
divide into more than one at a time, or the second division
may begin before the first is completed.

Less frequently the whole body breaks up simultaneously
into several pieces (Fig. C), assuming then the labyrinthine
appearance observed by Archer, the thickened nodes of the
meshwork being the separate energids of the protoplasmic body.

The present author, however, only observed the streaming of
the protoplasm out of such cysts as that depicted in Fig. B,
- not out of the Urococcus-like many-layered cysts, as seen by
Archer.

It is probable that this streaming and simultaneous division
are due to favourable conditions of temperature, and so forth,
supervening on a prolonged period of unfavourable circum-
stances, during which the forces concerned have become much
intensified. At different stages of division, or even before the
first division bas taken place, the amcebe ingest food particles,
mostly diatoms, and encyst themselves with these; the uni-
nucleate amcbe are only able, however, to ingest bacteria, or
very small green or blue-green Algze. More rarely they ingest
small pieces of Alga filaments, such as Gidogonium ; starch
grains they devour readily, and they will also take up grains
of sand or bits of decaying plants, though they soon egest these
again.

After encystment there is, in all cases, a rapid multiplication
of the nuclei.. In this condition they are generally found on

ABSTRACT AND REVIEW. 97

the outside of Sphagnum, grass leaves, etc., but sometimes
inside the Sphagnum cells with diatoms ingested, apparently
after their entrance into the cells. The ingested Algz are not
entirely digested, there being always left over the membrane,
and some grey or black granules, which are either egested
when next the creature creeps out as an ameeba, or if, as fre-
quently, a new cyst is formed inside the old one, are deposited
between the two. Once the contents of the cyst were observed
to divide, one half creeping out, the other re-encysting itself
with the previously ingested diatom ; and on another occasion,
after the cyst contents had divided, each half re-encysted,
the chromatophores of that half which contained the diatom
becoming quite pale (Fig. E).

Fie. K.—A cyst containing two daughter -eysts: one of these, which has
pale chromatophores, contains a large diatom; the other has dark brown
chromatophores.

It very frequently happens that the amebe re-eucyst without
having ingested any food at all. The cause of this is probably
to be sought in differences of temperature and moisture ; if, for
instance, certain conditions were to induce rapid multiplica-
tion of nuclei without a corresponding development of chro-
matophores, the creature might give up almost entirely a

voL. 42, PART 1.—NEW SER. G

98 J. W. JENKINSON.

holophytic mode of nutrition, and become for the time holozoic.
The paleness of the chromatophores at the period of ingestion
is probably to be explained on this assumption, that they then
become superfluous and go through a period of rest, during
which they are comparable to the leucoplastids of the higher
plants.

, 9
; sho
i 4 a ss bers
o ae Lae
ne fads V4 Vy
1 SOS Paes tee
i a) Sr ei
iy TT ; Vad
ag by

».\
Fic. F.—An ameeba with an extraordinarily large hyaline margin (ectosare).
Physodes are moving along the pseudopodia.

The ameebe generally creep on the surface of the Sphagnum
and other plants (Figs. A, C, and I’) ; very rarely they are seen

ABSTRACT AND REVIEW. 99

swimming (compare Lankester’s figure [9]). In this condition
they never divide, and are always circular or elliptical. The
dark central mass which contains the nuclei and chromatophores
is surrounded by a margin of clear protoplasm, from which
emanates a halo of hyaline pseudopodia, which continually
shorten and lengthen again, or may be altogether withdrawn
and then again protruded. These pseudopodia are similar to,
although much larger than, those of the creeping amcebe,
which latter are only developed on the forwardly directed end
(Fig. C). Along them move numerous round or spindle-shaped
physodes, carried hither and thither by the streaming of the
protoplasm. The emergence of the amcebez from their cysts,
and their subsequent division, only take place during the
summer months, and then only during the warm hours of the
day, and in the presence of light of sufficient intensity.

In the amebe the olive-green or red oil bodies are never
seen, nor the crystals of calcium oxalate; but both are fre-
quently found in deserted cysts, being sometimes egested at:
the moment of emergence.

6. The Cysts.—The form of the cyst, when on the surface
of leaves, etc., is generally spherical or ovoid, occasionally
elongated or irregularly lobed, the last when large Alge have
been ingested. In the case of those cysts, however, which are
inside the cells of the plants on which they live, the form is
adapted to that of the cell; but if this is too small for the
enclosed cyst, the latter may protrude (cf. Geddes, fig. 1) as a
hernia-like swelling into which a portion, or in some cases the
whole, of the protoplasmic contents migrate. A new spherical
cyst forms round the protruded protoplasm, while the old cyst
splits, thus allowing the newly formed one to escape without
becoming ameeboid.

Again, it often happens that the protoplasmic contents
divide into two or more parts, each forming a proper cell-wall
of its own. The large many-layered cyst described by
Archer, enclosing several small cysts, is probably a sort of
formation of resting spores. The repeated formation of
a new membrane round, and the subsequent division of the

100 J. W. JENKINSON.

whole cell contents, would produce a form quite similar to that of
Urococcus, except that it would be multinucleate and possess
no starch grains.

Two or more cysts are often found in the same Sphagnum
cell. These by their growth exert a pressure upon one another,
which may result in the death of one of them, or in the
bursting of the Sphagnum cell, and consequent protrusion
of the cyst. This mutual pressure also occurs in any cysts
which happen to lie closely together, producing thickenings of
their walls where they rest on one another. This latter also
occurs in solitary cells. The completely grown cysts contain
at least eight nuclei, sometimes as many as thirty-two.

The number of chromatophores, their size and distribution,
are extremely variable. The quite young cysts generally contain
few, but those large ; the full-grown cysts much smaller ones,
but in very much greater numbers. In the ripe cysts they can
only be distinguished from one another by squeezing out the
contents ; but if the cyst be unripe there are numerous large
vacuoles in the protoplasm, which make it quite possible to see
the separate chromatophores as well as the minute crystals of
calcium oxalate with which the vacuoles are crowded.

The over-production of this calcium oxalate seems to be in
some way deleterious to the organism; if the cell contents
migrate the crystals are egested, but the amceba stage may be
altogether prevented by their presence. If the organism be kept
in lime water calcium oxalate is produced in excessive quan-
tities, and many cysts are killed; the chromatophores are the
first to die, being transformed directly into brown or blackish
masses, and after them the nuclei become disorganised.

The vacuoles mentioned above as occurring in the unripe
cysts may run into one another. The protoplasm is then
divided up into numerous rounded or angular portions, each
containing a nucleus, and termed originally by Sachs (10) an
“energid.” They are connected with one another by threads of
hyaline streaming protoplasm, along which are carried the
glistening physodes (Archer’s spindles), and sometimes chro-
matophores as well, but never nuclei. ‘The separation of the

ABSTRACT AND REVIEW. 101

eyst contents into energids can be artificially effected hy expo-
sure to strong sunlight, when the protoplasm and chromato-
phores are found to have grouped themselves round the
individual nuclei, apparently to screen them from excessive
light ; and if this treatment be prolonged, many of the chroma-
tophores become transformed into first olive-green and then red
oil-drops, which always occupy a position outside the energids,
where they can shield the nuclei and the remaining chromato-
phores from thesun. By still further prolonging this treatment
all the chromatophores can be so transformed, but this is
followed by the death of the individual.

These oil-drops are also always most plentifully found in
material taken from dry or sunny places, and they are often to
be seen, as Geddes observed, between the layers of the cyst, in
which case it seems probable that they are rejected by the
organism when no longer of any use to it. The red colour of
the oil-drops finally changes to a deep brown or black, similar
in appearance to the granules of the digested Alge, whence
it seems probable that they may be used as food.

In favourable places the whole cycle of the life history may
be repeated several times in a summer. In winter, and in
dry or otherwise unfavourable places, the cyst becomes very
thick and stratified (Fig. D, 1).

IV. Tue Ceti Contents.

a. The Nuclei.—These can only be satisfactorily demon-
strated by appropriate fixing and staining reagents, although
in the living cysts the clear masses of protoplasm in which the
nuclei lie can sometimes be seen if the chromatophores and
other cell contents are not toonumerous. The calcium oxalate
crystals are also a great hindrance to accurate observation, and
must, before staining, be got rid of by treatment with hydro-
chloric acid.

The fixing reagents used were absolute alcohol, aqueous
picric acid, chromic acid, and various mixtures of these two
with acetic and osmic acids. Finally, and most generally,

102 J. W. JENKINSON.

iodine added to the water in which the creature was living.
The iodine material was gradually hardened in alcohol, while
that preserved in picric and chromic acid was transferred to
and kept in water containing numerous crystals of naphthalin.

The stains used were hzmatein-ammonia and Mayer’s
hemalum, the former of which gave the better results. The
objects were deeply over-stained, and then decoloured in alum
until only the granular portions of the nuclei retained the
stain; they were finally mounted in Canada balsam in the usual
way. Only those amoebe were preserved which happened to be
migrating into Sphagnum cells; but their nuclei, and therefore
presumably the nuclei of all other amoebz, exhibited the same
appearance as those of the cysts.

In the uninucleate amcebe or cysts the nucleus always
occupies a nearly central position, never being situated against
the cyst wall, or in the hyaline border of the ameebe. In the
multinucleate stages the nuclei are evenly distributed. They
may lie in a row if the cyst is compressed by the narrowness of
the Sphagnum cell, or, if the cyst is broader, they may
alternate with one another in two rows, or in large cysts may
lie equally in all directions (Fig. D, 2—4), and this uniformity
is quite uninterrupted by local thickenings of the cell-wall, or
by the protrusions of the cyst. The size of the resting nuclei
varies, the diameter being from 1} to 3m. They are gene-
rally round, almost spherical, more seldom lens-shaped, ovoid,
or faintly lobed. They lie embedded in a more or less thick
layer of hyaloplasm, which passes indistinguishably into the
nuclear membrane. The structure of the nucleus is reticulate,
though the threads of the reticulum are so excessively minute
that they can only be recognised by the rows of chromatin
granules that lie upon them.

In properly stained preparations the nuclei appear to be
dividedinto layers. In the centre space—about one third of the
diameter of the whole nucleus—are generally several, twelve
or more, but sometimes only a few, or even one, large, deeply
staining granules (Fig. D, 6). These are probably nucleoli; they
are not protein crystals, which do not stain with hematoxylin,

ABSTRACT AND REVIEW. 103

nor chromatin granules, for the small peripheral granules (see
below) go blue when stained with iodine-green fuchsin, or
methylene-blue acid fuchsin, while the central granules colour
red. Next there is a zone free from granules, and traversed only
by achromatic threads, between which is the so-called nuclear
fluid; this zone varies in breadth, but may occupy as much as
one fifth of the whole diameter. Around this clear zone, again,
is the outermost of all, which may be in breadth as much as two
thirds of the whole diameter, but which varies with the width
of the clear zone. It contains numerous small deeply staining
granules of chromatin, occasionally equalling the nucleoli in
size. The division of the nuclei is apparently amitotic.

b. The Chromatophores.—No “ diffuse chlorophyll” can
be detected either in the amcebe or in the cysts. The pigment,
yellowish or brownish green, is always contained in definite
though very minute corpuscles. These are either discoid or
lens-shaped, appearing fusiform in profile, or may be nearly
spherical, sometimes angular or lobed. In the largest chroma-
tophores the thickness varies from 14 uw to 2u, the diameter
from 4} to 54, but the size seems to vary inversely with the
number. In the amcebex, as also in the mature and in the
uninuclear cysts, they are generally very small indeed, about
3 in diameter, which may account for the diffuse chlorophyll
of other authors. They attain their maximum size in the adult
but still immature cysts.

In the uninjured cell very little structure can be detected
in the chromatophores, except now and then darker spots on
a lighter ground substance, due, perhaps, to thicker accumu-
lation of the oil which lies in the chromatophoric reticulum
(Fig. D, 7, 8). But when fixed, and with the oil and pigment
removed by alcohol, a skein-like structure of threads twisted
over one another exhibits itself. These threads are composed
of rounded segments, Mayer’s granules, and can be readily
stained. In the smaller chromatophores there is only one
twisted thread, but in the larger several. This fibrillar struc-
ture can be readily seen and the threads isolated by crushing
the chromatophores under the cover-glass, or by prolouged

104. J. W. JENKINSON.

treatment with strong salt solution. The granules are then
seen to be the vehicles of the pigment which lies in their
peripheral portion, the centre being colourless (Fig. D,9). The
yellow-green or brown-green pigment seems to be a com-
bination of two distinct colouring matters. If the chromato-
phores are treated with dilute alcohol they become of a grass-
green before they eventually fade; there is thus a brown or
yellow pigment, more soluble in alcohol or in water after the
death of the organism than the green. In absolute alcohol,
however, the yellow-brown pigment is the less soluble.

By the action of concentrated hydrochloric or sulphuric
acid the chromatophores are coloured bluish green or blue,
and by continued treatment with the former hypochlorin
masses are differentiated as dark green drops or crystals.
These reactions prove that the green colouring matter is
identical with, or very nearly related to, the chlorophyll
of the higher plants. The yellowish-brown pigment, on
the other hand, is probably itself a mixture of two: one of
these may be identical, as Geddes supposed, with carotin or
xanthophyll, or else with diatomin (phycoxanthin) ; the
other may be either the phycophzin of the Phzophycee, or
the phycopyrrhin of the Peridinez.

These chromatophores are almost certainly not symbiotic
Algee, neither a cell-wall nora nucleus being detectable in them.

c. The Oil-drops.—These, as above stated, arise by the
degeneration of the chromatophores, for when they are formed
the number of the chromatophores decreases, and if they be
decoloured in alcohol they exhibit traces of the chromatophoric
structure. They are at first olive-green, then red, and finally
brown or blackish. They are formed under the influence of
direct sunlight, and can be artificially produced by insolation ;
if the organism is then removed from the direct light, the
shade, which it appears to be the function of these oil-drops to
afford, is no longer necessary, and they become disorganised,
changing to the brown or black colour. They may, as Geddes
observed, be egested and deposited between successive layers of
the cysts; he figures them red, but the present author only

ABSTRACT AND REVIEW. 105

found them there in the brown condition. They may also be-
come enclosed inside the cyst by a proper cell-wall of their own.

The red colouring matter is probably one of the fatty pig-
ments (lipochromes). By treatment with sulphuric acid the
masses are coloured a bright blue, but the formation of crystals
of lipocyanin was not observed. The change of the red to a
brown colour possibly points to there being here also two
pigments in combination ; the red colour, also, is not always
the same, being sometimes rose or carmine, sometimes a brick-
red.

d. The Physodes and the Hyaloplasma.—The physodes
are glistening, highly refractile bodies, with sometimes, when
of considerable size, a blue sheen. In the cyst they are
spherical and arranged in rows, less frequently spindle-shaped,
angular, or lobed. They are Archer’s “ spindles,” taken by
later writers for nuclei. Archer observed in them a change of
form, which, however, has not been seen by the present author.
This change of form is probably to be explained, as Archer
believed, by supposing that the bodies become viscid under
stress of the varying forces exercised upon them from time to
time by the protoplasm.

The smallest of them exhibit no structure of any kind,
although the larger show a sort of stratification, there being a
central more refractile portion.

Crato (5) has recently figured similar cell contents as
occurring in many Fucacez, and also in Chlorophycee. To
these he has given the name physodes, and has asserted that in
many cases they contain phloroglucin. Now by treating
Chlamydomyxa with vanillin-hydrochloric acid a red colora-
tion, intensified by the addition of sulphuric acid, is found in
these bodies, a characteristic test for phloroglucin. <A similar
reaction can be obtained by treatment with alcoholic piperonal
solution and sulphuric acid. With iron chloride they stain a
pale greenish blue, and finally, in the living cell they take up
very strongly such stains as iodine green, methylene blue,
methyl violet, and methyl green, and retain these after the cell
has been killed in water or alcohol.

106 J. W. JENKINSON.

All these facts point to their being related to Crato’s
physodes. They are also partially soluble in absolute alcohol
and ether, though they do not entirely disappear in these
liquids as Crato’s physodes do. If killed in 80 per cent. alcohol
they only contract slightly, at the same time losing much of
their staining capacity, and refusing to show the characteristic
reaction with vanillin, so that the phloroglucin is apparently
dissolved out. By staining the fixed material with iodine-green
fuchsin or methylene-blue acid fuchsin, the ground substance of
the physodes appears kyanophilous ; it also takes up hematein-
ammonia as readily as the chromatin and nucleoli, but can be
destained with alum more readily than these.

In the living cell the physodes often lie in rows, sometimes
very closely together (Fig. D, 7), which rows have apparently
an oscillating movement. Along the protoplasmic threads, also,
which connect the separate energids of the cell, the single
physodes, as well as the rows, are seen to move. This may be
readily observed by intra vitam staining (Fig. D, 5) witha
very dilute aqueous solution of methylene blue ; if too strong a
solution be used the protoplasm becomes vacuolated, many of
the deeply stained physodes being ejected into the vacuoles,
where, together with the crystals of calcium oxalate, they are
seen to execute the so-called Brownian movements, evidence
that in the cyst, at any rate, they are of a more or less per-
manent consistence. The cell sap in the vacuoles becomes
at the same time faintly reddish, showing an acid reaction.
If the organism be placed in a dilute methylene-blue solution
for a very short time, so that a few only of the physodes
—those situated peripherally—are coloured, and be then left
for some time in pure water, then the more energetic
streaming of the protoplasm which ensues causes the physodes,
stained and unstained, to become closely packed together, and
finally deposited in one or more vacuoles which lie near the
cell-wall. These would probably, lke the oil-bodies, become
covered in time by a layer of cellulose. The arrangement in
rows of the physodes is probably due to frequent division, and
is seen best after the organism has been kept a short time only

ABSTRACT AND REVIEW. 107

in culture. They lie embedded in the threads (visible after
treatment with any reagent, and probably present though un-
detectable in the living state) of the hyaloplasma, which
themselves are coiled in a scarcely staining mass (Kupfer’s
paraplasma), the kyanophilous physodes being immediately
surrounded by an erythrophilous layer. This can be readily
made out with iodine, which stains this layer deeply, the
physode itself not at all.

The hyaloplasmic threads are closely packed round the nuclei,
less so round the chromatophores ; aggregations of them, with
their contained physodes, are especially common in places
where the cell membrane is undergoing a marked growth,
which would suggest that the physodes are reserve food
material, the nature of which cannot, however, be at present
more precisely affirmed. The hyaloplasmic threads never
branch or anastomose.

e. The Crystals.—These are found, in “ Brownian ” move-
ment, in the vacuoles. They are generally egested by the
organism when it assumes the amceboid stage, and are some-
times left behind in the cyst. That they are crystalline is
proved by polarisation, and that they are of calcium oxalate is
proved by their insolubility in acetic acid, whereas they readily
dissolve, even in weak solutions of hydrochloric, sulphuric,
and nitric acids, without the evolution of bubbles. If strong
sulphuric acid is used crystals of calcium sulphate are formed,
and finally, if they be heated, the calcium oxalate is converted
into what is probably calcium oxide, which is then soluble in
acetic acid.

The form of the single crystal is rod-shaped ; when cut it is
to be supposed that they are mono-symmetric; double and
triple crystals also occur, as well as aggregations, which stick
together and may completely fill up a vacuole. These crystals
are never found in the hyaloplasmic threads, and therefore
probably arise in the paraplasma, or may possibly be sometimes
formed in the vacuoles themselves.

f. The Cell-wall.—This gives the characteristic cellulose
reaction of blue or dark violet with iodine and sulphuric acid,

108 J. W. JENKINSON.

and with chlor-zinc-iodine. In the case of the many-layered
resting cysts tle outer layers were violet ; the inner, as also the
single membranes of the quite young cysts, blue. Thickenings
in the cell-wall stain more deeply and are less easily decolourised
than other parts with reagents like iodine green, Congo red,
safranin, etc.

V. Tue Systematic Position oF CHLAMYDOMYXA.

The views of Janisch, Archer, Geddes, Askenasy, and
Biitschli have already been stated. Lankester (9), while refrain-
ing from asserting its exact relation to the other Gymnomyxa,
believes, with Archer, its nearest ally to be Labyrinthula. It
is held by Hieronymus that this organism is on the border-land
between plants and animals, although nearer, by reason of its
chromatophores and cellulose membrane, to the former; on
the other hand itis, at times, certainly holozooic. This form of
nutrition is, however, known in Chromulina and Ochromonas
(12), and in Peridinez, even in chromatophore-containing
forms (11).

Had Chlamydomyxa no chromatophores it could have been
related to Vampyrella; also the paramylum grains, which
are present in this latter, have not been detected in Chlamy-
domyxa.

Assuming, with Biitschli, one single stem for the origin of
all organisms, one might (says Hieronymus) regard Chlamy-
domyxa asa specialised derivative of Vampyrella ; on the other
hand, if we assume with Nageli a polyphyletic origin for
organisms, there being between members of the different lines
of descent apparent relationships due to convergence, then
Chlamydomyxa must be placed in a separate family, which will
occupy the lowest grade among the yellow-brown Alga, leading
up to the Dinoflagellata and diatoms on the one hand, to the

1 Which is not surprising if we do not (as do a certain section of botanists)
go out of our way to make the indefensible assumption that the Flagellata
and Dinoflagellata belong to the vegetable series !—H. R, L.

ABSTRACT AND REVIEW. 109

Chromomonadinacez and Phzophycee on the other. It will
also have relations, in a wider and different sense, with the
Vampyrellaceee, Labyrinthulee, Myxomycetes, and Fungi on
one side and on the other with the Heliozoa, Rhizopoda, and
Radiolaria.

A parasite, probably Pseudospora maligna, sometimes
occurs in the cysts.

VI. Remarks.

[Hieronymus, of course, approaches his account of this
organism entirely from the botanical standpoint.' He adduces
chiefly the presence of chromatophores and of a cellulose cyst
as reasons for regarding the organism as a plant; but these
occur in several well-known instances in animals. It would
have been more to the point, perhaps, had he relied on
the presence of phloroglucin and calcium oxalate, and the
alleged periodic holophytism. Phloroglucin is unknown among
animals. Calcium oxalate occurs only, I believe, in the
Myxomycetes. Hieronymus may, of course, be mistaken in
his identification of the substance found in the spindles with
phloroglucin ; he certainly mentions a fair number of tests, but
these seem all merely to rest on our ignorance of any other sub-
stances which would give the same reaction. His figures of the
spindles (physodes) differ a good deal from those given by
Archer and Lankester.

It is also, perhaps, a curious coincidence that these physodes
should have so marked an affinity for the basic aniline stains ;
it is perfectly true, on the other hand, that they exhibit no
structure in the least comparable to that of a nucleus.

Hieronymus’s account of the nuclei seems sufficiently to

1 The notions indulged in by Hieronymus as to the relationship of Chlamy-
domyxa to the yellow-brown Alge, and of every yellow-brown organism with
every other, are devoid of any serious basis in fact. Whilst his paper contains
some observations of importance, ¢.g. as to the nuclei, and some the accuracy
of which seems to need further inquiry, e.g. as to the chromatophores, the
general views which dominate the author’s speculations appear to be those of
a botanical specialist whose knowledge of Protozoa is defective-—H. R. L.

110 J. W. JENKINSON.

explain why his figures show so little of the typical reticular
structure. He has, of course, omitted to emphasise reasons
which might induce a zoologist to claim this organism for his
own province, such as the ingestion of solid food, and the

existence of pseudopodia covered with streaming protoplasm.
—J. W. J.]

12.

LITERATURE.

. Ancuer.—“*Chlamydomyxa labyrinthuloides,” ‘Quart. Journ.

Mier. Sci.,’ xv.

. AskENAsY.—‘ Bot. Jahresberichte,’ 1882, x, i.
. B&rscuti1.—* Bronn’s Thierreichs,” ‘ Protozoa,’ vol. i, p. 145, note.
. CrenkKowsk1.—“ Bau und Entwickelung der Labyrinthuleen,’’ ‘Arch. f.

Mikrosk. Anat.,’ li, p. 274.

. Crato.— Die Physode, ein Organ des Zellenleibes,” ‘ Berichte der

Deutsch. Bot. Gesellsch.,’ x, 1892.

. GeppEs.—“The Resting Stage of Chlamydomyxa labyrinthu-

ides uart. Journ. Mier. Sci.,’ xxii.
loides,” ‘ Quart. Jo M Sci.,’ xx

. JANISCH.— Ueber Pleurostaurum acutum,” ‘ Hedwigia,’ vol. ii,

p. 25.

. Lanxester.—“ Haliphysema Tumanowiczii,” ‘ Quart. Journ. Mier.

Sci.,’ xix.

. LankesteR.—“Chlamydomyxa montana,” ‘Quart. Journ. Mier.

Sci.,’ xxxix.

. Sacus.— Physiologischen Notizen,” I, in ‘ Flora,’ 50, p. 57.
11.

Scuittinc.—“ Untersuchungen tiber die thierische Lebensweise einiger
Peridineen,”’ ‘ Berichten der Deut. Bot. Gesellsch.,’ ix.

Wysotzx1.—“ Mastigophora Rbhizopoda,” ‘Arbeiten d. Naturf.
Gesellsch.,’ xxi, 1887.

JUL i/ Lowy

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. I111

On the Development of the Parietal Eye and
Adjacent Organs in Sphenodon (Hatteria).

By

Arthur Dendy, D.Sc., F.L.S., ;
Professor of Biology in the Canterbury College, University of New Zealand.

With Plates 11—13.

——

TABLE OF CoNnTENTS.

PAGE
1. InTRoDUCTION 101
2. SysTEMAtTICc ACCOUNT OF ORhoNS ON Spay ton PUNC-

113

TATUS
3. Discussion or RESULTS:
(a) Structure of the Parietal Eye—Retina—Lens—supposed

Arthropod Characters. ‘ a ; i29
(4) The Nerve of the Parietal Hye gileill
(c) Relations of the Parietal Eye to the Parietal Stalk =! Le?

(d) Evidence of the Paired Origin of the Parietal Eye and Stalk 135
(e) Relations between the Parietal Stalk, the Epiphysis, and the

Brain ‘ ; 140

(f) The Paraphysis . . 148

(g) The Epiphysis of Sphewoda n and Raneeae . 145

(4) The Accessory Vesicle 2 : : . 146

4, GENERAL CONCLUSIONS . : : ‘ . 147
5, List or REFERENCES 148
6. DrscrIPTION OF FIGURES . : , : x AU

1. InrTRoDUCTION.

My attention was first directed to this investigation by ob-
serving a stage in the development of the parietal eye in an
Australian skink, in which the portion of the optic vesicle
behind the lens appears to consist of two very distinct layers,

VOL. 42, PART 2.—NEW SERIES. H

112 ARTHUR DENDY.

thus forming an optic cup structurally resembling that of the
ordinary paired eye. The conclusion that this cup had been
formed in the same way as in the paired eye was irresistible,
but fortunately I refrained from publishing until further
evidence was forthcoming, and in the meantime made arrange-
ments for securing a supply of Tuatara eggs, with a special
view to investigating this point. Thanks to the generous
assistance of Mr. P. Henaghan, I have obtained a consider-
able number of eggs, from which I have extracted a very good
series of embryos. I have recently published a general account
of the development of Sphenodon.! This will be followed
by more detailed accounts of the development of special
organs, by myself and other zoologists, and towards these the
present communication may be considered as a first instal-
ment.

My material was hardened in Kleinenberg’s picric acid,
followed by different grades of alcohol, and stained with borax
carmine in the usual way. The sections were cut by the
ordinary paraffin method.

Owing to the work having been carried out in Christchurch,
New Zealand, where, unfortunately, there is no proper
_ scientific library, the difficulties of the investigation have been
greatly increased. So much valuable work has been published
of late years on the ‘‘epiphysis” and parietal eye of various
types, that it was thought undesirable to publish my own
observations before having an opportunity of consulting the
more important recent papers on the subject by other writers.
In order to attain this end, great delay has been caused by the
necessity for communicating with England and America, but
it is hoped that this delay may be more than compensated for
by the increased interest attaching to the work when com-
pared with that of other investigators. As a result of this
comparison, I find that most of my conclusions have been
anticipated by one or other of the numerous writers on the
subject. Especially is this true in the case of the admirable
work of Béraneck (5, 6) on lizards, and Hill (16,17) on fishes.

1 This vol., p. 1.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 118

Sphenodon is generally recognised as being the most
ancient type of reptile still existing, and it might there-
fore well have been expected that the parietal eye in this
animal would be better preserved than in any other type.
Thanks to the classical works of Baldwin Spencer (338) and
de Graaf (15), this has long been known to be the case in
the adult, and I hope in the present memoir to be able to show
that its development has also undergone less modification than
in other reptiles, a fact which enables us to make a remark-
ably close comparison with the development observed by Hill
in fishes, and presents us with an important clue to the phylo-
genetic history of this remarkable organ.

As on so many previous occasions, I have to tender my
most sincere thanks to Professor G. B. Howes, LL.D., F.R.S.,
for his kindness in revising the proof sheets of these pages in
my absence from England, as well as for much valuable advice
and assistance in the matter of literature. I am also greatly
indebted to Dr. Gaskell, to Professor J. E. Reighard, and to
Professor W. B. Benham for sending me works which I was
unable to obtain in Christchurch.

The letters employed to designate the different stages are
the same as those of which I have already made use in my
general account of the development.

2. Systematic AcCOUNT OF OBSERVATIONS ON SPHENODON

PUNCTATUS.

Stage K.

The first indication of the parietal eye appears at the stage
in the development which I have called K, comparable with a
chick of about two days. The fore-brain is already bent
downwards and backwards, so that the mid-brain lies at the
anterior extremity of the body. The optic vesicles! of the

1 In the summary of my results recently sent to England I have, by an

unfortunate slip, stated (10) that the parietal eye first appears shortly after
the development of the optic lobes. I should have said optic vesicles.

114 ARTHUR DENDY.

ordinary paired eyes have begun to invaginate to form the
optic cups, while the lens already appears as a thickening of
the superficial epiblast.

A short way behind the ordinary paired eyes (actually in
front, owing to the cerebral flexure) a small round vesicle
appears as a bud on the roof of the fore-brain, slightly to the
left of the middle line. Owing to the cerebral flexure it
appears in sections on the ventral surface, as shown in fig. 2.
This vesicle I propose to term the primary parietal
vesicle. In the section figured, the roof of the fore-brain,
with the primary parietal vesicle, is cut almost tangentially,
in a direction not favourable for studying its histological
characters, and I therefore reserve the description of these
until later.

The position of the primary parietal vesicle to the left of the
middle line is very remarkable, and appears to be constant
(see also fig. 5, which is reversed as compared with fig. 2),
but the discussion of this problem must be postponed till the
sequel.

Stage L.

At Stage L, comparable with a chick of about three days,
the primary parietal vesicle has become conspicuous, when the
head of the embryo is examined in profile, as a small pro-
tuberance on the roof of the fore-brain between the paired
eyes. Longitudinal vertical sections of the head show it lying
a short way in front of the mid-brain (fig. 3). It still com-
municates by a wide aperture with the brain-cavity, but the
vesicle itself is somewhat flattened dorso-ventrally and extends
on both sides of this aperture, rather more forwards than
backwards (fig. 4), so that it appears to be attached to the
brain by a very short, hollow stem. Its wall is composed of
a layer of columnar cells similar to those forming the under-
lying wall of the brain, with large, oval, finely granular nuclei
and indistinct cell boundaries. The outer wall of the vesicle
lies immediately beneath the epidermis, and the mesoblast
cells are only just beginning to spread in between the two.
The cavity of the vesicle is filled with a finely reticulate

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 115

coagulum, in which at least one small nucleus is present
(fig. 4). A similar coagulum, with occasional nuclei, is visible
in the cavity of the fore-brain itself.

Stage M.

At Stage M, comparable with a chick at about the middle
of the fourth day, with the limbs just beginning to bud out,
the primary parietal vesicle still remains as a simple outgrowth
opening into the cavity of the fore-brain, slightly to the left
of the middle line (fig. 5).

Stage N.

The Parietal Eye and Stalk.—At Stage N, comparable
with a chick at about the commencement of the fifth day, a
great advance has taken place in the development of the parts
under discussion. Longitudinal vertical sections of the fore-
brain at this stage (figs. 7, 8) show the parietal eye as a hollow
oval vesicle lying on the roof of the fore-brain, in front and
slightly to the left of its so-called “ stalk.” The latter, which
I shall refer to in future as the “ parietal stalk,’ is a glove-
finger-shaped diverticulum of the roof of the fore-brain,
curving forwards and lying nearly or quite in the middle line.
The parietal eye appears to be nearly if not entirely separated
from the stalk. It lies wedged in between the superficial
epiblast and the roof of the fore-brain, and is somewhat com-
pressed dorso-ventrally ; its front or dorsal wall is already
thickening to form the lens (fig. 8). The cavity of the stalk
communicates freely with the cavity of the fore-brain. In
histological structure its wall closely resembles that of the
parietal eye, being composed of columnar cells similar to those
of the wall of the brain of which it is an outgrowth. Both
stalk and eye contain the usual coagulum.

It appears to me that there are two possible ways in which
the condition of the parietal eye and stalk at this stage may
have been derived from that which I have described as existing
at preceding stages.

(1) The primary parietal vesicle may have divided into two

116 ARTHUR DENDY.

parts, an anterior one forming the parietal eye and a posterior
one forming the stalk (or the eye may have budded out from
the stalk, which comes to much the same thing). This
appears to be the view generally adopted.

(2) It appears to me more likely that the parietal eye is
formed from the primary parietal vesicle, and that the parietal
stalk arises as a second, distinct evagination of the roof of the
fore-brain after the primary parietal vesicle has become almost
if not quite constricted off. This second evagination would
appear to take place slightly to the right of the primary parietal
vesicle, which seems to be pushed forward in the process.
This view of the case is strongly supported by the following
considerations :—(a) The primary parietal vesicle is already
almost constricted off from the brain, and has acquired the
characteristic flattened oval form of the young parietal eye
before the parietal stalk makes its appearance at all. At this
stage (fig. 4) it projects backwards as well as forwards from its
attachment to the brain, while the parietal stalk when it
appears projects forwards only. (6) The parietal eye, like the
primary parietal vesicle, at first lies to the left of the middle
line, while the parietal stalk is median, or nearly so.

The constancy of the left-sided position of the young
parietal eye will be discussed later on ; meanwhile I may state
that I have observed it in a sufficient number of cases to justify
me in attaching considerable importance to it, and I believe it
affords almost conclusive evidence in favour of my view that
the parietal stalk is a second and independent outgrowth of
the roof of the fore-brain. If its cavity ever communi-
cates with that of the parietal eye, which is doubtful, it is
probably because the second evagination involves the point of
attachment of the first to the brain before the opening of the
first has closed up.

The Paraphysis.—This structure does not appear until
some time after the appearance of the primary parietal vesicle.
It is certainly not present at Stage L, and I have not been

able to satisfy myself as to whether it is present or not at
Stage M.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 117

At Stage N, however, when the parietal eye and stalk are
both already clearly differentiated, the paraphysis appears as a
small, broadly rounded diverticulum of the roof of the brain,
close to the junction of the prosencephalon with the thalam-
encephalon. Owing partly to a not very well marked depres-
sion! of the roof of the brain just behind it, this diverticulum
already points backwards, as shown in fig. 6, towards the
parietal eye, from which it is separated by almost the entire
length of the roof of the thalamencephalon.

The cavity of the paraphysis on its first appearance commu-
nicates by a wide aperture with that of the fore-brain, and its
wall is composed of a layer of columnar cells similar to those
forming the wall of the brain in its immediate neighbourhood.
It is interesting to note that already a distinct blood-vessel is
conspicuous in sections just behind the paraphysis, lying on the
roof of the thalamencephalon (fig. 6).

Stage O.

The Parietal Eye and Stalk.—At Stage O, comparable
with a chick of about five and a half days, the parietal eye and
its stalk are conspicuous externally (as, indeed, they were
already at Stage N) when the head is examined as an opaque
object without staining. It will be seen, from an inspection of
fig. 12, that there is on top of the head a dark-looking, elon-
gated, quadrangular area,” tapering behind to an acute angle
which lies just in front of and between the optic lobes, while
anteriorly it ends in a less acute angle just behind and
between the cerebral hemispheres.

In the posterior angle of this area, just behind the paired
eyes and in front of the optic lobes, are visible two round spots
of an opaque white appearance. One of these lies just in front
of and slightly above the other, and also slightly to the left.

1 This depression, or slight fold of the brain-roof, appears to mark the

junction between prosencephalon and thalamencephalon, so that the para-
physis is to be regarded as an outgrowth of the hinder part of the prosen-
cephalon.

? This area belongs chiefly to the thalamencephalon, but its most anterior
portion, containing the parapbysis, belongs to the prosencephalon.

118 ARTHUR DENDY.

This is the parietal eye. It is further distinguished by the
presence of a distinct double circular outline, the inner circle
indicating the margin of the lens. The hinder of the two spots,
lying in the middle line or nearly so, represents the parietal
stalk, and, except that it sometimes appears dark in the centre
owing to its central cavity, it exhibits no further features of
interest.

It is important now to notice the position of the parietal eye
and stalk with regard to the paraphysis, shown also in fig. 12.
It will be seen that whereas the former occupy the posterior
angle of the roof of the thalamencephalon, the latter appears
as a round white spot between the cerebral hemispheres, being
separated from the parietal eye by almost the entire length of
the thalamencephalon. The same relations are shown in fig. 9,
which represents somewhat diagrammatically a longitudinal
vertical section of the head at this stage, the parietal eye being
shown as though it lay in the middle line instead of slightly to
the left.

The vesicle of the parietal eye is now somewhat deeper than
before, and the lens thicker (fig. 10). The ventral wall of the
vesicle is flattened against the roof of the thalamencephalon,
and its posterior wall against the anterior wall of the parietal
stalk. A few mesoblast cells have made their way in between
the lens and the superficial epiblast (epidermis), which appears
in the section figured to be slightly depressed just above
the parietal eye (probably owing to shrinkage in the course of
preparation).

The cavity of the parietal stalk still opens into that of the
fore-brain, but the section passes a little to the left of the
opening. Its front wall is much flattened against the posterior
wall of the parietal eye, so that it appears to project vertically
upwards from the roof of the brain.

Except for the thickening of the lens the wall of the parietal
eye at this stage has undergone no marked alteration. The
lens (fig. 19) is made up of more or less elongated columnar
cells with correspondingly elongated nuclei. These cells are
longest in the middle and much shorter towards the periphery

PARIETAL BYE AND ADJACENT ORGANS IN SPHENODON. ~— 119

of the lens. Here and there a clear-looking space appears be-
tween the elongated cells, containing what seems to be a much
smaller nucleus.

In the cavities of both stalk and eye the usual coagulum is
present with an occasional small nucleus. These nuclei pre-
sent a curiously lobed or agglomerated appearance, and appear
to me to be thrown off from the surrounding walls (figs. 19,
20, Nu.). In fig. 19 one appears to be just leaving the inner
surface of the lens. They are possibly identical with the
smaller nuclei noticed above in the interior of the lens, and
which also occur in the ventral wall of the eye, and perhaps in
the wall of the stalk, but I have not sufficient evidence to
decide this point. Karyokinesis is going on both in the wall
of the stalk and of the eye.

Just behind the parietal stalk the roof of the brain has
already become differentiated into two layers (fig. 10), the
outer finely punctate in longitudinal section and almost without
nuclei, the inner with numerous large oval nuclei.

The Paraphysis.—At this stage the paraphysis has elon-
gated backwards, and its wall has begun to be irregularly folded
(fig. 11). Its cavity still communicates freely with that of the
brain, and the blood-vessel is still conspicuous just behind its
apex.

I have already pointed out in dealing with the parietal eye
and stalk the relative positions of these organs and of the
paraphysis, which are readily seen on examining the upper
surface of the head of embryos at this stage (fig. 12).

Stage P.

Of Stage P I have only a single specimen, and it is so
similar in external characters to Stage O that I have not
thought it necessary to cut sections at present, especially as
its histological condition is doubtful.

Stage Q.
Of Stage Q I have also only a single specimen, which
was dead when it reached me, and useless for histological
purposes.

120 , ARTHUR DENDY.

Stage R.

Of Stage R, however, I have numerous examples, as the
embryo passes a long period in this condition. According to
my classification of the embryos, this is the last stage but one
before hatching. The embryo is very advanced. The limbs
and digits are well formed, and pigment is present in the
integument, forming a characteristic embryonic pattern of
transverse bands and longitudinal stripes.

The upper surface of the head of an embryo at about the
commencement of Stage R is represented in fig. 13. When
this figure is compared with fig. 12 it will be seen that owing
to the straightening out of the cerebral flexure the optic lobes
and cerebral hemispheres have become closely approximated,
and the length of the thalamencephalon thereby greatly
reduced, its roof appearing as a dark-looking, diamond-shaped
area between the two.! In the centre of this area the parietal
eye appears as a round white spot with an intensely black
border, the white centre representing the lens, and the black
border the now deeply pigmented margin of the retina. As
far as one can judge, the parietal eye has now become median
in position. |

Topography of the Brain inthe Neighbourhood of
the Third Ventricle.—As the brain has progressed greatly
in development since the last stage described, it will be
desirable to briefly describe the topographical relations of the
parts in the neighbourhood of the third ventricle before
proceeding any further, and this may best be done with the
aid of a series of transverse sections.

Fig. 22 represents a transverse section taken a short way in
front of the foramina of Monro. It shows the anterior
commissure (Com. Ant.) passing across above the anterior end
of the third ventricle. Above the anterior commissure is a
crescentic band of fibres (Com. Man.) uniting the two inner

1 This dark patch, however, does not exactly mark the limits of the roof of
the thalamencephalon, for it will be shown later on that the parietal eye lies

over the hinder part of the roof rather than over the centre as it appears
externally.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 121

mantle walls of the cerebral hemispheres. I identify this with
the mantle commissure or corpus callosum mentioned by
Hoffmann,! and I think a comparison of his fig. 4, pl.
clxill, representing a corresponding section of an embryo
of Lacerta, leaves no doubt as to the correctness of this
identification. Beneath the third ventricle is seen the optic
chiasma (Op. Ch.).

The section represented in fig. 23 passes through the
foramina of Monro, and is drawn to show especially the
choroid plexus of the lateral ventricle on either side springing
from a common origin between the inner mantle walls of the
cerebral hemispheres.

Fig. 24 represents a similar section immediately behind the
foramina of Monro, and the only feature about it which calls
for special comment in this place is the presence of a con-
spicuous transverse commissure (Com. F.) running across
above the third ventricle from the inner mantle wall of one
cerebral hemisphere to that of the other. This commissure,
here very strongly developed, appears to be identical with the
rudiment of the fornix discovered in 1881 by Rabl-Riickhard
(27) in Psammosaurus.” I shall refer to it in future as the
Commissura fornicis.

Fig. 25 represents a transverse section across the third
ventricle in the plane of the parietal eye.? The ventricle
itself (fig. 3) is partially divided into two parts, upper and
lower. The lower is a deep, slit-like space, bounded on either
side by the thick optic thalamus. The upper is a much wider
cavity, having its roof formed by the thin choroid plexus, and
its floor by the two ganglia habenule (Gang. Hab. R. and L.),
which lie upon the right and left optic thalami respectively,
on either side of the slit-like opening into the lower division of
the ventricle. There is no noticeable difference as regards
size between the two ganglia habenule. On either side of the

? (18), pp. 1977, 1979, et seq.

2 Vide also Hoffmann (18), pp. 1975 et seq.

3 The slightly asymmetrical position of the eye itself is evidently due to an
artificial crease in the integument.

122 ARTHUR DENDY.

thalamencephalon one of the cerebral hemispheres (C. H.),
with its large lateral ventricle (L. V.), is cut through.

Fig. 26 represents a section through the hinder part of the
third ventricle, the upper and lower divisions of which are seen
to be completely separated from one another in this region by
the development of a strong band of commissural fibres (Com.
Sup.) which runs across from side to side, between and beneath
the hinder ends of the two ganglia habenule. This commissure
I identify with the ‘‘superior commissure” described by
Burckhardt (9) in Lacerta, and by Hill (17) in various bony
fishes. Hull further identifies it with the superior commissure
found by Osborn! in Amphibia, and with a commissure found
by Balfour? in Elasmobranchs. Its position corresponds very
closely with what Béraneck (5) calls the “‘ parietal centre,” and
one can hardly help identifying it with this structure as
represented in his fig. 5.

Owing to its very strong development in Sphenodon, [ at
first mistook the superior commissure for the posterior com-
missure, which at this stage is very feebly developed.®

The upper division of the third veutricle is bounded beneath
by the gangha habenule, and arched over by the here much
folded choroid plexus (Ch. P.). The lower division is bounded
as before by the two optic thalami, and beneath it lies the
infundibulum (Jnf.).

Fig. 27 represents a transverse section through the entrance
to the iter, and shows the very feebly developed posterior
commissure (Com. Post.) running across just above the latter.
The extension of the upper division of the third ventricle
(V. 3), which runs back above the superior commissure, is cut
through between the anterior extremities of the two optic

lobes (O. L.).

1 «Journal of Morphology,’ vol. ii, pp. 51—96 (quoted by Hill).

? Vide Balfour’s ‘Comparative Embryology,’ vol. ii, p. 356, fig. 254.

3 This mistake occurs in my summary of results (10), in which the inferior
commissure is referred to as the posterior. I shall show later on, however,
that there is some reason for believing that the inferior commissure may unite
with the posterior commissure in the adult.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 1238

With the foregoing description of a series of transverse
sections should be compared the sagittal section represented
in fig. 15, especially with regard to the position of the
commissures. This figure shows very clearly the strong
upward arching of the thin roof of the third ventricle, which
is doubtless due to the compression of the thalamencephalon in
the straightening out of the cerebral flexure. The thin-walled
upward extension of the third ventricle thus caused is bounded
anteriorly by the commissura fornicis (Com. F.), and pos-
teriorly by the superior commissure (Com. Sup.).

The Parietal Stalk.—The parietal stalk (figs. 14, 15, 16,
27, Pa. 8S.) has now become considerably elongated. It is
attached to the roof of the brain, immediately behind the
superior commissure, and in front of the posterior commissure
(fig. 15). From its point of attachment it first runs upwards and
slightly backwards ; it then curves forwards over the bay formed
by the roof of the third ventricle, to which its anterior wall is
closely appled. Thus, in a transverse section, such as is
represented in fig. 27, the proximal part of the stalk is cut
longitudinally and tangentially below the thin-walled upper
division of the third ventricle, while the distal portion is cut
transversely above it.

At the commencement of Stage R the blind distal extremity
of the parietal stalk ends in a not very strongly marked and
somewhat flattened expansion close beneath the parietal eye
(fig. 14). In somewhat older embryos of the same stage it
ends in a glove-finger-shaped extremity separated by a con-
siderable interval from the parietal eye (figs. 15, 16), this
separation being due apparently to a forward shifting of the
parietal eye itself.

At its proximal end the parietal stalk becomes greatly
narrowed, as if by compression from the strong development
of the superior commissure in front of it; and its lumen, else-
where wide, is here completely obliterated, though its walls
can still be traced into direct continuity with the epithelial
lining of the brain. Except at its proximal extremity the wall
of the parietal stalk is thick and densely packed with large

124, ARTHUR DENDY.

oval nuclei, between which, on the inner aspect, the boun-
daries of columnar cells are faintly discernible. In one late
embryo of this stage I have observed the wall of the parietal
stalk to be clearly differentiated into two layers (fig. 21), an
outer thinner one composed of a single layer of short columnar
cells, and an inner very much thicker one, so densely crowded
with large oval nuclei arranged in many but irregular tiers,
that it is impossible to make out the cell boundaries satis-
factorily.

Towards the end of Stage R a patch of black pigment is
deposited in the antero-ventral wall of the parietal stalk, near
its distal extremity (figs. 15, 16, 21). This pigment resembles
that of the parietal eye, having the form of small black
granules lying between the innermost cells of the wall, next to
the lumen. It appears to be constant, for I have observed it
both in transverse and longitudinal sections of embryos at
this stage. De Klinckowstrém (19) has observed pigment in
the same situation in Iguana at a certain stage of develop-
ment, but apparently disappearing later. In Sphenodon
also it would seem to disappear in the adult, for none is shown
in Speucer’s figure (33). Ritter also (29) in Phrynosoma
coronata has observed pigment in what he terms the
‘‘epiphysial vesicle,’ which appears to be homologous with
the distal extremity of the parietal stalk in Sphenodon.

The Parietal Eye.—At the commencement of Stage R
we see the parietal eye in sagittal section lying over the
hinder part of the thalamencephalon, and above the flattened
distal extremity of the parietal stalk (fig. 14). The end of
the stalk lies wedged in between the eye and the brain, and
between the stalk and the eye there is a very thin layer of
connective-tissue cells. Above the eye, between it and the
superficial epidermis, is a rather thick layer of connective
tissue, forming part of the dermis, which, however, is here
thinner but denser than elsewhere. The upper surface
of the lens is pressed close against the under surface of the
dermis.

In the eye itself important changes have taken place. The

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 125

lens is very sharply marked off from the rest of the wall of the
vesicle. The latter, in fact, forms a thick-walled cup with a
sharp inturned edge, which appears to grasp the margin of the
lens just where the less convex outer surface of the latter joins
the more convex inner surface.

Histologically the lens shows little or no change. The rest
of the wall of the vesicle, however, forming what we may
fairly call the optic cup, has undergone marked alterations.
In the first place it has become distinctly differentiated into
two layers, an inner (Jn. W.), next to the cavity of the vesicle,
and an outer (Out. W., fig. 14). The inner layer is very much
thicker than the outer, and the two show a strong tendency to
separate from one another, except round the margin of the
cup, where no such separation takes place. The outer layer
is a single layer of rather short columnar cells with large
elongated oval nuclei; it contains no pigment. The inner
layer shows a series of columnar cells next to the cavity of the
vesicle, with very faint cell boundaries and large oval nuclei,
while its deeper part is densely packed with large oval nuclei
in several tiers, between which no cell boundaries can be
made out. The deepest nuclei, nearest to the outer wall of
the cup, are smaller than the others, but except for this no
differences can be made out between them. The nuclei are
not arranged in distinctly recognisable layers.

Small granules of black pigment, arranged in radiating lines,
are now conspicuous between the cells of the inner layer of the
wall of the optic cup, but almost confined to the inner half of
its thickness, and most abundant in the deeper part of that
half, i.e. about the middle of the inner layer. Round the
margin of the cup, however, the pigment is especially dense,
and lies more on the inner surface next to the cavity of the
vesicle.

Between the outer and inner layers of the wall of the optic
cup, in the space formed by their slight separation, a small
quantity of finely granular, non-staining material is present.
This appears to correspond to Spencer’s “ molecular layer,”
which in the adult parietal eye seems still to divide the retina

126 ARTHUR DENDY.

into an outer thinner and an inner thicker layer, as shown in
Spencer’s figures (38).

Towards the end of Stage R, as already noted, an important
change has taken place in the relative positions of the parietal
eye and stalk. The eye no longer lies over the end of the
stalk, but appears to have shifted forwards, so that it lies
slightly in front of the middle of the thalamencephalon and
over the tubules of the paraphysis, being separated from the
end of the parietal stalk by a wide interval. The relative
positions of the parts under discussion at this period are shown
in fig. 15, representing a longitudinal vertical section through
the thalamencephalon in the median plane. With the excep-
tion of the forward shifting of the eye the relations are much
the same as in the earlier part of Stage R.

As regards the parietal eye itself, the changes are not very
great as compared with younger embryos of this stage. The
optic cup, however, is deeper than before, and its axis is in-
clined backwards and downwards (fig. 16). The retina is still
clearly differentiated into two layers, as shown in figs. 16 and
17, and as described in the younger embryo. The black pig-
ment (Pig.) is more abundant, but is sti]] almost confined to
the inner half of the inner layer of the wall.

In both the earlier and later periods of Stage R a consider-
able amount of coagulated humour is present in the cavity of
the parietal eye, chiefly adhering to the inner surface of the
retina, as shown in fig. 17 (Coag.), with occasional nuclei, as
shown in fig. 18. It is seen also in much smaller quantity
adhering to the inner surface of the lens. In figs. 14, 15, and
16 it is omitted for the sake of clearness.

The Nerve of the Parietal Eye.—By far the most
important change which marks the later part of Stage R is
the appearance of the nerve of the parietal eye. This is seen
in longitudinal vertical sections as a faintly staining, deli-
cately fibrillated band attached somewhat posteriorly to the
outer layer of the retina, as shown in fig. 16 (Pa. N.). From
its point of attachment it runs backwards and downwards be-
tween the distal part of the parietal stalk and the tubules of

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 127

the paraphysis. I have not been able to trace it to a connec-
tion with the brain, but it certainly does not appear to be
connected with the apex of the parietal stalk. It is probably
not connected with the parietal stalk at all, though it lies
close under the antero-ventral surface of the latter. It
appears to contain elongated nuclei, but I am not quite certain
that these may not belong to surrounding connective-tissue
cells.

I suspect that the nerve of the parietal eye grows downwards
and backwards from the outer layer of the retina, and ulti-
mately becomes connected with the roof of the thalamen-
cephalon. If this should prove to be the case we shall have
here another noteworthy resemblance between the parietal
eye and the ordinary paired eyes.

The Paraphysis.—The relation of the paraphysis to the
other parts of the brain towards the end of Stage R is best
shown in fig. 15. It will be seen that the paraphysis arises
from the roof of the brain just in front of the commissura
fornicis. It curves upwards and backwards over the now
strongly arched and more or less folded roof of the third
ventricle. Its walls, which consist of short columnar cells
arranged in a single layer, are greatly folded and produced
into numerous blind diverticula, which form a mass of some-
what convoluted tubules lying in front of the parietal stalk
and beneath the parietal eye. Intermingled with these tubules
are numerous blood-vessels, easily recognisable by the cor-
puscles which they contain, and the whole is bound together
by loose connective tissue. Histological details are shown in
fig. 16. It is worthy of note that the blood-corpuscles in this
region contain numerous small black pigment granules similar
to those which occur in the retina of the parietal eye and in
the parietal stalk. These pigment granules occur also ina
blood-vessel lying above the end of the parietal stalk, as shown
in fig. 16 (compare Bernard [7}).

The paraphysis at this stage may also be advantageously
studied in transverse sections. Thus fig. 23 shows that it
originates from the same point as the choroid plexuses of the

VOL. 42, PART 2,—NEW SERIES. I

128 ARTHUR DENDY.

lateral ventricles, between which its opening (O. Par.) is
situated. Fig. 24 shows the paraphysis (Par.) cut through
above the commissura fornicis. Fig. 25 shows it cut through
above the roof of the third ventricle, and below the parietal
eye, and fig. 26 shows it cut through just in front of the
parietal stalk.

Accessory Vesicle.—At the commencement of Stage R
the tubules of the paraphysis do not extend so far backwards
as is represented in fig. 15, and between them and the parietal
stalk there lies a wide space of irregular shape (fig. 14, Ac. V.).
This space I propose to term simply an accessory vesicle;
its possible homologies will be discussed later on. It is at
once distinguished from the paraphysial tubules by the histo-
logical characters of its walls, which are composed of much
thinner epithelium, though not so thin as those of the blood-
vessels. It is further distinguished from the blood-vessels by
the absence of corpuscles from its cavity. It gives off a few
irregular diverticula or sacculations, but I have not been able
to detect any connection between its lumen and the cavity of
the third ventricle, nor does it appear to communicate with
either the paraphysial tubules or the blood-vessels. Although
so conspicuous at the commencement of Stage R, it has
completely disappeared in later embryos of this stage, its
place being taken by paraphysial tubules (figs. 15, 16).

Stage 8.

Of this stage, the last before hatching, according to my
classification, I have received only dead specimens, useless for
histological purposes.

In the adult Tuatara the parietal eye is no longer recognis-
able externally, although still very highly organised; but,
according to Thomas (88), in the recently hatched Tuatara it
still shows as a dark spot through the translucent skin over
the parietal foramen.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 129

3. Discussion oF RESULTS.

(a2) Structure of the Parietal Eye.

The Retina.—Certainly the most remarkable feature about
the retina of the parietal eye in Sphenodon is its differentia-
tion at Stage R into two very distinct layers, as shown in figs.
14—17. It was this feature, which I first observed in an
Australian skink embryo (Hinulia, sp.), which led me to
undertake the present investigation. The parictal eye at a
certain stage thus acquires an extraordinarily close resemblance
to the ordinary paired Vertebrate eye, consisting of a two-
layered optic cup grasping a cellular lens within its margin.
The developmental history shows, however, that this result is
arrived at in a totally different way in the two cases, for in
the parietal eye there can be no doubt that the lens is formed
simply by thickening of the front wall of an originally single-
layered vesicle. It will also be observed that the situation of
the pigment is different in the two cases, being in the inner
layer, next to the cavity of the cup, in the parietal eye, and
in the outer layer, away from the cavity, in the paired eye.
Baldwin Spencer (33) has also shown that the rods lie next to
the cavity in the parietal eye, instead of being formed from
the outermost part of the inner layer as in the paired eye.

It appears to me, from what Baldwin Spencer says of the
structure of the retina, that even in the adult parietal eye of
Sphenodon we can still trace the two embryonic layers.

This author describes no less than six series of histological
elements in the retina of the adult eye. His “ molecular
layer, consisting of fine punctated material, which takes the
stain (hematoxylin) with difficulty,” evidently marks the
embryonic separation of the wall of the optic cup into two
primary layers, as I have noticed previously. In the adult
the inner of the two primary layers is still much thicker, and
appears to have become differentiated into (1) a layer of rod-
like bodies enveloped in deep pigment, evidently formed from
the innermost layer of columnar cells which I have described
in the embryo, with the pigment granules between them ;

130 ARTHUR DENDY.

(2) a double or triple row of spherical nucleated elements,
apparently formed from the deeper cells of the inner embryonic
layer.

The outer of the two primary layers, consisting in the
embryo of a single layer of short columnar cells, appears to
have given rise to spherical elements, cone-shaped bodies and
spindle-shaped elements according to Spencer, but Hoffmann
(18) only recognises one layer in place of these three.

The Lens.—In the adult Sphenodon the lens of the
parietal eye has become much more strongly convex on its
inner aspect (by elongation of the columnar cells occupying
its centre) than it is in the latest embryo examined. In
advanced embryos the lens (figs. 14—16) is much more
sharply marked off from the retina than is indicated in Spencer’s
figure, and I have noticed the same fact in the development
of the parietal eye in an Anstralian Hinulia. Beard (4) has
also called special attention to the sharp delimitation of the
lens from the retina in Anguis, in support of de Graaf’s
original description (15).

Supposed Arthropod Characters.—The “ humour,”
so frequently noticed as forming a coagulum in the cavity of
the parietal eye vesicle, apparently persists in the adult, as
indicated in Spencer’s fig. 8. As a result of his observations
on Ammoceetes, Gaskell (12) has put forward what I cannot
help regarding as a mistaken view as to the nature of this
humour, at any rate so far as the types examined by me are
concerned. He says, “Everything seems to me to point to
the conclusion that the appearance of a large central cavity
[in the parietal eye] is brought about by the partial degenera-
tion of elements which originally filled it, and that their
remains have given rise to the impression held by Spencer and
Beard, that a large cavity exists which is filled by a coagulated
albuminous fluid.” Gaskell accordingly claims an Arthropod
structure for the parietal eye of vertebrates.!

It appears to me, however, that the observations of Beard

1 T learn from a paper by Prenant (26) that Leydig has also compared the
parietal organs with Arthropod eyes.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 181

(4), Spencer (84), Studniéka (85), and others, upon the
structure and development of this organ in Cyclostomes, as
well as those of the numerous observers who have studied it
in Lacertilia, afford conclusive evidence against Gaskell’s
views; and the weight of this is now only increased by my
own observations on Sphenodon, in which animal the
parietal eye certainly exhibits no trace of an Arthropod
character.

(6) The Nerve of the Parietal Hye.

The real nerve of the parietal eye appears, so far as I can
learn, to have been first discovered by Béraneck. I regret that
I have been unable to consult his earlier work on the subject,}
but fortunately he has given an excellent account of his ob-
servations, together with a discussion of the question, in a
later paper (5), which I have been able to obtain. From this
I learn that Francotte, Strahl, and Martin have also observed
this nerve in Lacertilia. It appears, according to Béraneck,
that in Anguis fragilis the nerve is connected distally with
the retina of the parietal eye, and proximally with the roof of
the brain at a point which corresponds very closely with the
superior commissure as described by Hill (1%) in fishes, by
Burckhardt (9) in Lacerta, and by myself in Sphenodon,
but which Béraneck terms the ‘centre ou noyau pariétal.”
De Klinckowstrém (19) has also described the nerve of the
parietal eye in Iguana, and states that the “ parietal centre”
from which it originates is situated asymmetrically on the
right side of the origin of the “ epiphysis” (= parietal stalk).
In this connection it is worthy of note that Gaskell (12)
describes the nerve of the parietal eye in Ammoceetes as being
connected proximally with the right ganglion habenule.

Baldwin Spencer (83), as is well known, believed the nerve
of the parietal eye in Sphenodon to be formed from the
distal end of the “epiphysis” (parietal stalk). This con-
clusion, strongly supported by Hoffmann (18), has, I under-
stand, been already challenged by Leydig, and my own

1 « Ueber das Parietalauge der Reptilien,” ‘ Jenaische Zeitschrift,’ 1887.

132 ARTHUR DENDY.

observations certainly seem to indicate that such an interpre-
tation is incorrect. As already stated, I found at Stage Ra
nerve which appears to correspond exactly so far as observed
with that described by Béraneck. It passes in front of and
beneath the distal extremity of the parietal stalk, and becomes
connected with the outer layer of the retina (figs. 15 and 16).
Unfortunately I have been unable to trace it to its point of
origin from the roof of the brain, but I have little doubt that
the connection takes place at or near a point immediately in
front of the origin of the parietal stalk, corresponding to what
Béraneck terms the parietal centre, and lying between the
posterior ends of the two ganglia habenule, above the fibres of
the superior commissure (fig. 26). It seems probable that the
nerve is actually connected with one or other of the ganglia
habenule, as in Ammoceetes, but I have been unable to detect
any such difference in size between the two ganglia as occurs
in the lamprey. The fact that the parietal eye itself in
Sphenodon originates on the left side of the middle line
seems to make it probable that the nerve is connected with
the left ganglion habenule. The observations of de Klinckow-
strém, however, point to the right ganglion. Possibly the
nerve crosses over from right to left, but this is a question
which cannot be decided in the present state of our knowledge,
and one which is well worthy of further investigation. We
may safely take it as an established fact, however, that the
parietal eye has a special nerve which lies in front of and is
not derived from the parietal stalk (‘ epiphysis”’).

(c) Relations of the Parietal Eye to the Parietal
Stalk.

Embryologists who have studied the question appear to be
divided into two schools on this subject. The first and older
school maintain that the parietal eye is either the distal
extremity of the stalk (“epiphysis”) separated from the
proximal, or that it is an outgrowth or diverticulum of the
stalk. The second and more modern school hold that the

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 183

parietal eye and the stalk originate as separate outgrowths
from the brain.

As one of the earliest of the first-named school I may cite
McKay (25), whose valuable work appears to have been over-
looked by many of the more recent writers. The figures given
by McKay of the early stages in the development of the
parietal eye in Grammatophora agree very closely on the
whole with what I have observed in Sphenodon, but he
interprets the appearances as follows :—“ The epiphysis cerebri
or pineal gland arises, as is seen, as an outgrowth of the
thalamencephalon. At this stage the outgrowth is composed
of a single layer of columnar cells with well-marked nuclei
(fig. 1). Second stage.—In the next stage the evagination or
vesicle undergoes the following changes. The anterior wall
begins to grow forward, and this soon leads to the formation
of a second evagination in the wall of the primary one (fig.
2, Pn.). Thus we have two vesicles formed, an anterior
larger (Pn.), destined to become the pineal eye, and a posterior
smaller one (Ep.). Since the anterior vesicle grows faster
than the posterior it bends forwards, and its inferior wall
rests on the superior surface of the columnar cells of the
thalamencephalon (fig. 2). The walls of both vesicles are
composed of a single layer of columnar cells with oval nuclei.”
Of his third stage he observes, “The anterior of the two
vesicles becomes constricted off to form the pineal eye (fig.
4, Pn.); while the posterior remains as the end of the epi-
physis (fig. 4, Ep.).”’

Unfortunately I have been unable to consult Hoffmann’s
original memoir! on the development of Lacerta, but this
author has also described and illustrated his results in his
memoir on the Reptilia in Bronn’s ‘ Klassen und Ordnungen
des Thier-reichs’ (18). From this description it appears that
in Lacerta also the early development of the parietal eye and
stalk is very similar to what takes place in Sphenodon and
Grammatophora. A hollow outgrowth is budded off from

1 «Weitere Untersuch. zur Entwicklungsgesch. der Reptilien. Morphol.,’
Jahrb. xi, 176 (quoted by McKay).

134 ARTHUR DENDY.

the hinder part of the roof of the thalamencephalon. This
vesicle, according to Hoffmann, becomes divided by a con-
striction into two parts, anterior and posterior, of which the
anterior develops into the parietal eye and the posterior into
the “eigentlichen Epiphyse” (= parietal stalk). Later on
the parietal eye vesicle no longer lies in the same plane as the
“epiphysis,” but somewhat obliquely in front and to the
right of it.

As a representative of the second school I may mention
Béraneck, who in his later papers (5 and 6) has contended very
vigorously for what he terms the individuality of the parietal
eye. He maintains that the two vesicles, observed by all who
have investigated the subject, originate independently of one
another from the brain, and that therefore the parietal eye is
not formed from the “epiphysis.” In support of this view
he lays special stress upon the fact that the nerve of the
parietal eye is developed independently of the so-called
epiphysis or parietal stalk, and not at the expense of the latter
as was formerly supposed.

In spite of the arguments advanced against Béraneck by de
Klinckowstrém (19), who has investigated the development of
the parietal eye and “ epiphysis” in Iguana, I must declare
myself strongly in favour of the “individuality” of the parietal
eye. Béraneck (6) has thoroughly argued the question in his
reply to de Klinckowstrém, and I think he has hit upon the
true explanation of the diversity of opinion in the following
passage :—“ Ces divergences assez importantes s’interprétent
facilement avec Vhypothése que l’ceil pariétal et l’épiphyse
sont des évaginations distinctes duthalamencéphale. En effet,
ces deux organes peuvent avoir un développement simultané a
successif, suivant certaines conditions embryogéniques, sans
que leurs caractéres morphologiques en soient pour cela
altérés.”

I have already given my arguments for believing that in
Sphenodon the parietal stalk and eye originate indepen-
dently. Béraneck (6) believes that the primitive evagination
of the brain described by Francotte in Anguis corresponds

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 185

not to the “ pineal gland,” but to the parietal eye only. This
is exactly my view with regard to this organ in Sphenodon,
in which it appears that the “stalk” or “ pineal gland” or
“epiphysis” (so called) develops a little later than the parietal
eye itself. The eye, it will be remembered, first appears as a
small round vesicle, which I have termed the “ primary pari-
etal vesicle,’ budded out on the left side of the middle
line. ‘This vesicle, according to my interpretation, soon
becomes completely closed, and a second vesicle appears as an
outgrowth of the roof of the fore-brain, beneath and slightly
to the right of the primary vesicle, pushing the primary
vesicle forwards. This second vesicle forms the ‘parietal
stalk.” For a long time its cavity retains free communica-
tion with the cavity of the fore-brain, but at Stage R it has
become completely shut off from the latter, a fact
which will be seen to have considerable importance when we
come to consider the relation which the parietal stalk bears to
the “epiphysis ” of the adult.

(qd) Evidence of the Paired Origin of the Parietal
Eye and Stalk.

The evidence in favour of the originally paired character of
the parietal eye in Sphenodon is derived principally from
the fact that the eye itself arises on the left-hand side of the
middle line, while the “ parietal stalk” appears almost or quite
in the middle line, and therefore a little to the right of the
parietal eye. These relations appear to be very constant.
Thus I have observed the first origin of the primary parietal
vesicle (= parietal eye) to the left of the middle line in the
only two embryos of which preparations suitable for demon-
strating this point have been made (vide figs. 2, 5) ; and later
on, when the parts in question have become conspicuous
externally, I have observed the young parietal eye lying to
the left of the stalk (fig. 12) in nearly every embryo examined,
though by the time Stage R is reached, as already noticed, it
appears to have become median (fig. 18).

136 ARTHUR DENDY.

As this is a point of some importance it may be well
to enumerate all the cases observed belonging to Stages N, O,
and P.

( Embryo 93
» 95 & Parietal eye lying to left of stalk.
» 96

» 94 Parietal eye lying directly in front of stalk.
» 97 Parietal eye lying pretty well in front of stalk, but
both to left of middle line.

1
|
1
iS
cee)
G
Stage O 2 s Parietal eye lying to left of stalk.

Stage P 5 87 Parietal eye lying well to left of middle line, stalk
indistinct (externally).

At very early stages in the development, immediately ante-
cedent to the first appearance of the primary parietal vesicle,
the roof of the fore-brain is in a markedly asymmetrical
condition, the left half overlapping the right for some distance.
This overlap is shown in fig. 1, in the region between the
optic vesicles of the ordinary eyes; it also extends to the
region in which the primary parietal vesicle will shortly
appear. It will be found more fully illustrated in my memoir on
the general development of Sphenodon. Granting that the
right parietal eye was once well developec, it seems not
unnatural to associate its suppression with this remarkable
overlapping of the left side of the roof of the fore-brain, which
at first seemed to me to be due to restraint exercised by the
very early developed pro-amnion upon the growing brain; but
the probability of this explanation is greatly diminished by
the fact that a very similar asymmetry of the developing
parietal organs occurs in fishes, as described by Hill (16, 17).
The part of the brain roof from which the right parietal eye
should arise is covered over at first by the overlap, which may
perhaps be supposed to retard its development.

If the right parietal eye ever develops at all it must be
represented by the parietal stalk, and there is certainly good
ground for considering this to be the case. In the first place
the parietal stalk originates, if I am right in my interpretation

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 187

of the development, in precisely the same manner as the left
parietal eye, of which it is quite independent. In the second
place it has for some time a very similar structure, though it
never acquires the same degree of perfection.

The similarity in structure of the parietal eye and the
parietal stalk is a point on which I am inclined to lay
considerable stress as an argument in favour of their paired
origin. Both have the form of hollow vesicles, which become
completely shut off from the brain cavity. Except for the
fact that no lens thickening is formed in the case of the
“stalk,”’ the histological characters of the two are, up to a
certain stage at any rate, identical. Fig. 21 represents a
transverse section through the “stalk” at Stage R, taken
near its distal extremity, just at the place where pigment is
deposited in its ventral wall as already mentioned. It will be
seen that the wall is divided into two layers, as in the left
parietal eye,—an outer thin one composed of a single tier of
short columnar cells, and an inner very much thicker one
containing many large oval nuclei not regularly arranged in
tiers, and with a layer of columnar cells next to the central
cavity. The pigment is deposited between the cells of the
inner half of the inner layer of the ventral wall, exactly as in
the left parietal eye. On comparing this figure with the
sections of the left parietal eye represented in figs, 14, 16, and
17, and bearing in mind its developmental history, the con-
clusion that the so-called ‘‘ parietal stalk ” or “ epiphysis ” in
Sphenodon represents a right parietal eye appears to me
inevitable.

The probability of this conclusion is greatly strengthened
when we come to compare certain observations of other
writers dealing with other types, by far the most important of
which are those of Hill (16, 17) on certain Teleostean fishes,
and on Amia.

In 1891 Hill described (16) in Coregonus albus two
“‘ epiphysial outgrowths” from the roof of the primary fore-
brain. He says, “On the roof of the brain—in the median
line, and in a plane passing through the middle of the optic

188 ARTHUR DENDY.

vesicles—is seen the posterior epiphysial outgrowth. It is
a small spherical body, having its lateral walls thickened so
that the cavity within it is laterally compressed. This cavity
is narrowest at the middle, on account of the greatest thickness
of the lateral wall of the vesicle falling at the middle of its
antero-posterior axis; consequently, in a dorsal view the
cavity has the form of a dumb-bell. Just in front of this
vesicle, and a little to the left of it, is a second similar
outgrowth. This anterior evagination is smaller than the
posterior one, and appears to be solid. It lies close against
the wall of the posterior vesicle, and is partly hidden by it.”

At earlier stages Hill found that each vesicle arises as a
separate outgrowth from the roof of the brain, and each contains
a cavity which opens separately into the cavity of the brain. -
“The anterior vesicle shows an increase in size for about
twenty days, and after that a decrease, while the posterior
vesicle shows from the beginning a gradual increase.” The
author further compares these two “ epiphysial outgrowths”
with those described by Leydig! in Lacertilia, and points out
‘‘ that while the early stages of these two epiphysial outgrowths
of Coregonus agree in many details with the corresponding
early stages of the two outgrowths in Lacertilia, as described
by Leydig, yet the ultimate fate of these two outgrowths in
the two forms is widely different. In Coregonus the anterior
outgrowth, which is the smaller, gradually disappears, while
in Lacertilia, according to Leydig, it develops into the adult
parietal organ.” He further thinks it probable that the two
outgrowths will be shown to be homologous with the primary
and secondary parietal vesicles described in adult Petromyzon
by Ahlborn (1), and in Lacertilia by Ritter (29).

In a later paper (17), published in 1894, Hill continues his
excellent researches on this subject, and arrives at closely
similar results in the case of various fishes. He shows that
there are two epiphysial outgrowths from the roof of the

1 « Tas Parietalorgan der Amphibien und Reptilien,’’ ‘ Anatomische-histo-

logische Untersuchung,’ Senckenberg. Naturf. Ges., Band xvi, p. 441. I
regret that I have been unable personally to consult this important memoir,

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 189

primary fore-brain of Salmo, Catostomus, Stizostedion,
Lepomis, and Amia. The anterior of these two is rudi-
mentary, and appears to lie constantly to the left of the
posterior. He concludes that the anterior epiphysial vesicle
is homologous with the parietal eye of Lacertilia, while the
posterior, or epiphysis itself, is homologous with the epiphysis
of Lacertilia ; and he thinks it probable that in their primitive
position the two vesicles were side by side.

I need hardly point out that my own observations on
Sphenodon in the main strongly support the important
results arrived at by Hill. Indeed, except that in Coregonus
the posterior vesicle appears to arise first, the agreement
between the early stages of the types investigated by us is
most remarkable; and I think there can be little doubt that
the parietal eye was not originally an unpaired sense-organ,
as usually supposed, but one of a pair, of which, in most cases
at any rate, the left at the present day has the eye-like
structure most fully developed. In bony fishes the right
parietal eye is evidently represented by the “epiphysis,” as
described by Hill. In Lacertilia it is represented by what I
still prefer to term the “ parietal stalk,” which is therefore
homologous with the ‘ epiphysis” of bony fishes; but how far
this latter is homologous with the “ epiphysis ” of Sauropsida
and Mammalia as that organ is ordinarily understood is
another question.

The fact that, both in the various types of fishes investigated
by Hill and Sphenodon, it is the left vesicle which alone or
first separates from the brain, and either degenerates or gives
rise to a parietal eye, appears to me very remarkable. It is not,
however, by any means certain that it is always the left parietal
eye which is best developed. Thus in Lacerta, according to
Hoffmann’s description (18), the parietal eye would appear
to be formed from the right vesicle, and the stalk from the left
one.

Gaskell also (12), who lays great stress upon the paired
origin of the parietal eye in the lamprey, believes it to be the
right parietal eye which is most perfectly developed, and which

140 ARTHUR DENDY.

overlies the much smaller left vesicle, a conclusion which
appears to be supported by the observations of Studni¢ka (35)
on the development of Petromyzon planeri. The latter
author, however, does not appear to realise the paired character
of the two vesicles, which he terms “ pineal ” and “ para-
pineal” respectively ; the former, which he supposes to be of
greater antiquity, apparently corresponding to the right, and
the latter, which he supposes to be of more recent origin, to
the left parietal eye. Why these two organs should so per-
sistently tend to alter their positions from the primitive trans-
verse to the sagittal plane is a mystery which cannot at present
be explained.

Before leaving the question of the paired origin of the parietal
eye, I should recall the very interesting and significant fact
that Locy (21) considers the epiphysis of Elasmobranchs to be
formed from a united pair of accessory optic vesicles ; and that
he has recorded (23) the discovery of accessory optic vesicles
in the chick, and put forward the view that the Vertebrate
eyes are segmental,

(ce) The Relations between the Parietal Stalk, the
Epiphysis, and the Brain.

The organ which I term in this paper the “ parietal stalk,’
as I have already pointed out, is commonly, if not invariably,
identified by writers on the subject as a portion, at any rate,
of the epiphysis cerebri or pineal gland. By those who
hold that the parietal eye is formed at the expense of the
stalk, the former is commonly regarded as the distal, and the
latter as the proximal part of the epiphysis.

In my summary of results I announced my belief that the
parietal stalk does not represent the epiphysis, basing this con-
clusion upon a more or less mistaken identification of the early
and strongly developed superior commissure which arises just
in front of the stalk as the posterior commissure.! Although my
conclusion was thus originally founded upon a mistake, and

1 [ still believe that the superior commissure may form the anterior part of
the posterior commissure of the adult.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 141

therefore requires considerable modification, I still believe that
there is very grave doubt as to the homology of the so-called
“ epiphysis” throughout the Vertebrate series. I am con-
vinced that in Sphenodon, at any rate, the parietal stalk does
not give rise by itself to what is usually regarded as the epi-
physis of the adult, as described, for example, by Hoffmann
(18), who begins his description of the epiphysis of Sphe-
nodon thus: “ Die epiphyse selbst bildet eine ziemlich weite,
schlauchformige Fortsetzung der dritten Hirnhohle.”

Perhaps one of the most important facts established by my
investigation of the development of Sphenodon is the com-
plete closure of the proximal end of the parietal stalk at Stage R,
so that its cavity is completely shut off from that of the brain.
This closure seems to be effected by compression between the
large superior commissure immediately in front of the base of
the stalk, and the posterior commissure immediately behind it
(fig. 15). It appears to me highly probable that these two
commissures ultimately unite at the point where the opening
of the parietal stalk was placed. If this be so, the so-called
posterior commissure of reptiles and higher vertebrates must be
regarded as a composite structure, the front part of which is
homologous with the superior commissure of fishes.

In any case it is clear that in Sphenodon the parietal stalk
cannot give rise to the main part of the epiphysis of the adult
as that organ is described by Hoffmann, though it may give
rise to a portion thereof.

This conclusion is strongly supported by the examination
of other types. Ritter’s observations on Phrynosoma are
especially instructive in this respect, and are capable of a very
simple interpretation in accordance with my views. Before
discussing these observations it may be well to quote his fol-
lowing paragraph (29) :

“In previous discussions of the nature and function of the
parietal organ, I believe sufficient attention has not been given
to the structure and development of the epiphysis and its
relation to the parietal vesicle, and especially its relation to
the so-called choroid plexus. I have designated the entire

142 ARTHUR DENDY.

structure found in connection with the roof of the thalamen-
cephalon as the epiphysis; but, as already said, I have con-
siderable doubt as to the wisdom of so doing. For the sake of
precision it would seem best that the term epiphysis should be
limited to the structure which arises as an evagination from
this portion of the brain. Certain it is that the large blood
sinus which I have described as a part of the epiphysis in
Phrynosoma cannot be regarded as forming an essential
portion of the structure, and I think it quite possible that
what I have called the epiphysial vesicle is not a portion of
the epiphysis, should the term be limited as I have suggested
that it ought to be. The distinctness of the epiphysial vesicle
from the proximal portion of the epiphysis in the adult
Phrynosoma is without exception, so far as my observations
have gone; and if it is regarded as having been derived from
the epiphysis, then we have two vesicles instead of one that
have arisen in this way, and the difficulty of explaining the
nature and function of the whole structure is correspondingly
increased.”

Now if we limit the term epiphysis as I understand Ritter
to suggest, we shall limit it to the comparatively late evagina-
tion of the thin roof of the third ventricle between the parietal
stalk and the paraphysis. There appears to me to be a good
deal in favour of this definition of the epiphysis; but if we
accept it, then it is quite clear that neither the parietal stalk
nor the parietal eye has anything whatever to do with the
epiphysis, but both develop much earlier, and also behind it.
It is also obvious that the epiphysis in Ritter’s sense is not
homologous with the epiphysis of fishes as described by Hill.

The parietal stalk in Phrynosoma is undoubtedly repre-
sented by what Ritter terms the “ epiphysial vesicle,’ whose
cavity, as in Sphenodon, is completely shut off from that of
the brain.

In comparing Ritter’s fig. 9, representing a sagittal section
through the thalamencephalon and adjacent structures in
Phrynosoma, with my fig. 15, representing a corresponding
section of an advanced embryo of Sphenodon, it is necessary

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 148

to notice one point very carefully. The commissure which
Ritter identifies as the superior commissure evidently does
not correspond with that structure as described by Hill in
fishes and myself in Sphenodon,. It corresponds rather to
what I have identified as the commissura fornicis in
Sphenodon. Where, then, are we to look for the superior
commissure in Phrynosoma? I believe that it has united
with the posterior commissure in the manner already indi-
cated. This interpretation renders Ritter’s figure directly
comparable with mine.

Further evidence in favour of my views on this subject is
afforded by my fig. 28, representing, in a slightly diagram-
matic manner, a sagittal section through the third ventricle
and neighbouring organs in Hinulia.

(f) The Paraphysis.

In the interpretation of the so-called “ epiphysis ” or * pineal
gland” of the higher vertebrates, the paraphysis, only discovered
within recent years, must in future play a very important part.
An excellent summary of the history of our knowledge of this
organ has been given by Studuicka (37). From this it
appears that Selenka, whose original paper! I have been
unable to see, gave the name “ paraphysis” to an unpaired
outgrowth of the membrauous roof of the prosencephalon in
embryos of reptiles and Selachians. He interpreted this
structure as a rudimentary sense-organ, and homologised it
with the auditory organ of Ascidians, as Spencer and de Graaf
had previously homologised the epiphysis with the eye.

The organ itself had been found previously by Hoffmann
in Tropidonotus and Lacerta, and by Francotte® in
Anguis, the latter interpreting it as the choroid plexus. It
has since been found in representatives of almost all vertebrate

1 “ Das Stirnorgan der Wirbelthiere,” ‘ Biol. Centralblatt,’ x.
2“ Weitere Untersuchungen zur Entwicklung der Reptilien,”’ ‘Morph.
Jahrb.,’ xi.
3 “ Recherches sur la Développement d |’Epiphyse,” ‘ Archiv. de Biologie,’
Vili.
VOL, 42, PART 2.—NEW SERIES. K

144 ARTHUR DENDY.

groups. McKay figured it in Hinulia (25, figs. 7, 8) as far
back as 1888, but described it only as an evagination in front
of the pineal eye, “ very similar to the eye itself.”

Leydig—as I learn from a short paper by Eycleshymer
(11)—finds instead of a single vesicle a group of five, which
later came into such close relation with the “ epiphysis” that
Spencer figured both as one structure. Eycleshymer himself
describes the paraphysis in Amblystoma as a median out-
growth of the posterior portion of the roof of the prosen-
cephalon. Lateral diverticula appear at its distal end, but its
cavity becomes obliterated proximally in a manner analogous
to that which occurs in the “epiphysis.” The two structures
in Urodela never come into close relation, as in Reptilia, but
remain widely separate.

The development of the paraphysis in Sphenodon has
already been described in these pages, and appears to agree
closely with what takes place in Lacertilia. It gives rise to a
mass of convoluted tubules, lined by short columnar cells and
intermingled with blood-vessels, which lies beneath the parietal
eye and in front of the parietal stalk (figs. 15, 16). These
tubules undoubtedly form a part of what is usually recognised
as the “ pineal gland.”

It thus appears that my observations with regard to the
paraphysis support the conclusions of Leydig already noticed.
Hoffmann also, according to Ritter (29), considers that the
paraphysis or “ ependyma” in the grown animal “‘comes to
take a not inconsiderable part in the formation of the epi-
physis.” There appears to be no reason whatever for regard-
ing it as a sense-organ, while its structure certainly suggests
a glandular function. In origin it is intimately connected
with the choroid plexus of the lateral ventricles, as I have
already shown.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 145

(g) The Epiphysis of Sphenodon and Lacertilia.

It thus appears that the “ epiphysis” of Sphenodon and
Lacertilia, as ordinarily understood, is a composite structure,
made up of the following constituent parts :

(1) An upward arching of the thin roof of the third ventricle
between the superior commissure and the commissura fornicis.
This outgrowth arises comparatively late in development, and
is probably the result, at any rate to a large extent, of the com-
pression of the thalamencephalon between the prosencephalon
and the mid-brain in the straightening outof thecerebral flexure.
It is evidently homologous with the Zirbelpolster described,
for example, by Burckhardt (9) in Lacerta. It is to this part
that Ritter (29) proposes to restrict the term ‘‘ epiphysis.”

(2) An outgrowth from the membranous roof of the pros-
encephalon arising in close relation with the choroid plexus
of the lateral ventricles, and known as the paraphysis. This
outgrowth grows upwards and backwards over the commissura
fornicis, and gives rise to a number of blind diverticula or
tubules, which lie beneath the parietal eye.

(3) A number of blood-vessels, which make their way in
amongst the diverticula or tubules of the paraphysis.

(4) The parietal stalk, which arises from the roof of the
thalamencephalon a short way in front of the optic lobes, and
passes upwards and forwards over the superior commissura
until it meets the backward-growing paraphysis. The parietal
stalk, however, does not appear to form any very important
part of the epiphysis of the adult, but is a vestigial structure
representing (in Sphenodon) the right ‘parietal eye. It
undoubtedly gives rise to the “epiphysial vesicle” of Ritter,
and to a part, at any rate, of what Spencer (33) calls the
*‘ portion of the epiphysis equivalent to the pineal stalk.” It
is clearly shown, for example, in Spencer’s fig. 41, representing
a sagittal section through the brain of Varanus bengalensis,
where it is labelled Ep.'; and in his fig. 25, representing a
corresponding section of Anguis fragilis.

This analysis of the so-called ‘‘ epiphysis ”’ into its constituent

146 ARTHUR DENDY.

parts is very obvious in the case of an advanced embryo of
Hinulia, as seen in sagittal section and represented in my
own fig. 28. It is also made very clear by Burckhardt’s
admirable figure of the brain of the embryo of Lacerta (9),
in which, however, he does not represent any commissura
fornicis.

(h) The Accessory Vesicle.

The accessory vesicle (fig. 14, Ac. V.), which appears between
the tubules of the paraphysis and the end of the parietal stalk
in the early part of Stage R of Sphenodon, is extremely
difficult to account for. The histological character of its walls
and the absence of any communication of its cavity with those
of the surrounding organs seem to indicate an independent
origin. It is conceivable that it may be formed from the out-
growth which I have identified as the commencing paraphysis
at Stages N and O (figs. 6,9, 11, Par.), and that the paraphysis
itself originates later, between Stages O and KR; but this view
appears to me to be highly improbable.

The absence of blood-corpuscles, the large size of the cavity,
and, to a less extent, the nature of the lning epithelium,
prevent us from regarding it as a blood-vessel. It may
perhaps be a large lymphatic sac, but this again does not
seem very probable. It may be homologous with the accessory
vesicle or “parapineal organ” described by Ritter (30) in
Phrynosoma, but it is very doubtful if it is homologous
with the “ parapineal organ” of Studnicka in the lamprey.

Burckhardt (9) figures a small vesicle in a similar position
in Lacerta vivipara, and, following Leydig, terms it the
“ Nebenscheitel-Organ.” Unfortunately I have not seen
Leydig’s work, and therefore am unable directly to compare
the organ as described by him with the transitory accessory
vesicle of Sphenodon. Judging from the observations of
Prenant (26), however, it seems to me hardly likely that the
accessory vesicle of Sphenodon is homologous with the
accessory parietal organs of Anguis fragilis, which are
supposed to be budded off from the ‘ epiphysis ’’ (parietal
stalk), and, like the latter, have thick walls,

“PARIFTAL EYE AND ADJACENT ORGANS IN SPHENODON. 147

4, GENERAL CONCLUSIONS.

As a result of my study of the writings of other embryo-
logists, as well as of my personal observations, I have been led
to formulate the following general conclusions as to the
homologies and phylogenetic history of the “epiphysis” and
associated organs.

(1) The “epiphysis” of Selachians is formed by a pair of
equally well-developed optic vesicles, originating as outgrowths
of the thalamencephalon, aud uniting together in the middle
line (as shown by Locy).

(2) In Teleosts and A mia there is also a pair of epiphysial
outgrowths arising in a similar way (with more or less dis-
placement), but the right vesicle alone gives rise to the
“epiphysis” of the adult, while the left.one separates completely
from the brain and degenerates (as shown by Hill).

(3) In Cyclostomes there is also a similar pair of epiphysial
outgrowths, which suffer displacement in such a manner that
the right vesicle comes to overlie the left. The right vesicle
forms a parietal eye, and the left one the “ parapineal organ.”
These two, together with the nerve of the parietal eye,
constitute the “epiphysis”? (compare Ahlborn, Beard, Gaskell,
Studnicka).

(4) In Sphenodon and Lacertilia the “epiphysis” is a
composite structure in which the paraphysis and “ Zirbel-
polster” take a very large share, while the parts which corre-
spond to the paired epiphysial outgrowths of fishes take a very
small one. These outgrowths originate, however, very much
as in fishes, and are subject to more or less displacement, and
one or other of them may give rise to a parietal eye. In
S phenodon it 1s the left parietal eye which is thus developed.

(5) The right parietal eye is represented in Sphenodon by
the “ parietal stalk.’ In Lacertilia the parietal stalk re-
presents either the right or left parietal eye.

(6) The parietal eye has no real connection with the
parietal stalk beyond that of fellowship, and is supplied with
a special nerve of its own not derived from the parietal stalk.

148 ARTHUR DENDY.

(7) The ancestors of existing vertebrates possessed a pair of
parietal eyes, which may have been serially homologous with
the ordinary vertebrate eyes.

5. List or REFERENCES.

The following list contains only the titles of such memoirs as I have been
able personally to consult, and is therefore by no means a complete biblio-
graphy of the subject.

1. Aucporn, F.—“< Untersuchungen tiber das Gehirn der Petromyzonten,”
‘Zeitschrift fiir wissenschaftliche Zoologie,’ Bd. xxxix, 1883, p. 191.
2. Batrour, F. M.—‘ A Treatise on Comparative Embryology,’ vol. i,
1881.
3. Bearp, J.—‘‘ The Parietal Hye in Fishes,” ‘ Nature,’ vol. xxxvi, 1887,
pp. 246 and 340.
4. Brarp, J.—‘‘ The Parietal Hye of the Cyclostome Fishes,” ‘ Quart.
Journ. Micr. Sci.,’ vol. xxix, 1888, p. 55.
5. Beraneck, H.—“‘ Sur le nerf pariétal et la morphologie du troisieme ceil
des Vertébrés,” ‘ Anatomischer Anzeiger,’ vol. vii, 1892, p. 674.
6. Brraneck, H.—* L’individualité de l'oeil pariétal,” ‘ Anatomischer An-
zeiger,’ vol. vill, 1893, p. 669.
7. Bernarp, H. M.—“‘ An Attempt to deduce the Vertebrate Hyes from
the Skin,” ‘ Quart. Journ. Mier. Sci.,’ vol. xxxix, p. 348.
8. BurckHarnpt, R.—‘ Die Homologien des Zwischenhirndaches, und ihre
Bedeutung fiir die Morphologie des Hirns bei niederen Vertebraten,”’
‘ Anatomischer Anzeiger,’ Bd. ix, 1894, p. 152.
9. Burckuarpt, R.—“ Die Homologien des Zwischeuhirndaches bei Rep-
tilien und Vogeln,” ‘ Anatomischer Anzeiger,’ Bd. ix, 1894, p. 320.
10. Drenpy, A.—“ Summary of the Principal Results obtained in a Study of
the Development of the Tuatara (Sphenodon punctatum),”
‘ Proceedings of the Royal Society,’ vol. Ixiii, 1898, p. 440.
11. KycLnsuyMer, C.—“ Paraphysis and LEpiphysis in Amblystoma,”
‘ Anatomischer Anzeiger,’ vol. vii, 1892, p. 215.
12. Gasket, W. H.— On the Origin of Vertebrates from a Crustacean-like
Ancestor,” ‘Quart. Journ. Mier. Sci.,’ vol. xxxi, 1890, p. 379.
13. GaskeLt, W. H.—‘ Opening Address, Physiological Section, British
Association,” ‘ Nature,’ vol. liv, 1896, p. 551.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 149

14.

15.

16.

Li.

18.

19.

20.

2l.

22.

23.

24.

25.

26.

27.

28.

29.

GAskKELL, W. H.—‘‘ On the Origin of Vertebrates, deduced from the
Study of Ammoceetes,” ‘Journal of Anatomy and Physiology,’ vol.
xxxii, 1898, p. 513.

Graar, H. W. pe.— Zur Anatomie und Entwickelungsgeschichte der
Epiphyse bei Amphibien und Reptilien,” ‘ Zoologischer Anzeiger,’
Bd. ix, 1886, pp. 191—194.

Hitt, C.—‘‘ Development of the Epiphysis in Coregonus albus,”
‘Journal of Morphology,’ vol. v, 1891, p. 508.

Hiut, C.—“The Epiphysis of Teleosts and Amia,” ‘Journal of Mor-
phology,’ vol. ix, 1894, p. 237.

Horrmann, C. K.—‘ Bronn’s Klassen und Ordnungen des Thier-reichs,’
Reptilien.”

Kiinckowstrom, A. pE.—“ Le premier développement de l’cil pinéal,
Peépiphyse, et le nerf pariétal chez Iguana tuberculata,” ‘ Anato-
mischer Anzeiger,’ Bd. viii, 1893, p. 289.

Leypie, F.—“ Das Parietalorgan der Wirbelthiere,” ‘ Zoologischer
Anzeiger,’ Bd. x, 1887, p. 534.

Locy, W. A.—‘‘ The Derivation of the Pineal Hye,” ‘ Anatomischer

Anzeiger,’ Bd. ix, 1894, pp. 169 and 2381.)!

Locy, W. A.—‘ The Mid-brain and the Accessory Optic Vesicles,”
‘ Anatomischer Anzeiger,’ Bd. ix, 1894, p. 486.

Locy, W. A.—‘‘ Accessory Optic Vesicles in Chick Embryo ” (Abstract),
‘Journ. of Royal Microscopical Society,’ February, 1898.)

Marsuaty, A. M.—‘ Vertebrate Embryology,’ 18938.

McKay, W. J.—‘ The Development and Structure of the Pineal Eye in
Hinulia and Grammatophora,” ‘Proceedings of Linnean Society
of New South Wales,’ series 2, vol. iii, 1888, p. 876.

Prenant, A.—‘ Sur l’ceil pariétal accessoire,” ‘ Anatomischer Anzeiger,’
Bd. ix, 1894, p. 103.

Rasi-RtcKkHarp, H.—“‘ Ueber das Vorkommen eines Fornixrudiments
bei Reptilien,” ‘ Zoologischer Anzeiger,’ Bd. iv, 1881, p. 281.

Rasi-Rickwarp, H.—‘ Zur Deutung der Zirbeldriiss (Epiphysis),”
“Zoologischer Anzeiger,’ Bd. ix, 1886, p. 405.

Ritrer, W. E.—‘‘ The Parietal Eye in some Lizards from the Western
United States,” ‘ Bulletin of the Museum of Comparative Zoology at
Harvard,’ vol. xx, 1890-91, p. 209.

? Full papers (which the author of the present memoir had no opportunity
of consulting) will be found in ‘Journ. Morphol.,’ vol. ix, p. 115, and vol. xi,
p. 497.—G. B. H.

150 ARTHUR DENDY.

80. Rirrer, W. E.—‘‘On the Presence of a Parapineal Organ in Phryno-
soma,” ‘ Anatomischer Anzeiger,’ vol. ix, 1894, p. 766.

81. Spencer, W. Batpwin.—‘‘ The Parietal Eye of Hatteria,” ‘ Nature,’
vol. xxxiv, 1886, p. 33.

82. Spencer, W. Batpwin.—* Preliminary Communication on the Structure
and Presence in Sphenodon and other Lizards of the Median Hye,
described by von Graaf in Anguis fragilis,” ‘ Proceedings of the
Royal Society,’ 1886, p. 559.

33. Spencer, W. BALDwin.—*“ On the Presence and Structure of the Pineal
Hye in Lacertilia,” ‘Quart. Journ. Mier. Sci.,’ vol. xxvii, 1886, p. 165.

34. Spencer, W. Batpwin.— The Pineal Eye of Mordacia mordax,”
‘Proceedings of the Royal Society of Victoria,’ vol. ii, 1890, p. 102.

35. Stupnicxa, F. K.—Sur les Organes pariétaux de Petromyzon
planeri,” ‘ Véstnik kral. éeské Spoleé nauk,’ 1893, ‘ Stzb. k. bohm.
Ges. Wiss.,’ i, Prag., 1893

36. StupnicKa, F. K.— Zur Lésung einiger Fragen aus der Morphologie
des Vorderhirnes der Cranioten,” ‘Anatomischer Anzeiger,’ vol. 1x,
1894, p. 307.

37. SrupnicKa, F. K.—* Zur Anatomie der sog. Paraphyse des Wirbelthier-
gehirns,” ‘Sitzungsberichte der Konigl-bohmischen Gesellschaft der
Wissenschaften,’ Prag., 1895.

38. Tuomas, A. P. W.—‘ Preliminary Note on the Development of the
Tuatara (Sphenodon punctatum),” ‘Proceedings of Royal Society,’
vol. xlviii, 1890, p. 152; and ‘New Zealand Journal of Science,’
vol. i (new issue), 1891, p. 27.

EXPLANATION OF PLATES 11—13,

Illustrating Mr. Arthur Dendy’s paper on the “ Development
of the Parietal Eye and Adjacent Organs in Sphenodon.”

EXPLANATION OF LETTERING.

Antr. Anterior. Au. Auditory vesicle. B.V. Blood-vessel. Cart. Cartilage.
C. H. Cerebral hemisphere. Ch. P. Choroid plexus. Ch. P. L. V. Choroid
plexus of lateral ventricle. Coagy. Coagulated humour in cavities of parietal
eye and stalk. © Com. Ant. Anterior commissure. Com. F. Commissura

fornicis. Com. Man. Mantle commissure. Com. Sup. Superior commissure.
Com. Post. Posterior Commissure. Corp. Séri. Corpus striatum. Derm.

PARIETAIL EYE AND ADJACENT ORGANS IN SPHENODON. 151

Dermis. pd. Epidermis. Hye. Ordinary paired eye. F. B. Fore-brain.
F. M. Foramen of Monro. F. N. P. Fronto-nasal process. Gang. Hab. R.
and L. Right and left ganglia habenule. H. B. Hind brain. H.C. Head
cavity. H. M. C. Hyomandibular cleft. nf. Infundibulum. Jnzé. Integu-
ment. Jz. W. Inner layer of wall of parietal optic cup and of parietal stalk.
Iter. Iter a tertio ad quartum ventriculum. JZ. Lens of parietal eye. L. P.
Lens of paired eye. JZ. V. Lateral ventricle. Mand. Mandibular arch.
M. B. Mid-brain. Mes. Mesoblast. Mo. Mouth. Mol. Molecular layer in
retina of parietal eye. ot. Notochord. Nw. Nucleus. O. LZ. Optic lobes.
O. Par. Origin of paraphysis between the choroid plexuses of the lateral
ventricles. Op. ch. Optie chiasma. Ow. W. Outer layer of wall of parietal
optic cup and of parietal stalk. O. VY. Optic vesicle of paired eye. Pa. EL.
Parietal eye. Padi. Pallium of cerebral hemisphere. Pa. V. Nerve of parietal
eye. Par. Paraphysis. Par. 7. Tubules of paraphysis. Pa. S. Parietal
stalk. PP. B. Pituitary body. p.m. Pia mater. P. V. Primary parietal
vesicle. Pig. Pigment. &. 7. Roof of thalamencephalon. 7ha/. Thalamen-
cephalon. Thal. Op. Optic thalamus. V.3. Third ventricle. XX. (Figs. 16
and 17.) Space due to separation of outer and inner layers of wall of parietal
optic cup. X. (Fig. 18.) Inner surface of retina of parietal eye. Z. Evagina-
tion of the roof of the third ventricle forming the proximal part of the
“epiphysis” of the adult, and sometimes known as the ‘ Zirbelpolster.”

Fics. 1—27. Sphenodon punctatus.

Fig. 1.—Stage J. Embryo 44. ‘Transverse section of head, showing over-
lap of the left side of the roof of the fore-brain on the apparent ventral
aspect. (Reversed as compared with Fig. 5.) Zeiss A, oc. 1, camera outline.

Fic. 2.—Stage K. Embryo 39. ‘Transverse section of head, showing first
appearance of the primary parietal vesicle to the left of the middle line on the
apparent ventral aspect. (Reversed as compared with Fig. 5.) Zeiss A,
oc. 1, camera outline.

Fie. 3.—Stage L. Embryo 50. Longitudinal vertical section of head
(nearly median), showing origin of primary parietal vesicle as an evagination
from the roof of the fore-brain. Zeiss A, oc. 1, camera outline.

Fic. 4.—Stage L. Embryo 50. Portion of the same section, more en-
larged, to show the histological structure of the primary parietal vesicle.
Zeiss D, oc. 1, camera outline.

Fic. 5.—Stage M. Embryo 81. Transverse section of head, showing
origin of primary parietal vesicle on the left side of the middle line, and
apparently ventral owing to cerebral flexure. Zeiss A, oc. 1, camera outline.

Fic. 6.—Stage N. Embryo 96. Median longitudinal vertical section
through the paraphysis at its first appearance, showing its origin as a backward-
pointing outgrowth of the roof of the fore-brain. Zeiss D, oc. 1, camera
outline.

152 ARTHUR DENDY.

Fie. 7.—Stage N. Embryo 96. Longitudinal vertical section through
parietal eye and stalk. The eye lies to the left of the stalk, and is cut on one
side of its median plane. Zeiss D, oc. 1, camera outline.

Fic. 8.—Stage N. Embryo 96. Similar section a little further to the left,
cutting through the middle of the parietal eye and escaping the stalk. Zeiss
D, oc. 1, camera outline.

Fie. 9.—Stage O. Embryo 92. Diagrammatic sagittal section through
the head, to show the relative positions of the parietal eye and stalk, para-
physis, and infundibulum. (The parietal eye really lies to the left of the
middle line.)

Fic. 10.—Stage O. Embryo 92. Sagittal section a little to the left of
the middle line, through the parietal stalk and eye, passing a little to the left
side of the opening of the cavity of the parietal stalk into the brain cavity.
Zeiss D, oc. 1, camera outline.

Fie. 11.—Stage O. Embryo 92. Sagittal section through the paraphysis,
showing its commencing convolution. Zeiss D, oc. 1, camera outline.

Fie. 12.—Stage O. Embryo 89. Upper surface of head, to show relative
positions of parietal eye aud stalk, paraphysis, cerebral hemispheres, and optic
lobes. x 10.

Fig. 13.—Stage R (early). Embryo 143. Upper surface of head, to show
positions of parietal eye, cerebral hemispheres, and optic lobes. x 5.

Fie. 14.—Stage R (early). Embryo 148. Part of sagittal section through
parietal eye and stalk, paraphysis, and accessory vesicle. Zeiss C, oc. 1,
camera Outline.

Fig. 15.—Stage R (late). Embryo 2. Sagittal section through the
thalamencephalon and neighbouring parts, combined from several of the same
series. Zeiss A, oc. 1 (assisted by camera).

Fie. 16.—Stage R (late). Embryo 2. Sagittal section through parietal
eye, end of parietal stalk, nerve of parietal eye, and paraphysial tubules.
Zeiss D, oc. 1, camera outline.

Fie. 17.—Stage R (late). Embryo 2. Part of retina of parietal eye in
vertical section, to show its histological structure. Zeiss I’, oc. 1, camera
outline.

Fie. 18.—Stage R (late). Embryo 2. Part of vertical section of parietal
eye, to show coagulated humour,! containing a group of nuclei and adhering
to the inner surface of the retina (X). The sharp bend in the surface of the
retina marks the bottom of the optic cup. Zeiss F, oc. 1.

Fie. 19.—Stage O. Embryo 92. Vertical section of lens (sagittal) with

1 In many of the drawings of the sections the coagulated humour has been
omitted for the sake of clearness.

PARIETAL EYE AND ADJACENT ORGANS IN SPHENODON. 158

adherent humour containing a nucleus, and with another nucleus apparently
escaping from the under surface of the lens. Zeiss F, oc. 1, camera outline.

Fic. 20.—Stage O. Embryo 92. Part of wall of parietal stalk from same
section as last, with adhering humour and nucleus. Zeiss F, oc. 1, camera
outline.

Fic. 21.—Stage R (late). Embryo 3. ‘Transverse section through the
parietal stalk near its distal extremity. Zeiss D, oc. 1, camera outline.

Fies. 22—27.—Stage R (late). Embryo 3. ‘Transverse sections through
the third ventricle and adjacent parts, arranged in order from before back-
wards. x 163. Camera outlines.

Fig. 28.—Hinulia, sp. Part of a sagittal section through the third
ventricle and adjacent parts of an advanced embryo, to show the relations
of the “epiphysis,” parietal eye, and associated parts. Zeiss A, oc. 1.
Slightly diagrammatic.

LGOe Taethearn: wee = Hike * rune it

ete, tT i aed ate a ha, ee
7 a, FU? yee ta 1 a gus) iy dials ae
a + ut “Hironp 10 Sala hat, ee Ci *
Ai ehia Yt aa TaN 0g fs j ‘ts ¢ ri dati wy pita a ial

“h i ‘ iz

are fe: = en : ; 9 : is
) | vg? tikfeA cree Ae riding tinh és ails oe
AHVINE etdin ef oh ee A ek vee ies a ti alt : 4
B + eae i f :
3 ’ 4 De Cali rc
’ ‘ .
= ah it 1a Todi a its
\ “Saepe |f
Lit } yy . ¢ >
= ‘
Fas }
« 4 a . .
{
* .
}
I
‘
: ,?
=
6

« = J std 4

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 155

The Molluscs of the Great African Lakes.

III. Tanganyikia rufofilosa, and the Genus Spekia.
By

J. E. S&S. Moore.

With Plates 14—19.

On the coast line formed by the red sandstone and con-
glomerate precipices which flank the picturesque shores of
Lake Tanganyika there are to be found a number of rock
molluses, which dwell upon the submerged stones, much
in the same manner as the periwinkles do upon the half-tide
rock faces of the ocean coasts. Of these small Gastropods,
what have appeared to be at least three distinct species have
been known, but hitherto only by the characters of their empty
shells, and in the absence of all other information they have
been regarded by the conchologists as approaching the genus
Lithoglyphus.! In the literature, one of these forms, L.
neritinoides, still bears that generic name, but the species
rufofilosa was eventually placed by Cross in what he supposed
to be an allied but separate genus, under the name of Tangan-
yikia, while the original L. zonatus was placed by Bourgui-
gnat in the genus Spekia.

No shells exactly similar to these have been found, hitherto,
outside the confines of the great lake in which they were
originally observed by Speke ; and from this very remarkable
fact we might suspect, without further investigation, that they

' That is as members of the family Hydrobiide.

156 J. HE. 8. MOORE.

belong to the singular marine fauna for which the lake is
famed.

In the present paper it is my purpose, in the first place, to
exhibit the wide morphological difference which exists between
these types by a description of the anatomy of Tanganyikia
rufofilosa as compared with that of Spekia zonata, both
forms having been collected during my expedition to the lake
in the summer of 1896. In the second, I wish to show that
the morphological characters of both these forms, like those
of all the halolimnic species, suggest that they are the con-
stituents of a fauna that some ancient sea has left behind.
And lastly, I wish to lay emphasis upon the fact that they
possess anatomical peculiarities, as a concomitant of their vast
antiquity, which are worthy of the closest study. For, quite
apart from the fact that these molluscs are zoological curiosi-
ties in the sense that they are relics of a departed biological
era, it would be difficult to instance any forms possessing a
higher scientific value as a means of clearing up the obscure
inter-relationship of numerous molluscan types.

Tanganyikia rufofilosa.

The shell of this form is represented in Pi. 14, fig. 1,
and was first described by Smith,! from the empty shells
obtained by Captain Speke and Mr. Hore. The animals,
which I observed alive, have a white, semi-transparent foot,
with a very broad, wrinkled, and pigmented snout. The
tentacles are not long, and the eyes are situated on the posterior
base of each. The proboscis is retractile but non-protrusible,
like that of Typhobia. The Littorinoid characters of the
operculum are apparent in fig. 1. The buccal mass in this
species is extremely small, and the whole radular apparatus is
so very little developed, and the radula itself proportionately
reduced to such a flimsy structure, that it is by no means easy to
find. It actually exists, however, in the bed of a very slight dila-

' ¢Proc. Zool. Soc.,’ 1881, p. 288, and ‘Ann. and Mag. N. H.,’ 1880, vi, p. 426.

THE MOLLUSCS OF THE GREAT AFRIOAN LAKES. 157

tation of the gullet (P1l.14, fig.3, rad.), and this dilatation must
be taken as the only indication of the almost entirely undeveloped
buccal mass. The radular dentition is interesting and some-
what peculiar; a single row of the delicate teeth is represented
in fig. 5. In many ways the dentition suggests physiological
adaptation. Both the rows of lateral teeth are very much alike,
and there does not appear to be any difference in the relative
size of the serrations on the head of either the outer or the
inner of these teeth. In this feature the lateral teeth
resemble those of Modulus lenticularis, as figured by
Troschel, and to a certain extent the whole row of teeth
approximates to this type; but on closer examination the
admedian, no less than the median tooth itself, differs widely
from any of the more normal melanioid forms. The predominant
denticle on the reflexure of the admedian teeth, which is such
a constant feature of so many radulas (see the radula of
S pekia zonata (PI.18, fig.3, pred.), is here quite wanting, but
it is clearly visible in Troschel’s figure of Modulus to which
I have referred. Neither is there any posterior denticle on the
median tooth of T. rufofilosa which would either correspond
to that of Modulus, or that on the median teeth of the forms
truly belonging to the genus Lithoglyphus. The admedians,
laterals, and to some extent the medians of this radula
approximate to those of Tympanotomus fuscatus given by
Troschel, and in a more general way to those of Planaxis
sulcatus (Lam.); and from these characters of the radula we
might conclude with certainty that the genus Spekia has little
or no connection with the Hydrobiide, to which family it
has been hitherto thought to belong.

In all the specimens of T. rufofilosa which I examined, the
mouth, esophagus, stomach, and rectum were crammed with
sharp rock fragments, about the size of a pin’s head; and it is
a curious fact that these animals should thus exhibit a pro-
pensity to fill themselves with stones, while at the same time
the radula, or normal rasping apparatus, is almost entirely
lost. The salivary glands are simple and somewhat long.

The cesophagus is narrow, and opens posteriorly in the

158 J. E. S. MOORE.

hinder chamber of a small stomach (PI. 14, fig. 4, @.). In
the anterior chamber of this organ (fig. 4, ant. ch.) there
is lodged a well-developed crystalline style, represented in
fig. 4, c.st. The intestine passes out of the left side of
the posterior stomachic chamber much in the same manner as
in Typhobia. It takes the course represented in fig. 4, int.
Towards the rectal extremity of the intestine there is a large,
round, glandular lump (figs. 4 and 9, R.g.), very similar at
first sight to that connected with the oviducts in the female
Littorina.

The liver is large, and opens by two very small hepatic
ducts on to the base of the posterior stomachic chamber (fig.
6, h. d.).

The kidney surrounds the heart, and extends posteriorly to
that organ. It opens by a minute aperture, represented in
Pl. 14, fig. 4, R., at the extreme upper end of the mantle cavity.

The heart has the normal tznioglossate characters. It les
in a large pericardial cavity (Pl. 14, fig. 4), and consists of —

1. A thin-walled auricle in direct connection with the pul-
monary vein.

2. A thick-walled ventricle which opens by a valvular aper-
ture into—

3. An aortic trunk, from which the anterior and _ posterior
aorte proceed forward and backwards (fig. 4).

The gill is large and somewhat peculiar (Pl. 14, figs. 2,14). It
occupies an elongated oval space upon the inner mantle wall,
and consists of numerous strong, broad, triangular gill leaves,
the projecting apices of which occupy the median line of the
structure. The apex of each leaf is prolonged into a filiform
process, which gives the curious appearance to the gill repre-
sented in figs. 2, 14.

These filaments do not project quite straight, but are more or
less inclined to the rectal (left) side of the gill. On this same
side, each gill leaf below the median process is slightly pig-
mented, the whole left side of the ctenidium appearing con-
sequently dark in comparison with the other.

The osphradium is long, and lies at the bottom of a groove

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 159

below the gill. It is a simple ropy structure at its base, but it
becomes very distinctly pectinated at its distal extremity (Pl.
14, figs. 2, 14, os.).

The nervous system of T. rufofilosa (Pl. 14, fig. 6) con-
sists of two large cerebral ganglia, closely united together as
in the genus Melania (PI. 15, fig. 58).

The cerebral ganglia are indistinguishably fused with the
pleural ganglia below, but from their respective loci there
spring the super- and sub-intestinal cords (Pl. 15, fig. 6), which
pass to the super- and sub-intestinal ganglia. From the locus
of the left pleural ganglion a nerve passes out towards the
mantle, and this nerve apparently anastomoses with a branch
from the subintestinal ganglion (Pl. 15, fig. 6). The nervous
system is therefore dialyneurous on the left.

In like manner there springs a nerve from the right pleural
ganglion, which also unites with a branch from the super-
intestinal. The nervous system is therefore entirely dialy-
neurous on both sides, and when viewed from above it is
practically indistinguishable from the similar view of the
nervous system of Melania (?) which I have given in Pl. 15,
fig. 5. On the under side of the cerebral ganglion, however,
there is a curious and, so far as I am aware, unique median
protuberance (Pl. 14, fig. 8, m.p.c.g.), but which is in itself
quite sufficient to distinguish the nervous system of T. rufo-
-filosa from that of any known Melania. The cerebro- and
pleuro-pedal commissures are of median length, and unite at
the upper extremity of the well-developed pedal ganglia (Pl. 14,
fig. 8, p. g.). On the upper posterior faces of the pedal ganglia
there are slight and peculiar projections (Pl. 14, fig. 8).

The otocysts are fairly large; they lie behind and slightly
above the pedal ganglia, the otocyst nerve passing diagonally
across the outer faces of the cerebro- and pleuro-pedal con-
nectives to the cerebral ganglia on each side (Pl. 14, fig. 8,
ot.). The otoliths are rectangular, small and numerous ;
there are a small osphradial and two visceral ganglia.

VoL, 42, PART 2,—NEW SERIES, L

160 J. E. 8S. MOORE,

Reproductive Apparatus.

In the female Tanganyikia rufofilosa the whole ar-
rangement of the genital apparatus is most interesting, and
there is little doubt that in both sexes of this species we are
dealing with one of the most archaic types of reproductive
apparatus that any of the Gastropods possess. The genital
gland is situated on the upper surface of the apical body-whorls,
and in the male (Pl. 14, fig. 4) it is united by a number of
small ducts to a rather long but not greatly coiled vas deferens,
which passes beneath the intestine, and opens by a fine aperture
immediately below the rectum (fig. 4, v. d.). Beyond this
aperture, in some male specimens, I was able to trace a fine
groove running forward along the lateral wall of the body,
and dying out beneath the eye (fig. 2, g. v.); while in others
this groove, which is well known in the males of several
Gastropods, was not to be found. Such grooves are known
to exist in the males of various representatives of the Struthio-
laride, and many other forms. In the Strombide and Litto-
rinidz they are to be traced in both the male and female.

In the female T. rufofilosa the ovary lies in the same posi-
tion as the testis in the male (PI. 14, fig. 10), and the ova are
collected by a number of small channels into a common long
non-convoluted oviduct (Pl. 14, fig. 10, g. d.).

The aperture of the oviduct is somewhat further forward in
the female than the corresponding opening in the male; and I
was surprised to find that the aperture of the oviduct in T.
rufofilosa is always related to a very strongly marked furrow
which runs forward in the same position as the spermatic groove
in the male (Pl. 14, fig. 14, g. v.), but it here terminates in a
little pit beneath the eye (fig. 14,2.). On the other side of the
head, that is on the left, there appeared in all the females that
were killed during the breeding season a relatively great pro-
tuberance, which had the appearance of a pathological swelling,
behind and below the left tentacle (Pl. 14, fig. 14,6.p.). On being
opened this swelling was seen to be full of small round bodies,
which I at first took to be parasites, but which when more

THE MOLLUSCS OF THE GREAT AFRICAN LAKES, 161

closely examined were found to be embryo rufofilosi in all
stages of their late development. By pushing a bristle into
the pit beneath the eye, on the right where the groove leading
from the aperture of the oviduct terminated, it was easy to
ascertain that this singular brood-pouch on the left was
connected with the base of the groove and the pit on the right
by a small tube (Pl. 14, fig. 10, ¢c.), which passed completely
through the foot beneath the buccal mass. Thus the external
groove passing forward from the opening of the oviduct, and
which unquestionably corresponds with the ciliated spermatic
groove in the male, is connected in the female with an
invaginated tube and brood-pouch, and we must consequently
regard all these structures as the correlated parts of an
accessory reproductive apparatus, which in the complete
form just described is not present, so far as I can ascertain,
in any of those Prosobranchs the anatomy of which is known.
In searching for the meaning of this singular apparatus I
discovered by accident among some of the so-called Melanias,
which the authorities of the British Museum had courteously
placed at my disposal for comparison, an obviously analogous
but not quite similar condition of the reproductive apparatus
in several of the viviparous females belonging to these forms.

In the so-called Melania episcopalis (Lea) which I
received through Mr. Smith we found, on removing the body-
wall, really in order to expose the nerves, that not a little to
our surprise we had cut into a sac unconnected with the
oesophagus, and which was filled with embryos in all their late
stages of development (PI. 14, fig. 18, B. P.). On examining
the position of the external opening of the oviduct, it was soon
seen that this was put into relation with a subocular pit (PI.
14, fig. 18, v.) by a most prominent groove (fig. 13, g. v.), and
it was easy to pass a bristle (B.) from the exterior opening of
the pit into the brood-pouch which had already been opened
up (Pl. 14, fig. 13, B. P.).

From this dissection it became at once evident that in M.
episcopalis there existed the same relations between the
oviduct in the female and an external groove connected with

162 J. E. 8S. MOORE.

an internal brood-pouch which had been observed in Tangan-
yikia rufofilosa. But in the M. episcopalis the pouch
is directly above the cesophagus, and median instead of being
below and to the left side as in the former case. (Compare Pl.
16, figs. 1 and 2.)

I do not think, however, that this relative change observed
in the position of the brood-pouch between T. rufofilosa and
M. episcopalis is of much morphological importance, for the
whole series of structures is far too similar in both cases to
admit of the slightest doubt that they must be regarded as
morphologically the same.

In two small Melanias (species?) which had been collected
by Mr. Cummings from the fresh waters of the Philippine
Islands similar grooves were present in the females, and in
like manner these were connected with median dorsal brood-
pouches.

In a male Faunus, also lent to me by the authorities of the
British Museum, there was a most pronounced groove (PI. 15,
fig. 11, g.v.); but, as I have had no opportunity of examining
the female, I am naturally ignorant whether the groove is
here connected with a brood-pouch or not.

From all this it is evident that in some Prosobranchs, T.
rufofilosa, M. episcopalis, and several other forms, there
exists in the female a complex accessory reproductive apparatus,
which consists of an external ovigenetic groove in direct rela-
tion with an internal brood-pouch, which is either median and
dorsal, or lateral and on the left side. In the males of these
forms these structures are represented by spermatic grooves
exactly similar to those of the Struthiolaridz, the genera
Strombus, Pteroceras, and the like. But so far as at present
known the complete female apparatus is only represented in
Spekia and the so-called Melanias which I have named. In
various female Prosobranchs, however, more or less of the groove
at any rate appears. In the female Strombus gigas there
is avery pronounced groove leading downwards across the foot,
which is figured by Haller. Also in the female Littorina I

1 * Morph; Jahrb.,’ Bd. xix, 1893; Taf. xix, fig. 17.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 163

have traced a similar groove, faintly representing the spermatic
structure relating to the penis in the male (PI. 15, fig. 15, g. v.).

In none of these cases, however, do these grooves open into
an internal chamber; and since in the female they can serve
no conceivable purpose, unless it be that of oviposition, we
must conclude that the groove, where it exists in those Proso-
branchs which do not possess a pouch, represents a vanishing
structure, and that the pouch, although it was probably pos-
sessed by the ancestors of these forms, has been completely
lost. In some forms, moreover, such as Ty phobia, all trace
of this complex accessory reproductive apparatus has entirely
disappeared, and there is no trace of either groove or pouch in
the female or the male.

In Tanganyikia, as in those forms of Melania which
possess the complete female apparatus I have described, there
is no penis, and only a slight spermatic groove represents the
apparatus in the male. But in Strombus (PI. 16, fig. 3, g.v.)
and Pteroceras the penis exists, and the groove is not only
pronounced in these genera, but, as is well known, it is pro-
longed along the posterior surface of the organ to the tip.
Modifications of the male intromittent apparatus can, however,
go even further than this. In some forms, such as Buccinum
(Pl. 16, fig. 4), there is no spermatic groove, nor indeed does
the vas deferens open in the mantle cavity at all. In these
forms the grooves have become enclosed, and as a simple
tubular continuation of the vas deferens pass through the
whole internal length of the penis, and open at the tip. (Com-
pare figs., diagram I, Pl. 17.)

Various modes of modification of the parts of the complex
apparatus existing inTanganyikiaandMelaniaepiscopalis
are thus seen to be widely distributed among entirely different
prosobranchiate forms; and although, so far as is at present
known, the entire apparatus exists fully developed only in the
four species I have named, it must be concluded, from the fact
that traces of these structures exist in most widely separated
types, that they are representatives of an archaic condition
of the reproductive apparatus, and we may almost certainly

164 J. E. S. MOORE.

conclude that the fully developed apparatus was present in the
remote ancestors of all these forms. Hitherto little or no atten-
tion has been paid to the spermatic grooves which have
long been known to exist in the males of many forms, but
directly their relation to the fully developed female apparatus
which I have described becomes apparent their morphological
importance becomes immense, for it will doubtless have been
seen how closely the fully developed apparatus in these
female Prosobranchs corresponds to the spermatic grooves and
invaginated penes of the Opisthobranchs; and indeed, when
we look further into the matter, the comparison becomes so
striking and so compiete that there can be little doubt of the
existence in this feature of an actual bond of morphological
unity between the two. I may, however, preface the remarks
which I have to offer on this subject by stating that it has
already been shown, as a result of Professor Pelseneer’s elabo-
rate and painstaking investigations! concerning the morpho-
logy of the Opisthobranchs, that the characteristic hermaphro-
ditism of these organisms appears to have been secondarily
acquired, and to have arisen in the female by the evolution
of a functional male gland.

If we bear in mind this result of a profound research while
comparing the genital ducts in the Opisthobranchs with those
of the Prosobranchs which I have just described, much of the
initial prejudice that would be likely to exist against such a
comparison will disappear, and it will be more readily seen
that in their simplest forms the external genitalia in both
orders are exceedingly similar.

In Aplysia, Pelta (PI. 16, fig. 5), and several other forms
the hermaphrodite genital aperture, as is well known, leads
into a forwardly extended groove (Pl. 16, fig. 5, g. v.), which
terminates anteriorly in a pit beneath the eye, and this pit
communicates with the cavity of an internal sac, the walls of
which can function as an introvertible penis when required.
Except in the addition of the muscles which introvert this sac,

1 «Arch. de Biol.,’ t. xiv, 1895. See also ‘ Quart. Journ. Mier. Sci.,’ vol.
37, p. 19, 1895.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 169

there is absolutely no difference in the relation of the curious
external genital apparatus in Pelta or Aplysia, and those
similar structures which T. rufofilosa and several Melanias
are now known to possess. (Compare figs., diagram I, Pl. 17.)
It is true that the genital opening in the Opisthobranchs is
that of an hermaphrodite duct, but we have seen that there is
every reason to regard this character as secondarily acquired ;
and just as would be expected from Professor Pelseneer’s view,
that the female first acquired the hermaphrodite character in
these forms, we find the sac, which in its origina! ancestral
prosobranchiate condition was a brood-pouch, here converted
with very little modification into an introvertible penis.
There seems, therefore, to me to be no admissible objection
to the conclusion that the parts of the external genital
apparatus which are present in T. rufofilosa and Melania
episcopalis are structurally homologous with the similar
parts of the external genital apparatus in the simpler Opistho-
branchs. Clearly, therefore, in the existence of these curious
structures which I found in T. rufofilosa and M. episco-
palis, we have still more evidence which will help to bridge
the rapidly diminishing gap between the two molluscan orders
in which they exist.

The ordinary external genital apparatus of the Opisthobranch
undergoes similar modifications to those occurring in the
Prosobranchs, but it exhibits these modifications in a more
pronounced degree. Thus the simple open grooves of Pelta
and Aplysia pass through modifications such as that appear-
ing in Auricula myosotis (Pl. 16, fig. 6), where an external
groove still exists, but the vas deferens has become invaginated,
so as to form a distinct internal tube. In Lobiga and
Actzon (PI. 16, fig. 7) the groove has apparently disappeared,
and the vas deferens, as in Buccinum among the Proso-
branchs, appears as an enclosed tube running to the extremity
of a permanent external penis (PI. 16, fig. 7).

Lastly, in Helix (see diagram, Pl. 17), both male and
female conduits may become enclosed as separate tubes, their
original connection with one another only appearing in early

166 J; £. 4S: MOOR.

ontogenetic life, and the line of their invagmation remaining
as a faint streak in the adult.

Whether we deal with the Prosobranchs or the Opistho-
branchs in this matter of the modifications which the original
external genital apparatus may undergo, it will thus be seen
that in both orders the modification is along the same lines,
and results in the conversion of an open groove into one or
two closed conduits as the case may be. In both orders the
original brood-pouch of the Prosobranchs and the introvert-
ible penis of the Opisthobranchs tend to become lost, and to be
replaced by permanent external penes, supplied with secondarily
acquired internal tubes. Very obviously, therefore, we are
here dealing with one of those numerous examples of parallel
modifications of an originally similar structure present in
two types which have become distinct. The whole apparatus
in its complete ancestral prosobranchiate form is too complex
and too peculiar in the arrangement of the different parts to
have been evolved twice from different things. We know
further that the Melanias to which I have referred, as well as
the T. rufofilosa of Tanganyika, are indubitably extremely
ancient forms, and that the complete structures they possess
were once widely distributed among their prosobranchial
ancestry, for parts of these structures as we have seen are
present in widely different prosobranchiate forms. It would
appear, therefore, that we are fully justified in concluding that
in all these cases we are dealing with an incompletely repre-
sented condition of the reproductive apparatus that was once
common to the Prosobranchs and the Opisthobranchs alike.

Lastly, it will have been seen that this view, drawn from
the study of Prosobranch morphology alone, and Professor
Pelseneer’s view respecting the origin and nature of the Opis-
thobranchiata, mutually confirm each other; for, as I have
already pointed out, Pelseneer has shown from an entirely
different line of inquiry—

1. That the Opisthobranchs arose from the Prosobranchs.

2. That their characteristic hermaphroditism was secondarily
acquired by the females of this group.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 167

How exactly this view accords with the results of the
anatomical comparison just given will be at once apparent.
And in conclusion I may add that, when I arrived at the
views above stated I was ignorant of Pelseneer’s work, and of
the conclusion at which he had arrived.

Spekia zonata.

The shell of this remarkable mollusc is represented in Pl. 18,
fig. 1, and it is certainly most curious that no attention has
been drawn by any of the conchologists to the extremely nati-
coid character which it presents, for the shell of this species 1s
so completely similar to that of numerous fossil naticoid forms
that, had it appeared fossilised instead of having been found
living in a great fresh-water lake, there is not the slightest
doubt that it would have been placed in one of the numerous
fossil genera which are supposed to group themselves about
the living Naticas. The oblique aperture, and the tendency
of the outer wall of the mouth to be continued as a cup-shaped
ring round the bases of the older shells, are exactly what is
observed in many fossil Naticas; while the presence of a very
pronounced umbilical opening, which is more or less filled up
with a deposit of callous substance, are features which are
generally regarded as almost diagnostic of naticoid shells.

The external appearance of this form is superficially similar
to that of T. rufofilosa. The foot is rather less broad, and
the snout is not so much pigmented; but, apart from the
naticoid appearance of the shell, it is only in the internal
anatomy that we begin to appreciate the wide morphological
differences which exist between these forms.

In S. zonata the buccal mass is well developed, and the
radular sac is conspicuous, but not of any considerable length
(Pl. 18, fig. 6). There are two very strong muscles attaching
the buccal mass to the body-wall, and the salivary glands are
long and simple. The radular dentition is characteristic, and
very strongly developed. A single row of teeth is represented
in Pl. 18, fig. 3. It will be at once seen that the characters of

168 J. KE. S. MOORE.

these teeth are highly suggestive of those of a number of well-
known forms. In gross detail the radular dentition is very
similar to that occurring in various forms of Anchylotus
figured by ‘Troschel. The predominant denticle on the
admedian tooth is well developed in zonatus, as indeed it is in
a very large number of dissimilar forms; and an exactly
analogous and widely prevalent feature is presented in the
difference of size and character between the denticles on the
heads of the outer and inner lateral teeth. Perhaps, however,
the most notable feature which the radula presents is the
peculiar structure of the median tooth. The outer surface of
this tooth is concave, like the median tooth in Anchylotus,
Thiara, Melania brevis (Dorb.),and Melanopsis. But it
differs from all these forms in having no predominant median
denticle, there being instead two lateral predominant denticles
and a median concavity. The only forms which appear to
possess this peculiarity of the median tooth are the different
species of the genus Sigaretus; and although in other respects
the radula differs widely from that of either a Sigaretus ora
Natica, I shall show in the concluding part of the paper
that this comparison is not nearly so far-fetched as it might
appear, since it can be clearly demonstrated that what we may
call the melanio-planaxoid form of radula which S. zonata
possesses is that of an extremely ancient type, and in all
probability, with the exception of the Rhipidoglossa, is ante-
cedent to that of all the other Prosobranchiates with which
we are acquainted.

In 8. zonata the esophagus leads into a peculiar stomach,
represented in Pl. 18, fig. 2, which contains at its anterior end
a body which must be regarded as representing a crystalline
style, although it is so small and so little elongated that at
first sight it does not seem to have the characters which this
structure usually presents.

The intestine takes the course represented in Pl. 18, fig. 6,
and communicates with a dilated rectum, opening in the usual
way just within the border of the mantle. The heart has the
regular tzenioglossate characters. The description of this

THE MOLLUSOS OF THE GREAT AFRICAN LAKES. 169

organ given for T. rufofilosa (p. 158) would stand equally well
for that of S.zonata. In S. zonata, however, owing to the
naticoid shape of the body, there is a forward displacement of
the internal viscera, which results in the pulmonary vein not
going directly forward to the base of the stenidium, as in T.
rufofilosa and most other Prosobranchs, but in its being bent
slightly backwards in a more or less acute curve before it reaches
the base of the gill (Pl. 18, fig. 6). This relative displacement
of the heart and gill is fully described by Haller! in the Naticas
Trochita radians, Ergaea plana, Crepidula_ peru-
viana(Lam.), Tanacus unguiformis (Lam.), and Crucibu-
lum (sp. ?), and it is curious to find that this displacement has
proceeded less in Spekia than in any of the forms which
Haller figured.

The gill in S. zonatais very much like that of rufofilosa.
The leaves are similarly broad, low, and triangular, and their
apices are prolonged into the same littorinoid finger-like
processes (P1]. 18, fig. 2).

Such processes are figured by Haller in the gills of Natica
lineata: they are present in the gills of Littorina (Pl. 15,
fig. 1), but curiously enough they are not represented in
Haller’s figure of the gill of Sigaretus neritoides
(Lam.). Neither are they present in the gills of the
Faunus which I examined, and which are represented in
Pi 15;, fig.19:

The osphradium is lodged in a groove beneath the gill; it
is simple, and slightly but distinctly tending to become pec-
tinated (Pl. 18, fig. 2, os.). Atits outer extremity it is curiously
bent downwards and back, exactly repeating in this the con-
dition of the same organ in Sigaretus neritoides and
Natica lineata as represented by Haller (loc. cit., pl. xiii,
figs. 17—20).

The nervous apparatus of S. zonata is the most interesting
feature which this form presents, the whole completely simu-

1 Haller, “Die Morphologie der Prosobranchia,” iii, ‘Morph. Jahrb.,’
xviii, 1892, p. 451, and plates.

170 J. E. 8S. MOORE.

lating in every detail that of Lamellaria perspicua, as
figured and described by Bouvier! and Haller.

The cerebral ganglia are united by a short but distinct
commissure, and are almost completely fused with the pleural
ganglia (Pl. 18, fig. 5, 6.). The right pleural ganglion gives
rise to a nerve-cord which passes upwards and directly across
the axis of the animal’s body. It almost immediately expands
into a ganglionic mass, which represents the supra-intestinal
ganglion (Pl. 18, fig. 5, Sup. ant. g.).

This ganglion is in turn connected with the left pleural by
the commissure (PI. 18, fig. 5, a. #.). The nervous system is
therefore completely zygoneurous on the left. From the left
pleural ganglion a nerve-cord passes almost parallel to and
below the supra-intestinal commissure to a ganglionic enlarge-
ment which represents the subintestinal ganglion on the
right (Pl. 18, fig. 5 6, Sd. int. g.).. This ganglion is in direct
connection with the right pleural ganglion by the connective 2’.
The nervous system is therefore completely zygoneurous on
both sides. All these details could be made out by dissection
alone, but the animals were so small that for confirmation I
was obliged to resort to sections.

In all essential details the nervous system just described
corresponds with that of Lame]laria, and as the modifications
it presents are most complete and peculiar, there can be no
doubt that the nervous system of S. zonata offers a true indi-
cation of the naticoid affinities of this form.

The reproductive apparatus very much resembles that of
Littorina, the genital gland occupies the upper part of the
apical body-whorl, and in the male is related to a simple vas
deferens which opens at the extreme upper end of the mantle
cavity (Pl. 18, fig. 2). This opening is, however, prolonged
as a ditch or groove (PI. 18, fig. 2, g. v.), which is overhung by
a flap, which latter structure terminates at the more usual
position of the genital opening.

An exactly similar state of affairs is present in Littorina
(Pl. 15, fig. 1, gy. v.), and I found it also very well defined in

1 «Ann. des Sciences Nat.,’ Series 7, “ Zoolog.,” tom. iii.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES, yt

the specimen of Faunus to which I have referred (Pl. 15,
fig. 2).

There is little or no trace of the spermatic groove in S§.
zonata beyond the termination of the flap, but this groove
is very plainly indicated in this position in Faunus (Pl. 15,
fig. 11, g. v.) and in Littorina (Pl. 15, fig. 1, g. v.).

In all essential details the female apparatus of zonatus
repeats that of the male, but there is neither an external
groove nor pouch.

Comparative.

In attempting to ascertain with what known molluscan
forms the foregoing species are to be placed, it will be abun-
dantly apparent from the previous anatomical descriptions
that each bears little if any relation to the other; and further,
that from the broad features of their general anatomy the
existing conchological determinations of their affinities with
the Hydrobiide must be at once dismissed. As regards the
so-called S. zonata, the general anatomical features of this
form, the gills, the position of the viscera, the nerves, and so
forth, all place it unquestionably among the simpler Naticoids ;
the only features which at first sight might militate against
such a determination of its phylogenetic relationships being
those presented by the radula and dentition. We have seen
that the dentition of S. zonata is in many ways identical with
that Melanio-planaxoid type which I have already described,
and that it only corresponds with that of any of the Nati-
coids in the minor features of the median tooth. It is there-
fore incumbent on us to ascertain whether this difference in
the radula is really an important feature, or whether it is
not wholly or partly due to the antiquity or the specialisation
of the Tanganyika form. But in order to obtain a clear con-
ception of this matter it is necessary to examine the radula
question from a more general point of view. I have long been
impressed by the possibility that the usually adopted methods
of estimating the gross relationships of these structures simply

172 J. E. 8. MOORE.

by the number of teeth which the transverse rows contain
may not be sound, for it was clearly shown by Troschel himself
that the Rhipidoglossate radula of Trochatella could be re-
garded as having been naturally evolved from the Teenioglossate
radula of such a form as Cistula through the splitting up into
numerous segments of the outer lateral teeth.

He says, “Im Gebisswurdesich wohl von den Tzenioglossen
zunichst ein Uebergang zuden Rhipidoglossen verleiten lassen
indem die Cyclostomaceen einerseits die Amphiperaside,
anderseits durch die tiefen Kammirtigen Hinschnitte der
Seitenplatten auffallend und die ganz zerschnittenen alisseren
Seitenplatten der Rhipidoglossen mahnen. Ich ziehe es jedoch
vor, die Abtheilungen mit kammirtigen Kiemen, wenn gleich
das Gebiss, sehr betrichtlich abweicht, zunachst zu behandeln.
Anatomische Erforschung aller iibrigen Verhaltnisse und die
Entwickelungsgeschichte gestatten uns vielleicht, kiinftig hier-
iiber eine sichere Entscheidung.”

If we accept this view—and, as I shall immediately show,
there is every reason why we should—it consequently follows
that in the Rhipidoglossate and Tnioglossate types of

Rae Rate Aa a ae Oe ee
D) ) a. Median zone.
Ss

d. Lateral zone.

Diagram showing the corresponding zones and denticles in the Rhipido-
glossate radula of Cocculina Rathbuni, Dall. (right figures), and in the
Tenioglossate radula of Melania Episcopalis, Lea (left figures).

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 173

radulas represented in the accompanying diagram the areas
lettered a, b,c, d, are morphologically the same. But if this is
so, the comparison can be pushed much further, since it is
apparent that there is a very real morphological distinction
between certain areas of the dentition, which is quite inde-
pendent of the number of denticles which the individual areas
may contain.

What is of primary importance, however, is this, that in a
very large number of Rhipidoglossa the radula presents a
simple division into separate denticular areas, three on each
side of a median line, there being precisely the same morpho-
logical distinction between the elements in these areas as that
which subsists between the molar and premolar teeth of
mammals. We know that the Rhipidoglossa stand in the
relation of ancestors to the great Tznioglossal group, and the
question naturally presents itself whether the 3, 1, 3 radula of
the Tzenioglossa is not to be interpreted as a condensation into
seven single denticles of the more numerous Rhipidoglossate
teeth ; for in the latter group the denticles are already diffe-
rentiated into a similar number of morphologically distinct
denticular areas. But although Troschel pointed out that the
Tzenioglossate radula in certain forms approaches that of the
Rhipidoglossa, he seems for some reason or other to have
missed this fundamental similarity, which becomes apparent
throughout the radule of both groups if we consider the
denticular areas, as we have every right to do, and not the sepa-
rate teeth. In some Pleurotomaridz! this division into lateral,
angular, and median denticular areas is marked, and as this
division is present in most Rhipidoglossa we are fully justified
in regarding it as a fundamental feature of molluscan tongues.
In fact, the presence of such a division in a group like the
Rhipidoglossa appears to be as strong, if not stronger evidence
for the fundamental character of the arrangement than that

1 Dall., ‘Bull. Mus. Comp. Zool. Havard,’ vol. xviii, 1889, Pleuroto-
maria adansoniana, C. and F., pl. xxxi, fig. 3. It is not so apparent in
Bouvier’s figures of P. quoyana, ‘Arch. Zool. Ex.,’ de série, vol. vi, pl. xii,

fig. 1.

174 J. E. 8. MOORE.

which may and ought to be obtained by the study of the
ontological formation of the teeth. Consider also the very
marked existence of this division revealed by the study of the
existing figures of the Dochoglossa, while in the Cephalopods
themselves we frequently encounter the regular Tzenioglossate
type of tongue.

From these considerations it is clearly incontestable that in
a large series of molluscs of very different types there exists a
fundamental similarity of plan, or symmetry, upon which the
dentition has been formed; and the conception of this unity
of groundwork seems to me important, in that it throws light
upon the means whereby the radule of those groups which
have emanated from the Rhipidoglossa, the Tzenioglossa, the
Rachiglossa, the Ptenoglossa, and the Opisthobranchs have
respectively been evolved.

In the light of these general conclusions respecting the unity
of nature, and the probableinter-relationships subsisting between
the different molluscan dentitions, we may pass now with advan-
tage to certain more particular considerations, which are perti-
nent to the determination of the true phylogenetic position of
the molluscs with which this paper is primarily concerned.

We have seen that the Rhipidoglossate radula may be made to
pass, by a series of gradations, through forms like Trochatella
into that of the Tzenioglossa such as cystula; while in the
genus Ovula, the teeth in both the lateral and angular areas
tend to be split up. If the above conception of the forma-
tion of the Tzenioglossate radula from the Rhipidoglossate type
be true, we should expect to find some definite approach to
the Rhipidoglossate condition in the Archi-tznioglossa as
they at present exist. Through the work of Bouvier and
Lacase-Duthier there seems much reason to believe that the
Cystostomide have originated from the ancient Littorinas, a
group which, if fossil identifications can be trusted, is as old
as any that we know.

Now there is a whole series of gradations between the quasi-
Rhipidoglossate radula of the Cyclostomes and that of the
Paludinas, another unquestionably Archi-tenioglossate group;

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. Io

and if we look into the matter a little more closely, we sce
that the only real difference between the single outer lateral
tooth of the Viviperas and that of the Cyclostomes lies in the
fact that the former is somewhat more elongated vertically,
and a little less numerously serrated on the crown.

From this form of Tenioglossate radula we may pass, by
inseusible gradations, to what I may call the Cerithio-planaxoid
radulas, in which the outer lateral has become still more
specialised and perpendicularly lengthened out. Thus the
Cerithio-planaxoid radula comes very close to the quasi-Rhipi-
doglossate radula of the truly Archi-tenioglossate forms, and,
as we should expect from the abundant geological evidence
respecting the antiquity of the Cerithide, what I have termed
the Cerithio-planaxoid radula would rank among the more
primitive of the existing Tzenioglossate types. It is this con-
clusion which I wish to emphasise, for it has a most direct
bearing on the nature and affinities of Spekia zonata, as
I shall now proceed to show.

- It has been seen in the descriptive part of this paper that
S. zonata has the anatomical peculiarities, and especially the
nerves, of several Naticoids, but, unlike any Naticoid hitherto
examined, it presents us with a well-developed Cerithio-plan-
axoid tongue. The radula of the more typical Naticoids is very
marked in type, and more generally resembles the_radule of
the Xenophoride, the Chenopodide, and their associates, than it
does that of the Cerithio-planaxoid forms. On the other hand,
it will at once be admitted by anyone acquainted with such
structures, that among these more highly specialised forms
there exist numerous individual examples which exhibit more
or less distinct traces of the retention of the Cerithio-planaxoid
teeth. The radule of Vermetus, Crucibulum, and Crepi-
dula all show undoubted stages in the gradual transformation
of the Ceritho-planaxoid radula into that common to the more
aberrant types to which these genera respectively belong.

Now, unquestionably, S. zonata is extremely old. It lives
in association with forms which have every appearance of being
identical with several Jurassic molluscan types. Therefore

VoL, 42, PART 2,—NEW SERIES. M

176 J. E. S. MOORE.

the co-existence in this species of a Naticoid organisation
with a Cerithio-planaxoid tongue is exactly what, judging
from the foregoing observations, we should naturally expect ;
for the complete retention of a Cerithio-planaxoid radula in
zonatus indicates, like the rest of this animal’s anatomy,
that we have in this form what is probably the most primitive
Naticoid at present known.

Concerning the Lamellarian nervous system of S. zonata
it has already been rendered evident, by the work of Haller,!
how such a modification has been derived from the more
normal Gastropod types. To illustrate this I have diagram-
matically represented the nervous system of Sigaretus
neritoides, in which it will be seen that the shortening of
the subintestinal cord is very pronounced (PI. 17, fig. I). In
Natica lineata (PI. 17, fig. 11), on the contrary, the nervous
system remains simply zygoneurous and bilaterally symmetri-
cal, with nothing remarkable about it except the simultaneous
shortening of the supra- and sub-intestinal cords. In fig. 4 is
represented the nervous system of S. zonata, and it will be seen
that the peculiarities which this form presents are primarily
due simply to a more complete shortening up of the supra- and
sub-intestinal cords. In fig. 5, which represents the nervous
system of Crucibulum, another line of modification is intro-
duced, for in this form it is the sub- and not the supra-intestinal
cord which is becoming short. A still further progressive
shortening of the subintestinal cord is witnessed in Trochita
radians (fig. 6), and the most extreme case of this progressive
development is represented in Crepidula. These modifications,
it will be observed, are the exact inverse of those obtaining in
Sigaretus (fig. 1). It is thus apparent that Sigaretus and
Crepidula represent the extreme terms of modification in
opposite directions of a mean type of nervous system, which is
exemplified by that of Natica lineata.

We witness thus the very instructive fact that both these lines
of modification have been carried to opposite extremes in genera
which are unquestionably close allies. S. zonata (fig. 4)

' Loe. cit.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES, 177

does not incline to either of these extreme types, but. like
Lamellaria (fig. 3), it represents a third modification, due to
the simultaneous shortening of both the supra- and sub-intes-
tinal cords. It would appear, therefore, that S. zonata approxi-
mates, both in its general anatomy and in the minutiz of its
nervous system, to those Naticoid forms which have been
already examined.

Unquestionably, if judged by the nervous system alone, it
would be placed, not only near, but within the genus Lamel-
laria; but, since it differs from this genus in its external form,
and from all other Naticoid genera in the characters of its
radular dentition, I think it will be well to keep it in a genus
by itself, which is to be regarded as closely related to, but
distinct from, Lamellaria. Unquestionably Spekia can
no longer be regarded as having any relation whatever to the
members of the family to which the genus Lithoglyphus
belongs, and with which 8. zonata was unhesitatingly placed
by the conchologists. We have here, therefore, obviously one
more example of the impossibility of making correct determi-
nations from the shell structure alone, while at the same time
it forms an equally striking instance of the marine nature of
the halolimnic fauna of the lake in which it lives.

Turning now to the question of the affinities of the genus
Tanganyikia, it will be seen that the nerves, no less than the
general anatomy of this genus as represented by T. rufo filosa,
correspond to those of a number of well-known molluscous
forms. I showed in the descriptive part of this paper that, but
for the ventro-median protuberance on the cerebral ganglia,
the nervous system of rufofilosa corresponds almost exactly
to that of Melania amarula, as described by Bouvier (loc.
cit.), and with that of the closely allied species of Melania
represented in Pl. 15, fig. 5. There is nothing in the general
anatomy, or in the reduced radula of T. rufofilosa, which need
lead us to suppose that it is not really closely related to both
these types. The only marked anatomical difference presented
lies in the possession by T. rufofilosa of the curious accessory
reproductive apparatus which I have described. But I have

178 J. E. 8. MOORE.

already shown that this apparatus is undoubtedly to be regarded
as indicatve of an archaic character in those forms which possess
it, and the absence of the complete structure in these Melanias
is therefore no indication that both they and the genus Tan-
ganyikia are not direct descendants of the same approximate
phylogenetic stock.

If, however, we were to assume from these considerations that
T. rufofilosa and the amarula type of Melania are repre-
sentative of the whole heterogeneous assemblage of organisms
which are at present regarded by the conchologists as belong-
ing to the single family of the Melaniade, we shall be speedily
disillusioned. In Pl. 15, fig. 3, I have represented the nervous
system of Melania episcopalis, and in fig. 4 that of
Melania aspirata, while in fig. 10 the similar nervous
system of Faunus is shown in the relation to the gills and the
surrounding parts. All these nervous svstems are obviously
constructed on a similar plan, and they are all completely
different from that type of nervous system which M. amarula
and T. rufofilosa both possess.

But not only do these types of so-called Melanias differ in
the character of their nerves; both series are easily distin-
guished by other anatomical features which they possess. In
M. episcopalis, M. aspirata, and Faunus, the radular
dentition, as represented in Pl. 15, fig. 14, is obviously of that
Littorinoid character which is so marked in the numerous
American forms, such as Io, and all the representatives of the
genus Pachychilus.

The conclusions to which the extensive researches of Bouvier
into the character of the nervous systems of the Cerithide and
the Melaniidz led him were briefly these: that the simpler
dyaloneurous Ceriths form the marine starting-point for what
may be called the Cerithio-melanias on the one hand, and for
such forms as the Potamids and the genus Tympanotomus on
the other. Secondly, that the Planaxids had nothing to do
with the Ceriths, but were to be referred to a Littorinoid
ancestry. This view of the matter, I believe, is in the main
correct, but it does not account for the characters which a

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 179

number of the Melaniide possess, for there are, as we have seen,
two distinct types of so-called fresh-water Melanias; one charac-
terised by the possession of a Cerithio-planaxoid type of radula,
the other by the dentition of the modern Littorinoid group.

Now if Bouvier’s view that the Planaxoide are related to
the Littorinide be correct, the modern Littorinas must have
become modified in their teeth, for the Planaxoid radula is the
older of the two (see p. 174). It follows consequently that
those Melanias which possess a Cerithio-planaxoid radula are
relatively the older fresh-water types; but they may be con-
sidered as having arisen directly either from the marine Ceriths
themselves, or from a Littorino-planaxoid of the older type.
In whatever way we view this matter, it is at any rate obvious
that the Melaniide are no real family group, but can certainly
be split into two broad divisions, one of which originated as a
fresh-water stock from the more modern Littorinas, and may
for present purposes be called the Littorino-melanoid
group. The other has a Planaxoid radula, and the exact
origin of its constituent forms is much more difficult to deter-
mine, for in the characters of their radule these species are
equally similar to both the Cerithidz and the Planaxide, nor
are there any other features at present known by which,
in the majority of cases, it would be possible to determine
satisfactorily from which oceanic stocks such Melanoids origi-
nally sprang. Unquestionably the shell and the internal
anatomy of many Melanias correctly indicate a Cerithioid
ancestry, as Bouvier supposed; but there are others, such as
M. episcopalis, which might equally well be supposed to
have originated from the older types of Littorinas which
possessed a Planaxoid radula, or even from the Planaxide
themselves. Until further investigation has been prosecuted
it seems, therefore, best to group all those Melanias with a
Planaxoid radula, whatever their origin, into one series, which
we may describe as the Melanio-planaxoid group.

In T. rufofilosa we have a form which in many ways would
typify the marine ancestry of many Melanio-planaxoids, and
the exact relation of this form to the living oceanic Planaxide

180 J. E. S. MOORE.

needs further working out. In the meantime, however, it
could certainly be regarded as a representative of that ancient
Littorino-planaxoid stock from which a portion of the Melanio-
planaxoids appear to have originated. This view appears to
me to receive a certain amount of support from the fact that
T. rufofilosa possesses a reproductive apparatus fully deve-
loped, which is at present only known in its complete form in
some of the Littorino-melanoids ; for this would appear to indi-
cate that both rufofilosaand some of the Littorino-melanoids
still possess characters which were once common to their
ancestral Littorinoid stock. Unquestionably the possession of
the complete reproductive apparatus present in rufofilosa
stamps a mollusc as an ancient form, for in this feature of its
organisation we have seen (p. 160) that it harks back to a time
and to those types which were in existence before the Opistho-
branchs had become separated from the primitive Tzenio-
glossate stock. I conceive, therefore, that in T. rufofilosa
we have before us a surviving example of a form that is closely
akin to that ancient Littorina with a Planaxoid radula, out of
which the modern Littorinas and the modern Planaxide with
their respective fresh-water and terrestrial derivatives have
individually sprung. To illustrate the relationships of these
forms, we may refer to the diagram given on Plate 19.

To recapitulate: we have seen that there are indications that
the Melaniide are certainly capable of being split into at least
two groups which have no proximate relation to each other.
But there is yet a third type of Melania represented by the
genus Melanopsis, and this third group is in many ways
more interesting than either of the other two. As Bouvier
showed, there is very little in common between the zygoneurous
nervous system of the genus Melanopsis and that of any of
the other forms associated together at present in the family of
the Melaniide. Nor yet does this nervous system approximate
to that of any of the marine Ceriths or their derivatives that we
know. On the other hand, there isa certain amount of resem-
blance both in the nerves and radula of the genus Melanopsis
to the similar parts of the Nassopsis of Tanganyika. In

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 181

another paper! I have shown that two of these halolimnic
genera, Paramellania and Nassopsis, are conchologically
indistinguishable from the genus Purpurina of the old
Jurassic seas; and although the genus Melanopsis does not
come at all sufficiently near Nassopsis to be placed within it,
I think there is some evidence for regarding this Tertiary
genus as having arisen as a modification of the old Purpurine
stock which died out in the sea during the later Secondary
formations, but which it is possible still lives in that part of
the old Jurassic seas to which Lake Tanganyika appears to
correspond.

As I have shown elsewhere, the most remarkable feature
about the genus Nassopsis is the very curious way in which
the nervous systems of the forms belonging to it foreshadow
those of the terrestrial Cyclophoride. Hitherto no satisfactory
explanation has been given as to the origin of the Cyclophoran
nervous system ; and it is in the highest degree probable, from
the close similarity which it bears to that of the Jurassic
Nassopsis, that the Cyclophoride originated from some true
fresh-water derivative of the Purpurinas, such as the genus
Pyrgulifera, which is found in the fresh-water deposits of
the chalk. In the table (Pl. 19) I have represented the
Purpurina of Tanganyika as the direct continuation of the
similar forms found in the old Jurassic seas; while it is
suggested that the more modified Pyrguliferas of the chalk
may represent a true fresh-water uon-halolimnic development
of these forms.

It will, I think, be obvious that the observations here collected
in no way finally dispose of the interesting problem of the
origin of the heterogeneous constituents composing the Mela-
niide as the family at present stands ; but the splitting of this
group into three broad divisions, based on the anatomical
characters of such of them as have hitherto been investigated,
is unquestionably in accord with the morphological facts of
the case. What is immediately required is a more detailed

1 * Quart. Journ. Mier. Sci.,’ Vol. 41, pp. 814—317, 1898.

182 J. E. S. MOORE.

knowledge of the morphology of a large number of the
tropical forms. When this is obtained it is in the highest
degree probable that the Melaniide will be still further split
up, for we are at present in complete ignorance of the nature
of the animals contained in a large proportion of the shells
belonging to this so-called group. As Bouvier so justly
remarks, shells which are identical may harbour animals that
are quite distinct; and if any shell is referred to in existing
conchological works as belonging to the Melaniide, one is
naturally led to inquire to which of the heterogeneous groups
lumped together in that family it should be referred, for there
is as much difference between these forms as that subsisting
between the members of such distinct families as the Aphoride,
the Chenopide, and the Strombide, and to all such questions
no answer is in the nature of the case to be obtained. What I
have endeavoured to do has been to obtain the materials where-
with it may be possible to view a certain portion of the subject
from a phylogenetic standpoint, since in dealing with such a
series it will certainly not be contested that this is the only
rational method it is possible to adopt. The conchological
method of grouping like shells together may, as Bouvier re-
marked, be useful for the compilation of museum catalogues,
but it is without meaning from a broader zoological point of
view ; and the extensive use of this purely conchological method
of classification, which has recently been so much in vogue,
is all the more mischievous because the same terminology
is used in it as that which is applied to the classification of
animals the morphology of which is fully known, and it thus
lends an appearance of completeness to the existing molluscan
system of classification that in reality is merely an illusion
and a snare.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 183

DESCRIPTION OF PLATES 14—19,

Illustrating Mr. J. E. S. Moore’s paper on ‘‘The Molluscs of
the Great African Lakes.”

Reference Letters.

B. P. Brood-pouch. @. Csophagus. s¢. Stomach. g.g. Genital gland.
g.d. Genital duct. g.v. External genital groove. Sp.d. Spermatic duct.
g.a. Genital aperture. os. Osphradium. rad. Radular sac. c.s¢. Crystal-
line style. 7. Renal aperture. iz¢. Intestine. azt.ch. Anterior stomachic
chamber. c.g. Cerebral ganglion. Sd. ivt.g. Subintestinal ganglion. Sup
int. g. Supra-intestinal ganglion. p/. Pleural ganglion. p.g. Pedal gan-
glion. p/.p.com. Pleuro-pedal commissure. C.p.com. Cerebro-pedal com-
missure. c.com. Cerebral commissure. «2. Zygoneurous connection. 4a.
Anus. Jnt.g.g. Internal genital groove. d. Dyaloneurous connection. M. p.
c.g. Median process on cerebral ganglia. B.m. Buccal mass. V.2. Visceral
nerve. V.g. Visceral ganglion. V.d. Vas deferens. R.g. Rectal gland.

PLATE 14.

Fic. 1.—Shells of Tanganyikia rufofilosa, showing the character of the
operculum.

Fie. 2.—Tanganyikia rufofilosa with the mantle cavity opened up so
as to show the relation of the mantle organs.

Fie. 3.—The radular apparatus and upper portion of the nervous system of
rufofilosa.

Fic. 4.—Dissection of the viscera of rufofilosa.
Fic. 5.—A single row of teeth from the radula of rufofilosa.

Fic. 6.—The stomach of rufofilosa, showing the anterior chamber and
crystalline style.

_ Fie. 7.—The gills of Melania episcopalis; at a. the gill filaments
are simple and triangular, while at 4. they have the filamentous character of
the Littorinoids.

Fie. 8.—Right and left lateral views of the anterior portion of the
nervous system of rufofilosa,

Fic. 9.—The large extremity of the intestine of rufofilosa, showing
rectal gland.

Fie. 10.—Semi-diagrammatic view of the reproductive apparatus of the
female rufofilosa.

184. J. E. 8S. MOORE.

Fic. 11.—The pericardial cavity and heart of rufofilosa.

Fic. 12.—Rudimentary spermatic groove in the male Melania epis-
copalis.

Fie. 13.—Dissection of the female Melania episcopalis, showing the
relations of the super-cesophageal brood-pouch. A bristle is passed through
the opening of the brood-pouch, 2., into its interior cavity.

Fic. 14.—The mantle cavity of a female Tanganyikia rufofilosa,
showing brood-pouch containing embryos similar to that in Melania epis-
copalis (see previous figure), but subcesophageal and on the left side.

PLATE 15.

Fic. 1.—Partial dissection of Littorina litteria.

Fig. 2.—Posterior portion of the mantle cavity of a Faunus species (?).

Fic. 3.—Nervous system of Melania episcopalis.

Fic. 4.—Nervous system and radular sac of Melania aspirata.

Fic. 5.—Nervous system of Melania (?).

Fic. 6.—Nervous system of Tanganyikia rufofilosa.

Fic. 7.—Melania aspirata, showing gills and genital groove.

Fie. 8.—Buccal mass, cerebral ganglia, and radular sac of Melania
episcopalis.

Fries. 9 and 10.—Gills and radular sac of Faunus species (?).

Fie. 11.—Lower portion of mantle cavity of Faunus, showing flab-like
covering of the internal genital groove, zz¢. g. g., and the external genital
groove, 7. v.

Fie. 12.—Nervous system of Littorina litteria.

Fic. 13.—Mantle organ of the Melanias, species (?).

Fic. 14.—A single row of teeth in the radula of Melania episcopalis.

Fic. 15.—Male Littorina litteria, showing the rudimentary spermatic

Zroove, J. Vv.

PLATE 16.

Fic. 1.—Semi-diagrammatic view of the reproductive apparatus in the female
Tanganyikia rufofilosa, showing the relation of the external genital
groove, g.v., to the internal suboesophageal brood-pouch, B. p.

Fic. 2.—Similar view of (a.) the female Melania episcopalis; (4) re-
productive apparatus of the male.

Fie. 3.—Semi-diagrammatic view of the reproductive apparatus in a male
Strombus, showing the relation of the spermatic groove to the permanent
penis.

THE MOLLUSCS OF THE GREAT AFRICAN LAK&S. 185

Fic. 4.—The same parts in a Buccinum, showing the enclosed spermatic
duct, v. d.

Fic. 5.—Diagrammatic representation of the reproductive apparatus in
Pelta, showing the similarity in position (in relation to the genital groove)
of the eversible penis to the brood-pouch of the Prosobranchs, figs. 1 and 2
(after Pelseneer).

Fic. 6.—The relation of the ducts in Auricula myosotis. Spermatic
duct, sp.d., is here enclosed.

Fic. 7.—The relation of the ducts in Acteon, showing similar modifica-
tion of the penis to that observed in Buccinum, fig. 4.

PLATE 17.

Diagrams showing—l, different modifications in reproductive apparatus
among the Gastropods. II, the different modifications in the nervous systems
of six Naticoids. Refer to pp. 163 and 175.

PLATE 18.

Fic. 1.—T'wo views of the shell of Spekia zonatus,

Fie. 2.—The gills and mantle organ of Spekia zonatus.

Fic. 3.—Single row of teeth in the radula of Spekia zonatus.

Fic, 4.—The snout and tentacles of Spekia zonatus.

Fig. 5.—Dorsal and left lateral views of the nervous system of Spekia
zonatus.

Fic. 6.—The relation of the visceral organs of Spekia zonatus.

Fic. 7.—Animal of Spekia zonatus removed from the shell.

Fic. 8.—Section through the genital groove of Spekia zonatus.

PLATE 19.

Diagram to show relationships of various genera.

: aay o pata he eA ee “eevee oie ag eed bt

ur’ rik tte.) se rz) iy ae he

ei 7 7, =
7 ele; Thy 7 ees eee ef iy >, rey le j

P pdt? eof Toe ora | ie Pero ee

=, , 4 a
io) an

4 . all i‘. :

ve be 7) ahi Mekal fC es ay

aa igs rel

ai Ay Re wr] Yer Vie: ed

: ; tune : eras mn

ey Le as:

r

il) ii ba ih é ait ipa Rae
7 ‘hs

p i tial

| 7 ary :
\ F 5 *
‘ : ‘1

a .*

.
”
%
-
U
*
’
‘
‘
rc
*'
S £
_—
4
*
-
7
“ é

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 187

The Molluscs of the Great African Lakes.
IV. Nassopsis and Bythoceras.
By

J. E. S&S Woore.

With Plates 20 and 21.

AmonG the numerous forms of molluscs that appear to be
peculiar to Lake Tanganyika at the present time, there are
three very marked genera—Paramelania, Nassopsis, and
Bythoceras, which, owing to the substantial similarity exist-
ing between their. shells, might, and have hitherto been
regarded as closely associated forms. The first of these to be
discovered, Nassopsis, was originally classed by 8S. P. Wood-
ward! among the Melanias, and as a member of the sub-genus
Melanella, his determination being made from some empty
shells which had been brought from Tanganyika by Captain
Speke, during Burton’s celebrated journey to that lake. The
second genus, Paramelania, was formed by Smith,” in 1881,
to include a somewhat similar form of empty shell which had
been brought from the same locality by the missionaries; and
to distinguish the members of this genus from the original
Melanella nassa of Woodward, Smith® placed the latter form
in the new genus of Nassopsis.

In 1896 I* dredged in the deep water of Tanganyika

1 «Zool. Soc. Proc.,”? 1857.

2 Ibid., 1881, p. 559.

3 «Ann. Mag. Nat. Hist.,’ 1890, vol. vi, p. 93.

4 The genus was first named in my paper in the ‘ Proc. Roy. Soc.,’ vol.
Ixii, p. 451. The full diagnosis is contained in the ‘ Proc. Mall, Soc.,’ 1898.

188 J. E. 8S. MOORE.

another form, which although it presents, when adult, the
peculiar horns above and below the mouth represented in
Pl. 21, fig. 3, still bears a most remarkable resemblance to
both Paramelania and Nassopsis. For this new form,
and before I was acquainted with more than the shells of any
of the three, I proposed the third generic name, Bythoceras.

Now, without further knowledge, anyone judging from the
appearance of the shells of these three different genera would
certainly regard them as being in all probability closely related
to one another. Therefore, holding the opinion that the con-
chological method of determining molluscan relationships is
utterly unsound, it is with the greatest satisfaction that I
am here enabled, by a detailed account of the anatomy of
Nassopsis and Bythoceras, to show that even in their
wider phylogenetic sense these genera bear no relation to
each other. The characters of both, however, are of far greater
importance to morphologists than that of affording them mate-
rial wherewith to exhibit the futility of attempting to determine
the nature of molluscan affinities from empty shells. The
anatomical features of Nassopsis are in many ways quite
unlike those of any forms hitherto described; and I shall
immediately make it clear that this form presents us with an
Archi-tznioglossa of an entirely new type. This being so,
when we consider the unquestionably vast antiquity of the
lake in which Nassopsis lives, and the fact that its shell,
along with numerous others still living in Tanganyika, is
specifically indistinguishable from forms which once abounded
in the old Jurassic seas,! it will be readily accorded by those
interested in prosobranchiate morphology that in Nassopsis
we are presented with an animal which in the future will
probably constitute one of the most important prosobranchiate
archetypes of which we are in search.

1 «On the Hypothesis that Lake Tanganyika represents an old Jurassic
Sea,” ‘Quart. Journ. Mier. Sci.,’ vol. 41, 1898, p. 303.

THE MOLLUSCS OF THE GREAT AFRICAN LAKHS. 189

(a) Nassopsis nassa.

During life this mollusc inhabits the surface rocks of Tan-
ganyika, and its shells are always richly encrusted with the
green alge which clothe the rocks for a considerable depth.
It is sluggish, and appears to browse within a very limited
area, like the Patellas of the ocean beach. The foot is broad,
somewhat pigmented, and quite white in places; the snout is
broad, black, and wrinkled, not protrusible, but retractile.
The tentacles are short and black, and the eyes are not carried

10.

Fic. I.—Stomach of Nassopsis nassa opened from above, showing relation
of crystalline style-sac to main chamber of stomach. 41. Bristle passed
from aperture of cesophagus along its interior. 2. Bristle passed from
aperture of intestine to the cut end of same. Crist. s. s. Crystalline
style-sac. a. crist. s. s. Opening between stomach and crystalline
style-sac. 0.d. Aperture of digestive gland. S.c. Spiral cecum.

190 J. E. S. MOORE.

on the tentacles themselves, but on secondary papille at their
posterior bases (Pl. 20, figs. 5, 12, 6). There is a well-de-
veloped mucous gland in the mantle cavity (PI. 20, fig.5, m.g.),
and the animal, unlike the genus Spekia, is viviparous. On
opening up the body from before there is found to be a tolerably
well-developed buccal mass (PI. 20, fig. 12, 5. m.) ; the radular
sac is of average length, and the salivary glands are somewhat
tortuous, simple saccular organs (fig. 12, s.g.). The radular
dentition is strong and of the Littorino-planaxoid type
(Pl. 20, fig. 8).

From a portion of the radula obtained by Smith, Guatkin
referred it to the types approximating to the genus Cerithium,
while Smith himself remarked upon its similarity to the
radula of the Planaxide. The cesophagus is long, narrow,
and simple, and leads into a large stomaehic chamber (PI. 20,
figs. 9 and 11, also Fig. I, page 189), on the walls of which
there are numerous glandular folds (fig. 11), and a very curious
and striking double hemispherical czecum on the floor (fig. 11,
and Fig. I, p. 189), between the folds of which the curiously
rectangular orifice of a single bile-duct opens (fig. 11, a.h.d.).
Besides this stomachic chamber there is an anterior diver-
ticulum into which the large stomach opens by a tubular aper-
ture, through which a bristle is represented as passing in fig.
11, w. #., and in this anterior stomachic chamber there lies an
almost spherical crystalline style. The intestine passes out of
the stomach beneath the tubular aperture between the posterior
and anterior stomachic chambers, as indicated by the bristle
represented in fig. 11, y. z., also Fig. I, p. 189. The intestine
is simple, almost straight, and towards its rectal extremity it
contains a number of glandular folds and striz.

The liver is large, and occupies the lower two thirds of the
last two whorls of the animal’s body. ‘There is a single bile-
duct, opening, as has already been stated, in the posterior
chamber of the stomach.

The heart has the normal tznioglossate characters, and con-
sists of a thin-walled auricle, a thick-walled ventricle, and a
short aortic trunk (Pl. 21, fig. 1). Between the auricle, ven-

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 191

tricle, and aortic trunks there are the usual valves. The gill
of Nassopsis is of average length, very simple in structure,
and consists of a large number of low, broad, triangular
leaves, the apices of which are not produced into filamentous
processes, nor ornamented in any way. The osphradium is
long and simple ; it lies in a groove at the base of the gill, and
shows no tendency to become pectinated or modified in any
way either before or behind (PI. 20, fig. 6, o.s.).

The nervous system of Nassopsis is extremely interesting,
and is certainly one of the most archaic tznioglossate types at
present known. The cerebral ganglia are widely separated
from one another (PI. 20, fig. 7, c.g.) ; and the pleural ganglia
are not only separated from the cerebral ganglia (fig. 10), but
quite below the cesophagus, the cerebro-pleural connectives
being consequently of great relative length. The supra-intes-
tinal cord springs directly from the right pleural ganglion,
passes up over the cesophagus, and carries the supra-intestinal
ganglion (fig. 7, sup. imt.g.). From the left pleural ganglion
there passes a fine nerve towards the supra-intestinal ganglion,
which appears to form a dyaloneurous connection with a deriva-
tive of the supra-intestinal nerves. Towards the right the
subintestinal connective passes from the left pleural ganglion
beneath the cesophagus straight to the subintestinal ganglion
(fig. 7, sub. int. g.). This ganglion is directly connected with
the right pleural ganglion by a thick cord (fig. 7, z.), and the
nervous system is therefore strongly zygoneurous on the right.
Above, the cerebral ganglia give off a number of anterior
nerves, which are distributed to the buccal mass and the
parietes of the head. Among these there are conspicuous the
tentacular nerves which pass separately to the tentacles and
ocular papille. The buccal ganglia are situated on the lateral
walls of the buccal mass, and are united to the cerebral ganglia
by connectives (Pl. 20, fig. 7, 6.9.). Near the origin of the
buccal nerves there arise two fine nerves, one from each cerebral
ganglion, which pass forward along the walls of the body, and
then bend down, uniting with each other below the mouth
(Pl. 20, fig. 7, L.com.). This connection appears, therefore,

VoL. 42, PART 2.—NEW SERIES. N

192 J. E. S. MOORE.

to be unquestionably the labial commissure described by
Bouvier as characteristic of a number of the Archi-tznio-
glossate and Rhipidoglossate types.

The cerebro-pedal connectives are very long (Pl. 20, figs. 7
and 10), and altogether the length of the cerebro-pleural, cere-
bro-pedal, and pleuro-pedal connectives gives to the nervous
system the longi-commissurate character described by Haller.

The pedal ganglia are united by a rather small connection,
and are prolonged into the foot along the course of two well-
developed scalariform pedal cords. Between these pedal cords
there exist ladder-like connections similar to those found between
the pedal cords of Cyclopherus.

The otocysts in Nassopsis are relatively immense (Pl. 20,
figs. 7 and 10, of.). They are situated well up on the course
of the pedo-pleural connectives, and the otocyst nerves pass
obliquely from them towards the cerebral ganglia, and are
not quite correctly indicated in figs. 7 and 10. The otoliths
are small, numerous, and rectangular, with the faces slightly
convex (fig. 10).

The reproductive apparatus in Nassopsis is similar in
many ways to that of Ty phobia, both male and female appa-
ratus occupying the same general position. In the male the
genital gland occupies the upper surface of the apical whorl in
the body, and is connected by several channels with a nearly
straight vas deferens, represented in Pl. 21, fig. 2,v.d. This
latter structure opens without any modifications along its
course by the slit-like aperture represented in PI. 21, fig. 5,
In the female the ovary occupies the same position as the male
gland, and in like manner it is connected with the nearly
straight oviduct, the lower portion of which, or that which lies
within the mantle cavity, forming a brood-chamber where the
eggs go through their later stages of development (Pl. 20,
fig. 6, 6. 8.).

(Bs) Bythoceras.

Like Nassopsis, the genus Bythoceras, so far as is at
present known, is exclusively restricted to Tanganyika, and as a

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 193

member of the halolimnic fauna of that lake is of considerable
interest. Firstly, because, although conchologically so similar
to the genera Nassopsis and Paramelania, it has, as we
shall immediately see, no morphological relation to the former
of these types. Bythoceras is interesting, secondly, because
it presents us with more numerous points of correspondence
with forms such as the genus Tympanotamus, which exists
elsewhere, and the anatomy of which is known, than is the
case with the majority of the halolimnic forms. Bythoceras
is at present represented in Tanganyika by a single species,
B. iridescens, which I dredged living at great depths! in the
southern portion of the lake. When young the shell does not
possess the characteristic spines represented in PI. 21, fig. 3
(compare PI. 3, fig. 4). Nor has it the peculiar pearly thicken-
ing of the mouth invariably present in the older forms. In the
young condition (PI. 21, fig. 4) the shell is extremely similar to
that of Paramelania, and I am inclined to think that the
figure of Paramelania crassilabrio given by Professor E.
von Martens in his work ‘ Beschalte Weichthiere, Deutsch.
Ost-Afrikas’ (PI. vi, fig. 38), is, in reality, that of a young
Bythoceras iridescens.

The outward appearance of the animal is extremely similar
to that of Cerithium vulgatum, with the exception that
there is less pigmentation of the foot, which is nearly white
in the Tanganyika species.

The snout is short, wrinkled, richly covered with black
pigment, and non-protrusible. The tentacles are short, and
the eyes are situated on the posterior bases of these organs,
and not separated from them on subsidiary papille as in
Nassopsis. The buccal mass is small, and the radular sac
short, being reduced, as in the case of Typhobia and Tan-
ganyikia, to a small swelling on the floor of the cesophageal
tube (Pl. 21, figs. 7—11, 7. s.).”

The radular dentition is extremely interesting, a single row

1 From 300 to 1000 feet.

2 Compare figures of Tanganyikia rufofilosa, preceding article (loc.
cit.).

194 J. Ri OS: “MOOR.

of teeth being represented in Pl. 21, fig. 9. The outer and
inner lateral teeth distinctly resemble those of the Melano-
planaxoid type which I! described in considering the relation-
ships of the genus Tanganyikia, and are something similar to
those of the “‘ Neomelanian” group of the brothers Saracin ;?
while the admedian tooth is peculiar, owing to the presence of
two small subsidiary denticles on the inner face (PI. 21, fig. 9), a
very peculiar feature, and one which is only exemplified in the
radula of Tympanotamus. The presence of this peculiarity,
together with the general character of the radula, should cer-
tainly be regarded as of weight in diagnosing the nearer affini-
ties of this form. And, as we shall see in placing it along
with Tympanotamus, it is certainly in accord with the rest
of the animal’s morphological peculiarities.

The cesophagus and salivary glands (Pl. 21, fig. 7, s.g.) are
in all ways similar to those of Ty phobia, but these characters
are common to so many different kinds of Gastropods that they
are of little value from a special morphological point of view.

The stomach has two chambers, the anterior of which (PI.
21, fig. 13) contains a style. The intestine is simple, and takes
the course represented in Pl. 21, fig. 13. The rectum is not
dilated, nor beset with any accessory gland. The bile-ducts
seem to open by two very small apertures upon the base of the
posterior stomachic chamber. Stomachic valves are feebly if at
all developed. The liver is large, and occupies much the same
position as in Nassopsis (loc. cit.). The excretory organ
occupies a place in front of and above the heart, and opens by a
minute pore at the extreme upper end of the mantle cavity
(Pl. 21, figs. 5 and 13, 7.a.).

The heart has the usual teenioglossate characters, consisting
of an auricle, ventricle, and aortic trunk; but the last struc-
ture is much less developed in Bythoceras than in many
forms,—as, for example, in the genus Typhobia.®

1 Loc. cit.

2 “Die Siisswasser-Mollusken von Celebes ” (Wiesbaden, C. W. Kreidel's
‘Verlag’), 1898.

? Loc. cit., ‘Quart. Journ. Micr. Sci.,’ vol. 41, 1898, p. 190.

THE MOLLUSOS OF THE GREAT AFRICAN LAKES, 195

The nervous system in Bythoceras is very interesting,
since it is absolutely unlike that possessed by the genus
Nassopsis, and closely simulates the type described by
Bouvier } as typical of the genus Cerithium. It also strongly
resembles that of the genus Tanganyikia. Viewed from
above (Pl. 21, fig. 6), the cerebral ganglia are seen to be
closely fused together, while the left pleural and subintestinal
ganglion, as in Cerithium, form a single massive trunk,
which at its hinder extremity gives rise to the subintestinal
and visceral nerve-cords, and to the right pallial nerve (fig. 6).
From the right pleural ganglion a nerve passes out to the
mantle, and a branch from this anastomoses with a branch on
the pallial nerve just described. In like manner on the left
the pleural ganglion gives birth to a nerve on that side (fig. 6),
which passes out and probably anastomoses with a twig given
off from the supra-intestinal ganglion, but I was not able to
trace this nerve throughout its entire course.

Unlike the subintestinal ganglion, the supra-intestinal gan-
glion is carried on a very long supra-intestinal connective
(Pl. 21, fig. 6, sup. int. g.) exactly as it is in Cerithium
or Aphorais. Viewed from the side (Pl. 21, fig. 10) the
cerebro-pedal and pleuro-pedal connectives are seen to be of
considerable length, rather longer than the same structures
in Voluta, but not so long as those in Nassopsis or in
Strombus. The pedal ganglion has the usual bulbous form
as in the true Cerithide, and in like manner there pass from
the lower extremity of each pedal ganglion two predominant
foot nerves (fig. 10).

The otocysts lie behind the pedal ganglia, and the otocyst
nerves pass directly between the cerebro-pedal and _ pleuro-
pedal connectives to the cerebral ganglia on each side (Pl. 21,
fig. 10). The otocysts are not large, and are round, as dis-
tinguished from those of Nassopsis. The otoliths are small,
rectangular or barrel-shaped, and numerous.

The reproductive apparatus is very simple, and in both
sexes consists of a genital gland which occupies the upper

1 *Ann. Dis. Sci. Nat.,’ 1887, pp. 181—155, pl. vii, and figs.

196 Js Sa SS MOORE:

surface of the last two whorls of the animal’s body. This
gland is put into connection with a large non-convoluted
oviduct or vas deferens, as the case may be, by a number of
fine tubes; and both ducts pass beneath the intestine and open
just behind the anus in a large slit (PI. 21, figs. 5 and 13,9. a.).

The genital duct in both sexes is much enlarged within the
mantle cavity, somewhat in the manner of the same structure
in the genus Typhobia. But in Bythoceras it is quite
destitute of the singular organ which I described as an evertible
penis in the male Typhobia.!

Comparative.

In considering the phylogenetic relationships of Nassopsis
and Bythoceras, it will be needless after the foregoing descrip-
tion to insist further upon the fact that these genera bear no
relation whatever to each other. Taking Nassopsis first, it
will have already been clearly seen that the characters of this
singular genus place it unquestionably among the archi-teenio-
glossate forms. Whatever opinion one may hold as to the
value of the characters of the radula, it will have been seen that
they too place it among the more primitive portion of the great
melano-planaxoid group ; and beyond this it is doubtful whether
the radula can be used in a diagnostic sense at all. The large
buccal mass, long radular sac, and the characters of the salivary
glands certainly recall the littorinoid group, and at once dis-
sociate Nassopsis from the early Stromboid or Xenophoran
type to which Typhobia, Bathunalia, and Tanganyikia
all appear to bear more or less distinct affinities. The ex-
tremely archaic and simple condition of the whole digestive
tract in Nassopsis requires particular attention, for it will
have been seen by any one acquainted with my former accounts
of the anatomy of Typhohia, Tanganyikia, Bythoceras,
and even Spekia, that the digestive apparatus, and especially
the stomachic portion of the digestive apparatus in ail these
molluscs is built upon the same general plan. They all possess -

1 * Quart. Journ. Mier. Sci.,’ vol. 41, 1898, p. 191, loc. cit.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 197

anterior stomachic diverticula which contain crystalline styles,
similar to those of the genera Strombus, Pteroceras, Ros-
tellaria, Murex, and Trochus; and it has already been
shown in aformer paper! that there is no reason whatever to
doubt that these structures, when they appear in the Proso-
branchiata, are like the heart when perforated by the rectum,
as in the Trochide, to be considered as the retention of a con-
dition common to the ancestors of both Prosobranchs and
Lamellibranchs alike.

The presence of these similar arrangements in so many of the
halolimnic Gastropods, which in all other ways are so widely
separated from each other, is thus only intelligible if we
suppose that peculiar stomachic apparatus was once universal
among the ancestors of the Prosobranchiata of the present
day, and that in the remnant of an old fauna, such as that
now existing in Tanganyika, we encounter a more abundant
representation of an archaic state of things.”

The presence of a well-developed mucous gland in Nas-
sopsis, the extremely simple character of the gills and the
osphradium, the simple reproductive apparatus, are features
which all further dissociate Nassopsis not only from the
hitherto known fresh-water forms, but also from the more
modified Teenioglossa at present existing in the sea.

All the preceding structural peculiarities which demonstrate
the archaic character of Nassopsis are, lastly, fully substan-
tiated by the curious condition of the nerves. The great rela-
tive length of the cerebro- and pleuro-pedal connectives suggests
the condition found in Haller’s so-called “ longi-commissurate
forms,” such as Strombus, Pteroceras, and their allies ;
while in the presence of a labial commissure we are confronted

1 ¢Quart. Journ. Micr. Sci.,’ loc. cit., pp. 198, 210.

2 In the case of Nassopsis we can go, however, further than this; for com-
paring the spiral cecum in the stomach of this genus with the well-known
and similar structure in the stomachs of the Rhipidoglossa, the two are found
to be apparently identical, and the retention of this typically primitive
feature in a form which also in other ways is so distinctly archi-tenioglossate
is a morphological fact of the highest interest.

198 J. E. 8 MOORE.

with a feature which, like the cecum, connects Nassopsis
on the one hand with the Rhipidoglossa, and on the other
with the Archi-paludines. In like manner the wide separation
of the cerebral ganglia from one another, and the comparatively
great length of the cerebro-pleural connectives, are certainly
features which recall the primitive scattered condition of these
parts. In fact, the position of the pleural gangliain Nassopsis
represents a condition of development which is really halfway
between what I may term the hypo- and the epi-athroid
types, the nervous system of Nassopsis marking, in fact,
a third, intermediate, or dystenoid type.

Again, in the scalariform pedal cords with which Nas-
SOpsis is provided we have another most important feature,
showing that this form is not far from the border-land between
the older groups of Tenioglossa and their Rhipidoglossate
ancestry.

Nassopsis in many ways is more primitive than Paludina,
but at the same time, as will have been seen, it bears no very
approximate relation to this form; neither is it very near the
ancient group of Littorinas, nor indeed to any of the indi-
viduals which at present are regarded as constituting the
archi-teenioglossate group.

The morphological interest of Nassopsis lies, therefore, in
this, that it presents us with a new starting-point from whence to
study the inter-relationships of the great Prosobranchiate
order.!

Nassopsis, therefore, presents us with a type of organisation
which there are conchological reasons for believing is similar
to that possessed by several species of a genus which was
once abundant in the sea, but which has long since become
extinct outside the confines of Lake Tanganyika, where,

1 I shall elsewhere emphasise more fully the importance of Nassopsis
as an archetype, and it is therefore needless for me here to do more than point
out the fact that the above conclusions respecting the great morphological
antiquity of this form are fully substantiated by the very unexpected simi-
larity which three of the rather marked varieties of the shells of Nassopsis
present to the genus Purpurina from the old Jurassic seas.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 199

strangely enough, it appears to have lived on, not alone, but
in company with several other usually extinct marine Jurassic
forms.

Turning now to the characters of the new genus Bytho-
ceras, we find no such peculiarities as those I have just
described as distinctive of Nassopsis nassa. But never-
theless this genus is not without much interest from a mor-
phologist’s point of view. Nowhere, so far as I am aware,
have we a better instance of the fact that shell structure as a
means of classification is a “* delusion and a snare ;”’ for in the
case of Bythoceras it is not only that the shell in no way
foreshadows the animal inside it, but that its surprising simi-
larity to Nassopsis is absolutely misleading, and, were we
ignorant of the animals contained in both, would lead to a
profound error in the close conchological association of the
two genera.

Who would have dreamed, when contemplating the heavy,
thick, and highly ornamental shell of Bythoceras Howesii,
that the animal it contained bore any close resemblance to
that enclosed in the strangely different shell of the genus
Tympanotamus? Yet in the most peculiar features of
its radula, nerves, gills, and viscera it strongly resembles
Cerithium, Tympanotamus, and their close allies.

The reproductive organs of Bythoceras are simple and
peculiar, and in many other ways the structure of this organism
will afford ample material for conjecture concerning the final
identification of the really primitive cerithoid type. For
which is more ancient, Bythoceras or Cerithium? I do
not feel competent to auswer this question, and I would refer
the reader to the broader discussion of this subject in my longer
paper on “The Prosobranchiate Mollusca” (in the hands of
the editor of the ‘Quart. Journ. Micr. Sci.’).

This much, however, may be affirmed with certainty, that
Bythoceras is a form closely related to our general concep-
tion of the genus Cerithium, with some of the minor features
in its radular dentition peculiar to and distinctive of the genus
Tympanotamus.

200 J. E. S. MOORE.

In conclusion it may be interesting to reflect how all the
evidence which has been collected concerning the nature of the
halolimnic Gastropods invariably points to the vast antiquity
of these forms. First we have the wide dissimilarity of their
empty shells from those of any living types; next their rigid
isolation to a solitary great lake, which, judged from whatever
standard we may choose to adopt, is unquestionably of an
enormous age. Next we have the wonderful similarity of the
halolimnic shells now living in Tanganyika to those which
have been left fossilised at the bottom of the old Jurassic seas ;
and lastly, there are the morphological characters of the
halolimnic animals themselves, whereby they become mentally
depicted like nothing so much as the incompletely developed
embryos of numerous living oceanic types.

EXPLANATION OF PLATES 20 and 21,

Illustrating Mr. J. E.S. Moore’s paper on *f The Molluscs of
the Great African Lakes.”

Reference Letters.

a. Anus. g.g. Genital gland. g.a. Genital aperture. m.g. Mucous
gland. Int. Intestine. os. Osphradium. col. m. Columellar muscle. op.
Operculum. 7s. Radular sac. s.g. Salivary gland. £B.m. Buccal mass.
eryst.s. Crystalline style. @. Gisophagus. a.h,d. Aperture of hepatic duct.
x. Kidney. &. Renal aperture. &.s. Brood-sac. g.a. Genital aperture.
st. Stomach. /f#. Foot. e.g. Cerebral ganglion. p/.g. Pleural ganglion.
p.g- Pedal ganglion. /.com. Labial commissure. 6.9. Buccal ganglion. of.
Otocyst. swb.int.g. Subintestinal ganglion. sup. int. g. Supra-intestinal
ganglion. os.g. Osphradial ganglion.

PLATE 20.

Fic. 1.—Two views of a variety of the shell of Nassopsis.

Fic. 2.—T wo views of another variety.

Fic. 3.—Embryonic shells and protoconch of Nassopsis.

Fic. 4.—Animal of Nassopsis removed from shell, showing the oper-
culum, op.

THE MOLLUSCS OF THE GREAT AFRICAN LAKES. 201

Fic. 5.—Male Nassopsis, showing the mucous gland, tentacles, and
genital aperture.

Fig. 6.—Mantle cavity of female Nassopsis, showing character of gill
and osphradium.

Fic. 7.—The nervous system of Nassopsis dissected from above.

Fig. 8.—Dissection of Nassopsis, showing relations of stomach, crysta-
line style, kidney, esophagus, and intestine.

Fic. 9.—The radular elements of Nassopsis.

Fic. 10.—Nervous system of Nassopsis dissected on the right side.

Fie. 11.—Details of the alimentary tract of Nassopsis.

Fic. 12.—The relations of the buccal mass, radular sac, and salivary glands
of Nassopsis.

PLATE 21.
Fic. 1.—The relation of the heart, pericardial cavity, and reno-pericardial
connection in Nassopsis.
Fic. 2.—The reproductive apparatus of a male Nassopsis.
Fic. 3.—Two views of the shell of Bythoceras Howesii.

Fic. 4.—Young shell of Bythoceras Howesii, showing the absence of
the spines above and below the mouth.

Fig. 5.—Dissection, showing the gill and mantle organs of Bythoceras.
Fic. 6.—The nervous system of Bythoceras dissected from above.
Fic. 7.—The buccal mass, radular sac, and salivary glands of Bythoceras.

Fic. 8.—The nervous system of Cerithium vulgatum dissected from
above.

Fic. 9.—The radular elements of Bythoceras.

Fic. 10.—The nervous system of Bythoceras dissected from the right
side.

Fie. 11.—Right side view of the buccal mass, salivary glands, and radular
sac of Bythoceras.

Fig. 12.—The operculum of Bythoceras.
Fic. 13.—Details of the alimentary apparatus of Bythoceras.

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 203

Further Study of Cytological Changes produced
in Drosera.

Part II.
By

Lily H. Huie,
Physiological Laboratory, Oxford,

With Plate 22.

Tuis research is an expansion of that previously undertaken,
of which an account has been given in this Journal, vol. 39.

Aims.

The present aim was to ascertain—

(1) Whether substances which differ chemically will produce
specific alterations in either the cytoplasm or the nucleus.

(2) To what extent the cytoplasm and the nucleus are
interdependent.

(3) Whether the rapidity with which changes occur depends
on the ease with which substances are absorbed.

(4) The relation between cytological changes and the irri-
tability of leaves, the latter being calculated from the rate of
closure of the tentacles.

(5) What effects can be produced by bringing healthy cells
in contact with supposed waste products of ordinary metabolic
activity.

204 LILY H. HUIE.

Methods.

To deal with these questions satisfactorily necessitates that
no reliance should be placed on any single fixing method, and
therefore the whole of my previous work has been repeated
with chemically very diverse fixatives, e.g. 5 per cent. and
10 per cent. trichloracetic acid, 5 per cent. and 10 per cent.
phosphotungstic acid, osmic acid vapour, picric acid, picro-
corrosive sublimate with tannin added, absolute alcohol,
Hermann’s fluid, Flemming’s strong mixture, and Flemming’s
weak mixture. All these methods fully confirmed what I
have stated in my previous paper, so that in pursuing my
new research I had every confidence in employing again the
methods I formerly used, viz. Mann’s aqueous picro-corrosive
fixative and Mann’s eosin and toluidin blue stain.

The following substances were placed on vigorous fully ex-
panded leaves :—Paraffin, white of egg, Kiithne’s pure ampho-
peptone, Witte’s peptone (albumoses), ammonium sulphate,
fibrin, Griibler’s fibrin peptone, globulin, casein, nuclein and
nucleic acid, milk, calcium phosphate, glutin, leucin, creatin,
urea.

In all cases except that of casein the species of Drosera
used was D. rotundifolia, as plants reared in greenhouses
cannot be considered normal.

The leaves after being fed were fixed at intervals varying
from one minute up to the time of reopening of the leaves.
As, however, in the case of white-of-egg feeding the changes
produced after one minute are so remarkable,’ shorter periods
varying from 5 to 60 seconds were tried with white of egg and
also with pure peptone.

Unfed leaves were fixed at the same time to serve as controls,
and to avoid all errors caused by irregular staining sections
of control leaves were always mounted alongside those of fed
ones on the same slide.

It may be well to quote here again a definition of the terms
applied by me to the nuclear organs.

1 See previous paper.

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 205

Chromosomes are the organs which show the well-known
affinity for alkaline dyes. They are situated at the periphery
of the nucleus, and consist of nuclein.

Nuclear plasm is the material which, in We) resting state,
forms the main bulk of the nucleus. It is neutrophile in
character, and is frequently called the nuclear sap, but corre-
sponds to “ cedematin ”’ globules.

Nuclear sap is the more watery fluid inside the nucleus, in
which is suspended the nuclear plasm, the latter being pre-
cipitable by HgCl, and other reagents.

Nucleoli are the spherical organs which show a special
affinity for acid dyes, and have a more or less central position,
and they consist of Flemming’s paranuclein.

Resting Gland Cells (fig. 1).—The gland cells in their
resting state, when fixed and stained by Mann’s methods,
appear as follows:

The cell wall is pale blue; the cytoplasm is arranged in an
open foam-like manner; it stains pure blue with a coarse
granulation embedded in it, which appears of a deeper blue
and represents the zymogen.

The nucleus is situated just below the middle of the cell.
It may be spherical oroval. It stains purplish, and is sharply
defined against the cytoplasm, partly because of the difference
in tint, and also because of the peripherally placed, very
minute, dark blue chromosomes.

The nuclear plasm is granular, always very dense, and of a
purplish tint. One (sometimes more) very distinct, deep red
nucleolus is always present, surrounded by a narrow From-
mann’s clear zone.

Experiments.

Paraffin (fig. 2).—The rapid changes which can be pro-
duced by certain foods suggested the idea of ascertaining
whether purely physical factors, such as contact with insoluble
substances, would not also produce effects. When pieces of
well-boiled cork were applied to the leaves vacuolation of the

206 LILY H. HUIE.

gland cells resulted; but as this alteration might have been
due to traces of soluble substances it was thought necessary to
use paraffin of a high melting-point (58° C.), which had been
carefully cleaned by boiling. Shavings of it were laid on the
tentacles, and pressed into thorough contact with them by
means of a brush. About one hour after the application of
the paraffin the alteration in the cells reaches its height. It
consists in a vacuolation of the cytoplasm, but with no
alteration in colour reactions. The nucleus remains quite
unchanged. In the course of twenty-four hours complete
recuperation has taken place.

White of Egg [stains Red] (fig. 3)—My former research
having dealt minutely with the changes produced in the gland
cells by white of egg, I have now only to add those resulting
in the exceedingly short time of five seconds. They are as
follows :

Cell wall very pale blue.

Cell plasm deep purple, frequently red round the nucleus,
vacuolated in the upper third of the cell, densely aggregated
below.

Nuclear plasm unaltered in arrangement, but staining dark
purple or red like the cytoplasm surrounding the nucleus.

Chromosomes as in controls.

Nucleolus staining somewhat less brightly.

Peptones (figs. 4—9).—The experiments were made on a
large number of leaves, which were fed and fixed on their
native moor. It is important to note this, as plants which
were taken home did not show the characteristic precipitates
about to be described, but only vacuolation.

Histological Changes.—In all cases the peptone ap-
peared as greenish-blue droplets adhering to the exterior
surface of the gland ceils. In the cells of that layer which
in my former paper I called the third layer of gland cells,
the cell sap is particularly rich in tannin, and here is formed
a heavy greenish-blue precipitate encrusting the peripheral
walls. When viewed with a low power the dark blue line
thus formed gives a perfectly characteristic appearance to all

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 207

the sections of material used in this experiment. The changes
in the apical gland cells are as follows:

Five Seconds after the Application of a 10 per
cent. Solution (fig. 4).—A precipitate (?) resembling minute
blue droplets covers the interior of the cell walls, especially the
lateral ones.

The cell walls stain very faintly blue, or show no colour, .

The cell plasm stains much more deeply blue than in
controls. It appears more dense and homogeneous, being
apparently almost free from vacuoles. The reticular arrange-
ment seen in controls has given place to what appears like a
thick, evenly distributed granular mass. Thus instead of
vacuolation there is found a direct increase in bulk. At the
periphery of the plasmic body are seen dark blue droplets or
granules resembling those on the walls.

The nucleus is normal as to position, size, shape, and clear-
ness of definition.

The nuclear plasm is often also bluer than in controls,
though it frequently stains normally. Its arrangement is
unaffected.

The nuclear chromosomes are unchanged.

The nucleolus is normal as to size, and also as to colour in
those nuclei in which the plasm stains normally. In the
bluer nuclei it stains more of a purple. Frommann’s zone
remains clear.

The nuclei of the third layer remain unchanged.

Ten to Thirty Seconds.—The above description holds
good. I have, however, observed a slight shortening of the
nuclei of the third layer of gland cells in leaves fed for thirty
seconds.

Specimens fed for one minute require no separate detailed
description, as they resemble the bluer types about to be
described.

Five Minutes after Feeding (fig. 5).—The precipitates
appear as in earlier stages. The cell walls stain blue with
extreme faintness. The cytoplasm remains as shown in fig.
4, or is more vacuolated, and stains less deeply (fig. 5). The

VoL, 42, PART 2,—-NEW SERIES. 0

208 Lily H. Bune:

vacuoles appear to be filled with a substance which stains
pale blue. In a few tentacles there is a reddish tinge in the
apical half of the cells. The nucleus is normal as to position,
size, and shape. The nuclear plasm stains occasionally bluer
than in controls (fig. 4); or again in those cells which show
a tendency in the cell plasm to stain red, the nuclear plasm
takes on the same tint. There is, therefore, at this stage a
decrease in the basophile affinity which was so marked a
characteristic of the earlier stages. The nuclear plasm is not
diminished in quantity or altered in arrangement.

The chromosomes are not perceptibly increased in size, but
appear to be slightly more numerous. In some nuclei they
appear as if united by fine threads of a substance which also
stains dark blue. The nucleolus is normal.

The nuclei of the third layer are either normal or very
slightly shortened.

One Hour after Feeding (figs. 6—8). — The droplets
adhering to the exterior of the tentacle and those on the
lateral walls of the gland cells generally stain more faintly
and are less conspicuous than in earlier stages.

The cell walls stain extremely pale blue.

The cytoplasm is paler blue and less dense than.in earlier
stages, but is quite as full or fuller than in most controls, and
its mass is often outlined with dark drops or granules.

The nucleus is sometimes normal in size and shape, and
sometimes shrunken. The nuclear membrane has quite dis-
appeared. The nuclear plasm stains like that of controls, and
in those nuclei which have not shrunken it is arranged as in
controls, while in the shrunken nuclei its arrangement cannot
be made out.

The chromosomes have increased enormously. In those
nnclei which have not shrunken they are seen to be ribbon-
shaped. I have not found it possible to determine whether or
not they exist in any fixed number.

The nucleolus is smaller than normal. In the unshrunken
nuclei it is surrounded by a clear zone.

In the third layer the precipitate is as heavy as before,

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 209

The nuclei vary in shape from being only slightly shorter than
normal to forms such as are shown in figs. 32 and 33 of my
previous paper.

One Day after Feeding (fig. 9).—After the leaves had
been fed it was unfortunately necessary to remove the plants
from their natural habitat. They were placed in a cool
greenhouse, where the leaves that were allowed to remain
growing reopened healthily in two to four days, but it is
possible that the transplanting may have affected the cyto-
logical appearances about to be described. It is remarkable
that there was nothing to be seen resembling the precipitates
of the earlier stages either externally or internally.

The cell walls are very pale blue, almost colourless.

The cytoplasm is purple, very granular, greatly vacuolated
in the apical half of the cells, dense in the basal half, though
showing here also small vacuoles. In the neighbourhood of
the large vacuoles dark blue granular matter collects.

The nucleus is normal in size and shape, and is surrounded
by a membrane, which facts suggest that recuperation has
begun in the nucleus, though not as yet in the cytoplasm.

The nuclear plasm is of the same tint as the cytoplasm, and
is irregularly disposed, instead of being regularly distributed
as seen in controls and in the earliest stages of peptone
feeding.

The chromosomes are large and angular, drawn out into
points, which frequently gives them a stellate appearance. I
have not ascertained whether their number is constant.

The nucleolus is not very much smaller than normal.

The nuclei of the third layer are globular or irregular in
shape, pale pink, nearly destitute of nuclear plasm, with a
few conspicuous chromatin granules and a diminished nu-
cleolus. They resemble fig. 32 of my former paper.

Ammonium Sulphate.—As Kihne’s amphopeptone is
prepared with ammonium sulphate, it was thought advisable to
ascertain whether the precipitates formed in the cells by feed-
ing with amphopeptone might be due to traces of this salt.
Ammonium sulphate in 10 per cent., 100 per cent., and 200

210 LILY H. HUIE.

per cent. solutions in distilled water was therefore applied to
the leaves. No precipitates resulted, and vacuolation of the
cytoplasm was the only effect produced in the cells.

Some experiments were also tried with Witte’s peptone,
which is very rich in albumoses. The tentacles did not close
rapidly, but in from one to four hours they had all closed.
The bending caused was more rapid than that induced by the
fibrin peptone. In about three days the leaves reopened.
The material fed with Witte’s peptone was accidentally
spoiled and rendered useless for histological examination.

Before describing, however, the changes induced by fibrin-
peptone it is best, for comparison, to detail alterations caused
in the gland cells by Gribler’s fibrin.

Fibrin (stains Red),—Maximum change shown in fig. 11.
The fibrin was reduced to a powder by crushing, and was then
applied either dry or after being moistened with distilled water.
If all the tentacles were loaded they closed in about ten
minutes, otherwise only those that were touched closed, and
did so more slowly. The leaf itself did not double up as after
feeding with white of egg. Cytological changes :

Five Minutes after Feeding.—The only observable
changes are that the cytoplasm is vacuolated in the upper half
of the cell, and the nucleolus somewhat diminished.

After One Hour (fig. 10 @).—The cell wall is paler blue
than normal. The cytoplasm pale blue, much vacuolated ; fre-
quently the cell is almost emptied. The nucleoplasm is aggre-
gated. The chromatin is normal. The nucleolus has dimin-
ished to about one half its original diameter.

After One Day (fig. 10 6).—The cell wall is almost colour-
less. The cytoplasm is reduced to a scanty pale blue reticulum.
The nucleoplasm is normal as to colour, or redder, aggregated
and diminished in amount, showing empty spaces. The chro-
mosomes have enlarged. The nucleolus is always diminished,
sometimes reduced to a mere point.

After Two Days (maximum change, fig. 11).—The cell
walls are colourless or slightly pink. The cell plasm is blue,
extremely scanty. The nucleoplasm is thin, pale purple. The

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 211

chromosomes are enlarged. The nucleolus is small and dull
red,

After Two to Four Days.—Recuperation takes place as de-
scribed for leaves fed with white of egg, but much more slowly,
and is complete in four to seven days.

Griibler’s fibrin-peptone stains red and produces the maxi-
mum change in one hour after feeding, as shown in fig. 12.
The food appears as pink drops adhering to the outside of the
gland. The cell wall is nearly colourless, or faintly greyish
blue. The cytoplasm is very scanty. In different leaves it
varies in colour from pale blue to pinkish grey. The nuclei
are more or less shrunken ; the nuclear plasm reduced to a
little granular cloud which stains violet. The chromatin has
increased enormously, and forms large segments with rounded
ends. Whether their number is constant or not I have not
determined. The nucleolus is small and of a dull red tint.

Milk (stains Blue).—A small drop was placed on the
centre of the leaf or drawn over the marginal rows of tentacles.
In the first case the leaf itself folded up. In the second case the
tentacles closed regularly inwards in ten minutes or less, and
after an hour or two the entire leaf doubled over as in the first
case. In from one to three days the leaves reopened perfectly
uninjured unless the tentacles had been here and there glued
together by the milk drying. Milk may be absorbed twice or
thrice and the leaf open as clean as a control.

Five Minutes after Feeding (fig. 13).—The cell walls
are somewhat deeper blue than normal. The cytoplasm has
contracted into dense strands, appearing coarsely granular and
staining a much deeper blue than normal. The nuclear plasm
is somewhat redder than in controls, and the individual
granules have fused (aggregated) to form dark masses alter-
nating with clear spaces. The nuclear chromosomes are
normal. The nucleolus appears more of a purple.

One Hour after Feeding.—The cell wall is normal. The
cytoplasm disappears completely except at the periphery of the
cell and round the nucleus. It stains pale blue. The nuclei
are sometimes slightly shrunken. The nuclear plasm is red-

212 TiLY e. | BOLE

dish and becoming more scanty. The chromosomes are normal.
The nucleolus is diminished.

One to Two Days after Feeding (usual type, fig. 14).—
The cell walls are very pale blue. The cytoplasm is extremely
scanty, and stains very pale blue. The nuclear membrane is
indistinct. The nuclear plasm is reddish and scanty. The
chromosomes have slightly enlarged. The description just
given is typical for the state of “maximum change”’ usually
produced by milk in one to two days. It will be seen that the
nuclear changes, especially those affecting the chromatin, are
comparatively small. I have found this to be characteristic of
milk. Ina few cases, however, the nuclear changes have pre-
sented the appearance shown in fig. 15, which must be regarded
as exceptional. The nuclear membrane is scarcely traceable.
The nuclear chromosomes are much enlarged, angular, and
very clear. They closely resemble those produced by calcium
phosphate feeding (see fig. 16). The nuclear plasm has almost
entirely disappeared. The nucleolus is pale red and very
small. Recuperation proceeds in the usual way, and is very
complete.

Calcium Phosphate (fig. 16).—The salt did not adhere
to the tentacles after fixation, and so does not appear in stained
material. When treated with a mixture of the two stains
employed it seems to take up both, assuming a violet tint.
The salt was applied as a dry powder. The leaves closed fairly
rapidly. In two to three days they reopened, showing still
the white powder apparently unaffected.

Changes induced in Five Minutes.—The cell wall
stains a deeper blue than normal. ‘The cell plasm shows a
slightly increased vacuolation. The nucleus is contracted to
about two thirds of its normal size. The nuclear plasm stains
deep blue. The chromosomes and nucleolus are normal.

One Hour after Feeding.—The cell wall is blue. The
cell plasm shows increased vacuolation. The nucleus is con-
tracted. The chromosomes are normal. The nucleolus is
diminished somewhat.

One to Two Days after Feeding (maximum change,

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 2138

fig. 16).—The cell walls are pale blue. The cytoplasm is
exhausted and very pale blue. The nucleus is contracted
irregularly. The nuclear plasm is scanty—very pale violet or
blue. The chromosomes are greatly enlarged. The nucleolus
is diminished, sometimes excessively small, and stains dull red.

The maximum changes thus described closely resemble con-
ditions produced by milk (fig. 16).

In Two to Three Days.—Recuperation occurs. The
cytoplasm remains rather poor. The nucleus becomes quite
normal.

Globulin (stains Red) (maximum change, fig. 17).—The
leaves were dusted with the dry powder and closed rapidly,
i.e. seven to ten minutes. The globulin was entirely absorbed,
and the leaves reopened in about three days quite clean.
Some of them were fed a second time.

Cytological Changes: Five Minutes after Feeding.
—The cell walls stain very pale blue or purple. The cyto-
plasm is diminished in amount, and large vacuoles appear in it.
It stains blue or slightly purple. The nuclear plasm shows
slight aggregations. The chromosomes and nucleolus are
normal.

One Hour after Feeding.—The cell walls do not stain
at all. The cytoplasm is grey and scanty. The nucleus loses
its sharply defined contour, and its outline is only indicated
by the dark blue circle of the somewhat enlarged chromo-
somes. The nuclear plasm isaggregated, and has become more
eosinophilous. The nucleolus has diminished, and stains less
brightly.

One Day after Feeding (maximum change, fig. 17).—
The food in immediate contact with the apex of the tentacle
appears at this stage striated with colourless lines issuing in a
radiate manner from the apical cells, and suggesting that
secretory products pass out through the apical walls. At the
same time the food seems to be passing in between the lateral
walls of the cells, which seem to stand apart, apparently sepa-
rated by the red-stained globulin. The cell walls stain red.
The cytoplasm is pale blue, or grey and scanty. The nuclear

214 LILY H. HUIE.

plasm is reduced to a few purplish granules. The chromatin
is greatly increased, and is distributed in a few large angular
segments, which, however, do not equal in size those resulting
from feeding with white of egg. The nucleolus is small, and
stains dull red.

In Two to Four Days.—Recuperation takes place as
in leaves fed with white of egg.

Casein prepared by “Hammersten’s” Method
(stains Red).—The experiments, owing to the time of year,
had to be made on Drosera capensis kept in a greenhouse,
and I therefore do not regard them as satisfactory.

The powder was dusted on the tentacles, which closed slowly,
taking several hours. The maximum change observed re-
sulted in two or three days, when the cytoplasm had entirely
disappeared with the exception of a very narrow red _ peri-
pheral layer, and a scanty remnant round the nucleus. The
nucleus remained normal in size and shape, but the nuclear
plasm was red and aggregated. The chromosomes showed
slight enlargement, and the nucleolus appeared to be slightly
diminished.

Nuclein (stains Violet).—Nuclein was applied to the
leaves dry, and also wetted with distilled water. It caused no
movement of the tentacles and no increased secretion, i.e.
apparently no effect at all.

Five Minutes after Feeding.—The cell walls stain a
deep violet colour. The cell plasm is somewhat vacuolated.
The vacuoles appear blue. The nuclear plasm is unaltered in
arrangement, but is sometimes of a bluer tint than normal.
The nuclear chromosomes and nucleolus are unchanged.

One Hour after Feeding (maximum change, fig. 18).
—The cell walls stain deep violet. The cytoplasm stains very
deep blue, and forms a network of thick strands enclosing
large vacuoles, which appear blue. The nuclear plasm stains
the same deep blue tint as the cytoplasm. It appears to be
slightly aggregated. ‘The nuclear chromosomes are normal.
The nucleolus is somewhat smaller and more purple. No
further changes occur, and the cells become completely normal

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 215

in one to two days. Fig. 19 shows such a cell nearly quite re-
cuperated after twenty-four hours.

Nucleic Acid (stains Blue) (figs. 20—23).—The acid was
prepared absolutely pure by Dr. Gribler. The powder was
made into a paste, and a piece about the size of a pin’s head
placed on the centre of a leaf which folded up and secreted
violently for one to two days. At another time dry powder,
or minute particles of the paste, were dusted on the tentacles,
which then bent inwards and secreted copiously for one to two
days.

In both cases the leaves commenced to reopen in one to
three days, and were perfectly uninjured.

Five Minutes after Feeding (fig. 20).—The cell walls
stain a much deeper blue than in controls. The cytoplasm
shows very great vacuolation, and has contracted into thick
strands, forming a network which stains very deep blue, and
has a coarsely granular or knotted appearance. The nucleus
has shrunk considerably. The nuclear membrane and nuclear
chromosomes are unchanged. The nuclear plasm stains bluer
than normal, is somewhat aggregated and less dense. The
nucleolus is normal in size, but stains purple.

One Hour after Feeding (fig. 21).—The cell walls and
cytoplasm show no further change. The nucleus has become
irregular in shape, and usually has elongated transversely to
the long axis of the cell. The nuclear periphery is distinct, the
membrane staining dark blue and remaining distended. The
nuclear plasm has contracted away from the nuclear membrane,
and forms a small, compact, purple sphere round the nucleolus,
suspended in the nuclear cavity by fine radiating threads
attached to the nuclear membrane. A homogeneous fluid
which takes a pale blue stain appears to fill the space be-
tween nuclear plasm and membrane. The chromosomes are not
distinguishable. The nucleolus stains dark crimson. From-
mann’s clear zone is obliterated by the closely contracted
nuclear plasm.

One Day after Feeding (fig. 22, a, 6).—The cell walls
stain paler blue, The cytoplasm is very pale blue, and is

216 LILY H. HUIR.

reduced to a mere vestige. The nuclei are frequently enlarged,
and may be spherical, oblong, or distended irregularly. The
nuclear plasm is absent or represented by a faint cloudiness in
the homogeneous pinkish fluid (?) which fills the nuclear
cavity. The chromosomes appear as extremely minute, peri-
pherally placed, blue granules. In some very clear nuclei they
appear to be connected by fine threads which also stain blue.
The nucleolus is slightly smaller than normal.

One to Two Days (fig. 23).—The cell walls are pale blue.
The cytoplasm is almost completely exhausted, very pale blue.
The nucleus is small. The nuclear plasm stains as in controls,
but deeper. The nuclear membrane is ill-defined and the
chromosomes are normal. Nucleolus diminished and staining
dull red.

One to Three Days.—Recuperation takes place in a per-
fectly normal way.

Glutin (stains Red).—In the first experiments glutin
was crushed into a powder and moistened with distilled water.
Minute particles of this were placed on the tentacles. The
tentacles closed rapidly, in about five minutes. The whole
supply of glutin given to each leaf did not exceed an ordinary
pin’s head in bulk. In half an hour the particles had swollen
and become white and translucent like milk. Dry glutin,
even if powdered, was not conveyed by the tentacles to the
centre of the leaf with anything like the same rapidity ;
indeed, it appeared to be indigestible.

All the tentacles and parts of the leaf which had been
touched by the moistened glutin, as just described, died, i. e.
became brown and withered, though the leaf as a whole
reopened. Therefore the experiments were repeated with still
smaller quantities of glutin. It was moistened and softened
with distilled water, and placed with great care on the tentacle
heads. There resulted complete absorption, and when the
leaves reopened they were perfectly clean and healthy.

These experiments show what very different results may
follow dissimilar applications of the same stimulus.

Five Minutes after Feeding, — The cell walls are

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 217

normal. The cytoplasm is normal as to colour, much vacuo-
lated in upper third. The nuclear plasm is reddish purple,
aggregated. The chromosomes and nucleolus are unchanged.

One Hour after Feeding. — The cell walls are pale.
The cytoplasm stains blue, and is much vacuolated. The
nuclear plasm is reddish purple and aggregated. The chro-
mosomes are more prominent than in controls. The nucleolus
is duller red and smaller.

One Day.—The cell walls are colourless or slightly pink.
The cytoplasm is very pale blue, extremely scanty, merely
lining the cell wall and surrounding the nucleus. The nuclei
are often shrunken. The nuclear plasm is almost absent, and
stains faintly blue or violet. The chromosomes are more or
less enlarged. The nucleolus is dull red and small. Recu-
peration is normal.

In the tentacles which died the cytological appearance was
the same as that caused by death from feeding with urea (figs.
26 and 27).

Creatin (fig. 24).—Creatin was employed dry and also
wetted with distilled water. The latter proved the better
stimulant. The tentacles bent inwards slowly. Secretion
was not profuse. Some leaves seemed quite unaffected, and
their tentacles did not bend. Those that closed reopened in
two days, and the white particles of creatin were adhering
to their tentacles apparently unaffected. The examination
showed conditions as follows :

One Hour after Feeding.—The cell wall is normal ; the
cytoplasm slightly pink, and greatly vacuolated in the upper
half of the cell ; the nucleus unaffected.

In the course of the one or two days in which the leaf
remains closed the further cytological changes are not great.

The nuclear plasm diminishes in quantity. ‘The chromo-
somes in some cases increase slightly. The nucleolus dimin-
ishes to some extent (fig. 24). Recuperation is complete in
two to three days.

Leucin (fig. 25).—Did not adhere to fixed tentacles, and
does not appear in stained sections. Leucin was applied

218 LILY H. HUIE.

both dry and wetted with distilled water; the former proved
most active. In about 15 per cent. of the leaves used some of
the tentacles bent slowly, taking about twenty-four hours to
close ; but as a rule the tentacles did not move at all, though
their secretory activity appeared to be stimulated. Some of
the leucin was dissolved and hung from the leaves in large
drops, which finally fell off. In some cases it was not dis-
solved, and fell off or was blown away. In any case the ten-
tacles were to all appearance left quite uninjured.

Leaves were fixed at the following intervals after feeding :—
Five minutes, one hour, one day, two days, three days.

The cytological changes seem to be produced very tardily.
I find none during the first two periods. After one day there
is considerable vacuolation of cell plasm and nuclear plasm,
and slight diminution of nucleolus. In tentacles that had
closed there is slight but quite distinct increase of chromatin
three days after feeding (fig. 25). This has not been seen to
occur in tentacles which remained unbent.

Urea did not adhere to fixed material, and so does not
appear in the stained preparations.

Crystals were powdered, and dusted on the tentacles. The
urea readily melted in the secretion, and seemed to increase
the exudation, but caused no bending of the tentacles. At the
end of half an hour the leaves looked as if they had been
under a shower of rain. In twenty-four hours the leaves
appeared yellow, flaccid, with weak and shrivelled tentacles.
Death of the leaf invariably ensued.

One Hour after Feeding (fig. 26).—The cell walls are
blue. The cytoplasm is still blue and granular in the lower
half of the cells, but is vacuolated in the-upper half. The
nuclear membrane remains irregularly distended, but the
nuclear contents have contracted to form a small dark
purple body in the centre of the nuclear cavity, in which
neither nuclear plasm, chromosomes, nor nucleolus can be
distinguished.

One to Two Days after Feeding (fig. 27).—The
withered tentacles show the cytoplasm reduced to a scanty

CYTOLOGICAL CHANGES PRODUCED IN DROSBRA. 219

reticulum of threads, staining blue, purple, or red. The
nucleus has almost lost its identity, having amalgamated with
its surrounding cytoplasm to form a more or less crescentic
body, which stains reddish purple, and exhibits no differentia-
tion whatever.

Summary.

1. By feeding with chemically different foods very cha-
racteristic alterations are obtained, both in the colour re-
action and morphology of the cell. For example, in five
seconds white of egg causes both the cytoplasm and nuclear
plasm to become more eosinophile, while pure amphopeptone
increases their affinity for blue stains. The former food quickly
causes great impoverishment of cytoplasm and nuclear plasm,
while the first effects of pure peptone are to increase their
bulk and density. Both the foods produce an enormous in-
crease of the chromatin element of the nucleus, while other
foods, e.g. nucleic and nucleic acid, produce no such result.

2. While the cytoplasm is the cell constituent most rapidly
and most constantly affected by external stimuli, the nucleus
is the seat of metabolic activity, and the state of the nuclear
organs indicates whether or not the food supply was of service
to the metabolism of the plant.

It is not surprising, therefore, to fiud a certain independence
of cytoplasm and nucleus in their behaviour to external
stimuli, coupled with great interdependence with respect to
all resulting processes of metabolism. Thus, substances such
as paraffin and nucleic acid act merely as stimuli to the secre-
tive activity of the cells; and they cannot be regarded as
foods, for they affect the nucleus only by causing a slight
drainage on the nuclear plasm and nucleolus. If the stimulus
is very transient the nucleus remains unaffected. On the
other hand, with highly nutritious food like egg-albumin and
peptone the nucleus is the seat of the greatest change, and
the chromosomes, i.e. the nuclein, undergo an enormous

220 LILY H: UIE:

increase, independently of the state of the cytoplasm. With
egg-albumin we do not get these great changes in the nuclear
organs till the cell plasm is thoroughly exhausted, while in the
case of 10 per cent. amphopeptone we get them while the cell
cavity is full of cytoplasm.

In all cases the process of recuperation begins in the nucleus.
The nuclear plasm first becomes abundant, and restoration of
the cytoplasm begins in contact with the nucleus, and spreads
thence to the remoter parts of the cell.

3. To answer the question, whether the rapidity with which
changes occur depends on the ease with which substances are
absorbed, we must deal only with such changes as are charac-
teristic of foods. Changes that are produced also by mere
contact with insoluble substances must not be taken account of.

Changes which are produced in five seconds are rapid, yet in
this short space of time we get quite specific alterations for
two foods so different in their diffusibility as egg-albumin and
peptone. But then only the plasm is affected, that is that cell
constituent which I have just shown to be most rapidly altered
by external stimuli. Nuclear changes, e. g. the increase of
chromatin, must depend for their rapidity on the constitution
of the food, 1. e. on the series of chemical changes the latter
must undergo before it is converted into basophile chromatin.
In the two cases just cited the length of time required to
produce this change differs widely, being about twenty hours
in the case of egg-albumin, while one hour suffices for peptone ;
and yet from the changes wrought in the cytoplasm in five
seconds we might argue that both foods entered the cells with
equal rapidity. We may conclude, then, that the rate of
plasmic changes depends on the rate of absorption, but that
the rapidity of nuclear changes is commensurate only with the
digestibility of the food.

4. In determining the relation between cytological changes
in the gland cells on the one hand, and the degree of irritability
in the leaves calculated from their rate of closure on the other
hand, cytoplasmic changes only constitute our legitimate
criterion; for I have shown above that the cytoplasm is the

CYTOLOGICAL CHANGES PRODUCED IN DROSERA. 221

cell constituent directly influenced by external stimuli, and
that nuclear changes are the secondary results of metabolism.
A careful comparison of the action of the substances applied
to the tentacles proves that there is a constant concord between
the rate of closure of the tentacles and the degree of vacuola-
tion produced in the cytoplasm. The citation of a few cases
will illustrate this :

Paraffin and nuclein caused no closure, and very slight and
transient vacuolation.

Pure peptone caused very slow inbending, and no vacuolation
for one or two hours.

White of egg and milk caused rapid bending and rapid
vacuolation.

Fibrin caused slower closure and slower vacuolation.

Creatin did not induce bending of all of the tentacles to
which it was applied ; and those that closed did so with extreme
slowness. The vacuolation was only transient.

Leucin sometimes, but not invariably, occasioned bending,
and this was extremely slow. Vacuolation of short duration
ensued.

Urea caused no bending, but killed the leaves. Being
therefore a poison, its action should not be compared with that
of the other stimulants named.

5. By bringing healthy cells into contact with the waste
products, creatin, leucin, and urea, urea was found to act as a
poison, creatin as a mild stimulus to movement, and leucin
caused active secretion without much movement. These two
substances, therefore, do not injure the leaves, and they seem to
be of some nutritive value.

At a future time I may ascertain the effects of feeding with
various carbohydrates, should the results obtained from
preliminary experiments give promise that such an investiga-
tion would be of any value.

In conclusion I desire to acknowledge the kindness of
Professor Gotch in allowing me to work in the Physiological
Laboratory at Oxford, and to express my indebtedness to
Dr. Gustav Mann for his constant advice and help.

222, LILY “H: “Hoth

DESCRIPTION OF PLATE 22,

Illustrating Lily H. Huie’s paper on “ Further Study of
Cytological Changes produced in Drosera.”

All figures were drawn with Zeiss’s camera lucida, Zeiss’s 3, apochromatic

immersion objective, and No. 8 compensating ocular. The magnification
equals 2000 diameters. All cells represent apical gland cells.

Fic. 1.—The resting condition (control). Vide p. 205.

Fic. 2.—After paraffin shavings, one hour. Vide p. 205.

Fic. 3.—White of egg, five seconds. Vide p. 206.

Fic, 4.—10 per cent. amphopeptone, five seconds. Vide p. 207.
Fic. 5.—10 per cent. amphopeptone, five minutes. Vide p. 207.
Figs. 6, 7, and §.—10 per cent. amphopeptone, one hour. Vide p. 208.
Fic. 9.—10 per cent. amphopeptone, two days. Vide p. 209.
Fic. 10 a.—Fibrin, one hour. 10 4. Fibrin, oneday. Vide p. 210.
Fie. 11.—Fibrin, two days. Vide p. 210.

Fie. 12.—Fibrin-peptone, one hour. Vide p. 211.

Fig. 13.—Milk, five minutes. Vide p. 211.

Fig. 14.—Milk, one to two days, usual type. Vide p. 212.

Fie. 15.—Milk, one to two days, occasional type. Vide p. 212.
Fig. 16.—Calcium phosphate, one day. Vide p. 212.

Fic. 17.—Globulin, one day. Vide p. 218.

Fie, 18.—Nuclein, one hour. Vide p. 214.

Vie. 19.—Nuclein, recuperating one day. Vide p. 215.

Fic. 20.—Nucleic acid, five minutes. Vide p. 215.

Fie. 21.—Nucleic acid, one hour. Vide p. 215.

Fies. 22 and 23.—Nucleic acid, one day. Vide p. 215.

Fig, 24.—Creatin, one to two days. Vide p. 217.

Fic. 25.—Leucin, three days. Vide p. 217.

Fie, 26.—Urea, one hour. Vide p. 218.

Fic. 27.—Urea, two days. Vide p. 218.

RECENT WORK ON THE PROTOCHORDA. 228

Remarks on some Recent Work on the Proto-
chorda, with a Condensed Account of some
Fresh Observations on the Enteropneusta.

By
Arthur Wiliey, M.A., D.Sc.

TueE eighth volume of the well-known ‘ Traité de Zoologie
Concréte,’ by Professor Yves Delage and M. Edgard Hérouard,
published in 1898, is devoted to what the authors style the
Procordés or Prochordata, a group which is made to include
three classes, namely, Hemichordia, Cephalochordia, and Uro-
chordia. It is a pleasure to turn to a text-book in which the
Enteropneusta are treated on equal terms with the Cephalo-
chorda and the Urochorda, especially when, as in this case,
the distinguished authors have aimed at impartiality—a
quality which in one or two places has led them beyond the
bounds of discrimination. It goes without saying that the
work is an excellent one, and admirably calculated, on the
whole, to give a just idea of these animals; but on the present
occasion I am neither concerned with its many excellences
nor with its few blemishes, but merely with the subject-
matter.

No doubt the classification employed by the authors of this
text-book answers not only their purpose, but that of their
readers also. At the same time it should be borne in mind
that it is in no sense a final classification, nor even one which
corresponds to the present state of our knowledge. By not
including the Pterobranchia in the same volume with the

VOL. 42, PARY 2,—NiuW SELIES. P

224, ARTHUR WILLEY.

Enteropneusta, the authors have neglected the opportunity of
pointing an interesting and instructive analogy.

What the Urochorda are to the Cephalochorda, such are the
Pterobranchia to the Enteropneusta.

The perpetual domination of the notochord in classification
constitutes a noteworthy example of the manner in which
zoological knowledge moves along well-worn grooves. There
is strong reason to suppose that the gill-clefts have the priority
of the notochord, or at least equal antiquity with it; and if
this supposition should prove to be correct in principle, there
ought to be some indication of it in the classificatory system.

I have treated this subject in some detail in a memoir on the
Enteropneusta collected by me in the South Pacific, which will
shortly be published ;' and the following is a simplified form
of the table of classification there constructed.

PHYLUM BRANCHIOTREMA, n. n.

I. HEMICHORDA, Bateson, 1884.

Class 1. PreroprancHiA, Lankester, 1885.
Class 2. Enruropneusta, Gegenbaur, 1870.

II. PROTOCHORDA, Balfour, 1882.

Class 1. Urocnorna, Lankester, 1877.
Class 2. CepHatocuorpba, Lankester, 1877.

Ill. VERTEBRATA,? Lamarck and Cuvier.

Class 1. AcraniA, Haeckel, 1866.
Class 2. Craniora, Haeckel, 1866.

In the above system the group containing Amphioxus
appears under two different names, Cephalochorda and
Acrania. I see no objection to this procedure, nor any other
way out of the difficulty.

1 In Part iii of A. Willey’s ‘Zoological Results’ (Cambridge University
Press).
2 Vertebrata = Holochorda, Gadow, 1898.

RECENT WORK ON THE PROTOCHORDA. DAS,

The justification for the new collective name, Branchiotrema,
is contained in my forthcoming memoir, where it is introduced
in connection with a new theory of gill-clefts, to which I have
been led by my study of the Enteropneusta. This theory may
be barely and briefly stated as follows:

The gonads and gill-slits were primarily unlimited in
number and co-extensive in distribution, the gonads having a
zonary disposition and the gill-slits occupying the interzonal
depressions.

The primary function of the gill-slits was the oxygenation
of the gonads, their secondary function being the respiration
of the individual—the change of function having taken place
pari passu with an elaboration of the vascular system.

Correlatively with the progressive regional differentiation of
the body, the gonads and gill-slits became limited both ante-
riorly and posteriorly. The anterior limitation of the gill-
slits behind the collar region is constant in all Enteropneusta,
but the posterior limitation is excessively variable.! The
emancipation of the gonads from their topographical relations
with, and functional dependence on, the gill-slits has taken
place in several ways, but the resultant tends to be, and
eventually actually is, the restriction of the gonads to a post-
branchial genital region.

Such, in outline, is the theory to which I have committed
myself. The Harmer-Brooks-Masterman theory of gill-clefts
does not, in my opinion, account for their prime origin, but it
does perhaps explain the retention of a single pair of gill-
clefts in forms like Cephalodiscus and Appendicularia.

In recording the fact that Spengel divided the species of
Enteropneusta into four genera, MM. Delage and Hérouard
prefer to describe them “
nation commune.” This conservative method of treatment is
somewhat foreign to the general spirit of the book, since an
endeavour has been made to incorporate the results contained
in the most recent publications. As a consequence, some

en bloc sous leur ancienne dénomi-

' Both anterior and posterior limits of the gonads are variable.

226 ARTHUR WILLEY.

structures of first importance are relegated to foot-notes, e. g.
the genital pleure (p. 5) and roots (p. 45) of Ptychoderide.

In a paper published in this Journal,! which our authors
have overlooked, I have given reasons for supposing that a
form like Ptychodera flava, in which the gill-slits open
freely to the exterior and not first into gill-pouches, represents
the most primitive existing type of Enteropneusta. Speaking
from personal experience, I may at least say that this species
has opened my eyes as to the significance of the enteropneustic
organisation.

The figure of Ptychodera clavigera, given on pl. i,
facing page 64, is apt to be misleading. The pharynx in this
figure appears to stand boldly forth as a cylindrical tube at the
base of the open chamber formed by the arching genital
pleure; and the parallel arcuate lines have the appearance
which 1s actually presented by the true gill-bars in Ptycho-
dera flava. The gill-bars of Pt. clavigera are, however,
quite invisible externally, the genital pleure have a dorsal
origin, the pharynx does not project beyond the level of the
floor of the peribranchial space,? and the gill-pores are
extremely small, lying at the base of the narrow branchial
grooves.

The authors have very naturally followed Spengel in their
explanation of the lateral septa of the Ptychoderide, as being
the outer walls of a pair of coelomic diverticula (p. 23, foot-
note).

The prolongations of the truncal coelom into the collar
region (viz. perihemal and peripharyngeal cavities) are intelli-
gible facts ; but how the truncal coelom could project a portion
of itself into itself was a mystery to my mind until I realised
that there is no question of a diverticulum at all.

1 A. Willey, on Ptychodera flava, Eschscholtz, ‘Quart. Journ. Mier.
Sci.,’ vol. 40, 1897, p. 165.

2 Excepting that the branchial tract or gill-area, i.e. the area enclosed
within the branchial grooves is somewhat arched. ‘The blue-lined structure
in the figure above referred to is simply the gill-area or Kiemenfeld of
Spengel.

RECENT WORK ON THE PROTOOGHORDA. 227

The fact that Ptychodera flava clears away this difficulty
is alone sufficient to entitle it to be regarded with particular
respect. This remark no doubt applies to the sub-genus
Chlamydothorax to which Pt. flava belongs.

In most Ptychoderide the lateral septa, as described by
Spengel, have a limited anterior extension; the point of their
proximal or mesial origin from the basement membrane gra-
dually approaches that of their distal insertion into the same
membrane until the two points coincide; and so the lateral
septum on each side comes to an end in the posterior branchial
region. In this way there actually exists a portion of the celom,
bounded mesially by the dorsal mesentery! and laterally by
the lateral septum, which ends blindly in front ; and as long as
this was all that was known on the subject there was perhaps
no other alternative than to propound some such formal ex-
planation as that put forward by Spengel.

In Ptychodera flava the lateral septa do not come to an
endin the posterior branchial region, but they are co-extensive,
in front and behind, with the genital pleure. It can therefore
hardly admit of question that the genital pleure and lateral
septa are causally related to one another.

Where the genital pleure are at their maximum the lateral
septa are entire. As the genital pleuree have become reduced,
the reduction always taking place from before backwards, the
the lateral septa have been subjected to the same process of
limitation, and exhibit the effects of it in a more marked
manner.

MM. Delage and Hérouard do not devote much space to the
difficult subject of excretion in the Enteropneusta, being con-
tent to state that the essential organ of excretion is the
glomerulus which forms part of the central complex of the
proboscis, while the excretory products are said to be got rid
of through the proboscis-pore.

1 This holds good only for the post-branchial region. In front of the last
gill-cleft on each side, the proximal origin of the lateral septum is transferred
from the wall of the gut to the basement-membrane of the epidermis.

228 ARTHUR WILLEY.

Bateson! found granules in nearly all the mesoblastic tissues,
and he says (p. 526) ‘they may perhaps be excretory, and it
is possible that they are more or less removed by the proboscis-
pore and collar-funnels respectively. This does not explain
their presence in large masses in the trunk body-cavity, from
which no pore has been observed to open.”

In the first place it should be remembered that it is not ab-
solutely necessary that excretory products must be removed
from the body; this is shown in the Ascidians, where there
are no excretory ducts. The so-called pericardium (Herzblase)
of the Enteropneusta which lies in the centre of the glomerulus,
i.e. between the two halves of the latter, appears, from the
curious way in which its endothelium proliferates into the
cavity (to such an extent as sometimes to completely block up
the cavity), to stand in functional as well as in topographical
relation to the glomerulus. If this is so, it would, in its
capacity as a closed sac associated with the renal function, be
physiologically comparable to the organ of Bojanus of the
Molgulide.

The glomerulus of the Enteropneusta is, so far as our present
knowledge goes,a structure sui generis, and itis quite clear that
itis the principal organ of excretion only in virtue of its having
superseded something else, namely the paired excretory canals.

The proboscis-pores are highly variable ; the collar-pores are
constaut; but neither the former nor the latter are any longer
mere excretory pores. The collar-pores especially seem to
promote locomotion by taking in water, and so causing the
collar to swell (Spengel) ; this may happen also in the case of
the proboscis-pores sometimes, but not always.

I have observed what I believe to be the vestiges of a pair
of truncal canals and pores in two species of the genus Spen-
gelia. In both Sp. porosa and Sp. alba, n. sp., there is a
pair of canalicular extensions of the first pair of gill-pouches
into the posterior end of the perihzemal cavities close to the
level at which the latter pass into the truncal celom. They
occur approximately at the same level as the collar-canals,

1 «Quart. Journ. Mier, Sci.,’ vol. 26, 1886.

RECENT WORK ON ‘''HE PROTOCHORDA. 229

which likewise arise as canalicular extensions of the first gill-
pouch into the collar ceelom on each side.’ These truncal
canals of Spengelia are such definite structures that I was for
a long time perplexed as to their significance.

I have referred above to the fact that the collar-canals are
actively functional; their walls consist of richly ciliated co-
lumnar epithelium, and they retain a uniform calibre from
their external orifice to the wide semilunar funnel by which
they open into the collar celom. The truncal canals, on the
contrary, taper towards their internal ends, their walls contain
ill-defined mucous cells, and, in short, they distinctly appear
to be in a vestigial condition. The perihemal prolongations
of the truncal cceelom usually contain merely virtual cavities ;
in other words, their cavities are quite blocked up with mus-
cular and connective tissue. This is the case in Sp. porosa,
whereas in Sp. alba a true space appears in the posterior
portion of the perihemal cavities, namely, in the region in
which the truncal canals occur.

I can neither state positively that there is an internal open-
ing nor that there is not; one thing only is certain, namely,
that the truncal canals are there. In Sp. porosa they are
longer than in Sp. alba, but they present more the appearance
of vestigial structures in a chronic state of mucoid degeneration
in the former species than in the latter.

A minute terminal pore is always difficult to find in trans<
verse section, or even in any kind of section, and it will be
remembered that there was the same difficulty in the case of
the atrio-ceelomic or brown funnels described by Professor
Lankester in Amphioxus.

If the truncal canals of Spengelia? and the brown funnels

1 This is Spengel’s view. Morgan says the collar-pore and first gill-slit
arise coincidently. I do not think this affects the present question. Bate-
son describes the collar-pores in B. kowalevskii as arising as thickenings of
the outer “atrial” wall which become perforated. The so-called atrial
cavities, formed by the overhanging lateral margins of the collar, are peculiar
to B. kowalevskii (Spengel).

2 It need be no cause for surprise that these structures only occur in one
genus of Enteropneusta, Lach species of Enteropneusta may and

230 ARTHUR WILLEY.

of Amphioxus be regarded as vestigial structures, the impor-
tance of their possession of an internal opening is diminished.

Sta
POR DAAE AN

it

Fie. 1—Diagram of anterior end of an Enteropneust, to show the regional
canals and pores. The proboscis-pores are indicated as they sometimes
occur in Ptychodera flava; the collar and truncal pores as in
Spengelia. c.e. Anterior and posterior margins of collar. c.f. Collar
funnels, d.c. Dorsal canals of proboscis celom. e. v. End vesicles of
proboscis (Hichelpforten). g. p’. First gill-pores (only the most dorsal
portion of the first gill-pouch is indicated). p.c. Proboscis colom.
ph. Perihemaf cavities. p. p. Proboscis-pores. t.¢. Truncal canals
(opening with the collar canals into the first gill-pouch).

In Ptychodera flava there are always two proboscis-
pores, one of which (either the right or the left) is smaller
than the other, and its terminal vesicle is usually not in

usually does present peculiar vestiges of structures which were presumably
associated together in the ancestral forms,

RECENT WORK ON THE PROTOCHORDA. 281

communication with the proboscis coeelom (see Fig. 1). Thus
the internal opening of the proboscis canal (end vesicle or
Kichelpforte) is lost, while the external opening remains ; and
it is probably a general rule that when these regional
pores change their function, or lose their function
and become vestiges, one of the first things to
happen is likely to be the closure of the internal
or celomic opening.}

In accordance with the above considerations, I regard the
truncal canals of Spengelia and the atrio-coelomic funnels
of Amphioxus as the vestiges of a pair of functional truncal
pores, which were homodynamous with collar-pores and
proboscis-pores. It is therefore of great interest to point
out that in Amphioxus there are also traces of the other
regional pores.

1 As I have referred to this loss of the internal opening of the end vesicle
of the proboscis canal, I will briefly state here what I believe to be some
of the potentialities of this structure.

i. The proboscis-pore is frequently well in front of the anterior neuropore.

ii. Sometimes it is closely associated with the neuropore.

ili, Sometimes it opens into the medullary tube of the collar behind the
neuropore.

iv. Frequently the end vesicle is prolonged behind the pore, as a cecal
follicle lying below the medullary tube.

v. Combining what sometimes happens into one phenomenon, we see the
neuropore leading into the medullary tube, anda subneural organ opening into
the latter.

vi. The entire medullary tube of the collar of Enteropneusta corresponds
to the cerebral vesicle only of Amphioxus and of the Ascidian larva.
‘he spinal cord is represented in Enteropneusta by the dorsal nerve which
lies in the skin, and is not closed in.

vii. The subneural gland of the Ascidian larva opens by the neuropore into
the dorsally placed mouth, and at the other end into the cerebral vesicle.

vill. The inner or cerebral opening of the subneural gland of the Ascidian
larva is thus seen to correspond to the proboscis-pore, which has lost all
relation to the ceelom.

ix. Hence the peculiar mode of development of the tunicate subneural
gland is explained, and the apparent absence of a proboscis-pore in the tunicate
larva is accounted for.

x. The roots of Ptychoderide are related to the epiphysial complex of the
thalamencephalon of Craniota.

282 ARTHUR WILLEY.

The structure in the larva of Amphioxus known as
Hatschek’s nephridium,! which opens at one end into the
buccal cavity, has been shown by MacBride? to be, at an
early stage, in open primary communication at its other end
with the left archenteric pouch, which he has suggestively
named the left collar-cavity. In spite of differences in the
method of development, I regard Hatschek’s nephridium as
being in principle the vestige of a pair of collar canals.

Bateson tentatively compared the collar-pores of the
Enteropneusta both to Hatschek’s nephridium and to Lan-
kester’s brown funnels. The comparison of the enteropneustic
proboscis-pore with the orifice of the amphioxine preoral pit
is of old standing, and likewise originated with Bateson,
who further compared them both to the craniate pituitary
body, without carrying the comparison into any great detail.

In the Enteropneusta the excretory function of the regional
pores has been superseded by the specialisation of the
glomerulus; in Amphioxus by the evolution of the nephric
tubules which were discovered by Weiss and Boveri.

It may indeed be said that in the Enteropneusta the
primordia of the nephric tubules are present in the form al a
minute diverticulum at the dorsal medial angle of each gill-
pouch, or in a corresponding position in those cases where the
gill-pouches are confluent, asin Pt. flava. These structures are
particularly well seen in sections through Spengelia alba.
Whether this be so or not there is undoubtedly a special
significance in the remarkable fact that Boveri’s tubules are
precisely co-extensive with the gill-clefts, and a renewed
importance should be attached to the connecting vessels
observed by Paul Mayer between dorsal aorta and sub-
intestinal vein in embryos of Pristiurus, which were shown
by Rickert to occur in the same segments with the pronephric
tubules and to furnish the latter with rudimentary glomeruli.

‘ If I understand them aright, MM. Delage and Hérouard have completely
misunderstood this structure (p. 121, foot-note).

? K. W. MacBride, “The Early Development of Amphioxus,” ‘Quart.
Journ. Mier. Sci.,’ vol. 40, 1897, p. 589,

RECENT WORK ON THE PROTOCHORDA. 238

If Boveri’s tubules open the way to a perception of the subse-
quent potentialities of the excretory system, Lankester’s brown
funnels, Hatschek’s nephridium, and the preoral pit furnish a
clue to its past history.

Fic. 2, A and B.—Portions of transverse sections through the caudal region
of Ptychodera ruficollis, n. sp. A. Through the anterior caudal
region. B. Through the mid-caudal region. em. Circular muscles of
body-wall. hg. Wall of hind gut. Im. Jongitudinal muscles. py.
Pygochord. vn. Ventral nerve-cord. vv. Ventral blood-vessel.

The substitution of nephric tubules in the truncal region
for regional pores in the archimeric (Masterman) regions,
which is displayed before our eyes in Amphioxus, is one of
the most striking examples of the working of the principle of
substitution that I can call to mind,

234, ARTHUR WILLEY.

MM. Delage and Heérouard retain the designation noto-
chord applied by Bateson to the diverticulum from the throat
which projects into the proboscis, where it acquires a rigid
consistency and sustaining properties. I prefer to call this struc-
ture by a non-committal name, and propose the term stomo-
chord. The stomochord is not the only skeletal product of
the gut wall inthe Enteropneusta. There is another structure
which finds no mention in the text-book under consideration,
but which is hardly second in interest to the stomochord itself.
It occurs along the entire length of the hind gut in the caudal
region on the ventral side in many Ptychoderide ; it is a solid
keel-like, ribbon-shaped band, with dilated distal border abut-
ting upon the ventral blood-vessel, and united at its dorsal
edge with the median ventral epithelium of the gut.

This structure, which I propose to call the pygochord,
was first seen by Spengel in Pt. minuta, then by Hill in
Pt. hedleyi, and I have found it in Pt. flava, Pt. carnosa,
n.sp., and Pt. ruficollis,n. sp.

In Spengelia alba, n. sp., it appears to be represented by
a vacuolar thickening of the ventral epithelium, which, how-
ever, retains its epithelial position, and is not drawn out into a
band.

I do not think that the enteropneustic stomochord corre-
sponds to any definite part of the true notochord. ‘The pre-
oral extension of the notochord, far beyond the anterior limit
of the neural tube in Amphioxus, is due to a forward growth
of the notochord as such ; whereas the preoral position of the
stomochord in the Enteropneusta is due to a forward projection
of a portion of the collar-gut or throat. Spengel calls it the
Eicheldarm, but although he intended this name to be in-
different, it is capable of misleading interpretation, since it
doves not belong to the proboscis at all in its primary quality of
integral constituent of the gut, but only in its secondary
quality of a skeletally metamorphosed derivative of the gut.
Moreover, whereas the notochord is essentially a uniform,
single, indivisible structure,' the stomochord exhibits strongly

1 Tue anterior tip of the notochord presents certain properties which may

RECENT WORK ON THE PROTOCHORDA. 235

marked regional differentiation (Fig. 3). It is therefore not
sufficient to say that any structure in other forms is compar-
able to the enteropneustic stomochord, but it must be specified
which portion of this structure is referred to.

Fic. 3.—Diagram of a complete stomochord of an Enteropneust from the
dorsal and lateral aspects, to show its regional differentiation. v.p.
Vermiform process. a.s. Auterior region of body of stomochord. 1.p.
Lateral pouch. v.c. Ventral cecum. n. Nuchal region. b.o. Buccal
orifice of stomochord. c.g. Throat or collar-gut.

The lumen of the stomochord is highly variable ; it may
safely be said that it is not the same in any two individuals of
a species. Sometimes it is in a fragmentary condition, some-

have an atavistic significance, but this does not affect my general proposition.
The most that could be looked for in the embryos of higher Chordata would
be a vestige of the buccal orifice of the stomochord, and perhaps this does
occur.

236 ARTHUR WILLEY.

times itis locally obsolete, and sometimes it is unusually capa-
cious. Jn all cases it is patently vestigial. In more than
one of my species (e.g. Pt. carnosa, n. sp.) the stomochord
undergoes fragmentation in the nuchal region, owing to the
invasion of skeletal substance. In Balanoglossus cana-
densis Spengel found that the entire nuchal region of the
stomochord was lacking.

The structure of which the stomochord, in its capacity as a
portion of the gut, is a vestige, must originally have been
post-oral,! and I have convinced myself that the structures
recently described by Masterman? in the Actinotrocha of
the Bay of St. Andrews are capable of being explained on this
basis.

In its middle reyion the stomochord is greatly dilated, both
dorso-ventrally and laterally. Spengel only emphasises the
ventral cecum of the stomochord in this region; but often
there is a pair of lateral pouches which are particularly well-
marked. In some cases the lateral pouches are less pronounced
than the ventral cecum, and sometimes the exact reverse is
the truth. Sometimes the lateral pouches do and sometimes
they do not unite with one another across the middle line by
the intermediation of the ventral cecum,

The interest of the situation has been increased rather than
diminished by the recently published observations of Roule on
the Actinotrocha of Phoronis sabatieri. Instead of the
paired lateral diverticula or pleurochords described by Master-
man, there is in Roule’s larva asingle median anterior ventral
diverticulum, whose cells likewise undergo vacuolar degenera-
tion, which gives it a semi-rigid consistency. According to
Roule this ventral diverticulum arises at the anterior end of
the intestine (in the same region as Masterman’s pleurochords),
and projects forwards ventrally below the cesophagus.

' T think it is universally admitted that the stomochord is a secondary
projection into the proboscis cavity pushing the cclomic epithelium or
splanchnotheca of Spengel before it.

* A. 'T. Masterman, ‘‘On the Diplochorda,” ‘Quart. Journ. Mier. Sci.,
vol. 40, 1897, p. 28h.

RECENT WORK ON THE PROTOCHORDA. 237

Thus in Masterman’s pleurochords we have, as I believe, the
representatives of the lateral pouches of the enteropneustic
stomochord ; while in Roule’s ventral diverticulum we have
the representative of the ventral caecum of the stomochord.!

With regard to the structure in Actinotrocha which
Masterman homologises with the so-called notochord described
by Harmer in Cephalodiscus and by Fowler in Rhabdo-
pleura, but which he (Masterman) labels ‘‘subneural gland,”
it is not easy to suggest a final explanation. Harmer? has
pointed out its resemblance to the vermiform process of the
stomochord which occurs in the Spengelide, and it may no
doubt be compared with this structure.

What is doubtful is its primordial significance. As a first
step in the discussion, the name arbitrarily given to it by
Masterman should be dropped, because it involves a simple
begging of the question. ;

The following proposition may therefore be stated cate-
gorically :—The ‘‘notochord”’ of the Pterobranchia represents
the vermiform process of the stomochord of the Enteropneusta
(Spengelide) ; but there is not sufficient evidence upon which
to found an opinion as to its antecedent history.

In the process of regional differentiation of the entire body,
which is such a characteristic feature of the Knteropneusta, the
part of this influence devoted to cephalisation has, as has been
already mentioned, led to serious changes at the anterior end

1 The above view gives an adequate explanation of Roule’s ventral diverti-
culum sufficient at least to show that the extraordinary theoretical excursion
which this author makes is quite superfluous. Roule compares Actinotrocha
directly with the Vertebrates because the Enteropneusta are ‘‘ quelque peu
aberrant.” As a fact there is, in my opinion, reason to suppose that the
Enteropneusta are even nearer the direct line of Craniate descent than
Amphioxus. To set aside the Enteropneusta as aberrant forms argues
ignorance of the group.

Reference.—Louis Roule, ‘Sur la place des Phoronidiens dans la
classification des animaux, et sur leurs relations avec les Vertébrés,”’ ‘C. R..,’
t. exxvii, 1898, p. 633.

2 §. F. Harmer, “On the Notochord of Cephalodiscus,” ‘Zool. Auz.,’
1897, p. 342. (Masterman’s reply thereto, ibid., p. 443.)

238 ARTHUR WILLEY.

of the trunk. Not only does a portion of the gut become pro-
jected into the proboscis, with the result that its lumen has
become vestigial and its walls rigid, but gill-slits have been
abolished from the anterior portion of the gut which lies in the
collar region.1 Masterman’s pleurochords lie in the collar or
lophophoral region, and from his writings * they appear to be
vestiges of the gill-clefts which still persist in Cephalo-
discus.

As we have seen, the stomochord of the Enteropneusta is
a derivative of the collar-gut, and retains vestiges of structures
formerly serving another function in the post-oral collar region.
Thus we may conclude, in accordance with the preceding
considerations, that the pleurochords of Actinotrocha, the
gill-clefts of Cephalodiscus, and the lateral pouches of the
enteropneustic stomochord are the persistent vestiges of primi-
tive gill-clefts belonging to that portion of the body which, in
the Enteropneusta, is now specialised as the collar region.
The great series of truncal gill-clefts is entirely lacking in
the sessile forms, just as in the Ascidians there are strong
grounds for the interpretation of the numerous branchial stig-
mata as having originated by the subdivision of a single pair of
gill-slits® which persist in their undivided condition in Appen-
dicularia.4

1 In Amphioxus the first larval gill-slit closes up.

2 A. T. Masterman, ‘On the Further Anatomy and the Budding Processes

of Cephalodiscus dodecalophus,” ‘Trans. Roy. Soc. Edin.,’ vol. xxxix,
1898, p. 507.

3 From the mode of origin of the primary branchial stigmata in Ciona
intestinalis I thought three primary gill-clefts were represented in the
Ascidians, but a study of their formation in Molgula manbattensis con-
vinced me that such an interpretation could not be upheld; and on this point
I modified my views, and am now disposed to recognise the truth of Vau
Beueden and Julin’s hypothesis as to the presence of one pair only of primary
slits (see A. Willey, ‘ Amphioxus and the Ancestry of the Vertebrates,’ 1894,
p. 232).

4 It follows from what has gone before that the anterior portion of the
body of the stomochord in Kuteropneusta, that is the part intervening be-
tween the vermiform process (when present) and the region of the cecal
pouches (Fig. 3, a. 8.), corresponds to the functional cesophagus of Actino-
trocha; not that Actinotrocha is itself an ancestral form, but it appears to

RECENT WORK ON THE PROTOCHORDA. 239

A word is necessary as to the development of the Entero-
pneusta and the significance of the direct and indirect methods
of development. In the text-book before us the authors
consider the direct development as the more typical. This may
be so in a certain sense, but it is necessary to bear certain facts
in mind.

Spengel showed clearly that the Enteropneusta are divisible
into three families, but he only named one of them, namely, the
Ptychoderidz. I propose to call the other two families the Spen-
gelide and Balanoglosside respectively. Spengel was of the
opinion that the Balanoglossidee comprise the most primitive
forms, and the Ptychoderide the highest or least primitive
forms. As I have before stated, I hold the Ptychoderide to be
the primitive family, emphatically not the Balanoglosside ; I
think the anatomy of Ptychodera flava shows this conclu-
sively.

The Balanoglosside are all northern forms (White Sea,
Greenland, Canada, Massachusetts), with relatively large eggs
(from 1 to 14 mm. in diameter), and from the size of the egg
alone we are justified in concluding that they develop directly
in the manner described by Bateson in B. kowalevskili. The
Ptychoderide+ have small eggs (rarely more than =, mm. in
diameter) which develop into a Tornaria larva, i. e. in-
directly.

have retained some of the characters of a primitive creature, just as the
Ascidian tadpole retains primitive features which have quite disappeared from
the larva of Amphioxus.

The difficulty will naturally arise as to how the portion of the primitive
gut, represented in the stomochord, could have been projected past the
mouth. I do not think we are obliged to make an obstacle of this difficulty.
The principle of segregation will temporarily account for it. That segregation
has taken place is shown by the origin of such a complex structure as the
stomochord from a simple primordium (the mode of development of the vermi-
form process of the stomochord in Spengelidz is unknown).

There is so much on the surface which demands explanation that I have
ventured on dangerous ground in the endeavour to collate the various facts.

1 Probably also the Spengelide (Schizocardium, Spengelia, Glandi-
ceps).

you. 42, PART 2.—NEW SERIES. Q

24.0 ARTHUR WILLEY.

It is therefore a striking fact that the more primitive forms
have the indirect method of development. ‘There is probably
a special significance in this seeming paradox, and in order to
get at the meaning of it, it is important to call to mind a
parallel case. The difference in the method of development
followed by Peripatus capensis and by Peripatus nove-
britannie is, allowing for the intra-uterine environment,
precisely the difference between direct and indirect develop-
ment. The egg of P. capensis (‘5 mm.) is about five times as
large as the egg of P. nove-britanniz, and there are not
wanting anatomical features which point to the more primitive
character of the latter species.

Thus both in the Enteropneusta and in the Onychophora
the most primitive forms pass through an indirect development ;
and in both cases it is the indirect development which yields
information about the proximal relationships of the respective
groups; while the direct development apparently instructs us
in the matter of the primordial significance of the organisation
(cf. blastopore of P. capensis and body-cavities of B. kowa-
levskii).

In their treatment of the development of Amphioxus, MM.
Delage and Hérouard expose themselves to criticism at several
points. They have apparently overlooked the works of Van
der Stricht and Sobotta,! especially the latter, in which the
question of the polar bodies of Amphioxus is practically settled.
These authors found that the first polar body is extruded while
the egg is still inside the ovary, and a portion of the egg-mem-
brane is constricted off with the first polar body, so that the
latter comes to lie quite outside the membrane. When the
latter springs away from the egg at the time of fertilisation,

1 Van der Stricht’s paper, “La maturation et la fécondation de l’ceuf
d’Amphioxus lanceolatus” (‘Arch. de Biol.’ xiv, 1895, p. 469), is
quoted in the bibliography at the end of the work, but Sobotta’s latest important
paper on this subject is not quoted; perhaps it is too recent. J. Sobotta,
“Die Reifung und Befruchtung des Kies von Amphioxus lanceolatus,”
‘Arch. f. mikr. Anat.,’ vol. 1, 1897, p. 15.

RECENT WORK ON THE PROTOCHORDA. 241

the first polar body is removed from the surface of the ovum
and so is lost.!

In their account of the later development, the authors of the
Traité have been seriously led astray by the recent work of
R. Legros.? I regret to say that this author has produced a
paper of a highly destructive character, from which it would
appear, to the uninitiated, that his predecessors are incapable
observers. As one of his principal results he seeks to show, by
transverse sections through embryos preserved in a sublimate-
acetic mixture, that the przoral pit, which is such a distinc-
tive feature of the larva of Amphioxus, arises as a solid
ectodermal proliferation which subsequently hollows out.

Hatschek’s account of the origin of the preoral pit from the
left head-cavity has recently been confirmed by MacBride (loc.
cit.) by transverse sections through embryos preserved in osmic
acid.

Hatschek’s nephridium, according to Legros, arises as an
outgrowth from the alleged ectodermal przoral pit, and the
whole apparatus is subjected to an obvious and well-fitting
comparison with the hypophysis of Ammoceetes, and hence
with the hypophysis of Craniates in general ; in fact, he makes
it identical with the hypophysis of Ammoceetes, thus leaving
no room for change of function nor for evolution.

The orifice of the preoral pit of Amphioxus, considered
as a ceelomic cavity opening to the exterior, has generally
been supposed to be related to the proboscis-pore of the En-
teropneusta.®> From what has been said above (p. 231) it
follows that the preeoral pit, like the mouth, has quitted its

1 Sobotta describes two membranes round the egg, an inner and an outer,
but makes no reference to the follicular membrane described and figured by
Langerhasn.

? Robert Legros, “‘ Développement de la cavité buccale de l Amphioxus
lanceolatus. Contribution a l’étude de la morphologie de la téte,”
‘Archives d’Anatomie microscopique,’ tome i, No. 4, 1897; and tome ii,
No. 1, 1898.

3 For full treatment of this difficult subject see my memoir entitled ‘ En-
teropneusta from the South Pacific, with Notes on the West Indian Species,’
now in the press; Part iii, Zool. Results.

242 ARTHUR WILLEY.

primary association with the neuropore, the notochord inter-
vening. I knew this had happened in the case of the mouth,
my views on this point being acceptable to MM. Delage and
Hérouard, but I had not, until recently, realised that a
similar change had affected the przoral pit.

It will therefore be seen that Legros has probably touched
upon the fringe of a fundamental truth, so far as the mor-
phology of the preoral pit is concerned, although led thereto
by erroneous premises ; moreover the same truth was broached
by Bateson, to whose work the author makes no reference.

The contradictory result as to the origin of the preoral pit
arrived at by Legros would, if true, very seriously discredit the
work of Hatschek, but it has nevertheless been well received,
Klaatsch,! for example, ingenuously and uncritically rejoicing
at the ‘* Correctur der Hatschek’schen Angabe;” and it has
been adopted by MM. Delage and Hérouard.

Hatschek’s discovery of the conversion of the left head-cavity
into the preoral pit, which has been confirmed, let it be re-
peated, by MacBride, was a matter of unbiassed observation, —
and must have come as a serious shock to his sense of coelomic
propriety. To contradict Hatschek on this point in ignorance
of the living transparent embryos, and to base the contradic-
tion entirely upon sections through material preserved in sub-
limate and acetic, is surely very rash. The Belgian author
has obviously, in ‘this matter, been a victim of the microtome,
but it is to be feared that his results will more or less dominate
the subject throughout the next decade, since they have already
found a home in a leading treatise.

The external orifice of the club-shaped gland is a minute
pore below the anterior end of the larval mouth ; it is invari-
ably to be seen in all living larve before the metamorphosis,
but is not always easy to find in transverse section. Of course
Legros does not find it, and he denies its existence.”

1 Hermann Klaatsch, ‘‘ Ueber den Bau und die Entwickelung des Tenta-
kelapparates des Amphioxus,” ‘Verh. Anat. Ges. (Anat. Anz.),’ 1898,

p. 184.
2 See also E. Ray Lankester and A. Willey, ‘The Development of the

RECENT WORK ON THE PROTOCHORDA. 243

In his remarks dealing with the larval mouth he attempts
to set aside my work on the later larval development alto-
gether; but he does not refer to a single figure of mine, so
that I do not know whether he doubts the accuracy of all of
them or only of some.

During the transition from the lateral larval mouth to the
median adult velum the mouth maintains its integrity, but
alters its shape and rotates through an angle of 90°.

Legros wastes pages of ink in denying this rotation, i.e. in
denying a self-evident fact.

He goes on to say that the larval mouth does not, in its
entirety, become converted into the definitive mouth, but only
its anterior portion; the posterior portion closes up by fusion
of the lips; and the evidence which he brings forward in sup-
port of this assertion is neither furnished by section nor by
direct observation, but by measuring the relative distance of
the posterior angle of the mouth from the tip of the snout. It
is quite true that this distance becomes somewhat shorter,
a fact which my figures and description render completely in-
telligible by the change of shape and position which the mouth
undergoes. No soldering of the lips whatever takes place, and
to assert that it does so on the evidence which Legros adduces
is mere trifling.

The Ascidians naturally take up the most space in this
volume of the Traité, and the treatment which they receive on
the whole leaves little to be desired. In the account of bud-
ding in the Botryllidz the authors follow Pizon’s work almost
entirely. On many points Pizon’s results are in opposition to
the work of Hjort. It is a subject which requires more in-
vestigation.

The volume concludes with a useful summary of facts and
of theories relating to the origin of the Vertebrata, from which
I will make one quotation only. On the subject of the pre-

Atrial Chamber of Amphioxus,” ‘Quart. Journ. Mier. Sci.,’ vol. 31, 1890,
where the external orifice of the club-shaped gland is figured both as seen in
toto and as seen in section.

VoL. 42, PAR’ 2.—NEW SERIES. R

244, ARTHUR WILLEY.

oral lobe (p. 815) the authors say, “Chez les Tuniciers
adultes, il n’existe rien de tel, mais chez leur larve on retrouve
des dispositions tout a fait comparables a celles de l’Am-
phioxus.”

I have been treated to harsh words for holding the view that
the organ of fixation of the Ascidian larva represents the pre-
oral lobe. It is therefore possibly a matter for satisfaction
that this view finds favour with the authors, but I could have
wished that they had been more explicit. It may be worth
while to add, in order to ward off possible misunderstanding,
that for my own part Lam more than ever convinced of its
essential truth.

Klaatsch (loc. cit.) has recéntly made the suggestion
“mit allem Vorbehalt,” “dass das kolbenférmige Driise [club-
shaped gland of the larva of Amphioxus] die Anlage des Ten-
takelskelets darstellen kénnte.”

This is an astounding suggestion to make, and it will not
survive criticism. The skeletal elements of the buccal cirri
commence to appear long before the disintegration and conse-
quent disappearance of the club-shaped gland.!

1 T venture to refer Dr. Klaatsch to my paper entitled ‘‘ The Later Larval
Development of Amphioxus,” ‘ Quart. Journ. Micr. Sci.,’ vol. 32, 1891; and
to the figures on Pl. 15 accompanying the paper. He may there see for
himself the vindication of what I have said above. If he doubts the accuracy
of these figures, possibly in his next publication he will kindly inform us why
he does so.

THE STRUCTURE OF XENIA HICKSONI.

245

The Structure of Kenia Hicksoni, nov. sp.,
with some Observations on Heteroxenia

Elizabethe, Kolliker.

By

J. H. Ashworth, D.Sc.(Lond.),

Demonstrator in Zoology, Owens College, Manchester.
With Plates 23—27.

ConrTENTS.

Introduction

External Characters of the Golanys Palas Tentacles, Banules

Diagnosis of the Species Xenia Hicksoni ;

General Anatomy: Stomodzum and Mesenterial Tilemente: Guten
of Polyps; Mesogloa, its Canals and Cells

Kctoderm: Nematocysts, Spicules

Mesenteries: Mesenterial Filaments, Cells in Musoplren of Mesenteric

Endoderm: Giant Flagella

Mesogloea; Superficial and Internal Gana Sesame Galles in Meee

Spermatogenesis

Nervous System

History of our icrawleiee of the Buds of the monte

External Characters of the Buds of Xenia Hicksoni. ; :

External Characters of Heteroxenia Elizabethe: of its young
Polyps and of its Siphonozooids .

Internal Anatomy of Heteroxenia, Autozooids, Siphonozowiae Stem

Origin and Internal Structure of the Buds of Xenia Hicksoni

General Summary

Literature

VOL, 42, PART 3.—NEW SERIES. R

PAGE
246
246
249

251
257
262
264
267
274
277
278
282

284
289
291
297
299

246 J. H. ASHWORTH.

Introduction.

At the suggestion of Professor Hickson I undertook to
examine a specimen of the Alcyonaceous coral Xenia, which
he had collected on the reefs of Talisse Island, North Celebes,
chiefly with the intention of working out the arrangement of the
canals of the colony. As the investigation proceeded the pre-
servation of the colony was found to be so exceptionally perfect
that it seemed desirable to make a study of its histology, and as
several new and interesting points were early observed I finally
decided to work out in detail the complete anatomy and histo-
logy of the colony. Although many authors have described
the external characters of various species of Xenia, few have
paid any attention to their internal structure. Bourne (1895)
has referred to the canal system, the mesogloea, and the
distribution of spicules in two species of Xenia, and in
Heteroxenia Elizabethe, and Kolliker (1874) described
the anatomy of his new Heteroxenia Elizabethe as far as
its very imperfect preservation would permit. These two
accounts contain practically the whole of our knowledge of
the internal structure of these two genera, and hence, when
the beautiful preservation of this colony of Xenia from Talisse
was apparent, there was a strong inducement to attempt a more
complete account of its detailed anatomy and histology.

The work has been carried on during the past two years in
the zoological laboratories of the Owens College. My best
thanks are due to Professor Hickson for the beautifully pre-
served specimen upon which most of my work has been done,
and for advice and criticism given during the progress. of the
work. Iam &.so indebted to Professor Lankester for a speci-
men of Heteroxenia Elizabethe, to Mr. J. S. Gardiner for
a specimen of Xenia from Rotuma, and to Dr. Arthur Willey
for fourteen specimens of Xenia from various reefs in the
Pacific.

External Characters of the Colony (Pl. 28).

The Xeniide are distinguished from all other Aleyonaria-
by their soft, fleshy consistency and non-retractile polyps. The

THE STRUCTURE OF XENIA HICKSONI. 247

former character is due to the fact that their spicules are very
minute rounded or oval discs, which have an organic basis
impregnated with only a small quantity of calcium carbonate.

This colony arises from a single stem, which is slightly
expanded at its point of attachment to the rock, and measures
about 15 mm. in diameter at that point. This basal portion is
very short and thick, and supports four main stems, all of
which divide into two or more, producing altogether thirteen
stems or branches ranging in length from about 10 mm, to 30
mm., and in breadth from about 4mm.tol0 mm. The total
height of the colony from the point of attachment to the tips
of the highest polyps is 50 mm.

The polyps are, for the greater part of their length, bound
together in bundles of about forty to sixty, each bundle forming
one stem of the colony. The free portions of the polyps arise
from the slightly expanded umbrella-shaped area at the distal
end of each stem. Many of the polyps stand almost perpen-
dicularly to the convex disc, but those near the edge of the disc
hang downwards towards the base of the colony. The polyps
are closer together near the edge of the umbrella, being here
‘5 mm. to ‘7 mm. apart, whereas in the middle of the umbrella
they are 1 mm. to 2 mm. apart.

Polyps (Pl. 24, fig. 2).—The polyps, or, more correctly,
the free portions of the polyps, are non-retractile and mode-
rately long and slender. The tentacles are half to two thirds
as long as the body of the polyp. Hach tentacle bears on its
inner side numerous short, conical elevations with rounded
ends. These correspond to the pinuules found on the tentacles
of other Alcyonaria.

The colour of the colony in spirit is ight brown.

As mentioned above, the free portions of about forty to sixty
fully developed polyps project from the umbrella-shaped area
at the distal end of each stem, but besides these there are several
younger polyps or buds in various stages of development, and
these are invariably situated on the edge of the umbrella.

In fully developed specimens the following are the measure-
ments :—‘‘ Body ” of the free portion of the polyp 4 mm. to

248 J. H. ASHWORTH.

7 mm. long, and 1:0 mm. to 1'2 mm. broad. Tentacles 2 mm.
to 5'7 mm. long, and ‘75 mm. broad. The total length of the
adult polyps is thus 6 mm. to 12 mm.

The body of the polyp is cylindrical and its wall moderately
strong. Inseveral of the Xeniide the body-wall of the polyp
is so weak that when the colony is taken out of spirit the polyps
fall together into a mass. In this species, however, the body-
wall is just strong enough to support the polyps in an upright
position, so that on removing the colony from spirit the polyps
do not hang limply, but remain standing approximately in their
natural positions.

Tentacles and Pinnules (Pl. 24, figs. 2 and 3).—Each
polyp bears eight tentacles, each of which is provided with
numerous pinnules. The pinnules on each side of the middle
line of the tentacle are arranged in three longitudinal rows,
and they form also somewhat oblique transverse rows of three
pinnules rising from the oral towards the aboral side of the
tentacle (fig. 3, D). When the tentacle is viewed from the
inner or oral aspect, all the pinnules are generally visible
(fig. 3, B, D), but on the outer or aboral side of the tentacle
only the outer longitudinal row of each side is, as a rule,
seen (fig. 3, 4, C). The pinnules are often clearly separated
into the two series of three rows in each by a narrow area
which extends along the middle line of the inner face of the
tentacle, from the base to within a short distance of the tip
(fig. 3, D). This area, free from pinnules, may be ‘25 mm.
across, and may often be traced to within 1 mm. of the tip of
the tentacle. In other specimens, however, it is entirely
obliterated, and the median pinnules of the two series are in
contact with each other, at any rate at their bases. The
width of this area varies, not only in separate individuals,
but in the different tentacles of the same individual. These
variations are probably due to the different degrees of con-
traction of the tentacles on killing, and the condition in which
the free area is well marked is seen only in those tentacles
which have been killed in an expanded condition.

At the tip of the tentacle the piunules are smaller than those

THE STRUCTURE OF XENIA HICKSONI. 249

in the middle. They are arranged more or less in two series,
one on each side of the middle line of the tentacle, but three
rows on each side are not distinguishable; at a distance of
about 1 mm. from the tip of the tentacle, however, the typical
arrangement of two series, with three rows of pinnules in each,
is gradually assumed (fig. 3, B).

At the base of the tentacle the pinnules are much smaller
but have the typical arrangement, except in the case of one or
two of the proximal transverse rows. Here the pinnules appear
to be in course of formation, the outer ones being formed first,
the inner ones developing from without inwards (see fig. 3, D).

On looking at the tentacle from the outer side only the
outermost row of pinnules is usually visible. These, which are
about twelve to twenty in number on each side, are set close
together and point towards the tip of the tentacle (fig. 3, A).
At the tip of the tentacle the arrangement ofthe pinnules may
be well studied, and, as in Aleyonium (Hickson, 1895), they
are not paired (Pl. 24, fig. 2). The pinnules are conical eleva-
tions with rounded ends. Those in the middle portion of
the tentacle are about ‘5 mm. long and ‘15 mm. to ‘2 mm.
broad, but those nearer the base and tip are smaller. The
pinnules when fully expanded are about three times as long as
they are broad, and each tapers gradually from its base to its
blunt, rounded tip. When slightly contracted the pinnule is
somewhat swollen at its base, and if further contracted becomes
more swollen and globular at its base as its length decreases.
Although the body of the polyp is non-retractile, the tentacles
are often found slightly contracted, being in many cases curled
inwards over the mouth. Several examples of tentacles in this
condition are shown in fig. 1.

Diagnosis of the Species Xenia Hicksoni.

The species of Xenia are distinguished from each other by
the general form of the colony, the size of the polyps and
tentacles, the number of rows and shape of the pinnules, and
the presence or absence of an area free from pinnules on the
inner face of the tentacle.

250 J. H. ASHWORTH.

After careful comparison with the accounts of all the hitherto
described species, I am unable to refer this specimen to any of
them, and therefore I have established for it a new species with
the name Xenia Hicksoni. Its characters are as follow:

The colony consists of several cylindrical, usually branched
stems, arising from a single thick stem or base. The stems
range in length from 10 mm. to 30 mm., and in breadth from
4mm.to 10mm. From the arched or convex summit of each
stem the free parts of the polyps arise. These are smaller and
moderately close together near the edge of the summit, but
larger and further apart in the middle of the arched end. The
polyps (including tentacles) measure 6 mm. to 12 mm. in length,
and 1mm. to1‘2 mm.in breadth. The tentacles are moderately
slender and 2 mm. to 5‘7 mm. long. Each tentacle bears on
the inner side two series of pinnules, each series consisting of
three rows of twelve to twenty pinnules in each row. There
is usually a narrow area free from pinnules extending along
the middle line of the tentacle to within about 1 mm. of the
tip. The pinnules are conical elevations with rounded ends.
Those in the middle of the tentacle are about ‘5 mm. long, and
are about three times as long as they are broad; those nearer
the base and tip of the tentacle are somewhat shorter. The
body-wall of the polyps is moderately thick, and is strong
enough to support the polyps in their natural position when
the colony is removed from spirit. The spicules are round or
oval discs measuring ‘012 mm. to ‘022 mm. in length, ‘006 mm.
to ‘013 mm. in width, and about ‘004 mm. in thickness. They
are numerous in the ectoderm of the stem and of the body of
the polyp, but are practically absent from the tentacles and
pinnules.

Round the edge of the umbellate summit of each stem are
a few buds or young polyps in early stages of development.
The specimen is light brown in colour (in spirit).

Habitat.—The reefs of Talisse Island, North Celebes.

This species appears to differ from most, if not all other
species of Xenia in the absence of spicules from the tentacles
and pinnules. In general form of the colony this specimen

THE STRUCTURE OF XENIA HICKSONI. 251

most resembles X. umbellata, Savigny, but it differs from
the latter in possessing smaller polyps, with much shorter and
stouter pinnules, which do not leave the axis of the tentacle
free along its whole length (cf. Klunzinger, 1877, pl. 3,
fig. 3a).

General Anatomy.

Stomodeum and Mesenterial Filaments.—In the
centre of the oral disc of each polyp there is a funnel-shaped
depression about one third of a millimetre in depth leading to
the mouth. This depression is formed by partial contraction
of the oral disc; if the polyp were fully expanded this depres-
sion would not exist, but the mouth opening would be level
with the oral disc. The mouth (Mo.) leads into the stomodeum
(S¢.), which is 1°8 mm. to 2°2 mm. long. The stomodeum is
long compared with the length of the free portion of the polyp,
and in longitudinal section presents a striking appearance,
running down, as it does, so far into the celenteron. The
stomodeum is oval in transverse section, being somewhat
flattened from side to side. It has a well-marked ventral
groove or siphonoglyph (Si.), the cells of the lower third of
which bear long flagella (F.). The groove is not as well
marked in the upper as in the lower portion of the stomodeum,
and is scarcely discernible at the mouth opening. The
columnar epithelial cells forming the siphonoglyph are, as is
usual, longer than those of the rest of the stomodeum, and
these cells bear very long flagella (‘07 mm.), which in some
examples extend almost to the centre of the cavity of the
stomodeum. The epithelium of the rest of the stomodeum is
smooth and not folded in any way. Many of these epithelial
cells bear short cilia on the free surface, but among these are
numerous cells (G.), which are, like goblet-cells, swollen or
flask-shaped, due to the presence of some secretion to which
they give rise. These cells generally appear to be empty,
having discharged their secretion, which in some cases can be
seen issuing from the cell into the cavity of the stomodzeum
(PI. 26, fig. 18). These secreting cells occur chiefly in the

252 J. H. ASHWORTH.

middle and lower portions of the stomodeum, and are most
abundant on the lateral walls near the siphonoglyph. They

i
=
‘=

A

=

Fie. 1.—Semi-diagrammatic view of one half of a polyp which has been cut
along the dorso-ventral line. Only the bases of the tentacles are shown x 20.

Fic. 2.—Transverse section through the lower third of the stomodeum
(about the level of the reference letter 7 in Fig. 1) x 80.

D.M.F. Dorsal mesenterial filament on the edge of the dorsal mesentery ;
F. Flagella of siphonoglyph; G.C. Gland cells of stomodeum; Z.M. Edge
of lateral mesentery (mesenterial filament absent) ; Mo. Mouth; S%. Siphono-
glyph; S¢. Stomodeum; 7. Tentacle; V.M@. Edge of ventral mesentery
(mesenterial filament absent),

THE STRUCTURE OF XENIA HICKSONI. 253

do not occur among the cells which form the siphonoglyph
(Pl. 25, fig. 10). Secreting cells have not hitherto been noticed
in the stomodzum of the Alecyonaria (Ashworth, 1898).

These goblet-cells of the stomodzum are the only secreting
cells connected with the digestive cavity, as the six thick short
ventral and lateral mesenterial filaments, which bear the gland-
cells in other Alcyonaria, are absent in all polyps of this
Xenia (see Woodcut, p. 252). Only the dorsal mesenteries
possess thickened edges, forming two mesenterial filaments
(D.M.F.) which have a similar course and structure to those
of Aleyonium. The free edge of the ventral and lateral
mesenteries is only very slightly thickened, and the thickening
is due entirely to the greater amount of mesogloa present at
the edge of the mesentery. The cells which cover the edge
differ in no way from those which cover the remaining portions
of the mesentery.

New points in the anatomy of this Xenia are the presence
of gland-cells in the stomodzeum, and the absence of the six
ventral and lateral mesenterial filaments usually present in the
polyps of the Aleyonaria. Wilson (in Kophobelemnon,
1884) and Hickson (in Alcyonium, 1895) have shown that
these mesenterial filaments bear the cells which produce the
digestive secretion. I would suggest that the absence of these
filaments in this Xenia is correlated withthe presence of gland-
cells in the stomodzeum, and that the latter, judging from
their appearance and position, perform some digestive function.
As mentioned above, the cells are most abundant on the ventro-
lateral walls of the stomodzum, near the siphonoglyph. As
the flagella of the siphonoglyph create an inward current of
sea water carrying food particles, which passes along the
ventral groove, the greater abundance of secreting cells
in the walls abutting on this groove is suggestive of the
digestive function of the secretion, which can readily be poured
out on to the ingoing food particles.

The siphonozooids which occur in some other Aleyonacea
(e.g. Sarcophyton) and in Pennatulids are the only re-
corded examples of polyps in which the ventral and lateral

254 J. H. ASHWORTH.

mesenterial filaments are absent. According to Wilson (1884),
these siphonozooids derive their food supply from the autozooids
or feeding polyps, the dorsal mesenterial filaments of the former
creating upward currents which cause a flow of fluid from the
autozooids to the siphonozooids, through the canals which
connect them together. Food undergoes digestion in the
autozooids, and some of the products are passed on to the
siphonozooids, hence the latter do not require cells to produce
a digestive secretion.

In this Xenia the secretion in connection with the digestive
cavity is formed, not by endoderm cells, but by cells which are
derived from the ectoderm, as from a study of the buds I have
found that the stomodeum is ectodermic in origin in this as
it is in other Aleyonaria (Wilson, 1883). Since the absence
of ventral mesenterial filaments in Xenia Hicksoni was
proved I have carefully examined all the other specimens of
Xenia at my disposal. These are sixteen in number, viz. one
from Talisse Island, North Celebes, one from Rotuma, Fiji
Islands, and fourteen from various reefs in the Pacific. In
all these the ventral and lateral mesenterial filaments are
absent, there being only a very slight thickening of the free
edges of the mesenteries, this thickening being entirely due to
a slight increase in the amount of the mesoglea along the free
edges. In all the specimens the two dorsal mesenterial
filaments are present, and have the typical course and struc-
ture. In Heteroxenia Elizabethe the two dorsal mesen-
terial filaments only are present. The absence of ventral and
lateral mesenterial filaments may, therefore, be considered as
one of the characters which distinguish the polyps of the genera
Xenia and Heteroxenia from those of other Alcyonaria.
In two at least of the specimensof Xenia! from Dr. Willey’s
collection some of the cells of the stomodzum are goblet-like,
and similar to those described above in the stomodeum of
Xenia Hicksoni.

Celentera of Polyps.—The other parts of the colony
present a structure similar, in the main features, to that

1 X. crassa, Schenk, and X. viridis, Schenk.

THE STRUCTURE OF XENIA HICKSONI. 255

of other Alcyonaria. The polyps are, for the greater part
of their length, bound together in bundles of about forty
to sixty, each bundle forming one stem of the colony.
The external characters of the free portion of the polyp
have been described above. The eight mesenteries of
the polyp are arranged as in typical Alcyonaria. On their
ventral faces they bear the retractor muscles, and on the
opposite faces the protractor muscles, neither of which are well
developed. This somewhat feeble development of retractor
and protractor muscles accounts for the non-retractile nature
of the polyps. The intermesenterial spaces are continued
upwards into each tentacle and into each pinnule. In the free
portion of the polyp, and particularly in the tentacles and
pinnules, these spaces are to a large extent filled up by
zooxanthellz. The specimen is a male, and in the lower part
of the free portion of the polyp, and in the upper part of the
stem, the celentera are crowded with sperm-sacs containing
spermatozoa in all stages of development, from the small
masses containing only two or four primitive sperm-cells to
the large mature sperm-sacs containing many hundreds of ripe
spermatozoa.

Mesoglea, its Canals and Cells.—In the stems the
ceelentera of the polyps are bound together by a moderately
large quantity of mesoglea. The mesoglea immediately round
each polyp is slightly denser than that further away, so that
in transverse sections of the stem, especially if the sections
be taken through the upper part, one can distinguish the more
deeply staining ring of slightly denser mesogloea, which defi-
nitely belongs to the ccelenteron within it, from the less dense
mass of mesoglea between these rings, which cannot be
assigned to any polyp or polyps (Plate 25, fig. 9). Traversing
the mesoglea of the stem are numerous canals, cords and
strands of cells, which place all the parts in intimate com-
munication with each other. The canals may be divided into
two systems—a superficial system and an internal] system.

The superficial canal system (figs. 8,9, Sup. Can.) is
formed by a plexus of numerous endodermic canals, which

256 J. H. ASHWORTH.

are situated in the outer portion of the mesoglea, just beneath
the ectoderm of the stem. This system of canals extends
all round the cylindrical stems, and also runs on the umbrella-
shaped areas from which the free portions of the polyps arise
(fig. 8). The cavity of this system of canals is invaded
throughout by zooxanthelle, which are especially numerous
in the canals in the upper part of the stem.

The internal canal system consists of a series of longitudinal
canals (figs. 8, 9, Long. Can.) which run generally in a
sinuous or zigzag course in the mesoglea, between the
celentera of the polyps. These canals le in the mesoglea,
almost equidistant from the surrounding ccelentera, and they
run in this position from the top of each stem to the base of
the colony. These canals are also endodermic, and have
usually only a small lumen. They communicate with the
ceelentera of the polyps, with the superficial canal system, and
with each other.

At the base of the colony the canal system is exceedingly
complicated, due to the numerous branchings and anastomoses
of the canals. The ccelentera are continued down to the base
of attachment of the colony, where they are in communi-
cation with the numerous branches of the canal systems.
The celentera are readily distinguishable from the canals by
their greater size, and the presence in them of eight small
ridges, to which the mesenteries in this region are reduced.
Besides the superficial and longitudinal canals briefly de-
scribed above, there are in connection with the longitudinal
canals numerous small lateral or transverse canals, with small
lumen, which pass from the canals either to other neighbouring
canals or to the celentera (figs. 8,9). The coelentera of the
polyps do not open into each other directly, but are indirectly
connected by these canals.

The base of the cylindrical main stem is somewhat flattened
out, and this basal portion is hard and horny; and, being
closely applied to the rock, provided a firm basis upon which
the other parts of the colony were supported,

THE STRUCTURE OF XENIA HICKSONI. 957

Ectoderm.

The ectoderm is a moderately thick layer, in which large
columnar and smaller interstitial cells may with difficulty be
distinguished. If a section of a tentacle or pinnule, which
was well expanded at the moment of killing, be examined, the
ectoderm is then seen to consist of a row of cells elongated at
right angles to the free surface, below which are smaller
rounded cells which probably correspond to the interstitial
cells of the ectoderm of Alcyonium and other ccelentera.
(PIS 255 fig <11):

The protoplasm of the ectoderm cells is very finely granular ;
occasionally cells are met with containing a few small vacuoles.
Many of the ectoderm cells of the tentacles and pinnules are
produced at their inner ends into muscle processes which lie
in the outer portion of the mesoglea, parallel to the free
surface. These muscles are all longitudinal in direction. The
ectodermic muscles of the tentacles are much more strongly
developed on the oral than on the aboral side, especially at the
base of the tentacles, where the muscles of the oral side form
a strong band beneath the cells quite eight times as thick
as the band of muscles on the aboral side. Nearer the tip
of the tentacle the muscles of the two sides become almost
equal.

In the body of the polyp ectodermic muscles are present
only in the distal portion around the base of attachment of
the tentacles and for a distance of about a millimetre below
this point. Myo-epithelial cells are absent from the ectoderm
of the stem. The absence of muscle-cells from the ectoderm
of the stem and the greater portion of the body of the polyp is
connected with the non-retractile character of these parts.
Their presence in the tentacles, pinnules, and distal portion of
the body of the polyp confers on these parts some power of
contraction, and an examination shows that the pinnules and
tentacles vary somewhat in length and shape, and that the
tentacles are often turned inwards over the mouth, due to the
contraction of the muscles of the oral side. The absence of

258 J. H. ASHWORTH.

spicules from the tentacles and distal portion of the polyp
renders the ectoderm softer and more pliable, and therefore
more readily acted upon by the contraction and expansion of
the muscle processes of its cells.

Among the ordinary columnar cells there are in the ectoderm
of the stem and the body of the polyp large swollen cells
which probably secrete the mucus which thinly covers the
external surface of most of these parts (Pl. 26, fig. 17,
Muc. C.). These mucus-cells are large and abundant in the
angle of the Y-shaped piece formed where a stem divides.

The ectoderm cells present on their outer side a moderately
plane surface, but on the inner side a more irregular one, as
numerous processes pass from the cells inwards and establish
communication either with the endoderm or with cells lying
deeper in the mesogleea (fig. 17).

On reaching the base of the colony the ectoderm curves
inward, and was applied to the face of the rock to which the
colony was attached.

The stomodzum is ectodermic in origin in this as in other
Alcyonaria (Wilson, 1883), as may be seen from a study of
the buds. The presence of a ventral ciliated groove and of
gland-cells and other features of its structure have already been
referred to (p. 251). The ectoderm of the stomodzeum and the
adjacent endoderm are connected by numerous cells or strands
of cells passing through the thin mesoglceal lamina which
separates the two cell layers (Pl. 26, fig. 18).

The ectoderm cells give rise to nematocysts and spicules.

Nematocysts.—The nematocysts are exceedingly numerous
in the ectoderm of the tentacles and pinnules (PJ. 25, fig. 11,
Nem.) ; there are large numbers in the ectoderm of the body of
the polyp, rather fewer in the ectoderm of the stem, and a few
in the ectoderm of the oral disc, in the funnel leading to the
mouth, and also in the upper part of the stomodeum. The
nematocysts are, as in other Aleyonaria, exceedingly small,
their length being ‘008 mm. and their breadth ‘002 mm. to ‘003
mm. They have bluntly pointed ends and are circular in trans-
verse section. Each nematocyst is formed in a cnidoblast cell

THE STRUCTURE OF XENIA HICKSON]. 259

(Pl. 25, fig. 12, Cn.C.), the nucleus and protoplasm of which lie
flattened against the capsule on one side and a little nearer
the inner than the outer end of the capsule. The filament or
thread lies coiled up inside this capsule, there being about
twelve coils distinguishable in the most favourable specimens.
The thread appears to be quite simple, there being no barbs
visible. Its length when shot out would probably be about
80 uw (08 mm.). The nematocysts are usually placed with
their long axes at right angles to, and their bluntly pointed
ends level with or slightly projecting from, the free surface of
the ectoderm. This can be especially well seen in sections of
the tentacles and pinnules (see fig. 11). The nematocysts
staindeeply with thionin, hematoxylin, and especially with iron
hematoxylin (Heidenhain). Iron hematoxylin is exceedingly
useful for staining the small nematocysts of Alcyonaria; in
fact, without some good staining reagent it would be very
difficult to find the minute capsules in many cases. Moseley
(1881, p. 119) wrote that “no nematocysts were found in
Sarcophyton,” but by using the iron hematoxylin stain I
have found them in sections passing through the ectoderm of
the tentacles. They are very similar in shape to those of
Xenia Hicksoni, but smaller in size, being only 6 uw to 7 wu
long and 2 wide. The nematocysts of Alcyonaria are all
very small, as may be seen from the following table:

Sarcophyton pauciflorum! 6 w— 7 p» long and 2 4 wide (J. H. A.)

Alcyonium digitatum . . 7354p 3 9) 2p—sp,, (Hickson)
Menia Hickson). 2.5. .- Sip alee 2 op gy, (ESA)
Heteroxenia Hlizabethe . Iu in. Game pe seme tis.)
Clavularia viridis . .. 9p—lOp ,, ,, 2u—dp,, (J.H.A,)
Heiiopora cerulea . . . 9p ie eee t= CN Loseley)
Clavularia prolifera . .10p—l5p ,, ... =. =. . (v. Koch)

Spicules (Pl. 25, figs. 13—15; Pl. 26, fig. 16).—The
spicules, the form of which was compared by Kolliker to that
of red blood-corpuscles, are rounded or oval discs, which are,
however, sometimes bilobed (fig. 13). They are ‘012 mm. to

1§, pauciflorum, Ehrenberg = Lobophytum pauciflorum, v.
Marenzeller, ‘ Zool. Jahrb.,’ i, 1886.

260 J. H. ASHWORTH.

‘022 mm. loug, ‘006 mm. to ‘013 mm. broad, and ‘003 mm. to
‘005 mm. thick.

Spicules are absent from the tentacles and pinnules of all .
the polyps examined except one. In the latter a few small
spicules (about 8u to 12 4 long) are present in the ectoderm of
the tentacles, but only in the proximal ‘4 mm.

In the deeper part of the ectoderm, just below the point of
attachment of the tentacles, spicules are present in small
numbers, while in the middle and proximal portions of the
body of the polyp they are very numerous, a tangential
section of the body-wall in this region showing that the
spicules form an almost complete layer in the deeper part of
the ectoderm. Spicules are numerous in the ectoderm of the
stem, and especially numerous in the angle of the Y-shaped
piece formed at the point of division of a stem. In the lowest
parts of the colony, i.e. in the portion attached to the rock,
they are present in enormous numbers, practically filling the
entire mesogloea in that region. The largest examples of
spicules are to be found in the basal portion of the colony.
(Those shown in fig. 13 are from this region.) The absence
of spicules from the tentacles, pinnules, and from the body of
the polyp around the base of the tentacles is correlated with
the power of contractility, slight though it is, which is pos-
sessed by these parts. The presence of an almost complete
layer of spicules forming a more or less rigid cylinder in the
middle and proximal portions of the body of the polyp and in
the stem, together with the absence of muscle processes from
the ectoderm cells of these portions, sufficiently accounts for the
non-retractile character of these parts of the colony. The
increased number of spicules in the angle between two branches
gives the required rigidity to this part, and prevents flexure of
the branches and consequent closure of some of their canals
and coelentera.

Besides the extraordinary number of spicules present in the
ectoderm and mesoglea of the last few millimetres of the base
of the colony, this part is further strengthened by much of the
mesoglcea becoming converted into a dense and horny substance,

THE STRUCTURE OF XENIA HICKSONI. 261

so that the part of the colony attached to the rock is hard and
quite different to the touch from any other part of the colony.
This hard base would afford a firm attachment and a rigid
support to the branches which arise from it.

The spicules are formed within cells which at first lie in the
deeper parts of the ectoderm or have migrated into the outer
parts of the mesoglea (fig. 16). The arrangement of the
spicules is not very regular, as they appear to present to the
free surface their edge or their flat face indifferently. In
nearly all preparations the nucleus and remains of the proto-
plasm of the spicule-forming cell can be seen. The spicules,
which have a horny consistency, have an organic basis im-
pregnated with only a very small amount of calcareous matter.
They stain deeply with hematoxylin, especially with hemalum.
They do not dissolve when treated with acids, but shrink very
slightly, probably owing to the solution and extraction of the
small amount of calcareous matter which they contain. They
do not offer any difficulties in section-cutting, as the razor cuts
through them with little resistance, and sections 2 to 4 in
thickness may be readily obtained, although the specimen has
not been previously decalcified. On this account Xenia offers
exceptional facilities for the study of the development of
spicules. As mentioned above, each spicule is formed within a
cell lying in the deeper part of the ectoderm. When the
young spicule first makes its appearance in the cell its substance
is scarcely distinguishable from the protoplasm of the cell; in
fact, it is not until the spicule has attained a diameter of 3 p to
4 that it is possible to clearly differentiate it (Pl. 25, fig. 15).
The young spicule is then a smail disc which stains with
hematoxylin like the protoplasm of the cell, but its homo-
geneous structure enables the observer to distinguish it from
the finely granular cell-protoplasm which surrounds it. From
this stage the spicule grows regularly in length and thickness,
and the protoplasm of the cell covering it becomes gradually
thinner until, in a fully formed spicule, this protoplasmic
sheath forms an exceedingly thin investment, in which at one
part may be seen the small, somewhat flattened nucleus em-

von. 42, PART 3,—NEW SERIES. s

262 J. H. ASHWORTH.

bedded in a small mass of protoplasm (figs. 13 and 14). In
all the specimens, many hundreds in number, which I have
examined, the spicule develops in a single cell with one nucleus.
I have not been able to find any examples which showed that
two cells or two nuclei were concerned in the formation of the
spicule (cf. v. Koch’s account of the development of the
spicules of Clavularia prolifera, ‘ Morph. Jahrb.,’ vii,
p. 473, 1882).
Mesenteries (Pl. 26).

Mesenterial Filaments.—The stomodeum leads into
the celenteron of the polyp, which is subdivided by the usual
eight mesenteries (Pl. 25, fig. 10). Of these only the two dorsal
ones possess thickened edges or meseunterial filaments (Pl. 26,
fig. 19). The free edge of the remaining six mesenteries is
only very slightly thickened, this being due entirely to the
presence of a slightly greater amount of mesogloa near the
free edge of the mesentery. The cells which cover this
thickened portion differ in no way from those covering the
other parts of the mesentery (PI. 26, fig. 20). The six ventral
and lateral meseuterial filaments usually present in the polyps
of the Alcyonaria are not found in this genus. The dorsal
mesenterial filaments arise from the lower edge of the stomo-
dzeum and run in a sinuous course along the dorsal side of the
cceelenteron. In the primary polyps they may be traced to the
base of the colony. In transverse section the filament is
slightly bilobed, i. e. there is a groove (of slightly varying depth)
extending all the way down the middle of the free surface of
the filament (fig. 19). The cells on each side of this groove
bear long cilia (‘0075 mm.). The dorsal mesenterial filaments
are quite typical, and agree well with the accounts given of
those of other Alcyonaria by Wilson and Hickson. They are
probably ectodermic in origin, as they appear to be formed as
two downgrowths from the inner end of the stomodeum. The
cells of the filaments agree in structure with those of the
stomodeum, being finely granular and non-vacuolated, and
differing markedly from the much vacuolated neighbouring
endoderm cells,

THE STRUCTURE OF XENIA HICKSONI. 263

Muscles.—The retractor muscles are situated on the ventral
faces of the mesenteries and the protractor muscles on the
dorsal faces, as in Aleyonium. These muscles are some-
what feebly developed, as might be expected from the non-
retractile nature of the polyps. Shortening of the retractor
muscles produces a slight contraction of the oral disc and
consequent formation of the funnel-like depression leading to
the mouth, to which reference has already been made (p. 251).

Cells in Mesoglea of Mesenteries.—On examining 2
transverse section through the mesenteries, there is seen to be
a considerable quantity of mesoglea between the two endo-
dermic lamelle covering the mesentery (fig. 20). In this
mesogloea there are cells which have the reticulate protoplasm
and general appearance of endoderm cells. These cells migrate
into the mesogloea from the endoderm covering the surface of
the mesentery, and even in a young polyp ‘8 mm. long a few
cells have already taken up their position in the mesoglea.
In older polyps there is a larger number of these cells in the
mesoglea of the mesentery, though they are not equally
numerous in all parts. In the upper portion of the polyp,
about the level of the stomedzum, the mesogleeal cells are few
in number and small in size, but from this part downwards
their number and size gradually increase, until in the mesen-
teries in the upper portion of the stem they are large and
numerous, and in some cases completely fill up the mesoglea,
so that the mass of cells is in close contact on both sides with
the endoderm covering the two sides of the mesentery. Towards
the base of the stem the cells become fewer in number and
slightly smaller in size. These cells are found in the mesogloea
of all the mesenteries, but they are less numerous in the dorsal
mesenteries than in the remaining six. Many of the cells are
at first somewhat elongated or pear-shaped, with one or more
processes in connection with, or pointed towards, the endoderm ;
but later many of them become rounded and larger, and their
nuclei become much larger.

The primitive genital cells are derived from these large
rounded cells in the mesoglozea of the mesenteries at the base

264 J. H. ASHWORTH.

of the polyp and in the upper portions of the stem. That this
is the case is shown by the following:

(1) These cells are most numerous in those parts of the
colony where gonads occur in greatest numbers.

(2) In many cases one of these cells, surrounded by a thin
film of mesogloea, may be seen enclosed in a follicle of endoderm
projecting from the edge of the mesentery. These are exactly
similar to the neighbouring follicles which contain spermatozoa
in various later stages of development.

(8) The nuclei of most of these cells are large and spherical,
more vesicular than the nuclei of the adjacent endoderm cells,
and resemble the nuclei of the genital cells (see Pl. 27, fig. 30).

(4) The ripe spermatozoa are situated in a follicle covered
by a thin mesogleeal lamina, as well as by the endoderm cells
outside this, i.e. the ripe spermatozoa are situated in the same
layer as these cells, viz. in the mesoglea.

The migration of genital cells from the endoderm of the
mesenteries into the mesoglcea is similar to that described by
O. and R. Hertwig in Actiniz (‘Die Actinien,’ Jena, 1879,
p95; and» pls 7):

Endoderm (PI. 26).

The endoderm cells lining the celentera and the cavities
of the tentacles have a similar structure throughout the
colony. They are cubical or columnar, and contain many
small vacuoles which give the protoplasm a reticulate appear-
ance.

Cells which bear Flagella (figs. 20—25, 27).—Among
the ordinary endoderm cells there are numerous cells, the
inner or free end of which is produced ito a long process,
which is from four to eight times as long as the basal portion of
the cell. This process may be slender or moderately stout,
and its length may vary in different specimens from ‘015 mm.
to'12 mm. The basal part of the cell from which the process
arises has the reticulate protoplasm of an ordinary endoderm
cell, and the nucleus of the cell is situated in this portion.
The process is not vacuolated, and for the greater part of its
length its protoplasm exhibits a homogeneous or very finely

THE STRUCTURE OF XENIA HICKSONI. 265

granular structure. Its basal part, i.e. the part in continuity
with the vacuolated portion of the cell, stains deeply with
hematoxylin, and in most cases shows very faint longitudinal
striations which are visible only in the proximal third of the
process.

The processes usually taper towards their free end, but in
one instance this end is slightly broadened and flattened
(fig. 25, A). In one case the process, which is a very large
one, bears a short branch near the middle of its length
(fig. 25, B). This is the only branched process found among
many hundreds examined. The processes of most of the cells
project outwards almost at right angles to the free surface of
the endoderm (figs. 20, 21, 23, 24), but there are many similar
to the one drawn in fig. 22,in which the process is strongly
curved and apparently moderately flexible.

These curious processes are very numerous, and are found in
all parts of the endoderm lining the celenteron and tentacles,
but are most abundant in the portion of the ccelenteron
situated in the body of the polyp and in the upper part of the
stem (Pl. 25, fig. 9).

The nature of these processes is difficult to determine. In
the preliminary note to the Royal Society (1898, written in
February) I called them pseudopodia, but further investigation
shows that the word flagella would probably better express
their nature. The processes appear to be permanent, to have
a moderately definite shape gradually tapering from base to
tip, and to be flexible. The word pseudopodia implies more
temporary structures with many different and continually
changing shapes, while the term flagella implies tapering whip-
lash-like processes of more permanent and definite shape. The
homogeneous structure of the greater portion of these pro-
cesses, differing so markedly from the vacuolated granular
protoplasm of the rest of the cell, is also more in accord with
their being flagella, as pseudopodia have the same structure as
the body of the cell from which they are protruded.

It is difficult to suggest the probable function of these giant
flagella. They are evidently motile organs, as they may be

266 J. H. ASHWORTH.

found in all stages of flexion, some being practically straight,
while others are bent almost into a semicircle. Their action
probably serves to keep the liquid in the ceelenteron in slow
motion, thereby securing a more equal distribution of the
nutrient substances contained therein to the cells in their
vicinity.

Many of the endoderm cells are provided with “muscle
processes,” but these processes are not numerous in the
pinnules and in the stem ; they are more numerous in the endo-
derm of the tentacles and of the free portion of the polyp.
The muscle-fibres (except those forming the retractors and
protractors) have a circular direction, and are similar to
those of Aleyonium. In Alcyonium, however, the endo-
derm cells of the tentacles do not possess muscle-fibres (Hick-
son, 1895, p. 376).

In teased preparations the muscle-fibres of the ordinary
endoderm cells may be clearly seen (fig. 26). On looking at
the flagella-bearing cells in the same preparations, it 1s seen
that most of these cells bear at their inner ends two processes
which appear to be less stiff than the muscle-fibres of the
ordinary endoderm cells, and which are never in a straight line
with each other, but are invariably bent more or less towards
each other (fig. 27). It is possible that these are the modified
muscle-fibres of the cell, as they appear to be homogeneous,
and, when treated with fuchsin or iron hematoxylin, stain
similarly to, though rather less deeply than, the muscle pro-
cesses of ordinary endoderm cells. The bending inwards of the
two processes from the inner end of the cell causes them to
become rather more deeply embedded in the mesoglea than
the muscle processes of the ordinary endoderm cells, and the
cell is therefore provided with a firmly fixed base upon which
the giant flagellum can work as on a fulcrum.

It is worthy of note that throughout the whole of the
colony the muscle processes of the endoderm cells (except the
protractor and retractor muscles on the mesenteries) are
circular in direction, whereas the similar processes (where
present) of the ectoderm cells are longitudinal in direction,

THE STRUCTURE OF XENIA HICKSONI. 267

Zooxauthelle are exceedingly numerous in the endoderm of
the pinnules, so numerous that, in many cases, the lumina of
the pinnules are entirely closed. They are also numerous in
the endoderm of the tentacles and of the free portion of the
polyp, but in the ccelentera of the stem there are few, except
near the upper end. There do not appear to be any large gaps
between the endoderm cells in the lower parts of the ccelentera,
as described by Hickson in Aleyonium, but the endoderm
cells in this part are more or less spherical in shape, and are
only loosely connected together.

Mesoglea.

1. Of the Free Portion of the Polyp.—The mesoglea
of the body of the polyp varies in thickness from 02 mm.
to ‘06 mm., while that of the tentacles is much thinner,
averaging ‘013mm. The mesoglea of the pinnules is exceed-
ingly thin, especially when they are expanded (cf. figs. 11,17).

Cells which connect the ectoderm and endoderm may be
seen crossing the mesogloza in all parts of the tentacles and
body of the polyps. In the tentacles, however, these cells
are few in number, but in the body of the polyp, where the
mesogloea is thicker, the cells are more numerous. They are
usually elongated or fusiform cells, having their outer
ends embedded in or connected with the ectoderm, and their
tapering inner ends passing into the endoderm (fig. 17). In
most cases the connection between the two cell-layers is estab-
lished by means of a single cell, but in some cases two or more
cells are placed end to end to form the connecting cord. In
most cases the larger portion of the cell and its nucleus are
situated in the ectoderm, and this, together with the nature of
the cell,which closely resembles an ectoderm cell in appearance,
points to the fact that these cells in the mesoglea are derived
from the ectoderm. Besides these moderately large cells, the
protoplasm of which 1s finely granular and often contains a
few small vacuoles, there are other cells of very much smaller
size which bear several or many processes. Some of the cells

268 J. H. ASHWORTH.

lie close to the ectoderm, while others are near the endoderm.
They are connected together by their exceedingly slender pro-
cesses, which traverse the mesogloea and unite with each other.
These cells resemble, and are probably homologous with, the
nerve-cells and nerve-fibres of Aleyonium and the Actinia.
They will be further described below (see p. 277). The
mesoglea of the mesenteries and its included cells have already
been described (see p. 263).

2. Of the Stem (see Pl. 25, figs. 8 and 9).—On examining
transverse sections of the upper portion of the stem there is
seen a slightly denser ring of mesoglea (Mg. D.) around each
of the coelentera. This ring of denser mesoglea is itself
moderately free from cells, being crossed only at intervals by a
cell or thin cord of cells, but it is bordered by an almost com-
plete cordon of cells, interrupted only for the passage of
endodermic canals. The canals of the stem are moderately
.arge and very numerous, being much more highly developed
than those of Aleyonium. The canals may be divided into
the two systems described below.

The Superficial Canal System.—This system of canals
is formed by numerous endodermic canals (Sup. Can., figs. 8
and 9) which are situated in the stem about ‘1 mm. beneath
the ectoderm. This system is really a fine network of
numerous canals, which have a similar structure and appear-
ance in all parts of the colony. The canals are about ‘08 mm.
in diameter. In the intervals between these canals there are
usually cords of ectoderm cells (Hct. Str.) which pass from
the ectoderm to cells in the deeper parts of the mesoglea. In
many cases the superficial endodermic canals are themselves
closely connected with the ectoderm by strands of cells passing
across the mesoglea from the ectoderm to the outer walls of
the canal lying beneath. From the inner wall of these canals
cords of cells frequently pass inward into the mesogloea, and
are connected with other cells, with the colentera, or with
longitudinal canals.

The superficial canals are also present on the convex summit
of the stem (see fig. 8), and form there a plexus of canals,

THE STRUCTURE OF XENIA HICKSONI. 269

with similar relations to those above described in the cylin-
drical portion of the stem. In this portion of the stem the
canals embrace or pass round each celenteron at a distance of
about ‘2 mm., and communicate by means of branches with
the celenteron and with the neighbouring longitudinal canals.
The superficial canals on the convex summit are continuous
at the edge of the summit with the corresponding canals of
the cylindrical portion of the stem. In the stem the super-
ficial canals frequently communicate with the neighbouring
colentera and longitudinal canals. Thus, by means of branch
canals, this superficial system of canals is placed in commu-
nication with the remaining cavities lined by endoderm, and
by means of strands of cells is placed in communication
with the ectoderm and with the neighbouring cells in the
mesogloa.

Where a stem divides there are extra canals which establish
thorough communication between the superficial canals of the
two branches.

As the base of the colony is approached there is a tendency
for the superficial canals to send inwards wide branches, and
in the lowest 2 mm. of the stem such branches are given off
in large numbers, and unite with the complicated anastomosis
of longitudinal canals present in that part of the stem.

This system of canals is of great importance, as all the young
buds produced in the colony are formed by enlargement and
growth outwards and inwards of one of these canals, the endo-
derm and lumen of the canal forming respectively the endoderm
and ceelenteron of the young polyp (see also p. 291).

Histology of the Superficial Endodermic Canals.
—The endoderm lining the cavity of the superficial canals
is always much thicker on the outer side of the canal than on
the inner side (Pl. 27, fig. 29). This is caused by the cells on
the outer side being columnar and longer than the cubical or
slightly flattened cells of the inner side of the canal.

The cells lining these canals resemble the endoderm cells of
the coelentera, but their protoplasm is somewhat less vacuo-
lated, and therefore does not so markedly present the reticulate

270 J. H. ASHWORTH.

appearance which is so usual in the endoderm of the ccelentera.
None of the cells of these canals bear flagella. Among the
bases of the cells there are small cells which are probably stages
in the formation of the larger ones. Some of the cells of the
canals appear to be provided with very slender muscle pro-
cesses. There are numerous zooxanthelle in the lumen of
the canals and embedded in the endoderm lining the cavity.

The Internal Canal System.—The canals forming the
main portion of this system are chiefly longitudinal in direc-
tion, and commence in the umbrella-shaped portion at the
top of each stem. ach canal runs in a sinuous or zigzag
course in the mesoglea, about equidistant from the surround-
ing celentera.

The longitudinal canal communicates with the superficial
canals lying around its origin (fig. 8). During its course
down the stem the longitudinal canal very frequently com-
municates by small transverse canals with the neighbouring
celentera and canals, and the longitudinal canals in the outer
portion of the stem communicate also with the superficial
canals. Owing to the frequent occurrence of branches, the
longitudinal canals are nearly always angular in transverse
section, one (or more) of the angles being usually produced
into a small branch canal. The branch canals vary greatly in
size; in the upper and middle portions of the colony being
small, and their lumen very small or obliterated altogether,
while near the base of the colony the branches are almost as
large as the main canal. At the base of the colony these longi-
tudinal canals give off several rather larger lateral branches on
all sides, some of which unite with similar branches from
adjacent canals, while others open into neighbouring ccelentera.
Owing to the presence of so many canals in this region of the
stem, the mesogloea is penetrated in all directions by a com-
plicated network of canals, which place all the cavities
lined by endoderm in intimate communication with each other.
Very close to the base of attachment of the colony, a canal
may usually be seen passing from each side of the lowest
portion of each primary ccelenteron, so that in longitudinal

THE STRUCTURE OF XENIA HICKSONI. PMN

section the celenteron and its two canals appear |-shaped.
The canal opens into the base of a neighbouring ccelenteron.
These canals are probably the representatives of the original
stolon from which all the primary ccelentera grew out.

The well-developed longitudinal canals running parallel to
and between the ccelentera of the polyps of Xenia remind
one of the ccenenchymal tubes of Heliopora ccerulea,
described by Moseley (1881) and by Bourne (1895). The
resemblance is more striking when we consider that in both
cases the longitudinal tubes are lined by endoderm, and are
connected near the upper surface of the colony with a network
of superficial endodermic canals.

The differences between the canals of Xenia and the coenen-
chymal tubes of Heliopora are chiefly due to the fact that in
the latter only the outer portion of the coral is living, the
internal parts consisting of calcareous skeleton only, whereas
in Xenia the whole colony is penetrated by living cells. In
Heliopora, therefore, the ccclentera of the polyps and all canals
running into the colony must terminate within about 2 mm.
of the surface, as this is the lowest limit of the living substance.
The ccenenchymal tubes of Heliopora have an exactly similar
course to the longitudinal canals of Xenia, as they run parallel
to the celentera from their point of origin from the superficial
canal system (just beneath the ectoderm covering the
free surface) to the base of the living portion of the colony,
where they and the cclentera terminate blindly.

Moseley (1881) believed the ccenenchymal tubes of Helio-
pora to be degenerate siphonozooids from which the mesen-
teries had disappeared, but Bourne (1895) has shown that
they cannot be so regarded.

In order to find a parallel to these coenenchymal tubes of
Heliopora, Bourne thought it necessary to go outside the
Alcyonaria and compare the tubes with certain longitudinal
canals of Millepora, described by Moseley (1876). After
carefully studying the canal system of Xenia, it appears to me
that this course is not now necessary, as the comparisons insti-
tuted above between the coenenchymal tubes of Heliopora

272 J. H. ASHWORTH.

and the longitudinal canals of Xenia are perfectly justifiable.
It is true the coenenchymal tubes of Heliopora are more
numerous than the longitudinal canals of Xenia, but this
also is probably due to the different modes of growth of the
two corals. In Heliopora the living part is, as it were,
spread out in a thin film, and the polyps are a considerable
distance (1mm. to 2mm.) apart, several coenenchymal tubes
being therefore required to place the coelentera in communica-
tion with the intervening ectoderm, calicoblasts (or skeleton-
forming cells), and mesoglea. In Xenia each stem is an
elongated, cylindrical, compact mass of living tissues, and the
polyps are much closer together, the ccelentera being only
about ‘25 mm. to ‘4 mm. apart in the stem, and therefore one
longitudinal canal, situated in the mesogloea between adjacent
celentera (together with the auxiliary strands and cords of
cells which are so well developed in this Xenia), is sufficient
to provide efficient communication between the ceelentera and
all parts of the narrow mesogleal column between them.

Histology of the Longitudinal Endodermic Canals.
—These canals are lined by a single layer of cells of equal
thickness on all sides (cf. the superficial canals). The cells
are more or less cubical in shape, and closely resemble the
endoderm cells lining the cclentera. None of the cells in
the canals bear flagella, but some bear slender muscle pro-
~ cesses which, like those of the endoderm of the ccelentera, are
chiefly circular in direction. In the canals near the base of
the colony many nematocysts, each in its cnidoblast cell, may
be seen lying among the endoderm cells. In the upper por-
tions of the colony nematocysts are seldom seen in the canals.
Zooxanthelle are present only in those longitudinal canals
which are situated in the circumferential portions of the stem
and in the upper portion of the canals, where they approach
the summit of the stem.

Cells in the Mesogloa.—On examining a transverse
section of the upper end of a stem of the colony, an almost
complete chain of cells is seen surrounding the denser ring of
mesoglea round each ccelenteron (fig. 9, Ect. Ch.). Hach

THE STRUCTURE OF XENIA HICKSONI. 273

ring of cells is situated at a distance of about (06 mm.—‘07
mm. from the endoderm of the ccelenteron within. These
cells are the ectoderm of the portion of the polyp enclosed in
the stem, as can be seen on examining a longitudinal section
through the upper part of the stem (see fig. 8), when this
cylinder of cells is seen to be continuous and in line with the
ectoderm of the free portion of the polyp. That these cells
are ectodermic is further shown by the fact that spicules and-
nematocysts are found in them in moderate numbers, especially
in the upper portion of the stem (figs. 8 and 9, Sp. Nem.).
Adjacent cylinders of cells are placed in intimate communica-
tion by numerous cords of cells which traverse the mesoglea
between them.

Sections taken further down the stem show that the ring of
cells becomes less complete and less definite, and each ring
widens and recedes into the mesoglea a little further from its
celenuteron. As they recede further and further, the rings of
cells often coalesce in the mesogloea at a point almost equi-
distant from their coelentera ; and, as the longitudinal canals
also lie in this region, the cells come to lie in relation with,
and form a plexus around, the canals. There is, therefore,
still a cylinder of cells round each celenteron, but the cylinder
is not quite so regular as in the upper portions of the stem,
being interrupted at frequent intervals, and is further away
(15 mm. to *2 mm.) from the enclosed cclenteron,.

In the upper portion of the stem the cylinder of cells has
the same relation to the endoderm of the celenteron within it,
as have the ectoderm and endoderm of the free portion of the
polyp to each other. Crossing the denser ring of mesoglea
between the cylinder of cells and the endoderm are cells (viz.
the vacuolated and granular cells, and the small nerve-cells)
exactly like those observed in the mesogloea of the free portion
of the polyp.

Lower down the stem the cylinder of cells has been pressed
further away from its ceelenteron, probably by the later growth
of the mesoglea, and is interrupted at intervals for the pas-
sage of canals and cords of cells which place all parts of the

274 J. H. ASHWORTH.

mesoglea in intimate communication with the celentera, the
canals, and the external ectoderm. Bourne (1895) has shown
that in Xenia umbellata these rings of cells are ectodermic
on account of their connection with, and general resemblance
to, the ectoderm of the free part of the polyp, and the occur-
rence of spicules and nematocysts in some of them.

The cords of cells in the mesogloea are then chiefly ecto-
dermic, as they arise from the cylinders of cells around the
coelenteron, or, in the outer portion of the mesoglea, migrate
inwards from the inner irregular surface of the external ecto-
derm. The cells all present a similar appearance, being either
rounded or rather elongated in shape, with somewhat vacuo-
lated protoplasm. The elongated cells frequently taper at
their ends into long slender processes which become connected
with similar processes of adjacent cells.

Some of the cords of cells are, however, obviously formed
by obliteration of the lumen of a small canal ; these cells are,
of course, endoderm.

Spermatogenesis (Pl. 27, figs. 30—35).

The specimen is a male, and shows beautifully all the stages
in the development of the spermatozoa.

Gonads are most numerous in the upper portion of the stems
of the colony, but many sperm sacs are found in the basal
part of the free portion of the polyps, and a few also in the
lower portions of the colony. Sections through the upper
portion of the stems show that sperm sacs are so numerous
that they practically fill up the cavity of the colentera of
several of the older polyps (Pl. 25, fig. 8, S. S.). In these
cases most of the sperm sacs are uo longer spherical, but by
mutual pressure have become angular, being usually penta-
gonal or hexagonal in section.

The genital cells are derived from the cells which lie in the
mesogloea of the mesenteries near their inner or free edge.
These cells, as shown above (see p. 263), have migrated to
their present position in the mesoglea from the endoderm, so

THE STRUCTURE OF XENIA HICKSONI. 275

that the gonads are, in this as in other Alcyonaria, endo-
dermic in origin.

Each sperm sac originates as a slight projection at the side
or free edge (except in the dorsal mesenteries) of the mesen-
tery, and consists of one of these genital cells covered by a
thin sheet of mesoglea, and by a single layer of endoderm cells
continuous with the endoderm of the sides of the mesentery.
The genital cell, the protoplasm of which is finely granular,
or sometimes contains small vacuoles, is spherical and about
‘01 mm. in diameter, and in the centre has a large spherical
nucleus whose diameter is about half that of the cell.

The nucleus and protoplasm of the genital cell undergo
division, which at first is apparently regular, as many cases of
four or eight cells so produced may be seen (fig. 30). The
divisions of the genital cell are accompanied by divisions of
the endoderm cells covering it, and very soon the increase in
size of the sperm sac causes it to project as a spherical or oval
body from the mesentery, to which it always remains attached
by a stalk consisting of a thin cord of mesogloea surrounded
by endoderm. When the sperm sac has reached a diameter of
about ‘06 mm. to ‘(08 mm. there appears in the centre a small
cavity, free from nuclei, but containing some coagulable sub-
stance (fig. 31, Cy.). This central cavity continues to enlarge
for some time, along with the growth of the sperm sac (fig. 32),
and attains its maximum size in sacs about °2 mm. to °25 mm.
in diameter, in which the central cavity reaches a diameter
about one fourth that of the sperm sac. After reaching this
size it is gradually encroached upon by the heads of the
ripening spermatozoa (fig. 33, Spz.), and in the fully developed
sperm sac, which is about ‘35 mm: in diameter, the cavity has
completely disappeared, the whole of the interior of the sperm
sac being filled with a mass of somewhat loosely packed
spermatozoa, many hundreds in number, produced by the con-
tinued division of the single primitive genital cell. There are
no examples of karyokinesis in the many hundreds of sperm
sacs I have examined.

The head of each ripe spermatozoon (fig. 34) consists

276 J. H. ASHWORTH.

of a blunt, conical, anterior piece fixed to the spherical
nucleus. The length of the head is 7 y, the nuclear portion
being 4 in diameter. The tail is a slender filament about
27 mw long.

The spermatozoa closely resemble those of Aleyonium in
their structure and development (Hickson, 1895). ‘lhe endo-
derm covering the sperm sac appears to undergo certain
changes as the sperm sac grows, and in thin sections from two
to five nuclei may be counted in many of the endoderm cells.
One of these nuclei is sometimes larger than the others. In
the left upper cell of fig. 35 the large nucleus near the centre
occupies the position of the original nucleus of the cell. The
other four nuclei have probably been produced from it by
division, but there has been no corresponding division of the
protoplasm. The cell with four nuclei which was figured by
Hickson (1895, pl. 39, fig. 45, f.) from the teased prepara-
tions of the sperm sacs of Alcyonium, was probably one of
the cells of the endodermic follicle.

At first it appeared that the sperm sacs were situated on
ventral and lateral mesenteries only, although, as pointed out
above (see p. 263), the cells in the mesogleea, from which the
genital cells are derived, are found in the dorsal mesenteries
as well. After examining a large number of sections, I have
found only two clear cases of sperm sacs occurring on the
dorsal mesenteries. In each case the sperm sac is situated
on the side of the mesentery a little distance from the free
edge, so that, although the sac is of considerable size (15 mm,
in diameter), it does not push the dorsal mesenterial filament
out of position, and, judging from the sections, would not
impede its action.

Empty sperm sacs the walls of which are collapsing may be
seen in several sections. The spermatozoa are discharged into
the celenteron by bursting of the follicle of the sperm sac, and
are then swept out to the exterior through the stomodeum,
In sections of two polyps in which spermatozoa were escaping,
the spermatozoa are found along the dorsal side of the stomo-
deum, being doubtless driven out by the upward or outward

THE STRUCTURE OF XENIA HICKSONT. ef

current produced by the cilia of the dorsal mesenterial filaments.
Although escaping in a mass they are not enclosed in the
follicle, but lie loosely aggregated in the dorsal portion of the
stomodeum. These two examples support the conclusion
expressed above that the spermatozoa are discharged into the
ceelenteron by rupture of the follicle, the collapsed remains of
which retain for some time their attachment to the mesentery.

As spermatozoa are present in all stages of development, it
is likely that the discharge of ripe spermatozoa continues over
a considerable period,—in fact, probably throughout the year.
This is perhaps due to the fact that, living on reefs in the
shallow waters of tropical seas, this coral is not subject to any
great variations in temperature and food supply. In this
respect Xenia differs from Alcyonium digitatum, which
occurs in the colder seas of Northern Europe. In the latter
all the sperm sacs of a colony have reached a similar stage
of development and are all ripe about the same time of the
year, viz. December, and therefore the discharge of ripe
spermatozoa occurs only over a limited period, probably over
about a month (Hickson, 1895).

Nervous System.

In several sections there is a plexus of fine fibrils in the
mesogloea connected with very small cells in relation with
the ectoderm and endoderm. This plexus appears to be
homologous with the similar plexus described by Hickson in
Alcyonium (1895, p. 371), and compared by him to the
**Nervenschicht ” of the Actinie.

In this Xenia the plexus is best seen in sections of a polyp
in which the ectoderm is cut slightly obliquely. On examin-
ing in such a section (PI. 26, fig. 16) the part where the ecto-
derm passes into the mesogloea, very fine fibrils (N. F.) may be
seen, forming an open network, upon which cells (N. C.) are
situated at intervals. The fibrils can be best seen in the
mesogloea, but can be traced close to the ectoderm and endo-
derm. The cells of the nervous system are exceedingly small,

voL. 42, PART 3.—NEW SERIES, .

278 J. H. ASHWORTH.

and usually fusiform, triradiate, or stellate in shape, the angles
of the cell being produced into nerve fibrils.

On tracing the fibrils outwards from the mesogleea into the
ectoderm, they are seen to be in connection with small cells
which are situated in the deeper part of the ectoderm. Owing
to the irregularity of the inner face of the ectoderm, and to
the presence of spicules in the portion of the layer where the
nerve-cells are situated, it is not possible to obtain a section
which shows the ectodermic nerve plexus clearly, but small
portions of it may be seen where the spicules are slightly less
numerous.

In the case of the endodermic nerve plexus this difficulty
does not exist, and an oblique section through the wall of a
polyp shows that the plexus of fibrils in the mesogloea is con-
nected with minute stellate cells situated upon the outer face
of the muscle processes of the endoderm cells. Hematoxylin
(especially Heidenhain’s iron hematoxylin and Meyer’s acid
heemalum) stains the nerve-cells and fibres most clearly.

History of our Knowledge of the Buds of the
Xeniide.

Besides the fully developed polyps, the description and
measurements of which are given above, there are on most of
the branches of the colony young polyps or buds in various
stages of development. These buds are invariably found at
the edge of the umbrella-shaped area at the end of the stem.
On one stem (fig. 1, left) there are ten small buds varying in
length from +5 mm. to 3 mm., and about fifty larger polyps
from 5 mm. to 10 mm. in length. The smallest buds have
simple tentacles devoid of pinnules, and all stages between
these and the adult polyps may be found. As the polyps (both
young and old) of Xenia are non-retractile, this genus offers
considerable advantages for the study of the development of
the polyps, and the young polyps have been noticed by many
observers.

Quoy and Gaimard (1838) first observed these small polyps in

THE STRUCTURE OF XENIA HICKSONTI. 279

the Xeniidz. They noticed them in Cornularia viridis,
which appears to belong to the genus Xenia, and they
suggested that these small polyps, the tentacles of which were
devoid of pinnules, were young forms which had not yet
attained the adult characters.

In 1874 Kolliker described Heteroxenia Elizabethe, in
which small individuals are very numerous. He regarded
these small individuals as being of two kinds, some being young
polyps in various stages of development, others being “ zooids.”
Heteroxenia, therefore, is dimorphic. According to K6l-
hiker, there are several essential differences between these two
kinds of individuals.

I. The Polyps.—These are of large size, the adults measur-
ing 20 mm. to 55 mm. in length, and about 3 mm. in breadth.
There are also obviously younger polyps, 5 mm. to 20 mm.
long, all of which are situated round the edge of the disc. The
tentacles of the adult polyps bear two series (in each of which
four rows are distinguishable at the base of the tentacles) of
long cylindrical pinnules, one on each side of the middle line of
the tentacle. The ccelentera of these polyps extend a consider-
able distance into the stem of the colony, and are crowded
with ova.

II. The Zooids.—These are much more numerous than
the polyps and much smaller, measuring only 8 mm. to 5 mm. in
length, and from ‘7 mm. to 1 mm. in breadth. The tentacles
are eight short simple lobes, -14 mm. to ‘2 mm. in length, and
they bear no pinnules. The celentera of these zooids extend
at most only 3 mm. into the stem, and contain no gonads.

It should be noted that Kélliker saw two kinds of small
individuals, and described the differences between them in size,
structure, and position, viz.: (1) the young polyps, 5 mm. to 20
mm. long., found only round the edge of the disc; and (2) the
“ zooids,” 3 mm. to 5 mm. long, found all over the disc among
the bases of the larger polyps.

Klunzinger (1877), who described the Xeniidz of the Red
Sea, saw small polyps in a specimen of Xenia umbellata;
he called them bud-like polyps, and said that their tentacles,

280 J. H. ASHWORTH.

which are at first simple, very soon show indentations or
pinnules. He regarded such individuals rather as young
polyps than as zooids. They were more numerous in the
outer part of the arched end of the stem. In a new species,
Xenia fuscescens, Klunzinger described the small polyps as
very numerous, outnumbering the large polyps, and filling up
the intervals between the bases of the latter. He wrote that
these small individuals do not appear to develop into fully
formed polyps, but to remain in the bud-like stage, with short,
simple, mostly incurled tentacles. They are 1 mm. to 2 mm.
long and *5 mm. broad, and are cylindrical or club-shaped.
On account of the large numbers of these small indivi-
duals, Klunzinger placed his Xenia fuscescens near the
Heteroxenia Elizabethe of Kolliker. There are no
transition stages between these small individuals which do not
appear to develop into Jarger polyps and the adult polyps, and
from the description and figures they appear to be quite as
distinct as the two kinds of individuals described by Kolliker
in Heteroxenia.

Haacke (1887), who examined some of the Xeniidz in
Torres Straits, says that the small individuals are merely
young polyps, and all stages of development between them and
the adult polyps may be met with. Therefore he denies the
occurrence of heteromorphism in the Xeniide.

Wright and Studer in the ‘Challenger Report’ (1889)
record the observations of Klunzinger and Haacke noticed
above. They agree with Haacke, and therefore propose the
provisional abandonment of Kolliker’s genus Heteroxenia.

Bourne (1895) observed in his new species, Xenia Garcia,
numerous imperfect polyps or buds in all stages of growth at
the edge of the polyp-bearing summits of the stems. He re-
marked that “these are not siphonozooids, but stunted or
developing polyps.”

Bourne also described a specimen which he referred pro-
visionally (being unable to procure Kolliker’s original descrip-
tion) to the species Heteroxenia Elizabethe. In his
description were noted—

THE STRUCTURE OF XENIA HIOKSONI. 281

(1) The larger polyps with well-developed tentacles, with
three rows of lateral pinnules on their margins, their coelentera
continued to the bottom of the stem or nearly so, and filled
with ova. At the edges of the arched end of the stem there
were numerous young polyps in all stages of development,
many of which showed distinct pinnules on their tentacles.

(2) Closely applied sterile zooids which have no tentacles,
but only eight radiate lobes round the mouth. The ccelentera
of these individuals extend only a little way into the stem, and
then communicate with the ceelentera of the polyps by anasto-
mosing endodermic canals. Among these zooids there are
never any individuals which show signs of pinnate tentacles nor
which contain gonads. Bourne concludes that in this form
there is distinct dimorphism.

Schenk (1896) mentions the occurrence of buds in eight
new species of Xenia from Ternate, which he has described.
He briefly describes the external characters of the buds in
Xenia viridis, and figures three stages of development.
The first figure represents a small bud about 3 mm. long, in
which the tentacles are simple finger-shaped lobes. The other
two are drawings of slightly larger polyps the tentacles
of which bear, in one case six, and in the other about twelve
pinnules on each side of the tentacle (seen from the outer
side). In all the examples mentioned by Schenk the small
individuals are young polyps in course of development, as
pinnules have already appeared on the tentacles of many of
them. Schenk (loc. cit., p. 53) remarks that he believes the
zooids of Kolliker are merely young polyps, and therefore
Heteroxenia Hlizabethe does not exhibit dimorphism ;
hence he renames it Xenia Hlizabethe, and places it near
X. fuscescens, Klunzinger. Thus Schenk supports Haacke’s
view that there is no dimorphism of the polyps of Xenia.

From this short account of previous observations it will be
seen that the nature of these small individuals in the X eniidz
is not yet determined. That they are all buds or young
polyps is strongly affirmed by Klunzinger, Haacke, Wright
and Studer, and Schenk; while the opinion of Kdlliker and

282 J. H. ASHWORTH.

Bourne is that they are of two kinds, some being young polyps
and others quite different individuals, termed zooids.

With a view to coming to some definite conclusion regarding
the nature of these small individuals, I have examined them
in the specimen of Xenia Hicksoni in great detail. Ihave
also examined the sixteen other specimens of Xenia at my
disposal, and a specimen of the Heteroxenia which Bourne
has described and figured. These will be briefly described
below.

External Characters of the Buds of Xenia Hicksoni.

In Xenia Hicksoni the small individuals are all buds or
young polyps, as in this one specimen every stage in develop-
ment may be seen, from the youngest polyp only ‘32 mm. in
length to the large adult polyp 12 mm. long; and the series is
perfectly complete, there being no break at any point which
would justify the division of the individuals into two kinds.

The buds are invariably found on the edge of the arched
end of the stem. Their numbers vary on different stems,
there being usually from six to ten buds less than 3 mm. long
on each stem (Pl. 23, fig. 1, left). The smallest bud (Pl. 24,
figs. 4, 4a) measures *32 mm. in length and ‘44 mm. in width.
It is situated just under the edge of the umbellate summit of
the stem. It isa short cylindrical outgrowth, at the distal end
of which the developing tentacles are indicated as eight small
rounded lobes about *] mm. long, divided from each other by
shallow furrows. <A slight depression in the centre of the distal
end indicates the position of the future mouth, which is not yet
open to the exterior. The tentacles increase in length rather
more rapidly proportionately than the body of the polyp. In
the smallest specimen they are less than one third the total
length of the polyp, but in rather larger polyps they form from
two fifths to one half the total length of the polyp. The
tentacles remain simple lobes until the polyps attain a length
of nearly 1 mm. In a specimen ‘95 mm. long (fig. 5) the

THE STRUCTURE OF XENIA HIOCKSONI. 283

tentacles have reached a length of ‘34 mm., and the tip of each
tentacle is distinctly trilobed when seen from the outer side,
i.e. there is an indication of the formation of the first two
pinnules, one on each side of the axis of the tentacle.

From this point onwards the formation of pinnules takes
place regularly as the tentacles increase in size, as may be
seen from the table below. When the polyp (fig. 6) has
reached a length of 1°42 mm. the tentacles, which are ‘6 mm.
long, are seen from the outer aspect to bear four pinnules on
each side. If one of the tentacles be examined from the inner
side it will be seen that an inner row of pinnules has already
been formed (fig. 6a). The polyp from which fig. 7 was
drawn was 2°27 mm. long, and its tentacles 1:30 mm. long.
Seen from the outer aspect the tentacles show eight pinnules
on each side. For a distance of about two thirds of a milli-
metre from the tip of the tentacle, there are on the inner face
two rows of pinnules on each side of the middle line, but a
little nearer the base of the tentacles the three rows of pinnules
on each side, characteristic of the tentacles of the adult, may
be seen (fig. 7).

After the earliest stages of growth are passed the polyps
grow quite regularly, and the length of the tentacles bears to
the total length of the polyp an almost constant proportion.
From the appended table of measurements it will be seen the
tentacles form rather more than two fifths of the whole length
of the polyp. With the increase in length of the tentacles
there is a corresponding increase in the number of pinnules
which the tentacles bear, and, as the following table shows,
there is a perfect series of examples, beginning with the very
young specimens, in which the tentacles are devoid of pinnules,
and ending with the largest polyps, whose tentacles show from
the outer aspect nineteen or twenty pinnules on each side of
the middle line. The gradual transition from the youngest to
the oldest polyps shows that the small individuals in this
colony are undoubtedly young buds in various stages of
development. They are all “‘ polyps,” using the word in the
sense in which it was used by Kolliker.

284, J. H. ASHWORTH.

— TT

Total length Width _ Number of
Reference] of Polyp of Length of | Pinnules on each
Number. (Body and Polyp. Tentacles. | side of Tentacle
lentacles), (outer aspect).
mm. mm mm.
1G "32 "44 | ‘10 0 Pl. 24, figs. 4, 4a.
1B "43 eke 0 In section, Pl. 27, fig. 28.
ik "56 ron ai) 0 —
NG 8 “4, 31 =| trace of first —
Ae oo "35 34 | Trilobed =1} Pl. 24, fig. 5.
VES 60 34 ‘43 | Trilobed = 1] In section, Pl. 25, fig. 8.
BOS kak 5 hey | 2 —
WHE), 1-49 87 Oa 4 Pl. 24, figs. 6, 6a.
DX) 6 “960K iF 4—5 a
X.| 2:0 "86 86 | 6 =
Ee sek: 8 1-03 7 —
MUMS Var 7) 26 i era. ed 8 Pl. 24, figs. 7, 7A.
CUM ORE MIE: || TR 8 ~
XIV. 3°2 AS eo TP el 8—9 aaa
OV le de@ 10 18 10
VE. | =621 “y 24 10—11 =
XVEE | 26:5 + 2°5 11—12 =
VIII || PGS. | 1G eee 13 =
|| OE | PMs ie ea 14—15 =
XX.| 190 |19 | 51 17 =
XXL) 165 [12 57 | 1920 | PL 94, fig. 9.

I find in all the sixteen other specimens of Xenia at my
disposal, young buds in all stages of growth, similar to
those described and figured in Xenia Hicksoni. Most of
these young buds occur on the arched end of the stem, but
in two of the colonies examined a few buds are found in
the middle portion of the summit between the bases of the
larger polyps. These buds, however, differ in no way from
those round the edge of the summit.

External Characters of Heteroxenia Elizabethzx
(Pl 27, tig soc:

Bourne (1895) has already described the external characters
of a colony similar to this one, but a brief description will be
given here. The colony is somewhat triangular in shape, the
base of the triangle being formed by the polyp-bearing summit
of the stem. The stem is slightly flattened, and is narrow

THE STRUCTURE OF XENIA HICKSONI. 285

at its lower end, gradually widening from below upwards.
The total length of the stem is about 30 mm., its breadth at the
base 12 mm. by 6°5 mm., its breadth at the top 23 mm. by 6mm.

The non-retractile polyps (A.) are long and slender, and
measure on an average about 15 mm. in length (including the
tentacles), though a few specimens reach almost 30 mm. in
length. The polyps are 1 mm. to2 mm. broad. The tentacles
of the polyps measure 4 mm. to 5 mm. in length, and bear on
the inner side two series of rather slender pinnules, each
series consisting of three rows of about sixteen to twenty-four
pinnules in each row. In one or two of the tentacles examined,
the pinnules at the base are so arranged that it is difficult to
say whether they are in three or four rows on each side of the
middle line. The pinnules in the middle of the tentacle
measure about ‘5 mm. in length, and are about four times as
long as they are broad. The middle line of the tentacle is free
from pinnules along the greater part of its length. The polyps
are 1 mm. to 3 mm. apart, but they are closer together round
the edge of the summit of the stem than on the middle portion
of the summit. Around the edge of the summit of the stem
there are many small polyps in all stages of development,—in
fact, all the polyps near the edge of the summit are small ones,
the large ones being situated in the middle portion of the
polyp-bearing area.

The walls of the polyps are thin, and when the colony is
removed from spirit the polyps fall together and form a con-
fused mass on the end of the stem.

Between the bases of the polyps there are many closely
apposed small zooids (fig. 37, S.), quite different in appearance
from buds or young polyps. These siphonozooids are from six
to ten times as numerous as the polyps or autozooids, and they
occur all over the summit of the stem. The siphonozooids are
all similar in appearance, being cylindrical or club-shaped,
2 mm. to 5mm. in length and *5 mm. to 1 mm. in diameter.
The tentacles are eight small rounded lobes arranged round
the mouth; they are never longer than ‘25 mm., and never
possess pinnules.

286 J. H. ASHWORTH.

Bourne provisionally referred this specimen to Kolliker’s
species Heteroxenia Elizabethe, and, after comparing the
specimen with Kolliker’s original account and figures (1874),
I can confirm his conclusion. This specimen is much smaller
than Kolliker’s, its polyps being only about half the size, but
there are many points of agreement between them, viz. general
anatomy and proportion of parts, size, and structure of zooids;
size, colour, and distribution of spicules; canals, &c.

After carefully examining the two kinds of small individuals
(viz. young polyps and zooids), I have no doubt that they are
entirely different in nature. The external characters are quite
different, and there are important differences in their internal
arrangements, to which reference will be made.

External Characters ofthe Young Polyps of Hetero-
xenia Hlizabethz.—The youngest polyp, which is also by
far the smallest individual on the colony, is ‘64 mm. in length
(fig. 38). Its tentacles are finger-shaped lobes °24 mm. to ‘3
mm. in length. The tip of one of the tentacles is slightly
three-lobed, but all the other tentacles have simple rounded ends.

A polyp 1*4 mm. long, the tentacles of which have attained
a length of -4 mm., are trilobed at their ends, i.e. the first
pinnule on each side of the axis is being formed. The
formation of pinnules proceeds regularly with the growth in
length of the tentacles, and the adult size and characters are
gradually attained. From the appended table it will be seen
that there isa complete series of stages from the smallest polyp,
‘64 mm. long, with simple tentacles devoid of pinnules, to the
largest polyp, nearly 30 mm. long, the tentacles of which show
on the outer face twenty-four pinnules on each side of the
middle line. In this Heteroxenia the ratio between the length
of the tentacles and the total length of polyp is not quite as
constant as in the Xenia considered above. The tentacles
form about one fourth to one third the length of the whole

polyp.

THE STRUCTURE OF XENIA HICKSONI. 287

Measurements of the Autozooids of Heteroxenia

Elizabethe.
Helerente ene of iain pene of Pines ai oe
umber. Polyp. Polyp. entacles. pee cee
mm. | mm. mm,
AC 53 64 | 3 | *24—3 0—1 Pl. 27, fig. 38.
Ol #3 Vege 3 os “4 Trilobed = 1 —
NEON OF Olen ok ag) 3—4 A 3, fig. 37.
Ara: 28a | 9 1:0 5 —
ee 44 ‘7 | 13 7 A 5, fig. 37.
IASON = 58 leit We¢/ 8 =
/ oy (aime 6:0 Pak 2°2 12—13 —
AS8.. 9-1 2°0 3'1 15 —_
ACO. 12°8 12 3°5 15—16 —
ACHOM 140 24 43 16 —
AVI. 20°2 2°0 5:0 18 _
AW UOT 28:4 2°4 bed 24—25 —_—

External Characters of the Siphonozooids of
Heteroxenia Elizabethe.—The siphonozooids, like the
autozooids, are non-retractile. These vary in length from
18mm. to 5°0 mm., and are ‘5 mm. to 1:0 mm. wide. Their
tentacles are all in the same condition, and are mere lobes
from °2 mm. to °25 mm. in length, and never show any traces
of the formation of pinnules. It will be seen from the
appended table that there is no development of the tentacles
corresponding to the increase in length of the individuals, so
that the tentacles of a zooid 5 mm. in length are very
similar to those of a specimen less than half its size (compare

S 1 and 8 7 in table).

288 J. H. ASHWORTH.

Table of Measurements of the Siphonozooids of
Heteroxenia Hlizabethe.

|
Reference Total length Width of | Length of
Number. of Zooid. | Zooid. Tentacles,

=
~
3
i=]
i=}
=]

BORO MO-~3!

mm.
1) ae
oy) Ss
23 =
24, =
aw) | oS
25 =
9 Figs. 37, 39.
25 =

CoM Ook Ww

OCR OWE

SOoOmnNowoo!
je

AARNNARNN

By means of their rudimentary, short, rounded tentacles
the zooids may be readily distinguished from young polyps of
the same size, whose tentacles are longer, pointed, and bear
pinnules. Compare, for example, the young polyp A5 (see
table, p. 287) with the siphonozooidS 5. They are both about
the same length, but the tentacles of the former are 1°3 mm.
long, and, examined from outer aspect, show seven pinnules
on each side of the middle line, whereas the tentacles of the
siphonozooid are only *2 mm. long, and are simple rounded
lobes. Now if, as urged by Schenk and others, these two
individuals are both young polyps, how can the differences in
the size and character of their tentacles be accounted for?
If the zooids are stages in the development of polyps, it is
very difficult to account for so many being in the same
stage of development (as the tentacles of all the specimens
examined are in the same simple condition), when, on the
same colony, other individuals of the same length, or even
smaller, have already acquired some of the adult characters,
viz. the pinnate tentacles. On the colony drawn in fig. 37
there are over two hundred zooids, all of which are similar in
appearance, having simple round tentacles. The examples
indicated in the above table are chosen haphazard from this
large number, being arranged in order of length merely for
convenience of reference, If these were young polyps we should

THE STRUCTURE OF XENIA HICKSONI. 989

expect to find a more or less constant increase in the length
of, and alteration in the character of, the tentacles, with the
increase in length of the individuals; but this is not the case.
We should also expect them to resemble other individuals of
the same length on the colony which are undoubtedly young
polyps; but, as we have seen, they are very different.

It must be concluded, then, that these zooids are different
in nature from young polyps, and that there are in Hetero-
xenia two kinds of individuals, polyps and zooids, or, to use
Moseley’s terms, autozooids and siphonozooids.

The Internal Anatomy of Heteroxenia Elizabethe.

Owing to the very imperfect preservation of the specimen, it
is impossible to give a detailed account of the internal struc-
ture of the colony, and therefore attention will be chiefly
directed to the points in which Heteroxenia differs from the
Xenia described in detail in the former part of this paper.

Autozooids.—Spicules are numerous in the ectoderm of
all parts of the body of the autozooid, and are very numerous
in the ectoderm of the outer face of the tentacles. They are also
present in the pinnules. They are similar in size and shape to
those of Xenia Hicksoni. They are whitish or bluish white
by reflected light, but of a reddish-brown tinge by transmitted
light. Nematocysts are exceedingly scarce and difficult to find.
They measure 9 p in length and 23 u in width. The ecto-
dermic muscles of the tentacles are well developed ; the muscle
band of the oral face is very much thicker than that of the
aboral face.

The oral disc of each polyp is slightly contracted, producing
a funnel-shaped depression leading to the mouth. The stomo-
dzeum is only 1 mm. to 15 mm. long, and is wrinkled or folded
transversely. It bears a ventral groove or siphonoglyph, the
cells of the lower half of which bear moderately long flagella.
There are the usual eight mesenteries, bearing rather feebly
developed retractor muscles on their ventral faces. Only the

290 J. H. ASHWORTH.

dorsal mesenteries bear mesenterial filaments, and these have
a typical course and structure. Owing to imperfect preserva-
tion it is impossible to say anything definite about the cells of
the other mesenteries, but no ventral or lateral mesenterial
filaments are visible.

The ccelentera of the polyps may be traced a considerable
distance into the stem, those of the primary polyps being con-
tinued down to the base of the colony. In the upper portion
of the stem the ccelentera contain many ova, each of which is
surrounded by an endodermic follicle attached to the mesen-
teries. The largest of these ova measure ‘3 mm. to ‘4 mm. in
diameter.

Siphonozooids (fig. 39).—Spicules are not so numerous
in the ectoderm of the zooids as in that of the polyps.

The stomodzum is very badly preserved in most of the
specimens. A few of the zooids situated near the edge of the
summit, however, are rather better preserved, and two of
these (S 3 and 8 7 in the table, p. 288) have been sectioned and
the stomodzum examined. In the zooid 2°3 mm. long the
stomodeum measures ‘6 mm., and shows a_ well-marked
siphonoglyph, the cells of the lower ‘2 mm. of which bear
flagella. The other specimen (S 7 in table, sce also fig. 39)
is5mm. long. Its stomodzum measures *8 mm. long, and the
cells of the lower ‘4 mm. of the siphonoglyph bear flagella.

Thus both autozooids and siphonozooids of this specimen
possess a siphonoglyph, the cells of the lower half or third of
which bear flagella. This does not agree with the observations
of Hickson (1883, p. 696) on a specimen of Heteroxenia, in
the siphonozooids of which he found a well-marked siphono-
glyph, but in the autozooids a complete absence of siphono-
glyph.

The siphonozooids possess the usual eight mesenteries, but
they are extremely thin, and retractor muscles are not visible
upon them. The dorsal mesenteries bear mesenterial fila-
ments (fig. 39, D. M. F.), which run in a sinuous course down
the dorsal side of the polyp, and a short distance into the
portion of the ceelenteron contained in the stem. The ccelen-

THE STRUCTURE OF XENIA HIOKSONIT. 291

tera of these siphonozooids cannot be traced more than two
millimetres into the stem. At this depth the celentera ter-
minate in connection with one or more of the numerous
endodermic canals. Ova are not present in any of these
celentera.

Stem.—No definite superficial endodermic canal system is
distinguishable on the summit of the stem, except at its edge,
where the young buds appear to originate from it. It is, how-
ever, present in the cylindrical portion of the stem, and has a
similar appearance and relations to the corresponding canal
system of Xenia Hicksoni.

The longitudinal canals are developed to a much less extent
than in Xenia Hicksoni. They are small, and their lumen
is often almost obliterated. Canals are most uumerous in
the upper part of the stem, especially in and immediately
below the portion penetrated by the ccelentera of the siphono-
zooids. In this region of the stem there are very numerous
short transverse canals which place the various ccelentera in
intimate communication with each other.

The Origin and Internal Structure of the Buds of
Xenia Hicksoni.

In this specimen the buds or young polyps arise on or just
under the edge of the umbrella-shaped area at the end of the
stem. In most other species of Xenia the buds generally
arise in a similar position, but in two of the colonies I have
examined, and in three of the new species described by Schenk
(1896), buds occur not only round the edge, but a few also on
the middle portion of the expanded end of the stem. The
buds are much more numerous on the edge of the summit of
the stem and in Xenia Hicksoni, and, in fact, in most
specimens the buds are found only in this position.

On examining a transverse section of the upper portion of a
stem of the colony, the cclentera in the peripheral portion of
the stem are seen to be smaller than those in the central
portion, and some of them are obviously the celentera of very
young polyps (Pl. 25, fig. 9, right).

292 J. H. ASHWORTH.

1. The buds arise in connection with that portion of the
superficial endodermic canal system situated at the edge of the
summit of the stem. When a bud is about to be formed, the
superficial canal becomes enlarged and the outer wall of the
canal becomes pushed outwards towards the surface of the
stem. This produces a small tubercle upon or immediately
under the edge of the expanded end of the stem. The
tubercle increases in size, and its distal end soon becomes
divided into eight small pouches which become the tentacles
of the polyp (Pl. 24, fig. 4). The divisions between the
pouches are the mesenteries of the polyp, and they gradually
grow inwards into the celenteron. The eight mesenteries are
already formed in the young polyp shown in fig. 4.

In the centre of the free end of the bud there is a slight
depression, and a small darker area which marks the position
of the future mouth (fig. 4, Mo.), and also the ingrowing plug
of ectoderm which forms the stomodzeum. Sections of this
bud show that the stomodeum is ‘14 mm. long, and rather
flattened laterally. Its oral end is still solid, but the inner
portion has become tubular, and its cavity opens into the
celenteron of the polyp. The distance from the depression
indicating the position of the mouth to the innermost portion
of the ccolenteron is ‘4 mm. Mesenteries are distinguishable
only in the upper three fifths of the celenteron. In other
respects this polyp resembles the one described below, and
drawn in fig, 28.

2. A slightly older polyp (No. II in table, p. 284, and drawn
in section, fig. 28), -43 mm. long, possesses tentacles which are
not all of the same size. The dorsal and ventral ones are
largest (about ‘12 mm. long), the two lateral ones a little
smaller, and the remaining four still smaller. These differ-
ences in size are very well seen in transverse sections through
the polyp at the level of the mouth. A second specimen of
the same size was cut longitudinally, and shows several inter-
esting points (fig. 28). The stomodzum is about ‘32 mm.
long, and is tubular along the greater part of its length. The
mouth is, however, closed partly by approximation of the walls

THE STRUCIURE OF XENIA HICKSONI. 293

of the stomodeum and partly by a small plug of mucus. The
siphonoglyph is not yet differentiated. The distance from
the mouth to the inner end of the cceelenteron is about ‘7 mm.
The mesenteries may be traced almost to the inner end of the
ceelenteron, and their retractor muscles are just distinguish-
able. The dorsal mesenteries are slightly thicker than the
others, but their mesenterial filaments have not yet been
formed. The celenteron is formed by enlargement of the
superficial endodermic canal. An evagination of the outer
wall of the canal forms the ccelenteron of the free portion of
the polyp, and pushes outwards the mesogloea and ectoderm,
thus giving rise to the protuberance which forms the free
portion of the young polyp. A diverticulum produced by the
bulging inwards into the mesogloea of the inner wall of the
canal forms the inner part of the ccelenteron.

The endoderm of the outer wall of the superficial canal is
much thicker than that of the inner wall (fig. 29). This dif-
ference enables one to see in section (fig. 28) that the endo-
derm of the whole of the free portion of the polyp is derived
from the thick outer wall of the canal. The endoderm of the
portion of the ccelenteron situated in the outer part of the
stem is formed directly from that of the canal, while the endo-
derm of the inner portion of the ccelenteron is derived from
that of the thin inner wall of the canal, which has grown in-
wards into the mesogloea. The ccelenteron gives off a large
canal at its base, which opens into the adjacent longitudinal
canals. During its short course in the stem the celenteron
communicates several times by means of short canals with the
neighbouring superficial and longitudinal canals.

Among the endoderm cells of the middle portion of the
ceelenteron there are a very few flagella-bearing cells. The
flagellum is a short conical or finger-shaped process which in
these sections is never more than 15 w in length.

3. A slightly longer bud (No. III in the table, p. 284), -56
mm. long, possesses, like the preceding example, tentacles of
different sizes. In this specimen also the dorsal and ventral
tentacles are largest, the lateral ones being next in point of

VOL, 42, PART 3.—NEW SERIES. U

294. J. H. ASHWORTH.

size, and the four intermediate ones somewhat smaller. Does
this indicate the order of development of the tentacles? This
bud resembles the preceding one in all essential particulars,

4. There are several important new structures noticeable in
a young polyp ‘8 mm. in length (No. IV in table, p. 284).
The tentacles are very slightly indented a short distance from
the tip; this is the first indication of the formation of pin-
nules upon the tentacles.

The stomodzeum is ‘57 mm. long, and oval in transverse section.
It is open throughout its whole length, thus placing the
coelenteron in communication with the exterior. There is now
also a marked ventral groove, which, however, is not distin-
guishable in the outer third of the stomodeum. The cells of
the lower two thirds of the groove bear flagella, except for a
very short distance near the inner end. Some of the cells of
the lower half of the stomodzum contain a large cavity, and
are similar to the goblet-cells described in the stomodzum of
the adult. They are more numerous on the lateral walls of
the stomodzeum, and do not occur among the cells forming the
siphonoglyph. The ectodermic muscles of the tentacles and
the retractor muscles on the mesenteries are now quite
obvious. The endoderm covering the mesenteries and lining
the body-wall is very thick in the free portion of the polyp,
(as in fig. 28), and the intermesenterial spaces are consequently
very small. The mesenteries may be traced nearly to the
end of the cceelenteron, i.e. about 1 mm. below the lower end
of the stomodeum. Flagella-bearing cells are present in the
middle and lower portions of the ccelenteron, but the flagella
are still few in. number and of small size, never exceeding
20 uw in length.

The most novel feature in this polyp is the presence of dorsal
mesenterial filaments, which may be traced more than halfway
(‘6 mm.) down the free edge of the dorsal mesenteries. They
have already acquired their typical structure, i.e. each is a
band of ciliated cells on the somewhat thickened edge of each
of the dorsal mesenteries, and there is a longitudinal groove
down the middle of the band, so that in transverse section

THE STRUCTURE OF XENIA HICKSONI. 295

it is V-shaped. The filaments have a very sinuous course, and
are therefore cut across two or three times each in the same
section. These filaments are undoubtedly derived from the
lower end of the stomodzum, with which they are perfectly
continuous. Their cells are exactly like the cells of the stomo-
dum, being small and having homogeneous or finely granular
deeply staining protoplasm and dark nuclei. Their cells differ
markedly from the cells of the surrounding endoderm, which
are larger, have reticulate protoplasm, stain more lightly, and
have less distinct nuclei.

There is another very interesting feature worthy of note in
this polyp, viz. that sexual cells are already clearly differen-
tiated. Sections through the mesenteries in the lower part
of the celenteron show that a considerable number of the
endoderm cells covering the mesenteries are more spherical,
and two or three times as large as their neighbours. They
have clearer protoplasm than the ordinary endoderm cells, and
their nuclei are large (5 uw in diameter) and vesicular, and
possess a well-marked, deeply staining nucleolus (fig. 86 shows
them in a slightly older polyp). These cells occur in all the
mesenteries, but are more numerous in the ventral and lateral
than in the dorsal ones. Some of the cells have already
migrated into the mesogloea of the mesenteries, but most of
them still retain their position in the endoderm. These modi-
fied cells are present only in the portion of the mesenteries
situated in the lower two thirds of the celenteron. _

On comparing these cells with the primitive sperm cells of
the adult the resemblance is so striking that I am convinced
these modified endoderm cells are the sexual cells which have
become differentiated at this early stage in the development of
the polyp. This view is supported by the following facts :

a. They are at first endoderm cells covering the mesenteries,
which, on becoming differentiated, migrate into the mesogleal
lamina of the mesentery. This agrees with the origin,
migration, and position of the sexual cells traced in the adult
polyps (see p. 263).

6, These cells are found in the mesenteries some distance

296 J. H. ASHWORTH.

below the lower end of the stomodeum. They are not present
in the mesenteries of the free portion of the polyp, but in the
portion enclosed in the stem. They are therefore in the
position in which the genital products of the adult polyps are
most numerous (see p. 263, and fig. 8).

c. They agree in size and are similar (especially their large
vesicular nuclei) in appearance to the primitive genital cells of
the adult.

Fig. 36 shows a section through a ventral mesentery of a
slightly older polyp (No. V in table, p. 284). The genital
cells are still more clearly marked than in the preceding polyp,
and, as shown in the figure, are readily distinguished by their
large vesicular nuclei. Many of them have already taken up
their position in the mesoglea.

On comparing the polyp *8 mm. long with one about
two thirds its size (‘56 mm. long) it is seen that during the
period in which the elongation of :24 mm. was accomplished,
many important adult structures were acquired, viz. the
first indication of pinnules on the tentacles, the open mouth,
the siphonoglyph, the goblet-cells in the stomodeum, the
dorsal mesenterial filaments, and the primitive genital cells.
A polyp which has reached this stage of development would
probably be quite capable of supporting itself, as, by means of
its siphonoglyph and dorsal mesenterial filaments, it would be
able to create the currents necessary for its nutrition and for
keeping up the circulation of liquid in the ccelenteron.

In a polyp ‘95 mm. long (V in table, p. 284) the first two
pinnules are distinguishable upon each tentacle. The stomo-
dzeum is 6 mm. long, and the siphonoglyph extends along the
ventral side of the lower,‘34 mm. The lips of the stomodzum
are widely open below. Flagella-bearing cells are much more
numerous in this polyp than in any of the young polyps
previously described. They are fewer in number in the upper
part of the colenteron than in the deeper portions. Their
flagella are still short, not exceeding 80 mw in length (fig. 36,
G8).

A polyp 1 mm, long (VI in table) is seen in section in PI,

THE STRUCTURE OF XENIA HICKSONI. 297

25, fig. 8. Its tentacles bear near the tip one pinnule on each
side of the axis. The stomodzum is*74 mm. long, and exhibits
a well-marked siphonoglyph, bearing flagella in its lower half.
The dorsal mesenterial filaments are well developed, and extend
in a sinuous course nearly 1 mm. down the ceelenteron. The
ceelenteron extends 1:4 mm. into the stem, curving so as to be
parallel to the neighbouring ccelentera. Its connections with
the longitudinal and superficial canals may be seen in the
figure. The flagella of the endoderm cells attain 40 m in length.
The distribution of spicules, nematocysts, and zooxanthelle is
shown in the figure. In such a polyp all the adult structures
are differentiated. In the growth of the polyp, from this
point onwards to the full adult size, there are few features
which call for comment. The chief of these are the formation
of a greater number of pinnules on the tentacles; the elonga-
tion of the ccelenteron outwards (as the free portion of the
polyp grows in length) and inwards into the mesoglea, where
it curves, so as to lie parallel to the neighbouring cclentera
(Pl. 25, fig. 8); the increase in number and size of the giant
flagella on the endoderm cells, and the development of sperm
sacs on the mesenteries.

GENERAL SUMMARY,

The following is a recapitulation of the new points to which
reference is made (paragraphs 1, 2, and 6 have been already
published in ‘ Proc. Roy. Soc.,’ 1898) :

1. The absence of ventral and lateral mesenterial filaments
usually present in the polyps of the Aleyonaria. The only
previously recorded examples of the absence of these filaments
are the siphonozooids which occur in some other Alcyonacea
and in Pennatulids.

2. The absence of these filaments is correlated with the
presence of gland cells in the stomodzum, which occur espe-
cially in the ventro-lateral walls which abut on the siphono-
glyph. Their position is suggestive of the digestive function
of the cells, as their secretion can be readily poured out on
to the ingoing food particles.

298 J. H. ASHWORTH.

3. The non-retractile nature of the bodies of the polyps
and of the stems is accounted for by the absence of muscle-
fibres from their ectoderm cells, and by the presence of
numerous spicules in these parts. The presence of ectodermic
muscles in the tentacles, pinnules, and distal millimetre of the
bodies of the polyps, together with the absence of spicules
from these parts, confers the power of contractility, slight
though it is, upon these parts. The muscle processes of the
ectoderm cells (where present) are longitudinal in direction,
while those of the endoderm cells are circular (except the pro-
tractor and retractor muscles on the mesenteries).

4, Nematocysts were found in Sarcophyton.

5. An extraordinary number of spicules is present in the
basal part of the colony, and much of the mesoglcea is converted
into a dense horny substance. Thus a firm base of attach-
ment is provided which would afford a rigid support for the
branches which arise from it.

6. Many of the endoderm cells lining the ceelentera and
tentacles bear giant flagella, which may attain 120 » in
length.

7. In adult polyps the primitive genital cells are formed by
differentiation of some of the endoderm cells which cover the
mesenteries. These genital cells migrate into the mesoglea
of the mesenteries, and then move outwards, one at a time,
each cell pushing the endoderm and a thin film of mesoglea
before it, and so forming a small tubercle on the side or
end of the mesentery. By division of the genital cell the
spermatozoa are produced. They remain until ripe, surrounded
by a thin film of mesoglea, and by a layer of endoderm cells,
many of which contain from two to five nuclei.

8. The longitudinal canals which traverse the mesogloa
of the stem are very well developed, and are physiologically,
and possibly morphologically, equivalent to the coonenchymal
tubes of Heliopora cerulea.

9. The nervous system is similar to that of Aleyonium;
the stellate cells immediately outside the endodermic muscle-
fibres are very clearly seen.

THE STRUCTURE OF XENIA HICKSONI. 299

10. A complete series of developing polyps is described,
beginning with very young specimens, in which the tentacles
are devoid of pinnules, and ending with the adult polyps with
multi-pinnuled tentacles.

11. A similar series of polyps of Heteroxenia Eliza-
bethz is described and compared with the siphonozooids of
the same colony. It is concluded that Heteroxenia is un-
doubtedly dimorphic, as stated by Kolliker and Bourne.

12. The origin of the buds of Xenia Hicksoni from the
superficial canal system, their subsequent growth, and the
appearance in them of the mouth, stomodzum, siphonoglyph,
gland cells in stomodeum, mesenteries, dorsal mesenterial
filaments, and flagella of the endoderm cells, are traced. It
is remarkable that the sexual cells are already differentiated in
a young polyp ‘95 mm. long.

LITERATURE.

1833. Quoy rt Garmarv,—‘ Voyage de |’Astrolabe,’ tome iv, “ Zoologie,”
Paris, 1833.

1874. Kouurker, A.—“ Heteroxenia,” ‘Festschrift der Physical.-Med.
Gesellschaft in Wirzburg,’ p. 13, 1874.

1876. Mosetey, H. N.—*On the Structure of a Species of Millepora
occurring at Tahiti, Society Islands,” ‘ Phil. Trans.,’ vol. clxvii, part
i, 1876.

1877. Kiunzincer, C. B.—‘ Die Korallthiere des Rothen Meeres,’ erster
Theil, ‘‘ Die Aleyonarien und Malacodermen,” Berlin, 1877.

1881. MoseLtey, H. N.—‘ Challenger Reports,’ Zoology, vol. ii, Corals,
1881 (Heliopora and Sarcophyton).

1883. Hicxson, 8. J.—On the Ciliated Groove (Siphonoglyphe) in the
Stomodeum of the Alcyonarians,” ‘ Phil. Trans.,’ part iii, 1883.

1883. Witson, E. B.—‘‘ The Development of Renilla,” ‘ Phil. Trans.,’ part
ili, 1883.

1884. Witson, E. B.—‘The Mesenterial Filaments of the Aleyonaria,”
‘ Mitt. Zool. Stat. Neapel,’ v, 1884.

1887. Haackzs, W.—‘‘ Zur Physiologie der Anthozoen,” ‘ Zoologischer
Garten,’ Jahrg. xxvii, p. 284, 1887.

300 J. H. ASHWORTH.

1889. Wrieut, E. P., and StupEr, T. H.—‘ Challenger Reports,’ Zoology,
vol. xxxi, ‘‘ Aleyonaria,”’ 1889.

1895. Bourne, G. C.—“ On the Structure and Affinities of Heliopora
ccerulea, Pallas, with some Observations on the Structure of Xenia
and Heteroxenia,” ‘ Phil. Trans.,’ 1895.

1895. Hickson, 8. J—‘The Anatomy of Aleyonium digitatum,” ‘ Quart.
Journ. Mier. Sci.,’ vol. xxxvii, part 4, 1895.

1896. Scunnx, A.—‘ Clavulariiden, Xeniiden, und Alcyoniiden von Ternate,’
Frankfurt-a.-M., 1896.

1898.—AsnwortH, J. H.—‘ The Stomodeum, Mesenterial Filaments, and
Endoderm of Xenia,” ‘ Proc. Roy. Soc.,’ vol. Ixiii, 1898.

EXPLANATION OF PLATES 238—27,

Illustrating Dr. J. H. Ashworth’s paper on “'The Structure
of Xenia Hicksoni, nov. sp., with some Observations
on Heteroxenia Hlizabethe, Kolliker.”

List or RerereNce Lerrers.

A,A3,A5. Autozooids of Heteroxenia (the numbers refer to the table
on p. 287). Cg. Coagulum in central cavity of sperm sac. Cz.C. Cnido-
blast cell. D. JZ, Dorsal mesentery. D.I/./. Dorsal mesenterial filament.
Ect. Ectoderm. Let. Ch. Chain or cylinder of ectoderm cells round cylinder
of denser mesoglea. Het. Str. Strands of ectoderm cells. xd. Endoderm.
End. Can. Endodermic canals. F. Flagella of siphonoglyph. G.C. Gland-
cellsin stomodeum. G.F. Giant flagellaof endoderm cells. Gen. C. Genital
cells in various stages of development. Long. Can. Longitudinal endodermic
canals. JZ. M. Lateral mesentery. MZ. Mesentery. My. Mesoglea. Mg. D.
Denser cylinder of mesoglea around each ccelenteron in stem. Jo. Mouth.
Muc.C. Mucus cells of ectoderm. J/.P. Muscle processes of endoderm
cells. MV. Nucleus. Nem. Nematocysts. N.C. Nerve cells. WV. F. Nerve
fibrils. 2. 12. Retractor muscles in mesenteries. §.S7. Siphonozooids of
Heteroxenia (the number refers to the table on p. 288). Sec. Secretion of
gland-cells of stomodeum. S%. Siphonoglyph. Sp. Spicules. Spz. Sper-
matozoa. S¢. Stomodeum. Swp. Can. Superficial canal. §.8. Sperm sacs.
V. M. Ventral mesentery. Z. Zooxanthelle.

PLATE 23.

Fic. 1.—View of the colony of Xenia Hicksoni, From one of the stems
on the left side all the larger polyps have been removed, so that the umbellate
summit of the stem, with the young polyps growing round its edge, may be

THE STRUCTURE OF XENIA HICKSONI. 301

more clearly seen. Note also that the polyps are smaller and closer together
near the edge than nearer the middle of the summit. The stem immediately
to the right of this also shows that the young and small polyps are on the
edge, while the large polyps are near the middle of the summit. ‘The colour
of the drawing is, as nearly as possible, the colour of the specimen in spirit.
x 3.

PLATE 24.

Fic. 2.—View of the polyp XXI in the table, p. 284. This is one of the
largest polyps of the colony. The tentacles show from the outer aspect a
single row of pinnules on each side, but from the inner aspect three rows on
eachside. Note that at the tip of the tentacles the pinnules are not arranged
in pairs, X 12.

Fic. 3.—Views of the tentacle of a large polyp. 4 and B are views of
the tip, @ and D of the base of a tentacle. 4 and C show the outer or aboral
side, B and D the inner or oral side. The narrow area free from pinnules,
which extends along the middle line of the oral face of the tentacle, is well
seen in D. At the tip of the tentacle, B, the pinnules are in two rows only
on each side of the middle line. xX 24.

Fic. 4.—Lateral view of the youngest polyp (32 mm. long) in the colony
(I in table). x 40.

Fie. 44.—Oral view of the same polyp. The tentacles are eight rounded
lobes. The depression (Mo.) in the centre of the oral disc indicates the
position of the future mouth, and the darker area below it is the ingrowing
plug of ectoderm which forms the stomodeum. x 40.

Fic. 5.—An older polyp °95 mm. long (V in table), The tentacles are
trilobed at the end, i.e. the first pinnules are being formed. Xx 24.

Fic. 6.—A polyp 1°42 mm. long (VIII in table). x 24.

Fie. 6a.—View of the oral face of a tentacle of the polyp shown in fig. 6.
Two rows of pinnules on each side of the middle line are now present. x 24.

Fig. 7.—A polyp 2°27 mm. long (XII in table). x 24.

Fic. 74.—View of the oral face of a tentacle of the polyp shown in fig. 7.
In the middle part of the tentacle, three rows of pinnules on each side of the
middle line are now distinguishable. xX 24.

PLATE 25.

Fic. 8.—A thick longitudinal section through the upper part of one of the
stems. The section passes along the dorso-ventral axis of a young polyp (VI
in table, p. 284) growing out just under the arched summit of the stem. On
the dorsal or upper side of the polyp one of the dorsal mesenteries (D. I.) is
cut through obliquely for a short distance. The stomodseum (S7.), siphonoglyph
(Si.), the course of the dorsal mesenterial filaments (D. 1, F.), the thin edge

3802 J. H. ASHWORTH.

of the ventral mesentery (V. W.), and the connection of the ceelenteron with
the canal system are shown. ‘Three celentera of older polyps are also shown.
Two of them are crowded with sperm sacs (8.8.), many of which, owing to
mutual pressure, have lost their original spherical shape and are pentagonal
or hexagonal in section. The other ccelenteron contained a similar number of
sperm sacs, but they have been omitted in order to more clearly show the
dorsal mesenterial filament (D. W. F.) and the thin edge of the ventral mesen-
tery (V. M.). The superficial (Sup. Can.) and longitudinal (Long. Can.) canal
systems, and their relation to each other and to the ccelentera; the denser
cylinder of mesoglea (Mg. D.) with its surrounding ectoderm cells enclosing
each coelenteron; and the distribution of spicules, nematocysts, and zooxan-
thelle are also shown. xX 35.

Fie. 9.—A thinner transverse section through the upper part of one of
the stems. The ccelentera in the peripheral part are smaller than those
nearer the centre. The cclenteron of a very young polyp is seen on the
right. The cells in the mesoglea of the mesenteries and sperm sacs in
various stages of development, attached to the mesenteries of the older ccelen-
tera, are also shown. The number of flagella cut across in one section is seen ;
thus an idea may be formed of their abundance and of their size relative to
the endoderm and to the celentera. Many of the features to which attention
was drawn in the description of the previous figure are also shown here.
x 50.

Fic. 10.—-Transverse section through a polyp at the level of the lower
third of the stomodeum. The chief features shown are the gland-cells and
siphonoglyph of the stomodeum, the somewhat feebly developed retractor
muscles on the mesenteries, and the distribution of the spicules, nematocysts,
and zooxanthelle. x 50.

Fie. 11.—Transverse section of the outer wall of a pinnule showing the
ectoderm containing nematocysts, the mesogloea, and the endoderm cells with
their reticulate protoplasm. xX 800.

Fic. 12.—A nematocyst in its enidoblast cell, from a section. One half of
each coil of the thread in the upper part of the nematocyst was cut away in
sectioning. x 2000.

Fic. 13.—Eight large spicules selected from those in the base of the
colony. Each spicule is situated in a small cavity in the mesoglea. The
nucleus and remains of the protoplasm of the spicule-forming cell are seen.
From sections. Xx 800.

Fic. 14.—A portion of the outer part of the mesoglca from a transverse
section of a polyp. Four of the spicules present their edge and two their
flat face to the observer. x 800.

Fic. 15.—Two very young spicules in their respective cells. The spicule
in 4 is 33 » long and 23 » broad. The spicule in B presents its edge to the
observer. Itis 7 »longand 134m thick. x 1000.

THE STRUCTURE OF XENIA. HICKSONI. 303

PLATE 26.

Fic. 16.—Section of the wall of a polyp which has passed very obliquely
through the mesogloea, to show the nervous system. Fine nerve fibrils in the
mesoglcea are connected on the outer side with cells in the outer part of the
mesogloea, some of which send processes into the ectoderm. On the inner
side, the fibrils pass into small stellate nerve-cells situated just outside the
muscle processes of the endoderm cells. x 300.

Fic. 17.—Longitudinal section through the body-wall of a polyp, to show
the general character of the ectoderm and endoderm cells, and also the
processes of ectoderm cells which penetrate the mesogloea and establish
connection with the endoderm. x 400.

Fic. 18.—Longitudinal section of the ventro-lateral portion of the lower
third of the stomodeum. The chief feature shown is the large gland-cells,
some of which have discharged their contents and appear empty, while others
are in the act of discharging their secretion. x 600.

Fic. 19.—Transverse section of the end of a dorsal mesentery showing the
V-shaped mesenterial filament bearing cilia. x 600,

Fic, 20.—Transverse section of the end of a ventro-lateral mesentery to
show the general character of the endoderm cells, the giant flagellum of one of
them, and the cells in the mesoglea. x 600.

Fic. 21.—Transverse section of the end of the same mesentery taken
°25 mm. higher, showing three flagella occurring close together. x 600.

Fies. 22, 23, 24.—Three flagella from sections of polyps to show their
various degrees of flexion. x 600.

Fic. 25.—Two abnormal forms of flagella from sections. 4 from the
tentacle of a polyp, B from the portion of a celenteron ina stem. x 600.

Fig. 26.—Endodermic myo-epithelial cells from teased preparations. In
connection with the one on the right there is a small stellate cell with three
long processes. ‘This is probably one of the nerve-cells of the endodermic
portion of the nerve plexus. x 500.

Fig. 27.—Isolated cells bearing flagella, from teased preparations. x 500.

PLATE 27.

Fic. 28.—Section of a stem at the edge of the umbellate summit, to show
the formation of a young polyp ‘43 mm. long (II in table, p. 284). Note
that the endoderm of the free portion of the polyp is thick, while that of the
inner part of the ccelenteron is thin. The neighbouring cclentera are cut very
obliquely. x 50.

Fig. 29.—The superficial canal indicated by the asterisk in Fig. 9 enlarged.

The cells of the outer wall are longer and more columnar than those of the
inner wall. x 500.

804, J. H. ASHWORTH.

Fic. 30.—Thin transverse section (3 thick) of the end of a ventro-lateral
mesentery, to which three very young sperm sacs are attached. In the one
to the right the primitive genital cell has divided into four, three only of
which are visible. x 500.

Fig. 31.—Section of a sperm sac a little older than the largest one shown
in the preceding figure. The central cavity containing a coagulum has now
appeared. x 150.

Fie. 32.—Section of an older sperm sac. x 150.

Fic. 33.—Section of a mesentery, on the end of which is a sperm sac in
which the central cavity, having reached its greatest size, is now being
encroached upon by the heads of ripening spermatozoa. Note that the
spermatozoa are surrounded by a thin film of mesogloea (represented in the
distal part of the sac by the line inside the endoderm), and by an endodermic
follicle, some of the cells of which contain from two to five nuclei. x 150.

Fic. 34.—Ripe spermatozoa.

Fic. 344.—From a section of the stomodum, through the dorsal portion
of which spermatozoa are escaping. Only the head and point of attachment
of the tail are seen. x 800.

Fic. 348.—An entire ripe spermatozoon from a teased preparation of a large
sperm sac. X 800.

Fie. 35.—Cells from thin sections (4 w thick) of large sperm sacs, to show
the two to five nuclei which may be present in each. x 809.

Fic. 36.—Transverse section of a ventral mesentery of a young polyp ‘95
mm. long (V in table, p. 284; see also Pl. 24, fig.5). The section is taken
about °5 mm. below the lower end of the stomodeum. Note the already
differentiated primitive genital cells migrating from the endoderm into the
mesoglea, and the short finger-shaped flagella of three of the endoderm cells.
x 500.

Fic. 37.—View of a colony of Heteroxenia Hlizabethe. The colony
was split in two longitudinally and the proximal half drawn, hence only half
the summit of the stem is seen. Note the autozooids with pinnate tentacles,
and the very numerous short siphonozooids with short rounded tentacles,
aud also the young polyps growing near the edge of the summit. x 3.

Fig. 38.—View of the youngest autozooid on the colony, 64 mm. long
(A 1 in table, p. 287). The specimen has been flattened out by pressure
against the side of the bottle which contained the colony. Note the long .
finger-shaped tentacles; the right proximal one is slightly indented near the
tip, but the others have simple rounded ends. The stomodzum is indicated.
x 40.

Fic. 89.—The largest siphonozooid on the colony (S 7, fig. 38, and table,
p. 288). The short simple tentacles, the stomodeum, and the two dorsal
mesenteries with their filaments are shown. xX 15.

NOTES ON BATRACHIANS OF THE PARAGUAYAN CHACO. 305

Notes
Ch

on the Batrachians of the Paraguayan
aco, with Observations upon their Breed-

ing Habits and Development, especially with
regard to Phyllomedusa hy pochondrialis,
Cope. Also a Description of a New Genus.

By

J. S. Budgett,
Trinity College, Cambridge.

With Plates 28-—32.

List of Batrachia collected by J. S. Budgett in the Paraguayan

Chaco, 1897.

Length.
I. Leptodactylus ocellatus, L. : . 120mm.
II. Leptodactylus typhonius, Daud. . : 45 mm.
Ill. Leptodactylus bufonius, Boul. . : 55 mm.
IV. Leptodactylus pecilochilus (Cope). 50 mm.
V. Phryniscus nigricans, Wiegm. : : 33 mm.
VI. Paludicola fuscomaculata, Steindachn. . 40 mm.
VII. Paludicola signifera, Boul. . : : 25 mm.
VIII. Paludicola falcipes (Hensel) . : : 15 mm.
IX. Eugystoma ovale, Schn. . : . 9° 40mm., > 25 mm.
X. Eugystoma albopunctatum, Boul. . ; 18 mm.
XI. Pseudis paradoxa,L. . 5 : : 50 mm.
XII. Pseudis limellum (Cope) . : ‘ ; 20 mm.
XIII. Bufo marinus, L. . ; : : ; 150 mm.
XIV. Bufo granulosus, Spix. ; F : 50 mm.
XV. Phyllomedusa hypochondrialis, Cope . 40 mm.
XVI. Phyllomedusa Sauvagii, Boul. ‘ 5 70 mm.
XVII. Hyla spegazinii, Boul. . : : ; 80 mm.
XVIII. Hyla venulosa, Laur. ‘ : 4 ; 70 mm,
XIX. Hyla nana, Boul. ; : ; ‘ ; 22 mm.
XX. Hyla phrynoderma, Boul.. ‘ ‘ : 43 mm,
XXI. Hyla nasica, Cope . : ; ‘ ; 28 mm,
XXII. Ceratophrys ornata, Bell. : ‘ : 120 mm.
XXIII. Lepidobatrachus asper, n. sp. i : 80 mm.
XXIV. Lepidobatrachus levis, n. sp. : ; 80 mm,

306 J. 8S. BUDGETT.

I. LEPTODACTYLUS OCELLATUS, L.

An extremely common frog, frequently found in the streets
of Concepcion at sunset and on both sides of the river Para-
guay.

At Concepcion, black markings on a greyish-green ground.
At Waikthlatingmayalwa the ground is usually of a brighter
green.

A triangular black spot at the back of the eyes is very
constant.

The natives, who are Lengua Indians, name this frog Nukk-
mikkting, and use it largely for baiting their hooks. The
largest measure 50 mm. from snout to vent.

The call is regularly repeated, beginning on a low note and
ending on a high one, and is constantly heard in wet weather.

There is, however, another call, which is heard immediately
after rain ; this is a drumming like that of a snipe.

A large variety found in the Chaco is called by the Lenguas
Yattnukkmikkting; these measure up to 120 mm., and are
only found down in the swamps. I think this may be L.
bolivianus.

II. LeEpropActyLus TYPHONIvS, Daud.

Not nearly so common as L. ocellatus; I procured only
two specimens, though I saw a few others. These were all
seen at Caraya Vuelta on the river bank. ‘The general colour
is lighter than ocellatus, the spots are more numerous and
smaller, and there is a bright gold band on either side running
from the eye to the hips.

It appears to be about the same size as ocellatus.

No Lengua name was obtained for it.

III. LEerTropactyLus BUFONIUS, Boul.

Small brown frog with blackish spots above, beneath pale

NOTES ON BATRACHTANS OF THE PARAGUAYAN CHACO. 807

yellow. Most inconspicuous on a background of earth. It is
very agile and extremely shy.

In damp waste places on the outskirts of Concepcion I
found it in great numbers, but very difficult to capture.

The call is a shrill sharp “ ping” kept up constantly until
approached, when it immediately ceases. The croaking of so
many of them at a time produces an almost continuous sound.

Though only one specimen was secured, it was frequently
heard on both sides of the river. This is probably the young
form of L. bufonius, which grows to about the same size as
L. ocellatus. I never detected large individuals of this form
calling, and I am convinced that during the continuous calling
described above the individuals about were of the small form
almost entirely.

It would appear, then, that either young forms have the
habit of calling to one another, or that there is a small and a
large variety. Lengua name Ukksaliapertikk. In Lengua
Uksaelia means a coin or disc. The name refers to the
spots.

IV. LepropacryLus pacitocnitus (Cope).

This frog is much less common than L. ocellatus. It is
of a more slender build ; the toes are thin and long, especially
the second toe. ‘The markings are all in the form of stripes
rather than spots. These are dark brown on a greyish-brown
ground, At the side yellowish. One broad dark stripe runs
down the back on either side at the edges of the transverse
processes of the vertebrae. One specimen was found at Con-
cepcion and one at Waikthlatingmayalwa.

I do not know if it has a native name.

V. PHRYNISCUS NIGRICANS, Wiegm.

This is a brilliantly coloured frog of toad-like appearance.
The ground colour is black, and is irregularly spotted with

308 J, (Ss BUDGHIT;

yellow, or sometimes with large yellow blotches on the upper
surface. Beneath it is black, with scarlet blotches; the palms
of the hands and the soles of the feet are scarlet.

The variety found at Concepcion had on the under surface
scarlet blotches extending to the throat, while the variety
found at Waikthlatingmayalwa had the scarlet confined to the
lower part of the abdomen.

This form, too, had yellow blotches irregularly arranged on
the back, while the Paraguayan form had small yellow spots
more regularly arranged.

On the journey between these two regions I twice met with
large numbers of small black frogs which seem to be of this
species. They were characterised by their smallness, and by
the absence of either yellow or red markings.

At the breeding season the males and females have a call
which consists of two clear musical “ pings,” followed by a
long descending “trill” like that of our British greenfinch.
The eggs are laid in separate globules of jelly, which float
freely on the surface of the water, and are heavily pigmented.

This frog, which at ordinary times is the slowest and most
bold of frogs, is now active and excessively shy. Swimming
rapidly between the blades of grass it climbs a tuft, and,
dilating its throat, repeats its call, but if in the least dis-
turbed it is suddenly gone. This change of habit is very
remarkable.

The spawn is found in quite temporary pools in grassy
ground ; the development is excessively rapid. Segmentation
beginning at 10 a.m., they were hatched and wriggling about
by 7 a.m. the following day. They probably are washed down
into deeper pools by the retreating waters, and for this purpose
the manner in which the eggs are laid, i.e. in separate globules
of jelly, seems especially suited.

The native Lengua name is “ Pithpaya.”

The eggs and larve do not seem to differ in any great degree
from those of Rana. There is, however, a very large yolk-
plug, which remains evident after the closure of the neural
groove,

NOTES ON BATRACHIANS OF THE PARAGUAYAN CHACO. 309

VI. PALUDICOLA FUSCOMACULATA, Steindachner.

This is the largest of the genus that I found in the Chaco.
It is a short-limbed frog, with spreading slender toes and small
head.

The upper surface is marked with characteristic marblings,
which vary, however, greatly in colour. The metatarsal tu-
bercles are large, horny, shovel-shaped, and black.

The peculiar cry which is so constantly heard in the neigh-
bourhood of shallow pools, and resembles that of a kitten, is
produced by the alternate inflation of throat and abdomen.

When fully inflated the frog appears to be the size of a golf
ball, but, if startled, instantaneously shrinks to one fifth of
that size, so that it seems to have vanished. It has also the
power of ventriloquising.

In the spawning time it was found at night floating on the
surface of pools in the distended condition, and crying to the
females in a most mournful way. On coming to the surface it
fills its lungs with a few gasps, greatly distending the walls of
the abdomen, and then drives the air into the throat diver-
ticula of the pharynx, causing them to become distended as the
stomach collapses, and giving rise to a kitten-like cry.

The eggs are chiefly laid in January, and are found em-
bedded in a frothy mass floating upon the surface of the water.
The eggs themselves measure 1 mm. in diameter, and are
without pigment and with extremely little yolk. They become
free-swimming within from eighteen to twenty-four hours of
the time of the first segmentation. When ready for hatching
they wriggle their way through the froth to the water below,
and hang into it from the floating froth.

In this rapidly hatching, free-swimming larva many of the
processes of development are blurred, and as it were hurried
over. The external gills never reach a high state of develop-
ment. The cell layers are many cells deep and diffuse, and the
involutions and evolutions are difficult to follow.

vou, 42, PART 38.—NEW SERIES. x

310 J. S. BUDGETT.

The natives call this frog “Zing Ye,” which of course
applies to the genus generally, for the species differ very
slightly.

In this species the testes are much pigmented and lobulated.

Its food consists largely of water-beetles.

VII. PatupicoLa sSIGNIFERA, Boul.

This is considerably smaller than fuscomaculata, and is
usually an olive-green on the back without conspicuous
markings.

Its general habits seem to be the same as those of P. fus-
comaculata, as also its cry. It is most agile. I put this
species into a cage in which were many brightly coloured
frogs, including Phryniscus nigricans and also Phyllo-
medusa hypochondrialis. In this cage was also a small
grass snake. Hitherto it had taken no interest at all in the
gaudy frogs in its cage; but as soon as the little Paludicola
made its first spring, it was caught in mid-air by the snake.

VIII. Patuptcota FauciPEs (Hensel).

Only one specimen found at Concepcion by the river side.
Its toes are even, long, and slender. Many of the specimens
in the British Museum are marked with one broad light band
running from nose to vent. But by no means all have this,
neither does it depend on sex. In the specimen which I pro-
cured this stripe is very much marked.

IX. EneystomMa ovALE, Schn.

This frog has a small head and pointed nose. The eyes are
set far forward, and there is an encircling fold just behind the
eyes. The fore-limbs are very small, and the general shape of
this frog proclaims it at once to be a burrower. The skin is
perfectly smooth. It is greenish brown above, yellow beneath,
and a bright yellow band passes up the thighs and over the

NOTES ON BATRAOHIANS OF THE PARAGUAYAN CHACO. 311

vent. In the male this band is bright red. The male is
somewhat smaller. The natives call it “Po it,” being convinced
that the cry which sounds to them thus proceeds from this
frog. However, in each case that I tracked down, the frog
calling with this cry I found a Leptodactylus ocellatus.
The cry was heard everywhere, but I only found one male and
one female.

I think the native boys were here mistaken again; they
pointed out to me holes in the ground beneath fallen tree
trunks, of the size of a cricket ball and lined with a froth con-
taining white eggs and also tailed larve. The entrance to the
whole was about a centimetre in diameter. This they said was
the nest of the “ Po it.”

I reared some of the eggs, and one as far as the four-legged
stage, when the young frog bore a very strong resemblance
to a Paludicola, but unfortunately escaped from my tank
before it had lost its tail.

Though the information obtained from the natives generally
turned out to be fairly accurate, yet I feel sure that in some
instances they were quite wrong.

To whatever frog these nests belonged, it is certain that they
were a most ingenious contrivance for collecting water and
keeping the eggs and larve at least moist, between the storms
of the wet season. They were always found within the forest
belts which lay on the highest ground.

I found with these larve that they would exist for a very
long time in a small quantity of water without increasing in
size, but that when removed to a tank they grew enormously,
and very soon left the water.

These eggs were somewhat larger than the minute eggs of
Paludicola, 1; mm.,and pigmentless. As far as my investiga-
tions have gone these eggs develop much as Paludicola, though
they are rather more heavily yolked.

X. EneysToMA ALBoPUNCTATUM, Boul.

About half the size of E. ovale, and found under a heap of

312 J. S. BUDGETT.

decaying vegetation in the forest. Plum-coloured and very
glossy above, and greyish with white spots below. One speci-
men found. Native name unknown.

(The specimens collected by Bohls, in Paraguay, are all
brightly spotted above.)

XI. Pszupis PARADOXA, L.

A water-frog never seen on land, and extremely shy.
Though often seen floating in a shallow pool, it was caught
with great difficulty.

In life most beautifully coloured with bronze, bright green,
and black markings above; underneath a satiny sulphur-
yellow, with brown spots on the trunk and brown stripes on
the thighs. On killing, all the brilliant colours of the back
turned to a dull uniform brown in a few minutes.

Though there were a pair of these in a pool all through the
early part of the wet season, yet the pool did not contain any
of the well-known gigantic larve with reference to which the
frog is named.

No native name known to me.

XII. Psrupis LIMELIUM (Cope).

Small green frog abundant on the camelota leaves at Con-
cepcion. Capable of changing its colour greatly from bright
green to dull brown, underneath silvery. Two white streaks
run backwards from the eyes. The call is a succession of
sharp croaks or vibrations resembling the sound made by
castanets. The throat is inflated for each series.

They hop quickly over the surface of the water, and perch
on the camelota leaves and stems. They are enabled thus to
hop on the surface of the water by reason of the very large
webs of the hind feet. The tips of the toes also have dilated
discs. Their food consists mainly of small fresh-water Gas-
teropods. Females larger than the males.

No native name known to me.

NOTES ON BATRAOHIANS OF THE PARAGUAYAN OCHACO. 313

XIII. Buro marinus, L.

The common toad of South America, up to 150 mm. in
length from snout to vent. It feeds on all kinds of insects,
and is very useful in helping to keep down the mosquitoes.
One half-grown toad, sitting by one man’s foot, picked off
fifty-two mosquitoes in the space of one minute, flicking them
up with his tongue as they settled.

This toad, which may be found in every shed or outhouse, is
called by the natives “ Pinnikk.” Its call consists of three
bell-like notes, the middle one being the highest. The parotid
glands are enormously developed, and, if the toad is roughly
handled, are discharged like squirts. When wet weather
comes it hops out from its hiding-place, and proceeds to sit in
a puddle, with its head out.

XIV. Buro GRANULOsUS, Spix.

A very common small toad, found in great numbers near
water after rain. Dark above, with black, brown, and greenish
blotches, and a light vertebral line. Skin much tuberculated.
Calls with a continuous bell-like tinkle, the vocal sac being
greatly distended. A great deal of variety in colour.

Native name “ Kelaelik.”

This species forms the chief food of the two newly de-
scribed species of Lepidobatrachus.

XV. PHYLLOMEDUSA HYPOCHONDRIALIS, Cope.

A brilliantly coloured grass frog, which I found breeding
freely in the Paraguayan Chaco, about 120 miles due west of
Concepcion (fig. 34). Above it is brilliant green, which may
become brown, grey, or bluish at will; below granular white.
The flanks are scarlet with black transverse bars, and the
plantar surfaces are a deep purplish black.

314 J. §. BUDGETT.

The “ Wollunnkukk,” as it is called by the Lengua
Indians, from the call of both male and female at pairing time,
is extremely slow in its movements, and is active only at night.
At this time, if it is seen by the aid of lantern as it slowly
climbs over the low bushes and grass, it is very conspicu-
ous, as shown in the figure. In the daytime, however,
nothing is seen but the upper surfaces of the body as it
hes on the green leaf or caraguata plant, and here it is most
inconspicuous.

This small Hylid has a remarkable power of changing the
colour of its skin to harmonise with its surroundings, and can
effect a change from brightest green to a light chocolate ina
few minutes. The skin is also directly sensitive to light; for
if the frog is exposed to the sun while in a tuft of grass in
such a way that shadows of blades of grass fall across it, on
removal it will be found that dark shadows of the grasses
remain on the skin, while the general colour has been raised
to a lighter shade. Its food consists largely of young locusts.
The ovaries on each side are divided into five distinct clusters.
The rectum has a large saccular diverticulum, which is very
heavily pigmented.

In the breeding season—December to February—this beau-
tiful grass frog collects in considerable numbers in the neigh-
bourhood of pools. During the night-time they call inces-
santly to one another, and produce a sound as of a dozen men
breaking stones, well imitated by the native name “ Wol-
lunnkukk.”

As regards the native names for frogs, most species had their
separate names; for instance, two species so closely like one
another as Leptodactylus ocellatus and L. bufonius
had their names respectively “ Nukkmikkting” and “ Ukselia-
pertic,” but with the Tree frogs it was not so. I could get no
name for any frog with dilated discs but “ Wollunnkukk,”
whether they had a call resembling this name or not, and
whatever their form, colour, and size. I may mention also
that they had no general name for frog, though they had a
general name for bird and fish,

NOTES ON BATRACHIANS OF THE PARAGUAYAN CHACO. 315

Breeding Habits.—On November 30th, 1896, I caught
six of these frogs at the edge of a shallow pool late at night,
and put them with some leaves in a tin until the morning.
Next morning I discovered batches of white eggs, in masses of
firm jelly, lying about at the bottom of the tin. I put some of
these in water, and some I kept damp. Those which I put in
water died immediately ; those which I kept merely moist I
watched segmenting and developing until December 5th, and
preserved several eggs of each stage, but on this day the last
of the embryos died, and I tried hard to get some more, and to
find out how they were laid in nature.

On December 81st I discovered a small leaf overhanging a
pool of water, and containing a batch of the Wollunnkukk
eggs. At this same pool I found within the next three weeks
about twenty leaves enclosing batches of eggs, in no case more
than two feet from the water.

On January 15th I had an opportunity of watching the
process of egg-laying. About 11 p.m. I found a female carry-
ing a male upon her back, wandering about apparently in
search of a suitable leaf. At last the female, climbing up the
stem of a plant near the water’s edge, reached out and caught
hold of the tip of an overhanging leaf, and climbed into it.
With their hind legs both male and female held the edges of
the leaf, near the tip, together, while the female poured her
eggs into the funnel thus formed, the male fertilising them as
they passed (fig. 35). The jelly in which the eggs were laid
was of sufficient firmness to hold the edges of the leaf together.
Then moving up a little further more eggs were laid in the
same manner, the edges of the leaf being sealed together by
the hind legs, and so on up the leaf until it was full.

As a rule two briar leaves were filled in this way, each con-
taining about 100 eggs.

The male hurried away immediately the laying was over,
and he did not embrace the female except during the act of
laying eggs. The time occupied in filling one leaf was three
quarters of an hour.

Life History.—Development proceeds very rapidly ; within

316 J. S. BUDGETT.

six days the embryo increases from 2 mm. (the diameter of the
egg) to 9 or 10 mm., when it leaves the leaf as a transparent
glass-like tadpole whose only conspicuous part is its eyes (fig.
30). These are very large and of a bright metallic green
colour, so that when swimming in the water all that is seen
are pairs of jewel-like eyes.

The newly hatched tadpole has also a bright metallic spot
between the nostrils somewhat in front of the pineal spot.
This is the point which touches the surface of the water when
the tadpole is in its favourite position. Whether it is a pro-
tective coloration, or some mechanical arrangement for hold-
ing the surface, I cannot say.

The leaves containing the eggs are not always directly over
water, and the newly hatched tadpole has often to make his
way many inches to the water.

This migration to the water usually takes place during a
shower of rain, when the larve tend to be washed into the
pool, but they also assist themselves by jumping several inches
into the air. They are intensely sensitive to light and shock.

During the embryonic development the jelly surrounding
the embryo becomes more and more dilated by the growth of
the embryo, and also by the accumulation of fluid within.
Towards the close of embryonic life the embryo comes to lie
quite freely within a membranous capsule.

The eggs are very heavily yolked, and some yolk persists
until the tadpoles are ready to leave the capsule.

On the third day external gills are well developed, and the
red blood-corpuscles may be seen coursing through them, and
the heart beating rapidly. These external gills reach their
highest state of development about the fifth day, when they
extend beyond the vent, and are of course bright red (fig. 27).

The tadpole is hatched without a trace of yolk, the external
gills have completely disappeared, there is a median spiracle,
and the lungs are already clearly visible shining through the
transparent body-wall (fig. 30).

The day after the tadpoles are set free, pigment begins to be
developed about the head and upper surface of the body.

NOTES ON BATRACHIANS OF THE PARAGUAYAN OHACO. 317

There is a conspicuous absence of pigment for some time over
the pineal body (fig. 26).

Black pigment appears first, then green. At the end of
about five more weeks the tadpole has begun to develop its
hind limbs. During this period it has grown to a length of
8 cm. The upper surfaces are now a glossy green, beneath
silver and rosy ; the tail is still transparent, and the red blood-
vessels give it a bright red colour (fig. 31).

At the time of the development of the hind limbs there is a
very great accumulation of black pigment at the middle of the
tail, especially below (fig. 31).

The tail is absorbed very rapidly up to this point ; the final
absorption of the proximal part of the tail is postponed for
some days.

The young frog, having now grown both pairs of legs, leaves
the water and betakes itself to the blades of grass close by
(fig. 32).

Here it sits during the time of absorption of the remainder
of the tail. When lying in the blade of grass, only the bril-
liant green upper surfaces are visible, and the tail helps to
make the young frog still less noticeable by shading off the
body, and causing it to become merged in the green of the
grass blade.

The young frog at the close of its metamorphosis is two
thirds the length of the adult frog, and at this time acquires
the red flanks barred with black (fig. 38).

There is a certain stage in the life of this larva when it will
not bear transferring from the pool to aquaria. If the larvee
are transferred at the time when pigment in the tail is just
beginning to accumulate, that is when they are 3 cm. in
length, they invariably die, though both younger and older
larve stand the change quite well.

Development.

External Characters. — Segmentation is _holoblastic,
though not so regular as in Rana and most Batrachia (figs.

318 J./ 8S. BUDGET.

1—6). The blastopore is formed by a more general over-
growth of epiblast, and is from the first circular; it is only just
before closure that it is possible to tell from an external view
which is the anterior and which the posterior side of the pore
(figs. 6,8). It is quite during the last stages of gastrulation
that the closing mouth of the gastrula swings up to the
posterior edge of the egg. When the blastopore has reached
this position it becomes pointed at the anterior end, and there
can now be seen running forward from this point a groove
showing unmistakably the line of fusion of the edges of the
mouth of the gastrula (fig. 9).

Finally the yolk retreats, and a slit-like open blastopore
remains at the posterior end of the pear-shaped neural plate
(fig. 10).

While this fusion has been taking place the centre of
gravity has been continually shifting; for along the line of
fusion there is a greater accumulation of yolkless protoplasm,
i.e. epiblast and mesoblast, than elsewhere. Yolk is heavy,
therefore the fused edges of the mouth of the gastrula come to
the upper side.

Finally the entire egg has rotated through 180°. The
anterior end of the archenteron is now in the position that the
posterior end occupied at the beginning of gastrulation. The
area occupied by the neural plate has been formed chiefly by
the downgrowth of the lateral and anterior edges of the
blastoporic rim; however, the posterior edge here takes a
greater share in gastrulation than in Rana, and in consequence
the blastopore comes to lie not at the extreme posterior end of
the archenteron, but further towards the middle, while the
neural plate extends beyond the blastopore. The anus, how-
ever, makes its appearance at the extreme posterior end of the
archenteron, far from the position now occupied by the blasto-
pore (Sections II, III, az.).

The centre of the neural plate becomes slightly depressed,
and here the blastoporic scar is seen running forwards from
the edge of the blastopore along the whole plate as the
“‘ primitive streak ”’ (fig. 11).

NOTES ON BATRACHIANS OF THE PARAGUAYAN OHACO. 3819

The neural folds now begin to approach one another at the
anterior and posterior ends of the groove, but there is no well-
marked anterior transverse fold (fig. 12).

Posteriorly the folds enclose the remains of the blastopore,
which then opens only into the neural canal formed by the
complete fusion of the edges of the two foids. A tail fold
develops of a crescentic form encircling the posterior end of
the neural plate, on the posterior convex side of which the
anus is formed (Section III, a@z.).

Between the blastoderm and the egg membrane there is now
present a considerable space, filled with a milky fluid (fig.
12, sp.).

When the neural folds have completely met, i.e. fifty hours
after laying, then the anterior end of the neural plate expands
to form the optic vesicles, and an elevation extends forwards
from them, homologous with the so-called ““Sense-plate” of
Morgan. Behind the optic vesicles extending laterally and
anteriorly on either side is seen the gill-plate or branchial fold.

Later this grows to completely encircle the sense-plate, which
now shows a depression at the anterior end, the rudiment of
the stomodzum (fig. 20, Stom.).

The right and left halves of the “‘ sense-plate ” thus divided
are very conspicuous features at this stage in development, and
for some time later. A little later they become formed into a
regular pair of mandibular bars, which only just meet below
the stomodzum (fig. 21, Mnd.).

In section they appear quite like the succeeding hyoid
arches, which are very slightly developed, and also the larger
first and second pair of branchial arches.

There is a total absence of suckers such as are borne behind
the mouth in most Batrachian larve, and the embryo has now
more the appearance of a young larva of Acipenser than of
Rana.

The gill-p late in life appears as a single elevation on either
side, but after fixing with appropriate reagents it may be seen
almost from the first to consist of three branchial pouches of
the pharynx ; the two anterior of these alone persist. The

320 J. 8S. BUDGEET.

first pouch is between the hyoid and the first branchial arch.
The second pouch is between the first and second branchial
arch. The third pouch is between the second and third
branchial arch (fig. 14, 3rd br. f.).

The optic bulbs early begin to bud out from the fore-
brain, and now just behind the gill-plate is seen the first rudi-
ment of the pronephros, a slight but defined elevation tapering
posteriorly ; mediad to this are seen four or five mesoblastic
somites (fig. 14, mes. som.).

The auditory vesicles are not easily visible until after the
appearance of the external gills.

Up to this time the embryo has lain almost flat upon the
surface of the yolk, preserving in all a spherical form; now,
however, it begins to rise from the surface of the yolk, both
anteriorly and posteriorly, but the yolk is still nearly spherical
(fig. 15).

The eye-bulbs increase greatly in size, and are exceedingly
large in comparison with what is found in Rana at a corre-
sponding stage. ‘lhe ocular muscles are developed very early,
and the eye may be seen to be rotated by them on the fourth
day of development. A very conspicuous feature of this stage is
the dilated condition of the double gill-pouch (fig. 15). Viewed
from the dorsal surface, the head region has now in outline the
form of a trefoil.

The tail now begins to grow back from the surface of the
yolk, the dilation of the branchial folds ceases, and in propor-
tion to the latter the head portion increases greatly (fig. 16).

The first pair of external gills now may be seen budding out
from the first branchial arch. Below the cleft post-oral region,
formed from the sense-plates, the rudiment of the heart is
clearly visible (figs. 21, 24, At.).

In a side view, more or less transparent as in life, there are
to be seen the heart, first pair of external gills, well-formed
eye with conspicuous choroid fissure, auditory vesicles, somites,
caudal notochord, and extended cloaca. The yolk-sac still
retains its spherical form (fig. 24).

The changes in external form which now take place mainly

NOTES ON BATRACHIANS OF THE PARAGUAYAN OHACO. 3821

consist in the appearance of the second pair of external gills,
which do not reach nearly so high a state of development as
the first pair; also in the rapid growth of the first pair of gills,
so as to extend beyond the vent as blood-red filaments through
which the corpuscles stream along, propelled by the now rapidly
pulsating heart (figs. 25, 27, 17—19).

A dense plexus of vitelline veins ramifies over the surface of
the yolk, while the dorsal aorta and cutaneous veins give to the
elongated tail a copious supply of blood (fig. 29). Indeed, so
noticeable is this, that I am quite inclined to agree with Mr.
Kerr’s suggestion that the tail of this larva is an important
organ of respiration. This view is further strengthened by
my observation that in hatched larve the tail often remains
motionless as a whole, while the extremity of the tail is
kept rapidly vibrating. As the larve are not propelled
by this motion through the water, I am tempted to think
that the object of it is that a stream of water may be
kept constantly running along the surface of the proximal
part of the tail.

The operculum now grows down from the hyoid arch and
encloses the gill arches. The external gills become rapidly
absorbed (but I think that a study of the origin of the fila-
ments of the internal gills shows them to be really of the same
nature as the external gills), the stomatodzal aperture breaks
through, and the young frog has reached the end of its em-
bryonic life (fig. 26).

Internal Characters.—On account of the short space of
time at my disposal it seems advisable not to attempt a con-
tinuous account of the internal phenomena of development,
but merely to figure and describe sections illustrating some of
the more important points of interest, leaving the fuller
account for a future occasion.

Section I? passes transversely through the blastopore before
the formation of the neural groove. The main points to be
noted are the smallness of the archenteron, the absence of a

1 The numbers here given correspond with those of the figures of Plates
31 and 32.

B22 J. 8. BUDGETT.

yolk-plug, the abundance of yolk, and the mesoderm extending
only as far as the equatorial region of the egg.

Section IT passes longitudinally through the blastopore (6/.),
the walls of the neural groove being closed anteriorly, but not
yet posteriorly. The points to be noted are the anterior posi-
tion of the blastopore, the fusion of the embryonic layers before
and behind the blastopore, and anteriorly the beginning of the
first branchial pouch (dr. f.).

Section IIT, a sagittal section of an embryo after the closure
of the neural groove, showing the comparatively anterior posi-
tion of the neurenteric canal (”. en.), the brain vesicles, the
notochord not yet differentiated posteriorly, the archenteron,
and anteriorly the branchial fold (dr. f.) of the pharyngeal
wall, which is continuous across the middle line; also
posteriorly the depression which will later become the anus
(an.).

Section IV is of the same series as the last, but further from
the middle line, showing the large lateral cavity of the arch-
enteron caused by the upraising of the branchial pouch (67. f.).
The section also passes through the optic vesicle (op. ves.).

Section V is a transverse vertical section passing through
the head region of an embryo in which the body of the embryo
is just beginning to rise up off the yolk. ‘To be noted are the
regularly formed optic vesicles (op. ves.) and stalks, there being
as yet no trace of the lens. Below is seen the pharyngeal
region of the archenteron (ph.).

Section VI is of the same series as the last, passing between
the optic and auditory region; it shows the single branchial
pouch (dr. f.) and the accompanying epidermal thickening.

Section VII of the same series, passing through the auditory
region. There are seen in section tbe front end of the noto-
chord (NV. ch.), the auditory thickening of the epidermis (Awd.)
and the rudiment of the pronephros being differentiated off from
the general mesoderm (Px.). There is seen also a faint indi-
cation of a neural crest (NV. er.).

Section VIII of the same series, passing through the pos-
terior end of the archenteron, where it is seen to be constricted

NOTES ON BATRACHIANS OF THE PARAGUAYAN CHACO. 323

into two portions, the upper being the opening of the neuren-
teric canal (NV. en.), the lower the rudiment of the rectum
(Rect.).

Section IX shows the fusion of the layers in the region of
the neurenteric canal, and the separation of the latter from the
rectum.

Section X of the same series, through the tail and vent,
shows the opening of the neurenteric canal into the neural
tube; also the fusion of the epidermis with the hypoblast at
the anus (A7z.). ;

Section XI, a transverse vertical section quite at the an-
terior end of the head of an embryo, in which the first pair of
external gills are beginning to bud. The section passes
through the bottom of the stomodzeum (Stom.), and obliquely
through the mandibular arches at the point where they meet
(Mnd.).

Section XII is of the same series, but further back, and on
the right side in the figure passes through the centre of one of
the eyes, showing the attenuation of the posterior wall of the
optic cup (Op. w. p.) and the thickening of the anterior wall
(Op. w. an.), the rudiment of the retina. The lens is also
seen arising as a regular involution of the nervous layer of the
epiblast, the epidermal layer (/.) remaining stretched across as
a very thin membrane. The section also passes through the
middle portion of the mandibular arches (Mnd.). The
pharynx and pericardium are also cut through (Ph., P.c.).

Section XIII of the same series passes through the centre
of the opposite eye. The proximal parts of the mandibular
arch are here cut through (Mnd.), The formation of the peri-
cardium and heart, with its mesodermal membranous lining,
is well seen (P.c., At.).

Section XIV of the same series passes a very little further
back through the infundibulum, pharynx, and the two lateral
extensions of the pericardium overlying the sinus venosus
(S. v.).

Section XV is of a slightly older embryo, passing trans-
versely through the eyes. The lens is now nearly completely

324 J. 8. BUDGETT.

constricted off, and the back wall of the lens has begun to
thicken and fill up the hollow of the lens (/.).

Section XVI is a sagittal section through the pineal eye of
an embryo about two days before hatching. It shows the
pineal stalk, still allowing free communication between the
pineal body and the brain-cavity; this passage is now dis-
tinctly ciliated (cd. st.). Blood-sinuses are seen in front and
behind.

Section XVII is a transverse section of an embryo just
before hatching. It passes through the root of the first ex-
ternal gill (zt. G.), and shows the developing first and second
internal gills (Iné. G.).

From this and similar sections there certainly does not seem
to me to be any very marked difference in the nature of the
external and internal gills.

As regards the development of the Pronephros and its
duct, my sections indicate that there is considerable variation.
Though by the time the external gills are developed there are
invariably three nephrostomes, as in Rana, the first and third
being lateral, the second dorsal, yet previously to this stage
I find often only two nephrostomes, and in some instances two
on one side with three on the other, and in one case but one.
This seems to me to indicate that the pronephric tubules do
not arise in the way usually described for Rana, namely, by
the primitive pronephric groove becoming a closed tube and
remaining in open communication with the coelom at three
points, but rather as a solid rod of mesoderm (Section VII),
which later becomes hollow and acquires perforations into the
celom, at first one, later three points—the nephrostomes.

In comparing the development of Phyllomedusa hypo-
chondrialis with that of Rana, Bombinator, Pelobates, and
other Batrachians with free-swimming larve, the first thing
that strikes one as regards external characters is that, through-
out, this embryo maintains a greater similarity to ichthyic
forms, especially Ganoids, on the one hand, and to the Uro-
dela on the other, than do the free-swimming larve of other
Batrachians.

NOTES ON BATRACHIANS OF THE PARAGUAYAN CHACO. 325

Again, we find this difference in general development of
the young larva intensified in such forms as undergo a
still more abbreviated embryonic development; for instance,
in Paludicola fuscomaculata, where the embryonic de-
velopment is shortened to something between twelve and
twenty-four hours. All the points in which Rana appears
to be a more modified form of development than Phyllo-
medusa are intensified, and the external characters are ill-
defined. However, a minute comparison cannot yet be made
until I shall have had time to study more carefully the details
of development in Paludicola. The study of the internal
anatomy leads to the same conclusion, namely, that in this
protracted development we do not find the course of develop-
ment distorted and blurred, but on the contrary every organ,
so far as I can find, develops as in the ordinary frog, only
more clearly and more definitely, and at the same time more
as we see it develop in other great groups, Elasmobranchs,
Ganoids, and the higher Vertebrates.

Take for instance the eye of a free-swimming batrachian
larva, and compare it with the eye of Phyllomedusa. The
evolution of the optic cup and lens is hurried over and
blurred in the former, so that they are often difficult to trace,
while here in Phyllomedusa it is as regular and diagrammatic
as in any Vertebrate there is. Contrary to what we find in
most Batrachia, the lens develops as an involution of a single
layer of nervous epiblast rather than a mere thickening of that
layer. In the free-swimming form the eye has been required to
become functional as rapidly as possible, while here it has been
suffered to go through its normal course of development in peace.

Take again the suckers of the free-swimming forms.
They are evidently new adaptations without phylogenetic sig-
nificance. Through the presence of these structures the form
of the mandibular arches has become quite obliterated, while
here in Phyllomedusa these would compare favourably with
those of an Elasmobranch, reptile, or bird.

The peculiarly symmetrical gastrulation that this egg
exhibits must be supposed, I think, to be the effect of a large

VOL. 42, PARY 3,—NEW SERIES, Y

326 J. S. BUDGETT.

amount of food-yolk, as it can hardly be supposed that, at a
stage previous to hatching in either mode of development,
Phyllomedusa should be more primitive than the free-swim-
ming forms.

I think the median spiracle may also be looked upon as
a primitive feature.

The manner in which the branchial fold encircles the
head reminds one strongly of Salensky’s figure of Acipenser
at a similar stage.

From the study of the development of Phyllomedusa, of
which I have described the points of more general interest, I
am distinctly inclined to think that we are not always war-
ranted in attributing to alecithal free-swimming larve a
greater biological importance, as far as retaining ancestral
characters is concerned, than to heavily yolked embryos.

I think, moreover, that this is what we should expect, for
from the time that the larva is hatched onwards it is subjected
to the influence of natural selection.

Indeed, in this particular case of Batrachian development it
would seem rather that the shortening of the embryonic period
may be a specialised and not a primitive condition.

The fact that the majority of frogs have a shorter embryonic
life does not seem to me to prove that the minority are the
specialised forms in this respect. This particular mode of
development is not confined to this species.

Von Jhering has described the oviposition of Phyllome-
dusa Jheringii, which agrees very closely with that here
described. The eggs were laid between two or more leaves
instead of being rolled in one, as with Phyllomedusa hy po-
chondrialis. Von Jhering did not, however, work out the
development of this species; in all probability it would not
differ from this one.

This year S. Ikeda, of Tokio, has published an account of
the oviposition in a species of Rhacophorus; from what he
mentions of the appearance of the embryos which develop in a
froth, much as is the case with Paludicola, I think the deve-
lopment of this form will be found to be quite like that of

NOTES ON BATRACHIANS OF THE PARAGUAYAN OCHACO. 3827

Phyllomedusa ; indeed, Professor Mitsukuri, who has seen
them both, assures me that this is so.

To a paper by Gasser published in the ¢ Sitz. d. Kon. Ak.
Marburg,’ 1882, upon the development of the midwife toad,
Alytes obstetricans, I have not yet been able to get access,
but I feel quite prepared to find that it exhibits the same
features that characterise the development of Phyllo-
medusa hypochondrialis.

XVI. PHyLtomEDusA Sauvaatt, Boul.

This handsome tree-frog was brought to me in the Chaco,
but [ am not able to state anything about its habits.

XVII. Hyza specazinit, Boul.

This fine Hyla was fairly common ; I often caught or saw
young specimens swimming from stem to stem of the Papyrus
grass as we travelled through the reed-choked swamps. The
full-grown specimens, however, were always taken either from
palm tops just felled or from the trees overhead.

When caught in the water by daylight they were a bright
light yellow, but at night they turned to a darker shade, and
became marbled on the upper surface with brown markings.
The full-grown specimens did not in this way become dark at
night.

The largest specimens taken measured 80 mm. The eggs in
the cloaca appear to be quite like those of Rana in size and
colour, and are probably laid and reared in the same way.

One full-grown specimen I obtained in Central Paraguay on
the Tibicuari, the rest in the Paraguayan Chaco.

XVIII. Hyzta venvutosa, Laur.

In life the markings are olive-green or grey upon a whitish
ground. When taken from amongst foliage the whitish
ground colour is suffused with green. It is a powerful and
energetic frog, the large toe-discs having a tenacious sucking
power.

328 J. S. BUDGETT.

The skin glands are strongly developed, emitting a very
sticky white slime.

XIX. Hyta nana, Boul.

This small frog was abundant in the swamps, usually found
by moonlight sitting on the broad-leaved plants of the swamp,
and calling with a rather highly pitched scraping note.

The upper surfaces, as in H. spegazini, are light straw-
colour by day, but brown by night. Flanks and underneath
pigmentless.

XX. Hyta PHRYNODERMA, Boul.

A light golden colour, shaded with darker above. The discs
fairly strongly dilated. The skin is warty and extremely
delicate, and it is not easy to catch one uninjured.

They are not common, but make themselves known by their
constant call, which is just like the quacking of a duck. All
the specimens I obtained were about the palm fencing and
sheds.

XXII. Hyta nasica, Cope.

This is the most common Hyla in the Chaco. It is found
everywhere, usually upon palm or palm fencing, where it is
most inconspicuous.

Its call and habits are much like H. phrynoderma; the
note is, however, somewhat lower. The colours are chiefly
olive-green and brown, but the markings are variable. It is
of more slender build than H. phrynoderma, the body being
longer in proportion to the width of the head.

XXII. CeratTopurRys ornaTA, Bell.

I obtained some half a dozen specimens of this curious and
well-known South American frog, commonly known as the
“Escuerso.”

Its ferociousness is its most striking characteristic, If it is

NOTES ON BATRACHIANS OF THE PARAGUAYAN CHAOCO. 329

approached to within two feet, it will make a vicious spring at
one with its gigantic mouth wide open. If it succeeds in
seizing any part of its tormentor, it holds on like a bulldog.
The habit it has of distending its lungs to their fullest when
teased has given rise to the idea amongst the Argentine
people that if teased sufficiently it will burst.

It is perhaps needless to say that I was disappointed in my
efforts to obtain this end. The Ceratophrys lives chiefly off
other frogs and toads, but it is said also that it will seize and
devour young chickens. The largest I saw was 120 mm. in
length. |

LerrposatTracuus, J.S.B. N.g.

Pupil horizontal. No vomerine teeth; transverse processes
of sacral vertebre not dilated. Large teeth in upper jaw;
also two large teeth in dentaries of lower jaw. Tongue cir-
cular and free behind. Nostrils the most elevated portion of
the head. Eyes close together, not more than the diameter
of the eye apart. Fontanelles in the parietal region. Outer
metatarsal tubercle very large. Great development of mem-
brane bones in the head ; width of jaw very great. Tympanum
fairly distinct.

XXIII. Leripopatracuus AsPpER, J.8.B. N. sp.

Hind legs carried forward, toes reach barely to the eyes.
Tips of toes horny. Skin of dorsal surface a dull leaden
colour, much tuberculated and tough.

This frog lives continually in muddy pools. Its habit is to
float with just the eyes and nostrils above the surface. If
disturbed it slowly sinks to the bottom, leaving no ripple on
the surface of the water. It feeds largely on Bufo granu-
losus.

XXIV. LeripopatrRAcuus Lavis, J.8. B. N. sp.

This may possibly be the same species as the last, but

330 J. S. BUDGETT.

differs from it in the greater width of the skull, greater length
of the hind legs, which carried forwards reach tip of snout, and
in the skin being smooth, thin, and slimy, with the organs of
the lateral line showing clearly upon it. Also the tympanum
is larger and more evident. The tips of the toes do not bear
horny caps as in the preceding species.

Below is a comparative list of measurements in millimetres
in two specimens of XXIII and one specimen of XXIV.

Total Length. Hind Legs. Width of Jaw. Eyeto Eye. Eyeto Har. Har to Har.

BOON or 62 . d4 3 4 : 8 - 24
‘aaa { B10). 60 7, Soo : Beet TES ca 23
XXIV. 1c. 60%: 70 ests) : 5 : 9 : 28

It is a source of great regret to me that I am obliged to
abandon for the present my work in this direction. I have a
considerable amount of material at my disposal of the develop-
mental stages of several of the species of frogs, concerning
which I have here merely stated the observations which I
made a note of while yet in the Paraguayan Chaco. I sin-
cerely hope that I may be able to return to this work at a
future date.

Concerning the species Phyllomedusa hypochondri-
alis I should state that, although I have gone more fully
into its development than others of my collection, here also my
work has been cut short.

In concluding, I should like to say that I am very greatly
indebted to Mr. Graham Kerr for the opportunity he afforded
me of obtaining my material, and also for much help and
advice in my work.

NOTES ON BATRACHIANS OF THE PARAGUAYAN CHACO. 331

EXPLANATION OF PLATES 28—82,

Illustrating Mr. Budgett’s “ Notes on the Batrachians of the
Paraguayan Chaco.”

All the figures relate to the same species—Phyllomedusa
hypochondrialis, Cope.

Pirate 28.
(See next page.)

PLATE 29.

The figures on this plate are all drawn unaer a magnification of eighteen
diameters.

Fics. 1—6.—Illustrating the character of segmentation. Figs. 1—4 are
views from above; Figs. 5 and 6 from the side.

Fies. 7 and 8.—Views of egg from below, showing diminution in size of
the blastopore.

Fic. 9.—Hgg seen from below, at a time when the blastopore is much
reduced in size, and has nearly reached the level of a horizontal plane passing
through the centre of the egg. From the pointed anterior end of the blasto-
pore there passes forwards a distinct groove, indicating the line of fusion of
the gastrula lips.

Fic. 10.—View of egg from above and behind, showing the continuation
forwards of the slit-like blastopore as a faint groove along the axis of the
medullary plate. f

Fic. 11.—View of egg from above, showing neural plate and early condition
of neural folds. ;
PLATE 30.

The figures on this plate are all drawn under a magnification of eighteen
diameters.

Fie. 12.—View of anterior end of embryo, with well-formed neural folds.
Sp. Space between embryo and egg membrane.

Fic. 13.—Similar view where the neural folds are arching over towards one
another.

Fie. 14.—View of middle of trunk region of an embryo in which the pro-
nephros has appeared on each side (p.z.). | Mes. som. Mesoblastic somites.
3rd Br. f. Position of third branchial pouch.

Fies. 15—19.—Figures illustrating the further development in general form
of the embryo.

332 J. S. BUDGETT.

Fic. 20.—View of anterior end of embryo, showing the first trace of stomo-
deum (Stom.).

Fie. 21.—Oblique view of embryo, showing the mandibular bars (Mnd.)
and rudiment of heart (A¢.).

Fies. 22 and 23.—Views of anterior end of head, showing the fusion of the
mandibular bars (mad.) in the mid-ventral line.

Fics. 24—26.—Side views of larve, showing the further changes in form
up till the time of hatching. ¢. in Fig. 24, rudiment of heart. In Fig. 25
the external gills are at about their maximum.

PLATE 28.

(The explanation is given here as the figures run on from Plates 29 and 30.)

Fics. 27—80 are drawn under a magnification of eight diameters.

Fies. 31—35.—Natural size. It will be noticed that Fig. 35 has been
wrongly orientated by the lithographer; the plant stem should be vertical.

Fies. 27—29.—Side view of larve during the last day of intra-oval develop-
ment.

Fic. 30.—Larva just after hatching.

Fic. 31.—Side view of larva at time of development of hind limbs, showing
accumulation of pigment in the tail.

Fie. 32.—Young frog after leaving water.

Fic. 33.—Young frog after completed metamorphosis.

Fic. @2—Pair of adults during the process of oviposition. ERY

Fie. 35—Adult specimen. This figure by comparison with Fig. & illus-
trates the extent of reflex colour change.

PLATE 31.

Fic. 1.—Trausverse section through blastopore before formation of neural
groove. 01. Blastopore. ep. Epiblast. mes. Mesoblast. Ayp. Hypoblast.

Ki1e, 2.—Longitudinal vertical section of embryo with neural groove closed
in anteriorly. 27. Blastopore. 4. Depression marking position where anus
will appear. Arch. Archenteron. Mes. Mesoblast.’ Br,f. Branchial out-
growth of archenteron.

Fic. 3.—Longitudinal vertical section of an embryo after the closure of the
neural groove. V.en. Neurenteric canal. az. Anal depression. otoch. Noto-
chord. yp. Hypoblast. Br. Branchial outgrowth of archenteric wall.
Ves}, Ves \, Ves i, Brain vesicles.

Fic. 4.—Section parallel to the last figured, but more lateral in position.
Op. ves. Optic vesicle. Br,f. Branchial outgrowth from archenteron.

NOTES ON BATRACHIANS OF THE PARAGUAYAN CHACO. 333

Fic. 5.—Transverse section through head of an embryo which was just
beginning to be folded off the yolk. Op. ves. Optic vesicle. P#. Pharynx.
mes. Mesoblast.

Fic. 6.—Section of same series as that shown in Fig. 5, through the single
branchial pouch (Br,f) and the ectodermal thickening accompanying it.

Fic. 7.—Section of same series through auditory region. dud. Commencing |
auditory invagination of ectoderm. W.ch. Notochord. Neur. cr. Neural crest.
P.n. Rudiment of pronephros.

Fie. 8.—Section of same series through posterior end of archenteron.

Rect. Rectum. NV. ex. Neurenteric canal opening into this. Ayp. Hypoblast.
N.ch. Notochord.

Fig. 9.—Section of same series further back. .ex. Neurenteric canal.
Rect. Rectum.

Fie. 10.—Section of same series showing opening of neurenteric canal (JV.
ex.) into neural canal; also anus (4z.).

Fig. 11.—Transverse section through anterior end of an embryo, in which
the first pair of external gills were beginning to develop. Stom. Cavity of
stomodeum. Mad. Mandibular arch close to its junction with its fellow.

Fig. 12.—Section of same series passing through the rudiment of the eye.
Op.W.an. Anterior layer of optic cup. Op.W.p. Posterior wall of ditto. JZ.
Outer layer of epiblast passing continued over mouth of lens invagination.
Ph, Pharynx. P.c. Pericardium. mzd. Mandibular arch.

PLATE 32.

Fic. 13.—Section of the same series as that shown in Plate 31, Fig. 12.
mnd, Mandibular arch. 4¢. Heart. P.c. Pericardium.

Fic. 14.—Section from same series through infundibulum (Ivf). P.c. Peri-
cardium. Sv. Sinus venosus.

Fic. 15.—Transverse section through head of slightly older embryo, show-
ing later stage in the formation of the lens (/.).

Fie. 16.—Sagittal section through pineal body (Piz.) of embryo about two
days before hatching. Cz. S¢. Pineal stalk with ciliated lining. Siz. Blood-
sinus.

Fic. 17.—Portion of transverse section of embryo just before hatching,

passing through the origin of the first external gill (#z¢.G.). Int. G. Internal
gill.

ee
- iv . a
> -
eo
wie St ay -
*,
1 2 y
= ral =
2

re
aa 7 t
a = Pee
kana
ar Aye. " ae =
- ee! ey sss
t ir he } a
eh hae ez
-- P er)
Pr = t inn
= ; =v. 7)
= 7 7 owe

Py api As ADibatas

‘ ‘ = PT Tl

ye us idl prong idee

je
- ¢ e bi
Lect ine? PIF ee
s ’
= - - t
= 7%
Bie, “Al of a Pr uh2 es
ares eh ox AAS _
teh) Paved 24 Pn ARSE re
oe veil tS \ ae >
‘ it
‘
© ~
4 : r > met ’ “Sls
" : .
= i ty. “ihe ifn +
é
a s
¥
yi rie : er + |
i
-
- , i 4 .
y dma
— ®
= Fe uy « &.
A
its pa
‘ ‘
a? ae i] at? om
‘
:
a: Re ag 8S | Pa od
* 4
om ’
’
, = a Ay
=
‘.
= ‘
_ + .
“
*\ wl (heats, §
«
*
= is
;
a ~ .
a ‘
; “y *
J
\
. -
7
cee a)
“a nee -.

Poeun fh of « he
.
%
-
Tr
4% = =
2 od ph
tie ¥
- chai ae
ie ei ‘
P > (ee
:
*L
t
~
\!
a)

or ir Gd rien P K,

inn? (ES eee he
4) J aii ie i .
Lido’ one Oe ee
- Fe ie =
* birt bStule eed

ce
-

THE DEVELOPMENT OF ECHINOIDS. 300

The Development of Echinoids.

Part IL—The Larve of Echinus miliaris and Echinus
esculentus.

By

E. W. MacBride, M.A.,
Professor of Zoology in McGill University, Montreal.

With Plate 33.

Tue development of Echinoids is up till the present very
imperfectly known, our information on the subject being in a
most unsatisfactory condition. The foundation was laid by
Johannes Miiller (3), but although he observed and described
the external features of the metamorphoses, he was unable to
refer with any certainty any one of the larval forms which he
described to the adult species from which it was derived. So
far as I am aware, Prof. Théel (4) and H. Bury (1) are the only
persons who have hitherto reared any species from the egg until
the conclusion of the metamorphosis. Through the researches
of the former we have obtained an accurate knowledge of the
characters of the larva of Echinocyamus pusillus at all
stages ofits growth. The youngest larval forms of many species
have been determined by rearing, nevertheless Mortensen (2),
in his most valuable summary of all descriptions yet pub-
lished of Echinoid larve, points out that our knowledge of the
later pluteus stages is most imperfect, and that it is not possible
to assign the older plutei which have been described to the
species from which they have been derived.

336 E. W. MACBRIDE.

Some three years ago I formed the project of making an ex-
haustive study of the development of one of the British species
of Echinus on similar lines to my work on the development of
Asterina gibbosa. After several unsuccessful attempts I
succeeded this spring in rearing the larve of Echinus escu-
lentus and Echinus miliaris from the egg to the latest
pluteus form, when the processes (arms) are all fully developed,
and the first pedicellarize of the adult have already appeared. In
every stage of their development these two larve are easily dis-
tinguishablefrom one another, and, in view of the interest attach-
ing to the question as to how far allied species are distinct at all
stages of growth, it seemed worth while to publish an account
of these larve before dealing with the more general questions
of Echinoid development. My best thanks are due to Mr. Allen,
the Director of the Plymouth Biological Laboratory, and to
his staff, for the assistance and advice they rendered me. To
the mechanical arrangement for continuously agitating the
water, invented by Mr. Allen, such success as I have hitherto
obtained is in large measure due.

The eggs of Echinus miliaris are smaller and more trans-
parent than those of E. esculentus, and as a consequence the
stage to which it is possible to rear them without special pre-
cautions is much less advanced than is the case with E. escu-
lentus. lLarve of E. miliaris have lived for a month without
showing unhealthiness, but also without developing a trace of
the oral disc (the first trace of adult structure to appear), or
even the full number of larval processes. The stage at which
larve of E. esculentus under similar circumstances cease to
progress, is that at which there is an unmistakable rudiment of
the oral disc of the adult. This curious difference throws light
on the extent to which the larve depend for support on material
stored up in the ovum.

The blastule of E. esculentus are nearly spherical ; those
of E. miliaris, on the contrary, distinctly ellipsoidal in form.
When the invagination which forms the gut has taken place,
it is seen that the oval outline of the latter larve is due to
their possession of an enlarged preoral lobe. This enlarged

THE DEVELOPMENT OF ECHINOIDS. Bat

forehead, as it may be called, long remains a feature of the E.
miliaris larve ; it is seen as a lip overhanging the mouth in
plutei with four processes completely developed (see fig. 1, fr.).

In addition to this feature the four-armed pluteus of E.
miliaris is distinguished from that of E. esculentus by its
more pointed posterior portion, and by the smaller length of
the processes in comparison to the body (comp. figs. 2 and 3).

In later stages the posterior pole of the E. miliaris becomes
rounded, but then the pluteus is broader in proportion to its
depth than is the case with EH. esculentus, approximating
more to the shape of an Ophiurid pluteus, or ophiopluteus, as
Mortensen terms it.

The ciliated epaulettes develop about the middle of the third
week of larval life. A new distinctive feature is now added to
those already possessed by E. miliaris. Just anterior to each
epaulette there is a large mass of bright green pigment, The
epaulettes of E. esculentusare, on the other hand (pig., fig. 4),
loaded with reddish-yellow pigment.

So far as I could observe, the four primary epaulettes at no
time form part of the general ciliated band, but are of inde-
pendent origin. In the pluteus of Echinus esculentus,
in addition to these, however, in the fourth week two posterior
ciliated epaulettes are formed, one on each side by abstriction
from the ciliated band just where it bends from the postero-
dorsal to the post-oral process (Mortensen’s notation). I did
not succeed in bringing the larve of KE. miliaris quite as far.
These posterior epaulettes are mentioned for the first time in
Mortensen’s work cited above; they are there described in two
undetermined forms of larve. The drawings are very poor,
and it is not possible to be sure whether these larve are
identical with one or other of the forms I have described, but
it is worthy of notice that E. esculentus is placed by Mor-
tensen amongst forms distinguished by not possessing these
epaulettes. It is therefore quite possible that here, again, the
old error has been made of mistaking two stages in the same
life-history for two different species.

The oldest stage of E. esculentus which I succeeded in rear-

338 E. W. MACBRIDE.

ing is represented in fig. 6. In this stage of E. esculentus
there are three pedicellariz, one at the posterior pole and two
on the right side posterior to the ciliated band; the oral disc
of the adult has encroached very much upon the stomach, and
both the anterior and the posterior ciliated epaulettes are well
developed: the posterior by coalescence have formed a con-
tinuous ring. The cilia covering these epaulettes are much
more powerful than those on the ciliated band, and the epau-
lettes form in later larval life the main organs of locomotion ;
their appearance in the living animal when expanded recalls
the trochal dise of Rotifera.

The “echinoplutei” larve (to use Mortensen’s term) are
distinguished from the more primitive bipinnariz by an im-
mense reduction of the proral lobe (Mortensen’s frontal
area). Under these circumstances it is interesting to find a
remnant of this primitive structure surviving in the larve of
E. miliaris, which develop from small eggs, and it is further
interesting to note that this primitive feature is most strongly
marked in the early stages of development.

Montreal; Oct. 15th, 1898.

WoRKS REFERRED TO IN THIS PAPER.

1. Bury, H.—‘‘ The Metamorphosis of Echinoderms,” ‘ Quart. Journ. Mier.
Sci.,’ 1895.

2. Mortensen, Tuo.—‘ Die Echinodermen-Larven der Plankton Expedition,’
Kiel and Leipzig, 1898.

3. Miter, Jon.—* Die Larven der Nchinodermen,” several papers, ‘ Ab-
handlungen der Kgl. Akademie der Wiss. zu Berlin,’ 1848—1855.

4, Tube, H.—< The Development of Echinocyamus pusillus,” ‘Trans.
Roy. Soc. Upsala,’ 1892.

>

THE DEVELOPMENT OF ECHINOIDS. 339

EXPLANATION OF PLATE 33,

Illustrating Mr. E. W. MacBride’s paper on “ The Develop-
ment of Echinoids.”

List of Abbreviations employed.

a. ep. Anterior ciliated epaulette. az. Anus. ca. cceelomicsac. Heh. Rudi-
ment of oral dise of young Echinus. /r. Frontal area. m.p. Madreporic pore.
ped. Pedicellarie. yp. ep. Posterior ciliated epaulette. pzg. Mass of green
pigment. pr. pre. o. Preoral process of ciliated band. pr. ant. lat. Antero-
lateral process. pr. post. dors. Postero-dorsal process of the ciliated band.
pr. post. o. Post-oral process of the ciliated band. [These last four names are
Mortensen’s notation for the pluteus arms.| ect. Intestine.

The magnification of all the figures is about that obtained by a Zeiss, obj. A,
oc. 2.

Fig. 1.—Larva of Echinus miliaris six years old, side view. Note the
preoral lobe (/r.) overhanging the mouth.

Fie. 2.—Larva of Echinus miliaris in the second week, dorsal view.
rect. Intestine. az. Anus, seen through; third pair of arms beginning to
appear.

Fic. 3.—Larva of Echinus esculentus seven days old, dorsal view.
Commencement of third pair of processes.

Fic, 4.—Larva of Echinus miliaris three weeks old, dorsal view. ped.
First rudiment of a pedicellaria. pig. Characteristic green pigment.

Fie. 5.—Larva of Echinus esculentus three weeks old, dorsal view.
Ech. First trace of oral dise of adult.

Fic, 6.—Larva of Echinus esculentus four weeks old, dorsal view.
p.ep. Posterior ciliated epaulettes.

¥

a : >= :
¢ ee > eID 8a
> ss =i
ec. gs Rese _ _ 4 p"
* -
_ - =a
‘ «
- 6 e ’
cf? - 2 as vw e
> a * 7 eof aut Bat =>
: . - ' + r es
. <i = j 7 = j fap, 2 42y &
> = 7 ‘ ~ ~ P be
st * y risk
»
* 4 ét
* _ = yy, r) 4é ‘ a
. . i J i ra
‘ * ‘ . ; who
= a ; ania 2 _ = *h
- 2 A é » f ke 7 ’ A
. Mi mt) \ é < Pig ij a) ore, ~*~
i Le = ot, ohaeier Tere mates Ae
. , oe | : 7 7 nm
rée Tul Whew tit: (ube Ba
: ey : 2 pret re ~ st .
a, PRE v eat SRSA ae

a x : ¥ . ail ia eae
’ } A : it4e) itty oes 2S ee
, : 1 ‘

; rai i’
ir, ne

i ‘1 F ei ie idee i

= * he .-T 4
f Pe A . ‘ a . mA hc it 8
7 fy = ae i= 2o4 th) cee RD eae
. af Rol) : hntres nhs i 2a iu
‘oe és _. IMEI
i £ ? | vi f
‘
. fi . } ~ i vi4% bec 74 ¢
ace i aaa i} j Va
— . -

. Le sen i hey ;
o
: ‘ oc peo ae ones

; ; - Meal Tees Wi a0

: ee i ; ; Yee 77 ie
: . ae

ae oh

Ne
Bie &

’ ’
a
e =5 :
é 7 <=
. : :
. >:
+ ;
-
>
.
.
< , .
‘ ~
.
. .
. -
~ “9
a - ba -
~ , -
F -

HYDROIDS FROM WOOD’S HOLL, MASS. 341

Hydroids from Wood’s Holl, Mass.

Hypolytus peregrinus, a New Unattached Marine Hydroid :
Corynitis Agassizii and its Medusa.

By

L. Murbach.
With Plate 34.

Hypolytus peregrinus, a New Marine Hydroid.

INTRODUCTORY.

Durine the last part of the summer of 1895, while searching
for the larve of Gonionemus! in tow and dredgings from
the eel pond at Wood’s Holl, a curious little hydroid polyp
was found. At the time it was sketched, and a few notes
made on the supposition that it was the larval form of some
other hydroid, perhaps a Tubularian. In the following
summer forms of the same kind with gonophores were found.
This left no doubt that it was after all a mature form, but the
limited number of specimens warranted no further conclusion
than that the animal would probably prove interesting on
account of its unattached condition, a character which would
not have been remarkable in a larval form. Nothing like a
perisarc was at first thought to be present, for, as was after-

' Although this medusa breeds regularly every summer in the eel pond, yet
after three summers of careful searching I have found only few metamor-
phosing larval stages, but have not been able to get the stages intermediate ;
nor have I been able to raise any beyond the stage of the polyp with four
tentacles.

VoL. 42, pART 3,——NEW SERIES. Z

342 L. MURBACH.

ward learned, there was simply a tubular secretion, only shreds
of which remained on the captured polyps.

Last summer (1897) more specimens were obtained by
allowing the tow-net to scrape gently over the eel-grass.
These seemed to warrant my previous conjectures, and
although specimens were sometimes taken in clear water, I
now concluded that they usually are temporarily attached to
submerged refuse, eel-grass, &c., by means of their perisarcal
secretion.

The attached condition of marine hydroids is so universal,
and those able to move from place to place are so few, that up
to the present time not much stress has been laid on this
characteristic. More cases of this kind would establish such
character as very primitive or as reversion to an ancestral
type, a free polyp, perhaps Actinula-like.

To the writer’s knowledge there are only two marine forms
so far known which may be considered free. They are
Protohydra Leuckartii and Halermita cumulans.

The former, long ago discovered by Greeff,! was con-
sidered, as its name indicates, to be an ancestral form of
hydroids, but its foot is adapted for fixation, and is therefore
of permanent character. Furthermore, no sexual reproduction
has up to the present time been observed, and partly for this
reason Schaudinn? suggests that it may be the larval form
of a more highly organised polyp.

Halermita, discovered by Schaudinn’® in the Berlin
Aquaria, is certainly a remarkable form, and as his report of
the finding may not be accessible to all, I shall recount: the
principal features of the polyp. Its name indicates its solitary
mode of life. In form it is short and conical (stumpf kegel-
férmig). There are no divisions into hydranth and hydro-
caulus; the tentacles (the figures show only one circlet) are
usually four in number, but never more than five, and these

1 Greeff, R., “Protohydra Leuckartii,” ‘Zeitschr. f. Zool.,’ Bd. xx,
1870.

; Schaudinn, F., “ Halermita cumulans,” ‘Sitzber. d. Gesellsch. Nat.
Freunde,’ Berlin, 1894.

HYDROIDS FROM WOOD’S HOLL, MASS. 348

are not knobbed at the end. The pear-shaped nettling organs
are of only one kind, and they are evenly distributed, i. e. not
in groups. The longest tentacles are 8 mm.; the endoderm of
the tentacles is a solid axis. From the paper I infer that the
foot end is in no way modified for fixation, but simply sticks
in the accumulated débris, hence its species name cumu-
lans.

Schaudinn places Halermita between Hydride and all
known hydroids, but this is only tentative, since he has not
been able to observe the sexual reproduction ; he admits that
it may also be the larval form of a more specialised hydroid.

Up to the present time, then, all marine hydroids known to
be adults are permanently fixed; and even if we consider
Protohydra and Halermita to be mature forms, then the latter
would be the only one so far recorded which does not seem
to be specially modified at any point of its foot for fixation.
Naturally the case of Corymorpha suggests itself as a form
that might be an exception, but may be dismissed on account
of the processes at its foot end, which are undoubtedly rem-
nants of a Hydrorhiza.! To the new polyp found at Wood’s
Holl I have given the name Hypolytus peregrinus.’

General.

The polyps of Hypolytus peregrinus are found tem-
porarily attached by the secretion of their ectoderm to some
foreign object several feet below the surface of the water, or
having become detached, probably by withdrawing from the
perisarcal tube (in which case a new one is quickly secreted),
they may be found floating at the surface of the water. This
no doubt accounts for their being occasionally taken with the
tow-net in clear water. Their temporary attachment is again

1 Korschelt and Heider (‘Comp. Embryology’) suggest that these may
indicate a previous colonial condition of Corymorpha.

2 From wt76, under, below; and Avw, loosen; peregrinus, travelling.
Should the name here proposed for this new genus be preoccupied, I propose
instead Gonohypolytus.

344 L. MURBACH.

easily effected anywhere along the tapering foot end. No
other part of the hydrocaulus seems to be used to “ make
fast.”

The predominant colour of the polyp is pale to bright pink,
resembling many Tubularians in this respect. As commonly —
occurs elsewhere, the colour is localised mostly in the endoderm
cells of the body, showing through the more transparent
ectoderm of the periphery. In another part of this paper it
will appear that greater activity in any part of the body is
marked by greater depth of colour. In a general way this is
evident in the more active digestive region of the hydranth.
An apparent exception are the intense pigment spots at the
free ends of the gonophores.

The regions of the body are well marked into hydranth,
bearing besides the mouth two circlets of tentacles and the
gonophores, and into hydrocaulus (cf. fig. 1). At the union
of these two main divisions of the body there is a thickened
collar-like portion studded with nettling organs. From this
structure to the free rounded foot end the hydrocaulus is
covered with a kind of rudimentary perisarc. The hydro-
caulus is never branched.

In size the average adult animal is from 1 to 13 em. long,
and 1 to 14 mm. thick. Of course the size varies with the
degree of expansion or contraction ; the measurements were
therefore made from a moderately expanded animal.

Locomotion is slow though definite, and not very extensive.
The animal seems to progress by leaving its tubular secretion
behind, stepping on it, as it were, so that a relatively long
piece of the tube, plainly marked by adhering foreign matter,
indicates its progress. The movements of parts of the body
are slow,—its tentacles, for example, swaying to and fro in
search of prey. When disturbed the hydranth and tentacles
contract first, and if the irritation is continued the whole
animal contracts into a small mass.

HYDROIDS FROM WOOD’S HOLL, MASS. 345

General Anatomy.

The hydranth is terminated at its free end by the usually
conical hypostome, containing considerable pigment at its
highest point (fig. 1, 4.). It is pierced by a small mouth-open-
ing leading directly into the celenteron. Immediately below
the hypostome is the set of oral tentacles, ten in number, and
placed at regular intervals like radii (fig. 1, 0.¢.). Ten being
the largest number commonly present, I take it to be the
normal. These tentacles are one third to one half shorter
than those of the lower circle, but are otherwise of the same
shape and structure. There are no scattered tentacles on the
hydranth, and the lower or aboral set occurs nearly two thirds
the length of the whole hydranth from the oral one. They
are in no way different from those of the oral circle except
that they are longer. ‘There are usually fourteen, though
frequently a smaller number has been observed (fig. 1, a. ¢.).

In general the tentacles are stouter in appearance than is
usually the case in such small polyps. They are slightly
enlarged at the end, though there is no knob present except
in the young animal, where they are somewhat knobbed. The
larger appearance of the tentacles is no doubt due to the
prominent ridges of nettling organs which run in circles and
short spirals, pushing their cnidocils considerably above the
surface. The ectoderm of the tentacles is very transparent,
and not easily separated from the mesoglea. The endoderm
forms a solid axis through the centre of the tentacle, and in
polyps somewhat reduced by fasting, much black pigment
collects in these cells, giving the tentacles the appearance of
being hollow; even in ordinary specimens some pigment may
be present.

The gonophores (fig. 1, g.) spring from the hydranth just
above the aboral circle of tentacles, and number in adult
polyps from one to three, never more than three having been
observed. They present some peculiar features, which will be
more fully described under reproduction.

The predominant colour of the hydranth is located in its

346 L. MURBACH.

endoderm, the ectoderm being quite clear. Both these layers
as well as the mesogloea have the typical celenterate character,
aud will not need further notice.

The-hydranth does not terminate immediately below the
aboral tentacles, but, as is evident from its internal and its ex-
ternal character, it extends over one third its length farther
down to where it unites with the body. Just below the point
where the aboral tentacles are attached there is an enlargement
in the digestive cavity, looking like a deeply pigmented band
running across the ccelenteron, for which I have so far found
no adequate explanation. Where the hydranth joins the
hydrocaulus there is a ring-like expansion, which gives the
appearance of the former being stuck on to or slipped over the
end of the latter, like a collar or flange (fig. 1,c.). In an
expanded condition of the body it is nearly obliterated, becom-
ing more prominent again after contraction has returned the
body to its normal. It marks the upper limit of the perisarcal
tube, and this being a rather tightly fitting structure may in
part account for the changeable character of the collar.

Large numbers of nettling organs are present in the lower
edge of the collar, and although they are apparently complete
for use are nevertheless not destined to be used here, for there
are no cnidocils present. They migrate from the collar toward
the tentacles, and are of no service until they reach these and
become erect. Of the nettling organs in general it may be
here added that there are at least two kinds, similar in shape
but differing in size and structure. They are very short
ovals.

The hydrocaulus is somewhat more slender than the adjacent
portion of the hydranth, and gradually narrows down to the
taper-pointed foot end, which is generally curved and forms a
better rest for the polyp (fig. 1,2.¢.). The character of its
layers is practically the same as that of the hydranth, added
only to this that the ectoderm cells differ physiologically in
that they secrete the perisarc-like tube. A portion of the
foot end is frequently of a deeper pink hue, indicating greater
activity here; but as this has to do with reproduction it will

HYDROIDS FROM WOOD'S HOLL, MASS. 347

be again referred to under that head. There is no special
differentiation at the end for attaching the animal. The ccelen-
teric cavity extends to the tip of the foot end, and in it con-
stant circulation may be scen, due to the flagella on the endo-
derm cells.

A very delicate perisarcal envelope covers the whole hydro-
caulus from the collar to the foot end, or it may even extend
farther beyond in a collapsed condition, adhering to foreign
objects, showing the distance the animal has travelled. It
invests the body so closely and is so thin that it can scarcely
be distinguished from the transparent ectoderm which secretes
it, except when favourable conditions of illumination show it
thrown into folds on the concave side of the body as the latter
bends in any direction. When polyps are roughly handled
with the pipette it is torn into shreds; in such specimens it
first came to my notice. Of course indisputable evidence of
its presence and its tubular nature is found in the remains
left behind on which foreign matter has collected. Frequently
several of these may be found radiating from near the same
point, usually a mass of débris, which then marks the place
where several polyps from one parent leaving these tubes
originally stood (cf. fig. 10).

The temporary nature of the perisarcal tube, which is easily
lost or even left, and quickly replaced, indicates that it simply
is a somewhat hardened mucous secretion serving for support
and protection, and not a true chitinous perisarc, such as other
hydroids usually possess.

Sexual Reproduction.

In Hypolytus the sexes are separate, and the males seem to
preponderate. Sexual reproduction probably takes place in the
latter part of summer, for by the middle of August sperm and
ova were just beginning to mature in some individuals. Only
one specimen was found with what appeared to be nearly
mature ova, and an attempt was made to fertilise them, but
was not successful (fig. la, g’”’).

The gonophores (fig. 1,9’, 9’) are limited to a narrow zone

348 L. MURBACH.

just above the aboral set of tentacles, standing at unequal dis-
tances apart, and were not more than three in number on any
of the individuals observed. The first sign of a budding gono-
phore is a slight elevation with a deep pink pigment-spot on
the hydranth. Both older and younger stages have a spindle
or elongated oval form, which in the mature ones becomes
distorted by the growth and aggregation of the sexual products
in the ectoderm of the outer wall (fig. 1, 9’). The general hue
of the gonophores is bright pink. In length the older ones
equal the part of the hydranth between the two circles of
tentacles, but being less contractile may appear longer.

A narrow neck connects the gonophore with the hydranth,
and just at the junction there is a small curved process directed
aborally (fig. 1, p.). It is hollow, and appears to belong to the
gonophore, its cavity being connected with that leading from
the gonophore into the celenteron of the hydranth. In small
specimens these processes are not yet present. Their nature
and significance have remained an enigma to me. I do not
know of a homologue anywhere among the hydroid polyps.
The coelenteron is continued through the gonophore to its tip,
where a bit of bright pigment is visible. Active circulation
may be observed in gonophores as well as in processes at their
proximal ends.

Asexual Reproduction.

My attention was first attracted to the remarkable mode of
asexual reproduction by the peculiar appearance of the foot
end of a few specimens in a lot of about twenty, taken July
26th. One or two constrictions (cf. fig. 2) marked off
deeper pink portions of greater diameter. When these seg-
ments were freed from the body of the adult they looked not
unlike the large planule of Pennaria, obtained at the time in
considerable numbers. Indeed, the same day such a planula-
like body was found in the tow. It was isolated and watched,
to determine if it were the detached foot end of Hypolytus
ora planula. It moved about for some time, and then slowly
erecting itself, attached by its narrow end (making it at once

HYDROIDS FROM WOOD’S HOLL, MASS. 349

evident that it was not the planula of a hydroid), and developed
into a young Hypolytus, thus settling the question beyond a
doubt.

As the segments freed from the foot end of Hypolytus are
destined to form new polyps directly, and differ from any kind
of bud heretofore described, as will be shown later, I shall
call them blastolytes.

From a large number of cases of asexual reproduction ob-
served, the following record of the typical course of events is
made.

The first signs of the process are seen in the deeper hue
taken on by the free end of the polyp, which is no doubt due
to the concentration of material for future use. A slight
thickening also takes place at this time, and both these phe-
nomena may be due to a very slow mass-contraction at the
foot end, or they may be due to constructive metabolism.
This point is an important one, but must be deferred until
microscopic examination of tissues is made. Next a constric-
tion is seen about two and a half times its diameter from
the foot end (fig. 2,4). The fact that it forms very gradually
and without any marked contraction of the body at this point
warrants the conclusion that this and the subsequent process
of complete fission are purely cellular activities. Frequently
before the first blastolyte is entirely constricted off a second
circular groove marks off another blastolyte (fig. 2,a); and
even a third has been seen in close succession to the other two,
but not more than two have been observed at one time.

Just as soon as one blastolyte is freed, its oral end (its
polarity, judging from all my observations, remains the same
as that of the parent) becomes rounded and somewhat thicker,
while the aboral is drawn out to be more slender,—probably a
shifting of material to a point where it will be soon needed for
the rapid development of the two sets of tentacles, the first
necessary organs for securing food.

From the usually curved position of the foot, as indicated in
the anatomical portion, the blastolyte lies almost horizontal or
at most somewhat inclined. From this position it rises up as

350 L. MURBACH.

soon as free, and apparently dissolves the portion of the peri-
sarcal tube immediately above. In specimens kept in glass dishes
the whole process of fission took place in about six hours. The
blastolytes given off by specimens in confinement did not show
much disposition to move about, as did one found free in some
tow. After rising up on the pointed end the tentacles begin
to bud out on the enlarged upper end as minute knobs (figs. 5,
6), generally two oral ones first, then two aboral, and almost
simultaneously with the two aboral ones the second pair of
oral tentacles develops.

When about 2 mm. in length the mouth opening is present
and the nematocyst collar begins to show. ‘There are five oral
and nine aboral tentacles, all somewhat knobbed. The odd
number of tentacles shows that after the first two they do not
continue to develop in pairs. At this stage the foot end rests
curved like in adults, a character which is also evinced by the
larve of other hydroids. The perisarcal tube is present, being
fully developed up to the collar. The tentacles are solid ata
very early stage.

When several blastolytes are given off in succession, a group
of polyps may arise, and remain close together for some time.
So situated, débris collects on the remnants of the perisarc,
and the individuals seem to stick in the accumulated mass.
The parent meanwhile has moved a considerable distance from
its offspring.

This is a brief account of the normal process of asexual re-
production as it takes place in the larger number of cases ; but
in some, such a pronounced modification was observed as to
warrant a separate description.

In the first case noticed (fig. 4) the constricting mass, the
second of two starting out apparently normally, began to show
a decided lateral thickening, evidently an accumulation of
material for some future use. The first blastolyte continued
its normal development, while the enlargement on the second
one increased, evidently at the expense of the two ends, for
which their attenuation speaks (fig. 4, a), At this point my notes
read :—After 9.20 a.m., or about two and a half hours after the

HYDROIDS FROM WOOD'S HOLL, MASS. 351

first blastolyte was seen free, the second one came off from the
polyp and, shortening somewhat, became arched (fig. 5, @), and
as it was before thickened in the middle, the two ends ap-
proached more and more, and formed the foot end of the
polyp. Even the next day the forked foot of this little animal
could be plainly seen (fig. 5,5). Here, then, the anterior or
oral end had formed from the side of the parent. On the next
day I saw the same result accomplished in another way (figs.
7—9). A constricting segment was found at the end of the
nearly severed parent, and it had a large hump on one side.
This lateral protuberance became larger as the constriction
proceeded, then it grew still more at the expense of the foot end,
being now severed from the parent ; it became the greater bulk,
and the former foot-mass became a narrow process (fig. 9, 0’).
The constricted end was also gradually drawn in until the
whole assumed the shape of the ordinary blastolyte (9, 5”).
Here again the oral end of the blastolyte was formed from the
side of the parent polyp, and as a lateral outgrowth.!

The normal process of asexual reproduction of Hypolytus
is different from any of the cases of fission described among
hydroids. Comparison with strobilation as it occurs among
the Scyphozoa, as furnishing a parallel case among the
Hydrozoa, seems too strained, especially since it is at the
wrong pole, and the resulting products are different.? It is
different from the frustulation of Schizocladium ramosum
described by Allman,® since there fission takes place at what
would be the oral end.

The saccule of Schaudinn’s Halermita represent freed lateral
buds, and resemble the blastolytes of Hypolytus only in that

1 Not thinking of the possible siguificance of these phenomena, no attention
was paid to the relative time it took such a blastolyte to produce tentacles,
and it is reserved for future observation to see if the explanation given on
another page will be borne out.

? The well-known phenomenon of “decapitation” among hydroids does not
come into consideration here, since it is not a process of reproduction.

3 Allman, G. J., ‘A Monograph of Gymnoblastic or Tubularian Hydroids,’
1871-2.

oon L. MURBACH.

they may develop directly into polyps. The processes con-
stricted off at the basal end of Corymorpha, which, according
to Allman,! develop into new polyps, if not the remnants
of a Hydrorhiza might correctly be compared with the blasto-
lytes.

Nearest of all, perhaps, comes the asexual reproduction of
Protohydra, L., recently more fully investigated by Chun.?
But this polyp is more primitive, and fission takes place at
almost any point on the body, the fission zone not being
constant as it is in Hypolytus.

Normal asexual reproduction in Hypolytus by spontaneous
fission of a definite portion of the free foot end is unlike any
reproductive process heretofore described among Hydrozoa, and
since it precludes an attached condition it probably represents
an ancestral mode retained by this form. It may have gone
through some such stage as Protohydra now does.

The modification of the normal process described under
asexual reproduction remains unexplained, and I offer the
following as a possible one. It would be of great advantage to
the young polyp to have the organs for obtaining a livelihood
developed as early as possible. If, now, the material from
which the hydranth and the tentacles are to be developed can
be accumulated and differentiated (the lateral enlargement)
while constriction and fission are going on, the young polyp at
the close of this operation could the sooner be ready for the
activities of life. This involves formation of a hydranth from
a lateral portion of the hydrocaulus instead of from the axial ;
in fact, just what does take place in Hypolytus. It furthermore
suggests a possible explanation of the origin of lateral budding
among marine hydroids, by assuming that the precocious
development of a hydranth made the separation from the
parent unnecessary.

To account for the unattached condition of Hypolytus we
may assume that it is secondary, or, on the other hand, that it

1 «A Monograph of Gymnoblastic or Tubularian Hydroids,’ 1871-2.

2 Chun, Carl, ‘Bronn’s Klassen u. Ordnungen d. Thierreichs,’ Bd. ii,
p. 115.

HYDROIDS FROM WOOD'S HOLL, MASS. 353

is primitive. In the first case it would have been preceded by
a fixed condition of the polyp. Then the polyp by some
process of fission managed to sever itself from a part of its
foot end, and attaching itself again went through the same
process until fixation was entirely dispensed with, and thus
reverted to the ancestral free form.! This is, however, with-
out parallel, unless the case of Corymorpha furnish one.
One other consideration seems to outweigh the above, viz.
that the peculiar mode of asexual reproduction in Hypolytus
involves fission of the free end of the parent. It seems to
me, then, that it is a phylogenetic character.

Summing up the characters of Hypolytus peregrinus,
we have—a single unbranched polyp of the Tubularian type,
with two circles of tentacles, ten in the upper and fourteen in
the lower ; a primitive perisare enveloping the hydrocaulus, at
whose free end buds are given off by spontaneous fission, and
these in turn develop into polyps like the parent directly ;
sexual reproduction by means of ova and spermatozoa,
developed in gonophores situated just above the aboral circle
of tentacles; on account of its unattached condition it is free
to move from place to place, which it does slowly. These and
some minor ones are characters that will have bearing on the
ultimate classification of our animal, which is not attempted in
this report. It is intended to bring out only those characters
that have to do with phylogeny and some other problems, such
as fission and budding.

Detroit, Micuican; March, 1898.

1 In this case we might expect the progeny to form at least a temporary
hydrorhiza, which, however, does not take place here as it does in Cory-
morpha. It may be that the embryology of Hypolytus may furnish some
further evidence on this point. Another way of looking at the same question
is, that Hypolytus was an attached colonial form, in which spontaneous fission
took place first just below and then above a lateral bud, and this becoming
permanent the lateral thickenings on the blastolytes are to be interpreted as
the last remains of budding.

354 L. MURBACH.

Corynitis Agassizii, McCrady, and its Medusa, Gemmaria.

Last summer, while examining some sargassum driven into
Vineyard Sound from the Gulf Stream, I found a small polyp
which proved to be Corynitis Agassizil, McCrady. In his
description of Halocharis (Corynitis), Agassiz! says the
medusoid stage was not found, but later he found his Halo-
charis identical with Corynitis of McCrady, who had observed
the meduse. But according to Allman® the medusa as-
cribed to Corynitis by both McCrady and Agassiz has four
marginal tentacles, each with a clavate extremity beset by
nodulated pads of thread-cells, and “ four overarched spaces
between the roots of the radiating canals,” while the immature
medusa possesses only two tentacles and no ‘ overarched
spaces.” Allman accepts the general correctness of McCrady’s
observations with some reservation, pointing out that McCrady
captured his four-tentacled medusa in the open sea. He,
therefore, has inferred its relationship to Corynitis by inter-
mediate stages.

As my polyps possessed numerous medusa buds, they were
kept under observation to determine the question raised by
Allman, and finding no previous record of the occurrence of
Corynitis in the vicinity of Wood’s Holl, I append a short
description of the polyps to better establish their identity, and
to add a few new points.

They are found most abundant on Membranipora incrusta-
tions below the low water mark, probably because on the
reddish calcareous deposit they have very good colour protec-
tion, as is evident from the difficulty of readily seeing them.

The hydrorhiza is deep pink, while the tiny hydranths have
a delicate, translucent, white shade enveloping body and ten-
tacles, with pink between the lighter edges. The hydrorhiza
is slender and thread-like, and anastomoses frequently, form-

1 Agassiz, L., ‘Contr. Nat. Hist. U.S.,’ vol. iv, pp. 239, 240, 1862.

2 Allman, G. J., ‘A Monograph of the Gymnoblastic or Tubularian Hy-
droids,’ p. 286, 1870-72.

HYDROIDS FROM WOOD'S HOLL, MASS. 355

ing a coarse network. The polyps generally arise singly from
the hydrorhiza, and do not branch. While the hydrorhiza is
covered with a delicate perisarc, I could not demonstrate it
with certainty on any part of the polyps, and so cannot verify
Allman’s prediction that a rudimentary hydrocaulus will be
found in Corynitis.

As a rule the polyps are slender club-shaped bodies, from
14 to 2 mm. in length (fig. 12). There is no marked division
into hydranth and hydrocaulus, as Agassiz has pointed out,
except that the proximal third is free from tentacles. The
oral end is quite blunt, but the hypostome as well as the
body is flexible and contractile. Distally the polyp is beset
by from thirty to forty-five short knobbed tentacles, which are
not arranged in regular circles, but in somewhat oblique rows,
giving rise to the spiral arrangement described by Agassiz.
The longest tentacles are not more than ;4; mm. in length,
being nearest the oral end, while the aboral ones are repre-
sented by mere elevations on the body. The upper tentacles
do not form a circlet around the hypostome, there being a
single one higher than the rest. The longer tentacles bear
definite large nettling knobs at their ends; a solid row of
endoderm cells forms their axes. Nettling organs are also
found in the ectoderm of the body migrating? toward the
tentacles from the base of the polyp where they are developed.

Medusa buds appear most numerous in a zone where the
rudimentary tentacles are, though scattered ones may also be

1 This summer I have found what appears to be a second species of Cory-
nitis. It differs from C. Agassizii in the presence of a well-developed peri-
sare on the hydrorhiza and the short hydrocaulus, forming imperfectly
annulated cups about one fourth the length of the polyp, and in the fact that
the medusa-buds are on branched stalks. The colony was not in good enough
condition to be sketched, and no meduse were freed, so it must be left for
future observation to determine its relationship.

2 In a recent article, ‘ Biol. Centralblatt,’ Bd. xvii, No. 18, 1897, v. Len-
denfeld has thrown doubt on the fact of the migration of nettling organs. In
this connection it is sufficient to state that for several summers in succession
this phenomenon has been observed on fresh Pennaria by our students in the
laboratory here.

356 L. MURBACH.

found higher up on the polyp. The largest, and therefore the
oldest buds are borne below the middle of the length of the
body on single short stalks. From one to ten buds may be
found on one polyp.

Scattered everywhere among the polyps bearing medusa
buds are others that appear to be sterile individuals. When
there are only a few of them they are not conspicuous, and at
a time when none of the polyps of a group had any medusa
buds they might not at all be noticed. They resemble the
others in all respects except that they are more slender and
taller, being often 2 to 3 mm. in height, and that they lack
any traces of medusa buds, while those around them are very
prolific. I cannot understand this sterility of individuals so
nearly like the reproductive ones, unless it be a functional one,
and is to be interpreted as the beginning of a division of labour
in these simple polyps, which in time will lead to a more
striking polymorphism. Allman! in his Gemmaria im-
plexa has observed a similar difference, of which he says,
“Tn nocase can it be regarded as reducing the hydranth to the
condition of a blastostyle.”

Shortly before the Corynitis polyps were found here I had
been taking in the tow a small medusa bearing on its two
tentacles some peculiar stalked organs, not unlike stalked
Protozoa. When I found them to be an integral part of the
tentacles it became evident that the medusa before me was
Gemmaria, especially since other characters agreed. The ex-
planation of the presence of this form in our harbour was
soon apparent when the meduse, freed from the Corynitis
polyps, were found to have exactly the same characters, and so
proved to be Gemmaria. To leave no doubt whatever, one
medusa was observed continuously while freeing itself from its
polyp nurse by repeated contractions, and until it had suffi-
ciently expanded to recognise its distinctive characteristics. Its
umbrella was more spherical, and its tentacles more contracted
—one shorter than the other,—as were also its stalked organs

1 © A Monograph of Gymnoblastie or Tubularian Hydroids,’ 1871-2.

HYDROIDS FROM WOOD’S HOLL, MASS. 357

on the tentacles, than is usually the case in older meduse.
Ova were present on the manubrium.

Older medusz (ef. fig. 11) measure from 1 to 2 mm. in
diameter; the umbrella is obovate, deeper than broad, the
walls rather thin. Just opposite the four radial canals on the
outside of the bell, and extending up about one fifth its
meridian, are swellings (fusiform sacs of Allman, p. 224),
filled with large nettling organs. The manubrium, extending
through over half the length of the bell, is cylindrical, becom-
ing conical when its walls are distended with sexual products.
The velum is well developed, with opening rather small. The
two tentacles on opposite perradii are quite long and slender
when fully expanded, and are provided with long slender
filaments, bearing thick-walled oval capsules, each of which
contains from three to five oval glistening bodies, and is beset
by stiff hair-like processes. At least the proximal portion of
each tentacle is hollow, as is evident from the circulation of
food particles, while farther out there appear to be separate
vacuoles, each containing minute granules exhibiting active
Brownian movement. The bases of the two tentacles are
enlarged, and bear irregular pigment masses. At the two
remaining perradii are slight prominences (below the swellings
before mentioned) filled with small nettling organs and some
pigment, representing, no doubt, two rudimentary tentacles.

The very unique feature of the genus Gemmaria, as Allman!
has pointed out, is the stalked organs on the tentacles.
These organs one is tempted to compare with the nettling
batteries of some of the Siphonophora, not only on account of
their containing a number of nematocysts” in one receptacle,

1 Thid., p. 225.

2 Although each nematocyst showed a central body looking like folded barbs,
i was at first inclined to doubt their nettling function; for while all other
nematocysts of the medusa responded to mechanical or chemical stimuli, these
were most obdurate. But finally I succeeded in causing threads to be dis-
charged, and now the evaginated thread showed that the appearance in the
intact capsule was due to a number of small folded barbs occurring just below

a vesicular enlargement of the thread (cf. fig. 13). Allman (ibid., pl. viii) has
also figured such a nematocyst from his Gemmaria polyp.

VOL. 42, PART 3.—NEW SERIES. AA

358 L. MURBACH.

but because the stalks themselves resemble very much the
elastic filaments found in nettling batteries.

in the contracted condition the stalked organs seem to beset
the tentacles on all sides, but during expansion they are all
directed more or less aborally. The nematocysts in the stalked
organs are developed in the bases of the tentacles, and migrate
outward to where the capsules of the stalked organs arise as
evaginations of the ectoderm (fig. 14). At such points the
ectoderm is already supplied with the hair-like processes which
later stand on the capsules.

When somewhat expanded the stalks are thick and have a
wavy outline, while, expanding still more, they look not unlike
an unfolding zigzag line or spiral. This gives rise to the
granular appearance described by Allman (p. 225), and is in all
probability due to optical sections of the joints or spirals of the
unfolding stalks.

Finally, there is a very fine smooth filament, not much thicker
than a nettling thread, and about as long as the diameter of
the medusa, bearing the quivering capsule on its end. During
contraction these several appearances occur in reversed order.

The capsules (fig. 13) are thick-walled and somewhat wavy
in outline, as if they were made up of segments, and are pierced
by a number of openings for the emission of the threads from
the contained nematocysts. Covering at least two thirds of
the outer portion of each capsule are stiff hairs capable of
vibrating so as to impart a peculiar quivering motion to the
capsule. They do not wave as cilia usually do, and so can
hardly be compared with them. Allman has called them
vibratile cilia, and Agassiz! does not mention or figure them.
The function of the vibratile cilia may be to move the capsules
through more space, or may also be tactile.

As to the identity of our medusa with Gemmaria gem-
mosa, McCrady, there can be no doubt, and that it agrees as
nearly with the European form as with the one figured and
described by Agassiz! may be attributed to age, sex, and con-
dition of expansion of parts.

1 Agassiz, A., ‘Ill. Cat. N. A. Acaleph,’ p. 184, 1865.

HYDROIDS FROM WOOD’S HOLL, MASS. 359

The polyps of the European and American Gemmaria, how-
ever, exhibit differences enough to remain generically separate ;
such are the form of the polyp, the degree of differentiation into
hydranth and hydrocaulus, the relative development of perisarc,
and the arrangement of tentacles. Here, again, as has been
pointed out by others, we are confronted by the anomalous con-
dition of two medusz, almost identical, being produced from
polyps generically separated.

Marine BioLoeicaL LABORATORY,
Woon’s Hout, Mass.; August, 1898.

EXPLANATION OF PLATE 34,

Illustrating Mr. L. Murbach’s paper on “ Hydroids from
Wood’s Holl, Mass.”

HYpouytus.

Figures 1—10 are eight times enlarged.
"

Fic. 1.—Adult male polyp. a@.¢. Aboral tentacles. c. Nettle collar. g”.
An immature gonophore. g’. Gonophore with sperm nearly ripe. Ae. Hydro-
caulus. f. Hypostome. Ay. Hydranth. o.¢. Oral tentacles. p. Processes
at the bases of the gonophores.
mn

Fic. 1 a—g'". A female gonophore. o. Ova. jp. The process at the base.
Fic. 2.—¥oot end of a polyp showing constrictions (a, 6) preceding fission.

lic. 3.—The same with one blastolyte (0) free, and the other (a) showing
a lateral thickening.

Fic. 4.—The same with both blastolytes free.

Fig. 5.—The same showing the hydranth of the young polyp (4) formed
from the side of the adult. The forked foot was formed by approximation of
tapering ends of the blastolyte (a).

Fig. 6.—Young polyp normally formed from the blastolyte (4).

Fic. 7.—Foot end of another polyp (@) showing lateral thickening on
blastolyte (4).

360 L. MURBACH.

Fic. 8.—The same (a) with freed blastolyte (4) showing increasing lateral
thickening.

Fic. 9.—é'. The same blastolyte as 4, Fig. 8, showing the increased growth
of the lateral thickening. 6”. Final result of 6, Fig. 7.

Fic. 10.—Portion of hydrocaulus showing perisarcal secretion.

CorYNITIS AND GEMMARIA.

Fie. 11—Gemmaria gemmosa, representing one tentacle fully ex-
panded.

Fic. 12.—Polyp of Corynitis Agassizii without medusa buds.

Fic. 13.—(a@) Capsule of one of the stalked organs, from the tentacle of
medusa, showing one nematocyst discharged; (4) large nematocyst discharged
from one of the swellings on the bell.

Fie. 14.—Portion of tentacle, showing origin of stalked capsules and
vacuoles in the axis.

ARHYNCHUS HEMIGNATHI. 361

Note.

Arhynchus hemignathi.

In the 39th Volume (New Series) of this Journal, p. 207, 1
described a new Acanthocephalon taken from inside the skin
in the neighbourhood of the anus of a Sandwich Island
bird, Hemignathus procerus. I suggested the name of
Arhynchus for this form, since its chief characteristic is the
absence of a proboscis. Recently both Professor C. Wardell
Styles and Professor A. Hassall have pointed out that this
name is preoccupied, having been used by Dujean in 1834 for
a beetle. I therefore now suggest the name of Apororhyn-
chus for the new genus which I described in 1896.

ARTHUR KE. SHIPLEY.

ZooLocical, LABORATORY,
CAMBRIDGE ;
July, 1899.

VOL. 42, PART 3.—NEW SERIES. BB

; ; . ; a] - »*, 7
a 2 a : a eS 7 oe

r 6 Py sh

i LPP Ast Peat es We Pipe

oo! oe
: a ear
a ke §
Led oe 4 id
i
:
‘
-
. i.
‘ a , ‘
Teoh woe sy
*
ij at tLe 5
Ls (* is
: ce
é
7 :
i
, ;
Far | 1th I a. \
1% Pel) we
— J be

7 il &Y gl weniis "

wa

At ae 7
; : Vitra =

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS.

363

The Structure and Metamorphosis of the Larva

D

of Spongilla lacustris.
By
Richard Evans, B.A.,
Of Jesus College, Oxford.

With Plates 35—41.

[This memoir was awarded the ‘ Rolleston Memorial Prize” of the

University of Oxford for the year 1898.—E. R. L.]

ConrTeENTS.

I. InTRopDUCTION
A. Historical review ;
B. Summary of the author’s Pesalis
(a) Histology of the larva
(4) Metamorphosis and subsequent development
Il. Deraitep Account oF THE DEVELOPMENT
A. Histology of the larva .
(a) Flagellated cells
(4) Inner mass
Characters of inner mass in ne A

bE) ”? ” B
29 3 3? C
D

33 39 ”
B. The fixation, metamorphosis, and further development .
(a) Features common to the metamorphosis of all
larvee
(4) Special features of the metamarnlos
(1) Of type C
(2) Of type D
(c) Further remarks on the formation of the seneral
canal system

vol. 42, PART 4,—NEW SERIES, Gre

PAGE
364
364
367
367
370
374
374
376
378
383
384
385
390
394.

398
408
409
415

418

364 RICHARD EVANS.

Appendices : PAGE
A. Nutritive vacuoles and yolk bodies . - 422
B. The origin of the cell groups found in types
B and C ‘ : 425
C. Development and structure of Poe ; 430
D. Division of collar-cells and multiplication of
chambers : ; : 432
EH. Mitotie division of the aisle ‘ fs 434
F. Technique . : 436
III. Critica Review or PREVIOUS ecooNrs AND Canennaene . 443
TV. THEORETICAL . ; ; 5 : : : 455
Conclusion . : ; : : ; : 461
VY. BIBLIOGRAPHY . : ; : : : ; 462
VI. Description oF PLATES . ‘ ‘ ; ; ‘ 463

I. INTRODUCTION.

Tus piece of work was begun about eighteen months ago
on the recommendation and under the superintendence of my
teachers, Professor E. Ray Lankester, M.A., F.R.S., and Mr.
E. A. Minchin, M.A., Fellow of Merton College, Oxford. The
development of Spongilla was suggested to me as a subject
worthy of study, on account of the extreme confusion and
startling contradictions contained in the published accounts,
rendering it almost impossible to make out what was true or
what was false. It was with the hope of determining the true
history and eliminating incorrect statements that I embarked
upon this study.

In order to show what was the state of our knowledge, I
begin with a short historical account, which will be followed by
an abridged account of my own results.

A. Historical Review.

There is no need to go further back than the account given
by Ganin (4) in the year 1879.

Ganin distinguished three layers of cells in the free-swim-
ming larva, which he called “ ectoderm,” ‘‘ mesoderm,” and
“ entoderm” respectively. The ‘‘ ectoderm ” consisted of the
layer of flagellated cells at the surface, and the “ entoderm ”

STRUCTURE OF THE LARVA OF SPONGILLA LAOUSTRIS. 365

was made up of the lining of the larval cavity and its diver-
ticula extending into the “mesoderm,” by which term he
included the remaining cells of the larva. He described the
larval cavity as being produced by liquefaction and breaking
down of some of the central cells.

When the larva became fixed the ‘‘ ectoderm” was said to
flatten out and to become the flat epithelium. The ‘ meso-
derm ” was supposed to produce the connective tissue and the
wandering cells, while the ends of the diverticula of the ‘‘ en-
toderm ” gave rise to the flagellated chambers, the diverticula
themselves producing the exhalant canals, and the larval cavity
becoming the gastral cavity.

The next observer who studied the development and meta-
morphosis of Spongilla was Gotte (5), who wrote in the year
1886. Gdotte divided the cells of the larva into two classes,
which he called ‘ ectoderm” and “entoderm.” The “ ecto-
derm ”’ consisted of the flagellated cells of the surface layer,
while the term “entoderm ”’ included all the cells enclosed
within this layer. He derived the larval cavity by mere sepa-
ration of the cells at the centre, and held that the flagellated
cells were thrown away when the larva became fixed, and con-
sequently that the whole sponge was built up from the “ ento-
derm.”” He was of opinion that the granules contained in the
large cells in the interior of the larva were yolk bodies, which
became filled with chromatin, and developed into nuclei, the
future nuclei of the cells of the flagellated chambers. Therefore,
according to Gotte, the flagellated chambers arose from the
large cells of the interior,—that is, the cells with vesicular
nuclei; the remaining cells of the “ entoderm” giving rise to
the rest of the sponge.

The third observer to study the development of Spongilla
was Maas (7), who wrote in the year 1890.

Maas describes the larva as consisting of ‘ ectoderm,” ‘‘ me-
soderm,” and “ entoderm,” terms which had the same meaning
as was given them by Ganin. The results of his investigations
agree to such an extent with those of Ganin as to need no
further statement of them than to say that he considered

366 RICHARD EVANS.

the granules in the cells with vesicular nuclei to be purely
vitelline.

The fourth observer to study the structure of the larva of
Spongilla and its metamorphosis was Yves Delage (1), who
wrote in the year 1892.

Delage wisely discarded the terms which had been used by
all previous observers, and described the larva as consisting of
four kinds of cells, which he called flagellated or ciliated cells,
epidermal cells, amoeboid cells, and intermediate cells.

The flagellated cells consist of the layer of cells which cover
the surface, and which pass into the interior, and are taken in
and ejected again by the ameboid cells, finally becoming the
collar-cells of the chambers.

The epidermal cells were described as forming a complete
layer of cells underlying the flagellated cells, which during
metamorphosis travel to the exterior, and give rise to the cells
of the flattened epithelium.

The intermediate cells he described as being embedded in
the inner mass, and as forming a lining to the larval cavity.
They become the lining of the canals in the interior of the
sponge, in addition to giving rise to connective-tissue cells.

The ameeboid cells are those which possess vesicular nuclei,
and which during the metamorphosis take in the flagellated
cells by a kind of phagocytic action, ejecting them again later
on. On being set free the flagellated cells become the collar-
cells of the chambers.

Maas, who in the meantime had, from his studies upon other
siliceous sponges, arrived independently at embryological con-
clusions similar to those put forward by Delage, retracted in
1893 (8) his former statements with reference to Spongilla,
and recognised in the larva three kinds of cells, viz. the flagel-
lated cells, the ameeboid cells, and all the remaining cells. He
denied that the flagellated cells are taken in by the ameeboid
cells, and held that they become the collar-cells without passing
through the peculiar changes described by Delage.

The last observer to study the development of Spongilla was
Noldeke (13), who wrote in the year 1895,

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 367

Néldeke described the larva as consisting of “ ectoderm”
(flagellated layer) and “ entoderm ” (all the inner mass). He
agreed with Delage that the flagellated cells pass into the
interior and are taken in by the ameeboid cells, but he differs
from him in stating that they are completely digested, and not
set free again to form collar-cells. He therefore follows Gotte
in describing the whole sponge as being developed from the
cells of the “ entoderm,”’ but he derives the chambers either
from a single cell or by the coming together of cells.

B. Summary of the Author’s Results.

This brief account is introduced here with the object of
making it easier for the general reader to grasp the results,
and in order to emphasise the main facts brought out by my
investigations ; in the first place as regards the histology of
the larva, and in the second place as regards its metamor-
phosis and subsequent development.

(2) Histology of the Larva.—The free-swimming larva
is egg-shaped, with a broader anterior and a narrower posterior
pole. The surface of the body is covered all over with a uniform
layer of flagella. At the anterior pole is lodged the larval
cavity, while the posterior pole is a solid mass of cells.

In the histological composition of the larva we can dis-
tinguish—

(1) A layer of flagellated cells at the surface.

(2) The inner mass.

(1) The flagellated cells are arranged in a complete
layer over the whole surface of the larva, and are uniform in
character. Hach cell carries a flagellum, which can be traced
down to the nucleus. The cell body is elongated and some-
what constricted in the vicinity of the nucleus, which is onion-
shaped, and situated at the base of the cell (Pl. 35; fig. 29a,
Pl. 40, &c.).

(2) The inner mass may consist of as many as three

kinds of cell elements, of which two at least are always
present.

(a) Cells with granular nuclei, always present as an irregular

368 RICHARD EVANS.

layer under the flagellated cells as well as lining the larval
cavity, and it may be in other parts of the inner mass. They
are irregular in shape, but may be very flattened when they
border a cavity.

The nucleus is spherical or subspherical, with small irregu-
larly shaped chromatin granules, more or less equal in size,
situated at the nodes of an even reticular framework.

The cytoplasm may contain a few yolk bodies (figs. la, 5,
and 6 5 16..g5.7-):.

(3) Cells with vesicular nuclei aggregated chiefly towards
the centre of the solid posterior region of the larva. They are
massive cells, spherical or oval in form, especially in the
younger larve, but sometimes in later stages quite irregular in
shape. The cells with vesicular nuclei represent a class of un-
modified celis derived from the blastomeres from which the
other cell elements arise, and which therefore diminish in
number during the progress of the development. Their nucleus
possesses a large central corpuscle suspended in a delicate
nuclear reticulum which always contains granules, varying
in number, size, and distribution. The cytoplasm contains
from one to four “ nutritive vacuoles ” and several or numerous
“‘ yolk bodies ” (figs. 1, 2, 5a and J, 6a, and 7; ¢.v. n.).

(y) Small cells arranged in groups, which may be termed
briefly “cell groups.” They are always situated in the interior
of the solid posterior part of the inner mass. In many cases
the component cells of a group are not completely separated
from one another, but present the appearance of nuclei arranged
near the surface of an incompletely divided mass of cytoplasm.
The nuclei are small, and at a certain stage in their develop-
ment resemble those of the flagellated cells, but they undergo
a process of change which causes them to have a slightly
different appearance in different larve (figs. 5 a, 6, and c, 7
and: 7 @; 11; Vbhe;-12, 13, 13a and; gac.):

The composition of the inner mass may be very different in
different larvee as regards the relative quantities of the cell
elements above enumerated. Four main types of larve may
be distinguished, which are connected by transitions, but which

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 369

may, nevertheless, be conveniently described apart from one
another.

Type A.—In these larvee the inner mass consists of only two
kinds of cells; namely, the cells with granular nuclei (a) and
with vesicular nuclei (3), the latter alone making up the central
part of the solid posterior region (fig. 1).

Type B.—The inner mass consists of three kinds of cell
elements above enumerated. The cells with granular nuclei
(a) are smaller and more numerous towards the interior. In
the cells with vesicular nuclei ((3) the ‘‘ yolk bodies ” are some-
what smaller. The “cell groups” are almost always in the
incompletely divided stage (figs. 5, 5 a, 6, and c, and 9, 9a
and 0).

Type C.—The composition of the inner mass is the same as
in type B, but with the following differences as regards the
development of the cells. The cells with vesicular nuclei (3)
are much smaller than in types A and B, and contain fewer
“yolk bodies,” and as a rule not more than one “ nutritive
vacuole.” The “ cell groups ” have the cells completely divided,
forming in many cases distinct flagellated chambers, and even
going so far as to develop collars and flagella (figs. 7, 7 a, 11,
11 a, 12, 18, and 13.4 and 3).

Type D.—The inner mass contains the same three elements
as in type B, but in very different proportions, and the cells
with granular nuclei (a) are reduced in size and much more
numerous. The “cell groups” (y) are exceedingly few in
number, though always present. The cells with vesicular
nuclei ((3) are much as in type C (figs. 6 and 6a).

The relationship of the four larval types is probably as
follows :

_ Type A represents an early form, from which type B may be
derived directly by further differentiation of the cells of the
inner mass. Type C may be considered as a further develop-
ment of type B. Type D,on the other hand, must be regarded
as having arisen directly from type A. Two divergent lines of
development can therefore be distinguished in the larva, of
which the two culminating points are represented by types C

370 RICHARD EVANS.

and D, type B being an intermediate stage between types A and
C. Type D is not necessarily younger than type C. In the
development of type C the vesicular cells (3) give rise to the
“cell groups” (vy) rather than to the cells with granular nuclei
(a). In the development of type D the exact opposite takes
place. The relations can be expressed graphically by the fol-
lowing diagram :

=

2 D
C

(6) Metamorphosis and Subsequent Development.
—The larva fixes itself either by the anterior pole, or by a
point not far from that pole, and these differences in the fixa-
tion are the causes of transitory variations in the form of the
newly fixed stages. In the case of type C the larval cavity
becomes obliterated soon after fixation. In the case of type
D, however, it seems practically certain that the larval cavity
is not obliterated. This may be inferred from the fact that a
cavity of considerable size is often found persistent in speci-
mens with fully formed flagellated chambers, with the flagel-
lated layer almost absent from the surface, with the flattened
epithelium almost complete all over, and with the developing
exhalant canals in some cases already opening into the
cavity.

The flagellated cells pass into the interior either individually
or in groups, which present a fan-like appearance, and which
in some cases consist of a great number of cells (figs. 14, 15,
and 15a and 0). ‘They pass in more rapidly on the lower than
on the upper surface, and this difference is more accentuated
in a larva of type C than in one of type D (comp. figs. 15 and
29). Simultaneously with the passing in of the flagellated
cells, the cells with granular nuclei pass out to form the
flattened epithelium of the upper and lower surfaces, as well
as to form the marginal membrane. ‘The cells with granular

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 371

nuclei appear to pass out individually over the whole surface,
and when they have got to the exterior they flatten out and
become soldered together by their margins to form the layer
of flattened epithelium which covers the surface.

The flagellated cells in all cases, after passing into the
interior, enter into the formation of peculiar associations in
which no cell limit can be seen except just at first. These
associations, the ‘‘ polynuclear groups”’ of Delage, will be
described in this paper as plasmodial aggregations (figs.
16, 16a, 29, 29 a and 5). More than half the mass of the
bodies in question must be made up in many cases of the
flagellated cells which have entered them by a process of active
migration, and which later on separate themselves from them
in the same manner.

The nuclei of the flagellated cells undergo peculiar modi-
fications. They contract, and the chromatin becomes re-
arranged and increases in quantity, so that by the time this
change has reached its extreme limit the nuclei of the flagel-
lated cells are very difficult to distinguish from the “ yolk
bodies.” In each plasmodial aggregation the external
limit of the cell mass as a whole is at one time sharp and well
defined; but this condition is transitory, lasting but a short
time, and is therefore not often met with. The plasmodial
aggregations become ill-defined and run into one another,
thus presenting a syncytium-like arrangement, in which it
is most difficult to make out cell boundaries. During this
period the nuclei of the flagellated cells change in their cha-
racters, and pass by degrees into a condition which represents
the definitive state of the nucleus of the collar-cell; their
framework becomes looser in texture, owing to the appearing
of threads and ultimately of small irregularly shaped granules,
and a well-defined nuclear membrane is formed. The nuclei
of the flagellated cells. while undergoing these changes, are as
a rule arranging themselves in rings in the cytoplasm round
cavities of a circular shape, and in each such ring the cyto-
plasm becomes divided into cell bodies corresponding to the
nuclei. The cells so formed are independent of one another,

372 RICHARD EVANS.

and develop collars and flagella (figs. 16—22, fl. c. = c.c¢.).
In the meantime the cells with granular nuclei which are still
in the interior are arranging themselves to form a lining to the
cavities which have appeared in the syncytium-like parenchyma
of the young sponge. The spaces later on become the sub-
dermal cavity, the inhalant and exhalant canal systems, and
the gastral cavity; in short, all the cavities of the sponge,
other than the chambers, are lined by cells with granular
nuclei. This description of the fate of the flagellated cells
applies equally to the larval types C and D. The history of the
flagellated cells, however, is more easily made out in the type
D, owing to the almost complete absence of the small cells
which compose the cell groups, than in type C, where the cells
in question are numerous.

In the type C, flagellated chambers are in many cases fully
developed at the time of fixation, possessing cells which are
adorned with collars and flagella (figs. 7 and 7a). These cells
are found in groups in type B, and are seen to possess at a
certain stage a nucleus so exactly like that of the flagellated
cells as to be indistinguishable from it (figs. 5¢ and 9a and 6).
During the change from type B into type C the nucleus alters
in character, and assumes the definitive structure of the nucleus
of the collar-cells. Hence in the complete history of the
flagellated cells on the one hand, and of the small cells in the
“cell groups” on the other, we have two stages in which their
nuclei are exactly like one another. First, the nuclei of the
small cells of the groups at the time when the cytoplasm is as
yet incompletely divided up are exactly like the nuclei of the
flagellated cells, so long as the latter retain their position in
the superficial layer of the larva. Secondly, they resemble one
another so much as to be indistinguishable when both have
attained the definitive structure of the nucleus of the collar-cell.
Though these two developmental stages correspond exactly,
the intermediate conditions through which they pass are very
different ; the nuclei of the cell groups pass gradually from one
stage into the other without any interruption, while those of
the flagellated cells pass through a most extraordinary series of

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 373

changes in connection with the plasmodial aggregations
of cells.

The megascleres are produced in the cells with vesicular
nuclei, and arrange themselves in various ways at the meta-
morphosis. Many of them place themselves almost vertically
and raise up the dermal membrane, giving the upper surface
of the young sponge an irregular conformation of hills and
valleys, as it were. Below the dermal membrane is the sub-
dermal cavity, lined by cells with granular nuclei above and
below, while in the centre of the little sponge are the large
spaces of the exhalant system opening by the osculum.
Along the surface, on the outer aspect, are the inhalant ostia
opening into the subdermal cavity. The osculum, which
occupies a central position, is at first on a level with the sur-
face, but soon becomes situated at the tip of an erect oscular
tube. All the cavities in the interior are lined by flat epi-
thelium cells with granular nuclei, and other cells of this class
are also seen in the spaces between the chambers accompanying
the cells with vesicular nuclei. The microscleres are secreted
by cells with granular nuclei.

Now that a brief historical account and a summary of the
results of the work embodied in this paper have been given, the
problem to be solved in the more detailed account which is to
follow can be stated.

It will be necessary to show that there are several types of
larvee which differ considerably from one another as regards
the structure of the inner mass, and that owing to these dif-
ferences the metamorphosis of Spongilla may take place in
more than one way,—that is, the metamorphosis depends on
whether the larva at the time of fixation has developed in the
direction represented by types B and C, or in that represented
by type D.

It will be shown that in the first case there will result two
slightly different methods of metamorphosis, the differences
between which will depend on the age of the larva. If the
larva fixes in the condition of type B, there will be small cells
in the interior not yet developed into chambers. If, on the

374 RICHARD EVANS.

other hand, it has reached the condition of type C, there will
be chambers completely formed which have been derived from
the vesicular cells of the larva. In both cases the flagellated
cells of the larva develop into collar-cells, passing through
the same series of changes as they do in a larva of type D.

In the second case, due to the fixation of a larva of type D,
the flagellated chambers arise almost entirely from the flagel-
lated layer of the larva, and scarcely at all from the cell groups,
which are very few in number and possibly absent.

The divergences in the metamorphosis will be seen to be
differences of degree which merge into one another, but which
uevertheless produce strikingly different appearances in the
early stages of development. Further, it will be evident that
the variations in the metamorphosis are due only to a very
slight degree to differences in the age of the larve, and depend
almost entirely on the divergent courses of evolution which
will be described in the account of the larva.

II. DetarLep Account or THE HistToLocy oF THE LARVA AND
OF THE METAMORPHOSIS.

A. Histology of the Larva.

Introductory Remarks.—In describing the histological
structure of the free-swimming larva and the changes through
which it passes before, during, and after metamorphosis, it
will be necessary to bring into line all the phenomena pre-
sented by the developing larva and the young sponge, and to
avoid having recourse to the easy method of explaining diffi-
culties as abnormalities. Further, it will be convenient to
avoid using the words ectoderm, mesoderm, entoderm, and
such terms, because they are liable to lead to confusion, and
to prejudge the questions of homology which may arise.

By describing the histology of the larva, with a special view
to the changes which go on during the free-swimming part of
the life history, many of the errors of previous observers will
be avoided, and many of their mistakes will be explained; while

STRUOTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 375

several statements, which appear at present to be conflicting,
will be brought into line, and a more complete description
than has hitherto been given will result.

The larve do not appear to hatch, or, at least, to swim out
of the mother colony, at the same age. It is quite possible
that they pass into the excurrent canals at approximately the
same stage of development, but owing to obstructions are
unable to swim out. They leave the mother colony by way of
the osculum, and are carried along by the current which issues
from that opening. They, however, soon gain control over
their movements, and swim to the surface of the water, darting
down instantaneously should they be disturbed in any way.
In all their movements the glistening broad end is directed
forward. When they reach the surface they either swim about
for a time, or settle almost immediately on the side of the
vessel, though they do not at first fix themselves to it. The
greater number will, if left undisturbed, fix to the side of the
vessel, while others will fix to the film of air at the surface of
the water. The interval of time between the actual swimming
out of the mother colony and the fixing is not constant. They
appear to fix in greater numbers between three and six o’clock
in the afternoon than at any other time.

The larva is oval in form, or, perhaps more correctly, egg-
shaped, and is covered with a uniform coat of flagella. The
broader anterior end has a glistening appearance, owing to its
containing a large cavity, the “larval cavity,” filled with a
kind of clear jelly. The narrower posterior end presents an
opaque appearance, owing to its being composed of cells which,
in the younger larve, are full of food material, which, as deve-
lopment proceeds, becomes used up.

In the following more detailed description of the cellular
elements of the larva the same plan will be maintained as was
followed in the abridged account given above.

In the histological composition of the larva we can distin-
guish—

(1) The flagellated layer at the surface (figs. 9 and 11).

(2) The inner mass (figs. 9 and 11).

376 RICHARD EVANS.

(a) The Flagellated Layer.—In contrast with what
occurs in the greater number of sponge larve, the flagellated
layer completely surrounds the inner mass, each of its consti-
tuent cells being provided with a flagellum (figs. 88 a, b, c, and
d). All the flagellated cells present the same characters,
only differing slightly from one another in length; and the
nuclei, which in all cases are situated near the bases of the cells,
are consequently not on the same level (figs. la, 5, 6, 7, and
38 a and 6). Their general form is, on the whole, a charac-
teristic one so long as the flagellated layer retains its position
at the surface. The cell body, which lies almost altogether
externally to the nucleus, is constricted so as to present the
appearance of having a waist. The narrower part measures
no more than 14 » across; while the broader part, which is
situated externally, measures from 2 to 24 mu, the length of the
cell varying from 5} to 74 p.

Owing to the constricted condition of the middle portion of
the cells, spaces are often seen between them. The nuclei,on the
other hand, are closely wedged against one another, while the
external ends fuse with one another to form what might almost
be described as a thin membrane in which no cell outlines can
be distinguished. The cytoplasm contained in the external
position is more opaque than that existing in the waist of the
cell. This condition is probably due to the presence of very
fine granules, which give it a denser consistency. The cells
appear as if they were suspended from this membrane, only
touching one another at their inner ends where the bulging
nucleus is situated (figs. 1 and 5).

The flagellum with which each cell is provided lies in part
outside and in part inside the cell. The portion which lies
outside is at least as long as the cell itself, and appears to
taper to a point, its base being in some cases surrounded by a
cone-like elevation of the cytoplasm. The portion which lies
inside the cell passes down to the neighbourhood of the nucleus,
and in many cases presents a small swelling which lies on
the nuclear membrane, and possibly represents the centro-
some. Sometimes the flagellum, instead of ending in a small

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 3877

body situated close to the nuclear membrane, seems to spread
over the membrane in question, which appears to be drawn out
like the outer coat of an onion. The internal part of the
flagellum can be traced along its whole length in all the larval
stages, as well as in some young fixed stages in which the
flagellated layer has not completely migrated into the interior
(figs. 29 a and 38 a—d).

The nucleus of the flagellated cell, when cut tangentially to
the surface of the larva, presents a circular section nearly 2 p
in diameter, but when cut radially it has an oval appearance,
or, perhaps more correctly, that of the outer half is almost
cone-shaped, while that of the inner half is semicircular. It
measures about 2 mu across and 24 win length. By combining
these two sections it is evident that the shape of the nucleus
is that of a cone with a hemisphere at its base, or, in other
words, it is onion-shaped (figs. 1 and 29 a).

I must here apologise for going still further into detail con-
cerning the flagellated cells. This is necessary, however, in
order to compare the cells here described as “cell groups ”
with the flagellated cells.

The cytoplasm of the flagellated cells is clear, and contains
at most no more than three or four small round granules,
which cannot be seen in ordinary preparations ! (figs. 38 a—d).

The nucleus has a thicker and better defined nuclear mem-
brane than any other class of nuclei found in the larva. In-
ternally the nucleus contains a few small and irregularly shaped
chromatin granules, one of which may exceed the others in
size, scattered at the nodes of a somewhat coarse nuclear re-
ticulum. When there is one larger granule, as is often the
case, it usually occupies a central position, and the threads
pass straight from it to the nuclear membrane; but when there
are several granules approximately equal in size, which is the

1 These small granules could not be seen in sections of larve preserved
either in Flemming’s fluid or in Perenyi’s fluid, or in absolute alcohol, and
mounted in Canada balsam. However, they were distinctly seen in prepara-

tions of larvee preserved in osmic vapour, stained in picro-carmine, and
mounted in glycerine.

378 RICHARD EVANS.

usual rule, the granules are evenly distributed throughout the
whole nucleus, and are united to one another and to the
nuclear membrane by means of threads. These differences,
however, are only differences of degree, aud not of quality, for
there are always a great number of intermediate stages.

Tosum up, the flagellated cell may be described as elongated,
and constricted at the middle. ‘The nucleus, situated at the
base of the cell, has a thick membrane enclosing a network
of threads, at the nodes of which from one to five chromatin
granules are situated.

(5) The Inner Mass.—The inner mass may consist of as
many as three kinds of cells, of which two at least are always
present. Without any further general remarks I shall pro-
ceed to describe these three kinds of cells, leaving certain
questions to be dealt with in the appendices, e. g. the origin of
the “cell groups,” the enclosures found chiefly in the cells
with vesicular nuclei, and the development of the spicules and
changes which take place in the nucleus of the scleroblast.

(a) The Cells with Granular Nuclei.—These cells are
found to occur in two positions, at least, out of the three which
they may occupy. They are always found as a more or less
complete layer immediately under the flagellated cells, and as
the larva grows older the completeness of the layer becomes
more evident. Secondly, they are always present in the imme-
diate vicinity of the larval cavity, especially the anterior moiety
of that cavity ; and here, again, the layer becomes more com-
plete with age. And thirdly, they may occur in the interior
of the solid part of the inner mass, where, however, they do
not appear to be present at first, save in exceptional cases, and
even then only in very small numbers.

In the first and second of these positions the cells with
granular nuclei tend to become flattened, and consequently in
a radial section present an oval form; while in the third they
present an irregular shape, pushing their processes between the
other elements of the inner mass. As the larva becomes older
they increase in number at the expense of the cells with vesicular
nuclei as well as by their own division. The proportion which

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 379

they bear to the number of other cells present varies consider-
ably. The size of the cell, which probably depends on the
number of divisions that have taken place, is far from constant,
even in the same larva. They become smaller as the larva
grows older, measuring from 8 to 10 over one of their flat-
tened surfaces, but they are really too irregular in form to admit
of proper measurement. Sometimes a spiny microsclere appears
in these cells, even in the larva.

The cytoplasm is usually clear, the cell body having at most
only a few enclosures. A ‘‘ nutritive vacuole ” is seldom seen
in them, but the “ yolk bodies” are often present, though few
in number.!’ They also contain some of the small refringent
granules to which reference was made in describing the flagel-
lated cells (fig. 38 e).

The nucleus is either spherical or subspherical. Its usual
tendency is to assume the latter form when the cell is in either
the first or the second of the above-mentioned positions, and
the former when it is in the third. The above facts have no
importance other than that the general configuration of the cell
influences the shape of the nucleus,—that is, when the cell
becomes flattened the nucleus acquires a slightly compressed
form. The size of the nucleus, like that of the cell itself, and
probably for the same reason, is variable, ranging from 3} to
54 wu in diameter.

The nuclear membrane is much thinner than that of the
nuclei of the flagellated cells, often to such an extent as to be
difficult to distinguish from the surrounding cytoplasm. The
interior of the nucleus is occupied by numerous small and
irregularly shaped granules, placed at the nodes of a close and
fine nuclear reticulum.

There is no reason that I can see to be found in the characters
of these nuclei which would justify the division of this class of
cells into two, namely, “ epidermal cells”? and “ intermediate
cells” of Delage. ‘They appear to be only one class of cells,
which are capable of being modified into flat epithelium

1 For a further discussion of the “nutritive vacuoles” and “ yolk bodies ”
cf. Appendix A, pp. 422—425.

VOL, 42, PAR’ 4.—NEW SERIES. DD

880 RICHARD EVANS,

whenever they are situated near asurface. Whether the surface
in question is internal or external seems to be immaterial.

To sum up, the characters of the cells with granular
nuclei are as follows :—an irregular or flattened form ; clear
cytoplasm, which may contain a few enclosures; and a fairly
large nucleus with a thin nuclear membrane, which encloses
a great number of small granules placed at the nodes of a fine
reticulum.

(8) The Cells with Vesicular Nuclei.—These cells
for the most part occupy the posterior moiety of the inner
mass. They are seldom found in the anterior region,
between the layer of flattened cells which line the larval
cavity and the flagellated layer. Inthe youngest larve they
are not separated from the larval cavity situated anterior to
them, nor from the flagellated layer at the posterior end, but in
older ones a layer of cells with granular nuclei is developed in
both of these positions. Consequently the cells with vesicular
nuclei, mingled with cells of other kinds, become restricted to
the internal part of the solid posterior end of the inner mass.
These cells are by far the largest in the whole larva, and con-
trast with the cells which possess granular nuclei in having a
perfectly definite outline as well as an oval or circular form,
at least in the younger larve ; in older specimens they seem
to lose to some extent their regularity of outline and compact-
ness of form. Another point of contrast between these two
classes of cells is that the cells with granular nuclei contain but
a few enclosures, while those with vesicular nuclei are always
possessed of one or more “ nutritive vacuoles ”’ and several
“yolk bodies.” The number of these enclosures found in the
individual cells depends on the stage of development of the
larva. In young larve the cells often contain as many as
three or four nutritive vacuoles and numerous yolk bodies,
while in older larve they almost invariably contain only one
nutritive vacuole and but a few yolk bodies. These cells,
in common with the two classes of cells already described,
possess some of the small refringent granules mentioned in
deseribing the flagellated cells,

STRUCTURE OF THE LARVA OF SPONGILLA LAOUSTRIS. 381

The nuclei of these cells are the largest in the whole larva,
measuring sometimes as much as 7 in diameter, but their
size, like that of the cell itself, depends on the state of deve-
lopment of the individual. The nuclear membrane, like that
of the granular nuclei, is thinner than that of the nuclei
of the flagellated cells, but, as a rule, is easily made out. The
centre of the nucleus is occupied by the central corpuscle—
the so-called nucleolus,—which, again, like the cell and its
nucleus, varies considerably in size. The space between the
central corpuscle and the nuclear membrane is occupied by
a variable number of granules situated at the nodes of the
nuclear reticulum. When the number of granules is small
they are comparatively large in size, and the threads of the
nuclear reticulum are few and coarse. On the other hand,
when the number of granules is large they are small in size,
and the threads of the reticulum are correspondingly close and
fine. The granules in these nuclei vary also as regards position.
In some nuclei the granules, whether exceedingly numerous
and small, or few and comparatively large, are concentrated
in the neighbourhood of the nuclear membrane, leaving a clear
zone round the central corpuscle ; in others the granules are
evenly distributed throughout the whole space, and consequently
the clear zone is absent (fig. 41 a, and fig. 40 a).

These differences in the structure of the vesicular nuclei
seem to suggest that the class of cells here described consists
of several different kinds of cells. The term “ cells with vesi-
cular nuclei” may, in fact, be regarded simply as a convenient
one to hide our ignorance of the division of labour that has
already come into existence among these cells. The fact that
megascleres are being developed in some of these cells may be
taken as evidence in favour of this view (figs. 36 a, 6, and c).

To sum up, these cells are the largest in the whole larva,
occupying the interior of the inner mass, and containing a
number of enclosures in the form of “ nutritive vacuoles,”
‘yolk bodies,” and “ refringent granules.”’ They possess a
nucleus which contains a large central corpuscle, sometimes
surrounded by a clear zone. The nuclear membrane is thin,

382 RICHARD EVANS.

and encloses a variable number of small granules situated at
the nodes of the nuclear reticulum.

(y) The Cell Groups.—These groups of cells are situated
in the interior of the solid posterior part of the inner mass.
The number of cells which constitute a group is highly
variable. The groups seem to be present in all the older
larvee, though they are far more numerous in some individuals
than in others. They are so few, however, in some older
larvee that they might have been overlooked had it not been
that in some individuals they are more numerous. In some
cases the groups are well defined and isolated, while in others
they run into one another owing to their close proximity.
Sometimes they are surrounded by a well-developed membrane,
formed, apparently, by the cells with granular nuclei, this
being especially true of the groups situated near the centre of
the solid posterior end of the inner mass. In the youngest
larvee in which the groups occur the cytoplasmic bodies of
the individual cells cannot be distinguished from one another ;
but in older larve the cells are perfectly independent, and may
go so far, even in the free-swimming larva, as to form collars
and flagella. In the latter case they enclose a cavity, which
is that of the flagellated chamber. Just as the cells them-
selves exhibit a progressive development from an incompletely
divided condition to one in which the cells are free from one
another, so also the nuclei pass through a series of changes.
At one stage they resemble those of the flagellated cells to
such an extent that the same description might be made to
apply to both, while at a later stage they assume the definitive
characters of the nuclei of the collar-cells.!

Now that a general description of the cell elements which
enter into the histological composition of the larva has been
given, it is necessary to discuss the relative quantities, occurring
in the different types of larva, of the elements above enume-
rated. Four main types of larve may be distinguished, which

1 For a further discussion of the possible origin of the cell groups ef.
Appendix B, pp. 422—425.

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 383

are connected by transitional stages, but which may never-
theless be described apart from one another. The flagellated
cells are alike in all the types, and need no further discussion.
The only difference that exists between their early and late
condition is that their nuclei are nearly on the same level, and
consequently that they are almost the same length. In addi-
tion to this, in the older larve the cells with granular nuclei
are seen to push their processes between the flagellated cells,
indicating that fixation and metamorphosis cannot be long
delayed. The differences which exist between the several
types of larve to be considered concern the inner mass,
and the special features of each type will be described in the
following pages.

Special Features of Type A.

The inner mass in this type consists of only two kinds of
cells, namely, cells with granular nuclei (a) and with vesicular
nuclei (3).

The former are almost entirely confined to two positions,
occurring first as an incompletely developed layer of cells
immediately below the flagellated layer; and secondly, in the
vicinity of the larval cavity, where, however, they are almost
completely limited to its anterior border. They have not yet
assumed the flattened shape which they acquire in later stages in
either of these positions.

The latter have the monopoly of the interior of the solid
posterior end of the inner mass. They enclose in many cases
three or four nutritive vacuoles, together with numerous yolk
bodies, and occasionally a developing spicule is found in them.

Microscleres and cell groups are not found in this type,
which is evidently the youngest of all free-swimming larve
(fig. 1). From it all the others are developed, and it is highly
probable that the larva never fixes while in this early state of
differentiation of the inner mass, but proceeds to develop into
one or other of the types which remain to be described.

384 RICHARD WVANS.

Special Features of Type B.

In this type the inner mass consists of three kinds of cells,
namely, cells with granular nuclei (a), with vesicular nuclei
((3), and cell groups (y).

The cells with granular nuclei are now sufficiently developed
to form fairly complete layers in the two positions in which
they were recognised in type A, and have also appeared in the
interior of the solid posterior end of the inner mass. ‘They
have also become flattened save in the interior, and their nuclei
are in some cases subspherical in form.

The cells with vesicular nuclei are far less numerous than
in type A. They have become slightly smaller in size, and
less definite and regular in outline. The nucleus also is
smaller, and the central corpuscle is evidently diminishing
in size and breaking up. ‘These facts point to the gradual
change of some of the cells with vesicular nuclei to such as
have granular nuclei.

The cell groups which occur in this type are a new and
most important feature which did not exist in type A.
The cytoplasmic bodies of the individual cells which make up
the groups are as yet incompletely divided from one another,
so that the nuclei present the appearance of lying at the
periphery of a mass of cytoplasm. It is only when a group is
looked at in surface view that anything resembling a dividing
line between the nuclei can be seen, a fact which indicates
that the division of the multinucleated cytoplasmic mass
proceeds slowly from the surface towards the centre (figs.
Qaand b,g.c.). The amount of cytoplasm corresponding to
each nucleus is not more than that contained in a flagellated
cell. The nuclei in a radial section of a group present the
same onion-shaped form as those of the flagellated cells. The
nuclear membrane is thick, and the granules are small and few.
In short, these cells, while differing completely, on the one
hand, from the cells with granular nuclei, both as regards the
size of the cell and the characters of the nucleus, are almost
identical, on the other hand, in both these points with the

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 3880

flagellated cells. These considerations suggest irresistibly the
conclusion that they are developed from cells identical in
character with those which gave rise to the flagellated cells,
and possibly in the same way (figs. 5, 5 a—-c, 9 a, 0).

I have no reason for thinking that this type of larva often
fixes itself, though it may occasionally do so. As a rule, how-
ever, it develops further, aud gives rise to the larva of type
C, the special features of which will be described next.

Special Features of Type C.

In this type of larva the same three kinds of cells are found
as in type B, but, owing to the differentiation and development
which have taken place, the differences of the cell character-
istics are considerable.

The cells with granular nuclei are much more irregular in
shape than in the larve described above. In this larva they
branch extensively, and unite by their processes to form mem-
branes which are still more or less incomplete, both under the
flagellated layer and as a lining to the larval cavity. Many of
those which are situated in the interior have changed con-
siderably in shape, having in many cases flattened out so as to
surround small lacunar spaces or canals (fig. 14 a). The
lacune in question are destined to become the exhalant system,
and the cells surrounding them to become the flat epithelium
of the same. Simultaneously with the flattening of the cell
body the shape of the uucleus is also changed to that of a bi-
convex disc. In this type some of the cells with granular nuclei
may develop a spiny microsclere (fig. 37 a).

It is necessary to point out here that the spaces and canals
lined by the cells with granular nuclei are not in any way
comparable to the spaces surrounded by the cell groups. In
the latter case, the spaces are the cavities of future chambers
which are either appearing or have already appeared in the
larva. The small spaces and short canals here described are
lacunar cavities which have the same relation in many cases
to the cavities of the cell groups as the exhalant canals have

386 RICHARD EVANS.

to the chamber in an adult sponge. In reality the communi-
cation between these two kinds of spaces is an already deve-
loped exhalant pore or apopyle. It may further be pointed
out that these spaces, whether canals or chambers, are struc-
tures formed in situ from thecells of the solid pos-
terior end of the inner mass, and not as ingrowths
into it from the layer of cells which line the larval
cavity. Should some of the short canals appear to open into
the larval cavity, the communication is one which has been
secondarily acquired during the development.

The cells with vesicular nuclei are less numerous, more
irregular in shape and outline, and smaller in size than they
are in types A and B. They become less numerous owing,
on the one hand, to their conversion into cells with granular
nuclei, and, on the other hand, to the formation out of them
of the cell groups. They become more irregular in shape
owing apparently to the stored-up food material being used up,
giving them greater facility for change of form and the
exercise of their wandering function. They become smaller
in size owing to repeated cell division. In this type it is
quite exceptional to find a cell measuring more than 10 pu
across, while in type A they often measure from 12 to 15 pn,
and in type B from 10 to 13 pw. A nucleus measuring 7 pu
across is often seen in types A and B, but in type C the
largest vesicular nucleus seldom exceeds 5} uw in diameter.
On the other hand, a vesicular nucleus measuring less than 5
across is scarcely found in types A and B, but in type C they
constantly occur. The same reduction of size is noticeable in
the central corpuscle (or nucleolus) of the vesicular nucleus.
In types A and B it often measures 2 « across, but in the type
now under consideration it averages only 1} windiameter. In
types A and B these cells contain one or more nutritive
vacuoles and several yolk bodies, but in type C they seldom
contain more than one of the former, which have also become
slightly reduced in size. The number of the yolk bodies is
also greatly decreased, though they are by no means completely
absent. Moreover, the cells in question often produce spicules

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 387

which have a tendency to protrude through the flagellated layer,
though they are still completely surrounded by the cells which
produced them. The various changes undergone by the cells
with vesicular nuclei suffice, apart from any other reason, to
prove that the order of the types above adopted is the true one
as regards age: i.e. that type A is the youngest of all, because
the interior of the solid posterior end of the inner mass is
monopolised by the large cells with vesicular nuclei; and that
type B must be younger than type C, because the characters of
the remaining cells with vesicular nuclei are almost the same as
the characters of those cells in type A, though the numerical
proportion which they bear to the other cells has changed con-
siderably (figs. 11, 11 a, 12, 18, and 13 @ and 8).

The individual elements of the cell groups which have been
described in type B as being incompletely divided, and as
having their nuclei lying near the periphery of a common
mass of cytoplasm, are, in this type, completely independent of
one another. ‘The cell groups now are as numerous as they
were in their incipient condition in type B, a state of things
markedly different from what exists in the type of larva still
to be described (figs. 7, 7a, 11, and lla). The cells which
constitute the groups are fairly uniform in size. They
possess a nucleus which measures about 25 in diameter,
showing a slight increase in size as compared with the nuclei
found in the incompletely divided groups of type B, as well as
a greater number of chromatin granules and a higher degree
of complexity in the nuclear reticulum. Sometimes, however,
there may be seen in a group one cell larger than the others
with a nucleus of corresponding size. ‘The existence of such
cells is probably to be explained as being the result of inde-
pendent growth after the cells have become liberated from the
incompletely divided groups of type B (fig. 11 a).

The number of cells which constitute a group varies con-
siderably, just as the number of nuclei vary in the multi-
nucleated masses of cytoplasm in type B. There may be no
more than four or five cells in a radial section of a group, or
there may be a dozen or even more, but it is probable that the

388 RICHARD EVANS.

number of cells that can take part in the formation of a group
cannot pass a certain limit. Consequently, as the individual
elements of the cell groups multiply by division, it seems a
well-founded conclusion that the groups themselves multiply
in the same way.’ .

The large number of cells found in the bigger groups might
be accounted for on the supposition that the individual cells
have divided since their liberation from the groups of type B.
Though this supposition may be to some extent true, it is
more probable that the difference in the number of cells which
constitutes a cell group or a flagellated chamber in type C
corresponds to a similar difference in the number of nuclei in
the multinucleated cytoplasmic masses of type B.

The cytoplasm of the individual elements of the cell groups
is usually clear, but may contain a few granules which are
either reduced yolk granules or bodies of the same nature as
the refringent- granules already described as existing in the
flagellated cells, and also in all the cells of the inner mass.
Another very remarkable fact is that they occasionally contain
a small nutritive vacuole, a fact which points to their origin
from the cells with vesicular nuclei.

Another feature of some of the cell groups in this type is
that their individual cells develop collars and flagella in all
respects like those of the collar-cells in the adult sponge. It
is true that the greater number of these groups consist of
cells which have not as yet developed these organs ; and it may
be further stated that unless great care is taken in the pre-
servation of the larve, all of them, without exception, will be
without collars, and will present only a kind of process possess-
ing a more or less conical shape, and pointing towards the
cavity inside the group of cells. The collars unite by their
margin to form the so-called “ membrane of Sollas” (figs. 7
and 7 @).

These cells, which from this stage onwards may be called
collar-cells, present the same general arrangement and the

1 For further remarks on the multiplication of the collar-cells and of the
flagellated chambers ef. Appendix D.

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 389

same relation to the exhalant pore already described as they
do in the fully metamorphosed larva or adult sponge. The
cells have a horseshoe arrangement when seen in a radial section
of agroup. The opening, which points towards the lacunar
spaces already mentioned, is a true exhalant pore or apopyle
(fig. 7 a, B).

The nucleus of the collar-cells is but very slightly larger than
it was in the incompletely divided condition seen in type B.
In the interior the chromatin granules have become more
numerous, and the nuclear reticulum more complicated. Con-
sequently there is at present no very striking difference between
these nuclei and some of the nuclei of the cells which have
been described as having granular nuclei. But for the dis-
covery of the origin of these cells in the larve of type B, their
existence as aseparate class would probably have remained un-
discovered and even unsuspected, and they might easily have
been placed among the cells with granular nuclei—an error
which would have almost certainly led the way to another
mistake, that, namely, of describing the cavities of the cell
groups as incipient exhalant canals.

So important are the main features of this type that they
may be briefly summarised. In the first place there are two
kinds of cavities, those of the flagellated chambers on the one
hand, and the lacunar spaces of the incipient exhalant system
on the other hand. In the second place these two kinds of
cavities are surrounded by two different kinds of cells,—the
former by the individual elements of the cell groups or collar-
cells, and the latter by cells with granular nuclei or flat epithe-
lium. The two cavities above described are, in certain cases,
already in communication by means of the exhalant pore.

The important point to emphasise is the fact that the
collar-cells, in this type of larva, are developed in the interior
before any flagellated cells have made their way there;
and further, that they have been developed by fragmentation
of the nucleus and the subsequent division of the large cells
with vesicular nuclei found in type A, and have passed through
the condition found to be characteristic of type B.

390 RICHARD EVANS.

Special Features of Type D.

In this type of larva the same three kinds of cells take part
in the formation of the inner mass as in types B and C, but
the numerical proportion is very different. In types B and
C the individual elements of the cell groups are as numerous
as either the cells with granular nuclei or the cells with vesi-
cular nuclei; but in type D the number of the cell groups is so
small that, had they not been discovered in the other types,
they would most probably have been overlooked. However,
they seem to be present always, though the groups often
consist of only four or five cells, and are few in number
(fig. 6a, g.c.).

The cells with granular nuclei occupy the same position as
in type C, and present individually the same characters. A
noticeable feature in connection with them in both larvee is
their irregularity of form when they are examined from the
flattened surface. In consequence of this, the so-called mem-
branes, one of which lies under the flagellated layer and the
other lines the larval cavity, are far from being complete even
when they are best developed (fig. 10). Another feature,
equally noticeable with the above, is that the cells with
granular nuclei situated in the solid part of the inner mass
scarcely ever flatten out to surround lacunar spaces or canals,
which are almost completely absent from this larva. Their
nucleus varies considerably in both shape and size, measuring
from 3 to 44 across, ‘but remains uniform in structure,
whatever their position. The chromatin granules are always
small in size, irregular in shape, and situated at the nodes of a
nuclear reticulum which consists of fine threads enclosing
small meshes.

The ‘cell groups” are few, and consist of a small number
of cells, as has been mentioned above. Their individual
elements average about 5 in diameter, and the nucleus varies
from 1? to 24 in diameter, a size which agrees closely with

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 391

that of the nuclei of the flagellated cells. Further, they
possess the same structure as the nuclei of the latter cells—that.
is, a few granules situated at the nodes of a coarse network of
threads and a thick nuclear membrane. ‘There is no doubt but
that these cells belong to the same class as those of the cell
groups found in types B and C, and their presence, however
few they may be, accounts for the existence of an occasional
chamber in a larva of type D, immediately after fixation.

The cells with vesicular nuclei present much the same
characters and general distribution as they do in type C.
They are both less numerous and smaller than they are in
types A and B. The number of yolk bodies which they
contain is much smaller than in types A and B. The
nutritive vacuole of course is a constant feature of the cell,
with a vesicular nucleus in all stages, but it appears to
decrease in magnitude simultaneously with the decrease in
size of the cell. The inner mass, therefore, in this larva
consists of a fair number of cells both with granular and with
vesicular nuclei, together with a few of the elements of the cell
groups, in contrast with the large number of them found in
types B and C (figs. 6 and 6a).

General Remarks upon the Relationship of the
Larve.—tThe author’s views with regard to the relationship
of the four larval types already described have been stated
above (p. 869). It only remains to discuss these relations in
greater detail.

There can be no doubt that type A is the youngest larva of
the four types described above. The fact that the cells with
vesicular nuclei are here relatively more numerous than they
are in any other type supports this view, because they are the
most primitive cells of the larva. They retain, in fact, in
type A almost all their blastomeric characters, which in the
other types they gradually lose, so far as the contents of the
cells are concerned, though not perhaps in the physiological
sense.

Histogenesis seems to advance or to be retarded in separate
regions of the larva at different times. At the close of

392 RICHARD EVANS.

segmentation, while the embryo is in the maternal follicle, all
the cells possess the same characters ; but between the above
stage and the time of hatching, the cells at the surface become
differentiated into flagellated cells externally, and cells with
granular nuclei beneath, while the cells in the interior of the
solid posterior part of the inner mass seem to be arrested in
their development. As the results of these changes we obtain
the larva which has been described above as type A. It has
a flageilated layer at the surface, and cells with granular nuclei
beneath and in the neighbourhood of the larval cavity. The
posterior end, however, is occupied by the large cells with
vesicular nuclei, which represent both morphologically and
physiologically a number of unmodified blastomeres. After
this stage has been attained, further differentiation seems
to take place in the interior, while the flagellated cells, situated
as a layer completely surrounding the larva at the surface,
are, for a while, retarded in their development.

The changes in the interior consist in the further differen-
tiation of the cells with vesicular nuclei which have, as above
described, retained the characters of blastomeres, but which
no longer give rise to flagellated cells on the exterior as well
as to cells with granular nuclei. The place of the flagellated
cells is taken now by the individual elements of the cell groups.
Thus another argument in favour of the homology of the small
incompletely separated cells, seen in type B, with the flagel-
lated cells, is furnished by the fact that, when the cells with
vesicular nuclei cease to give rise by differentiation to flagel-
lated cells, they begin to give rise to cell groups. This is an
important point when taken, not by itself, but in conjunction
with the arguments already put forward, based upon the identity
in size of the individual elements of the cell groups with the
flagellated cells, and the similarity of their nuciei.

There results, as the outcome of these internal changes, the
larva described above as type B, from which type C is derived
by further differentiation along the same line. The multi-
nucleated masses of cytoplasm in the former divide into as
many corpuscles of cytoplasm as there are nuclei in the whole

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 3893

mass, and in this way give rise to the chamber-like groups
found in the latter. The chief difference between the elements
of the cell groups in type B and the flagellated cells at the
surface is the lack of a flagellum in the former and its presence
in the latter. But the cells which lack the flagellum in type B
surpass their rivals, so to speak, in type C, and develop not
only a flagellum, but a collar as well.

In order to lay more emphasis on the arguments which favour
a homology between the cell groups and the flagellated cells, it
seems advisable to summarise them briefly. First, the cells in
question are almost exactly of the same size; secondly, the
nuclei of the cell groups in type B are identical in characters
with those of the flagellated cells; thirdly, when the cells with
vesicular nuclei cease to give rise to flagellated cells they
begin to give rise to cell groups; and fourthly, though at first
devoid of a flagellum, the individual elements of the cell
groups develop both a collar and a flagellum, becoming the
collar-cells of type C.

The relationship of types A, B, and C to one another seems
to be a simple problem to solve, for the passage from A to B
and from B to Cis very gradual; but when type D is taken
into consideration, the problem of the relationship of the
larvee to one another becomes much more difficult to solve.
Type D cannot, apparently, be fitted anywhere into the above
series of larve, either in an intermediate position or at the end,
but can only have been produced directly from type A. The
almost complete absence of cell groups in type D is conclusive
against its origin from types B and C. On the other hand, it
cannot possibly have given rise to type B,in which many of the
cells with vesicular nuclei still retain the structural characters
of blastomeres. It would be equally impossible to imagine
type D giving rise to type C, for were the cells with vesicular
nuclei to proceed to divide and to produce cell groups, the
result would be a larva possessing quite different characters
from those of type C. ‘There would be a great number of cells
with granular nuclei on the one hand, and of cell groups on the
other, but hardly any cells with vesicular nuclei, It may there-

394. RICHARD EVANS.

fore be fairly concluded that type D has been produced directly
from type A.

From these considerations it seems that two divergent lines
of development can be distinguished in the larva of Spongilla,
of which the two culminating points are represented by types
C and D, type B being an intermediate stage between types A
and C. Type C is not necessarily older than type D, or vice
versa, but both of them have attained that stage at which
they usually fix themselves. The differences exhibited by
these larvee are in no way more striking than those which will
be found in the newly fixed stages, as the results of these
variations. In fact, the structure of the larve seems to cast
its shadow, as it were, over the whole period of metamor-
phosis, even up to the appearance of the young sponge.

B. The Fixation, Metamorphosis, and Further
Development of the Larve.

General Remarks on the Fixation of the Larve.—
There are two facts at least which tend to make the study of
the larve of Spongilla during fixation and metamorphosis a
laborious task, and to render difficult a correct interpretation
of the phenomena observed. In the first place there must be
taken into account the difference in the structure of the larvee
at the time of fixation, a difference which is the result of
divergent variation culminating in the types C and D. In the
second place there are found other differences, due to the fact
that some of the larvee fix themselves at an earlier stage than
others ; such differences are exemplified by the types B and C.
In the former case we are confronted with a diversity of the
most fundamental nature, one in which the numerical propor-
tion of the several classes of cells which build up the inner —
mass vary, and which no amount of delay with regard to
fixation can rectify. In the latter case the variation is not so
far-reaching in its effects, and the fixation need be delayed but
a very short time to obliterate them.

The types of larvee which appear to be ripe for fixation are
Cand D, but type B may occasionally settle down and undergo

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 395

metamorphosis. Type A seems to be incapable of fixation,
owing to the fact that the cells with granular nuclei are as yet
few in number. The immediate result of the fixation of types
B and C is absolutely different from that obtained when type
D settles down to its sessile life. When a larva fixes after
attaining the stage of development described above in type C,
flagellated chambers which have acquired their definitive
structure are present as well as the rudiments of the exhalant
system. But when a larva possessing the structure described
in type D becomes fixed, both the flagellated chambers and the
rudiments of the exhalant system are almost completely absent.

The Actual Fixation of the Larva.—The larva fixes
either by the anterior pole or by a point on the side not further
back than the line of separation of the larval cavity from the solid
posterior part of the inner mass. No larva was observed to fix
by the posterior pole, or by a point near it. The fixation is
brought about by the passing out of the cells with granular
nuclei that lie beneath the flagellated cells. The larva just
before it settles down turns about much in the same way as a
spinning-top does when about to end its spin. The point at
which the cells with granular nuclei make their way out
corresponds to the peg of the spinning-top in the above
comparison. That point in the flagellated layer seems to be,
as it were, paralysed. Atthe time when the cells with granular
nuclei have actually penetrated the flagellated layers, and are
beginning to spread themselves along the surface of fixation,
the motion of the larva as a whole inevitably ceases, though
the flagella of the flagellated cells situated elsewhere may
continue in motion for some length of time.

The Obliteration of the Larval Cavity after Fixa-
tion.—The persistence or obliteration of the larval cavity
depends upon the structure of the larva. When type D fixes
itself by the anterior end, the solid part of the inner mass, which
lies at the posterior end, approaches the surface of fixation, and
the larval cavity is thereby reduced to mere slits (fig. 19),
and ultimately disappears completely (fig. 16), the fixed larva
flattening out fairly symmetrically in all directions. But when

you. 42, PART 4.—NEW SERIES, EE

396 RICHARD EVANS.

the same larva fixes by the side the solid posterior part of the
larva topples over on to one side, and the larval cavity is
obliterated by the coming together of the opposite sides. In
this case the larva flattens out as before, but owing to the fact
that the greater number of cells are situated on one side,
it lacks the radial symmetry characteristic of the larva that
fixes by the anterior pole, and presents a temporary variation
of form, which, however, is soon lost. On the other hand, it is
very doubtful if the larval cavity is ever completely obliterated
at the fixation of type C, whether the larva settles down by the
anterior pole or by the side. Intype C either the chambers are
already formed—their individual elements being adorned with
collars and flagella—or they will be formed soon after fixation.
In either case the young sponge can ill spare, so to speak, the
time necessary for the destruction of the larval cavity, and the
subsequent new formation of the inhalant and exhalant systems
of canals. Besides, it has been shown above that in some
larve the chambers, being fully formed, already open into small
spaces and lacunar canals, which represent the beginnings of the
exhalant system, and that these canals open in turn into the
larval cavity, at least in some cases. Individuals possessing
the above characters have been seen soon after fixation to
possess so large a cavity, evidently derived from that larva,
that it is impossible to believe that the larve in question
would have lost it had they been allowed to develop to
maturity. The fact that the flagellated chambers and the
finer portions of the exhalant system have been developed
already, and that nearly all the food material stored in the egg
cell has been used up, render it absolutely necessary that the
canals along which the current carrying food material for the
young sponge is to pass should be in working order as soon as
possible. Taking these facts into consideration, it may be
fairly concluded that the larval cavity is not usually obliterated
after the fixation of type C, but becomes a portion of the
exhalant canal system and gastral cavity.

The Metamorphosis.—In describing the metamorphosis,
which is a process involving the reversion of the layers that

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 397

takes place during and after fixation, it will be necessary to
consider two extreme cases which will be the result of the
settling down to a sessile life of types C and D.

As far as possible the following arrangement will be pur-
sued. First, the features common to both types will be
described ; secondly, the special features of the two types will
be considered—in the first place of type C, and in the second
place of type D. The formation of the cavities and canal
systems in general will be followed in the subsequent pages.

The larva described above as type B, so far as its changes
need be considered, can be taken along with type C, for they
both present in common the important feature of possessing
cell groups,—either as small cells more or less incompletely
divided, or quite independent of one another, but not pro-
vided with collars and flagella; or as flagellated chambers
in which the cells are adorned with collars and flagella, and
have attained their definitive arrangement. There is also
another difference between types B and C when they are
about to fix themselves, which will be of necessity the cause
of a variation occurring in the structure of the young fixed
stages. The difference in question is the presence of a great
number of yolk bodies in the cells with vesicular nuclei in
type B, while in type C they are always few, though invari-
ably present. This is an important feature, the bearing of
which is only rightly appreciated when it is recognised that
there is a stage during the metamorphosis at which it is
almost impossible to distinguish the nuclei of the flagellated
cells from the yolk bodies.

The larva described above as type A need not be taken into
consideration, as it probably never fixes itself at so early a stage
in the development. Type D, on the other. hand, presents
extremely important features which must be specially described.
It agrees with type C to the extent of possessing cells with
vesicular nuclei which contain but few yolk granules. It
must be admitted, however, that there is nothing impossible in
a larva fixing itself which was in an earlier stage than type D,
but situated on the line of development from types A to D, and

398 RICHARD EVANS.

therefore comparable to type B, intermediate between types A
and C. Such a larva when settled down would contain
yolk bodies in the same way as type B does soon after
fixation. Hence it is to be recognised that, whether we deal
with larvee developed along the line of histological differentia-
tion passing from type A to type C, or from type A to type
D, we may come across specimens which possess cells with
vesicular nuclei which contain numerous yolk granules instead
of a few, as in types C and D.

(a) The Features Common to Both Larve at the Time
of Fixation and Metamorphosis—the Disap-
pearance of the Flagellated Layer from the
Surface, the Formation of the Flattened Epi-
thelium and of the Marginal Membrane.

The flagellated cells may pass into the interior either
individually or in groups of several cells. They generally
tend to pass in groups from the lower surface, that is the
surface of fixation, and individually from the upper sur-
face. The usual result of this difference is, that all the
flagellated cells which once occupied the lower surface are
well within the body of the young individual, while those of
the upper surface still form a more or less complete layer, for
a time retaining their flagella, though the cells are by no
means so closely packed as in the free-swimming larva
(fig. 29). However, there is a considerable amount of variety
in the mode of flattening out on the part of the different larvee.
Some appear to flatten so quickly that the best way to describe
it is to say that the larva appears to tumble into pieces almost
instantaneously. These differences are due, probably, to the
number of points on the larva through which the cells with
granular nuclei make their way out. If these cells break
through the flagellated layer in several places at the same
time, the result is a quick and rapid metamorphosis. If, on the
other hand, they burst out merely at one point on the surface,
that point will become the area of fixation, from which the

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 399

cells with granular nuclei will only spread gradually. The
flagellated layer will in consequence remain for a long time on
the upper surface, and immigrate from it slowly, the cells pass-
ing in one by one. This is the natural result of the breaking
out of the cells with granular nuclei at one point rather than at
many points. The internal pressure has thereby been decreased
to such an extent that the resistance of the flagellated layer is
enough to prevent the cells with granular nuclei spreading
over the upper surface until the pupa,! by its expansion, has
decreased the cohesion of the flagellated layer.

These two processes goon at the same time. While the flagel-
lated cells are passing in, the cells with granular nuclei struggle
to the exterior, The process is, apparently, a reciprocal one in
which both classes of cells take an active part. The reversion
of the layers does not take place so quickly on the upper as it
does on the lower surface, and this is especially true of type C.
There is also a slight difference between the cells with granular
nuclei situated at the upper and lower surfaces respectively,
the nuclei of those in the former position being smaller than
those of the cellsin the latter. The cells with granular nuclei,
after passing through the layer of flagellated cells and arriving
at the surface, become flattened out, their edges meeting one
another. In this way a continuous layer of cells is formed on
both upper and lower surfaces, derived from those cells which
in the free-swimming larva are situated below the flagellated
layer; that is to say, they formed originally a part of the
“inner mass.”

Owing to the large size of the nuclei of those cells which
pass to the lower surface, a most satisfactory and conclusive
proof is obtained that the flattened epithelium of the young
sponge is formed from cells which once lay in the interior
of the larva, and not from the flagellated cells at the
surface. The nuclei of the latter are small, and whatever
other change they may undergo at this stage, it is certain that

1 This term is used by Mr. HK. A. Minchin to indicate that stage in the

development which occurs between actual fixation and the appearance of the
dermal pores and osculum.

4.00 RICHARD EVANS.

they do not increase in size, which they would have to do
were they to become the nuclei of the surface layer of cells.
For these reasons the flagellated cells cannot possibly give
rise to the flattened epithelium of the lower surface, where
the nuclei of the cells forming it are many times as large as
those of the surface layer of the larva (fig. 16 a).

The cells with granular nuclei not only give rise to the epi-
thelial layers of the upper and lower surfaces, but also produce
the marginal membrane, which differs considerably in thickness
and compactness according to the length of time which has
elapsed since the fixation of the larva took place. At first
only a few cells are seen to creep outside the limits of the
body of the larva which is in the act of fixing itself, but the
number of cells that wander out seems to increase with a won-
derful rapidity. In a short time they present the appearance
shown in fig. 23, in which the cells have not as yet arranged
themselves so as to form a complete layer, for there are large
spaces to be seen between them. In their outward course they
seem to struggle on, passing over and across one another. The
outer margin of the as yet incomplete membrane is quite irre-
gular, and the cells which form it appear to be absolutely in-
dependent of their neighbours, the limit of each cell being well
defined (figs. 23, 23 a).

The cytoplasmic mass of the cell body is exceptionally clear,
and presents the appearance of an alveolar structure in which
the meshes are slightly elongated in the direction of motion.
Pseudopodia are not always produced by these cells as they
move out; in many cases they present at least a complete and
uninterrupted margin. Sometimes the cells, as they creep out,
carry with them some of the flagellated cells, and in many
cases they contain yolk bodies.

As the fixed individual becomes older the marginal membrane
becomes thicker and more compact in structure. It now consists
of two or three superposed layers of cells, save at the extreme
margin, where it is only one cell thick. The outer margin is
still quite irregular in places, owing to the continued outward
movement of the cells which constitute it; but later on this

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 401

irregularity disappears (figs. 24.and 24a). When the marginal
membrane has become so thick as to consist of two or three layers
of cells, there often appears a space between these layers, and
into this space flagellated cells, accompanied either by cells
with granular nuclei or by cells with vesicular nuclei, find their
way. The nuclei of the lower layer of cells, constituting the
marginal membrane, display the same difference of size from
those of the upper layer as was described above as existing
between those of the lower and upper surfaces in general. The
clearly defined limit characteristic of its cells at an earlier
stage is now no longer visible, and the extreme margin
presents in all cases a regular and unbroken edge. The mar-
ginal membrane seems to be nothing more than a continuation
of the dermal epithelium of the upper and lower surfaces, into
the inner portion of which the internal substance of the
sponge enters comparatively late, never passing into the outer
margin, which later on is retracted.

The Changes through which the Flagellated Cells
passat Fixation.—The passage of the flagellated cells to
the interior has been already described, but it must not be for-
gotten that there is considerable variation in the rate of change
of position, which affects, in its turn, the time necessary for
the complete enclosure of the cells in question in the interior
of the pupa. It often happens that they have completely dis-
appeared from the lower surface, while they still remain as a
fairly complete layer on the upper.

The flagellated cells after passing to tle interior undergo a
most extraordinary series of changes, during which the nuclei
seem to lose their internal structure so completely that at one
stage they are almost indistinguishable, except by chemical
reactions, from the yolk bodies. The changes in question have
already commenced in the larve represented in figs. 15 @ and
29 6, where a few cells are seen adherent to the surface of
some of the cells of the inner mass, i. e. to cells with vesicular
nuclei, and in some cases even to cells with granular nuclei
as well. The first sign of the disappearance of the ordinary

402 RICHARD EVANS.

structure of the nuclei of the flagellated cells is a certain
amount of contraction, which in itself suffices to account to
some extent for the density of structure found in them during
these stages, though by no means sufficient to explain the whole
of it. Another reason is probably to be found in the fact that
when the flagellated cells pass to the interior they come within
reach of a quantity of food material stored up in the cells with
vesicular nuclei, and available for their use. When at the
surface they are to all intents and purposes starved, the result
being the arrest in the progress of their development that has
already been mentioned when describing the larva. In conse-
quence of this partial starvation, the flagellated cells, on passing
to the interior, are attracted to the stores of food material
which they find there. The nucleus being the main instrument,
if not the only instrument of constructive metabolism, is pro-
portionately affected and altered in structure; while the cell
body itself, which was never very big and is not destined to
grow toa very great extent, becomes plastered to the body
of the cell in which the food material is stored up. Hence
the changes which go on in the flagellated cells after emigra-
tion seem to result from absorption of food material and con-
sequent increase of nucleoplasm, especially of the chromatic
portion of the nucleus.

Simultaneously with the contraction of the nucleus de-
scribed above the nuclear threads become thicker, and the
chromatin rearranges itself. Instead of being scattered about
in small granules, it appears, as a rule, as irregular patches
lying against the nuclear membrane, though at this stage it
does not invariably conform to a definite type of arrangement.
At the commencement of the above process of adhesion the
outlines of the flagellated cells are visible (fig. 29 5); but
later on they become so closely adherent that they are indis-
tinguishable as separate units in the morphological sense. As
a rule they fix themselves to the cells with vesicular
nuclei, but it often happens that a number of them become
attached to cells with granular nuelei, which in some
cases contain several yolk bodies, and which therefore act as

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 408

centres of attraction in possessing stores of food material. At
this stage it is almost impossible to distinguish the nuclei
of the flagellated cells from the yolk bodies always present in
greater or less number.

The cell aggregations compounded each of numerous flagel-
lated cells, and one cell with a vesicular nucleus or a granular
nucleus, as the case may be, must be carefully distinguished
once and for all from the “cell groups” of type B, in
which the cells were incompletely divided from one another,
and never had a separate existence. The groups here de-
scribed will be referred to as ‘‘ plasmodial aggregations,” for
they follow in their formation the principle involved in the
building up of a plasmodium rather than that which has to be
considered as connected with phagocytic action. In the former
case no cell is subordinated to the other in any way, but in
the latter one cell takes in the other, and seeks to destroy it.
From the bionomic point of view the action of the flagellated
cells in this case is comparable rather with that of the com-
mensal, which feeds, as it were, at the table of another, but
does not directly harm the host. Similarly the fiagellated cells
feed on the food material which has been stored up in the
cells with vesicular nuclei, and which they inherited from the
egg cell in the course of their development. If the formation
of these groups were a case of phagocytic action the large
central cell would have to be considered as the phagocyte,
taking in all the flagellated cells it could lay hold of, and
endeavouring to absorb and destroy them. ‘This view has
been put forth, but, as will be seen, it obtains no support from
the subsequent development of the plasmodial aggregations.

The outline of the plasmodial aggregations is at a certain
stage as well defined as that of the cells with vesicular nuclei
in the youngest larva, that is in type A. During the pupal
life there is a stage, however transitory, during which the
young individual consists of only cells with granular nuclei
at the surfaces and plasmodial aggregations inside, provided
always that such an individual has been produced from the
larva described above as type D (figs. 16, 16 a, 26, and 27).

4.04. RICHARD EVANS.

In an individual produced by the metamorphosis of type C,
the plasniodial aggregations will of necessity be mixed up with
the flagellated chambers derived from the cell groups.

Since individuals occur with neither chambers nor free cells
capable of forming chambers, but with the interior full of
plasmodial aggregations surrounded by cells with granular
nuclei at the surfaces, the important question of the origin of
the cells which later on become the collar-cells of the flagel-
lated chambers forces itself upon us. Are they developed
anew from the cells with vesicular nuclei, or do the flagel-
lated cells—no longer flagellated, it is true—separate themselves
from the plasmodial aggregations into the composition of
which they have entered, in order to develop into collar-cells?
If the latter be the actual course of the development, it would
furnish a further proof of the homology of the flagellated
cells with the constituent cells of the “ cell groups,” which are
characteristic of types B and C, and which in the latter have
developed into flagellated chambers. To answer this question
it is necessary to trace further the morphological changes
undergone by that constituent of the plasmodial aggregations
which owes its origin to the immigration of the surface layer
of cells in the larva.

The small nuclei of the flagellated cells which have already
been traced through a series of changes, leading them to
acquire a structure almost indistinguishable from that of the
yolk bodies, now embark upon a similar series of transforma-
tions, but in the reverse order, as the result of which they
revert to a condition slightly different from that which they
presented as the flagellated locomotor layer at the surface of
the larva.

The small nuclei—that is the nuclei of the cells which were
once flagellated—in the plasmodial aggregations commence
this series of changes by increasing slightly in size, simul-
taneously with some internal changes in disposition of the
chromatin and nuclear reticulum. At one stage they appear
as oval masses, uniformly coloured in stained sections, but
now the chromatin becomes looser and aggregated into small

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 405

granules, and the threads of the nuclear reticulum become
visible. The chromatin granules are smaller and stain faintly.
They are more numerous than they were in the primitive con-
dition of the cell, and the nuclear reticulum is not so coarse
(figs. 17—20).

In the description of the larva given above reference has
been made to a slight change in the character of the nuclei of
the individual elements of the cell groups, taking place as they
pass from the condition of structure found in type B to that
which exists in type C. It was pointed out that the nuclei
of the cell groups in type B are indistinguishable from those
of the flagellated cells at the surface. As a matter of fact,
the nuclei of the collar-cells in type C are equally indistinguish-
able from those of the flagellated cells at the period when
the latter—by their own energy, apparently—emancipate
themselves from the plasmodial aggregations into which they
entered, and again assume their individual form. Here, there-
fore, is another argument, in addition to those already brought
forward, in favour of the view that the flagellated cells at the
surface and the cell groups in the interior are homologous,
and really belong to the same class of cells.

As it is impossible to observe the above changes in the living
pupa, we are forced to the method of studying them in sections,
and so drawing our conclusions. In our examination of sections
we may follow one of two courses. It happens sometimes
that the changes in the structure of the nuclei of the flagel-
lated cells, as they pass from the condition in which they are
difficult to distinguish from the yolk bodies to that found in
fully developed collar-cells, can be traced while they are still
inside the plasmodial aggregations,—in other words, the
changes can be followed in one individual, or even in one
section of such an individual.

Fig. 17 illustrates the changes which go on in these nuclei
as they pass from one condition to the other, and shows a
number of transitions in one individual. The nucleus marked
a has become enlarged, and the chromatin is scattered almost
uniformly in it; while in the nuclei marked } the nuclear

406 RICHARD EVANS.

reticulum begins to appear, and the general structure is looser
than in the nucleus a. The nuclei labelled d show a further
change, and, though they are still enclosed in a plasmodial
aggregation, they are so like the nuclei marked e as to be
indistinguishable from them, though the latter are undoubtedly
the nuclei of free cells which will later on develop into collar-
cells. An unprejudiced examination of the above figure.can
hardly fail to satisfy the most sceptical person that the plas-
modial aggregations contain the nuclei of the flagellated cells
in a state which is only an intermediate condition
between that which they possessed when the cells
were free at the surface, and that which later on they
assume as collar-cells in the interior.

While the above proof seems conclusive, it is perhaps advis-
able, though it may appear superfluous, to confirm our results
by the comparison of several individuals preserved at different
stages in their development, in order to trace in them the series
of changes which have been described above as taking place in
the same pupa. In the pupa represented in figs. 16, 16a, and
26 there are scarcely any nuclei or cells save those enclosed
in the plasmodial aggregations and those of which the flattened
epithelium consists ; while in that represented in figs. 18, 18 a,
and 18 4, which is an older pupa, the nuclei of the flagellated
cells are emancipating themselves at all points from the plas-
modial aggregations, which are losing their sharp and well-
defined outline. At this stage and slightly later such a thing
as a cell outline or limit can hardly be discerned. The nuclei of
the flagellated cells can be seen clearly becoming looser in
structure; the linin threads are well developed; and the
chromatin in general is rearranging itself preparatory to the
subsequent stage which is represented in fig. 19, and especially
in fig. 20, in which the nuclei have attained their definitive
characters.

It would not be out of place at this juncture to lay special
emphasis on a fact which has been already mentioned, namely,
that the plasmodial aggregations always contain yolk bodies,
besides several nuclei. ‘The former may vary considerably in

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 407

number, and are indistinguishable from the latter at that
stage at which the nuclei of the flagellated cells have reached
the extreme limit of the modifications which they undergo
during metamorphosis. When, however, this limit has been
passed, and the nuclei in question are gradually assuming the
ordinary structure of those of the collar-cells, the difference
between them and the yolk bodies becomes more marked
stage by stage. Fig.16 a illustrates a condition in which it is
impossible to distinguish between “ yolk bodies ” and nucleif;
in fig. 18 this is less difficult, while in fig. 19 and still more so
in fig. 20 the difference, as a rule, is well marked.

The similarity at a certain stage between the nuclei and the
yolk bodies makes it difficult to say whether all the flagellated
cells are set free from the plasmodial aggregations or not.
Judging, however, from certain pupe, such as that represented
in figs. 18, 18a, and 184, there appears to be no reason what-
ever for the supposition that they are in any way absorbed by
the central cell of the plasmodial aggregation. Those bodies
which appear to be reduced in size, and not to be gradually
acquiring the ordinary nuclear characters, can be more than
accounted for from the number of yolk bodies occurring in the
larva at the time of fixation. There is no reason, however,
why a flagellated cell should not be completely absorbed if by
any mishap it was injured during the interchange of position
taking place at the time of fixation and subsequent metamor-
phosis. Any such cell would probably fall an easy prey to the
cells with vesicular nuclei, which in the young sponge are
ameceboid and nutritive in character. Nevertheless it appears
almost certain that the vast majority, if not all the flagellated
cells emancipate themselves from the plasmodial aggregations.

The results obtained by a comparison, on the one hand, of
nuclei in a single individual, and on the other hand, by follow-
ing the different stages in several pupz, may, therefore, be
summarised as follows. The plasmodial aggregations contain
both yolk bodies and a number of nuclei. The former appear
to decrease in size as development goes on, and have therefore
supplied an argument in favour of the view—evidently incor-

408 RICHARD EVANS.

rect—that the nuclei of the flagellated cells became completely
absorbed. The latter, on the other hand, become nuclei of
the young sponge, and this is true equally of the central vesi-
cular nucleus and of the numerous smaller nuclei belonging
to the flagellated cells.

Simultaneously, as a rule, with the above changes in the
characters of the yolk bodies, and the small nuclei contained
in the plasmodial aggregations, the cytoplasmic bodies of
the groups in question lose their sharp outline. They spread
out and become irregular in shape and almost indistinguishable
from one another. By the time the nuclei have attained the
definitive structure of the nuclei of collar-cells, the internal
arrangement of the cytoplasm belonging to the various groups
of plasmodial aggregations may be described as being syncytial.
Spaces begin to make their appearance in the undifferentiated
cytoplasm (fig. 18), which soon develop into large cavities
(fig. 19), lied by cells possessing granular nuclei. Mean-
while the nuclei of the flagellated cells arrange themselves in
the rings of the cytoplasm, which are at first quite irregular
and ill-defined (fig. 19a), and do not appear to consist of
individual cells ; but this syncytial condition soon passes away,
and the separate individual collar-cells make their appearance.
They develop collars and flagella at the time of separation
as free cells, and in this way the plasmodial aggregations give
rise to the flagellated chambers. The cell with vesicular or
granular nucleus which occupied the central position retires
outside the chamber, and takes no part in its formation.

(6) Special Features of the Metamorphosis of the
Different Types of Larve.

The phenomena of the development common to all the
types having been described in the previous section, in the
following pages only those features which characterise the
metamorphosis of the fully developed types, namely, C and D,
will be specially considered, Reference must be made also to

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 409

type B in cases in which the appearances figured are such as
would result from the fixation of individuals of that type.
But as type B is only a younger stage of type C, and nota
special variation of the fundamental type of larva, so to speak,
no special description of the changes taking place in it appear
necessary.

(1) Special Features of Type C during Metamor-
phosis.-—It is characteristic of the metamorphosis of all the
types that the flagellated cells disappear more quickly from the
lower surface of the pupa than from the upper. However,
this feature is strongly emphasised in the pupa formed from
type C, as compared with that formed from type D. But
it is highly probable that this difference is not so great in all
the pupz derived from larve of type C as it is in the one
actually represented in section in fig. 29. The structure of
the pupa in question is such that it is even possible that it
could thrive if the flagellated layer at the surface were
thrown off altogether, as described by Gotte. I have seen,
however, no evidence of such a procedure on the part of any
fixed larve. In the case of the pupa figured, the flagellated
layer at the surface is as complete as in a free-swimming larva.
Both the nuclei and the flagella present the same appearance
as in the larva, and are equally well defined. On the other
hand, the flagellated cells which were at one time situated
at the lower surface have disappeared completely from that
position, and have migrated into the interior. They have be-
come plastered to the surfaces either of the cells with vesicular
nuclei, or of those with granular nuclei, as the case may be,
to form the plasmodial aggregations, which have already been
described in the previous section of this paper. The contrast
between the appearance presented by the flagellated cells
at the upper surface, and those which have at this stage
travelled to the interior from the lower surface, may be seen on
comparing figs. 29 a and 29 4, two figures drawn from the upper
and lower surfaces respectively of the same larva as fig. 29.

Another feature characteristic of the pupa which results
from the fixation of type C, is the presence of fully developed

410 RICHARD EVANS.

flagellated chambers (figs. 29 and 29 a), in which the individual
cells are provided with a collar and a flagellum (fig. 29 c),
resembling in every respect those of the collar-cells of the adult
sponge. There can be no reasonable doubt, therefore, as to the
origin of the chambers in question from the cell groups of
types Band C. Inasmuch as the cell groups found in type
B are developed from the cells of the inner mass, that is,
from the cells with vesicular nuclei; and since, further, these
chambers are produced from the cell groups; it follows that
the flagellated chambers found in the young pupa derived from
type C, and represented in figs. 29 and 29a, are produced by
the multiplication of the cells of the inner mass, and not from
the flagellated cells which migrated into the interior, and
which have as yet, even at the lower surface, only gone so far
as to form plasmodial aggregations (fig. 295). These facts
furnish a clear proof that cells of the inner mass are capable
of giving rise to collar-cells, and that the flagellated chambers
do actually so arise in the development. The cells arranged
in groups surrounding spherical cavities in the larva do not
flatten out to form the canals of the sponge. On the other
hand, it has been shown that the cells which in the larva become
flattened to form the lining of the lacunar spaces, belong to
an entirely different class of cells, i.e. the class which has been
described as consisting of cells with granular nuclei, from
which arise the epithelial membranes of the sponge in general.

Having traced the origin of the flagellated chambers from
cells of the inner mass, it remains to inquire what happens to
the flagellated cells after they have entered the plasmodial
aggregations in those pup in which the cells of the inner
mass do indubitably take part in the formation of the flagel-
lated chambers.

It has already been pointed out that the flagellated cells
migrate into the interior and form plasmodial aggregations
(figs. - 29, 296, 246, and 25), though they migrate far
more slowly from the upper than from the lower surface. In
figs. 830 and 30a the flagellated layer, which has not com-
pletely disappeared from the surface even at this stage,

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 411

presents an extraordinary change in the arrangement of the
cells. Its constituent cells are migrating into the interior
individually, and consequently the layer itself becomes less
dense, the cells shorten, and at the same time become broader,
owing to the superficial expansion of the sponge asa whole. The
above changes in the characters of the flagellated cells might
seem at first sight to support the view, held by many authors,
that the flagellated cells become the flattened epithelium.
But a more careful study of the figures is sufficient to refute
any such idea, and to force us to the conclusion that what-
ever happens to the flagellated cells, they do not flatten
out to become the constituent cells of the dermal epithe-
lium. In the first place, the nuclei of the flagellated cells
are seen to be passing in, though it is almost impossible
to make out the cell body, and the portion nearest the margin
of the pupa is full of them (figs. 80 and 30a). In the second
place, the cells with granular nuclei have come to the surface
already in the portion nearest the margin (fig. 30), while
nearest the centre they are seen engaged in the struggle, so
to speak, of passing to the exterior between the few remain-
ing flagellated cells (fig. 30a). And in the third place, the
nuclei of the cells, which before metamorphosis were situated
in the interior, retain their large size and granular character,
while those of the flagellated cells, which were once at the
surface, are undergoing the usual changes through which they
pass when they are about to enter into the composition of the
plasmodial aggregations.

There is, therefore, no reason whatever to suppose that the
flagellated cells flatten out and become indisting uishable-from
the cells with granular nuclei, which they wouid have to do
were they the cells from which the dermal epithelium is
developed.

The flagellated cells have been traced to the interior already,
and their nuclei have been shown to be undergoing the changes
usual in the formation of plasmodial aggregations. It remains
to show what furt her changes they will pass through. In the
pupa which is represented in fig. 31 the flagellated cells are

VoL. 42, PART 4.—NEW SERIES, FF

412 RICHARD EVANS.

no longer independent. Fully formed plasmodial aggregations
are found to exist, in addition to chamber-like rings of cells
derived from the “ cell groups” of thelarva. The nuclei of the
plasmodial aggregations, however, in a short time emancipate
themselves, and the cells consequently, becoming free and
sharply individualised, give rise to the collar-cells of a new
series of flagellated chambers. The collar-cells of the sponge
are therefore developed, on the one hand, from the cells of the
inner mass, and, on the other hand, from the flagellated cells.

In fig. 28 two flagellated chambers are represented ; in one
of them the cells are perfectly separate and independent,
with the nuclei all alike, and possessing the same structure
as the nuclei both of the collar-cells—i.e. chamber cells,
for they have not as yet developed collars, represented in
figs. 24.6, 31, and 31 a—and of the cells which have already
developed collars and flagella as seen in fig. 29a. In the
other chamber in the same figure the cells are not so dis-
tinct, and the nuclei are much smaller and more irregular in
size. In fact, both the arrangement of the cells and the size
and structure of the nuclei resemble those of the flagellated
chambers represented in figs. 19, 19a, 20, and 21, which are
drawn from specimens developed from type D, in which, how-
ever, flagellated chambers derived from “cell groups ” are
practically absent. It is evident that the cells of the lower
chamber in fig. 28 have emancipated themselves from one or
more plasmodial aggregations, which they formed in combina-
tion with the large cells with vesicular nuclei, situated close to
the chamber -in question. With regard to these cells with
vesicular nuclei, it may be pointed out that those in the
immediate neighbourhood contain very few yolk bodies, which
at this stage are of necessity much reduced in size, while
those further off contain more yolk bodies. This fact points
to the view that the yolk bodies in these cells have been
almost completely used up by the flagellated cells which grouped
themselves round the cells with vesicular nuclei to form the
plasmodial aggregation. As has been stated above, the nuclei
of the flagellated cells in the plasmodial stage can be distin-

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 4138

guished from the yolk bodies only by their affinity for a differ-
entiating stain.) The nuclei then pass during this stage
through a series of changes in rapid succession, and are there-
fore not constant in structure. The volk bodies also change
from stage to stage, owing to their being used up in the forma-
tion of living protoplasm. These changes, both in the nuclei
and in the yolk bodies, render interpretation of the phenomena
observed a most difficult task. It is evident, however, from the
characters of the chambers represented in fig. 28, that the
flagellated cells give rise to a series of chambers in the pupa
derived from a larva of type C, in the same way as they do in
the one of type D. The former statement is far more difficult
to prove than the latter, owing to the complication brought
about by the presence of a number of chambers derived from
the cells of the inner mass in type C, and consequently in the
pupa derived from that type of larva. But the presence of yolk
bodies in the pupa on the one hand, and of flagellated cells
on the other, suffice to explain the conflicting statements
which have been made by various observers, and to reconcile
them in the following manner. Some of the bodies contained
in the plasmodial aggregation, namely, the yolk granules,
become used up and disintegrate, whilst others of a different
nature emancipate themselves and give rise to the nuclei of the
collar-cells.

The question of the formation of the flattened epithelium
from the cells with granular nuclei has been incidentally men-
tioned, but it must be described here at a greater length from
the point of view of the structure of the pupa derived from
the larva described as type C. The flattened epithelium of the
lower surface forms much more quickly than that of the upper
surface, in correspondence with the different rate at which the
flagellated layer disappears from these surfaces. The difference
in question is brought out very sharply in fig. 29, while in

1 With carmine and bleu de Lyon the nuclei are stained red, while the yolk
bodies are stained blue. This reaction is difficult to bring about owing to the
thickness of the nuclear membrane, which stains blue in the same way as the
yolk bodies,

414 RICHARD EVANS,

fig. 30, which represents an older stage, the difference is not
so marked, as the cells on the upper aspect have in some cases
made their way to the surface, and in fig. 31 the epithelial
membrane of the upper surface is as complete as that of the
lower.

In the interior of the pupa the cells with granular nuclei
continue to line the persistent larval cavity, which in some
larve may be of great size (figs. 29 and 30). The cells with
granular nuclei, situated in the solid part of the inner mass,
flatten out to form the -lining of the exhalant canal system.
The canals, which are short at this stage when they exist, may
be seen to communicate on the one hand with the larval
cavity, and on the other with the cavity of the flagellated
chambers. These connections explain why some observers
have made the mistake of describing the flagellated chambers
as developed from the cells of the inner mass at the blind
ends of outgrowths from the flattened layer of cells lining the
larval cavity, i.e. from the so-called “ entoderm.”

The special features of the pupa derived from the larva
described above as type C may be summed up as follows :

1. The flagellated cells pass in at different rates from the
lower and upper surfaces, and consequently the flattened epi-
thelium forms much more slowly on the latter than on the
former. The difference in question is much more marked in
this pupa than in the one derived from the larva described as
type D.

2. Flagellated chambers are always present, appearing in
section as rings. In many cases the individual cells are pro-
vided with a collar and a flagellum even at the time of fixation.

3. The flagellated cells, after passing into the interior, take
part in the formation of plasmodial aggregations, from which
they later emancipate themselves and give rise to flagel-
lated chambers, the cells of which for a time can be distin-
guished from the cells of the chambers derived from the
“inner mass” by their nuclear and other characters (figs.
28, 31).

4. The canal systems appear as early as the time of fixation,

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 415

The larval cavity is not obliterated, but takes part in the for-
mation of the exhalant system and gastral cavity.

(2) Special Features of Type D during Metamor-
phosis.—There remain but a few statements to make with
regard to the metamorphosis of the type of larva now under
consideration, owing to the completeness of the description
given above of the features common to the metamorphosis of
all the larvee.

The interchange of position between the flagellated cells on
the one hand, and the cells with granular nuclei on the other,
takes place much in the same way as in the larva described as
type C. There are two features, however, which are perhaps
worthy of further notice: first, the flagellated cells migrate
almost at the same rate from the upper and lower surfaces ;
and secondly, they tend to form fan-like groups at the lower
surface.

In fig. 15 only a few plasmodial aggregations have been
formed, but notwithstanding this fact the flagellated cells have
disappeared from both surfaces, and the flattened epithelium
is complete almost everywhere. The arrangement of the cells
is, therefore, quite different from that represented in fig. 29,
which has been drawn from a pupa derived from the larva
described as type C. In figs. 15 @ and 64 the second point,
namely, the formation of fan-like groups, is illustrated. The
groups in question may in some cases consist of so many cells
as to make it possible that an actual invagination has taken
place, though for other reasons this is not probable (fig. 15 4).

The individual cells of the groups in question appear as if
they had been drawn inwards by some internal force, the
body of the cell being in consequence greatly elongated
and attenuated. The dark streak which was described above
as being situated externally to the nucleus of the flagellated
cells in the free-swimming larva is seen clearly in many
of the cells contained in the groups, though there is no
sign whatever of the portion of the flagellum that lay out-
side the cell. The nuclei of the flagellated cells do not appear
to change in character so long as the cells remain in these

416 RICHARD EVANS.

groups, but immediately after breaking away from them, and
coming into contact with the cells with vesicular nuclei
situated in the interior, they undergo the structural changes
which have been already described.

The peculiar arrangement of the flagellated cells in the
groups under consideration appears to be due to the pressure
exerted upon them by the cells with granular nuclei, which
are struggling towards the exterior at all points on the surface
of the individual. The two processes, namely, the passing in
of the flagellated cells and the passing out of the cells with
granular nuclei, go on at the same time, both classes of cells
taking an active part, apparently, in effecting the interchange
of position.

The next feature to be considered is one of the greatest
importance, and consists in the existence of a stage in which
there are no free cells derived either from cell groups or from
the flagellated layer of the larva. The pupa at this stage
consists of plasmodial aggregations inside and flattened epi-
thelium on the outside (figs. 16, 16 a, and 26). There are no
signs whatever of the larval cavity. The formation of the
plasmodial aggregations has been so fully dealt with already,
that it is unnecessary to say anything further with regard
to them. However, it must be pointed out that the nuclei
of the flagellated cells sometimes—though this is not the
usual rule—undergo the changes which have been described
while still remaining in close association with the central
cell which possesses a vesicular nucleus. The nuclei marked
n. fl. c. at the left-hand corner of fig. 20 illustrate this
point. Further, the nuclei labelled a, 6, and d in fig. 17
exemplify the successive stages in the series of changes
through which they pass. The nucleus marked a in fig. 17
has already increased in size, and contains the chromatin in
an evenly distributed condition; while in those labelled 4,
linin threads of considerable thickness, covered with chro-
matin, are appearing. ‘The nuclei marked d present the ulti-
mate structure of those of the collar-cells, though they are
still situated inside a plasmodial aggregation. In the face

STRUCTURE OF THE LARVA OF SPONGILLA LAOCUSTRIS. 417

of such evidence it seems impossible to doubt that a great
number of the bodies contained in the plasmodial aggregations
give rise to the nuclei of the collar-cells.

Fig. 17 illustrates the commencement of the breaking up
of the plasmodial aggregations, and the change which takes
place in the characters of the nuclei of the flagellated cells.
Both of these changes are seen to have advanced still further
in figs. 18, 18 a, and 18 4, in which, however, there is no sign
as yet of a flagellated chamber. In figs. 19 and 19 a, which
still lack flagellated chambers, the plasmodial aggregations
have disappeared as completely as if they had melted away. A
stage is therefore produced in which the cells present a syncytial
arrangement. In fig. 20 a stage is illustrated in which the
nuclei of the flagellated cells are becoming arranged in rings
round cavities which in some cases are more or less circular
in form, but the general arrangement is still syncytial. Fig.
21 represents a further change. The cytoplasm, surrounding
the nuclei arranged in rings in the stage represented in fig. 20,
becomes divided into masses correspouding to the nuclei, and
the individual collar-cells provided with a collar and flagellum
make their appearance.

The changing nuclei described above must not be confused
with the yolk bodies which exist inside the cells with vesicular
nuclei, and which become smaller stage by stage, while the
nuclei of the flagellated cells, on the contrary, become larger.
The number of yolk bodies found in the cells with vesicular
nuclei varies considerably, according as the larva fixed at an
early stage of development or a late one. In fig. 26 the yolk
bodies are coloured blue, while the small nuclei are red. In
figs. 18, 19, 20, and 21 the distinction between yolk bodies and
small nuclei is such that it is impossible to mistake the one for
the other. ‘The important point for our purpose, however, is
that the nuclei of the flagellated cells and the yolk bodies exist
side by side in the plasmodial aggregations, the former giving
rise to the nuclei of the collar-cells, while the latter are used up
during the development as food material.

To sum up, the special features of the pupa derived from the

418 RICHARD EVANS.

larva described in the early part of this paper as type D are
briefly as follows:

(1) The flagellated cells pass in at a rate which is but slightly
different for the upper and lower surfaces.

(2) There is a stage, however brief, during which the pupa
consists only of plasmodial aggregations inside and flattened
epithelium outside. During this stage there are no signs either
of the larval cavity or of incipient canals. The former has
been obliterated, while the latter have not as yet made their
appearance.

(3) The flagellated chambers are derived for the most part,
if not entirely, from the flagellated cells of the larva.

(4) The canal system only appears after the plasmodial
aggregations break up and form the flagellated chambers.

(c) Further Remarks on the Formation of the General
Canal System.

In specimens derived from the larve of type D all the canals
and spaces must be formed anew, for the larval cavity is entirely
obliterated, and there are no lacunar spaces or canals such as
were found in the pupz derived from typeC. The canals appear
as spaces in the interior soon after the flagellated cells have
begun to emancipate themselves from the plasmodial aggre-
gations, and have passed into the syncytial condition described
above. Itis needless to say that the canals at their first appear-
ance do not possess a proper lining in the form of flattened inde-
pendent cells. Later on, however, they become lined with
cells possessing granular nuclei, which are derived from two
sources. In the region of the subdermal cavity they are derived
toa great extent from the cells with granular nuclei which had
been already produced at the time of metamorphosis. In the
interior, on the other hand, they are produced as in the larva by
the gradual conversion of the cells with vesicular nuclei, as can
be seen from the presence of small central corpuscles in many
of these cells.

Owing to the continued formation of the cells with granular
nuclei from those with vesicular nuclei, many of the former

STRUOTURE OF THE LARVA OF SPONGILLA LACUSTRIS. «419

are found in the spaces between the chambers where the latter
are always situated. In these positions, especially in the
neighbourhood of the developing skeletal fibres, they form a
sort of connective tissue, and the spongin occurring in the
sponge seems to owe its origin to these cells. Wherever a
canal is formed, it is, of course, lined by these cells. The cells
with vesicular nuclei also frequently take up a position adjacent
to the canals. Whenever they do so the character of their
nuclei is doomed to change just as the nuclei of the cells
which give rise to the megascleres become modified.

It makes no difference, so far as the various classes of cells
are concerned, whether the gastral cavity is a new formation,
as in the young sponge developed from a larva of type D, or is
derived mainly from the larval cavity, as was shown to be the
case in the development of the larva of type C. In each case
the cavity is lined by the same class of cells, that is, by those
with granular nuclei. This fact does away with any difficulty
that might arise from the histological point of view.

The cells with vesicular nuclei, after the flagellated cells
have separated themselves from them and have become ar-
ranged in chambers, remain in the neighbourhood of these
structures and constitute the amoeboid elements of the young
sponge. They often adhere to the surface of the chambers,
being in many cases situated at a point not far from the
apopyle as well as elsewhere, and to a certain extent spread
over them. It is quite possible that some of them in this
position become perforated to form the openings of the in-
halant canals into the flagellated chambers. This point is
further explained in the descriptions of figs. 32, 33, and 34,
in which inhalant pores are figured. It is difficult to prove
that this pore is intra-cellular because of the great superficial
expansion of the cells in the neighbourhood of the pore, and
the distance which, as a consequence, intervenes between the
nucleus of the cell and the pore. If these pores are really
intra-cellular, then the cells in question would correspond to
the porocytes which have been described by Mr. Minchin in
the Ascons. The only explanation I can give of the appear-

4.20 RICHARD EVANS.

ances represented in fig. 33 is that the pore is intra-cellular
passing through the large cell, the vesicular nucleus of which
is drawn in fig. 83d. If some of these cells do become poro-
cytes we would expect the nucleus to become granular. In
fig. 246, which represents a surface view drawn at different
levels, there are two extremely large nuclei situated deep in
the tissues of the individual among the small cells which are
evidently becoming grouped to form chambers. The cells to
which these two nuclei belong seem to lie more or less loosely
on the cells of the chamber, and it is quite possible that they
are cells similar to that containing the large vesicular nucleus
drawn at the top of fig. 33d, in which the nuclei have lost
their vesicular character. The porocytes in the Ascons are
described by Mr. Minchin (11) as large cells, situated near the
surface when the sponge is fully expanded, which in the con-
tracted condition of the sponge push their way inwards between
the collar-cells, and ultimately come to lie inside the gastral
layer as a granular axis to the Ascon tube. The granules,
according to Topsent (14), consist of reserve food material. If
so, the cells probably collect it when the sponge is expanded,
and distribute it during the contracted condition to the other
cells. The passing in of the porocytes to the interior would
seem, therefore, to have a physiological meaning. In Ascons,
contraction or expansion may happen at any time in the history
of the sponge. Let us now turn to the pupa of Spongilla. Here
the large cells with vesicular nuclei, and containing food
material in the larva at the time of fixation, would correspond
to the porocytes of the Ascons in the expanded condition of the
sponge. During metamorphosis, however, they become sur-
rounded by the flagellated cells, that is by cells which are po-
tentially collar-cells—a condition of things found in the Ascons
during the contracted stage, with the difference, in the two
cases, that in Ascons the gastral layer is continuous, while in
Spongilla it is broken up into groups of cells. Hence during
the contracted stage in the Ascons we find a continuous axis
surrounded by an uninterrupted gastral layer; but in Spon-
villa, during the stage with plasmodial aggregations, we find the

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 421

individual cells with vesicular nuclei surrounded by groups of
flagellated cells. When the plasmodial aggregations break up,
the cells with vesicular nuclei pass out of the group of cells
and plaster themselves to the outer surface of the flagellated
chambers subsequently formed, in the same way as the pore
cells of the Ascons pass out of the gastral cavity to the surface
when the sponge expands. From the physiological point of
view the correspondence between these two classes of cells
seems pretty complete. Of course it is quite needless to say
that the cells with vesicular nuclei do not all become porocytes,
for many of them retain their amoeboid character as well as
their blastomeric properties, and serve later on for the building
up of the gemmule. This discussion regarding the physiolo-
gical correspondence of these cells strengthens the argument in
favour of their morphological homology ; in other words, in
favour of the view that some of the cells with vesicular nuclei
become porocytes in Spongilla.

The Subdermal Cavities.—These spaces are situated
below the dermal membrane, and communicate on the one
hand with the exterior by way of the inhalant ostia, and on
the other with the flagellated chambers by way of the inhalant
canals and prosopyles. The cavities in question have a super-
ficial position, and they and their ostia are arranged on the
slanting sides of the young sponge, while the osculum takes up
a more or less central position. Deeper down, and nearer the
centre, the exhalant canal system and gastral cavity are found.

The spaces which have been figured in such profusion in the
region between the marginal membrane and the body of the
young sponge are by no means all of them true ostia opening
into the subdermal cavities. Many of them are merely the
nutritive vacuoles found in the cells which have wandered into
the space between the two layers of the marginal membrane.
The vacuoles in question in many cases present the exact
appearance of an intra-cellular opening, the nuclei of the cells
lying close to them, but on focussing the microscope more care-
fully the cells of the flattened epithelium can be seen overlying
them. However, this does not do away with the presence of

422 RICHARD EVANS.

true ostia, which are small openings measuring about 15 mu
across, and which appear to be intercellular, and not intra-
cellular. They occur all over the surface, even close to the
osculum (fig. 8).

The osculum is much larger than an ostium, and is situated
near the centre of the young sponge. At first it appears as a
funnel-shaped opening, the rim of which is on a level with the
general surface. Its sloping sides are perforated by one or
more smaller openings which communicate with the exhalant
system beneath. The thickened rim which surrounds the
osculum soon grows up, and forms a tube carrying the opening
at its extreme end.

AppenDIx A.

On the Nutritive Vacuoles and Yolk Bodies.

The enclosures found chiefly in the cells with vesicular
nuclei are of two kinds, and must be described separately. The
less numerous and larger enclosures may be termed the “ nutri-
tive vacuoles,” while the smaller and more numerous will be
given the name “ yolk bodies.”

To have described them fully at the same time as the cells
which contain them would have burdened with too much detail
the description of the more important elements of the larva,
and also would have caused a certain amount of the technique
to be mixed up with the description. For these reasons I
thought it advisable to describe these enclosures, as well as
certain other processes which go on during the development,
in a number of appendices.

(1) The “Nutritive Vacuoles.”’—These structures are
almost invariably restricted—in the free-swimming larva at
least—to the cells with vesicular nuclei, which in the earlier
stages of the development may contain as many as three or
four of them, though one is the rule. Owing to the large size
of these cells it often happens that the nucleus and the vacuole
must be looked for in different sections. These structures have
not been described by previous authors, save as an “ occasional

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 4238

vacuole,” but they appear to be present in all cells with vesicular
nuclei, and occasionally in the cells with granular nuclei, and
even in the collar-cells in the pupa and young sponge. Their
occurrence in the last two kinds of cells shows them to have
been developed recently from the cells with vesicular nuclei.

The vacuoles here described behave very differently under
the action of various preserving reagents. When the specimen
is preserved with absolute alcohol, or a mixture of corrosive
sublimate and glacial acetic, or Perenyi’s fluid, these en-
closures appear as clear vacuoles. When Flemming’s fluid is
used for a short time some of them are clear while others are
black, but when it is used for a long time they are all as black
as ink, as they are, also, when the specimens are preserved in
Hermann’s fluid. Whether as clear vacuoles or as black en-
closures they are always circular in section, and equal or even
exceed the vesicular nucleus in size (figs. 7, 11, and 11a).

When a section of a specimen preserved in Hermann’s fluid
is subjected to the bleaching action of chlorine, these bodies
appear as clear spaces, just as in sections from specimens
preserved in either absolute alcohol, or corrosive sublimate
and glacial acetic, or Perenyi’s fluid. This proves conclu-
sively that the clear vacuoles observed in specimens preserved
in these reagents correspond to the black enclosures found in
material preserved in Hermann’s fluid. By stopping the bleach-
ing action short of its completion the black enclosures can be
seen in different stages of decoloration (fig. 2).

In sections stained according to Heidenhain’s iron hema-
toxylin method they become pale, though with other stains
they remain black. In sections of larve preserved in absolute
alcohol, or in bleached sections, the vacuoles were never stained,
though a kind of a membrane which surrounded the vacuole
always stained with safranin, when followed by gentian violet
and iodine, or even with gentian violet alone, as well as with
hematoxylin followed by fuchsin S (fig. 2). Whatever substance
these vacuoles contained in the living condition, it may be
concluded that it has been dissolved out of the preserved
specimens; and further, the blackening under the action of

4,24, RICHARD EVANS.

Hermann’s or Flemming’s fluid tends to prove that it was of a
fatty nature.

(2) The Yolk Bodies.—These bodies are far more numer-
ous than the nutritive vacuoles, but the same remarks apply to
them as to the vacuoles in regard to their occurrence in differ-
ent classes of cells. They are chiefly found in the cells with
vesicular nuclei, which, as a rule, contain only one nutritive
vacuole, while they contain several yolk bodies, and in many
cases they may be described as being full of them. Their
average size in the youngest larve is about 24 mu across, and
their shape is either spherical, oval, or plano-convex (figs. 1—4).
As the larva becomes older, and cell-differentiation progresses,
they become smaller in size, and some of them seem to be used
up completely, though this is not true of all of them. No
larva, even of types C and D, was observed to be absolutely
devoid of yolk bodies; on the other hand, they are often very
numerous in the fixed stages (figs. 25 and 26), a fact which
points to the early fixation and metamorphosis of such in-
dividuals.

They almost invariably stain like the nucleus, but when
sections are successfully stained with carmine, and subsequently
with bleu de Lyon, the nuclei take the former, whilst the yolk
bodies take the latter (figs. 1, 3—6 a, 25, and 26). The outer
layer stains more deeply than the central part, which, asa rule,
is devoid of any stain, the colourless space being almost circular
in shape, and slightly eccentric in position. It happens some-
times that this space, instead of being clear, is stained red,
while at other times there is a red patch near the border. This
space seems to be only a vacuole in certain cases, while in others
it is occupied by a substance which stains differently from the
bulk of the granule. When sections are stained with Bismarck
brown, followed by malachite green, there can be seen in the
clear space inside the yolk bodies one or more refringent
granules which are devoid of crystalline structure. The
granules in question vary considerably in size and shape,
having the form of dumb-bells, spheres, or rods, all of which
are represented in fig. 4.

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS, 425

As has been stated already, they are present in all the types,
though they are far more numerous in types A and B than in
types C and D. In types A and B they show great affinity for
bleu de Lyon, while in types C and D the substance which dis-
played the above affinity for the blue colour seems to have been
used up almost completely, so that had they not been observed
in types A and B they would have been probably overlooked in
the other two types, owing to the difficulty with which they are
stained. They can be stained, however, with bleu de Lyon in
these types as well as in types A and B, provided the specimens
have been preserved in absolute alcohol, or in corrosive subli-
mate and glacial acetic (fig. 6a). Butif they have been pre-
served in Hermann’s or Flemming’s fluid they often display
an affinity for the carmine rather than for the bleu de Lyon,
and are consequently stained red (fig. 7).

Aprenpix B.

On the Origin of the Cell Groups found in
Types B and C.

The groups, the origin of which will be described in this
appendix, must have originated from one or other of the three
kinds of cells described in type A. They may have been
derived from the flagellated cells by immigration, or from the
cells with vesicular nuclei, either directly by division or frag-
mentation, or indirectly by passing through the intermediate
condition of the cells with granular nuclei.

There are two points in favour of the view that they have
been derived from the flagellated cells by immigration. In the
_ first place, the nuclei of the cell groups on the one hand and of
the flagellated cells on the other are almost identical both in
structure and size, while in both these respects they differ
widely from the granular nuclei. In the second place, the size
of the cells of the incompletely divided groups agrees with
that of the flagellated cells rather than of any others. But, on
the other hand, there are several points which go against this

4.26 RICHARD EVANS.

view. Inthe first place, the peculiar grouping and incompletely
divided condition of the cells does not fit in with the view
under consideration. In the second place, no cells have been
observed in the act of migrating from the flagellated layer in
type B. When all these facts have been considered, the
balance of evidence seems to be against the view that they
have originated from the flagellated cells by immigration.

The second possible view of their origin is that they have
been developed indirectly from the cells with vesicular nuclei,
which were first of all transformed into cells with granular
nuclei, which, in their turn, divided and subdivided to pro-
duce the groups in question. There is only one argument in
favour of this view, the fact, namely, that the nuclei could
be described as being granular, which is of doubtful importance,
since the nuclei of the flagellated cells could also be described
as being granular. On the other hand, there are several
reasons for rejecting this view. In the first place, the cells
with vesicular nuclei as well as those with granular nuclei
divide by mitosis, but at no time was a karyokinetic figure
found in these groups of small nuclei, in spite of the fact that
these cells themselves later on divide by mitosis. Inthe second
place, when the cells with granular nuclei divide, the daughter-
cells become completely separated from one another, the con-
stricted nuclear spindle being often seen as a fine thread con-
necting the two cells; but here a great number of nuclei lie in
amass of cytoplasm, which as yet shows hardly any sign of
dividing (figs. 5a and b, and 9a and 6). In the third place,
these nuclei are remarkably uniform in size, which would not
have been the case had they originated from those of the cells
with granular nuclei. The nuclei of the latter are much
larger, and differ among themselves in size. This uniformity
in size is a most remarkable thing, for in later stages, when
these cells become the collar-cells, their nuclei are not uniform,
since they divide repeatedly, and consequently the size of the
nucleus varies. The equality of size alone, apart from any
other argument, would incline us towards the opinion that the
nuclei of the cell groups in type B have been produced not by

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 427

successive division of any kind of nuclei into two, then into
four, and so on, but by the breaking up or fragmentation of
one nucleus, the fragments becoming the nuclei of the cell
group.

The discussion so far has had but a negative tendency, my
aim being to show that the cell groups have not originated
either by immigration from the flagellated layer or by division
of the cells with granular nuclei. It remains, therefore, to
bring forward some positive proof of their origin by the third
possible method,—that is, directly from the cells with vesicular
nuclei by fragmentation of the nucleus.

It has been pointed out that there is a considerable amount
of variation among the vesicular nuclei, and it is quite pos-
sible that a certain amount of it is connected with the produc-
tion of the nuclei of the cell groups. There are also in type
B a certain number of cells which possess an irregular and
blotchy nucleus, as well as certain cells which apparently do not
possess a nucleus at all. In the cells just mentioned there are,
however, a number of bodies quite distinct from the yolk
bodies, and usually exhibiting a certain amount of structure.

It will be my purpose at present to show how these frag-
mented nuclei give rise to those of the cell groups, the changes
involved in their production being represented in fig. 40. The
cell marked a presents a rather unusual condition of the vesi-
cular nucleus. The granules, instead of being small and
numerous, are large and few, while the network of linin fibres,
instead of containing small meshes, encloses large ones, and
consists of a few threads instead of several. The cell 4 repre-
sents the next stage, in which the central corpuscle has lost the
sharp outline it had in the previous stage, seen in cell a, and
the granules are situated nearer the centre. The cell c illus-
trates the next stage, in which there is no sign of a central
corpuscle, but the nucleus contains a great number of small
granules of about the same size as those found in the cells a
and 6. The cell d exemplifies the next stage, in which the
granules have grown so as to give the nucleus a blotchy
appearance, and in some cases the chromatin can be seen

VOL, 42, PART 4,—NEW SERIES, G a

428 RICHARD EVANS.

passing out of the nucleus through the nuclear membrane,
which may occasionally, so to speak, heal up again. In
such cases, though the chromatin has been cast out of the
nucleus to form the small nuclei, a kind of skeleton of the old
nucleus is often seen to accompany the smaller ones in the
incompletely divided mass of cytoplasm (fig. 56). The cell
marked d’ is a peculiar one, not often seen, and had I not
been quite familiar with mitotic figures I might possibly have
made the mistake of explaining it as a stage in mitotic divi-
sion. But the great number of the granules is a sufficient
reason for rejecting such a view. It is far more probable that
it is a cell in which the chromatin of the nucleus has become
ageregated into granules before the nuclear membrane gave
way for its extrusion. The stage represented in the cell d
passes gradually into that illustrated in e, in which the chro-
matin appears in the form of small and spherical bodies which
display no structure whatever, and which are distributed
throughout the cell, the nuclear membrane having disappeared.
By a gradual transition the condition found in the cell e passes
to that represented in f, in which the chromatin bodies show
a certain amount of structure. The fig. g represents a still
later stage, in which the chromatin bodies have grown con-
siderably, and are not very different from the nuclei of the cell
groups illustrated in fig. 9a. If the group of small nuclei
represented in fig. 40 g be compared with that in fig. 9a, and
the one represented in fig. 40 9’ with fig. 9, the resemblance
in all their characters will be found to be a most striking one.
The additional fact that the figs. 40g and g’ represent the
same cell in successive sections, and the figures of the “ cell
group” in 9a and 6 do the same, increases the importance of
the resemblance between them, and the only conclusion pos-
sible is that the one becomes the other, and that the changes
above described prove the nuclei of the cell groups to have
originated from the vesicular nuclei by fragmentation.

Since the cell groups are produced in the way described, it
follows that they must be considered as a category of cells
quite apart from the cells with granular nuclei. Though they

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 429

originate from the same cells, they are different in their mode
of origin as well as in their destiny.

The above conclusion may be further strengthened by the
analogy of the early development of the larva while still within
the maternal follicle. At the close of segmentation the young
embryo consists only of cells with vesicular nuclei. Subse-
quently the outer layer of these divides up to produce the
flagellated cells, while some of the cells more deeply situated
give rise to cells with granular nuclei. In this way the youngest
free-swimming larva, described in this paper as type A, origi-
nates. In its development, cells which were quite similar in
character gave rise to two different classes of cells, namely,
flagellated cells at the surface and cells with granular nuclei
within. In the later stages of development, that is in the pro-
duction of types B and C, these two processes seem to be going
on, but instead of going on at the surface they go on in the
interior of the solid part of the inner mass. The formation of
the cell groups in the interior corresponds to the formation,
during the earlier stages, of flagellated cells at the surface,
and is, in fact, a continuation of the same process, just as the
formation of cells with granular nuclei deep in the interior of
the inner mass is a continuation of the same process that gave
rise to similar cells underneath the flagellated layer.

If this view be correct, a process of differentiation is con-
tinually going on, producing, on the one hand, cells with nuclei
exactly similar in their character to those of the flagellated
cells ; and, on the other hand, adding to the number of the cells
with granular nuclei. It might be argued that these cells do
not develop the flagella characteristic of the flagellated cells,
and cannot, therefore, be homologous with them, or even belong
to the same class of cells. The reply to this argument is that
they do produce flagella, and that later on they develop a
collar. As collar-cells their shape is not very different from
that of the flagellated cells, the nuclei in both being situated
at the base of the cell.

The conclusion adopted, after all these arguments for and
against, is that the flagellated cells at the surface, and the cell

430 RICHARD EVANS.

groups in the interior, are to be regarded as one and the same
class of cells, while the cells with granular nuclei form another
class. Both classes are derived directly from the cells with
vesicular nuclei which retain their blastomeric characters, and
are therefore capable of giving rise to any tissues of the
sponge.

This view of the larva enables us to give a rational explana-
tion of the development of the gemmule of Spongilla into the
adult sponge. Zykoff has shown that all the cells of the
gemmule of Spongilla are alike. Hach gemmule cell possesses
a vesicular nucleus, and in all its characters, morphological
and physiological, may be compared to a blastomere of the
ovum. Hence the gemmule cells are capable of giving rise on
the one hand to collar-cells, and on the other to cells with
granular nuclei. The same is true of the cells with vesicular
nuclei in the larva, and the developmental processes are strictly
comparable in the two cases.

The gemmule, therefore, is an aggregation of these cells
brought together in its formation from the neighbouring parts
of the adult sponge. Its constituent cells, having retained
their blastomeric characters, are capable of giving rise to the
whole sponge. From this it follows that the gemmule of
Spongilla cannot in any sense whatever be described as a bud.

AprEnDIx C.
The Development and Structure of the Spicules.

The facts already known of the development of the spicules
of the Monaxonida have been so well summarised by Mr.
Minchin in his paper (11) published in this Journal at the
beginning of the year 1898, that I shall simply refer the
reader to that account.

There are two kinds of spicules in the species, the develop-
ment of which has been the subject of this paper, termed
respectively “ megascleres ”’ and ‘‘ microscleres.”” The former
are smooth, either straight or slightly curved, and sharply

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 431

pointed. The microscleres, on the other hand, are covered
with spines. The megascleres form the main skeleton, and
have been for this reason called the skeletal spicules; while
the microscleres take no part in the formation of the main
skeleton, but are scattered about loosely in the sponge tissues,
especially in the epithelial membranes, and have consequently
been termed the flesh spicules.

The Megascleres.—These spicules always make their first
appearance in the cells with vesicular nuclei. They are found
in the very youngest free-swimming larva, and are placed at an
angle of less than 90° to the line which joins the two poles of
the larva. In the pupa, and subsequently in the young sponge,
they become arranged in various ways, but the majority of them
are placed almost vertically and slightly tilted towards the
periphery of the young individual. As they grow above the
general surface of the flattened or cake-like young sponge,
they carry with them the dermal membraue, which therefore
becomes raised up from the bulk of the tissues which lies
underneath. This lifting up of the dermal membrane on the
points of the spicules gives rise to the subdermal cavity. In
individuals which have been fixed for a number of days the
spicules, instead of appearing singly as at first, become aggre-
gated together to form fibres in which the axes of the spicules
are parallel to one another. They are surrounded by a layer
of cells which secretes a substance, presumably spongin, which
stains deeply red with fuchsin S.

Figs. 36 a, 6, c, and d illustrate successive stages in the
development of the megascleres. At their first appearance
and for a considerable time afterwards they are enclosed
completely by the scleroblast, but what happens to the secreting
cell ultimately I have not been able to make out. The cyto-
plasm of the cell is exceptionally clear, but occasionally contains
a few yolk bodies and one or two nutritive vacuoles. So far
as my observations go, the nucleus of the cell does not divide,
but seems to lose its vesicular character, and to become
granular as the spicules increase in size. It is almost certain
that when a cell with a vesicular nucleus begins to secrete a

432 RICHARD EVANS.

spicule, it is destined to change in character, and to become a
cell with a granular nucleus.

The spicules when they have grown somewhat in size show
very clearly an axial thread, which traverses their longitudinal
axis from end to end, but which cannot be seen in the early
stages of their development. ‘The thread in question appears
to consist of some form of organic material which lies in the
canal of the spicule, and which stains red with fuchsin S near
the broken ends of pieces of spicules. The stain seems to
penetrate into the canal from the broken end of the spicule
and fades away gradually, showing clearly the difference
between the stained and the unstained portion of the thread.

The Microscleres.—These spicules are not formed so
early in the development as the megascleres. However, they
are found soon after fixation,—for example, the spicules re-
presented in fig. 87 a have been drawn from a specimen which
still retained the flagellated layer almost complete on the
upper surface. They are not developed in cells with vesicular
nuclei, but in cells with granular nuclei, that is in those cells
which give rise to the flat epithelium of the sponge. And,
moreover, they are often seen inside the cells of the flat epi-
thelium, two cells of which are represented in fig. 37 6, one
containing a flesh spicule.

It is seen from the above description of the development of
the spicules that cells belonging to two out of the three
classes found in the youngest larva, i.e. type A, are capable
of secreting spicules.

AprEenpDIx D.

On the Division of Collar-cells and the Multiplica-
tion of Chambers.

There occurs in Mr. Minchin’s able paper on the position of
sponges in the animal kingdom the following statement :—
«The fact remains, that both the multiplication of collar-cells

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 433

and the formation of new ciliated chambers in a growing
sponge are scarcely known, and no satisfactory observations
have been recorded with reference to this point.” In the
present appendix it is my purpose to give a brief description
both of the multiplication of collar-cells and of the formation
of new chambers.

The collar-cells, when they have been fully formed either by
the fragmentation of a cell with a vesicular nucleus or by the
immigration of the flagellated cells during metamorphosis,
divide by mitosis. It is a rather remarkable thing that these
cells, after having been developed, in one case at least, from cells
with vesicular nuclei by fragmentation, begin to divide by
mitosis.

The nuclei of the collar-cells are situated at the base of the
cell, but when they are going to divide the nuclei travel to
the other end of the cell, and are seen situated close to the
collar, which is gradually withdrawn (fig. 33 6, the cells c.c.).
The nucleus soon loses its ordinary structure, and travels into
the middle, and there goes through all the changes of mitotic
division. The longitudinal axis of the nuclear spindle is placed
tangentially to the wall of the flagellated chamber. Conse-
quently the cell must divide radially with respect to the
chamber (fig. 35, the cell ¢.c.). The presence of the spindle
in the collar-cell marked c.c. in fig. 85 proves, beyond any
doubt, that the collar-cells divide, and consequently increase
in number. It might therefore be argued, on a priori ground
alone, that the chambers also must divide, because the number
of cells which go to form a chamber has its limits. But there
is no need to have recourse to a priori reasoning, for flagel-
lated chambers which are actually dividing have been often
observed in sections. One such chamber has been drawn in
fig. 29 a, and is marked C. The chamber in question is
actually being constricted into two, the collar-cells having
already become arranged in two groups round two different
points as centres, and the cells with granular nuclei, especially
from the outer side, are making their way in with a view to
the formation of the lining of an exhalant canal, which will

434, RICHARD EVANS.

be formed by the separation of the daughter chambers. The
other chambers situated further in illustrate a slightly more
advanced stage in the division of a chamber. The space be-
tween these latter chambers is not furnished as yet with its
lining cells, or at least they have not flattened out to form flat
epithelium. The large cell with a granular nucleus, situated
close to the nutritive vacuole (m.v.) on the inner side, is about
to flatten out to form the lining of the exhalant canal that is
being developed between the two chambers.

From the above facts the following conclusions may be formu-
lated :—The cells of the flagellated chambers divide by mitosis.
They become too numerous to constitute one chamber, and
consequently are separated into two groups or daughter
chambers, which have their apopyles facing each other, and
at first, at least, open into a common exhalant canal lined by
cells with granular nuclei.

APPENDIX KE.

On Mitotic Division in the Cells of the Free-
swimming Larve of the Young Sponge.

Though it was not the object of this paper to describe the
mitotic division of the nuclei, still, owing to the fact that in
sponges little has been done on this line of inquiry, and that
I have from time to time seen phases in cell division and
drawn them, it is deemed advisable to record them. Though
they are in no way complete, they may serve as the bases of
further research.

The chromatin, immediately after the nucleus has lost its
ordinary characters, presents the form of small granules—
about a dozen in number—which are by no means easily made
out (fig. 42, 1 a).

During the next stage the chromatin granules become
arranged in the form of an equatorial plate in which the indi-
vidual chromosomes are difficult to identify. The threads of

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 495

the nuclear spindle have not yet appeared, or are indistinctly
seen. It seems that they are in the process of development
(fig. 42, 2a, 26). In the next stage the chromatin assumes
the form of a well-defined equatorial plate. The individual
granules (or chromosomes), which were difficult to identify
in the previous stage, are much more distinct, being appa-
rently twelve in number. Lach one of these chromatin
granules seems to divide, so that a double plate of granules
appears in the median plane of the spindle (fig. 42, 3 a—3 e).
In this, the metaphase stage, the spindle threads are dis-
tinctly seen, but the poles of the spindle seem to be em-
bedded in a cloudy, ill-defined area of the cytoplasm. The
nuclear spindle has evidently arisen inside the nucleus, for the
nuclear membrane in many cases still exists. Though I
have often suspected the presence of a centrosome in
the cloudy area of the cytoplasm, I have not yet been
able to demonstrate its existence with certainty. The two
small bodies in fig. 42, 4a, have probably no connection
of any kind with the centrosome, supposing such a body
to exist. The chromosomes after dividing at the centre pass
in two groups towards the poles of the spindle (fig. 42, 4 a,
4b). When they have reached the poles of the spindle, they
are arranged at first more or less in the form of a ring in
which the individual granules are distinct (fig. 42,5 a and5 3).
During the passage of the chromosomes from the centre to
the opposite ends of the spindle the spindle threads are lost
sight of, but after the passage they again become distinct.
Concurrently with the changes represented in fig. 42, 4a—4c,
the nuclear membrane disappears. Whether the threads
visible in fig. 42, 5 a and 5 6, are the same as those seen in
3 a—3 e, or are developed from the nuclear membrane which
has disappeared, I am. uncertain, but incline towards the latter
view. The chromosomes after reaching the poles of the
spindles soon lose their individuality, to all appearance at
least, and become united together so as to form a kind of cap
of chromatin, which at first gives some indications that it has
been formed by the coming together of several bodies. From

436 RICHARD EVANS.

this cap of chromatin short processes pass along the inter-
zonal fibres for a short distance, thus giving them the appear-
ance of being thicker at the ends than in the middle. The
fibres during this and the previons stage are more distinct
than they are at any other time (fig. 42,64,60). The cap of
chromatin becomes gradually more uniform, and shows fewer
signs of its multiple origin from a number of chromosomes,
but the interzonal fibres still retain their sharp outline (fig.
42,7 a, 7 5b).

The next change observable is the constriction of the cell.
The interzonal fibres, at the same time becoming much elon-
gated and attenuated, are difficult to make out, though lying
parallel with one another and still keeping their independence
(fig. 42, 8 a, 8 BD).

The chromatin presents, in some cases, a blotchy appearance,
while in others it shows signs of preparation for the final break-
ing up into granules. In the next stage the cell has become
completely divided, the daughter cells, however, being con-
nected with each other by means of a fine thread which consists
of the constricted remains of the interzonal fibres. The chro-
matin shows further signs of the final breaking up (fig. 42, 9 a,
9 5). .

Finally the two cells become separated, and all traces of the
fibres connecting the two masses of chromatin in the previous
stages have disappeared. The chromatin itself presents the
appearance of a granular blotch lying in a clear area, and
subsequently breaks up so as to produce the small granules
characteristic of the resting nucleus (fig. 42, 10 a).

AppEenpDIx F,
Technique.

The colonies of Spongilla from which the larve were
obtained were procured from the river Cherwell, in which
Spongilla grows in large quantities on the roots of the trees

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 457

which line the bank of the river, on logs of wood in the river,
and on the pillars which support the bridges. The sponges
were taken from the river in the early morning. ‘The roots on
which they grew were cut, and the whole of each colony was
transferred from the water into a glass jar full of river water
in as complete a condition as possible, and taken into the
laboratory of the University Museum, where, through the
kindness of Professor E. Ray Lankester, this piece of work
was done. Great care was exercised so as not to leave the
mother sponge for any length of time out of water. The
sponges, having been brought into the laboratory, were placed
in glass vessels with wider openings than those in which they
were carried in. The vessels were then put in the tanks, and a
slow current of water was allowed to flow over them. In this
way as many larve as was desirable could be collected.

Owing to the continual supply of fresh water the sponge was
kept in an active and healthy condition. The oscula were
always open, and a stream of water could be seen to issue out
of them, carrying with it at intervals the larve. The larve
were almost invariably produced on the same day as the mother
sponge was taken from the river, often on the same morning.
The larve appear to be carried helplessly out of the oscula by
the current, but in all cases they soon gain complete control
over their movements, and swim towards the surface. They
dart downwards instantly when disturbed. As they swam at
the surface they were easily caught with a pipette and removed
into watch-glasses. Some of the larvz so obtained were pre-
served at once, while others were reared in watch-glasses, allowed
to fix and to undergo metamorphosis, and were preserved at
various stages in their development.

The watch-glasses in which the larve were placed measured
about 18 cm. across, and were about two thirds full of river
water. The watch-glasses were first of all carefully cleaned
with strong hydrochloric acid and with ether successively.
They were then covered over with a thin layer of glycerine,
except for about half an inch near the edge, and finally with a
thin layer of paraffin of low melting-point. The glycerine

438 RICHARD EVANS.

prevented the paraffin sticking to the glass in the centre, while
near the margin, where there was no glycerine, it became firmly
adherent.

The larve, after they were placed in the watch-glasses, used
to swim about for a time, as a rule, near the surface, though a
few were always found to sink immediately and to move about
at the bottom. Those which sank, whether sooner or later,
usually became fixed and underwent metamorphosis. Some
larve, however, never went to the bottom, but continued to
swim near the surface, and would fix ultimately to the film of
air at the surface of the water if allowed to do so. In order
that some use might be made of the larvee which fixed at the
surface, glass cover-slips were floated on the surface that they
might settle ou them instead of fixing to the film of air. The
majority, however, fixed at the bottom of the watch-glasses.
After a certain number of larve had become fixed, the watch-
glasses were sunk in a larger vessel, so that those undergoing
metamorphosis might have a greater supply of water, which
was always obtained direct from the river.

Various reagents were used for the preservation of both
the free-swimming Jarve and the fixed stages. The former
were merely dropped into the preserving fluid with a
pipette. The specimens which had fixed to the glass cover-
slips were similarly treated by dropping the cover-slip with
the individual fixed on it into the preserving medium. In no
case were these specimens removed from the cover-slips.
Those specimens which had fixed to the paraffin were removed
from the watch-glass together with a piece of the paraffin to
which they had become fixed by running a needle round them.
The piece of paraffin along with the specimen settled on it
floated to the surface, and was then dropped into the preserving
fluid in the same way as the cover-slips to which some larve
had become fixed.

The reagents used for preservation were the following:

(a) Absolute Alcohol.—The specimens were placed in
absolute alcohol for five minutes, and were then removed into
90 per cent. alcohol. They were afterwards stained with a

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 439

weak solution of borax carmine and subsequently washed with
spirit and acid, being left in both of these liquids for about
fifteen minutes. They were usually left in the spirit and acid
for the same time as they were left in the borax carmine solu-
tion, but the specimens on the cover-slips had to be left in
longer, as they were to be examined whole. They were after-
wards treated with 90 per cent. alcohol, in which they were
kept to await further treatment.

The specimens on the glass cover-slips, which were to be
examined whole, were further stained with a weak solution
of bleu de Lyon in 90 per cent. alcohol. To guard against
overstaining they were taken out from time to time and exa-
mined, and when sufficiently stained they were passed through
absolute alcohol, cleared in pure xylol, and finally mounted
between two cover-slips in Canada balsam dissolved in xylol.
Being thus mounted they could be examined from both sur-
faces.

(6) Corrosive Sublimate and Glacial Acetic Acid.—
This mixture was made up of four parts saturated solution of
corrosive sublimate in 90 per cent. alcohol and one part glacial
acetic. The specimens were left in this fluid until they became
whitish in colour, and were afterwards treated in essentially
the same way as those preserved in absolute alcohol.

(c) Flemming’s Weak Solution.—The specimens were
left in this fluid for about ten minutes, and were then washed
in a current of water for an hour and finally brought up through
the alcohols into 90 per cent., in which they were left to await
further treatment.

(d) Perenyi’s Fluid.—The specimens were left in this
fluid for an hour and were then transferred into 70 per cent.
alcohol, in which they remained for two hours, at the end of
which they were removed into 90 per cent., which was changed
several times during the subsequent twenty-four hours.

(e) Osmic Vapour and Miiller’s Fluid.—The specimens
were held in the vapour given off from a 2 per cent. solution
of osmic acid. The bottle which contained the solution was
previously warmed. They were afterwards placed in Miiller’s

44.0 RICHARD EVANS.

fluid for twenty-four hours, and then washed for an hour in a
current of water, and finally brought up through the alcohols
into 90 per cent.

Some of the free-swimming larve which were held in osmic
vapour as above described were afterwards stained in picro-
carmine and mounted in glycerine. When transferred into
the glycerine the larva tumbled into pieces, and the cells
separated owing to the maceration that had gone on in the
picro-carmine. In this way free cells were obtained.

(f) Hermann’s Fluid.— Though I did not use this reagent
myself, I obtained from Mr. Minchin a number of valuable
specimens both of free-swimming larve and of fixed stages.

After the specimens had been preserved in these various
ways, and had been brought into 90 per cent. alcohol, they
were then mounted on thin sections of liver.

The larve were placed directly on the liver, that is without
anything intervening between them, and were covered with a
small drop of glycerine albumen, which coagulated in the alcohol.
The liver was then cut parallel to the longitudinal axis of the
larva, and thus enabled the orientation of the specimen to be
made out when sections had to be cut from it.

The fixed stages, on the other hand, were placed on the
liver sections, with the piece of paraffin on which they had
settled intervening between them and the liver. In no case
were the specimens removed from the paraffin. In this way
the whole specimen was mounted, without danger of any part
of it being broken away, and the lower surface being always
nearest to the liver the manner of fixation could be easily
made out in the early stages. After the specimen and the
paraffin to which it had fixed itself had been covered with
glycerine albumen, the liver section was cut off on all sides as
near as possible to the specimen so as to reduce its size.

When the specimens had been placed on liver they were
dehydrated by passing them through three successive changes
of absolute alcohol, leaving them for about two hours in each.
They were afterwards transferred to tubes containing some
chloroform at the bottom and absolute alcohol at the top.

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 441

The specimens when first put in floated in the intermediate
layer, but gradually sank to the bottom. The liquid above
them was being continually drawn away with a pipette, so
that by the time they had sunk to the bottom they were in
almost pure chloroform, after which they were transferred to
the pure liquid.

After the specimens had been brought into pure chloroform
they were transferred into small watch-glasses, which con-
tained some chloroform together with a certain amount of
high-melting paraffin (135° F.), and were placed on top of the
water-bath. The chloroform soon evaporated, and the paraffin
simultaneously melted. They were then placed in the water-
bath, where the last traces of chloroform were driven away.
They were left in the water-bath only for a very short time, in
order that they might not be subjected to high temperature
for a long interval, which is an important thing if histological
details have to be considered.

The paraffin used for embedding was that which melts at
135° F,

The sections were always cut with a Jung microtome, the
block being painted over with a mixture of collodion and gum
mastic dissolved in ether when exceptionally thin sections were
required. The usual thickness was about 3 pu, though sections
varying from 23 to 6 u were occasionally cut.

So many methods of staining have been tried, that it would
be going too far even to enumerate them all. However, the
most important and generally useful in connection with the
development of Spongilla will be given.

(a) Borax Carmine followed by Bleu de Lyon.—This
is a most important method, as it has been used by previous
observers, and has given different results in different hands.
The staining of the specimens with borax carmine has already
been described ; there is therefore no need but to give the
last part of the process. Several ways of staining carmine-
coloured sections with bleu de Lyon have been tried, but the
following has proved to be the most useful. The sections
were overstained with carmine, and could on that account be

442 RICHARD EVANS.

washed over with spirit and acid, being subsequently immersed
for some time in 90 per cent. alcohol. They were afterwards
placed in a weak solution of bleu de Lyon dissolved in alcohol
of the same strength, and were left in it for from ten to fifteen
minutes. The sections were subsequently washed with 90 per
cent. alcohol in order to take away the excess of bleu de Lyon,
which sometimes tended to mask the carmine. The differen-
tiation went on slowly, and could be watched, therefore, under
the microscope. When this washing process was complete they
were washed with absolute alcohol and subsequently with pure
xylol, and finally mounted in Canada balsam dissolved in
xylol.

(6) Carmalum and Bleu de Lyon.—Sections of material
preserved in Flemming’s fluid and also in Hermann’s fluid
were stained for twenty-four hours in carmalum, and were
then treated in the same way as the sections stained in borax
carmine, omitting, however, the washing with acid, and to
some extent varying the time as well as using stronger solu-
tion of bleu de Lyon. .

(c) Hematoxylin and Fuchsin S.—The sections were
first stained by Heidenhain’s method in iron alum and hema-
toxylin. After differentiation in the alum solution they were
washed in a current of water for twenty minutes, treated with
absolute alcohol, and stained for nearly a minute in a saturated
solution of fuchsin S in absolute alcohol, and then mounted in
the usual way.

(d) Safranin and Gentian Violet.—The sections were
stained in safranin for twenty-four to thirty-six hours. They
were then washed with spirit very slightly acidulated, and
subsequently with absolute alcohol. Afterwards they were
stained with gentian violet for from three to five minutes,
when they were washed with absolute alcohol and placed in a
solution of potassium iodide and iodine dissolved in water
until they were quite black, the time required for this purpose
being usually four or five hours. They were then slowly
differentiated in absolute alcohol and mounted in the usual

way.

STRUCTURE OF THE LARVA OF SPONGILLA LACGUSTRIS. 443

The staining solutions which were used had the following
composition :
90 c.c. distilled water saturated with aniline oil.
10 c.c. absolute alcohol.
1 gramme of the stain (safranin or gentian violet).

(e) Gentian Violet.—The sections were left in the stain-
ing solution for fifteen minutes, and were then subjected to
the action of absolute alcohol and clove oil until they were
sufficiently differentiated, being then mounted in Canada
balsam as usual. The solution used for staining purposes in
this case was the same as in the previous method.

Though the above are the chief methods which have been
used, several other methods have been tried, both in the way
of combining the above stains in a different manner and
of using other staining solution not mentioned above.

The following may serve as a few examples of the numerous
combinations which have been tried :—first, safranin, followed
by gentian violet and orange G; secondly, Bismarck brown,
followed by malachite green; thirdly, iodine green, followed
by fuchsin S; and lastly, hematoxylin, followed by orange G
or eosin or picric xylol instead of fuchsin S.

The above enumeration of the staining methods which
have been tried suffices to show that no trouble has been spared
to test the correctness of the conclusions arrived at in this

paper.

III. Comparison OF THE ABOVE ACCOUNT WITH THOSE

oF Previous AUTHORS.

When I embarked upon the study of the structure and
metamorphosis of the larva of Spongilla I hoped to be able to
show that one or other of the accounts which had already
been published was in the main correct. As I had no theory
whatever to uphold, I started on my work with an open mind.
However, I never expected that the result of my work would
be to show that nearly all the accounts were in the main
correct, though they were all incomplete.

vol. 42, PART 4.—NEW SERIES. HH

44,4. RICHARD EVANS.

In the following discussion of previous works we shall take
first each class of cells in the larva separately, and consider
the various theories that have been held with regard to it.
This method of treatment will result in less repetition than
that of dealing with the various accounts in the order in which
they were published. We may consider, in the first place,
the fate of the flagellated layer; and, in the second place, the
differentiation of the inner mass, both during the free-swimming
period and the pupal stage.

A. The Fate of the Flagellated Layer.

The views held as to the fate of the cells which constitute
the layer in question may be divided at the outset into two
classes: first, that of Gdtte (5), who holds that they are cast
off during the metamorphosis and lost altogether ; secondly, the
views of all other authors, who hold that they become of use to
the young sponge in some form or other. |

This second class of views may be divided into three sub-
classes: first, the view of Ganin (4) and Maas (7), that they
become flattened out to form the cells of the flat epithelium ;
secondly, the view of Delage (1) and Maas (8), who hold that
they become the collar-cells of the young sponge, though these
two authors differ considerably as to the details of the process ;
and thirdly, the novel view of Néldeke (18), who holds that
they are devoured by the cells with vesicular nuclei.

In considering the view held by Gotte (5), it must be
admitted that larve such as that represented in section in
figs. 29 and 29a could possibly dispense with the flagellated
layer, for the inner mass already contains all the elements
necessary for the building up of the young sponge. But I have
never seen any signs of these cells being thrown off. The
only explanation of what seems to be an error on the part of
Gotte is that he observed specimens which had in some way or
other become injured. However, the credit belongs to him of
discovering that the flagellated cells do not become the flat
epithelium or “ ectoderm.”

STRUCTURE OF THE LARVA OF SPONGILLA LAOUSTRIS. 445

The second view to be discussed is the one which was for a
long time almost universally held to be the fate of the flagel-
lated cells in most sponges. Ganin (4) and Maas (7) are the
authors who have given expression to this view with regard
to the fate of the flagellated cells of the Spongilla larva. Further
comment on the view under consideration would be unneces-
sary, as it is practically obsolete, had not Mr. H. V. Wilson
(17) given it his support as late as the year 1894 in his obser-
vations on the gemmule and egg development of marine
sponges. However, in spite of Wilson’s attempt to reinstate
this view in its former position, all that is necessary is to point
out that he makes some very important admissions: first, that
he has never seen the flagellated cells being transformed to
the cells of the flat epithelium or so-called “‘ ectoderm ;” and
secondly, that some of the ectoderm cells (meaning flagellated
cells) of the larva migrate into the interior during metamor-
phosis. These admissions, together with others made in
Wilson’s account, are fatal to the view which he upholds, not
to speak of the fact that his figures in spite of himself tend to
support the opposite view. However, with regard to Spongilla
the only statement I wish to make is that I consider that such
figures as 15 a and 15 of the present paper cannot be otherwise
explained than as a most convincing and final proof that the
cells with granular nuclei pass out to form the flattened epithe-
lium, and that the flagellated cells pass to the interior, which
is the view now held by Maas (8), and from the first by Delage
(1) and Noldeke (18).

Though these three authors agree as to the immigration of
the flagellated cells, they differ considerably as to their ultimate
fate. Maas and Delage, however, support the view that they
become the collar-cells of the young sponge, but differ widely
as to the details of the process of transformation.

Maas holds that they become the collar-cells directly, and
without passing through any such series of changes as have
been described by Delage in the formation of his “ polynuclear
groups.” Maas undoubtedly saw the “ polynuclear groups”
of Delage, the plasmodial aggregations of the present

44.6 RICHARD EVANS.

account, and came to the conclusion that the granules contained
in them were purely vitellime. He must, therefore, have
missed that peculiar stage of the development in which there
are no small cells in the interior, but only plasmodial aggrega-
tions. Maas always found small cells in the interior, either
loose or forming flagellated chambers, and hence arrived at
the conclusion that the granules, which he supposed to be
inside the cells with vesicular nuclei, are all vitelline, thus over-
looking the distinction between the yolk bodies, which are
really inside the ameboid cells, and the small nuclei of the
flagellated cells plastered to their surface. Another fact helped
to establish him in his error, namely, the continual decrease
in size of the yolk bodies contained in the ameeboid cells.
Delage (1), on the other hand, goes to the opposite extreme,
for, according to him, the granules in his polynuclear groups
are all nuclei of flagellated cells, and are situated inside the
cells with vesicular nuclei. He is not in any way impressed
by the decrease in size and the complete disappearance of
some of the granules in his polynuclear groups, facts which alone
drew Maas’ attention, and which, as we shall see later on, have
influenced Néldeke. The following expression, which occurs on
p. 356 of Delage’s valuable paper, may throw a certain amount
of light on his failure to distinguish between the vitelline con-
stituent and the nuclear one in the granules of his “ poly-
nuclear groups :”—“ Chez la larve libre, les cellules amoeboides
ne contiennent rien autre chose que leur noyau propre.” I
can hardly understand how such an able observer as Prof.
Delage made such a statement. But his method of treating
the mother sponge and of obtaining the larve may account
for it. He kept the mother sponges in vessels, and instead
of providing them with a continuous current he changed
their water every twenty-four hours, and thus obtained his
larve only at the time he gave them fresh water. In conse-
quence of not providing the sponges with a continuous current
they were for nearly twenty-four hours in a contracted condi-
tion, and no current passed through them. Therefore no
larvee could be hatched, though they might be ready to emerge.

STRUOTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 447

When the sponge again expanded on being supplied with fresh
water the larve came out, but owing to their detention inside
the mother colony many of them had used almost all their
food material, and were so far abnormal. It is highly probable,
however, that had they been preserved in some other fixing
reagent than absolute alcohol Prof. Delage would have found
the granules in question, though less numerous and reduced
in size.

Another detail of Prof. Delage’s account, with which I
cannot agree, is that the flagellated cells are actually taken
in by the amoeboid cells. After the most careful consideration
of the point in question I am of opinion that they are plas-
tered to the surfaces of the ameeboid cells rather than enclosed
by them. If, Delage’s view were correct we should not expect
to be able to see the outlines of the flagellated cells after they
had entered these associations, but they are easily made out
for some time, as can be seen from fig. 294. Delage was led
to this view by the presence of the yolk bodies, which he did
not distinguish as such, and which are undoubtedly inside the
cells with vesicular nuclei which occupy the centre of the
groups.

It appears, therefore, that owing to his failure to detect the
yolk bodies in the larva Delage concluded that all the small
bodies in the “ polynuclear groups ” were nuclei; and further,
owing to the presence of true yolk bodies inside the cells, and
his failure to detect the outlines of the flagellated cells soon
after they had entered the groups, he came to the additional
conclusion that the small nuclei were inside the central
cell. The fact is that these aggregations contain both yolk
bodies inside the central cells, and the small nuclei of the
flagellated cells plastered to their surfaces.

The last view with which we have todealis that of Noldeke(18).
The peculiarity of this view is that it involves as a consequence,
like that of Gotte (5), the formation of the whole sponge from
the inner mass, or so-called ‘‘endoderm” of certain authors.
Noldeke’s observations seem to be far too limited to enable
anyone to come to such drastic conclusions. He seems never

44.8 RICHARD EVANS.

to have seen individuals in which the interior is full of
plasmodial aggregations alone. He has also failed to distin-
guish between the yolk bodies and the nuclei of the flagel-
lated cells which have been taken in by the large ameeboid cells
in the same way as is described by Delage. But, on the other
hand, he has seen some of them decreasing in size and disap-
pearing, which Delage did not see, and consequently comes to
the conclusion that all the bodies figured in his large ameeboid
cells, which are the same as the ‘‘ polynuclear groups” of
Delage and the “ plasmodial aggregations” of the present
memoir, are completely absorbed, a conclusion which is in no
way warranted.

Moreover, the magnification to which he has drawn his figures
is not sufficiently high to enable him to distinguish between yolk
bodies on the one hand, and nuclei on the other. The figures,
therefore, which scarcely show any nuclear structure, save what
is sufficient to distinguish a vesicular nucleus as such, have no
value whatever for deciding such a point. These facts detract
considerably from the value of Néldeke’s account, and cast a
doubt upon his conclusions.

This brings to a close what we have to say as to the fate of
the flagellated cells, which, in our opinion, migrate into the
interior and become plastered to the surfaces of the cells with
vesicular nuclei. Ultimately, however, they break away from
these associations, and become the collar-cells of the young
sponge.

B. The Differentiation, &c., of the Inner Mass.

The views held with regard to the constitution and differen-
tiation of the inner mass may be divided into two classes:
first, those according to which it consists of both ‘‘ mesoderm”
and “ endoderm,” held by Ganin (4), and by Maas in his first
account (7); secondly, those according to which it contains
only an aggregation of cells called by some authors ‘ endo-
derm,” which is divisible, however, into several kinds of cells.

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 449

The second class of authors may be further subdivided: by
some the inner mass is supposed to give rise to the collar-
cells as well as to the rest of the sponge—the view held by
Gotte (5) and Noldeke (18) ; while according to another view,
supported by Delage (1) and by Maas (8) in his later paper,
the inner mass only gives rise to the ‘‘ dermal layer,” including
in that term the ameeboid cells.

Since the views of Ganin and Maas (7) are practically
the same with regard to the structure of the inner mass
and the fate of its constituents, it will suffice to discuss
Maas’ account alone. Before making any statements with re-
gard to the above account it is necessary to refer to Delage’s
criticism of it, but for the present I shall reserve what I have
to say on Delage’s views of the structures described by Maas
as flagellated chambers.

Delage says that the larvee observed by Maas are abnormal,
or rather unusual and pathological. I have shown in the
account already given that the larvae which Delage describes
as abnormal are capable of developing to the adult sponge.
He is, therefore, not justified in describing such a larva as a
pathological one incapable of righting itself. The larva in
question is not an abnormality, but a variety, which is
capable, and also does give rise to the same end result as
Delage’s normal larva, which is the same as type D of the
present account.

The chief peculiarity of both Ganin’s and Maas’ view of the
development of the flagellated chambers is that they derive
them from the layer of flattened cells (“endoderm”) which
lines the larval cavity. The “ endoderm” grows out into the
layer which intervenes between it and the flagellated “ ecto-
derm,” i.e. into the “mesoderm” of the triploblastic larva.
The swollen or expanded end of the evaginations in question
are held to be the flagellated chambers, while the intervening
canals become the exhalant system. What is peculiar about
this view is that it is absolutely right as to the fate of the
parts in question, but equally wrong as to their origin. It
appears that Maas’ mind was at that time so dominated by the

a

450 RICHARD EVANS.

idea that the sponge larva must be subjected to the dogmas
of the “ germ-layer theory,” that, having discovered both
“endoderm” and “ mesoderm”’ in the inner mass, he could not
conceive the canals and chambers as developing in situ, which
is their true origin. In many cases the short canals which
have been described in the present account do not seem to
communicate with the larval cavity at all, save in exceptional
cases, and for that reason cannot be produced as evaginations
from it.

The swollen or expanded ends described by Maas are really
the cell groups of the present memoir, and become the flagel-
lated chambers. They are derived, as has been shown in
Appendix B, from the cells with vesicular nuclei, by fragmen-
tation of the nuclei and susequent division of the cell body.
The canals which he described as communicating with the
chambers on the one hand, and with the larval cavity on
the other, have been shown to have their lining formed by
cells with granular nuclei developed in situ. They con-
sequently belong to the same class as those which line the
larval cavity, as stated by Maas, though for erroneous reasons,

In the next place we have to examine Gotte’s view, namely,
that the cells of the imner mass (“endoderm”) give rise
to the whole sponge, the “ectoderm” being completely
lost. The flagellated chambers are described by him as being
produced by groups of cells—each of which has arisen from
a single cell—enclosing a cavity, just as has been described at
full length in the present account. Gdtte did not, however,
distinguish clearly between yolk bodies and fragmenting nuclei.
A somewhat similar view of the origin of the flagellated cham-
bers was put forward by Saville Kent (6), though he did not
understand the true nature of sponges, and held that they were
Protozoa. Dendy (2) also described a similar origin for the
flagellated chambers in some horny sponges (Stelospongus).

With regard to the development of the lining of the canals
from cells of the inner mass I agree with Gotte. But for
several reasons I hold that the cells which line the canals
belong to quite a different class from those which become the

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS, 451

collar-cells. It is true that both classes are developed origi-
nally from cells with vesicular nuclei, but their mode of origin
is absolutely different, as well as their fate. The “cell groups”’
have been shown to be developed by the breaking up of one
cell, while the cells with granular nuclei increase in number by
the gradual transformation of the cells with vesicular nuclei.

Delage divides the cells of the inner mass into three classes,
which must be compared with the similar classification I have
adopted above. Delage’s three classes are the following :—
*‘cellules épidermiques, cellules amceboides, et cellules inter-
médiaires;”’ while my three classes are termed “ cells with
granular nuclei,’ “cells with vesicular nuclei,’ and “cell
groups.” ‘The second class in each system is identical. There-
fore only the other two remain to be examined and compared.
Since Delage considers the larva described by Maas, and in
the present memoir referred to as type C, to be an abnormal
form, he has probably not taken the class of cells described
above as cell groups into account, and has included what
there was of them in the larva which he considered normal,
which is type D of the present account, among his “cellules
intermédiaires.” As has been said already in describing the
larva type D, the cell groups are so few that probably they
would have been missed by me had I followed Delage in con-
sidering type D abnormal.

It seems, therefore, that Delage’s “ cellules épidermiques et
cellules intermédiaires ” are equivalent to the class here termed
cells with granular nuclei, together with a few cell groups
which might have existed in his normal larve, but which he
did not recognise. There is no valid reason for the division of
this class of cells into two. The “ cellules intermédiaires ”
are separated from the “ cellules épidermiques” for no other
reason than that their nuclei are slightly larger and their
situation deeper. The former statement, which is true only
of the cells lodged in the interior of the solid part of the
inner mass, and not of the cells which line the larval cavity,
and which are also included in this class, results from their
more recent origin from the cells with vesicular nuclei. How-

452 RICHARD EVANS.

ever, great credit is due to Delage for showing that the flagel-
lated cells of the larva became the collar-cells of the young
sponge, though he is in error in denying that they may also be
developed from the cells of the inner mass; and the division
of the class of cells with granular nuclei into two is, to say
the least, needless.

The last account to be considered is that of Néldeke (11),
who derives the whole sponge, like Gétte, from the inner mass.
Noldeke describes the inner mass as “endoderm,” and dis-
tinguishes between endoderm cavity and endoderm nucleus.
The latter varies according to the age of the larva. In
the youngest larvae, which correspond to type A of the
present account, the endoderm nucleus consists of large cells
which contain several food granules. In older larve differen-
tiation has set in, so that the inner mass consists of
“« Bildungszellen ”? and ‘‘ Amoeboidzellen,” the ‘former class
being equivalent to the two classes described in the present
memoir as cells with granular nuclei and cell groups, the latter
class being equivalent to the class of cells here described as
cells with vesicular nuclei. The most important error in this
grouping of the cells is the classifying together of the cells
with granular nuclei and the cell groups. We may repeat, in
fact, the remark made in reviewing Gotte’s work; their mode
of origin is different as well as their ultimate fate. Con-
sequently they cannot belong to the same class.

Noldeke says there are two methods by which the flagellated
chamber originates: first, from one mother cell by division ;_
secondly, by the coming together or migration of many distinct
cells to one spot. These two statements are highly suggestive.
If the flagellated chambers were derived from cells of the inner
mass alone we would hardly expect to find these two methods of
origin occurring side by side. But if the flagellated chambers
are derived on the one hand from flagellated cells, and on the
other hand from the cell groups, as has been shown in this
paper to be the case, we would almost expect to find that they
were formed in two different ways. Nodldeke’s two methods
evideutly correspond to the two kinds of cells which give

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 453

rise to them. The method by which the cells come together
to form chambers corresponds to their origin from the flagel-
lated cells, and that by which the cells are derived from one
cell to the formation of the cell groups described above. It is,
therefore, highly probable that Noldeke has seen the formation
of flagellated chambers from the flagellated cells of the larve,
as well as from the cell groups, the peculiar mode of origin of
which he did not recognise, and consequently classed them
along with the cells with granular nuclei as ‘* Bildungszellen.”

This brings to a close our remarks on the accounts already
published, and what most of all impresses us is their incomplete-
ness in every case. Ganin and Maas appear to have recognised
larve belonging to type C alone—perhaps type B as well,—
while Gétte found types A, C, and D. But he seems to have
drawn type C as a larva, and described the metamorphosis of
type D. Delage was aware of the existence of types C and D,
but considered the former to be an abnormality, and believed
the latter to be the only one that pursued the normal course of
development ; while with regard to the former, so far as he
described it, he arrived at a quite erroneous conclusion with
regard to the origin and fate of the rings of cells occurring in
it. G6tte and Néldeke are the only two who were acquainted
with type A as a free-swimming larva. Noldeke, however, seems
to have completely missed the larva described in this memoir
as type D. The larva which I have described as type B seems
to be an entire stranger to the literature on the question.
However, it is one of the best examples possible of the value
of intermediate stages. It may be that Maas saw it, for he
draws ameeboid cells with yolk granules side by side with
incompletely divided cell groups.

Now that the accounts already published have been compared
with that offered in the present account, it is necessary to make
a brief summary of the points which are considered to have
been proved.

454 RICHARD EVANS.

Conclusions.

(1) That there are different types of free-swimming larve,
which have been described as A, B, C, and D. Type A is the
youngest form of all, type B is an intermediate form between
types A and C, while type D is a variation derived along a
different line of development from type A.

(2) That the flagellated cells of the larva in all cases become
the collar-cells of the young sponge.

(3) That certain cells of the inner mass, distinguished by
their vesicular nuclei and blastomeric characters, are capable
of giving rise to collar-cells, via the cell groups, as well as to
the flat epithelium, &c. ; i.e. both to the dermal and the gastral
layer.

(4) That consequently, during the metamorphosis of type C,
flagellated chambers are developed from the “ cell groups,”
derived originally from cells with vesicular nuclei situated in
the inner mass, as well as from the flagellated cells.

(5) That in type D hardly any cell groups are formed, and
consequently the gastral layer is developed almost completely
from the flagellated cells of the surface layer.

(6) That both the cell groups and the flagellated cells of
the larva are to be considered as belonging to the same class,
the latter heing developed on the outside, and consequently
producing only flagella, while the former originate inside, and
ultimately develop both collars and flagella.

(7) That all the cavities, canals, and surfaces are lined by
cells possessing granular nuclei, which are capable of produc-
ing microscleres, even when they are situated in the flat epi-
thelium of the surface layers.

(8) That the megascleres are produced at first in cells with
vesicular nuclei, which later on become granular.

(9) That some of the cells with vesicular nuclei plaster them-
selves to the surfaces of the flagellated chambers in the young
sponge, and become pore cells, their nuclei subsequently chang-
ing in character and becoming granular.

STRUCTURE OF THE LARVA OF SPONGILLA LAOUSTRIS. 455

(10) That there is always a residue of cells with vesicular
nuclei which retain their blastomeric characters, and which are
therefore capable of giving rise to the whole sponge. Some of
these or perhaps all of them become wandering cells, and ulti-
mately give rise to the gemmule which is capable of producing
both the dermal and the gastral layers of the sponge.

(11) That the collar-cells multiply by karyokinetic division,
and that owing to the multiplication of the collar-cells in the
flagellated chambers the latter become separated into two
groups, and so produce two daughter chambers.

IV. THEORETICAL.

I do not intend to embark on a complete discussion of the
position of the sponges in the animal kingdom. However, I
have a few considerations to bring forward in favour of what I
consider to be their true relation to the Protozoa on the one
hand, and to the Metazoaon the other. My chief reason for
not wishing to debate this question at length is the publica-
tion, only a year and a half ago, of a very able article by
Mr. Minchin. I shall take all he has said for granted, and
refer the reader to his article (12). However, I must quote a
few expressions from the concluding paragraph of his paper.
After giving an account of the views held by different ob-
servers, and the arguments for and against such views, he says,
‘We have two theories to choose between ; either to regard
sponges as descended from choano-flagellate ancestors indepen-
dently of the Metazoa, or to regard them as true Metazoa
composed of the two primary layers, ectoderm and endoderm,
which have become reversed in position in the adults.” He
then states that the choice ‘‘ will depend on which of these two
assumptions is the most difficult. If sponges are Metazoa, the
collar-cell occurring in no other Metazoa must have been inde-
pendently acquired. If sponges are descended from choano-
flagellates, then the sexual reproduction, the segmentation of
the ovum, and the formation of two germ layers must be
processes analogous and not homologous with the similar

456 RICHARD EVANS.

processes in the Metazoa.”” Mr. Minchin finally concludes
that the “theory of the Metazoan nature of sponges offers in
the present stage of our knowledge fewer difficulties than the -
theory of independent descent from Choanoflagellata.”

In the course of the few remarks added here I propose to
point out the difficulties attending a theory of the Metazoan
nature of sponges, and to lay stress on certain points which
favour the theory of their independent descent from the
Choanoflagellata.

In the above quotation from the last paragraph of Mr.
Minchin’s article three points are mentioned as the mainstays
of the Metazoan theory, namely, (a) sexual reproduction; (8)
segmentation of the ovum ; and {c) the formation of germ layers.

Those who uphold the Metazoan theory of the nature of
sponges have strained every nerve to find some parallelism
between the development of sponges and that of the Metazoa,
while they have neglected to seek for points of comparison
between their development and certain processes which occur
in the multiplication or reproduction of the Protozoa, especially
in the higher and most differentiated members of the latter
sub-kingdom.

We may consider in the first place the arguments derived
from sexual reproduction. Nothing more is meant by this
term than that two cells, specially developed, unite together to
form one cell, which is capable of giving rise to an animal
similar to those which produced the original cells, which by
their fusion became the cell in question, or fertilised egg-cell.

In the Metazoa the development and maturation of the germ
cells are more or less uniform, allowing for certain variations,
when the processes are looked at from a theoretical point of view,
especially in the mode of reduction of the chromosomes. ‘The
only account existing of the maturation of the ovum in
sponges is that of Fiedler (3), in which the ovum of Spongilla
is described as showing remarkable deviations from the type
usually considered as normal for the Metazoa. Perhaps the
most important peculiarity is the origin of the polar bodies
by a process more akin to direct division than to mitosis.

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 457

They are, as it were, budded off, after disappearance of the
nuclear membrane, from the central corpuscle of the vesicular
nucleus. The importance of such a difference is evident
without further discussion. So remarkable is the difference,
that we are almost inclined to state that the process wit-
nessed in the preparation of the micronucleus of such
Protozoa as Parameecium presents greater resemblance to
what occurs in Metazoa than does that occurring in sponges.
In the Protozoon just mentioned the micronucleus divides a
number of times before conjugation is brought about, and the
interchange of micronuclear substance takes place. Some of
the products of division in Paramcecium degenerate and come
to nothing, much in the same way as the polar bodies do in
the sponges; but the important point is that the micronucleus
divides always by mitosis. There is, therefore, greater simi-
larity, as regards the point in question, between what may be
termed the maturation of the micronucleus in the Ciliata, and
that of the nucleus of the egg-cell in Metazoa, than there is
between the maturation of the latter and that of the egg-cell
in sponges. It seems, therefore, that the Metazoan theory
derives no support from the maturation of the ovum, and the
facts appear to cut both ways.

Again, we find among the Protozoa, e. g. some of the Volvo-
cinez, both male and female cells. The male cell in Volvox
and Kudorina is small, active, and motile, being comparable
to the spermatozoa of the Metazoa; while the female cell is
large, inert, and sought for by the male cell in the same way as
is generally the rule among the Metazoa. I do not mean to
assert that the fusion of the male and female cells in Volvox is
comparable in all its details with what occurs in Metazoa, but
the very existence of such a process does away with any difficulty
—based on the presence of sexual reproduction in the sponges—
of adopting the view of the independent origin of the Porifera
from the Protozoa. Since these processes occur in the Protozoa,
their transmission during the phylogenetic evolution of the
sponges from that group does not after all appear so im-
probable. In short, the incipient methods of sexual reproduc-

458 RICHARD EVANS.

tion found in the Protozoa will serve as a sufficient reason for
its full development in the Porifera.

The second argument that will have to be examined is the
segmentation of the ovum.

The division of the cell produced by the fusion of a male and
female cell is a phenomenon which occurs in both Protozoa and
Metazoa. In Eudorina, one of the Volvocinee, the egg-cell
divides regularly into two, four, eight, sixteen, and sometimes
thirty-two cells, which lie in a kind of jelly and constitute the
colony. In Volvox, again, the fertilised egg-cell divides with a
regularity almost unknown among the Metazoa, and develops
a colony of numerous individuals. The colony, becoming
spherical in shape, might be described as a veritable blastula
if it were a young stage of a multicellular animal instead of
being an aggregation of unicellular ones, somewhat more
closely bound together than Protozoan colonies usually are.
Since such regular division of the egg-cell takes place in
Protozoa, the strength of the argument based upon the seg-
mentation of the egg in sponges appears to be completely lost.
Seeing that such division takes place in Protozoa, we are
tempted to ask, what difficulty is there in concluding that it
was transmitted along one line of evolution to the Porifera and
along another to the Metazoa?

In the third place, the argument based upon the formation
of germ layers must be examined. It is perfectly evident that
we cannot argue back from the Metazoa and Porifera to the
Protozoa when examining the present argument in favour of
the Metazoan theory of the nature of sponges, as was done
when dealing with the other two arguments. It will be neces-
sary, however, to take some of the Protozoa into consideration
even here.

I am kindly informed by Mr. Minchin that the first histo-
genetic differentiation among the simplest and most primitive
of sponges, i.e. the Ascons, is the formation of a ciliated layer
on the one hand, and of ‘‘ posterior granular cells ” on the other.
In Clathrina the youngest larva consists of a ciliated layer and
from one to four posterior granular cells, while in Leucosolenia

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 459

there is a group of granular cells placed at the posterior pole
in the pseudogastrula stage, and afterwards lodged in the
interior of the larva. The granular cells in question are none
other than the mother cells of the amceboid wandering cells,
i. e. of the cells which later on become, either wholly or in part,
the reproductive cells. Clearly, therefore, the first division of
labour occurring among the cells of the Ascon larva is the
differentiation of locomotor and reproductive cells respectively.

Further, Mr. Minchin believes, and I fully agree with him,
that the same differentiation takes place in the Sycons. The
few granular cells at the posterior end in the youngest Sycon
embryo (pseudogastrula) are the same as those found in the
Ascon larvz, and they are situated, at the time of the invagi-
nation of the blastula to form the so-called pseudogastrula,
adjacent to the almost obliterated cavity of the blastula,—that
is, they are the innermost cells of all. The other non-ciliated
cells are developed later from the ciliated layer at the posterior
pole, thus producing the characteristic comphiblastula larva,
composed of ciliated cells anteriorly, non-ciliated (dermal) cells
posteriorly, and a mass of granular cells in the centre. The
central cells of the larva probably become the ameeboid cells of
the adult sponge, i.e. they become the cells which will give rise
to the reproductive cells. If this interpretation be true, the first
division of labour in the Sycons would be into mother cells of
the sexual cells on the one hand, and ciliated cells on the other,
the cells which are destined to become the dermal layer
developing later from the ciliated cells, in the same way as
has been described in Leucosolenia variabilis by Mr.
Minchin (8). It is highly probable that the methods of histo-
genesis which take place in the Ascon and Sycon larve are
essentially similar. The point I wish to emphasise, however, is
that the first cell-differentiation in these larvee is into ciliated
cells on the one hand, and reproductive cells on the other, the
latter being represented by the posterior granular cells.
Among the Metazoa this does not always occur; the repro-
ductive cells or their immediate precursors appear early in
the development only in a few cases. Asa rule, epiblast and

VOL. 42, PART 4,—NEW SERIES. 1s

4.60 - RICHARD EVANS.

hypoblast are formed, and then the reproductive cells become
separated from one of these along with the mesoblast where
this latter occurs. Again, we may invoke the aid of the
Protozoa, and notably of Volvox, a colonial Protozoon, in which
a phenomenon occurs which is on a par with what takes place
in sponges. In Volvox the reproductive cells become marked
out from all the others as male and female cells, but the vege-
tative cells remain all alike. Clearly the Volvox colony with
its reproductive cells is comparable with the larvee of calcareous
sponges at a time when they consist of only a few posterior
granular cells and of ciliated cells.

In Volvox no further division of labour takes place among
the ciliated or vegetative cells of the colony, but it would not
be very difficult to imagine the ciliated individual cells becom-
ing differentiated to two groups in the same way as those of
the larvee of Leucosolenia and Sycandra. I conclude, therefore,
that the order of differentiation of the sponge cells from the
phylogenetic point of view, as well as from the ontogenetic—in
the Ascons at least—is, first, into reproductive and locomotor
cells ; and secondly, the latter become differentiated into two
groups, one without flagella and collars, and another which
retains these cell organs.

These considerations seem to cast a doubt upon the whole
idea of germ layers in the sponge larva. Given a group of
cells such as is found in the sponge embryo at the close of
segmentation, two layers, an outer and an inner, are bound to
appear sooner or later. Such a thing might easily happen in
the case of Volvox, in the same way as it does in the case of
the sponge larve.

Schulze’s (18) great argument against the theory of the
independent origin of sponges from Choanoflagellata is that
they should have as a necessary consequence the occurrence
of choanocytes in the larva. Though I do not agree with
Schulze, rather than controvert the statement I am prepared
to meet hisdemand. I venture to point out that such cells do
often occur in the larva of Spongilla. They do not occur on
the outside, it is true, but in the inner mass in the form of

STRUCTURE OF THE LARVA OF SPONGILLA LAQUSTRIS. 461

flagellated chambers; and besides, it has been amply proved
that the ciliated cells of the larva are potential collar-cells.

It may be argued, perhaps, that the occurrence of flagellated
chambers in the larva of Spongilla is a case of precocious
segregation. This term is a highly convenient one, and the
principle involved in it is often found to operate in nature ; but
one cannot help thinking that its assistance is invoked far too
often to solve the riddles presented to us by embryology. In
the case now under consideration the appearance of choano-
cytes or collar-cells in the free-swimming larva of Spongilla
may be a case of reversion rather than of precocious segregation.

However, the existence of choanocytes or collar-cells in the
sponges, as well as their early appearance in the free-swimming
larva, together with the ontogenetic method of differentiation
found in the Ascons, the most primitive of sponges, inclines me
to believe that sponges have been evolved independently from
Choanoflagellata. This conclusion is further strengthened by
the consideration that collar-cells do not occur in any of the
various phyla of the Metazoa.

Conclusion.

As has been stated already, this piece of work has been
carried out under the supervision of my teacher, Professor E.
R. Lankester, M.A., F.R.S., and Mr. E. A. Minchin, M.A.
To the former, whose pupil I have the honour of being, I wish
to offer my heartfelt thanks for the free use of his laboratory
and all its resources, as well as for many valuable hints and
suggestions kindly given, both in connection with the work done
and its publication. ‘To the latter I am greatly indebted for
his most generous and invaluable assistance, especially in con-
nection with the technique, aud am glad of the present oppor-
tunity of expressing my thankfulness.

I have also to offer my sincerest thanks to the Principal and
Fellows of Jesus College, who not only continued my exhibi-
tion after I had completed the ordinary university course, but

462 RICHARD EVANS.

increased it to the full value of a scholarship, and thereby
enabled me to enter upon a research course of two years’
duration.

10.

ll.

12.

13.

List oF MEMOIRS REFERRED TO IN THE Text.

poe 4 ’
. Detacn, Yves.—‘ Embryogénie des Eponges,” ‘Arch. Zool. exp. et

gén.’ (2), x, pp. 8345—498, pls. xiv—xxi.

. Denny, A.—‘ Studies on the Comparative Anatomy of Sponges.

II. On the Anatomy and Histology of Stelospongus flabelli-
formis, Carter, with Notes on the Development,” ‘ Quart. Journ.
Mier, Sci.,’ n. s., xxix, pp. 825—358, pls. xxx—xxxiil,

. Frepier, K. A.—‘ Ueber Hi- und Samenbildung bei Spongilla fluyia-

tilis,” ‘ Zeitsch. f. wiss. Zool.,’ xlvi, pp. 85—128, Taf. xi, xii.

. Gantn.—“ Zur Entwickelung der Spongilla fluviatilis,” ‘Zool. Anz.,

1 Jahrg., No. 9, 1878.

. Gorre, A.—‘ Abhandlungen zur Entwickelungsgeschichte der Thiere,’

Hamburg and Leipzig, 1886.

. Kent, W. S.—‘ A Manual of the Infusoria,’ London, 1880-81.
. Maas, O.—‘* Ueber die Entwickelung des Siisswasserschwammes,” ‘ Zeit.

f. wiss. Zool.,’ }, pp. 527—554, Taf. xxii, xxiii.

. Maas, O.—*Die Embryonal-Entwickelung und Metamorphose der

Cornacuspongien,” ‘ Zool. Jahrb.,’ vii (‘Abth. f. Anat. u. Ontog.’),
pp. 431—448, pls. 19—23.

. Maas, O.—‘ Ueber die erste Differenzierung von Generations- und

Somazellen bei den Spongien,” ‘Verh. Deutsch. Zool. Ges.,’ 1893,
pp. 27—35.
Mincuin, EF. A.—‘ Note on the Larva and Post-larval Development of

Leucosolenia variabilis, H., sp., &.,” ‘Proc. Roy. Soc.,’ vol. lx,
No. 359, pp. 42—52.

Mincuin, EB. A.— Materials for a Monograph of the Ascons. I. On
the Origin and Growth of the Triradiate and Quadriradiate Spicules in
the Family Clathrinide,’”’ ‘Quart. Journ. Micr. Sci.,’ n. s., 40, pp. 469
—587, pls. 38—42.

Mrncnin, E. A.—‘ The Position of the Sponges in the Animal Kingdom,”
‘Sci. Prog.,’ n.s., i, pp. 426—460.

Né.prxe, B.—“ Die Metamorphose des Siisswasserschwammes,” ‘ Zool.
Jahrbiicher,’ viii (‘ Abth, f, Anat., &c.’), pp, 153—189, Taf. 8, 9,

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 463

14. Scuuuzy, F. E.—* Ueber den Bau und die Entwickelung von Sycandra
raphanus, Haeckel,” ‘ Zeitschr. f. wiss. Zool.,’ xxv, suppl., pp. 247—
280, Taf. 18—21.

15. Scuuuzn, F. E.—* Ueber der Verhaltniss der Spongien zu den Choano-
flagellaten,” ‘ Sitzungsber. Akad. Berlin,’ 1885, i, pp. 179—191.

16. Torsrnt, E.—“ Notes histologiques au sujet de Leucosolenia coriacea
(Mont.), Bow,” ‘Bull. Soc. Zool. France,’ xvii, pp. 125—129.

17. Witson, H. V.—“ Observations on the Gemmule and Egg Development
of Marine Sponges,” ‘ Journ. Morph.,’ ix, pp. 277—406, pls. xiv—xxv.

18. Zyxorr, W.—“ Die Entwickelung der Gemmula bei Ephydatia fluvia-
tilis,” ‘Bull. Soc. Imp. Nat. Moscow’ (1892), No. 1, pp. 1—16.

EXPLANATION OF PLATES 35—4l,

Illustrating Mr. Richard Evans’ paper “ On the Structure and
Metamorphosis of the Larva of Spongilla lacustris.

All the figures have been drawn with the camera lucida, some of them
being magnified about 350, the rest 1000 times.

SIGNIFICANCE OF THE LETTERING.
Cavities, CANALS, AND MEMBRANES.
C. Cavity of the flagellated chamber. D. Dermal membrane, MM. Mar-
ginal membrane. #.C. Exhalant canal. J.C. Inhalant canal. Z.C. Larval
cavity. S.C. Subdermal cavity. 7. ep. Flat epithelium.

CELLS.

c.c. Collar-cell. c.g.z. Cell with granular nucleus. c.v.z. Cell with a
vesicular nucleus. fi.c. Flagellated cell. g.c. Cells of the cell groups.
pl.a. Plasmodial aggregation. sp.c. Spicule cell, or scleroblast.

NUCLEI.
g.v. Granular nucleus. 2./.c. Nucleus of flagellated cells. z.g.c. Nucleus

of a cell of the cell groups. .sp.c. Nucleusof spicule cell. 7. c.¢. Nucleus
of collar-cell. ».. Vesicular nucleus.

OTHER STRUCTURES.

ce. Collar. fl. Flagellum. jl. c.c. Flagellum of a collar-cell. sy. Spicule.
e.p. Apopyle (or exhalant chamber pore). 7.p. Prosopyle (or inhalant

chamber pore). y.. Yolk body. wm. v. Nutritive vacuole. 4.7. v. Blackened
nutritive vacuole.

464 RICHARD EVANS.

PLATE 35.

Fic. 1.— x 1000. Portion of a section of a larva, type A, at the junction
of the larval cavity and the solid posterior part of the inner mass (ef. Fig. 9).
Below the flagellated cells (fl. c.) are scen the cells with granular nuclei
(c.g.m.) containing a few yolk bodies, and in the interior the cells with
vesicular nuclei (c.v.#.) full of yolk bodies. The vacuole (z.v.) seen in
the greater number of these cells is the nutritive vacuole, which blackens
with osmic or Hermann’s fluid. The yolk bodies (y. .) are deeply stained
with bleu de Lyon, and exhibit a clear area in the centre, which in some cases
has a red patch in it.

Fig. la.— x 1000. Portion of a section of the same larva as Fig. 1, showing
the layers of cells at the side of the larval cavity. Note the yolk bodies
(y. 6.) in the cells with granular nuclei, and that they are few in number.
Note also the absence of cells with vesicular nuclei, as well as of a flattened
layer of cells lining the larval cavity (Z. C.).

Fie. 2.—x 1000. A few cells with vesicular nuclei drawn from a larva
which had been preserved in Hermann’s fluid. The sections were afterwards
partly bleached with chlorine, and owing to this process of treatment the
nutritive vacuoles (4, 2. v.) became decolourised, presenting the same appear-
ance as in the previous figures.

Fie. 3.—A cell with a vesicular nucleus which has divided into two, the
cell itself being as yet undivided. Note that it also represents a transitional
stage from a cell with a vesicular to one with a granular nucleus. The
transitional characters are—a few yolk bodies (y. 4.), a reduced nutritive
vacuole (z. v.), and two nuclei possessing very small central corpuscles.

Fie. 4.—>x 1000. The yolk bodies as seen in a preparation stained with
Bismarck brown followed by malachite green. Note the small refringent
structures contained in the yolk bodies (y.0.), and also their variety of form.

Fic. 5.— x 1000. Portion of a section of a larva, type B, at the junction
of the larval cavity (2. c.) with the solid posterior end of the inner mass. In
addition to the flagellated cells (/.c.), the cells with granular nuclei (c.g. 2.)
and with vesicular nuclei (c.v..), there is also a group of small cells with
small nuclei (g.c.). The cells with granular nuclei are flattened to form a
kind of lining to the larval cavity, and their nuclei are beginning to assume
an oval form.

Fie. 5a.—A portion from the centre of a section from the same larva as
Fig. 5. The groups of cells with small nuclei have more or less run together.
The yolk bodies in many cases present no change from that represented in
Fig. 1; others are greatly reduced in size and present a circular section instead
of an oval one, besides showing a certain amount of structure in the interior
and a tendency to stain red rather than blue, a state of things due to the older

STRUCTURE OF 'THE LARVA OF SPONGILLA LAOCUSTRIS. 465

condition of the larva on the one hand, and the method of preservation with
Hermann’s fluid on the other.

Fic. 56.— x 1000. Section of the solid posterior end of the inner mass,
containing a cell group (gy. ¢.) in which the nuclei lie at the periphery, the
cytoplasm being incompletely divided. There is also at the top of the cell
group a curious structure, perhaps the remains of the old nucleus which
gave rise to the smaller nuclei.

Fic. 5¢e.—xX 1000. Section of a cell group from the same larva as Fig. 5.
In the centre is seen a structure which apparently represents the nutritive
vacuole, surrounded by small nuclei.

Fie. 6.—x 1000. Portion of a section at the side of the larval cavity,
type D. The cells are devoid of enclosures of any kind. ‘The cells with
granular nuclei have become quite flattened to form a lining to the larval
cavity. Their nuclei are slightly compressed, but otherwise do not differ from
those of the cells which lie between them and the flagellated cells.

Fic. 6¢4.— x 1000. Part of a section of the same larva as Fig. 6. Note
that the cells with granular nuclei (c.g. x.) are far more numerous than they
were in the larva represented in Fig. 5 a, type B, and that the ‘ cell groups ”
are far less numerous, being in reality almost a uegligible quantity. The
yolk bodies (y. 4.) are also greatly reduced in size and diminished in numbers.
All the cells in general as well as their nuclei are far smaller than they are in
either type A or type B.

Fic. 7.— xX 1000. Portion of a section of type C at the junction of the
larval cavity with the solid posterior part of the inner mass. Note the cells
with granular nuclei (c.g. x.) making their way towards the surface, and the
cell groups (g.¢.) in places arranged in chamber-like rings with developing
collars. The yolk bodies are few in number, and stain red owing to the
method of preservation. The nutritive vacuoles are intensely black owing to
the action of Hermann’s fluid during preservation.

Fie. 7a.—xX 1000. Two chambers drawn from the same larva as Fig. 7.
One chamber (4.) has been cut transversely, while the other (B.) has been
cut longitudinally, ‘The collar-cells have well-developed collars and flagella,
and are arranged to form horseshoe-like figures. The large opening towards
which the collars and flagella are directed is the apopyle (or exhalant pore).
It opens into the beginning of an exhalant canal (#. C.), which is lined by a
flattened cell with granular nucleus. Note that the collars unite by their
margins to form what is usually called Sollas’s membrane.

Fic. 8.—x 1000. Surface view of the osculum (osc.) and ostia (os¢.)
of a young sponge soon after their formation. The osculum is still on a level
with the general surface. The rim is formed of superposed layers of cells
with granular nuclei, which form the flat epithelium, Note also the presence
of an ostium (?) close to the osculum. It is a small opening measuring about

466 RICHARD EVANS.

15 » across, while the osculum measures from 60 to 70 ». Compare the
figures on Plate 39.

PLATE 36.

Fic. 9.— x 350. Entire section of a larva, type B. Note the flagellated
layer (fl. c.) covering the whole surface ; inside the flagellated layer the cells
with granular nuclei (c.g. z.) forming a lining to the larval cavity as well as
scattered about in the solid part of the inner mass; the cell groups (g. ¢.)
deeply embedded, as a rule, in the inner mass; and finally the cells with
vesicular nuclei (c. . z.), containing numerous yolk bodies and usually only a
single nutritive vacuole (w. v.), coloured black in one half but clear in the
other half of the figure. The section from which this figure was drawn was
bleached with chlorine, which rendered the nutritive vacuoles clear, as repre-
sented in the right half of the figure; in the other half they are represented as
seen in the section before bleaching.

Fies. 9a and 94.—x 1000. Figures drawn from the same larva as Fig. 9,
introduced to show the groups of small cells (g. c.) with their small nuclei
(z. g. ¢.), which are seen in the centre of Fig. 9 a asa patch, but in Fig. 94
as a ring round the periphery of the undivided cytoplasm. In these two
figures the same group of small nuclei is represented in successive sections.

In Fig. 9a the group has been cut tangentially, and both the common
membrane by which they are surrounded and the faint lines which separate
them can be seen. ‘These faint lines represent the first appearance of the cell
walls in the multinucleated cytoplasmic mass.

In Fig. 90 the group has been cut radially ; the nuclei, therefore, appear in
the form of a ring round the cytoplasmic mass in which they lie. There are
faint dividing lines to be seen passing between the nuclei which stretch across
the central space.

Fie. 10,— x 1000. Tangential section of a larva of type D, the fourth
section of the series. The nuclei of the flagellated cells (~.f.c.) are cut
transversely, and are therefore circular in section. The cells with granular
nuclei occupying the centre of the section are branching and irregularly
shaped cells, and are far from forming a complete membrane even at this
stage. The nuclei of the flagellated cells in some places appear to lie within
the cells with granular nuclei, but a radial section proves that this is a de-
lusion ; they merely lie close to one another.

Fie. 11.— x 350. Entire section of a larva of type C. The flagellated
cells are much the same as they were in Fig. 9. The cells with granular nuclei
(c. g. 2.) are smaller, and are extremely flattened towards the larval cavity.
The cell groups are in many cases completely divided and form chamber-like
rings, the individual cells of which are often provided with a collar and a

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 467

flagellum. The yolk bodies (y. 6.) are far less numerous, though always
present, as will be shown in more highly magnified figures. The nutritive
vacuoles are smaller, owing to the fact that the cells with vesicular nuclei
which contain them are smaller. This figure was produced in the same way
as Fig. 9.

Fie. lla.— x 1000. Portion of a section at the junction of the larval
cavity with the posterior part of the inner mass, from the same larva as
Fig. 11. The cell groups, which are very numerous, are arranged in chamber-
like rings. The cells with granular nuclei (¢. g. z.) lying towards the larval
cavity are highly flattened, forming a lining to it. The nutritive vacuoles
(4. 2. v.) are blackened by the action of the preserving reagents, and the yolk
bodies are few and reduced in size.

Fie. 12.— x 1000. A group of small nuclei such as is characteristic of
type B, from a larva which in other respects possesses all the characters of
type C. The difference between the small nuclei and the granular nuclei of
the cells which surround them is most striking, and their presence in this
stage of development in type C points to their production being a process
which goes on continuously.

Fies. 18, 18a, 136.—x 1000. Sections of a larva of type C to show
that the larve contain yolk granules (y. 4.) even in the oldest free-swimming
stage. Besides those in the cells with vesicular nuclei, even the cells with
granular nuclei and cell groups contain a few of them, a proof that the
number of granules in type C is a variable quantity.

In Fig. 13 there is a ring-like group of cells, among which a very interesting
cell occurs, showing a stage of the nucleus preparatory to division. The
nucleus (z. g. c.) las travelled from the base of the cell to the side, and has
lost its usual structure, and has an irregular blotched form.

Fie. 14.— x 1000. Section through the side of the cavity of a larva of
type C when on the point of becoming fixed; showing the region where the
cells with granular nuclei first, make their way out, and the flagellated cells
immigrate. The structure of the small nuclei (7.77. c.) is in some cases already
beginning to change. They become smaller in size, though the actual
quantity of chromatin seems to increase. The linin threads thicken, and the
chromatin shows a tendency to become aggregated into one or more irregular
patches, lying close to the nuclear membrane. Owing to these changes it
is already difficult to distinguish some of the small nuclei from the reduced
yolk bodies, though they are easily distinguished from those that are not
reduced.

Fic. 14a.— x 1000. Two cells from the same larva as Fig. 14. They
show how the cells with granular nuclei flatten out to enclose the rudiment,
of an exhalant canal in the free-swimming larva, and should be specially com-
pared with the cell groups found in the same type of larva.

468 RICHARD EVANS.

PLATE 37.

Fie. 15.—x 350. Complete section of a pupa derived from a larva type
D, with consequently very few ring-like groups of small cells. Plasmodial
aggregations have not yet been formed. It contains cells with vesicular
nuclei, in which there is a comparatively large number of yolk bodies (y. 4.).
The flagellated cells have disappeared from the upper surface as well as from
the lower, and the cells with granular nuclei have nearly formed a complete
layer on both surfaces. The remains of the larval cavity (Z. C.), not yet
obliterated, are seen as small slits near the lower surface, which is a proof
that this larva is fixed by the anterior pole. ‘The flagellated cells, especially
at the lower surface, present a fan-like arrangement.

Fic. 15 a.—x 1000. Portion of the lower surface of a section from the
same larva as that drawn in Fig. 15. It shows the fan-like appearance of
the immigrating flagellated cells and the cells with granular nuclei (c.g. 7.)
passing out between the groups. The nuclei of the flagellated cells which
have adhered to the surface of the cell marked c.g. . have already changed
in character. The outline of the cells, however, can be made out. The
cells with vesicular nuclei contain several yolk bodies, which, owing to their
large size, are easily distinguished from the nuclei of the flagellated cells ;
the latter are plastered to the surface of the cell (c.g.2.). There are also
a few cells, about half a dozen in number, which belong to the class described
as cell groups (g.c.), which in this larva are exceedingly few in number. A
portion of the larval cavity still remains (Z. C.).

Fie. 15 4.—x 1000. Portion of the lower surface of a section from the
same series as Fig. 15. ‘The larger group of flagellated cells, which have
travelled in, is quite exceptional, as is also the fully developed chamber situated
close to it. The presence of this chamber, however, shows that the differ-
ence between pups derived from type D and those derived from type C is
only one of degree, the flagellated chambers in the one being much more
numerous than in the other. The nucleus of one of the collar-cells is prepar-
ing to divide, the chromatin having travelled to the centre of the cell after
the nucleus has travelled from the base to the inner end of the cell (cf. Fig. 18,
2. Je C.).

Fic. 16.— x 350. Complete section of pupa slightly older than that drawn
in Fig. 15, derived from type D. There are no free cells in the interior,
which is filled with ‘‘ plasmodial aggregations ” (p/. qa.) in which no cell limit
can be detected, though that of the whole group is sharply defined. Spicules
(sp.) protrude at the upper surface.

Fie. 16a.— xX 1000. Portion of the same section as Fig. 16, showing
both the upper and lower surfaces. The plasmodial aggregations (p/. a.) are
seen depicted on a larger scale. They contain, as a rule, a vesicular

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 469

nucleus at the centre, together with smaller nuclei and yolk bodies nearer the
surface. ‘The yolk bodies are almost indistinguishable from the small nuclei
of the flagellated cells during this stage. Note also the large granular nuclei
situated on the lower surface, which owing to their large size cannot possibly
be derived from the nuclei of the flagellated cells.

Fic. 17.—x 1000. Portion of a section of a pupa slightly more advanced
than that represented in Fig. 16, derived from type D. The nuclei of the
flagellated cells in the plasmodial aggregations are changing in character; that
marked @ has become larger in size, and the chromatin in it is evenly dis-
tributed ; in that labelled 4 strands or thick threads appear; while in those
marked d@ the ultimate structure of the nuclei of the collar-celis has been
practically attained. All these four nuclei are still well within a perfectly
definite plasmodial aggregation, proving in a most satisfactory way that the
nuclei of the flagellated cells become the nuclei of the collar-cells, since the
four nuclei in question are hardly distinguishable from those marked e, which
are undoubtedly the nuclei of cells which later on become collar-cells.

Fies. 18, 18 a, 18 4.—x 1000. Section of a pupa which is slightly more
advanced than the one represented in Fig. 17. The plasmodial aggregations
( pl.a.) are losing their sharp outline and are becoming gradually united with
one another. The small nuclei of the flagellated cells, which in Figs, 16 and
16a are almost indistinguishable from the yolk bodies, are in this stage easily
made out, owing to their loose internal structure, and the thick strands of
chromatin which appear in them.

Fig. 18 represents a portion near the margin of the section, and though
there are several plasmodial aggregations it does not contain a single vesicular
nucleus. In Fig. 18a, which is further in, an occasional vesicular nucleus
is seen ; and in Fig. 18 4, which is drawn from near the middle of the section,
there are several. The cells with vesicular nuclei, therefore, are found at the
centre of the young sponge, which means that those near the margin have
been already transformed into cells with granular nuclei, which very likely
become pore-cells.

Figs. 19, 19 @.—xX 1000. Portions of a section of an individual which
is slightly more advanced than that drawn in Fig. 18, derived from
type D. The plasmodial aggregations of cells have completely lost their
individuality and have become indistinguishable from one another. In
Fig. 19a the cytoplasm is arranging itself in the form of a ring, the
cavity of which represents that of the future chamber, and the nuclei en-
closed in it are those of the future collar-cells. In Fig. 19 large spaces have
appeared inside the young individual. The upper one (S.C.) represents the
subdermal cavity, which has been produced through the pushing out of the
dermal membrane (D.) by the outgrowing spicules.

470 RICHARD EVANS.

PLATE 38.

Fic. 20.—x 1000. Portion of section of a pupa slightly more advanced
than that drawn in Figs. 19 and 19 a, derived from type D. The small flagellated
cells have extricated themselves almost completely from the plasmodial aggre-
gations, and are arranged in irregular rings (~./l. c.) round the cavities of the
future chambers. The cell limits are beginning to reappear. The future collar-
cells, i.e. the former flagellated cells, are in some cases beginning to develop a
collar which protrudes into the chamber cavity (C.). The subdermal cavity
(S. C.) has increased in size considerably, and the cavities in the interior are
becoming differentiated into inhalant and exhalant canals. The nucleus marked
v. 2. on the left side of the figure and the four small nuclei (w.//.c.) situated
close to it present a most interesting case of the nuclei of the flagellated cells
acquiring the characters of collar-cell nuclei, before breaking away from the
cell with vesicular nucleus.

Fie. 21—x 1000. Portion of a section of a still more advanced pupa
than that from which Fig. 20 was drawn, also derived from type D. The
flagellated chambers are well defined, and the inhalant and exhalant canals
are becoming lined with cells possessing granular nuclei, which are gradually
flattening out. The cells with vesicular nuclei are situated close to the
flagellated chambers in many cases. The collar-cells have almost completely
developed collars and flagella, though the latter are difficult to trace.

Fie. 22.— x 1000. Portion of a section of a slightly more advanced pupa
than that represented in Fig. 21, derived from type D. The inhalant and
exhalant canals (I. C. and #. C.) communicate with the cavity of the chamber
(C.) by means of a prosopyle (¢.p.) and the apopyle (¢.p.). The collars are
fully developed, and unite together by their margins to form the so-called
** membrane of Sollas.”

PLATE 39.

Fie. 23.—x 1000. Surface view of the marginal membrane of an indivi-
dual slightly more advanced than that represented in section in Fig. 15. The
individual cells of the membrane are quite separate, and all contain granular
nuclei. The flagellated cells, which in some cases are carried away by the
cells with granular nuclei, are plastered to the surfaces of the latter, and
now exhibit a definite outline.

Fie, 23a.—x 1000. A portion of the same preparation slightly nearer the
centre than that drawn in Fig. 23. Note how the flagellated cells become
attached to both cells with vesicular nuclei and cells with granular nuclei,
and that in many cases they are still free.

Figs. 24, 24a, 246.—x 1000. Surface views of an individual which had

STRUCTURE OF THE LARVA OF SPONGILLA LAGUSTRIS. 471

been fixed for approximately the same length of time as that represented in
section in Fig. 31. Like the pupa drawn in section in Fig. 31, it had not
advanced far from the condition found to be characteristic of the second larval
type when first fixed.

Fig. 24 represents an irregular portion of the marginal membrane, while
Fig. 24 represents a portion which has already acquired an unbroken outline.
In both figures several yolk bodies (y. .) are seen, and it may be that there
is here and there an occasional nucleus of a flagellated cell. But nearly all
of them are bodies of the same nature as those marked y. 4. in Fig. 23.

Fig. 244.—x 350. ‘This figure represents a surface view of a portion of
the larva close to the inner limit of the marginal membrane. The nuclei of
the flagellated cells and the yolk bodies seen in the plasmodia! aggregations
are not easy to distinguish from one another (comp. Fig.31). The small cells
derived from the cell groups, which are so characteristic of the larva described
as type B, are aggregated together in a loose fashion, and are proceeding to
develop into collar-cells.

At the right lower corner the dermal membrane (D) alone, consisting of
flattened epithelium, has been drawn, a large subdermal cavity being situated
immediately below it. The developing chambers, i.e. the cell groups, are
somewhat loosely held together by cells with granular nuclei (coz. ce//). The
nuclei of the flat epithelium—not seen in the same focus as those of the cell
groups—are drawn over or above the groups of chamber cells in the lower
part of the figure, but not in the upper part. The group in the right-hand
upper corner of the figure has been left clear without anything being drawn
above it (c.¢.,g.¢.). The two exceedingly large granular nuclei are possibly
nuclei of developing pore cells, which have already lost their vesicular
character (p. ¢. .).

Fig. 25.— x 1000. Portion of a section of a larva similar to that of which
a part is drawn in surface view in Fig. 23. The small cells in both figures
are the flagellated cells of the larva. The bodies enclosed in the plasmodial
aggregations are in some cases yolk bodies (y. 4.), and in others nuclei of
flagellated cells (7. ff. c.); but the formation of the plasmodial stage is as
yet far from complete. Note that the nuclei of the small cells (7. c.) are
much smaller than those of the cells c. c. in Figs. 246 and 31. The latter
have already attained the structure of the collar-cell nuclei, while the former
are only in a stage of preparation for the formation of plasmodial aggregations.

Fic. 26.—x 1000. Portion of a section of an individual slightly more
advanced than that drawn in Fig. 25, also derived from type D. This figure
should be carefully compared with Figs. 16 and 16a. The plasmodial aggre-
gations (pl. a.) are fully formed, and by means of the differentiating stain the
nuclei of the flagellated cells are easily distinguished from the yolk bodies,
though there is no sign of cell outlines.

472 RICHARD EVANS.

Fie. 27.— x 1000. Section of the lower surface of an individual which
contains a few loose cells in addition to the plasmodial aggregations, and is
derived from type D. ‘The small group of loose cells (g.c.) should be
specially compared with the similar groups in Fig. 7 a.

Fie. 28.— x 1000. Portion of a section of an individual slightly more
advanced than those represented in Figs. 24 dand31. It shows two chambers
which are exceedingly different in the character of their individual cells. In
the upper one all the cells and their nuclei are like one another, while in the
lower one they are not so. The nuclei of the cells (c. c., g.c.) of the upper
chamber resemble those of the chamber cells of Figs. 24 6 and 31, while those
of the cells (c.c., fl.c.) of the lower chamber are far more like the nuclei
of the chamber cells seen in Figs. 19, 20, and 21. The cells c.c., g.¢., have
evidently been derived from cell groups, while the cells c.c.,fl.c., have
been produced from flagellated cells, and have not as yet attained their
definitive structure, after passing through a series of changes in connection
with the formation of the plasmodial aggregations. |

PLATE 40.

Fre. 29.— x 350. Complete section of a pupa from a larva of type C. The
flagellated layer has completely disappeared from the lower surface, while it
is still complete on the upper surface. The flagellated cells which have
travelled in from the lower surface are already in process of forming plasmodial
aggregations. The flat epithelium of the lower surface is almost complete, but
the cells with granular nuclei still remain inside the flagellated layer on the
upper surface. Flagellated chambers (C.) are fully formed. The larval cavity
is still retained, and will probably become a part of the exhalant system and
gastral cavity.

Fic. 29 a.— x 1000. A portion of the upper surface of the same larva as
Fig. 29. It shows the flagellated layer absolutely complete at the same time
as fully formed flagellated chambers. The yolk bodies are rather numerous,
and their remains are seen even in the collar-cells. It is quite possible that
these small bodies seen in the collar-cells here are the same as the refringent
bodies seen in preparations mounted in glycerine, after maceration with osmic
and staining with picro-carmine (cf. Figs. 38 and 39).

The flagellated chamber marked C. is being constricted to form two chambers.
The two chambers situated further in have been produced in the same way,
the space between them becoming an exhalant canal.

Fie. 29 d.— x 1000. A portion from the lower surface of a section from
the same larva as that drawn in Fig. 29. The flagellated epithelium has com-
pletely disappeared, and the flat epithelium is well formed. The plasmodial

STRUCTURE OF THE LARVA OF SPONGILLA LACUSTRIS. 473

aggregations (p/. a.) are in an advanced stage of development, but the outline
of the flagellated cells adhering to their surfaces can be easily made out.

Fic. 29 c.—Three fully developed coliar-cells from the same larvaas Fig. 29.
The cell @ has a small nutritive vacuole (z. v.) which points to its recent origin
from a cell group, as well as three other small bodies, probably reduced yolk
bodies (y.4.). The nucleus (n. c. c.) is often onion-shaped, and the flagellum
which protrudes out of the collar passes down to it, and has usually a small
swelling at its base.

Fies. 30, 30a.—x 1000. These two figures represent portions of
sections of an individual slightly more advanced than the one represented
in Fig. 29, derived from a larva of type C. Fig. 30q@ is continuous
with the upper corner of Fig. 30. The flagellated layer has not yet com-
pletely disappeared from the surface, and its cells are becoming widely
separated owing to individual immigration, and also owing to the superficial
expansion of the pupa as awhole. The cells with granular nuclei (c.g. x.)
are in some cases at the surface, while in other cases they are seen in the act
of passing to it. At the lower surface (Fig. 30) these cells form an almost
complete layer. The larval cavity (Z. C.) is still very large, and is lined by a
layer of cells with granular nuclei similar to those at the surface, but smaller
and more flattened. The cavity shows no sign of disappearing.

In Fig. 30 the nuclei of the flagellated cells are changing in character pre-
vious to the formation of plasmodial aggregations, the portion nearest the
margin being full of them.

In Fig. 30 @a flagellated chamber derived from the cell groups has been cut
almost tangentially (C.).

Fic. 31.—x 1000. Portion of a section of an individual fixed for a
longer time than that drawn in Fig. 30. The flagellated layer has completely
disappeared, and the flat epithelium is well formed. The flagellated cells have
entered into the formation of plasmodial aggregations, and are indistinguish-
able from the yolk bodies. Flagellated chambers are in the process of develop-
ment from the cell groups, the collars and the flagella not having been produced.
This pupa here represented must have fixed during the stage of development
described as type B, or else the cells of the cell groups would have developed
collars before the fully formed condition of the plasmodial aggregations had
been attained.

Fie. 31 a.— x 1000. Portion near the margin of a section of the same
individual as Fig. 31. It shows how the flat epithelium of the upper and lower
surfaces passes into the marginal membrane, and how all kinds of cells make
their way into the cavity which exists between the two layers of epithelium
close to that membrane,

474 RICHARD EVANS.

PLATE 41.

Fics. 32 a—d.— x 1000. Four successive sections of the same flagellated
chamber, drawn to show the nature of the inhalant canal and pore. The canal
which is drawn is an exceedingly short one, and passes straight from the sub-
dermal cavity above into the flagellated chamber below.

a shows how the layer of flat epithelium (7. ep.) forms a depression round the
entrance of the canal, the cell below forming, apparently, the actual wall of
the canal itself.

6 shows the canal (7. p.) along its whole length, save a small portion near
the surface, where it is covered over by the flat epithelium which lines the
subdermal cavity (S.C. in a). On one side a flattened cell is seen to line the
canal, but there is no nucleus anywhere.

ce shows the same canal or pore (%. p.) opening into the flagellated
chamber. It appears to be surrounded by the same cell on all sides, and the
cell in question is distinct from the flat epithelium of the subdermal cavity.

d shows no sign of the canal, but a large cell with a vesicular nucleus
occupies the position taken up by the canal in figs. and c. The cell in
question (c.v. 2.) is the same cell as lines the canal in the other figures, and is
perforated by it.

Fie. 33.—A flagellated chamber with a short inhalant canal and inhalant
pore (i..). Apparently both the canal and pore are lined by the same cell, the
nucleus of which is not seen.

Fic. 34.— x 1000. The same inhalant canal (J. @.) and pore (¢.p.) in the
section succeeding the one drawn in Fig. 33. In this section the nucleus of
the pore cell is seen, and is evidently a vesicular nucleus in the process of
transformation to the granular condition.

Fic. 35.—This figure represents a section of a flagellated chamber, and
shows two things:

(1) A collar-cell which has withdrawn its collar and flagellum, and contains
a nucleus in the spindle form, showing clearly that the cells of the flagellated
chambers divide by karyokinesis. Note that the longitudinal axis of the
spindle lies in a plane tangential to the wall of the flagellated chamber. The
cell must therefore divide longitudinally.

(2) A transverse section of a spicule (sp.) lying in the scleroblast, a cell
with a vesicular nucleus (v.2.) against which the spicule presses giving it
a crescentic appearance.

Fies. 836a—d.—x 1000. Four stages of growth of the megascleres,
The spicule in each case is completely enclosed by the scleroblast. In
Fig. 36a the nucleus is vesicular as well as in 364; but in Fig. 36c it
begins to lose its vesicular character, and tends to become granular. In

STRUCTURE OF THE LARVA OF SPONGILLA LAOUSTRIS. 475

Fig. 36 ¢ this stage has been reached, the nucleus being similar to that of any
flat epithelial cell.

Fies. 37 a, 6—xX 1000. Microscleres in their epithelial scleroblasts,
completely enclosed by the cells which secrete them.

Fig. 37a represents spicules from a larva belonging to type C, preserved
immediately after fixation, i.e. while the flagellated layer was still complete at
the upper surface.

Fig. 37 4 represents two cells of the flat epithelium, one of which contains
a microsclere, drawn from a somewhat advanced young sponge.

Fies. 38, 39.— x 1000. These figures represent the cellular elements of
a larva which was preserved in osmic acid, stained in picro-carmine, and
mounted in glycerine. Owing to the maceration that had taken place
during fixation and staining it fell immediately into pieces, and the cells
separated.

Figs. 38 a—d represent the flagellated cells, and show small granules
which were not seen by the aid of any other method of preparation.

Fig. 39 shows a portion of the flagellated layer from within, and the
granules are seen in all the cells.

Figs. 38 a, 6 show the difference in length which obtains between the
flagellated cells, ¢ the flagellum passing down to the nucleus, while in d it
could not be traced any further than a small and irregularly shaped granule
situated about halfway down to the nucleus.

The cells e and / are extremely irregular in shape, while the outline of the
cell g is much more definite. The cells e and / contain far less granules than
the cell g, a fact which may have some relation to their shape, the cell g
being rendered more inert than the others by the granules.

The cells f and g, which are provided with a vesicular nucleus, contain a
large nutritive vacuole, blackened by the osmic acid. The cell g contains
several large yolk bodies (y. 4.), while the cell e has only one.

All these cells contain small refringent granules, probably reduced yolk
bodies, though they are not shown in the flagellated cells of the larva by
any other method of preservation. Compare Fig. 29 a.

Fies. 40 a—g'.— x 1000. Development of the cell groups characteristic
of the larva of type B. For description see Appendix B, on pp. 425—430.

Fie. 41¢a.—x 1000. Figure showing the difference of structure between
the vesicular nuclei; one, with a few but large granules, is in the same state
as the cell a of Fig. 40; the other represents the structure of an ordinary
vesicular nucleus.

Fie. 414.—x 1000. Two cells which, to judge from the number of yolk
bodies contained in them, originally had vesicular nuclei. The nucleus,
however, has become fragmented, and is represented by a number of small
granules situated in the centre of the cells. The condition of the nucleus

VOL, 42, PART 4,—NEW SERIES, KK

476 RICHARD EVANS.

in these cells probably represents a stage between the cells b and ¢ of
Fig. 40.
Fies. 42, 1a—10a.—x 1000. Various stages of mitotic division found in

the cells of the free-swimming larva and of the young sponge. See Appendix
K, pp. 434, 435.

C@LOM AND VASCULAR SYSTEM IN THE LEECH. 477

On the Communication between the Celom and
the Vascular System in the Leech, Hirudo
medicinalis.

By

Edwin 8S. Goodrich, B.A.,
Aldrichian Demonstrator of Comparative Anatomy, Oxford.

With Plates 42—44,

ContTENTs.
PAGE
INTRODUCTORY 4 , : 3 ; a ATT
Tue EvipEence oF INJECTIONS : ; F . 481
THE KvIDENCE OF SECTIONS. : : ’ . 482
SUMMARY AND CONCLUSION . : : : . 490

THE investigation, of which an account is given in the
following pages, was undertaken with the object of ascertaining
for certain whether, in the medicinal leech, the cavities of the
so-called sinus system do, or do not, communicate with those of
the contractile vascular system.

This question, which at first sight seems so simple, has for
many years given rise to much controversy, some authors
believing the communication to exist, others, on the contrary,
holding the view that the two systems of cavities are quite
distinct. Since the champions of neither theory have brought
forward conclusive evidence in support of the view they uphold,
the question remains to this day unanswered.

Several reasons have tended lately to arouse renewed interest
in the relations of the vascular system of the leech, and to

478 EDWIN 8S. GOODRICH.

add fresh importance to the controversy. For, since it has
become gradually established that, in the Invertebrata in
general, the ccelom is quite distinct from the vascular system,
that even in the Molluscs and Arthropods the two systems of
cavities are of quite separate origin, it is clear that a communi-
cation between the two would be a very exceptional pheno-
menon. Now, Birger has shown by his embryological researches
(2) that at all events the ventral sinus, the perinephrostomial
sinuses, and the branches immediately deriving from them are
of ceelomic nature and origin in Hirudo, just as they have been
shown to be in the Rhynchobdellid leeches; and, moreover, it
has recently been contended by Oka (10) and Johansson (6)
that in the adult Rhynchobdellidz the vascular system remains
distinct and closed off from the coelomic cavities. This interpre-
tation is now supported by the evidence derived from the
structure of the interesting leech Acanthobdella, in which
Kowalevsky describes a closed vascular system filled with red
blood, distinct from the spacious ccelomic cavity (7).

All these facts are so unfavourable to the view that in the
Gnathobdellid leeches, alone amongst the Invertebrata, the
ceelom is in open communication with the blood-vascularsystem,
that many authors refuse, and I think justly refuse, to admit
its truth without further and more definite proof. For instance,
Mr. Sedgwick (11), in the excellent ‘Text-book of Zoology’
which he is now publishing, says ‘it is still generally held that
they [the vascular and the sinus systems] are continuous
through their finer branches. This view is, as we shall see,
based on insufficient evidence ; and having regard tothe state-
ments of Oka and Biirger, it seems safe to assert as a fact that
the two systems are separate” (p. 517); and after further
arguments against their communication he adds (p. 519), “A
continuity between the vascular system and the undoubted
celom of the sinuses would be a unique phenomenon in the
structure of the animal kingdom.”’ !

It was on reading these passages that I finally determined

1 Apparently adopting the view that in the Vertebrates the lymphatic
system is not in real continuity with the coelom,

C@LOM AND VASCULAR SYSTEM IN THE LEECH. 479

to try and solve the question ; and although, at first, I was
strongly in favour of the view held by Sedgwick, the unmis-
takable evidence of the facts has forced me finally to adopt the
older interpretation. The result of these researches is to prove
without doubt that the continuity exists.!

It was Leydig who first showed that in the leeches there are
two systems of vessels, a contractile and a non-contractile. De
Quatrefages and Leuckart compared the former with the
vascular system, the latter with the body-cavity of the
Chetopods. Ina classical paper on the structure of leeches,
Professor A. G. Bourne carefully traced out the relations of
the main cavities of the two systems throughout the Hiru-
dinea (1). Heconcluded that in all leeches a continuity exists
between the two sets of cavities. Further, that in the
Rhynchobdellidze there are four main longitudinal blood-
vessels, lying in four longitudinal sinuses of coelomic nature ;?
the nephrostomes lie in coelomic spaces, more or less cut off
from the ventral sinus containing the nerve-cord. In the
Gnathobdellidz, Bourne believed that “ (1) al] trace of lateral
sinus and of the dilatations connected with it has vanished ;
(2) all trace of the dorsal and ventral vessel has vanished ;
(3) the lateral vessels with their connections and the dorsal and
ventral sinus system are placed in communication only through
a new development, viz. botryoidal tissue.”

The blood system of the leeches has been studied by a large
number of investigators ever since the time of Cuvier. We
need not give a detailed historical review of the work of the
early authors, of which an excellent account has already been
given by Moquin Tandon (9) and Gratiolet (4) ; it is sufficient
to point out that in 1862 Gratiolet (4) gave an admirable de-
scription of the vascular system of Hirudo, to which little has

1 The fluid contained in the channels is, therefore,a hemolymph in the
true sense of the word, and the combined contractile and non-contractile
systems may justly be called the hemolymph system.

2 The distinction of these spaces by the terms ‘‘vessel”’ and ‘‘sinus”’ is
not a very convenient one, but it is difficult to find a better. The word

channel may be used as an indifferent term. There are some channels, the
vascular or ceelomic nature of which cannot at present be determined.

480 EDWIN S. GOODRICH.

been added since.| What concerns us at present is the
evidence of continuity between the contractile and non-
contractile systems in Hirudo.

The earlier authors all believed in the continuity, and based
their conclusions mainly, if not entirely, on injection. Jaquet
(5), who has recently re-investigated the blood system of
Hirudo by means of injections, confirms this view, describing
small branches communicating from the longitudinal vessels
to the longitudinal sinuses in front, and also from the lateral
vessel to the perinephrostomial sinus. But the evidence
brought forward by these authors is not clear and convincing,
and, in fact, they do not treat the question as one requiring
definite proof.

A. G. Bourne, who also studied sections, agrees with Grati-
olet in almost every particular, and believes that the two
systems communicate by means of their finer branches ending
in capillary networks. But here again we miss the necessary
convincing evidence of the continuity.

Since considerable doubt arose as to the correctness of the
generally accepted view, Mr. A. KE. Shipley in 1888 re-investi-
gated the subject with the help of sections, and wrote a short
paper without figures in which he says (12), ‘A fragment of
the brown tissue of a leech shows at once the connection of the
lumen of the botryoidal tissue with that of the thin-walled
vessels. And my sections through Clepsine and Hirudo show
in numerous places the large openings by means of which the
botryoidal tissue is put into communication with the sinuses.” ?

1 In this work Gratiolet first showed the continuity of the botryoidal
channels with the vascular system (sinus or vessel), though the botryoidal
tissue was first correctly described and so named_ by Lankester, and its con-
tinuity with capillaries and their mode of growth in connection with it
figured (8).

2 A statement which it is difficult to reconcile with that found farther on,
that “certain large corpuscles which occur in the sinuses of Clepsine and
Pontobdella are not found in the blood-vessels, being . . . too large to pass
through the communicating channels.” As far as Clepsine is concerned,
Oka’s results are, of course, directly opposed to those of Bourne and
Shipley.

C@LOM AND VASCULAR SYSTEM IN THE LEECH. 481

The evidence brought forward in this paper is derived from
two sources: firstly, from some injection experiments ; secondly,
from the study of serial sections. The former researches were
carried on last summer, with the help of my friend Mr. L. J.
Picton ; the part dealing with injections must, therefore, be
considered as the joint work of Mr. Picton and myself. For
the remainder of this paper I am alone responsible.

Tue EvipENcE or INJECTIONS.

Two reasons may be assigned why the work of Gratiolet,
Jaquet, and others has not carried conviction. These natu-
ralists, who, as I have already remarked, aimed rather at
tracing out the distribution of the hemolymph channels than
settling the question of the communication between sinus and
vessel, do not seem to have taken any special precautions
against the forcing of the injected fluid through any thin walls
which might be supposed to separate the two systems, nor do
they appear to have been careful as to the state of preservation
of the leeches they injected. Gratiolet, indeed, frequently
macerated his leeches before injecting them, finding the process
easier to accomplish when the animals had begun to decay.

For these reasons we determined to avoid, as far as possible,
all such sources of error.

The leeches, before injection, were not killed, but anzsthe-
tised with a mixture of chloroform, ether, and alcohol; then
spread out on a cork, and opened up so as to expose the lateral
vessel, by means of which the injection is accomplished.
When necessary, the leech was kept in the anesthetised state
for any length of time by placing on its head a pad of cotton
wool dipped in the anesthetic.

A filtered solution of Berlin blue was used for injecting, and
the apparatus consisted of a very fine glass cannula fixed to a
long india-rubber tube leading to a small funnel, into which the
solution was poured. The pressure of the fluid and the flow
from the nozzle of the cannula could be regulated at will by
fixing the funnel at any given height. The pressure used was

482 EDWIN 8. GOODRICH.

so slight, and the tip of the cannula so fine, that the blue fluid
only came out drop by drop from its extremity.

When the cannula is introduced into the lateral vessel, the
blue fluid mixes with the hemolymph and spreads throughout
the system of channels, apparently often as much by the natural
contractions of the vessels as by the pressure exerted from the
cannula,

A large number of such injections were made both in
Hirudo medicinalis and Aulostoma gulo, the operation
lasting from thirty minutes to twenty-four hours. The leeches
were then killed, hardened, and cut.

We found that the injection from the lateral vessel of one
side passed easily into the lateral vessel of the opposite side.
It also very soon reached the dorsal and ventral, and the peri-
nephrostomial sinuses, and the capillary networks of the body-
wall. On the other hand, it seemed to penetrate into the
botryoidal channels only with some difficulty, and in the last
place.

Transverse sections show all these spaces filled with hemo-
lymph tinged with the Berlin blue. The injection seems to
have flowed in quite natural channels, and shows no signs of
having been forced into spaces not belonging to the contractile
and non-contractile systems, or through thin walls of sepa-
ration.

In fact, we fully convinced ourselves that, both in Aulostoma
and in Hirudo, blood-vessels, sinuses, and botryoidal
tissue arein free communication. At the same time we
realised that the evidence of injections alone can never be
placed entirely beyond criticism, and that some other method
would have to be adopted to convince the sceptical, and remove
all possibility of doubt.

Tuer EvipENcE or SEcTIONS.

A careful reconstruction of a series of sections seemed to me
the only way of obtaining the end in view. The method I
adopted was as follows. Having anesthetised the leeches,

CH@LOM AND VASCULAR SYSTEM IN THE LEECH. 483

they were preserved in a mixture of one volume of 4 per cent.
formaldehyde solution, and one volume of a saturated solution
of corrosive sublimate containing 5 per cent. glacial acetic
acid. This fixative gives excellent results for section cutting.
The sections were stained on the slide in a mixture of methy]
blue and eosin, according to a method suggested to me by Dr.
A. Mann which has proved most useful. When successful,
the combination stains the tissues blue or purple, and the
hemoglobinous coagulum, the hemolymph, brilliant scarlet.
This striking contrast enables one to follow out the minutest
capillary with comparative ease.

The work of reconstruction had then to be undertaken.
For the larger vessels and sinuses (figs. 1, 2, 3, &c.) I made
use of a series of 600 transverse sections, 10 u thick, from the
middle region of the body. Camera drawings (x 25) were
made of the first and every tenth section, and these were
plotted out on paper ruled to scale. For this purpose arbitrary
fixed lines had to be adopted; a vertical line was, therefore,
drawn in each case through the nerve-cord and dorsal sinus,
and another horizontal line at right angles to this through the
nerve-cord. The measurements were then taken from these
lines. In this way a certain element of arbitrariness is no
doubt introduced into the reconstruction of the curves of the
vessels ; but this is only very slight, and really of no import-
ance, since, of course, it does not in any way alter the true
relations and communications of the spaces. These were
always verified by examining the intermediate sections.

In reconstructing the systems of smaller vessels, sinuses,
_and capillaries, the difficulties were much greater. Here no
fixed points were available to measure from, no practicable
arbitrary lines could be drawn, owing to the extremely com-
plicated character of the network visible only with a compara-
tively high power, the field of which includes but a small
portion of the section.’

1 Even had a fixed point been available, no purely mechanical process of
reconstruction, such as the wax-plate method, could be trusted, sinee the
capillaries are so numerous and so near to each other that it would be scarcely

484 EDWIN S. GOODRICH.

Camera drawings on a large scale were, therefore, made ;
then combined by means of transparent tracing paper, the
communications of the capillaries being verified in every case
with the greatest care under the high power.

Before describing these important capillary systems, in
which the communication really takes place, it is necessary to
give an account of the large vessels and sinuses, in order to
show that amongst these there is no continuity of the two
systems.

The Lateral Vessels and their Branches.—A large
longitudinal vessel, with muscular contractile walls, extends
in a sinuous course along each side of the body. This is the
well-known lateral vessel, and is said to communicate in
frout aud behind with the corresponding vessel of the opposite
side. In every segment the lateral vessel gives off a pair of
large dorsal branches (figs. 1, 2,6). The first of these is the
short latero-lateral vessel of Gratiolet, passing almost
vertically upwards to break up into smaller branches, and lead
to the superficial cutaneous capillary network on the dorsal
and lateral regions of the body.

The second is the more important latero-dorsal vessel of
Dugés. This vessel soon divides into two large branches, the
anterior of which passes over to the opposite side to join its
fellow above the dorsal sinus, and the other, the posterior, runs
towards the median line, but does not communicate with the
vessels of the opposite side. Both these branches give off smaller
vessels running outwards to the superficial capillary networks.

The anterior branch of the latero-dorsal vessel also gives off
vessels passing downwards to the wall of the gut.

The entrance of the large, and contractile, latero-lateral
and latero-dorsal vessels into the lateral longitudinal vessel is

possible to avoid all sorts of errors on superimposing the camera drawings.
The slightest inaccuracy in such drawings—due, for instance, to the somewhat
different position of the eye in making them, or perhaps to a slight shrinking
or expansion of the section itself—would be sufficient to vitiate the whole
result by leading any given capillary to wrongly fit on to any of its nearest
neighbours.

CHLOM AND VASCULAR SYSTEM IN THE LEECH. 485

constricted, being provided with sphincter muscles (fig. 16),
and on its inner side, within the lateral vessel, is a valvular
arrangement composed of a mass of long-staiked cells situated
round the aperture. Although it is somewhat difficult to
judge certainly as to the action of these valves, yet [ think
there can be little doubt that they prevent the hemolymph
from returning into the dorsal branches when the lateral vessel
contracts. For the bunches of cells would block up the narrow
opening if the fluid tended to return into these branches, as in
the case of similar valves in other annelids. On the other
hand, on the hemolymph flowing into the lateral vessel, the
valvular cells would merely hang freely in its wide lumen.

The only other branches coming from the lateral vessels are
the latero-abdominal vessels of Dugés, arising about
midway between the dorsal branches (figs. 2,3, and 5). They
bend downwards and bifurcate, each branch joining its fellow
from the opposite side below the ventral sinus. A lozenge-
shaped figure is thus formed by the right and left branches,
generally more regular than in fig. 3. The lozenges are
joined together by short median vessels.

The latero-abdominal vessels give off branches supplying
the nephridia, and the capillary cutaneous plexus of the
ventral and ventro-lateral regions.

There are no valves round the aperture of the latero-
abdominal into the lateral vessel, and the entrance is not much
constricted.

This description of the main trunks of the contractile
vascular system agrees in all essential points with that of
Gratiolet (4), excepting for the valves described above, which
apparently were missed by previous observers.

Gratiolet would seem to have been mistaken in thinking
that the anterior branches of the latero-dorsal vessels joined
across to form a complete arch above the dorsal sinus only in
the region of the intestine; this occurs also in the region of
the sacculated crop,! and here, as elsewhere, these arches give

1 It is possible that individual leeches vary in this respect, since Jaquet (5)
also states that there is no union from side to side in the anterior region.

486 EDWIN S. GOODRICH.

off on either side a branch passing downwards to spread over
the wall of the alimentary canal.

The Dorsal Sinus.—Running along close to the wall of
the alimentary canal (figs. 1, 2, 4, and 6) the dorsal sinus
gives off a pair of branches in each segment. One of these
loops underneath the anterior branch of the latero-dorsal
vessel in a peculiar manner (figs. 1 and 6). Both ultimately
break up into small capillaries passing into the cutaneous
plexus of the dorsal and dorso-lateral regions.

Small sinuses also run ventrally from the dorsal sinus to the
wall of the alimentary canal, both to the crop and intestine,
although Gratiolet only found those supplying the latter (figs.
6 and 15).

According to previous observers, the dorsal sinus communi-
cates with the ventral sinus in front and behind.

The Ventral Sinus.—This large sinus, which contains
the nerve-cord, gives off two pairs of lateral branches in each
segment. The most important of these are the short canals
leading on either side into the perinephrostomial sinus,
entering at its anterior end. From the opposite extremity of the
sinus a branch is given off to the nephridium (figs. 2, 3, and 5).

In the region of the nerve ganglion a sinus branches out,
following the posterior lateral nerve for some distance (figs. 3
and 4), This sinus divides into two branches, one going to
the body-wall, in the ventral region, and the other passing
vertically upwards to the dorsal cutaneous plexus (fig. 4).
The latter is the abdomino-dorsal of Dugés.

According to Gratiolet, a similar sinus passes up from the
perinephrostomial sinus; but I have not been able to find it.

A pair of vertical channels in each segment extend from the
ventral cutaneous plexus to the dorsal network.

The chief point to notice about the systems of larger vessels
and sinuses is that the two do not communicate with each
other. It is only by means of the complex capillary systems
that the continuity is established.

The Capillary Systems,—Gratiolet (4) divided the capil-
lary systems into three sections: (1) an inner deep layer, the

C@LOM AND VASCULAR SYSTEM IN THE LEECH. 487

vessels of the botryoidal tissue arising from branches of the
lateral vessels; (2) an intermediate layer, being the capillaries
winding amongst the muscles, derived from the same vessels,
and communicating with the botryoidal vessels ; (3) the.super-
ficial cutaneous layer, divided into a right and left plexus,
communicating with capillaries arising from the intermediate
layer, and also supplied by fine branches from the sinuses.
This last section, the cutaneous plexus, has been shown by
Professor Lankester to extend into the epidermis itself (8).

The results of my observations agree fairly closely with
Gratiolet’s description, the chief differences being with regard
to the supply of the capillary systems.

It has already been mentioned that the latero-lateral and
the latero-dorsal vessels give off small branches passing radi-
ally outwards to the skin. So far as I have seen, these radial
vessels have no direct communication either with the botryoidal
tissue, or with the intermediate layer of capillaries amongst
the muscles, but pass right through these to near the epidermis
(fig. 7). Here they branch, forming what I shall call annular
vessels,! running round a little below the epidermis, on the
one hand towards the median dorsal line, and on the other
towards the ventral surface (figs. 4 and 7).

At short intervals the annular vessels give off small branches
to the delicate epidermal plexus. Although the annular vessels
of the right side never directly join those of the left, two such
vessels from different branches of the latero-dorsal may run
into each other, forming a complete loop (fig. 7).

At the sides these vessels may pass round, describing more
or less complete semicircles, and reaching sometimes to the
ventral surface. Exceptionally, as shown in fig. 4, they may
turn inwards again so as to open into the ventral botryoidal
channels.

Now the superficial epidermal plexus, into which
the small branches of the annular vessels open, is

1 These annular vessels may be the small vessels termed ‘‘ branches verti-
cales superficielles” by Gratiolet, and described as branching at both ends
into the plexus.

488 EDWIN S. GOODRICH.

directly continuous with capillaries coming from the
intermediate intermuscular plexus, and this network
amongst the muscles passes on its inner side into
larger channels, which are, in fact, offshoots from the
lateral branches of the dorsal sinus (figs. 7—11).

The epidermal network may, therefore, be considered as the
region where the sinus system opens into the contractile vas-
cular system. In figs. 8—11 three consecutive sections and
a reconstruction of the same are drawn, showing this connec-
tion in a small portion of the region represented in fig. 7. A
similar continuity can, of course, be found throughout the
other parts.

There can be no doubt whatever that the dorsal sinus com-
municates with the lateral vessels by means of small branches
given off by each system to the superficial plexus. Indeed,
the distribution of the radial capillaries going to the skin
being what it is (figs. 7 and 12), it is evident that sucha
connection must exist; the blood brought by one set of
capillaries must necessarily be carried off by the other. Once
the disposition of these small channels was ascertained, their
continuity was a foregone conclusion. But, of course, it
is satisfactory, and necessary for the sake of dispelling all
doubt, to have the actual evidence of sections before us.

The communication of the contractile system with the ven-
tral sinus must now be established.

A reconstruction of a portion of the capillary system of the
ventral region is given in fig. 12. It will be seen at once that,
although the actual disposition of the capillaries is quite
similar to that described in the dorsal region, yet the conditions
are reversed. For here the annular channel comes directly
from the ventral sinus (ep. figs. 7 and 12), and the radial
capillaries derive from the contractile system, being branches
of the latero-abdominal vessel.

A small portion of the region represented in fig. 12 has been
reconstructed, so as to be shown on a larger scale in fig. 13.
Here the continuity between the two systems is again
proved.

C@LOM AND VASCULAR SYSTEM IN THE LEECH. 489

The annular channels of the dorsal region always derive
from the contractile system. Those of the ventral region
generally belong to the sinus system. The two sets inter-
digitate, an annular sinus passing upwards between two
annular vessels, and they only communicate by means of fine
capillaries of the epidermal plexus. There are a considerable
number of annular channels to each segment.

It has already been mentioned that a dorsal annular vessel
may reach round to the ventral region ; this happeus in the
vicinity of the nephridiopore, and is shown in fig. 14. In
such cases the annular vessel comes into communication with
capillaries from the latero-abdominal vessel, which then bear
just the same relation to it as the capillaries from the dorsal
sinus in the upper region.

These ventral annular vessels of dorsal origin bend inwards
towards the ventral sinus, and pass upwards by the side of the
gut to the dorsal region, where they break up into capillaries
to be distributed to the botryoidal tissue and dorsal epidermal
plexus. They form, in fact, the vertical channels already men-
tioned.

There remain to be described the relations of the botryoidal
tissue. The channels are lined within with the well-known
yellowish-brown cells, the outer region of which is filled with
coarse pigmented granules, and the inner deeply-staining half
formed of comparatively clear protoplasm.!

The botryoidal vessels lie chiefly in a dorsal and ventral
mass on each side, between the alimentary canal and the
muscles of the body-wall in the general parenchyma. On the
outer side the botryoidal vessels communicate with the inter-
mediate capillary plexus of sinus origin (figs. 7 and 8). Occa-
sionally in the ventral region (and perhaps elsewhere) they
open into capillaries of the contractile system (fig. 14). On

1 It is, apparently, this deeply staining region of the cells which has been
mistaken by Graf (8) for an inner coat in Nephelis. There can be no doubt
that in Hirudo, Aulostoma, and Nephelis there is no such inner lining, and that
the previous observers were quite correct in describing the brown cells as
bathed by the hemolymph.

490 EDWIN 8S. GOODRICH.

the inner side of the botryoidal tissue its vessels open here and
there into the small sinuses passing outwards to the interme-
diate layer.

In the ventral region the botryoidal channels also open
directly into the perinephrostomial sinus (fig. 17).

I must finally describe the communications between the
contractile and non-contractile systems which occur in connec-
tion with the vascular supply of the alimentary canal.!

Gratiolet gave a correct description of the vessels of the in-
testine. They are derived from short vertical branches coming
off from the latero-dorsal arches, and passing into a longitudinal
lateral vessel on each side, which is itself connected with a
median ventral channel. From these lateral intestinal vessels
a fine plexus of capillaries extends over the wall of the alimen-
tary canal; joining again into larger trunks, the capillaries
open at intervals into the dorsal sinus (such a connection is
shown in the reconstruction given in fig. 15).

It may be added that a similar communication exists on the
wall of the crop, but the capillary plexus is there less elaborate
(fig. 6).

I have also noticed a peculiar connection between the latero-
abdominal vessel and the abdomino-dorsal sinus by means of a
tortuous capillary shown in fig, 4 at C. This may be excep-
tional.

SUMMARY AND CONCLUSION.

According to the foregoing account, the evidence of carefully
executed injections strongly favours the view that a continuity
exists between the contractile vascular system and the non-
contractile sinus system in Hirudo. This continuity is proved
to exist in various regions of the body by means of serial
sections. The communication takes place through the capillary
systems.

The hemolymph system of Hirudo consists of four main
longitudinal trunks, sending out transverse branches to the

1 T have no doubt a similar continuity exists on the walls of the nephridium,
as mentioned by Gratiolet.

C@HLOM AND VASOULAR SYSTEM IN THE LEECH. 491

body-wall. The dorsal branches of the lateral vessels pass into
small annular vessels communicating with the plexus of minute
capillaries in the epidermis. From these, again, arise capil-
laries going to small sinuses which run into the lateral trans-
verse sinuses, and so into the dorsal sinus.

Similarly the ventral sinus sends annular sinuses along the
ventral region of the body-wall opening into the epidermal
plexus, whence arise capillaries joining the latero-abdominal
vessels.

Continuity between the two systems has also been shown to
take place by means of capillaries on the wall of the alimentary
canal, and probably exists on the other internal organs of the
body.

Two questions still remain to be solved: firstly, as to the
circulation of the hemolymph; secondly, as to the exact
homology of the channels in which it flows.

With respect to the first of these problems, I have no direct
observations to record; but it may be pointed out that the
presence of the valves described above show, at least, that the
hemolymph must flow in a constant direction—that there is a
real circulation, not a mere motion backwards and forwards.
It seems to me extremely probable that the annular vessels
collect the oxygenated blood from the epidermal plexus, and
carry it into the latero-dorsal and latero-lateral vessels, whence
it would be pumped into the lateral vessels. From these some
of the hemolymph must be carried by the latero-abdominal
vessels to the various organs of the body, and to the ventral
cutaneous plexus. ‘The annular sinuses would collect it from
this plexus andcarry it into the ventral sinus. The abdomino-
dorsals and the dorsal sinus would appear to supply the dorsal
and lateral cutaneous plexus.

We are left in considerable uncertainty as to the true nature
of some of the spaces. That the lateral vessels belong to the
real vascular system, and that the ventral sinus and perine-
phrostomial sinuses belong to the true cceelomic system, seems
to be clearly established both by comparative anatomy and by
the embryological researches of Birger (2). This observer,

voL. 42, paART 4,—NEW SERIES. LL

492 EDWIN 8S. GOODRIOH.

however, could not trace the dorsal sinus to a coelomic origin,
and since its branches bear the same relation to the cutaneous
plexus as those of the latero-abdominal vessels, I am inclined to
think that the dorsal sinus may represent the dorsal vessel of
other annelids. In that case the ccelomic cavities do not
persist dorsally, or have never reached the median dorsal
region in the Gnathobdellide.

The annular channels may possibly represent the annular
ceelomic lacunz so well described and figured by Oka in
Clepsine (10), and it may perhaps be through them that the
chief communication between the cclom and the vascular
system has been established. The observation of the some-
what variable relations of these annular channels tends to
support this view.

With the very imperfect knowledge of the development of
the coelom and blood-vessels in Hirudo at our disposal, we
cannot say for certain at present where the one ends and the
other begins, nor whether a given capillary really belongs to
the one or the other. Nor can we safely conjecture how the
continuity has actually taken place. But one thing seems
fairly certain, namely, that it is not only by means of the
botryoidal channels that the communication has been brought
about.

It is very tempting to compare the leech with the Verte-
brate, in which a third system of spaces—the lymphatic
system—has been interpolated, allowing a communication to
take place between the originally distinct coelom and blood-
vascular system.! But the botryoidal tissue is not so inter-

1 The structural analogy between the lymphatic system of the Vertebrate
and the botryoidal tissue of Hirudo is in some respects very close. The former
develops as an independent set of cavities in the mesoblast, which subse-
quently open into the veins and the ccelom (Balfour, ‘ Comparative Embryo-
logy’). The latter, according to Birger (2), develops also as a number of
independent channels, hollowed out in strings of cells of mesoblastic and even
peritoneal origin, which later come into connection with the hemolymph
system. ‘The functions of the lymphatic and botryoidal systems must be
quite different, since the latter is not in any way specially related to the
alimentary canal.

C@LOM AND VASCULAR SYSTEM IN THE LEECH. 493

polated in the case of Hirudo; if it were obliterated, the two
systems would still be in free continuity by means of capil-
laries. The botryoidal channels would seem to be rather of
the nature of a by-path, through which the hemolymph does
not necessarily circulate. In this connection it should be
mentioned that in sections they are rarely seen to be as much
distended with the fluid as the neighbouring capillaries of
similar size.

Whatever may be the process whereby the continuity be-
tween the ccelom and vascular system has been established in
the Gnathobdellide, there can be little doubt that it is a
secondary condition, and that the structure of such a form as
Acanthobdella, in which a closed blood-system lies in a normally
developed ccelom, is really the more primitive.

List oF REFERENCES.

1. Bourne, A. G.—“ Contributions to the Anatomy of the Hirudinea,”
‘Quart. Journ. Mier. Sci.,’ vol. xxiv, n. s., 1884.

2. Bireer, O.—‘‘ Neue Beitrage zur Entwickl. des Hirudineen,” ‘ Zeit. f.
wiss. Zool.,’ vol. lviii, 1894.

3. Grar, A.—‘ Beitrage zur Kenntniss der Exkretionsorgane von Nephilis
vulgaris,” ‘Jen. Zeit. f. Naturw.,’ vol. xxviii, 1894.

4, GratioLet, P.—‘‘ Recherches sur |’organisation du Systéme vasculaire
de la Sangsue médicinale,” ‘Annales des Sci. Nat. Zoologie,’ vol.
xvii, 4th ser., 1862.

5. Jaquet, M.—“ Recherches sur le systéme vasculaire des Annélides,”
‘Mitth. Zool. St. Neapel,’ vol. vi, 1886.

6. Jouansson, L.—‘‘ Ueber den Blutamlauf bei Piscicola und Callobdella,”’
‘Festskrift Lilljeborg,’ Upsala, 1896.

7. Kowatevsky, A.— Etude sur l’anatomie de l’Acanthobdella pele-
dina,” ‘ Bull. Acad. Imp. des Sci. de St. Pétersbourg,’ vol. v, No. 4,
1896.

8. LanKEsTER, HE. Kay.—‘‘On the Connective and Vasifactive Tissues
of the Medicinal Leech,” ‘Quart. Journ. Micr. Sci.,’ vol. xx, n. s.,
1880.

9. Moquin-Tanpon, A.—‘ Monographie de la Famille des Hirudinées,’
Paris, nouv. édit., 1846.

494. EDWIN 8S. GOODRIOH.

10. Oxa, A.—‘ Beitrage zur Anatomie der Clepsine,” ‘ Zeit. f. wiss. Zool.,’
vol. lvii, 1894.

11. SepewicK, A.—‘ Student’s Text-book of Zoology,’ vol. i, London, 1898.

12. Suiptey, A. E.—“On the Existence of Communications between the
Body Cavity and Vascular System,” ‘ Proc. Cambridge Phil. Soc.,’
vol. vi, 1888.

EXPLANATION OF PLATES 42—44,

Illustrating Mr. Edwin S. Goodrich’s paper “‘ On the Com-
munication between the Colom and the Vascular System
in the Leech, Hirudo medicinalis.”

All the figures are of Hirudo medicinalis. The contractile vessels and
their branches are coloured pink, the sinuses and their branches are coloured
blue.

Fries. 1, 2, anp 3.—Reconstructions, from a series of 600 sections, of the
main trunks of the contractile and sinus systems. Figure 1 shows a dorsal
view of the lateral vessels and dorsal sinus. Figure 2 shows a left-side view
of the lateral vessels, dorsal sinus (without its branches), ventral sinus, and
perinephrostomial sinus. Figure 3 shows a more complete view from above
of the ventral sinus and its branches, with the nerve-cord visible by trans-
parency, the lateral vessels with their latero-abdominal branches.

Fic. 4.—Reconstruction as seen in transverse section of sections 50—150,
front view. A capillary, C, is seen to join the right latero-abdominal vessel
with the abdomino-dorsal sinus.

Fic. 5.—Reconstruction as seen in transverse section of sections 213—360,
front view. The position of the gonads is indicated by a dotted line.

Fre. 6.—Portion of section 340 partially reconstructed backwards, showing
the connection, at C, between a small sinus and a branch of the latero-dorsal
vessel. Dotted lines indicate the general course of the latero-dorsal vessel,
which is several times cut through.

Fie. 7.—Reconstruction showing the communication between a branch of
the latero-dorsal vessel, and a branch of the dorsal sinus ; also capillaries
opening from the intermediate plexus into the botryoidal tissue. From another
series.

Fries. 8, 9, 10, and 11.—Camera drawings of three consecutive sections,
and a reconstruction of the same, showing the continuity between the two

CG@LOM AND VASCULAR SYSTEM IN THE LEECH. 495

systems in a small portion of the epidermal network included in Fig. 7 between
two X.

Fic. 12.—Reconstruction showing the communication between a branch of
the ventral sinus and the latero-abdominal vessel.

Fic. 18.—Reconstruction on a larger scale of a small portion of the super-
ficial plexus included in Fig. 12 between two x. The dotted line repre-
sents the basal limit of the epidermis.

Fie. 14.—Reconstruction of a portion of the ventral capillary plexus in the
neighbourhood of the nephridiopore, showing the connection (not followed
out in detail) between the annular vessel of dorsal origin and capillaries
derived from the latero-abdominal vessel; also the communication of the
botryoidal channels with the intermediate plexus,

Fic. 15.—Reconstruction showing the communication between the two
systems on the wall of the intestine, the position of which is represented by a
dotted line.

Fie. 16.—Section showing the opening of the latero-dorsal into the right
lateral vessel (445th section of the first series).

Fic. 17.—Section showing a botryoidal channel opening into the perine-
phrostomial chamber.

Fie. 18.—Section showing the opening of a small sinus into the botryoidal
channel (one of those represented in Fig. 7).

7 7
7
-
i>
=
=
‘ohn
is :
2
we
.
}
A
4
«

aay
_

i yy - a Li : Z ’
Fe a2 = =
a a oC.
zs n a _
te a - a Dt
Vi-> ae
— -
om a 7
- =e. 7 o
se a ; -
ae “? 7 _ -
: wn 7 -
- ae ‘ :
te M wir
ioe 3 f ATM Mee
es 7
=
ral, Oe WY Breed |i ell
\ ae? £- € 7
ds iA eS yy Dw?’ eis a “
*% te i eke
= 7 %
4 ss Perea Y fhe vier ee .
a: q = b white ei) ¥
. 1
—s > Lilie fay
od ‘ iL
: TP oar
Ee
‘
s pt it |
ti 4
:
{ -
=
: -
i
5 -
Aa Se 1 >
. a
pa ll

BALANOGLOSSUS OTAGOENSIS. 497

Balanoglossus otagoensis, n. sp.
By

W. Blaxland Benham, D.Sc., M.A.,
Professor of Biology in the University of Otago, New Zealand.

With Plate 45.

HirHerto no example of the Hemichorda has been re-
corded from New Zealand; and, indeed, it is only quite
recently that any member of the group has been demon-
strated to occur on the shores of Australasia. In 1898,
Mr. T. P. Hill,’ of Sydney University, recorded the capture of
a specimen of Ptychodera Australiensis, of which he gave
a detailed account in 1895.. But the genus Balanoglossus
has hitherto not been discovered in the Southern Hemisphere
(fide Spengel, 1893), hence the species about to be described
gains in interest.

On February 16th of the present year, Mr. Hamilton, the
Registrar of the University, accompanied me on a short collect-
ing expedition to the Otago Harbour at Port Chalmers; we
were examining, in a boat, seaweeds, &c., from a depth of
about one fathom, close inshore, which here descends abruptly
into the sea to about this depth ; while overhauling a piece of
great kelp, Macrocystis, we observed a bright red worm
creeping along the stem. At first I thought it a Nemertine, till 1
had placed it in a bottle of water, when I was delighted to find
it tobe an Enteropneust. A few minutes afterwards we obtained
a second specimen, smaller, and of an orange tint, which

1 «Proc. Linn. Soc., N.S.W.,’ 1895.

498 W. BLAXLAND BENHAM,

turned out to be a quite young individual of the same species.
A few days afterwards I returned to the spot, but was unsuc-
cessful in repeating the capture.

The following account is founded on an examination of the
larger individual, which, after examining it alive and making
a sketch of it in natural colours, I killed and stained. I then
examined it in oil of cloves, and finally cut it into a series of
transverse sections.

The animal measured during life, when moderately extended,
30 mm. from the tip of the proboscis to the anus, its greatest
diameter at the collar 2:4 mm.; the proboscis was 9°5 mm., about
one third the whole length (fig. 1). The colour is rich carmine-
red, the proboscis and collar being deepest in tint, the colour
fading gradually towards the hinder end; in the last half of
the body the thin body-wall allows the brown intestine to show
through. The hinder part of the body is flecked with white
(? glands), visible through a hand lens. The extreme tip of
the proboscis is brown, which passes gradually and imper-
ceptibly into the red.

The proboscis is the most remarkable feature in this species,
for it is traversed throughout its length by a deep groove along
its dorsal surface (figs. 1, 2, a, and figs. 5,6, 7,G). It was only
after I had killed the worm that I became aware of the interest
attaching to this groove, and I regret that I did not examine it in
greater detail during life. I am not quite certain whether it is
a permanent structure, or whether it can be flattened out
during the movement of the animal. My notes read as
follows :—‘* The animal was freely creeping on the weed, .. .
creeping by the use of its very contractile proboscis, which
becomes deeply grooved along one surface.’” However, this
fact can easily be ascertained on the next occasion on which
the worm is captured.

The worm, at any rate, appeared to use its proboscis as a
temporary organ of fixation during progression, for it attached

1 The measurements of the specimen when killed are as follows :—Proboscis
6:25 mm., branchial region and collar 2°4 mm., genital region 3 mm., total
length of worm 20 mm.

BALANOGLOSSUS OTAGOENSIS. 499

the tip of it to the surface of the weed or bottle, and drew its
body after it, with a good deal of wriggling ; probably this act
of fixation was effected by the abundant sticky secretion which
is discharged by the epidermis.

The collar is moderately long and is triangulate, being marked
by two transverse furrows, of which the second is the deeper.

The branchial region is relatively short: I counted twelve
double gill-slits on each side of the pharynx, i.e. there are
twelve pairs of branchial apertures opening into the branchial
grooves (fig. 1). Iwas unable to see these apertures in life, but
I counted the gills when the animal had been stained and
cleared (fig. 2) ; the gill-bars are without synapticula (fig. 8).
I further confirmed this observation by counting the gill-pores
in this series of sections.

The post-branchial region is not “ winged;” the dorsal
surface is slightly depressed along the branchial and _post-
branchial regions, and the gonads lie, as usual, in the lateral
ridges. Of these there are about sixteen on each side, in this
case ovaries. .

There are no hepatic diverticula; the post-genital region of
the body is cylindrical, and exhibits a narrow, but deep,
ventral furrow for some distance (fig. 11).

In the genital region the intestine is narrow and its wall is
a good deal folded, but posteriorly it widens out and comes to
fill the body-cavity (fig. 12).

The internal anatomy, as derived from study of the sections,
agrees in all points with those on which Spengel! lays stress
in characterising the genus Balanoglossus, viz.—

The absence of circular muscles in the trunk ; the absence of
synapticula in the gill-bars; the length of the divergent limbs
of the subnotochordal skeleton, which in the present species
extend backwards to the hinder end of the collar, being cut
through in the sections that include the collar-pore (fig. 4, ¥F).
The longitudinal muscles of the collar region have a fan-
shaped arrangement around the end of this limb (fig. 4, @).

1 Naples monograph, 1898.

500 W. BLAXLAND BENHAM.

So much for its generic characters. There is no doubt that
it belongs to the genus Balanoglossus. With regard to
specific characters, it belongs to the same section of the genus
as B. Kowalevskii and B. Mereschkovskii, as is shown
by the following features:—The great relative length of the
proboscis ; indeed, this length is greater in B. otagoensis than
in either of these; there is only a single proboscis pore; there
are no median gonads. Further, it possesses paired intestino-
tegumentary canals (Darmpforte of Spengel), as do these two
species ; but whilst in them there is at least six pairs of these
peculiar structures, there is but a single pair in B. otago-
ensis (as in the genus Schizocardium). It may not be
quite safe, perhaps, to place much reliance on this point from
an examination of so few specimens, since Spengel states that
the number is not constant in B. Kowalevskil, and suggests
that it increases with age, as do the gills and gonads. Never-
theless it seems to me probable that a single pair is constant
in B. otagoensis from the following fact: not only is there
only a single pair observable in the sections of the adult
animal, but in the smaller younger specimens, in which no
gonads are as yet developed, the canals are already present,
although the gill-slits have not reached more than half their
total number; the canals, then, appear early in life.

Another distinction of specific importance is noticeable in
the sections, in regard to the longitudinal muscles of the pro-
boscis. In both the species referred to above, Spengel de-
scribes the muscles as being arranged in several concentric
layers, encroaching considerably on the connective tissue
which more or less fills the proboscis. In B. otagoensis,
however, the longitudinal muscles of this organ are confined
to a very narrow band (fig. 5, D), close to the circular muscles,
which, as in Balanoglossus generally, are very feebly de-
veloped.

From these facts there is no doubt but that the New Zea-
land species is quite distinct from the American and European
species. But there is a Japanese species, briefly described by
Spengel, which agrees with B. otagoensis in the most

BALANOGLOSSUS OTAGOENSIS. 501

noticeable external character, namely, in having a grooved
proboscis. Spengel had portions of three specimens of this
species, B. sulcatus, Spengel, which he was unable to further
investigate owing to an accident that happened to them. He
gives a drawing, however, on page 347, from which it will be
seen that it bears a considerable resemblance to the New
Zealand species. All that is known of B. sulcatus is as
follows:—It has a groove along the proboscis; it has no
synapticula ; it appears, from the drawing, to possess ten or
eleven pairs of gill-slits. This drawing is about eight times
natural (preserved ?) size, from which we may conclude that
the proboscis is about 8 mm. Jong and the width of the collar
15 mm., both measurements agreeing fairly well with the
proportions of these organs in B. otagoensis.

It is, from consideration of geographical distribution, possible
that they are identical '—there is nothing to oppose identity in
this respect. Nevertheless, as we know practically nothing of
the anatomy of B. sulcatus, I give a new name to the New
Zealand species, which may be characterised as follows :

Balanoglossus otagoensis, n. sp.

Hab., coast of Otago, New Zealand.

The proboscis is deeply grooved along the whole of its dorsal
surface. ‘The proboscis cavity extends right to the tip of the
organ. The longitudinal muscles of the proboscis form a very
narrow band close to the wall. There is but a single proboscis
pore. There are no median gonads. There is a single pair of
intestino-tegumentary canals. The arms of the subnoto-
chordal skeleton reach backwards to the level of the collar-
pore.

I have considered it unnecessary to enter into details of
anatomy or histology, for these are to be found in Spengel’s

1 T hear from Prof. Dendy, of Christchurch, N.Z., that he has captured a
species of Nchiurus on the coast, which agrees very closely with the peculiar
Japanese E. unicinctus.

502 W. BLAXLAND BENHAM.

great monograph; but there is one small point to which I
would draw attention, as I do not find anything quite like it
recorded by Spengel. It has reference to the relation of the
cardiac vesicle (Herzblase, Spengel: ‘‘ sac of proboscis gland ”
of Bateson) and central sinus (Bateson’s “ heart”’).

In the anterior region of the ‘‘ basal complex” of organs, the
condition of affairs in B. otagoensis is quite in agreement
with that described for other species; the central blood sinus
projects upwards into the cardiac vesicles, so that the cavity of
the latter is more or less crescentic in section. The greater
part of the sinus appears, in section, as a subcircular or semi-
circular space filled with blood; but it is prolonged right and
left into an arm, which passes upwards outside the cardiac
vesicle, and downwards around the notochord. This arm
gives off a number of blind diverticula, arranged one above
the other (fig. 9). The whole series of outgrowths is covered
by a layer of cells, the nuclei of which take the stain deeply.
It constitutes Spengel’s “ glomerulus” (the ‘proboscis gland”’
of Bateson). Such is the normal arrangement, and such it
is in the anterior part of B. otagoensis; but further back the
relative sizes of the parts undergo a peculiar change—the
central sinus becomes greatly dilated, bulging upwards more
aud more into the cardiac vesicle, which it almost entirely
fills, so that its cavity is reduced to a very narrow cleft
(fig. 10). Further back still, the usual condition is again
assumed till the central sinus disappears as such.

Looking through the ‘monograph,’ I have been unable to
see any account of this condition for Balanoglossus; but in
Ptychodera minuta it occurs, and is figured on pl. 111,
fig. 18.

I do not know that any great importance is to be attached
to this greatly dilated condition of the central sinus; it may be
that the worm happened to be killed during a local contrac-
tion which drove the contents of the sinus into this temporarily
dilated region, but I did not observe any corresponding restric-
tion of its dimensions, and since the condition of these parts
has been dealt with by Spengel at length, and some stress is

BALANOGLOSSUS OTAGOENSIS. 503

laid on the variations presented by the different genera, it
seems worth while to note the peculiarity in the present
species.

EXPLANATION OF PLATE 45,

Illustrating Mr. Blaxland Benham’s paper on “ Balano-
glossus otagoensis, n. sp.”

Fie, 1.—Dorsal view of Balanoglossus otagoensis (x 8) from a
sketch of the living animal; the gill-pores have, however, been inserted from
later study. (a) The characteristic groove on the proboscis.

Fig. 2.—View of the left side of the specimen after staining and clearing.
_ The depth of the groove (A) is indicated. (B) The “cardiac vesicle.” (c)
Branchial region. (D) Genital region.

Fig. 3.—Sketch of gill-bars.

Fie. 4.—Half a transverse section through the region of the collar-pore, to
show the end of the subnotochordal skeleton. x 125. (a) Dorsal nerve-cord.
(8) Dorsal blood-vessel. (c) First gill-sac. (p) Collar funnel and pore opening
into it. (D) Parts of gill-bar. (#) Pharynx. (8) Limb of subnotochordal
skeleton. (G) Longitudinal muscles. (H) Collar cavity (camera).

Fig. 5.—A transverse section of the proboscis near the anterior end (stained
with picro-carmine). X 125 (camera outline). (A) Hpidermis. (B) Nerve tissue.
(c) Basement tissue. (D) Longitudinal muscles, confined to a very narrow
area. (&) Connective tissue. (F) Cavity of proboscis. (G) Dorsal groove.

Fie. 6.—Outline of transverse section of proboscis (camera) near the middle.
xX 125.

Fic. 7.—Transverse section of proboscis near the base (camera). x 125.
(a), (c), (D), (£), (F), as before. (G) Central sinus. (H) Cardiac vesicle.
(1) Glomerulus. (s) Notochord.

Fic. 8.—A portion of the section of the wall of the proboscis in the region
of the square on Fig. 5. x 450 (drawn with camera). (A) Epidermis. (B)
nerve layer. (c) Basement tissue and circular muscle. (p) Longitudinal
muscle-fibres. (£) Connective tissue with nuclei.

Fics. 9, 10.—Two transverse sections of the “ basal complex ” to illustrate
modifications in form of cardiac vesicle. (G), (H), (1), (J), as in Fig. 8 (both
drawn with camera). x 150.

504 W. BLAXLAND BENHAM.

Fie. 11.—A diagrammatic transverse section through the genital region
in the region of the single pair of ‘intestino-tegumentary canals” (B)
(compiled for a series of outline drawings). (A) Intestine. (B) ‘ Canal.”
(c) Its pore. (D) Gonad. (£) Ventral blood-vessel. (F) Ventral nerve.
(G) Ventral groove. (H) Dorsal nerve. (1) Dorsal blood-vessel.

Fie. 12.—Sketch of anterior part of the body of the small specimen, stained
and mounted entire, to show the position of the single pair of intestino-
tegumentary canals (F). (a) Collar. (B) Branchial region. (c), (D) Intestine
in genital region. (£) Post-genital intestine.

THE MOVEMENTS OF COPEPODA. 505

The Movements of Copepoda.
By

E. W. MacBride,
Professor of Zoology, McGill University, Montreal.

Stncez the time of Brady it has been generally supposed that
amongst the Copepoda the most important locomotor organs
were the antennules. This belief is categorically stated in
Huxley’s text-book on ‘The Anatomy of Invertebrated
Animals,’ p. 235.

During a stay of several weeks in the Plymouth Biological
Station I had almost daily opportunities of examining the
numerous Copepoda captured in the tow-net, and my observa-
tions are by no means consonant with the popularly accepted
idea. The movements of the species I examined were of two
kinds; there was a slow gliding movement, and a sudden dart
of lightning swiftness. During the prosecution of movements
of the first description the antennules, or first antenne, are held
rigidly extended at right angles to the long axis of the body,
and their appearance suggests the idea that one of their func-
tions may be to act as hydrostatic sense-organs. Movement is
effected principally by means of the second antenna, the
gnathites likewise assisting, notably the second maxilla. It
seems probable that feeding is carried on during these slow
movements.

The quick movements are effected, on the other hand, entirely
by the simultaneous action of the thoracic feet. A sudden
blow executed by all the four powerful pairs of paddles is suffi-
cient to propel the animal for a very considerable distance.
The animal moves so quickly during the longer darts that it is

506 BE. W. MAOBRIDE.

impossible to see exactly what happens to the antennules, but
by carefully examining the shorter darts, which are carried out
at a more moderate speed, it is seen that the antennules are
held as rigidly as during the slow movement, and there is
therefore no ground for attributing any share in the produc-
tion of the movements of these animals to the first antenne.
Naturally when the animal is suddenly propelled forward the
tips of these appendages will be mechanically dragged back by
the resistance of the water; and a careless observation of this
phenomenon, joined to the undoubted fact that in the fresh-
water Cyclops the first antenne do assist in the slow move-
ments of the animal, may have given rise to the belief that it
was the rule among Copepoda to propel themselves by means
of the first antenna.

In a paper published some years ago (Sedgwick’s theory of
the embryonic phase of ontogeny as an aid to phylogenetic
theory, ‘ Quart. Journ. Micr. Sci.,’ 1895) I put forward the view
that the progressive development of the Crustacea was corre-
lated with a passing of the function of locomotion backwards
along the series of appendages. Thus, in the ancestor repre-
sented by the Nauplius, the first, and more especially the second
antennz were the main locomotor organs; in the stagejrepre-
sented by the Zooea the maxillipeds had acquired the function ;
whilst in the ancestral condition corresponding to the Mysis
larva, motion was effected chiefly by the hinder thoracic legs,
as is still the case with Schizopoda; finally, the lower Macrura
swim by means of the abdominal appendages alone. The result
of this process has been that the appendages of the anterior
segments have been one by one relieved of the function of loco-
motion, and have become specialised for masticatory and sensory
purposes. Now in the Nauplius the main brunt of the work of
locomotion is borne by the second antenna ; the first is already
mainly a sensory organ, and if the Copepoda really did propel
themselves chiefly by the first antennz they would have retro-
graded from the condition represented by the Nauplius. It is
interesting to note that not only is this not the case, but that the
second antenna is still an important locomotor organ. Func-

THE MOVEMENTS OF COPEHPODA. 507

tionally, indeed, Copepoda seem to stand on pretty much the
same plane of development as the Protozocea larva.

Phylogenetic hypotheses have too often been based on mere
resemblances in form, apart from a consideration of function.
This seems to me to be a wrong methodof attacking the problem.
Function is the all-important thing—that which determines
structure; and I hold that if ever we are able to sift the primary
from the secondary elemeuts in ontogeny, it will be by the recog-
nition of the fact that the persistence of ancestral structure is
caused by the retention of ancestral habits, and that the habits
at all periods of the life-history demand the closest study.

Montreal; Oct. 15th, 1898.

VOL, 42, PART 4.—NEW SERIES. MM

od i ay Pees
E30 T Sts er 7

innae ait nae
. Ma
ag >. nen Oe la ied
: mwa 4 :
y vi, 73 ar 6 4 ts at
- nfs Sahih g
* uh ; ; | |
fs A
i « art 7¢ we :
‘ ; a Crees
7 , -
:
if i. ia
: | is bend i? Si4ae
| ms
ier oe
| ’
"
= |
| =
>
‘
.
=
i . |
‘
’

a

y ae me -
Leifegrgay a

er comin tear bia
+ at) oF
ee ov) i= be is |

Fs Tied

.
=
€ Ly =

ENDER X PO] NOt AZ,

NEW SERIES.

Arhynchus hemignathi, note on,
by Shipley, 361

Ashworth on Xenia and Heteroxenia, |
| Goodrich on the ceelom and vascular

245

Balanoglossus otagoensis, by
Benham, 497

Batrachians of the Paraguayan
Chaco, by Budgett, 305

Benham on Balanoglossus ota- |

goensis, 497

Budgett on Batrachians of the Para- |

guayan Chaco, 305

Chlamydomyxa, by Hieronymus, 89
Copepoda, movements of, by
MacBride, 505

Dendy, development of Sphenodon,
1

Dendy on the parietal eye of Sphe-
nodon, 111

Drosera, cytological changes in, by
Huie, 203

Echinoids, development of, by
MacBride, 335
Enteropneusta,

on, by Willey, 223
Evans on the larva of Spongilla
lacustris, 363

e |
fresh observations |

Hye, parietal, of Sphenodon, by Dendy,
lll

system of Hirudo medicinalis,

477

Hieronymus on Chlamydomyxa, 89

Hirudo, ccelom and vascular system
of, by Goodrich, 477

Huie on cytological changes in Dro-
sera, 203

Hydroids from Wood’s Holl, Mass.,
by L. Murbach, 341

Leech, cceelom and vascular system
of, by Goodrich, 477

MacBride on the development of
Echinoids, 335

MacBride on the movements of
Copepoda, 505

Molluses of the Great African Lakes,
by Moore, 155 —

Moore on the Molluses of the Great
African Lakes, 155, 187

Murbach on Hydroids from Wood’s
Holl, Mass., 341

Parietal eye of Sphenodon, by Dendy,
111

510

Protochorda, recent work on, by
Willey, 223

Shipley, note on
hemignathi, 361

Arhynchus

Sphenodon, development of, by Arthur |

Dendy,-1

|
|

Sphenodon, parietal eye of, by Dendy, —

111

INDEX.

| Spongilla lacustris, the larva of,

by Evans, 363 .

Tanganyika and Spekia, by Moore,
155

Willey on Protochorda and Entero-
pneusta, 223

Xenia and Heteroxenia, by Ash-
worth, 245

PRINTED BY ADLARD AND SON,
BARTHOLOMEW CLOSE, E.C., AND 20 HANOVER SQUARE, W.

; bi aay
; Lo ok ye A vay W fi
nad

TRAY

eR ag

Mi

nies r
bit iy

Ly
| a )
AUN

one
f] rae
ATA tan
f une

ih ¥
uy ae ;

Heit a
ta
FA Mi

ne,

ROR ea tah i p
Deut iipauih aks Aw
Uy ed Saye Ut any
‘ yeas hy

EC Ran

Ce
A # val
ava } ‘

bia) ae

ee

re
Bey Leta)
y

AN Na

) ip ;
Wi
el
,

eee oe

pape

sida salto

—

erect: foe muee