Fa eS vi stad, Sh fated tod. 003-¥Suded
en : wg th grees teh -
PoRADAPal aineWRbanb hss Lace bbe Pmenee el ee eee Hh TERS TP OTRO

(QUA G 204

[#49
r +21
HARVARD UNIVERSITY.

Lb rR Aus

OF THE

MUSEUM OF COMPARATIVE ZOOLOGY.

GIFT OF

ALEX. AGASSIZ.

y,
:
ih

‘ar

yin ?

y i ver mh si hid \

a | vane! ua <

re

vee si ‘iy
a A at

ek

Se al

’ ‘i, ‘
} 1 ‘ i 7 ay
¥ ahs Wey rhe ine ie SY
ne ‘ i* ’ } (esa x at tgs Pris
yy ’ j Ri ann iti oot Ma)
¢', (@ i ued fig ra iA Mah “a ae
aes § Fe wh ¥Y ry o“ fi a i
Pa AN Peete 7, Perey: t
‘ Poa. dob TA tigen t (a! Pace
hy Poe Pee CAV PAE as
ecw, PO RR ve Gs aL
Pith pat THR, ae Sind av ee Cee
av ie y
wv iy AP yn Oa vil: a : Less
seuilys idl alietabe 1+ a haba: nh ety tt Oi? Calais ya aly ai “
Mite - 7 “a ‘ek ey tes A a ie UP) “4
CA A) ae Re
ee een

i

he f i;

wir

hy
a A.
Ani
i) og a 5)
' Py v i
7 ; Le ty 1
i We
A
‘2 i, wy
vis ; ' “ } , i
i, o is
ia), tL ash i
ay) ¥) Hi or tr ais ht ey
‘Ae - } ry : A Ta 7
wa ah r ; in 7) Wie ay
Ma, , / yh ag a Vy A be a
i, i. 7 ot 7 A es ; i ”
YY ee teeny ey Oe) ery

S429
j~2/

QUARTERLY JOURNAL

OF

MICROSCOPICAL SCIENCE.

EDITED BY

E. RAY LANKESTER, M.A., LL.D., F.B.S.,
Linacre Professor of Comparative Anatomy, Fellow of Merton College, and Honorary
Fellow of Exeter College, Oxford; Corresponding Member 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 ; Associate Member of the Biological Society
of Paris; Fullerian Professor of Physiology
in the Royal Institution of London.

WITH THE CO-OPERATION OF

ADAM SEDGWICK, M.A., FE.RS.,
Fellow and Tutor of Trinity College, Cambridge ;

AND

W. F. R. WELDON, M.A., F.R.S.,
Jodrell Professor of Zoology and Comparative Anatomy in University College, London;
late Fellow of St. John’s College, Cambridge.

VOLUME 40.—New Srnrizs.
With Aithographic Plates and Engrabings on Wood,

LONDON:
J. & A. CHURCHILL, 7, GREAT MARLBOROUGH STREET.
1898.

CONTENTS.

CONTENTS OF No. 157, N.S., APRIL, 1897.

MEMOIRS :

The Constitution and Development of the Society of Termites:
Observations on their Habits; with Appendices on the Para-
sitic Protozoa of Termitide, and on the Embiide (concluded
from Vol. 39, p. 315). By Professor B. Grassi in collaboration
with Dr. A. Sanpias

On the Structure of Hydractinia echinata. By Marcargt
C. CoticuTt, Zoological aumee University College, London.
(With Plate 1) . = - : . -

On the Histology of the Ovary and of the Ovarian Ova in certain
Marine Fishes. By J. T. Cunnincuam, M.A.Oxon. (With
Plates 2—4) : - : :

On Ptychodera flava, acheohelts By ArtHuR WILLEY,
D.Sc. (With Plate 5) d : .

On the Nepliridia of the Polycheta. Part I1—On Hesione,
Tyrrhena, and Nephthys. By Epwin 8. Goopricn, B.A., As-
sistant to the Linacre Professor, Oxford. (With Plates 6—9) .

The Pre-ocular and Post-ocular Tentacles and Osphradia of Nau-
tilus. By ArtHuR WittEy, D.Sc. (With Plate 10)

On Heteroplana, a New Genus of Planarians. By AnrtHuUR
Witey, D.Sc.

CONTENTS OF No. 158, N.S., AUGUST, 1897.

MEMOIRS:

The Adhesive Tentacles of Nautilus, with some Notes on its
Pericardium and Spermatophores. By ArntHuR WitteEy, D.Sc.
(With Plate 11).

PAGE

101

165

185

197

203

207

iv CONTENTS.

On some Modifications of Structure subservient to Respiration in
Decapod Crustacea which burrow in Sand; with some Remarks
on the Utility of Specific Characters in the Genus Calappa, and
the description of a New Species of Albunea. By WattEer
Garstanc, M.A., Fellow and Lecturer of Lincoln College,
Oxford. (With Plates 12—14).

Notes on the Anatomy of Sternaspis. By Epwin 8. Goopricn,
B.A., Assistant to the Linacre Professor, Oxford. eS Plates
15 and 16)

On the Relation of the heer Head | to the Nene Pee!
mium. By Epwin 8. Goopricu, B.A., Assistant to the Linacre
Professor, Oxford

On the Development of the Californian Hag-fish, Biaiieaeie
Stouti, Lockington. By Basurorp Dean, Ph.D., Columbia
College, N.Y. (Preliminary Note.) (With Plate 17) .

On the Diplochorda.—1l. The Structure of Actinotrocha. 2. The
Structure of Cephalodiscus. By A. T. MastEermay, B.A.,
B.Sc., Lecturer and Assistant Professor of Natural History in
the University of St. Andrews, N.B. (With Plates 18—26)

CONTENTS OF No. 159, N.S., DECEMBER, 1897.

MEMOIRS :

Note on a New British Echiuroid Gephyrean, with Remarks on the
Genera Thalassema and Hamingia, By W. A. Herpmay,
D.Se., F.R.S., Professor of Natural History in University College,
Liverpool. (With Plates 27 and 28) : ‘

The Placentation of Perameles. (Contributions to the Embryology
of the Marsupialia—I). By Jas. P. Hitz, Demonstrator of
Biology in the University of re N.S.W. (With Plates
29—33) . . : :

On the Green Pigment of the Intestinal Wall of the Annelid
Chetopterus. By E. Ray Lanxzster, M.A., LL.D., F.RS.,
Linacre Professor and Fellow of Merton Bie iae Oxford. (With
Plates 34—37) d é ;

PAGE

211

233

247

269

281

367

385

447

CONTENTS.

CONTENTS OF No. 160, N.S., JANUARY, 1898.

MEMOIRS :

Materials for a Monograph of the Ascons. I. On the Origin and
Growth of the Triradiate and Quadriradiate Spicules in the
Family Clathrinide. By E. A. Miycuty, M.A., Fellow of
Merton College, Oxford. (With Plate 388—42)

The Early Development of Amphioxus. By E. W. MacBrinz,
M.A., Fellow of St. John’s College, Cambridge; Professor of
Baeloay in the McGill University, Montreal. (With Plates
43—45) . : ;

On Drepanidotenia hemignathi, a New Species of Tapeworm.
By Artuur K. Surptey, Fellow and Tutor of Christ’s College,
Cambridge, and University Lecturer in the Advanced Morpho-
logy of the Invertebrata. (With Plate 46). .

Spengelia; a New Genus of Enteropneusta. By AntHUR WILLEY,
D.Sc., Balfour Student of the University of eek (With
Plate 47)

On a Prorhynchid Turbellarian from Deep Wells in New Zealand.
By Witttam A. Haswett, M.A., D.Sc., F.R.S., Professor of
Biology, Sydney University. (With Plate 48)

Note on the Development of the Atrial Chamber in Amphioxus.
By E. Ray Lanxester, M.A., LL.D., F.R.S.

Titte, INDEX, AND CoNTENTS.

PAGE

469

589

613
623

631

647

” \ ’ se
wt ¢
«
— - , *
7 * . z . r Sf. “ 7
A .
7 t a *
“ \ og * &
ad Pe ' “ ’
. ax ‘ - . : <*
} > ad : = . yop — Ny F
" - / \
: ’ 7
- ; 2
1 bs * hg
) > + + | ad \ a
a * i - if <Z
. s
<a . r - in ‘i

MAY 22 1897

The Constitution and Development of the Society
of Termites: Observations on their Habits;
with Appendices on the Parasitic Protozoa
of Termitide, and on the Embiide.

(Concluded from Vol. 39, p. 315.)
By

Professor B. Grassi in collaboration with Dr. A. Sandias.!
With Plates 16—20 in Vol. 39.

The Mode of Development of the Castes in Termes and
Calotermes.

I. DEVELOPMENT OF THE SUBSTITUTE OR COMPLEMENTARY
Royat Forms.

A. Experiments and Observations on Calotermes
under Natural Conditions.

1. Tue royal pairs of about eighty nests were removed at
different times from January to July.” After the following
October each nest was found to contain a substitute pair, i.e.

1 [By a misunderstanding, for which the translator is not accountable, the
paragraphs containing the observations of Dr. Sandias referred to on p. 258
were enclosed in Part I of this paper between square brackets. All such
paragraphs occurring in the text of that part should be credited to him; in
the following pages they are distinguished by being placed between asterisks.

On the other hand, all such footnotes of both parts, and such portions of
the text of the present instalment, as are contained in square brackets
have been added by the translator, some notes not contained in the original
having been supplied for this translation by Professor Grassii—W. F. H. B.]

2 All these nests were left with a more or less numerous population of
soldiers, larve of different ages, and nymphs.

vou. 40, paRT 1,—NEW SER, A

2 B. GRASSI AND A. SANDIAS.

forms of a golden-yellow colour, with the compound eyes almost
always pigmented, with or without wing-buds and with the
abdomen dilated, particularly in the female, which as a rule
did not possess the genital appendices. In the majority of
examples the wing-rudiments were absent or very short, but
exceptionally they approximated in length to those of a nymph.
Sometimes one member of the royal pair did, and the other
did not possess wing-buds.

The nests not infrequently contained eggs, but never a true
royal pair.

2. Hither the king or queen alone was removed in the
spring from about a dozen nests. After October each nest
contained a yellow, i.e. substitute king, associated with a
black queen if the king had been taken away ; if the queen, a
black king with a yellow queen. Therefore the true royal
form which had been removed was still wanting.

Corollary.—After loss of the royal forms the orphaned
nest furnishes a pair of substitute forms; if one only is
missing, it is replaced by a substitute of the same sex, and in
such cases a black example mates with a yellow.

Individuals capable of reaching sexual maturity are selected
for development into substitute forms, but the most advanced,
the nymphs, are not preferred.

3. Many nests orphaned in July, August, and September
provided themselves with a substitute royal pair, notwithstand-
ing that perfect examples, that is, black and with fully formed
wings, were abundant in those months.

Corollary.—Orphaned nests do not have recourse to perfect
insects in order to provide themselves with a king or queen.
There is no justification for the belief that nests orphaned at
the swarm-period adopt a pair of perfect insects, because
thickly populated nests are never found naturally to contain a
king and queen of small size.

4. Certain nests orphaned in March and April and reopened
in June and July contained a limited number only (six or
eight to twenty) of royal substitutes, which were still sexually
immature.

CONSTITUTION AND DEVELOPMENT OF TERMITES, 3

5. Five orphaned nests were re-examined after not less than
two years’ interval. They possessed substitute pairs of the
usual type, except that the yellow colour was deeper, and the
abdomen, especially of the queens, more distended. Several
of these royal forms had no sign of wings.

Others were opened after three or four years, and contained
substitute forms precisely similar, save for a more marked
enlargement of the abdomen, just as in true kings and queens
of the same age.

The progeny in all the above-mentioned nests consisted as
usual of soldiers, nymphs, &c.

Corollaries.—There is no subsequent growth of the wings
in the royal substitutes.

Their progeny is identical with that of the perfect royal forms.

6. Two nests orphaned a couple of years previously contained
no royal pair, but merely some five or six examples in process
of becoming substitutes. Neither eggs nor new-born larvze
were present, but the rest of the brood was normal. This
was observed in April.

These nests had clearly been orphaned a second time after
having once possessed a substitute royal pair.

7. The substitute pairs were removed from several nests
which were furnished with them ; the colonies provided others.

8. It is not uncommon in nature to find substitute pairs in
trees which have been felled or broken by a gale or other
accident. In such cases the true royal pair was evidently left
behind in the severed portion, and the nests were consequently
left orphaned.

Corollary.—Loss of the royal couple is a frequent natural
phenomenon.

B. Experiments and Observations on Termes under
Natural Conditions.

1. I have never found a single nest containing a true royal
pair, though on one occasion such a pair was discovered
under natural conditions, but without offspring. By keeping
twenty or thirty winged examples of both sexes in a glass jar

4 B. GRASSI AND A. SANDIAS.

filled with rotten wood, I have often got a pair to lay eggs, and
once I obtained larvee in various stages of growth.

Lespés claims often to have found true royal pairs in France.
But serious doubt has been thrown on this statement, to which
I shall refer in the section on Historical Accounts."

2. In Sicily, Termes multiplies by means of complementary
or substitute sexual forms.

3. If two colonies of Termes, both containing comple-
mentary queens, are taken from different trees at some distance
away and intermingled, they are found to fraternise without
the least difficulty.

This has been demonstrated for Termites collected respec-
tively at places six and eight kilometres apart.

It is certain that the insects commingled in the last two
experiments belonged to absolutely distinct colonies, and it
must be inferred that the Termites of different nests fraternise.
And therefore, in spite of their mutual recognition, the appli-
cation to them of those criteria, which are so effective in
deciding whether bees belong to the same hive or not, becomes
excessively difficult. One can satisfy oneself that friendship
takes place equally whether complemental queens are present
in one or in both of the united populations; and this proves
that such queens live peaceably together without signs of
jealousy even when derived from separate nests.

4, Termites readily migrate from tree to tree, and carry

1 [Perris, who was a very accurate observer, records (‘Ann. Soc. Ent.
Fr.’ [5], vi, 1876, p. 201) the discovery in the Landes on June 10th of two
pairs of Termes lucifugus buried in the ground at the stump of a pine, and
each accompanied by a few (four) eggs. Pérez (‘Comptes Rendus,’ cxix, 1894,
pp- 804—806) also performed the experiment of furnishing winged T. lucifu-
gus with an artificial habitat. The couples were put into a jar containing earth
and a large block of decayed wood. They buried themselves in the earth, and
selected domiciles under it in the cracks of the wood. Pairs established on
April 29th were found to be flourishing on July 4th and August 30th, and
the abdomen was gradually becoming distended. They had penetrated more
deeply into the wood by the latter date, and were found with difficulty. On
October 15th six sexual forms were found in the same cavity together with two

young workers and an egg. They avoided the light, and no further growth of
the abdomen had taken place. ]

CONSTITUTION AND DEVELOPMENT OF TERMITES. 5

their eggs and young with them. Consequently many trees
which contain a brood are devoid of complementary queens.

5. A trunk containing a numerous population but no queens
is cut down and removed acertain distance away. The colony,
finding itself orphaned, at once sets to work to rear substitute
royal forms. The same thing happens when the comple-
mentary queens are killed; and if a few only are destroyed,
a few are formed.

All these experiments were begun from November to Feb-
ruary, and I did not search for the substitute queens till ten
months later. Ina single case I began the experiment at the
end of March, and looked for substitutes in June; instead of
the usual kind I found partially infuscate forms with torn
wings, such as I have previously described as originating from
imagos, which are still white, and as yet incapable of flight.

I have met with similar examples on various other occasions,
when my knowledge of Termitide was still too imperfect to
allow me to complete the observations then made.

Corollaries.—Termite societies inhabiting trees devoid
of complementary or substitute forms, either developed or in
process of development, are certainly in connection with
colonies in other trees which contain such forms. So far is
this true that, if the former are removed to a distance from
the latter, they at once set themselves to rear complementary
or substitute examples. The insects are therefore fully aware
when they have royal forms at hand, and regulate themselves
accordingly. Complementary queens are only found in nature
in one out of every fifteen or twenty trees infested with
Termites; and it must be recollected in connection with this
fact that the inhabitants of a trunk destitute of such queens
bestir themselves to form them whenever the means of com-
munication with the parent nest containing them are in-
terrupted by distance or other inconvenienccs, or whenever
they think fit to do so.

It will consequently be seen that there actually exists a

1 Since this work was written I have been able to establish four other
cases altogether similar to the one described above.

6 B. GRASSI AND A. SANDIAS.

certain number of Termite nests, each of which extends into
several trees; and it is impossible in a particular case to
decide without great difficulty whether any given examples
represent an entire colony, and whether the inmates of a tree
destitute of parental forms belong to one community rather
than to another.

c. Anatomical Observations on the Generative Organs
and their Connection with the Alimentary Canal.

Though I have already said a good deal about the substitute
and complementary king and queen, it becomes necessary to
give certain details about their generative organs, which my
investigations prove beyond doubt to be precisely identical
with those of the perfect insect.

I begin with those of Calotermes. The substitute queen
possesses a bundle of six ovarian tubes in each ovary (Pl. 19,
fig. 1), which increase in length concomitantly with her growth
in bulk, and eventually become over a centimetre long (PI. 19,
fig. 2), whereas those of recently developed substitute queens
measure scarcely half a centimetre. As a rule but one tube
of each ovary contains a nearly ripe ovum, which is always
larger than the most advanced ova of the remaining tubes.
The ovarian tubes may occasionally be bifurcated at the lesser
(anterior) extremity. Vitellogenous cells are absent, but an
ovarian follicle exists. The receptaculum seminis is elongated,
and never contains a large quantity of semen. The accessory
colleterial glands (Pl. 19, fig. 3) are also long; the oviducts
are short. The genital appendices, as we said before, are
absent, save in very rare cases, in which they differ from those
of the male by their less proximity to each other.

All these characteristics, with the exception of the occa-
sional presence of the appendices, are met with alike in the
true queen.

The number of follicles which compose each testis of the
king is difficult to determine, but is about nine. However, it
may confidently be stated not to differ in the true and sub-

OONSTITUTION AND DEVELOPMENT OF TERMITES. T

stitute kings respectively. Like the ovaries, the testes increase
with the growth in bulk of their possessor, and are therefore
notably smaller-in a small than in a large king (PI. 19, figs.
4, 5). The vesicule seminales and genital appendices are
present, but there is no intromittent organ. In all these
respects the true and substitute kings are identical.

Both kinds are found to contain ripe spermatozoa, though
always few in number, in the vasa deferentia. There is a
special liquid which rarely exhibits spermatozoa in the struc-
tures which, by analogy with other insects, I call the vesi-
culz seminales ; and spermatozoa are always absent in both the
vasa deferentia and vesicule of examples which still possess
fully formed wings and of young substitute kings.

No writer has hitherto observed the spermatozoa, on account
of their unusual character and small size; in describing them
~I shall limit myself to those features which can be made out
without special preparation, as I have not investigated them
minutely.

In order to make sure of their complete maturity, spermatozoa
were taken for examination from the spermatheca of the queen
(Pl. 19, fig. 12). They are relatively short, varying from 12
to 20 » in length, non-motile, flattened, and therefore differing
in shape according as they are viewed from the face or the
side. When seen from the face they are elongate and sub-
quadrangular, like a grain of rye, or better, a seed of zinnia;
sideways they appear more or less regularly linear. One ex-
tremity usually presents a characteristic thickening, and there
are certain cilia-like appendages which are better visible when
the spermatozoon is seen sideways. There is no difference
between the spermatozoa of the true and those of the substitute
king.

I now pass to Termes lucifugus. In the queen the ovarian
tubes are so numerous as to be counted with difficulty, but
there are about thirty-six on each side; though they do not ex-
ceed half a centimetre in length in the largest individual, their
shortness in comparison with those of Calotermes is more than
compensated for by their number (Pl. 18, figs. 1—4). They

8 B. GRASSI AND A. SANDIAS.

appear at first sight to be less numerous in young substitute
and in perfect queens; but this is deceptive, because many are
so far arrested in growth as to be with difficulty made out.
The remainder of the female generative organs (Pl. 18, fig. 10)
present no marked differences from those of Calotermes.}

In Termes alike the number of testicular follicles is difficult
to count, but is about eight (Pl. 18, figs. 5—8). Neither in
this nor in any other respect is there any difference between
the sexual organs of the substitute and of the perfect king.*

Spermatozoa are present in Termes lucifugus in the
spermatheca (in small quantity, from 150 to 1000 in round
numbers) and vasa deferentia, just asin Calotermes. For
examination they were collected by preference from the sperma-
theca (Pl. 18, fig. 9). Their form is round, about 2 to4 min
diameter. The immature spermatozoa in the testis possess a
tail, which is lost as they ripen. They are also non-motile.

The degree of development of the generative organs in
examples in process of becoming substitute kings or queens
varies concomitantly with the stadium in which they are. In
Calotermes, in which the phenomena can be better followed,
I find that the development of the gonads proceeds very slowly
in the royal substitutes, and that the pigmentation of the com-
pound eyes is almost always the earliest diagnostic character.
The female genital appendices are usually shed at the next
succeeding ecdysis.

I have spoken of the substitute forms in course of growth,
but which have not yet moulted, as larve of such forms; they

1 In Termes numerous unicellular glands, each furnished with a cuticular
duct, are crowded under the columnar epithelium of the spermatheca and
open into its lumen (Pl. 19, fig. 6). Ihave not examined Calotermes for
them. The accessory colleterial glands are two in number, one usually long
and full of secretion, the other shorter and narrower, its secretion being
generally very scanty.

These glands have a common aperture in connection with the ninth true
sternite, whilst that of the vagina, into which the spermatheca opens, is
in the eighth true sternite. These observations apply to Termes only.

2 The intromittent organ is wanting, and the male genital opening lies in
the ninth true sternite.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 9

are, as a rule, distinguished by the pigmented eyes and the
presence of genital appendices in both sexes.

The gonads are sexually differentiated, and the vasa deferentia
and oviducts are present at the time of birth, but there is no
trace of the external genitalia, which do not become evident
until the larvee are divisible by the greater or less size of their
head.

The soldiers and workers are of both sexes, as Lespés dis-
covered for Termes, and Miiller confirmed for the soldiers of
Calotermes. Observations made by means of longitudinal
and transverse sections allow of my giving some additional
particulars. The gonads are evident in every soldier and
worker, and persist in a nearly uniform stage of development,
regardless of age ; and all the component parts of the generative
apparatus of one or the other sex are invariably present, though
of very small size. Both the external and internal genital
organs are less reduced in the Calotermite soldier than in the
soldier and worker of Termes, and are better developed in
these forms of both species than in the larve at the time when
the head is beginning to vary in size.

The greater part of the abdominal cavity is occupied by the
gonads and the alimentary canal (the hinder part of the fore-
gut, the chylific ventricle, and intestine). The intestine
presents a cecal ampulla, which is sometimes greatly dilated,
but is much less so at other times, and may even be contracted.
In the first case it occupies a large part of the abdomen, and
must undoubtedly exercise an indirect pressure on the genital
glands, except during the earliest stages of life. The amplitude
of the cecum stands indisputably in an inverse ratio to that of
the degree of development of the gonads, and its greater or less
size is correlated with the presence or absence of certain
parasitic Protozoa. These are fully described in an appendix.
They are here mentioned with reference only to the singular
fact of their absence in substitute or complementary royal
forms, whereas they are present, at least in small numbers,
in the perfect insects.

10 B. GRASSI AND A. SANDIAS.

The ampulla also contains bacteria (Spirilla and Lepto-
thrix), which I have not specially studied, for they are not
absent in the complementary or substitute forms, and occupy
but little space. But it is crammed, so to speak, with Protozoa,
either Cercomonadide (Monocercomonas, Dinenympha),
Pyrsonymphide, or Lophomonadide. ‘Termites free from
Protozoa are generally distinguishable by the naked eye, on
account of the immaculate abdomen ; in those which possess
them the abdomen presents a more or less distinct yellow patch,
due to the cecum being visible through its transparent walls.
This patch derives its colour from the vegetable aliment of the
Termites, with which a large number of the parasitic Protozoa
are filled (see the Appendix) ; it is indistinguishable when the
food is colourless, and a Termite may then be supposed to be
free from Protozoa, whilst it is really harbouring an enormous
number.

Protozoa are absent in the new-born larve, and make their
appearance in Calotermes concomitantly with the appearance
of the four secondary Malpighian tubules, and with the ac-
quisition of a twelfth antennal joint, the third and fourth being
glabrous. These examples, described in the preceding chapter,
vary from 2°5 to 3 mm. in length, and may possess either a
large or a small head. If the former is the case their am-
pulla is comparatively dilated, and contains both Monocer-
comonads and Lophomonads; whereas those with a small head
contain Monocercomonads only, the Lophomonads first appear-
ing in a later stadium.

Monocercomonads, Dinemympha, and Pyrsonymphide begin
to appear in Termes lucifugus when there are twelve all
pilose antennal joints and eight Malpighian tubules, four large
and four small, and the body averages a little more than 2°25
mm. in length; and they are sometimes present and sometimes
absent in such specimens. More precisely, they are generally
absent in small-headed and present in large-headed individuals;
but they are constantly to be found when the body length
exceeds 2°5 mm.

The Lophomonadide do not appear, at least as a rule, until

CONSTITUTION AND DEVELOPMENT OF TERMITES. 11

the body exceeds 2°5 mm. in length, and the eight Malpighian
tubules have become similar.

In general, Protozoa may be said to make their earliest
appearance in the large-headed forms both of Termes and
Calotermes, at a time when the Malpighian tubules are
evidently divisible into two groups, large and small.

If we disregard the absence of Protozoa in the very young
larvee, they may be said constantly to be present in the winter,
when the insect development is arrested, in all members of the
community except the complementary or substitute forms, and
to be then very abundant in the soldiers and workers, some-
what less so in the larve and nymphs, and scarce in the true
kings and queens. I should add that they are found, usually
in small numbers, in every winged example which leaves the
nest.

In summer the Protozoa are somewhat differently distributed
on account of the frequent ecdyses of the Termites, before
each of which they die off, to be subsequently re-acquired.

That Protozoa are absent in the complementary or substitute
forms is a fact that has been verified in hundreds of cases.
But it is apparently contradicted by the following obser-
vations :

1. A certain number of Protozoa were found in queens
of Termes lucifugus, evidently of great age, as their
shrunken ovarian tubes and empty spermatheca showed.
That these examples had certainly borne eggs was demon-
strated, firstly, by the presence of a granular yellowish sub-
stance, never found in examples which have not oviposited,
and situated at the mouth of the ovarian tubes in the oviducts ;
and secondly, by the similarity in condition of the upper and
lower ovarian tubes, the former being always much the smaller
in virgin examples.

2. Whenever Protozoa were present in any numbers in a
substitute king or queen of Calotermes, the host was found to
be depauperised (e. g. by infestation with acari), and exhibited
an imperfect development of the testes or ovaries.

12 B. GRASSI AND A. SANDIAS.

3. Numerous Protozoa were found in a few substitute kings
and queens of Calotermes which had been kept some days
in a glass jar containing a little wood, together with many
other examples of the same species taken from different nests.
But during this time the royal specimens had been taken out
and put back several times, and may be regarded as having
been disorganised.

4. Lastly, a very few kings were found to contain a
moderate number of Protozoa in midwinter,—that is to say, at
a time very far removed from the period of copulation.

Allowing all due weight to these exceptions, we may still
say that they are far from disproving, but rather confirm the
rule, which may be thus formulated:—When the genital
organs are mature in individuals of which the wings
are not fully developed, Protozoa are absent, and the
dilated cecum is therefore of minimal size. The
absence of Protozoa does not, however, connote the
incapacity of the insects in question to possess them.

But here a grave question presents itself. Is the absence of
Protozoa a cause of the ulterior development of the genital
organs, or simply a consequence? Observations for the pur-
pose of solving this question were undertaken on comple-
mental or substitute forms in process of development (larvz or
immature examples). In these Protozoa were usually, but not
invariably, absent. The significance of this last fact is not
easily estimated, but after careful consideration I feel jus-
tified in offering the following explanation:—Examples in
process of becoming complementary or substitute sexual forms
do not habitually possess Protozoa; they may from time to
time become infected with them, but they rapidly get rid of
them independently of the moults. Possibly the development
of the genitalia is arrested during the infection.

It follows from all this that in complemental and sub-
stitute forms there is an accelerated or anticipated
maturation of the genital organs which appears to
be in intimate relation with the loss of the parasitic
Protozoa from the cecum. Such a loss in growing

CONSTITUTION AND DEVELOPMENT OF TERMITES. 13

specimens, in which the generative glands are still
very immature, or before they are in a condition to
undergo compression bythe cecal ampulla, suggests
that the absence of Protozoa may not be unconnected
with the maturation of the glands.

The experiments and observations described below were
undertaken by me in order to determine better the importance
of these parasites.

p. Experiments and Observations on Nests kept in
Glass Tubes.

(a) Calotermes.

1. Numerous small nests were made in glass tubes, each
containing from fifteen to forty examples of different ages, but
no adults and no royal pair.!. After a few days from two to
six Incipient substitute forms, characterised by the pigmented
eyes, were found in each tube; they were usually obtainable
in about a week if the tubes were carried in the waistcoat
pocket, and in summer they often appeared as soon as the
fourth day. In fact, after as short a time as thirty or forty
hours one could tell which individuals were capable of acquir-
ing the ocular pigment.

The abdomen of these specimens presents a characteristic
appearance, being hyaline and destitute of the tinge of colour
caused by the food (wood). They evidently possess no Protozoa.
But those which assume this appearance do not always acquire
the ocular pigment,—that is, they do not always become royal
substitute larvee. The eyes become (invariably?) pigmented
without an antecedent ecdysis.

The formation of such incipient royal substitutes never
takes place in any tube containing a royal pair. But if the
king and queen are removed for twenty-four or forty-eight
hours, or are injured, a few substitute forms appear at times.

1 The wood in such nests should not be white, or the presence or absence
of Protozoa cannot be readily distinguished by examination of the specimens
with the naked eye or a simple lens.

14 B. GRASSI AND A. SANDIAS.

These facts have been confirmed on very many occasions.

2. Furthermore, a nest was made up of three large larve
only; after a fortnight one of these exhibited pigmented
eyes, and was therefore in process of becoming a substitute
form.

3. The formation of substitutes may take place at any
time of the year, and is never delayed for more than a fortnight
if the tube is carried in the waistcoat pocket.

4. If a nest containing growing examples is furnished with
adults ready to fly, or with the wings shed in the act of capture,
the formation of royal substitutes takes place as usual, while
the perfect insects bore through the cork and escape in search ©
of fresh quarters. This confirms the observation that Calo-
termes does not make use of perfect insects capable of flight
as substitute forms; and the reason of this is certainly to be
found in that prepotent instinct towards quitting the nest
which arises in all Termitidz as soon as they become fully
coloured imagos.

5. If a nest is formed only of examples with not more than
twelve antennal joints, substitute forms are not developed
within the usual period. Butif a few small-headed larvze, with
from fourteen to sixteen antennal joints, are included, they are
promptly transformed into larve of royal substitutes.

6. A colony of about two hundred Calotermites is collected
~ and separated into some fifteen nests, the royal pair being
purposely killed. Every nest will furnish two, three, or four
substitute forms.

Another colony of equal size is collected and put into a
single glass jar after destruction of the royal pair; it will
form from four to six substitutes.

The same number of Calotermites will therefore produce a
very large or a very small quantity of substitute forms accord-
ing to circumstances.

These simple experiments at once suggest that the ordinary
members of the colony are capable of development into these
forms; and further justification for this view is found in the
fact that specimens in different stages of development (without,

CONSTITUTION AND DEVELOPMENT OF TERMITES. 15

or with more or less evident rudiments of wings) may be found
in process of elevation to this dignity.

7. In a few tubes (2, 3, or 4 per cent.) no substitute
forms are developed, in spite of the presence of individuals
apparently capable of undergoing that destiny.

T have established the fact that there is no development of
royal forms during May and June in tube-nests which contain
only nymphs.

And here I ask the reader to admire the talents of these
little animals !

What is the use of furnishing royal substitutes when all the
inhabitants of the colony will soon become fully winged, and
will disperse to found new nests? An occasional nymph may
be sacrificed to become a substitute form. But that takes
place only at a time remote from the swarm-period, or when
there are younger examples present.

A few nests, indeed, form no substitutes, and that without
any justifiable motive. I am unable to decipher their inten-
tions, but the fact remains that these republics, imbued, as
they seem to be, with Malthusian doctrines, find their parallel
in certain beehives which cease to concern themselves with the
possession of a queen.

8. In nests formed of individuals presumably free from Pro-
tozoa (that is immediately before, during, or after an
ecdysis) a certain number of incipient substitute forms (at first
larval, and later immature forms) appeared without any marked
delay. They preserved themselves free from Protozoa for some
time at least, but the other occupants became infected simul-
taneously with the appearance of these incipient substitutes.

9. Even if all the inmates of the nest contain Protozoa,
some of the usual substitute forms are produced without
material delay.

10. In the eighth case the appearance of Protozoa was due
either to their unsuspected presence in some individual which
infected the others, or to the accidental inclusion in the nest of
fresh dejecta of examples containing Protozoa (the nests having
been made with wood previously inhabited by Calotermites).

16 B. GRASSI AND A. SANDIAS.

The experiment was therefore repeated several times with
increased care in the selection of specimens, and with the em-
ployment of wood which had been occupied by Termes (many
parasitic Protozoa of which are not conspecific with those of
Calotermes).

In this way it was possible successfully to establish several
nests which remained free from Protozoa for over a month.
Nevertheless only a few examples from among all those of
suitable age became substitute forms.

It is thereby demonstrated that the mere ab-
sence of Protozoa is insufficient by itself to stimu-
late the maturation of the genital organs. But it
is certain that the first indications which point
to the development of a substitute form date from
their disappearance. However, this disappearance
also takes place in examples approaching an ecdysis,
and by the use of tube-nests it can be made out with
certainty that the incipient substitute royal forms
(larve) undergo a moult a few days after the eyes
begin to darken. The ecdysis may therefore be re-
sponsible by itself for the disappearance (death) of
the Protozoa, which would then be indicated as a
secondary phenomenon.

11. Nests of examples free from Protozoa are difficult to
keep alive. This raises the question whether the cause of
death is due to the absence of the parasites, or simply that the
inmates are in process of moulting, or are just entering or
quitting a moult. Recently moulted examples have an espe-
cially tender cuticle, and undoubtedly require a greater amount
of moisture in the air and food; they are therefore more deli-
cate, and this alone may sufficiently explain the difficulty in
question. But I shall return to this point later.

12. Examples in process of becoming royal substitutes
reached maturity in the tubes in very few cases (after three or
four months; the tubes in question were packed in a box and
taken to North Italy and Germany). Asa rule, the nest dies
away before they are fully developed.

CONSTITUTION AND DEVELOPMENT OF TERMITES, iG

So considerable a delay in maturation appears only natural
if it is recollected that examples are selected for the throne
at a time when their generative organs are very backward in
development.

Forms destined to become substitutes may be said generally.
to become re-infected with Protozoa at the end of a week or
ten days ; they lose them in about a month at the succeeding
moult, but may subsequently re-acquire them. Too much
importance should not be attached to these facts, because
they are exhibited when the nests are mostly in bad con-
dition.

From the foregoing observations I infer that the transfor-
mation of ordinary into substitute forms must be dependent
on a change, either quantitative or qualitative, in the cha-
racter of the ordinary diet. This is a necessary sequence from
what can be made out in the tube-nests, in which, in fact, we
have seen that a small quantity of decayed wood suffices not
merely to keep the Calotermites alive, but to enable them to
rear up substitute forms.

I shall therefore record a series of observations on the
aliment of Calotermes, which have been carried out by
means of tube-nests. '

Their food and drink consist of—

1. Wood.

2. The matter excreted or disgorged by other Caloter-
mites.

3. The exuviz of each other.

4. The corpses of other Calotermites.

5. Their own saliva or that of others.

6. Water.

*Calotermites triturate dead or dry wood (and cork) with
their mandibles, and devour the fine meal which is thus
obtained. This meal is not wholly used for food, but, as we
have mentioned more than once, is employed partly for build-
ing, and is also partly regurgitated (vide infra). Wood is
therefore the basis of the alimentation of Calotermes.

The insects are consequently nourished on a material very

vou, 40, paRT 1.—NEW SER. B

18 B. GRASSI AND A. SANDIAS.

deficient in nitrogen, and their slowness of growth must be
correlated with this fact. They must certainly digest cellu-
lose or lignin, or both of these ternary substances.

The mandibles of the soldiers are incapable of gnawing up
wood. This explains why these forms die after a short time
if kept in isolation—a fact I have repeatedly established by
forming nests entirely of soldiers (videinfra). Nevertheless,
they eat wood which has been triturated by other members of
the colony.

The dejecta play an important part in the nutrition of
Calotermes, and may be said to form the ordinary diet.
Lignin (by the phloroglucin reaction) and cellulose are
found to be present in the feces, which appear under the
microscope as a very fine detritus, in which traces of fibres. or
spiral vessels are to be made out with difficulty. They appear
to the naked eye as cylindrical scybala, usually a little over
half a millimetre, but sometimes as much as a millimetre in
length. Their colour varies in accordance with that of the
wood in which the Calotermites are living, and is commonly
dirty white if derived from cactus, red-brown or brown if from
almond, &c. The scybalum is often brown with one extremity
reddish, and is usually quite brown when stale. If carefully
examined with a lens it is seen to be prismatic rather than
cylindrical, owing to the impressions it receives from the rectal
plicee.

Calotermites ingest this substance either fresh or old, and
may use it for building, either directly or after ingestion and
subsequent regurgitation. By prolonged observation of the
living animal, or by suitable examination of the contents of the
alimentary canal, it is easy to satisfy oneself that but a very
small proportion of what is ingested is regurgitated.

The dejecta are greatly preferred at the time of elimi-
nation, probably because, as is obvious, they are then less dry
(Calotermes does not habitually drink, as I shall mention
later).

In order that they may be obtained in this condition, their
elimination is provoked by means of the antenne and maxil-

CONSTITUTION AND DEVELOPMENT OF TERMITES. 19

lary, possibly also the labial palpi (I have not been able
definitely to establish the latter point).

When a Calotermite wishes to feed, he accosts one of his
fellows and caresses the abdomen with the antennz and palpi.
If the one thus accosted is prepared to eliminate, he at once
extrudes the scybalum from the anal aperture. The other
then removes it, chiefly by aid of the maxillary palpi, and
usually in two operations separated by a short interval, draw-
ing it at first half, and then completely out. -He then rapidly
seizes it with the mandibles, suspending his caresses for this
purpose, and when he has possessed himself of it, nibbles at
and ingests it little by little.

If the insect accosted is unable to furnish what is required,
he is at once abandoned and another pursued, and in this way
several examples are solicited in turn until one produces the
substance desired. But a single scybalum is rarely sufficient,
and consequently the one which has devoured it immediately
sets off to procure another.

A scybalum, of which the extrusion has been caused by an
individual even of royal stock, is sometimes seized and quickly
devoured by another.

If the example solicited is prepared to eliminate, it stands
still, otherwise it runs away; and the thing becomes more
curious when the dejecta are either not quite ready for extru-
sion, or when but a small.amount is ready. The one solicited
runs away and the other pursues ; -and this most probably is
the explanation of the so-called amorous passages seen in
many termites and described in the preceding chapter.

The operation of caressing the apex of the abdomen to
procure elimination of the dejecta is performed alike by the
soldiers, but with some difficulty, on account of the large size
of their mandibles. The soldier achieves his purpose ‘by
placing himself so that his head forms nearly a right angle
with the body-axis of the example caressed, and the’ buccal
apparatus situated on the under side of the head comes into
contact with the anal aperture of his fellow, who in turn is
held steady by the soldier’s mandibles resting on the upper

20 B. GRASSI AND A. SANDIAS.

side of the posterior extremity of the abdomen. Palpation is
accomplished to some extent by the mandibles, but principally
by the fore-legs, as is sometimes observed in other members of
the colony.!_ The soldier ingests slowly and with difficulty.

Other individuals do not fail to perceive what is taking place,
and make attempts at robbery, which are sometimes successful.

After provoking the elimination of the dejecta, a large
example sometimes retires to allow a little one to partake;
and the latter has been occasionally noticed to have previously
twitched the legs, and especially the tibie of the larger one,
perhaps as a warning that it was hungry.

It should be added that much fecal matter is sponta-
neously deposited in the middle of the nest, and, if not in-
gested, is then carried away by its eliminators, who employ it
in forming barricades, or store it up in uninhabited parts of
the nest.

When an individual has finished eating, another approaches
and licks over the front of the head and antennz with rapid
movements of the palpi, evidently to clean them; not un-
commonly two clean each other alternately. Then one may
stop for a moment while the other continues the process, and
passes to the occiput and then to the legs, going from their
base to their apex. The recipient of these attentions stands
still for some time, and then rapidly takes to flight, probably
because the other has bitten one of its tarsi on reaching it.
This cleansing operation is usually resorted to after a meal,
but is sometimes manifested without that preliminary.

If the mouth of certain examples is attentively observed, it
will often be seen to exhibit a microscopic reddish-brown
pellet, which gradually increases in bulk till it forms a rounded
mass of about a millimetre in diameter. This pellet is obviously
composed of regurgitated food, and is sometimes used for
building ; at other times another example approaches, seizes
and devours it.

The excreted or disgorged substances are occasionally

1 Excretion may also be stimulated by gently stroking and compressing
the abdomen with a feather.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 21

yellowish and semi-fluid, and the exact method of nutrition
is then more difficult to ascertain.

Though ecdysis is a very frequent phenomenon in Calo-
termes, it is difficult to find exuvie in the nest. This,
coupled with the fact that the alimentary canal sometimes
contains portions of cast skins, suggested to me that they are
habitually devoured. Attentive observation shows that this is
really though not invariably the case. Sometimes the assist-
ants at a moult eat the exuvie bit by bit as they are shed ;
and at other times an example seizes the skin directly it is
cast, and carries it away for a greater or less distance to eat
part or the whole. The proctodzeal cuticle may or may not
be shed with the rest of the exuvie. If it is not, it may be
passed either spontaneously, or as a result of the palpation
and pressure employed to produce defecation by another
insect, which then draws out and devours, instead of the
dejecta, a white substance, which is the cuticular lining in
question.*

Besides the practice of eating the cast skins, I have men-
tioned that at times a Calotermite may apparently finish the
operation of licking another by a bite. The carnivorous pro-
pensities of these insects do not stop there. Any example not
in normal condition (e. g. which is shrivelled, or which shows
sigus of being ill by remaining long motionless for some un-
decipherable reason, or which is unable completely to moult
the skin) is eaten before death by its companions. It is not
uncommon to see a soldier decapitate a living nymph, or to
find an example with the abdomen eaten and the legs still in
motion. At times a victim attempts to retreat, but is then
usually decapitated at a stroke by a soldier. But this mode
of execution is not always resorted to, and the antenne may
be bitten off first, &c.

*I may add in passing that so great is the ferocity of the
soldiers that one may be occasionally seen furiously to assault
five or six other individuals and bite off heads and abdomens,
&e. The object of these massacres, which take place only
when the nest is thrown into disorder, is not quite clear.

22 B. GRASSI AND A. SANDIAS.

Probably the soldier imagines that his companions are the
cause of the disturbance.

The king of a queenless nest was observed to be dead, but
a few hours later the body could not be found—a clear proof
that it had been devoured. This king had shown signs of ill-
ness a few days previously, and the population had provided
a certain number of substitute larvee.

On one occasion nine examples, of which one was a soldier,
were detected at night in the act of devouring a royal sub-
stitute in process of moulting. The wretched animal was
still alive, and writhed all over its body to free itself from the
torture. The nine assassins were probably annoyed at the
light to which they were exposed, for they at once stopped
eating, and jointly carried off the victim by means of their
trophi to a darker part of the nest. Meanwhile many others
crowded up, evidently to partake in this feast of royal flesh.

Occasionally one is seen to lick the leg of another for some
time, and then suddenly to bite off the tarsus.

The tubes often contain nymphs with imperfect wings
which have been nibbled off by their companions; and it is
remarkable that the right fore-wing is usually selected in
those destined to become substitute forms. This has been
already referred to, as has the constant mutilation of the
antenne in the royal pairs. The preceding observations suggest
as an explanation of the latter curious phenomenon that the
antennz have been gnawed off. Possibly the royal examples
bite off each other’s antenne, for they are found to be already
mutilated in solitary royal pairs, that is without offspring.

An individual which has been some time dead is no longer
eaten.

The saliva next claims attention: it issues in connection
with the labium as a colourless and distinctly alkalme liquid,
which contains no structures discoverable with the microscope.
This liquid collects on the labium as a small drop, which may
be employed either as a cement in building or as food for
others. These may either possess themselves of the drop and
then retreat a little way to swallow it gradually, or they may

CONSTITUTION AND DEVELOPMENT OF TERMITES. 23

receive it from the one which secretes it and clearly provides
it for them as an article of diet.

The assimilation of a drop requires a certain number of
acts of deglutition, which may be counted, usually four or five.

Examples may sometimes be seen to perform several such
acts in order to swallow their own saliva as it trickles from the
labium.

Termitide do not habitually drink; but Calotermites
(imagos, soldiers, &c.) which are moribund from drought may
be seen to apply the mouth to wood saturated with water and
suck the latter up, keeping the mandibles still and moving the
rest of the mouth parts. When the water is drained from one
spot they move to another.

Other parched examples (soldiers kept some time in a dry
place) have been observed to have recourse to a drop of water
placed in a watch-glass, in which they were found. When the
watch-glass was examined from below, the soldiers were seen
to move both the mandibles and maxillary palpi when drink-
ing, holding the legs as erect as possible, and the abdomen
turned up so as not to wet the body.

We may disregard these exceptional cases to consider how
and when the different forms are nourished.

Wood-meal is eaten by all except the very young, which
first begin to do so when the apex of the mandibles and
maxille becomes coloured. This statement applies equally
to the dejecta, vomit, exuvice, and dead bodies, except that the
young appear to feed on the two former substances earlier
than on wood.

I shall specially mention the following phenomena as having
been observed : the soldiers are sometimes supplied with vomit
standing mouth to mouth; they also regurgitate it; the royal
forms feed alternately on each other’s dejecta; large examples
may solicit from small.

Saliva is given or yielded in abundance to larve which are
too young to eat wood, and to those in course of becoming
royal substitutes, and a certain quantity is also given to the
other small-headed forms.

24 B. GRASSI AND A. SANDIAS.

It is only with much patience that such phenomena can be
determined, but it can be done with complete certainty. It
has been especially noticed that an incipient royal substitute
was approached after a recent ecdysis by different individuals
which administered saliva to it; and it should be added that
after the last moult the nymphs make repeated acts of deglu-
tition to swallow the saliva secreted by their own glands.

Moreover saliva is seen to trickle from the labium of, and
to be swallowed by incipient royal substitutes which have just
finished moulting, if they are not supplied with it by their
companions. The secretion may stop for a few minutes and
then recommence. The forms which secrete saliva for others
are large larve or nymphs.*

Individuals fed with saliva exhibit a great transparency of
the abdomen, an indication that they are in progress of becom-
ing royal substitutes. They contain no Protozoa, or contain
them dead, and their death or disappearance is most probably
a direct result of the action of the saliva.

It is a moot point whether the maturation of the generative
organs is due solely to the saliva or to the absence of Pro-
tozoa as well; but the latter, as my previous remarks show, is
not by itself a sufficient cause.

I have frequently asked myself whether the Protozoa have
not an important digestive function, since the comminuted
wood passes almost entirely through their bodies. It is pro-
bable, but not proved.

*Termitide can go without food for several days; the
soldiers especially eat much less than other forms, and can
endure more than eight days’ abstinence. Exercise, therefore,
costs the colony little.

As is only natural, when the feces have repeatedly passed
and repassed through the bodies of the insects, they are no
longer sufficient to support life. For this reason a nest of
nothing but soldiers dies, because they are incapable of gnaw-
ing up fresh wood. But the addition of a single large larva
to a colony of ten or a dozen soldiers is enough to keep them
alive.*

CONSTITUTION AND DEVELOPMENT OF TERMITES. 25

(5) Termes lucifugus.

Experiments on this species are much more difficult, and
can only be carried out in large glass receptacles, which do
not permit of the necessary observations. Substitute forms
were easily obtainable in greater or less numbers in the few
cases in which the Termites survived in abundance and in
good health for some months. On this point I have nothing
to add to what is recorded in other parts of the present work.

*The complementary queens are objects of tender care not
only on the part of the workers, but of the larve, and are
much more assiduously cleaned than the substitute queens of
Calotermes. Five or six companions will stand round one
of these queens at the same time, one cleaning her legs, another
the antenne, a third the abdomen, &c.*

Their nutrition exhibits the same features as those related
for Calotermes.

CONCLUSION.

The facts recorded justify the conclusion that the
saliva of both Termes and Calotermes exercises a
marvellous influence on individuals in process of
becoming perfect insects; it effects their transfor-
mation into substitute orcomplementary royal forms.
This is substantially a most remarkable phenomenon
of neoteinia. But whereas neoteinia, or the sexual
maturity of forms retaining larval characteristics, is
dependent in Amphibia on the environment, in Ter-
mitide it is essentially subordinate to the nutrition.

II. DEVELOPMENT OF THE SOLDIERS AND WORKERS.

Termitide possess three special castes—workers, soldiers,
and neoteinic forms (according to the expression used in the
preceding paragraph). The latter are not only arrested in

26 B. GRASSI AND A. SANDIAS.

development, but may present characters peculiar to themselves,
and differing from those of the perfect insects (e. g. the long
outstanding abdominal setx and black maculation in Termes).
Special and still more pronounced characteristics are exhi-
bited by the other larve of arrested growth, that is the castes
of soldiers and workers. The proof just given that one, the
neoteinic caste, is dependent on nutrition, is enough to suggest
on a priori grounds that this may be equally true of the
others. ‘

Strictly a priori the soldiers may be regarded as further
differentiated workers, and indeed at the beginning of their
development they exhibit but a single known characteristic,
the enlargement of the head as in the worker. The soldier
therefore begins by possessing the characters of the
worker. In other words, we may say that an individual
destined originally to become a perfect insect is capable of
undergoing a paranomalous development, of acquiring
certain characteristics which, by absolutely destroying its
fertility, no longer allow of its becoming a perfect insect. If
the differentiation of the head is arrested when certain
structural modifications have been acquired (simple increase
in its size, a special form of pronotum, &c.), and the individual
is limited further to uniform growth in stature, except for the
augmentation in the number of antennal joints, we have a
worker. But if, instead of this limitation, the mandibles and
labrum become elongated while the maxille and labium do not
alter, then we have a soldier. In short, worker and soldier
follow a common path for some distance; at a given point
one, the worker, continues along it, while the other, the soldier,
diverges from it. And with this are connected the facts that
young soldiers and workers are indistinguishable, and that
certain Termitide (Calotermes) possess soldiers, and others
(Anoplotermes) workers only. ;

But these inductions all require the control and support of
direct observations ; and mine indeed are not quite complete,
as I have been unable to keep Termes alive for a sufficient
time.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 27

Nevertheless a few crucial tests have given results entirely
in accordance with the above induction. These are :—

1. Transformation into a soldier larva, and subsequently
into a soldier, can take place at very various ages.

As in Calotermes, the soldier may originate from ex-
amples with from twelve to seventeen antennal joints, and
therefore of very different sizes and with or without rudiments
of wings (see § 2). Newly formed soldiers are therefore to
be found of different dimensions (small, medium, and large),
and with a variable number of antennal joints.

The small soldier is an inhabitant of newly established
nests which experience the want of soldiers, and therefore
anticipate their development.

2. Various observers have found soldiers with more or less
well-marked wing-rudiments in the nests of exotic Termitide,
and have regarded them as an inexplicable abnormality. My
belief, on the contrary, is that their development is the result
of unusual or extraordinary nutrition, and is in harmony with
my previous statements. If it be correct, it should be possible
to put Termitide under such conditions as would compel them
to produce similar monstrosities. That such a thing is really
possible is shown by the following experiments.

*In winter small Calotermite nests of nymphs alone, and
therefore free from soldiers, are established in tubes containing
triturated wood, and are kept in a warm place, in the pocket
or near astove.! After a time they are found to contain not
merely a certain number of nymphs in process of becoming
royal substitutes, but others as well, which are being trans-
formed into soldiers, and are distinguished especially by an
elongation of the mandibles and labrum as well as by the
greater size of the head. The wings are simultaneously re-
absorbed until barely a vestige remains. These nymph-soldiers,
as they may be termed, become perfect soldiers after an ecdysis.
It cannot be positively stated that all succeed in reaching the

1 Tt should be added that the development of Termitide kept in a warm

place proceeds even during the winter, the queens prematurely laying eggs,
which hatch precociously, &c.

28 B. GRASSI AND A. SANDIAS.

goal, because they remain unchanged in some nests for months
together. The pigmentation of the eyes found in some of
these nymph-soldiers is curious but intelligible.

Similar experiments were made on natural nests in trees by
removal of a large number of soldiers, and after some time a
few nymph-soldiers were found in them.

I was led by this to suspect that such forms may at times
be normally found in intact nests, and by paying particular
attention to the origin of the large soldiers I actually found
that normal nests in which the soldiers originate from larve
with wing-buds, or nymphs, are not rare; this phenomenon has
escaped the most accurate investigators, including Fritz Miiller,
on account of the reduction of the wing-rudiments, which
perhaps disappear completely in the case of larve.*

The following fact shows that these phenomena hold good
equally for Termes.

In a small nest of Termes lucifugus kept in a glass jar
I obtained a nymph-soldier which had almost exactly the
thorax of a nymph of the second form, and the head of a soldier.
At the time of formation the nest contained a certain number
of perfect insects, many workers, a few soldiers, and some
undifferentiated larve. This nymph-soldier was found in it
six. months later, together with a number of workers, an
ordinary soldier, and a nymph of the second form. Death
had evidently claimed many victims during the six months
that the nest lasted.

I have met with similar nymph-soldiers on other occa-
sions. All these observations are sufficient to show that the
casts of workers and soldiers denote merely a paranomalous
development of individuals capable of becoming perfect in-
sects.

3. Further evidence is afforded by the fact that the newly born
larvee are relatively alike inter se,—relatively only, because
they are not all identical in bulk. This was observed by taking
Calotermites in the act of hatching (they emerge at one pole
of the egg) and comparing them after preservation by a uni-
form method.

CONSTITUTION AND DEVELOPMENT OF TERMITES, 29

*It is hardly practicable to discover Termes in the act
of quitting the egg ; consequently a large number of examples
must be collected, subdivided according to their different
stadia, and then compared.

*4, The newly hatched larve are fed on saliva, to which,
after some time, is added a clear yellowish vomit, probably
largely diluted with saliva; later still they begin to adopt
other articles of diet.

Examples of which the head is beginning to enlarge must
receive but a very small amount of saliva, as is shown by the
woody or fecal colour of the intestinal contents. But those
in which the head remains narrow continue for a longer time
to receive pure or nearly pure saliva, which is afterwards
always administered in some quantity.

5. If a large number of small and narrow-headed larve are
added to small tube-nests, some without and others containing
soldiers, part of those in the former nests are generally found
to develop a large head, which is not normally the case in the
latter.

It is clear, therefore, that the colony short of soldiers tends
to furnish itself with them, as it does with royal forms.

But proof is required that the small large-headed larve
really become soldiers. A calculation of the proportionate
numbers of the castes in colonies of Calotermites shows that
of the soldiers to be very small—at most one in every fifteen
or twenty in a very populous nest, or one to every five or six
others in a very small colony. Now, in the aforesaid soldier-
less nests one large-headed larva appears to every five or six
inmates, and in natural nests the small large-headed larve are
very few in proportion to those with a small head.

All this goes far to furnish the necessary proof. And the
matter becomes a certainty if we study the nests of Termes,
in which the young, with already differentiated heads, show a
large percentage of large-headed forms, in agreement with the
fact that a considerable number of workers is developed in
addition to the soldiers.

Evidently, then, the large-headed larve are des-

30 B. GRASSI AND A. SANDIAS.

tined to become workers and soldiers, or the latter
alone in Calotermes, in which workers do not exist.

Although details cannot be given, it may be concluded that
the development of the soldiers and workers is consequent on
the less quantity of saliva which they receive; and with this
is associated the earlier appearance of Protozoa, and their con-
stant presence in great abundance.

The method by which the soldier is further differentiated
from the worker is a more obscure matter to determine. Is
the larger amount of nutriment which the latter receives the
cause of such a phenomenon? In any case it is certain that
the essential factor must be one of nutrition.

Numerous experiments have shown that the soldiers cannot
become royal substitutes, and further that they themselves
are incapable of transforming larve or suitable nymphs into
substitute forms.

GENERAL CONSIDERATIONS AND RerrosPectivE NoTE ON
THysanura.!

1. The soldiers, workers, neoteinic forms (complementary
or substitute kings and queens), and the true perfect insects
are all derived from similar ova; in other words, every ovum
must be regarded as inherently capable of giving rise to any
one of these four classes of forms. It follows that the dif-
ferences between them connote something absolutely unrelated
to the distinctions of sex; and the existence of Termite
castes has therefore no bearing on the theory which
postulates that every somatic cell is derived partly
from the mother and partly from the father.

The following points are also of importance :

(1) The neoteinia of Termitide is no isolated phenomenon,
but is paralleled in many other insects, including the Orthoptera
(sensu lato).

Sexually mature forms, some with well-developed, others
with short wings, and that independently of sex, are met with .

1 (Grassi, ‘ Atti Acc. Lincei’ (4), iv (1887), pp. 543—606, pls. i—v; with
bibliography of the author’s antecedent memoirs on Thysanura. ]

CONSTITUTION AND DEVELOPMENT OF TERMITES. 31

in certain species of grasshoppers. Similar phenomena occur
in Psocidee, Perlidze, &c.

(2) The Embiide, which, as will bé seen later (Appendix IT),
have been regarded as near allies of the Termitide, though
they are nothing of the sort except in so far as they belong
to the same order, the Orthoptera (s. lat.), exhibit a sexual
dimorphism comparable with that subsisting between the
soldiers and other Termite forms, namely, a much greater
development of the mandibles in one sex than the other.

As the marvellous effects produced by differences
in nutrition remain always indisputable, this would
lead us to believe that the intimate structure of
Termitide may form, so to speak, a nidus specially
adapted for their manifestation.!

2. The great influence exercised on the genitalia by the

salivary secretion is a fact which finds its analogue in bees. And
inasmuch as these insects are very remote from Termitide, and
are of independent phylogeny, it must be inferred that we
are in presence of a marvellous phenomenon of con-
vergence.
- 8. If the aliment can exercise a profound influence on the
genital organs in forms so remote from each other as Termi-
tide and bees, we may fairly suppose it to produce a direct
effect on these organs in many other animals. And
this seems to me to form a serious argument in
favour of the possibility of inheriting acquired
characteristics.

No light is thrown on the mode of development of the

1 [In other words, the intimate structure of the protoplasm in other
species of Orthoptera already exhibits a tendency towards the production of
better developed mandibles in certain individuals (males). We may therefore
suppose that at one time the ancestors of Termitide exhibited a much
greater development of the mandibles in one sex than the other. Similar
suppositions, and others which readily present themselves to one’s mind
facilitate, in my opinion, the understanding as to how this character may
present itself in the soldiers without the invocation of direct inheritance.
(See the introduction to my memoir on “The Anatomy of the Thysanura.”)
G. B. Grassi, December, 1896. }

32 B. GRASSI AND A. SANDIAS.

workers, soldiers, and neoteinic forms, either by the compari-
son instituted by Darwin or by the existence of worker bees
capable of oviposition.!

I regard it as possible that the characteristics of these three
classes of forms may be suddenly manifested, because, as I have
previously suggested, the latent tendencies may be pre-existent
in the as yet undifferentiated Termite larve.

4. Since the development of workers, soldiers, and neo-
teinic forms is arrested, as many of their characters show,
and since the metamorphosis of Termitide is incomplete, the
development of castes in this family is entirely different from
that which occurs in bees, ants, &c.

Further, the fact remains that the workers, soldiers, and
neoteinic forms on the one part and the perfect insects on the
other part are not separated by that enormous difference of
instincts which is manifested between the queen and worker
bees. On the contrary, the instincts of the former group of
forms are possessed generally by the other members of the
Termite colony.

5. In the desire to estimate the importance which attaches
to the facts disclosed of my investigations on Termitide, I
propose to begin by the consideration of a preliminary question.
In earlier geological periods the Termitidze were represented
in Europe by numerous species, which are now reduced to
two, and confined to 8. Europe, whilst a large number still
exist in the tropical and subtropical zones.

This and other facts already discussed indicate that the
different European distribution of Termitidz at other periods
was probably dependent on the warmer climate which then
prevailed over this part of the world.

These considerations, together with the fact that our termite

1 [Of late I have changed my mind, having had occasion to meet with a
nymph-soldier (vide p. 14) of Termes lucifugus with well-developed
ovarian tubes, and return to the supposition that the phenomena of inheritance
in the sterile casts of Termitide (workers and soldiers) can be interpreted
as I have pointed out in the introduction for the bees, i.e. by the exceptional
existence of workers and soldiers capable of oviposition.—G. B. Grassi,
December, 1896.]

CONSTITUTION AND DEVELOPMENT OF TERMITES. 33

queens are very far from attaining tle colossal dimensions so
well known in exotic species, suggest that our Huropean species
may be in process of degeneration.

Such a belief would rest on the behaviour of the royal forms
in the Termes society; but is this foundation really valid ?
At first sight Termes lucifugus certainly appears to be
very different from the species hitherto described, but an exa-
mination of the many brief and incomplete accounts we
possess allows me to assert that all the Termitide of other coun-
tries really fall into the two primary types I have observed in
Calotermes and Termes. These are—

(1) A colony presided over by a king and queen,
which have possessed and shed fully developed wings.
When orphaned it is headed bya pair of royal sub-
stitute or neoteinic forms.

(2) A colony at the head of which are numerous
neoteinic queens, the kings, also neoteinic, being
present for short periods only. The colony is not
founded by the royal forms which govern it; or, more
exactly, the neoteinic forms have been raised by a
detached portion of a pre-existing colony, which thus
founds a new and independent society.

Hagen discovered, but could attach no meaning to, a certain
number of examples in other species, e.g. Termes morio,
Latr., which must be interpreted as neoteinic royal forms.’
Fritz Miiller states? that Hageu wrote to him that the Asiatic
and African queens are all true imagos with wing-stumps,
but that all those from Brazil and America generally are in
the nymph garb; but this is contradicted by Miller, who found
hundreds of true queens in Brazil, as well as those in nymph
form.

I am myself acquainted with nymph-like forms of several
tropical species; I therefore conclude that all Termitide are
reducible to the two main types above mentioned, and see no

1 (A better example of a neoteinic form is given by Hagen in Termes
debilis, Heer, ‘ Linn. Ent.,’ xii, p. 207, pl. i, fig. 26.]

2 [‘Jen. Zeitschr.,’ 1873, p. 457, note 3.]

vou. 40, part 1.—NEW SER. C

34 B. GRASSI. AND A. SANDIAS.

further reason for regarding the European species as degene-
rate.

6. This question settled, another arises as to which of the
two European forms must be considered the more primitive.
Evidently Calotermes, for the following reasons :

(1) The ovary is composed of seven ovarian tubes only, and is
therefore in a relatively archiac condition (see my studies on
Thysanura) ;

(2) The soldiers are still furnished with eyes ;

(3) The colony is headed by individuals which have pos-
sessed wings suitable for flight ;

(4) The art of building is little developed in all species of
the genus ;

(5) The queens of this and other species are relatively small,
and therefore resemble the females of other insects ;

(6) There are no workers.

This aggregate of facts leads one to say positively that
Calotermes is the more primitive, while none of them justifies
the supposition that it is a degraded form.

7. What interpretation is to be placed on the singular fact
that all the winged forms of Termes are inexorably lost?
Does this loss take place everywhere, and have not numerous
cases occurred in France in which they have been found to
establish new nests, as I have shown for Calotermes?
These are questions which admit of no definite answer, because
the observations made in France by Lespés are not entirely
reliable, as Hagen has pointed out.

Fritz Miller, whilst denying that the winged forms can
possibly found new colonies, and supposing, on the contrary,
that they enter orphaned nests, furnishes the clue to the pro-
blem by a most original comparison.!

He compares the winged forms to perfect flowers, and the
substitute nymphs, as he regards them, to cleistogamic flowers.
The object of the winged forms, like that of the perfect flowers,
would: be to render interbreeding difficult; and that of the
substitute nymphs, like the cleistogamic flowers, would be to

1 (‘Jen. Zeitschr.,’ 1873, Beitr. iii, pp. 451—463.]

CONSTITUTION AND DEVELOPMENT OF TERMITES. 30

furnish a means of escape from the considerable danger to
which the perfect insects are necessarily exposed.

In other words, according to Miiller, Termites furnish royal
substitutes only if orphaned at a time when there are no
winged forms. Their procedure is really different, but the
comparison between them and plants possessing both perfect
and cleistogamic flowers still holds good. Calotermes may
be compared with a hypothetical plant, in which seed is pro-
duced by perfect flowers, and not by cleistogamic flowers unless
the others fall; Termes (in Sicily) to a plant in which the
perfect flowers do not produce seed, but the cleistogamic flowers
seed abundantly.

The separate swarming of the sexes, discovered by myself,
fully confirms this conclusion.

Lastly, the phenomena exhibited by Termes indicate an
advance on those found in Calotermes.

8. It will be seen from the second appendix to the present
work that there is no direct evidence of relationship between
Termitide and Embiide if we disregard the wings, which,
according to Westwood,! Hagen,’ and Redtenbacher,’ are
somewhat alike. But Redtenbacher himself, the most com-
petent authority of the three, concludes that the Embiide are
separated by the wing-structure as an entirely special group,
which exhibits relationship with the Termitide and Blattide,
as well as with the Perlid.t Thus there cannot be said to be
any close connection between the Termitide and HEmbiide
even in wing-structure, while the number of the Malpighian
tubules, the arrangement of the testes and ovaries, the sper-
matozoa and alimentary canal, all present differences of the
highest importance.

The Embiide, it is true, lead a sort of gregarious life, but
they do so under conditions very different from those found in
Termes, since they show no sign of division into castes.

' (*Trans. Linn. Soe.,’ xvii (1837), pp. 369—378, pl. ii.]

[‘ Canad. Ent.,’ 1885.]

3 (Ann. K. K. Nat. Mus. Wien,’ i, pp. 153—232, pls. ix—xx.]

* [A similar conclusion is arrived at by Brongniart, ‘ Insectes Fossiles,’
p- 215.)

36 B. GRASSI AND A. SANDIAS,

9. If the Termite colony were headed only by royal forms,
such as have been described, without sign of wings, the
phenomenon would occasion no surprise, but all valid proof
would be wanting that the Termitide themselves had ever
possessed wings.

This consideration may throw some light on the origin of
the Thysanura. My researches on these insects showed that
their characters are partly primitive, and partly connect them
intimately with the Orthoptera (s. str.). Such a connection
seemed, however, to be contra-indicated by the hypothesis,
which I accepted, that the Thysanura have never possessed
wings, structures of which I have been unable to find the least
indication.

But I am compelled to reject this hypothesis, owing to what
I have seen take place in Termitide under my eyes, and to
admit that the Thysanura may originally have possessed
wings, which they have subsequently and entirely lost. They
must be descended from primitive forms in which the first
abdominal segment remained independent, with its sternite
complete and not fused with the metasternum as in Termi-
tide ; and not a few winged Orthoptera (s. lat.) are known to
exist which have the first abdominal sternite separate as in
Thysanura.

And this leads me toregard the division of Insecta
into Apterygogenea and Pterygogenea as unfounded,
whatever Brauer may assert.}

In other words, 1 admit that all existing species of
insects may once have possessed wings.

Historical Accounts and Developments.

I leave to the writer of a complete monograph on the Ter-
mitide the task of elucidating point by point the contribu-
tions to a knowledge of them which Dr. Sandias and I have
made and set forth, but I cannot here pass in silence over
some preceding accounts of special significance.

* © Zool. Anz.,’ 1888, pp. 698—600.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 37

In a preliminary note I have recorded what has been made
out about the neoteinic forms from the works of those who
have ‘previously studied the subject, and especially from the
paper by Fritz Miller, and I repeat my remarks textually.

“To the genius of Fritz Miller belongs the credit of having
conceived the novel, brilliant, and most plausible hypothesis of
the royal supplemental pairs in the Termite kingdom, an hypo-
thesis supported partly by the observations of other writers,
and partly by his own.!

“Tn Termes lucifugus, a South European species, Lespés
discovered two kinds of nymphs, which he called respectively
the nymph of the first and the nymph of the second form.
The former was more active and slender, and possessed long
and broad wing-rudiments, which completely covered the base
of the abdomen; it became a perfect insect and left the nest
from about the 15th to the 20th of May. The nymph of the
second form was much rarer, its abdomen was larger and
heavier, and the wing-rudiments were short, narrow, and
situated at the sides of the thorax. When first found in
February the latter nymphs were equal to the others in
length (6—7 mm.), but subsequently exceeded them (8—10
mm.), simply through enlargement of the abdomen, especially
in the female. The abdominal tergites were then insufficient
to cover the sides of the body, and were distinctly separated
from each other by soft membranous interspaces. In fact, a
dilation of the abdomen had taken place which corresponded
with a much greater development of the testes and ovaries
than that existing in the nymphs of the first form. Thenymphs
of the second form remained unaltered till July, when they
turned brown, but they gradually became very much rarer.
Although Lespés did not continue his observations beyond that
month, he supposed that these nymphs were metamorphosed in
August into winged males and females, and swarmed like those
of the firstform. From the latter he derived the small kings
and queens occasionally found in the nests; from those of the
second form he derived the large kings and queens. This

1 €Jen. Zeitschr.,’ 1873, Beitr. iii.

38 B. GRASSI AND A. SANDIAS.

supposition was based on the lesser development of the gonads
in the small king and queen, just as in the nymph of the first
form, and on their much greater development, as in the
nymph of the second form, in the large king and queen, or
as he terms them simply, the king and queen. But these
degrees of development may be explained, as Miiller remarks,
simply by differences of age and of the seasons when the
observations were made.

“Moreover Hagen and Miiller observe that the royal forms
possess wing-stumps, and this presupposes a degree of deve-
lopment of the wings which appears unattainable by the
nymph of the second form, of which the wing-buds are still
very short in July. To the writers quoted must be added
Bobe-Moreau,! who studied a species, presumably Termes
lucifugus, in South Europe, but failed to observe the second
swarming conjectured by Lespés to take place.

“ According to Miller, the nymph of the second form
remains wingless and never abandons the nest, in
which he believes it to become sexually mature
under certain circumstances. He adds that fertile
examples with the appearance of a nymph have been described
as the queen iu different species, not only in Termes luci-
fugus (Joly),? but in T. flavipes, T. arenarius, and Calo-
termes flavicollis (?). Miller does not believe that the
swarming of Termitide can lead to the foundation of new
nests; whilst he does not positively (geradezu) deny the
possibility of this happening in Calotermes, he absolutely
excludes it for all species of Termes, Hutermes, and Ano-
plotermes, studied by himself. The effect of swarming in
his opinion is simply to provide royal pairs for unoccupied
thrones, and the colony must avoid the enormous amount of
labour and the serious consumption of its members entailed
by swarming, and be certain of possessing a king and queen by
the retention in the nest of a royal pair produced therein.
Now this pair is the offspring of the same parents, inasmuch

1 [* Mémoire sur les Termites observés 4 Rochefort, &.,’ 1843. |
2 |«Mém. Ac. Sci. Toulouse’ (3), v (1849), pp. 1—87, pis. i—iii.]

CONSTITUTION AND DEVELOPMENT OF TERMITES. 39

as a nest usually contains but a single royal couple, and their
interbreeding must therefore vitiate the stock. Swarming
brings about the intercourse of examples from different nests,
and diminishes the evils of interbreeding by leading to the
pairing of non-consanguineous royal forms. This, therefore,
is its proper function.

“Tn the realisation of this advantage it may too easily
happen that the orphaned population is unsuccessful in pro-
viding a fresh royal couple for the throne. In this case the
royal substitute pairs, or sexually mature nymphs of
the second form, step in and thus ensure the safety of
the colony. Their slow development is correlated with this
function, and their reduction in number during July may
possibly indicate that they are killed off when no longer
required, and that the colony does not keep alive the large
number originally provided.

«This hypothesis was supported by the following observation
made by Miiller himselfin Brazil. He found in the solid core
of a Eutermes nest no less than thirty-one substitute queens,
which were seen to lay eggs, together with a single true king,
possessing wing-stumps; a true queen was wanting. These
supplemental queens bore a general resemblance to workers,
but were twice as large; their wing-buds were mostly very
short (about half the length of the segment from which they
took origin), but were markedly longer in avery few examples.
The antennz were fourteen-jointed as in the workers (those
of the soldiers having thirteen, of the perfect insects fifteen
joints). Their head might be taken for that of a worker,
except for the presence of small pigmented compound eyes.

‘These are the whole of the particulars given in Fritz Miiller’s
paper. Hagen, quoted by Miiller, takes a different view, and
believes that all African and Asiatic queens originate from
perfect insects, and those in America directly from nymphs.

* Latterly von Jehring (in Brazil) has published two notes
on alternation of generations in Termitide.! He regards the
substitute queens, found on a single occasion by Miller, and

1 ‘Ent. Nachr.,’ xiii (1887), pp. —4, 179—18y.

40 B. GRASSI AND A. SANDIAS.

never by himself, as abnormal forms (oviparous workers), like
the soldiers with rudimentary wings described by Hagen, and
as of no importance in the termite economy. He considers
that the nymphs of the second form found in Termes luci-
fugus are to be explained either as the result of seasonal
dimorphism, or as belonging to a species inquiline with the one
to which the nymphs of the first form belongs (he claims to
have observed the latter phenomenon in many American
species !).

“ Miller! has declared von Jehring’s criticism to be insuffi-
cient, though without advancing any fresh facts, and they do not
appear to me to deserve further consideration.”

From these quotations it will be seen that Hagen and Fritz
Miiller have endeavoured to correct the grave errors into which
Lespés fell; and Miiller in particular has nearly reached the
solution of the problem of the nymphs of the second form on
theoretical grounds, but has confined himself simply to sug-
gestions.

The principal new points in the present work of Dr. Sandias
and myself are the following :

“1, Numerous fully-winged examples emerge annually from
the nests of Calotermes flavicollis and Termes luci-
fugus. A certain number of those belonging to the former
species succeed in founding new colonies.

“Those of Termes lucifugus are all irretrievably lost in
nature, at least in Sicily. This is shown by researches ex-
tending over some seven years.

«92. The males and females are accustomed to swarm sepa-
rately, and consanguineous pairing is thereby rendered difficult.

«3, A certain number of winged Calotermes settle after
swarming upon decayed spots on tree-trunks.

«There they skilfully get rid of their wings if not previously
assisted by chance, and then begin to burrow into the decayed
area. At this time the sexes meet and pair, and each pair
begins to found a fresh colony.

1 «Ent. Nachr.,’ xiii (1887), 177-8.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 41

“The so-called amatory passages are not exhibited by
Calotermes; they are seen in Termes, and possess no sexual
significance, but are simply attempts to solicit dejecta.

“The pairs which begin to form fresh colonies have the
antenne mutilated; this is probably effected reciprocally on
each other. Furthermore, these organs are never found intact
in any royal example, true or neoteinic, of either Calotermes
or Termes.

“4, Termes and Calotermes communicate among them-
selves chiefly by a jerking convulsion of the whole body, which
may be accompanied (in the soldiers of T. lucifugus) by a
readily perceptible crepitus, produced by friction between the
head and pronotum.

“The tibial organ, discovered by Fritz Miller, is tympanic
and most probably auditory. As far as we can make out,
Termites hear the sounds produced by the convulsive movements.

“Members of the same nest recognise each other.

“5, The food of Termitidz consists of—

**(1) Triturated particles of dead or decayed wood ;

(2) The material disgorged by their fellows, and composed
of triturated wood mixed with saliva ;

** (3) The dejecta of their fellows; this constitutes the
favourite food of these insects, which show great skill in
soliciting it from each other ;

“ (4) Dead examples of their own or other colonies, but of
the same species,—moribund, healthy, but superfluous indi-
viduals (superfluous royal substitutes or soldiers, &c.) ;

“‘ (5) The salivary secretion of their fellows (a transparent
alkaline liquid).

“Termites also imbibe water.

“6, The colony can modify the development of a certain
number of individuals which are destined normally to become
perfect insects, by varying the proportion and quantity of their
nutriment. It thus obtains workers, soldiers (which may be
considered as further differentiated workers), and neoteinic
forms, which may or may not exhibit special characters, such
as long sete. The neoteinic forms become sexually mature

42 B. GRASSI AND A. SANDIAS.

without fully acquiring the perfect instar, and thus preserve
the facies of the larva or nymph; they consist of substitute
and complementary kings and queens.

“All this may be rigorously proved by the observation of
colonies which have been deprived of the king, queen, or
soldiers, &c.; and the insects can thus be forced to produce
soldiers, workers, or neoteinic forms at will.

«These transformations can take place without limitation
to a specified age in the individuals selected. Soldiers may
be derived from larve and nymphs of different ages, and neo-
teinic forms from these or from perfect insects which are still
white.

“‘ Nevertheless the colony prefers to select examples for
transformation at the ages specified in the present memoir.

«7, The larve and nymphs administer a large quantity of
saliva to examples destined to become neoteinic; it causes the
disappearance of the parasitic Protozoa. The importance of
this disappearance is not clearly understood, but it is certainly
insufficient by itself to produce neoteinia.

“ Newly-born larve receive nothing but saliva, but little
or none is administered later to those in progress of becoming
workers or soldiers.

“8, The colony of Termes lucifugus is headed by
hundreds of complementary queens; the existence of the
complementary kings is precarious. Both appear invariably
to be derived from the nymph of the second form.

“Orphaned colonies of Termes are found after a certain
time to contain substitute instead of complementary queens.
They are frequently derived from entirely wingless larve,
but not seldom from perfect insects, which are more or less
completely white.

“The colony of Calotermes flavicollis is headed by
a royal pair derived from perfect insects. When they are
absent a substitute pair is furnished. Strictly speaking, several
such pairs are provided, but only one survives the ferocious
conflicts and banquets of royal flesh which take place.

“Termes lucifugus readily migrates from one tree to

CONSTITUTION AND DEVELOPMENT OF TERMITES. 43

another, and carries eggs and young with it. But the com-
plemental queens never change their situation, and are there-
fore not to be found in many of the dead trees inhabited by
the species in different stages.

“Tf a detached offshoot established in a tree without royal
forms loses its communication with the main colony, it imme-
diately provides such forms in hundreds. New colonies of
Termes lucifugus arise in this way.

“These phenomena depend entirely on the promptness with
which Termites guard against the want of royal individuals.

*‘Calotermites readily receive strangers of the same species
into a nest, even a royal pair, if they are orphaned. Jealousy
is most conspicuously manifested between the royal forms, but
much less rapidly than in bees.

APPENDIX I.

Tue Parasitic Protozoa or TERMITIDA.
(Vol. 39, Plate 20.)

I propose to complete the account of my investigations on
the protozoa Parasitic in Termitidz by a concise description of
the forms observed,' which all belong to the class of Flagellata.

Calotermes flavicollis harbours two species alone, one
belonging to the family Cercomonadide, Grassi, the other to
the Lophomonadide, Grassi.

In Termes lucifugus I have been able positively to
determine six species, of which two belong to the Lophomona-
didze, two to the Cercomonadide, and the last two to a new
family, Pyrsonymphide, mihi.

Adopting as far as possible the nomenclature employed by

1 Similar parasites have been observed wherever Termites are found. Thus
Leidy (‘ Proc. Ac. Sci. Phil.” 1877 and 1881) and Saville Kent (‘Manual of
the Infusoria,’ ii, pp. 551—556, pl. xxviii) have published incomplete descrip-
tions of some species, and a new form has recently been described by Frenzel
(‘ Arch. f. mikr. Anat.,’ 1891, pp. 301—316). [A paper by Seip (‘Am. Mic.
Journ.,’ ii, p. 288) is not referred to by Prof. Grassi.] These parasitic
Infusoria were indicated in Termes by Lespés.

4,4, B. GRASSI AND A. SANDIAS.

Kent and Leidy in their well-known works, I have employed
the following names for the seven species studied.

Fam. Lophomonadide.
I. Joenia annectens, gen. et. sp., Grassi (in Calotermes
flavicollis).
II. Trichonympha agilis, Leidy (in Termes luci-
fugus).
III. Microjoenia hexamitoides, gen. et sp., Grassi, =
immature Trichonymphaof Leidy (in Termes lucifugus).

Fam. Cercomonadide.
IV. Monocercomonas termitis, Grassi (in both Calo-
termes flavicollis and Termes lucifugus).
V. Dinenympha gracilis, Leidy (emend.); probably =
Pyrsonympha vertens, Leidy (pro parte) (in Termes
lucifugus).

Fam. Pyrsonymphide.
VI. Pyrsonympha flagellata, Grassi (in Termes
lucifugus).
VII. Holomastigotes elongatum, gen. et sp., Grassi
(in Termes lucifugus).

te

JOENIA ANNECTENS, gen. et sp., Grassi.) (PI. 20, figs. 6—9.)
Atti Acc. Lincei (5), p. 36 (1892).

Of relatively gigantic size, usually exceeding 130, in
length and 40 in width. Variable in shape, sometimes
pyriform with the broad end anterior, sometimes constricted
in the middle with the anterior extremity the smaller.

Like Lophomonas, it has a large tuft of numerous fla-
gella at the anterior extremity, is devoid of cytostome or

1 The genus Joenia is dedicated by me to the memory of the distinguished
naturalist Cavaliere Gioeni.

? [Although the genera and species described by Professor Grassi are
indicated as new in the original of the present memoir, the descriptions were
actually published in the preceding year. }

CONSTITUTION AND DEVELOPMENT OF TERMITES. 45

contractile vacuoles, and possesses a large nucleus in the
neighbourhood of the flagellar tuft. It exhibits, however, the
following important differences :

1. Instead of the denser and more opaque protoplasmic
region occupying the anterior half of the body of Lopho-
monas, it possesses a complex endoskeleton, which seems of
cuticular substance, and lies nearly in its long axis. This
consists of—

(a) A rod similar to that found in Trichomonas, and
tapering posteriorly. Anteriorly it is widened and bilaterally
symmetrical, owing to the presence of a recess which lodges a
large part of the nucleus.

(6) Numerous curved and claviform rodlets, which appear to
be inserted into the anterior end of the rod by their lesser
extremities, so as to form an encircling ring, which is incom-
plete, owing to their absence over a small portion. This circle
of rodlets sometimes exhibits a bilateral symmetry, which does
not, however, correspond with that of the rod. At times a
filament may be observed apparently to connect the rod with
the base of the flagellar tuft ; Ido not know whether this should
not be considered part of the endoskeleton.

2. The posterior part of the body is ciliated in Joenia, and
the cilia are never motile.

Joenia feeds on particles of wood, which it ingests in a
manner not yet determined, and which may exceed half its
body in length or bulk. Ingestion and elimination of dejecta
probably take place at the anterior half of the body, except
over the flagellate area, and are accompanied with amcboid
movements.

TE

TrIcHONYMPHA AciLiIs, Leidy (Pl. 20, figs. 1—5.) Leidy,
‘Proc. Ac. Sci. Phil.,’ 1877, p. 147.

Reaching the dimensions of Joenia, variable in shape,
sometimes oval, sometimes transversely constricted into two
unequal portions at the level of the front of the nucleus. The

46 B. GRASSI AND A. SANDIAS.

anterior is the smaller, and may show traces in its turn of
one or two transverse constrictions.

The species is more commonly mammiform (shaped, e. g.,
like the udder of a goat), with a small nipple-shaped anterior
process, which more or less resembles a teat, even when the
general shape is not mammiform, and will be briefly referred
to as the mamilla.

The Protozoon, therefore, exhibits a mamilla and a base
from which it springs and appears to be delimited by an evident
constriction. This mamilla may be curved towards the body
in various ways, or may be spirally twisted.

Most of the varieties of form here described appear to be
constant, and no change of shape is discernible under the
microscope. But this may be due to the unfavorable cir-
cumstances for observation, and it must be left undetermined
whether this diversity of form is dependent on contractile
phenomena or not.

The mamilla and the anterior half of the body are certainly
much more variable in form than the posterior half; and
Trichonympha in general is far more variable than Joenia.
In any case the mamilla is undoubtedly flexible.

Both ectoplasm and endoplasm are distinguishable. The
ectoplasm of the mamilla and the anterior half of the body
may conveniently be spoken of as the striated zone (although
the apex of the mamilla is not striated); it differs from
that of the posterior half, which constitutes the unstriated
zone. The limit between these two zones may be median, or
may be situated at the junction of the anterior third or fourth
part of the body with the remainder.

The striation of the striated zone is longitudinal, and clearly
due to the alternation of shallow sulci and ridges in the cor-
tical layer of ectoplasm; that of the mamilla is coarser and
more remote than that of the body mass. The ectoplasm of
the striated zone usually appears dense and homogeneous, and
its inner layer is traversed towards the front of the body by
circular transverse lines; their meaning is obscure, but they
are probably myonemes.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 47

At the summit of the mamilla the ectoplasm is reduced to a
very thin layer, like a structureless membrane, without trace
of striation : this is more fully explained below.

Passing to the unstriated zone, the ectoplasm of the posterior
extremity appears somewhat different from the rest, and forms
a layer variable in thickness and aspect, but-as a rule not
dense. I suppose that this hinder extremity can execute
amoeboid movements for the purpose of ingesting the food,
which ordinarily consists of more or less bulky particles of
wood, sometimes more than half the length of the Protozoon.

The ectoplasm over the rest of the unstriated zone is dense
and homogeneous, forming a thinner and more granular layer
when the endoplasm is loaded with food particles. Externally
it presents a distinct double outline, which is never distin-
guishable in that of the posterior extremity.

There is no distinct demarcation between the ectoplasm and
endoplasm of the unstriated zone; but the ectoplasm of the
striated zone encloses a clear space, which I regard as a
lacuna filled with liquid or semi-fluid protoplasm. The ar-
rangement of this lacuna will be made clearer by a further
description of the mamilla. Its anterior unstriated extremity
is separated from the rest by means of a transverse diaphragm,
which gives it the form of a skull-cap. This diaphragm is
membranous, and possesses a circular median orifice, which is
occupied by the rounded and closed apex of a cylindrical tube,
situated in the long axis of the striated portion of the mamilla.
This tube does not completely occlude the orifice, but is
separated from it by an annular space.

Under the thin membranous ectoplasm of the apical portion,
which has been compared to a skull-cap, is a clear space, which
I take also to be a lacuna filled with liquid or semi-fluid
protoplasm, like the one previously described. And these
two lacune communicate by means of the annular space.

The walls of the above-mentioned cylindrical tube are also
membranous in appearance, and contain a liquid or semi-fluid
protoplasm, like the lacune. Behind, at the base of the
mamilla, this tube widens out considerably ; its walls become

48 B. GRASSI AND A. SANDIAS.

thinner from this point until they are no longer distinguish-
able, and the contents of the enlarged portion do not remain
liquid or semi-liquid, but change abruptly to a very dense pro-
toplasm, full of fine uniform granules.

This enlargement gives rise to a structure which may be
compared with a bottle, of which the cylindrical tube forms the
neck. The bottom of this bottle corresponds posteriorly with
the limit between the striated and unstriated zones, and is
hollowed out, exactly as in a wine-bottle, to receive the nucleus.

Continuing the comparison, it will be seen from this de-
scription that the neck of the bottle is filled with a liquid or
semi-fluid substance, whilst the body contains a granular pro-
toplasm ; but it fails in the fact that the bottle is devoid of
walls at the part corresponding to the bottom, or rather that
its walls become so attenuated towards the foot as to be in-
distinguishable. It will also be seen that this bottle-shaped
structure is surrounded with the liquid or semi-fluid proto-
plasm of the lacunz except at its base.

The concave base protects the nucleus by reception of its
anterior half. The posterior half lies in a sort of cage of
curved rodlets, somewhat remote from each other and connected
by granular protoplasm. I am not certain as to their exact
arrangement, but they terminate almost in connection with the
periphery of the foot of the bottle, and most, if not all, are
curved in apposition with the posterior half of the nucleus.
I cannot give fuller details, but I may add that I believe them
to be somewhat asymmetrical, and not alike in all individuals.
They appear to consist of very dense protoplasm, and resist the
action of acetic acid for some time.

The nucleus is very large, and oval or rounded. It may
measure as much as 14 or 16 in diameter; when oval its
major axis may measure from 16 to 18 yw, its minor axis 14 yp.
It has a well-marked limiting membrane, and usually contains
numerous small deeply-staining bodies, resembling large, thick,
and more or less curved bacteria.

I have mentioned that nutrition and elimination of fecal
matter most probably take place at the posterior extremity,

CONSTITUTION AND DEVELOPMENT OF TERMITES. 49

and that the endoplasm may contain more or less solid food
material; when this is not the case it is abundantly filled with
granules of different sizes, usually very minute. The apparatus
described at the anterior pole cannot possibly be regarded as a
buccal organ.

The flagella remain for description. They arise from the
striated zone in distinct longitudinal rows, and I believe from
the examination of many preparations that they take origin
from the longitudinal ridges which separate the sulci. Their
number is very large; they are shortest in front, and gradually
increase in length from before backwards, so that the longest
posterior flagella greatly exceed the body in length.

The anterior flagella are variable, and may be directed either
forwards or backwards, whereas the posterior are always
directed somewhat obliquely backwards, so as to describe a
spiral figure round the body: in spite of their length, they
rarely arise from the posterior extremity. The animal’s move-
ments are rapid, and may follow a helical course.

The dense granular protoplasm forming the mass of the
- bottle-shaped structure is very scanty in many examples, so
that the nucleus lies in a more forward position.

III.

MIcROJOENIA HEXAMITOIDES, gen. et sp., Grassi (Pl. 20,
fig. 10). ‘Atti Acc. Lincei’ (5) i, p. 86 (1892).

Of relatively small size, never exceeding 45 in length,
and consequently less than the shortest specimens of Tri-
chonympha agilis. Oval in shape, prolonged behind into
a more or less prominent mucro. Intermediate in characters
between Joenia and Trichonympha.

As in the latter genus, the anterior extremity is destitute of
flagella, and is delimited by a very thin layer of dense ecto-
plasm, beneath which is a clear space or lacuna. The flagella
originate from a striated zone corresponding with that of
Trichonympha. A subaxial skeletal rod can be exception-
ally made out, as in Joenia. The nucleus is in the neigh-
bourhood of the anterior pole.

voL. 40, part 1.—NEW SER. D

50 B. GRASSI AND A. SANDIAS.

Leidy observed the species here briefly diagnosed, but regarded
it, without evidence, as a young Trichonympha.
It also assimilates solid food.

IV.

MonocERCOMONAS TERMITIS, Grassi. ‘ Atti Acc. Lincei’ (5),
i, p. 36.

The anterior pole of this Monocercomonas is furnished
with at least six very long flagella, of which one is directed
backwards and the others forwards. A subaxial skeletal rod is
present, as in Trichomonas. Average length 15. Solid
food is ingested.

ANG

DINENYMPHA GRACILIS, Leidy. (Pl. 20, figs. 11—17.) Leidy,
‘Proc. Ac. Sci. Phil.,’ 1877, p. 148.

Body uniaxial, dissimilar at the poles, elongate, subcylin-
drical or vittate, often -shaped or readily becoming clavate,
the posterior extremity being the larger. Movements usually
very rapid, consisting of alternate flexion and extension in
the long axis of the body. Locomotion usually helicoid in
direction.

The ectoplasm and endoplasm are contradistinguished only
by the absence of solid food-particles in the denser peripheral.
layer. There is no trace of contractile vacuoles or mouth.
The nucleus is situated anteriorly, and is pyriform or clavate,
with the larger part posterior; there is no paranucleus.

Flagella are absent, and their place is taken by delicate
undulating membranes similar to those of Trichomonas, or
more closely, Paramecioides, their flagellar origin being
equally doubtful. The free margin of the membrane is cer-
tainly thickened, but as it is not united to the body by a very
delicate attachment, as in Trichomonas, it never becomes
separated so as to look like a flagellum. These apparently
homogeneous membranes extend from one pole to the other,
sometimes in straight lines, at other times so as to form one
or more spiral turns. When their direction is straight, or

CONSTITUTION AND DEVELOPMENT OF TERMITES, 51

forms a single turn, it is easy to see that there are four, but
I cannot say that this number is constant in all individuals.
They are delicate in some and thicker in other examples.

Lastly, as in Trichomonas, a rod extends backwards from
the anterior extremity without being strictly axial; it is ex-
tremely elastic, and appears to be bent by the rapid sinuations
of the body.

Either or both extremities or the entire surface may also be
furnished with appendages which may be taken at first sight
for flagella, as has been done by Leidy. But after much
hesitation I have satisfied myself that they are parasitic
Spirilla, such as occur free in the intestinal contents of Ter-
mitidz. Their shape, size, inconstant presence, and irregular
distribution all justify this statement.

NAL:

PYRSONYMPHA FLAGELLATA, Grassi. (Pl. 20, figs. 18—20.)
‘ Atti Acc. Lincei,’ (5), i, p. 36.

This and the following form are assigned to a new family,
the Pyrsonymphide, characterised by the body covered with
flagella, disposed more or less exactly in spiral lines, the
situation of the nucleus towards the anterior extremity, the
absence of a paranucleus, mouth and contractile vacuoles,
and lastly by the elliptic monaxial body, with antero-posterior
asymmetry, moving, at least as a rule, in a helicoid path.
In the family I distinguish two genera, each with a single
species: Pyrsonympha flagellata, Grassi; and Holo-
mastigotes elongatum, Grassi. lLeidy considers these
forms to be young Trichonymphe, a view I absolutely
reject.

Pyrsonympha flagellata, which will first be dealt with
in detail, is subelliptical, and may measure as much as 98 in
the major and 40 u in the minoraxis. The anterior extremity
terminates in a more or less evident papilla; the posterior is
variable, and may be rounded or produced, &c.: it probably
serves for ingestion and elimination, as in Trichonympha.
Nutrition is effected as in that genus.

52 B. GRASSI AND A. SANDIAS.

The surface of the body is covered with flagella about 18
in length ; if it is viewed lengthways, the flagella are seen to
be arranged in oblique lines which run in opposite directions
on the two faces, as an alteration of the fine adjustment shows.
Appropriate observation suggests that they are disposed uni-
seriately in spiral lines, but their mode of origin at the anterior
extremity and the number of turns made by each line appear
impossible to determine, on account of the close juxtaposition
of these turns over the anterior papilla. The spirals may or
may not extend to the posterior extremity, and the lines of
origin of the flagella appear subelevated throughout.

The ectoplasm differs from the endoplasm only in its greater
density and freedom from food-granules.

In connection with the anterior extremity is a tubular
orgav, as in Trichonympha.

All the larger examples possess five or more rodlets, which
may be somewhat bent at the level of the nucleus, as if to
embrace it; they converge in front of it and lie in close proxi-
mity to each other, being disposed according to a slightly
curved surface, and thus giving the animal a bilateral
symmetry.

VI.

HoLoMASTIGOTES ELONGATUM, gen. et sp., Grassi. (PI, 20,
figs. 21—24.) ‘Atti Acc. Lincei’ (5), i, p. 36.

Reaching a length of 60 u, and a corresponding width of
20 to 24 4. Oval, very elongate, with the posterior extremity
usually the more attenuated. The anterior papilla wanting;
the flagella arranged as in the preceding species in spiral
lines, easily seen, during contraction of the body, to start from
the anterior pole. During elongation other spiral lines, which
do not extend to the anterior extremity, are seen to be inter-
calated between those bearing the flagella.

Ectoplasm as in the preceding species: the endoplasm con-
stantly contains very numerous, scarcely refringent granules,
which might be taken at first sight for nuclei; they disappear
with acetic acid, and do not stain. Similar granules may

CONSTITUTION AND DEVELOPMENT OF TERMITES. 53

sometimes be found in the endoplasm of Pyrsonympha:
their nature is unknown.

GENERAL NotTEs.

It seems incredible, even to myself, but it is nevertheless
true that, in spite of long and fruitless search, I have
never been able to find any of these Protozoa encysted, or
evidently in process of reproduction.

On a single occasion I found two examples of Tricho-
nympha agilis fused together by the hinder extremities.
They separated under my eyes after being united for an in-
stant by a slender thread of hyaline protoplasm. This thread
became much attenuated in the middle, and then broke,
leaving each of the examples with a peduncle which was quickly
retracted. They were of medium size, and both belonged to
the form in which the nucleus is situated far forward.

This may have been either a casual act of fusion, brought
about perhaps by the method of preparing for examination, or
a process of division. But never having been able to find any
other example in similar circumstances, or in any other stage
connected with reproduction, I must adopt the former suppo-
sition.

In Dinenympha gracilis I have observed a condition
which may be connected with reproductive phenomena.

Motionless examples are common, with or without food-
particles, sometimes full of vacuoles and with the marginal
membranes thrown into folds and quiescent, or even absent.
The nucleus and skeletal rod appear to be unchanged.

These examples are attached to the intestinal epithelium,
like Gregarines, by a kind of cuticular peduncle arising from
the anterior extremity.

It was not till I had completed these observations on the
Protozoa that it occurred to me that they must be ingested in
large quantities by the Termitide with the feces; they may
possibly pass without encystment into the stomodeum and

54 B. GRASSI AND A. SANDIAS.

chylific ventricle, or may perhaps multiply therein before
entering the proctodeum.

I propose to return to this point in a future work, in which
I shall deal more particularly with the minute structure of the
protoplasm and nucleus in these Protozoa.

The various rods and rodlets of these parasitic Protozoa, as
well as the dense and granular protoplasm of Trichonympha,
evidently serve as an endoskeleton for the support of the body
and protection of the nucleus. This is shown by their entire
absence in Holomastigotes, the only form in question that
does not take in solid food-particles (fragments of wood), which
would seriously endanger the nucleus.

The special apparatus at the anterior extremity of Tricho-
nympha allows it freely to thread its way through the
surrounding swarms of its own and other species without
compression of the nucleus, to which the mamilla acts as a
sort of buffer. I have sometimes thought that the structure
in question may have the further function of a sucker, by which
the animal can attach itself to the intestinal walls. The
papilla of Pyrsonympha is probably similar in function.

As I have previously said, all these parasitic forms must be
classed among the Flagellata, and cannot be referred to the
Ciliata, chiefly owing to the absence of a micronucleus (para-
nucleus).

The Lophomonadide, in which a large area of the body
is covered with flagella, evidently form a family osculant
between the other Flagellata and the Pyrsonymphide, in which
the entire surface is thus covered.

The presence of the endoskeleton is the principal reason
which leads me to associate all these forms with the group of
Flagellata.

Catanra; October, 1890.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 55

APPENDIX II.
CoNTRIBUTIONS TO THE Stupy OF THE EMBIID&.
(Vol. 39, Pi. 19).

The Embiidz have been consigned by systematists to a
position in the zoological scale scarcely a degree lower than that
occupied by the Termitide in the Corrodentia. If this is
their position, they might be expected at least to manifest
something of the extraordinary perfection of qualities ex-
hibited by the Termitide, and for this reason I decided to
include them in my studies.

A preliminary survey of the literature on the subject, which
is fully given in Dr. Hagen’s “ Monograph of the Embidina,”?
at once showed me that almost nothing was known about
either the anatomy or the biology of the Embiidz, and I was
therefore compelled to make original investigations. These,
as will be seen, force me to conclude simply that the family
forms a separate branch of the Orthoptera (s. lat.) of uncer-
tain position, and in any case without direct relationship to
the Termitide. I propose, therefore, briefly to epitomise the
results of my studies, and to limit myself to a sketch of
the anatomy and biology of the Embiide.

The species I have investigated is widely distributed in
Italy, and is probably, as I shall show later, Embia solieri,
Rambur.2. Hitherto the larva alone of this species has been
known very imperfectly; and that it has even been doubtful
whether it acquired wings or not.®

The external features of the species in question will first be
dealt with.

1 Canad. Ent.,’ xvii (1885), pp. 141—155, 171—178, 190—199, 206—229.

2 [* Hist. Nat. des Insectes—Neuropteéres,’ p. 313, No. 4.]

3 [The fact that the larva only of E. solieri has been previously recorded

from various parts of Europe appears to have depended on the unfounded
hypothesis that the insect must necessarily be winged in the perfect state. ]

56 B. GRASSI AND A. SANDIAS.

The adult female attains a maximum length of 12 mm.;
the adult male is usually a little smaller; neither possesses
the slightest trace of wings.

The adult male is a dull ferruginous-brown, except the
head, which is sometimes lighter; the flanks, which are
markedly lighter; and the prothorax, which is reddish-yellow.
The legs are concolorous with the prothorax, the dorsum of
the posterior femora alone being red-brown. The antenne
and cerci are ferruginous.

The adult female is similar in colour to the male, differing
only by the lighter ferruginous tint of the dorsal and ventral
surfaces, though the former is sometimes as dark as in the
male.

The larva is ferruginous above except for the prothorax, which
is more clearly yellow, and sometimes the head, which approxi-
mates to the prothorax in colour, and the abdomen, which may
be bright castaneous. The ventral surface is always lighter ;
the antenne, legs, and cerci are testaceous or ferruginous.

Both adult and larva are pilose, with sparse and more or less
long hairs ; some longer setz are present, especially at the side
of the body. The basal joint of the cerci is clothed at the
sides with long, fine, vibratile hairs, and at least one such hair
is present at the middle of the distal joint on its dorsal side.

The body is somewhat flattened. The head is nearly hori-
zontal, flattened and subhexagonal, with rounded angles, and
the Y suture is conspicuous in the larva. The eompound eyes
are pyriform, scarcely prominent, and situated anteriorly in
approximation with the anterior angles of the hexagon; each
consists of more than thirty facets. Ocelli are absent. The
antenne are twice the length of the head, filiform, and
inserted in front of the eyes. Their joints are rather stout
and elongate, cylindrical towards the base, subelliptical to-
wards the apex ; the basal joint is markedly broader and often
a little longer than any of the others ; the fourth is generally
the shortest, the fifth a little longer than the fourth, the
second than the fifth, but occasionally it is the shortest of all.

The remainder are relatively long and differ little from each

CONSTITUTION AND DEVELOPMENT OF TERMITES. 57

other, but careful comparison shows the last to be a little
longer than the penultimate, and the third to be very variable.
The antenne, which do not differ in the sexes, are very fragile.
The maximum number of joints in the adult is nineteen, and is
less in the larva.

[The translator has studied the arrangement of the antenna
joints in a very large species of Embiide from Trinidad,
recently described hy M. de Saussure (‘ J. Trinidad Club,’ ii,
p- 293, 1896) under the name Embia urichi, but really
discovered by Mr. J. H. Hart, F.L.S. Their number does not
exceed twenty-five in any example; they are blackish in the
male, fuscous in the female with pale articulations ; the apical
joints are white, but the last is tipped with black and differs
slightly in shape. Its presence serves to show whether the
antenna is intact at the time of examination. The number
of joints present in the intact antenna may vary, irrespec-
tive of sex, from fifteen to twenty-five, and may differ on
the two sides; when it is small there is an increase in the
length of the individual joints, so that the average length of
the organ does not vary proportionately with their number.
It may, perhaps, be assumed that antenne presenting
unusually few joints have been broken at an early period,
and subsequently regenerated; but it is a curious circum-
stance that the number of white apical joints is roughly
proportional to the whole number present. Antenne of
15—18 joints have only two white apical joints (counting the
last) ; antennee of 19—20 joints have three (rarely four) white
joints ; and antennz of 22—25 joints have four (rarely five)
white joints. So generally is this the case that in one
example the left antenna has twenty-three joints, of which
the apical half of the nineteenth is pale, and the twentieth to
the twenty-third white; and the right antenna has twenty-two
joints, of which the entire nineteenth is pale, and the last three
are white. The blackish (basal) and white (apical) portions
thus possess a certain constant ratio of length to each other
independently of the number of joints, which may vary in
each portion; and it therefore appears probable that if this

58 B. GRASSI AND A. SANDIAS.

variability is due to regeneration, it cannot be explained by
any process so simple as that of proliferation by division of
the third or fourth joints, but that when the shaft of the
antenna is regenerated, its division into a basal dark and an
apical light part must be antecedent to the meristic division.—
Waugh. B.]

The buccal organs are of the Orthopterous type. The
epistoma is short and broad, and is connected with the
labrum by a membranous rhinarium. The labrum is
broader than long, rounded in front and furnished beneath
with two rows of strong spines; it possesses a median
suture.

The mandibles are three-toothed, the two posterior teeth
being more or less distinctly denticulate. In the male
imago they are slender, elongate, andcurved. But in
the female, and the larva of both sexes, they are
robust, short, and not curved; and the teeth are
generally stronger.

The inner lobe of the maxille (Pl. 19, fig. 8) is somewhat
thin, bidentate at the apex, and furnished on its inner portion
with numerous species like those of the labrum. The outer
lobe (galea) consists of a single joint, which carries, as far as
I can see, a fringe of uniseriate hairs along its margin, and is
supported by a well-developed base. The maxillary palpi are
five-jointed, and somewhat exceed the maxille in length; the
first joint is a little the widest, and the last is subconical ; the
first, fourth, and fifth are nearly equal in length, and a little
longer than the second and third, which are subequal. The
palpiger is scarcely distinguishable. The stipites of the
labium (PI. 19, fig. 7) are fused together in the middle line,
which merely presents a hairless sulcus. There are two pairs
of lobes, the inner being very much the smaller. Both pairs
are separated from the stipes by an evident suture. The
labial palpi are three-jointed, the apical joint being much the
longest, and reach, when directed forwards, to the anterior
extremity of the outer lobes; they are carried by a well-
developed base which simulates a fourth joint. The ligula

CONSTITUTION AND DEVELOPMENT OF TERMITES. 59

is well developed, somewhat conical, and flattened at the tip.
The mentum is subtriangular, the submentum subquadran-
gular; both are broader than long.

The pronotum is narrower than the head, subquadrate, and
deeply sulcate behind the anterior fourth, at right angles to a
shallow median sulcus which extends backwards to the base.
The intersegmental membrane between the pronotum and
mesonotum is well developed, pilose, and somewhat thickened,
except at the anterior and posterior margins.

The mesonotum is broader than the pronotum, subquadrate
with rounded angles. The intersegmental membrane between
it and the metanotum is very similar to the previous one.
The metanotum is subrectangular, as wide as, but shorter than
the mesonotum, with rounded angles.

The thoracic sterna exhibit various transverse and oblique
sulci. The mesosternum is longer than the prosternum, and
the limits between the metasternum and the first abdominal
segment are not clearly recognisable.

The legs are long and robust, the middle pair being the
least developed. The respective pairs are equidistant from
each other, and arise from the hind part of their respective
segments ; -the insertions of the anterior and middle legs are
remote and almost lateral; those of the hind legs are some-
what approximated. In all the coxa is very short and conical ;
the trochanter is short, narrow, and subcylindrical, and the
femur laterally compressed.

The anterior femur, tibia, and first tarsal joint are of nearly
equal length. The latter joint (Pl. 19, fig. 9) is enlarged,
and possesses a curved sulcus running from the middle to the
apex of its upper surface; below it is furnished with a
very large number of small spines, and a smaller
number of setw, tubular and slightly curved at the
tip, which sometimes exhibits a minute drop of trans-
parent secretion. The second and third tarsal joints are
comparatively small ; the former is widened and subtriangular,
the latter narrow, cylindrical, and inserted on the upper side
of the second towards its apex. The second is armed below

60 B. GRASSI AND A. SANDIAS.

like the first with spines and sete; the third is furnished with
two claws, and has no plantula.

The femur and tibia of the middle legs are subequal; the
latter is furnished beneath, towards its apex, with peculiar
short spines. The first tarsal joint (Pl. 19, fig. 10) is slender,
cylindrical, shorter than that of the anterior pair, and fur-
nished below with numerous spines similar to those of the
tibia, and with a bare! colourless papilla at its apex;
the second is very short, subcylindrical, and similar be-
neath to the first; the third is about twice as long as the
second, but markedly shorter than the first, and possesses two
claws as in the anterior legs.

The hinder femora are stout and longer than the tibie.
The tarsi (Pl. 19, fig. 11) resemble those of the second pair,
except that (1) the basal joint is more robust; (2) it is
shorter in comparison with, and therefore nearly equal to the
third ; (8) it has a median ventral papilla similar to the one
near the apex; (4) the second joint is not spinose below, but
the base of the distal papilla is furnished, as far as I have seen,
in the larva and the adult female with small spines on its inner
side, which are absent in the male; (5) the claws are more
robust. '

The abdomen is much longer than the thorax, and its
tergites are better developed than the sternites; the former
are ten in number, the last three being the smallest.

The tenth tergite of the female and of the larva of both
sexes is triangular and very broad at the base. In the adult
male it is produced backwards into two asymmetrical apophyses
[titillatores], a right and a left (Pl. 19, fig. 67), the former
being the larger. As the smaller process is entirely covered
below by the penis, it is not easily made out, except at its apex,
which is spirally twisted.

The cerci are bi-articulate, symmetrical in the adult
female and larve, asymmetrical in the adult male, in which

1 With high magnification the base of the papilla shows a few minute
spines.
2 [In this figure the drawing is reversed; it is so in the original plate. ]

CONSTITUTION AND DEVELOPMENT OF TERMITES. 61

the left cercus is excavated on its inner side for partial recep-
tion of the penis (vide infra).

Nine sternites, the second to the tenth, can be made out,
but the tenth is scarcely distinguishable in the adult male. It
exhibits the anal fissure in both sexes. Sexual appendices
are wanting inthe female. The male imago possesses
a long, conical penis, traversed by an internal canal
(Pl. 19, fig. 6); it arises from the ninth sternite
and is prolonged on to the tenth, beyond which it
projects some distance. It is asymmetrical in
position, and directed to the left so as to terminate
under the left apophysis of the tenth tergite and
in the cavity of the left cercus. The right apo-
physis can be directed downwards and to the left,
so as to be flexed under the penis.

Probably the excavation of the cercus and the
two apophyses serve to fix the penis during the
act of coitus, the left apophysis acting above, the
right below, and the cercus laterally.

From the external characters I pass to a description of the
habits of the Embia.

This insect is tolerably common at Catania, and inhabits
localities which are neither very moist nor very dry; it is
particularly fond of very stony ground. In some places it
occurs singly, in others eight or ten together may be found
under a stone.

From November to May they remain under stones in silken
galleries constructed by themselves, much longer than their
body and ramifying. These galleries are attached by means
of threads or small webs to vegetable rubbish (dead grass or
dry leaves), or to stones on the ground; every gallery may
exhibit one or more exit-holes, and those of different examples
living together under the same stone communicate with each
other.

From May to July they live in similar galleries, but at a
depth of 10 to 15 cm. below the surface of the ground, so as

62 B. GRASSI AND A. SANDIAS.

to avoid the dryness of that part. They do not dig in order
to bury themselves, but make use of natural cracks caused by
drought.

In order to gain a clearer idea of the galleries and of the
mode of life of these animals, I kept a certain number in
a corked glass jar a third full of earth and vegetable rubbish,
and in this they flourished for several months.

They constructed galleries in the middle of the contents_or
between them and the inner surface of the jar, most of which
below the level of rubbish was spun over with threads or small
webs of silk, with occasional galleries. They generally remain
motionless in these, and probably emerge only for pairing,
feeding, defeecation, or to escape their enemies.

If an Embia is put into a glass jar with a little earth it
runs about in search of a suitable situation, either in a corner
or against a lump of earth; then in two or three minutes it
begins to cast out silk threads in order to make its gallery.
This is sketched out in about a quarter of an hour, but takes
from twelve to fifteen hours to complete. The Embiaclearly
accomplishes the task with its fore-legs, standing still
on the chosen spot and moving them in the most
varied directions, downwards, sideways, forwards,
and backwards, always with great rapidity. Fre-
quently it alternates the work of one leg with that
of the other, but at times both work together with
intervals of rest.

The silk produced is seen under the microscope to consist
of a network of subcylindrical fibres, which appears to the
naked eye as a thin semi-opaque white membrane.

Here and there are scattered silk threads of variable length
and thickness, which serve to fix it in position. These threads
can often be seen to merge gradually into the network, of
which they are clearly a modified form; under the microscope
they are found to be composed entirely of fibres. The com-
ponent fibres of the silk are more or less slender, and vary
from extremely fine fibrils scarcely visible with a high power
to fibres readily distinguishable with low magnification. A

CONSTITUTION AND DEVELOPMENT OF TERMITES. 63

distinct striation indicative of secondary fibrillation is entirely
wanting.

The fibres of the web cross in every direction, but those of
the securing threads are more or less regularly parallel.

The Embia, when moving in its gallery, bends up
the apical joint of the hinder pairs of tarsi, in order
that the claws may not come into contact with the
silk, and thus progresses without entangling itself.
This explains the physiological function of the
papille I have described on the second and third
joints of these tarsi; it is by meaus of them that the
feet come into contact with the gallery.

The silk is probably extruded as a liquid, and its structure
and the method in which the insect works lead me to
conclude that it is secreted in the anterior legs, which,
as will be seen farther on, are furnished with strongly de-
veloped glands.

As an additional proof of so unusual and unexpected a
source for the silk, I made the following experiment.

Ten intact Embie were put into a glass vessel, and ten
more, whose anterior legs had been very carefully cut off, into
a similar one. The unmutilated examples all began to pro-
duce silk in half an hour to an hour, and had completed and
were comfortably living in their galleries at the end of twelve
or fifteen hours. The amputated Embiz lived and managed
to hide themselves between particles of earth, but produced no
silk. If the silk issued from the head, as one would theoreti-
cally suppose, some threads at least should have been found
in spite of the amputation.1

1 [The translator has kept examples of the Trinidad Embia urichi alive in
England for some time. Unfortunately they were sluggish, and would not
feed on anything that could be procured for them, but the little that was
observed of their web-producing methods was in accordance with Prof.
Grassi’s account. Mr. J. H. Hart, F.L.S., by whose courtesy the specimens
were obtained, was incredulous as to this origin for the silk, but though he
observed the species under more favorable conditions he has communicated
no facts antagonistic to Prof. Grassi’s observations. It may be added that
the salivary glands of E. urichi are apparently far too small to produce the

64 B. GRASSI AND A. SANDIAS.

The galleries obviously serve as a protection to the body
against excessive loss of moisture, or to keep the insect sur-
rounded by not too dry an atmosphere. They are made both
by the young and adults.

The Embiz all become adult about the middle of June.
Pairing takes place about the end of the month, and they
probably deposit their eggs a few days later, dying perhaps in
the course of the summer.

I imagine that they then die, as this would explain why
young ones only, at most from 7 to 9 mm. in length, were to
be found from November to June. Unfortunately my obser-
vations were interrupted from July to October. That they can
acquire wings is entirely out of the question. In fact, at the
end of June I have found impregnated females with the recep-
taculum full of semen, and others which had already laid eggs,
and at this time neither sex exhibited the slightest trace of
wings.

The Embiz cannot jump, but run actively forwards or
backwards, and, unlike Thysanura, can climb with ease up a
glass surface.

They feed on vegetable matter, but perhaps do not disdain
small arthropods (this I have been unable to determine).!

I now pass to their anatomical characters.

The chitinous cuticle is mostly thin and soft, especially in
the larva, and its external surface is furnished with numerous
small points. The hypodermis is pigmented.

There are ten pairs of stigmata (Pl. 19, fig. 2) as in the

necessary amount of silk, and there are no other glandular structures in the
body except those of the anterior legs to which its origin can possibly be
attributed. ]

1 [No observations in England on the food of Embia urichi were success-
ful; but, according to Mr. Hart, the species probably feeds largely on bark,
lichens, or fungous growths, and also on Coccide. Specimens of the latter
on leaves which accompanied the insects appeared clearly to have been eaten;
the contents of the alimentary canal in preserved examples were mainly of
vegetable origin, but contained nothing of which the structure was definitely
to be identified. ]

CONSTITUTION AND DEVELOPMENT OF TERMITES. 65

Lepismide, and the respiratory apparatus is therefore holo-
pneustic. The anterior pair is situated latero-ventrally: seen
from above, they lie on the line of junction between the pro-
notum and mesonotum; seen from below, they lie in front
towards the hind margin of the prosternum. The second pair
is also latero-ventral; seen from above they appear to lie be-
hind the anterior margin of the metanotum, and from below
behind the anterior margin of the metasternum. Further, the
position of these two pairs varies with the degree of contrac-
tion of the body. The remaining pairs are also lateral, but lie
nearer the dorsal than the ventral surface; the line con-
necting any given pair would cut the tergite at the junction
of its anterior fourth with the remainder.

The trachez are very numerous, without any dilation.
They anastomose by longitudinal and transverse branches,
which are repeated in most of the abdominal segments. The
transverse branches are ventral and unpaired ; they unite the
tracheal system of one side with that of the other, or rather
form a transverse canal between each pair of bronchi (as I call,
with many writers, the trunks which originate from the stig-
mata), e. g. between the sixth right and the sixth left bronchus.
Similar anastomoses are met with in the Lepismide, Ter-
mitide, &c. The remaining anastomotic branches are paired
and longitudinal, and of two kinds, ventral and dorsal. The
former serve as a communication between the transverse
anastomotic branches, their anterior terminations lying more
laterally (externally) than their posterior. The dorsal branches
connect the dorsal transverse tracheal branches, and form
longitudinal dorsal trunks, such as exist in the Lepismide.
These anastomoses will be more clearly understood by observing
that each stigma (e.g. the sixth right) gives rise to a single
bronchus, which turns forward, and after a short course gives
off a branch to form the transverse ventral anastomosis with
the corresponding branch of the opposite side (continuing the
illustration, from the sixth left bronchus). A little in advance
the same bronchus gives off a trunk which runs transversely to
the dorsum, and furnishes branches of the second kind, forming

vot. 40, PART 1.—NEW SER. E

66 B. GRASSI AND A. SANDIAS.

the longitudinal dorsal anastomoses. Thus the transverse
anastomosis is formed by a branch springing directly from the
bronchus, whereas the longitudinal dorsal and ventral anasto-
moses, on the other hand, are formed by branches derived
indirectly from the bronchi. Similar anastomoses are found
in other Orthoptera (s. lat.), including the Termitide.

On one occasion I thought I had found a second dorsal
longitudinal anastomosis, but I have been unable to detect it a
second time, and think I must have been mistaken.

The general arrangement of the thoracic trachez is a little
different from that in the abdomen, as can be partly seen in
the plates attached to this work. Two large trunks on either
side, as in Thysanura, run to the head.

The anterior pair of stigmata are, as usual, larger than the
others, and the cuticular margins are obviously more developed,
so as to act perhaps as a sort of valve. I have been unable to
detect anything of the kind in the other stigmata. The spiral
thread of the trachee is very distinct.

I have found nothing of importance in the vascular system.
The blood-corpuscles are usually subelliptical and crowded
with shining granules.

The nervous system exhibits, as usual, a supra-cesophageal
ganglion, a ventral ganglionic chain (Pl. 19, fig. 2), and a
stomato-gastric system; the sympathetic is not distinct. The
supra-cesophageal ganglion possesses the usual median constric-
tion ; the optic lobes are small, lateral, and give off a fine and
rather long nerve to the compound eyes. The ventral gan-
glionic chain presents a sub-csophageal, three thoracic and
seven abdominal ganglia. The three thoracic ganglia are large,
and correspond with the respective thoracic segments. The first
six abdominal ganglia are moderately small, the seventh is
larger. The first corresponds with the segment médiaire;
the second, third, and fourth with the second, third, and
fourth abdominal segments, the latter ganglion lying in the
posterior part of the segment; the fifth segment has no
ganglion, so that the fifth, sixth, and seventh ganglia lie
respectively in the sixth, seventh, and eighth segments. The

CONSTITUTION AND DEVELOPMENT OF TERMITES, 67

commissures between the fourth and fifth ganglia are rela-
tively very long.

There is a small frontal ganglion connected as usual with
the supra-cesophageal ganglion. From it there comes off an
azygos nerve which turns back to run along the esophagus ;
along it is a series of small ganglia, some before the point
where the csophagus begins to dilate, another in the first
thoracic segment where it is somewhat dilated, and lastly, one
in the third thoracic segment where it is much dilated. This
supra-cesophageal nerve gives off numerous small twigs to the
cesophagus. There are special nerves to the salivary glands,
but I have not been able to determine their origin. More
minute researches will probably lead to the detection of other
ganglia in the neighbourhood of the csophagus; I simply
suppose them to exist.

The eyes are compound and euconic; there is a retinal
ganglion (ganglionic lamina), The cones and rhabdoms
are very short, and the cornea very thin. Ocelli are absent.

~——* The anterior portion of the alimentary canal (PI. 19, fig. 8)
is very narrow at its commencement, with well-marked longi-
tudinal folds. It begins to dilate from about the posterior
margin of the head, and the folds become less distinct. About
the beginning of the mesothorax it becomes very large, often
after presenting a slight constriction, and the folds are no
longer discernible. It persists in this condition to the seg-
ment médiaire, where it again becomes very narrow by an
almost sudden constriction to finish a little further on. Four
longitudinal folds arise at the point of constriction, and are
continued back over a short and narrow tract, where they are
intercalated with others. The anterior three fourths of this
tract are free, and the hinder fourth is invaginated into the
chylific ventricle. The cuticle lining the dilated portion of the
fore-gut is spinose.

The chylific ventricle is enlarged in front and narrowed
posteriorly. Like the esophageal portion it runs straight, and
is continued to the seventh abdominal segment. Its epi-
thelium is cylindrical, with a striated cuticular margin, and at

68 B. GRASSI AND A. SANDIAS.

the base of the cylindrical cells are interposed scattered
small cells (replacement cells ?). Ventricular cca are
absent.

The intestine is more or less dilated in front, rather nar-
rowed and more curved in the middle (giving a curvature to
the whole region), and dilated again posteriorly (rectum).
The rectum possesses six longitudinal folds (plice rectales)
of large size and rich in glandular cells; each of them receives
the ramifications of a tracheal branch.

The Malpighian tubules are long, with a somewhat ample
lumen filled with solid excrementitious matter. Their number
varies with age, and as many as twenty may be made out in
the adult ; perhaps the number may be still larger. Further
appendages of the alimentary canal are the two salivary glands.
These lie on each side of the thorax, and are lobulated; each
has a special efferent duct dilated towards the middle, the ducts
uniting in front to form a common canal which opens on the
labium. These efferent ducts possess a spiral thread lke the
trachee.

I may here mention the sericigenous glands of the anterior
legs. As I have stated, the lower border of the first tarsal
joint is set with abundant short spines, among which is a row
of less numerous sete, curved at the tip, and perforated by a
very fine canal. These canals open externally, in my belief,
at the somewhat blunt apex of the sete, and extrude from
time to time a drop of transparent secretion ; at their base they
terminate in a complicated manner in multicellular glands,
a number of which fill the greater part of the joint. Tubular
setze are also found beneath the second tarsal joint, but I have
not detected the corresponding glands; perhaps they com-
municate with glands in the basal joint.

Other special organs are the papille of the middle and pos-
terior tarsi. They possess a rather thick cuticle, more or less
clearly divisible into two layers, and marked in places with
distinct striz (pore-ducts?). ‘Threads of silk may often be
noticed adhering to these papille.

The function of these tarsal organs, which show a certain

CONSTITUTION AND DEVELOPMENT OF TERMITES. 69

analogy with the adhesive organs of the feet of many insects,
has already been explained.

I pass to the generative organs. There are two ovaries
(Pl. 19, fig. 4), and each is pectinate, presenting five uni-
lateral ovarian tubes, which, when mature, occupy most of the
abdomen except the posterior segments. In the larva they
lie at a certain distance from each other, in about the fourth,
fifth, and sixth abdominal segments.

The oviducts are long and straight, and join about the
hinder part of the seventh abdominal segment to form an
unpaired very short and wide canal (oviduct-uterus), which is
unlined by chitin. To this succeeds the still shorter and
wider vagina, which has a cuticular lining, and into it opens
a large spermatheca furnished with a similar delicate lining.

This spermatheca lies on the dorsal side of the vagina, and
extends before and behind it; it communicates with its dorsal
wall by a very short duct, also with achitinous lining. Colle-
terial glands are absent. The vulva, like the vagina, is un-
paired, and belongs to the eighth sternite.

Each ovarian tube possesses a very delicate investing mem-
brane, and each ovum is surrounded by a follicle which is
united to its fellows, when the ova have reached a certain size,
by short peduncles formed of small cells, which are a con-
tinuation of those composing the follicle itself.

The ripe eggs are subelliptical, relatively large, with a very
large micropyle, which is developed at the anterior extremity
of the egg as it lies in the maternal body.

The male has five testes on each side, or ten in all, each
consisting of a certain number of capsules, provided with
separate investing membranes (Pl. 19, fig. 5). They lie in
the third, fourth, and fifth abdominal segments. This arrange-
ment in the male and the very similar disposition of the
ovaries in the female may possibly explain the apparent
absence of the ganglion in the fifth abdominal segment. When
mature the testes occupy a large part of the abdomen, and,
like the ovaries, are arranged unilaterally on the vasa de-
ferentia, which are paired, like the oviducts. Each vas is

70 B. GRASSI AND A. SANDIAS.

dilated posteriorly to form a vesicula seminalis; it then be-
comes narrowed again, and opens finally behind into a common
ejaculatory duct. Just before the origin of this duct each vas
receives the discharge of two glandular sacs, which sometimes
contain a transparent secretion, and at other times semen.!
The male genital opening is evidently situated in the larva at
the hinder part of the ninth sternite. In the adult it is less
distinct, as the penis is developed beyond that point by a direct
prolongation of the ejaculatory duct, the external meatus of
which corresponds nearly with its apex.

The spermatozoa, like those of most insects, possess a
readily distinguishable head and an elongate tail.

[The internal anatomy of Embia urichi, with the omis-
sion of the frontal ganglia, the structure of the testes, and the
histological features, which the translator has not investigated,
corresponds precisely with Professor Grassi’s description, ex-
cept in the following points. The stomato-gastric nerve ends
exactly as in Blattide in a triangular ganglion at the back of
the proventriculus, from which two nerves pass obliquely back-
wards under the invagination between it and the chylific
ventricle (no other ganglia are evident along the course of the
cesophagus). The arrangement of the abdominal nerve-cord
is identical. The chylific ventricle is relatively much larger
than in Professor Grassi’s figure, and extends from about the
mesothorax to the eighth abdominal segment. At its anterior
extremity on the dorsal aspect there is a short rudimentary
pouch, which projects forward over the deeply invaginated
portion of the proventriculus, and looks like an imperfectly
developed cecum. The Malpighian tubules vary in number
from about twenty to twenty-six.

The ovaries are identical in structure; they become very
large in the fertile female, and their arrangement cannot then
be made out, as the tubes become coiled in a complex manner,
and may extend even up to the prothorax, from which ripe
ova have been extracted. As many as twenty-six ripe eggs of
equal size have been found in the same female. They are oval,

1 The nature of their contents requires further investigation.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 71

with avery large flattened micropylar extremity, which is placed
obliquely; behind it the egg has a very slight twist, so that
its shape is almost exactly that of the body of a retort which
has been cut off obliquely at its junction with the neck.

The structure of the testes cannot be made out in the adult,
and the larval forms available were not sufficiently well pre-
served.

As Embia urichi is as far removed from Professor
Grassi’s species as any form in this very small family, it may
be inferred that there is no material difference between the
internal organs of the existing known species.—W. F. H. B.]

The systematic position of the form I have studied now
claims consideration. At the present time seventeen species
of Embiide are known (a few appear to be doubtful) ; of these,
fourteen are winged and three wingless. The latter may be
considered in relation to the form investigated. They are
Embia (Oligotoma) antiqua, Pictet, found as a fossil in
Prussian amber; Embia (Olyntha) Miilleri, Hagen, from
Brazil (known only by a single example, probably a female, in
bad condition) ; and Embia solieri, Rambur, found accord-
ing to various authors in the south of France and Spain, but
as yet only in the larval state. I have already stated that my
species is probably Embia solieri, which is in its turn not
clearly distinguishable from Embia antiqua. But further
comparison is necessary before its identity can be indisputably
established. In the length of the antenne and the asymme-
trical hinder extremity of the male my species approaches the
genus Oligotoma, and differs from Embia (s. str.) by the
symmetry of the hinder extremity in the female. It is
remarkable for the sexual difference in the mandibles, which
does not appear to exist in any other species, and I therefore
think it should be referred to a new genus.

[There is very great confusion about the sexual and generic
characters of Embiidz, which Professor Grassi has not
unravelled, owing to his necessarily relying on Hagen’s state-
ments. An analysis of the descriptions, a comparison of all

72 B. GRASSI AND A. SANDIAS.

available species, and his own dissections, convince the trans-
lator that in the known species of Embiide—1, no winged
female is known; 2, the apex of the abdomen is always asym-
metrical in the male; 3, it is never asymmetrical in the
female. The number of described species has now reached
twenty, of which ten are known by winged males alone, mostly
described from single specimens. The only species in which
a winged female has been positively asserted to exist is Hmbia
mauritanica. Lucas! describes a winged imago and a
wingless larva (as large as the imago), and states that he
opened the abdomen of several imagos and found them to be
females. As his very poor description of the ovaries is abso-
lutely irreconcilable with Professor Grassi’s, or with the
writer’s dissections of E. urichi, and as his figure of the
winged form has palpably the asymmetry of the abdomen due
to the presence of the penis and titillatores, it must be irresis-
tibly concluded that he mistook the testes for the ovaries, and
that his so-called larve were presumably the females of the
species. Embia persica, McLach., is said to have a winged
female, but only very doubtfully. On the other hand, wingless
females (which are much rarer in collections than the males)
exist in six described species. In Embia solieri the so-
called wingless larve may have included imagos of both sexes,
and the species is probably, as Professor Grassi conjectures,
the one with which he has dealt. As for the remaining
generic characters, the sexual differences of the mandibles can
be evaluated only when they have been studied in both sexes
of other species; antennal differences are absolutely fallacious
in forms in which the organs vary so excessively in the number
of joints, and the only remaining character of any value
appears to reside in the neuration which distinguishes the
wings of the males in Oligotoma and Embia (with which is
included the somewhat different-looking subgenus Olyntha),
and in the absence of wings in the males of Professor Grassi’s
species, Embia solieri, and (apparently) Embia antiqua.
But this applies to one sex only, and the writer sees no
1 «Explor. Alger.,’ iii, pp. 11I—114; ‘Neur.,’ pl. ili, fig. 2, a—n,

CONSTITUTION AND DEVELOPMENT OF TERMITES. 73

present possibility of separating the females generically. It
may be added that, like the antenne, the neuration can vary
within limits in the same species, and even on opposite sides
of the same individual. Embia urichi, de Sauss., which
has been several times referred to, belongs properly to West-
wood’s subgenus Olyntha.—W. F. H. B.]

The Embiide are very remote from the Termitide, as I
have pointed out in the body of my memoir, and they have no
more definite affinity with the Perlide. Even the much-in-
voked relationship with the Psocide is very problematical ;
the concentrated nervous system of the latter and their tracheal
system (according to researches communicated by Dr. Rovelli)
establish a wide difference between the two families. To sum
up, the Embiide approach the Orthoptera (s. str.) more nearly
than any other group, though they are separated by a series
of important characters. In the present state of our know-
ledge I think that they should be ranked among the Orthop-
tera (s. lat.), as a special sub-order parallel with that of the
Orthoptera (s. str.).

It seems important to note that my species of Embia has
certainly degenerated, a process which has simplified the
organism without bringing out any of the characters which I
regard as primitive in Thysanura (eleven pairs of stigmata,
ocelli with rhabdoms, &c.). Even if we admit that the Thy-
sanura have undergone a certain amount of degeneration, and
that they once have possessed wings, it must always be recol-
lected that they are the degenerate descendants of Orthoptera
which retained certain primitive characteristics, which are lost
in the Orthoptera of the present day, as far as is known at
present.

My species of Embia is degenerated, or rather, simplified
by the loss of its wings. This is probably due to the fact that
the species are inhabitants of hot countries, and at the same
time of situations neither over-dry nor over-moist. As the
lack of a sufficiently warm climate prevents the European
Embiide from acquiring wings before the summer drought

74, B. GRASSI AND A. SANDIAS.

sets in, there is in my belief a precocious maturation of the
generative organs (neoteinia). This interpretation is sup-
ported by the fact that they become mature here in Sicily at a
time when the summer heat is well marked, and before the
ground has become unduly dry.

Catania; December, 1888.

[The transformations of the Embiide are practically un-
known, and the few facts recorded have simply increased the
confusion which has surrounded it. While reserving the details
of the metamorphoses of Embia urichi, it is sufficient to say
here that the females of Embiide are practically ametabolous,
the only essential morphological change during growth consist-
ing in an increase in the number of antennal joints. The male
of Professor Grassi’s species undergoes the changes in the form
of the mandibles and the posterior extremity which supervene
upon the last ecdysis. In addition, the males of the winged
species, such as Embia urichi, are normally hemimetabolic,
and pass through a nymph stage, as in other Orthoptera
(s. lat.) The form described by Mr. McLachlan! as anymph has
been conjectured by Hagen to be a micropterous form. This is
quite erroneous. The chitinous coverings thrown off at the
ecdyses are extraordinarily thin and transparent, and quite
pigmentless, except over the tergal and sternal plates. In the
nymph of E. urichi the pterothece are so extremely delicate
as to be indistinguishable without very careful examination, and
the young wings which they enclose appear quite distinct and
separate as in the imago, but of very small size. But if speci-
mens preserved in alcohol, which causes the body to shrink
away from the integuments, are examined, or if the nymph-
skin is removed intact by a longitudinal incision, like the
ecdysial fissure, the existence of the pterothecal coverings to
the small wings becomes evident. The structure of the thoracic
segments and the mode of attachment of the legs is alike in
both winged and wingless forms, and is typical of apterous

1 «Journ. Linn. Soce.,’ xiii (1877), pp. 383=4, pl. xxi.

CONSTITUTION AND DEVELOPMENT OF TERMITES. 79

insects, “so that the HEmbiide that have wings may be
described as apterous-like insects provided with two pairs of
inefficient wings” (Sharp), the attachment of which to the
thorax is feebler than and different from that found in other
winged insects. In these respects the Embiide must be re-
garded as absolutely the lowest type of winged insect, and the
present writer is inclined to assign them a position intermediate
between the Thysanura and the Orthoptera Cursoria. At
the same time he is fully impressed with the fallacies which
attend any endeavours to establish phylogenetic relationship
between the loosely knit surviving families which make up the
Thysanura, Orthoptera, and Neuroptera, and believes that
these three orders, which have given rise to so persistent a
difference of opinion among systematists, may best be treated
by their combination into a single order, which will be some-
what more than the equivalent of any of the more specialised
orders of insects, and its division into families, without recog-
nition of any larger aggregates.—W. F. H. B.]

N.B.—A second species of Embia, of which I know only
the larva, exists at Catania. Its habitat is similar to that of
the species described, but it is readily distinguished by certain
characters. Above it is uniformly dark red-brown, except for
the intersegmental membranes between the thoracic segments,
which are dull yellow; the antennz, cerci, and the greater
part of the legs are nearly concolorous with the dorsum. The
antenne are twenty-jointed. The body, especially the abdo-
men and the hinder thoracic segments, is relatively larger.
The papille of the second pair of tarsi are furnished on the
inner side with numerous small and easily distinguishable
spines; the same arrangement is repeated in the hinder pair,
except that the second joint has no papilla.}

Catania; June, 1889.

1 Dr. Rovelli has found an undetermined species of Embia at Rome.

1% St ‘J -3f bp
Tee thls aa er ay
7 a ad ae i Hs nag re
cp ’, ee 1ar eine
Re
beet a 34 lin eee ae
ving “Nels Net alten eS
. os eee ‘
a apuied -
ibid Meath me ie
a ee
see 4 am i a

ON THE STRUCTURE OF HYDRAOTINIA ECHINATA, 77

On the Structure of Hydractinia echinata.
By

Margaret C. Collcutt,
Zoological Laboratory, University College, London.

With Plate 1.

AN investigation of the structure and relative positions of
the coenosare and chitinous parts of Hydractinia echinata
was suggested to me by Professor Weldon, for whose generous
advice and assistance I cannot be too grateful.

From my observations on this hydroid I find that the chitin-
ous skeleton is for the most part a continuous irregular crust
attached to some foreign object, and overlaid by a ccenosare
consisting of two layers of ectoderm, enclosing between them
a number of branching and anastomosing endodermal tubes,
which are connected at intervals with the endodermal canals of
the polyps, the upper layer of ectoderm being continuous with
the ectoderm of the polyps.

This view is somewhat similar to that of Strethill-Wright,
who describes the chitinous skeleton as forming a continuous
crust below the ccenosarc, which, according to him, consists
of two ectodermal layers enclosing a single and continuous
layer of endoderm permeated by tubular cavities.

With other observers of Hydractinia the prevalent idea
as to the structure of the colony seems to have been that the
skeleton was tubular, the chitinous tubes enclosing the cceno-
sarc, which also extended as a continuous sheet over the surface

78 MARGARET ©. COLLCUTT.

of the skeleton, and here connected the polyps with one
another.

The earliest reference we have to Hydractinia is that com-
prised in a few lines by Fleming (1) in 1828; this observer,
however, regarded the horny crust of Hydractinia as being
a polyzoon ectocyst, and assigned to it the name Aleyonium
echinatum.

This view of the nature of the hydroid was also held by
Johnson (2) in 1888.

Van Beneden (4), in 1841, first gave the generic name of
Hydractinia to the hydroid, and wrote a paper on the
structure of the egg. He regarded the sporosacs borne round
the sides of the reproductive polyp as “eggs,” but gave no
description of the structure of the colony.

Hassall (5), in 1841, described, under the name “‘ Echino-
chorium clavigerum,” a hydroid found in Dublin Bay
adherent to empty univalve shells, and which is, according to
Allman (18), the “ Hydractinia, sp.” of Van Beneden.
According to Hassall the colony consists of a ‘ polypidom
muricated with rough, spinous papillze about a line in height.
There are numerous indentations on the surface of the polypi-
dom, in each of which the base of a polyp is inserted; this
latter is about a quarter of an inch in height, and is of a white
colour; its head is somewhat enlarged, and is surrounded with
numerous contractile club-shaped tentacula; the number of
these varies considerably, but frequently amounts to between
twenty and thirty. The tentacles are not arranged in any
determinate order, but are variously disposed. Whether the
polyps are separate or united at their bases I am unable to
say.” His accompanying figure represents nutritive polyps
and ridged spines on a reticulated base.

Philippi (6), in 1842, described as “‘ Dysmorphosa con-
chicola” a hydroid from the Bay of Naples; his description
is short, but he recognised the fact that the polyps are con-
nected at their bases.

De Quatrefages (7), in 1848, described Hydractinia under
the name of Synhydra parasites in the most comprehensive

ON THE STRUCTURE OF HYDRACTINIA ECHINATA. 79

paper on this hydroid that had so far appeared. He mentions
two polyps,—reproductive polyps without a mouth, and nutri-
tive polyps with a mouth; but describes the chitin as an endo-
skeleton developing in the substance of the common ccenosare.
He says that in the nutritive polyps the tentacles vary in
number, are distributed in two alternating circles, and are
very contractile. The polyps are all connected with a common
living tissue which is directly continuous with the exterior
layers of the body, and which, towards the edge, only consists
of a delicate pellicle enclosing no solid skeleton. Below the
horny base, and protected by it, there are little anastomosing
tubes, the central canals of which communicate with the
digestive cavity of the polyps.

De Quatrefages is the first observer who describes the
reproductive polyps in any detail.

Van Beneden (8), in 1844, published a second paper, in
which he describes the two sexes of Hydractinia as two
distinct species—H. rosea and H. lactea.

Johnston (9), in 1847, revived the specific name “ echi-
nata,” used by Fleming. He did not accept De Quatrefages’
view of the endoskeletal nature of the chitin; on the other
hand, he conveys no distinct impression as to how the indi-
viduals are connected together.

Strethill-Wright (10), in 1856, published a paper on
Hydractinia, containing the most correct account of the
structure of the colony which had so far appeared. He
describes the corallum (skeleton) as being secreted externally,
and forming a raised network between the spines. The poly-
pary (soft tissue) “invests the corallum and fills up the grooves
‘of its papille, the interstices between its reticulations, and
“the cavities of its hollow spines. It is often absent at the
*‘ summits of the papille. It secretes, renews, and extends the
*‘corallum, gives rise to new polyps, and is the seat of commu-
“nication between the polyps of the colony. . . . I conclude
“that the polypary of this zoophyte consists of a single layer
‘“of endoderm enclosed between two layers of ectoderm. That
“the lower ectodermic layer, as it grows over the shell, attaches

80 MARGARET C. COLLCUTT.

“itself by its colletoderm” (which is, according to Strethill-
Wright, an epidermis replacing the corallum at the growing
extremities of the branches), “and secretes the horny plate of
“the corallum. On this plate, by a further process of secretion
“from the lower ectoderm, the grooved spines are erected.
“That the upper layer of ectoderm is naked over the greater
“‘ nart of its surface, or only covered by a thin epidermis; but
“ occasionally this layer also takes its share in the secretion of
“the corallum, and in that event produces the smooth conical
“spines, the concavity of which it fills.”

Strethill-Wright conveys the impression that the endoderm
of the polypary is a continuous layer, permeated by tubular
excavations, which are connected with the digestive cavities
of the polyps.

With respect to the edge of the colony he says, “On the
less exposed parts of the shell the polypary frequently passes
beyond the papillary corallum as a thin membranous expan-
sion, or breaks up into a loose network of delicate anasto-
mosing tubes...... Propagative stolons are given off by
these “tubes; . .7) 0s a delicate chitinous investment may
also be detected on the creeping tubular fibres, from which the
stolons of Hydractinia take their rise; but I have not
satisfied myself as to its presence on the entire upper surface
of adult polyparies.”’

Two kinds of nematocyst, differing in size, are mentioned
by Strethill-Wright, who is the first observer of the spiral and
tentacular polyps. He describes the spiral polyps as mouth-
less, with rudimentary tentacles and a highly developed
muscular coat, large nematocysts being crowded in the ecto-
derm of the tentacles and the whole body. The tentacular
polyps he regards as always present. “ Their tips are covered
with a dense pavement of the larger thread-cells, and a few
of the same bodies are thinly scattered along their whole
length.”

Van Beneden (11), in 1866, published another paper on
Hydractinia, in which he says that the tentacles are gener-
ally absent in the reproductive polyps. What he formerly

ON THE STRUCTURE OF HYDRACTINIA ECHINATA. 81

described as “eggs” he now designates ‘sporosacs,” and
asserts that the skeleton is external and similar to that of
Campanularia and Tubularia. The reproductive polyps
have no mouth and only rudimentary tentacles. He confuses
Hydractinia with Podocoryne, and distinguishes three
new species of Hydractinia.

Hincks (12), in 1868, also mentions tentacular polyps ; he,
as well as Strethill-Wright, describes fixed reproductive sacs.
He speaks of the basal ccenosarc as consisting of “ a number
of anastomosing tubular stolons closely packed together, and
filling in the tubular orifices of the chitinous skeleton, which
latter appears to consist of a series of tubes laid side by side
on a plate of chitin, and closely appressed one to the other.”

Allman (18), in 1871, published a full account of Hy-
dractinia, in which he affirms that the basal part of the
colony consists of a number of closely approximated chitinous
tubes containing ccenosarc, which consists mainly of endoderm
with only a thin investment of ectoderm, the latter secreting
the chitin. “ At the free surface of the ccenosarcal expansion
its intercommunicating canals are only partially invested by
chitin, this excretion being in the superficial layer of canals
confined to the deeper parts, thus forming open channels in
which the canals are lodged, so that when the soft parts are
removed the chitinous perisare forms on the surface a multi-
tude of intersecting ridges, having between them the channels
which had contained the superficial ccenosarcal canals. Upon
the whole of the free surface, however, the ectoderm of these
canals forms a continuous and very conspicuous layer, having
acquired increased thickness and developed in its substance
abundance of thread-cells. ‘The whole free surface of the
common basal expansion of the colony thus presents an abso-
lutely naked layer of ectoderm. . . . There can be no doubt
that the whole hydrophyton of Hydractinia must be re-
garded as consisting of a set of ccenosarcal, freely intercom-
municating tubes, which haye excreted from their free surface
a chitinous perisarc, and have intimately coalesced with one
another.”

voL. 40, parr 1.—NEW SER. F

82 MARGARET C. COLLCUTT.

An injured part of the hydrophyton repairs the injury by
a network of ccenosarcal tubes invested by a chitinous perisarc,
the meshes of which are ultimately obliterated by the thicken-
ing and coalescence of their chitinous walls. The ectoderm
and endoderm of the hydranths is continuous with the ecto-
derm and endoderm of the basal ccenosarc.

The spiral zooids are long, cylindrical, mouthless hydroids,
with a crown of rudimentary tentacles crowded with large
nematocysts; they have a tubular endodermal cavity, and
possess the power of coiling themselves into a spiral: the
mesoglea fibres are strongly developed. The blastostyles may
have a mouth.

Allman believes that the tentacular polyps are only rarely
present, and regards them as abnormal modifications of other
hydranths.

Weismann (15), in 1883, described the blastostyles and
migration of the sex-cells in Hydractinia. He figures a blasto-
style with a mouth and ciliated endoderm, and mentions the
occurrence of food granules in the endoderm of the upper
region of the body.

Miss Bunting (17), in 1894, published an account of the
origin of the sex-cells and the development of Hydractinia.
She first observes the ova in the endoderm of the blastostyle,
and concludes that they are probably endodermal in origin.

My observations on the structure of Hydractinia echi-
nata have led me to agree to a certain extent with Strethill-
Wright with respect to the relations of the coenosare and
chitin, and to differ from Allman and Hincks as to their views
of the tubular nature of the adult skeleton.

Moreover I have found a dactylozooid with a mouth, the
existence of which has hitherto been overlooked or denied,
while other observations show that there is a migration of
ova between ectoderm and endoderm in the blastostyle.

Colonies of Hydractinia are generally situated on the surface
of shells, which are commonly whelk-shells inhabited by hermit-
crabs. A large colony may cover the whole shell except for
a small roundish patch where the shell rubs along the ground

ON THE STRUCTURE OF HYDRAOTINIA EOHINATA. 83
as the hermit-crab drags its home about; a small colony is
usually situated near the edge of the shell.

A Hydractinia colony is comprised of four kinds of polyps:

1. Gasterozooids, or Nutritive polyps.
2. Blastostyles, or Reproductive polyps.
3. Dactylozooids, or Spiral polyps.

4. Tentacular polyps.

The Gasterozooids are the most numerous of the polyps ; in
the early spring and in the summer, however, the Blastostyles
increase greatly in numbers, and at these times give rise to
the generative products, the colonies being either male or
female.

Round the shell mouth are situated the Dactylozooids,
which are capable of coiling themselves spirally, and may
function as defensive polyps.

Strethill-Wright (10) and Hincks (12) also mention “Tenta-
cular polyps,’ which are scattered on the outskirts of the
colony, and which they regard as constant in occurrence.
Allman (18), on the other hand, does not believe that they
are universally present, and supposes that they are merely
abnormalities when they do occur.

However, in all those colonies which I have closely exa-
mined with a view to finding these polyps I have been
successful; in one specimen they were particularly abundant,
occurring principally at the growing edge of the colony. In
other specimens they were rare, while on one very flourishing
colony I only found about half a dozen. I doubt if they are
abnormal dactylozooids, as Allman suggests, for I have never
found them situated on that part of the colony frequented
by dactylozooids.

These various kinds of polyp are all connected together at
their bases by a common ccenosareal expansion, which is a
white semi-transparent investment completely covering the
chitinous skeleton.

I succeeded in keeping several colonies alive for some time
by suspending the shells on which they were situated in small
tanks, the other occupants of the shells being removed. One

84 MARGARET C. COLLCUTT.

of these colonies lived in this way for about three months,
when the zooids gradually dwindled away, and finally all disap-
peared. This colony, as is general with Hydractinia, ex-
tended itself round the shell mouth into the interior of the
shell, and apparently began to absorb this enclosed portion of
the shell, which became thin and brittle.

In other colonies kept in confinement the dactylozooids
which were congregated round the edge of the shell disappeared
after a few days.

A point of some interest which I noted in examining whelk-
shells tenanted by hermit-crabs and bearing Hydractinia
colonies on their surfaces is that a Polychete worm, Nereis
bilineata, was always found living inside the shell in com-
pany with the hermit-crab. This worm lived in the coils of
the tip of the shell, and could completely withdraw itself from
observation.

The most successful killing reagents used were Flemming’s
solution and picric acid solution; Hermann’s solution also
gave good results for histological purposes. The staining re-
agents principally used for sections were Delafeld’s hema-
toxylin and borax carmine. For surface views of the thin edge
of the colony aniline orange was found to be an excellent chitin
stain; the preparations were left for a few minutes in a 90 per
cent. alcoholic solution of aniline orange, when the chitin was
stained bright yellow, the protoplasm being but slightly
stained.

GENERAL ANATOMY.

The Skeleton.—This is a continuous unevenly deposited
layer of horny chitin, so closely attached to the rough surface
of the shell that the latter must be decalcified in order to
isolate the hydroid colony. It is, for the most part, secreted
by the lower layer of ectoderm which forms part of the cceno-
sarc of the colony. In the central parts of the colony it is
in many places of considerable thickness, with irregular
lacune, and thickly beset with small chitinous spinules, while
it is frequently raised up into a number of large conical

ON THE STRUCTURE OF HYDRACTINIA ECHINATA, 85

grooved spines. Even in these mature parts of the colony,
that part of the skeleton which connects the spinules and
spines together is often exceedingly thin and inconspicuous.

The spinules are merely solid chitinous thickenings, irre-
gular in shape and size, projecting upwards from the chitin
which is attached to the whelk-shell. They occur all over the
skeleton, and are very conspicuous towards the thin edge of
the colony (Pl. 1, fig. 1, a and 0).

The spines are one of the characteristic features of the
colony; they are mostly deeply furrowed, having longitudinal
serrated ridges between the grooves. The ridges are con-
nected with one another by chitinous cross-bars, and fuse
together at the tip of the spine. Central chambers or a single
chamber communicate with the grooves. The whole spine is
covered by the general ccenosare, which is also continued into
the meshes of the framework, though it may often become
rubbed off from the tips of old spines where the shell has
come in contact with the ground.

Allman (18) has figured and described a typical spine in
his account of H. echinata. Sometimes the spines bifurcate
towards their apex, or are otherwise irregularly shaped.

Between the spines and spinules the skeleton is in some
places deposited as a single thin layer, but it is usually secreted
in thin, irregular, reticulated layers, one above another. In
the older parts of the colony these layers, in vertical section,
have the appearance of chitinous strands varying in thickness
and running horizontally or obliquely (P1.1, fig.1,c), but towards
the thin growing edge of the colony they are more regular.

At intervals throughout the skeleton there are spaces sur-
rounded on all sides by chitin, and containing degenerating
masses of coenosarc; such masses have become constricted off
from the rest of the ccenosare by unequal growth of the chitin,
the various stages of such constriction being demonstrable.
In some cases where the constriction and isolation are not com-
plete the degenerating ccenosare is seen to be in direct con-
tinuity with the coenosarc of the colony (PI. 1, fig. 2). In other
cases the spaces are empty.

86 MARGARET C. COLLCUTT.

The chitin which is situated beneath such a degenerating
mass becomes somewhat cup-shaped by unequal growth (PI. 1,
fig. 2). Gradually the edges of the cup grow over, only leaving
the mass in the cup attached by a narrow bridge to the upper
cceenosare, until finally the edges of the cup meet and coalesce,
and so isolate the degenerate mass of ccenosarc. These con-
stricted masses consist only of the deeper parts of the cceno-
sarc, i.e. the lower ectodermal layer and the endoderm, and as
they degenerate the boundaries between the cells become
obliterated, the protoplasm becoming very granular, and the
nuclei losing their distinctive appearance.

The Edge of the Colony.—At the edge of the colony
where growth occurs the skeleton is very thin, though thick-
ened at frequent intervals into short, solid spinules ; here, too,
the coenosarc is generally thinner, and the polyps are few and
small.

Extending over the surface of the coenosarc of this region is
a thin membranous layer of chitin, which is connected at
many points with the tips of the spinules of the skeleton.

In this region the skeleton is occasionally raised up into
conical smooth spines, which are fenestrated at their bases, so
that portions of the basal ccenosare, consisting of lower ecto-
derm and of endoderm, can penetrate into the interior of the
spines (Pl. 1, fig. 3).

These spines are partially covered by ccenosarc, which, how-
ever, does not extend over their tips; a thin layer of chitin
spreads over the surface of the ccenosarc, but is not continued
over the naked tips of the spines (PI. 1, fig. 3).

Strethill-Wright (10) mentions hollow chitinous spines filled
with ccenosarc, but he describes them as being derived only
from the upper layer of ectoderm.

The colony is derived at its growing edge from a number of
ccenosarcal tubes clothed with a chitinous perisarc; these
tubes branch about irregularly over a large portion of the
whelk-shell, into the cavity of which they extend ; ectoderm
covers their tips, and they appear to secrete their perisare as
they grow.

ON THE STRUCTURE OF HYDRACTINIA ECHINATA,. 87

The lower surface of the perisarcal tubes is thicker than the
upper surface ; this and the side walls are frequently thickened
into vertical, conical spinules, and form the foundation of the
characteristic skeleton of older parts of the colony. The thin
upper surface has no chitinous thickenings, and usually stains
more readily than the rest of the perisare. As the coenosarcal
tubes continue to grow they branch repeatedly and anastomose
with one another until the tubular character of the chitinous
investment is obliterated; in this way a sheet of coenosarc,
enclosed between two layers of chitin, is formed.

When two such tubes anastomose, it is probable that part
of the internal lining of ectoderm in each tube which lies
nearest to the region of contact absorbs the chitinous wall
along this region ; consequently the two tubes are here placed
in communication with each other, and the ectoderm of the
upper half of one tube becomes continuous with the ectoderm
of the upper half of the other tube, while the ectoderm of the
lower half of the one becomes continuous with the ectoderm
of the lower half of the other. Hence arises the differentia-
tion of two ectodermal layers so characteristic of the maturer
regions of the colony (Pl. 1, figs. 1 and 4).

The endoderm retains its tubular character.

The chitinous spinules along the side walls of anastomosing
tubes do not all become absorbed, but many remain as part of
the permanent skeleton, and are attached at their tips to the
thin upper membrane of chitin (Pl. 1, fig. 1, a, ¢, 0).

As the colony grows this membrane gradually weakens, and
finally is lost (Pl. 1, fig. 1, 6); the ccenosare then grows up
over the tips of the spinules, and so becomes a continuous
expansion over the surface of the skeleton.

Here and there are empty chitinous tubes from which the
protoplasm seems to have withdrawn itself; frequently all that
remain of these tubes are the small chitinous spinules.

At intervals a small polyp branches from this growing
region, breaking through the upper chitin, which does not
extend over the polyp above its base. In some of the colonies
I examined I found many “ tentacular polyps” in this region.

88 MARGARET C. COLLCUTT.

The Basal Cenosare of the Colony.—As mentioned
above, this consists of two layers of ectoderm, enclosing be-
tween them a ramifying mass of endodermal tubes, and forms
that part of the colony which connects the polyps with one
another. The lower layer of ectoderm secretes the greater
part of the skeleton, forming a continuous layer in contact
with it; the upper layer of ectoderm also forms an unbroken
expansion, and is continuous with the ectoderm of the polyps.
The two ectodermal layers thus form two sheets of cells, one
above the other, and are in close contact with one another,
except where the tubes of endoderm are interposed between
them, such tubes serving to place the cavities of the polyps in
communication (Pl. 1, fig. 1, ¢).

These endodermal tubes are more or less oblong in cross-
section, and are always surrounded by a narrow layer of
mesoglea, which separates the endoderm from the ectoderm
(Pl. 1, fig. 5), and is continuous with the mesoglca of the
polyps. avs «h

At frequent intervals thread-like strands of mesoglea from
the lower surface of this endoderm pierce the lower ectoderm,
and attach themselves by slightly widened bases to the
skeleton (Pl. 1, figs. 1, c, and 5).

Similar strands fix the polyps to the chitin, but these are
thicker, more numerous, and, since the lower ectodermal layer
is generally much shallower below the polyps, they are often
shorter. These processes thus exhibit the singularity of being
surrounded on all sides by ectoderm cells. Similar strands
are figured by Weismann (15) as attaching the coenosare of
Eudendrium ramosum to the perisare.

The lower ectoderm and the endodermal tubes penetrate
into and fill up the intercommunicating chambers and the
grooves of the large spines, the superficial ectoderm extending
over all.

Strethill-Wright (10) and Allman (18) both mention the
fact that this basal coenosarc is capable of communicating the
effect of any shock it may receive to the various members of
the colony.

ON THE STRUCTURE OF HYDRACTINIA ECHINATA. 89

The Polyps.

1. The Gasterozooids.—The external appearance of these
polyps has been frequently described and figured; they are
cylindrical in shape and very contractile, and may attain to a
height of about a quarter ofan inch. They are provided with
a conical hypostome, terminating in a mouth, and having
round its base two closely approximated series of tentacles,
which increase in number with the age of the polyp, and vary
from ten to thirty in different colonies. These tentacles are
very contractile, and are provided with numerous nematocysts.
The polyps are capable of everting their hypostomes. The
outermost cell-layer of a Gasterozooid is a layer of ectoderm,
which is continuous at the base of the polyp with the upper
ectodermal layer of the basal ccenosare; the alimentary canal
is lined by a single endoderm layer, which forms a wide
cylindrical tube, connected at its base with the plexus of en-
dodermal tubes which permeates the basal ccenosare (PI. 1,
fig. 1,c). Between the ectoderm and endoderm is the well-
marked mesoglea.

2, The Blastostyles.—These are specially modified re-
productive zocids, giving rise to “ sporosacs.”? The blasto-
styles are smaller than, and differ considerably in appearance
from the Gasterozooids ; they have a small mouth at the apex
of a conical “ head,” and at a short distance from this terminal
mouth there are two or more closely approximated circles of
tentacles. These tentacles vary in number from ten to thirty ;
they are rudimentary knob-like structures, containing prolon-
gations of the endoderm and mesoglea of the polyp, and their
ectoderm is crowded with nematocysts (Pl. 1, fig. 6).

Below the tentacles the body narrows gradually, becoming
externally constricted into a narrow “neck ;” then it widens
considerably into a globular dilatation, from the walls of which
arise the round or oval sporosacs. The sporosacs are out-
growths of the walls of the polyp, all three layers of the polyp
being involved in their formation: generative cells arise in
the body of the blastostyle and migrate into the sporosacs.

90 MARGARET CG. COLLCUTT.

Below the globular region the body again narrows and
becomes connected with the basal coenosare. Miss Bunting
(17) has fully described and figured the sporosacs and gene-
rative cells of Hydractinia. She mentions that she was unable
to trace egg-cells in the ectoderm or mesoglea of the blasto-
style. I have, however, observed small cells situated in the
ectoderm of the blastostyle close to the mesogloea, and indis-
tinguishable from undoubted egg-cells in the endoderm of the
same region of the blastostyle (Pl. 1, fig. 6).

Also in two cases I have found egg-cells in the mesoglea;
the endoderm of the same region already contained several
eggs (Pl. 1, fig. 7). In one of these cases the blastostyle had not
yet formed sporosacs.

3. The Dactylozooids.—These are cylindrical through-
out their length, and are furnished at the distal extremity with
a circle of knob-like rudimentary tentacles, from ten to sixteen
in number, and crowded with nematocysts.

They have no hypostome, and are described by Allman (18)
and Strethill-Wright (10) as having no mouth. I have, how-
ever, found a distinct mouth situated in the centre of the
tentacle circle, and leading down through a short tubular canal
bounded by ectoderm into the endodermal cavity (Pl. 1, fig. 8).
These polyps are exceedingly muscular, and are capable of
coiling and uncoiling themselves.

In one colony which I examined at Plymouth the dactylo-
zooids were frequently branched once or twice, each branch
terminating in a typical dactylozooid head.

4. The Tentacular Poly ps.—These polyps are principally
situated towards the outskirts of the colony, though they may
also occur in the older regions; most frequently I have found
them arising from the network of tubes at the growing edge.
They are exceedingly slender, though often longer than the
other zooids comprising the colony, and are contractile, for
their ectoderm is frequently marked by transverse folds. They
possess a tubular internal cavity lined by large endoderm cells :
I have not succeeded in demonstrating a mouth in these
individuals.

ON THE STRUCTURE OF HYDRACTINIA ECHINATA. 91

The distal extremity is slightly enlarged into a club-shaped
tip, covered all over with nematocysts, which also occur in
considerable numbers throughout the ectoderm of the polyp.

In examining colonies of Hydractinia for these polyps I
came across two abnormal individuals. One of these had the
appearance of a tentacular polyp which had given rise to two
branches situated at different levels, and each terminating, like
the main polyp, in a rounded tip covered with nematocysts.
(Process, Fig. 1.)

The other abnormality was a gasterozooid from which, just
below its tentacles, branched a tentacular polyp. (Process,
Fig. 2.)

Ime, Be

Histo.oey.
a. Histology of the Lower Ectoderm.

This layer extends, renews, and secretes the greater part of
the skeleton ; occasionally, in very limited areas, it is more
than one cell deep.

The cells comprising it are not uniformly equal in size;
beneath the polyps and towards the edge of the colony they
are more or less square and regular in shape, but in the adult
parts they are frequently deeper than they are broad. Asa
consequence of the great variation in the depth of the cells of

92 MARGARET C. COLLCUTT.

this layer the thickness of the ccenosarc varies considerably
throughout the colony ; over the tips of the chitinous spinules
and the jagged ridges of the spines this ectodermal layer is
often so shallow as to be almost inappreciable.

The nucleus is large, and situated in the centre, or rather
towards one side of the cell; it is oval in shape, and has a
dense, deeply staining reticulum with several nucleoli.

In some regions of this ectodermal layer the cells are very
vacuolated; the nuclei of these cells are easily discernible,
and are surrounded by a mass of coarsely granular protoplasm,
from which protoplasmic strands extend to the cell borders
(Pld fie),

In other places, however, the cells are closely crowded
with deeply staining roundish corpuscles, which are often
present in such numbers as to obliterate from view the
boundaries between the cells and hide the nuclei. Probably
the presence of these corpuscles is accounted for by the fact
that the lower ectoderm apparently secretes some of the nema-
tocysts which frequently occur in great abundance in these
parts of the lower ectoderm (Pl. 1, fig. 1, ¢).

As was mentioned by Strethill-Wright (10), there are two
kinds of nematocyst in Hydractinia, small and large; both
kinds aré found in the lower ectoderm.

The ectoderm of the gasterozovid tentacles is crowded with
the small variety, while large nematocysts occur in the ecto-
derm of the blastostyle and dactylozooid tentacles, and in the
ectoderm of the tips of the tentacular polyps; nematocysts
are also found in small numbers throughout the ectoderm of
the bodies of the gasterozooids, blastostyles, and dactylo-
zooids, and in large numbers in the ectoderm of the bodies of
the tentacular polyps, but I have never found both the large
and small variety occurring in the same individual.

B. Histology of the Upper Ectoderm.
1. Ectoderm of the Basal Cenosare.—This forms a
continuous and regular layer of cells, which are approximately
equal in size, and appear cubical in section.

ON THE STRUCTURE OF HYDRACTINIA ECHINATA. 95

These cells are apparently not characterised by muscle tails,
but interstitial cells are sometimes wedged in between them
at their bases (Pl. 1, fig. 5). There are no sensory cells in
this layer, but nematocysts are occasionally found. The nucleus
is situated in the centre of the cell ; it is round or slightly oval,
and has always one large nucleolus: surrounding it is a mass
of finely granular protoplasm which almost fills the whole cell.

2. Ectoderm of the Polyps.—This layer is in many
respects similar to the basal ectoderm; the cells are often
somewhat narrower, more columnar in shape, and more variable
in size.

They are characterised by exceedingly long muscle tails,
which can be demonstrated by macerating a hydroid for about
an hour in + per cent. solution of acetic acid, washing in water,
staining with picro-carmine for an hour or more, and examin-
ing in weak glycerine.

The protoplasm of the muscle tails appears more homoge-
neous and less coarsely granular than that of the body of the
cell. Inseveral of these isolated ectoderm cells large vacuoles
were observed. (Process, Fig. 3.)

Fic. 3.

94, MARGARET C. COLLCUTT.

The muscle tails in non-macerated material are most easily
demonstrable in sections of dactylozooids, where they are
greatly developed (PI. 1, fig. 8).

c. Histology of the Mesoglea.

The mesoglea extends between ectoderm and endoderm
throughout the colony as a structureless, homogeneous layer.
In the basal part of the colony it is very thin, and, as men-
tioned above, thin strands of its substance serve to fix the
coenosare to the chitinous skeleton; in the polyps it forms
a much thicker layer (PI. 1, fig. 1, c).

pd. Histology of the Endoderm.

1. Endoderm of the Basal Cenosarce.—The endoderm
lining the cavity of the ramifying tubes consists of a single
layer of cubical cells, with large central oval nuclei. The proto-
plasm is coarsely granular and abundant, and there are many
food granules. Other deeply staining gland-like cells occur
at intervals throughout the endoderm of these tubes (PI. 1,
fig. 1, 6); these are probably similar to certain so-called gland-
cells which occur in the endoderm of the Gasterozooids.

2. Endoderm of the Gasterozooids.—This endodermal
layer consists of long narrow cells, closely pressed against one
another, and somewhat irregular in shape; they vary in length,
so that their free ends are not all on the same level. They
are greatly vacuolated, especially towards their free ends,
where they widen slightly. The nucleus is situated in the
middle region, and close to the border of the cell; it is oval,
and has a granular reticulum, with one or more nucleoli.

In the hypostome the endoderm is thrown into longitudinal
ridges, which project into its cavity; there are usually from
four to eight of these ridges, which then pass downwards and
lose themselves in the endoderm of the stomach. A transverse
section of the hypostome consequently represents the lumen
of the gut as being star-shaped. Allman (14) mentions that
similar endodermal ridges occur in the endoderm of other
gymnoblastic hydroids.

ON THE STRUOTURE OF HYDRACTINIA ECHINATA. 95

In a gasterozooid which has had no food for some time the
protoplasm of the endoderm cells is exceedingly vacuolated,
there being only a small mass of protoplasm congregated round
the nucleus, and a thin protoplasmic cell lining; no other
endodermal elements can be demonstrated, at all events in
dead material.

On the other hand, an examination of the endoderm of a
gasterozooid which is digesting food clearly shows that there
is a certain amount of differentiation between the cells. The
ordinary vacuolate endoderm cells are crowded with food
vacuoles; the nuclear plasm appears more granular than in
non-digesting endoderm cells. But besides these vacuolate
cells, there are numbers of pyriform cells which are situated
singly between every two or three of the vacuolate cells, and
which stain very readily. These cells are rather wider in
their middle region than the vacuolate cells, and taper off
basally where they come in contact with the mesoglea; the
nucleus is situated towards the base of the cell and is sur-
rounded by adense mass of protoplasm. The rest of the cell is
crowded with small spherical bodies, which are so abundant as
to hide the protoplasm in which they are embedded (Pl. 1, fig.9).

Miss Greenwood (16) describes similar cells in the endoderm
of Hydra under the name of “ gland-cells.”” Among the cell
contents of the endoderm are numerous roundish bodies,
smaller than nuclei, and especially abundant in the endoderm
of the hypostome; some of these stain equally throughout,
but others only stain partially, a large slightly stained
region being ieft. They stain deep blue with hematoxylin
but yellow with borax carmine; they are colourless in the
living animal, and stain with iodine less than the other cell
contents. The most likely hypothesis seems to be that they
are proteid in nature ; they may correspond to the “ nutritive
spheres ” of Hydra described by Miss Greenwood (16). Endo-
derm is continued into the tentacles as a single row of cells
which are very vacuolated and regular in shape; the nucleus
is central and surrounded by protoplasm, which forms a central
band in the cell.

96 MARGARET C. COLLCUTT.

3. Endoderm of the Blastostyle.—The endoderm
lining the cavity of the blastostyle is mostly richly ciliated
with very long cilia; the cells of the “ head” are large, and
each contains a nucleus like that in the endoderm cells of the
rest of the colony; these cells digest food material, apparently
becoming ameeboid at their free ends (Pl. 1, fig. 6), the blasto-
styles undoubtedly taking in nutritive material by their
mouths.

The cells of the “ neck ” are very long, regular, and narrow,
and there are numerous small nuclei, each containing several
distinct nucleoli in this region, which is richly ciliated; egg-
cells are here present in abundance (PI. 1, fig. 6).

Below this region the body of the blastostyle dilates, and
sporosacs arise; the cells here are smaller, have large nuclei,
and cilia are not so apparent. Gland-cells are nowhere
present in the endoderm of blastostyles. The tentacles have
a solid endodermai core as in the gasterozooids. Egg-cells
have been observed in the endoderm of the base of the blasto-
style.

4. Endoderm of the Dactylozooid.—This endoderm is
very vacuolated ; the cells are approximately equal in size, but
are rather smaller towards the distal extremity of the polyp.
The nuclei are situated in the middle of the cells. The endo-
derm of the upper half of the body is often crowded with food
granules, but I have never observed food masses in the lumen
of the endodermal gut, nor are gland cells present.

5. Endoderm of the Tentacular Polyps.—This endo-
derm is also crowded with food granules throughout its extent.
The cells are very regular, long, and narrow, and the lumen
of the gut widens considerably towards the base of the polyp.

ON THE STRUCTURE OF HYDRACTINIA ECHINATA. 97

BIBLIOGRAPHY.

. Fremine, J.—* Aleyonium echinatum,” ‘ British Animals,’ 1828.

2. Jounston, G.—“Alcyonidium echinatum,” ‘ British Zoophytes,’

p. 304, pl. xlii, figs. 3, 4, 1838.

. Jounston, G.—‘*Coryne squamosa, var.,” ‘British Zoophytes,’

pl. ii, figs. 4, 5, 1838.

. Van Benepen, P. J.—“Hydractinia, sp.,” ‘Bull. de Acad. Roy. de

Bruxelles,’ tom. viii, 1841.

. Hassatt, A. H.—‘ Echinochorium clavigerum,” ‘ Ann. Nat. Hist.,’

vol. vii, p. 371, pl. x, fig. 5, 1841.

. Putureer, AA—“ Dysmorphosa conchicola,” ‘ Wiegman’s Archiv,’

1842.

. DE QuatReFAGEs, A.—‘‘ Synhydra Parasites,’ ‘Ann. des Sci. Nat.,’

vol. xx, p. 230, pls. 8, 9, 1843.

. VAN Benepen, P. J.—‘<Hydractinia lactea and Hydractinia

rosea,” ‘ Rech. sur l’Embryogénie des Tubulaires, Mém de |’Acad.
Roy. de Brux.,’ tom. xviii, p. 104, pl. ix.

. Jounston, G.— H. echinata,” ‘ Brit. Zooph.,’ p. 34, pl. 1, figs, 4, 5,

1847.

. STRETHILL-Wricut, T.—‘‘ Hydractinia echinata,” ‘Edinb. New

Phil. Journ.,’ April, 1857.

. Van Benepen, P. J.—“H. echinata,’ ‘Recherches sur la Faune

littorale de Belg.,’ p. 134, pl. xl, figs. 1—8, 1866.

. Hincxs, T.—“ H. echinata,” ‘ Brit. Hydr. Zooph.,’ p. 23, pl. iv.
. Attmay, G. J.—‘‘ H. echinata,” ‘Gymnoblastic Hydroids,’ pp. 220,

342, pls. xv and xvi, figs. 10, 11, 1871.

. Atiman, G. J.—‘ Report on the Hydroida collected by H.M.S. ‘ Chal-

lenger,’ p. ix, 1888.

. Weismann, A.—‘ Enstehung der Sexualzellen bei den Hydromedusen,’

Jena, 1883.

. GrEENwoop, M.—“ Digestion in Hydra,” ‘Journal of Physiology,’ vol.

ix, 1888.

. Buntine, M.—* Origin of Sex-cells in Hydractinia,” ‘ Journ. Morph.,’

ix, 1894.

vou. 40, PART 1.—NEW SER. a

93 MARGARET C. COLLCUTT.

EXPLANATION OF PLATE 1,

Illustrating Margaret C. Colleutt’s paper “On the Structure
of Hydractinia echinata.”

(Several different colonies were examined, hence there is a slight variation in
the size of cells, &c., figured.)

LIST OF REFERENCE LETTERS.

up.ect. Upperectoderm. J/.ect. Lowerectoderm. ezd. Endoderm. mes.
Mesoglea. i.c. Interstitial cell. . Nematocyst. .c. Nematocyst cell.
gl. Gland-cell. plp. Polyp. spz. Spinule. p.m. Perisare membrane. c. ¢.
Ceenosarec tube. p.¢. Perisarc tube. e. Egg-cell. fm. Food material.
mo. Mouth. s&. Skeleton. v. Vacuole. m.¢. Muscle tail. mes. p. Meso-
gloea process. d.c. Degenerating ccenosare. zw. Nucleus.

Fig. 1.—Series of vertical sections illustrating the transition from the
growing edge to the older parts of the colony. a. Section at extreme edge,
showing to the left a coenosarc tube surrounded with perisarc, and towards
the right tubes which have anastomosed. 4. Section of somewhat older
part of the colony, where the perisare membrane tapers off and disappears.
c. Section of old part of colony passing through portion of the wall of a
polyp, and showing the communication between the polyp and an endodermal
tube of the basal ccenosarc. The section also shows the tabulated growth of
the chitinous-skeleton. D. 3 cam.

Fic. 2.—Vertical section of base of colony, showing the degeneration of
ccenosare and a degenerating mass isolated in a chitinous lacuna. D. 3 cam.

Fic. 3.—Diagrammatic section of a smooth spine at the colony edge.
A. 3 cam.

Fie. 4.—Optical horizontal section of two anastomosing tubes, showing
both commencing anastomosis and complete anastomosis. The left-hand tube
has given off a branch. D. 3 cam.

Fie. 5.—Vertical section through a growing chambered spine illustrating
the relations of the basal ccenosarc to the skeleton, and showing a chamber of
the spine containing ccenosare, D. 3 cam.

Fie. 6.—Longitudinal section of upper part of blastostyle ; on the right-
hand side is an egg-cell, e’, in the ectoderm, and two egg-cells in the endo-

ON THE STRUCTURE OF HYDRACTINIA ECHINATA. 99

derm of the “neck.” Food is being digested by the endoderm cells of the
“head.” D.3 cam.

Fic. 7.—Transverse section through a blastostyle showing an egg-cell,
e’, in the mesogloea of the blastostyle, and four egg-cells in the endoderm of
the sporosac. D. 3 cam.

Fie. 8.—Longitudinal section of a dactylozooid, showing food granules in
the endoderm of the upper part. D. 3 cam.

Fic. 9.—Longitudinal section of the endoderm of gasterozooid. D. 3 cam.

t
ms 1 Lee rig?
vee J 3 ip L2ee
, <b ha tae Ge j
d j 7 ‘
Pe iemapnay 4. ' ] : i spue bali 8. he by
; f Fie fi)
iW Aly Ue} . ‘ = ha wid ‘ 4 sr he rs i rue i.
? n . tt at/ a] ‘ i weeds
, ’ pnb bodes tH le Be bt bar ies: 3
f ’
eT ai
= t “
mr ‘
“a ‘
b .
iv
é . ; ~<" : .
a -_ , ~ i
° ,, Luss
i eS : p bd :
—~ Se ; .
' ' a aA, al ats ' fet ‘
2 = a. | o* : a a= = ade”
a aS, . » 4 " ie M
* , ; a 5 a “aay ¥
Xe se

. ~ ¢ ; 7 “a .' - ee

>
.
* 5 4
ra 7
- a
s* ‘
Pi
6
]
: cS
ee =.
A]
: & .
. iP * y
val *
We as
a : a ’ ae oy =
* P a

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 101

On the Histology of the Ovary and of the Ovarian
Ova in certain Marine Fishes.

By

J.T. Cunningham, MW.A.Oxon.!

With Plates 2—4.

THE DEVELOPMENT OF THE YOLK.

In my paper in the ‘Journal of the Marine Biological
Association,’ vol. iii, No. 2, p. 154, I have described the
development of the yolk in the plaice and dab as commencing
by the deposition of yolk granules in a thin layer near the
surface of the egg, and proceeding by the continual extension
of this layer towards the centre. I have also described and
figured the appearance of minute refringent globules dis-
tributed sparsely throughout the protoplasm of the egg, in
specimens of the sole which were examined a little time before
the commencement of the spawning season. Such globules
were also seen in spent ovaries of the sole, and seemed to
indicate that the formation of yolk occupied more than a year.
in some fishes, while in the plaice and dab no sign of the
development of yolk was seen until some months after the
spawning season.

I have since ascertained that among fishes with pelagic ova
the one type of development is characteristic of eggs which
have oil globules, the other of eggs that are not provided with
such elements. Certain fishes, as for example the gurnards,
have an extended spawning period, and ripen their eggs not

1 The researches described in this memoir were carried out in the service
of the Marine Biological Association in 1895 and 1896,

102 J. T. CUNNINGHAM.

simultaneously but in succession. In the ovary of a gurnard
accordingly, at the commencement of the spawning season, ova
in all stages of development are found. In figs. 1 to 5 are
shown the appearance under a low power in the fresh state, of
a number of eggs of the common or grey gurnard (Trigla
gurnardus) at successive stages in the development of the
yolk. In fig. 1 the protoplasm is transparent, and contains a
small number of globules scattered singly. In the next stage
a dark zone is seen commencing to form around the nucleus,
and in the outer region of the egg are globules similar to those
of the previous stage, but smaller and more numerous. The
contrast between the two concentric layers or regions is most
marked in the next stage, fig. 8, in which around the light
central region indicating the position of the germinal vesicle,
there is a very opaque layer sharply delineated externally from
an outer more transparent layer. In the next stage, fig. 4, the
contrast between the two layers has disappeared, and except
for a somewhat lighter ill-defined central region, the whole is
nearly uniformly opaque. Eggs which are larger than this,
begin to grow transparent again, and the appearance they
present is shown in fig. 5. It is easy to perceive from exami-
nation of this last stage, that the internal globules are oily, the
external vitelline. As the egg approaches the ripe condition,
both the oil globules and the vitelline globules begin to fuse
together. The large oil globules in the interior of the egg,
formed by the fusion of the smaller, are seen in fig. 5. The
vitelline globules in fusing form a continuous liquid, in which
the still unfused globules remain suspended. The opacity
of the egg is due to the small size and great abundance of the
globules at the earlier stages, and we learn, therefore, that the
greater opacity of the inner layer is due to the smaller size and
greater refracting power of the fat globules, which are formed
close to the germinal vesicle.

Similar features have been described by Emery in the deve-
lopment of the ovum in Fierasfer acus. The mature
ovum of this fish resembles that of the gurnard, having a
homogeneous transparent yolk, and a single large oil globule.

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 108

Emery figures a portion of an immature ovary as it appears
under a low power in the fresh condition. In the largest eggs,
scattered, highly refringent globules, exactly like those de-
scribed by me in the eggs of the sole, &c., have begun to
appear, and are identified by Emery as adipose globules.
Sections of later stages from material prepared with picro-
sulphuric acid are figured, in which the oil globules have become
larger by fusion, and the yolk globules are developing, the
latter forming an external layer, the former situated near to
the germinal vesicle. There is a difference, however, to be
mentioned. In Fierasfer the eggs are not spherical, but
oval, and the germinal vesicle is situated nearer to one pole of
the oval than to the other. At the pole which is farthest from
the germinal vesicle the yolk globules are formed first, and the
vitelline nucleus is situated there also, so that the yolk glo-
bules are formed around it. The yolk layer during its increase
continues to be thickest at the same pole.

Emery describes the course of development in the ovary
which culminates in the annual spawning. He found that
from the commencement of autumn to ihe end of spring, the
ovary contained transparent eggs only, of which the largest
contained a few small adipose globules, but no vitelline; in all
the ova the vitelline nucleus was still visible. As the time of
spawning—which extends over July, August, and September—
approached, the vitellus in the larger eggs was rapidly formed.
When the mature eggs had been discharged the ovary was
collapsed, it showed traces of hemorrhage in the form of
extravasated blood, or masses of pigment red or yellow in
colour, and besides young ova contained others which, not
having reached perfect maturity in time for the spawning
process, were not expelled, and were undergoing adipose
degeneration.

Dr. Robert Scharff published a paper on the development of
the egg in Teleosteans some years ago (this Journal, vol.
XXXvili), in which he refers especially to the ovarian eggs of
Trigla gurnardus. He discusses somewhat briefly the
formation of the yolk spherules and oil globules. His figures

104 J. T. CUNNINGHAM.

represent the appearance of eggs and sections of eggs with
considerable accuracy, but he interprets them as leading to
the conclusion that the vitelline elements are derived from
the nucleus, a conclusion which I hold to be entirely erroneous.

I will proceed now to a detailed description of the develop-
ment of the yolk in the various species which I have examined.
I will deal first with—

Pelagic Eggs containing one or more Oil Globules.

In the same family of fishes, some species may be charac-
terised by eggs with oil globules, others by eggs in which
these structures are absent. In the Pleuronectidz it is curious
to note that all the left-sided species have a single oil globule
in the egg, namely, the turbot and brill (Rhombus), the
megrim (Lepidorhombus megastoma), and the topknot
(Zeugopterus). The soles, which are right-sided, produce
eggs in which there are numerous small oil globules, and in
the remaining right-sided species forming the genera Pleuro-
nectes, Hippoglossusand Hippoglossoides, the eggs are
without oil globules. In the family Gadide a similar state of
things occurs, in certain genera, e.g. Molva, the ling, an oil
globule being present in the eggs, while in the genus
Gadus it is absent.

In immature specimens of the brill examined during the
spawning season, the scattered globules are seen in the
larger eggs. These oil globules first appear in contiguity
to the membrane of the germinal vesicle. Fig. 6 shows the
appearance of a portion of the ovary of a specimen of this
species 12? inches long, examined at Grimsby on May 30th,
1895. The same condition was observed in a number of
specimens examined on this date; they were from 10% to 122
inches long, and captured near the German coast off the Sylt
Island. A piece of one of these ovaries was preserved in a
mixture of picro-sulphurie acid and spirit, and in sections of it
stained with Mayer’s carmalum the oil globules, or the va-
cuoles in the cytoplasm in which the oil was originally con-

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 105

tained, can be at once recognised, having retained their form
and position.

Even in the brill the oil globules at their first appearance
are not entirely confined to the neighbourhood of the germinal
vesicle, some of them being scattered in the rest of the cyto-
plasm. In the gurnard as seen in fig. 1, they are irregularly
scattered through the cytoplasm, while in the sole the greater
number are situated near the surface of the ovum. For the
present I will confine myself to pelagic eggs with a single oil
globule, leaving the sole to be considered separately.

In the turbot in ripening ovaries examined in the fresh state
the same contrast between inner and outer zone in the deve-
loping eggs is observed as in the gurnard; I believe it occurs
also in the brill. The fact that the outer more transparent
zone consists of yolk spherules, indicates that, as might be
expected, the yolk is formed first in these cases, as in the
species of Pleuronectes, at the periphery. More minute exa-
mination of the process must be made by means of prepared
sections.

Different modes of preparation make great differences in the
visibility of the vitelline elements in sections. Thus portions
of the ovary of a brill 17; inches long were preserved imme-
diately after death, at sea, on September 29th, in chromic acid
z percent. This ovary is evidently in a much more advanced
condition than that previously mentioned, it contains eggs
which have reached a much larger size than any in the latter,
and in which the yolk is considerably developed. Yet in the
younger eggs, of sizes corresponding to those in the specimen
previously mentioned, the oil globules are quite invisible.
The chromic acid appears to have the effect of closing up the
oil-containing vacuoles.

From what has been stated above concerning the ovary of
Trigla gurnardus, it is evidently possible to study all stages
of the development of the yolk in this species, in sections from
the same ovary taken at the beginning of the spawning season.
The earlier stages are better preserved in material treated with
picro-sulphuric acid and spirit. I have sections from material

106 J. T. CUNNINGHAM.

so prepared taken from a specimen obtained in Grimsby
market on April 26th, and from portions of an ovary preserved
directly after the death of the fish at sea on July 22nd. Com-
parison of the two shows that the material from the market
specimen is in fairly good condition, the vitelline elements
being well preserved. The principal defect due to the stale-
ness of the material is the contraction of the protoplasm of the
younger yolkless eggs, and the presence of a layer of granular
matter around them.

The yolk globules are first recognisable in eggs ‘23 mm. in
diameter, and have the form of minute granules situated near
the surface of the ovum. The yolk granules can be distin-
guished from the oil globules by the fact that they are solid.
The process of preparation coagulates the yolk substance, and
it retains the staining matter used to a greater or less degree,
while the oil globules are not coagulated but dissolved, and
the spaces they occupied appear merely as empty vacuoles.
At the stage now under consideration there is a zone of such
vacuoles of rather large size separated by a zone of relatively
homogeneous protoplasm, both from the surface of the nucleus
and from the yolk zone. This stage corresponds to that of
fig. 3, among those representing the appearance in the fresh
condition.

Since in earlier stages, before the appearance of yolk
granules, oil globules occur quite close to the wall of the
germinal vesicle, it would seem that some change has occurred
which has driven them outwards. This might ‘be explained
by supposing that new protoplasm was forming in the neigh-
bourhood of the germinal vesicle. The egg is rapidly increas-
ing in size, and it seems probable enough that the deposit of
food material is taking place at the periphery in the older
protoplasm, while the growth of the protoplasm is taking
place in the neighbourhood of the germinal vesicle. The zone
around the latter, however, has a granular appearance, and
contains I believe very numerous but extremely minute glo-
bules of oil, although they cannot be recognised with certainty
as such in the sections.

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 107

It is improbable that the great opacity of the inner zone in
the ova seen in the fresh condition, could be due to the some-
what large and not very numerous oil globules which are most
conspicuous in the sections, and in fact the appearance in the
fresh condition indicates very numerous minute globules.
Fig. 7 represents the appearance of an egg in the stage here
described. The preparation from which it was taken was
made from portion of an ovary of Trigla gurnardus, ob-
tained in the market on April 26th. The material was pre-
served with a mixture of picro-sulphuric acid and spirit, and
the sections were stained with Delafield’s hematoxylin. Sec-
tions from material similarly treated, preserved at sea imme-
diately after the death of the fish on July 22nd, contain eggs
in a similar stage, and show in them the same structure, but
for some reason or other the outline of such eggs in these
sections is distorted, and I have therefore preferred to draw
from the others. In the sections from the fresh material the
inner zone of protoplasm shows more distinctly the minute
cavities which I believe to be the smallest oil globules. In
sections from some of the same material preserved in chromic
acid 4 per cent., the structure at this stage is not so well
shown. The yolk and oil globules are less distinct, and there
is a broader internal zone of more homogeneous protoplasm
with a rather distinct boundary just within the zone of larger
oil globules.

In the later stages, i.e. in larger eggs as seen in the sec-
tions, the zone of yolk globules has increased in thickness and
extends to the zone of oil globules. Fig. 8 shows the appear-
ance of a section of an egg ‘37 mm. in diameter, preserved
immediately after death in picro-sulphuric acid and spirit.
The yolk globules themselves are now larger, and under a
high power are seen to be solid, 1. e. coagulated spheres, often
with granules in their interior; usually some slight space is
seen between the outline of the globule and that of the vacuole
which it occupies. The yolk globules are stained a light
yellow by the picric acid, but do not usually take much colour
from the staining reagents used. ‘The outer oil globules are

108 J. T. CUNNINGHAM.

considerably larger than the yolk globules, and are seen as
empty vacuoles in the protoplasm; they are not very nume-
rous, forming an irregular ring, and within them is a pro-
toplasmic zone for the most part destitute of yolk globules, and
containing the minute vacuoles which probably were occupied
by small oil globules in the fresh condition. Here and there,
however, scattered yolk globules are seen between the larger
oil vacuoles, and in the inner zone of protoplasm.

I have found that the reason why it is difficult to recognise
the smaller oil globules in the prepared sections, is because
the oil is entirely dissolved and removed in the process of pre-
paration. The reagents used for removing the paraffin from
the sections, and for clearing and mounting, namely, benzole
or turpentine, necessarily remove the oil originally contained
in the eggs themselves. When a portion of an ovary is pre-
served with osmic acid, either alone or in combination with
other reagents, the oil in the eggs is blackened. I found that
some of the blackened oil remained in the sections after the
processes of dehydration, soaking in benzole, imbedding, cut-
ting, and dissolving the paraffin from the cut sections on the
slide, but by the time the sections were mounted with Canada
balsam all trace of the blackened oil had disappeared. By
washing the sections with alcohol after the paraffin had been
removed, and then mounting them in glycerine, a portion of
the blackened oil was kept permanently in situ. In order to
preserve it completely it would be necessary to cut the sections
by an aqueous method, e.g. by the freezing method, and
mount them in glycerine, without dehydrating them at any
Stage.

The thickness of the zone of yolk spherules continues to
increase until it includes the whole of the egg except the
germinal vesicle, and a small number of large oil vacuoles,
into which the numerous small oil globules of former stages
have united. These vacuoles are situated close to the surface
of the germinal vesicle. The condition here described is
illustrated in fig. 9, which is taken from a preparation made

from material preserved at sea on July 22nd in chromic 4 per

OVARY AND OVARTAN OVA IN CERTAIN MARINE FISHES. 109

cent., the sections being stained with safranin. The longer
diameter of the egg represented is '53 mm.

My sections from the gurnard do not show any later stages
of the egg in which the changes immediately preceding the
ripe condition are taking place, namely, the fusion of the yolk
globules, and the passage of the nuclear elements to the
periphery. I will pass on now, therefore, to the consideration
of eggs of similar kind in other species.

The eggs of the turbot and brill are so similar that they may
be described together. They differ from those of the gurnard
in three particulars: (1) their smaller size; (2) the greater
uniformity of condition among the eggs in a single fish; (3)
the less conspicuous appearance of the oil vacuoles in prepared
sections.

I have prepared sections from a portion of the ovary of a
turbot, 1 foot 9} inches long, which was obtained from
Grimsby market on April 12th, 1895. The material was pre-
served with picro-sulphuric acid. The condition of the yolked
ova in these sections is good, the protoplasmic ova show the
contracted condition usually seen in sections from market
material. When fresh this ovary was found to contain
opaque white eggs advanced in development, but none in the
ripe transparent condition. The contrast between the internal
and external zone was not visible, the stage of development
being too advanced. .

Figs. 10, 11, 12 represent eggs in three different stages as
they appear in these sections. The great majority are in the
condition seen in fig. 12, in which the extra-nuclear region is
everywhere crowded with round yolk globules, excepting the
space occupied by the oil vacuoles near the nucleus. The egg
section drawn is ‘35 mm. in diameter. This is the stage a
little before the fusion of the yolk globules commences. A
proportion of younger eggs in successive stages are present in
the sections. In the stage represented by fig. 11, the large
size of the yolk globules is remarkable. This stage in the
turbot corresponds with that of fig. 8 in the gurnard, and it
will be seen that in the turbot the yolk globules are larger and

110 J. T. CUNNINGHAM.

less numerous, and the oil globules in the zone next to the
nucleus much smaller and less conspicuous. Fig. 10 repre-
sents the stage at which the formation of yolk granules at the
periphery is just commencing, small oil globules being visible
in the inner zone ; eggs in this stage are comparatively few in
number. There are, as in all stages of the ovary, a certain
number of yolkless ova.

Sections from the ovary of a specimen 19% inches long,
obtained in Grimsby market on April 24th, show a less
advanced condition of maturation. The majority of the ova
are in a stage a little earlier than that of fig. 12, the oil
vacuoles being smaller and more numerous.

I have sections from another stage in the history of the
turbot’s ovary, the material having been taken from a fish
2 feet 13 inches long, on July 23rd, and preserved in picrosul-
phuric acid and spirit at sea immediately after death. The
preservation is not quite satisfactory, but the sections show
many points worthy of mention. With the exception of a
small number in young stages, all the eggs are in two stages,
namely, ripe, with the yolk in a continuous mass surrounded
by a thin envelope of protoplasm, and nearly ripe, with the
yolk spheres still separate. There are scarcely any inter-
mediate stages between these two; but in the nearly ripe con-
dition the protoplasmic strands separating the yolk globules
are not visible toward the centre of the egg. It is evident,
therefore, that the fusion of the yolk spherules takes place
almost simultaneously. The nearly ripe eggs in these sections
are nearly all ruptured on one side, the zona radiata being
much thickened on the opposite side, and the contents parti-
ally escaping. This may be attributed to the effect of the
picro-sulphuric acid, but whether its action has been to contract
the zona radiata, or to burst the latter by swelling the yolk, I
cannot say. Curiously enough the ripe eggs are not burst, but
in them the zona radiata is thinner, and the diameter of the
egg larger. It is evident that the yolk swells considerably
during the final stage of maturation.

The sections of the ripe eggs exhibit an internal homo-

ao
5

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 111

geneous mass of transparent almost unstained yolk surrounded
by a thin envelope of stained protoplasm. The latter contains
numerous vacuoles, and its inner surface is very uneven; but I
have failed to detect in it any structures representing the
nucleus of the ovum, which ought to be present in it some-
where, and which is distinct enough in the nearly ripe eggs in
the same sections. The oil globules, one or more in number,
are generally visible just within the protoplasmic envelope.
Sections of a portion of the ovary of a brill preserved in
chromic acid + per cent., on September 29th, show an early
stage in the development of the eggs for the next spawning
season. The largest eggs are ‘°3 mm.in diameter, and are in a
condition somewhat more advanced than that of the turbot’s
ege represented in fig. 10, which is only ‘17 mm., and similar
to that of the gurnard’s egg represented in fig. 7. There is in
these eggs a broad inner layer of rather large oil globules, and
a narrower external zone containing minute yolk granules.
Although the eggs of the mackerel, like those previously
considered, have a single oil globule and homogeneous yolk, I
have not seen any indication in them of the presence of scat-
tered small oil globules before the commencement of yolk
formation. I examined several mackerel 12 to 13 inches in
length at Lowestoft, on October 8th, and although they were
evidently mature fish which had spawned the previous season
(May and June), under the microscope all the eggs were yolk-
less, and quite transparent, without any refringent globules.
On April 27th, 1896, I examined the ovary of a large
mackerel in London. This ovary was much enlarged and
approaching the ripe condition, but had not begun to dis-
charge its ova. Microscopic examination in the fresh condi-
tion showed that the majority of the eggs were much advanced
in development. They were full of spherical yolk globules,
and in the central region could be seen in many the oil of the
egg already fused into one large globule. In younger and
smaller eggs, a central more opaque zone contrasted with an
outer more translucent, as in the eggs of the gurnard, but
there were no transparent eggs with separate scattered oil

112 J. T. CUNNINGHAM,

globules. The earliest stages seen were nearly transparent,
with minute globules at the periphery, and also minute
globules not much more opaque around the nucleus. I con-
cluded that the latter consisted of oil, and that in the mackerel
the deposition of oily substance and yolk commences in these
different regions of the egg simultaneously, and does not take
place in any eggs until some time after the previous spawning
season. In the eggs of the sole, I have not observed at any
stage the marked contrast in the fresh condition between the
darker and lighter zones which is seen in the eggs of the
gurnard and other species. In immature and spent specimens
a considerable proportion of the transparent young eggs contain
numerous scattered globules which are certainly oil globules.
They have a yellowish colour, and most of them are situated
near the periphery of the egg, forming a somewhat well-
marked layer. ‘They can be seen even in the earliest stages in
prepared sections, but are much more conspicuous in the fresh
condition.

The yolk proper begins to be developed after the spawning
season. In a specimen examined at Lowestoft, on September
20th, it had made considerable progress. It begins as usual at
the periphery of the egg outside the oil globules, and the latter
in portions of the ovary examined in the fresh state remain in
all stages visible through the yolk, and are conspicuous on
account of their somewhat large size, refringent. character, and
yellowish colour. Fig. 13, Pl. 2, shows the condition of the
most advanced eggs, as seen in prepared sections, in the ovary
of the specimen mentioned above, which was 16 inches long,
and was obtained in the market. The external portion of the
ovum contains yolk in the form of minute granules; on the
inner side of the zone of yolk are the oil globules in groups
projecting into the protoplasmic zone, which is rather deeply
stained.

In the nearly ripe egg of the sole as seen in sections pre-
pared from an ovary which was in process of spawning, oil
globules are not easy to distinguish. In the figure of such an
egg which I have given in my Treatise on the Sole, they have

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES, 113

been overlooked altogether. ‘They are separate small vacuoles
distributed singly in the yolk. The yolk itself consists
throughout of rather large spherical globules, which in my
sections contain in their interior refringent granules.

Pelagic Eggs containing no Oil Globules.

In figs. 14 and 15 are represented two stages in the deve-
lopment of the yolk as seen in sections prepared from the ovary
of a large plaice taken from the aquarium, and killed on
August 25th. The zone in which the yolk is already de-
veloped is much more sharply defined from the inner proto-
plasmic zone than in eggs which contain oil globules. The
yolk layer continues to become thicker in proportion to the
layer of cytoplasm internal to it, until it reaches the surface of
the germinal vesicle. The formation of yolk globules com-
mences at the periphery of the cytoplasm, and in consequence
of the absence of oil globules, when eggs of this kind are
examined in the fresh state, no dark inner zone is seen. The
eggs become more and more opaque, but their opacity is nearly
uniform, except that the centre is usually more transparent in
consequence of the presence of the germinal vesicle. In the
ovaries of plaice and other fish, whose eggs. do not contain
oil globules, transparent ova containing scattered refringent
globules are never seen. In the immature fish all the ova are
perfectly transparent, with the exception of the occasional
aborted ova described in a subsequent part of this paper. In
the spent ovary, too, all the ova are transparent, except those
whose maturation has been arrested, and which are dead and
in process of absorption.

Structure of the Spent Ovary.

A spent ovary, that is one from which the annual crop of
ripe eggs has just been extruded, may be recognised from
several symptoms, of which the most certain is the presence of
a few ripe eggs detached from the walls of the ovary, but still
remaining in the cavity. When a portion of the germinal
tissue of an ovary in this condition is microscopically examined

vou. 40, pART 1,—NEW SER. H

114, J. T. CUNNINGHAM.

-in the fresh state, the most conspicuous characteristic is seen
to consist in a variable number of opaque, granular bodies of
irregular shape and structure. Investigation has proved that
these are eggs which have not fully matured, which have died
in situ, and which are not discharged from the ovary by the
bursting of their follicles, but are removed by absorption. In
the fresh material, when a number of fish are examined from
time to time, various stages in the disintegration of these eggs
are seen, from the condition in which the original structure is
but slightly altered, and which can only be distinguished from
that of the healthy eggs by the great opacity, to a condition in
which nothing is left but a small mass of opaque granules.
The appearance of two such dead yolked eggs as seen in the
spent ovary of a sole is shown in fig. 16.

The presence of dead aborted eggs is not, however, the most
essential characteristic of ovaries which are spent. Such eggs
occur, as will be explained below, in other stages of the ovary.
Aborted eggs are absorbed in situ, but healthy ripe eggs
escape from the ovary by the bursting of their follicles, and
only ripe eggs escape in this way. The essential characteristic
of spent ovaries is, therefore, the presence of empty collapsed
follicles from which the eggs have escaped. So far as my
observations go, these empty follicles are not to be detected by
microscopic examination of fresh material; in this all condi-
tions and stages of the eggs are easily seen, but the condition
of any part belonging to the connective tissue of the ovary is
very difficult to distinguish. In mounted sections, however,
the history of the follicle, after the escape of the egg, can be
studied with some success.

In the fish ovary the follicle, after the escape of the matured
egg, passes through changes similar to those which are known
to occur in the so-called corpora lutea of the mammalian
ovary. In the fish ovary the, degenerating follicle is always
found in connection with the superficial membrane of the
germinal lamellae. When the egg escapes, the interior of the
follicle opens on to the surface of the ovarian lamella, and
the wall of the follicle is thus restored to the condition from

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 115

which it started when the egg first began to develop—namely,
that of a portion of the superficial tissue of the germinal fold
or lamella.

The opening of the follicle, however, soon closes up, and
the whole cavity disappears by the contraction of the walls.
In follicles from which the eggs have only recently escaped, a
somewhat indefinite cellular tissue is seen, containing numerous
round nuclei. The appearance of the collapsed follicles in a
newly-spent plaice, as seen under a low power, is shown in
fig. 17, Pl. 3, while fig. 18 shows a single follicle of the
ovary of the grey gurnard more highly magnified. The
internal cellular tissue is detached from the inner surface of
the wall of the follicle, and presents a different appearance
from that of the wall itself. I think there can be little doubt
that this tissue is the remains of the follicular epithelium. 1
have not observed any indication in later stages of hypertrophy
or proliferation of this epithelium—in fact, it soon ceases to be
distinguishable. It has recently been maintained by J. Sobotta
that the corpus luteum of the mammalian ovary is produced
chiefly by the hypertrophy of the follicular epithelium, but this
conclusion, if correct, does not appear to me to apply to the
Teleostean ovary.

The wall of the empty follicle has the same structure as the
outer layer of the stroma of the ovary, of which it forms part.
It consists of connective tissue, somewhat fibrous in appear-
ance, containing numerous nuclei, and furnished with blood-
vessels. In its final stages, the follicle forms merely a globular.
projection inwards from the surface membrane of the ovarian
tissue. At this stage it is a solid mass, consisting of fibro-
cellular connective tissue, quite similar to that which forms
the wall of the follicle at the earlier stage. I have not
observed anything in its structure at the later stages which
appears to be derived from the follicular epithelium.

The absorption of the empty follicles proceeds rather rapidly,
and all trace of them has disappeared in ovaries which have
begun to mature the ova for the next season. On July 15th
I killed at Plymouth a plaice 16 inches long, of which the

116 J. T. CUNNINGHAM.

ovary appeared from its size and flaccid condition to have
previously spawned. The tissue was opaque white, and the
microscope showed it to contain numbers of partially yolked
eggs, the largest of which was *5 mm. in diameter. In
sections prepared from this ovary no trace of degenerating
follicles can be seen. They are also absent in sections from
an ovary of a large plaice killed on August 25th, which was
known to have spawned the previous season, having been kept
under observation.in the aquarium.

A female flounder which spawned in March in the aquarium
was killed on July 11th. The ovary contained only trans-
parent yolkless ova—that is to say, the formation of yolk in the
eggs to be shed in the following spawning season had not
commenced. But in prepared sections a few remnants of the
degenerating follicles from the previous spawning could be
discovered. It may be concluded, therefore, that in plaice and
fiounder the “corpora lutea” are entirely absorbed in about
three or four months.

In preparations from portions of the ovary of a spent sole,
preserved on board a fishing vessel fifty miles eastward of the
Humber, on July 24th, the degenerating follicles are numerous
and distinct. They are, however, small and solid, the cavity
having been obliterated. This ovary contains only a compa-
ratively small number of yolkless ova, the production of new
eggs having not yet taken place to any considerable extent.
The spawning of soles in the North Sea takes place chiefly in
May and June. In preparations from a sole taken off Lowe-
stoft, on September 24th, the degenerating follicles cannot be
detected, but there are distinct though small remnants of them
in sections from a specimen obtained on September 20th.

Tn a haddock 19 inches long, portions of whose ovary were
preserved at sea immediately after death on July 23rd, the
empty follicles had been entirely absorbed. The haddock
spawns in the North Sea principally in March, and there are
no immature specimens over 16 inches long.

OVARY AND OVARIAN OVA IN CERTAIN MARINE FIisHi“s. 117

Reabsorption of Aborted Eggs in Spent Ovaries.

I will proceed now to the consideration of the opaque granular
masses which are so frequently seen in the fresh germinal
tissue of fishes’ ovaries. As was previously stated, these masses
are eggs in which the development of yolk has proceeded to
some stage, and which have then died, and are in process of
reabsorption. On August Ist, 1895, I made a careful examina-
tion of the ovary of a sole 17 inches long, obtained in
Grimsby market, A sole of this size is invariably mature, and
it may be presumed, therefore, that this specimen had spawned
in the preceding spawning season—that is, about May or June.
The roe was thin and narrow, though of considerable length,
and when cut open presented the appearance suggestive of the
spent condition. It was red and congested, and here and
there were opaque yellow spots, which were evidently dead
yolked eggs. Microscopic examination of a portion of the
fresh tissue showed that it consisted chiefly of transparent eggs,
the larger of which were ‘16 mm. in diameter, and contained
scattered minute oil-globules. There were no eggs containing
yolk, except those which were evidently dead and had been
left behind in an incomplete state of maturation when the
spawning process was over. Fig. 16 represents the appear-
auce of a portion of the tissue under a low power. The
dead yolked eggs were from ‘29 to °33 mm. in diameter,
and were in various stages of yolk development, the
largest being quite opaque. As seen in the figure, they
had shrunken away from the walls of their follicles very
considerably.

I have not been able to make a complete histological study
in prepared sections of the history of these dead eggs and the
process of their absorption. I can only describe briefly the
condition in which I find them in a few specimens of different
species. The appearance they present in sections differs very
much according to the stage at which their development has
been arrested. In cases where they have ceased to develop at

118 J. T. CUNNINGHAM.

an early stage they are much less conspicuous in sections than
they are in the fresh condition. Among my preparations they
are most conspicuous in sections from a specimen of Trigla
hirundo caught on September 24th; the portion of the ovary
from which the sections were made was preserved in chromic
acid + per cent., at sea, immediately after death. In these
sections the dead eggs are very numerous, and contain a large
quantity of yolk. Each consists of a mass of yolk globules of
various sizes, contained in a follicle; between the yolk globules
are nuclei and cells, the latter not distinctly defined. The
process that is taking place in these dead eggs is evidently
closely similar to that which I have described as occurring in
an empty follicle. A proliferation of cells has taken place
from the walls of the follicle towards the interior, the cells
penetrating into the interior of the mass of yolk, and doubtless
effecting its absorption. The question arises whether these cells
are derived from the follicular epithelium or from the connec-
tive tissue of the wall of the follicle, and I consider the latter
alternative is probably correct. In these sections none of the
diminishing empty follicles are to be seen, and therefore I am
not sure that the specimen was normal; possibly for some
reason or other the discharge of ripe eggs had not taken place,
and they were all being reabsorbed in situ.

In sections from spent ovaries of the plaice preserved in
February, as in portions of the same examined in the fresh
condition, small ova in which the formation of yolk has com-
menced are to be seen. It is known from the condition of the
evary of the fish in later months that such eggs die and are
absorbed. But I have no preparations showing stages of the
absorption in the spent ovary in this species.

In preparations from part of the ovary of a weever (Tra-
chinus draco), preserved immediately after death at sea on
September 25th, both empty follicles and aborted eggs can
be seen in process of absorption. The empty follicles are
much reduced, and small, and consist of thick-walled capsules
full of cellular tissue. Here and there among the rather
small ova at the surface of the ovarian lamelle is seen

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 119

one which is undergoing retrogressive change. Neither in
these nor in the neighbouring normal ova can yolk granules
be distinctly recognised, but there are a few vacuoles which
probably represent small oil globules. The chief abnor-
mality in the degenerating ova is the condition of the ger-
minal vesicle, which consists of one Jarge nucleolar mass ;
the rest of the vesicle. membrane and clear contents, having
disappeared almost entirely. In the central part of the ovarian
lamellz were distinct masses of irregular shape and yellowish
colour, which had not taken the stain of the hematoxylin.
These consisted of yolk-like substance, with nuclei scattered
through it. Each was surrounded by connective-tissue fibres,
but not by a distinct follicular wall. These masses are the
remains of eggs which had reached an advanced stage of
maturation when spawning took place, and which have been in
great part reabsorbed. It is evident, therefore, that in the
spent ovary, eggs in various degrees of development die, and
are absorbed.

On September 17th I examined, in the fresh condition, an
ovary of the greater weever in which spawning had more
recently occurred, and in which, therefore, the reabsorption of
abortive ova had scarcely begun. Its condition supports the
interpretation given above. It contained (1) perfectly trans-
parent yolkless ova, (2) slightly opaque, apparently healthy
eggs containing a zone of granules, (3) dark yellowish granu-
lar masses, without refringent globules, evidently dead eggs
which had not been ripe when spawning was completed, and
(4) large rounded masses, quite opaque, and containing large
oil globules; these were evidently eggs which had nearly
reached the ripe condition before they died.

Reabsorption of Aborted Eggs in Immature
Ovaries.

I at first supposed that the opaque granular masses, identi-
fied as partially matured but aborted ova, only occurred in
spent ovaries. But it was afterwards found that they occur

120 J. T. CUNNINGHAM.

very commonly, if not always, in immature ovaries. I have
seen them in abundance in numbers of specimens of plaice
examined in November and December, in which no healthy
yolked eggs were to be seen; and at that time of the year the
ovaries of all mature specimens are full of eggs in which the
development of yolk is far advanced. At the same time none
of the plaice have begun to spawn, and no spent fish are to be
found. The specimens, therefore, whose ovaries have not begun
to mature in those two months, could not spawn until the next
season, and are therefore immature.

I have prepared sections from an ovary with dead yolked
eggs belonging to a plaice obtained in Grimsby market on
August 8rd, 1895. The specimen was 11% inches long, and was
caught on grounds not far from the Humber. As no speci-
mens less than 13 inches long from that part of the North Sea
have been found to be mature, it may be concluded that this
specimen had never spawned. In the sections, eggs in three
different conditions are seen. (1) There are small eggs of
healthy appearance, entirely protoplasmic, and, as is usual with
such young eggs, deeply stained with hematoxylin. These are
from ‘04 to:11 mm. in diameter, and have a normal and healthy
structure. The germinal vesicle in them has a regular definite
circular outline with nucleoli arranged round the periphery.
These young ova are not very numerous; they are distributed
singly just within the surface of the ovarian lamelle. (2)
There are next a number of larger eggs much less deeply
stained, and having a granular appearance; the largest of
these are about ‘18 mm. in diameter. There can be no doubt
that these are the largest transparent eggs seen in the fresh
condition of the same material. No yolk was visible in the
fresh condition, and none can be seen in the sections. But
the granular condition of the protoplasm, especially at the
periphery, indicates that these eggs have reached that stage
of maturation at which the deposition of yolk is about to com-
mence. The material was obtained from the market, and
therefore not preserved till some time after the death of the
fish. This doubtless accounts for the fact that in many of

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 121

the eggs of this kind the protoplasm has separated from the
germinal vesicle, leaving a space containing scattered granules,
and in some cases the germinal vesicle is absent altogether,
having been washed away in the process of preparation. But
where it is present it shows the usual structure, having a
distinct membrane, on the inner side of which are the nucleoli.
In many of these eggs the vitelline nucleus is very distinct,
situated at the periphery of the protoplasm.

The remaining eggs are in a remarkable condition, which is
illustrated in fig. 19. The shape is variable. ‘The remains of
the egg are seen forming a shrunken dense mass within the
follicle. The mass is more deeply stained than the eggs in
the condition previously described, and less granular in appear-
ance. Within the mass is a more translucent area containing
a single nucleolus. The nucleoli of the healthy condition
have apparently fused together, and the germinal vesicle has
degenerated. On the outside of the mass can be seen a
shrunken and crumpled membrane, evidently the membrane
of the ovum which had begun to form at the time when the
maturation was arrested. Within the wall of the follicle is
seen a distinct granular lining with nuclei here and there, but
no definite cell outlines. This I take to be the follicular epithe-
lium in a degenerating condition, although it is thicker and
more distinct than in either of the conditions previously de-
scribed. In the walls of the follicle are seen fibres of connec-
tive tissue and nuclei. There can be no doubt that these
structures are the opaque dead eggs, or “ masses”? seen in
examination of the fresh material, and it is evident that the
egg having reached the stage of maturation at which yolk
formation commences, has died, and forms within its follicle a
contracted opaque mass which is undergoing a process of
absorption.

That the condition of the germinal vesicle seen in these
aborted eggs in prepared sections is not artificially produced
by the process of preparation is proved by the fact that the
same condition has been observed in aborted eggs in the fresh
state. In fig. 20, Pl. 3, is shown the appearance presented

122 J. T. CUNNINGHAM.

by an aborted egg in the ovary of a plaice 124 inches long,
examined at Lowestoft on September 21st, 1895. The egg
has evidently died at a very early stage, when scarcely any
yolk had been deposited, and consequently it is not very
opaque. It appears also to have only recently died, and there-
fore to have undergone but little of the process of retrogres-
sive change. The nucleus, it will be seen, shows an irregular
indistinct outline, and a single nucleolus towards the centre.
The appearance of this aborted egg presents a marked contrast
with that of the three normal younger eggs figured with it.

Tne History or THE GERMINAL VESICLE AND YOLK
NUCLEUS.

To give a complete history of these structures and their
relations to one another, it would be necessary to follow con-
tinuously the history of the egg from its origin in the germinal
epithelium onwards. But unfortunately it is very difficult to
obtain a distinct and definite view of the parts of the young
ova in their earliest stages, in consequence of their small size
and the indefiniteness of the limits between different eggs and
different parts of each egg. I have endeavoured to trace the
origin of the ova in the germinal epithelium, both in the
ovaries of very young immature fish and in the spent ovaries
of mature fish, at the stage in which the new crop of ova is
beginning its development. Up to the present I have not
succeeded in tracing satisfactory indications of the division of
the germ-cells by which the ova must be produced, nor have I
been able to detect the yolk nucleus in the very young ova still
within the germinal epithelium, or but recently separated
from it.

If indirect or mitotic division of the nucleus were universal,
or if it occurred at least in all undifferentiated rapidly mul-
tiplying cells, it ought to be the mode in which the germ-cells
divide in the germinal epithelium. We know that it occurs
and is obvious and conspicuous enough in the division of

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES, 123

spermatocytes. The question of the occurrence of direct or
amitotic division in the multiplication of germ-cells is dis-
cussed by O. vom Rath, in a paper published in 1894 (18).
He states that many authors consider that this mode of
division occurs in genital cells. He points out that a decision
of the controversy is of great importance, since all modern
views on the nature of fertilisation and heredity are based
on the assumption of the continuous succession of mitotic
or indirect divisions in the history of genital cells. He
argues that if amitotic division did take place, an exact
division of the chromatin between the two daughter-cells
would be impossible, since the division of the threads does not
occur in that process as it does in mitosis. Nevertheless, vom
Rath has himself seen phases of amitotic division in genital
cells in the ovary of the Salamander, especially in very young
females, together with stages of regular mitosis; in his figure,
however, he shows none of the latter. He considers that
all the ova or germ-cells which undergo amitotic division
are abortive, are undergoing retrogressive changes, and are
about to be absorbed. In the ovaries of older specimens
of Amphibia, amitosis is more seldom seen, but degeneration
of ova without amitosis is common. Vom Rath considers
that the amitoses observed by other investigators in generative
organs were in many cases divisions of follicular cells, which
come to an end of their history when the genital cells are ripe or
are absorbed, and not of true germ-cells destined to become
ova or spermatozoa. In fact, he maintains the general theory
that amitosis only occurs in cells which are approaching the
termination of their capacity for division, and that when
amitosis of the nucleus once occurs noné of its descendants
ever again pass through the processes of mitosis, but soon
cease to divide at all, and ultimately die. Mitosis is, accord-
ing to this view, essentially connected with the continuity and
persistence of cell life.

The youngest stage of the Teleostean ovary which I have
examined is that of a plaice 3 inches long. The specimen was
killed in March, and must have been hatched in the previous

124 J. T. CUNNINGHAM.

year. The ovary was preserved with a mixture of chromic
acid =, per cent., and osmic acid #4; per cent. The eggs in
the sections from this ovary measure from ‘01 to ‘06 mm. in
diameter. In none of these eggs have I been able to distin-
guish the yolk nucleus with certainty. All of the eggs men-
tioned have passed the stage of subdivision, they are no longer
multiplying germ cells, but definite eggs in the stage of
maturation. The germ cells and germinal epithelium in
these sections are too much shrunken by the preserving
reagents, and too little differentiated by the staining to be
studied.

The nucleus of the smallest eggs exhibits a single rather
large nucleolus 2°5 « in diameter in the central part, though
not exactly at the centre, and some smaller, not very distinct
nucleoli which appear to be thickenings of the nuclear mem-
brane. In the larger eggs there are several nucleoli of very
different sizes; usually one is very large, reaching 7°5 m, or
even ‘Ol mm. in diameter, while the rest are much smaller. All
the nucleoli are situated close to the nuclear membrane, and
they are usually spherical in shape, but occasionally they are
hemispherical, the flat side being in contact with the nuclear
membrane. It is possible that the large nucleolus divides,
but I have only once seen a condition which could be con-
sidered a stage of division, namely, two nucleoli of equal size
almost in contact. Occasionally in these, as in other sections,
a nucleolus is seen outside the nuclear membrane, but when-
ever I have seen this I have seen evidence that the external
nucleoli had been removed from their proper position me-
chanically in the process of preparation. The central space
of the nucleus is occupied with a network of fine threads,
according to the prevalent view representing the chromatin,
the nucleoli being composed of a material somewhat different
in properties.

The sections I have next to describe were prepared from the
ovary of a plaice 7} inches long, killed in March. The tissue
was fixed in a mixture of chromic and osmic acid, each to the
strength of ;; per cent., the tissue remaining in this mixture

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 125

for about fifteen hours. The largest eggs in these sections are
‘15 mm. in diameter. In these the vitelline nucleus is dis-
tinct. The form of the egg is well preserved. The cytoplasm
is very homogeneous in appearance, and exhibits no trace of
yolk or of granular texture. The vitelline nucleus is situated
close to the periphery of the egg, and consists of a rounded
stained mass containing vacuoles in its substance. Its limit
from the cytoplasm is not perfectly definite, as it is evidently
continuous with the substance of the latter. ‘The germinal
vesicle 1s somewhat excentric in position in the egg, has a
distinct membrane, and a number of small nucleoli arranged
at nearly equal intervals inside the membrane. The reticulum
of the internal cavity of the nucleus is extremely fine, and
takes stains very slightly. The smallest egg in which I have
seen the vitelline nucleus in these sections is ‘08 mm. in
diameter, and in this case it is also situated at some distance
from the germinal vesicle. I have not been able to make it
out in any egg in contact with the membrane of the germinal
vesicle, although it is seen in that position, in what is evi-
dently its earlier stage, in fresh material treated with acetic
acid on the slide. I am obliged to conclude, that although
the preservation of the material in the sections above described
appears to be excellent, yet the mixture of chromic and osmic
acid mentioned is not capable of differentiating the vitelline
nucleus in the earlier stages of its history, and of demonstrat-
ing its origin. In these sections, as in those previously men-
tioned, the germinal epithelium is thin, flat, and indistinct, so
that it is vain to attempt to trace the origin of the ova
from it.

In the ovary of a mature flounder which had spawned in
March, and which was killed on July 11th, it is evident from
the sections that the production of the new crop of ova from
the germinal epithelium was going on. At numerous points
in the epithelium are seen groups of germ-cells, consisting of
two or more, each having an oval shape in section, and con-
taining a large nucleus, which in structure is similar to that of
the smallest eggs described in the ovary of the plaice 3 inches

126 J. T. CUNNINGHAM.

long. These germ-cells are about 8 w in their larger diameter,
but some of them are somewhat larger. Around them can be
seen the small nuclei of the ordinary non-germinal epithelial
cells, which in most cases extend over the surface of the groups
of germ-cells. The epithelium in the intervals between the
spots where germ-cells are seen appears to be only one cell
thick, and the cells composing it are extremely small. They
are not, however, so flat and thin as they are in other condi-
tions of the ovary, but show some thickness in section.

Some of the sections now under consideration are from
material fixed with corrosive sublimate and acetic, some from
material fixed with the strong mixture of Flemming. In none
can I distinguish actual figures of mitotic division in the germ-
cells, although I am strongly inclined to hold that vom Rath’s
view is correct, that the germ-cells always divide by typical
mitosis. I have seen division figures in these cells in the
germinal epithelium of Myxine and Conger. The preparations
of the generative organ of Myxine to which I refer were
obtained from very young specimens, in which the organ was
a very thin narrow fold. It contained anteriorly minute eggs,
and in the posterior part small testicular capsules. The
material was fixed in Flemming’s mixture, the sections stained
with hematoxylin. The germinal epithelium in these Myxine
sections is limited to the extreme edge of the genital lamina,
and in section is somewhat crescentic. It is several cells
deep, but the cells are irregularly arranged, and are poly-
gonal in shape, not flattened. The germ-cells in the resting
state are about 13 » in diameter, the greater part of which is
taken up by the nucleus ; the interstitial cells are somewhat
smaller.

No investigator appears yet to have traced out completely
the division of the germ-cells in the germinal epithelium of the
vertebrate ovary, and their conversion into definite ova
surrounded with follicle cells. Vom Rath figures only two
sections, one of the ovary of a young female Salamander, one
of an undifferentiated reproductive ridge of the same animal.
In the former the structure represented seems to me to corre-

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 127

spond to the stroma of the ovary containing definite ova, and
not to include the germinal epithelium properly so-called.

G. Born (7) mentions briefly that in the ovary of Triton
teniatus, between the larger eggs are primitive ova, or nests
of them; that he often saw in these primitive ova stages of
mitosis, and that he also saw lying in contact with the resting
nuclei of primitive ova a mass of finely granular protoplasm
containing some more distinct granules. But he gives no
further consideration to these points, his paper being devoted
to the later history of the ovum.

R. Fick (10) and Riickert (8) also confine themselves to the
discussion of the history of the definite ovum after it has left
the germinal epithelium and ceased to divide.

I have not been able to trace the centrosome in the primitive
ova still within the germinal epithelium, nor have other ob-
servers described it in this stage in Teleosteans. It is still a
disputed point whether the centrosome is a permanent organ
of the cell, that is to say whether it is always present in the
cytoplasm throughout all the successive phases of cell-life,
dividing before the nucleus when division takes place. The
majority of authors believe in the persistence of the centro-
some, and hold that it is always outside the nucleus. But
O. Hertwig maintains that this is true only in certain cases,
and that usually after mitosis the centrosome passes into the
interior of the nucleus, and only emerges again into the proto-
plasm when the cell is preparing to divide again.

Beneath the epithelium in the ovarian lamellz in these
sections is a layer of young ova, all destitute of yolk, of
various sizes, the largest being ‘14 mm.in diameter. In the
larger of these the vitelline nucleus is very definite and con-
spicuous, forming a rounded or elliptical body, somewhat
deeply stained and having a granular appearance. Sometimes
the granular appearance is seen to be due to vacuoles, and
occasionally I have fancied I saw a minute corpuscle in the
centre of the mass. As usual, the edges of the body are not
sharply defined, but are seen to be connected with the strands
of the surrounding cytoplasm. In the smaller eggs there is

128 J. T. CUNNINGHAM.

no body separate from the germinal vesicle, but there is
frequently to be detected a cap of differentiated cytoplasm
applied to the external surface of the membrane of the ger-
minal vesicle. There can be no doubt that this cap of
cytoplasm, which is granular and deeply stained, is the earlier
condition of the vitelline nucleus. The two stages described
are represented in fig. 21, as they are seen in two contiguous
ova in one of the sections. The smaller ovum is ‘037 mm.
in diameter, the larger ‘098 mm.

In a previous paper (20) I have briefly described the relations
of the vitelline nucleus in the ova of the common pipe-fish,
Syngnathus acus, so far as they can be ascertained by the
examination of fresh material. Although I have not been
able to trace the history of the ova in this form completely,
I am able, after examining several ovaries by means of sections,
to confirm and extend the results previously recorded. The
ovary of Syngnathus is a cylindrical tube of narrow diameter,
and its structure is remarkable on account of the narrow limits
to which the proliferating germinal epithelium is confined.

There is but one germinal lamina which extends along the
ovarian tube lengthwise, and germ-cells are present only at
the extreme edge of this lamina. The ova, when separated
from the germinal epithelium, pass in succession towards the
base of the lamina, and then into the wall of the ovary as
they grow larger. At least, it is certain that the largest and
most advanced eggs are found in the wall of the ovary, and
that the free projecting lamina contains a row of eggs diminish-
ing in size, and descending in degree of maturation towards
the extreme edge. The arrangement may be partly or wholly
due to the growth of the germinal lamina outwards, and not
to the passing of the ova inwards, as the fact that the oldest
parts of a herbaceous plant are nearest to the root is due to
the fact that growth takes place at the apex. I have not
attempted hitherto to determine positively the mode of growth
in the ovary, but the resulting arrangement is that represented
by the section of the germinal lamina shown in fig. 22. The
section from which the figure was taken was prepared from

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 129

material preserved in a mixture containing ;4, per cent. chromic
acid and 40 per cent. picric acid. The condition of the yolk
in the three largest eggs is the effect of the picric acid, which
destroys the original form of the yolk globules, and in sections
prepared with this reagent the vitellus usually exhibits the
laminated appearance shown in the figure. But it generally
happens that when the yolk in the larger eggs is well pre- .
served, the young protoplasmic eggs are shrunken and dis-
torted, and vice versa. In one of the younger eggs there
are seen two very distinct vitelline nuclei or corpuscles. The
sections were treated with the triple stain recommended by
Flemming—gentian violet, safranin, and orange G,—and the
vitelline nuclei are brightly coloured with the safranin, while
the remaining parts show only a light tinge derived from the
orange. ‘The vitelline nuclei are round, and differ from those
seen in the ova of Pleuronectes in their sharpness of contour.
Their outline is very definite, and appears to have no connec-
tion with the surrounding cytoplasm.

The ovum in which the two vitelline nuclei are seen is
"16 mm. in diameter. Although the presence of two of these
bodies in one ovum is not uncommon, it is more usual to find
only one, and there can be little doubt that the presence of
two is due to the division of a single one. In the examination
of fresh material, treated with acetic acid, I have sometimes
seen three, and even four of these bodies ; but when there are
several they are proportionally smaller than when only one is
present. The earliest condition in which I have been able to
demonstrate the structure is, as in other cases, that in which
it is in contact with the outer surface of the membrane of the
germinal vesicle. Fig. 23 shows the smallest ovum in the
section represented in fig. 22 more highly magnified. The
greatest diameter of this ovum is ‘(049 mm. The vitelline
nucleus is seen as a little, hemispherical, highly-stained body
attached to the membrane of the germinal vesicle; around it
is a region of denser cytoplasm. It may be remarked that
even if it should prove that the vitelline nucleus is derived
from the germinal vesicle, it is evidently not of the samc com-

vot. 40, part 1.—Nuw SER. I

130 J. T. CUNNINGHAM.

position as the nucleoli, as in these sections it has taken a
very different colour in staining. In none of my sections of
ovaries of Syngnathus are the germ-cells sufficiently differen-
tiated to enable me to trace the centrosome in them, or decide
as to the origin of the vitelline nucleus.

Henneguy (11) has recently studied the vitelline nucleus in the
ova of Teleosteans and other classes of Vertebrates. In young
ova of the trout he finds it is a round body at some distance
from the germina! vesicle, and consisting of a central part
deeply stained, and an external zone in whicb the stain is less
intense. In Syngnathus acus he finds its earliest condition
in very young ova is that of a refringent corpuscle in contact
with the germinal vesicle. In his text he does not connect
the structure with anything previously existing in the germ-
cells either in a state of division or in the resting state, but
concludes that it arises from the germinal vesicle, and gives to
it a very far-fetched and fanciful interpretation—namely, that
together with the nucleoli of the germinal vesicle, it repre-
sents the macro-nucleus of the Infusoria, the chromatic net-
work representing their micro-nucleus. But Henneguy gives
a figure of a section from the ovary of a newly-born kitten
which goes far to prove the identity of the vitelline nucleus
with the centrosome. At the time he wrote his paper, Bal-
biani’s paper (12) mentioned below was not published, and the
doctrine of the centrosome was not so far developed as it has
been since. It is not surprising, therefore, that the author
overlooked the significance of the figure to which I refer.

In this figure are shown germ-cells in process of mitosis
with a centrosome at each end of the spindle, and others in
the resting-stage with a minute stained corpuscle at the side
of the nucleus. There is nothing to distinguish this corpuscle
from the centrosome of the dividing cell on the one hand, and
on the other from the vitelline nucleus or corpuscle shown in
other figures of the same plate in ova contained in definite
follicles, except that the vitelline nucleus is somewhat larger.

The history of the structure has also been studied by Mr.
Jesse W. Hubbard (15) in the ova of Cymatogaster aggre-

OVARY AND OVARIAN OVA IN OERTAIN MARINE FISHES. 131

gatus, Gibbons, one of the viviparous Embiotocidz of the
Pacific coast. The eggs of this fish are very small, and develop
scarcely any yolk. The earliest stage of the vitelline corpuscle
which Hubbard could discover was, like that which I have
described, a crescent-shaped body, fitting closely to one side
of the nucleus. The eggs in which this condition was observed
were 20 uw, or ‘02 mm., in diameter. In later stages the body
was found at a distance from the germinal vesicle. It tra-
velled towards the periphery of the ovum, and remained visible
even in the period of segmentation. Hubbard concludes that
the body originates from the nucleus.

The later history of the vitelline nucleus in Pleuronectide
is as follows:—It moves away from the germinal vesicle to-
wards the periphery of the ovum, as seen in fig. 21. When
the deposition of yolk commences in the peripheral layer of
cytoplasm, the vitelline nucleus is seen close to the inner limit
of the yolk, and the body then assumes somewhat the form of
a sphere at the apex of a cone. The conical portion is in con-
tact with the yolk layer at its base, and careful examination
shows that it is continuous with the strands of cytoplasm
which separate the yolk globules. This condition is shown in
figs. 10 and 14. Usually the cone can be seen to give off
divergent strands which pass into the cytoplasmic network,
and then the vitelline body reminds one forcibly of the form
of an octopus, with its arms extending into the yolk. In the
spherical portion of the body vacuoles appear, and these ap-
pear to be similar to those which contain the yolk globules,
but I am not certain whether they actually contain yolk sub-
stance or not. As the yolk layer increases in thickness, the
vitelline body becomes completely surrounded by it, and is
then detected with some difficulty as a little island of cyto-
plasm rather more deeply stained than the rest, and compara-
tively free from yolk. After the yolk has reached the surface
of the germinal vesicle, I have not been able to detect the
vitelline nucleus.

The structure in question is very much less conspicuous in
the ova of the grey gurnard in my preparations than in those

i332 J. I. CUNNINGHAM.

of Pleuronectide. 1 have been able after some trouble to
convince myself that it exists, and has similar relations in this
species, but have only been able to detect it occasionally in
the larger of the yolkless ova, which were lightly stained.
My material was fixed with picro-sulphuric acid, and with
chromic acid alone; the latter is not very effective for the
demonstration of the body in Pleuronectid ovaries, but the
former brings it out very clearly in them in certain conditions
of the ovary. It would perhaps be more easily seen in the
ova of the gurnard after the use of a fixing mixture containing
acetic acid.

In Syngnathus I have not seen any trace of the vitelline
body or bodies after the appearance of the yolk. As described
above, the yolk globules in this form appear uniformly through-
out the cytoplasm from the first, not as a peripheral zone
increasing in thickness, and I have been unable to discover
the remains of the vitelline bodies among them. The largest
ovum in which I have seen it was ‘25 mm. in diameter, and in
this the formation of the yolk was just commencing.

According to Boveri, both the centrosomes of the first division
spindle in the fertilised egg arise from the spermatozoon,—
that is to say, the spermatozoon becomes in the egg a male
pronucleus and a centrosome, and the latter divides to form
the two centrosomes of the first segmentation spindle. The
ripe egg on this view possesses no controsome, and for this
reason is incapable of self-division.

Balbiani, the original discoverer of the vitelline nucleus, in
1893 identified this body as the centrosome of the ovum. He
concludes that it arises from the nucleus as a little bud at the
moment when the ovum quits the germinal epithelium. He
points out that the vitelline nucleus condenses around it a
portion of the cytoplasm more dense than the rest; and this
surrounding layer he compares to the archoplasm or attractive
sphere around the centrosome. The vitelline nucleus may be
double, and the centrosome is also often seen to be double in
resting cells. The increase in size of the vitelline nucleus is
interpreted as hypertrophic degeneration. Balbiani considers

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 133

that the degeneration of the vitelline nucleus, that body being
the centrosome of the egg, accounts for the absence of the
centrosome in the fertilised egg; and for the fact, asserted by
Boveri, that the centrosomes of the segmenting egg are both
derived from the spermatozoon.

These views are open to several serious objections. There
is first the uncertainty with respect to the origin of the vitelline
nucleus. Balbiani argues that since the vitelline nucleus arises
from the nucleus proper, the centrosome in an ordinary cell
has a similar origin, a fact which he thinks explains the im-
portant part taken by the centrosome in the division of the
cell. But the latest and most definite researches indicate the
constant presence of the centrosome in the cytoplasm, and if
this is correct then it ought to be possible to trace back the
vitelline nucleus into continuity with the centrosome of the
germ-cell. An equal difficulty presents itself in the interpre-
tation of the later history of theegg. There is ample evidence
that the vitelline nucleus disappears in the yolk, and takes no
further part in fertilisation or segmentation. Yet the polar
bodies are formed by mitotic division before fertilisation. If
the vitelline nucleus is the effete centrosome of the ovum,—and
if centrosomes are such a constant and important feature in
mitotic division,—how is it that the nucleus of the ovum is able
to divide twice in the formation of the polar globules? Boveri
(3) concluded that the centrosome of the ovum disappeared
before fertilisation, and pointed out that in some eggs the direc-
tive spindle was destitute of polar radiations at its extremities,
But numerous observers have seen radiations and centrosomes
at the extremities of the directive spindle, and Boveri suggests
that the female centrosome, when present, is no longer capable,
after the formation of the polar globules, of performing its
functions, i.e. of taking part in cell division; it disappears,
and the male centrosome takes its place. This suggestion has
been demonstrated to be actually correct by A. D. Mead (16)
in a paper published in 1895. ‘This observer states that he
followed the degeneration and disappearance of the egg
centrosome,—that is to say, the inner centrosome of the

134 J. T. CUNNINGHAM.

second directive spindle. These facts, assuming them to
be firmly established, have a very important bearing on
Balbiani’s theory of the significance of the vitelline nucleus.
Since, at any rate in the eggs of many animals, the directive
spindles are provided with centrosomes, of which the one
belonging to the female pronucleus disappears before the for-
mation of the first segmentation spindle, it is clear that the
disappearance or degeneration of the vitelline nucleus can have
nothing to do with the absence of the female centrosome in
fertilisation. We must consider the question of the identity
of the vitelline nucleus with a centrosome on other grounds.
If the centrosome were a permanent organ of the cell, and
the vitelline corpuscle were identical with the centrosome of
the ovum, the vitelline corpuscle having disintegrated, or, at
any rate, having been removed from the neighbourhood of the
nucleus, fixed in a distant part of the cytoplasm, and usually
surrounded by yolk, the directive spindle could not possess
centrosomes. For it is certain that the vitelline nucleus does
not return to the vicinity of the germinal vesicle. and again
take part in its changes. If follows, therefore, either that the
centrosome is not a permanent part of the cell, or that the
vitelline nucleus is not to be identified as the centrosome. Now
if the vitelline nucleus is not the centrosome it is certain that
there is no other body visible in the ovum outside the germi-
nal vesicle which can be identified with the centrosome. The
suggestion that the centrosome exists within the germinal
vesicle is inconsistent with the theory of the persistence of the
centrosome outside the nucleus. The history of the ovum, then,
between its first definite constitution and the formation of the
polar bodies, actually disproves the theory of the unbroken
continuity of the centrosome as an extra-nuclear body. It is
an observed fact that the directive spindle, whether first or
second, possesses centrosomes, and yet no body can be dis-
covered in the egg at an earlier stage which ean be identified
as a centrosome, except the vitelline nucleus, which is known
to degenerate and disappear. The centrosomes of the directive
spindle, then, must be formed as such at the time of their

OVARY AND OVARIAN OVA IN OERTAIN MARINE FISHES. 135

appearance—must bea new modification of some part of nucleus
or cytoplasm.

If centrosomes can be formed anew thus on the occasion
of one mitosis, there is apparently no reason why they should
not be newly formed at each mitosis. Further, we know that
the centrosome of the ripe egg—that is to say, the inner centro-
some of the second directive spindle—degenerates. It may be
urged that this is a special case, that this centrosome dege-
nerates as a preliminary to fertilisation, because the centro-
somes of the first segmentation spindle are supplied by the
sperm. But since the centrosomes of the directive spindle are
formed anew, the preceding centrosome must have disappeared.
Therefore there is nothing to contradict the suggestion that
this preceding centrosome is the vitelline nucleus, which dege-
nerates. And if the centrosome degenerates after one mitosis,
there is no reason why it should not degenerate after every
mitosis. On these grounds the theory might be suggested
that the centrosome is not persistent, but is a structure formed
by the changes which precede mitosis, and that after division
it disappears, more or less gradually in the cytoplasm. This
theory does not, however, appear to be applicable to all mitoses,
for in the segmentation of the fertilised ovum, according to
the latest observations, the centrosome and archoplasm persist
after each division, remain during the reconstitution of the
resting nucleus, and then divide to form the new spindle of
division and the centrosomes of its two extremities. It is
definitely and particularly stated by A. D. Mead that this
occurs in the segmentation of the ovum of Cheetopterus.

We must conclude, then, that although the extra-nuclear
persistence or continuity of the centrosome has been observed
in the segmentation of the ovum, and may occur in the mitosis
of all somatic cells, it certainly does not occur in the matura-
tion of the ovum. The centrosomes of the directive spindles
are not formed from a pre-existing extra-nuclear body. In
this case the centrosome of the ovum must have passed into
the interior of the.germinal vesicle, or must have degenerated
in the cytoplasm, There is no evidence of the inclusion of

136 J. T. CUNNINGHAM.

the centrosome within the germinal vesicle, while there is
some, although not complete, evidence of the identity of the
vitelline nucleus with the centrosome left at the last division
of the germ-cell. The centrosome of the ovum then de-
generates and disappears before the two polar divisions, and
this suggests the question whether there is anything in the
process of spermatogenesis which corresponds to this occur-
rence in the ovum.

It has been found that the reduction divisions of the sper-
matocyte take place in the same way as in the ovum—that is
to say, before the last two divisions there appear half the
proper number of chromosomes, each of which consists of
four distinct particles, and after the divisions one of these
particles is contained in each spermatozoon.

According to O. vom Rath there are four generations or
divisions of the male germ-cells before the last two, or reduc-
tion divisions, which take place by what he terms “heterotypical
mitosis,” because of its special character. He finds the cell
of the fourth generation larger than that of the third, and
believes that there is a resting and growing phase after the
third generation. But he does not mention any degeneration
or disappearance of the centrosome between the fourth division
and the first reduction division, and this is the interval where
such an occurrence is to be sought. He figures one of the
large resting cells of the third generation, and says that often
near the resting nucleus at a spot where the cytoplasm has a
coarse granular appearance and dark colour, he has seen two
round bodies, which he takes to be two attraction spheres
with their centrosomes, as they are too large for centrosomes
alone.

According to vom Rath, the tetrads or groups of four chro-
matin bodies which are present at the commencement of the
reduction divisions in the spermatocyte are formed directly
without a resting phase from the dyaster of the previous
division. ‘Thus the process does not agree with that which
takes place in the ovum. But he points out that in the growth-
period of the ovum, according to certain observations, the

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 137

nucleus does not pass through a genuine resting phase. ‘Thus
Val. Hacker believed that in the ova of certain fresh-water
Copepods throughout the growth-period a double chromatic
thread could be traced in more or less evident form, and that
it was directly derived from the dyaster of the preceding
division, so that the division of the chromosomes (tetrads) in
the first reduction division was prepared in the previous
mitosis. Riickert has reached similar conclusions in the
study of the ova of Selachians. No further information being
apparently available concerning the history of the centrosome
in the spermatocyte in the period just before the reduction
divisions, we may proceed to consider more precisely the
history of the germinal vesicle in the ovum, and may con-
veniently refer to Riickert’s observations on the Selachian
ovum as affording the most definite and comprehensive view of
the subject, and then inquire how far they harmonise with
what I have been able to see in the Teleostean egg.

Many authors have maintained that the nuclear network
temporarily vanishes in the germinal vesicle. Riickert re-
marks that the reticulum which is considered to be the vehicle
of heredity, and has been studied to the farthest possible
detail in the final stage of its development, was not known
with certainty, when he began his investigation, even to exist
in the young stages of the egg. In 1890 Boveri, considering
each tetrad as a single chromosome, proved that the reduc-
tion to half the normal number of chromosomes had already
taken place when the first directive spindle appeared, and
maintained that the reduction must take place in the ger-
minal vesicle, if not at some earlier stage. But he had not
minutely traced the fate of the chromatic substance in the
germinal vesicle.

O. Hertwig, on the other hand, considered that the number
of chromosomes was doubled, which means, of course, that
each member of the tetrad represents a chromosome. Then
the reduction follows from the two divisions without the
interposition of a resting phase or further splitting of the
chromosomes.

138 J. T. CUNNINGHAM.

Riickert’s investigations were carried out principally on
ovaries of Pristiurus at Naples, though he also examined
ovaries of Scyllium and Torpedo. For fixing he employed
Hermann’s osmic mixture, sublimate, and sublimate with 5
per cent. acetic. The youngest ova he describes are 28 u in
diameter. In these eggs a membrane was not seen around the
germinal vesicle in material fixed with sublimate, but was
distinct enough in osmic preparations. The cavity of the
vesicle contained some small gleaming nucleoli, and a distinct
and still stainable chromatin reticulum. The latter consisted
of separate unbranched chromosomes of fairly uniform thick-
ness and bent into undulating curves. These formed a tangle
(knaiiel). In this respect the structure of the germinal
vesicle differed from that of the ordinary resting nucleus, and
approached the tangle phase of mitosis ; from which, however,
it was distinguished by the less compact form of the chromo-
somes. There was never in the growing egg a branched chro-
matin network like that which can be demonstrated in the
ordinary resting nucleus, although this period in the history
of the egg extends over a long period of time; in the case of
eges which ripen late in the fish’s life, over several years.

Riickert was not able to count the chromosomes at this
stage exactly, but estimated their number at thirty to thirty-
six, which was about the same as in the somatic cells of the
embryo of the same species. The chromosomes in the ger-
minal vesicle were not all of the same size, some being remark-
ably small and slightly stained.

The first period of the development of the egg is from the
stage described above to that of eggs which are 14 to 2 mm.
in diameter, in which the germinal vesicle is at its maximum
size, namely, mm. in diameter. The membrane of the
vesicle becomes thicker, and is thrown into grooves and folds,
which last,if examined in section, might be taken for outgrowths
of the vesicle. The nucleoli increase in number and size, and
after being at first chiefly peripheral, gradually aggregate in a
particular part of the vesicle, usually that towards the surface
of the egg. The network of the vesicle becomes almost in-

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 139

visible, and stains scarcely at all. It is this condition which
has led to the conclusion that the chromatic network dis-
appears ; but Riickert says that the chromosomes are converted
into much expanded structures, consisting of numerous trans-
verse threads along an axis, so that each chromosome resembles
the cylindrical brush used for cleaning lamp chimneys. The
change is effected by the microsomes first becoming elongated
into transverse rods, or perhaps discs, and then further breaking
up into the thin curved threads. In eggs } to 1 mm. in diameter
the tangle of chromosomes again becomes more distinct. To
see the general distribution of the chromosomes, whole germinal
vesicles must be isolated and examined, because a section
takes in only a very thin slice of the large structure. It is
now found that the chromosomes are arranged in pairs, and
there are as many pairs as there were chromosomes in the
earlier stage, which shows that each pair has been produced
by the division of a chromosome. The position of the members
of each pair indicates that the division of the chromosome has
been longitudinal.

The second period of the development is to be followed in
eggs from 2 or 3 mm. in diameter to the full-grown condition
when they are 14 to 16 mm. in diameter. The vesicle becomes
smaller, and reaches the surface of the egg, against which
it flattens itself to some degree. The chromosomes remain
in pairs, but become much shorter and thinner, and they
concentrate towards the centre of the vesicle.

In eggs of 14 mm. diameter the coil or tangle formed by the
chromosomes is only 36 « by 8 u in dimensions, while the whole
vesicle is 296 uw by 148 uw. The external part of the feathery
structure of the chromosomes becomes stainless, while the
central part stains more deeply. Rickert regards the changes
in this and the preceding period as to a great extent due toa
divergence and convergence of particles. Finally, the chromo-
somes become converted into portions of chromatin, consisting
of distinct granules, but having the form of dense masses, not
elongated loops or cords; in some of the masses there are
indications that each consists of four rods.

140 J. T. CUNNINGHAM.

The author finds that during the enlargement there is a great
increase of substance, during diminution a great loss of sub-
stance, in the chromosomes. He regards this temporary sub-
stance as the somatic plasm, whose function is to govern or
cause the growth of the egg and the accumulation of the yolk;
when this object is accomplished it merges in the cytoplasm,
i.e. becomes ordinary cytoplasm, and the chromatin which is
left is that which is concerned in fertilisation and heredity.

The nucleoli experience in the second period of the develop-
ment of the egg a reduction in mass, which proceeds at a rate
equal to that of the diminution of the chromosomes. The
nucleoli diminish in size, become paler, and then vanish alto-
gether, following the chromosomes during the process towards
the centre of the vesicle.

According to B. Holl, in the human ovum the chromosomes
disappear altogether, and the nucleolus (here single) takes
their place, breaking up into little spheres which become the
chromosomes of fertilisation. But Riickert considers that
this is an error.

The third period in the history of the ovum, according to
Rickert, is the formation of the polar bodies. At the be-
ginning of this period the chromatin-figure is surrounded by
a remnant of the germinal vesicle without any membrane, and
much smaller than the original vesicle. The chromatin forms
a dense heap, 6 in diameter. It seems at first to be homo-
geneous, but is really a dense coil formed of the rods of the
previous stage. The heap separates again into separate cor-
puscles, and these form the equatorial plate of the first direc-
tive spindle.

In a later paper (1892) Rickert has described how he en-
deavoured to ascertain the stage at which the doubling, or
longitudinal division of the chromosomes took place. He
found that they were already double in eggs of § to 4 mm., in
which the chromosomes are most indistinct, and, indeed, in
ova fixed with sublimate, invisible. They can, however, be
traced in the form previously described in eggs fixed with
Flemming’s mixture. He points out that in younger eggs the

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 141

chromatin network, although distinct and well stained, is so
dense, and exhibits so many crossings in all directions, that it
appears to have the structure characteristic of the resting
nucleus. He declares, however, that by means of the oil-
immersion lens he was able to resolve the network into a
tangle (kniuel) of chromosomes, and that in thin sections of
the germinal vesicle some of these being included in their
whole length can be seen to be double. He was even able to
see traces of doubling in the daughter tangles of the last
mitotic division of the germ-cells, and comes to the conclusion
that the germinal vesicle in the egg during its enlargement
and growth is not a resting nucleus, but a daughter tangle of
the germ-cell enlarged to enormous dimensions.

The history of the germinal vesicle in Teleostean ova,
according to my observations, agrees in many respects with
that described by Riickert in Selachians. There are certain
well-known features which are common to both cases, in
particular the conspicuousness and deep staining of the nu-
cleoli, and the indistinctness and unstained condition of the
nuclear network during a considerable period of the egg’s
growth. As I have already described, in the smallest ova in
my preparations, about ‘Ol mm. in diameter, the germinal
vesicle exhibits one large conspicuous nucleolus, with others
much smaller, and a nuclear network. The structure of these
germinal vesicles or nuclei appears to resemble that of a resting
nucleus in an ordinary cell, and 1 have not been able to dis-
tinguish chromosomes in them as described by Rickert in the
ovum of Selachians. I cannot assert that the network does
not consist of separate chromosomes, but although I have used
sections from material fixed in Flemming’s mixture, and have
studied them with an apochromatic oil-immersion of 2:0 mm.
by Zeiss, I have not succeeded in resolving the network into
chromosomes. Riickert maintains that with sufficient magnify-
ing power and careful scrutiny the apparent network can
always be seen to consist of a tangle of chromosomes, and
that the germinal vesicle is, in fact, an enormously enlarged
tangle-phase or spirem, and therefore essentially different from

142 J. T, CUNNINGHAM.

an ordinary resting nucleus. But it may be suggested that
possibly the individual chromosomes could be distinguished in
the ordinary resting nucleus, if it was re-examined with suffi-
cient care. It may, on further investigation, be found that
the fibres of the network of the resting phase are only appa-
rently branched and anastomosed, and that it really consists of
chromosomes, the limits of which do not entirely disappear. On
the other hand, if the network of the germinal vesicle consists
of chromosomes, we have to inquire what is the origin of the
membrane and nucleoli. The existence of these structures
cannot be ignored. The theory that the chromosomes after
the last division of a germ-cell place themselves in a convoluted
series to form a tangle or spirem, without entirely losing their
individual independence, does not account for the appearance
of the nucleoli or of the membrane. There are two views of
the reconstruction of the resting phase after the mitosis of an
ordinary nucleus. According to the most detailed recent
descriptions of the process in the blastomeres of a segmenting
ovum, the chromosomes become vacuolated, and form little
vesicles, which fuse together, and so give rise to the nuclear
network, membrane, and nucleolus. According to Flemming’s
original scheme, the stages of reconstruction were the same as
those of division, but occurred in the opposite order: the
chromosomes united into a continuous thread, which acquired
the intricate convolutions of the tangle-phase or spirem, and
then resolved itself into nuclear network, membrane, and
nucleoli. According to Rickert’s views, neither of these modes
of reconstruction occurs in the germinal vesicle, the chromo-
somes being always distinct. But perhaps while preserving
their individual existence these bodies give up and afterwards
regain a large portion of their substance, which goes to form
the nucleoli.

Riickert describes the daughter-tangle after the last division
of a germ-cell, as consisting of a number of bent chromosomes
with their bends all converging towardsa “ polarregion,” in which
a large nucleolus was often to be seen. He says nothing concern-
ing the origin of this nucleolus, but it is certain that the chromo-

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 148

somes without a nucleolus are not the same as chromosomes
plusanucleolus. It is not probable that the nucleoli are formed
from the achromatic elements; on the contrary, there is much
evidence indicating their connection with the chromatin.

If the chromosomes are of great physiological importance,
it must be because they influence the life of the cell, they
must take part in the general metabolism of the cell. If they
were merely fixed elements which were divided and trans-
mitted in permanent form from generation to generation of
cells, it would be difficult to understand how they could affect
the properties and qualities of the cell. To say that the chro-
matin is the substance of heredity, merely means that judging
from the history of the spermatozoon in fertilisation, all the
peculiar qualities which are transmitted from parent to off-
spring must be contained in the chromosomes of the pro-
nucleus. Even this is not strictly true, because there are the
centrosomes to be taken into account. But the number and
form of the chromosomes are nothing unless we know how
they act upon the cell and determine its behaviour.

The small ovum in fig. 21, Pl. 3, shows the appearance of the
germinal vesicle in ova of the flounder of ‘037 mm. diameter.
The reticulum is but slightly stained, more by hematoxylin than
by safranin, the large nucleolus is deeply stained, and the cyto-
plasm also takes stains very deeply at this stage. In this respect
the cytoplasm of the ovum at this stage differs from that of
other cells, and from that of the ovum at other stages. The
cytoplasm is very dense and almost homogeneous in appearance.
The other ovum represented in fig. 21 is ‘(098 mm. in dia-
meter. The germinal vesicle in it is much larger than in the
smaller egg, there are several small nucleoli in contact with
the membrane of the vesicle, and there is a nuclear network,
some strands of which are more stained and more conspicuous
than the rest. The cytoplasm is granular, and after the
action of chromic and osmic acids scarcely stained, but in
material fixed with picro-sulphuric or sublimate it stains con-
siderably. In the stage here considered there is a distinct
membrane.

144, J. T. CUNNINGHAM.

The following stages are to be seen in sections made from
the ovary of a plaice killed on August 17th. The development
of yolk was advanced in many of the ova. The best prepara-
tions are from portions fixed with chromosmic mixture or
Kleinenberg’s picro-sulphuric acid. In ovaa little larger than
those last mentioned—namely, from about ‘16 mm. in diameter
upwards—in the interior of the germinal vesicle there can be
distinguished a central and a peripheral region. In the central
region can be seen, even with a low power, indications of
stained fibrils, which are wanting in the peripheral region.
Examination with higher powers shows that these fibrils form
apparently separate lengths or loops, whose direction is very
irregular ; sometimes they can be seen to be in pairs, and with
the immersion lens they have a feathery appearance. These struc-
tures clearly, therefore, correspond to those described by Riickert
in the Selachian ovum, and identified by him as chromosomes.
The impression I have obtained by comparing the ova in this
condition with the younger is that the vesicle has expanded, while
the fibrillar network has remained of the same size, so that a
space has been left between the border of the network and the
membrane of the vesicle. In the younger ova the nucleoli are
all in contact with the membrane of the vesicle, but in later
stages they travel towards the centre, and are found in the
region of the chromosomes. The substance between the fibrils
and in the peripheral region of the vesicle appears to be finely
reticular. Fig. 24 shows the germinal vesicle of an ovum ‘27
mm. in diameter from a chromosmic preparation, as seen with
Zeiss immersion 2:0 mm., compensation ocular 8. The re-
ticular appearance of the achromatic substance is not very
evident, the ovum represented being one of those in the internal
part of a piece of material, in which the achromatic substance
is very pale. In the ovum from which this figure is taken yolk
formation had not yet commenced, and only three small
nucleoli are seen in the region occupied by the chromosomes.

When the period of yolk formation sets in, the following
changes occur in the germinal vesicle. The membrane becomes
thinner and less distinct, and loses its regular contour. In

OVARY AND OVARIAN OVA IN OERTAIN MARINE FISHES. 145

sections it is seen to be more or less wrinkled, folds of it
projecting inwards between the nucleoli. This is not a con-
dition produced entirely by the action of the fixing reagents.
I have often observed in perfectly fresh material that the surface
of the vesicle instead of being smooth consisted of numerous
bulbous projections, as seen in fig. 25. In consequence of this
condition it frequently happens that a nucleolus appears in a
section to be outside the vesicle, when examination of consecu-
tive sections shows that it is really contained in a pocket or
diverticulum of the vesicular membrane. This is, I believe,
one of the circumstances that have led observers to describe the
migration of nucleoli from the vesicle into the external cyto-
plasm. It is, however, certain that during the period now
under consideration the nucleoli migrate from the periphery to
the central region of the vesicle, where they are for the most
part found around the tangle of fibrils, though some are scat-
tered among the fibrils. Fig. 26 shows a section of a vesicle
from the same preparation as fig. 25 in an ovum ‘39 mm. in
diameter, in which the yolk is at the stage seen in fig. 15, Pl. 2.
The ovum figured being near the edge of the preparation,
where the osmic and chromic acids have acted most strongly,
the fibrils are not well stained, and are therefore less distinct ;
the reticular appearance of the achromatic substance is, on the
other hand, conspicuous. Fig. 27, Pl. 4, shows the appearance
of the vesicle of an ovum ‘36 mm. in diameter, in a section
fixed from material with picro-sulphuric acid, the material being
a portion of the same ovary of plaice from which the chromosmic
sections were derived. The vesicle itself is -16 mm. in its
longest diameter. The section was stained with Delafield’s
hematoxylin. The fibrils are distinct, but appear at this
stage to form a continuous convoluted thread, a true spirem,
not a number of separate chromosomes.

I have not been able to follow out the history of the ger-
minal vesicle in the later stages in the plaice and flounder, and
must therefore proceed to the description of certain stages
which I have examined in other species. In figs. 10, 11, 12, Pl. 2,
will be seen the appearance of the vesicle in various stages of

vou. 40, part 1,.—NEW SER. K

146 J. T. CUNNINGHAM.

the egg of the turbot, all from one series of sections, the
material for which was fixed some hours after the death of
the fish in picro-sulphuric acid. The chromatic fibrils in
these sections are very indistinct, in many of the oldest ova
they are not to be distinguished, the structure appearing
uniformly granular. In the younger ova, and in some of the
older, however, distinct traces of the fibrils are to be seen.
The sections from which figs. 28 and 29, Pl. 4, are taken were
prepared from material taken from a fish just captured and
fixed immediately ; but the mixture used, picro-sulphuric acid
and spirit, has not given very good results. The fibrils are
very indistinct.

It will be seen that in the turbot ovum in the stage shown
in fig. 12, when the whole of the cytoplasm is filled with yolk
globules, the nucleoli are very large and deeply stained ; in
many cases each consists of a number of large coarse gran-
ules. In the ovary to which figs. 28 and 29 refer, the ova are
nearly all in two stages, the ripe stage in which the yolk glo-
bules have fused together, and the nearly ripe stage in which
the yolk is a little more dense than in fig. 12. In the former
stage I have not been able to detect the germinal vesicle, or
any remnant of it at all. In the latter stage the vitelline
membrane has burst at one point, in consequence of the action
of the fixing reagent, which has caused the yolk to swell, but
the germinal vesicle is not much distorted. In fig. 28 it will
be seen that the nucleoli are distributed irregularly about the
nuclear cavity, with the exception of the centre, where there
are one or two indications of fibrils. The nucleoli are more
numerous than at the preceding stage, having presumably
subdivided. In some of the eggs, although these are not
quite the largest, the germinal vesicle presents the remarkable
structure shown in fig. 29. There is a ring of stained bodies
having the form of rods of various shapes, and within the
ring are faint indications of fibrils. The stained rods must
obviously be the transformed nucleoli, and at the same time
they possess considerable resemblance to chromosomes. Ac-
cording to Riickert the nucleoli vanish altogether, and are not

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 147

converted into any of the structures of the directive spindle,
but he suggests it as probable that they form reserves of
chromatic substance, which they afterwards give up to the
chromosomes. His own description is not very consistent
with this suggestion, since he states that the nucleoli increase
and decrease synchronously with the increase and decrease in
size of the chromosomes. But although, according to various
descriptions, the chromosomes of the directive spindle are
very much shorter than the chromosomes of the earlier stages
of the germinal vesicles, they are much thicker, and much
more deeply stained. They contain, therefore, in all proba-
bility, more chromatin, and possibly they take up chromatin
from the nucleoli when the latter disappear. Possibly the sub-
stance of the nucleoli is absorbed into the fibrils. The condi-
tion shown in fig. 29 may be a stage in such a process. The
stained bodies in this condition have a resemblance to chro-
mosomes, but as observers are agreed that the nucleoli are
not converted into chromosomes, but that these are identical
with the nuclear fibrils, and since the fibrils are present
within the ring of stained bodies, it may be inferred that
the substance of the latter is about to be transferred to
the fibrils. This suggestion can only be tested by further
investigations.

In figs. 7,8, 9 are shown some stages of the germinal vesicle
in the ova of the grey gurnard (Trigla gurnardus). Fig. 7
is a section of an ovum +28 mm. in diameter. The prepara-
tion from which it is taken was made from material obtained
in the fish market on April 26th, and fixed with picro-sulphuric
acid. The largest yolked ova are for the most part burst, in
consequence of the action of the fixing reagent on the yolk
and the vitelline membrane, but the younger ova are well
preserved. In those of the stage figured there are numerous
small nucleoli on the inner side of the membrane, and very
distinct feathery separate fibrils in the central region of the
vesicle. These are of various shapes, sometimes V-shaped,
sometimes even circular. The ground substance appears very
finely granular. The feathery fibrils are even more distinct

148 J. T. CUNNINGHAM.

in ova smaller than that figured, and are visible as separate
elements in the larger up to those of a diameter of *43 mm.

The sections from which fig. 8 is taken were prepared
from material fixed immediately after the death of the
fish in a mixture of picro-sulphuric acid and spirit. The
feathery fibrils in these can only be faintly traced with diffi-
culty, as they are scarcely stained at all. They are more distinct
in the younger eggs. Fig. 30, Pl. 4, represents the vesicle of
the most advanced egg in these sections. Although there are
empty follicles showing that ripe eggs have escaped, there are
no stages more advanced than that figured. Fig. 9 shows the
structure of the most advanced egg in sections from part of
the ovary of a gurnard fixed with chromic acid + per cent.
In the younger eggs the feathery fibrils can be seen, but in
this stage in these sections they cannot be distinguished.
The nuclear substance appears pale and granular. The ovum
represented is ‘53 mm. in diameter.

The history of the nucleoli in the ova of other animals has
been much debated, and although the transformations of the
germinal vesicle have been recently traced completely and in
detail in certain cases, there is still room for doubt concerning
the significance of the changes which are observed. In 1887
O. Schultze stated that in the nearly ripe egg of Rana fusca
there was no chromatin network ; that the nucleoli formed a
rounded group in the germinal vesicle, in the centre of which
were very minute corpuscles. He believed that these corpuscles
were derived from the disintegration of nucleoli, and that they
united together to form the chromosomes, which afterwards
united into a convolution. He also saw some nucleoli dissolve
into liquid. me

Professor G. Born, in 1892, made an exact investigation of =
the development of the first directive spindle in the ovum of
another Amphibian, Triton teniatus. He found that the
chromatic convolution was not derived from the nucleoli, but
as Riickert described in Selachians, from chromatic fibrils
previously present. These fibrils are very imperfectly seen
after staining in mass with borax carmine, but are to be

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 149

demonstrated by fixing with chrom-acetic acid and staining
with hematoxylin and orange G. After such treatment
in the germinal vesicle, when the nucleoli are still at the
periphery, there is seen a more deeply stained and more coarsely
granular body in the central region. ‘This is the appearance
with a low magnification ; with higher powers the body is seen
to consist of broad winding chromatin cords, composed of
threads transverse to their dircction, i. e. having the structure
which I have called feathery. These feathery threads contract
into thinner, denser threads which break up into separate
chromosomes: during this process the nucleoli move in a
centripetal direction, surround the aggregate of chromatic
fibrils, and gradually disappear, some breaking up into pieces,
others becoming pale and dissolving.

In 1893 Rudolf Fick (10) described the corresponding trans-
formations in the ovum of the axolotl. He observed that the
nucleoli vanished completely, but leaves open the question
whether there is any connection between their disappearance
and the development of the chromosomes.

In astill more recent paper (1895) E. Korschelt (17) describes
the maturation of the ovum in the chetopod Ophryotrocha
puerilis, and states that the single nucleolus present vanishes
by dissolution while the convoluted chromatic thread is
developing, but leaves undecided the question whether the
substance of the nucleolus is transferred to the thread.

In one of the Foraminifera the chromatin of the nucleus
is in the form of elements which have a curious resemblance
to the nucleoli seen in the germinal vesicle of the Teleostean
ovum; but in the present state of our knowledge it is impossible
to say whether any great importance is to be attached to this
resemblance. Fritz Schaudinn (18) has described the history
of the nucleus in Calcituba polymorpha, and the following
isa summary of it. The protoplasm of the animal is multi-
nucleate. The largest nuclei are 10 to 35 m in diameter, and
consist of a vesicle bounded by amembrane. The central part
is filled only with structureless nuclear sap, while the chro-
matin lies on the inner surface of the membrane in the form

150 J. T. CUNNINGHAM.

of homogeneous compact spheres of different sizes (1 to 5 1)
and numbers (20 to 100). The membrane of the nucleus in
this stage disappears, and the chromatin masses wander apart,
forming the smallest nuclei lying free in the plasma. These
dense homogeneous nuclei send out processes and embrace
vacuoles of the plasma, until they contain several of these, and
then they again assume a smooth outline. The nuclei in this
stage have thus a membrane of chromatin and a framework of
partitions between the vacuoles which appears optically as a
reticulum. This process recalls the vacuolation of nucleoli
seen in the germinal vesicle of Teleostean and Amphibian eggs,
and also the vacuolation of the chromosomes described in
the reconstitution of nuclei after mitosis in segmenting ova.
The chromatin is seen in granules in the network, and it accu-
mulates into a single mass in the centre, or fixed to one side
of the membrane. From the mass extend linin threads; then
the mass of chromatin breaks up, granules passing in suc-
cession along the threads to the points where these join the
membrane, and there forming the spherical masses; the
achromatic threads then break or dissolve, and so we have the
recurrence of the condition from which-we started, and the
process. begins over again. All this suggests strongly that
the multiplication of nucleoli in the Teleostean ovum, and
their disappearance at a later stage, are effected by the stream-
ing of chromatin along threads of linin.

The Murenide.

In consequence of the peculiarities in their reproduction
the generative organs of the conger and eel will be here
considered separately. For a fuller discussion of their
natural history I must refer to my papers in the ‘ Journal of
the Marine Biological Association,’ and the literature to which
references are there given. It is sufficient to mention here
that they do not spawn annually, nor periodically, but only
once, after which they die, and that their adolescent or imma-
ture stage extends over several years.

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 151

In the eggs of the conger and eel, as in those of the gurnard,
turbot, brill, &c., above described, the formation of oil globules
in the cytoplasm precedes that of the yolk proper. But the
deposition of fat occurs at a certain stage not merely in the
eggs but in the stroma or interovular tissue of the ovary, and
suggests that there is some important relation in the nutritive
processes between fat and proteinaceous vitelline substances,
Whether the fat is actually combined with other substances to
form nitrogenous compounds, or is merely used up separately
to supply the energy required by the body while the vitelline
constituents are being produced from other sources, is a phy-
siological question I am not prepared to discuss. Whatever
the function of the fat, there are three distinct periods in the
history of the ovary: (1) a stage in which the organ is very
small, and contains only protoplasmic ova destitute of deuto-
plasmic elements, and in which the stroma of the ovary is
small in quantity and destitute of fat-cells; (2) a stage in
which fat is largely deposited in the stroma in the form of fat-
cells, and the eggs are isolated from one another by this adi-
pose tissue, and in which the eggs themselves contain oil
globules ; (3) a stage in which the fat in the stroma is gradu-
ally absorbed, and yolk globules are developed in the eggs: at
the end of this stage a condition is reached in which the ovary
resembles in the structure of the stroma the earliest stage (1),
and consists almost entirely of eggs, now much larger and full
of yolk, crowded together.

The first of the stages above defined is seen in the ovaries
of female conger which are 2 feet in length. The ovary in
such specimens is 3 to 7 mm. in width, and the breadth of the
lamelle, measured from the attached base to the free edge,
does notexceed‘6mm. These lamellz contain small protoplas-
mic eggs entirely without secondary deposits (fig. 31, Pl. 4). The
greater number of these eggs are of nearly uniform size, and
these are the largest eggs present. Their diameter in my
preparations from one specimen of 2 feet in length is (05 mm.
There are asmaller number of smaller eggs, and at the surface
of the lamelle is seen the germinal epithelium, which in a

152 J. T. CUNNINGHAM.

great many places is proliferating, and so giving rise to large
numbers of new eggs. These young eggs, or germ-cells, are
collected in small groups, which project inwards from the
germinal epithelium towards the interior of the lamella, and
these groups are doubtless formed by the subdivision of one or
a few cells.

Of the intermediate or second of the stages defined above I
have only one preparation. It was made from material pre-
served in a mixture of chromic acid and platinum chloride.
The appearance of a portion of one of these sections is shown
in fig. 32, which represents part of the transverse section of
alamella. It will be seen that the lamella consists of adipose
tissue in which are embedded two rows of cells, one correspond-
ing to each surface of the lamella. The largest eggs are
‘25 mm. in diameter. Near the periphery of these eggs there
is a zone of vacuoles, which in the fresh state were oil globules.
No yolk elements are distinguishable.

In his paper on the sexual organs of Murenide! J. Brock
has given an interpretation of the structure of the ovary which
differs from mine, and which I believe to be erroneous. With-
out mentioning the size of the fish whose ovaries he examined,
he describes a stage of the conger ovary in which the lamelle
are no higher than they are broad. In each lamella, at this
stage, there are eggs towards each surface in two or three
layers, while in the axis of the lamella, cavities are beginning
to appear in the supporting tissue. He then describes ovaries
of Conger and Myrus, which were apparently advanced in
development. The lamelle in these were2—3 mm. high. He
found that when stained in the mass these ovarian lamelle
retained their opaque white colour, as though they consisted
merely of fat, showing only small stained dots scattered at
intervals. He states then that microscopic examination ex-
plained the mystery; that abortion of sexual cells must have
taken place on a large scale, for the whole ovary consisted of
a narrow-meshed network of empty follicular membranes, in

1 “Untersuchungen tiber die Geschlechtsorgane einiger Murenoiden,”
‘Mitt. Zool. Stat. Neapel,’ Bd. ii, 1881.

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 153

which here and there perfect eggs were distributed either
singly or insmall groups. Brock remarks that these specimens
show that the abortion of sexual cells takes place in the female
sex as well as in the male, but that at the same time the fact
that the condition in question was only observed in isolated
cases leads to the conclusion that it is not a constant and
regular stage in the history of the ovary.

My own observations, on the other hand, show that the
condition of the ovary here in question is a perfectly constant
stage, and that it is due to the remarkable and abundant
deposit of fat in the connective tissue of the ovary. The.
meshes of the network described by Brock are, in reality, fat-
cells. In sections prepared by the paraffin method the fat has
been dissolved out, but in the fresh state the oily nature of
the contents of these cells is perfectly obvious. There is no
indication at all of the abortion of ova. The process of such
abortion and reabsorption, as it occurs in the ovaries of other
species, has been shown in this paper to be manifested in par-
ticular appearances produced by the presence of the remnants
of the disintegrated ova. No trace of such remnants has
been seen in the ovary of the conger, and their absence is in
accordance with the special history of the ovary in the eel
family—namely, a slow but continuous development culminat-
ing in a single and final act of spawning.

In the conger the fatty condition of the ovary is found from
the time when the fish is a little over 2 feet long to that when
it is full-grown, 4 to 6 feet or more long, and the ovary begins
to assume the ripe condition. In nearly all specimens that are
caught in the sea, except the small, in which the ovaries are in
the stage previously described, the ovaries are in this condi-
tion.

The third stage in the development of the ovary of the
conger is that exhibited by the ovaries of specimens which
have died in aquaria with these organs much enlarged, and
evidently almost ripe. In these ovaries there are no fat-cells,
they have all disappeared, and each lamella consists chiefly of
two layers of large eggs not differing from one another very

154 J. T. CUNNINGHAM.

much in size or stage of development. These eggs are close
to one another, only separated by the thin membranes of the
follicles containing them, the interstices being occupied by
strands of connective tissue and blood-vessels. I have pre-
pared sections from three specimens which died in the aqua-
rium of the Laboratory of the Marine Biological Association
at Plymouth, and the history of which is recorded in my
paper on the conger in the ‘Journal’ of the Association,
vol. ii (new series).

The structure of the largest and most advanced eggs in
these sections is shown in fig. 33, which is from a preparation
made from material preserved in corrosive sublimate and
acetic acid. It will be seen that the vitellus has shrunken
away from the vitelline membrane, which is very thin. The
follicular epithelium also lying upon the vitellime membrane
is very much thinner than in the eggs of other fishes at a
similar stage. The yolk consists of small vitelline globules
which extend throughout the extra-nuclear region, and are
closely crowded together. There are also a number of scat-
tered oil vacuoles, situated principally about midway between
the germinal vesicle and the surface of the egg. The germinal
vesicle is somewhat contracted, so that there is an artificial
separation between it and the surrounding vitellus. There
is no nuclear membrane, the nucleus consisting of a finely
granular or really reticulate achromatic substance containing
a small number of deeply-stained nucleoli.

It is an interesting question whether any inferences can be
drawn from the structure of this nearly ripe egg, concerning
the condition of the ripe egg after deposition. The fact that
among pelagic eggs none have been obtained in British waters
which could possibly belong to conger or eel, is an obstacle to
the supposition that the egg of the conger may be pelagic. Yet
the structure of the egg, as above described, certainly agrees
rather with that of nearly ripe pelagic eggs than with that of
those which are heavy and adhesive. Eggs of the latter kind
generally have a thick vitelline membrane, showing a division
into two layers, while that of the conger is remarkably thin.

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 155

The condition of the yolk in the conger’s egg resembles that
of nearly ripe pelagic eggs, and there is nothing in it to indicate
that the yolk spherules might not fuse together more or less
completely, and so give rise to the condition of a pelagic egg
with one or more oil globules. But there is a third possibility,
that the ripe egg is neither pelagic nor adhesive, but remains
unattached on or near the bottom.

My preparations of the ovaries of conger do not supply any
evidence of great importance concerning the more minute
features in the structure of the ovum which I have discussed
in connection with other species. They were not made for the
purpose of minute investigation, and the material was fixed in
either corrosive sublimate or Perenyi’s mixture. I have not
been able to trace in them the vitelline nucleus, and although the
nuclear network is visible in the younger stages, in eggs up to
a diameter of -27 mm., I have only seen a faint trace of fibrils
in the nearly ripe eggs. The presence of one, or occasionally
two nucleoli of very great size is noticeable in the younger
eggs up to the size just mentioned. In an ovum of that
diameter I found that the large nucleolus measured ‘03 mm.
There are in addition very numerous small peripheral nu-
cleoli.

The history of the ovary in the eel is very similar to that in
the conger. I have not examined the earlier condition in very
young eels, but I have preparations showing the fatty stage
from a specimen 1 ft. 114 in. long, killed in June. The ova
in these sections do not exceed ‘13 mm. in diameter, and they
are surrounded by fat-cells, as in the conger ovary represented
in fig. 32. On the other hand, in sections from the ovary of
an eel | ft. 93 in. long, killed on November 15th, the largest
ova reach ‘2 mm., and there is no fat. One of the ova in
these sections is represented in fig. 34. As shown in the
figure, the cytoplasm contains numerous oil globules, but no
yolk.

The testes of the conger, like the ovaries, also at a certain
stage consist in large proportion of adipose tissue, in which
small islands of germ-cells are scattered. As in the case of

156 J. T. CUNNINGHAM.

the ovary, J. Brock has described the fat-cells as empty follicles
of which the germ-cells have aborted. He states that in a
testis of Myrus he saw near the empty follicles altered germ-
cells which were opaque and had lost their nucleus, and con-
siders that this leads to the conclusion that in the empty
follicles the germ-cells have disappeared altogether. He
proceeds to state that in osmic acid preparations the conclusion
becomes a certainty, for the follicles in these are usually not
empty, but contain shrunken cells which have reduced the
osmic acid very strongly as only fat and nerve-cells usually do.
He infers that the germ-cells have undergone a fatty dege-
neration. It is clear enough that all this is better explained
by the simple fact that what Brock calls follicles are merely
true fat-cells, developed in the connective tissue, since evidence
that abortive germ-cells turn into fat is entirely wanting.
With the development of the adipose tissue, the testis, like the
ovary, becomes much larger in comparison with the younger
fatless stage, and therefore there is no reason to suppose that
the germ-cells, although they are scattered, have been reduced
inuumber. Fig. 35 shows the proximal portion of a section
of a testis of the conger in the adipose condition, as seen under
alow power. At the base of the section is seen the lumen of
thevas deferens. The total breadth of the testis, whose structure
is shown in this figure, from the attached base to the distal
edge, was about 5 mm.

As the testis becomes more developed the narrow cords of
germ-cells or spermatogonia, seen in section in fig. 85, become
much enlarged, and form spermatic tubes filled with spermato-
cytes, of which those in the centre are smallest, and therefore
most advanced in development. ‘The tubes are still separated
by a certain amount of adipose tissue, especially near the
distal edge of the organ, while near its attached base they are
more closely approximated. ‘This is the condition observed in
sections of an organ about 7 mm. broad. In a much larger
organ, something over 1 cm. broad and 5 mm. thick from side
to side, there is still some adipose tissue between the large
tubes crowded with spermatocytes. But as I have shown in

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 157

my paper in the ‘ Journal of the Marine Biological Associa-
tion’ that the ripe conger lives for about six months without
taking food, there can be no doubt that all the fat in the testis
is absorbed long before the milt is all shed.

SUMMARY AND CONCLUSIONS.

In fishes which have pelagic ova and an annual spawning
season, the formation of yolk in the developing ova to be shed
at a given spawning season commences some months after the
close of the preceding spawning season. The active deyelop-
ment of the annual crop of ova does not take much more than
six months.

The formation of yolk always commences near the surface of
the cytoplasm and extends inwards.

In those eggs which develop separate oil globules, a few of
these of small size are present long before the formation of
yolk commences. The eggs of the mackerel form an exception
to this statement. In immature specimens of sole, turbot,
brill, &c., examined during the spawning season, the largest
ova in the ovaries are found to contain scattered oil globules,
and these are also present in the largest transparent ova in
spent ovaries of these species. When the formation of yolk
takes place in such eggs the oil globules form a zone internal
to that of the yolk.

The essential peculiarity of the spent ovary is the presence
in it of the ruptured follicles, from which the ripe eggs have
escaped. The follicular epithelium in these appears to dis-
integrate and dissolve. The cavity is obliterated by the con-
traction of the follicle, which forms a mass of cells and fibres,
and is finally absorbed soon after the commencement of the
formation of yolk in the eggs for the following season.

In the spent ovary there are a number of eggs which have
not reached the ripe condition, which die, and are not dis-
charged from their follicles, but absorbed in situ. In the
fresh state they are visible as opaque amorphous masses.

Similar opaque, masses are also seen in immature ovaries in

158 J. T. CUNNINGHAM.

which spawning has never occurred. Here also they are aborted
dead ova, which are undergoing disintegration and absorption.
They are scattered singly in the ovarian tissue, and their de-
velopment is arrested at an early stage, before the formation
of yolk has made any progress, if even it has commenced.
In those in which death has only recently occurred the germinal
vesicle is seen to be shrunken, and to contain a single large
spherical nucleolus.

The vitelline nucleus is first seen as a stained corpuscle in
contact with the germinal vesicle. I have not been able to
follow the mitosis of the germ-cells, or to trace back the vitel-
line nucleus, but consider it most probable that the latter is
identical with the centrosome which remains in the ovum after
the last division of the germ-cell.

In ova of plaice and flounder the vitelline nucleus separates
from the germinal vesicle, moves towards the surface of the
ovum, and is afterwards found at the inner border of the yolk-
layer. It becomes surrounded with yolk, and ceases to be
visible.

In Syngnathus acus there are often two or even more
vitelline nuclei in a single ovum. These are probably pro-
duced by the division of a single body.

If the vitelline nucleus is the centrosome, its disappearance
forms an interruption to the persistence of the centrosome as
an extra-nuclear body, since the directive spindle is provided
with new centrosomes.

The germinal vesicle in the Teleostean ova examined con-
sists at first of a single large nucleolus, a nuclear network,
and a surrounding membrane. I could not resolve the net-
work into a continuous filament, or into separate chromo-
somes.

At the next stage the vesicle is larger, and there are several
nucleoli in contact with the nuclear membrane.

In still larger ova the nucleoli are still at the periphery, but
there is a central region in the vesicle distinguished by the
presence of separate feathery fibrils, the centrosomes of
Rickert.

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 159

After the formation of yolk has commenced the membrane
of the vesicle is wrinkled. The nucleoli migrate from the
periphery of the vesicle, and are found around and among the
central fibrils.

There are indications in the ova of the turbot that the
substance of the nucleoli is absorbed into the central fibrils
to form the chromosomes of the polar mitoses, but the actual
formation of these chromosomes was not followed.

List oF Papers AND Books ciTED, IN CHRONOLOGICAL ORDER.

(1 Emery, Carto.—‘ Fauna und Flora des Golfes von Neapel:’ Ite
Monographie, “ Fierasfer,” Leipzig, 1880.

(2) Scnutze, O.—* Ueber die Reifung und Befruchtung des Amphibieneies,”
‘ Zeitschr. f. wiss. Zool.,’? 1887.

(8) Bovert.—“ Zellenstudien,” ‘Jenaische Zeitschr.,’ 1888.

(4) Bovert.—“ Zellenstudien: ueber das Verhalten der Chromatischen
Kernsubstanz, &c.,”’ ibid., 1889.

(5) Hertwic, O.—*< Vergleich der Hi- und Samenbildung bei Nematoden,”
‘Arch. f. mik. Anat.,’ Bd. xxxvi, 1890.
(6) Hennecuy, L. F.— Nouvelles Recherches sur la Division cellulaire
indirecte,” ‘Journ. de ]’Anat. et de la Physiol.,’ t. xxvii, 1891.
(7) Born, G.—* Die Reifung des Amphibieneies und die Befruchtung
unreifer Kier bei Triton teniatus,” ‘Anat. Anz.,’ Bd. vii, 1892.
(8) Rickert, J—‘ Zur Entwickelungsgeschichte des Ovarialeies bei Sela-
chiern,” ‘ Anat. Anz.,’ Bd. vii, 1892.
(9) Rickert, J.—‘ Ueber die Verdoppelung der Chromosomen im Keim-
blaschen des Selachiereies,” ‘ Anat. Anz.,’ Bd. viii, 1892-3.
(10) Fick, R.—* Ueber die Reifung und Befruchtung des Axolotleies,”
‘ Zeits. f. wiss. Zool.,’ Bd. lvi, 1893.
(11) Hennecuy, L. F.—‘‘Le Corps Vitellin de Balbiani dans l’ceuf des
Vertébrés,” ‘Journ. de l’Anat. et de la Physiol.,’ tome xxix, 1893.
(12) Barpranr, E. G.— Centrosome et Dotterkern,” ‘ Journ. de l’Anat. et
de la Physiol.,’ tome xxix, 1893.
(13) Ratu, O. vom.—‘ Beitrage zur Kenntniss der Spermatogenese von
Salamandra maculosa,” Parts I and II, ‘Zeits. f. wiss. Zool.,’
Bd. lvii, 1894.
— (14) Cunninenam, J. T.—‘‘The Ovaries of Fishes,” ‘Journ. Mar. Biol.
Ass.,’ vol. iii, No. 2, 1894,

160 J. T. CUNNINGHAM.

(15) Hupparp, Jesss W.—“ The Yoke-nucleus in Cymatogaster aggre-
gatus, Gibbons,” ‘ Proc. Amer. Philos. Soe.,’ vol. xxxiii, 1894.

(16) Mzap, A. D.—* Maturation and Fecundation in Chetopterus perga-
mentaceus, Cuvier,” ‘Journ. of Morphology,’ vol. x, 1895.

(17) Korscuett, E.— Hireifung und Befruchtung bei Ophryotrocha
puerilis,” ‘Zeits. f. wiss. Zool.,’ Bd. Ix, 1895.

(18) Scuaupiny, Frivz.—‘ Untersuchungen an Foraminiferen: I. Cal-
cituba polymorpha, Robaz,” ‘Zeits. f. wiss. Zool.,’ Bd. lix, 1895.

(19) Wizson and Marrnews.— Maturation, Fertilisation, and Polarity in
the Echinoderm Egg,” ‘ Journ. of Morphology,’ x, 1895.

(20) Cunninenam, J. T.—“‘The Development of the Egg in Fiat fishes
and Pipe fishes,” ‘Journ. Mar. Biol. Ass.,’ vol. ili, No. 4, 1895.

(21) Sonorta, J.—‘ Anat. Anz.,’ 1895.
(22) Henneevy, L. F.—‘ Lecons sur la Cellule,’ Paris, 1896.

DESCRIPTION OF PLATES 2—4,

Illustrating Mr. J.T. Cunningham’s paper, “ On the Histology
of the Ovary and of the Ovarian Ova in certain Marine
Fishes.”

PLATE 2.

Fics. 1—5.—Ovarian eggs of Trigla gurnardus in various stages of
erowth and yolk formation, as seen in fresh condition with Zeiss’s obj. A,
oc. 3. From a specimen examined April 26th, 1895, at Grimsby.

1. Transparent ovum with scattered small globules, probably oily.

2, Large ovum with internal darker zone of oil-globules, external more
transparent zone, containing yolk.

3. Later stage, in which the contrast between the yolk layer and the
oil-layer is at its greatest.

4. Later stage, in which the yolk layer has become so opaque as to
couceal the oil layer.

5. Stage in which the ovum is approaching maturity, and begins to
grow transparent again.

Fic. 6.—Portion of ovary of immature brill, 123 inches long; examined in
fresh condition May 30th, 1895, at Grimsby. Zeiss A, oc. 3.

Fic. 7.—Ovum of Trigla gurnardus, ‘28 mm. in diameter. From roe
containing some ripe eggs, obtained in the market, fixed with picro-sulphuric

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 161

acid. Sections stained with hematoxylin. Zeiss C C, oc. 3, camera.
y.g- Yolk globules. o.g. Oil globules.

Fic. 8.—Ovum of Trigla gurnardus, ‘37 mm, in diameter. From ripe
ovary fixed with picro-sulphuric acid and spirit. Sections stained with
Delafield’s hematoxylin. Zeiss C C, oc. 3, camera.

Fic. 9.—Ovum of Trigla gurnardus, ‘53 mm. in diameter. From rip
ovary fixed with chromic acid + percent. Sections stained with safranin
Zeiss C C, oc. 2, camera.

Fic. 10.—Ovum of Rhombus maximus, 175 mm. diameter. From
ovary of a fish 1 ft. 92 in. long, obtained in the market April 12th. Fixed
with picro-sulphuric acid. Stained with Delafield’s hematoxylin. Zeiss
CC, oc. 3.

Fre. 11.—Ovum of same species, *25 mm. diameter. From same series of
sections as Fig. 10. Zeiss C C, oc. 3.

Fie. 12.—Ovum of same species, ‘35 mm. diameter. From same series of
sections as Fig. 10. Zeiss C C, oc. 3.

Fie. 13.—Ovum of Solea vulgaris, ‘67 mm. diameter. From ovary of
a fish 163 in. long, obtained in the market September 20th, 1895. Fixed
with picro-sulphuric acid and spirit. Stained with hematoxylin.

Fie..14.—Ovum of Pleuronecles platessa, ‘33 mm. diameter. From
ovary of a large specimen taken from the Plymouth aquarium on August 17th.
Fixed with Lobianco’s chromosmic mixture. Stained with Hhrlich’s hema-
toxylin. Zeiss C C, oc. 3. v. 2. Vitelline nucleus.

Fic. 15.—Ovum of same species, *34 mm. diameter, in which the develop-
ment of yolk is more advanced. From the same ovary as that shown in the
preceding figure, but fixed with Kleinenberg’s picro-sulphuric acid and stained
with Delafield’s hematoxylin. Zeiss C C, oc. 3.

PLATE 3.

Fic. 16.—Portion of ovary of a spent sole, 17 in. long, from Grimsby
market August Ist, 1895. Examined in fresh condition. Zeiss A, oc. 3.
d, o. Dead eggs, in which yolk development has proceeded to a certain stage.

Fic. 17.—Portion of a section from an ovary of a spent plaice, 153 in.
long, obtained in the market at Plymouth February 2nd. Fixed with Flem-
ming’s mixture. Stained with safranin. Zeiss A, oc. 3. fol. Empty
follicles, from which the ripe eggs have escaped. /f. e. Follicular epithelium.

Fic. 18.—An empty collapsed follicle from a section of the ovary of a
specimen of Trigla gurnardus, captured July 22nd in North Sea. Fixed
in picro-sulphuric acid and spirit. /. e. Follicular epithelium. 2 w. Wall of
the follicle. 4. v. Blood-vessel.

voL. 40, part 1.—NEW SER. L

162 J. T. CUNNINGHAM.

Fic. 19.—A dead aborted ovum, from a section of the ovary of a plaice
112 in. long, obtained in Grimsby market August 3rd, 1895. Fixed in
chromic acid } percent. Stained with Mayer’s hemalum. fe. Follicular
epithelium. g.v. Germinal vesicle in altered condition, showing oue large
nucleolus.

Fic. 20.—Ova from the ovary of a plaice 123 in. long, examined in the
fresh condition September 21st, 1895, at Lowestoft. The largest ovum has
just died at a stage when the formation of yolk has scarcely commenced.
g. v. Germinal vesicle.

Fic. 21.—Two ova from section of the ovary of a flounder 11 in. long,
taken from Plymouth aquarium July 11th. Fixed in Flemming’s strong
mixture. Stained with safranin. Zeiss F, oc. 3. v. ”. Vitelline nucleus.

Fic. 22.—Portion of a section of an ovary of Syngnathus acus, fixed
with chromic acid ;4, per cent. and picric acid 40 per cent. Stained with
safranin, gentian violet, and orange G. Zeiss A, oc. 3. g.e. Germinal
epithelium at the edge of the ovarian lamella. v. ~. Vitelline nuclei.

Fic. 23.—Ovum °049 mm. diameter, from the section shown in Fig. 22,
showing the vitelline nucleus in contact with the germinal vesicle. Zeiss E,
oc. 38. »v. 2. Vitelline nucleus.

Fic. 24.—The germinal vesicle of an ovum ‘27 mm. diameter, in section
of the ovary of a plaice from which Fig. 14 is taken. In this ovum yolk-
formation had not commenced. Zeiss, apochr. immersion, 2 mm., compens.
oc. 8. ch. Feathery fibrils, the chromosomes of Riickert. x. Nucleoli.

Fic. 25.—Ova from the ovary of a plaice 133 in. long, examined in the
fresh condition September 16th. The formation of yolk has just commenced
in the largest ovum, which also shows the wrinkled or mulberry-like form of
the membrane of the germinal vesicle. Zeiss C C, oc. 3.

Fic. 26.—Germinal vesicle of an ovum ‘39 mm. in diameter, from the same
series of sections from which Fig. 14 is taken. Zeiss EF, oc. 3. ch. Central
fibrils. 2. Nucleoli.

PLATE 4.

Fie. 27.—Germinal vesicle of an ovum ‘36 mm. in diameter, from the same
series of sections from which Fig. 15 is taken. ch. Central fibrils. 2.
Nucleoli.

Fie. 28.—Germinal vesicle of an ovum ‘46 mm. in diameter, in a section

from the ovary of a turbot 2 ft. 13 in. long. Fixed with picro-sulphuric
acid and spirit. Stained with hemalum. Zeiss C C, oc. 3.

Fic. 29.—Germinal vesicle of an ovum ‘39 mm. in diameter, from the same

series of sections as Fig.28. Zeiss E, oc. 3. 2. Nucleoli in the form of
rods of various shapes.

OVARY AND OVARIAN OVA IN CERTAIN MARINE FISHES. 163

Fic. 30.—Germinal vesicle of an ovum '55 mm. in diameter, in a section of
the ripe ovary of a specimen of Trigla gurnardus. Fixed with picro-
sulphuric acid and spirit. Stained with hematoxylin. Zeiss C C, oc. 3.

Fic. 31.—Portion of a section of the ovary of a conger 2 ft. long, killed
October 24th, 1888. The figure shows two ovarian lamelle. Zeiss A, oc. 3.

Fie. 32.—Portion of a section from the ovary of a conger of moderate size.
Zeiss A, oc. 3. f. Fat-cells. 0. ova.

Fie. 33.—Ovum from section of portion of the ovary of a conger, which
died with nearly ripe ovaries in the Plymouth aquarium March 21st, 1890.
Diameter, 65 mm. by 56 mm. Fixed with corrosive sublimate and acetic.
Stained with borax-carmine. Zeiss C C, oc. 2.

Fie. 34.—Ovum from section of the ovary of an ecl 214 in. long, killed
November 15th, 1895. Fixed with chromic acid } per cent. Zeiss C C, oc. 3.
Fie. 35.—Section of testis of conger in the intermediate condition.

Zeiss A, oc. 2. f. Fat-cells. sp, Cords consisting of spermatic cells. vv. d.
Vas deferens.

A. ' hPaie 15 > oe i wel eye
ei ie Stier - “v4 + i 4 ‘ > }
wsscert ih Me si

hx eee ot gies : eA 5p Wits ree

hye ey é
hij haar id heel aD) am ee a

- in get? 4 Wd. ang PRD is
. Us > of . ‘ ae | we. eg

Bers ee aot LV / cao oo tek eae
pa ; 4. ( ‘ A a" ;

ee Sa ! Gas ahe Sa Spies

) : we on ie, Ve: 1 ig CGE S
‘ ' A a i (pee
«Ss * i pal ) na iA ‘ca, » ri

ane aad
we. \- ea bt i £2" eae
s) b 2 S 7 Le

iw oe bal tee

ON PTYCHODERA FLAVA. 165

On Ptychodera flava, Eschscholtz.
By

Arthur Willey, D.Sc.

With Plate 5.

PrycHODERA FLAVA has the distinction of being the oldest
as well as one of the least known of the Enteropneusta, having
been recorded and figured for the first and only time by
Eschscholtz in 1825, from material obtained from the Marshall
or Rumanzow Islands, in the Pacific, north of the equator.

Eschscholtz, as quoted by Spengel,! regarded his Ptychodera
as a worm-like animal belonging to the group of the Holo-
thuride. Otherwise, however, his description was very de-
fective, and far from being a specific diagnosis. Still, in so
far as Eschschoitz stated that the body of the animal was “der
Lange nach gespalten,”’ and also from the figure which accom-
panied his description, and is reproduced in Spengel’s mono-
graph, indicating the presence of genital pleurz (see below),
Spengel was enabled to retain the species in his amended
genus Ptychodera.

Since 1825 this species has not been re-discovered in the
true sense of the term, although Spengel makes the suggestion,
which the present contribution provisionally” supports, that
the fragments of a Balanoglossus obtained by Dr. Francois in

1 J. W. Spengel, ‘Die Enteropneusten des Golfes von Neapel, &c.,”
‘ Fauna und Flora des Golfes von Neapel,’ Berlin, 1893.

2 I say provisionally because absolute certainty can only be arrived at
when the form from the first recorded habitat (Marshall Islands) comes to be
re-investigated.

166 ARTHUR WILLEY.

the vicinity of Nouméa, New Caledonia, belonged to P. flava.
Francois! simply mentions that his native servant one day
brought him some fragments of a Balanoglossus, and he makes
no further reference to it.

I have found a Ptychodera which may probably be identified
with P. flava, especially if it may be assumed that Eschscholtz’s
figure represents approximately the natural size of the object,?
in great abundance near the low-tide mark on the small islet
of Amédée, upon which stands the lighthouse, some twelve
miles out from Nouméa and eight or ten miles inside of the
great Barrier Reef of New Caledonia.

It occurred near the surface of the sand, chiefly underneath
loose stones, often adhering to the latter, and creeping into
the holes with which the coralline blocks are riddled. On a
later occasion I found it to be, if possible, still more abundant
on the rocky platform of coral limestone which surrounds a
great part of the Isle of Pines. This platform is, in places,
much excavated, and, while it is exposed at low water, there
are numerous rock-pools scattered over it, in some of which
many different kinds of seaweeds luxuriate. In the shallower
pools Ptychodera flava occurs in great numbers in the sand
at the base of or in the neighbourhood of the tussocks of sea-
weed, being often involved in the roots of the latter. Several
species of Nemertines occur in the same locality, but are
much rarer than the Ptychodera.

External Features (cf. Fig. 1).

When first taken from their native habitat the individuals
of P. flava average in length approximately from 14 to 2 or
even 3 inches, and the intestine is often full of sand. But

1 Ph. Frangois, “ Choses de Nouméa,” ‘ Archives de Zool. expér.’ (2), t. ix,
1891, p. 232.

2 Spengel (loc. cit., p. 190) suspects that the figure given by Eschscholtz
represents the animal on a reduced scale. Judging by the material obtained
by me in New Caledonia, this need not have been the case.

3 The exact spot on the Isle of Pines where I found the Ptychodera was
situated at the point on the opposite side of the harbour to that on which
the military buildings are placed.

ON PTYCHODERA FLAVA. 167 -

after being kept for a short time in captivity they discharge
the sand, and the larger specimens may then stretch them-
selves out to a length somewhat exceeding 5 inches. The
collar region may be upwards of a quarter of an inch in
length, while the proboscis is normally somewhat shorter. The
relatively great length of the collar region is characteristic of
the genus Ptychodera. The gill-region measures from half
to three quarters of an inch in length, being about as long as
the proboscis and collar regions put together. The hepatic
region may measure about one and a half inches.

Eschscholtz correctly described the body as_ presenting
numerous transverse folds or annulations. When the intestine
is free from sand and the body is consequently not swollen
out, these annulations are very prominent ridges, and are
characteristic for the species. They are not continuous all
round the body, but are interrupted along the dark yellowish
or reddish-yellow coloured lines which mark the course of the
dorsal and ventral nerve-cords. Here and there, especially in
the posterior dorso-lateral region, the annulations are sub-
divided into islets. The more faintly marked ridges on the
outer surface of the genital pleure are often similarly sub-
divided and also branched.

On either side of the dorsal nerve-cord may be seen a dark
longitudinal band corresponding in position to the ciliated
grooves in the intestinal wall, described by Spengel in other
species, and more recently by Hill! in P. australiensis.

But these externaily visible lines do not cause any interrup-
tion in the annulations or islets of the integument, as they do
in P.australiensis, according to Hill’s description, and as
does the single asymmetrical band, present only on the left
side, in P. minuta, as described by Spengel.

As indicated by the specific name, flava, the colour of this
species is a nearly uniform dull yellow, somewhat deeper in

1 Jas. P. Hill, “On a New Species of Enteropneusta (Ptychodera
australiensis) from the Coast of New South Wales,” ‘Proc. Linn. Soe,
New South Wales,’ vol. x (2), 1894. Id., “ Preliminary Note on a Balano-
glossus from the Coast of New South Wales,” ibid., vol. viii (2), 1893,

168 ARTHUR WILLEY.

the more opaque regions of the proboscis and collar. Some-
times the body has a more brownish tinge. The anterior liver
sacs, however, offer a relief to the general yellow ground
colour, in that they are of a dark greenish-brown colour, while
the sacs about the middle of the hepatic region are of a light
brown, passing posteriorly into the usual yellow colour. The
liver sacculations pass quite gradually behind into the ordinary
annulations of the body-wall, and it is not always easy to say
which is the last hepatic diverticulum.

In cases where the body has evidently been broken in two
in the hepatic region, and the anterior portion of the body,
including the whole of the branchial region has been lost at
no very distant period, a new collar and proboscis have been
added by regeneration immediately in front of the liver-sacs,
while the branchial region would no doubt be regenerated later.
In such regenerated individuals the collar and proboscis are
white and unpigmented.

The proboscis in the normal condition is distinctly grooved in
the dorsal middle line, and in this respect P. flava may be
compared with Balanoglossus sulcatus, Spengel (cf. fig. 1).

The liver-sacs are not always simple smooth outgrowths, but
the larger ones are distinctly lobed, and sometimes present a
digitate appearance. An intensification of this lobed structure
would probably lead to such a diffuse arrangement of the
liver-sacs as is met with in P. erythrea, Spengel.

In only two individuals out of the many that have passed
under my observation have I observed them to be infested with a
curious parasite (? Ive balanoglossi, Paul Mayer), originally
remarked by Spengel in P. minuta, and more recently by
Hill in P. australiensis. As described by Hill in the latter
species, the parasite occurs in one of the genital pleure (in the
example here figured on the right side), where it forms a very
prominent tubular enlargement ” (cf. fig. 2). Hill states that
in P. australiensis “a large proportion of the individuals ”
are infested with the parasite.

ON PTYCHODERA FLAVA. 169

Anal Respiration.

I should like to direct the attention of those zoologists who
may have future opportunities of observing living Balano-
glossus, to the possibility of the occurrence of anal respiration.
When the posterior end of the body is protruding from a mass
of sand and seaweed I have observed the anal orifice in P.
flava to open periodically, widely, and slowly for a second or
two, and then to close up again. This may occur two or three
times to the minute, and it has apparently no relation whatever
to the evacuation of feces.

In the case of the large Balanoglossus occurring at the
Islands of Bimini, in the Bahamas (species not stated), whose
development was studied through the Tornaria stage by Morgan,}
the author states that generally the posterior end of the worm
protrudes from the surface of the sand, sometimes as much as
aninch. “If,” he says, “ the spade is thrust rapidly into the
sand before the worm has been disturbed, it is easy to cut off
from six inches to a foot of the hind end of the body, but im-
possible to get more of the worm.”” When it is so deeply em-
bedded in the sand, it is conceivable that the branchial respira-
tion would not entirely suffice for the needs of the animal, and
that anal respiration may occur as an accessory to the former.

With regard to the tenacity of life exhibited by P. flava,
it cannot compete with Balanoglossus Kowalevskii in this
respect, according to Bateson’s account (quoted by Spengel,
loc. cit., p. 341). Ina dish of P. flava, in which the water
became slightly turbid overnight, the Ptychodera were nearly
all dead, those that were not dead being moribund, and they
were outlived by several Annelids.

Genital Pleure.
The genus Ptychodera is distinguished from the other genera
of Enteropneusta, established by Spengel,—above all by the
possession in the anterior region of the body (the branchio-

' T. H. Morgan, “The Development of Balanoglossus,” ‘Journ. Morph.,’
vol, ix, 1894.

170 ARTHUR WILLEY.

genital region of Spengel) of lateral wing-like expansions,
which can be folded over so as to meet one another in the
dorsal middle line, and so to completely embrace the branchial
region and the most anterior portion of the hepatic region.
These are the genital wings (Genitalfliigel) of Spengel, so
called because they contain the gonads. We may conveniently
refer to them as the genital pleure.

In P. flava the genital pleure have a very low origin,
arising from the ventro-lateral margins of the body, and they
constitute remarkable structures. They are very mobile, and
in life can be spread out laterally nearly flat ; while, as already
stated, they can meet over the pharynx in the mid-dorsal line,
thus producing a most effective peripharyngeal cavity or
atrium, opening to the exterior posteriorly in the neighbour-
hood of the anterior hepatic region. The genital pleure of P.
flava attain their maximum development within the branchial
region, and maintain it for some distance into the post-bran-
chial region, behind which they gradually decrease in size, and
finally die out on the outer sides of the liver-sacs (fig. 1). In
P. australiensis, according to Hill, they reach their maxi-
mum size somewhat posterior to the gill region.

The Pharynx (cf. Fig. 3).

When, in the living animals kept under observation, the
genital pleurze are spread out laterally, a complete and
beautiful view of the entire pharynx is to be obtained. The
latter is then seen to stand up, erect and independent, in the
middle of the peripharyngeal area, and the branchial bars are
visible nearly if not quite throughout their whole length.
Dorsally, on either side of and adjacent to the dorsal nerve-
cord, two whitish pigmented bands extend throughout the
length of the pharynx. These are the bands which in most
Enteropneusta form the inner or median boundary of the lon-
gitudinal grooves into which the gill-pores open.

In P. flava, however, the U-shaped gill-slits open freely to
the exterior throughout their whole extent, and their external
openings are therefore not reduced to minute circular or ellip-

ON PTYCHODERA FLAVA. 171

tical pores as they are in most other species of Enteropneusta,
and indeed in most other species of the genus Ptychodera
itself.

In the possession of this remarkable free pharynx P. flava
exhibits its close affinity with P. erythrea, Sp., from the
Red Sea, and P. bahamensis, Sp., as described by Spengel,
especially, as it would appear, with the latter.

As in all species of Ptychodera, so here, the anterior portion
of the alimentary is subdivided by a deep longitudinal con-
striction into a dorsal, branchial, and a ventral cesophageal
portion.

From what has been said above it is obvious that P. flava
is a very favorable species for studying the structure of the
pharynx, since the latter can be easily removed and examined
under the microscope.

As might be supposed, there is not much anatomical detail
to be added to the exhaustive account given by Spengel of
the Enteropneustan pharynx; but there is a point of import-
ance in any comparison between the latter and the pharynx of
Amphioxus, which is not emphasised in Spengel’s monograph ;
in fact, so far as I can ascertain, he makes no reference to it
whatever, and yet it is of prime significance.

On examining the pharynx of P. flava one cannot fail to
be astonished at the relatively enormous size of the tongue-
bars as compared with the primary or septal bars (fig. 3).
The former are wide, opaque, dark brownish coloured struc-
tures, while the former are narrow and semi-transparent.
The contrast between the primary and tongue bars in point of
size and appearance could hardly be much greater than it is in
P. flava.

In fact, it may be stated categorically that in the Entero-
pneusta in general (as shown by Spengel’s figures), and in P.
flava in particular, the tongue-bars are much larger than the
septal bars ; while in Amphioxus, as is well known, the reverse
condition obtains, in that the primary bars are larger (but not
so much larger) than the tongue-bars.

This is a most important difference, not only in an ana-

172 ARTHUR WILLEY.

tomical and physiological sense, but in a morphological sense
also, because while no zoologist would conclude from it that
the corresponding structures in the respective types were mor-
phologically different, yet it serves to explain most of those
differences in detail which Spengel so elaborately enumerates.

In correlation with the great size of the tongue-bars in the
pharynx of the Enteropneusta, it is not surprising to learn
the important fact from Spengel that they, rather than the
primary bars, are hollow, containing a wide prolongation of
the celom. “In Folge dessen,” says Spengel himself (loc.
cit., p. 725), “kann die Zunge der Amphioxus-Kieme nicht,
wie die der Enteropneusten, zwei Zungenzinken enthalten,
sondern nur eine, die allerdings aus zwei gleichen Halften
zusammengesetzt erscheint.” .

Thus, according to Spengel’s own assertion, the presence of
two skeletal rods instead of one only, in the tongue-bars of
the Enteropneusta, stands in correlation with their hollow
character; while the latter, in its turn, is correlated with the
great size of the tongue-bars. In consequence of the occur-
rence of two separated skeletal rods in the tongue-bars, the
dorsal arcuate extremities of the branchial skeletal structures
are not continuous as they are in Amphioxus, but are in-
terrupted at each tongue-bar (cf. Spengel, loc. cit., Taf. ii,
fig. 21).

The fact that the skeletal rod of the tongue-bar is single in
Amphioxus and double in the Enteropneusta is an anatomical
difference of importance, but not necessarily and, it may be
confidently asserted, uot in fact a morphological difference.
But it accounts, on the principle of correlation, for other
differences upon which Spengel lays such stress. It fully
explains the difference which Spengel has had printed in spaced
type—namely, that “ beim Amphioxus gehort jede Skeletgabel
einer einzigen, bei den Enteropneusten aber zwei Kiemen an.”

Before coming to the conclusion that ‘die Kiemen der
Enteropneusten und. des Amphioxus . . . . wesentlich ver-
schiedne, morphologisch einander nicht entsprechende Bild-
ungen sind,” Spengel makes a serious attack upon the synap-

ON PTYCHODERA FLAVA. 173

ticula or cross-bars of the pharynx. He says (loc. cit., p. 726),
“Bei den Enteropneusten sind die Synaptikel stabformige
Sprossen, welche zwischen einer Zungen- und einer Septal-

zinke ausgespannt sind. . . . Anders beim Amphioxus. Dort
sind... . die Synaptikel . . . . zwischen zwei Septalzinken
ausgespannt und der dazwischen liegenden Zungezinke nur
angelagert.””

A glance at fig. 3 accompanying this paper will, I think,
show conclusively that the above quotation represents merely a
subjective mode of expression. In P. flava the synapticula
on one side of a tongue-bar are approximately often quite
opposite to those on the other side. As the skeleton of the
wide tongue-bar is separated into two halves, the synapticula
must necessarily likewise be separated.

By insisting on detailed differences, and more or less neglect-
ing the broader distinctions to which they are subordinate,
and which would in great measure account for the former,
one can really arrive at any conclusion to which the individual
mind is inclined.

To see the synapticula in P. flava the pharyngeal wall
must be viewed from the inside, since, as pointed out by Spengel,
these structures are placed towards the inner side of the gill-
bars, and not on their outer face asin Amphioxus. I have found
it a good method to kill in dilute formol, and having removed
and opened out the pharynx to mount it in the same liquid.
It is well to cut away portions of the genital pleura before
preservation in formol, as they are otherwise liable to become
glued together in the dorsal middle line.

The number of synapticula in a vertical or dorso-ventral
series in P. flava is from ten to thirteen. In this species, as
in the majority of Enteropneusta, the gill-bars are not straight
as they are in Amphioxus approximately, but are markedly
bowed, the convexity facing outwards.

Gonads of P. flava.

Another remarkable feature of the species under considera-
tion, which it presents in common with P. erythrea, P.

174 ARTHUR WILLEY.

bahamensis, and the post-branchial region of P. aurantiaca,
is the diffuse arrangement of the gonads. They are not in the
remotest degree arranged one after the other, in a manner
resembling a paired metameric series, as they are more or
less in most other Enteropneusta, but they are scattered in the
most irregular way in the substance of the genital pleura
(cf. figs. 4and 5). In correspondence with this multiplication
of the gonads, Spengel has shown in the above-named three
species of Ptychodera, which he had at his disposal to examine
by sections, that several gonaducts may be involved in a single
transverse section, each gonad having its own duct, opening to
the exterior on the inner surface of the genital pleura. This
is also the case in Balanoglossus canadensis, Spengel,
in which, however, there is a multiple series of gonads, both
medial and lateral, of the gill-pores (cf. Spengel, loc. cit.,
Taf. 17, fig. 22). Spengel states that he has never found
gonads mediad of the gill-pores, either in Ptychodera or
Schizocardium.

It has been quite impossible for me, under the circumstances,
to prepare a series of sections, and I have had to make the
best of hand preparations and dissection. But the diffuse and
irregular arrangement of the gonads in P. flava can perhaps
be even better realised in in toto preparations than in sections.

Figs. 4 and 5 represent a few of the gonads in male and
female individuals respectively, as seen under a low power in
small detached portions of the genital pleura. The gonads, as
shown in the figures, have in both sexes the most variable
outline. Their appearance naturally varies somewhat with the
state of contraction or extension of the animal or portion of
the animal. Detached fragments of the genital pleura will
creep from under the cover-glass like a Planarian.

The integument over the testes on the inner face of the
genital pleura in P. flava is characterised by patches of dark
brown pigment, and on this account it is possible to distinguish
the males from the females (fig. 4a).

The female gonads (fig. 5) contain a variable number of
ova, which do not take up the whole volume of the gonads, but

ON PTYCHODERA FLAVA. 175

are surrounded by a mass of small refringent globules. As the
ova in the individual from which the preparation represented
by fig. 5 was taken appeared to be sub-mature, it seems not
impossible that these globules are of the nature of mucous
granules. Spengel says that in P. minuta, when the forma-
tion of sexual cells commences the fat-like substance begins to
disappear, and is finally quite replaced by ova and spermatozoa.
The conditions in P. flava appear to differ from this.

The ova (fig. 6), when obtained free by artificial rupture of
the gonads, are seen to be surrounded at an interval by a hyaline
double-contoured membrane, the follicular membrane. They
are opaque, being filled with fine yolk granules. They measure,
apart from the membrane, ‘006 mm. in diameter.

With regard to the shape of the gonads in the branchial and
post-branchial regions of the genital pleura, there is no difference
whatever in P. flava (cf. fig. 4a). Butin P. minuta and
P. Sarniensis, Spengel states (loc. cit., p. 653) that in the
branchial region the gonads are almost always simple un-
branched sacs, while in the post-branchial region their form
becomes more complicated.

Tn one instance of a male individual of P. flava I observed
a much elongated gonad, as long as four or five of those on
either side of it taken together.

Systematic Position of P. flava, Eschscholtz (char.
emend. mihi).

As might be expected, the short description given by Esch-
scholtz, beyond indicating by the presence of the genital pleura
that his species belonged to the genus Ptychodera in Spengel’s
system, fell far short of being a satisfactory specific diagnosis.

In consequence of this, Spengel has wrongly placed this
species in his genus or sub-genus Tauroglossus. He does not,
however, finally assume this, but puts a mark of interrogation
against it.

Spengel has, as it seems with justice, subdivided the genus
Ptychodera, suggesting the formation of the family Ptycho-
deridz, with the three genera, Ptychodera, Tauroglossus, and

176 ARTHUR WILLEY.

Chlamydothorax. But, to avoid confusion, it is more con-
venient at present to regard these as sub-genera of the genus
Ptychodera. :

Ptychodera (sensu stricto) has rudimentary genital pleura ;
to it belong P. minuta, Kowalevsky, and P. Sarniensis,
Koehler.

Tauroglossus is distinguished by the dorsal origin of the
genital pleura; and to it belong T. apertus, Spengel, T.
claviger, Delle Chiaje, T. gigas, Fr. Miiller,T. aurantiacus,
Girard, and T. australiensis, Hill.

Finally, Chlamydothorax is characterised by the ventral
origin of the genital pleura; and to it are assigned Ch.
erythrzus, Spengel, Ch. bahamensis, Spengel, and probably
Ch. ceylonica, Spengel, although the last-named species is
only referred to in Spengel’s monograph, and not fully
diagnosed.

From the account given above of P. flava, it is at once
evident that its place is under the sub-genus Chlamydo-
thorax.

The fact of its close affinity to P. bahamensis instead of
to its neighbour, P. australiensis, of New South Wales, is
interesting in connection with the fact that the Amphioxus
(Asymmetron caudatum) which I obtained from the Loui-
siade Archipelago, and described in a previous communication,
is likewise much more closely related to the Bahama species
(A. lucayanum, Andrews) than to the Australian forms.

CoNCLUSIONS.

My investigation of P. flava, necessarily somewhat super-
ficial, has nevertheless sufficed to establish its systematic posi-
tion, but would hardly allow me to engage in an extended
morphological discussion. Still there are a few points upon
which one might venture an opinion, especially since it is
impossible to have once seen the free, erect, upstanding
pharynx of P. flava without being deeply impressed.

Moreover the account, admirable enough, which Spengel

ON PTYOHODERA FLAVA. L7Z

has given of the other two described species of the sub-genus
Chlamydothorax was based in each case on a single specimen
preserved in spirit, so that an Enteropneust with an eminently
free pharynx has never been studied in the living condition
before. And this makes a difference.

The conclusions arrived at by Spengel, based as they were
upon such laborious and prolonged researches, are entitled to
the profoundest respect. Still, with the best will in the world,
I cannot follow him in his adverse criticism of the theory as
to the relationship of the Enteropneusta to the Chordata. One
might conceivably be able to relinquish the idea of the existence
of a notochord or its representative in the Enteropneusta, but
the gill-clefts are a perpetual fact, and it seems little less than
perverse not to recognise it.

Indeed, in his remarks directed against the assumed Chor-
date affinities of Balanoglossus, it would almost appear that
Spengel has carried the analytical method of argumentation
to an extreme, and that he is unable to see a general corre-
spondence or homology through the veil of differences in detail.
The other extreme is to imagine correspondences where none
exist. But it is certainly not necessary to force matters in
any way in order to clearly recognise an affinity between the
Enteropneusta and the Chordata.

Unfortunately we are here in the presence of one of those
distressing instances, so common in the realm of morphology,
in which two entirely opposed views can be more or less equally
supported. This is due, as Spengel himself points out (loc.
cit., p. 722), to the lack of a definite method or criterion in
attempting to answer morphological questions.

There is, however, a principle which should be of service in
this connection, namely, the principle of correlation between
structural and topographical features on the one hand, and
physiological or functional peculiarities on the other.

Spengel lays great stress upon the dorsal position of the
gill-pores in the Enteropneusta and their ventral (sic) position
in Amphioxus, this difference in position being especially in-
dicated by the relations of the vascular system, the propelling

vot. 40, parts l.—new sER. M

178 ARTHUR WILLEY.

vessel being dorsal in the former and ventral in the latter.
This fact, according to Spengel, is in itself almost enough to
prove that the gills of the Enteropneusta and of Amphioxus are
essentially different structures, and that they do not correspond
with one another morphologically. It may, indeed, be said to
be Spengel’s strongest point in his objection to the supposed
Chordate affinities of the Enteropneusta. But, by applying
the above-mentioned principle of correlation to the elucidation
of this problem, these and other differences may be viewed
from quite another aspect.

The question, with what other characteristic in the organi-
sation and mode of life of the Enteropneusta the dorsal posi-
tion of the gills and gill-pores may be correlated, is not con-
sidered by Spengel.

Balanoglossus (employing the name in the wide sense) is a
creeping animal,? and the ventral surface, as in all creeping
animals, is the locomotor surface. Some animals may swim
on their backs and others on one side, but all who creep do
so on their ventral surface. It is inconceivable that gills or
gill-pores could occur on the locomotor surface.

On the contrary, Amphioxus, when active, is essentially a
swimmer, and it can no more creep than Balanoglossus can
swim. There is, therefore, no such locomotor surface in
Amphioxus, and the dorsal region of the primitive alimentary
canal is converted into a skeletal support for the body, viz.
the notochord.

The gill-slits and gill-pores of the Enteropneusta are placed
dorsally, therefore, in correlation with the locomotor function
of the ventral surface, the latter not having such a function in
Amphioxus ; and the general homology between the pharyn-
geal apparatus in the two types is not thereby prejudiced.

1 That such a difference in the direction of flow of the blood should not be
overrated in the Protochordata is shown by that very well-known faculty of
the Tunicate heart of reversing its action and consequently the direction of
propulsion.

2 The kind of burrowing undertaken by Balanoglossus is a variety of
creeping, but it creeps too, apart from its burrowing propensities,

ON PTYCHODERA FLAVA. 179

To return to Ptychodera flava, the formation of a tem-
porary atrial cavity round the branchial sac by the mutual
approximation of the genital pleura is a most striking fact.
Spengel calls attention to this, and rightly urges that the
peripharyngeal cavity so formed is more readily comparable
to the atrium of Amphioxus than anything else that has been
suggested. He then goes on to add, however, that in his
opinion it is nothing but an entirely superficial resemblance.

Nevertheless it is a real resemblance. The branchial sac
being dorsally placed in accordance with the principle above
referred to, the peribranchial cavity must be also dorsal in
Ptychodera. Presumably there can be no doubt that there is
a general homology between the atrium of the Ascidians and
that of Amphioxus; and yet in the former it is a dorsal
structure (except in the free-swimming Appendiculariz), and
in the latter ventral.

With regard to the synapticula or cross-bars in the branchial
skeleton, Spengel (loc. cit., p. 725) draws attention to the fact
that they are not present in the genera Glandiceps and Balano-
glossus, but are present in Ptychodera and Schizocardium,
which, he says, are probably younger phylogenetically. But
it is very much open to doubt whether Ptychodera is phylo-
genetically younger than the other genera of the Entero-
pneusta.

On page 357 of his beautiful monograph Spengel gives a
list of what he regards as signs of a primitive organisation in
the group. These are open to the criticism that they are,
without exception, all negative properties,—the lack of this,
that, and the other.

Then with regard to Ptychodera he says (p, 358), “ Als die
héchste Form erweist sich endlich Ptychodera.”” For my part,
T deny this, and oppose the view on the following grounds.

In the first place, the positive fact of the diffuse arrange-
ment of the gonads, which is a characteristic of P. flava and
of the other species belonging to the sub-genus Chlamydo-
thorax, bears all the marks of an archaic type.

Secondly, it seems only reasonable to suppose that the

180 ARTHUR WILLEY.

elements which compose the branchial skeleton, namely, pri-
mary or septal rods, secondary or tongue-rods, and synapticula
or cross-rods, were developed at a time and in a type in which
their presence was absolutely necessary to prevent the collapse
of the branchial sac. It is not so easy to see that their
presence is directly necessary to those forms in which the
septa between adjacent gill-slits are fused with the thick
parenchymatous tissue of the body-wall, and the slits only
open to the exterior by minute pores. But they are there not-
withstanding, namely, because they are derived from forms in
which the presence of skeletal supports for the much-perforated |
pharyngeal wall was a sine qua non. Such a form is P.
flava, with its free and otherwise unsupported pharynx.

If so much is admitted, then the presence of the genital
pleura, covering over the unprotected pharynx, needs no
special comment.

Thirdly, the fact that in Schizocardium and in Glandiceps
Hacksi the anterior region of the alimentary canal is not
subdivided into branchial and cesophageal portions militates
strongly against these genera being regarded as more primitive
than Ptychodera.

Finally, the habitat of Ptychodera in the littoral zone,
often between the tide-marks, is another positive indication of
the primitive character of the genus. The greatest depth
recorded by Spengel for a species of Ptychodera is 20 feet
for P. minuta in the Bay of Rio de Janeiro. Schizocardium
(S. brasiliense, Spengel) descends to 18 to 20 fathoms;
Glandiceps (G. talaboti, Marion) descends to 450 metres,
while another species (G. abyssicola, Spengel) was obtained
by the “Challenger” from 2500 fathoms; Balanoglossus
Kupfferi, von Willemoes-Suhm, was obtained from 12 to 16
fathoms.

It is very possible that the forms which have migrated into
deeper water may have retained some primitive features which
are lost to the littoral or tidal forms, just as many of the
Elasipoda among the Holothuroidea have retained the
primitive connection of the stone-canal with the exterior,

ON PTYCHODERA FLAVA. 181

which has been lost by all other recent Holothurids.! But it
seems clear enough that the occurrence of diffuse gonads, and
the free, open pharynx in the sub-genus Chlamydothorax, and
particularly in P. flava, are facts which point conclusively to
the archaic character of Ptychodera.

Summary oF Principat Resvtts.

1. This is the first time that an Enteropneust with a free
pharynx has been studied in the living condition.

2. The Ptychodera flava of Eschscholtz (char. emend.
mihi) is rightly assigned by Spengel to his amended genus
Ptychodera, as shown by the presence of the genital pleura,
of external liver saccules, and by the length of the collar
region.

3. P. flava belongs to Spengel’s sub-genus Chlamydo-
thorax, as shown by the ventral origin of the genital pleura,
the diffuse gonads, and the free pharynx.

4. In the fact of the gill-slits being open directly to the
exterior throughout their entire length, P. flava is more
closely related to P. bahamensis than to any other described
species. This is also indicated by the simple rows of paired
liver saccules as opposed to the irregular multiple arrangement
met with in P. erythrea.

5. The genus Ptychodera (referring more especially to the
sub-genus Chlamydothorax) probably represents an archaic
type, as shown by the diffuse arrangement of the gonads, the
free pharynx, and its littoral habitat ; and it is probably not,
as Spengel supposes it to be, phylogenetically younger than
the other genera of Enteropneusta.

6. The gill-slits, branchial skeleton, and the temporary atrium
formed by the apposition of the genital pleura in Ptychodera,
offer a general homology to the corresponding structures in
Amphioxus and the Ascidians, while presenting many differ-
ences in the details of their structure and relations.

7. Some of these differences are comparatively unimportant,

1 Cf. Hjalmar Théel, “Report on the Holothuroidea,” part ii, ‘ Chall.
Rep. Zool.,’ vol. xiv, 1886.

182 ARTHUR WILLEY.

and such as might well be expected to occur in distantly re-
lated forms with such totally different habits of existence,
while others are to be accounted for by a wide interpretation
of the principle of correlation between structure and function.

8. Many differences of detailed structure in the pharyngeal
wall and its skeletal supports between the Enteropneusta and
Amphioxus are to be correlated with the fact that, in the
former, the tongue-bars are larger (often, as in P. flava, very
much larger) than the primary bars, while in the latter the
reverse condition obtains.

Additional Note.

As the Marshall Islands are distant about two thousand
miles from New Caledonia, and as the species figured by
Eschscholtz renders possible the interpretation that its genital
pleura had a more dorsal origin than in the species above
described from New Caledonia, it is advisable provisionally
to name the latter P. flava-caledoniensis, or simply P.
caledoniensis, until the form from the Marshall Islands
comes to be re-examined.

IsLE oF PinEs;
August 2nd, 1896,

ON PTYCHODERA FLAVA. 183

EXPLANATION OF PLATE 5,

Illustrating Mr. Arthur Willey’s paper “On Ptychodera
flava, Eschscholtz.”

Fic. 1.—Dorsal view of Ptychodera flava. The annulations of the
genital pleura are indicated anteriorly on the left side; those of the body
proper behind the hepatic region. pr. Proboscis. c. Collar region. 4
Branchial region ; the genital pleura are slightly divaricated, and the pharynx
is visible. p.4. Post-branchial region. 4%. Hepatic region. J.s, Dorso
lateral streaks marking the course of the ciliated grooves in the intestinal
wall. d.z. Line marking the course of the dorsal nerve-cord. a, Anus.
Drawn from the living object.

Fic. 2.—Anterior portion of another individual from the dorsal side, show
ing the tuberosity caused by a Copepod parasite. The cross-lines over the
infested region represent continuations of the annulations of the genita
pleuron.

Fic. 3.—Portion of the pharyngeal wall of P. flava, including the lower
portions of three tongue-bars, to show the larger size of the latter as com-
pared with the primary bars. The lower free extremities of the tongue-bars
are differentiated from the rest of the bar, being marked off in each case by a
pigmented line of demarcation. ¢.4. Tongue-bars. p.d. Primary or septal
bars. sy. Synapticula. g.s. Gill-clefts. @.s. Cisophageal ridge (=
Grenzwulst of Spengel), forming the boundary between the branchial and
cesophageal portions of the alimentary canal. Drawn from a preparation in
formol. Zeiss cam. luc., oc. 3, obj. A.

Fic. 4.—Groups of male gonads of P. flava sketched from a detached
piece of the genital pleura.

Fic. 4¢.—A single testis from the branchial region with the pigment
patches over its surface.

Fic. 5.—Groups of female gonads of P. flava with ova. The portions of
the gonads not occupied by the ova are filled up with coarse mucous (?)
granules.

Fic. 6.—An artificially liberated ovum of P. flava surrounded by the
double-contoured follicular membrane. Zeiss cam. luce., oc. 3, obj. D.

A?

, a Gas Ge . '
ae ‘ ; arr cls tReet es We
: ~-
i
*
=
‘ (i~® J i : ’
, : ? 7 rrivi tas P
> ee ’ ‘ f G' 7
‘ E
7 : .
A ij = :
« . . is al j - rine F) ahi ‘1
: ae
anv. ; ;
‘ j f j é
we 4 ii
} pee DO a
; J T Aa! af
. P
“ , pind
+. 7 .
i i 3 ‘ A ; i
® J ;
, > 2
a ‘
, : on
»é ’ ,
x -
ie s Aa ter
7 )
ie -
! " \ 4
’ , 2 F rat ay py
Te ote aap ts ‘
* S - é i 7 :)
: a: 7 inyrye?
= ; 7
Mbhthac ' ad TT) , i ;
2 ’ So) v " :
. a ; >
i i wish F ;

ai; Tine. i 7 3 ; ibe he at OL eh eae
; ( sae or \ ge ir: ‘ A, ares yoga
: 4
Vea, .) ity hace Ts ai (ae
‘ ; «ti am
i We ote © I Pes iat bin Ole wi | ache Lehit ers ‘i
aan nggkag

eet pa ew Lal an ia Mea aticitebeg
iy :
f'n A et Mog a DR Vea ee ai act § tn
- - - <<
\ * aan Ma ie ie Ee Geol ss
. y LP) s re ’ P J 4 ee S ; E sil
on y. Ars ily pels SMe oie
iA *
‘i
-
:
| ’
¢
| 7 nh 4

ON THE NEPHRIDIA OF THE POLYOHATA. 185

On the Nephridia of the Polycheta.
Part I.—On Hesione, Tyrrhena, and Nephthys.
By

Edwin 8. Goodrich, B.A.,
Assistant to the Linacre Professor, Oxford.

With Plates 6—9.

Tue following observations on Hesione sicula, Dch.
(Fallacia sicula), and Tyrrhena Claparedii, Quatref.,
were begun last year at Naples, and completed at Oxford ;
those on Nephthys scolopendroides, Dch., and Nephthys
ceca, Fabr., were begun at Roscoff, continued at Naples, and
finished here.! The material was examined in all cases both
fresh and preserved.

HESIONE.

The Ciliated Organ.—When full grown, Hesione
sicula is of about the size and appearance of Peripatus
capensis, and can be easily dissected. On opening up the
worm dorsally, and turning back the cut edges of the body-
wall, the segmental dorso-lateral blood-vessels are seen ex-
tending outwards from the intestine to large bundles of oblique
vertical muscles attached to the body-wall between each pair
of parapodia. On reaching these bundles the vessels pass
downwards amongst the muscles, and reappear near the median

1 I must express my sincere thanks to the directors and staff of the Zoolo-
gical Stations at Naples and at Roscoff for the kind and helpful way in which
they received me.

186 EDWIN 8S. GOODRICH.

ventral line, entering the ventral vessel of that side above the
nerve-cord. Dr. Hisig has accurately described and figured
the vascular system of Hesione.! I can refer the reader to his
paper for further details. It is at the point where the dorso-
lateral vessel (A of Hisig) first touches the hinder edge of the
oblique muscles that the “ ciliated organ” is situated.

Fig. 1 isa careful drawing of an inner view of the left side
of a portion of a worm hardened and cut in half. It shows
the position and relative size of the ciliated organ, c.0., in two
segments about the mid-region of the body in front of the
posterior extremity of the long pharynx, which has been
removed, together with the longitudinal dorsal vessels.
Below are seen the longitudinal ventral vessels (v.v. of
Hisig), into which enter the ventro-lateral vessels, vl. v.
(v. vv. of Hisig). The ciliated organ, of considerable size, is
crescent-shaped, with two free horns or limbs, the internal
one alone being visible in this drawing. The external horn
curls round to the outer side of the bundle of muscles.

A microscopic examination shows that that surface of the
organ which faces backwards and away from the muscle is
furrowed with deep grooves, alternating with sharp-edged
ridges, running transversely to its long axis (figs. 3, 14, and
20). The ridges (fig. 3, c. 7.) extend up to the very edge along
the dorsal free margin ; ventrally, on the contrary, they stop
short of the edge, leaving a narrow smooth strip beyond. The
whole of this grooved surface is densely covered with fine cilia.

A comparison of three transverse sections (figs. 9, 10, and 11),
taken from before backwards, of three sagittal sections (figs.
14, 15, and 16), taken from within outwards, of three frontal
sections (figs. 20, 21, and 22), taken from above downwards,
and of the reconstruction in fig. 2 will make clear the structure
of the ciliated organ. It is free along its dorsal margin; but
for about the middle third of its length the lower edge is
attached to the muscles by a thin septum, formed by a double
layer of peritoneum (figs. 3, 21, and 16, at v.e.). About half-

1 H. Hisig, “ Ueber das Vorkommen eines schwimmblasen ahnlichen Organs
bei Anneliden,”* ‘ Mittheil. Zool. St. Neapel,’ vol. u, 1881.

ON THE NEPHRIDIA OF THE POLYCHATA. 187

way the ventral edge is produced downwards into a sort of
hollow groove (figs. 2, 10, 13, 16, v. prol.), which becomes
directly continuous with the lip of the nephridial funnel, neph. f.
This point will be dealt with again later. The ventral pro-
longation of the ciliated organ (not ciliated at its lowest ex-
tremity) lies very near the epidermis lining the intersegmental
groove, being attached to it by a fold of peritoneum (fig. 13).

The histology of the ciliated organ presents uno striking
peculiarity. Like that of Nereis,! it is similar throughout,
and shows no trace of glandular structure. No cell walls are
visible ; the cytoplasm is even, but stains more deeply towards
the ciliated surface (fig. 5). Numerous small round or oval
nuclei are situated near the outer surface, chiefly along the base
of each ridge. The anterior surface of the organ, which is
turned towards the muscle, is lined with flat coelomic epithe-
lium (fig. 5, ¢. ep.).

A pair of ciliated organs occurs in the anterior region of
every segment of the body after the third parapodium. The
organs are largest in the median segments, where they attain
a length of 2 mm. from the end of one horn to that of the
other, and possess from fifty to eighty transverse ridges.

The Nephridium.—The nephridium of Hesione opens
internally into the colom by means of a simple funnel, pro-
vided with long, stiff, curved cilia (fig. 17). The lips are
reflected all round the aperture, on one side curving round
the dorso-ventral blood-vessel, to which the nephrostome is
attached (figs. 2 and 13). A narrow neck of varying length
leads from the funnel to a wide and somewhat twisted tube
(figs. 2, 16, and 17, neph. ¢.), which in turn becomes narrower
and more convoluted, forming a mass flattened dorso-ventrally,
stretching backwards on the floor of the segment (fig. 2).
The nephridial tube does not branch in its course. Finally it
widens slightly, and runs into the body-wall, through which it
opens by a small pore immediately below the base of the
parapodium (fig. 2, neph. p.).

1 “On a New Organ in the Lycoridea, &c.,” ‘Quart. Journ. Mier. Sci.,’
vol. xxxiv, 1893.

188 EDWIN S. GOODRICH.

Sections show that, as already mentioned, the lip of the
nephrostome is directly continuous with the ventral prolonga-
tion of the ciliated organ. Since this connection is, I believe,
of considerable morphological importance, I have figured it in
detail. A transverse section of the two organs is shown in
fig. 12 (an enlarged view of part of fig. 10), a frontal section
in fig. 22, a sagittal section in fig. 16, and a transverse in
fig. 13. From these it will be seen that the ventral prolonga-
tion of the ciliated organ is distinguishable from the lip of the
nephridial funnel,—the cells of the former being small, deeply
staining, and not ciliated; those of the latter being ciliated,
less deeply staining, and larger. The convex outer surface of
the large cells forming the funnel can be seen in the living
tissue (fig. 17).

Passing downwards, a section through the wide tube (fig. 4)
shows that the lumen is intra-cellular, and the cilia arise from
various places round the inner surface; cil., fig.6, represents a
section through the convoluted mass, where the lumen is cut
through seven times. The whole organ is, of course, covered
with ccelomic epithelium (c. ep.).

Such nephridia occur in all the segments of the body
occupied by the ciliated organs.

TYRRHENA.

In Tyrrhena Claparedii the condition of the ciliated
organ and of the nephridium is very much the same as in
Hesione. As in the latter, the nephridial funnel is in direct
continuity with the ventral prolongation of the ciliated organ.
The nephridium is of simpler structure, the tube being very
little convoluted. The ciliated organ itself is in the same
position, but it is less elongated in shape, and its ciliated
surface has fewer ridges.

NEPHTHYS.

The Ciliated Organ.—When describing the ciliated
organ of the Lycoridea! I stated that I had been unable to

1 Loe. cit.

ON THE NEPHRIDIA OF THE POLYOHATA. 189

discover it in Nephthys. At that time I had only been able
to study poorly preserved material. After the examination of
a large number of living and well-preserved worms, I am able
to say that what I had before mistaken for the nephrostome is
really the “ ciliated organ.”

In fig. 7 is given a careful enlarged drawing of the inner
view of the right side of some segments taken from the mid-
region of the body of a large hardened specimen of Nephthys
ceca, and cut in half. The musculature is much more de-
veloped and complicated than in Hesione. Thin strands of
muscle stretch from the dorsal surface of the intestine out-
wards to the body-wall, forming an incomplete longitudinal
oblique septum, 0b/. sept. Two powerful strap-like longitudinal
muscles lie beneath the intestine on either side of the ventral
blood-vessel, v.v., and give off transverse muscles at every
segment. This ventral band overlies the thick muscular
transverse septa, sept., which extend upwards to the dorsal
longitudinal muscles, d./.m. Various bundles of vertical
oblique muscles extend outside the main longitudinal muscles.

The dorsal vessel, d.v., gives off a dorso-lateral vessel at
each segment, d./.v. (dorso-pedal of Jaquet”), which passes
down to the large bundle of oblique muscles, 04/.m., corre-
sponding to those described above in Hesione, and on which
are situated the ciliated organs, c.o. Thence the dorso-ventral
blood-vessel (branch @ of Jaquet) runs downwards and inwards
to join the ventral subintestinal vessel, vv. An offshoot of
the ventro-lateral vessel goes to the neural longitudinal vessel
of that side.

The ciliated organ when dissected out resembles somewhat
the shell of Pecten (fig. 19). It is smaller and more rounded
in shape than that of Hesione. The ciliated surface of the
organ faces outwards and forwards; it is raised into about
twenty sharp ridges, alternating with deep grooves, which

1 A specimen from St. Andrew kindly given to me by Dr. W. B.
Benham.

2M. Jaquet, “Recherches sur le Systéme vasculaire des Annélides,”
‘ Mittheil. Zool. St. Neapel,’ vol. vi, 1886.

190 EDWIN 8S. GOODRICH.

converge towards the ventral and posterior extremity. The
edge of the ridges is extremely thin and jagged (figs. 19 and
26). The clear protoplasm of which it is formed contains a
number of fine refringent granules.

Sections show that the organ is composed of the loose tissue
characteristic of these worms, which has a fibrous appearance
when preserved (figs. 238, 24, and 25, c.o.). The nuclei are
scarce and scattered irregularly.

In Nephthys the position of the ciliated organ is essentially
the same as in Hesione (figs. 7 and 8). Its upper expanded
portion rests on the same dorso-lateral vessel, d./.v., and its
lower end forms a sort of deep groove or ventral prolongation,
v. prol., running down the body-wall near the intersegmental
groove. There appears to be no communication whatever with
the lumen of the nephridium.

These ciliated organs occur in both sexes throughout the
body in every segment except about the first ten. They are
more fully developed in mature than in young specimens.

The Nephridium.—The nephridium of Nephthys is of
very remarkable structure, representing, indeed, an entirely
new type of Chetopod nephridium, unlike that of any
member of that group hitherto described.

The small external aperture lies on the ventral surface of
the body below the parapodium, and a little beyond the outer
edge of the ventral longitudinal muscles. Leading from this
nephridiopore (fig. 8, neph. p.) is a narrow canal running
upwards, then obliquely forwards through the muscular
septum, at which point it becomes narrower still (at all events
in sections). Emerging from the septum the nephridial canal
runs inwards and forwards, clinging closely to a blood-vessel
from the body-wall which joins the ventro-lateral vessel.
Ascending the dorso-ventral vessel, and increasing slightly in
diameter, it passes along the inner and anterior edge of the
ciliated organ on the posterior non-ciliated surface. Finally it
emerges on the top of the ciliated organ, where it divides into
free branches forming a sort of plume.

The structure of the minute nephridial tube is simple. The

,.

ON THE NEPHRIDIA OF THE POLYOHATA. 191

wall is composed of loose fibrous-looking tissue (an appear-
ance probably due to the presence of a large quantity of water
in the living cells), A denser layer surrounds the lumen
(figs. 24 and 25, neph. ¢.). Nuclei are seen in the wall here
and there, but the lumen is probably intra-cellular. As in
Nereis, the cilia are disposed along one side of the tube only
(occasionally, however, in two opposite rows), and in a nephri-
dium freshly dissected out, the characteristic undulation pro-
duced by such an arrangement of cilia is very marked.

The path of the canal as it passes up behind through the
substance of the ciliated organ is shown in fig. 19 by the dotted
line, p. neph. ¢., and in the sections drawn in figs. 23, 24, and 25,
neph. ¢t. On reaching the upper edge of the ciliated organ the
tube, with its lumen, divides into three, four, or five branches,
more generally four, which float freely in the celom. One or
more of the branches usually has a T-shaped extremity, and
the multiplication of the branches (to the maximum number
of five) would appear to take place by the splitting of such a
T-shaped branch to its base. The lumen of the nephridial
canal ends blindly at the tip of each branch (as far as I have
been able to make out). But the chief interest lies in the
minute structure of the branches themselves.

Roughly speaking, each branch may be said to consist of a
double row of cells, with swollen bases containing the nuclei,
enclosing the lumen of the canal (figs. 26, 27, and 28). Each
cell tapers off into a long narrow neck, ne., which stretches out
at right angles to the axis of the branch on which the cells are
set. At its distal extremity the neck-like process becomes
slightly swollen, and sharply bent round towards the corre-
sponding cells of the other side. This sort of crook bears at
its extreme end a long narrow tube, ¢w., which runs down
parallel to the neck towards the nephridial canal at the base of
the cell. Piercing the wall of the canal the delicate tube
leads directly to its lumen, even projecting slightly into it.
A very long and slender flagellum undulates freely in the tube.
Attached by its base at the distal end of the tube, the flagellum
passes downwards and out from the tube into the lumen of the

192 EDWIN 8. GOODRICH.

canal, where it is continued for some length, fl. Undulations
swiftly pass along the flagellum from its base to the free end
in the canal.

The protoplasm of the tube-bearing cells is very granular,
many of the granules being probably of an excretory nature.
At the distal curved end of the neck are generally seen delicate
protoplasmic projections, often of great length, floating in the
ceelomic fluid, pr. These are not cilia, but appear to be rather
of the nature of ameboid processes. I have not observed
them moving. The tube itself is rather narrower at its base
than at the end which enters the canal. It is quite straight
as arule, and oval in section. The wall is composed of a clear
refringent substance, apparently of a cuticular nature, which
resists the action of caustic potash longer than the proto-
plasmic parts of the cell.

The large oval nuclei (figs. 26—29, 7.) have in the fresh
tissue a vacuolated appearance. They are remarkable for the
extreme avidity with which they take up ordinary nuclear
stains, such as carmine or hematoxylin. So pronounced is
this tendency that in a preparation or section they become
deeply stained when the other nuclei are hardly yet affected,
and become intensely overstained by the time the other nuclei
are sufficiently coloured (see fig. 25, term. pl. neph.). A stained
preparation of a whole branch of the terminal plume, some-
what flattened out, is figured (fig. 29), showing the nuclei closely
packed in an irregular double row.

So far as I have observed the tube-bearing cells are never
placed singly, but are ranged in pairs along each side of the
canal (fig. 28). The bases and necks of two adjacent cells are
closely applied to each other along one side to near the distal
extremity, where they diverge to form the terminal crooks.
Although the cells are thus firmly fixed to each other, yet a
clear line of demarcation can always be detected separating
them along the middle line. Occasionally three cells are joined
together, as shown in the middle of the branch in fig. 26.
The nephridia occur throughout the body of the worm, ex-
cepting in the first and last few segments.

ON THE NEPHRIDIA OF THE POLYCHATA, 193

From the above description it appears that in Nephthys
there is no internal opening to the nephridium, which ends in
a bunch of short blind branches. The current in the lumen
of the tube produced by the flagella of the tube-bearing cells
and the cilia along the canal travels from the terminal branched
organ towards the external pore. It seems obvious, then, that
excretion must take place through the walls of the nephridium
as it does in these organs in the Platyhelmia and Nemertina.
Possibly the thin-walled tubes in which the flagella work act as
osmotic filters, allowing liquid to pass through from the cclom.
Solid excretory products are more probably conveyed to the
lumen of the canal by the cells themselves. The wall of the
nephridial canal often contains such a number of granules as
to appear of a distinctly brown or greenish hue.

A discussion of the bearing of the facts described above on
the question of the morphology of the nephridium and ciliated
organ is reserved for a second paper, in which it will be shown
that the nephridium of the Glycerids is built on essentially
the same plan as that of Nephthys, with flagellated “ tube-
bearing ” cells.

EXPLANATION OF PLATES 6—9,

Illustrating Mr. Edwin S. Goodrich’s paper “ On the Nephridia
of the Polycheta.”

List oF REFERENCE LETTERS.

ac. Aciculum. at.v.e. Attached ventral edge of the ciliated organ. 4. .
Body-wall. c.ep. Coelomic epithelium. c.f. Ciliated furrow. cil. Cilia.
c.o. Ciliated organ. cal. Colom. cel. corp. Celomic corpuscles. ¢.7.
Ciliated ridge. cut. Cuticle. c¢.w.int. Cut wall of the intestine. d./.m.
Dorsal longitudinal muscles. d/.v. Dorso-lateral vessel. dp.v. Dorso-pedal
vessel. d.v. Dorsal vessel. dv.v. Dorso-ventral vessel. epid. Epidermis.
j.d.e. Free dorsal edge of the ciliated organ. ff. Flagellum. go. Gonads.
int. Intestine. irr.m. Iridescent muscle. /.v.v. Left ventral vessel. /w.
Lumen. m. Muscle. 2. Nucleus. z.c. Nerve-cord. ze. Neck-like pro-

von. 40, PART 1.—NEW SER. N

194, EDWIN 8. GOODRICH.

cess of the cell. xeph.f. Nephridial funnel. xeph.p. Nephridial pore.
neph.t. Nephridial tube. od/.m. Oblique muscles. 00/. sept. Oblique mus-
cular septum. ar. Parapodium. ph. Pharynx. p.xeph.t. Path of the
nephridial tube. yr. Protoplasmic process. 7.v.v. Right ventral vessel.
sept. Septum. sabint.m. Subintestinal muscular band. term. pl. neph. Ter-
minal plume of the nephridium. ¢w. Tube. »./.m. Ventral longitudinal
muscle. v./.v. Ventro-lateral vessel. v.pro/. Ventral prolongation of the
ciliated organ. v.v. Ventral vessel.

PLATE 6.

Figs. 1—6.--Hesione sicula.

Fig. 1.—Enlarged inner view of the right half of two segments from the
mid-region of the body. Drawn from fresh and hardened specimens.

Fic. 2.—Diagrammatic reconstruction of the ciliated organ, the nephridium,
and accompanying blood-vessels, as seen from in front. ‘The cilia are not.
represented.

Fic. 8.— Portion of the ciliated organ much enlarged. From the fresh.
< 95, cam.

Fic. 4.—Section through the wide part of the nephridial tube. x 400,
cam.

Fie. 5.—Section through the ciliated organ, across the ridges, and show-
ing an accumulation of ccelomic corpuscles. x 400, cam.

Fie. 6.—Section through the narrow and convoluted region of the nephridial
tube, cutting the lumen seven times. x 400, cam.

Fic. 7.—Enlarged inner view of the right half of four segments from the
mid-region of Nephthyscerca. A portion of the intestine is represented
in front. Drawn from a hardened specimen.

Fie. 8.—Diagrammatic reconstruction of the ciliated organ, nephridium,

and accompanying blood-vessels of Nephthys scolopendroides, as seen
from in front. The cilia are not represented.

PLATE 7.

Figs. §—17.—Hesione sicula.
Fics. 9, 10, 11.—Pertions of three transverse sections, taken from before
backwards, showing the relation between the ciliated organ and the nephridium.
x 20, cam.

Fic. 12.—More enlarged view of portion of Fig. 10, showing the connection
between the ciliated organ and the nephridial funnel. x 130, cam.
Fie. 13.—Similar figure of a portion of a transverse section from another

series, showing the continuity between the ventral prolongation of the
ciliated organ and the lip of the nephridial funnel. x 400, cam.

ON THE NEPHRIDIA OF THE POLYCHATA. 195

Figs. 14, 15, 16.—Portions of three sagittal sections, taken from within
outwards, illustrating the relation of the nephridial funnel to the ciliated
organ. X 250, cam.

Fig. 17.—Enlarged view of the nephridial funnel freshly dissected out.

Fic. 18.—Piece of the dentated edge of a ridge of the ciliated organ of
Nephthys scolopendroides, much enlarged.

Fic. 19.—Enlarged view of the ciliated organ of Nephthys scolopen-
droides, freshly dissected out. The cilia are not represented.

PLATE 8.
Figs. 20—22, Hesione sicula. Figs. 23—25, Nephthys scolopen-
droides.

Fics. 20, 21, 22.—Portions of three frontal sections, taken from ahove
downwards, showing the ciliated organ and its ventral connection with the
nephridium. x 40, cam.

Fic. 23.—Portion of a frontal section of a female, showing the ciliated
organs and nephridial tubes. x 100, cam.

Fic. 24.—More enlarged view of the ciliated organ and nephridial tube,
from the same series as Fig. 28. x 400, cam.

Fic. 25.—Portion of a transverse section, showing the nephridial tube cut
twice, the ciliated organ, and a piece of the terminal branch of the nephridium.
x 180, cam.

PLATE 9.
Figs. all of Nephthys scolopendroides.
Fic. 26.—Enlarged view from a fresh specimen of a small piece of the

ciliated organ and a branch of the nephridial plume, showing the tube-bearing
cells.

Fic. 27.—Semi-diagrammatic transverse section of a branch of the terminal
nephridial plume.

Fig. 28.—Semi-diagrammatic longitudinal section of the same.

Fic. 29.—Branch of the nephridial plume stained and somewhat flattened
out. x 500, cam.

a , - : <a
‘ \ aoe 7 waa
sf ’ c 4 4 ¢
, . /
~
. .
d -
>
‘ iy '
J - bs é [=
? Wott ‘1s Tae et ay

eur : ' ior Smee
| r ' * ;
A ¥ ¢ ‘
a , : ‘ ay! teh
$ .
4 * i y |
‘
« ‘ 4
; a
a j ,
- z
vet 1
.
~ 1
’
i ‘1 ’
.
4 J
; ‘ 2
i
i 5
_ Ne ’ j ru
' ¥_
4 “
'
.
i i
‘ .
Pa 7 - a -
* as . . + ’ f ee ( i a
p 7 ; ‘
j “ie é

8 etal qa ri)
: ‘ ye :
. + - ales bel Vila uk ati » > Mgab! yal raids 18 ae m

b. Lives eas Su) rh a a a,
Z 3 : > y es » >t Jig a jj

at ee ait
, ww Fd :

' ¢ Work 4 :
’ a A * bt Cea 4 ¢ *

Bras hi all,’ as AY oe
|

**) . ; . +

“4 ' , * - ~ =et@e
: . : -
a A Seo
.
Ls : : ary s © ee
; 2 ares are aes,’
im rit + ed y
, ~- : . > a ae Je Las (
* = 7 : = ‘a . oe
a4 rs ie). “Sgt ‘we ats 5 wat ee
} pYY. J a
lay - Se ate ee AS oe ah hee ee ae
7 -

PRE-OCULAR AND POST-OCULAR TENTACLES OF NAUTILUS. 197

The Pre-ocular and Post-ocular Tentacles and
Osphradia of Nautilus.

By
Arthur Willey, D.Sc.

With Plate 10.

Tue occurrence of a special tentacle in front of the eye and
another behind the eye in Nautilus is well known. These
tentacles resemble the large number of remaining tentacular
appendages in being ringed, and also in being retractile within
sheaths, but differ from them in almost every other respect.
In the first place, most of the tentacular appendages of
Nautilus have essentially an adhesive function, to which is
related a prehensile function. They are employed for seizing
hold of food and for attachment to surfaces.

Attachment is effected by the definite suctorial ridges on
their lower and inner sides (fig. 3). When attached by its
tentacles, Nautilus holds on with considerable tenacity ;! and
sometimes in forcibly detaching it some of the tentacles
break off, and remain fixed to the surface of attachment. The
shape of a section of the tentacular processes of Nautilus is
that of a spherical triangle, the base of which corresponds to
the suctorial ridges, while the apex and sides are directed

1 In the next number of this ‘Journal’ Dr. Willey’s figure of a pearly
Nautilus crawling on a glass surface by means of its tentacles will be pub-
lished, as also some notes on the ccelom and pericardium of Nautilus. His
discovery of the deposited’ eggs of the Nautilus has been published with
figures of the eggs in the ‘ Proc. Roy. Soc.’

198 ARTHUR WILLEY.

outwards, and are distinguished by a deep brown pigment.
The suctorial ridges have a pale neutral tint. We may, there-
fore, speak of those tentacles which are used for prehension
and attachment as the adhesive tentacles. Their function
speaks for the pedal nature of the tentacular processes of
Nautilus, as does also their innervation; and further, their
function allows them to be compared with the arms of the
Dibranchiata. If the comparison is carried into detail the
suctorial ridges of the former would correspond with the
definite suckers of the latter.

That prehension as well as adhesion is a function of the
Molluscan foot is well illustrated by the method of capture of
a species of Oliva at Lifu. This species can be obtained in
large numbers by employing a line baited with the animal of
a land-shell (Placostylus). The Oliva wraps its foot round the
bait, and so can be lifted out of the water and landed.

Recently arguments have been brought forward by Kerr!
against the supposed pedal nature of the Cephalopod arms in
general and the tentacular processes of Nautilus in particular.
If embryological data are not to be trusted on account of the
large quantity of food-yolk in Cephalopod ova, we are obliged
to consider, among other things, function and innervation.
With regard to the latter, Kerr throws doubt on the generally
accepted identification of the sub-cesophageal ganglionic
masses of Cephalopods. It is well, however, to remember
that in Dentalium the ganglionic centres have the same
topographical relations as in Nautilus, the undoubted pedal
ganglion being placed far in front of the pleural and visceral
ganglia.

Returning to the tentacular appendages of Nautilus, it will
not be surprising to learn that the adhesive tentacles are not
ciliated; but it is necessary to mention this negative fact,
because the pre-ocular and post-ocular tentacles are ciliated.
On the side corresponding to the suctorial ridges of the ad-
hesive tentacles the annulations of the pre- and post-ocular

1 J. Graham Kerr, “ On some Points in the Anatomy of Nautilus pom-
pilius,” ‘Proc. Zool. Soc.,’ 1895.

PRE-OOULAR AND POST-OCULAR TENTACLES OF NAUTILUS. 199

tentacles form deep grooves, between which the ridges project
as prominent lamelle. The upper and lower surfaces of the
lamell and the bases of the grooves are covered with vibratile
cilia (fig. 2). There can be little doubt that the pre-ocular and
post-ocular tentacles of Nautilus represent tentacular pro-
cesses, homologous with the adhesive tentacles, which have
been modified to serve an accessory olfactory function. We
will therefore speak of them as the olfactory tentacles, in
contrast to the adhesive tentacles. As is well known, there is
a rhinophore in Nautilus, placed directly below the eye, corre-
sponding to the rhinophore or olfactory groove of the Di-
branchiata. In Nautilus there is a small tentacle as well as a
fossa in connection with the rhinophore, but it is not annulated
and not retractile.

The olfactory tentacles (apart from the rhinophore) when
extended stand out from the body nearly at a right angle, the
pre-ocular tentacle being directed slightly forwards, and the
post-ocular tentacle usually tending backwards (fig. 1). The
ciliated olfactory lamelle are directed strictly forwards.

In the living Nautilus the olfactory tentacles otherwise offer
a strong contrast to the adhesive tentacles by their almost
uniform white colour. When examined under the microscope
there is found to be a little brown pigment in the annulations
and at the edges of the lamellae, but when viewed in toto
under water the general colour effect is white.

Moreover the adhesive tentacles can be touched without
necessarily being retracted, but at the slightest contact with a
foreign body the olfactory tentacles are instantly retracted
within their sheaths.

The presence of accessory olfactory tentacles in Nautilus
can, I think, be related to an essential bionomical difference
between the existing Tetrabranchiata and the Dibranchiata.

Nautilus finds its food chiefly by the sense of smell, while
it is a matter of more or less common observation that the
Dibranchiata with their remarkably perfect eyes pursue their
quarry by the sense of sight. This difference, which is to a
certain extent evident from the facts of organisation, is further

200 ARTHUR WILLEY.

emphasised by the different modes adopted by the natives for
trapping these animals.

One of the surest ways of obtaining Nautilus, and, in fact,
the method by which I have obtained most of my specimens
at Lifu, is to bait the fish-basket with the cooked and bruised
exoskeleton of Palinurus or an allied form. The strongly
scented “ potage” so produced is then wrapped up in cocoa-nut
fibre like a small parcel, and placed in the fish-trap overnight.
There is therefore nothing to be seen, but on the other hand
there is somethiug to be smelt, and by this means I have
obtained as many as ten Nautilus at one time.

For taking Octopus the natives of Lifu employ a very dif-
ferent method. A rounded oval piece of stone backed by a
well-fitting piece of the shell of a species of Cyprzea, to which
are added pieces of leaf to simulate legs and tail, is dangled
along the surface of the water at the end of a line. The
natives say that the Octopus mistakes this for a rat, against
which it has a special grudge; but whatever the reason may
be, the fact remains that Octopus attacks this singular non-
scented contrivance, and so is captured.

Tue OsPHRADIA OF NAUTILUS.

In an article published in ‘ Natural Science’ for June, 1895
(vol. vi, pp. 405—414), I suggested that the post-anal papilla
represented a pair of osphradia—namely, the inner osphradia,
in addition to the outer osphradia which were originally de-
scribed by Lankester and Bourne. The nerve to the outer
osphradium on each side is bound up together with the nerves
to the branchiz into a common trunk, the respective nerves
separating out from the trunk towards the base of the branchiz.
The nerve supplying the inner osphradium has a generally in-
dependent course close beside the above-mentioned common
nerve-trunk. I cannot believe that this slight difference in
the behaviour of the osphradial nerves constitutes an obstacle

PRE-OCULAR AND POST-OCULAR TENTACLES OF NAUTILUS. 201

to the identification of the post-anal papilla as a pair of
osphradia, as has been recently suggested.

However, by means of macroscopic sections of fresh material
the presence of vibratile cilia on the sensory epithelium of both
the inner and outer osphradia can be demonstrated, and this I
regard as the final proof of the osphradial character of the
so-called post-anal papilla (figs. 4, 5). The sensory epithelium
of both osphradia is distinguished from the surrounding ecto-
derm both by the presence of the cilia and by the general
absence of goblet-cells.

The olfactory lamelle of the accessory olfactory tentacles
and the sensory epithelium of the osphradia are the only
places where I have observed vibrating cilia in Nautilus
hitherto.

EXPLANATION OF PLATE 10,

Illustrating Dr. Arthur Willey’s paper on “ The Pre-ocular and
Post-ocular Tentacles and Osphradia of Nautilus.”

Fic. 1.—View from above of the “hood” and tentacles of Nautilus
during life.

Fic. 2.—To show the ciliated ridge of the olfactory tentacles.

Fic. 3.—To show the suctorial ridges of the ordinary tentacles.

Fic. 4.—The ciliated structure of the surface of one of the anterior
osphradia.

Fic. 5.—Ditto of the posterior.

Fig. 6.—Outline sketch of the Gastropod Aplustrum to show “laminate
organ ” comparable to the olfactory tentacles of Nautilus.

1 W. Garstang, “The Morphology of the Mollusca,” in ‘Science Pro-
gress,’ vol. v, March, 1896.

vou. 40, PART 1.—NEW SER. )

oe
7 A nee
foetal eit! JAS 1) A ep igre atth st baile
Jatt: Ha | , ay: nile reg fis (hte hate Er ed imjifed hid cen ook a ee
er ‘ ; rim aa) ier corpggt Payne f, Ne Pratt
: a?
a4 FACT teeth ia wasit ef kk a We hl Vit ete

. t = gt an] i
hae! ee ree > Ss Titi mal? fi. eras vEar ae ey eae 8 nS

i ~ 0 a OL TIEN als. i“ eee Ae Td elgoeeine ap & alee

3 by
' 3 5 so A é. ty oe
: agh pt: <b Lead Fi Ass Pie f ati | N : i"
canitedunes sata OLEAIeE aod) Little te
eer rr ei ee) ee
4 ‘ 7 P A ea Tels
Linco aibeeeee, Mite ey Pd 15a re® “
: - : Ay bee gay 1a See
ae P ’ ¥ :4
al babrh , yaa = > 7 i” « = ih is a: ry VIAL
Dt ae vii? BP ‘3 Tue Sipe Res RF Polina Ys ve!
"s Lateral ie weal ie eietiie ! | ls he Sige;
- S ,
‘
* j 5 :
rE Cy i vey ie} F } Ai oF)
- y : -
—— fig «Pen aks chi ps Teal
yD ingiple WP Nie =
| <> . : 2 ; w y
' ; ; er . ;
Pik ae Ss) METAS. wit hie ih VE 7 J al re 2
Pee) Eat Nish bps ahet pls Maersiue hile at lh Bone See
4 i = ; : ve -_
; isaac | Gurbeosul’ 1 a Ret i ite
rian o) oe ie, PD ei ae .. mic: an labia» a =
i i — y
P So 9), Coaeel ca)
; ee
r 7 rN
leads PONY 4 af iriepaeiat re ihe
; ke coal nt =
, ad ee at oe ee
=) H 6 SP" eat OU
von . .
/ = ‘
r :
i,

ON HETEROPLANA. 203

On Heteroplana, a New Genus of Planarians.
By
Arthur Willey, D.Sc.

Since I have been in Lifu I have obtained one specimen
only of a very remarkable flat-worm which must form the
type of a new genus and a new family. The quite unexpected
relations of symmetry presented by this form, which I will
call Heteroplana, speak for the accuracy of the observations,
although I should be more satisfied if I could obtain more
examples of it.

Heteroplana presents externally a sub-symmetrical ap-
pearance, but in its structure it is characterised by what may
be roundly described as the atrophy of the left half of the
body. The left intestinal diverticula are aborted. The
mouth is placed at the middle of the length of the body, but
approximated to the left margin. The cerebral ganglion is
similarly approximated to the left margin. Through the
whole body, and especially prominent in the anterior and
posterior regions, is a close reticulum formed by the anasto-
mosis of fine moss-like tubules which probably constitute the
genital apparatus.

I had the specimen under observation for a comparatively
long time, and made several drawings from it, one of which I
enclose. On account of its remarkable relations of symmetry
I should place this genus in the order Archiplanoidea, esta-
blished by me for the reception of Celoplana and Cteno-
plana, because, although it is very different from the two
latter forms, yet it would appear to be more nearly related to
a biradial type than to a bilateral type like the Planarians.
This seems to follow from a consideration of such a form as

204 ARTHUR WILLEY.

Ctenoplana. I have proved, as I think conclusively, that
the tentacle axis of Ctenoplana corresponds to the longitu-
dinal axis of the flat-worms. But when Ctenoplana
creeps, it does so, not with one of its pinnate tentacles
directed forwards, but with one side foremost.

In Heteroplana as indicated in the accompanying poe
the locomotion is usually conducted in a somewhat one-sided
fashion, and the number of marginal eyes on the
forwardly directed lobe is more than twice that on
the corresponding lobe on the opposed side of the frontal
region. There are no tentacles in Heteroplana.

Apart from the presence or absence of pinnate tentacles I
think it is fairly evident that Heteroplana is almost directly
derivable from a biradial type of the same grade of organisa-
tion as Ctenoplana.

But in Heteroplana the direction of locomotion (creeping)
has been definitely localised, and the side (namely, the right
side) to which the preference has been given has for that very
reason predominated over the other (left) side. In other
words, in Heteroplana there is hypertrophy of the right
side and atrophy of the left.

Heteroplana is perhaps the completest novelty that I
have obtained. I trust that my contributions to a knowledge
of the primary relations of symmetry in animals in respect
of this form and of Ctenoplana (see ‘ Quart. Journ. of Micr.
Sci.,’ vol. xxxix) will prove of value. There can be no doubt
that some form of symmetry is primarily present in all
animals from the Protozoa (ameboid animals excepted) to
the Vertebrata; amorphous groups like the Gastreidz
(Physemariidz) and the Porifera being of a secondary or
derived nature. Radial symmetry of every description seems
to have been preceded by what may be called polar sym-
metry or directive polarity, as seen in the ciliate Infu-
soria, the Dicyemidz, and the larve of Actinians, &c.

Heteroplana Newtoni was found by me on the lower
side of a Madrepora-stock in Sandal Bay, Lifu, on August
12th, 1896. The general colour, especially of the mid-region,

ON HETEROPLANA. 205

is orange-yellow. The expanded anterior region and the less
expanded posterior region are transparent and almost colour-
less, whilst the moss-like, dendritic, anastomosing gonadial
tubules are to be plainly seen.

These transparent extremities of the body (especially the
anterior region) possess great adhesive power. There are
marginal eyes in front and behind, but not at the sides of the
body. The shape of the anterior portion of the body is con-
stant and highly characteristic. The two posterior lobes, one
smaller and one larger, are also constant.

Anterior marginal eyes

Cerebral eyes

~._ Right intestinal
+ adtverticula

Position of mouth _
on. ventral surface ~ ~~

‘

Left rudimentary intestinal
WIVEFHCUIRS ~~ 2 2 N

Biral eyes

eek
il mar

Posterio.

Heteroplana Newtoni, nov. gen. et sp. (The arrow indicates the usual
direction of locomotion.)

vou. 40, part 1.—NEW SER. P

THE ADHESIVE TENTACLES OF NAUTILUS. 207

The Adhesive Tentacles of Nautilus, with some
Notes on its Pericardium and Spermatophores.

By

Arthur Willey, D.Sc.

With Plate 11.

Adhesive Tentacles—In the plate (Pl. 11) accompanying
this note I have given a sketch (fig. 1) taken from a living
nautilus showing the way in which the ordinary tentacles
are applied to a foreign object lying in the water (in this case
a glass vessel) so as to adhere very firmly to it; whilst the
pre-orbital and post-orbital tentacles have no such adhesive
power, and remain erect on either side of the ocular bulb. As
I mentioned in a previous note, the ordinary tentacles some-
times adhere so tightly to a foreign body as to become torn
away from the animal when the foreign body is forcibly re-
moved.

Spermatophore—In Plate 11, fig. 3, is given a figure of
Nautilus pompilius, showing the spermatophore sac in
situ at the dorsal base of the buccal cone. This position
was originally discovered by Van der Hoeven; but it has not
been seen since, or at least not described, and Van der Hoeven’s
figures are a little wanting in clearness.

Pericardium.— The structure of the vena cava of
Nautilus, discovered by Owen, was again very exactly
described and figured by Keferstein in 1865. This was ap-
parently overlooked by Mr. Kerr in his work “On some
Points in the Anatomy of Nautilus pompilius” (‘ Proc.

voL. 40, PART 2,—NEW SER, Q

208 ARTHUR WILLEY.

Zool. Soc.,’? 1895). This important paper, and one by Béla
Haller (‘Beitrige zur Kenntniss der Morphologie von
Nautilus pompilius”) in Semon’s ‘ Forschungsreisen in
Australien, &c.,’ 1895, have appeared since I left England.
In a former paper I dealt with the question of the genital
arteries of Nautilus.

There is another matter concerning Mr. Kerr’s work to
which I wish to refer. Mr. Kerr (op. cit.) says, “The true
ccelom (viscero-pericardial sac, Owen) has received compara-
tively little attention from previous investigators, Grobben and
Lankester being the only authors who devote to it more than
a few passing words.”

Haller (op. cit.), referring to the orifices in the pallio-
visceral ligament leading from the pericardium into the vis-
ceral portion of the celom, says, “doch so viel ich aus der
Literatur ersehe, wurden diese Oeffnungen zwischen den beiden
Célomtheilen weder ihrer Zahl noch ihrer Grésse und Lage
nach, beschrieben.”’

What I wish to point out is that both these authors have
evidently overlooked Huxley’s brief but pregnant memoir
«¢ On some Points in the Anatomy of Nautilus pompilius”
(‘Journ. Linn. Soc.,’ “Zool.,” vol. iii, 1859, pp. 36—44), in which
the delimitations of the colom and the position and features
of the three openings leading from the pericardium into the
visceral portion of the ccelom are described with perfect
accuracy and with marvellous clearness. It was not necessary
for Huxley to have illustrated his article, so lucid is his
account, and as a matter of fact the figures he does give do
not help much. But I might simply transcribe Huxley’s
description as an explanation of the figure (PI. 11, fig. 2) which
I have drawn from a dissection showing the three openings in
question. I have rarely read an account of somewhat compli-
cated anatomical relations so perfectly and immediately intel-
ligible as the one above referred to by Huxley, and it is
undesirable that such work should be ignored.

OCT 11 1897

THE ADHESIVE TENTACLES OF NAUTILUS. 209

EXPLANATION OF PLATE 11,

Illustrating Dr. Willey’s note on “ Adhesive Tentacles,
Spermatophore, and Pericardium of Nautilus.”

Fie. 1.—Nautilus macromphalus attached to the side of a vessel by
its cephalic (pedal) tentacles; not by their extremities, but by their recurved
sides. Lifu, September 9th, 1896.

Fig. 2.—Nautilus pompilius dissected to show the pericardium by
throwing back the ventral integument. June, 1896. p. v. Z. Pallio-visceral
ligament (Huxley). c. 7. Cardiac ligament. o. p. /. Outer or posterior peri-
cardial ligament. 7. p./. Inner or anterior pericardial ligament. o.p.g. Peri-
cardial follicles of outer or posterior renal organ. o. 7. Outer or posterior
renal sac. 7. p. g. Pericardial follicles of the inner or anterior renal organ.
t.7. Inner or anterior renal sac. 7. m. Reflected flaps of median ventral
integument. m. Mantle.

A,B, c. “The three apertures of communication between the two divisions
of the fifth chamber” of Huxley—that is to say, between the pericardial
ccelom and the perivisceral ccelom (the four other chambers treated of by
Huxley being the two pairs of renal sacs).

u. Systemic heart (ventricle).

Fic. 3.—Male Nautilus pompilius, to show spermatophore sac in
position. Blanche Bay, New Britain, May 24th, 1895.

CORRIGENDA.

(To be bound opposite to Plate 11.)

The following corrections should be noted in the lettering of fig. 2 of the
plate (PJ. 11) accompanying Dr. Willey’s note on “ The Adhesive Tentacles
of Nautilus, &c. &c.,” in Part 2 of Vol. 40 of this Journal :

1. For “ovarian artery” read anterior pallial artery.
2. For “siphonic artery” read posterior pallial artery.
3. The bristle is passed through the left external viscero-pericardial orifice.

MODIFICATIONS OF STRUCTURE IN DECAPOD CRUSTAOBA, 211

On some Modifications of Structure subservient
to Respiration in Decapod Crustacea which
burrow in Sand; with some remarks on the
Utility of Specific Characters in the genus
Calappa, and the description of a new species
of Albunea.

By

Walter Garstang, M.A.,
Fellow and Lecturer of Lincoln College, Oxford.

With Plates 12—14.

A coop deal of scepticism has been expressed in recent
years by various writers as to the utility of the more trivial
features which distinguish the genera and species of animals
from one another. I do not think that such scepticism can
excite much surprise if one remembers that the vast majority
of “‘ biologists” are almost exclusively engaged in the study
of comparative anatomy and embryology. The amount of
attention paid to these branches of biology has long been
utterly out of proportion to the scant attention devoted to
the scientific study of the habits of animals and of the func-
tion of the organs and parts composing their bodies. With
isolated and noteworthy exceptions, the only naturalists who
seriously add to our knowledge of the latter subjects are
those who travel in distant countries, and who are thus thrown
into close relations with animals in their native haunts. Yet
all the time there are thousands of forms living on our own
coasts and almost at our very doors of whose detailed habits
and life-conditions we know practically nothing. I venture
to think that the time has come for consideration whether the

212, WALTER GARSTANG.

subject of bionomics (in Professor Lankester’s sense of the
word) should not receive more adequate recognition than it
does at present in the biological curriculum of our universities.
That such recognition would almost immediately produce
effects in a rapid extension of our knowledge is certain; and
the subject is invested with so much intrinsic interest, as well
as with such important bearings on the problems of evolution,
that I believe such recognition would also have the effect of
attracting many students to the pursuit of morphology who
at present avoid it as a region of mere comparative anatomy.

The present paper contains an account of some modifications
of form in certain exotic Crustacea upon which a new light
appears to be thrown by my recent researches upon the habits
of certain less specialised forms which inhabit British seas
(1896, 1897). My thanks are tendered at the outset to Pro-
fessors Lankester and Poulton for the facilities which they
have kindly placed at my disposal during my study of these
and numerous other forms of Decapod Crustacea. I am par-
ticularly indebted to Professor Lankester for the services of
his artist, Mr. Bayzand, from whose beautiful drawings my
figures are copied.

1. Calappa granulata, Linn.

The figure (Plate 12, fig. 1) represents a front view of a
specimen of this crab as seen when lying flat upon a plane
surface.

The genus Calappa is distinguished, among other points,
by the extraordinary size and shape of the chelipeds, which in
flexion are pressed tightly against the inferior surface of the
carapace, and by interlocking with one another form a sort of
buckler, the anterior and upper margins of which exactly
coincide with the anterior and lateral margins of the crab’s
carapace. If the “ hands” (propodites) of the chelipeds were
of the simple form usual among crabs, the anterior part of
the buccal apparatus would be visible even when the chelipeds
were pressed against the carapace, as is the case, for example,
in the species of Atelecyclus or in the species of Matuta

MODIFICATIONS OF STRUOTURE IN DECAPOD ORUSTACEA. 213

figured in this paper (Plate 12, fig. 1). But in Calappa the
buccal apparatus is completely covered by the chelipeds when
flexed, owing to the fact that these appendages are here pro-
vided with a pair of remarkable cockscomb-like crests on the
upper (anterior) margin of the hands (fig. 1, d). The margins
of these crests are serrated, but otherwise coincide with the
outline of the anterior region of the carapace during flexion
of the chelipeds.

What is the meaning of this extraordinary opercular appa-
ratus furnished by the chelipeds in this genus, and why do the
chelipeds fit so nicely to the carapace during flexion ?

I have found no satisfactory explanation from a study of
the literature dealing with the habits of these crabs. Risso
(1816, p. 18) states that crabs of this species ‘‘ inhabit holes in
steep rocks 20 to 30 métres deep. When they are obliged to
abandon their usual retreats they withdraw their feet under
the carapace, draw their chelipeds together, and let themselves
fall like balls to the bottom of the water.” He implies, in
fact, that the arrangement is an adaptation for defence,
similar to the ball-forming habits of the common wood-louse
(Oniscus), or of its marine relative Armadillo.

But it is now known that crabs of this genus do not usually
live upon rocky, but upon sandy shores, and that they possess
markedly fossorial habits (Macgillivray, 1852, p. 102; Hender-
son, 1893; Schmidtlein, 1879, pp. 24,25). I understand that
the sand-burrowing habits of this particular species have been
frequently observed in the tanks of the Naples aquarium.

The only other suggestion which I have met with as to the
function of the chelipeds is contained in Schmidtlein’s inter-
esting observations on specimens in the Naples aquarium
(l.c., pp. 24, 25). Schmidtlein states that the chelipeds serve
“gum EKinwithlen und zum Schutze.” It would appear from
his observations that the crab, if placed in a tank containing a
number of hungry fishes, protects from their thievish attacks
any morsels of food which it may be engaged in devouring, by
hiding them beneath its tightly closing chelipeds. Neither
this use, however, nor the act of burrowing, provides any

214 WALTER GARSTANG.

explanation of the wonderful exactitude with which the che-
lipeds fit the carapace, nor of certain peculiar relations which
they bear to the respiratory channels to be now described.

It is well known that in most, if not in all, of the Oxy-
stome crabs, the exhalant orifice of the respiratory canal is
carried to the tip of the snout by a prolongation of the endo-
podites of the first maxillipeds, either alone as in Calappa
(fig. 1, a, 6), or with the co-operation of the endopodites of the
third maxillipeds, as in Matuta (fig. 2, a, 6).

The inhalant orifice has been shown, on the other hand, by
Milne-Edwards to vary considerably in position in the same
group of crabs. In most forms it is situated at the base of
the chelipeds, between these appendages and the adjacent edge
of the branchiostegite ; but in Dorippe,! as already shown by
Milne-Edwards (1884, i, p. 89), it is situated further forwards,
and has the form of a deep emargination of the edge of the
pterygostomial portion of the branchiostegite. In Ebalia
and the Leucosiide in general, the inhalant apertures occupy
a position which has hitherto been regarded as unique, being
situated beneath the orbits on the outer sides of the exhalant
orifice. They lead into a pair of deep afferent gutters exca-
vated in the external wall of the pterygostomial? plates. The
gutters are converted into closed canals by opercular expan-
sions of the exopodites of the third maxillipeds. In the

1 Examination of a specimen of this crab has convinced me that in this
form also the chelipeds furnish an operculum for the peculiar afferent orifice,

but in a way quite different from that described below for Calappa and
Matuta.

2 Not in the prelabial plate, as stated by Milne-Edwards (1839, p. 135)
and Dana (1852, p. 62). The language used by the former naturalist is
somewhat vague, but Dana at any rate has completely misinterpreted the
relations of the afferent gutter to the carapace in this family. I may here
draw attention to the fact that in the crab Myctiris platycheles of
Australian seas, the edge of the pterygostomial plate also exhibits a deep
gutter, which extends from the infra-orbital region to the afferent aperture at
the base of the cheliped, and thus closely resembles the afferent canal of the
Leucosiide. The gutter is clearly subservient to the respiratory function,
and interesting results will undoubtedly reward the first careful investigator
of the respiratory processes in this aberrant representative of the Catometopa.

MODIFICATIONS OF STRUCTURE IN DECAPOD ORUSTACEA. 215

remaining Oxystomata the exhalant orifices have been re-
garded as situated at the base of the chelipeds, i. e. as occupy-
ing the normal position.

It is perfectly true that the proximal aperture by which
water enters the branchial chamber does occupy this position;
but an examination of the crabs Calappa and Matuta has
revealed to me that before the water enters the branchial
chamber by this aperture, it has in all probability previously
traversed certain accessory channels formed by the apposition
of the chelipeds to the inferior surface of the carapace.

In the case of Calappa granulata it may be observed
that on each side of the median exhalant orifices there exists
a well-marked slit-like aperture between the infra-orbital
margin of the carapace, and the serrated crests of the apposed
chelipeds (fig. 1, e). If one of the chelipeds be now with-
drawn from its flexed position, it may be noticed that this
slit-hke aperture leads downwards and backwards as a well-
marked channel to the inhalant aperture at the base of the
cheliped. The channel is bounded internally by the ridge-
like outer edge of the third maxilliped ; its other boundaries
are furnished by the approximated surfaces of the inner face
of the chelipeds, and the pterygostomial region of the cara-
pace. It is in fact an accessory channel formed by the
cheliped ; and the inhalant aperture is carried by its means to
the tip of the snout in precisely the same way that in the
Leucosiide the corresponding aperture is carried forward by
means of the exopodite of the third maxilliped.

I propose to call this accessory water-channel the “ exo-
stegal canal.” In its situation on the external side of the
pterygostomial portion of the branchiostegite it contrasts with
the more primitive branchial canals, which are endostegal in
position. It differs from the characteristic afferent canal of
the Leucosiide in requiring the participation of the chelipeds
to complete its external wall, whereas in the Leucosiidz the
third maxillipeds are alone concerned in bringing about the
same result.

Attention may now be directed again to fig. 1, which shows

916 WALTER GARSTANG.

that in Calappa granulata of the Mediterranean, the antero-
lateral margins of the carapace are free from denticulations,
while the propodial crests of the chelipeds are conspicuously
serrated.

On the other hand, it will be remembered that the presence
of teeth along the antero-lateral margins of the carapace is a
conspicuous feature of a great number of the less specialised
types of crab (Cyclometopa), as, for example, in Carcinus
and the Portunide in general.

Why are these marginal teeth so commonly found among
other crabs, while they are absent in Calappa granulata?

I have recently determined (1897) by experiments upon forms
such as Atelecyclus and Bathynectes, which possess well-
developed marginal teeth, and which adopt sand-burrowing
habits, that when these crabs are partially or wholly buried
in sand, the chelipeds are approximated to the branchial
regions of the carapace, as in Calappa and Matuta, but in
such a manner that the marginal teeth of the antero-lateral
regions of the carapace exactly overhang the elongated slit-
like orifice between chelipeds and carapace. Moreover, during
life a current of water can be demonstrated incessantly
pouring into this orifice between the marginal teeth of the
carapace, whence it traverses the accessory channel between
chelipeds and carapace, in order to reach the afferent branchial
aperture at the base of the chelipeds.

These types, therefore, possess a pair of functional exostegal
canals, which differ from those which I have described in
Calappa merely in their greater extent and in their less
specialised form. I have, moreover, shown that the marginal
teeth which overhang the orifices of these canals act as a
“coarse sieve or grating”? which prevents the accidental in-
trusion of foreign bodies, such as grains of sand, into the
respiratory canal.

It seems to me to be accordingly probable that the absence
of spines and teeth from the antero-lateral margins of the
carapace in Calappa granulata is functionally correlated
with the restriction of the exostegal orifice in this form to the

MODIFICATIONS OF STRUCTURE IN DECAPOD ORUSTACEA. 217

infra-orbital region of the carapace, where the teeth on the
margin of the propodial crests of the chelipeds have taken
over the sieve-function which in more primitive types is dis-
charged by the marginal spines of the carapace.

A similar argument may be employed to explain the absence
of marginal teeth on the carapace in Ebalia and the Leucosiidz
in general, In these forms, also, there is a very highly
specialised exostegal afferent canal, the aperture of which is
restricted to a very narrow area beneath the orbit of the crab.
Since the canal is completed in this group by the exopodite of
the external maxilliped alone, the respiratory process is in-
dependent of chelipeds and carapace margins alike, and there
is consequently no necessity for sieve-forming teeth on either
of these parts of the body. Whether, however, this inde-
pendence was maintained throughout the whole ancestral
history of the Leucosiide is another matter ; the considerations
advanced in the present paper seem to me to render it probable
that the peculiar respiratory adaptations of these forms have
also been derived from the more generalised type of adaptation
found in the Cyclometopa. In that case the Leucosiide have
lost the spines on the carapace margins pari passu with the
restriction of the area occupied by the exostegal afferent current
of water ;! and the chelipeds have re-acquired their inde-
pendence simultaneously with the expansion of the third
maxillipeds to form an opercular floor to the exostegal gutter.
In Calappa, both the chelipeds and the third maxillipeds are
concerned in forming the walls of the gutter. It is quite con-
ceivable how the maxillipeds could gradually usurp the whole
opercular function to the exclusion of the chelipeds, especi-
ally as the specialisation of the Leucosiide has clearly been
accompanied by a gradual diminution of size, rendering possible

1 It is interesting to note that the larger types of Leucosiid appear to
have acquired anew set of denticulations at the anterior (infra-orbital)
extremity of the narrow afferent gutter. I have observed the presence of
such denticulations in species of Ilia, Iphis, and Philyra. Their function
is probably the same as that of marginal spines in the Portunids, and of the
teeth on the crests of the chelipeds in Calappa.

218 WALTER GARSTANG.

a continuous relative reduction of the water-supply for the
branchiz, and consequently of the area occupied by the afferent
current.

Utility of Specific Characters in the Genus Calappa.

The most recent revision of the genus Calappa, that of
Alcock (1896, pp. 140—148), although confined to the nine
species found on the Indian coasts, shows that the characters
which are employed to discriminate the different species are
principally the following :

1. Proportion of length to breadth of carapace.

2. Extent of postero-lateral clypeiform expansions of
carapace.

3. Serrations of carapace margins :

i. Antero-lateral margins.
ii. Margins of clypeiform expansions.

4, Hairiness of pterygostomial regions.

I make no pretence to be able to explain the possible utility
of the varied combinations of these features which the different
species of Calappa present, when my only material is the
literature upon the genus and some spirit specimens of several
species. Nevertheless I venture to point out certain correlations
which are not without their significance in this connection.

1. On the whole the more elongated species are restricted
to deeper water than the broader species.

2. The clypeiform expansions are largest in the shallow-
water species and smallest in those which inhabit deep water ;
cf. small size of expansions in C. pustulosa (25 fms.), C.
W oodmasoni (34 fms.), and C. exanthematosa (100 fms.).

This correlation is confirmed by the fact that the species of
the allied genus Mursia, “ which is practically Calappa
without the wings to the carapace,” are found exclusively in
the deep sea (e.g. M. bicristimana, 150—400 fms.).

3. The denticulations of the antero-lateral margins I have
shown to subserve a sieve-like function in British crabs with an
elongated orifice to the exostegal canal. It is probable that

MODIFICATIONS OF STRUCTURE IN DECAPOD ORUSTACEA, 219

the ancestors of Calappa had a similarly elongated orifice to
the exostegal canal, and that the serrations of the antero-
lateral margins in the modern species of Calappa are the last
relics of the marginal spines which covered the afferent
orifices in their ancestors. In C. granulata they have com-
pletely disappeared. In support of this contention I may
point out that the antero-lateral border of the carapace, which
in C. gallus is merely “ crenulate” in the adult, is “ sharply
serrate ” in the young (Alcock, 1896, p. 147).

The denticulations of the clypeiform expansions are also
known to present a similar process of alteration from youth to
maturity (Alcock, ]. c.; Henderson, 1893). No adequate ex-
planation, however, of the function of the clypeiform expan-
sions has been yet put forward. In view of Henderson’s re-
marks concerning the prevalence of protective tints among the
arenicolous crabs of the Madras shores (1. c., 1898), the further
suggestion may be hazarded that the flattening and expansion
of the carapace in the shallow-water species of Calappa may
possibly indicate a process of selective assimilation towards the
appearance of empty bivalve shells. I am inclined, however,
to think that the explanation when found will probably in-
volve something more than mere protective resemblance.

4, Upon the hairiness of the pterygostomial regions I can,
1 think, throw some positive light. Just as the marginal
Spines serve as a sieve for fragments of sand and shell, so the
pterygostomial hairs serve as a sieve for the finer particles of
mud. In Calappa granulata the pterygostomial region
forms a triangular area, bounded on all three sides by a dense
row of fine hairs—a submarginal series beneath the antero-
lateral margin of the carapace, an internal series along the
exopodite of the third maxilliped, and a posterior transverse
series along the interior edge of the meropodite (arm) of the
cheliped. The submarginal series is supplemented by a
carpet of fine hairs on the outer side of the branchiostegite,
the afferent channel being alone free. In my specimens these
hairs are full of mud, indicating that they have efficiently
discharged their sieve function during life,

220 WALTER GARSTANG.

The same remarks apply to my specimens of C. hepatica
from Honolulu and the Sulu Sea. On the other hand, Alcock
states that the pterygostomial region of several species (C.
exanthematosa, gallus, pustulosa, and Woodmasoni)
is characterised by possessing but ‘‘ few scanty hairs.” When
the study of habits is considered worthy of the attention of
men of science, we shall perhaps learn whether or not it is
true, as I venture to believe, that the species with few ptery-
gostomial hairs live in cleaner ground than those having the
outer part of the pterygostomial regions “ densely hairy.”
That Calappz do inhabit mud, as well as sand, is certain
from Macgillivray’s remark (1852, vol. i, p. 102) that on a
coral reef off the coast of Queensland containing all varieties

of coral, mud and sand, “‘ smooth Calappz seek refuge... .
in the shallow muddy pools . . . . by burrowing beneath the
surface.”

2. Matuta picta,! Miers.

The figure (Plate 13, fig. 2) which I give of this species
has been carefully drawn by Mr. Bayzand from a specimen
(an adult male) brought back by Mr. G. C. Bourne from Diego
Garcia in 1892. It represents the crab lying in a somewhat
inclined position, the anterior part of the body being elevated
so as to display more effeetually the infra-orbital and Tees
gostomial regions of the carapace.

The genus Matuta is closely related to Calappa, but in
the form of the carapace and chelipeds is less specialised than
the latter type. In the broad swimming plates of the hind-
most pair of thoracic legs, in the obsolescent teeth on the
antero-lateral margins of the carapace, and in the great pair of
epibranchial spines, Matuta betrays obvious signs of deriva-
tion from an early progenitor of the Portunid type, such as

1 My specimen is identical with M. picta of Miers (1877) and the
synonymous M. lunaris of Leach (1817). It would be referable to the
more comprehensive M. Banksii of Alcock’s recent revision (1896), were it
not for the fact that the “ posterior granulated ridge” is prolonged as far as

the epibranchial spine, on the posterior border of which it dies away. Alcock’s
distinctions under this head are far from satisfactory,

MODIFICATIONS OF STRUCTURE IN DECAPOD ORUSTACEA. 221

Lupa or Callinectes. All the species of this genus are
remarkably specialised for burrowing in sand, as indicated by
the great compression of the four hindmost thoracic pairs of
legs, and by the spade-like modification of their terminal
joints (cf. Rumphius, 1705, and Krauss, 1843),

The chelipeds are curved and moulded to fit the sides of the
carapace during flexion, exactly as in the genus Atelecy-
clus. They are destitute of the cockscomb-like crests
which furnish such a characteristic feature of the genus
Calappa. The anterior part of the buccal apparatus is con-
sequently exposed even during flexion of the chelipeds. The
protection of the mouth parts is, however, ensured by the
forward prolongation of the external maxillipeds (fig. 2, ), a
feature which is not to be observed in the genus Calappa,
but which is a marked characteristic of the allied Leucosiide.

If I am right in my interpretations, the exostegal canal in
this crab has a most extraordinary course. The orbits, in
which the eye-stalks are retractile, have the form of a pair of
stony cups. The outer and inferior angle of the orbit is,
however, incomplete, and its cavity is continued downwards
and outwards over the pterygostomial regions by a deep semi-
cylindrical gutter (fig. 2, e). This gutter is converted into a
tube by two dense rows of hairs which arise from the inner
and outer edges of the gutter, and by their interdigitation
beneath the cavity of the gutter furnish it with a complete
hairy floor. So closely are the hairs set to one another, and
so intimately do they interlock, that upon a cursory examina-
tion the hairy nature of the floor of the orbital gutter is not
at first suspected, and Hilgendorf actually figures the gutter
(1869, fig. 2) as a completely tubular passage excavated in
the substance of the thick calcareous wall of the carapace.

The orbital gutter, as soon as it loses its tubular appear-
ance, turns obliquely outwards and backwards (fig. 2, f) until
it is lost in a thick carpet of hairs, which is especially well de-
veloped in front of and to the outer side of the afferent aperture
at the base of the chelipeds, but which is also continued back-
wards along the whole inferior edge of the branchiostegite,

Pad WALTER GARSTANG.

The chelipeds are smooth and concave on their inner face,
and are capable of being closely apposed to the inferior wall
of the carapace, so that they cover the afferent aperture and
the whole carpet of hairs in this region, as is well shown in
the figure here provided. Their upper margins do not coin-
cide with the edge of the carapace, as they do in Calappa,
but during flexion appear to come as far forwards as the
posterior aperture of the orbital gutter.

Water clearly seems to enter the orbits, travelling back-
wards through the orbital gutter into the carpet of hairs
(which, when the chelipeds are flexed, must furnish a most
efficient sieve for the finer particles of mud and sand), through
which it no doubt eventually makes its way to the afferent
aperture at the base of the chelipeds.

This aperture is also furnished with a special hair-sieve of
its own, since the edge of the branchiostegite which forms its
anterior wall is fringed with a special line of stiff hairs, and
there is a corresponding series of hairs on the opposing sur-
face of the basal joint of the cheliped. When the cheliped is
apposed to the carapace, the two sets of hairs interdigitate and
constitute a sieve completely covering the aperture. A similar
arrangement exists in Calappa also, but in the latter form
the basal portion of the epipodite of the third maxilliped
furnishes a much more obvious operculum to the aperture
than is the case in Matuta.

The remarkable course of the exostegal canal in Matuta,
with the restriction of its principal aperture to the cavity of
the orbit, appears explicable to me only on the view which I
have set forth in the case of Calappa and the Leucosiide,
viz. that the common ancestor of all three types possessed a
continuous waterway along the whole extent of the antero-
lateral margins of the carapace to the base of the chelipeds ;
that the antero-lateral margins were denticulated; and that a
process of restriction of the inhalant orifice began, by which
the closure of the whole inhalant gap between chelipeds and
carapace became gradually effected, except in the infra-orbital
region, This process of restriction was effected as a con-

MODIFICATIONS OF STRUCTURE IN DECAPOD CRUSTACEA, 223

tinuous process of adaptation to a sand-burrowing existence
The marginal denticulations becoming useless, gradually lost
their sharp and prominent form, until they assumed the blunt,
irregular, variable and obsolescent character which they ex-
hibit in the modern species of Matuta. The great epibran-
chial spine is in itself evidence of the validity of this view, for
it clearly represents the posterior spine of the antero-lateral
series in such genera as Bathynectes, Callinectes, and
Lupa. The same spine is again met with in the allied but
less specialised genus Mursia; and a comparison of M.
armata (De Haan, 1833, pl. xix, fig. 2) with M. cristata
(Milne-Edwards, ‘ Régne Animal,’ pl. xiii, fig. 1) confirms the
views I have put forward. In M. armata the epibranchial
spine is longer, while the antero-lateral teeth are absent; in
M. cristata both are present, but the epibranchial spine is
less elaborately specialised and still forms part of the marginal
series.

The reason for the great elaboration of this epibranchial
spine in Mursia armata and Matuta is less clear, and can
scarcely be found without special study of the living animal.
In its initial stage, however, as presented in Bathynectes
longipes, I have every reason to believe it functions as
a stay or barrier to the cheliped during apposition to the
carapace, thus mechanically maintaining the arm of the cheliped
in the right position for the closure of the exostegal canal
(1897, p. 400). It seems to discharge this function also in
Matuta picta, but I am doubtful whether this function is
the only one which it discharges in cases where it is so highly
developed.!

I make no suggestions as to the utility of specific characters
in this genus, owing to the fact that the species of Matuta,

1 Krauss (1843) remarks on the frequency of similar spines in arenicolous
animals of various groups, e.g. fishes and molluscs, as well as crabs. One
function may be to protect the crab from the danger of forcible dislodgment
from the sand by wave-currents, as ably maintained by my friend Mr. Hunt in
the case of the spiny species of Cardium (‘ Proc. Linn. Soe.,’ xviii, “ Zool.,”
p. 269).

voL. 40, PART 2.—NEW SER, R

224 WALTER GARSTANG.

and the range of variability in the different species, are as yet
very inadequately determined. I would only remark that the
beaded ridge which bounds the posterior and postero-lateral
borders of the carapace, and which frequently bears a pair of
tubercles in its course (see fig. 2), is clearly the homologue of
the posterior dentated ridges of Calappa and Hepatus, and
is probably degenerate in character. If this is true, the
variations presented by this ridge during the stages of its
disappearance are little likely to furnish the satisfactory
characters for specific discrimination which some systematists
have ascribed to them.

3. Albunea symnista, microps, and scutelloides, n. sp.

The problem of a pure water-supply in the case of sand-bur-
rowing crabs has been solved in certain instances, as I have
elsewhere shown (1896, 1897) in a manner even more original
than that which I have illustrated for Calappa and Matuta.
In the forms to which I refer (Corystes cassivelaunus,
Portumnus nasutus, and Atelecyclus heterodon of the
British coasts) the normal respiratory current of water—the
constancy of whose direction has been anaccepted maxim among
naturalists since Milne-Edwards’ classical elucidation of the
process nearly sixty years ago,—the normal current is actually
reversed in direction, and flows through the branchial chamber
from before backwards, instead of from behind forwards. In
Corystes cassivelaunus I have shown that it enters the
chamber through a long tube formed by the apposition of the
second antennz, whose double rows of hairs interdigitate with
one another in a most effective manner (Plate 14, fig. 3).

I now show that the structure of the first antennz in the
genus Albunea is strikingly similar to that of the second
antennze in Corystes, and that the tube formed by their
apposition has the same relations to the branchial chamber as
in Corystes. The species of Albunea are known to have
sand-burrowing habits of life, so that in all probability a
reversal of the branchial current takes place in this genus as
in Corystes.

MODIFICATIONS OF STRUCTURE IN DECAPOD CRUSTACEA. 2206

Figs. 3a and 30 on Plate 13 illustrate the arrangement
of parts in Corystes cassivelaunus, for a detailed descrip-
tion of which I refer to my paper on that animal (1896). The
only point that I need emphasise here is that the tube is
formed by the outer or second antenne, the first antennze
being situated in the interior of the tube.

Figs. 4a and 46 illustrate the structure of a Madras speci-
men of Albunea symnista, Fabr., which belongs to the
Hippidea. The systematic position of this group of Crustacea
is discussed by Miers (1879). The various naturalists who
have previously examined specimens of this genus have all
failed to recognise the fact that the hairs on the antennules
are arranged along two longitudinal lines, and that they are
directed towards the axial line of the body. The figures which
have been published are all ludicrously conventional in this
respect, and represent the irregularly hairy antenne of a
Palinurus less incorrectly than they do the antennules of an
Albunea (see Milne-Edwards, ‘ Régne Animal,’ pl. 42,
fio; “Crustacés, pl.-21, fig. 9; Miers, 1879, pl: 5;
Henderson, 18938, pl. xxxviii). The converging double rows
of hairs interdigitate naturally to form a tube, as I have
recognised in A. symnista, Fabr. (Pl. 14, fig. 4), Albunea
microps, Miers, and another Albunoid form (Pl. 14, fig. 5)
which I have not been able to identify with any described
species, and which I here name Albunea scutelloides,
Nn. sp.

The antennular tube expands at its base into a prostomial
chamber, as does the antennal tube of Corystes. In the
latter case the floor of this chamber is formed by the third
maxillipeds (fig. 34); but in Albunea it is formed by the
broadly ovate terminal lobes of the endopodites of the first
maxillipeds (fig. 46)—the homologues of the organs which in
Calappa form the opercular floor of the exhalant passages
(fig. 1d). The prostomial chamber communicates on each side
by a wide aperture with the branchial chamber. The channels
of communication are bounded externally and ventrally by
large lamellate expansions of the basal joints of the first and

226 WALTER GARSTANG.

second antenne. The scaphognathite is of unusual size; both
in A. symnista and in A. microps its anterior edge touches
the basal joint of the antennules, while its posterior extremity
is level with the base of the second thoracic leg. It is, in fact,
half as long as the body (excluding the antennules) in the
attitude represented in fig. 4a.

In Corystes the roof of the prostomial chamber is largely
furnished by the projecting frontal area of that animal
(fig. 3a). In Albunea, however, the frontal area is emargi-
nated (fig. 4a), and the roof of the prostomial chamber is
furnished by the eye-peduncles, whose flattened scale-like form,
varying in shape in the different species, is one of the most
characteristic features of the genus (figs. 4a@ and 5).

In Corystes the three stout basal joints of each antenna
are disposed at right angles to one another in the vertical
plane, bringing about a characteristic bend in the basal part
of the antenna (figs. 3a and 34), a feature which is functionally
correlated with the reversal of the branchial current and its
course through the antennal tube.

A precisely similar arrangement is recognisable in the species
of Albunea, but in connection with the antennules instead of
the antenne (figs.4a@and5). In A. symnista (fig. 4a) the
joints are disposed at right angles to one another, as in
Corystes cassivelaunus; but in A. scutelloides (fig. 5)
the distal joint is pressed much further back than in either of
these forms, thereby greatly reducing the angles of inclination.
This difference may be readily seen to be correlated with the
fact that in the latter form the part played by the frontal
region in covering the prostomial chamber is very much less
than in Corystes cassivelaunus or A. symnista. In
Corystes the roof is provided by the prominent frontal area
(fig. 3a); in A. symnista by the apposed plate-like optic
peduncles (fig. 4a), but in A. scutelloides the optic plates
are so small and short that they scarcely project from the
orbital emarginations. The increased backward bend of the
antennules in the latter species compensates for this deficiency.

In the figure of A. scutelloides (fig. 5) the antennules

MODIFICATIONS OF STRUCTURE IN DECAPOD CRUSTACEA. 227

are represented after being pulled forwards to some extent, in
order to show the cavity of the prostomial chamber beneath and
behind them.

Enough has, I think, been said to justify my view that
many of the characters which distinguish the species of
Albunea, both from one another and from their allies, are
correlated with the function of respiration under arenicolous
conditions of life. The verification of this inference must rest
with those who have the opportunity of examining these
animals alive under the proper conditions.

It must in any event, however, remain clear that the great
problems which Darwin left us as his heritage, after so greatly
illuminating them, are not to be solved by the exclusively
morphographical researches which occupy the time and zeal of
the great majority of naturalists to-day. Even in the best of
hands such researches are capable, as I have shown from the
history of the forms discussed in this paper, of obscuring even
the simple facts of structure which they profess to elucidate ;
while the study of the functional relations of parts, side by
side with the anatomical elucidation of the parts themselves,
provides not only the data for generalisations of intrinsic
importance, but assistance of an invaluable character in the
field of morphological criticism.

APPENDIX.

Description oF THE New Species or ALBUNEA (A. SCU-
TELLOIDES, N. SP.) MENTIONED IN THE FOREGOING PaPER,

By Walter Garstang, M.A.
With Plate 14, fig. 5.

This new species of Albunea closely resembles Albunea
microps, Miers, in size, colour (in spirit) and shape. I found
a single male individual among a number (placed at my dis-

228 WALTER GARSTANG.

posal by Professor Lankester) of Albunea microps (both ¢
and ¢) in the Oxford Museum, labelled “Sulu Sea, H.M.S.
Nassau, 1871-2,” and the accompanying description is derived
from an examination of this single specimen.!

Length of carapace, 9 mm. Sculpture on back closely
resembling that in A. microps, but readily distinguishable by
the following points:—The principal M-shaped transverse
line across the middle of the carapace is relatively more con-
spicuous in A. scutelloides, and the remaining transverse
interrupted step-like ridges are relatively much more nu-
merous (quite twice as numerous). They are consequently
more closely set and give the carapace a still rougher appear-
ance than in A. microps. Under a lens the ridges are seen
to have a minutely tuberculate or beaded character, which is
not seen in specimens of A. microps.

Mid-frontal area emarginated, broadly concave, but the
emargination is wider than in A. microps, and therefore
appears less deep; provided with a median tooth, as in A.
microps, and a pair of small admedian teeth, as in the same
species. Antero-lateral margin (from the admedian tooth to
the antero-external angle of the carapace) divided into two
approximately equal halves by a sublateral prominence; each
half presents a slightly concave curvature. The inner and
outer halves of the antero-lateral margin correspond with the
bases of the antennules and antenne respectively, and may
therefore be termed the antennular and antennal curves.
Antero-lateral margin without teeth, but with twelve or thir-
teen minute close-set tubercles distributed along the anten-
nular curve and around the border of the sublateral promi-
nence (thus differing from A. microps).

Antennules long; each provided with two rows of hairs
which interdigitate with those of the other. Antennules pre-
senting a marked double bend at their basal joints.

Antenne provided with an accessory joint (aciculus), as
long as the joint of the flagellum to which it is approximated,

1 The type specimen of this species will be deposited by Professor Lankes-
ter in the British Museum.

MODIFICATIONS OF STRUCTURE IN DECAPOD CRUSTACHA. 229

i.e. of the same relative length as in A. symnista and A.
microps.

Third maxillipeds. The carpal lobe is not produced beyond
half the length of the propodite (in this agreeing with A.
microps).

Optic peduncles scale-like, elliptical, broader than long,
presenting a deep emargination at their antero-internal angles
which lodges the cornea. The peduncles occupy the lateral
compartments of the median emargination of the frontal area,
which may accordingly be termed the orbital emargination.

Telson in the ¢ somewhat like that of A. microps, but
more elongated and slender, the broadest part being a little
nearer the base of the telson, and the sides more distinctly
concave. No transverse rows of hairs on back of telson like
those in A. microps and A. Guérinii; but a double ad-
median longitudinal series, and a group at each of the basal
angles.

This new species approaches in certain features the species
Albunea scutellata, described by Milne-Edwards (18384,
il, p. 204, pl. 21, figs. 9—-13), which, with two other forms
(venusta and myops), has been referred by Stimpson to a
new genus, Lepidops (1858, p. 230; Miers, 1879, p. 281).
These features are—(1) the absence of frontal denticulations ;
(2) the broad, scale-like optic plates. In fact, were it not for
his remark concerning the truncation of the optic plates,
Milne-Edwards’ short description would be perfectly applicable
to the present species. His figures, however, are not appli-
cable. In his figure the carapace breadth is greatest in front ;
in my specimen it is greatest across the middle (as in A.
microps). He figures no accessory joint to the antenna;
and if the truncated plates in front of the carapace represent
the optic plates, as his account implies, the possible identity
of the two forms is altogether precluded. Moreover, there is
no projecting lobe at the base of the sickle-shaped dacty-
lopodite of the third pair of thoracic legs in my specimen,
while such a lobe is clearly indicated in his figure.

Dana’s description of A. scutellata (1852, i, p. 406) is

930 WALTER GARSTANG.

also inapplicable in regard to the form of the frontal margin.
My figure (Pl. 14, fig. 5) does not indicate the median orbital
emargination as deep as is actually the case.

Stimpson’s descriptions (1858, p. 230) of the genera Al-
buna (sic) and Lepidopa (sic) render the identity of the
two forms still more improbable, owing to the characters of
the antennal aciculum and of the carpal lobe of the third
maxillipeds which I have described for my specimen. These
characters are such as also exist in the species of the restricted
genus Albunea, from which Stimpson removes the species
scutellata. On these grounds, therefore, I refer my speci-
men to a new species of Albunea, distinguishable from the
other known species by the character of the carapace-sculpture,
frontal margin, optic plates, and telson.

On the other hand, Milne-Edwards’ figure is clearly bad
{note the absence of any distinction between the abdominal
tergites and their lateral lamellate expansions); and it is
possible that if his (or Desmarest’s) specimens could be re-
examined, some, if not all, the distinctions upon which I have
relied would vanish. If the form of the ocular plates is de-
termined in this genus, as I have rendered probable in the
preceding paper, by their opercular relations to the prostomial
chamber; and if the emarginations of the frontal area are
functionally correlated with the play of the optic peduncles,
antennules, and antenne, as appears to me to be the case after
examination of three species of Albunea; then it is perfectly
clear that Milne-Edwards’ fig. 9 does not correctly represent
his specimen in these respects.

BIBLioGRAPHY.

Atcock, A. (1896).—‘ Materials for a Carcinological Fauna of India ”
No. 2, “The Brachyura oxystoma,” ‘Journ. Asiatic Soc. Bengal,’
Ixv, pt. 1, No. 2, pp. 184—296.

Dana, J. D. (1852).—* Crustacea,” ‘U.S. Exploring Expedition, 1838—
1842,’ vol. xiii.

MODIFICATIONS OF STRUCTURE IN DECAPOD CRUSTACEA. 231

Garstanc, W. (1896).—‘'The Habits and Respiratory Mechanism of
Corystes cassivelaunus,” ‘Journ. Mar. Biol. Ass.,’ iv, No. 3,
pp. 223—232.

Garstane, W. (1896).—‘ On the Function of certain Diagnostic Characters
of Decapod Crustacea,” ‘ Report Brit. Ass.,’? Liverpool Meeting,
pp. 828—830.

GarstaneG, W. (1897).—‘* The Function of Antero-lateral Denticulations of
the Carapace in Sand-burrowing Crabs,” ‘Journ. Mar. Biol. Ass.,’ iv,
No. 4, pp. 896—401.

GarstaneG, W. (1897).—*“‘ The Systematic Features, Habits, and Respiratory
Phenomena of Portumnus nasutus” (Latreille), ‘Journ. Mar.
Biol. Ass.,’ iv, No. 4, pp. 402—407.

Haan, W. bE (1850).—* Crustacea,” Siebold’s ‘ Fauna Japonica.’

Henperson (1893).—‘‘ A Contribution to Indian Carcinology,” ‘ Trans.
Linn. Soe.’ (2), v, ‘ Zool.,” pp. 325—458.

Hiueenporr (1869).—*‘ Crustaceen,”’ Baron C. C. von der Decken’s ‘ Reisen
in Ost-Africa,’ ii, pp. 93, 94, Taf. 3, fig. 2.

Krauss, F. (1843).—‘ Die Siidafricanische Crustaceen,’ Stuttgart, p. 16.

Leacu, W. HE. (1817).—* On the Characters of Matuta, with Descriptions of
the Species (lunaris, Peronii, Lesuerii, Banksii),”’ ‘Zool.
Miscellany,’ ii, pp. 12—14, Tab. 127.

Maceinutvray (1852).—‘ Narrative of the Voyage of H.M.S. “ Rattle-
snake,”’ vol. i, p. 102.

May, J. G. pr (1881).—* Remarks on the Species of Matuta,’ ‘ Notes from
the Leyden Museum,’ ili, pp. 109—120.

Mizrs, EH. J. (1877).—‘‘ Notes upon the Oxystomatous Crustacea,” ‘ Trans.
Linn. Soe. Zool.,’ (2) i, pls. 39, 40.

Miers, W. J. (1879).—* Revision of the Hippidea,” ‘Journ. Linn. Soc.,’ xiv.

Mitne-Epwarps, H. (1834).—‘ Histoire Nat. des Crustacés,’ 3 vols.

Mityz-Epwarps, H. (1839).— Recherches sur le Mécanisme de la Respira-
tion chez les Crustacés,”’ ‘ Ann. Sci. Nat.’ (2), xi, pp. 129—142.

Mitne-Epwarps, H. (1849).—‘ Le Regne Animal, Crustacés.’

Risso (1816).—‘ Hist. Nat. des Crustacés des Environs de Nice.’

Rumpuivus, G. E. (1705).—‘ D’Amboinische Rariteitkamer,’ Amsterdam,
pp. 11, 12.

ScumMIpTLeIN, R. (1879).—‘ Beobachtungen tiber die Lebensweise einiger
Seethiere innerhalb der Aquarien der zoologischen Station,’ ‘ Mitth.
Zool. Stat. Neapel,’ i.

Stimpson, W. (1858).—“ Prodromus descriptionis animalium evertebratorum,
&e.,” pars vii, pp. 225—252, ‘ Proc. Acad. Nat. Sci.,’ Philadelphia.

202 WALTER GARSTANG.

EXPLANATION OF PLATES 12—14,

Illustrating Mr. W. Garstang’s paper on “‘ Some Modifications
of Structure subservient to Respiration in Decapod Crus-
tacea.”

PLATE 12.

Fie. 1.—Calappa granulata, Fabr., from the Mediterranean. Front
view, showing the closure of the chelipeds beneath the carapace and the
respiratory orifices above them. a. Exhalant orifice. &. Branch of endo-
podite of first maxilliped, forming opercular floor of exhalant canal. c.
Antero-lateral margin of carapace. d. Dentated crest on propodite of
cheliped. ¢. Inhalant orifice of exostegal canal. . Second antenna, forming
part of orbital wall.

PLATE 13.

Fic. 2.—Matuta picta, Miers, from Diego Garcia. View from above,
the anterior part of the crab’s body being elevated to display the buccal
region. a. Exhalant orifice. 6. Third maxilliped. c. Carpet of hairs on
pterygostomial plate. d. Propodite of cheliped. e. Orbital gutter, having a
ventral floor of interlocking hairs. Postero-lateral extension of orbital
gutter on pterygostomial plate.

PLATE 14.

Fie. 3a.—Corystes cassivelaunus from Plymouth. Frontal area,
showing tube formed by second antenne (dorsal view).

Vic. 36.—Corystes cassivelaunus from Plymouth. Ventral view,
showing floor of prostomial chamber.

Fic. 4a.—Albunea symnista, Fabr., 9, from Madras. Dorsal view,
showing tube formed by first antenne, and the ocular plates which form the
roof of the prostomial chamber.

Fic. 46.—Albunea symnista, 2, from Madras. Ventral view, show-
ing basal part of antennular tube, lamellate terminal expansions of first
maxillipeds, and the pediform second and third maxillipeds.

Fie. 5.—Albunea scutelloides, n.sp., ¢, from Sulu Sea (Oxford Mu-
seum). Dorsal view. The antennules are pulled slightly forwards, showing
the prostomial chamber beneath. Also showing the double bend of the basal
joints of antennules and the broad elliptical optic plates, and eye-spots.

N.B.—The figure does not represent the median emargination as deep as
is actually the case. The small admedian teeth, situated on the outer sides
of the optic plates, are also inadvertently omitted.

NOTES ON THE ANATOMY OF STERNASPIS. 233

Notes on the Anatomy of Sternaspis.
By

Edwin 8S. Goodrich, B.A.,
Assistant to the Linacre Professor, Oxford.

With Plates 15 and 16.

In the beautiful works both of Professor Vejdovsky (5) and
of M. Rietsch (8) on Sternaspis we find certain statements
which, if correct, would place that worm in a very exceptional
position. These authors describe the excretory organ as a
lobulated sac with neither internal nor external opening, and
the genital organ as a somewhat similar sac entirely shut off
from the ceelom, but opening to the exterior by two long ducts
with which the sac is directly continuous. Thus Sternaspis,
in having a completely closed excretory organ (nephridium ”),
and in having the ovary or testis situated in a special cavity
without communication with the ceelom, would differ from all
known Polycheta, of which group it is no doubt a highly
modified member.

It was, therefore, with a view to either confirm or correct
these descriptions that I began a study of Sternaspis thalas-
semoides, Otto, during a recent visit to Naples. I may say
at once that they both proved to be erroneous.

Tio these observations have been added some notes on the
cuticle and muscular system.

The Genital Organs.—Fig. 1 represents, somewhat dia-
grammatically, a ventral view of the ovary, or ovisac, as it

234 EDWIN S. GOODRICH.

would be more correct to call it,! removed from a female
Sternaspis, the oviducts having been cut through near their
attachment to the body-wall. These ducts pass forward and
open to the exterior at the end of the processes seen in fig. 16,
in front of segment 8.

When viewed from the dorsal surface the two ducts appear
to be simple tubes, uniting and passing into, or expanding to
form, the lobulated genital sac.

When viewed from the ventral surface, however, we see that
on either side a small blood-vessel, Jat. v., fig. 1, comes off
from the large ventral vessel, v.v., passes along the inner
surface of the wall of the sac for a short distance, and then
emerges on the outer surface of the duct down which it runs
to the body-wall. Now this blood-vessel passes out from the
genital sac by an open canal formed by the folding of a ciliated
membrane, op. The edge of this membrane forms a sort of
ciliated funnel, cil. mb., opening into the ccelom, and is pro-
duced as a ciliated ridge down the outer and ventral side of
the duct for about two thirds of its length.

The exact conformation of the ccelomic opening of the
genital sac will be better understood on looking at the series
of sections represented in figs. 2—8. Fig. 2 is taken through
the narrow region connecting the sac with the ducts; fig. 3
is through the same region, but nearer their point of origin.
Although the blood-vessels lie in the wall of the sac, the ciliated
epithelium never covers them. In fig. 4,asection immediately
after the bifurcation, the beginning of a ridge is visible pro-
jecting into the lumen of each duct. In fig. 5 this ridge is
seen to project far inwards in the upper section, whilst in the
lower section it has cut the lumen completely into two. The
blood-vessel, Jat. v., has passed into the ventral and smaller
lumen. The sections represented in fig. 6 show the blood-
vessel lying near the edge of the ciliated membrane, so as to
close the aperture of the funnel, which is seen to be widely

1 The ova and spermatozoa are developed on the walls of the branches of
the ventral vessel, which enter the sacs near the point of bifurcation of the
genital ducts.

NOTES ON THE ANATOMY OF STERNASPIS. 235

open farther forward in figs. 7 and 8. More forward still, the
membrane, which is for a considerable distance attached to the
tube by one edge, becomes quite free. The membrane and
the wall of the duct are ciliated on one surface only, and this
is continuous from one to the other. This description applies
to both sexes.

As for the structure of the wall of the lobulated sac itself,
previous observers do not seem to have noticed that it also is
ciliated internally,—if not over its entire surface, at all events
along extensive tracts reaching up the lobes. Like the ducts,
it is covered on its outer surface with flat coelomic epithelium,
shown in fig. 10, a drawing of a fragment of the wall treated
with silver chloride and stained. In certain regions, separated
from the outer epithelium by a very thin connective-tissue
layer with a few muscular cells, is the ciliated internal lining,
cil. epith., formed of elongated cells with oval nuclei. The
cilia are short and closely set along narrow longitudinal tracts
directed towards the base of the organ where the ducts come
off (fig. 9). The direction of the ciliary current is from the
tip of the lobes to the base, and down the ducts to the external
openings. The cilia on the membranous funnel, and on the
wall of the narrow canal leading from the cclom into the
genital sac, produce a current running inwards towards the
cavity of the sac. In this way, although the ducts and the
sac may be full of ova or spermatozoa, none are allowed to
stray into the body-cavity.

The Nephridia.—These are two lobed sacs of a yellowish-
brown colour, situated in front of the genital organ (fig. 11).
Their general form and relations have been well described and
figured by Rietsch (3)!. The main mass of each organ lies
closely applied to the osophagus, and is connected with the
body-wall at the level of the intersegmental groove between
segments 6 and 7 (fig. 20) by a rapidly diminishing stalk,
with a narrow lumen, apparently ending blindly; for, like
Vejdovsky and Rietsch, I failed to find any external opening.

1 Jai vainement cherché a leur surface des entonnoirs vibratiles, les
faisant communiquer avec la cavité générale.”—Reitsch (3).

236 EDWIN S. GOODRICH.

After a careful search I found a small ciliated funnel opening
into the celom, and situated on the narrow stalk a little way
above its point of attachment (figs. 11, 12, 13, cil. fun.). The
blood-vessel which accompanies the stalk (figs. 12 and 18,
bi.v.) sends a branch into the lip of the funnel, running im-
mediately below the ciliated epithelium, as in many Poly-
chete nephridia. Fig. 13 is a view of a funnel in a still
living condition, whilst fig. 14 shows the edge of the lip of
another funnel. The cells of which it is composed are vacuo-
lated and slightly granular; bear numerous cilia, and occa-
sionally irregular processes. Flat coelomic epithelium covers
the outside of the whole nephridial organ. The internal
epithelium is entirely composed of large cells, protruding far
into the lumen, and loaded with “ granules ” of peculiar struc-
ture, to be described farther on. In fig. 21 is represented a
fragment of the wall of the nephridium, showing the outline
of the cells with the nuclei situated at their base.

Vejdovsky denies the presence of an internal cavity and of
cilia:— Ein innerer Hohlraum, sowie die Bewimperung
fehlen hier ganzlich.” There can be no doubt that a cavity
of considerable size really exists inside the organ. As to the
presence of cilia, which is also denied by Rietsch, I must
confess that I feel by no means convinced of their absence
even in the main sac. When these soft-walled organs are
placed under the microscope the cells of the internal surface,
which are so full of granules, and bulge inwards into the
reduced lumen, become inevitably pressed against each other,
and being very thin-walled, they have a great tendency to
burst, so that a few scattered cilia would be very difficult to
detect. When teased up the cells break off, and present the
appearance shown in fig. 23 a. On the other hand, the cells
which line the narrow duct leading towards the funnel are less
bulky, and preserve their shape better. In this region I have
seen cilia producing a current from the funnel towards the main
cavity of the organ.

The “ granules” filling the cells of the internal epithelium
of the nephridium are of curiously complex and quite con-

NOTES ON THE ANATOMY OF STERNASPIS. 237

stant structure (figs. 23@ and 6). They consist of an outer
transparent sphere filled with clear liquid, in which some
minute granules are occasionally seen floating. In the centre
of the sphere is situated a highly refringent yellowish body,
which I shall call the concretion. ‘This body is composed of
two halves, one of which is slightly larger and darker than the
other. In each cell spheres may be found of varying sizes,
from a maximum diameter of about 15 « to a mere speck.
But even when quite minute they always contain, as far as I
have been able to see, a central concretion of structure similar
to that described above. When teased out and brought into
some foreign medium the spheres always burst and disappear ;
no fixative, as far as I am aware, will preserve them. The
concretions, on the other hand, are much more resistent.
Treatment with divers reagents reveals the fact that the two
halves have different properties, and, moreover, that they are
cup-shaped, enclosing a third element, which may be called
the central granule. Distilled water, ammonia, alcohol, and
ether have no effect on the concretion. Caustic potash (5 per
cent.) dissolves the small half, and apparently the central
granule, but not the large half, which resists even when
heated. Acetic acid, on the other hand, dissolves the large
half, and the central granule after prolonged action; the small
half remains unaffected. Weak hydrochloric acid dissolves
the large half first, and then the smaller, whilst the central
granule remains. Strong mineral acids destroy the whole
concretion. Neither osmic acid nor iodine stains the concre-
tion to any marked extent.

Fig. 22 shows the smaller half and central granule (the
larger half having been dissolved) of concretions in cells pre-
served with Hermann’s fluid, and stained with alum carmine,
a stain which the concretions readily take up.

Vejdovsky (5) briefly mentions the refringent granules in
the cells of the nephridium of Sternaspis; but Rietsch curiously
enough seems to have mistaken them for nuclei.!

1 “Ta couche épithéliale interne se compose de cellules trés inégales (fig.
52): les unes, volumineuses, presentent de nombreux noyaux dont la plupart

238 EDWIN S. GOODRICH.

It will be seen from the description here given that these are
no ordinary excretory granules.!. On comparing them with
the granules found in the nephridia of other Polychetes, we
find in some forms, such as Trophonia and Pectinaria, trans-
parent spheres containing concretions ; but the latter are often
numerous, and are composed of irregularly aggregated smaller
concretions without constant shape.

Of course, there is no direct evidence that these granules in
Sternaspis are of an excretory nature; they certainly do not
appear to be got rid of by the animal, since the nephridium is
not known to open to the exterior. It is, therefore, not im-
possible that they may be stored up in the organ to serve some
further purpose in the economy of the worm.

In the foregoing account the anterior paired brown sacs
have, for convenience’ sake, been called nephridia. It is quite
possible, however, that they are not true nephridia, but peri-
toneal funnels peculiarly modified, and retaining but a small
ciliated opening into the general body-cavity. The question
of their exact homology can only be answered by the help of
embryological data which we do not yet possess.

The same may be said concerning the homology of the
genital organs. Yet in this case the anatomical evidence
seems to point more clearly to the ducts and lobed sac being
formed by the modification of a pair of ciliated peritoneal
funnels. Cases of the overgrowth of the gonad by the funnel
of the genital duct (or a fold of the peritoneum), so as to
almost or entirely close it off from the general coelomic cavity,
are by no means rare amongst the Celomata. The male
organs of Lumbricus offer a familiar example in which the
testis and genital funnel become enclosed in a sac surrounded
by, yet separate from, the general celom. The peritoneal

sont en voie de division. . . . A premiere vue on pourrait prendre pour des
vésicules adipeuses les nombreux noyaux trés réfringents de ces cellules;
mais ils ne se colorent nullement par lacide osmique, et il est facile aussi de
constater que la plupart d’entre eux sont en voie de segmentation” (8).

+ The “‘chloragogen” granules in the cells of the intestine do not resemble
them.

NOTES ON THE ANATOMY OF STERNASPIS. 239

sacs of the Eudrilide described by Beddard (1) are an in-
stance of the same thing occurring in connection with the
female organs. Throughout the higher Vertebrata the male
genital cells are shed directly into the ducts ; whilst in certain
cases—as, for instance, many Teleostean fish and the mouse—
the ovary is likewise shut off from the body-cavity, being
enclosed in a sac in direct continuity with the duct to the
exterior. It is, indeed, amongst the Mammalia that we find
perhaps the closest parallel with the state of things described
above in Sternaspis. In the female racoon (Procyon) or
badger (Meles) the ovary is overgrown by a fold of the peri-
toneum (“ broad ligament’’) and the funnel of the Fallopian
tube, leaving only a narrow aperture communicating with the
general ceelom (cf. A. Robinson, 4). The steps intermediate
between the ordinary condition in Polychzetes, in which the
gonad is situated close to the wide-mouthed peritoneal funnel,
and the condition in Sternaspis, in which the gonad is enclosed
by the funnel, are easy to conceive.

The Cuticle, Ventral Shield, and Chete.— The
cuticle of Sternaspis has been well described by Vejdovsky
and Rietsch. It is very thick, and consists mainly of inter-
crossing fibres. On the outer surface, however, is a thin layer
of somewhat different nature, to which are fixed numerous
sandy (chiefly siliceous) particles (fig. 24). In this respect
Sternaspis resembles the Chlorhzmids, in which the cuticle is
also covered with sand. These foreign particles are chiefly
grouped round the base of the numberless little papillee which
cover the surface of Sternaspis, and are especially numerous
and coarse behind the seventh segment. On the anterior re-
tractile segments the particles are finer and fewer in number.

Both inner and outer layers of the cuticle are insoluble in
alcohol and ether. The thin outer layer is insoluble in
hydrochloric acid, but soluble in strong solution of caustic
potash. On the other hand, the thick inner layer is soluble
not only in potash and hydrochloric acid, but also in boiling
distilled water. It therefore resembles the cuticle of the
earthworm (2).

vot. 40, PART 2.—NEW SER, 8

240 EDWIN S. GOODRICH.

The thick brown ventral shield is insoluble in boiling water,
caustic potash, alcohol, and ether. Placed in cold concen-
trated hydrochloric acid, it gives an orange-yellow solution ; a
transparent colourless portion remains, which dissolves with
difficulty on boiling.

The cheete are insoluble in water, caustic potash, alcohol,
and ether. As in the case of the earthworm (2), an outer
shell and distal cap remain undissolved when they are boiled
in hydrochloric acid.

The shield, then, is probably formed of the same substance
as the cheetee—not of true chitin.

The Muscular System.—Since the musculature of Ster-
naspis has been only very slightly dealt with by previous
observers, a somewhat detailed account of the highly modified
muscular system of this Polychete is here given. It is well
known that, when irritated, Sternaspis can rapidly withdraw
the first seven segments into its hind body ; this introversion
is brought about by a complicated system of muscles, derived
chiefly from the longitudinal layer.

A side view of an expanded specimen is shown in fig. 16,
and of a retracted specimen in fig. 15. In the latter the
whole of the anterior region has been withdrawn to the level
of the genital papille (gen. p.), which remain projecting from
the front edge of the body. The branchiz at the hind end are
not withdrawn, but merely contract into close spiral coils.
Their contraction is, however, quite independent of that of the
body.

Externally we notice the first segment, bearing the mouth
(m.) and the rounded prostomium (prost.), followed by seg-
ments 2, 3, and 4, armed with strong chete. Each row of
cheete is set on a crescentic and slightly elevated area on each
side at the hinder margin of the segment. As has been shown
by previous observers, the bundles of chet in segments 8 to 14
are sunk in the body-wall, and completely hidden from view.
The grooves separating the first seven segments are shallow,
and completely surround the animal; those separating the
posterior segments are deeper (in the expanded worm), and do

NOTES ON THE ANATOMY OF STERNASPIS. 241

not reach right round, but leave a narrow smooth strip on the
mid-dorsal and mid-ventral regions. The posterior bundles
of cheetee are set round the lateral and posterior edges of the
ventral shield; the lateral bundles are held by small conical
parapodia.

The anterior end of the body is extruded (pleurecbolically)
by the contraction of the circular layer of muscles in the hind
body. These muscles are disposed in a thin layer, interrupted
dorsally and ventrally at the smooth strips already mentioned
(fig. 16, d. a. and v. a.). The inner layer of longitudinal
muscles, passing from one intersegmental groove to another,
lines the body-wall within (figs. 17—19, long. m.). The muscles
which serve to move the large anterior chetz are disposed
as follows (figs. 17—20) :—The chetz are bound together at
their inner ends with connective tissue into three bundles on
each side. A short thick muscular band joins these three
bundles together, and a similar band attaches them to the
posterior end of the pharynx. A larger strand of muscles runs
forwards to the body-wall near the base of the prostomium,
passing along the side of the pharynx (fig. 20). A slender
band of muscle, starting from the inner end of the outer bundle
of chzetz, passes almost horizontally outwards to the body-wall,
where it is attached at the hinder limit of the fifth segment
(Jat. m.). This muscle draws the ends of the chet inwards
and towards the body-wall, whilst the other bundles just
described push them outwards and towards the pharynx.
Each bundle of chet is, moreover, provided with proper
protractor muscles (protr.). The retractor muscles are joined
together in two large bundles (reé7.) (not three, as figured by
Vejdovsky and Rietsch), which pass backwards to their point
of attachment on the anterior edge of the ventral shield on
either side. Not only the chetz, but also the whole anterior
region of the body, are withdrawn by these powerful muscles.

Large strands of longitudinal muscles extend dorsally from
behind the prostomium to the median dorsal region, from the
anterior seven segments to segments 8 to 14 (figs. 17 and 18,
d. retr.). Similar strands extend ventrally from beneath the

242 EDWIN S. GOODRICH.

pharynx to segments 8 to 14, and the front edge of the shield
(v. retr.) on either side of the nerve-cord. A retracted specimen
cut in half vertically (fig. 18) shows these muscles plainly ;
dorsally and ventrally they are attached to the smooth areas
interrupting the intersegmental grooves on the outer surface
of the worm. They are the chief retractor muscles of the
anterior region of the body; the sharp bend in the dorsal
retractors (fig. 18) is due to the pressure of the viscera. This
figure also shows the position of the prostomium and pharynx
in a retracted specimen, and the nerve-cord bent back at a
sharp angle. It will be noticed that owing to the length of
the nerves running to the body-wall from the ventral surface
of the nerve-cord, the latter organ does not closely follow the
curve and folds of the body-wall, and thus is possibly saved
from injury during the rapid process of retraction.

The fourth set of retractor muscles are seen in fig. 17, and
in fig. 19 a retracted Sternaspis cut horizontally above the in-
trovert. These are narrow muscular ribbons running from
the posterior margin of the fifth segment to the grooves
between the 7th, 8th, 9th, 10th, llth, and 12th segments
(lat. retr.).

It is obvious that all these muscles, by co-ordinate and suc-
cessive action, form a very perfect apparatus for the acrembolic
introversion of the first seven segments.

Posteriorly the rectum is provided with paired dorsal re-
tractors (figs. 17 and 18, d. rect. m.), and with paired ventral
retractors (figs. 17 and 19, v. rect. m.), attached to the ventral
shield.

One of the functions of the ventral shield seems to be to
act as a sort of fulcrum for the attachment of the main re-
tractor muscles.

The ten large lateral bundles of chet set round the sides
of the shield are provided with protractor muscles, and with
peculiar slender retractors attached above the nerve-cord
(retr. lat. ch., fig. 19). No such muscles belong to the small
posterior bundles of chetz as figured by Vejdovsky. They
have small retractors attached to the shield.

NOTES ON THE ANATOMY OF STERNASPIS. 243

SUMMARY.

It has been shown in the foregoing account that the cavity
of the genital sac of Sternaspis communicates with the body-
cavity ; that the nephridium is provided with a small ciliated
funnel, possesses a lumen, and in one region, at all events, is
ciliated internally. The complex granules of the nephridial
cells have been described in detail; experiments have been
recorded as to the solubility in certain reagents of the granules,
the cuticle, the ventral shield, and the chete. A detailed
account of the muscular system has also been given.

List or REFERENCES,

1. BepparD, F. E.—‘ Monograph of the Order Oligocheta,’ Oxford, 1895.

2. Goopricu, EK. 8.—* Notes on Oligochetes,” ‘ Quart. Journ. Micr. Sci.,’
vol. xxxix, 1896.

3. Rretscu, M.—“ Etude sur le Sternaspis scutata,” ‘ Ann, Sci. Nat.,’
6e sér., Zool., vol. xiii, 1882.

4. Rosinson, A-—“On the Position and Peritoneal Relations of the
Mammalian Ovary,” ‘Journ. Anat. and Phys.,’ vol. xxi, 1887.

5. Vespovsky, F.— Untersuchungen ther die Anatomie, &c., von Ster-
naspis,” ‘Denkschr. d. Wien. Akad. Math. Naturw. Cl.,’ vol. xliii,
1882.

244, EDWIN S. GOODRICH.

EXPLANATION OF PLATES 15 & 16,

Illustrating Mr. Edwin S. Goodrich’s paper ‘‘ Notes on the
Anatomy of Sternaspis.”

List oF REFERENCE LETTERS.

a. Anus. /.v. Blood-vessel. 47. Branchie. ch. Chete. cil. Cilia.
cil. epith. Ciliated epithelium. cil. fun. Ciliated funnel. cil. mb. Ciliated
membrane. e7/. 77. Ciliated ridge. cal. epith. Colomic epithelium. d@. a.
Dorsal area. d. rect. m. Dorsal rectal muscles. d. re¢v. Dorsal retractors.
extr, Extremity of the nephridium attached to the body-wall. gez.d. Genital
duct. gen. p. Genital papilla. gr. Granule. iz. gr. Intersegmental groove.
lat. retr. Lateral retractors. Jat. m. Lateral muscle. Jat. par. Lateral para-
podium. Jat. v. Lateral vessel. 7. gen. s. Lobes of the genital sac. long. m.
Longitudinal muscles. dw. caz. Lumen of the canal leading from the body-
cavity to the cavity of the genital sac. m. Mouth. z. Nucleus. z.c. Nerve-
cord. wzeph. Nephridium. o. cu¢. Outer layer of cuticle. op. Opening lead-
ing into the genital sac. ov. Ovum. ph. Pharynx. post. ch. Posterior
cheete. prost. Prostomium. profr. protractors. rect. Rectum. retr. Re-
tractors. eér. lat. ch. Retractors of lateral cheetee. sd/. p. Siliceous particles.
sph. Transparent sphere. s¢. Stalk of nephridium. v. a. Ventral area.
v. rect. m. Ventral rectal muscles, v. ve¢r. Ventral retractors. v.v. Ventral
vessel. ww. gen. d. Wall of genital duct. 7. gen. s. Wall of genital sac.

PLATE 15.

Fic, 1.—Somewhat diagrammatic enlarged view of the genital sac and
ducts of Sternaspis dissected out, and viewed from the ventral surface.
showing the openings leading from the body-cavity into the genital sac.
On the right side the blood-vessel which accompanies the duct is not
represented.

Fies. 2—8.—Series of transverse sections through the base of the genital
sac and origin of the ducts, showing the communication with the coclom by
means of the narrow canal formed by the folding of the ciliated membrane.
< 400, cam.

Fie. 9.—Inner surface of a piece of the ciliated epithelium lining of the
genital sac, drawn from the fresh and stained tissue. Xx about 400.

Fie. 10.—Outer surface of a piece of the wall of the genital sac, stained
with silver nitrate and carmine. The ceelomic epithelium has been partially
torn off, exposing the inner ciliated epithelium. Xx about 400.

Fic. 11.—General enlarged dorsal view of the nephridia, showing the
slender stalk inserted into the body-wall and the small ciliated funnel,

NOTES ON THE ANATOMY OF STERNASPIS. 245

Fic. 12.—LExtremity of the stalk of a nephridium, with its accompanying
blood-vessels, and the ciliated funnel. From a stained preparation. x 400,
cam.

Fie. 13.—Enlarged side view of a ciliated nephridial funnel, from the fresh.

Fic. 14.—Enlarged view of the edge of a nephridial funnel, from the fresh.

PLATE 16.

Fic. 15.—Retracted Sternaspis, enlarged side view of a living specimen.

Fic. 16.—Expanded Sternaspis; enlarged left side view from living and
preserved specimens.

Fic. 17.—Enlarged inner view of the right half of a hardened expanded
specimen. The viscera have heen removed, and the branchie are not repre-
sented.

Fic. 18.—Similar view of a retracted specimen.

Fie. 19.—Enlarged inner view of the ventral portion of a retracted speci-
men, cut through horizontally.

Fic, 20.—Similar view of the anterior end of an expanded specimen.

Fic, 21.—Wall of the nephridium in optical section, showing the base of
the granule-bearing cells with their nuclei. From a stained preparation.

Fie. 22.—Front and side view of nephridial granules, from a preparation
preserved in Hermann’s fluid and stained with alum carmine. Cam. x 400.

Fic. 23.—a. A portion of a nephridial cell teased out, showing the con-
tained granules. From the fresh. 4. Isolated granules, more enlarged.

Fie, 24.—Portion of the outer layer of the cuticle, with the foreign
particles fixed on to it. Cam. x 400.

ft ie P yuaeey aa sah ¢

; -

Gi MAT TS CTA, AO OP Meee

4's of
ree = sPAtawe
my iAE as Ae nL ig fil f i A i ine aL. hy ar sva¢

4 =
mn ae! ar.” creel CAL TPL Pe ae Sn ae
Mt A. oie. Lan ly fast Oe ee eee b

eile iv ae :
(i
; ne gee ad ] Pte rd . ; . Y nde s a hy hs ity
Oi “Ale 1 oo a ; i é 3 ite) en

, ; Ay ru '
ba ae LL re THE iL I j ¢
= {
yin be v7 “2-4 j re
4
im , : x ‘ ee 7
7 «! 4 om - 9 ve *
on ! by ! et pa’ sh ~ | awe aie ae
Ps - ; , rere igh :?
> ell
. a oo -
pip i ant anil ne 76 A aie”
sc® » ed . * 2 ;

" j = ‘ .
ay! oh Sa 7 Me CUT ar" ve =? Vie” Ws

a > ae ? : 7 rf » ew Ai) edie
ae Oy he ‘4 v1 hs

= "
‘ Py) eee pe

Roya, bis a Pry) hs ' i A Ls ye ee .
pit re wm | oe | v 4 Y arpeile A rat i \|
a0) «(A teet ‘ e oy pip aleg hast ie

; A

——
te val on"
f =

- =
’
ee)
,
“«
’
,
=
’
4
+

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM. 247

On the Relation of the Arthropod Head to the
Annelid Prostomium.

By

Edwin S. Goodrich, B.A.,
Assistant to the Linacre Professor, Oxford.

Tue question of the segmentation of the Arthropod head,
and of the homology of the preoral region in Arthropods and
Annelids, has for long excited the interest of naturalists,
giving rise to much discussion, and leading investigators to
the discovery of many important facts. The present paper,
written at the suggestion of Professor H. Ray Lankester, does
not claim to be a contribution to our knowledge of the pro-
blems involved, nor an exhaustive history of the subject; it
aims neither at originality nor completeness, but is merely an
attempt to give a plain account of the questions at issue, and
the advance that has been made towards answering them, for
the benefit of those who have not devoted special attention to
the subject.

If we wish to compare the preoral region of an Arthropod
with that of an Annelid, it is necessary first of all clearly to
understand the relation of the prostomium and the peri-
stomium, or buccal segment, to each other, and to the other
segments of the body of an Annelid worm.

It was Professor Huxley who first introduced the word
prostomium in his ‘Lectures on General Natural History,’
published in 1856 (5). “The body of the Polynoé,” says
Huxley, ‘is composed of a series of twenty-six ‘somites,’
terminated anteriorly by a ‘segment,’ the prestomium!
(Kopf-lappen, Grube), and posteriorly by another, the pygi-

1 'Tne modified form prostomium was introduced by Lankester (8).

248 EDWIN S. GOODRICH.

dium, which may or may not represent single somites.”” The
first somite with parapodia, chete, and acicula he calls the
* peristomium.”

Fie. 1.

Fic. 1.—Ventral view of a young Nereis.

Fig. 2.—Diagrammatic plan of the anterior segments of the Polycheta. The
nervous system is represented in black, the mesoblastic tissues are
dotted, the ccelomic cavities are left white. /. P. Frontal process.
m. Mouth. Pa. Parapodium. Pr. Prostomium. 7. Tentacle. A.
Archicerebrum. The Roman numbers refer to the segments.

The prostomium (Kopflappen, lobe céphalique), then, is a
median anterior process lying above and in front of the
mouth. It may be small and insignificant, as in most Oligo-
cheetes, or it may be large and of great physiological im-
portance, asin many Polychetes. The prostomium may be
long and produced into a moveable process (Stylaria), or
distinctly annulated (Glycera). In the Polycheta it contains
the brain or supra-cesophageal ganglion, and often bears
specialised organs of sense, such as dorsal tentacles and eyes,
and ventral palps. In the Oligocheta, on the other hand, the
brain recedes from the prostomium (except Molosoma) into

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM, 249

the first, second, or third segment; whilst in some forms the
prostomium is quite rudimentary (Diacheta), in others it
extends backwards dorsally (as marked by a groove) to the
hinder limit of the peristomium (Lumbricus).

In the Amphinomids it grows backwards over several seg-
ments.}

The peristomial segment, to which the prostomium is at-
tached, is almost always considerably modified in connection
with the mouth, and may even be sharply marked off from. the
posterior segments. It is often called the cephalic segment
(Kopf segment, segment céphalique, buccal segment, head
segment, &c.). The peristomium and prostomium together
constitute the head in Oligochetes and some Polychetes.
In many Polychetes, however, several anterior segments may
become so modified as to contribute to the formation of the
head.

Having thus briefly reviewed the structure and relations of
the prostomium in the Chztopoda, we must proceed to a more
accurate study of its homology.

The prostomium can be one of three things: (1) a modified
or reduced segment; (2) an incipient segment, growing on the
anterior surface of the peristomium ; (3) not a segment at all,
but a structure of different and special nature.

Before attempting to prove that the last interpretation is
the true one, it must be clearly established what we mean by
a true segment or metamere, and then how such a metamere
differs radically from the prostomium.

It is comparatively easy to give a serviceable definition of a
typical segment: it is a region more or less distinctly marked
off from the rest of the body by transverse grooves, surround-
ing the alimentary canal, containing a special coelomic cavity
(more or less completely separated off from the celom of
adjoining segments by means of transverse septa), a pair of
nephridia and of peritoneal funnels communicating with the
exterior, a pair of ganglionic enlargements of the ventral

1 An excellent discussion of the structure and morphology of the prosto-
mium and brain of the Polychta has lately been given by M. Racovitza (14).

250 EDWIN S. GOODRICH.

longitudinal nerve-cords, and (in Polychetes and Arthropods)
a pair of appendages. But we know very well that such fully
equipped segments are rarely found in nature. Intersegmental
grooves frequently disappear (head and thorax of Arthropods) ;
there are segments without coelomic cavity (the thoracic seg-
ments of insects, for instance) ; and again, there are segments
with neither peritoneal funnels nor nephridia (most. Arthro-
pod and Chetopod anterior segments). Some metameres
have no ventral nerve-cord (the first two segments of Lum-
bricus, the posterior abdominal segments of many insects) ;
appendages are often absent.

It is clear, then, that the examination of adult structure
will help us little in deciding whether a debatable region re-
presents a true segment or not, though a careful comparison
with allied types may often be of use. Embryology is our
best guide in these cases, and generally furnishes a decisive
answer. We find, as a matter of fact, that the segments,
which lack in the adult those structures most essential, possess
them at some time during early development, and lose them
at a later period. Yet here, again, it must be admitted that
undoubted metameres may have lost even during development
one or more of the structures characteristic of true segments
(for instance, no distinct ccelomic cavity is known to occur
in some of the anterior segments of many Crustacea). We
cannot, therefore, assert that a given region is not a metamere
because it does not possess this or that character. The only
dogmatic statement we are justified in making is, that when a
region exhibits during development a sufficient number of the
essential structures of a typical segment, it may be assumed
to be a true metamere.! What is “ sufficient’ has to be de-
cided in each case. It should be added that one positive fact
outweighs many negative ones ; the known presence of a certain
characteristic of a segment in a certain region of an Arthropod,
for instance, is of far greater importance than its ascertained
absence in numerous other cases. A good example of this

1 Theories as to the origin of metameric segmentation do not concern us
here ; at any rate, I do not propose to include them in this discussion.

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM, 251

kind is offered to us in the case of the loss of the limbs in
snakes. The argument which might be urged—that the an-
cestors of the Ophidia were legless, since no obvious vestiges
of limbs are seen in by far the greater number of snakes at
any time in their development—is entirely disproved by the
few instances, such as the Python and Tortrix, in which such
vestigial hind limbs are known to occur.

Segments may be suppressed, either temporarily in the
young, as in the zocea larva; or, on the contrary, in the adult ,—
as, for instance, the first abdominal segment in Arachnida. On
the other hand, neglecting the special cases of reproduction
by fission, new segments are never intercalated between old
ones, except in the normal process of growth at the tail end
of the animal. This brings us indeed to one of the most
important characters of the segmentation of Annelids and
Arthropods, namely, that new segments, during the develop-
ment of an individual, are invariably added between the last
segment or telson, and the one immediately in front of it.
All apparent exceptions to this rule, often called the law of
Milne-Edwards, seem to be due to retardation in development,
as in the case of the zocea already mentioned.

At the risk of wearying the reader, it has been necessary to
indulge in these commonplace and well-known remarks for the
sake of clearing the ground. We may now return to the
discussion of the morphology of the peristomium in Annelids.

Careful modern researches (Vejdovsky, Wilson, &c.) have
shown that in Oligochetes the peristomium exhibits the essen-
tial characters of a true segment. It develops as a region
surrounding the mouth, in which are formed a pair of meso-
blastic somites which become hollowed out to form the
celom ; a ganglionic thickening is produced ventrally, which
soon fuses with that of the succeeding segment ; a nephridium
(head kidney) is developed. In the Polychzetes, on the other
hand, where the head in the larva is so often enlarged to a
disc-like shape, it is generally more difficult to trace the origin
of the celom in the peristomium, as indeed also in the seg-
ments behind it. In some cases, at all events, it has been

252 EDWIN 8. GOODRICH.

shown that a pair of somites are formed in the peristomium,
become hollowed out, and even give rise to peritoneal funnels
(E. Meyer, 11). Nephridia are almost invariably developed
in this segment. In Polychztes, moreover, a pair of lateral
appendages are often developed, although they generally

WiGanee

Fig. 4.

™m

Pee c.2-

‘fo

phil fd. ]

Fig, 3.—Ventral view of an embryo of Allolobophora putra (after

Vejdovsky).
Fic. 4.—Diagrammatic plan of the anterior segments of the Oligocheta.
Pr.l. Procephalic lobe; other letters as in Figs. 1 and 2.

become highly modified. In fact, it becomes evident, when
we examine the development and the adult structure of the
peristomium in the various groups of the Annelids, that it is
really a metamere strictly comparable to the posterior seg-
ments, even when much modified owing to its position at the
anterior end of the animal.

The prostomium, on the contrary, presents none of the
characters essential to a segment. It never surrounds the
alimentary canal; it never possesses a pair of mesoblastic
somites.! The cavity which it contains is primitively of the

1 Considerable confusion has been introduced into this question, apparently
by the misunderstanding of Kleinenberg’s results (6). The terms ‘cephalic
germinal streak,” “head segment,” “ head cavity,” “cephalic zoonite,” used
by that author all seem to refer to the peristomium (not prostomium).
Describing the development of the first pair of somites, he says, “ ‘lhe

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM, 253

nature of a blood-space, most clearly seen in trochosphere
larvee, where it is much enlarged. Although later in develop-
ment in both Oligocheta and Polycheta, the prostomial cavity
becomes confluent with the ccelom of the peristomium. It is
only ceelomic by virtue of this connection.! No nephridia are
developed in the prostomium, and on its upper surface is
formed the primitive brain or supra-csophageal ganglion. This
brain may develop from the first as a single median structure,
or may originate from several centres connected with the
organs of sense, which subsequently become fused (Kleinen-
berg, Racovitza). We see that not only does the prostomium
in Annelids differ from a metamere in size and shape, but it
never at any time during its development exhibits the cha-
racters of a true segment. It therefore cannot be considered
as a reduced or vestigial metamere. Can it be considered as
an incipient segment? Obviously not, except on the very
strongest evidence, since such a fact as the growth of a new
segment in front of the first metamere would be opposed to
the rule which is known to hold good for Annelid segmentation
(Milne-Edwards’ law). Such evidence is entirely wanting.

The only other opinion that can be held is that the pro-
splanchnic layer of the cephalic ring, which at first covers only the upper side
of the buccal fossa and cesophagus with a thick layer of mesoderm, extends
gradually its lateral parts towards the central [ventral ?] surface, and embraces
the ingestive aperture completely.” And again, ‘‘The anterior end of the
head segment becomes more and more prominent, and is transformed into a
cylindrical process, the upper lip, a kind of proboscis ” = prostomium ?

1 As Vejdovsky says (p. 320, 16), “die Kopfhohle [peristomial ccelom]
sowohl von Rhynchelmis als der Lumbriciden weicht also genetisch nicht von
Leibesholile der nachfolgenden Segmente als. Sie wachst erst nachtraglich
zum sog. Prestomium aus, welches letztere daher nicht als Kopf, sondern
als ein Kopf-fortsatz oder Kopflappen aufzufassen ist. Dafiir sprechen
zuerst die embryologischen Thatsachen bei Lumbriciden, wo der Mund am
vorderen .Korperende terminal nach aussen miindet, und erst nachtraglich
durch den sich verlangernden Kopflappen von der Riickenseite verdeckt wird.
Dies allerdings erst sehr spat, nachdem das Kopfganglion langst angelegt
ist und sich somit nicht im Praestomium bilden kann, wie unlangst von einer
Seite behauptet wurde.” With regard to the latter fact, the brain neverthe-

less arises from the prostomial region, although the prostomium may be
retarded in development.

254 EDWIN S. GOODRICH.

stomium, being neither a reduced nor an incipient segment, is
a special region not of segmental value. Further, we may
take two views of this question: the first, and the one
generally held, is that the prostomium represents the region
lying in front of the mouth of the primitive unsegmented
ancestor; the second is that the prostomium is a new growth
from the first segment, or region surrounding a terminal
mouth in the primitive ancestor.1 According to the first, the
prostomium is a region of great morphological and phylogenic
importance. According to the second, a more recent addition
of relatively little significance. A comparison of the merits
of these two views would land us in the midst of theories into
which there is no need to enter here; it is sufficient for our
present purpose to have shown that the prostomium is not a
true segment.

In the Arthropoda we find a region in front of the mouth
of varying size, bearing as a rule antenne and well-developed
sense-organs, and containing the brain. A certain number
of segments behind bear appendages connected with and
modified in relation to the mouth. These, together with the
preoral region, constitute the “head.” Long ago the evidence
of comparative anatomy and embryology convinced naturalists
that the head of Arthropods, both in front and behind the
mouth, is composed of several metameres, more or less fused
together. It is with the preoral region that we are most
directly concerned in our comparison with the Annelid.

How do true segments come to lie in a preoral position ?
is one of the first questions we have to answer. If the first

1 Such a supposition would lead us, perhaps, to somewhat modify our con-
ception of the peristomium (first segment) as being merely a metamere,
since it would have the property of producing an anterior prostomial out-
growth (and brain). It must be remembered, however, that, in the cases of
reproduction by fission, other and posterior segments possess the same power.
Lankester has assumed that theoretically every segment should develop a
prostomium, and is only as it were withheld from this completion of itself by
a “longitudinal cohesion or integration.” ‘“ In most Annulosa this longitudinal

cohesion counteracts entirely the opposing tendency to produce a head and to
separate” (8).

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM. 255

segment behind the mouth in Arthropods represent the peri-
stomium in Annelids, then those between it and the anterior
extremity must be new metameres—a supposition which,
if true, would contradict the general law of segmentation and
the evidence of ontogeny.! Professor Lankester in 1878
argued that these preoral segments must originally have been
post-oral, and that they have since moved forwards in front of
the mouth—or, in other words, that the mouth has shifted
backwards.2_ This fertile suggestion is supported by the facts
observed in the ontogeny of Arthropods generally ; and even
in the Polycheta there is often a tendency for the primitively
post-oral segments to shift forwards in front of the mouth,
as in many Amphinomids and in Aphrodite. Lankester’s
explanation has been generally adopted, the only difference of
opinion being as to how many of the “ head segments” are
true metameres, and therefore of post-oral origin.

Of the highest importance in connection with this problem
is the study of the structure and development of the brain.
It is well known that the Arthropod brain presents the
appearance, either in the embryonic or adult state, of being
formed of several segments. Here again the suggestion made

1 Segments develop from before backwards. Although the sequence of the
differentiation of the anterior segments of Arthropods may be somewhat
obscured by what is almost certainly secondary modification and retardation
(cheliceral segment in Arachnids), yet we never find the germ bands, after
the first segment has been formed, growing forwards beyond it to give rise to
new segments in front.

2 «The segmentation of the prostomial axis in Arthropoda and some Anne-
lids, which has an appearance of being a zooid segmentation comparable to
that of the metastomial axis, on account of the identity in tie character of
the appendages with those of the metastomial axis, has yet to be explained.
It may be suggested that it is due to a distinct breaking up of this axis like
the posterior one into zooid segments or zoonites; there is much against this
supposition (see ‘Trans. Linn. Soc.,’ 1869, “On Chetogaster and Aolo-
soma”). Much more likely, it seems, is the explanation that the oral aper-
ture shifts position, and that the ophthalmic segment alone in Arthropoda
represents the prostomium, the antennary and antennular segments being
aboriginally metastomial, and only prostomial by later adaptational
shifting of the oral aperture” (9).

voL. 40, PART 2.—NEW SER. T

256 EDWIN S. GOODRICH.

by Lankester in 1881, that the ganglia of primitively post-oral
metameres have shifted forwards, to fuse with the primitive
Annelidan brain, the archicerebrum, to form a syncerebrum of
compound structure, has been amply supported by the facts
of comparative anatomy and embryology.' But this suggestion
must not be pushed too far; not every lobe, not every epidermic
invagination or centre of proliferation in the embryo must be
taken for a metameric ganglionic mass. We know, as already
mentioned, that the unsegmented archicerebrum of Annelids
may be much lobed and differentiated, and may even arise
from several separate centres in the prostomium itself. An ap-
parent “neuromere” can only be accepted as of metameric value
when the interpretation is supported by evidence derived from
other parts, such as the mesoblastic somites and the appendages.
Let us now examine the various groups of Arthropods.

The Peripatoida (figs. 5 and 6).

In Peripatus, the most worm-like Arthropod, we find a head
bearing a pair of antenne and eyes, and two pairs of more
posterior appendages modified in relation to the mouth—the
mandibles and oral papille.

A study of the development has shown that the head of
Peripatus is formed of three segments. All observers are
agreed that the posterior two, to which the oral papille and
mandibles belong, are genuine metameres; but some doubt
exists as to the nature of the preoral segment bearing the eyes
and antenne, many writers having compared the antenne to
the prostomial tentacles of Annelids. Von Kennel and Sedg-

1 “Tn the Chetopoda, the pre-cesophageal ganglion appears always to
remain a pure archicerebrum. But in Crustacea (and possibly also all
other Arthropoda, though there is a case to be considered for Peripatus and
for the Hexapoda and Myriapoda, on the supposition that their antenne are
not the equivalents of Crustacean antenne, but of the processes of the cephalic
lobe of Cheetopoda) the pre-cesophageal ganglion is a syncerebrum, con-
sisting of the archicerebrum and of the ganglion masses appropriate to the
first and second pair of appendages, which were originally post-oral, but have

assumed a preoral position whilst carrying their ganglion masses up to the
archicerebrnm to fuse with it.”—H. Ray Lankester (10).

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM. 257

wick do not express a definite opinion on this point, but they
show most conclusively that in its development the preoral
segment resembles a true metamere. It has paired mesoblastic
somites, developed post-orally as the first of the series which
shift forwards in front of the mouth, As in the posterior
segments, each of these somites becomes hollowed out to form
the ccelom, from the wall of which is developed a rudimentary
“nephridium” (peritoneal funnel). ‘It presents exactly the
same relations as do the nephridia of posterior somites,” says
Mr. Sedgwick (15), and adds, ‘‘ The first somite, therefore,
behaves exactly as do the posterior somites.”

As pointed out by Professors Korschelt and Heider, in their
excellent text-book of embryology (7), it can now hardly be

| Bie. 5.

Cicto
Once,

castles

Ventral view of an embryo of Peripatus capensis (after Balfour).
Pr.l. Procephalic lobe. m. Mouth.

doubted that the antenne of Peripatus were primitively post-
oral, and that the segment to which they belong is therefore

258 EDWIN S. GOODRICH.

homologous with the first segment or peristomium of Annelids.
Heider has further suggested that the two small processes
found in front of the head, near the median line in the embryo,
represent the prostomial tentacles.

Fic. 6.

Diagrammatic plan of the anterior segments of the Peripatoidea. Ant.
Antenna. Md. Mandible. O.P. Oral papilla. P. Protocerebruim.
D. Deutocerebrum. Other letters as in Figs. 1 and 2,

The study of the development of the brain in Peripatus
confirms the conclusions derived from that of the mesoblastic
structures. It is formed by the fusion of two pairs of gan-
glionic masses derived from the first two segments. Whether
the archicerebrum is still to be distinguished, perhaps in the
median anterior region, in connection with the pair of small
processes mentioned above, is a question which remains to be
solved, and requires a renewed investigation.

The first and largest segment of the brain, which supplies
the eyes and autenne, is developed from the ectoderm of the
large procephalic lobes (first metamere). The process is aided
by the formation of a crescentic pit, a fold of the surface on
either side. The second segment of the brain supplying the
mandibles is smaller. The oral papille are innervated from
the ventral nerve-cord.

We conclude, then, that in Peripatus the first segment has
become much enlarged in development, and that the mouth

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM. 2659

has shifted behind it. The first pair of ganglia have attained
a great size, and differentiation in connection with their
anterior position and relation to the sense-organs. ‘The pro-
stomium is insignificant, and the archicerebrum no longer
clearly distinguishable. The antennz are not prostomial ten-
tacles, but outgrowths of the first metamere.

The Myriapoda.

Unfortunately the development of the Myriapods is very
imperfectly known; but, according to the account we have of
Julus (Heathcote, 1), it resembles exactly that of Peripatus
as regards the segmentation of the head. In the adult there
are three pairs of cephalic appendages—the antenne, the
mandibles, and the labial plate (fused maxilie). The last two
pairs belong to undoubted metameres. The first pair, the
preoral antenne, are developed on the large procephalic lobes,
which give rise to the main segment of the brain (cerebral
grooves are formed here also). As in Peripatus, the antennary
segment contains the first pair of mesoblastic somites.

The Hexapoda (figs. 7 and 8).

Four pairs of appendages are borne on the adult insect’s
head. A study of its development shows that, in reality, it
is composed of six regions. Of these the three posterior,
belonging to the labium, maxille, and mandibles, are univer-
sally considered to represent true metameres. The next,
counting from behind forwards, the recently discovered pre-
mandibular segment (Wheeler, 18), although possessing in the
earlier stages a distinct pair of coelomic somites and cavities,
and in some cases rudimentary appendages, becomes reduced,
and disappears in the adult. The next anterior segment,
bearing the antennz, was for long considered to be not only
preoral in position, but prostomial in origin. Here, again,
embryology shows that, like the posterior segments, it has a
special pair of mesoblastic somites, with well-developed
coelomic cavities (as a rule).' Moreover, since in the early

1 «The deutocerebrum [antennary segment] in all the Orthoptera which I

260 EDWIN S. GOODRICH.

stages of development the antennary segment, together with
its appendages, is distinctly post-oral in position, most authors
are now agreed that it is a true metamere of primitively post-
oral origin,

R-2 Wie; 7/5

ley 2 Fie. 8.

Fic. 7.—Ventral view of an embryo of Anurida maritima (after Wheeler).

Fic. 8.—Diagrammatic plan of the anterior segments of the Hexapoda (the
numbering of the segments is doubtful, as explained in the text). And.
Antenne. #. Eye. Zb. Labrum. Md. Mandible. Mz. Maxilla. Pr. 7.
Procephalic lobe. P. Protocerebrum. D. Deutocerebrum. T. Trito-
cerebrum. Other letters as in Figs. 1 and 2.

We now come to the first serious difficulty in the interpre-
tation of the Arthropod head. In front of the antennary seg-
ment, in the embryo Hexapod, extend the large procephalic

have examined is provided with a pair of true mesodermic somites and with a
pair of appendages, the antenns. Hach mesodermic somite sends a hollow
diverticulum into an antenna.”— Wheeler (18).

“Vas Coelomsiickchenpaar des Antennensegments wor bei den von mir
untersuchten Embryonen stets in typischer Weise ausgebildet.”—R. Hey-
mons (3).

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM. 261

lobes, from which region are developed the anterior segment
of the brain, the optic ganglia, and the eyes. Three views
may be maintained with regard to the homology of the proce-
phalic lobes: (1) they represent the prostomium of Annelids ;
(2) they are merely the specialised anterior region of the
antennary segment, due to its secondary subdivision; (3) they
represent a true metamere, and the first.

In answer to the first suggestion, it may be said that they
would be a strangely large rudiment (anlage) for a prosto-
mium ;! that they are differentiated before the posterior seg-
ments; that, although in the Insecta no special ccelomic
cavities have been found as yet to develop in this region, they
are known to occur in the similar procephalic lobes of some
Myriapods and Arachnids; that in their development they
strikingly resemble in general shape, position, and in their
markedly bilobed character the first segment of Peripatus
(antennary) ; and finally, that they give rise to the same and
largest segment of the brain, which includes the optic
centres.

The same arguments may be used to refute the second view,
though perhaps not quite so convincingly. On the other
hand, it must be remembered that embryologists are all agreed
in considering the antennary segment of insects as complete
in itself, and therefore as not including the procephalic lobes.

The third view, that the lobes represent the first metamere,
remains as the most probable. It has recently been held, if I
understand him rightly, by Heymons (8).

Viallanes? has shown, by his very careful researches on the
structure of the adult brain (17), that it consists in insects of
three segments. This conclusion is thoroughly supported by
embryological evidence. The first or protocerebrum, includ-

1 The procephalic lobes are also distinctly paired. The rudiment of an
Annelid prostomium is unpaired (except Rhynchelmis, Vejdovsky).

? Although Viallanes speaks of these segments as belonging to three
** zoonites,” he draws a distinction between the first as preoral, and the second
and third as originally post-oral. Such a view is difficult to reconcile with what
we know of Peripatus.

262 EDWIN S. GOODRICH.

ing the optic centres, corresponds to the first segment in
Peripatus. The second or deutocerebrum, supplying the an-
tenne, corresponds to the mandibular segment; whilst the
third, or tritocerebrum, represents the segment in Peripatus
which supplies the oral papille.

It would appear, then, that in the Hexapoda the prosto-
mium and archicerebrum have not been plainly distinguished ;*
that the large ophthalmic segment represents the primitive
peristomium, or first metamere, which has shifted in front of
the mouth together with the antennary or second metamere.

The Arachnida (figs. 9 and 10).

In the Arachnids the head appears to be formed of two
segments—the anterior represented by the procephalic lobes
in the embryo, and the posterior by the segment bearing the
cheliceree. The syncerebrum is formed by ganglionic masses
from these two regions.

Concerning the metameric nature of the cheliceral segment
there can be no doubt. It is primitively distinctly post-oral
in position, and contains a pair of mesoblastic somites with
coelomic cavities extending into the appendages. These are
innervated by a pair of primitively ventral post-oral ganglia,
which subsequently move forward and dorsally to form the
second and posterior segment of the syncerebrum.

The procephalic lobes, on the other hand, offer almost the
same difficulties of interpretation as in Hexapoda. Asa rule,
they are separated from the cheliceral segment only after the
appearance of several more posterior segments. No distinct
appendages are produced, ‘This large ‘ procephalic” region,

1 Mr. Wheeler says “it is extremely improbable that so highly important a
structure as the Annelid brain should have completely disappeared in the
Arthropods” (18). This no doubt is quite true; yet it must not be taken for
granted that in the remote Annelidan ancestor of the Arthropoda the brain was
as important and fully differentiated an organ as in certain modern Polychetes ;
and anyhow it must be admitted as a fact that some of the functions of such
an archicerebrum have come to be shared, if not usurped, by the ganglia of
posterior metameres.

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM. 2638

which in certain forms occupies a peristomial position in the
embryo, bears the same relation to the brain as the first seg-

Fic. 9.

Fie. 10.

Fic. 9.—Ventral view of an embryo of Agelena labyrinthica (after
Balfour).

Fre, 10.—Diagrammatie plan of the anterior segments of the Arachnida. Ch.
Chelicera. £#. Eye. Pr. /. Procephalic lobe. P. Protocerebrum. D.
Deutocerebrum. Other letters as in Figs. 1 and 2.

ment in Peripatus and insects, giving rise to the large proto-
cerebrum including the optic centres. Mesoblastic somites
are present, containing in scorpions and spiders (according to
Balfour, Metschnikoff, Kowalevsky and Schulgin, Laurie, and
Schimkewitsch) a pair of distinct coelomic cavities. In Limu-
lus and some scorpions (according to Kingsley, Kishinouie,
and Brauer) the ceelom of the procephalic region is formed by
the forward extension of the cavities of the cheliceral segment.
It is unnecessary to repeat the arguments concerning the
homology of the procephalic lobes already used in the case of
the Hexapoda; one may add that the presence of a celomic

264 EDWIN S. GOODRICH.

cavity in the case of the Arachnida somewhat strengthens the
evidence in favour of this region representing the primitive
peristomial metamere.

The Crustacea (figs. 11 and 12).

The Crustacean head is composed of six regions. The last
three are obviously true metameres, post-oral in position, and
innervated from ganglia on the ventral nerve-cord; they bear
the two pairs of maxillz and the mandibles.

The next two regions, as we go forwards, are preoral in
position, carrying the two pairs of antenne, but are now
almost universally considered to be metameres of primitively
post-oral origin which have shifted in front of the mouth.
In the lower Crustacea (Apus, &c.) the second pair of an-

Fie. 11.

Ventral view of an embryo of Astacus fluviatilis (after Reichenbach).
Ant) and Ant2 First and second antenne. Md. Mandibles. Pr. /.
Procephalic lobe.

tenn are still innervated from the cesophageal commissures

(Pelseneer, 13). In the higher forms the brain supplies both

pairs. These two segments assume the preoral position during

development, and their ganglia fuse with those of the most
anterior region to form the deuto- and trito-cerebrum of the
adult brain.

There remains in front the sixth region, which bears the
large paired compound eyes. Ever since Milne-Edwards, in

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM. 265

1834, put forth the view that the stalked eyes of Crustacea
represented a pair of metameric appendages, observers have
been divided into two camps; some supporting this theory
(Huxley, Reichenbach, Nussbaum, &c.), others opposing it.
The latter maintain that the eyes, whether stalked or sessile,
are merely prostomial sense-organs.

In the embryo we find, in this region, two large procephalic
lobes from which are developed the eyes and optic centres and
the anterior segment of the brain, corresponding to the pro-

Gem

Diagrammatic plan of the anterior segments of the Crustacea. Azf.! and
Ant? First and second antenne. #. Hye. FF. P. Frontal process.
Md. Mandible. Mz.' and Mz. First and second maxille. Pv. 7. Pro-
cephalic lobe. P. Protocerebrum. D. Deutocerebrum. T. Tritocere-
brum. Other letters as in Figs. 1 and 2.

tocerebrum of other Arthropods (Viallanes). Unfortunately,
no definite evidence has been obtained with regard to the
metameric nature of this region from the study of the meso-
blast, since no distinct somites or ceelomic cavities have been
traced with certainty in the anterior segments of most Crus-
tacea. It is evident, however, that although the procephalic
lobes may represent the first metamere, the stalked eyes need
not necessarily be its true metameric appendages. Neverthe-

266 EDWIN S. GOODRICH.

less some evidence for this interpretation is afforded by the
cases brought forward by A. Milne-Edwards of a Palinurus
(12), and by Hofer (4) of an Astacus, in which the eye-stalk
on one side was produced into a jointed flagellum ; also by
. some recent experiments of Dr. Herbst, who, having cut off
the eyes of Palzemon, finds that jointed antenna-like append-
ages are regenerated in their stead (2).

The prostomium itself may have to be sought forin the median
anterior region in front of the procephalic lobes. It has been
suggested that the median eye and little frontal processes of
the Nauplius larva represent prostomial sense-organs, and it
is possible that the anterior region of the brain in connection
with these represents the archicerebrum.

In the foregoing pages the view that the procephalic lobes
are homologous throughout the Arthropoda, and represent
the peristomial segment of Annelids, has been consistently
favoured, not because this interpretation can be considered as
firmly established, but from a conviction that the best way of
presenting the problem is to uphold a definite theory. Thus
both the weakness and the strength of the position become
clearer. It is quite plain, however, that much evidence is still
needed bearing especially on the presence of distinct meso-
blastic somites in the procephalic region in several groups, and
on the possibility of distinguishing the true prostomium and
archicerebrum in the Arthropoda.

RELATION OF ARTHROPOD HEAD TO ANNELID PROSTOMIUM. 267

"el[IxeU puodag

“RITXBU 4SALT

BOAIPURY

“tmn1qa1a0
-O}11} ‘euuayue ptuodeg

“wun1qe1990
-ojnep ‘KBuUayUe 4SITT

“wn1qe1e0
-oqord ‘saqoyorpeydaoorg
¢ sassaooad [eyuory

G

“VEOVISOA)

ow “elixeut puooag
oUt “ay TXBUL 4ST g

|
OW “soyqrpuryy |

"UN1Ga1990jII} ‘ase

‘quauises yunap|-puedde Aaeyuawipnyy
“unaq “uanaq
-a1900jnap + ‘B1VdI]aYO]-ate00jnep = ‘wuueyuy,
"wun.1qa1a0 "wn1g9.190

-oyo1d ‘saqoy o1yet}da00rq]-0401d ‘saqoy o1peydoo01g

|

“YCINHOVUY *YLOASNT

“ON
“ONT

‘quawses yun,

‘eijided [eg

“wunaq

-91900jnap “saTqrpueryy

"uniqar90040Id ‘aeuua4
“ue ‘saqoy orjeydaoo1g

"9g yuoUseg
"g quaulsag

‘Pp quousag
"e JuaUseQ
*g yuaUIseg

“CUNTUE
-OSII9q 10 [ Juawsag

¢ sassaoord [equa y
d

*VACIOLVAINA

“TUNIGa.1IOIYOIW
*saToBJU9} JNOYIIM
IO YAM wUNIMO4SOIg

‘VAITANNY

So ea a i ee SS eS See

o*

268 EDWIN S. GOODRICH.

List oF REFERENCES.

. Heatucote, F. G.— The Post-embryonic Development of Julus ter-

restris,” ‘Phil. Trans. Roy. Soc.,’ vol. clxxix, 1888.

. Herzst, C.—‘‘ Ueber die Regeneration von antennenahnlichen Organen,

&e.,”’ * Arch. f. Entw.-Mechan.,’ vol. ii, 1896.

. Heymons, R.—‘ Die Segmentirung des Insecten-Korpers,” ‘ Phys. Abh.

K. Ak. Wiss.,’ Berlin, 1895.

. Horer, B.—‘ Hin Krebs mit einer Extremitat statt eines Stielanges,”

‘Verh. d. deutschen zool. Gesellsch.,’ 1894.

. Huxtey, T. H.— Lectures on General Natural History,” Lecture VI,

‘ Medical Times and Gazette,’ vol. xiii, N. 8., 1856.

. Kuernenserc, N.—“*The Development of the Earthworm,” ‘ Quart.

Journ. Micros. Sci.,’’ vol. xix, 1879.

. Korscuett and Heiper.—“ Lehrbuch d. vergl. Entw. d. Wirbellosen

Thiere, Jena, 1893.

. Lankester, EH. Ray.—“ A Contribution to the Knowledge of the Lower

Annelids,” ‘Trans. Linn. Soc.,’ vol. xxvi, 1870.

. LANKESTER, EH. Ray.—“ On the Primitive Cell Layers,” &c., ‘Ann. and

Mag. Nat. Hist.,’ vol. xi, 4th S., 1873.

. LANKESTER, FE. Ray.—‘‘Appendages and Nervous System of Apus

cancriformis,” ‘Quart. Journ. Micros. Sci.,’ vol. xxi, 1881.

. Meyer, E.—“ Studien tiber den Korferbau der Anneliden,” ‘ Mitth,

Zool. Sta. Neapel,’ vol. viii, 1888.

. Mitnz-Hpwarps.—* Transformation du pédoncule oculaire en une an-

tenne,” ‘Compt. Rendus,’ vol. lix, 1864.

. PELSENEER, P.—‘‘ Nervous System of Apus,” ‘Quart. Journ. Micros.

Sci.,’ vol. xxv, 1885.

. Racovitza, E.—* Le Lobe Céphalique et l’Encéphale des Annélides

Polychetes,” ‘Arch. Zool. Exp.,’ vol. iv, 3rd Ser., 1896.

. SeDewick, A.— The Development of the Cape Species of Peripatus,”

Part III, ‘Quart. Journ. Micros. Sci.,’ vol. xxvii, 1887.

. Vespovsky, F.—‘ Entwickelungsgeschichtliche Untersuchungen,’ Prag,

1888-92.

. ViatLanes, H.—‘‘ Centres nerveux et Organes de Sens des Animaux

Articulés,” ‘Ann. Sci. Nat. Zool.,’ vol. xiv, 7th Ser., 1893.

. WHEELER, W. M.—“<A Contribution to Insect Embryology,” ‘Journ.

of Morph.,’ vol. viii, 1893.

DEVELOPMENT OF THE CALIFORNIAN HAG-FISH. 269

On the Development of the Californian Hag-fish,
Bdellostoma Stouti, Lockington.

By

Bashford Dean, Ph.D.,
Columbia College, N.Y.

(Preliminary Note.)

With Plate 17.

Dr. Howarp Ayers, in a lecture before the Wood’s Holl
Biological Laboratory,' has given a detailed account of the
occurrence of the Pacific Hag-fish, Bdellostoma Dombeyi

= Stouti), in the Bay of Monterey. Its great abundance in
this region gave promise of material for the study of the
development of a Myxinoid; or, at all events, the chances to
obtain these valuable stages seemed far more favorable here
than in the European localities, mainly in the North Sea,
where the eggs of Myxine had long been sought. Among the
zoologists who endeavoured to secure hag-fish embryos in
California Professor G. C. Price, of the Leland Stanford, Junior,
University, was the first to succeed, and he has already pub-
lished two accounts of his studies.” During the past summer
additional developmental stages, including a number of the
earlier ones, were collected by the present writer during a
visit (July, August, September) at the Hopkins Marine
Laboratory.

1 1893, Ginn and Co., Boston.

«Zur Ontogenie eines Myxinoiden,” ‘Sitzungsberichten der mathematisch-
physikalischen der K. Bayer. Akad. d. Wiss.,’ Bd. xxvi, 1896, Heft 1, S.
67—74.

“On some Points in the Development of a Myxinoid,” ‘ Verhandlungen
der Anat. Gesellschaft.’

270 BASHFORD DEAN.

At the request of Mr. J. T. Cunningham, well known for his
interest in the study of the Myxinoids, the following paper
has been prepared, with a view of presenting in outline the
main facts relating to the external development of this re-
markable Chordate type.

The hag-fish finds a natural spawning ground in the Bay of
Monterey. Here the spawning appears to take place over a
wide area, since eggs have been secured at many and distant
points. The embryos in the writer’s collection, however, were
taken from one particular neighbourhood (about a mile off
shore, in twelve fathoms of water, sp. gr. 1:020, temp. 50°—60°
F.), which represented doubtless a favorable spawning ground.
About one fifth of the eggs here collected were found to yield
embryos, while eggs from other regions rarely contained them,
the worthless eggs averaging nearly 98 per cent. In view of
the present method of collecting, success in obtaining embryos
is to a large degree a matter of accident; for the eggs can
very rarely be dredged, probably because they have been
deposited among rock fragments or deep in the mud, as Plate
suggests.! It is upon trawl-line fishing, as Price has noted,
that the collector must finally depend. By this means many
scores and even hundreds of hag-fishes may be caught during
a day’s fishing ; but among these there will only rarely be one
which has entrapped an egg-string in its encasing slime.
Such an instance is shown in the accompanying figure (fig. 1),
from a water-colour sketch by the writer. The hag, reddish
purple in colour, has twisted itself into a knot around the
trawl-line; the slime mass enclosing it is translucent, whitish,
here and there blood-stained ; the eggs are tangled securely
in the meshes of the slime, suggesting a string of yellow beans.
The eggs, as has often been noted, are fastened together at
either end by clumps of wiry filaments, whose tips interlock.
The tips, shown in fig. 2, vary notably in size and form (con-

1 “Ueber die Hier von Bdellostoma Bischoffii, Schneider,” ‘ Sitzungs-
berichten der Gesellschaft naturforschender Freunde zu Berlin,’ Jahrg.
1896, No. 1, 8. 17—21.

The writer’s material was collected almost entirely from rocky bottom.

DEVELOPMENT OF THE CALIFORNIAN HAG-FISH. 271

trast Plate’s fig. 3); they are typically flower-shaped, their
broadly everted petal-like margins enabling them to interlock.

Time AND Mope or OvIPosITION.

Both Ayers and Price have agreed that in the Bay of
Monterey the spawning season of Bdellostoma is an extended
one. These conclusions agree fully with the writer’s observa-
tions. He has obtained embryos of many ages from the same
locality at the same time ; he notes, however, that during the
present season the embryos were most abundant of a particular
size, a fact which suggests that a time of maximum spawning
had occurred. As to actual oviposition, the following note is
suggestive :—A fisherman brought in (August 25th) twenty-
three eggs, which he said had been spawned in the boat
shortly after the mother, which he also brought, had been
taken. Hach egg was encased in a transparent glistening
outer capsule, delicate but elastic. One would be inclined to
believe that the fisherman had observed the normal method of
spawning, i. e. that the eggs were discharged at the same time,
extruded from the abdominal pores the more readily on account
of the sheathing capsule. ‘The rupture of the capsule at or
before fertilisation would set free the fastening processes, and
thus enable the eggs to become attached, either together or to
other eggs in the neighbourhood. This view, as opposed to
that of Ayers, is supported by the fact that in a number of
cases embryos of widely different ages have been found in the
same egg-string.

GeneRAL Monet or DeveELopMent.

The material collected by the present writer includes an
almost complete series of embryos from a stage in which the
blastopore is closing, to one about the time of hatching;
together with these are several possible stages of segmenta-
tion. A few of the most typical of these embryos are shown
in the accompanying figures (figs. 3—8), and afford a con-
venient basis for an outline of the mode of Myxinoid develop-
ment.

vou. 40, PART 2.—NEW SER. U

O72 BASHFORD DEAN.

As Price has shown, and as others have indicated, the
development of a Myxinoid is typically meroblastic ;’ it might
reasonably be surmised that the large amount of yolk-material
would enable the embryo to attain an advanced stage before
the period of hatching, and that the duration of development
in the egg should be a long one. These conjectures are
clearly confirmed by the writer’s observations upon embryos
reared for a time in aquaria; he believes that the eggs do not
hatch within two months, and thinks it possible that several
more months may be taken. By this time the young Bdello-
stoma has attained a length of over 6 cm., and has outwardly
the characters of the adult. Price notes that a distinctly
larval period is absent.

There can be little doubt, as Price has noted, that Bdel-
lostoma is not protandric in the sense demonstrated by
Cunningham in Myxine.? The eggs are in all probability
fertilised after deposition. An egg nucleus is then present
immmediately below the single large micropyle at the opercu-
lated (animal) pole of the egg.

Segmentation. — Until the writer’s material of the
youngest stages shall have been sectioned, he can refer but
doubtfully to a single egg. Of this transverse sections near
the animal pole show on one side a number of nucleus-like
structures embedded in a clear peripheral germinal area.
There are no cell outlines, although the stage is probably
early. One cannot doubt, however, that a condition of asym-
metry with reference to the longitudinal axis of the egg
appears in very early stages. And from the conditions of the
following embryo it will be inferred that the dorsal lip of the
blastopore appears on one side of the egg, not far from the
rim of the opercule, and that from this region it closes fissure-
wise backward.

Very Early Embryo (fig. 3).—At this stage the embryo
is thread-like. The very narrow neural cord, Nn, represents

1 Cunningham, Retzius (1889), ‘ Biol. Foren. Forhdgr.,’ Bd. i, pp. 22 —28.
? Cunningham, J. T., “Reproductive Elements in Myxine,’ ‘Quart.
Journ. Micros. Sci.,’ vol. xxvii, 1886.

DEVELOPMENT OF THE CALIFORNIAN HAG-FISH. 273

the fused lips of the blastopore, whose fissure is still to be
made out behind the tail region, Be. Anteriorly the long
narrow brain mass, B, is indicated, scarcely deeper and wider
than the myelon into which it merges. No canalis centralis
can be determined in surface view. Notochord and meso-
blastic somites have not yet arisen. A mesoblastic thickening
is apparent in surface view only at *.

Early Embryo (fig. 4).—This embryo is scarcely longer
than the earlier one, but it is notably broader, and it has made
marked advances in organogeny. Anteriorly the neural cord
has acquired a lumen; its walls, thickened and folded asym-
metrically, mark out vaguely the divisions of the brain. Au-
ditory vesicles, aU, are prominent ; optic vesicles present but
very indistinct. Mesoblastic somites, ps, about ninety in
number, appear close to the neural cord, and extend sideways
but a very short distance. They are here differentiated in
situ, but surface view cannot demonstrate definitely that gut
cavities are present.

Interesting is the condition of the fore-gut, which now
dilates under the definite head region, and is already, as far as
one can judge from surface view, pierced with several gill-slits.
The heart, H, is vaguely indicated in front of the head, extend-
ing directly forward.

Moderately Early Embryo (fig. 5).—The trunk has
become outlined in this stage; on each side it is separated
from the yolk region by a marginal artery which passes back-
ward as it breaks into branches in the neighbourhood of the
tail. Body length has somewhat increased, due apparently to
growth in both directions, the head now extending under the
operculum, the tail somewhat nearer the opposite pole of the
egg. Advances in the development of the neural cord include
anteriorly the enlargement and definite outlining of the brain
parts which have now grown under the auditory vesicles and
over the eyes, and the now prominent and fused nasal pouches,
and posteriorly the extension of the canalis centralis into the
region of the tail. The asymmetry of the foldings of the
neural tube is now visible only in the region of the medulla.

274 BASHFORD DEAN.

Somites are prominent, ps. Pronephric tubules are apparent
in connection with all the mesoblastic somites, PN, each nar-
rowing out laterally into a delicate contorted tube, but as yet
there has appeared no pronephric duct. In the gill region
about the normal number of gill-slits have now been formed.
The number increases from behind, but remarkably enough
those latest formed (i.e. hindmost) are largest. The gill
region is marked out clearly on either side of the embryo as a
conspicuous lappet-shaped outgrowth.

Late Embryo (figs. 6, 7).—Continued growth has carried
the head of the embryo around the animal pole as far on the
ventral side as past the rim of the operculum; the tail, now
narrow and tapering, has elongated until it has almost reached
the hind end of the egg. Head, neck, and tail are separated
below from the yolk-sac, although they lie in grooves, and are
deeply sunken into it. Again, a marginal artery, pa, separates
the trunk from the yolk-sac. Somites are now clearly marked ;
pronephric structures are prominent, pn, and a segmental
duct, sp, is found. In the head region the principal advances
include the enlarged size of the nasal pouch, n, the appearance
of supporting tissue and of mouth, the latter originating as a
single invagination. ‘This is very definitely shown in one of
the writer’s preparations, who thus differs from the conclusion
of Price as to its probable origin as a paired structure. The
gill-slits at this stage, es, may be seen through the back
when this region is viewed as a transparent object. Since the
last stage these have been drawn forward, and at the same
time inward toward the ventro-median line by the growth of
the head. The heart is at this stage a well-marked tube,
passing straight toward the head from the hinder pole of the
egg; on either side it receives asymmetrically the vessels
from the yolk region, carrying the venous blood to the gills.
The vessels are very conspicuous in the living object, the
crimson threads in brilliant contrast to the rich yellow of the
underlying yolk.

Latest Embryo (fig. 8).—The figure illustrates the latest
stage in the writer’s material. The young Bdellostoma

DEVELOPMENT OF THE CALIFORNIAN HAG-FISH. 275

measures 6 cm. in length, and will (probably) shortly hatch.
It exhibits occasional writhing, and its mode of growth sug-
gests that in the process of hatching the operculum will be
detached by movements of the tail. The colour at this stage
is purplish, save in the region of the yolk-sac. This, as before,
is bright yellow, closely traversed by vitelline blood-vessels.
The yolk-sac is now attached to the embryo only in the middle
portion of its trunk; both the anterior and posterior trunk
regions seen in the figure are separate, therefore, from the yolk-
sac, although pressed tightly against it, and embedded in deep
grooves. The tail lies on its right side, and is growing over
the head toward the opercular end of the egg. The head,
lying somewhat on its left side, has now grown backward till
it has nearly reached the posterior pole of the egg. Struc-
turally this late embryo closely resembles the adult; cirri
are present around the mouth as in the adult condition,
although they cannot be seen in the figure, the mouth being
pressed far backward on the ventral side. The anus is indi-
cated at a, segmental mucous pits at mM. Dermal-fin rays are
well developed. The gills, Gs, are now the characteristic
pouches with infolded walls and with external tubular ducts.
Their final position will be seen to be far backward of the head.
It is evident, accordingly, when figs. 4,5, 6, and 8 are con-
trasted, that the gills have not been growing forward at an
equal rate with the head region. They early arise near the
first slit even in front of the auditory sacs (fig. 4); they are
seen later (fig. 5) in the process of being drawn forward under
the head, but are now well behind the ear capsules; still later
(fig. 6) they are seen in their normal position on the ventral
side of the throat, but further back behind the ear sacs, sepa-
rated from them by a distance of several somites; and finally
(fig. 8) they appear in their adult conditions, still further
caudad, now separated from the ear sacs by ten or more
somites. By the process of unequal growth, therefore, the
change in the position of the gills is explained. As at no

1 As has also been shown to be the case in Amphioxus by Lankester and
Willey. (Hditor.)

276 BASHFORD DEAN.

stage in the writer’s material has he found any traces of either
anterior or posterior gill-slits to increase the normal number,
he is unable to accept Price’s suggestion, based upon the
number of spinal ganglia and the change in the position of the
gills, that the embryos of Bdellostoma have as many as thirty-
five pairs of gill-slits.

CoNCLUSiONS.

The Myxinoid differs widely from all other Chordates in its
developmental type. Its sharply marked differences from the
mode of development of the lamprey emphasise the wide
divergence of these branches of the Marsipobranchian stem ;
and this alone forms a strong ground for belief in the antiquity
of the Cyclostomes, and for rejecting even a most remote
Teleostean ancestry. Developmentally the two branches cer-
tainly differ as widely as sharks and Ganoids. There can be
no doubt, furthermore, that the embryonic conditions of
Bdellostoma are not to be derived directly from those of the
lamprey. ‘To what degree the converse is to be accepted must
remain for further study. We may now reasonably believe that
the ontogeny of a Myxinoid, when fully studied, will enable the
interrelationships of the Marsipobranchs to be broadly out-
lined: and it is possible that many valuable suggestions will
follow as to the general relations of these to Protochordates
on the one hand, and on the other to the ancestral Gnatho-
stome.

Some of the more striking features in the development of
Bdellostoma, as above described, may finally be summarised.

I. The neural tract is laid down, nearly in its entire length,
before the appearance of somites, and without any indication
of neuromeres.

II. The neural tract, as in Petromyzon, acquires a lumen
by dissociation of cells (as shown by sections), proceeding
antero-posteriorly.

III. The brain is distinctly a tubular structure, and differs
little in calibre from the spinal cord up to a relatively late
period in development, i.e. to the time of the appearance of

DEVELOPMENT OF THE CALIFORNIAN HAG-FISH. PAPE

paired sense-organs, of gill-slits, and of nearly the adult
number of somites.

IV. The brain portion of the early neural tube is of very
great length, one fifth the entire length of the nerve-tube
(one quarter if the foldings in the brain wall be taken into
account), at a stage when nearly the adult number of somites
are present. It is to be inferred, therefore, that the craniote
brain is composed of a far longer region of the anterior neural
tract than has hitherto been supposed. It is thus probably
homologous with the branchial region of the cord, as well as
with the so-called brain of Amphioxus.

V. The numerous foldings (asymmetrical) of the brain wall,
by which the regions come to appear, indicate a primitive
condition of the craniote brain, i.e. that the latter was
originally a tube of many vesicles! (eight of these at least
between cerebellum and thalamus), which in the ontogeny of
higher forms have become merged into fewer and larger ones.

VI. There is an almost entire absence of cranial flexure.

VII. Although the developmental type of Bdellostoma is
distinctly meroblastic, it is noteworthy that there appears no
trace of a germ ring; except in the immediate region of the
tail, the somites arise in situ at the sides of the neural axis.

VIII. The early appearance in each segment of pronephric
tubules, similar to each other in essential characters, demon-
strates, as Price has shown, that the entire excretory system of
Myxinoids must, from the standpoint of ontogeny, be regarded
as univalent. If we accept, therefore, Spengel’s criticisms of
the results of Semon on. the morphology of the excretory
system in Myxine, we must nevertheless admit Semon’s a
priori view as to its homology as a pronephros.

Note on the above by J. T. Cunningham, M.A.—It is,
I think, of some interest that the method of fishing by which
alone eggs of a Myxinoid have been obtained in California is
1 The writer believes that these may be distinguished from the neuromeres

of Locy and recent authors, which are not distinctly vesicular, and which
accur in cord as well as in brain.

278 BASHFORD DEAN.

commonly practised off the mouth of the Firth of Forth, and
elsewhere along the east coasts of Scotland and England,
although eggs of Myxine have never been obtained by that
method. Some years ago I made strenuous endeavours to
obtain fertilised eggs of Myxine, and received grants from
the Royal Society to defray the expenses of the investigation.
I frequently made excursions on fishing-boats engaged in fish-
ing for haddocks with long lines, which the Americans call
trawl-lines. On these lines large numbers of Myxine were
always taken on the hooks, just as in the fishing described in
the Bay of Monterey. But no extruded or fertilised eggs
were ever seen by me entangled in the slime of the Myxine.
The only explanation of this which seems possible is that off
the English coast the fishing-lines have not been shot on a
ground where Myxine spawn, all endeavours to obtain the
eggs from the lines or from the fishermen having hitherto
failed. I cannot agree with Mr. Bashford Dean in his re-
marks on the mode of oviposition. The sheathing capsules to
which he refers as enclosing the spawned eggs observed by a
fisherman must be, I think, the follicular capsules belonging
to the ovary, in which the eggs are developed. I have fre-
quently seen the eggs of Myxine enclosed in these capsules
pressed from the fish, but I showed in my paper on the gene-
rative organs that the eggs normally escaped from the ovary
by the rupture of these capsules, and the separation of the
capsules without rupture is due to some violence or pressure
in the capture or handling of the female. In Myxine it is
certain that the eggs when extruded are enclosed only in the
egg membrane provided with threads at the poles, this mem-
brane being of the nature of a chorion produced in the folli-
cular capsule. It is probable that in this respect the eggs of
Bdellostoma Stouti are quite similar to those of Myxine
glutinosa.

DEVELOPMENT OF THE CALIFORNIAN HAG-FISH. 279

DESCRIPTION OF PLATE 17,

Illustrating Dr. Bashford Dean’s paper “ On the Development
of the Californian Hag-fish, Bdellostoma Stouti,
Lockington.”

Fie. 1.—Bdellostoma taken on trawl-line, showing egg-string entangled in

encasing slime. xX &.

Fie. 2.—Tips of horn-like filaments of eggs. x about 30.

Fies. 3, 4, 5.—Stages of early embryos. x 44. as. Anterior end of
brain. au. Auditory vesicle. 3B. Brain. sp. Blastopore. Gs.
Gill-slits. u. Heart. w. Spinal cord. o. Rim of operculum. oP.
Optic vesicle. pn. Pronephric tubules. PS. Somites. UDZ. Un-
differentiated tail mass. * Appearance of mesoblast.

Fies. 6, 7, 8—Two stages of late embryos. x 44. A. Anus. av. Audi-
tory vesicle. pa. Dorsal aorta. Gs. Gill-esacs. oH. Heart. ™M.
Mucous pouch. op. Hye. n. Nasal sac. pn. Pronephric tubule.
sD. Pronephric duct.

« >
= = -
‘ ~~ '
*
* ;
'
é =
$ J
ae EL Latr ete
— .*
. a . - -
5a A @*
Q 4 a as
; .
« ¥
“<)
r
a cr
, =
’ j
‘ ' hy »
wey’, ;
Ja opal athe ol
et Se - ‘ ap trel
~~ 3
- . i
, . =
iy i) fei i
- wal? - ait
£ = 4 1 ‘i
ae)
ae Ss :
a oo
pt ‘
>"
{< ae) |
' : 4 ‘
Skee etre iew S
' : 7
aes
oa f
-
°
‘ .
- = .
.
= ’
ts ‘
»
’
. oa
7 '
ea) ve

pe eu ae
Pad ial nne us

ON THE DIPLOCHORDA. 281

On the Diplochorda.

1. The Structure of Actinotrocha.
2. The Structure of Cephalodiscus.

By

A. VT. Masterman, B.A., B.Sc.,
Lecturer and Assistant Professor of Natural History in the University
of St. Andrews, N.B.

With Plates 18—26.

INTRODUCTION.

Amonest the many and diverse types of animal structure
which are familiar to the zoologist there are some, both’ simple
and complex in character, which have never offered any serious
difficulty with regard to their systematic position in the natural
classification ; and there are others which ever since their first
discovery have provided material for discussion concerning
their systematic relationships, and whose true genetic con-
nection has not yet been clearly elucidated. From the nature
of the case these latter animals are usually generalised types,
and a certain proportion of the difference of opinion has arisen
by different observers laying greater stress upon special struc-
tural features.

On the other hand, possibly on account of the doubtful
nature of their genetic relationships, these animals are com-
monly either left out of zoological text-books altogether, or
are merely referred to casually as incerte sedes, although
the study of their structure and development always gives the
promise of renewed light upon the many morphological pro-
blems of the day. In other words, these animals form the
anomalies or exceptions which in other sciences, as well as

282 A. T. MASTERMAN.

that of biology, have always been the keys with which to unlock
the secrets of the natural system.

Amongst these types the genus Phoronis holds pre-
eminently the position here referred to; for though its first
discovery only dates back some forty years, or, including its
extraordinary larval form, a little over fifty years, and the
genus only comprises some half-dozen known species, yet a
long series of investigators have devoted their energies at one
time and another upon it, and the divergence of opinion re-
sulting from this and the attempt to point out its affinities
has been truly remarkable.

The larval form, Actinotrocha, first discovered by Muller
in the North Sea, was conjecturally assigned by sundry
zoologists to the Turbellaria, Rotatoria, Polyzoa, Mol-
lusca, Sipunculoidea, Annelida, and so on; and since
its transformation into the adult Phoronis was proved it
has been variously enligned with the Polyzoa, Sipuncu-
loidea (Gephyrea), Chetopoda, Brachiopoda, and so on.

The whole history and literature of the Phoronidea have
been so exhaustively dealt with, firstly by Professor McIntosh
(15), and secondly, so recently as 1892, by Dr. Cori (7), that it
is scarcely justifiable to repeat the account here, especially as
the memoir by Dr. Cori is the latest publication till within
the last year dealing with the genus. Suffice it to say that
the generally accepted opinions of zoologists appear to be
either in favour of regarding Phoronis as an aberrant
Sipunculid! or as an aberrant Polyzoan.? An intermediate
position between these two groups, or between one of them
and some other distinct groups, such as Mollusca or Bra-
chiopoda, has also been held by many.’

It was thus felt that the last word had not been said upon
Phoronis, and that a renewed investigation of the group,
especially with the assistance of modern methods, might be
attended by profitable results.

1 Krohn, Schneider, Wilson, Metschnikoff.

2 Ostroumoff, Barrois, Gegenbaur, McIntosh, Lankester, Cori.
3 Kowalevski, Claparede, Caldwell.

ON THE DIPLOCHORDA. 2838

The contents of the following pages are the result of more
or less intermittent work since 1894, which has led me to hold
that, although Phoronis has some features in common with
both Gephyrea and Polyzoa, its more immediate genetic
allies are not usually found in either of these groups, but are
commonly placed together under the title Hemichorda.

Thus whilst the more pronounced community of structure
holds between Phoronis and Cephalodiscus, yet the like-
ness of the former to Balanoglossus in certain other cha-
racters is so marked that it rather tends to strengthen the
alliance between the two latter.

No one can doubt that Phoronis is, in the adult stage,
greatly specialised for a sedentary life, and that this specialisa-
tion is, as in the case of the Tunicata, a degeneration involving
a loss of certain organs, degradation of others to a lower
structure, and yet again hypertrophy of others which are not
so prominent a feature in the archaic type. Thus a study of
the adult alone, with a view to phyletic conclusions, would be
embarrassed by these attendant complications; and were
Phoronis to reproduce itself by a foetal method, these could
not be surmounted. Fortunately, in Actinotrocha we have
a free larval form, which not only must be construed, with
certain reservations, as exemplifying the structure of the
undegenerate free-swimming ancestor of Phoronis, but which
has a very well-marked culminating point in differentiation,
with a sudden decline in morphological status to the adult
condition. This being so, an investigation of the fully
developed Actinotrocha should offer the readiest solution of
the problem before us.

In Part I, I have gone over the whole structure of the
Actinotrocha larva, as can be discerned by an examination
of preserved specimens, both whole and in section. The
method of examination by serial sections has enabled me to
correct some misconceptions with regard to the structure of
this unique larval form. It is purposed to show that Actino-
trocha has so close a similarity in structure to the three
members of the Hemichorda that the assumption of a

284, A. T. MASTERMAN.

genetic connection appears to be the only conclusion. Never-
theless, for reasons stated below, I would suggest that
Phoronis be considered as constituting a distinct subdivision
of the Chordata, for which the name Diplochorda is
proposed.

In this group the leading anatomical characters of the highest
Chordata have their rudimentary homologues, and the dis-
tinctive features of the Vertebrata are reduced to so general
a type that the transition to the true Invertebrata is but a
small step.

Whatever value this research may have as an addition to
our knowledge cannot therefore be summed up by the looked-
for result that the Chordate affinities of Phoronis will be
definitely acknowledged, but will lie in the fact that the
recognition of this relationship will bring the base of the
Chordate tree into contact with certain Invertebrata which
have always been more or less connected with Phoronis.
My suggestion of the group Archicelomata is merely the
embodiment of this conception, and the apology for its creation
will be forthcoming if it may lead specialists in the groups of
Sipunculids, Polyzoa, Brachiopoda, and Echinoder-
mata (our great sedentary groups) to hold in view the possible
occurrence of pre-chordal characters in the ontogeny of their
investigated types. The discovery of the close connection of
the Tunicata with Amphioxus served rather to remove
them from their former alliance, and cannot be said to
have pointed to any link between the Chordata and the
Invertebrata.

The removal of Balanoglossus, and later the association
with it of Cephalodiscus and Rhabdopleura, allowed the
Chordate ancestry to be traced back to pre-chordal (Lankester)
times ; and the study of Phoronis, and especially its develop-
ment, takes us back to the definite meeting-place between the
two great types of segmented animals, the segmented In-
vertebrates and the higher Chordata.

During the progress of this work a detailed comparison
was made between the several organs of Phoronis and Acti-

ON THE DIPLOCHORDA. 285

notrocha with the Hemichorda (24 and 25). In this
comparison it was assumed that the structure of Balano-
glossus had been modified by its peculiar habits in such a
way that the notochord and other organs had become secon-
darily shifted into the pre-oral lobe. This led to the difficulty
that the ‘‘ notochord” of Cephalodiscus (Harmer, 10) was
in the same position, and yet the burrowing habit is not in this
case indulged in.

The discovery of the “subneural gland” (see below) in
Actinotrocha, however, seemed to complicate the difficulty ;
for this organ, although epiblastic, occupies the same position
and has the same relationships to other organs as the “ noto-
chord” of Cephalodiscus. The natural conclusion was that
the “notochord ” of Cephalodiscus was a subneural gland,
and the notochord itself was yet to be found. Withsuch an end
in view [examined Cephalodiscus by sections, and there can
be no doubt that this animal possesses a paired notochord on
the dorso-lateral part of the pharynx. In the light of this
and other facts given below, granting that the structural and
phyletic relationships of animals are to be expressed in our
classification, it seems reasonable, and indeed neces-
sary, that Phoronis and Cephalodiscus be included
in one group, Diplochorda; whilst their relation-
ship to Balanoglossus (or Hemichorda) may be ex-
pressed by including the two groups under the name
Archichorda.

The present paper will therefore be confined to an account
of the structure of Actinotrocha, and a note upon the
structure of Cephalodiscus. Part I.—The Structure of
Actinotrocha. Part II.—The Anatomy of Cephalodiscus.

286 A. T. MASTERMAN.

PART I:
The Structure of Actinotrocha.

Literature.

As is well known, the larva, Actinotrocha, was first de-
scribed by J. Miller (16) as occurring in the North Sea. He
gave an account of the general characters, the form of body,
number of tentacles, &c. This was followed a year later by a
research of Wagener, who, assuming the animal to be an adult,
gave a description of its anatomical features. His work was
conducted, as far as can be ascertained, without the assistance
of sections, and it will be referred to in the following pages.
It is evident that some of his statements are erroneous, more
especially in the section referring to the nervous system.

In 1850, Siebold (19) contended for the larval nature of
Actinotrocha, drawing a comparison with Bipinnaria.

In 1854, C. Gegenbaur (9) notified the occurrence of a
similar form from Messina, in winter, of a small size (35 mm.),
and remarks on the differences from Miiller’s. He traced its
growth and differentiation to ‘4 mm., with an accompanying
marked elongation of the hind region.

In 1858, Krohn (11), examining the same species as Gegen-
baur, came to the conclusion that it had specific distinctions
from that of J. Miiller, mainly in the perianal ciliated band
and the pigmentation. He also discovered the fact that its
metamorphosis resulted in a gephyrean worm.

In 1862, A. Schneider (18) described and figured a new
species, A. pallida, and gave an account of the metamor-
phosis into the worm-like ‘ Sipunculid ”” afterwards identified
as Phoronis. His description of the larval anatomy con-
firmed in most particulars that of his predecessors, but he cor-
rected Wagener on some points, such as the blood-system.
His description and that of Claparéde (5) of the “ notochord ”
are of special interest. His observations on the adult worm
led him to find its nearest allies in Phascolosoma, Sipun-
culus, and Aspidosiphon.

ON THE DIPLOCHORDA. 287

In 1867 Actinotrocha again came under detailed notice,
this time at the hands of Kowalevski (12), who gave an
account of the developmental changes and metamorphosis.
He identified the adult worm with Phoronis (syn. Crepina,
van Beneden). Written in Russian, this memoir would be a
sealed book to most morphologists were it not for a German
epitome. Yet again, in 1871, E. Metschnikoff (14) obtained
a number of early stages from the Mediterranean Sea, which
he described and figured, and also gave further particulars upon
the metamorphosis. Like his predecessors, he did not adopt
the method of sections in his work, or it is probable that some
further important facts on the structure of his young stages
would be to hand. After a period of some ten years the
Actinotrocha of Chesapeake Bay was investigated by E.
Wilson (22), who discriminated two species, and followed
out the metamorphosis. Disagreeing in minor points with
Metschnikoff, he confirmed in the main this naturalist’s
account. A large part of his paper is devoted to a discussion
of the significance of the metamorphosis and of the systematic
position of Phoronis.

In 1883 a preliminary note upon the development of
Phoronis, besides other points, was issued by Caldwell (2).
So far as I am aware, nothing further has been contri-
buted by this worker except a short paper (3) on the early
stages of the embryos, although no doubt the expectation
of further contributions has caused the investigation of
Actinotrocha and its earlier stages to be neglected for the
last dozen years or so. In comparing my results with those
of Caldwell there has been an important difficulty to contend
with. In these days of rush for priority it is questionable
how far a worker can be expected to stand by every assertion
made in a “preliminary” note, and it is also difficult in a
paper of this kind to discern how much of the statements
therein contained are claimed as original. If Caldwell’s
paper does not, therefore, receive as much recognition below as
it would seem to be entitled to, its preliminary and therefore
tentative nature must be given as the reason and excuse.

voL. 40, PART 2.—NEW SER. x

288 A. T. MASTERMAN,

To these we must add that Cobbold (4) describes, in 1857,
the occurrence of Actinotrocha in the Frith of Forth. His
paper is principally notable for the statement that the larve
fixed themselves by their anal extremity.

Two years after (1859), Dyster (8) noted the eggs and early
stages of Phoronis. We may add that in 1888 McIntosh (15)
gave, in connection with his anatomical account of Phoronis
buskii, two figures and a short description of some early
stages of Actinotrocha from the tentacles of the adult.

We may only add to this short account the fact that
Actinotrocha, after it was once proved to be the larval form
of Phoronis, has in various text-books and general zoological
articles been usually interpreted as a much “ modified trocho-
sphere” larva, whilst comparisons have been drawn between
its structure and that of the Echinoderm larve. Although
several authors, to be mentioned later, have drawn a com-
parison between the adult Phoronis and the Hemichordata,
Harmer (10) is the only author, as far as I have been able to
ascertain, who has attempted a tentative comparison between
the organs of Actinotrocha and those of Balanoglossus.
His deductions were limited in extent, owing to the supposed
absence of the notochord, gill-slits, and mesenteries separat-
ing the cavity of the pre-oral lobe from those of the lopho-
phoral region.

External Form.

The species of Actinotrocha referred to in this paper
does not appear to differ in any essential respect from that
originally described by J. Miiller near Heligoland, and named
by him Actinotrocha branchiata. The specimens here
figured were caught by the bottom tow-net in St. Andrews
Bay. In the absence of knowledge with respect to the adult
in both cases the minor difference need not concern us here.
Although the general appearance of Actinotrocha is familiar
to every zoologist from the figures of Muller (16), Wagener
(20), Metschnikoff (14), and many others, copied more or less
accurately into nearly every illustrated text-book, yet I have

ON THE DIPLOCHORDA. 289

drawn the general appearance of the larva intact, and in
several positions mainly for the demonstration of several
newly described organs.

Pl. 18, fig. 2, gives the appearance of a young Actino-
trocha with twenty arms, as seen in side view; the hood is in
the position which perhaps may be described as normal. The
front parts of a later stage are indicated, as seen from the
dorsal surface, in Pl. 18, fig. 3; and a side view of the same
parts at a yet more advanced stage is seen in Pl, 21, fig. 38.

Lastly, the general shape of the body is shown in Pl. 18,
fig. 2. ‘This is the side view of a larva with eighteen tentacles,
introduced mainly to give an indication of the distribution of
the principal nervous tracts. The posture of the pre-oral hood
is very characteristic; indeed, most Actinotroche die in
this position, or else with the hood turned still further back.
The larva when alive is very active, and on provocation is
capable of extraordinary contortions. Figs. 4 and 5 are two
outline sketches which indicate two extreme positions assumed
by one lively little fellow who lived for some hours in a small
glass vessel, the only living specimen I have yet observed.
The mobility of the pre-oral hood and its great prominence in
the anatomy of the animal are remarkable features of its
general anatomy.

It is not proposed here to follow out the development to
any great extent, but there are indications that there is a
great increase of size in the later stages, confined mostly to
the trunk portion of the body. It is difficult to speak with
certainty, but it appears likely that this increase is further
followed by a contraction in size, a phenomenon of the same
nature as that described for Tornaria by Morgan (17); at
least, certain of the larve are smaller than usual, and in
section all their epiblastic cells are closely aggregated, and
form a thick epithelial wall, in marked contrast to the greatly
attenuated cells which form the outer covering of the trunk in
the larger larve. In these smaller forms the organs are
closely packed together, and the coelomic cavities are very
reduced. The natural supposition would be that they are the

290 A. T. MASTERMAN.

earlier stages of the larger larve; but this cannot be the case,
for the structure of the notochord, of the nervous and circu-
latory systems, besides the number of tentacles, show clearly
that they are at a more advanced stage. In the absence of
direct observation of transition stages, there are only two
alternative conclusions. These smaller larve are either later
stages, reduced in size as above suggested, or they belong to a
different species. Wilson (22) observed two distinct species
of Actinotrocha, which he called A and B, differing mainly
in the fact that the form A had a shorter intestine and stouter
body.

The whole body of Actinotrocha has three natural divi-
sions :

1. The pre-oral hood (syn. pre-oral lobe, cephalic lobe,
protomere), whose position is essentially pre-oral, and from
Metschnikoff’s figures (14) of early stages it is primarily
so. This hood is homologous in every particular with the
“‘ proboscis ” of Balanoglossus, the “buccal dise”’ (buccal
shield) of Cephalodiscus, and the “ epistome”’ of Rhabdo-
pleura.

This pre-oral lobe, whose shape and relationships are shown
_ in the figures already referred to, is produced backwards into
two lateral horns (cf. Balanoglossus), which fuse on to the
body-wall, laterally to the mouth-opening.

2. The lophophoral area, or the part bearing the arms. It
commences at the hind end of the pre-oral hood, the line of
junction being marked by a radial nerve (see Nervous System),
and extends backwards a very short way in the dorsal region,
but a greater distance in the ventral. It is limited posteriorly
by the line of tentacles, or rather by the nerve-ring lying
posteriorly to them; and as this passes diagonally from the
front dorsal region (nerve-ganglion) to the ventral side (see
Pl. 18, fig. 2) it is easy to see that this lophophoral segment
consists of a cylindrical area surrounding the part immediately
behind the mouth, the ventral length of the cylinder being
many times longer than the dorsal. This segment or area is,
I hope to show, homologous with the “ collar” of Balano-

ON THE DIPLOCHORDA. 291

glossus, and the lophophoral segment (already sometimes
termed the “collar”) of Cephalodiscus and Rhabdo-
pleura. The whole segment may well be termed in all these
animals the ‘‘ mesomere ”’ (see below).

3. The trunk, extending as an elongated cylindrical portion
backwards from the mesomere, and bearing terminally the
perianal band and the anus. It will probably be a matter of
no great difficulty to show that this “ trunk ” segment (“ meta-
mere”) is homologous with the trunk of Balanoglossus,
Cephalodiscus, and Rhabdopleura.

The three segments are sufficiently well defined externally,
but the arrangement of the internal organs, such as ccelomic
cavities, is even more marked, and corresponds exactly with
the divisions here indicated.

There are three prominent ciliated bands, the pre-oral (or
prototroch), the collar-band following the line of the tentacles
(mesotroch), and the trunk band (perianal or metatroch)
surrounding the anus. Lach is defined not only by the pre-
seuce of very long cilia, but of elongated and densely aggre-
gated epithelial ectodermal cells bearing them, and well
supplied with nerve-tracts.

Of the three the perianal band is the most prominent, and
functions as the locomotor organ of the larva.

The other general relations of the three divisions are
diagrammatically shown in Pl. 22, figs. 47 and 50.

The Organs of the Epiblast.

The dorsal and ventral surfaces of the pre-oral hood are
sharply separated by the pre-oral band of cilia and its thickened
cells covering a nerve-ring. The whole dorsal surface is covered
with thick glandular epiblast cells, which bear minute cilia
(Pl. 19, fig. 6, and Pl. 20, figs. 18, 19, 20, 28, and 29). Each cell
is elongated, with a nucleus near its middle. In some parts
the longest cells are found in the mid-dorsal part, but in others
the part next to the ciliated band is thickened.

In the small specimens referred to above, this very thick
glandular epiblast is more conspicuous than in the larger

292 A. T. MASTERMAN.

forms, and in preserved specimens is greatly folded on itself.
On either side of the middle line, near the nerve-ganglion, a
pore leads by a short canal into the coelomic cavity of the pre-
orallobe. These two pores and canals are described in further
detail below, and seem to be comparable with the “ proboscis-
pores” of Cephalodiscus.

The ventral surface of the pre-oral lobe has an epithelium of a
very different character from that already described forthe dorsal.
The body-wall is extremely thin, and consists of attenuated
cells (Pl. 19, fig. 6, and Pl. 20, figs. 18, 19, 20, 28, and 29),
except in the region immediately anterior to the mouth.
Here they are rathe rmore cubical and bear cilia (fig. 6, v.
c.6.). I have not been enabled to detect any cilia upon the
rest of the ventral surface of the pre-oral lobe.

Almost the whole surface of the collar segment is covered
with cilia, and the cells are fairly thick and cubical. Running
down the mid-ventral line from the mouth region is found a
ridge of cells with strong cilia, and on either side of this ridge,
for some way backwards, the cells of the ventral surface of the
collar are not ciliated, and are somewhat attenuated. Dorsally
the thick epithelium of the pre-oral lobe is continued on to
that of the collar on either side of the nerve-ganglion, and
(Pl. 19, fig. 7) the dorsal surface of the tentacles is also
covered with thick epithelial cells bearing long cilia.

In the larger larve the epiblast of the trunk becomes so
attenuated as to defy complete analysis with the highest ob-
jective at my disposal (Pl. 21, fig. 89). The only exception to
this is the epiblast of the perianal band, which is formed by the
longest cells of the whole body, aggregated in enormous num-
bers, and bearing long flagelliform cilia. Seen in cross-section
the ciliated ring or band appears lenticular, and just anteriorly
to it the cells are slightly thickened, and form a nervous ring
(Pl. 20, figs. 22 and 23).

This attenuation of the epiblastic cells of the trunk segment
seems to be attended by and connected with an elongation of
this region. Gegenbaur (9) has noticed that in the earlier
stages (‘35 mm. long) the trunk region was short, and that by

ON THE DIPLOCHORDA. 293

the time a length of ‘5 mm. was reached it had greatly elon-
gated, and formed a long cylindrical portion behind the tenta-
cles. This was confirmed by Metschnikoff (14) and by others.
A similar phenomenon is found in Tornaria, according to
Morgan (17), the area between the post-oral and the perianal
bands being very small in his younger stages, but elongating
greatly afterwards.

It is a very necessary process in Actinotrocha in order
that the requisite space for the accommodation of the ventral
diverticulum may be forthcoming.

An inspection of the figs. 18—27 inclusive (Pl. 20) will suffi-
ciently indicate the further relationships of the areas in which
the epiblast is thickened to those in which it is attenuated. In
fig. 28 is seen the marked contrast in this character of the
cells on the dorsal and the ventral surfaces of the pre-oral
lobe. In fig. 29 are to be noticed the two oral grooves (0. g.)
passing out from the mouth. They are longitudinal depres-
sions in the ventral collar area, and are also found in Cepha-
lodiscus. In fig. 26 the right-hand groove is seen entering
the side of the mouth. They can be traced on to the general
surface of the collar. On the ventro-lateral side of the pre-oral
lobe are also two grooves (fig. 24, a. g.), which pass from the
antero-lateral corners of the mouth round the edge of the pre-
oral hood. Their bounding cells, like the rest of the ventral
surface of the hood, are without cilia and much attenuated.
They may be termed the atrial grooves, and no doubt serve for
the removal of the water which is brought into the mouth
by the action of the ciliated oral grooves. In fact, they are
the true analogues of the gill-slits of the Chordata, though
perhaps not homologous with these.

The histology of the several areas is sufficiently indicated in
a series of transverse sections (Pl. 19, figs. 6—10, 16, 17, and
Pl. 21, fig. 39).

Nervous System.

The nervous system of Actinotrocha has been little
studied, and some of the observations made with respect to it
are without doubt erroneous. Wagener (20) describes in some

294, A. T. MASTERMAN.

detail the appearances which he considered to be connected
with the nervous system. He noticed, on the dorsal surface
of the pre-oral hood, a “ horn-shaped protuberance,” and,
posterior to this, a wart-like thickening. This latter is the true
nerve-ganglion, aud the former is the apical sense-organ, but
they were not recognised by him as such. On the contrary,
he describes a swelling (Wulst) on the ventral surface of the
pre-oral lobe, which he conjectures to be the ganglion. Ra-
diating from this, he says, are found a number of lines, and
he indicates the course of these lines, and even figures the
nervous system, as it appears when isolated (fig. 6). His
“Wulst ” is the aperture of the subneural gland, to be de-
scribed later, and the radiating lines and commissures are, at
any rate in large part, the meseuteries separating the cavities
of the pre-oral lobe and the collar, together with the mesentery
bounding the vascular space below the true nerve-ganglion.
A comparison of P]. 18, fig. 3, with Wagener’s figure will make
this clear.

One can readily imagine how an incautious inspection of
the larva intact might cause this mistake, but how the observer
referred to managed to “isolate”? such a heterogeneous and
unconnected collection of structures is not easy to understand.

Caldwell (2) speaks of the ectoderm becoming thickened in
two regions :

1. In the pre-oral lobe.

2. In the form of a post-oral ring round the mouth.

The former, he states, becomes the future ganglion, and the
latter the circum-cesophageal nerve-ring. Again, in the fully
developed larva he remarks that “the nervous elements of
the ectoderm of the pre-oral lobe in all species are concentrated
into a ganglion. In some species a large number of nerve-
fibres pass forwards from it to a sense-organ.”’

The central nerve-ganglion (figs. 1—3, 18, 14, 15, 20, 23,
28, n.g.) is a lenticular mass lying on the mid-dorsal line at
the base of the pre-oral lobe. In fig. 1 it is seen to be at the
front end of the body, but when the hood is directed forwards
its true situation at the base of the pre-oral hood is evident

ON THE DIPLOCHORDA. 295

(Pl. 18, fig. 2, and Pl. 20, fig. 20). It consists of a mass of
ganglion-cells and fibres in which the inner ends of the over-
lying epiblast take part.

The mesoblastic posterior wall of the pre-oral lobe can be
seen passing downwards from its front border (figs. 1, 20, 23,
&c., mes.), and the anterior mesoblastic wall of the collar
cavities is in like manner to be observed passing from its
posterior border inwards to the wsophagus (PI. 18, fig. 2). The
space lying immediately below the ganglion is therefore a large
hemoceelic cavity (fig. 2, s.v.s.). The position of this ganglion
is exactly similar to that of the central nerve-ganglion of
Cephalodiscus, as described by McIntosh, and the homology
of the two cannot be doubted.

In early stages this ganglion les just under the epiblast,
which is continued forwards over the hood (Pl. 18, fig. 1); but as
development proceeds a depression of the epiblast immediately
in front of the ganglion takes place. The ganglion having a
convex anterior border, the opening of this epiblastic pit is at
first crescentic in shape, but afterwards it becomes only
elongated from side to side. The depression becomes deeper,
and eventually forms a long sac-like diverticulum lying under
the nerve-ganglion (Pl. 18, fig. 2, Pl. 20, fig. 20, &c., . p.).

The ganglion is, as already stated, a proliferation of the
inner cells of the epiblast, and in almost every direction there
radiate nerve-tracts formed by the inner ends of the epiblastic
cells. This is especially the case in the anterior direction
over the pre-oral hood. Here there are three main nerves,
which run parallel to each other, one median and the other two
on either side of it (Pl. 18, fig. 3, a. m.); and on either side of
these again are a great number of smaller trunks, the course
of which is indicated in fig. 2. They are seen to run forwards
and outwards, and then to bend backwards and take a course
to the posterior corners of the hood. All these are involved
in the invaginated pit here referred to, and they appear in
surface view to radiate from its front border. They are con-
tinued backwards, however, along the lower wall of the pit,
and then round its base and up to the ganglion. The pit is

296 A. T. MASTERMAN.

therefore a depression of nervous epiblastic tissue immediately
in front of the main nerve-ganglion. It appears to be homo-
logous with the ‘“‘neuropore” of the Chordata in its
inceptive stage. The same process of depression involving the
nerve-ganglion, and carried further backward, would result in
a tubular dorsal nervous system of the same type as that in
Balanoglossus and the higher Chordata.

The three dorsal anterior nerves run forward, and the
median one (PI. 19, fig. 6, a. m. n.) continues into a large median
nerve-swelling lying immediately under the sense-papilla.
Beyond this, however, it can be traced together with the two
dorso-lateral nerves (fig. 6, &c., a. J. n.) till they are lost in the
pre-oral nerve-ring. This is a thick ring of nervous tissue,
found immediately under the pre-oral ciliated band throughout
its length. At the posterior corners of the pre-oral hood this
ring terminates as such, and appears to branch out in several
directions. One branch passes up again to the nerve-ganglion
(Pl. 18, fig. 2), and numerous fibres also appear to pass on to the
ventral surface of the collar region (Pl. 19, fig. 10, p. 0. n.).
The mass of nerve-tracts already referred to as leading out from
the ‘‘ neuropore,” and bending back to this spot, appear also
to pass on to the ventral surface of the collar, forming a pair
of nervous areas on either side of the mouth. These spread
out over the whole ventral surface of the collar as a series of
branching fibres.

Posteriorly the nerve-ganglion passes gradually into a pair
of thick nerve-tracts, which run along the dorsal collar region
and then diverge. The main mass of fibres passes diagonally
downwards immediately behind the tentacles, and, as is seen
in sections, exactly at the posterior border of the collar
region (Pl. 20, fig. 19, ¢.”.s.). The rest of the fibres pass mid-
dorsally as a pair of tracts, giving off branches to the body-
wall, and terminating in a nervous ring just anterior to the
perianal band.

The collar-ring behind the tentacles is continued in the
mid-ventral line to the same nerve-ring in front of the perianal
band, and supplying the ciliated cells of this area with nerves,

ON THE DIPLOCHORDA. 297

In addition to these main nervous tracts the inner ends of
the whole epiblast form a fine fibrillar nervous network in every
direction. In the thin epiblast of the trunk I have not been
able to make these out, but in surface view this area shows,
as already described, fine nerves branching out in all directions
from the ventral and dorsal trunks.

Pl. 18, fig. 2, is a surface view from the side of the whole
larva, with the chief nerves indicated in black, whilst in fig. 1
are indicated the chief nerves of the hood in dorsal view at a
slightly earlier stage, and figured as they actually appear, 1. e.
as fine unstained tracts.

The appearance of the nerve-tracts, as seen in section, is
shown in Pl. 19, figs. 6 to 17.

In fig. 6 are to be noticed the three main dorsal nerves of
the hood (a.d.n. and a.m.n.), which at the level of fig. 11
commence to pass down the ventral floor of the neural tube,
and in fig. 13 the median nerve is seen to pass into the
ganglion.

In fig. 15 the posterior end of the ganglion is reached, aud
it is seen to diverge into two main dorsal trunks situated
(fig. 7) upon the dorsal surface of the collar area. These
may be traced through fig. 8 at the level of the mouth, and
fig. 9, which is almost at the posterior dorsal termination of
the collar area. Here they diverge, and are found at the base
of the tentacles, and they are seen to give off branches to the
tentacles.

In fig. 9, immediately below the mouth, may be seen the
ventral nervous area formed by the partial fusion of two tracts
situated lateral to the mouth in fig. 8.

Fig. 10 shows the collar-ring (c. 7.7.) still more diverging
into two lateral branches, and the mid-dorsal region shows in-
dications of the fibres running down to the perianal band,
though the attenuation of the cells makes them difficult to
follow. On the right of this figure is seen (p.0.m.) the
termination of the pre-oral nerve-ring, and branches from it
may be noticed passing dorsally and ventrally.

In fig. 16 the foregoing features of the collar-ring, &c.,

298 A. T. MASTERMAN.

are seen, and in addition the ventral nerve-tract is seen to
be now clearly separated into two ventro-lateral areas (v.c.n.)
with columnar ciliated cells, though the nerve-fibres may be
also traced somewhat prominently in the mid-ventral line. In
Pl. 21, fig. 39, the trunk-area alone is seen in section, and
ventral nervous areas can be just discerned, though their
demonstration is easier in surface views. Thus the nervous
system of Actinotrocha consists of—

1. A central ganglion lying in the front collar region and
between this and the pre-oral lobe, with the epiblast imme-
diately in front, depressed to form a neuropore.

2. A ring round the posterior part of the collar, continued
from the ganglion dorsally and ventrally, giving off fine double
groups of nerve-tracts to the anal end of the body.

3. Groups of fine nerve-tracts continued dorsally along the
trunk from the hind end of the collar to the anal end of
the body.

4, A ring around the anal end of the trunk, into which the
dorsal and ventral tracts lead.

5. A ring round the edge of the pre-oral lobe, joined at
each side to the ganglion, and in the median front region by
three main tracts running in the mid-dorsal line forwards from
the ganglion.

6. A diffuse plexus of fibres at the base of nearly all the
epiblastic layer, conspicuous amongst which are the fibres of
the ventral collar area, which pass forwards and dorsally to
meet the ganglion.

A comparison of this nervous system with that of Balano-
glossus and Cephalodiscus will be instituted later.

Sens e-organs.—At a stage later than fig. 1 (vide fig. 2)
the mid-dorsal surface of the pre-oral lobe is elevated into a
cone-shaped process, which has slightly columnar cells, im-
mediately under which is a swelling of nervous fibres pro-
truding into the cavity of the pre-oral lobe (fig. 20, s.p.). The
examination of fresh specimens is required for determination of
the further structure of this organ.

Pre-oral Ciliated Ring.—Round the edge of the pre-

ON THE DIPLOCHORDA. 299

oral hood is an area of epiblastic cells which bear long and
prominent cilia. The cells themselves are rather longer than
broad (PI. 21, fig. 44), and fine fibrils are readily distinguished
passing into the mass of fibres which forms a nerve-ring
immediately under the epiblast. These cells come to a sudden
termination at the posterior angles of the pre-oral lobe, where
they are continued on to the ordinary cells of the body.

Tentacles.—The epithelium of the tentacles is of two
types. Pl. 19, fig. 17, shows a succession of tentacles cut in
transverse section, so that various stages are indicated. Starting
from the right-hand side the tentacles are seen to arise as pro-
tuberances of the thickened ciliated cells of the collar area,
with a diverticulum of the collar celom running down the
centre. The nervous area at the base of the cells is prominent,
especially at the outer angle of the thickened cells. As the
tentacle becomes completely separated from the surface of the
collar its inner wall is compieted by a very thin non-ciliated
area of cells. The transverse section of a tentacle (Pl. 21, fig.
43) is therefore somewhat ovoid, two-thirds of its epithelium
having thickened ciliated cells, and the posterior one-third
consisting of a thin cellular lamella. At the outer angle isa
nervous area, which is probably a tentacular nerve seen in cross-
section ; and immediately internal to this is a hemoceele space
or vascular trunk, which, however, is only present at the base
of the tentacle (cf. fig. 43). The rest of the cavity is cclomic,
and is lined by a mesoblastic wall, between which and the
epiblast is a mesoblastic skeletal lamella, referred to later.

The Perianal Band.—The perianal band is composed of a
dense aggregation of very long columnar cells (Pl. 18, figs. 1
and 2, p. a.) in direct continuity with the epiblastic cells of the
trunk. In cross-section the band appears lenticular in outline
(Pl. 20, figs. 18 to 23, and Pl. 21, fig. 45),-and the cilia appear
to be grouped more or less into masses. They are many times
longer than any other cilia on the body, and are slightly
curved towards the posterior end. Immediately in front of the
band is a nerve-ring in a shallow depression, and fibres from
this can be seen to enter the bases of the ciliated cells,

3800 A. T, MASTERMAN.

Proctodeum, Stomodeum, and Subneural Gland.

From a want of definite knowledge upon the point one
cannot speak of a proctodeum with any degree of certainty,
but Caldwell refers to a “ slight invagination of ectoderm” in
the formation of the anus. From a figure of an early stage,
given by Metschnikoff (14), there can be little doubt that the
organ usually referred to as the oesophagus is epiblastic in its
origin. In Metschnikoff’s figure (14, Taf. xix, 1) the connection
between the cesophagus and stomach is apparently not yet
established. In view, however, of no certain statement, the
cesophagus will be described later.

Subneural Gland.—There remains an important organ
about the origin of which there can be no doubt. During

1 2. 3. 4.,

PS ON LSS,

1. Early stage of subneural gland in longitudinal section. 2. Later ditto. 3. Later
than 2. 4. Latest stage ditto.
development the epiblast becomes tucked in at the mouth, and
in the mid-ventral line of the hood, anterior to the mouth, there
is furmed a small depression (woodcut 1). As the depression
deepens and increases in size it is carried further inside the
buccal cavity (see woodcuts 2, 3, and 4), until the latest con-
dition I have observed, seen in PI. 21, fig. 38, s.., is arrived at.
In median sagittal section the same organ is shown in Pl. 20,
fig. 20(s. .). Its walls at this stage still exhibit a structure
closely similar to that of the oesophagus. As referred to below,
this may be termed the subneural gland, and may be compared
to the organ of the Tunicata bearing the same name, and

ON THE DIPLOCHORDA. 301

possibly to the hypophysis of the Vertebrata. Its relation
to the blood-sytem, &c., will be described under that head.

Organs of the Hypoblast.

The alimentary canal is ciliated throughout, consisting of a
tube of varying character, from the ventral mouth to the
posterior anus. The mouth opens immediately under the pre-
oral lobe, between it and the collar region. A mid-ventral
ciliated area leads into it from the pre-oral lobe in front, and
a broad ciliated area, depressed into two oral grooves (Pl. 20,
fig. 29, 0. g.), approaches it from the ventral surface of the
collar area.

From the dorso-lateral corners of the mouth lead outwards
the atrial grooves. The general arrangement, as seen in ventral
view, is indicated in woodcut 5. At the edge of the mouth

D.

5. View of mouth-parts of Actinotrocha, from ventral surface. In the oral
grooves the arrows point towards the mouth, in the atrial grooves they
point away, and indicate the course of currents.

the cells on all sides pass by gradation to the long columnar

ciliated cells of the esophagus. The nuclei of these are evenly

arranged at their centre (Pl. 19, fig. 8), and the inner half of
each cell is clearer, and not so readily stained as the outer.

In fig. 7 may be noticed the base of the subneural gland,
leading off from the “ cesophagus ; ” and in figs. 9 and 10 there
is shown the typical cross-section of the cesophagus below the
mouth. On their outer face the cells are covered by a thin

302 A. T. MASTERMAN.

layer of the mesoblastic lining of the collar cavities. The
course of the cesophagus is first upwards towards the dorsal
surface, and then curved backwards (Pl. 20, fig. 20; Pl. 21, fig.
38), and has a straight direction till it opens into the dorsal
anterior end of the ‘‘ stomach,” which is partially differentiated
into a pharynx.

The part of the ‘stomach”’ lying in the collar region is
separated by a constriction from the part behind, and in fact in
some cases the constriction forms a short tube running from the
anterior part or pharynx to the posterior part or true stomach.

The pharynx has small cubical cells with regularly disposed
nuclei; the protoplasm stains homogeneously. They have
small cilia, but the contrast of these cells with the “ ceso-
phageal”’ cells is marked and the transition abrupt (cf. Pl. 19,
figs. 9 and 10).

The cells situated in the mid-dorsal region of the pharynx
are, however, for some way columnar and strongly ciliated
(fig. 16), and altogether resemble more the “ cesophageal”
type. In the antero-lateral region the pharyngeal wall is pro-
duced into two remarkable diverticula, which in the fully
developed larva lie, as a pair of elongated organs (notochords),

laterally to the cesophagus (figs. 9, 22, nch.).
~The cells are columnar, with homogeneous protoplasm, but
in the course of development they undergo a remarkable
transformation. The earliest stage yet found by myself (Pl. 21,
fig. 30) presents the diverticulum as a hemispherical protube-
rance. Sections at the level of a,b, and c exhibit the characters
shown in figs. 30a, 300, and 30ce.

In fig. 30a is seen a section through spherical vacuoles with
a few nuclei squeezed in between them. The vacuoles are
filled with a clear, homogeneous, non-staining substance. (I
have failed to stain them in the slightest degree with hema-
toxylin, carmine, picric acid, iodine, and numerous aniline
stains, such as eosin, methyl green, safranin, &c.)

In fig. 30¢ the central cavity is reached, and the cells are
seen in longitudinal section. This reveals that the vacuoles
are regularly disposed at the distal extremity of each cell, one

ON THE DIPLOCHORDA. 303

to each, and that the nucleus lies exactly internal to the
vacuole in each case. The contrast between the even ring of
light vacuoles and the inner ring of dark nuclei makes a beau-
tiful and symmetrical section, which can be only poorly indi-
cated in the figure. We cannot doubt that the presence of
the vacuoles is due to the secretion of some fluid product at
the distal extremity of the cell, which is further enclosed in a
cell membrane. The close contact of the nucleus with the
vacuole is suggestive in this connection.

In a later stage (figs. 81 and 32) there has arisen another
vacuole in each cell, internal to the former. The whole organ
is also more elongated. In fig. 31 the vacuoles are mostly
alternate to each other, but in certain stages there can be dis-
cerned, especially at the area later developed (below v. in
fig. 31), that the inner vacuole is first formed exactly internal
to the first, and that later on a shifting takes place till the
alternate arrangement is arrived at. The latter is evidently
the more stable condition, and the assumption of it by these
vacuoles tends to indicate that the vacuoles are turgid. In
this stage the nuclei are still regularly disposed, the majority
situated in an even row internal to the vacuoles, but a few are
caught between the latter and are seen squeezed in the chinks
here and there.

These changes may be indicated semi-diagrammatically by
figs. 33, 34, and 35. The stage with three rows of vacuoles is not
apparently to be obtained, because the vacuoles are later dis-
posed rather more irregularly, and the next stage of import-
ance is seen in fig. 36. Here the vacuoles have become
squeezed together, like fish-eggs allowed to settle slowly in
spirit, and some are no longer spherical. The majority of the
nuclei are still confined to a small area of protoplasm at the
inner end of each cell; they are not very regularly arranged.
Fig. 387 is added to show the condition of this organ as seen
in oblique section, within the young Phoronis, soon after
the metamorphosis. Its subsequent fate is unknown, but it is
not present in the adult. We may note in this final stage
that the vacuoles are still further distorted by mutual pressure,

vot. 40, part 2.—-NEW SER, Y

304 A. T. MASTERMAN,

and altered in outline. I have found no indication that the
fusion of the vacuoles with each other takes place. The position
of the organs in situ is seen in PI. 18, fig. 1, and Pl. 21,
fig. 88 (nch.), in lateral view, whilst fig. 3 gives the dorsal
view of the two notochords (nch.). In PI. 20, fig. 22, their
appearance in coronal section can be followed, whilst fig. 19
shows a sagittal section of one of them.

We may emphasise the fact that development of the noto-
chords proceeds in two ways :—l. The organs themselves arise
from a pair of evaginations of the antero-lateral walls of the
pharynx, which gradually become longer and deeper till they
extend forward in close contiguity with the mesentery between
the collar and the pre-oral lobe. They have no connection
whatever with the epiblast. 2. At the same time as this
growth in length the cells undergo a remarkable metamor-
phosis into vacuolated tissue.

These organs will be compared to the notochord of the
higher Chordata. As regards previous observations upon
them, we may say at once that the majority of the numerous
workers on Actinotrocha were quite content to describe
them as “liver diverticula.”

Thus Wagener (20) refers to the “ liver-blind-darme,” and
Gegenbaur (9) as heaps of cells (liver-cells ?) : Caldwell (2)
remarks, “The stomach at its anterior end is produced into
one or two ventral processes. In the vacuolated walls of
these structures brown concretions are present” (p. 378).
Claparéde (5) described it as a dark mass with light balls
embedded in it, no doubt in allusion to the appearance in the
live animal, in which the dark areas are formed in the same
way as the black rim of an air-bubble under the microscope.
Schneider (18) accepts Claparéde’s interpretation, and gives
up his former view that they were fat globules, but does not
clear up the point any further; he remarks that the “ balls”
are surrounded in Actinotrocha with black pigment, in
allusion no doubt to the refractive effects above mentioned.

Metschnikoff (14) gives a figure (fig. 6) of a larva with four
pairs of arms, in which the vacuolisation of the antero-lateral

ON THE DIPLOCHORDA. 805

region of the ‘‘ stomach” has commenced, but the wall has
not yet begun to protrude. He refers to the area as the
“brown specks.”

Lastly, Wilson (22) figures these organs in several positions,
and refers to them as “ glandular lobes of the stomach.”

It wili thus be readily seen that the organs have been
repeatedly noticed by others, and a few sections could not
have failed to reveal their true structure.

At its posterior end the pharynx becomes constricted, and
then leads into the true stomach. The walls of this part con-
sist, for the most part, of a single layer of hypoblast; the
cells are cubical, with small cilia, and differ from the cells of
the pharynx in that the nuclei are not so regularly arranged
(Pl. 19, fig. 17). In the ventro-lateral region on each side
they lose their cell walls, and are heaped up into two
ridges or mounds extending the whole length of the stomach
(Pl. 19, fig. 17, d.a.). Hach of these ridges consists of an
amoeboid mass of nucleated cells without cell walls. They
project some distance into the lumen of the gut, and there are
always embedded in them the residue of various food-particles,
usually surrounded by a clear vacuole. ‘These ridges are their
special organs for intracellular digestion, and may be termed
the digestive areas. The mid-veutral wall of the gut between
them differs slightly from that of the rest of the stomach, the
cells being rather more columnar and numerous, and the
nuclei more regularly arranged. In addition to this a certain
portion of them, exactly opposite the “ diverticulum ” (PI. 20,
fig. 20, v. ch.), undergo a modification into vacuolated tissue
in all essential respects like that of the notochords; but it
does not apparently extend beyond the two-vacuole-deep stage.
This area of chordoid cells is evidently developed as a sup-
porting tissue to the parts in this region.

The stomach contracts to a small opening at its hind end
which leads into the intestine, a fairly long bent tube which
passes to the anus situated at the centre of the area sur-
rounded by the perianal band. The cells of the intestine
are single-layered, columnar, and ciliated, and call for no special

306 A. T. MASTERMAN.

mention. Jnst before the anus is reached the lumen of the
gut widens out into a larger vestibule, which may be termed
the rectum.

The alimentary canal thus consists of a wide mouth leading
? which is most likely epiblastic, and
should be more correctly termed the stomodeum ; the pharynx,
with its notochords, leading by a short cesophagus to the
stomach, with its two “ digestive areas; ’’ and the intestine,
with the distal end widened into a rectum and terminating in
a posterior anus.

into an “ cesophagus,’

Organs of the Mesoblast.

The mesoblastic structures of Actinotrocha are highly
developed, and indicate that this larva must be regarded in
its highest development as at a high morphological grade.
Thus there is little indication of “ mesenchyme,” and the
celom is present with splanchnic and somatic layers. In
addition to this the vascular and muscular systems are well
differentiated.

Celom.

The mesoblast is in the condition of a thin layer of proto-
plasm (Pl. 19, figs. 6-—10, and 15, &c.), with a few nuclei
dotted here and there, but at certain places this layer is thick-
ened by a massing together of mesoblast cells which form
simple contractile muscle-cells, and at others again are found
protoplasmic processes of the same nature, extending across
the body cavity. This thin layer of mesoblast is closely
applied to the hypoblast along the gut, except at certain places
where the blood lacunz are formed by the spaces left between
the two layers. On the outer wall the somatic mesoblast is
also in close contact with the epiblast, though between them is
a thin layer of mesoblastic chondroid tissue.

The ccelom is segmented into three parts,—the ccelomic cavity
of the pre-oral lobe, that of the collar, and that of the trunk.

The cavity of the pre-oral lobe (Pl. 19, fig. 6) is of the same
size and shape as the hood itself. Posteriorly it is produced

ON THE DIPLOCHORDA. 307

into two horns running back laterally (Pl. 20, fig. 24) to meet
the collar celom. Here the mesoblastic walls of each cavity
meet each other and form a well-marked mesentery, which is
of considerable firmness, and probably has in its centre a thin
lamina of chondroid tissue.

In the mid-dorsal line the posterior wall of the pre-oral celom
borders the subneural sinus (PI. 20, fig. 20, mes.) ; and just
where the pre-oral mesoblastic wall slopes away on either side
of the sinus (Pl. 21, fig. 42) there are a pair of thickenings, p. p.,
which, traced forwards, show themselves to be the commence-
ment of a pair of internal openings. Further forward they
become cellular tubes (PI. 20, fig. 28, p. p.), and after a short
course parallel to the body-wall they open into grooves on
either side of the nerve-ganglion. A cross-section of the tube
shows a single layer of columnar ciliated cells, closely similar
to the cross-section of the collar nephridium (PI. 21, fig. 41).
The pre-oral coelom, therefore, opens to the exterior by a pair
of ciliated pores, identical in position and structure with the
proboscis pores of Cephalodiscus (see below).

The pre-oral coelom is traversed in every direction by fine
protoplasmic filaments, with here and there, especially at the
unction of the threads, a nucleus (PI. 20, fig. 28, b. ¢., 1).

The collar coelom is a spacious cavity separated in front by
the mesentery (mes. in the figures) from the pre-oral ccelom,
and behind by the mesentery (mes’. in figs. 22 and 23) from the
trunk cavities. It is produced into each tentacle, and is every-
where bounded by typical coelomic epithelium. A dorsal
mesentery is present throughout their length, in connection
with which is a dorsal vessel. We may specially notice that
as the notochords grow forward they push the splanchnic
coelomic epithelium before them, and they are thus, like the
rest of the hypoblast, surrounded by a layer of mescblast, in
this case forming a mesoblastic sheath to the notochordal
diverticula (Pl. 21, figs. 30, 31, sh.). The mesentery (mes’.)
separating the collar coelom from that of the trunk starts just
posterior to the tentacles and collar ring, and runs forwards
and inwards to meet the gut. Its course is shown in trans-

308 A. T. MASTERMAN.

verse sections in PI. 19, figs. 16 and 17, and in longitudinal
sections in PI, 20, figs. 19—23 and 27, mes’.

At the ventral posterior part of this collar area are seen in
the view of the whole larva (fig. 17, c. 2., in section) a pair of
large bodies, apparently aggregations of ameeboid cells with
large nuclei. Their presence has been observed, figured, and
remarked upon by almost all the workers upon this larva, and
various interpretations have been put upon them. It has been
noticed that the masses break up at the metamorphosis, and
the constituent cells have in different cases been seen to float
freely in the cavity, pass into the tentacles, or into the vascular
trunks.

In fig. 17 there is seen nothing but a mass of protoplasm
and large nuclei, but in sections further back we notice that
the cells are grouped round a few intercellular canals which
open into the collar cavity by broad funnels, join each other,
and pass outwards and downwards between the two layers of
the collar-trunk mesentery (mes'.) to open to the exterior near
the mid-ventral line immediately on either side of the “ diver-
ticulum.”

This is the organ referred to as the “ nephridium” by
Caldwell (2), as follows :—‘‘ The pair of nephridia lie on either
side of the body, their numerous excretory cells floating freely
in the body-cavity in front of the septum. 'The external open-
ings are placed one on either side of the opening of the foot.’
He further describes in some detail the excretory cells, and
remarks, ‘‘ At no time during the free-swimming life of larva
does the excretory canal system open into the body cavity.”
This statement does not apply to the species [ have investi-
gated, and it is improbable that it does so to any other
Actinotroche. The internal funnels, without doubt, open
into the collar cavities, and not into the trunk cceelom, as do
the adult nephridia. Pl. 21, fig. 40, shows a longitudinal section
of one of these collar nephridia as constructed from a series of
sections. Each is suspended and held in position by a long
cord of mesoblastic cells running from the front wall of the
collar. A trausverse section of the duct is seen (fig. 41) to

ON THE DIPLOCHORDA. 309

consist of a single layer of columnar ciliated cells enclosing a
fine lumen.

Wagener (20) figures one of these organs, and describes its
appearance in the live larva. He says that “from its free
broad sides there project many small stumpy pointed knobs on
long stalks, which spread about in all directions like a bouquet
of flowers” (translation). This description plainly indicates
that the mass of cells seen in prepared sections around the
tubes is formed by a number of cells attached by long pro-
cesses to the mouth of the funnels. In fact, a few preserved
specimens give indications of this arrangement. We thus
arrive at a pair of collar nephridia resembling almost identi-
cally the structure of the nephridia as described by Boveri (1a)
in Amphioxus. <A point which cannot be fully entered
into here is the fate of these collar nephridia, so evidently homo-
logous with the collar pores of Cephalodiscus and Balano-
glossus. If they become the nephridia of the adult the open-
ings into the collar region must be lost, and fresh openings
must be acquired into the trunk ceelom. Caldwell (2) says,
“The large canals remain as the paired nephridia of
the adult.” We may, however, recollect that the adult
nephridia open on the dorsal side and the collar nephridia on
the ventral side. Further than this, there are described below
other organs, which may possibly be the rudiments of the true
nephridia.

Trunk Celom.—The celom of the trunk has the same
characters as that of the collar. It is continued dorsally
forwards for some way almost to the level of the notochords.
There is apparently no dorsal mesentery, but a ventral me-
sentery is continued throughout the length of the trunk, and
suspends the intestine. At the perianal area are a pair of
organs which I have not fully made out, but they may be the
rudiments of the trunk nephridia. They lie in the hemocele
space immediately below the ciliated band, and are thin-
walled. They have an internal opening into the trunk celom,
and apparently open to the exterior on either side of the anus.
Their walls are of the same nature as those of the celom,

310 A. T. MASTERMAN.

and appear to be portions of the celom, with the trunk
nephridial tubes cut off from the rest of the trunk ccelom.

These two organs are seen in a longitudinal section in
Pl. 21, fig. 45. A study of other stages can alone show their
true significance.

Skeletal System.—The skeletal elements are difficult of
demonstration, but a very thin layer of mesoblastic chondroid
tissue, homologous with that of the adult, appears to be pre-
sent immediately under the epiblast. In carefully macerated
specimens the epiblast of the tentacles can be removed, and
the skeletal system is then evident as a thin hyaline layer. In
the trunk region, again, it is difficult to account for the rigidity
of the body-wall, considering its extremely attenuated epiblastic
and thin mesoblastic layers, without the assumption of the
skeletal supporting layer.

Muscular System.—The muscles are all in the condition
of simple mesoblastic cells with long contractile processes.
They are most prominent in the interior of the hood, where
they form a fine meshwork. The thickest and most regularly
arranged are inserted in the mesentery, between the pre-oral-
and collar-cavities, and radiate outwards on either side to the
edge of the hood (figs. 23 and 50). A few strands also connect
the further side of this mesentery with the sheath of the
notochord. In the collar there are two muscular bands,
already referred to, attaching the collar nephridia to the collar
wall, and there are a few muscle-cells forming a sheath around
the ventral blood-vessel (Pl. 19, fig. 17). In the trunk region a
few fine circular muscle filaments can be discerned in surface
view, and a dorsal longitudinal trunk (fig. 16). The trunk
mesoblast, forming a sheath round the “ diverticulum” (“ foot,”
Caldwell), is very thick, and mesoblast cells are becoming
differentiated into true muscular elements (PI. 20, fig. 19, div.).

The splanchnic mesoblast over the dorsal blood-vessel is
thickened by an accumulation of simple contractile cells (figs.
16, 17, and 20, d. d. v.). The contractile nature of the dorsal
blood-vessel has been noticed by many workers. ‘Thus
Schneider found it to be rhythmically contractile, and remarked

ON THE DIPLOCHORDA. ott

upon the presence of transverse lines due to the contraction.
He compared the structure of the muscle-cells to that of those
lying over the “ diverticulum.”

Blood System.—The vascular system has received a great
deal of attention by former observers. Wagener’s first obser-
vations were corrected in some particulars by Schneider. The
latter found a dorsal vessel on the stomach with indications of
its double nature. He also described a number of blind sacs
which, as finger-like processes, project into the coelom at the
junction of stomach and intestine. Their presence had been
noticed prior to this by Leuckart and Pagenstecher.

Schneider, along with some others, states that the blood-
corpuscles arise from the masses of cells referred to above in
connection with the collar nephridia.

Metschnikoff (14) states the vascular system to be in direct
communication with the body cavity, apparently as an inference
from the fact that these masses of cells, at first in the body
cavity, are found later in the blood-sinuses. Krohn, again,
gives a description of the dorsal and ventral trunks and
the cxcal vessels, he observed the contraction and expan-
sion of the cecal vessels; and Wilson (22) figures the main
features of the vascular system as determined by his prede-
cessors.

Caldwell (2) confirmed the results of his own work and that
of previous workers as follows :—“ 1. Blood-corpuscles aggre-
gated in two or more masses, lying free in the body cavity of
the pre-oral lobe, i.e. in front of the septum. 2. A blood-
vessel formed on the dorsal wall of the stomach—a marked
structure in the larva. 38. The splanchnopleure sac, which in
the region of the stomach forms a loose sac surrounding the
gut. 4. Cecal prolongations of this sac. 5, Cecal prolonga-
tions into the rudiments of the adult tentacles.”

I have been enabled not only to confirm some of the above
observations, but to complete the account of the blood system.
It consists essentially of a system of sinuses and fissures lying
outside the mesoblast and between it and the epiblast or hypo-
blast. These may be regarded as vestiges of the segmentation

312 A. T. MASTERMAN.

cavity. In every case in which a true vessel is present it is
formed by the thickened mesoblastic walls surrounding the
vessels.

The vascular system is, therefore, morphologically in a very
primitive condition.

There is just under the nerve-ganglion a large sinus caused
by a want of contiguity between the mesoblastic walls of the
pre-oral cavity in front and the collar cavities behind. This
sinus, which may be termed the subneural sinus (s.”.s.), 18
seen in figs. 1, 3, 20, 23, and 38, in which its relationships can
be readily made out. Into its cavity projects upwards from
below the inner end of the subneural gland, so that the
lining cells of this organ are in direct contact with the hemal
fluid. At the posterior extremity of the sinus it leads by a
chink or fissure between the gut wall and that of the two
collar cavities, along the dorsal wall of the cesophagus (figs. 7,
8, d. 6.v.) to the hind extremity of the collar region. Here it
falls into the large dorsal vessel with contractile walls. This
vessel is continued along the dorsal wall of the pharynx and
stomach (PI. 19, figs. 16 and 17, d. db. v.) till it meets the intes-
tine. Here a small ring-sinus is formed connecting it with the
ventral vessel with cecal prolongations into the body cavity.
At the anterior extremity of the dorsal vessel, at the front
end of the pharynx, two lateral trunks are given off, which,
passing downwards round the base of the cesophagus and
between this and the notochords, meet again in the mid-ventral
line, forming a post-oral ring sinus (PI. 20, fig. 21,7.s.). From
the point of junction (fig. 20, v. d. v.) there originates a mid-
ventral vessel which may be traced throughout the length of
the gut (figs. 16 and 17, v.d.v.) into a large hemal ring
immediately internal to the perianal ciliated band. Lastly,
there are some not very distinct indications of hzmal sinuses
passing down the tentacles; but these are not very decided.

This blood system is closely comparable to that of Balano-
glossus and of Cephalodiscus, as I hope to show later.

Caldwell makes the statement, ‘‘ Free communication exists
between the body cavity in front of the septum and the split

ON THE DIPLOCHORDA. Sle

in the splanchnopleure, which will form the blood-sinus and
vessels of the adult.” I can find no confirmation of this
statement ; on the contrary, I find the vascular system a com-
pletely closed system of sinuses, bounded in various parts by
epiblast, mesoblast, or hypoblast, but in no way connected with
the coelomic spaces.

GENERAL CONSIDERATIONS.

In the foregoing pages has been given a brief account of the
structure of the larval Phoronis, the investigation of which
was undertaken mainly to assist in deciding the true systematic
position of this group. As before stated (24, 25, 26), the facts
clearly point to a close genetic relationship with Balano-
glossus and Cephalodiscus, especially the latter.

We therefore have, firstly, to institute a comparison of
Actinotrocha and Phoronis with these; and secondly, to
see whether, this comparison being held valid, the anatomical
facts will throw any further light upon the relationship of
these groups collectively to others above and below them in
the morphological scale.

Comparison of Phoronis and Balanoglossus.

Under this head we may compare (1) Actinotrocha with
Balanoglossus, (2) Actinotrocha with Tornaria, and
(3) Phoronis with Balanoglossus.

(1) Actinotrocha and Balanoglossus.

General Shape of Body.—In each case we have an
organism of an elongated shape, and showing clearly external
indications of a division into three segments, following one
another from before backwards. One of these, the pre-oral
lobe, is situated in front of the mouth, and forms a prominent
organ overhanging the oral aperture. It is narrowed at the
base, presenting a stalk in Balanoglossus, and a contracted
portion bounded by the atrial grooves in Actinotrocha (ef,

314 A. T. MASTERMAN.

Pl. 22, figs. 50 and 51). This is followed by a post-oral segment,
which in Actinotrocha is produced into tentacles, and in
Balanoglossus is not. This at first sight is a marked dif-
ference, but the fact that the collar region in Cephalo-
discus is produced into tentacles, and also that in some
Tornariz tentacular processes are found post-orally as well
as pre-orally, must be taken into consideration, and the secon-
darily acquired adult habitat of Balanoglossus renders an
atrophy of tentacular processes at least a plausible hypothesis.
Behind this is the third portion of the body or “trunk,” an
elongated cylindrical segment with a terminal (or sub-
terminal) anus.

These segments are, in each type, covered with a unicellular
layer of epiblast or ectoderm, which is for the most part
ciliated and glandular.

Nervous System.—The nervous system is in both cases
of a primitive type. The inner ends of the outer layer cells
form a fine diffused network of fibres underlying the whole
surface, but special concentrations can be distinguished in
various parts.

In both Actinotrocha and Balanoglossus (Pl. 22, figs.
46, 47) the chief nervous concentrations are found in the dorsal
region of the collar, and form a thick nervous mass extending
throughout this region. The differences are of degree only,
for in Actinotrocha the dorsal collar region is narrowed
from before backwards, and with it the nervous area, and the
more or less hollow tubular condition in Balanoglossus is
seen in a more nascent condition in the neuropore of Actino-
trocha (cf. figs. 48, 49). In each there extend forwards
from the main ganglion, on to the pre-oral lobe, three main
nerve-tracts, one median dorsal (in the condition of three
secondary groups in Actinotrocha) and two lateral bands,
skirting the lateral edges of the pre-oral segment and meeting
in front. |

Posteriorly the collar nervous area gives off a pair of lateral
branches which encircle the extreme hind edge of the collar,
with a nervous ring, and from the mid-ventral point of this

ON THE DIPLOCHORDA. 315

ring are given off fibres forming a ventral nervous tract in the
trunk. In addition there is a mid-dorsal nervous tract along
the trunk, concentrated into a cord in Balanoglossus, in a
more diffused condition in Actinotrocha. Lastly, there is
the perianal nervous ring in the latter.

It is difficult to follow this comparison with the semi-dia-
grammatic figures (figs. 46, 47), and to account for the resem-
blances as accidental or homoplastic.

Alimentary System.—We have seen that there is some
reason for suspecting the presence of a stomodeum in Acti-
notrocha, whereas, so far as evidence goes, such is scarcely
present in Balanoglossus. Again, the hypoblastic system
is complicated in the latter, by a forward movement of certain
of the organs, into the proboscis, giving them a secondary
pre-oral position. Thus it cannot be denied that the noto-
chord is secondarily displaced into the proboscis, probably in
connection with a burrowing habit. The proboscis is pri-
marily pre-oral, and therefore de facto has no hypoblast in its
cavity ; any hypoblastic organ must, therefore, invade it second-
arily. Thus Bateson states that the notochord arises in the
collar region and moves forward later in development.

This moving forward phyletically of the notochord must
necessarily involve the displacement of other structures, and
together with the absence of a stomodzeum must be allowed
for in our comparison.

At the front end of the pharynx in Actinotrocha area
pair of organs which have been described and referred to above
as “ notochords.”

The presence of a “ notochord” has always been so insisted
upon as an essential character of the Chordata that at first
sight one might be led to suppose that the proving of the
identity of these organs of Actinotrocha with the notochord
of Balanoglossus is a sine qua non for the assumption of
their close genetic connection. This does not, however,
appear to be the case, for the anatomical likeness is so marked,
quite apart from the ‘ notochord” question, that it merely
serves to form an additional reason for regarding these latter as

316 A. T. MASTERMAN.

of notochordal value. However, the question of the actual
homology of these organs has been carefully investigated, and
the balance of evidence seems distinctly in favour of regarding
them as the primitive paired representatives of the notochord
of Balanoglossus.

There can be no question of their analogy to notochords.
Their histological structure and their relationship to the
pre-oral mesentery indicate a function of support, as in the
case of the true notochord. They are undoubtedly formed
from hypoblastic tissue, and arise normally as diverticula of
the primitive gut wall. They also arise in the front end of
the collar region, a position identical with that of Balano-
glossus. The point of difference, therefore, resolves itself
into this,—that there are in Actinotrocha two lateral
organs, whereas in all the other Chordata, as far as is known,
there is but one median dorsal notochord.

If we can assume a fusion of two lateral organs in the
median line as having taken place, the change from lateral to
dorsal would follow asa matter of course, so that the difference
resolves itself into the difference between ‘‘ paired” and
“median unpaired.”

To any one who can appreciate the principles upon which a
bilaterally symmetrical animal is constructed this difference will
not bear any importance. Indeed, it is more difficult to con-
ceive in such an animal of the presence of an organ which
arises primarily in a strictly unpaired condition ; and it is at
least a defensible hypothesis that in the higher Vertebrata
every asymmetrical unpaired organ arose by atrophy of its
fellow, and every symmetrical one either by survival from a
radially symmetrical condition or by fusion, in the middle
line, of paired rudiments.

The atrium in Tunicata arises by paired lateral rudiments,
in Amphioxus by a median ventral invagination, in spite of
which difference we do not find the homology of the two called
in question, and one might multiply instances from the study

1 The two lateral notochords of Cephalodiscus were found subsequently
to the writing of this statement.

ON THE DIPLOCHORDA. 317

of the arteries, veins, central nervous axis, skull, swimming
bladder, genitalia, &c.

Scarcely any process in the phylogeny of the higher animals
is more clearly indicated than this, unless it be the ease with
which primitively unpaired organs may become impressed with
bilateral symmetry.

In the case of the notochord itself we may remark that in
Balanoglossus it tails offin the hind region (collar) into paired
rudiments, which are more marked in the young stages. The
figures of Morgan (17) in this connection are very striking, and
we cannot explain the fact that the part of the notochord which
is least displaced from its primitive position is found in the paired
condition, otherwise than as indicating a paired phyletic origin.

It may be urged that if the notochord of the Chordata
arose as a paired organ there should be indications of this in
the ontogeny of the higher types.

I think there is one reason why the vertebrate notochord
does not as a rule show traces of the paired condition, and
this is that in the higher Chordata the development of the
notochord is greatly hastened. Thus in Amphioxus and in
several higher types the rudiment of the notochord is pinched
off from the gut at quite as early a stage as are the mesoblastic
pouches. A transverse section of a larval Amphioxus at a
suitable stage shows the pair of dorso-lateral mesoblastic
pouches still in continuity with the gut wall, and in the median
line the notochordal rudiment is in the same condition.

So much is this the case in some Vertebrata, e.g. the
chick, that certain observers have claimed that the notochord
arises primitively as a median rod of mesoblast.

A careful study of the more primitive forms, however,
clearly shows that the notochord naturally arises from the
hypoblast which remains after the mesoblast has been de-
veloped from it (secondary endoderm). Without offering any
explanation of this precocious development of the notochord,
it is plain that, with such ontogenetic conditions, the retention
of a phyletic process of development from paired rudiments is
impossible. The mesoblastic pouches force the notochordal

318 A, T. MASTERMAN,

rudiments into the median line from their very inception,
though even in Amphioxus the organ is formed by the
“ dovetailing”’ together of two lateral rows of cells.

I am indebted to Dr. Gadow for calling my attention to the
work of Mitsukuri (15a) on the ontogeny of Emys, in which
he has described paired rudiments of the notochord, which he
styles “hemichords,”’ and which fuse together in the median
dorsal line to form the notochord. The observation is signi-
ficant in the light of this discussion.!

There is one other possible interpretation of these ‘‘ noto-
chords” of Actinotrocha, namely, that of a pair of abortive
gill-slits. In the light of the work on Cephalodiscus which
follows, it will be seen that this is perhaps a tenable hypo-
thesis, but it will be discussed under that head.

The significance of the ventral chordoid tissue will be
referred to later. Assuming, then, that these organs may be re-
garded as primitive paired rudiments of the Chordate notochord,
I have depicted for the further comparison of the two types
one of these notochords in fig. 49, as bent into the mid-dorsal
line, in order to have it included in a median sagittal section.

In fig. 48 is seen a similar section of Balanoglossus, com-
piled, in a semi-diagrammatic way, from the works of Spengel
(19a), Kohler (12a), and Morgan (17). A comparison of the
two will show the way in which it is here suggested that the
notochord of Balanoglossus has moved forwards. In doing
this it has come into peculiar relation with the pre-oral
celom, which will be referred to again.

We have in Balanoglossus the “ proboscis vesicle” of
Morgan (17) (‘sac of proboscis gland” of Bateson[1]and Kohler
[12a], Herz-blase” of Spengel [19a]), with its relationship to the
front end of the dorsal blood-sinus and to the pre-oral ceelom,
and in Actinotrocha the subneural gland with precisely
similar relationships to these parts. This is even more clearly

1 Davidoff (7a) has shown that in certain Tunicata the notochord and
nervous system arise from paired rudiments, and Brooks (1b) finds a similar
origin for the eleoblast of Salpa. These facts may have more phyletic
significance than Brooks is inclined to allow.

ON THE DIPLOGCHORDA. 319

shown in coronal sections of the two types (figs. 50 and 51). »
In each case the organ has precisely similar relationships to the
pre-oral coelom, collar ccelom, and subneural sinus (or heart),
though there is a striking difference in the fact that the sub-
neural gland has a duct to the exterior, whilst the proboscis
vesicle has none in the adult, though it arises in close contact
with the epiblast, even if not actually from this layer. We
have assumed a forward movement of the notochord in Balano-
glossus, and this movement must have considerably altered
the environment of the proboscis vesicle. It is not incon-
ceivable that this organ might be carried forwards and lose its
connection with the exterior, as is shown in fig. 48.

The subneural gland is undoubtedly epiblastic in origin,
and arises in the mid-ventral line, whereas there is considerable
difference of opinion regarding the origin of the proboscis
vesicle. Spengel (19a) maintains that it is epiblastic in origin,
and Morgan (17) is inclined to believe that it is mesoblastic.
Morgan’s figures show the early rudiment of the vesicle in
close contact with the epiblast; and I think that although
Morgan (17) interprets the organ as formed from mesenchyma,
there is as much testimony in his figures to its epiblastic
origin as is found in the case of the segmental duct of Verte-
brata and the nerve-cord of Teleostei. Although the organ
in Actinotrocha arises ventrally, it comes into connection
with the dorsal sinus and other parts in such a way that it
would not be surprising to find the mid-ventral invagination
method in this type altered to the modified process of develop-
ment in Balanoglossus. With the forward movement of the
notochord a mid-ventral development of the proboscis vesicle
would be an ontogenetic process of great difficulty. (We might
here compare Willey’s [23] suggestion of the alteration in posi-
tion of the mouth of Amphioxus, correlated with a forward
growth of the notochord.) The structure of Cephalodiscus
is especially instructive with regard to the subneural gland.

Gill-slits.—I have not as yet been able to find any trace
of gill-slits or their homologues in Phoronis or Actino-
trocha, though their analogues are present in each case,

vot. 40, PART 2.—NEW SER. Z

320 A. T. MASTERMAN,

represented in Actinotrocha by the atrial grooves already
described.

Stomach.—After the pharynx is found the digestive por-
tion of the stomach; and though the intracellular digestive
areas may be probably regarded as the first rudiments of the
hepatic diverticula of Balanoglossus, this suggestion cannot
be enlarged upon here.

Intestine and Anus.—An intestine suspended by a ventral
mesentery terminates in each case in a posterior anus.

Mesodermic Organs.

Vascular System.—As in the case of the nervous system,
the vascular tissues in both Balanoglossus and Actino-
trocha are in what is undoubtedly a primitive condition.
The veins and arteries, in fact, are merely sinuses and chinks
between the mesoderm on the one hand and the ectoderm or
endoderm on the other.

In each case (figs. 50, 51) there is a large sinus (the sub-
neural sinus[s. 2. s.] of Actinotrocha, the heart [Aé.] of Bala-
noglossus), which lies just below the junction of pre-oral lobe
and collar, a position with a special significance. This is con-
tinued along the dorsal endodermic wall between the collar
walls as a small fissure into a larger sinus, with more or less
muscular walls, differentiated from the trunk celomic walls,
which are produced forward (perihzemal, Spengel). In each
case there is a ring sinus in the collar-region which is con-
tinuous with a ventral median sinus below the gut. Such
resemblances, in so plastic and adaptable a system as the
vascular, speak for themselves (cf. figs. 48 and 49).

Celom.—The condition of the celom in Actinotrocha
bears so remarkable a resemblance to that of Balanoglossus
and Cephalodiscus that a genetic relationship might be
claimed upon this count alone.

‘he pre-oral lobe is in each case lined internally by the
mesodermic wall of the pre-oral cclom, the cavity of which
is itself partially filled by muscular tissue, passing in various
directions, though in each case the main strands can be

ON THE DIPLOCHORDA. 321

traced from the lateral walls in a converging manner to the
posterior angles, where in Actinotrocha they are inserted
in the “ pre-oral collar ” mesentery, and in Balanoglossus in
the sub-notochordal skeletal rod. The arrangement of these
muscles will be shown to be in Cephalodiscus identical
with that of Actinotrocha. In Actinotrocha there
are two proboscis pores opening on either side of the sub-
neural sinus (fig. 51) in identically the same manner as in
Cephalodiscus. In Balanoglossus the pores are some-
what further away from the middle line, but in an homologous
position. The collar-cavities are closely similar, with a ten-
dency in each case to a production of the dorsal part forwards
into the neck. The collar pores have already been compared
to the collar nephridia of Actinotrocha. The relationships
of the collar coelomic walls to the vascular system and the gut
are practically identical.

The trunk cavities are continuous dorsally, and ventrally
they form a mesentery. From their walls are derived in each
case the gonads (Phoronis). The front dorsal part of the
trunk ccelom is produced into a pair of perihzmal spaces,
embracing the dorsal blood-vessel.

We thus see that not only is there in both types a segmen-
tation of the celom into five pouches, one pre-oral and
unpaired (leaving out of consideration for the present the
possible coelomic value of the proboscis vesicle), and two pairs
of post-oral pouches, but these pouches are similar, to a large
extent, in their anatomical relationships, though perhaps the
presence of tentacles in Actinotrocha tends to make the
resemblance to Cephalodiscus even closer.

Thus a very general and brief comparison of the nervous,
alimentary, vascular, skeletal, and coelomic systems brings out
a close agreement, the genetic origin and value of which can
hardly be denied.

(2) Actinotrocha and Tornaria.

The comparison of these two larve is interesting in that it
leads one to the conclusion that the fully developed Actino-

322 A. T. MASTERMAN.

trocha is more nearly allied to Balanoglossus than to its
larval form, Tornaria. Thus a closer agreement would pro-
bably be found between the latter and an earlier stage of
Actinotrocha than is here described. There are, however,
certain general points of similarity which bear out the con-
clusions already arrived at in the former comparison of
Balanoglossus and Actinotrocha.

Each has a mouth overhung by a large pre-oral hood, at the
posterior mid-dorsal border of which is a thickened nervous
area. From this runs a pre-oral ciliated band round the edge
of the hood, and a post-oral band behind the mouth. In
Tornaria the bands are more convoluted, but the arrangement
is well seen in Stage tv, figured by Morgan (17). The terminal
anus is surrounded by a powerful perianal band, which has
long lash-like cilia, and forms the main locomotive organ of
the body (cf. 26).

The condition of the ccelom is similar, though Tornaria is
less developed in this respect. It, however, shows an unpaired
pre-oral ccelom with a dorsal pore, a pair of collar cocloms, and
a pair of trunk cavities. In the course of development the
trunk region is in each case greatly elongated.

(3) (The Comparison of Phoronis with Balanoglossus
must be left to another Paper.')

We may now inquire if the acknowledgment of Phoronis
as closely connected with the chordate stock sheds any further
light upon the origin of the Chordata, or of organs peculiar
to this group.

As regards the notochord a very primitive condition is met
with. In Actinotrocha the notochords are little more
than vacuolated areas of the gut wall. Their lumen is in
continuity with that of the gut, and the process of vacuolisation
is very simple and instructive. Each cell can be observed to
be gradually filled with turgid vacuoles, and these are not

1 We may state that the differences in habitat and their resulting modifi-

cations obscure most resemblances in protomere and mesomere, but that the
arrangement of the parts in the metamere or trunk is closely similar. (See 24.)

ON THE DIPLOCHORDA,. 323

further complicated by fusions. Just as in tracing down, e. g.,
the digestive organs into lower types, we find that stages are
reached in which the function is more diffuse, and is confined
only by the extent of the hypoblast layer (indeed, not even by
this in Coelenterata), so in tracing down the hypoblastic
skeletal function we find it in the Diplochorda distributed
over certain areas in which it is no longer found in higher
types. Thus in Actinotrocha it is found in the mid-ventral
wall (in the form of chordoid tissue) exactly at the spot where
most support is required, in addition to the normal area of
chordoid tissue in the collar region.

Again, in Cephalodiscus, chordoid supporting tissue is
developed round the pharyngeal clefts, where, again, it is
obvious that support is specially essential (see next paper).
It seems that in the Diplochorda the supporting function of
the hypoblast has been active, and chordoid tissue developed in
whatever position was requisite for the needs of the organism.

The development of turgid vacuoles for support seems to be
the special character of the hypoblast (or endoderm), and is
also found in the Coelenterata, e.g. in the tentacles of
certain Hydroids and in some larve, such as that of Lucer-
naria (R. S. Bergh). We may suppose that a feature of the
chordate stock is the retention of this ability to form hy po-
blastic skeletal tissue by vacuolisation.

The subneural gland is apparently to be compared with that
of Tunicata bearing the same name; and if we assume that
the line from subneural gland to nerve-ganglion marks the
posterior boundary of the pre-oral lobe, we may be helped in
the comparison of these Urochorda with other Chordata.
(I see no sound reason for holding that the large suboral pro-
cess with papillze for attachment, found in the larvie of sedentary
Ascidians, is the pre-oral lobe, but would rather regard it as an
hypertrophied ventral sucker belonging to the post-oral region.)

Lastly, this organ may be compared with the hypophysis of
Vertebrata, and the relationship of the latter to the front
end of the notochord and to the subneural mesodermic tissue
would thereby receive a ready explanation.

324 A. T. MASTERMAN.

The notochord must be regarded as arising in the collar-
region, and the prolongation forward of this organ in Am-
phioxus is usually acknowledged to be a secondary adapta-
tion, as is also the pre-oral position of Balanoglossus.

The true Vertebrata, primarily free from a burrowing habit,
have probably retained at least the front end of the notochord
in its primitive position. The front end of the notochord,
therefore, indicating the anterior collar-region, the hypo-
physis is in the same relation to it as the subneural gland in
Actinotrocha. In the former case its inner end comes into
close relation with mesodermic vascular tissue lying close under
the brain, and in the latter case the subneural gland has similar
relations with the subneural sinus under the main ganglion. It
is even possible that the base of the “ neuropore” might be com-
pared with the infundibulum. Asin the case of the Tunicata,
it is probable that the line from brain (infundibulum) to
hypophysis marks the posterior boundary of the pre-oral lobe.

It is not so safe to attempt to draw the posterior boundary
of the collar-region. In both the Cephalochorda and
Vertebrata there can be no doubt that the trunk region is
hypertrophied, and also that the position of the anus has
changed, perhaps more than once. At the same time, in these
forms and in the Urochorda there are always traces of
hypoblast other than the notochord (post-anal gut, &c.), which
indicate that the prolongation of the notochord in a posterior
direction was part of the process of a movement of the whole
gut towards the posterior end of the body. There is no
difficulty in reconciling the relations of the notochord to gut
and blastopore in the embryos of Vertebrata, and the idea
of a primitive condition of collar notochord. In fact, the
elongation of the notochord posteriorly by a proliferation of
cells at the anterior end of the blastopore is readily interpreted
by this assumption.

Again, a careful consideration of the structure of Actino-
trocha may make it possible to trace the chordate stock
practically to the Colenterate stage, or at least to a rudi-
mentary triploblastic type.

ON THE DIPLOCHORDA. 325

Rather than work back to simpler organisms, I would
attempt to follow the indicated course of evolution from simple
to complex.

In woodcuts 6, 7, and 8 I have depicted the leading characters
of Stage 1, a pelagic, triploblastic, radially symmetrical ancestor

6. Ventral view of Stage I. 7. Ventral view of ditto. 8. Lateral view of
ditto. N.B.—In this and succeeding woodcuts the general ectoderm is
unshaded, the endoderm shaded, nervous areas black, and mesoderm
dotted. The ecelomic pouches are numbered.

of the Coelomata, or at least of those Cceelomata to which

I wish to refer. With four ccelomic pouches it is quite easy

to imagine this organism as derived from a pelagic Coelenterate

with four gastric pouches. The origin of the radial symmetry
must be assigned to the conditions of free pelagic (floating)
existence, and the presence of the apical nervous system
precludes a sedentary origin. The environment, in a hori-
zontal plane, of a free pelagic organism with only perpendicular

automatic locomotion should be as capable of impressing a

radial symmetry upon an organism as that of a sedentary

habit.

We may note that there is an apical ganglion with four
radial nerves, meeting a nerve-ring round the mouth.

Below the nerve-ganglion is a subneural sinus which com-

326 A. T. MASTERMAN.

municates with a system of interccelomic sinuses, the primitive
vascular system.

The four coelomic pouches all retain the muscular, repro-
ductive, and nutritive functions appertaining to the primitive
celom.

The axis of symmetry is perpendicular, and the central
nerve-ganglion is situated at the termination of this axis in the
direction of motion.

I would specially emphasise the fact that the mesoblastic
pouches are evolved in this stage (Protaxonia, Hatschek),
so that their segmentation into four is due to radial deter-
mining factors, and it follows from this that they cannot strictly
be regarded as “ paired.”” This segmentation of the mesoderm,
for reasons given below, we may term the “‘ archimeric
segmentation.”

With horizontal motion forwards the axis of symmetry is
replaced by a perpendicular plane of symmetry in the direction
of motion. The Heteraxonia (Hatschek) stage is now com-
menced, and, as has been shown by Sedgwick, the blastopore
elongates and forms mouth and anus.

Under these conditions the mesoblastic ‘ archimeres” are
placed under fresh symmetry-determining factors, and they
are now no longer under the same environmental conditions.
In woodcuts 9, 10, and 11 are seen three views of this stage
(Stage 11). The new mouth and anus are almost completed,
and it is now seen that one of the archimeres (1) is pre-oral
in position, two are lateral (2, 2’), and the last (3) is now
post-anal.

The first of these (the protomere) is now subjected to the
environment incidental to the front end of the body and to
the pre-oral position. No longer in a suitable position with
respect to the digestive part of the gut, it will lose its nutritive
and digestive function, which implies also a loss of that of
reproduction. In other words, its vegetative functions will be
lost whilst its animal functions will be markedly developed.
This protomere is evidently homologous with the pre-oral
lobe, hood, proboscis, &c., of so many of the Celomata. It

ON THE DIPLOOCHORDA. eM

appears to be destined to play a very secondary part in the
higher Ceelomata in any case. In animals which take up a
sedentary existence it is no longer required to any great extent,

9. Ventral view of Stage II. 10. Dorsal view of ditto. 11. Lateral
view of ditto.
for the “animal” functions are at a low ebb. Again, in those
which do not thus retire from the arena of pelagic or littoral
life it is replaced by the shifting forwards of the “ metameric”’
segments from behind.

The cavities (2, 2’) of the two lateral segments are the most
suited, by their position, for performing the functions of
digestion and excretion ; and at the same time their position
immediately behind the mouth causes these segments to be
adapted (by tentacles, &c.) for ingestive functions. They may
be termed the ‘“‘ mesomeres,”’ and are recognised in Hemi-
chorda as the collar-segment. The last (3) or post-anal is
only found in this position in rare cases, for the next stage
is soon reached.

The nerve-ganglion is seen to be making its way forwards

328 A. T. MASTERMAN.

to the new animal apex, and with it the radial nerve-tracts
are altered somewhat in position, as also are the ccelomic
pouches, in correlation to the forward motion of the sub-
neural sinus.

In the next stage, Stage 111 (woodcuts 12, 13, and 14), further
changes are indicated. The nervous system is somewhat

13.

12. Ventral view of Stage III. 18. Dorsal view of ditto. 14. Lateral
view of ditto.

altered. The nerve-ganglion has still progressed forwards,
and it is seen that the longitudinal nerve-ring and radial
nerves are now tending to form a secondary system of rings
in planes at right angles to the axis of motion. Thus two
radial nerves and part of the primitive ring form the “ pre-
oral ring,” two radial nerves form the ‘‘ post-oral ring,” and
part of the primitive ring forms the “perianal ring.” In
other words, these rings are formed to correspond to the three
archimeric segments—protomeric, mesomeric, and metameric.

ON THE DIPLOCHORDA. 329

We note that the anus has travelled to the posterior end,
in obedience to general laws of symmetry.

This change from radial to bilateral symmetry, though
hypothetical, is paralleled in a remarkable manner in the
Echinoids, and the above woodcuts will forcibly recall the
transitional symmetry of such forms as Spatangus.

The change in position of the anus is important, for it
places the metameres (3, 3’), now separated by the alimentary
canal into two, in an advantageous position in regard to
the processes of nutrition, and hence we find that not only
do these metameric coelomic cavities early monopolise the
reproductive function to the exclusion of the protomere and
the mesomeres, but they elongate enormously ; and in them
is instituted in higher forms a secondary bilateral seg-
mentation, which is already termed “metameric.” The
study of the segmented forms leads one to believe that the
metameric segments grow forward, at least the cclomic
elements in them, and to a large extent replace the archimeric
protomere, and mesomeres.

The stage we have now reached is practically identical with
that exhibited by Actinotrocha (without taking into con-
sideration the notochordal rudiments). I need not go into a
detailed comparison of the two types,—Actinotrocha is so
closely represented by this stage that its study suggested the
hypothesis here laid down. Woodcuts 12, 13, and 14 would
almost pass for diagrammatic representations of Actino-
trocha. These are the bare outlines of a suggested ancestry of
the Chordata, tracing the lines of evolution nearly to the
Coelenterata. We must assume that the archimeric seg-
mentation is predominant till above the Hemichorda,
and that in Amphioxus the metameric segmentation comes
into existence, and the protomere and mesomeres become
of secondary importance or disappear altogether. I find, in
looking at the literature of Balanoglossus, that Morgan
has suggested that the condition of the mesoderm in Am-
phioxus is arrived at through secondary segmentation
of the trunk of Balanoglossus, so that I must adopt

3800 A. I. MASTERMAN.

his hypothesis, and put no claim to originality in this par-
ticular point.

If the metameric or secondary segmentation took place
between the stages of Balanoglossus and Amphioxus it is
evident that, as Bateson has emphasised, the “segmented ”
Invertebrata and the segmented Vertebrata must be
genetically distinct, in spite of the most elaborate anatomical
resemblances. The recognition of this archimeric segmenta-
tion and its relationships to the secondary or “ metameric”
segmentation will, I venture to think, not only clear up many
difficulties in tracing the origin of Vertebrata, but will
furnish a sound basis for the phyletic classification of the
Celomata.

Although the archimeric segmentation would form the basis
of the group Archicelomata, here suggested, a primitive
condition of the other systems is characteristic. Thus we might
state the following characters of this group:

1. The body is divided into three archimeric segments, the
protomere, the paired mesomeres, and the metamere, either
throughout life or in early stages,! when the relative promi-
nence of one or more of these may be altered.

2. The metamere is never definitely divided up by a secondary
bilateral segmentation, usually known as “ metameric” seg-
mentation.

3. The nervous system is usually in a primitive condition,
and is still in continuity with the ectoderm throughout life.

1 The coronal section of Bateson’s early larva of Balanoglossus clearly
shows indications of primitive equivalent radial value of the four pouches
though the mesomeric elongate very early. In ontogeny one need not
expect the four pouches to, in every type, be separately invaginated from the
gut. A difference in the relative time of development of different parts would
readily account for the variations in this respect seen in the different groups
of Echinodermata and the Chetognatha, though the separate invagina-
tion of each from the hypoblast must be regarded as the more primitive con-
dition, just as the elongation of blastopore to join both mouth and anus is
regarded by many as the more primitive condition, and the survival of this
aperture as the anus or mouth only is conceded to be an ontogenetic
adaptation.

ON THE DIPLOCHORDA. Spi

4, The vascular system is primitive, consisting, when present
at all, of sinuses and splits between the coelomic walls.

5. The protomere (pre-oral lobe) is usually present through-
out life, or is distinctly in evidence at one time in the ontogeny
of the individual.

6. The ceelom is segmented into a protocele in the pre-oral
lobe, a pair of mesoceeles, and a metacele which tends to be
secondarily more or less divided up into two. Each ccelomic
cavity has primitively a pore leading to the exterior, which
primarily acts as a gonaduct and nephridiopore. The proto-
coelic pore is divided usually into two proboscis pores, the
mesoccelic pores are usually termed collar pores; all these
early lose their gonaducal function. The metaccelic pores
usually become specialised for gonaducts.

These will indicate a few characters of the Archicelomata,
and it will be seen that the Archichorda (Cephalo-
discus, Phoronis, and Balanoglossus), Echinodermata,
Chetognatha, and Brachiopoda, all readily group them-
selves under this head, and there can be little doubt that the
Sipunculoidea and Polyzoa will be found to conform to
the same character.

As might be expected of animals preserving a primitive
type of structure, nearly the whole of these groups are essen-
tially of sedentary or burrowing habits, though the Chetog-
natha are a remarkable exception. The position of the anus
and the retention of the reproductive function in the meso-
coeles in this group indicate an early assumption of the group-
characters, and the very existence of this group shows that,
under certain conditions,’ the archiccelomate type has been
able to maintain its existence in competition with metameri-
cally segmented forms. Amongst these Archicelomata
we may notice one group, the Archichorda, which are nearest
the direct line of the true metamerically segmented Chordata;
and others again, such as the Sipunculoidea and Brachio-
poda, which show, in their ontogeny, stages in the evolution

1 The Chetognatha are probably a primitively pelagic group.

332 A. T. MASTERMAN.

of the metamerically segmented Invertebrata, the Annelids
aud Arthropoda.

One cannot here enter further into these ideas, but they may
be appropriately expressed by the accompanying tree.

Though it would be difficult to imagine a more rudimentary
condition of the notochord than that of Actinotrocha, the
ontogeny of the Echinodermata and Brachiopoda, or even
the Sipunculoidea and Polyzoa, may yet furnish interest-
ing rudiments of chordate organs.

Another suggestive line of thought is the gradual abolition
of the archimeric segmentation and elaboration of the
metamericin the Annelida, Arthropoda, and Eu-chorda.
The nerve-ganglion of Actinotrocha is probably homologous
with the trochophoral apical ganglion, and hence with the
supra-cesophageal ganglion of the Annelida.

The protomere is most likely represented by the prosto-
mium, the mesomeres by the peristomium, and the post-oral
or mesomeric nerve-ring by the circum-cesophageal ring. The
mesomeric ceelomic cavities may possibly break down, leaving
the trocophoral head-kidneys as mesomeric nephridia, and
_ there are indications that all the coelomic cavities (and appen-
dages in Arthropoda) are post-oral, probably metameric."

It is not difficult to understand why the metamerically seg-
mented animals have replaced, in nature’s highways, the
Archicelomata, retaining only the archimeric segmenta-
tion into three parts.

Just as a limit is placed to the organic capabilities of single-
celled individuals, which is easily surpassed by the Metazoa,
so the animals starting with a greater number of segments,

1 «Tt appears that the antenne of Peripatus and the antenne of Myriapods
and Hexapods, as well as the two pairs present in Crustacea, are all of them
post-oral parapodia shifted in relation to the mouth...... The conclusion
to which embryological study (as to segments of body-cavity, so-called cere-
bral ganglion, and position of early rudiments of antennary appendages of
Arthropods) tends is that in the various branches of Arthropod descent ... .
there have been forward movements or shiftings of post-oral segments and
their parapodia.”—H, Ray Lankester, ‘ Natural Science,’ April, 1897, p. 265.

ON THE DIPLOCHORDA. 333

and other suitable conditions, are capable of more elaborate
differentiation to suit a more complex environment than those
which are limited in the number of their morphological units,
i.e. segments in this case.

The same environmental factors acting on the metameric-
ally and the archimerically segmented animals produce a close
homoplastic resemblance in the two groups of organisms.
Thus the metamerically segmented Arthropoda, and even
Vertebrata, show a grouping of segments into three regions,
the head, thorax, and abdomen, which in the general plan of
their several form and functions have a close resemblance
to the protomere, mesomere, and metamere of the Archi-
colomata.

On the other hand, in the parallel evolution of metameric
segmentation in the Eu-chorda, the tracing of archimeric
segments in them is fraught with interest.

334 A. T. MASTERMAN.

Holochorda (Vertebrata).

Annelida and Ku-chorda.
Arthropoda. hte tees)
Cephalochorda.
Urochorda.
Archichorda.}
(Actinotrocha.)
Hemichorda.
Diplochorda.

Chetognatha.———— > (Arehiceelomata. )

Sipunculoidea. we Stage ILI. aN
Poly zoa. | Echinodermata.

Brachiopoda.
Stage II. (Pelagic nekton.)

Stage I. (Pelagic plankton.)

Diploblast.

In the above tree the idea of the ancestry of the Chor-
data is a long line of pelagic organisms up to the stage
reached in Actinotrocha, at which stage two groups took
to degraded habits, the Hemichorda and the Diplochorda,
and thereby saved themselves from extinction. Balano-
glossus, with its burrowing habitat, shows a few remarkable
modifications and further differentiation in organs, and even a
commencing metameric segmentation, and is less degenerate
than the Diplochorda. In this group Cephalodiscus
has sought refuge in a deep-sea habitat and a protecting house ;
whilst Phoronis also resorts to a sedentary and, in most
cases, tubicolous habit. Hence the former retains a more

1 Differing in the main from Stage III by the evolution of paired chordoid
rudiments, rudimentary pharyngeal clefts, and a subneural gland.

ON THE DIPLOCHORDA. 335

primitive character, whilst the latter exhibits extreme degenera-
tion in the adult.

The main chordate stem then underweut metameric segmen-
tation, and gave rise to the Hu-chorda. Of these, those
still retaining an active pelagic or littoral life gave rise to the
Vertebrata or Holochorda (Gadow) ; whilst others again,
falling out of the ways of progress, suffere d the same fate as the
Hemichorda and Diplochorda. The Cephalochorda,
taking toa burrowing habitat, suffered like modifications (such
as pre-oral extension of the notochord) to the Hemichorda ;
whilst the Urochorda adopting the sedentary habits of the
Diplochorda, and especially those of Phoronis, like this
group suffered extreme degeneration. ‘The metameric segmen-
tation of the U rochorda is not beyond dispute, but the con-
sensus of workers on the group appears in favour of regarding
them as metamerically segmented, at least primarily.!. The
views here propounded may be expressed in this classification.

Chordata.
I. Archichorda.

Archimeric segmentation into protomere, paired mesomeres,
and metamere; little or no metameric segmentation. Noto-
chord in primitive continuity with the walls of the gut
throughout life. More or less connected with the protomere,
the main animal organ of the body. Nerve-ganglion between
protomere and mesomere or dorsal to mesomere. Main nerves
are protomeric ring, mesomeric ring, and dorsal and ventral
trunks. A mesoblastic chondroid skeleton and an ectodermal
chitinoid tube or skeleton.

1. Hemichorda.—Notochord fused im middle line and
protruding far into the protomere. Commencing metameric
‘segmentation in gill-slits and gonads.

2. Diplochorda.—Notochord in primitive paired condi-
tion. In close connection with the two posterior protomeric

1 The position of Urochorda is undoubtedly nearer the Eu-chorda than
the Archichorda.

voL. 40, PART 2,—NEW SER. AA

336 A. T. MASTERMAN.

mesenteries. Mesomeres produced into numerous tentacles.
Metamere with dorsal flexure.
(1) Phoronidea.—Loss of pre-oral lobe and notochord
in adult. No gill-slits.
(2) Cephalodiscida.—One pair of gill-slits with chordoid
walls ; persistent notochords.
3. Rhabdopleurida.’

II. Eu-chorda.

Archimeric segmentation replaced by metameric. Single
dorsal notochord loses connection with the gut-wall, and dorsal
nervous system with the ectoderm. Protomere and mesomeres
reduced. Notochord extends into metamere (Urochorda, Ho-
lochorda), and also into protomere (Cephalochorda).

1. Urochorda (?).

2. Cephalochorda.

3. Holochorda.

REFERENCES.

1. W. Batrson.—‘ Quart. Journ. Mier. Sci.,’? xxiv—xxvi.

la. Bovert.—‘ Anat. Anz.,’ vii, 1892, pp. 170—181.

1b. W. Brooxs.—‘ The Genus Salpa,’ Baltimore, 1893, p. 303.
2. HE. H. Catpweti.—‘ Proc. Roy. Soc.,’ xxxiv, pp. 371—383.
. KE. H. Canpwett.—‘ Quart. Journ. Micr. Sci.,’ xxv, 1885.

. T. 8. Coppotp.—*‘ Quart. Journ. Mier. Sci.,’ vol. vi, p. 50.

. E. Crarparrpr.—‘ R. und D. Archiv,’ 1861, p. 537.

. EK. CraparrzpE.— Beobach. tiber Anat. und Entwick.,’ Leipzig, 1863,
p. 83.

7. C. J. Cornt.—‘ Zeitschr. f. wiss. Zool.,’ xli (1890), pp. 480—568.
7a. M. Daviporr.— Mittheil. Zool. Stat. Neap.,’ 1889-91.
8. F. D. Dyster.—‘ Trans. Linnean Soe. Lond.,’ vol. xxii, p. 251.

9. C. GecenBauR.—‘ Zeitschrift,’ Siebold and Kolliker, 1854, pp. 347—
350.

10. 8. F. Harmer.—‘ “ Challenger ” Reports,’ vol. xx, pp. 46—47.
11. A. Kroun.—‘ Miiller’s Archiv,’ 1858, pp. 293—298.

Qa a P &

1 For further details regarding the Archichorda see end of Part II.

ON THE DIPLOCHORDA. SB We

12. A. Kowa.evsx1.—St. Petersburg, 1867. (‘ Leuckart’s Bericht.,’ ‘Archiv
f. Naturfesch Jahrg.,’ xxxiii.)

12a. R. Kouter.—‘ Internat. Monatsschr. Anat. u. Hist.,’ Bd. iii, 1886.
13. R. Levexart.—‘ Weigmann’s Archiv,’ 1859,

14. HE. Merscunrxorr.— Zeitschrift, Siebold and Kolliker, 1871, pp. 244—
251.

15. W. C. McInrosu.— “ Challenger ” Reports,’ vol. xxvii, p. 28.
15a. I. Mirsuxur1.—‘ Journ. Imp. Coll. Sci.,’ 1896.

16. J. Mijitter.— Archiv fiir Anat. und Phys.,’ 1846, pp. 101—104.
17. T. Morcan.—‘ Journ. Morph.,’ vol. ix, January, 1894, vol. v, 1891.
18. A. ScunerpER.—‘ Miiller’s Archiv,’ 1862, pp. 47—65.

19. T. Strzotp.—‘ Weigmann’s Archiv,’ 1850.

19a. J. W. Spencen.—‘ Fauna v. Flora des Golfes von Neapel,’ xviii, 1893,
and previous works.

20. R. Wacener.—‘ Archiv fiir Anat. und Phys.,’ 1847, pp. 202—206.
21. R. Wetpon.—‘ Proc. Roy. Soc.,’ xlii, p. 146.

22. E. B. Wirson.— Quart. Journ. Micr. Sci.,’ 1881, pp. 202—218.
23. A. WILLEY.—‘ Quart. Journ. Micr. Sci.,’ 1893.

24, A. T. Masterman.— Proc. Roy. Soc. Edin.,’ March, 1896.

25. A. T. MastermMan.—‘ Proc. Roy. Soc. Hdin.,’ June, 1896.

26. A. T. Masterman.—‘ Zool. Anzeig.,’ No. 505.

EXPLANATION OF PLATES 18—22,

Illustrating Mr. A. T. Masterman’s paper ‘On the Diplo-
chorda: Part I. The Structure of Actinotrocha.”

List of Abbreviations.

a.g. Atrial grooves. a.z. Anterior nerve. a. m.xz. Anterior median
nerve. a./, 2. Anterior lateral nerve. .c.! Pre-oral body-cavity. B.c.?
Collar body-cavity. &.c.° Trunk body-cavity. ¢.p. Collar-pore. c¢. x.
Collar nephridium. c.z. 7. Collar nerve-ring. d. 4. v. Dorsal blood-vessel.
v. b.v. Ventral blood-vessel. d.x. Dorsal nerve. d.a. Digestive area.
ht. Heart. mes. Mesentery between pre-oral and collar cavities. mes’. Mes-
entery between collar and trunk cavities. m. Mouth. xch. Notochord.
n.g. Nerve ganglion. zp. Neuropore. «@s. Gisophagus. 0. g. Oral grooves,

338 A. T. MASTERMAN.

pb. Proboscis vesicle. p.a. Perianal band. p.a.z. Pre-oral nerve-ring.
p.s. Perianal sinus. p.p. Proboscis pore. 7.s. Ring sinus. sh. Sheath of
notochord. s. 2. Subneural gland. s. z. s. Subneural blood-sinus. s. p.
Sense papilla. s¢. Stomach. ¢. Tentacle. v.c. Ventral collar area. v. ¢. r.
Ventral ciliated ridge. v. ch. Ventral chordoid area. v. Vacuoles. v. ¢. x.
Ventral collar nerves. v. m. Ventral mesentery.

PLATE 18.
Fie. 1.—Lateral view of entire larva with ten pairs of arms.

Fie, 2.—Lateral view of later larva, showing the distribution of the main
nervous tracts.

Fic. 3.—Dorsal view of larva a little later than Fig. 1.

Fies. 4 and 5.—Two positions commonly assumed by living larva.

PLATE 19.

Fic. 6.—Transverse section of fully developed larva in region of pre-oral
hood.

Fie. 7.—Transverse section of same larva behind the nerve ganglion.
Fic. 8.—Ditto, through mouth.

Fig. 9.—Ditto, through notochords and cesophagus.

Fic. 10.—Ditto, through collar region.

Fies. 11—15.—Transverse sections of dorsal wall of same larva, between
Fig. 6 and Fig. 7, to show nerve ganglion.

Fic. 16.—Transverse section of same larva through collar and trunk.
Fic. 17.—Ditto ditto, further back.

PLATE 20.
Fies. 18 —20.—Sagittal sections of fully developed larva. Fig. 20 is nearly
median.
Fies. 21—27.—Coronal sections of ditto, with hood turned back.
Fic. 28.— Transverse section of hood.
Fic. 29.—Obliquely transverse section to show oral grooves.

PLATE 21.

Fic. 30.—Surface of notochord in early larva.

Fics. 30a, 308, 30c.—Transverse sections of notochord in Fig. 30 at a,
B, and © respectively.

Fie. 31.—Transverse section of later stage of notochord.
Fie. 32.—Longitudinal section of later stage of notochord.

ON THE DIPLOCHORDA,. 339

Fics. 33—36.—Series of semi-diagrammatic figures illustrating the pro-
cess of vacuolisation.

Fie. 837.—Oblique section of notochord in young Phoronis.

Fic. 38.—Side view of front parts of fully developed Actinotrocha.

Fic. 39.—Transverse section of same larva as Figs. 6—17, in trunk
region.

Fic, 40.—Longitudinal section of collar nephridium, as reconstructed
from serial sections.

Fic. 41.—Transverse section of duct of collar nephridium.

Fic. 42.—Transverse section of nerve ganglion and subneural sinus.

Fic. 43.—Ditto of tentacle (larval).

Fie. 44.—Ditto of pre-oral nerve-ring.

Fie. 45.—Longitudinal section of hind end of larva to show the doubtful
organs,

PLATE 22.

Fic. 46.—Semi-diagrammatic dorsal view of nervous system of Balano-
glossus,

Fic. 47.—Ditto of Actinotrocha.

Fic. 48.—Semi-diagrammatic median sagittal section of front end of
Balanoglossus.

Fie. 49.—Ditto of Actinotrocha.

Fic. 50:—Semi-diagrammatic median coronal section of Balanoglossus,

Fic. 51.—Ditto of Actinotrocha.

List of Abbreviations.

b. c.) Pre-oral body-cavity. 4. ¢.8 Trunk body-cavity. 4. s. Blood-sinus.
c. p. Collar-pore. d. mes. Dorsal mesentery. d. 4. v. Dorsal blood-vessel.
ep. Epistome. g.s. Gill-slit. iv¢. Intestine. 7. 6. v. Lateral blood-vessel.
l.c.c. Left collar cavity. m. Mouth. 2.g. Nerve ganglion. 2. Nuclei.
nch. Notochord. ov. Ovum. yp.o./. Post-oral lamella. pA. Pharynx.
p. p- Proboscis-pore. 7. c.c. Right collar cavity. sf. Stomach. sh. Sheath.
s. 2.8. Subneural sinus. s. z. g. Subneural gland. 7. Tentacle. ». d. v.
Ventral bluod-vessel. v. mes. Ventral mesentery.

340 A. T. MASTERMAN.

PARI it,
On the Structure of Cephalodiscus dodecalophus, McIntosh.

As already indicated in the previous work upon Actino-
trocha, the comparison of some of its organs with those of
Balanoglossus indicated that a renewed study of Cephalo-
discus and its “notochord” was desirable. The work here
referred to was done upon material from the “ Challenger”
collection.

The former work upon Cephalodiscus is confined to a
description of the chief anatomical features by Professor
McIntosh (2) in the “ Challenger” series, and an appendix to
it by Mr. S. F. Harmer (1), the principal feature of which is
a detailed comparison of this species with Balanoglossus,
as a result of which the group Hemichorda has by many
been made to include these two species in addition to Rhab-
dopleura.

As already mentioned in the preceding paper on Actino-
trocha, work on this animal and considerations leading there-
from caused me to suspect the presence in Cephalodiscus
of a notochord different in form and position from that already
described as such by Harmer. I must here express a deep
sense of gratitude to Professor McIntosh, LL.D., F.R.S., for
the gift of specimens of this unique animal, taken from the
few now remaining in his hands. I hope to be able to publish
later the results of labour upon the buds and young forms, but
this note must be confined to the description of some organs
not hitherto noticed, and to the inferences to be drawn there-
from. It need only be mentioned that my pre-conceived view
that Harmer’s “ notochord ”’ is the homologue of the subneural
gland, and that Cephalodiscus has a paired notochord, has
been confirmed to a remarkable degree.

ON THE DIPLOCHORDA, B41

Of the two preceding workers, the keynote of McIntosh’s
(2) views was a close resemblance to Phoronis and Rhab-
dopleura; of Harmer’s (1), an alliance with Balano-
glossus. Whilst accepting the views of the latter, I think
the work below will justify the placing of Phoronis and
Cephalodiscus together under the title Diplochorda.

We may add that Professor A. Lang (4), in an article upon
the organisation of Cephalodiscus, accepts its affinities with
Balanoglossus. Further than this, he considers that the
organs of the former show an earlier stage of evolution than
those of the latter. On the other hand, he comes to the con-
clusion that the simplicity in structure of Cephalodiscus is
not primitive, but has been secondarily acquired by a degenera-
tion due to its habit. I fail to see any indications that this is
the case. It seems to be the more natural conclusion, and
that most in accordance with the facts—more especially since
I hope to show (below) that a distinct vascular system and
also paired notochords are present—to regard this animal as
retaining a primitive Archichordate organisation, always ex-
cepting the dorsal flexure of the gut, and the hypertrophy and
forward direction of the tentacles, which cannot be regarded as
characters of a free-swimming chordate ancestor. This will be
further referred to below.

Lang considers the resemblance of Cephalodiscus to
Phoronis and the Bryozoa to be convergent. Whilst
leaving out of the question at present the latter, I hope to show
sufficient anatomical resemblance between Cephalodiscus
and the free-swimming Actinotrocha to outweigh such
suggestions.

The absence of pharyngeal clefts apparently extends to
Rhabdopleura as well as Phoronis, but the value of this
feature cannot be gauged till we know more concerning the
origin of these organs.

Nervous System.— McIntosh (2) described the main
nervous ganglion on the dorsal side of the collar region, and
remarked, “It extends for a considerable distance laterally
on each side along the basal region whence the plumes

3842 A. T. MASTERMAN.

spring, and for some distance on the dorsal surface of the
buccal disc.” He also noticed that the general arrangement
was similar to that of Phoronis. Harmer (1) showed further
that nervous branches passed up the dorsal sides of the arms.
As will be seen by inspection of the accompanying figures, in
which the nervous system is coloured yellow, my sections
confirm the previous results, and make further statements
possible. The main nerve-ganglion (Pl. 23, fig. 3, 2. g.) lies
over the subneural blood-sinus (s. 2. s.), which is a hemal
space immediately between the mesodermal wall of the pre-
oral body-cavity (epistomial) and those of the collar cavities.
In other words, it is in identically the same position as the
nerve-ganglion of Actinotrocha. On the walls of the sub-
neural sinus arise the inner ends of the “ proboscis pores”
(or canals) (p. p.), which then lead forwards and upwards on
either side of the ganglion to the exterior (see Pl. 23, figs. 2
and 38,p.p.). This arrangement of proboscis canals, subneural
sinus, and ganglion is precisely similar to that of Actino-
trocha.

Traced backwards (fig. 4), the main ganglion gives off the
thick lateral branches to the arms, whilst further back it con-
tinues its course, overlying the thin dorsal blood-sinus between
the two collar cavities (figs. 5—8). By the time fig. 9 is
reached (cf. Pl. 24, fig. 13) the pharynx, or more possibly the
basal part of the subneural gland, comes to lie close under
the nervous mass, which then divides into two lateral nerves
(fig. 12, 7. 2.). These give off a post-oral ring at the posterior
edge of the collar region (Pl. 25, fig. 15), which becomes
lost ventrally in a mass of nerve-fibres, scattered over
the inner surface of the ectodermal layer of the post-oral
lamella (fig. 13). The lateral nerves are then continued as a
pair of rather broad nerve-tracts, lying laterally in a position
similar to those of Phoronis.

Auteriorly, the ganglion is continued as a mass of fibres
along the upper surface of the epistome, in the mid-dorsal line
and somewhat to each side of it (figs. 1, 18, 14), round the
apex, and then unites with a very broad flat ring which lies at

ON THE DIPLOCHORDA. 343

the base of the long glandular cells, covering the “ front ” of
the epistome.

Laterally and anteriorly, the ganglion gives off a pair of
large nerves (Pl. 24, fig. 18, p. 0. ”.) which pass downwards
and outwards along the “inner” surface of the epistome,
where they unite (fig. 12), and passing round the base, they
meet the broad ring.

This arrangement, allowing for the peculiar modification of
the epistome, is closely similar to that of Phoronis (Actino-
trocha) and B alanoglossus.

Lastly, along the mid-ventral surface of the trunk there are
several nerve-fibres running longitudinally, and some of these
pass into the ‘‘ pedicle” in a peculiar manner, where they
appear as fine nervous cylinders. The detailed description of
these and of the pedicle will be given in future work.

The mid-dorsal and two lateral epistomial branches have
their homologues in Balanoglossus and Actinotrocha,
whilst the p re-oral nerve-ring of the latter can be compared to
the “front”? ring of Cephalodiscus, especially when the
relationship of both to the coelomic muscle-strands is con-
sidered. One can assume that the condition of the pre-oral
nerve-ring of Cephalodiscus has been arrived at by a pro-
cess of centralisation by which the ring, and with it the origin
of the muscles, has been moved into the centre of the disc, for
the formation of a sucker.

I do not feel justified in giving a complete diagram of the
nervous system of Cephalodiscus until I have confirmed
certain points not dealt with here. It has been shown that
the lateral trunks of Phoronis may be regarded as the two
halves of the dorsal nerve of Balanoglossus, in the trunk
region, and that their peculiar position can be accounted for
by displacement due to reduplication of the trunk. The same
reasoning would apply to the case in hand, the lateral nerve
trunks of Cephalodiscus occupying the same position as in
Phoronis.

Sense-organs.—The consideration of the curious masses
of pigment surrounding the oviducts, which were at first

344 A. T. MASTERMAN.

described as eyes, together with that of the reproductive organs,
must be left to a later paper (in which I hope to give an
account of the buds and young forms), but a careful examina-
tion has led me to believe that Cephalodiscus is provided
with no less than a dozen large eyes of a very primitive com-
pound type.

Each plume terminates in a globular enlargement, the
appearance of which has been described and figured by
McIntosh (3). Many of the cells of this enlargement appear
to contain a large clear globule of an ovoid shape. McIntosh (3)
remarks, “The rugose appearance, however, is due to large
gland-cells containing granules and globules, which are
arranged in a somewhat regular manner round a central
cavity, and which present a deep yellowish tint in the prepara-
tions” (p. 11). If the parts be subjected to partial maceration
the clear globules can be obtained free (PI. 26, figs. 28 and 29),
and they remind one irresistibly of a crystalline refractive
lens. ‘Two different common shapes are here indicated. In
longitudinal section of the bulb (fig. 30) it is seen that the
epithelium is single-layered, and consists of elongated tapering
cells, with their nuclei mostly situated at the base. All have
fine pigment granules scattered throughout their interior, and
a great number of them contain the crystalline lenses referred
to. With Meyer’s carmine and with hematoxylin these latter
stain readily, one or more areas in the centre staining more
deeply than the rest. An axial part of the protoplasm imme-
diately internal to these bodies, and containing the nucleus,
usually appears darker than the rest of the protoplasm. The
lens may often be observed protruding through the cuticle to
the exterior, evidently an abnormal condition.

Fig. 31 gives a view of one of the cells in longitudinal sec-
tion (unstained). It is bounded externally by a definite but
fine cuticle, and its inner end tapers to a fibre-like thread,
which I believe to have in some cases traced into the main
nerve of the plume.

The whole structure here described seems to indicate that
these organs are rudimentary monostichous compound eyes,

ON THE DIPLOCHORDA. 345

which bear a remarkable resemblance, both in appearance and
structure, to the “‘ branchial organs” found in the sedentary
Annelids, such as Potamilla and Sabella. The figures and
description of these organs as given by Andrews (8) (and a
comparison of them with these organs of Cephalodiscus)
can leave little room for doubt that the functions in each case
are similar. It seems most reasonable to regard them tenta-
tively as primitive eyes, though the presence of compound
eyes in the Chordata is rather remarkable.

Andrews conducted experiments with Potamilla to deter-
mine if the “‘ branchial organs” were possibly phosphorescent
organs, but apparently with a “negative result.” The pre-
sence of hairs (sensory ?) in the branchial eyes of some of the
Annelida, such as Myxicola and Filigrana, led him to
suspect a possible sensory function other than that of vision,
but there is no indication of hairs or cilia in the branchial eyes
of Cephalodiscus.

Branchial Plumes (Lophophore).—The collar region, as
is well known, is produced on each side into six arms or
plumes, and each of these has a double row of smaller pinna,
borne laterally.

The structure of a plume is very similar to that of the rest of
the collar region. A layer of epithelial cells rests upon the meso-
blastic chondroid tissue (PI. 26, fig. 26), which encloses a well-
defined cavity. Across this are stretched many fine strands of
protoplasm, with nuclei dotted here and there in their course.
The transverse section of a plume is crescentic in outline, and
the dorsal side is convex. The epithelium on this side is
thinner than that on the concave ventral surface, and consists
of closely set cubical cells ; it is continuous with the epithelium
of the dorsal collar region. In the mid-dorsal line a flattened
nerve runs throughout the length of the plume, terminating
distally in the “ branchial organ,” and joining proximally with
the other five to enter the nerve-ganglion. Immediately internal
to this nerve (in the plume) runs a main blood-vessel, usually
somewhat triangular in transverse section (fig. 26), and con-
tinuous with the main dorsal blood-vessel in the collar region.

346 A. T. MASTERMAN.

On the ventral side the ectoderm consists of long glandular
cells closely crowded together, and sometimes thrown into
folds. They are covered with minute cilia, and are continuous
ventrally with the epithelium lining the mouth-cavity.

Ventro-laterally are given off, from each plume, the rows of
pinne on each side. A cross-section of a pinna is shown in
fig. 27. The dorsal and ventral surfaces of the pinna are
covered by similar cells respectively to the same two surfaces
of the plume, and under the dorsal cells is a crescentic (in
transverse section) blood-vessel whose lumen is continuous
with that of the plume. The rest of the internal cavity is
ceelomic, and is also continuous with that of the plume. A
few intersecting strands of mesoderm are to be seen. There
does not appear to be any definite nerve in the pinna.

The way in which the six plumes are arranged is peculiar,
and although one or two may be, and usually, in sections,
are slightly displaced, it is easy to demonstrate the principle
upon which they are disposed. Pl. 26, figs. 82—386, are
meant to illustrate this, and they are selected from a series
of sections cut perpendicular to the long axis of the buccal
shield.

The areas occupied by glandular ciliated epithelium, and be-
longing morphologically to the ventral surface, are indicated
by the thicker lines, whilst the thinner lines indicate the
morphological dorsal surface with its characteristic non-ciliated
epithelium. In fig. 32 the right side shows the base of the
united plumes and the post-oral lamella behind them, whilst
the left side shows indications of a separation of the two most
posterior plumes (5, 6). In fig. 33 these two are seen to be
free from the main axis, and very little further up (fig. 34) the
next (4) branches off. Higher up still these three arrange
themselves in linear series, so that their morphological dorsal
surface is opposite to that of the main axis, which has not yet
divided into three. Fig. 35 shows the separation of (3), and
further up again the last two diverge. Fig. 36 indicates the
arrangement some way further towards the tips, and the six
plumes are here seen to be arranged in more or less of a circle,

ON THE DIPLOCHORDA. 347

though they still show indications of two rows, an anterior and
a posterior.

These sections are all transverse to the epistome, and are
not quite parallel to the dorsal surface of the collar. [t is pro-
bable that in a section in the latter direction the six plumes
would diverge from the main branch at about the same level,
a conjecture further borne out by the examination of uncut
specimens.

Practically the plumes and their pinne form a large funnel-
shaped framework, with the interior lined by the thinner or
atrial epithelium, and the exterior with its twelve grooves
running down into the mouth, covered by the glandular and
ciliated branchial epithelium.

In Phoronis a precisely analogous arrangement holds, but
in this case the tentacles surround the mouth in the mid-
ventral line, and are then produced laterally to form a double
row, which, bending backwards, forms an imperfect atrium, in
which is situated the anus. In this case there is a similar
disposition of branchial and atrial epithelium, but, as the
tentacles here surround the mouth, the former lines the
branchial cavity, with the mouth at its base, whereas the latter
lines the less complete atrium. It is evident that if the outer
row of tentacles, meeting mid-ventrally in Phoronis, were
removed, and the epistome restored to something like its
Actinotrochan proportions, the similarity would be striking.

One cannot doubt that in Cephalodiscus the cilia of the
ventral surface cause currents down the ventral grooves of the
plumes, and thence into the mouth, as has already been sug-
gested, carrying food particles, entangled in slime, into the
alimentary canal, a mode of alimentation analogous to that of
Phoronis, and indeed of the Urochorda.

Organs of the Mesoderm.

Muscular System.—The muscles are little differentiated,
and often consist histologically of simple protoplasmic strands
with nuclei. In the pre-oral lobe, or protomere, a great number
of these strands cross the coelomic cavity in a definite radiat-

348 A. T. MASTERMAN.

ing direction (figs. 1,9, 14),—cf. McIntosh (3). They originate
from the thickened chondroid skeleton lying between the
posterior ccelomic wall of the epistome and the anterior wall
of the collar, and run forwards and outwards to be inserted
into the thin chondroid tissue immediately underlying the
anterior nerve-ring of the epistome (Pl. 25, fig. 15). Their
contraction would result in the drawing inwards of the central
part of the epistomial ventral surface ; and hence, if this organ
were applied to a flat surface, they would convert it into an
efficient sucker.

The coelomic cavity of the collar and plumes is also traversed
by a number of protoplasmic filaments which are much finer
than those of the epistome ; they may or may not be contrac-
tile (figs. 15,16). In all cases they run from one part of the
chondroid tissue lining the ceelom to another, and appear
indefinite in arrangement. The largest and best defined run
from the lateral skeletal bodies to the head of the notochords
(fig. 15).

In the trunk there is (contrary to Phoronis and Balano-
glossus) a marked absence of muscular tissue, with the
exception of the mid-ventral region, in which there are definite
muscular bands running from behind the mouth to be inserted
into the “ ventral sucker ” (Pl. 25, fig. 17, v.m.), as described
by McIntosh. In the ventral sucker itself these muscles show
a fibrous structure.

Skeletal System.—Closely underlying the ectodermal cells
is found the so-called ‘‘ basement tissue,” which varies greatly
in thickness in various parts. A thin layer of this tissue lines
the whole cavity of the epistome, and where its mesodermic
wall is in contiguity with that of the collar ceelom there is
formed on either side (fig. 15) a thick mass of the same tissue,
from which long muscular fibres run forwards through the
epistomial coelom and backwards to the front end of the
notochords. They have been referred to as the lateral skeletal
masses.

The chondroid tissue is also easy to trace in the collar region,
—in the post-oral lamella, the plumes and pinne, and the

ON THE DIPLOCHORDA. 349

proximal area of the collar. In the trunk it is thinner, though
slightly thickened in the ventral sucker. It is only present
in connection with the “ somatic” part of the mesoderm in
contiguity with the ectoderm, except in the case of the gonads
and the lateral masses.

At the inner posterior end of the gonads there is a thick wall
of this tissue surrounding their tapering extremities. The
structure of this skeletal tissue is the same throughout, a
hyaline supporting but elastic mass, staining easily. There
can be little doubt that it is mesoblastic in origin, considering
that parts of it, such as around the gonads and in the lateral
skeletal bodies, are far removed from the ectoderm or endoderm,
and that the tissue in question is in such intimate connection
with the muscular system. In this case it could be homologised
with the chondroid tissue of Phoronis, and even with the
mesoblastic cartilaginous skeleton of the Vertebrata.

In the regions in which there is no chondroid tissue, such
as the trunk mesenteries and mesodermic (splanchnic) layer
over the endoderm, the mesoderm appears in the form of a fine
protoplasmic layer (Pl. 25, figs. 18, 19, sh.), with nuclei here and
there. In the parts directly enclosing blood-sinuses there is often
a thickening of this layer, due to an accumulation of branched
cells, which are probably contractile (see Vascular system).
It is possible that a thin layer of mesoderm may line the
coelomic cavities internally to the chondroid tissue, but the
protoplasmic filaments have the appearance of actually arising
from the substance of the chondroid tissue.

With the exception of the muscle-fibres of the ventral
sucker, therefore, the skeleto-muscular system of Cepha-
lodiscus appears to remain in the undifferentiated condition
of protoplasmic strands or branched cells and a_ hyaline
skeletal matrix, as in the early stages of other forms, such as
Phoronis.

McIntosh (8) has suggested that the elasticity of the chon-
droid tissue (“ basement membrane ”) may compensate for the
absence of circular muscles in the ventral sucker and other
parts, and, in addition, it evidently imparts a rigidity to the

350 A. T. MASTERMAN.

plumes—and, in the case of the two lateral skeletal masses,
forms a fulcrum for the pre-oral muscles. These lateral
masses, by their structure, origin, and function, can be
homologised with the chondroid tissue of Balanoglossus
(Marion).

Vascular System.—There is no prior reference to a vas-
cular system. A system of sinuses, bearing a close similarity
to that of Actinotrocha, is capable of easy demonstration.
The walls of the sinuses are in all cases formed by the meso-
derm of the cclomic cavities, and in some cases either the
ectodermal or endodermal wall. In certain parts, especially in
the dorsal vessel, there are thickenings of the walls caused by
nucleated mesodermic cells which bear a close likeness to those
described in Actinotrocha, and it can hardly be doubted
that these are contractile as in the latter. Immediately under
the nerve-ganglion lies a large sinus, which, as in Actino-
trocha, we may term the subneural sinus. It is very con-
stant in outline (figs. 2 and 38), and has, in transverse section
of the epistome, a quadrilateral shape. Its dorsal wall is
formed by the nerve-ganglion, and the other three sides of the
rectangle (i.e. laterally and ventrally) are formed by the
coelomic wall of the epistomial cavity. Anteriorly it is blind,
and posteriorly it tapers off into the mid-dorsal blood-vessel,
the greater part of which is filled up by the long cecal sub-
neural gland (s.n.g.) (see below). Its course can be followed
in the series of sections from fig. 1 to fig. 9, in which it is
tinted red. The dorsal blood-vessel in this region (d. 6. v.) is
bounded laterally by the collar walls, and these are thickened
aud, as already mentioned, probably contractile.

At the level of fig. 10 this vessel separates into two, which
encircle a part of the gut to be referred to later, and again
join up beyond this to form a single median vessel (fig. 15,
d.b.v.). Dorsally to the pharynx this vessel can be traced
(fig. 16) till it reaches the stomach, round which it breaks up
into a system of sinuses, which, more or less disconnected in
the preserved specimens, is probably one large blood-sinus in
the living animal. From this sinus branches, with indications

ON THE DIPLOCHORDA. 351

of contractile walls, are given off to the gonads, and there are
traces of a ventral vessel which also passes out into the pedicle
(fig. 17, v. 6. v.). ‘There is a hemal vessel in each pinna of the
tentacles, but I have not traced these further (see “ Branchial
plumes ”’).

Many of these vessels have undoubtedly shrunk in size, and
in some cases lost all lumen; so that in the living animal,
especially if the blood be tinted, they probably form a very
prominent system. ‘The same condition is met with in pre-
served Actinotrocha, in which the vessels are difficult to
trace, though followed with ease in the living larva.

The vascular system here described corresponds, even to
minute detail, with that of Actinotrocha and Phoronis.

Subneural Gland.!—In his comparison of this species
with Balanoglossus, Harmer (1) drew up a list of features in
which it closely resembled the latter. Amongst these he com-
pared an anterior diverticulum of the gut to the notochord of
Balanoglossus. Whilst confirming the other features he
described, and further adding thereto the very similar vas-
cular system, I think that the homology of this organ to the
enteropneustan notochord (or to that of any other animal)
cannot be maintained. Its origin (from epiblast or hypoblast)
cannot be determined till the sexual development is followed,
for I have every reason to believe that the entire gut in the
young buds is formed from the ectoderm of the parent. It
presents no histological features resembling those of every other
notochord yet described, and it is difficult to conceive of its
performing any supporting function. As already suggested,
I propose to homologise it with the subneural gland of
Actinotrocha (and in part to the proboscis vesicle of Bala-
noglossus.”

In view, however, of its “‘ notochord ” claims, I have worked

1 Described here to follow ‘vascular system,’ though of course not meso-
dermal.

? In section the subneural gland consists of a single layer of elongated
ciliated cells surrounding a small lumen, and in many cases there may be seen,
in the centre of the lumen, a rod of hardened darkly staining mucus.

VOL. 40, PART 2.—NEW SER. BB

852 A. T. MASTERMAN.

out its relationships in the adult with some detail. Figs. 1 to 12
are drawings, by camera, of sections selected from a consecutive
series, cut nearly transversely to the long axis of the epistome.
In the sections in front of fig. 1 the whole interior of the
epistome is lined by the coelomic epithelium of the pre-oral
body-cavity (6. c. 1). At fig. 1 are cut the tips of the two
collar cavities, and in the centre the front end of the subneural
sinus. In fig. 2 there is further seen a space in the centre of
the subneural sinus, which is continuous with the pre-oral body-
cavity (see Pl. 24, fig. 14), and the walls of which are thickened
and contractile. In life this space, in all likelihood, only
exists when its walls are in a state of contraction. In fig. 3
the tip of the subneural gland is seen, and in fig. 5 this organ
is to be noticed embedded in the sinus. The condition seen in
figs. 6, 7, and 8 may be found in a number of sections, and
probably represents the state of affairs when the walls are
collapsed. They all indicate the dorsal blood-vessel, with the
subneural gland lying in its lumen, though mostly resting
against the ventral eclomic wall. In fig. 9 the tip of the gut-
wall is recognisable, and in Pl. 24, fig. 10, the subneural gland
is observed to pass into it. Figs. 11 and 12 illustrate further
stages, which are readily understood with the help of fig. 138.
I would regard this dorsal diverticulum of the gut which
underlies the collar nerve-mass, and into which the subneural
gland is continued, as the enlarged base of the gland, and it is
at least possible that the stomodeum extends as far as the
dotted line in fig. 13.

The relationships of this subneural gland and its structure
are so closely identical with those of the similarly named
organ in Actinotrocha that a want of definite proof of an
epiblastic origin can hardly militate against the homology of
the two organs. If further evidence were required, the dis-
covery of other organs which have good claims to be regarded
as of notochordal value can be added.

What little light is thrown upon the question by study of

1 The above description has been checked by sagittal and coronal sections,
which, however, it would be superfluous to reproduce.

ON THE DIPLOCHORDA. 353

the buds tends to show that the subneural gland is, from the
first, pre-oral in position, and does not move forward from the
collar region, like the notochord of Balanoglossus,

Organs of the Endoderm.—There are a conspicuous pair
of grooves converging to the mouth (Pl. 24, fig. 12, 0.9.),
which correspond with the oral grooves of Actinotrocha.

The mouth (m.) leads into an area which I would consider a
stomodeum, giving off its subneural gland, and thence into a
pharynx (pd.). This is a spacious chamber which extends
some way down the body and finally opens through a short
cesophagus into the large stomach (s¢.), which, in its turn,
passes to the exterior by a broad intestine (¢vé.). These parts
have all been described in more or less detail elsewhere, and
the pharynx alone need here detain us.

Its walls are folded transversely into many sac-like out-
growths which present a somewhat regular arrangement, though
they may be largely due to shrinkage (fig. 13), but throughout
its whole length extends a pair of lateral vacuolated areas
which I would compare to the notochords! of Actinotrocha.
Though almost lateral in the extreme front end (Pl. 25,
fig. 15, ch.) they come to occupy a dorso-lateral position at
the posterior end (fig. 17, mch.), the whole pharynx being
in this region, partially divided into ventral and dorsal halves
by lateral ridges.

In tracing out the course of one of these notochords, we
first observe it in transverse section at the extreme front end of
the pharynx, covered by a mesodermic sheath formed mainly
by the collar celom, and in this sheath originate strong
muscular bands which extend forward to be inserted in a
thickened mass of chondroid tissue lying between the epi-
stomial and collar coelomic walls. In the opposite face of this
chondroid skeleton are inserted the muscles which are con-

1 In this case, and in Actinotrocha, the term notochord is, strictly
speaking, a misnomer ; but, considering Bateson’s name for Enteropneusta,
we can scarcely adopt the term ‘“‘hemichord” (Mitsukuri). With those who
object to the term notochord, applied to a lateral organ, the name “ pleuro-
chord” might find favour.

354. A. T. MASTERMAN.

nected with the central disc of the epistome, and which no
doubt subserve the function of attachment by suction. Thus,
in an indirect way, the front end of the notochord acts as a
fulcrum for the epistomial muscles, and before the evolution of
the chondroid tissue it would do so directly.

Lower down (fig. 16, left side of), at the level of the mouth,
the notochord is seen as a diverticulum abutting on the nerve-
tract and the collar-pore (c.p.); whilst a little below this
(fig. 16, right side of) the pharyngeal cleft is seen to lead
forwards and downwards in front of the lower end of the collar-
pore (c.p.). This will be referred to later.

With very little change in character the notochord is con-
tinued throughout the length of the pharynx, as a vacuolated
longitudinal groove, with a wide lumen in continuity with the
pharyngeal cavity. Here and there it is folded or thickened,
but this is probably an artefact. Fig. 17 shows very nearly
the hind end of the pharynx with the notochords. The mid-
dorsal pharyngeal wall is thickened, arched into a ridge, and
strongly ciliated, whilst in the mid-ventral line a part of the
pharynx tends to become constricted off as a spacious ventral
groove.

Behind this the notochords gradually lose their vacuolated
cells, giving place by degrees to the ordinary cells of the
pharynx, and a very few sections after this the pharynx passes
into the narrow cesophagus.

In fig. 19 is seen the appearance of the notochord in trans-
verse section. ‘The vacuolisation extends throughout, and the
nuclei are seen dotted about, especially at the junctions of the
vacuoles.

Amongst those who accept the genetic relationship of Bala-
noglossus and Cephalodiscus, I do not think there will
be any hesitation in regarding these as a pair of notochords
(pleurochords), together homologous with the single notochord
of the other Chordata. The remarks made concerning the
notochords of Actinotrocha apply equally well to these
organs, and the homology to each other cannot be doubted,
especially as the small differences, such as the presence of a

ON THE DIPLOOHORDA. 300

thin unmetamorphosed layer of protoplasm in the notochord
of Actinotrocha, can be traced to the fact that only un-
developed stages are found. The notochord of the young
Phoronis most nearly approaches that of Cephalodiscus in
structure.

The same supposition as was put forward in the case of
Actinotrocha will hold for the comparison of Cephalo-
discus and Balanoglossus. It seems likely that in Bala-
noglossus the two notochords have protruded still further
forward into the proboscis and have fused in the middle line,
and that the two thickened masses of mesoblastic skeletal
tissue (chondroid) in Cephalodiscus have met in the mid-
ventral line in Balanoglossus to form the skeletal rod.

Pl. 26, fig. 20, gives a diagrammatic sagittal section of the
front end of Cephalodiscus, comparable to those of Balano-
glossus and Actinotrocha, previously given. A compari-
son of the three figures will show the close similarity between
the three types.

The only previous observation, so far as I am aware, of the
chordoid tissue of Cephalodiscus is in the work of
McIntosh (2), in which, referring to the walls of the pharyngeal
clefts, which are also of chordoid character, he speaks of “ the
translucent wall of the slits which seems to be a modified con-
tinuation of the pharyngeal mucous membrane. The granules
are finer, and the whole tissue is more translucent ” (p. 16).

As a matter of fact, the walls of the pharyngeal cleft present
identically the same histological structure as those of the noto-
chords, except that the vacuoles are smaller. The chordoid
tissue of the notochord is continued directly into that of the
pharyngeal cleft (fig. 16), and meets the ectoderm at the distal
end of the cleft. In fig. 19 the left-hand end of the notochord
is continued into the commencing pharyngeal cleft.

Fig. 18 is a transverse section of the cleft, drawn from a
section of the same series as, and a little further out than,
fig. 14. Just as the notochords in the pharynx subserve a
supporting function, so here the chordoid tissue of the gill-slits
eyidently has also a similar function analogous to that of the

356 A. T. MASTERMAN,

tracheal thickenings of Insecta and Vertebrata. The
maintenance of an open pharyngeal cleft must be an impor-
tant factor in the well-being of this little animal, or the water
currents would be diverted into the stomach.

The existence of this chordoid tissue surrounding the
pharyngeal cleft is, I think, very instructive in its bearing upon
the evolution of the chordate organism.

Firstly, in regard to the later history of gill-slits in the rest
of the Chordata, we note that in every type of vertebrate
the gill-slits are supported by branchial bars developed from
mesoblast. The mesoblastic cartilaginous skeleton consists of
a central axis of cranium and vertebral column, and of an
appendicular skeleton of visceral bars. The former are
usually regarded as a secondary skeleton, replacing phyle-
tically in great part the primary hypoblastic axis or noto-
chord; but so far as my knowledge extends, there has not
been described a primary hypoblastic chordoid basis preceding
the branchial bars of the gill-slits. This may be partly ac-
counted for by the earlier evolution of these bars in com-
parison with the vertebral column.

The chordoid skeleton of the pharyngeal clefts of Cephalo-
discus can, however, be interpreted as the primary hypoblastic
skeleton of this region, which remains the permanent sup-
porting structure of the pharyngeal cleft in this type, just as the
notochord remains as the permanent supporting axial structure
of such forms as Amphioxus and the lowest Vertebrata.

In other words, Cephalodiscus has the chordoid fore-
runner not only of the vertebrate axial, but also of the appen-
dicular skeleton.

In the absence of any ontogenetic evidence, this chordoid
condition of the pharyngeal clefts continuous with the true
notochordal tissue, renders possible two theories with regard to
the evolution of gill-slits (primarily pharyngeal clefts).

In the first place we may assume that, in earlier forms, cer-
tain areas of the pharyngeal wall became chordoid for the pur-
pose of support, and that these diverticula came in close
contact with the ectoderm, as in the case of the base of the

ON THE DIPLOCHORDA. 357

subneural gland and the notochordal diverticula (fig. 16), and
that a breaking-down of the terminal cells gave rise to the
clefts. Vacuolisation is certainly a process of cell-degene-
ration, and disintegration of the vacuoles would be but a step
further in the process. On this hypothesis the notochordal
diverticula of Actinotrocha might possibly be interpreted as
a pair of blind pharyngeal clefts in process of formation.

The objection to this theory is that it implies a discontinuity
in function, the pharyngeal clefts being incapable of performing
their true function until they were actually formed, an objection
which applies equally heavily to the hypothetical formation of
a new mouth and a new anus in the Vertebrata, and the
annelid derivation of this group.

The other alternative is to suppose that the pharyngeal clefts
were already present without chordoid support, and that the
lining cells became vacuolated later.

It is in these lowest Chordata that one would expect the
chordoid tissue to be diffuse and occurring in any part of the
hypoblast where it is specially needed (cf. Actinotrocha).
This explanation of the chordoid pharyngeal clefts appears to
me to be safe, and to leave out of the question the primary
origin of gill-slits.

It is remarkable that there are no gill-slits in Phoronis,
though reasons have been given elsewhere for their absence in
the adult (see below). In Actinotrocha there are a pair of
atrial grooves (see Actinotrocha) which function for the re-
moval of water brought along the oral grooves by ciliary action.
These grooves, therefore, have the function which the pharyn-
geal clefts perform in Cephalodiscus, and it is conceivable
that these grooves travelled back on either side in the course
of phyletic history of the Chordata, performing their function
the more efficiently as their point of exit from the pharynx
became carried further backwards. The tissue behind their
track would close up, and no trace of the migration of the grooves
(now clefts) would remain. There seems to me to be no more
difficulty in this process than in the migration of the origin of
a blood-vessel from one point to another of the parent vessel.

358 A. T. MASTERMAN.

In ontogeny one would hardly expect to find the exact process
repeated, for other structures may have been evolved in the
areas lying in the course of the migration, which appear early
in the ontogeny, but there are certain facts in the development
of gill-slits which may be explained in the light of this hypo-
thesis. Thus the gill-pouches of Tornaria first appear as
paired sac-like outgrowths of the cesophagus close to the mouth
region, so much so that if they reached to the overlying epi-
blast immediately on their appearance and perforated it, the
gill-slits would open through the lateral wall of the proboscis
(cf. Morgan, ‘ Journal of Morphology,’ January, 1894). Later
on they shift along with the gut itself posteriorly, so that by
the time they reach the epiblast they do so at a spot behind the
collar-pores, and in the anterior part of the trunk region. This
is the kind of modified repetition we should expect, in ontogeny,
of a phyletic migration of apertures from one part to another
(cf. formation of new mouth in Antedon and other types).

At any rate tentatively, I would regard the atrial grooves of
Actinotrocha as the early rudiments of pharyngeal clefts as
found in Cephalodiscus.

The structural condition of the endoderm in the pharyngeal
region of Actinotrocha, Cephalodiscus, and Balano-
glossus may be compared as follows:—In PI. 26, fig. 21, is
seen a diagrammatic transverse section of the pharynx of
Actinotrocha. In this case differentiation of the endoderm
has taken place. Two lateral areas are modified into chordoid
tissue for supporting function, whilst the ventral ciliated area
is more directly connected with alimentation. The two crosses
indicate the position homologous to that in which pharyngeal
clefts are situated in the other two types.

In Cephalodiscus (fig. 22) the same two lateral chordoid
areas are seen, though more approximated in the mid-dorsal
line. In addition, the two chordoid pharyngeal clefts open at
their ventral border, and below these, in the mid-ventral line,
is the special alimentary part or true gut.?

1 The right-hand side is supposed to indicate the condition behind the gill-
slits (cf. Actinotrocha, fig. 21).

ON THE DIPLOCHORDA. 359

In Balanoglossus there are very similar structures (fig. 23),
but the chordoid structure is no longer formed in the two
lateral diverticula of the respiratory portion, nor in the
pharyngeal clefts, which have now acquired the function of
gill-slits. In the former the chordoid tissue is only formed in
the anteriorly situated coalesced part, and in the latter case
the chordoid walls have been superseded or forestalled by the
cuticular branchial skeleton.

Fig. 24 is a diagrammatic transverse section of the pharynx
of a vertebrate, in which the development of the notochord is
assumed to resume its primitive order of appearance after
the mesoderm has been developed from the endoderm and has
taken up its proper position. The two rudiments of the diplo-
chordate condition have fused into one median dorsal notochord,
still, however, in continuity with the gut. Below this on either
side are the widely open gill-slits, and in the mid-ventral line
the ciliated groove, destined to be later reduced into the thyroid
gland (Gegenbaur).

In fig. 25 the adult condition of the notochord is reached, in
which it has (Eu-chorda) separated from the gut, and be-
comes an organ entirely distinct therefrom in form and func-
tion. In doing so it pushes dorsalwards into the hemocele
space, and the aorta is thus formed beneath it. (A portion of
the pharynx [stomodeum?] of Cephalodiscus in similar
manner protrudes dorsalwards, to come in contact with the
ectoderm of the dorsal surface, and in doing so divides the
dorsal blood-vessel into two lateral vessels [ef. Pl. 24, fig. 12].
An extension of this process would lead them to fuse in the
mid-ventral line. See Vascular system.) A further stage could
be instanced from the higher Vertebrata in which the thyroid
also, under the changed conditions of alimentation in the
Gnathostomata, loses its connection with the gut-wall.

We may thus trace the stages from the archichordate pharynx
with its walls supported by chordoid tissue, with pharyngeal
clefts for escape of the water-current, and with a ventral
alimentary portion for the conduction of food particles to the
highest Eu-chordate pharynx with the pharyngeal clefts formed

360 A. T. MASTERMAN.

into true gill-slits (branchial) with mesoblastic skeletal bars,
the chordoid tissue evolved into a supporting organ for body-
muscles, and eventually itself replaced by chondroid vertebral
tissue, and the ventral groove, the true alimentary part (gut)
in the Archichorda, losing its connection with the gut, and
becoming more or less vestigial as the thyroid gland. We
can clearly see some of the lines in relation to function along
which these changes have proceeded. In the Archichorda
the method of food ingestion is, typically, by ciliary action,
causing a current of water and minute food particles. ‘l'he
former has to be removed by pharyngeal clefts (Harmer, Brooks),
or by atrial grooves, in Actinotrocha, whilst the latter is
entangled in currents of slime, and thus retained and passed
down the gut for digestion. In Phoronis the same method
of ingestion is effected, but the water-current is got rid of by
a special adaptation of the epistome, to be described later,
before the mixed currents reach the true mouth, so that true
pharyngeal clefts should be superfluous ; and not only so, but
as the pharynx is an organ especially evolved for the effectual
separation of these two currents, it has entirely disappeared
in Phoronis, carrying with it its chordoid walls.

The same method of ingestion is pursued in the Uro-
chorda and even the Cephalochorda, though in each of
these, as in the Hemichorda, the simple pharyngeal clefts are
elaborated into a complicated system of branchial slits.

In the Holochorda the mesoblastic gill-bars are impressed
into the service of food ingestion, and the Gnathostomata
are evolved. The true jaws and elaborated locomotory system
enable more bulky prey to be secured, and the primitive func-
tion of pharyngeal clefts, that of removal of the water-current,
is no longer existent, so that the more recently acquired
branchial function alone warrants their maintenance. A few
exceptions are noteworthy. In the herrings and their allies,
the diet of Copepoda has given rise to a return of this func-
tion, and the gill-slits and gill-rakers are again requisitioned
for removal of water-current. Again, in the Cetacea, a
return to the pelagic diet involves the elaboration of a system

ON THE DIPLOCHORDA. 361

of whalebone bars, which, in an animal whose ancestors long
since lost their pharyngeal clefts, perform the primitive func-
tion of these organs.

To this change of diet (made possible by the mesoblastic
branchial bars!) we may trace the change of function of the
pharyngeal clefts, and upon a loss of this later function (bran-
chial) their extinction. On the other hand, the notochord
may be traced from chordoid tissue supporting the ingestive
pharynx to a fused chordoid rod forming the main skeletal axis
until it also is replaced by a mesoblastic skeletal organ.

The mid-ventral alimentary area of the Archichordate
pharynx, adapted for the secretion of mucus and the passage
of it, together with food-particles, eventually down the gut
(cf. endostyles, hypopharyngeal groove), loses its function in the
Gnathostomata, and retaining some other function of which
little is known, and which is not connected directly with that
of alimentation, is no longer in continuity with the gut.

As already mentioned, I propose to include Balanoglossus,
Cephalodiscus, and Phoronis together in one group, the
Archichorda, dividing this secondarily into two sub-groups:

1. Diplochorda,—Cephalodiscus, and Phoronis.

2. Hemichorda,—Balanoglossus.

Such differences as may hold between these three types can
be traced to their differing environment. Their common
meeting-ground is in a pelagic ancestor, very nearly represented
by Actinotrocha. We may suppose that this ancestor on
taking to life on the bottom developed a sucker on the ventral
surface, by which it was capable of fixing itself, more or less
permanently, to foreign objects. In the case of Phoronis,
the sedentary habit became complete, with a consequent
loss of pre-oral lobe, notochord, and sense-organs, and an
approximation of mouth and anus at the end furthest away
from the foreign object. In this particular case, the fixing
organ being ventral, the approximation is dorsal. In the case

1 Other factors in the evolution of the Gnathostomata are the increase in

size, made possible by greater correlation of parts and elaboration of the
muscular system,

362 A. T. MASTERMAN,

of Cephalodiscus the ventral sucker was also adapted for a
fixing organ, but in addition the pre-oral lobe became modified
for a like function, and the animal, protecting itself in a
spacious house or ccenecium, is most probably enabled to
travel about, somewhat after the manner of a leech, although
the budding function of the ventral sucker (or pedicle) points
to the fact that it is largely sedentary and fixed by this organ.
It is possibly the very intermediate character of this animal’s
habits, partly sedentary, partly locomotory, that has given it its
peculiar structure. In the reduplication of the gut and trunk
and the pre-oral position of the branching tentacles are to be
seen very marked sedentary characters, whilst the persistence
of the epistome and the notochords are to be accounted for by
a still functional locomotory capacity.

In the case of Balanoglossus, although the presence of
the ventral sucker in the young individual, and of a like organ
of attachment on the proboscis (Bateson), indicates a transitory
adoption of the distomial mode of progression, a burrowing
habit and locomotion in a longitudinal direction seems to have
been early adopted, with a loss of the ventral sucker and of
tentacles. At the same time, the burrowing habit is corre-
lated with the migration forwards of the notochord, with its
connected structures ; whilst the elongation of the trunk-region
and commencing metamerism are at least made possible by a
locomotory habit.

It is possible that the Echinodermata may be descended
from a form in many respects not unlike Actinotrocha, in
which fixation took place, not by the ventral sucker, but by the
pre-oral hood.

Cephalodiscus, isolated in its deep-sea ccencecium, has
therefore been removed from the active arena of life, where
fresh organs are evolved and primitive organs are modified,
whilst its own peculiar method of progression has saved it
from the fate following upon a completely sedentary habit, a
fate involving the loss of many important organs, as is shown
by the anatomy of its sedentary ally, Phoronis.

The resemblances of Cephalodiscus to Phoronis and to

ON THE DIPLOCHORDA. 363

Actinotrocha will have been noticed throughout this paper,
and that upon Actinotrocha. A detailed comparison of the
two species could be given, but the same purpose will be served
more succinctly by the subjoined list of characters of the
Archichorda and its two sub-groups.

ARCHICHORDA.

Body composed of three archimeric segments, protomere,
mesomeres, and metamere. Ectoderm simple, in great part
ciliated and glandular, secretes mucoid exoskeleton (tube,
cenecium). Nervous system still in connection with the ecto-
derm, consisting of central dorsal ganglion, pre-oral ring, post-
oral ring, dorsal and ventral cords, and, in addition, a more or
less diffuse nervous plexus. Mesoderm in four coelomic pouches,
the protoccele and metaceele showing secondary indications of
a paired condition. Protoccele opens to exterior, usually by
two proboscis-pores ; the mesocceles and collar-pores and the
metacceles have either paired nephridia, functioning as genital
ducts, or closed genital ducts. Muscular system prominent in
protoceele (the “ animal” organ), and in some a circular and
longitudinal layer in the metacceles. A mesodermic skeleton
of chondroid tissue—a vascular system of hemocele spaces,
consisting mainly of subneural sinus (heart)! near the dorsal
ganglion, dorsal and ventral vessels, and a sinus round the
gut. A simple digestive tube, with paired lateral (or early
fused into one) notochords, never free from the gut, and one or
more pairs of pharyngeal clefts. A subneural gland, opening
primarily into stomodeum, gonads confined to metacceles.
Metamere bears a ventral organ of attachment, ventral sucker.
Habitat burrowing or sedentary.

1. Hemichorda.—Body, especially the metameres, elon-
gated, and the latter showing in gonads and gill-slits traces of
true metameric segmentation. Well-developed muscular
system in metameres. Ventral sucker present only in young.
Notochords fixed and protruding forwards into protomere.

1 Only in the Eu- chorda, when the branchial gill-slits appear, is the typical
ventral heart of the Vertebrata found.

364 A. T. MASTERMAN.

Distal portion of subneural gland detached to form “ proboscis-
vesicle.” Burrowing habitat.

2. Diplochorda.—Mesomeres produced laterally into a
number of ciliated branchial tentacles, which in the adult point
upwards in front of the mouth, are supported by a chondroid
skeleton, and subserve ingestion of food. Metameres redupli-
cated by adorsal flexure. Stomodzum with subneural gland
still opening to exterior, and extending into the subneural
sinus. Paired proboscis-pores near median dorsal line, arising
internally along the wall of the subneural sinus. Paired
notochords in pharynx, not displaced forwards. A short
cesophagus, stomach, and intestine. One pair of pharyngeal
clefts may (Cephalodiscus) or may not (Rhabdopleura,
Phoronis) be present, with chordoid walls. Ventral sucker
forming the organ of attachment throughout life.

(1) Cephalodiscida.—Protomere persistent throughout life
as adhesive organ. Twelve pinnate plumes with eyes. Noto-
chords and chordoid gill-slits persistent. Ventral sucker forms
budding organ. Habitat, creeping, sedentary, and coeneecial.

(2) Phoronida.—Loss of protomere, atrial grooves, sub-
neural gland, and notochords in adult. Great development of
lophophoral tentacles (unbranched) and of chondroid tissue.
Paired nephridial apertures in metameres. Metamere elon-
gated, with circular and longitudinal muscles (as in Balano-
glossus). Permanent fixation by ventral sucker. Habitat
sedentary and tubicolous. Phoronis.

(3) Rhabdopleurida.—Protomere persistent. No noto-
chord (?)! nor pharyngeal clefts (?) in adult. Two pinnate
plumes.’ Attached by hypertrophied ventral sucker. Habitat
creeping, tubicolous. |

1 In the light of this interpretation of the organs of Cephalodiscus the
“notochord” of Rhabdopleura requires renewed investigation, as it evi-
dently corresponds to the subneural gland of the former. Rhabdopleura
should, at one time in its life, have paired pleurochords, as in Cephalo-
discus.

srs

ON THE DIPLOCHORDA. 365

List oF REFERENCES.

F, Harmer.—‘ “ Challenger ” Reports,’ vol. xx, Appendix to (2).

2. W. C. McInrosu.—‘ “ Challenger”? Reports,’ vol. xx, on Cephalo-

8. E.

discus dodecalophus,
A. AnpREws.— Journ. Morph.,’ vol. v, 1891.

4. A. Lane.—‘ Jenaische Zeitschrift,’ xxv (1890), pp. 1—12.

KEY TO PLATES 238—26,

Illustrating Mr. A. T. Masterman’s paper “On the Diplo-

Fig.

chorda: Part II. The Structure of Cephalodiscus.”

PLATE 23.

1.—Transverse section of the epistome of Cephalodiscus dode-

calophus, showing only the central area.

Fie.
Fic.
Fic.
Fic.
Fic.
Fic.
Fic.
Fic.

Fic.
Fic.
Fic.
Fic.

to cut

Fic.

2.—Ditto, from same series, next section.
3.—Ditto, ditto, third section.
4,.—Ditto, ditto, fourth section.
5,—Ditto, ditto, sixth section.

6.—Ditto, ditto, ninth section.
7.—Ditto, ditto, twelfth section.
8.—Ditto, ditto, seventeenth section.
9,—Ditto, ditto, eighteenth section.

PLATE 24.

10.—Ditto, ditto, nineteenth section.

11.—Ditto, ditto, twentieth section.

12.—Ditto, ditto, twenty-third section.

13.—Sagittal longitudinal section (nearly) to one side of median line
notochord, &e.

14.—Ditto, of same series as Fig. 13, left side nearly median.

PLATE 25.

Fics. 15, 16, 17.—Nearly transverse sections, in order from before back-

wards.

Fic.

18.—Transverse section of pharyngeal cleft, from a sagittal section of

same series as Figs. 13 and 14.

366 A. T. MASTERMAN.

Fic. 19.—Transverse section of lateral notochord, passing to pharyngeal
cleft on the left and pharyngeal wall on the right.

PLATE 26.

Fic. 20.—Diagrammatic median sagittal section of front part of Cepha-
lodiscus.

Fic. 21.—Diagrammatic transverse section of pharynx of Actinotrocha.

Fie. 22.—Ditto of Cephalodiscus.

Fic. 23.—Ditto of Balanoglossus.

Fic. 24.—Diagrammatic transverse section of stage in evolution of Verte-
brate (Eu-chordate) pharynx.

Fic. 25.—Ditto of pharynx of Hu-chordata.

Fic. 26.—Transverse section of a plume of Cephalodiscus.

Fie. 27.—Ditto of a pinna of Cephalodiscus.

Figs. 28, 29.—Refractive globules isolated from the “eyes.”

Fic. 30.—Longitudinal section of branchial eye of Cephalodiscus.

Fic. 31.—Ditto of single cell from branchial eye of Cephalodiscus.

Fics, 32—36.—Serial transverse sections of plumes and collar region.

ABBREVIATIONS.

b.c.1. Coelomic cavity of epistome. 4.¢.3. Trunk celomic cavity. 0.5.
Blood-sinus. c.p. Collar-pore. d..v. Dorsal blood-vessel. d. mes. Dorsal
mesentery. ep. Epistome. gd. Gonaduct. g.s. Gill-slit (pharyngeal cleft).
int. Intestine. 7/.6.v. Lateral blood-vessel. /.c.c. Left collar-cavity
(coelomic). 7.2. Lateral nerve. m. Mouth. zch. Notochord. z.g. Nerve
ganglion. 2. Nucleus. o.g. Oral grooves. ov. Ovary. ph. Pharynx.
p.o.l. Post-oral lamella. p.o.. Pre-oral nerve. *.c.c. Right collar cavity
(ceelomic). sh. Sheath of notochord. s¢. Stomach. s.#.g. Subneural gland.
s.#.8. Subneural sinus. ¢. Plume.

FEB 2g 1898

NOTE ON A NEW BRITISH ECHIUROID GEPHYREAN. 367

Note on a New British Echiuroid Gephyrean,
with Remarks on the Genera Thalassema
and Hamingia.

By

W. A. Herdman, D.Sc., F.R.S.,
Professor of Natural History in University College, Liverpool.

With Plates 27 and 28.

On the dredging expeditions of the Liverpool Marine Biology
Committee we endeavour as much as possible to try new
methods of collecting, or to modify old methods, with the
object of securing animals whose special habitat renders them
difficult of capture. With this end in view, when dredging
last Easter (April, 1896) in the deep depression that runs
down between the Isle of Man and Ireland, it occurred to me
to try a haul of the large fish-trawl (50-foot beam) which we
had on board, but to steam very much more slowly than is
customary when trawling, so as to allow the foot-rope to dig
deeply into the stiff blue mud which I knew we were on. We
went dead slow with plenty of the heavy steel warp out, the
result being—as the event proved—that a good deal of the
warp lay on the bottom, and so made the pull on the trawl
practically a horizontal one. As a consequence of this we
must have sliced off the top six inches to a foot, possibly, of
the bottom deposit as we went along. When the trawl came
up it was half full of the blue mud, and we found it impossible
to get the net on board. After steaming round for a little a
good deal of the mud was washed out through the meshes,

vol, 40, PART 3.—NEW SER. co

368 W. A. HERDMAN.

and the cod-end was then hoisted on board by means of the
tackle, and ripped up so as to precipitate upon the deck a
seething mass of mud, spawning fish,’ and Invertebrates, which
spread over a considerable area to a depth of about a foot.

While the cod-end was still hanging from the tackle,
dripping mud and fish-spawn, I noticed and managed to secure
a soft, green-coloured, worm-like object which was protruding
from one of the meshes. A careful search through the mud
on deck afterwards gave me two other similar green objects—
evidently all of them pieces of a large Gephyrean worm.
They were still alive when we landed that evening at Port Erin,
and were at once transferred to a vessel of fresh sea water in
the Biological Station. It was too late to see anything more
of them that night, and the next morning they were so feeble
that after watching their sluggish movements for a time and
making a few sketches, I considered it best to kill them. I
put two of the pieces in a 5 per cent. solution of formol with
the view of preserving the beautiful green colour, while the
third was put in strong spirit for anatomical purposes.

The particulars of the locality where the specimens were
obtained are as follows: .

Twelve miles south-west of the Chicken Rock, Isle of Man,
bottom “‘reamy” mud, depth 40 to 50 fathoms, beam-trawl,
on L.M.B.C. expedition of April 5th, 1896.

IDENTIFICATION.

On the first glance I supposed that the specimens were a
species of Thalassema, and I spoke of them as such to Mr.
Gamble and the other naturalists who were with me. But
the following day, being struck by the difference in appearance
from the common British Thalassema Neptuni, and in-
fluenced by the striking green colour, and by the observation
that the proboscis was not perfect, I thought that they more
probably belonged to a Bonellia. So in writing shortly

1 This locality, on the mud, about March and April, is a very important

spawning ground for food fishes. On this occasion we obtained ripe haddock,
ling, hake, plaice, and witches, with the eggs running out freely.

NOTE ON A NEW BRITISH ECHIUROID GHPAHYREAN. 369

afterwards to Canon Norman I told him of the specimens,
and stated that I thought they were probably pieces of two or
three large Bonellia viridis. On looking at the specimens
more carefully, however, and on talking the matter over with
Professor Lankester, I came to the conclusion that, from the
proportions of the body and proboscis (even supposing the
latter to be imperfect at the extremity), it was impossible for
the species to bea Bonellia, and that it was much more likely
to belong to Thalassema, as I had at first supposed. The
detailed anatomical examination which I have since made
proves this latter view to be correct. The specimens belong
either to a new species of Thalassema, which comes near
T. gigas, M. Miller, from the Adriatic, or represent a new
genus close to Thalassema in the direction of Hamingia.
I incline to the former view (see below, p. 381), and shall
describe the species as a Thalassema.

ANATOMY.

The three specimens in my possession are all imperfect.
They are certainly pieces of two, and possibly of three distinct
worms. I believe they make up between them a complete
body, or very nearly so; there may be a very short piece in
the middle of the body missing, but it cannot be much, or of
any anatomical importance. Consequently the restoration
shown in fig. 4 must be fairly correct. The dimensions and
positions in the body of the fragments are as follows :

a. Anterior half (?) with proboscis (probably incomplete at
tip) ; body about 3 cm., proboscis about 9 cm. in length ;
preserved in formol (see Pl. 27, fig. 1).

B. Posterior two thirds (?) of the body, measuring about
8 cm. in length; preserved in formol (Pl. 27, fig. 2).

c. Anterior half or more (?) with proboscis (probably com-
plete) ; body about 8:5 cm., proboscis about 8 cm. in length;
preserved in spirit (Pl. 27, fig. 3).

Possibly a and B may be the two halves of the same speci-
cimen, and even if the body be longer than I take it to be,
B and c together probably represent the whole of it, even if

370 WwW. A. HERDMAN.

they do not belong to the same specimen. If the proboscis
of a, which seems to have a natural appearance and shape
(formol specimen), such as it had when alive, is incomplete at
the tip, that of c, which is somewhat distorted (spirit spe-
cimen) about the middle, has every appearance of being perfect
at its extremity. Consequently I believe that the three frag-
ments represent all parts, and show us the full characters of the
species so far as is shown by the adult female (see fig. 4).
The description is as follows:

Shape.—The body is elongated and irregularly cylindrical,
rather wider anteriorly than posteriorly, with somewhat undu-
lating outline along the sides, and with occasional irregular
swellings (see fig. 2). At the anterior end it is rounded, and
narrows to join the constricted base of the proboscis (fig. 3).
The posterior end is evenly rounded. The proboscis springs
from the anterior end of the body by a narrow neck, which
rapidly widens, while a slit appears in the median ventral line,
which widens to form a shallow groove, and then opens out to
become the flattened ventral surface of the greater part of the
proboscis (fig. 5). Along its whole length the edges of the
proboscis are slightly incurved, so that the ventral surface is
somewhat concave and the dorsal somewhat convex. The
distal portion may be rolled up (see fig. 1), and when unrolled
the tip is excavated in the middle, so as to leave two projecting
horns at the sides (fig. 8). In transverse section the proboscis
is at its base cylindrical, then becomes a deep groove nearly
closed in, then an open groove, and finally a nearly flat plate
with incurved edges (figs. 1 and 5).

Surface.—That of the body is tuberculated all over; that
of the proboscis is smooth, and when alive of glutinous and
slimy appearance.

Colour.—The body is of a beautiful and nearly uniform
apple-green,! which has the appearance of being quite on the
surface of the integument. The proboscis is not so deeply

' Described by different experienced observers as “chrome” green and

“apple” green. Professor Lankester tells me it is exactly the colour of
Hamingia.

NOTE ON A NEW BRITISH ECHIUROID GEPHYREAN. 371

coloured, and the colouring is not so uniform, but is rather
streaky. It is greener along the edges and paler (pale greyish
green, becoming in places almost a dirty white) along the centre.

Size.—The length of the body when complete was probably
about 12 em.; the thickness varies from 5 to 15 mm. The
length of the proboscis is fully 9 cm., and its breadth in the
flattened part varies from 15 to fully 20 mm. The narrow
basal part is about 5 mm. in diameter.

Apertures.—The mouth is placed at the anterior end of
the body, in the median ventral line, beneath the base of the
proboscis (which is a pre-oral lobe). The cloacal opening is
median and posterior (Pl. 27, fig. 6).

The two genital apertures are placed ventrally, one at each
side of the middle line, and a little behind the mouth. Each
genital aperture has, projecting from the integument beside it,
a large curved seta, or genital bristle, of a golden colour
(figs. 5 and 13).

Integument and Muscles.—The body-wall is thick,
especially at the posterior end. It varies from 1 mm. on the
thinnest part of the sides to as much as 4 mm, near the cloacal
aperture; and has the usual layers—cuticle, epidermis, con-
nective tissue, three muscle layers, and the lining of the ccelo-
mic cavity. Under the epidermis there is a thick gelatinous
dermal layer (Pl. 27, fig. 14, d., in which the green colour is
found), which may be up to 3.mm. in thickness. It is this layer
which gives rise to the variations in thickness of the body-wall.
The pigment is present in the form of accumulations of small
rounded masses (fig. 14, p.). These may extend to the bases
of the epidermal cells, or even up between them, but always
belong to the dermis. Inside the gelatinous dermal layer
come the muscular layers of the body-wall, two circular layers
separated by a longitudinal, which are not broken up into
definite regularly arranged bundles as in the case of some
Gephyrea, but form a practically continuous layer of muscle-
fibres over the whole inner surface, becoming rather thicker
at the posterior end around the cloaca. Special muscle
bundles run in radial fashion from the wall of the cloaca to the

3872 W. A. HERDMAN.

neighbouring body-wall. Special bundles are also attached to
the inner ends of the genital sete, which project freely for the
greater portion of their length into the ccelomic cavity.

Mesenteries containing delicate muscle-fibres project inwards
from various points on the body-wall, and sling the loops of the
alimentary canal in position (fig. 15). The mesenteries are deli-
cate and silky in appearance, and are much folded and crumpled
so as to be wisp-like. This appearance is partly caused by the
thickening of the celomic epithelium in irregular ridges
and masses on the mesenteries (fig. 12). The end of the
mesentery which arises from the body-wall is clear, trans-
parent, and membranous, while the end which is attached to
the wall of the gut is grey, opaque, and like a wisp of spun
silk (figs. 1O—12).

Alimentary Canal.—The gut is long and much convo-
luted (figs. 7—9, 15). It can scarcely be divided into
regions. The anteriorly placed mouth leads into a pharynx
which cannot be called dilated—one of the points in which
this form differs from Hamingia arctica, D. and K. The
gut performs several close convolutions in the neighbourhood
of the genital setz, and then stretches backward to the level
of the posterior ends of the anterior nephridia (uteri), where it
again coils upon itself (fig.9). It then extends backward once
more and enters the posterior much convoluted part of its
course, in which it runs almost to the posterior end of the
celom, and then forms first one and then another loop
directed anteriorly before becoming the rectum, which extends
down to the cloacal aperture (figs. 15 and 16). The intestine
as a whole is not so wide relatively to its length as in the case
of Hamingia arctica.

Posterior Nephridia.—These dense tufts of white twigs
are placed one at each side of the rectum, and open into the
cloaca. Each organ consists of a single, central, thick-walled,
opaque brown tube, about 12 mm. in length, which gives off
an immense number of delicate, opaque, white-coloured twigs,
which interlace with one another so as to give rise to the bush-
like appearance (figs. 15—18). These white twigs, however,

NOTE ON A NEW BRITISH ECHIUROID GEPHYREAN. 373

do not, so far as I can determine, ever branch again. They
are very long, and are closely intertwined in their proximal
parts (see figs. 19—21), so as to be difficult to follow ;
but all those I have teased out seem, like the one shown in
fig. 22, to be free from the neighbouring twigs. The appear-
ance of repeated branching seen in figs. 19 and 20 is due to
frequent crossing of the twigs. This is a case, then, of what
Rietsch! calls a simple ramification, in distinction from both
the unbranched tube of some other species of Thalassema?
(e.g. T. Neptuni) and the doubly ramified organ of Bo-
nellia viridis. Rietsch figures and describes the posterior
nephridium of Bonellia minor as being simply ramified, but
the twigs in that case are not nearly so long nor so numerous
as in the present species. In anatomical condition, then, so
far as this organ is concerned, the present form lies between
Bonellia viridis and B. minor, and is certainly in this one
character more like that genus than like the typical Thalas-
sema. The posterior nephridia of Hamingia arctica seem,
from the figures and description of Danielssen and Koren,’ to
resemble very closely our form.

Each of the milk-white twigs bears a little bell-shaped funnel
or nephrostome at its distal end (figs. 21 and 23). The margin
of the funnel has a thickened ciliated rim (fig. 24). This
arrangement is very like that of Bonellia minor as figured
by Rietsch. The further details of appearance are sufficiently
shown by my figures (PI. 28, figs. 19—24). The nephrostomes,
then, are very numerous, and open freely by wide ciliated
mouths into the body-cavity.

Anterior Nephridia.—These also are a single pair.
They lie one at each side of the alimentary canal ventrally, in
the anterior one fourth or so of the body (figs. 7, 8, 25, ).
They are large, and in both my specimens are distended with

1 Recueil Zool. Suisse,’ t. ili, 1886.

* T. gigas has nephridia which, from Miller’s figure, seem to resemble
those of our form.

3 ‘Gephyrea of Norwegian North Atlantic Expedition, 1876-8,’ Chris-
tiania, 1881.

374. WwW. A. HERDMAN.

ripe ova. ‘hey have the usual structure of uterine nephridia
in the Echiuroidea, and consist of (see fig. 26) (1) a simple
external aperture (nephridiopore), (2) a much-coiled slit-
like internal opening (nephrostome, n. s.), (3) a short wide tube
leading to (4) a globular dilatation (w.), in which the eggs lie,
followed by (5) a long, narrower czcal tube hanging down
posteriorly into the body-cavity. The eggs are chiefly in the
globular uterine dilatation (figs. 25, 26, and 31), and are very
distinctly seen through the transparent but rather tough
walls. The nephrostome is a striking object. The narrow
slit is drawn out to a great length to form two horns,
which are then each coiled up spirally (as shown in figs.
26—28), to form a structure closely similar to the dorsal
tubercle at the opening of the hypophysial gland in a Cynthiid
Ascidian. As in the case of the Ascidian dorsal tubercle,
however, there seems to be considerable individual variation.
My second specimen is not so much drawn out laterally, and
not so much coiled (see figs. 29 and 30). The external apertures
of the anterior nephridia (genital) are placed close together
near the ventral middle line of the body, and each is provided
with a strong genital seta, embedded in the body-wall and pro-
_ vided with special muscle bundles. These genital setze are of

a burnished golden appearance where they project to the ex-
terior. The shape is shown in fig. 18, and the entire length
of the seta is 5°55 mm. The inner end, which projects into the
body-cavity and has muscle bundles attached to it, is wider,
softer, and of a white colour.

Gonad, &c.—Both my specimens were females. The ovary
in each case was distinctly visible, running along the upper
surface of the posterior part of the nerve-cord and ventral
vessel (figs. 17 and 18, ov.). A few ova were found floating
freely in the ceelomic cavity, and, as has been noted above, the
anterior nephridia or uteri contained large numbers of ova.
The globular dilatation was in one specimen packed full (fig.
31), and in that case about 120 ova were visible on the one
side of the vesicle. The ovarian ovum not quite ripe has a
central more granular and opaque part, in which the germinal

NOTE ON A NEW BRITISH ECHIUROID GEPHYREAN. 375

vesicle and spot are distinctly visible (tig. 33), and a peri-
pheral clearer zone of considerable width. The ripe ova from
the uterus (fig. 32) are more opaque, and are almost filled
with yellow granules, the germinal vesicle being no longer
visible, and the clear peripheral zone very narrow (fig. 34).
I was able by slight pressure to squeeze a line of ova out from
the uterus through the nephridiopore, as shown in fig. 26.

Nothing specially noteworthy was noticed in connection
with the blood-vessels, the nerve system, and other organs.

No coloured corpuscles were found in the ccelomic fluid,
and no rudimentary males were seen. These and other mor-
phological points are more fully discussed below in determin-
ing the systematic position (see p. 378).

Tue Green PIGMENT.

One of the most remarkable and interesting points about
this worm is certainly its beautiful green pigmentation. When
alive the colour was a well-marked apple-green, and the speci-
mens preserved in formol have retained a good deal of the
colour, although the fluid has also become coloured green. The
specimen put in spirit has lost its colour altogether, and the
spirit has not acquired any green tint ; while in the formol, on
the other hand, the fluid has become coloured, and the specimens
have retained their green appearance. I have used the green
formol solution for an examination of the spectroscopic charac-
ters of the pigment, and my friends Professor Sherrington and
Dr. Noél Paton have kindly investigated samples for me, and
have given me their results. I am indebted to these physio-
logists and their assistants for the information that follows.
Professor Sherrington states :

“The dilute formol in which the Thalassema has been lying was clear, but
tinted a pale greenish blue,—in fact, a “sea” colour. Examined in a layer
20 centimetres deep before a Hilger single flint-prism spectroscope illuminated
by a Welsbach incandescent gas lamp, the following localisation of absorption
was obtained (see Pl. 28, fig. 35). From the violet end up to nearly as far as
the solar line F, from the red end up to the solar a, between the solar lines
C and D a single very definite broad band of shadow. By diluting the fluid
or diminishing the thickness of the layer examined until the band between

376 W. A. HERDMAN.

C and D just disappeared, the absorption at the ends was not greatly dimin-
ished. On again increasing the thickness of the layer the point of first
appearance of the single band was rather nearer to D than C, i. e. at \ 617.

«The measurements obtained were as follows :

Stronger solution, absorption from violet end to A 468.
absorption from red end to d 716.
absorption between A 630 and A 602.
Solution just too weak to give any absorption between C and D gave
absorption from violet end to A 428.
absorption from red end to A 725.

“The colour was not affected by weak reducing or weak oxidising agents,
and no evidence has been obtained indicating that the pigment is of respi-
ratory function. It is not easily bleached; in formol solution exposure at a
south window for nine weeks has made no perceptible difference to its depth
of tint, as compared with a similar tube-full preserved during that period in a
dark cupboard. The localisation of the absorption points to the pigment
being one not hitherto met with, at least not hitherto recorded.”

Later Professor Sherrington added:

“ Hemoglobin in formol solution exhibits the spectrum of reduced hemo-
globin. There is no similarity between the spectrum of the pigment here
examined and that of hemoglobin. On the other hand, the position of the
band recalls that of the strong band given by ‘bonellein,’ A 643 to A 617
(Sorby). But bonellein was not examined in formol solution. No other
definite absorption band was given by the Thalassema pigment in formol. ‘The
substance is not a respiratory pigment. The spectral band-shadow sug -
gests alliance with bonellein..... Of course there is no question of the
identity of this with bonellein; the only thing the spectral map does tell
us is that this pigment cannot be the same as bonellein, unless—which is very
unlikely—bonellein has a single band in formol solution, or this a multiple
shadow in CS, or C,H,0.”

Dr. Noél Paton writes to me:

“Dr. Milne Murray and I have examined the solution after evaporating to
about one half and placing it in a 3-inch tube. We find a single band
with ill-defined edges in the red, with its centre at A 640. It was impossible
to fix the exact position of the edges. There is very little absorption of the
red end of the spectrum, which can be seenup to A 790; but there is consider-
able absorption of the violet end—up to about 4 496. I shall send the fluid
back to you, unless you would like me to ask Miss Newbigin, who is working
at the pigments of Invertebrates, to see if it has any of the characters of a
lipochrome.”

_ Lasked Dr. Noél Paton to hand the sample of fluid over to
Miss Newbigin, who has since sent me the following report :

/

NOTE ON A NEW BRITISH ECHIUROID GEPHYREAN. 377

“ Report on GREEN PIGMENT OF THALASSEMA.

“The pigment was received in the form of a solution in formalin. The
solution was a dull green colour, and did not display any fluorescence, as do
solutions of bonellein or chlorophyll.

“Action of Heat.—The solution underwent no change on boiling. When
evaporated to dryness in a water-bath a dull green pigment was left behind.

“Action of Ether and Aleohol.—When shaken in a separation funnel
with ether, the ether did not remove any pigment from the solution. The
addition of alcohol to the solution produced a green stringy precipitate,
soluble in water, slightly soluble in excess of alcohol. The pigment obtained
by the evaporation of the formalin extract is soluble in alcohol.

“Action of Acids.—Acetic, hydrochloric, and nitric acids produced no
change. This is in marked contrast to the reactions given by bonellein to
acids. According to Krukenberg (‘ Vergleich. Physiol. Studien,’ Ilte Reile,
IIte Abtheil., pp. 70—80), that pigment in alcoholic solution turns violet with
strong acids, whether organic or inorganic; and on the further addition of acid
turns blue, both solutions having definite spectroscopic characters. The
green colour was restored on the addition of alkali.

“If the green Thalassema pigment be boiled with concentrated nitric
acid, the solution turns first yellow, and then evolves nitrous fumes and
becomes a clear green.

“Action of Alkalies.—Alkaline solutions, such as caustic soda and am-
monia, in large part precipitated the green pigment as a green stringy mass,
The precipitate was insoluble in excess of alkali, and only slightly soluble in
water or methylated spirit.

“When this methylated spirit solution was evaporated to dryness it left a
yellowish rather than a green residue. When this residue was treated with
strong nitric acid the acid became yellow, and then nitrous fumes were evolved
and the pigment became green. ‘The residue obtained by evaporating the
formalin solution to dryness did not give this reaction so readily.

«HS seemed to have no effect on the pigment.

“ Conclusion.—If the absence of the red fluorescence, and of the reaction
with acids described by Krukenberg, are to be relied upon, then this green
pigment is not bonellein. Again, the absence of fluorescence, the absence of
an associated lipochrome, and the colour are evidence against the supposition
that it is chlorophyll. ‘The only reaction given by the pigment which is at
all distinctive is the vivid green coloration with concentrated nitric acid. This
is a reaction given apparently by all of a little known series of pigments,
forming the hepatochromes of Krukenberg and the enterochlorophylls of
MacMunn, which occur in the livers and digestive glands of many Inverte-
brates, and are either of a green or a yellow colour. ‘The green does not give
the reaction with nitric acid so well as the yellow, but seems to be readily
convertible into the yellow. According to Krukenberg, there is no evidence

378 W. A. HERDMAN.

that they are really of the nature of chlorophyll. ‘They are apparently useless
substances, eliminated either by the gut or by the skin, or sometimes by both.
“M. T. Newsiain.”

Professor Sherrington and Dr. Noél Paton, as is shown
above, differ somewhat in their localisation of the absorption
band, but that may possibly be due to a change caused by the
evaporation, to about one half, of the formol solution.

The result appears to be that this substance, which may be
appropriately named ‘“‘ Thalassemin,” is a remarkable and appa-
rently unknown pigment, which is not allied to hemoglobin nor
tochlorophyll. It is not a respiratory pigment, and is apparently
nearer to “ bonellein,” described by Dr. Sorby in 1875,' from
the Gephyrean Bonellia viridis, than to any other known
pigment; but differs markedly from bonellein in several re-
spects, besides having a single in place of a multiple band, and
so cannot be identical withthat substance. Possibly thalassemin
is more closely related to the green pigment of Hamingia.

The chart showing the spectral characters drawn up by
Professor Sherrington is given in Pl. 28, fig. 35. I may add
that this pigment is so intense that even in the case of the
specimen from which the formol solution referred to above
had been extracted, the green colour is still distinctly visible
in the pigment masses of the dermis in thin sections (about
8 uw) under Zeiss’s ;4; oil immersion.

SysTEMATIC PosITION.

In 1883 Professor Lankester gave an account of some
specimens of Hamingia arctica, Kor. and Dan., from the
Hardanger Fjord, and discussed fully the relations of the
genus Hamingia to Thalassema and Bonellia. He con-
sidered Horst’s H. glacialis to be the same as Koren and
Danielssen’s H. arctica; and from the additional specimens
he and Canon Norman dredged in the south of Norway he
was able to add considerably to our knowledge of the struc-
ture of this form, and of the differences between it and the

1 ¢Quart. Journ. Micros. Sci.,’ N.S., vol. xv, p. 166.

NOTE ON A NEW BRITISH ECHIUROID GEPHYREAN. 3879

neighbouring genera Thalassema and Bonellia. He dis-
covered and figured the minute worm-like male living as a
parasite in the dilated pharynx of the female Hamingia.
Lankester regards Hamingia as having a closer resemblance
to Bonellia in internal organs, and to Thalassema in ex-
ternal characters; while it is quite peculiar in having the
genital pores on prominent papilla, and in the absence of
genital sete in the female. He gives a useful tabular state-
ment of the characters of the three genera, which I shall make
use of in discussing the position of our new form.

The specimens I am describing from the Irish Sea are inter-
mediate in their characters between the genera Thalassema
and Hamingia as defined by Lankester, but in my opinion
they come nearer to Thalassema—so near, in fact, that I
think they must constitute a species of that genus. I shall
take up the characters seriatim as given by Lankester in
his tabular statement.

1. In shape of body the new form agrees equally well with
Hamingia (since Lankester showed that that form possessed
a large proboscis, as long as the body) and with various species
of Thalassema. It perhaps most nearly resembles in size
and shape the T. gigas of M. Miller, from the Adriatic.

2. The proboscis agrees equally with the characters of the
two genera.

3. The female genital pores in Hamingia are one or two
in number, and open on well-marked papilla. In Thalassema
(according to Lankester) the female genital pores are four to
six, and are not placed on papille. In our new form there
are no papille, but only two pores are present. In this
character, then, our form is intermediate between the two
genera. In Thalassema gigas and T. faex, however, there
is said to be only one pair of ‘‘segmental organs” (see Greef,
Selenka, and K. Lampert), and therefore, I take it, only one
pair of female apertures. I should alter, then, this character?

1 «Ann. and Mag. N. H..,’ vol. xi, 1883, p. 42.
? This extension of the diagnosis is recognised by Rietsch, Selenka, and
others.

380 W. A. HERDMAN.

of the genus Thalassema, as given by Lankester, to read,
“ Uteri (enlarged nephridia) and female genital pores from one
to six pairs, not opening on papille.” That would admit
our new species.

4. In Hamingia the male is a minute parasite on the
female, asin Bonellia. In Thalassema (so far as is known)
the males and females are alike in size and appearance. The
three pieces of our new form all belong to mature females,
and after a careful search in uteri, pharynx, and proboscis, I
am unable to find any minute males. This character, then,
does not help us. Minute males or full-sized males may yet
be found.

5. A pair of strong genital sete are present in our species,
and in this respect it agrees with Thalassema, and differs
from Hamingia.

6. The ova are not enclosed in follicle cells. This is in
agreement with other species of Thalassema.

7. There are no distinct “zones” in the mature ovum, such
as are present in Hamingia.

8. The uterine pouches when distended have hyaline, trans-
parent walls, but still they are firm and resistent. In this
respect our form seems to combine the characters of Hamingia
‘and Thalassema, as given by Lankester.

9. The internal opening of the uterine pouch (nephridium)
is drawn out into a spiral trough, as in Thalassema.

10. The anterior part of the pharynx is not dilated, thus
agreeing with Thalassema.

11. The cloacal nephridia are densely covered with long
intertwined branches, so as to look like little bushes, and the
nephrostomes are on the ends of the twigs. In this respect
our form comes nearer to Hamingia and Bonellia, but does
not exactly agree with either genus as previously described.

12. No red-coloured corpuscles were found in the ccelomic
fluid. The fluid was slightly milky in colour. In the absence
of hemoglobin our form differs from Hamingia, and agrees
with at any rate some species of Thalassema.

To sum up: our new species agrees with Hamingia in

NOTE ON A NEW BRITISH ECHIUROID GEPHYREAN. 381

having branched cloacal nephridia; while in all other respects
it either agrees with or comes nearer to Thalassema.

Professor Lankester, after seeing one of my specimens and
hearing the details of structure as given above, has suggested
that I should describe this form as a new genus intermediate
between Hamingia and Thalassema, and forming a term
in the series: Echiurus, Thalassema, the new form, Ha-
mingia, Bonellia. Professor Lankester’s opinion must
always carry great weight, especially in regard to this group
which he had already done so much to elucidate; but in the
present matter I am inclined to think, from the consideration
of the characters given above, that my new form is distinctly
nearer to Thalassema than to Hamingia, and may without
any violence be included in the former genus. Consequently
I prefer to describe it as a new species of Thalassema,
related to T. gigas, M. Miller, rather than to form an inde-
pendent genus for its reception. I desire to associate Professor
Lankester’s name with this species. It seems appropriate, as
he has written on the two genera to which this form is related.
The specific diagnosis will run as follows :—

Thalassema Lankesteri, n. sp.:

Length about 20 cm. Proboscis nearly as long as trunk,
and in most of its extent wider. Tip of proboscis trun-
cated and slightly indented. Surface evenly tuberculated
all over. Colour apple-green on the trunk; paler on
proboscis. Longitudinal musculature not divided into
bundles. A single pair of anterior nephridia; nephro-
stomes spirally twisted. Cloacal nephridia branched, with
numerous ciliated funnels on the ends of the branches.
Females alone known. Off Isle of Man, fifty fathoms.

As a species, T. Lankesteri is undoubtedly distinct from

the previously described species of Thalassema. Apart from
the anatomical peculiarities noted above, the external charac-
ters sufficiently distinguish the species. According to Lam-
pert’s and Rietsch’s synoptic tables our form would be
grouped, from the characters of the body musculature and the

382 W. A. HERDMAN.

nephridia, along with T. gigas,! M. Mill., taken off Trieste,
from which it may be distinguished by the proportions and
shape of the proboscis, which is much wider compared with
the body than in T. gigas, and is not trilobed at the tip.

The other species which have been described as being more
or less of a green colour are—

(1) T. Baronii, Greef, from the Canaries; but there the
probascis is relatively much smaller, and, moreover, there are
important anatomical differences. T. Baronii has two pairs
of anterior nephridia, and a different arrangement of muscles.

(2) T. Moebii, Greef, from Mauritius, with three pairs of
anterior nephridia.

(3) T. viridis, Verrill, off north-east coast of America, with
small (6 mm.) swollen body and long slender proboscis.

The remaining species of the genus which have been found
in the North Atlantic or Mediterranean are T. Neptuni,
Gaertner (British), T. Frohmanni, Diesing (Sicily), and T.
faex, Selenka (between Scotland and the Faroe Islands). The
first of these has two pairs of anterior nephridia; the last
agrees with our species in having only a single pair, but
differs totally in the structure of the cloacal nephridia, as well
as in colour, shape, and appearance generally; while T.
Frohmanni, from the Mediterranean, is insufficiently known,
and may be the same as T. Neptuni. However that may be,
T. Frohmanni cannot be confused with T. Lankesteri; the
short description given by Diesing is sufficient to show that
our form differs from his both in colour and shape. As a
matter of fact, although several of the above-named species
have been referred to as more or less green, none of them,
judging from the coloured plates given by Greef and others,
are so intensely and completely green as our specimens. ‘T.
Lankesteri in its coloration is probably more like Ha-
mingia arctica than any other known form.

1 ‘Observationes anatomice de Vermibus quibusdam maritimis, Berlini.’
T. gigas was originally found at Trieste, and is still taken there. I believe
it has not been found elsewhere; and, as in the case of T. Lankesteri, only
females are known.

NOTE ON A NEW BRITISH ECHIUROID GEPHYREAN. 383

EXPLANATION OF PLATES 27 & 28,

Illustrating Professor W. A. Herdman’s paper on “A New
British Echiuroid, Thalassema Lankesteri, n. sp.
(Herdman).”

REFERENCE LETTERS.

cl. Cloacal aperture. ¢. Cuticle. ¢. 2. Cloacal nephridia. ep. Epidermis.
d. Dermal layer of skin. /f. Ciliated funnels of posterior nephridia. n¢.
Intestine. 7. ~. Left anterior nephridium. m. Mouth. mes. Mesentery.
musc. Muscular layer of body-wall. . Anterior nephridium., 2. c. Ventral
nerve-cord. z.s. Nephrostome. @s. @sophagus. ov. Ovary. p. Groups of
rounded pigment masses of a green colour in the dermal layer of skin. ap. /.
Pre-oral lobe or “ proboscis.” 7. 2. Right anterior nephridium. 7. Rectum,
s. Genital seta. «. Uterine part of nephridium. vv. v. Ventral blood-vessel.
w. Body-wall.

PLATE 27.
Thalassema Lankesteri, n. sp. (Herdman).

Fic. 1.—Specimen A. Natural size, from left dorsal side.

Fic. 2.—Specimen B. Natural size, showing the green colour.

Fic. 3.—Specimen C. Natural size, from ventral side.

Fic. 4,—Restoration, to show probable appearance of the species.

Fic. 5.—Ventral view of anterior end of Specimen A, to show base of pro-
boscis and genital sete.

Fic. 6.—Posterior end and cloacal aperture, to show papille on surface.
x 2.

Fic. 7.—Anterior end of spe¢imen, dissected from right side, to show
alimentary canal and other organs.

Fic. 8.—Anterior part of alimentary canal, with anterior nephridia, &c.,
isolated, from ventral surface.

Fic. 9.—Diagram of course of anterior part of alimentary canal.

Fic. 10.—A few convolutions of alimentary canal, to show the silky
mesenteries.

Fie. 11.—Piece of a mesentery, enlarged.

Fic. 12.—Small part of same mesentery, magnified (x 50).

Fie. 13.—One of the genital sete, enlarged.

Fic. 14.—Part of a section through the skin, to show the position of
the pigment masses (p.) which give rise to the green colour (x 300).

voL. 40, PART 3.—NEW SER. DD

384

W. A. HERDMAN.

PLATE 28.

Fig. 15.—Dissection to show posterior part of alimentary canal.

Fic. 16.—Dissection to show cloaca and posterior nephridia.

Fic. 17.—Rectum reflected to show posterior nephridia opening into cloaca.

Fic. 18.—Posterior nephridium and ovary.

Fie. 19.—Posterior nephridium isolated and enlarged.

Fig. 20.—Same cut across, enlarged.

Fig. 21.—Small portion of same teased and mounted (x 50).

Fic. 22.—Single branch bearing ciliated funnel, teased out and isolated
(x 100).

Fic. 23.—Two of the ciliated funnels.

Fie. 24.—Single ciliated funnel, highly magnified (x 300).

Fic. 25.—Dissection from right dorsal side, to show pair of anterior
nephridia in situ.

Fic. 26.—Single anterior nephridium, enlarged.

Fic. 27.—Vesicle and nephrostome from one end.

Fic. 28.—Same nephrostome from the side (x 50).

Fic. 29.—Vesicle and nephrostome of another specimen (x 50).

Fic. 30.—Same nephrostome from the other side.

Fic. 31.—Vesicle (uterus) of anterior nephridium, packed full of ripe ova.

Fic. 32.—Ova from uterus.

Fie. 33.—Immature ovum from ovary.

Fic, 34.—Ripe ovum.

Fic. 35.—Chart of the spectrum of the green pigment (‘‘ Thalassemin ”’)

of Thalassema Lankesteri (kindly drawn up for me by Professor Sher-
rington).

THE PLACENTATION OF PERAMELES. 385

The Placentation of Perameles.
(Contributions to the Embryology of the Marsupialia—I.)
By

Jas. P. Hill,
Demonstrator of Biology in the University of Sydney, N.S.W.

With Plates 29—33.

INTRODUCTION.

In a preliminary note (1) communicated to the Linnean
Society of New South Wales I recorded the occurrence
of an allantoic placenta in Perameles obesula, and
gave a short account of its structure based on an exami-
nation of a single stage, at that time the only one in my
possession. Since then, the acquisition of certain important
earlier and later stages has enabled me to study the placenta-
tion of Perameles in some detail, and the results of this
investigation are presented in the following pages. I am now
able not only to amplify the short statement of the preliminary
note, but to give a fairly connected account of what I believe
are the most important phenomena in the evolution of the
Perameles placenta. I am also now able to state that an
allantoic placenta in all respects similar to that of P. obesula
occurs in the closely related P. nasuta. The stages described
consecutively in the following pages have been derived from
these two species indifferently, just as pregnant individuals
chanced to come to hand; but so far as I have been able to
make out, there are no recognisable differences in the details

386 JAS..Ps Bilu;

of the development in the two forms. Even now, after nearly
two years’ collecting, the material at my disposal is by no
means large. It comprises altogether six stages, of which two
are post-partum. I take this opportunity of acknowledg-
ing my great indebtedness to the following gentlemen for
most generous aid in the by no means easy task of securing
the material on which this paper is based :—Messrs. J. B.
Cooper, A. G. Hamilton, A. M. Lea, Thomas Steel, and Dr.
A. E. Mills. To all these gentlemen I tender my sincere
thanks.

To my friend Professor J. T. Wilson I am under a very
deep debt of gratitude for invaluable help in the way of sug-
gestion, criticism, and advice during the whole course of my
work. I am also indebted to my friend Professor C. J.
Martin for much kind advice.

I have further to thank Mr. R. Grant, of the Physiological
Laboratory, for very great help in photography, and for assist-
ance in other ways.

Lastly, I desire to thank my honoured chief, Professor W.
A. Haswell, F.R.S., for his uniform kindness and considera-
tion, and for much kindly interest in my work.

Methods.—As fixing fluids picro-sulphuric and picro-nitric
acids were used. Most satisfactory staining results were got
by staining sections fixed on the slide by Mann’s albumen
method, first with a weak watery solution of Renaut’s hema-
toxylic glycerine for eighteen to twenty-four hours, followed
by an alcoholic solution of eosin. Even better results were
obtained by the substitution of hematéin for hematoxylin in
Renaut’s formula. By this method of double staining the
foetal and maternal vessels can be beautifully differentiated.

THE PLACENTATION OF PERAMELES. 387

GENERAL SUMMARY OF RESULTS.

Berore entering upon the detailed description of the various
stages, it may conduce to clearness ifa brief résumé of the main
facts of placental development be here presented.

I. CHanges IN THE UTERINE WALL.

The mucosa undergoes marked hypertrophy over its whole
extent ; the uterine glands increase both in transverse diameter
and in length; the interglandular connective tissue forms a
loose open network of anastomosing cells, and becomes per-
meated by abundant lymph; the vessels of the mucosa in-
crease greatly in size and in number. These changes in the
corium are accompanied by the transformation of the whole
of the uterine epithelium into a vascular syncytium. This is
ushered in by the disappearance of the cell outlines between
the cells and the active proliferation of the nuclei. The
uniform nucleated syncytial layer thus produced increases in
thickness by the growth of the protoplasm; the nuclei also
increase in number, and eventually become, in greater part,
grouped together in nests, situated in lobular projections of
the deeper surface of the syncytium. At the same time
maternal capillaries pass up between the syncytial lobules,
penetrate the syncytial protoplasm, and form a network on
and just beneath its surface.

The uterine wall is now prepared for the attachment of the
embryo.

II. Empryonic CHANGES.

(a) Fixation of the Embryo.—(1) The embryo becomes
attached to the maternal syncytium by means of the enlarged
ectoderm cells over the discoidal area of true chorion with
which the allantois fuses. The ectoderm, when the attach-
ment is complete, consists of a single layer of greatly enlarged
cells, roughly cubical or columnar in shape. Their irregular
outer ends accurately fit into the irregularities of the surface

388 JAS. P. HILL.

of the syncytium, and are firmly adherent thereto. In corre-
lation with this close adherence of the chorionic ectoderm, this
area of the uterine syncytium is markedly thicker than the
remainder, and forms the allantoic placental area. In the
allanto-chorionic mesenchyme, and in close relation to the
inner surface of the chorionic ectoderm, run the allantoic
capillaries. (2) Outside the discoidal allanto-chorionic area
a somewhat annular zone of the yolk-sac wall is also brought
into intimate relation with the maternal syncytium around the
above-mentioned allantoic placental area, by means of a close
approximation of its exceedingly thin ectodermal cells. This
annular zone of yolk-sac wall corresponds to the embryonic
vascular area, and at this stage the portion of the syncytium
in relation to it is more highly vascular than the allantoic
placental syncytium itself. This structural arrangement can
hardly be considered other than a yolk-sac placental forma-
tion, functional at a time when the allantoic placenta is yet
only being formed.

(B) Formationof the Functional Allantoic Placenta.
—This is brought to pass through the gradual degeneration
and resorption of the enlarged chorionic ectoderm cells over
the placental area proper. These cells thus take no further
part in placental formation. The allantoic capillaries can now
directly reach the vascular surface of the allantoic placental
syncytium, to which they become intimately attached, dipping
down into the depressions in its surface, and forming in places
a regular interlocking system. The foetal and maternal blood-
streams are now only separated by their thin endothelial walls,
and perhaps a thin layer of syncytial protoplasm.

III. Parturition.

At birth not only is there no loss of maternal tissue (i.e. no
decidua is formed), but the vesicular portion of the allantois
remains persistently attached to the placental syncytium, and
is gradually absorbed in situ along with the latter through
the agency of maternal leucocytes.

The foetus, whilst still connected with the placental area by

THE PLACENTATION OF PERAMELES. 389

the lengthened allantoic stalk, passes to the exterior, not by
way of the lateral vaginal canals, but by breaking through
along a median track leading backwards from a posterior
common portion of the two uteri.

DESCRIPTION OF STAGES.
Structure of the Non-pregnant Uterus.

The uterine wall is shown in transverse section in fig. 1.
Externally is the fairly thick serosa (s.) continued on from the
ligamentum latum. Internal to the serosa is the muscularis,
composed of circularly running non-striate fibres (c. m.). The
mucosa (m.) follows immediately on the muscularis, and is on
the whole sharply marked off from the latter; occasionally,
however, the terminal ends of the uterine glands may pene-
trate into the muscularis.

The mucosa varies considerably in thickness in different
uteri, averaging about ‘75 mm. Its free surface is thrown
into irregular longitudinal folds. The matrix of the mucosa
consists of fairly compact retiform connective tissue (c. ¢.), in
which are embedded the uterine glands and blood-vessels.

The uterine glands (gl.) are very numerous, straight or
greatly convoluted tubules, averaging ‘045 mm. in diameter.
They are lined by a lowcolumnar epithelium, outside of which
is a thin tunica propria derived from the surrounding con-
nective tissue. They open freely into the uterine lumen.

The blood-vessels enter the mucosa from the circular mus-
cularis. The majority of the superficial vessels of the mucosa
are of the nature of capillaries, with only an adventitious layer
of connective tissue surrounding the endothelium; in the
deeper portions of the mucosa, however, vessels with distinct
muscular walls also occur.

The lining epithelium of the uterus (ep.) consists of a layer
of low columnar cells with a thickness of about ‘(015 mm., and
continuous with the lining epithelium of the glands at the
gland openings.

390 TAS.) Ps HEL,

Stace A.—P. NASUTA.

Both uteri were somewhat enlarged, and presented a con-
gested appearance. In the right uterus an early blastocyst
was found.

The wall of the blastocyst was separated on one side from
the enclosing membrane (Selenka’s “ granulosa membran ”
[2], Caldwell’s shell membrane [8]) by a space, and had a
transverse diameter of °525 mm.; while, including the invest-
ing membrane, the whole blastocyst had a diameter of ‘675 mm.
This blastocyst has not yet been examined in sections, but it
probably nearly corresponds to a ten hours’ blastocyst of Di-
delphys, which, according to Selenka (2), has a transverse
diameter of about ‘5 mm.

Microscopical examination of the right uterine wall shows
that as a whole it has increased considerably in thickness as
compared with the non-pregnant uterus. This increase is
mainly due to the enlargement of the mucosa, which now
averages 1°5 mm. in thickness.

The uterine glands are closely packed together, causing the
interglandular connective tissue to appear greatly reduced.
They have increased both in length and in transverse diameter,
the latter now averaging ‘075 mm., and their epithelial lining
has undergone marked proliferation. It now consists of a
high cylindrical epithelium with numerous small deeply stain-
ing nuclei basally situated. The meshes of the connective-
tissue network are occupied by lymph coagulum, and numbers
of somewhat enlarged capillaries are also present, but as yet
in no great abundance.

The most important change in this uterus, however, con-
cerns the lining uterine epithelium. Through the disappear-
ance of the cell outlines between the cells, it has become
transformed into a continuous protoplasmic layer or syncy-
tium all over the surface of the mucosa (fig. 2, syn.), and at
the same time it has increased somewhat in thickness, now
measuring ‘025 mm.

Along with this fusion of the cell bodies, the nuclei have

THE PLACENTATION OF PERAMELES. 891

undergone active proliferation. They now form an irregular
band occupying the mid region of the syncytium, and are so
numerous as to frequently overlap even in very thin sections
(fig. 2). They vary considerably in shape, mostly ovalish or
elongated, and are evidently in a most active phase. Though
I have not been able to make out undoubted mitotic figures in
my preparations, there can be no doubt that marked pro-
liferation of the syncytial nuclei has taken place.

Minot’s description (4) of the early changes taking place in
the uterine epithelium of the rabbit prior to its complete
degeneration is equally applicable to Perameles. He says
“the thickening [of the uterine epithelium] is due to the en-
largement and fusion of the epithelial cells, and this enlarge-
ment of the cells is due to the proliferation of the nuclei, and
to the growth of the protoplasm which begins later, and con-
tinues longer (as later stages show) than the multiplication of
the nuclei.”” It may be pointed out, however, that the agree-
ment in the two cases goes no further than the earliest stages.
As we know from the researches of Minot (4), Duval (5), and
others in the rabbit, this nucleated protoplasmic layer formed
from the uterine epithelium soon degenerates and disappears ;
in Perameles, on the other hand, as will be abundantly evident
further on in this paper, the syncytial layer derived from the
uterine epithelium not only does not degenerate, but, increas-
ing in size and becoming vascularised by maternal vessels,
persists throughout the whole period of pregnancy, and takes
a most essential part in placental formation.

Stace B.—P. oBEsuULA.

The left uterus was somewhat larger than the right, measur-
ing 17 mm. in length by 11 mm.in breadth. It contained
two blastodermic vesicles, with the ‘‘ granulosa membran ” of
Selenka still in greater part persistent round them. The
embryo measures about 7 mm. in length, and possesses at least
fifteen mesodermal somites. It is characterised as follows :—
Anterior end strongly flexed and enclosed in the large pro-
amnion; medullary plate in anterior cerebral region still

392 TAGS Pe aE

unclosed, but just closed in trunk region ; distinct sinus rhom-
boidalis enclosing primitive streak; fore-limb buds; median
heart anlage ; blood circulating ; distinct anditory grooves.

Both uteri were examined microscopically, and were found
to have undergone exactly the same changes. It may here be
noted that such changes as have occurred are not limited to
any special region of the mucosa, but occur uniformly all
over it.

The general appearance of the uterine wall under a low
power is shown in fig. 3. Owing to the enlargement of the
uterus as a whole, the serosa and muscularis appear to be
somewhat thinner than in the preceding stage. The mucosa is
approximately of the same thickness as in that stage, but has
altered considerably in appearance. The uterine glands (fig.
3, gl.) are now for the most part widely separated from each
other, and the interglandular connective tissue appears greatly
attenuated. It consists of a very delicate retiform tissue, and
is permeated by abundance of lymph coagulum, while nume-
rous leucocytes are also distributed through it.

The glands appear the same as in the preceding stage. The
mucosa is now much more vascular than in Stage A. The
syncytial lining of the uterus has undergone further enlarge-
ment and differentiation. The layer has an average thickness
of (035 mm., i.e. it is somewhat thicker than in Stage A.
Further, its inner surface is now found divided up intoa series
of numerous close-set lobular projections of somewhat irre-
gular size (fig. 4, syn. /.). The greater number of the syncy-
tial nuclei are disposed in relation to these lobules, in many
cases filling them completely, in other cases forming an irre-
gular layer in the marginal protoplasm of the lobule. Scat-
tered nuclei also occur in the superficial portion of the syncy-
tium, but not abundantly. Like the syncytial protoplasm,
the ovalish or rounded nuclei stain deeply. They now present
the appearance of typical resting nuclei, a fact which suggests
that the further enlargement of the syncytium is not to any
great extent at least accompanied by active division of the

nuclei.

THE PLACENTATION OF PERAMELES. 3893

Another highly significant fact in connection with the
syncytium is that it is already vascularised. Both between
the lobules and enclosed in the protoplasm of the syncytium
itself small capillary vessels, with distinct nucleated endothelial
walls and containing maternal blood-corpuscles, can be readily
made out (fig. 4, syn. c.). These syncytial capillaries are
derived from the capillaries of the mucosa, which are seen to
pass up between the syncytial lobules, and from there to
ramify out in the syncytium itself. That the capillaries
actually penetrate into the syncytium by their own growth
seems beyond question, but no doubt the subsequent gradual
enlargement of the syncytium as a whole, and especially of its
lobules, also aids in bringing about the enclosure of the ca-
pillaries.

We can only regard the formation of the syncytial lobules
as the result of the enlargement and growth of the protoplasm,
and it seems probable that the direct invasion of the syncytium
by ingrowing capillaries may have been the inciting cause of
this mode of growth.

This transformation of the uterine epithelium into a vascular
syncytium is a highly distinctive and peculiar feature in the
developmental history of the placentation of Perameles. Such
a condition has hitherto never been met with in any other
mammalian form, and is especially interesting in view of the
wide-spread occurrence of degeneration of the uterine epithe-
lium prior to placental formation in so many diverse Eutherian
orders. The only form known to me which in the behaviour
of its uterine epithelium offers any points of analogy with the
above-described transformation of the epithelium in Perameles
is Sorex. Hubrecht has shown that, in this Insectivore, mo-
dification of the uterine epithelium over the placental area, by
way of proliferation, is the first and most important change
“that takes place in the maternal tissues preparatory to the
reception, fixation, and nutrition of the blastocyst” (7, p. 491).
But when one compares the details of the proliferation in the
two cases they are seen to be essentially different in character,
though offering interesting analogies. In Sorex, following

394 JAS. P. HILL.

Hubrecht’s account, we have to do with a proliferation of cells
from the under surface of the uterine epithelium. These pro-
liferated cells eventually form crypts, between which vessels
penetrate. The crypts, however, play only a temporary
role in the formation of the placenta, and take no ultimate
part in its development.

In Perameles, on the other hand, we have to do not with a
proliferation of cells, but of nuclei in a continuous syncytial
layer; and what is more important is the fact that here this
transformed epithelium persists to form the maternal portion
of the functional placenta.

Staces C anp D.

General Account of the Foetal Membranes.

Before proceeding to the detailed consideration of Stages
C and D, it is advisable here to give a general account of the
foetal membranes, so far as they can be made out from these
two stages.

In Perameles the foetal membranes have the same general
arrangement as in Phascoiarctus, my two stages exhibiting
characters corresponding to the stages described and figured
by Caldwell (12) and Semon (8).

Owing to the mode of growth and the development of an
exceedingly voluminous proamnion the embryo is found, at
the stage when the amnion is complete, sunk down into the
cavity of the yolk-sac, and partially surrounded by the upper
portion of the yolk-sac wall (text fig., y. spl.), which is thus
invaginated into the yolk-sac cavity. Semon distinguishes
this invaginated portion of the yolk-sac wall (or briefly
“ yolk-sac splanchnopleure”’) simply as “ inneres Blatt.”

The space in which the embryo, enclosed in its amnion, lies
is, of course, the extra-embryonic splanchnoceele (text fig.,
cw.), and is closed externally by a discoidal area of true chorion."
It is with this discoidal area that the allantois fuses, and over
it the allantoic placental connection is eventually established.

1 We use the term chorion here in the sense specified by Minot (‘Human

Embryology,’ p. 286), viz. the true chorion is that part of the extra-embryonic
somatopleure which remains after separation of the amnion.

THE PLACENTATION OF PERAMELES. 395

The periphery of this true chorionic area thus indicates the
limit of extension of the coelom, and beyond that limit chorion
and invaginated yolk splauchnopleure alike merge into the

QUAN. Coe.

bil.orph. y. spl.

Diagram showing the arrangement of the foetal membranes in Perameles.
amn. Amnion. adll.c. Allantoic cavity. all. mes. Allanto-chorionic me-
senchyme. al.s. Allantoic stalk. 42d. omph. Bilaminar omphalopleure.
ch. Marginal zone of true chorion around the allanto-chorionic area. ca.
Extra-embryonic splanchnoceele. c@.w. Inner or celomic wall of allan-
tois. proa.r. Persistent remnant of proamnion. s.¢. Sinus terminalis.
vasc. omph. Vascular omphalopleure. y.c. Cavity of yolk-sac. y. spl.
Invaginated yolk-sac splanchnopleure. The ectoderm is represented by
a thin line, the entoderm by a dotted line, and the mesoderm by a thick
lime.

unsplit wall of the blastodermic vesicle or primitive yolk
cavity (text fig., vasc. omph.). For this outer unsplit wall of
the yolk-sac no very suitable name exists in the literature of
embryology. Semon simply terms it “dusseres Blatt.” If
we restrict the use of the term chorion, as Minot has done,
then such terms as “ omphalo-chorion ” (Fleischmann), “ Dot-

396 JAS. P. HILL.

tersackchorion seu pseudo-chorion”’ (Selenka), are open to
objection. Hubrecht’s term “ omphaloidean diplotrophoblast ”
(9, p. 385) is also inadequate, since it is only properly appli-
cable to the somatopleural constituents of the as yet unsplit
yolk-sac wall. To avoid confusion it seems desirable to
employ a distinctive term descriptive of this unsplit yolk-sac
wall in its entire thickness and extent; and for this end I
propose to adopt, at the suggestion of my friend Professor
J. T. Wilson, the term ‘‘omphalopleure.”

The term omphalopleure, then, signifies the whole
of the wall of the blastodermic vesicle or primitive
yolk-sac, beyond the region of extension of the
splanchnocele. The employment of the term omphalo-
pleure will thus prevent the unnecessary use of the expression
‘‘blastodermic vesicle wall” in stages when the embryo is
completely folded off, and one no longer wishes to speak of a
blastodermic vesicle as such. According to the extension of
the unsplit mesoderm the omphalopleure may be trilaminar or
bilaminar in greater or lesser extent. Also a unilaminar con-
dition may be temporarily found in a position corresponding
to the lower pole of the blastodermic vesicle, prior to com-
plete ventral extension of the yolk-sac entoderm.

In Marsupials the trilaminar portion of the omphalopleure
is co-extensive with the vascular area, the sinus terminalis
marking not only the margin of the vascular area, but also the
peripheral limit of the unsplit mesoderm. We may thus refer
to the trilaminar omphalopleure in Marsupials as the “ vascular
omphalopleure” (text fig., vasc. omph.). Beyond the sinus
terminalis mesoderm is absent, and there the omphalopleure
consists solely of ectoderm and entoderm (text fig., d¢d.
omph.). We shall hereafter refer to this as the “ bilaminar
omphalopleure,”’ which seems more expressive than Semon’s
term ‘‘ Prokalymma” (8). We can thus distinguish in the
outer wall limiting the whole spheroidal embryonic formation
three areas of widely different structure: towards each pole a
discoidal area,—the one limiting the extra-embryonic cclom,
aud consisting of true chorion ; the other limiting the yolk-sac

THE PLACENTATION OF PERAMELES. 397

cavity, and consisting of bilaminar omphalopleure ; between
these and running round the mid region of the whole struc-
ture, a somewhat annular zone of vascular omphalopleure.

The allantois of Perameles consists of a long stalk (text fig.,
all. s.) and a terminal expanded and much-compressed vesicular
portion. The stalk, leaving the embryo immediately behind the
yolk-stalk, curves round its right side, and, extending through
the extra-embryonic celom, expands at its distal end to form
the flattened vesicular portion. In the stalk the allantoic
cavity is reduced to a narrow compressed canal opening dis-
tally into the continuous cavity of the vesicular portion (all. c.).
From the flattened form of this latter we may, for descriptive
purposes, term that portion of its wall next the ccelom the
inner or ccelomic wall (cw. w.), and that turned towards the
chorion the outer or placental wall.

In Stage C the mesoderm of the outer wall of the allantois
is found fused with the somatic mesoderm of the discoidal
chorionic area (all. mes.), the enlarged ectodermal cells of
which are firmly adherent to the vascular maternal syncytium.
For the chorion, after the allantois has fused with it, we shall
employ the term “allanto-chorion” (Fleischmann). The
allantoic vessels consist of a large vein and two smaller arte-
ries. They extend unbranched in the stalk, and, in fact, con-
stitute its greater bulk. At its distal end the arteries branch
out on the inner wall of the vesicular portion of the allantois,
while the vein is there formed by the union of two main factors
which accompany the main arterial trunks. The latter branch
in a dichotomous manner, each of the larger arterial branches
being accompanied by a corresponding venous channel. This
arrangement is, however, as Fleischmann (10) has pointed out
for the cat (cf. his fig. 7, Taf. iv), confined to the larger trunks ;
the finer branchings do not thus correspond. These vessels
ramifying on the inner wall of the allantois pass round the
margin of the allantoic vesicle into the allanto-chorionic
mesenchyme of the outer wall, and there form a rich capillary
network corresponding to the so richly developed network
described and figured by Semon (8) for Phascolarctus (cf. his

398 JAS: -P.-HILI.

fig. 38, Taf. iv). Of the vitelline vessels unfortunately I can
give only a very incomplete account. In opening up the uteri
in both Stages C and D the vascular area was partly destroyed,
so that I can only state the course of the main trunks. As in
fipyprymuus and Phascolarctus, according to Semon’s account
(8), the yolk-sac is supplied by a vitelline artery and a vitelline
vein.

The vitelline artery on leaving the yolk-stalk runs obliquely
backwards in the yolk splanchnopleure, and finally passes over
into the vascular omphalopleure, where it bifurcates into a right
and left trunk, which together constitute the circular sinus
terminalis (text fig., s.¢.). From the sinus smaller branches
pass off into the vascular area. Whether the two trunks
actually inosculate or are only connected by capillary anasto-
mosis I am unable to state.

The vitelline vein is formed close to the base of the yolk-
stalk by the union of two main factors which arise in the
vascular omphalopleure by the union of tributaries coming
from the capillary network of the vascular area. These two
main vitelline trunks coming from opposite sides of the
vascular area pass over from the vascular omphalopleure to
the yolk splanchnopleure, and there run posteriorly over the
left side of the head of the embryo. They gradually approxi-
mate as they approach the yolk-stalk, near the base of which
they unite to form the single vitelline vein.

The last point to which we need refer in this general
account concerns the existence of a persistent remnant of the
proamnion somewhat similar to the ‘‘ proamnion-rest” de-
scribed by Semon (8) for Phascolarctus. As in that form, the
persistent connection between the amnion and the here non-
vascular yolk splanchnopleure forms a small pear-shaped area
bounded laterally by the main factors of the vitelline vein,
and extending from their point of union up to about the level
of the eye on the left side of the embryo. A section through
this proamniotic area is represented in fig. 14. It will be
seen therefrom that both the ectoderm and entoderm over the
area are considerably modified.

THE PLACENTATION OF PERAMELES. 399

The ectoderm (fig. 14, ect.) is markedly thickened, and at
the margins of the area forms cushions several cells in thick-
ness. The superficial cells often project freely, and are club-
like in form, with the nucleus lying in the freely projecting
part of the cell. Apparently in Phascolarctus such a marked
thickening of the ectoderm does not exist.

The entoderm presents a somewhat varied appearance in
different sections. In the section drawn (fig. 14, ent.) it is a
quite irregular layer of some thickness. Many of its cells are
greatly enlarged and vesicular-looking, presenting quite a
degenerate appearance. Even in places where the entoderm
over the area does not differ greatly from the entoderm of the
yolk splanchnopleure, one often meets with similar isolated
enlarged and vesicular-looking cells. In Phascolarctus Semon
describes the entoderm over the area as “stark verdickt und
eigeuthitimlich gewulstet” (8, p.31). At the edges of the area
(fig. 14) the somatic mesoderm of the amnion (amn.) is con-
tinuous with the mesoderm of the splanchnopleure, so that the
continuity of the extra-embryonic ccelom is here definitely
interrupted. The mesoderm penetrates into the area for a
short distance peripherally; but, contrary to what Semon
describes for Phascolarctus, it does not form a continuous
layer extending right through and separating the ectoderm of
the amnion from the entoderm of the yolk-sac. In the centre
of the area these layers may be either in close apposition or
separated by a narrow cleft. This central portion of the area
thus consists of true proamnion.

Stace C.—P. oBESULA.

The left uterus was very much larger than the somewhat
enlarged right, and formed a large globular swelling containing
a single embryo (fig. 5) with a crown-rump measurement of
7mm. For details of the internal anatomy of this embryo
see table (appendix).

This is a most important stage, since it shows the mode of
attachment of the embryo. Microscopic examination of the

voL. 40, PART 3.—NEW SER. EE

400 JAS. P: HEIGL.

empty right uterus shows that it has undergone changes
parallel with those undergone in the pregnant left, a vascular
syncytium of some thickness having been formed all over the
surface of the mucosa.

In this and the following stages I propose to describe
separately (1) the changes in the uterus, and (2) the structural
features of the foetal membranes and their relations to the
uterine wall.

I. Left Uterus.—The discoidal allantoic placental area is
situated on the ant-mesometrial side of the uterus. The
uterine surface is thrown into a series of irregularly longi-
tudinal folds, this folding being especially marked in the
allantoic placental area. One may emphasise this by stating
that from edge to edge of the area in a straight line the
breadth in its mid region is only about 4 mm., while following
the folds the allantoic placental area has a breadth in sections
of 183—14 mm.

The muscularis is apparently somewhat thicker than in the
preceding stage.

So far as the corium is concerned no very sharp distinction
can be drawn between the portion of it lying beneath the
allantoic placental area and that outside the latter. Still for
descriptive purposes it is convenient to speak of these two
portions separately.

The corium beneath the allantoic placental area varies very
greatly in thickness over its extent owing to the greatly
folded nature of the surface. Measured from the bottom of
a depression between two folds the corium may have a thick-
ness of only ‘6 mm., while in the region of a fold the thickness
may reach as much as 2:7 mm. Outside the allantoic placental
area the mucosa is not so markedly folded, and is on the
whole thinner. The corium here varies in thickness from
‘5 tol‘7 mm. This difference in thickness is mainly due to
the expansion of the interglandular connective tissue below
the allantoic placental area, so that here the corium has a less
compact appearance, and the glands are on the whole more
widely separated from each other than in the corium outside

THE PLACENTATION OF PERAMELES. 401

this area. Abundance of lymph coagulum is found through-
out the meshes of the delicate imterglandular connective
tissue.

The uterine glands have not essentially altered since Stage B.
Some few of them, however, are now enormously enlarged,
and lined by a low cubical epithelium.

In the allantoic placental area the gland openings are
occluded by the chorionic ectoderm cells, which at this stage
form an almost continuous layer firmly attached to the surface
of this portion of the syncytium. In the region surrounding
the allantoic placental area the gland openings are similarly
occluded by the close contact of the vascular omphalopleure
with the syncytial surface.

The corium is now very richly supplied with blood; the
capillaries are numerous and greatly dilated.

Syncytium.—The syncytium in this stage is found to have
undergone differentiation into three fairly well-defined regions
corresponding to the three areas we have already distinguished
in connection with the foetal membranes, viz. the allanto-
chorion, the vascular omphalopleure, and the bilaminar ompha-
lopleure. The area of the syncytium to which the enlarged
ectoderm cells of the allantochorion are attached we shall
term the allantoic placental syncytium, as distinguished
from that portion of the syncytium in relation to the ompha-
lopleure.

(a) Allantoic Placental Syncytium.—This is at once
to be distinguished from the remainder of the syncytium, not
only by the fact that the chorionic ectoderm is at this stage
firmly adherent to it, but also by the facts of its greater thick-
ness, and the larger size and deeper staining qualities of its
nuclei (figs. 6 and 7, pl. syn.). It has now an average thick-
ness of 1 mm., i.e. it is just about three times as thick as the
syncytium in Stage B.

The outer surface of the layer is not smooth, but wavy and
irregular, while its inner surface is distinctly lobulated. The
large and deeply staining ovalish or rounded nuclei are for the
most part closely aggregated together in nest-like groups in

402 JAS. P. HILL.

the deeper lobular zone of the syncytium, while scattered
nuclei also occur in its superficial portion just as in Stage B
(figs. 7 and 8, pl. syn.).

The syncytium is now much more vascular than in Stage B,
numbers of enlarged capillaries with distinct endothelial walls
occurring throughout the protoplasm. As in Stage B, the
capillaries are found to enter the syncytium between the
lobules. They pass mainly into its superficial zone in order
there to ramify just beneath its surface, clothed at this stage
by the chorionic ectoderm.

(6) Syncytium beyond Allantoic Placental Area.—
The allantoic placental syncytium at its margin becomes
greatly reduced in thickness, and continues on as the syncytium
in contact with the vascular omphalopleure (fig. 6, ex. syn.).
Here it has a thickness of about ‘(05 mm. It has essentially
the same characters as the allantoic placental syncytium. The
nuclei are, however, smaller, and stain somewhat less deeply.
They are, as a rule, aggregated in nests in the lobules, though
here and there they tend to become more irregularly dis-
tributed throughout the protoplasm.

What, however, specially characterises this portion of the
syncytium in contact with the vascular area is the richness of
its blood-supply. At this stage it is distinctly more vascular
than the allantoic placental syncytium. The capillaries form
a rich network just beneath and upon its irregular free surface
(fig. 6, syn. c.). The significance of this fact will be pointed
out later, in connection with the description of the vascular
omphalopleure.

Beyond the sinus terminalis, where the syncytium is in
contact with the bilaminar omphalopleure, it is on the whole
just about half as thick as over the vascular area. Here,
though the characteristic arrangement of the nuclei in nests is
still to be seen, the nests are small and inconspicuous, and are
often separated by fairly wide intervals, in which the nuclei
are irregularly distributed in the protoplasm (fig. 6). The
nuclei are similar in size and in staining properties to those of
the syncytium in contact with the vascular area. The vas-

THE PLACENTATION OF PERAMELES. 403

cularity of this region is considerable, but much less so than in
the syncytial region in relation to the vascular omphalopleure.

Il. Faran Mempranes.—(a) Chorionic Ectoderm.—
The ectodermal cells of the discoidal area of allanto-chorion
are at this stage found firmly adherent to that portion of
the syncytium we have already described under the term
allantoic placental.

The chorionic ectoderm consists of a single layer of irregu-
larly columnar or cubical cells, of large size and with large
ovalish or rounded deeply staining nuclei (figs. 6, 7, and 8, ch.
ect.). The layer is on the whole sharply marked off from the
underlying syncytium by the fact that the protoplasm of the
ectoderm cells stains much less deeply than the syncytial
protoplasm. The outer ends of the ectoderm cells project quite
irregularly, and are found to be accurately adapted to the
irregular syncytial surface, dipping down into and filling up
every little depression in its surface. The connection, then,
between the chorionic ectoderm and the underlying syncytium
is of the closest and most intimate character. Such close and
accurate apposition could only have arisen through a mutual
process of growth and enlargement, affecting both the ectoderm
cells themselves and the underlying syncytium.

So far the above description of the chorionic ectoderm would
apply to the whole of that layer in a stage slightly earlier than
the one under consideration. In this latter, however, the
ectoderm is not uniform in character over its whole extent,
since over certain portions of it degenerative processes have
already set in. Such a portion is represented in fig. 9. It
will be there seen that the ectoderm no longer forms a con-
tinuous and uninterrupted layer of fairly regular cells, but is
irregular and definitely interrupted on the right side of the
figure, thus allowing the allantoic capillaries to come into direct
contact with the vascular syncytium. A comparison of the
chorionic ectoderm in fig. 8 (ch. ect.) with that shown in this
fig. 9 will render at once apparent the marked changes which
have taken place in the characters of the cells. Here they are

404 JAS. P. HILL.

of a very varying size and shape, and in places, through the
disappearance of the outlines between adjacent cells, large
multinucleate cells have been formed. The protoplasm of
these degenerating cells often stains just as deeply as that of
the syncytium, rendering it difficult to determine accurately
the limit between the two. In many of the ectoderm cells,
shown in fig. 9, the nuclei are also seen to be in various stages
of disintegration. Many of them stain only slightly; the
nuclear membrane is becoming indistinct, while the chromatin
is found broken up, and diffused in the form of small granules
throughout the delicate nuclear reticulum. Eventually the
position of the nucleus is only marked by a few straggling
irregularly thickened remnants, which finally become diffused
throughout the protoplasm, and lost to view.

It may be noted that degeneration does not take place in a
uniform manner over any given area, but quite irregularly in
patches, so that in a small portion of the allantoic placental
area (as in fig. 9) various stages in the degenerative process
are met with. In certain portions of the chorionic ectoderm
where it still forms a continuous layer of fairly regular cells,
and shows no signs of the degenerative process just described,
I have found that the inner ends of the cells are greatly va-
cuolated, a fact which suggests that a process of vacuolation
may also play a réle in the retrogression of the chorionic
ectoderm.

That the chorionic ectoderm is destined to disappear is
abundantly evident from this stage alone, and I am inclined
to believe that the allantoic capillaries, so closely related to its
inner surface, are by no means the least active agents in
effecting its removal. Of direct fusion of the degenerate ecto-
derm with the syncytium there can be no question. All the
facts negative such a view.

The rédle of the ectoderm is apparently merely that of
attaching the embryo to the previously prepared maternal
syncytium. Once the allantoic capillaries have spread out on
its inner surface, it degenerates and disappears in order to
allow of closer proximity between the foetal and maternal

THE PLACENTATION OF PERAMELES. 4.05

capillaries, and thus takes no part in the constitution of the
functional placenta.

(6) Allantois and Mesenchyme of Allanto-cho-
rion.—The allantoic stalk is ovalish in cross-section (fig. 10),
measuring in long diameter °3 mm., and in short '18 mm. In
it run the three allantoic vessels, two smaller arteries (all. a.),
and a larger vein (all, v.). The vessels are lined by a delicate
endothelium, round which the mesoderm is condensed to form
a thin sheath. Lying in the mesenchyme between the two
arteries is the small canal of the stalk (all. cl.) lined by a layer
of somewhat flattened entoderm. The stalk is covered ex-
ternally by a layer of mesothelium, internal to which and
forming the matrix of the stalk are branched mesenchyme
cells.

The stalk enters the body of the embryo behind and to the
right of the intestinal loop to which the yolk-stalk is attached,
and has the usual relations ; 1.e. the stalk, now consisting of the
entodermal canal and the two arteries, passes back in the
median line attached to the inner surface of the ventral abdo-
minal wall, and its canal finally opens into the cloaca. In the
abdomen and close to the body of the embryo, the entodermal
lining of the canal consists of a low cubical epithelium. At
its distal end the stalk expands into the flattened vesicular
portion of the allantois, the canal of the stalk being in direct
continuity with the cavity of the vesicular portion (fig. 11).
This latter is an exceedingly thin-walled sac, possessing a con-
tinuous cavity of very irregular form (fig. 7, all. c.). The
inner or ceelomic wall of the allantois (fig. 7, cw. w.) is fairly
smooth and unfolded, but its outer wall has grown out into
great hollow folds which enter the deep depressions of the
uterine surface, and thus is of far greater extent than the inner
wall,

The allantoic cavity is lined by a very thin layer of ento-
derm with small flattened or ovalish nuclei (fig. 7, all. ent.).
The mesoderm covering the inner wall (fig. 7, cw. w.) consists
only of a thin mesothelial layer, except along the vessels. The
mesoderm of the outer allantoic wall is now organically con-

406 JAS. (PB. BILE.

tinuous with the somatic mesoderm of the chorion, forming
the allanto-chorionic mesenchyme. This is, at this stage, an
exceedingly thin layer, which becomes somewhat thicker at the
margin of the allantoic placental area and around the larger
allantoic capillaries (figs. 6, 7, and 8, all. mes.). The mesen-
chyme consists of small branched cells, the delicate processes
of which anastomose with each other, and with the entodermal
lining of the allantoic cavity. I find that the allantois of a
late uterine embryo of Macropus dorsalis shows essentially
the same structural features as the vesicular portion of the
allantois of Perameles, only in the Macropus embryo the
vessels are less marked ; and there is, of course, no union with
the chorion. In both cases the allantoic wall is characterised
by its extreme tenuity.

In the inner or cclomic wall of the allantois, between the
mesothelium and the entoderm, run the main branches and
factors of the allantoic arteries and vein. Corresponding to
their characteristic mode of branching, one finds in sections
the larger trunks in pairs, a smaller and slightly thicker
walled arterial trunk accompanied by a larger venous channel.
In the inner wall the vessels have distinct thin sheaths of
condensed mesenchyme. In order to reach the outer surface
of the allantois these vessels, ramifying in the inner wall, turn
round the periphery of the flattened vesicle. Seeing the
allantoic cavity is a continuous and uninterrupted one, there
are no direct passages across the cavity by means of cellular
bridges, as Hubrecht describes for the allantois of Erinaceus
(9). The vessels of the inner wall gradually decrease in size
by repeated branchings as they approach the periphery of the
vesicle, their mesodermal investing sheath becomes reduced to
a layer a single cell thick, and they pass round into the
allanto-chorionic mesenchyme. There they again branch re-
peatedly, forming a network of small capillaries, with only
endothelial walls. The ultimate branches of this capillary
system come into very close contact with the inner surface of
the chorionic ectoderm, and even in places where cells of the
latter have disappeared, into contact with the vascular maternal

THE PLACENTATION OF PERAMELES. 407

syncytium (fig. 9). The true chorion at the margin of the
allantoic placental area yet remains to be mentioned (text fig.,
ch.). It will be seen from figs. 6 and 7, representing sec-
tions through the opposite margins of the allantoic placental
area, that the allantois does not at this stage completely
extend over the whole of the chorionic area, but leaves outside
its periphery a narrow marginal zone of true chorion, which
intervenes between the allanto-chorion and the vascular om-
phalopleure.

On the one side (fig. 6) the allantois does not even extend
completely to the outer limit of attachment of the chorionic
ectoderm, but leaves here a marginal attached portion of true
chorion. On the other side (fig. 7) the margin of the allantois
corresponds fairly accurately with the outer limit of attach-
ment of the chorionic ectoderm, so that here only a narrow
strip of free chorion remains. On both sides the ectoderm
cells of this marginal chorionic zone become gradually reduced
in size, and pass over into the thin ectoderm of the vascular
omphalopleure. Closely adherent to the inner surface of this
transitional chorionic ectoderm is the thin single layer of
somatic mesoderm of the chorion (figs. 6 and 7, som.), con-
tinuous on the one side with the allanto-chorionic mesen-
chyme (all. mes.), and on the other with the mesoderm of
the yolk splanchnopleure (y. spl.).

(ec) Yolk Splanchnopleure.—The line along which the
somatic mesoderm of the chorion is continuous with the meso-
derm of the yolk splanchnopleure marks the outer limit of the
splanchnocele (figs. 6 and 7, cw.), and thus also the com-
mencement of the vascular omphalopleure (vase. omph.). The
mesoderm of the yolk splanchnopleure is a thin layer, carrying
fairly numerous vessels, except in that portion of it included
between the two main factors of the vitelline vein, which is,
as Semon has pointed out (8), permanently non-vascular. Its
entoderm is similar to that of the vascular omphalopleure.

Traced round to its connection with the embryo, the yolk
splanchnopleure narrows to form the yolk-stalk, or vitelline
duct, which opens into the gut at the apex of the intestinal loop.

408 JAS. PL HILL.

(d) Vascular Omphalopleure and Yolk-sac Pla-
centa.—The vascular omphalopleure includes, as already
pointed out, the area of the yolk-sac wall between the peri-
phery of the true chorion and the sinus terminalis, and is
co-extensive with the vascular area. Its entoderm (figs. 6 and
7, vasc. omph.) consists of a fairly thick, somewhat flattened
layer of varying width, with oval or rounded nuclei staining
deeply like the protoplasm. The unsplit mesoderm is a thin
and delicate layer carrying the very numerous capillaries of
the vascular area.

The ectoderm is especially noteworthy on account of its
extreme tenuity (fig. 6, ect.). It consists of a delicate thin
layer of mostly flattened cells with small oval or fusiform
nuclei. Here and there at intervals, cells of a somewhat
triangular shape are found, with their outer pointed ends
projecting beyond the level of the general surface of the
ectoderm.

In Didelphys, according to the descriptions both of Osborn
and Selenka, the ectoderm of the vascular omphalopleure is a
much thicker layer than in Perameles. Osborn describes it
as consisting of “ elongated cells with amcoebiform processes
which are closely applied to the lining epithelium of the
uterus” (11, p. 378; cf. also his fig. 4, Pl. xvii) ; and Selenka,
describing the layer in a 7} days’ embryo, says (2, p. 1387),
‘Bei weitem der groésste Theil dieser Ektodermzellen dehnt
sich aus und nimmt blasige Form an, unter gleichzeitiger
Vergrésserung der Kerne (Taf. xxvii, fig. 5, d.). In diesem
Felde von blasigen Zellen bemerkt man gegen Ende der Incu-
bation vereinzelte Fleckchen kleiner nahezu kubischer Ekto-
dermzellen (¢ und a) und hier und da sah ich an Schnitten
sogar Andeutung von Zottenbildung (4) mit axialen Meso-
dermzellen.” Again, in a 5—6 days’ old Hypsiprymnus he
describes the ectoderm of the vascular omphalopleure in the
following words (2,p.184) :—“ Doch zeigten sich die Ektoderm-
zellen des Chorion auffallend vergrdéssert und gegen die
Uterushéhle Zapfen-oder kuppelartig vorgewolbt (Tafel xxxii,
Figur 2).”’ Caldwell, again, referring presumably to Phasco-

THE PLACENTATION OF PERAMELES. 4.09

larctus, says (12, p. 657), “The whole vascular area is covered
by flat cells of the subzonal membrane.” In a uterine embryo
of Macropus ruficollis at a stage slightly later than the one
of Perameles under consideration, I find that the ectoderm of
the vascular omphalopleure consists of a fairly thick flattened
layer, which more nearly resembles the entoderm of the vas-
cular omphalopleure of Perameles than the ectoderm of the
same.

So far, then, as the forms mentioned are concerned, it appears
that the extreme thinness of the ectoderm of the vascular
omphalopleure in Perameles is an exceptional and noteworthy
feature. A glance at fig. 6, which includes the whole of the
extent of the vascular omphalopleure on one side, will show
that, although the latter is separated in sections from the
syncytium, there is an accurate and fairly close correspond-
ence between the two, elevations in the one corresponding to
depressions in the other, and vice versa. The vascular
omphalopleure, in other words, appears moulded to fit the
irregular surface of the maternal syncytium, and without
doubt during life the two surfaces were accurately apposed
the one to the other, the above-mentioned projecting ectoderm
cells even serving as an actual attachment. It has already
been pointed out that the portion of the syncytium in relation
to the vascular omphalopleure possesses a very rich network
of maternal capillaries, on and just beneath its surface, and
that at this stage it is even relatively much more vascular than
the allantoic placental syncytium. The foetal capillaries of
the vascular area are thus only separated from the maternal
by a thin and delicate ectodermal layer, plus a thin layer of
syncytial protoplasm, the latter often absent indeed where the
maternal capillaries actually reach the surface.

These facts point to the conclusion that, prior to the period
of functional activity of the allantoic placenta, the placental
function is subserved by the close contact of the vascular
omphalopleure with the vascular maternal syncytium, an ar-
rangement which we are therefore justified in designating as
an actual yolk-sac placenta.

410 JAS. P. HILL.

That the yolk-sac placenta is of high functional importance
in the nutrition of the embryo at this stage is borne out not
only by the fact that the vitelline vein is nearly three times as
large as the allantoic, but also by the further fact that the
larger proportion of the purified and food-laden blood coming
from the vascular area passes directly to the heart. So far as
one can judge from structure alone, this veritable yolk-sac
placenta of Perameles appears to be more efficiently adapted
for respiratory and nutritive functions than the arrangement
found in other described Marsupials; e. g. in Macropods the
ectoderm of the vascular omphalopleure is a comparatively
thick layer, the uterine epithelium persists, though in a some-
what modified form, and the maternal capillaries existing
below it are not very numerous, and nowhere directly project
on the free surface.

(ec) Bilaminar Omphalopleure.—Beyond the sinus
terminalis (fig. 6, s.¢.) mesoderm is absent, the wall consisting
here solely of ectoderm and entoderm, for the most part in
close contact with each other.

The entoderm is, on the whole, slightly thicker than that of
the vascular omphalopleure. It consists of a layer of cells of
varying size and shape, so that its inner contour is somewhat
irregular (fig. 6, del. omph., and fig. 12, ent.). In places the
entoderm cells present a vacuolated appearance.

The ectoderm differs markedly from that of the vascular
omphalopleure. It is a very much thicker layer, the cells
are large, rich in protoplasm, and vary greatly in form and
size. Their outer ends project more or less freely in a quite
irregular manner, so that the free surface of the layer presents
a roughened irregular appearance (fig. 6, di. omph., and
fig. 12, ect.).

Like Semon, I see no evidence in sections of the existence
at this stage of “ pseudopodia-like” processes of these ecto-
derm cells, such as Caldwell (12) describes as serving to attach
the blastodermic vesicle to the uterus.

The significance of the persistence of this bilaminar portion
of the omphalopleure for a longer or shorter period in dif-

THE PLACENTATION OF PERAMELES. 411

ferent mammals has been already discussed by Semon (8).
He has come to the conclusion that it has an important phy-
siological meaning; in his own words (p. 55), “hier erfolgt
eben der Durchtritt der von der Mutter gelieferten Nahrungs-
stoffe in das Innere des Dottersacks, von wo aus weiterhin die
Aufnahme und Uebergabe an das Blut durch die Entoderm-
zellen der gefisshaltigen Zone ausgefiihrt wird.’ We may,
however, point out that in Perameles only a small proportion
of the uterine secretion need take this indirect way of reach-
ing the vessels of the vascular area. By far the greater pro-
portion of the secretion, no doubt, passes directly through the
thin ectoderm of the vascular omphalopleure into the yolk-
sac vessels.

Stace D.—P. oBESULA.

Upon the examination of this stage was based the prelimi-
nary account already published (1). It is specially important
since it shows the allantoic placenta well developed. Both
uteri were greatly enlarged; the left contained two embryos,
measuring respectively 8 and 8°25 mm., while the right con-
tained a single embryo measuring 8°75 mm. from crown to
rump. For the structural characteristics of this latter embryo
see table (appendix).

These three embryos all present substantially the same
features of placental connection. Fig. 13 represents the larger
of the three embryos attached to the placental area of the
uterine wall, and still partially enclosed in its membranes.
The dissection from which this figure was made was prepared
by opening up the uterus by a ventral longitudinal incision,
which also involved the closely adherent omphalopleural wall.
In the figure, then, we see the inner surface of the yolk-sac
wall lying on the inner surface of the uterus, which has been
spread out flat.

In the middle of the figure lies the embryo seen through
the amnion and the yolk-sac splanchnopleure (y. sp/.). At the
back of the embryo and partly concealed by the body of the
latter is the allantoic placental area (pl.a.), also seen through

412 JAS. P. HILL.

the yolk splanchnopleure. The allantoic stalk with its three
vessels (all. a. and all. v.) is seen to emerge from under the
embryo’s right side, and at its distal end the vessels spread out
on what is the inner wall of the vesicular portion of the
allantois. The distribution and mode of branching of the
allantoic vessels is clearly shown. As already described, the
vein (all. v.) is formed by the union of two factors which
accompany the corresponding arteries (all. a.).

The placental area is discoid in shape and, from the presence
of a prominent rim on the side visible, appears somewhat
depressed below the general surface of the uterus. In sections
it is found to commence a little behind the anterior margin of
the flexed head end of the embryo, and to terminate some
distance behind the level of its curved posterior end. Its
length, in the direction of the long axis of the embryo, is thus
about 9 mm., while in its middle region, following the folds,
it has a breadth of about 12 mm. transversely to the long axis
of the embryo.

The flattened allantoic stalk has an approximate length
of 8mm.

The course of the vitelline vessels is also shown in the figure.
Theaving the yolk-stalk, the artery (vit. a.) passes obliquely
backwards, traversing the yolk splanchnopleure (y. spl.) to
reach the vascular omphalopleure (vasc. omph.), and there it
almost immediately divides into two right and left branches,
which form the sinus terminalis (s. ¢.), a portion of which is
visible. The area outside the sinus (below) in the figure is
part of the bilaminar omphalopleure (d2/. omph.). From the
yolk-stalk the two factors (vit. v.) of the vitelline vein pass
anteriorly in the yolk splanchnopleure, and gradually diverging
from each other, they pass over into the vascular omphalo-
pleure, where they are formed by the union of lesser factors
coming from the capillary system of the vascular area.

I. Urerus.—The serosa, muscularis, and corium are es-
sentially the same as in Stage C.

(a) Allantoic Placental Syncytium.—This presents
the same general features as the corresponding area in Stage C.

THE PLACENTATION OF PERAMELES. 413

but is somewhat thicker (averaging ‘12 mm.) than in that
stage. It is also much more highly vascular (cf., e. g., fig. 15
with fig. 7). The capillaries entering between the syncytial
lobules ramify in great abundance at and just beneath the
surface, where they form a rich network. These superficial
capillaries are Jarge, and vary considerably in diameter, averaging
about ‘Ol mm., some attaining a width of ‘(028 mm. The
surface of the syncytium is here by no means smooth, but
owing to the bulging of the capillaries on the surface it pre-
sents an irregularly ridged structure (figs. 16 and 17, pl. syn.).

(6) Syncytium beyond Allantoic Placental Area.—
That portion of it in contact with the vascular omphalopleure
forms a fairly uniform layer, averaging ‘O9 mm. in diameter.
It is thus nearly double as thick as the corresponding portion
of the syncytium in Stage C, and is also rather more vascular.
These facts are significant in view of what has been said above
on the high functional importance of the yolk-sac placenta
prior to the complete formation of the allantoic.

The superficial capillaries of this portion of the syncytium
are markedly developed; many of them lie at the surface
(fig. 24, syn. c.), which here presents a wavy contour, but is
not ridged, as is the allantoic placental syncytium.

That area of the syncytium in relation with the bilaminar
omphalopleure presents the same features as were described
for the corresponding region in Stage C. It is, however,
somewhat thicker than in that stage.

Il. Fara, Mempranes.—(a) Chorionic Ectoderm.—
The chorionic ectoderm has now almost completely disappeared.
It is still, indeed, recognisable as a continuous layer of cells at
the margin of the allantoic placental area (fig. 15, ch. ect.),
but over the remainder of the latter it is represented only by
more or less isolated degenerating cells (figs. 19—21, ch. ect’.).

The persistent marginal zone of ectoderm (fig. 15, ch. ect.)
is narrow but of very variable width. Its most peripheral
cells, adjoining the ectoderm of the vascular omphalopleure,
are the least altered, but the remainder are irregular, and

414, JAS: Po IATEB:

vary both in size and shape. A comparison of fig. 15 with
fig. 7 shows at a glance the marked change in the character
of the marginal chorionic ectoderm in this stage; for the
process of degeneration and absorption which had set in, in
the central part of the placental area in Stage C, has now
extended nearly to the margin.

In connection with the disappearance of the chorionic
ectoderm at this stage it may be noted that the straggling
isolated cells of it which yet persist have become greatly
hypertrophied. In some cases they are multinucleated (figs.
20 and 21, ch. ect’.), or the single nucleus is also hypertrophied
and vesicular (figs. 15 and 19, ch. ect’.). Fig. 20 is worthy of
remark as showing how such isolated and hypertrophied cells
may be gradually undermined by the ingrowth below them of
an allantoic capillary; the area of contact of the larger cell
shown in the figure with the syncytium is in this way greatly
reduced, while the smaller cell is entirely separated from the
syncytium, and awaits its resorption in isolation in the allanto-
chorionic mesenchyme.

(6) Allantois and Allantoic Placenta.—In its general
features the allantois is essentially similar to that of Stage C.
The allantoic stalk with its vessels now, however, is just about
half as thick again as that of the preceding stage (cf. fig. 23
with fig. 10), thus showing that the allantoic circulation has
increased considerably in volume. In correlation with this
increase in the blood-supply, the capillary network (fig. 22,
all. cap.) in the allanto-chorionic mesenchyme is now much
more richly developed than in Stage C.

Now that the chorionic ectoderm has almost wholly dis-
appeared, the capillaries are able to attach themselves closely
to the syncytial surface. We have already laid emphasis on
the characters of the allantoic placental portion of the maternal
syncytium, and have pointed out that its surface is both highly
irregular and very vascular, and in certain patches exhibits an
irregular system of interlacing vascular ridges separated by
depressions. With this irregularly ridged and highly vascular
surface the allantoic capillaries are in most intimate contact ;

THE PLACENTATION OF PERAMBELES. 415

so close, indeed, is the attachment that the walls of the capil-
laries appear as if united with the syncytial protoplasm. The
capillaries dip down into and accurately fill up the depressions
between the vascular ridges, so that there is here and there
formed an actually interlocking system of vascular projections
of the syncytial and allantoic surfaces respectively (figs. 16
and 17). As already indicated, this interlocking does not
occur uniformly all over the placental area, but varies in its
degree of perfection in different places (cf., e. g., figs. 15, 18,
and 19 with fig. 17). The interlocking here referred to repre-
sents the highest state of placental differentiation realised in
Perameles.

It is thus evident that in this placental differentiation foetal
and maternal elements take an approximately equal share. In
the functional organ so produced, it will be noted that the
foetal and maternal blood-streams are separated from each
other only by the thickness of two endothelial walls, with at
most the addition of a thin layer of syncytial protoplasm.
We may here point out that in Perameles, contrary to what
obtains in most other mammals,! the uterine glands of the
placental area do not degenerate, but persist throughout the
whole period of pregnancy. In this stage some of the gland
openings in the placental area are still occluded by persistent
portions of the chorionic ectoderm ina more or less degenerate
condition ; others of the openings, however, are obstructed by
allantoic capillaries extending right over them (fig. 18, gl.).
It may be that nutritive substances derived from the uterine
secretion of such glands are directly absorbed by the allantoic
capillaries occluding their openings.

(c) Yolk Splanchnopleure.—This is essentially as in
Stage C. We may mention, however, that the yolk-sac cavity
no longer stands in open communication with the gut, as was
the case in that stage.

(d) Vascular Omphalopleure and Yolk-sac Placenta.

1 Strahl (15) and Vernhout (14) describe a similar persistence of the uterine

glands in the mole. According to Vernhout they are invaded by the “ plas-
modiotrophoblast ” shortly before parturition.

voL. 40, PART 3.—NEW SER. FF

416 JAS. P. HILL.

—The vascular omphalopleure is on the whole similar to that
of Stage C. Here and there, however, the flattened cells of
the entoderm give place to larger somewhat cubicai cells with
rounded free ends. In the protoplasm of some of these larger
cells there occur vacuoles. Over the sinus terminalis and the
larger vessels of the vascular area the entoderm cells are also
markedly enlarged, and much more so than in Stage C. The
entoderm cells in these positions are now somewhat club-
shaped in form, with their enlarged ends projecting freely and
containing the ovalish or rounded nuclei (fig. 24, ent. over
s.t.). Selenka has already described a similar condition of the
entoderm cells of this region in Didelphys. He says (2,
p. 138), “ Die Entodermzellen des Chorion verandern gleich-
falls vielfach ihre Gestalt wahrend der letzten zwei Tage des
Foetallebens. Sie werden cylindrisch oder birnformig, zumal
in der Nihe der grdsseren Blutgefasse. Streckenweise be-
halten sie aber ihre frihere abgeplattete Form bei oder
nehmen mehr oder wenig an Volumen zu.”

The unsplit mesoderm of the vascular omphalopleure is
exactly as in Stage C.

The ectoderm is, as in that stage, an exceedingly delicate
layer of greatly attenuated cells (fig. 24, ect.). In my prepa-
rations of this stage, not only is the vascular omphalopleure
very evidently stamped with the contour of the highly vascular
syncytial surface, but in places the two are in most intimate
attachment, thus affording support for the belief already ex-
pressed that we have here an actual yolk-sac placental con-
nection.

In this stage, then, we regard the yolk-sac placenta as being
in functional activity along with the allantoic, though now it
has diminished considerably in importance, as the examination
of the foetal circulation shows. The vitelline vein in this stage
is both absolutely and relatively smaller than in Stage C, and
now most of its blood has to pass through the capillary system
of the liver before reaching the inferior vena cava. These
facts, taken in conjunction with the already mentioned greater
size of the allantoic trunks in this stage as compared with the

THE PLACENTATION OF PERAMELBES. 417

preceding, show conclusively that with the advent of the
allantoic placenta the yolk-sac circulation is giving place to
the allantoic. The latter, indeed, is now the predominant one ;
and we may add that just as, in the preceding stage, most of
the blood coming from the yolk-sac placenta passed directly to
the heart, so now most of the blood coming from the allantoic
placenta passes by way of the left allantoic vein and the ductus
venosus Arantii directly into the inferior vena cava.

The question whether the yolk-sac placenta remains func-
tional, though in a diminished degree, throughout the whole
period of intra-uterine life of the embryo ; or whether, as seems
likely from comparison with other placental mammalian forms,
it soon after this stage gives entire place to the later appear-
ing allantoic placenta, can only be definitely decided when
further material is available. As tending to support the latter
alternative, it may be pointed out here that in the next (post-
partum) stage, while the allantois was still found adherent to
the syncytium of the placental area, no portion of the omphalo-
pleure was to be found in the uterine cavity. Further, the
syncytium outside the allantoic placental area no longer
showed a richly vascular surface, but was rapidly retrogressing,
and indeed was already partly covered by the regenerating
uterine epithelium. The syncytium of the allantoic placental
area, on the other hand, though in process of absorption, had
not altered to such a marked degree.

These facts render it probable that the omphalopleure breaks
up and disappears some time before the end of intra-uterine
life (cf. also next section).

Unfortunately I am unable to give any details as to the
relative dimensions of the vascular area in this and the pre-
ceding stage. It may, however, be mentioned that the vessels
of the vascular area in this stage are apparently not nearly so
richly developed as in a Macropod embryo of about the same
developmental stage.

(e) Bilaminar Omphalopleure.—This presents features
tending to suggest that it is even now in process of degenera-
tion. The ectoderm has on the whole become greatly flattened

418 JAS. P. HILL.

and attenuated. This is especially noticeable close to the sinus
terminalis (fig. 24, bt. omph.). Further out one meets with
scattered projecting cells of large size and of irregular form,
the protoplasm and nuclei of which stain deeply.

The entoderm has also become considerably thinner in places.
Where it has not undergone attenuation the cell protoplasm is
often found to be greatly vacuolated, with irregular deeply
staining nuclei. Here and there, also, unaltered entoderm
cells are met with either singly or in groups.

Stace E.—P. nasuta (post-partum).

The material available for this most important stage con-
sisted of the genital organs (less the cloaca) of a female P.
nasuta, together with two newly born young from the pouch.

The new-born young (fig. 36) had a crown-rump measure-
ment of 14 mm., and a head length of 6 mm. For the details
of their external characters and internal anatomy see table
(appendix).

Both uteri were considerably enlarged: the left, the larger
of the two, measured 17 mm. in length by 9 mm. in breadth;
the right had a length of 16°5 mm. and a breadth of 8°25 mm.

When the uteri were opened up it was found that parturi-
tion had been recently accomplished, and that in each uterus
the flattened vesicular allantois with its stalk attached was still
adherent over the placental area (fig. 25, pl.a.). This latter
formed a fairly sharply circumscribed ovalish area, bounded by
an almost continuous ridge, and differed from the rest of the
irregularly ridged uterine surface by its closer texture. It
was situated on the dorso-mesial inner surface of the uterus,
i.e. ant-mesometrially. The area measured 9 mm. in length
by about 5 mm. in breadth (i.e. without following the folds).
The allantoic vessels ramifying in the inner wall of the allan-
tois could not, in surface view, be very definitely made out.
Apart from the adherent allantois no other portions of the
foetal membranes were encountered in the uterus.

The left uterus alone was submitted to microscopic exami-
nation.

THE PLACENTATION OF PERAMELES. 419

1. Urervus.—The serosa and muscularis are of about the
same thickness as in the non-gravid uterus. The muscularis
is penetrated by numerous large vessels,

As im previous stages, the whole mucosa is folded, the folds
being especially marked in the placental area.

Corium beneath Allantoic Placental Area.—This
portion of the corium now differs markedly in appearance and
character from the remainder, being as a whole much denser
and more compact-looking; and its component parts—inter-
glandular connective tissue, uterine glands, and blood-vessels—
have all undergone important modifications.

In the preceding stages we have seen that the connective
tissue of the whole corium consisted of a very delicate retiform
tissue. Now, however, in this region the connective-tissue
cells have not only increased in number, but also very greatly
in size. From the large, usually rounded, deeply staining cell
bodies less deeply staining processes pass off, which anastomose
with similar processes of adjacent cells to form a much coarser
and closer network than that seen in preceding stages, or
even in the corium outside this region in the present stage
(fie. 26,.¢. t’.).

There can be no doubt that we have here to do with a
process of proliferation of the connective-tissue cells beneath
the placental area, accompanied by their subsequent hyper-
trophy.

This proliferation and overgrowth of the connective tissue in
Perameles offers interesting points of comparison with the
formation of the decidual cells in the pregnant human uterus,
which arise, as Minot has rendered certain, by the direct
proliferation and enlargement of the anastomosing connective-
tissue cells of the mucosa (4, p. 419).

Numbers of polynuclear leucocytes occur throughout the
proliferated connective tissue, especially in its superficial
portions immediately beneath the syncytium.

Many of the uterine glands of the placental area, and
especially their peripheral portions, are now in process of
marked degenerative change. Various stages in the degenera-

420 JAS: P. HLGL.

tive process are met with, from the first signs of alteration of
the gland epithelium to the almost complete obliteration of
the gland lumen, by an accumulation of cellular débris, derived
from the completely disintegrated epithelial lining of the
gland (fig. 26, d.g/.).

The deeper portions of the glands adjacent to the muscularis
are on the whole less altered. The gland epithelium here
still forms a distinct deeply staining layer, but it is somewhat
thinner, and not quite so regular as the gland epithelium of
former stages. The gland secretion not being able to pass
away, owing to the degeneration and occlusion of the peri-
pheral ends of the glands, is here often found in the lumen as
a deeply staining coagulum. The mouths of the glands still
open freely on the placental area. The glands would thus
appear to retain their function throughout the whole period of
intra-uterine development of the foetus.

The blood-vessels also show considerable alteration. The
vessel walls appear greatly thickened, and the endothelial cells
have increased in size (fig. 26, m.v.) and proliferated, thus
markedly diminishing the lumen, and in some cases occluding
it completely (fig. 26, m.v’.). In other more advanced cases
the whole vessel is found to have undergone fibrous degenera-
tion, and appears quite solid in section (fig. 26, m.v”.). In
the coagulated blood in certain of the vessels, polynuclear
leucocytes are found to occur.

Corium beyond Allantoic Placental Area.—Here the
corium has not undergone such marked change as that of the
placental area proper.

Though in parts the connective-tissue cells have under-
gone a considerable amount of proliferation, yet the tissue as
a whole presents the same open and loose appearance as in the
preceding stages.

The glands, too, present a more normal appearance, and
even where degeneration of their epithelium occurs it has
not advanced to such an extent as beneath the placental
area.

The blood-vessels of this region show the same essential

THE PLACENTATION OF PERAMELES. 421

alterations as in the placental region. Here, however, the
vessels are not nearly so numerous as in the latter.

Allantoic Placental Syncytium.—Following the folds
of the mucosa in section, this portion of the syncytium has a
breadth of about 9 mm.in the mid region of the placental
area ; it varies in thickness from ‘15 to ‘28 mm., and is thus
slightly thicker on the average than in Stage D.

The syncytium here is now in active process of absorption
and retrogressive metamorphosis.

The syncytial protoplasm is coarsely granular, and numerous
irregular spaces and clefts occur throughout its extent (figs.
27 and 28, sp.). Its nuclei are much less numerous, and no
longer form such distinct nest-like groups in the syncytial
lobules (fig. 29). They vary very greatly in size and in shape,
being often quite irregular, and occur in all stages of retro-
gressive change. Many of them are vacuolated and greatly
hypertrophied (fig. 27).

Throughout the protoplasm great numbers of leucocytes
occur. The majority of these are of the polynuclear variety,
possessing oval or rounded slightly staining cell bodies, in which
numbers of small deeply staining nuclei lie (fig. 27, p. leuc.).

It seems certain that these polynuclear leucocytes are the
active agents in the absorption and removal of the degenerat-
ing syncytium, and accordingly it may be observed that the
leucocytes greatly predominate in the more degenerate areas.

The placental syncytium is still vascular (figs. 28 and 29),
though the superficial capillaries are now not nearly so pro-
minent as in Stage D.

The persistence even in this advanced stage, of compara-
tively unaltered portions of the syncytium, shows that no
essential alteration takes place in the latter from Stage D up
to the time of birth, and that the features peculiar to the
present stage are solely characteristic of post-partum meta-
morphosis.

Syncytium beyond Allantoic Placental Area.—This
portion of the syncytium has undergone more marked altera-
tion in character than the placental portion. On its deeper

422 JAS: P. “BIL

surface in many places it is no longer sharply marked off from
the underlying connective tissue, but there is a gradual transi-
tion from the one to the other (fig. 30, ex. syn’.).

The small and often deeply-staining nuclei are irregularly
distributed throughout the altered protoplasm, which stains
deeply, and is often vacuolated. The maternal capillaries are
now very greatly reduced both in size and in number. They
occur quite irregularly in the protoplasm, and no longer form
a superficial network (fig. 30, syn. c.).

Polynuclear leucocytes occur in numbers in the connective
tissue immediately below the syncytium, but only sparingly
in the syncytium itself.

In regions where disintegration of the syncytium is well
marked, i.e. where its remains are practically incorporated
with the underlying connective tissue, regeneration of the
uterine epithelium has already commenced. Here, as in the
human uterus, the uterine epithelium is regenerated by the
growth of the gland epithelium at the openings of the uterine
glands. In fig. 30 the opening of such a gland is shown,
with its lining epithelium spreading out over the degenerate
syncytial surface (ez. cyn'.) to form the thin and somewhat
irregular uterine epithelium (7. ep.), better seen in fig. 31.
The regenerative process does not take place uniformly all
over the surface of the syncytium in this region, but in
patches, and is apparently conditioned by the stage of degene-
ration of the syncytium.

As we have already pointed out in connection with Stage D,
it seems probable that the much more degenerate condition of
this region of the syncytium, as compared with the placental,
is to be correlated with the presumed early retrogression of the
entire omphalopleure.

II. Faran Memsranes.—Allantois.—The allantois is the
only portion of the foetal membranes which is found intact
and persistent in the uterus. In its general relations it is
essentially as in the preceding stage. Its outer wall closely
follows the folds of the mucosa, and is about three times as
extensive as the inner, In parts the outer wall is still closely

THE PLACENTATION OF PERAMELES. 423

adherent to the irregular surface of the syncytium, while in
others it has become separated from the latter (figs. 28
and 29).

The allantoic cavity (fig. 28, all.c.) is distinct and con-
tinuous, but its entodermal lining is no longer distinguishable.
It contains here and there an irregular cellular detritus. The
walls of the allantois have altered considerably in character.
The inner (ccelomic) wall (fig. 28, cw. w.) is somewhat thicker
than the outer (all. mes’.), and is now composed of a dense
mesodermal layer, carrying embedded in it in pairs the branches
and factors of the allantoic arteries and vein.

The allanto-chorionic mesenchyme of the outer (placental)
wall (all. mes’.) has also become quite compact in appearance.
On its outer surface the allantoic capillaries project (figs. 28
and 29, all.cap.). They now contain enucleated fcetal blood-
cells, and can still in places be seen to fit in and adhere to
corresponding depressions in the syncytial surface (fig. 29).

It will be apparent from figs. 28 and 29 that no essential
change has taken place in the constitution of the placenta in
the period intervening between Stage D and the time of birth.

The question whether or not the allantois is resorbed in
situ is at once settled positively by the occurrence in the
outer allantoic wall, and to a lesser degree in the inner as
well, of numbers of polynuclear leucocytes similar to those
already described as existing so abundantly in the placental
syncytium (figs. 32—34, p. leuc.).

There is not the least doubt but that these leucocytes
migrate from the syncytium into the allantoic walls. In
sections the leucocytes are found not only at the surface of the
syncytium, but actually in the spaces which exist here and
there between the syncytium and the outer allantoic wall, and
they are even to be seen just in process of entering the latter.

Absorption of the allantois has, however, not yet actively
begun, still it is breaking up in portions of its extent (cf.
fig. 28), and the cellular detritus in the allantoic cavity can
only have arisen through disintegration of its walls.

And if, as we have seen, the foetal portion of the placenta is

424, JAS. P. HILL.

not expelled at birth, but absorbed in situ, it is obvious from
the nature of the case there cannot be any shedding of maternal
tissue, i.e. no decidua is formed.

The case of Perameles is thus in most striking agreement
with the condition in the mole, where, according to the ob-
servations of Hubrecht! (9 and 18), “ no afterbirth is shed,
although the animal has a discoid placenta;” and he has
further pointed out that “ not only is the mole not deciduate,
but that even embryonic tissue is left behind against the
uterine surface, and is gradually resorbed in situ” (18,
p: 117):

It is thus obvious, as Professor Hubrecht has pointed out to
me (in litt.) that the term non-deciduate as long ago used
by Huxley is altogether inadequate and misleading as applied
to the post-partum conditions obtaining in Talpa and Pera-
meles. In these two forms there is not only no complete
separation of maternal and embryonic structures at birth
(Adeciduata), but no maternal tissue is lost (Deciduata) ;
on the contrary, foetal tissue is actually resorbed by the
mother. For such a condition Professor Hubrecht suggests
the term Contra-deciduata.

The discovery of the contra-deciduate character of the
placenta of Perameles thus affords welcome testimony to the
rightness of Hubrecht’s opinion, based on a consideration of
Talpa alone, that this contra-deciduate condition is “not a
secondary modification which has arisen among mammals that
were already frankly deciduate, but [is], on the contrary, a
more primitive developmental phase ” (18, p. 118), a view with
which I am in complete agreement.

Possible Vestiges of other Foetal Membranes.—In
connection with the margin of the allantois there are, in some
sections, appearances which I can only interpret as greatly
altered remnants of the walls of the yolk-sac. These remnants
vary very greatly in their detailed relations and in their extent,

1 Later confirmed by Vernhout (14).

THE PLACENTATION OF PERAMELES. 425

and are often entirely absent in the sections. It is unnecessary
to enter into details.

I would simply refer to figs. 32 and 33, and point out that
from the knob-like projection (a) attached to the margin of
the allantois (all. m.) there arise two cell layers, one from each
side. The one (c) may possibly be a vestige of the yolk-sac
splanchnopleure, while the other (4) may similarly be a vestige
of the vascular omphalopleure.

Whether this be so or not the fact remains, that besides
these structures no other traces of the yolk-sac were found in
the uterus, and this fact, taken in conjunction with the already
described greater degeneration of the syncytium outside the
placental area as compared with that of the latter, renders it
almost certain that the omphalopleure breaks up and dis-
appears some time previous to parturition.

Finally I may direct attention to the cells marked ch. ect”.
in figs. 82 and 34, as they may possibly represent persistent
marginal chorionic ectoderm cells.

Parturition.

In fig. 25 the genital organs of this stage are shown
partially dissected. The left uterus (J. ut.) has been opened
by a ventral longitudinal incision, so as to expose the placental
area (pl. a.). The vaginal ceca (v. ce@c.) and the bladder (6/.)
are pushed over to the left side, and the allantoic stalk (a//. s.)
arising from near the centre of the inner wall of the adherent
allantois has been traced posteriorly. It was found to pass
backwards through a posterior common portion of the two uteri
(common uterine canal) into what sections show to be a
median cleft-like passage in the connective tissue lying between
the two lateral vaginal canals. Through this median passage,
or median pseudo-vaginal passage, as we may term it, it is
obvious that delivery must have taken place.

At the time of making this dissection I was unaware of the
existence of any median passage in Perameles. Owen (16,
p. 683), in his short description of the female genital organs
of P. obesula, makes no mention of such; and, indeed, in

426 JAS. P. HILL.

my own dissections of non-gravid genital organs I had dis-
covered no such median passage. I was therefore considerably
surprised to find the allantoic stalk extending straight back
into the connective tissue bet ween the lateral vaginal canals,
and not into one of the latter, which I had believed must
serve for the passage of the young at birth in spite of the
narrowness of their communications with the uterine cavities.
The novel features revealed in the dissection were, however,
further elucidated by series of transverse sections across the
urino-genital strand! (fig. 25, u.s.), which demonstrated the
existence of a slit-like passage enclosing the allantoic stalks,
one from each uterus.

On investigation the stalks could be traced down the median
pseudo-vaginal passage from the centre of the inner wall of
the allantois for a distance of about 3 cm. They did not
extend quite to the extreme posterior end of the urino-genital
strand shown in fig. 25, but this is no doubt to be accounted
for by tearing of the stalks in the process of removal of the
genital organs. In part of their course they were found to be
looped upon themselves.

In Stage D the allantoic stalk of the larger embryo mea-
sured only about 8 mm. in length, so that shortly prior to or
during parturition a very considerable lengthening of the
stalks must take place. The stalk no doubt becomes severed
from the embryo only at the moment of birth, leaving merely
an insignificant portion (in length about ‘5 mm.) attached at
the navel of the latter (fig. 36). Similarly, in Erinaceus,
Hubrecht (9, p. 347) has shown that “by far the longer
portion ”’ of the strand formed by the lengthened-out allantoic
vessels at parturition remains attached to the afterbirth, which,
though eventually shed, is found in the uterus shortly after
delivery.

A section of the mid region of the urino-genital strand of

1 This name is applied to the elongated mass of connective tissue in which
are embedded the lateral vaginal canals and urethra. It is united to its sur-
roundings by more areolar connective tissue, and is of very considerable
length.

THE PLACENTATION OF PERAMELES. 427

this stage is shown in fig. 35. It will be there seen that the
median pseudo-vaginal passage (med. p.) is simply a cleft-like
space in-the central connective tissue of the strand, lying
dorsally to the urethra (ur.) and between the lateral vaginal
canals (vag. l.). Its walls are entirely formed by the connec-
tive-tissue core of the strand, and they exhibit no histological
differentiation into coats, muscular or other. ‘The passage is
of somewhat varying outline, with a greatest long diameter of
about 1:2 mm., and a short diameter of ‘6 mm. In this cleft
together with the allantoic stalks there occur masses of coagu-
lated blood (c. 6/.), especially abundant along the dorsal
portion of the passage, where indeed the clot in certain sec-
tions forms a definite ovalish mass almost as large as the
allantoic stalks, and partially separated from the rest of the
passage by an imperfect fibrous septum. This clot, however,
is continuous anteriorly and posteriorly with that present in
the main subdivision of the channel, and also with the extrava-
sated blood so abundantly present in the surrounding connec-
tive tissue (fig. 35, c. 6l.). The allantoic stalks (fig. 35, all.
s.) are somewhat oval in outline, and measure ‘3 mm. by ‘2
mm. in diameter. They are now in process of histological
degeneration. In the centre of each the cells appear clear and
vesicular, and the nuclei are for the most part quite dege-
nerate; marginally they stain very deeply. The allantoic
vessels are either empty or are partly occupied by degenerat-
ing, mainly enucleated, foetal blood-cells, together with a
granular deeply staining detritus. Their endothelial lining
has disappeared, and their mesodermal wall is enucleated and
fibrous-looking. In some sections the allantoic canal can be
indistinctly made out, but no longer with an entodermal
lining.

Direct observations of the parturition phenomena in Mar-
supials are by no means numerous. I know of only three
accounts :—(1) Owen (16, p. 721) quotes from a paper by
Rennger to the effect that in Didelphys azare the young
‘in gestation make the circuit of the lateral canals in which
they are found to be deprived of their foetal envelopes ;”’ (2)

428 JAS. P.- HILL.

Osborn (11, p. 377) records finding in Didelphys virgi-
niana “the foetal membranes . . . crowded into the uterine
orifices of the vaginz, which indicates that they had been
detached from the embryo in the uterus itself;” (3) Stirling
(17) furnishes a valuable account of the parturition in Ma-
cropus robustus [Osphranter erubescens]. He has
shown that in this form the young one passes out through the
median vaginal canal; and that while the ventral portion of
the yolk-sac remains in the uterus, interdigitating with the
folds of the mucosa, its dorsal portion, remaining attached to
the foetus, becomes, as the latter passes down the median
vaginal canal, drawn out into a long stalk carrying the three
vitelline vessels. The two forms, wallaroo and bandicoot,
thus agree in giving birth to the young through a median
channel; but the median canal of the one with its definite
walls is by no means homologous with the median cleft-like
passage of the other; for while the former is morphologically
continuous with the lateral vaginal canals, and is a true epi-
thelially lined tube, the latter has no connection what-
ever with the lateral canals, at no time possesses an
epithelial lining, and in fact is non-existent prior to
the first parturition.

It may further be pointed out that in the behaviour of their
foetal membranes at parturition the two forms exhibit an
interesting parallel and contrast. In the wallaroo, while the
extra-embryonic allantois has disappeared at birth, the yolk-sac
remains persistent in the uterus, and is drawn out into a long
cord, which remains connected with the embryo in its passage
outwards. In the bandicoot, on the other hand, it is the
allantois which similarly remains attached to the embryo by
its stalk during its course down the median passage, and
which persists in the uterus, while the yolk-sac has entirely
disappeared. This parallel behaviour of non-homologous
structures, by means of which nutriment is conveyed to the
foetus, tends to suggest that the passage of the young outwards
is a quite gradual one.

The discovery of this unique mode of parturition in Pera-

THE PLACENTATION OF PERAMELES. 429

meles led to a re-investigation, by means of serial sections of
the structure of the female genital organs, especially with
reference to the question of the existence or otherwise of such
a median pseudo-vaginal passage in virginal and non-gravid
genital organs. The results of this investigation will be set
forth in detail elsewhere. Suffice it here to state that in the
virginal genital organs the two uteri do not open into each
other posteriorly, and there is no trace of a median vaginal
passage or of any epithelial or other track, which might indicate
the site of a future passage of any kind whatever.

In the non-gravid organs of animals with large pouch-young,
on the other hand, the median pseudo-vaginal cleft is found to
exist, but it neither stands in open communication with the
common uterine canal nor does it open into the cloaca. As in
the post-partum stage, the passage is wholly destitute of
any epithelial lining or any other specialised wall.

As to the mode of formation of this median passage in the
first instance I am unable to come to any definite conclusion.
It has just been stated that in the virgin the uteri do not com-
municate with each other posteriorly, and no median passage
exists. The latter is, then, evidently formed either just before
or at the first act of parturition. That the embryo should in its
passage out literally bore its way through the connective tissue
seems to me improbable, but at least it would seem as if the
hindrance to the exit of the foetus offered by the narrow opening
of the uterus into the lateral vaginal canals was actually
greater than the resistent power of the tissue between the
posterior ends of the uteri, and that rupture of the latter must
occur. That some such rupture does occur is evidenced not
only by the appearance of the false passage, but also by the
pretty extensive extravasations of blood found both in and
surrounding the track followed by the fetus during its egress,
i.e. the median pseudo-vaginal passage.

It is evident that the detailed character of the phenomena
of parturition, and above all the nature of the causes producing
the extraordinary condition above described, can only be
definitely ascertained by the examination of the genital organs

430 JAS. Pi HELE,

of a female immediately prior to the commencement of the
first parturition. But whatever may be the precise mode of
formation of the passage, this most remarkable method of
getting rid of the young would seem to be without parallel in
the whole mammalian class.

New-born Young.—Here! I need only point out that, so
far as my observations (cf. appendix) go, the new-born Pera-
meles does not appear to. differ to any very great extent, in its
degree of development, from the new-born young of undoubted
non-placental Marsupials, e.g. Didelphys.

Stace F.—P. oBESULA (POST-PARTUM).

The material for this stage consisted of the genital organs
(less the cloaca) of a female with two pouch-young measuring
from crown to rump 22 mm. The left uterus, the larger of
the two, measured 15 mm. in length by 6 mm. in breath.

Microscopical examination shows that the uterus has now
almost completely regained the resting condition.

The mucosa, however, is just about half as thick as that of
the resting uterus. The epithelium of the uterine glands has
been laid down anew, and now consists of a low cubical epithe-
lium with fairly large ovalish nuclei. The gland lumen is
nearly always occupied by a finely granular coagulum, which
may contain cellular constituents. The interglandular con-
nective tissue is in parts fairly open in appearance, consisting
of a network of anastomosing cells; in other parts it is quite
dense and compact owing to the presence of great numbers of
young connective-tissue cells.

The syncytium and allantois have completely disappeared,
and the uterine epithelium now forms a continuous layer all
over the surface of the mucosa. It consists of a low layer of
cubical cells with rounded closely packed nuclei in a single
row.

Over the greater portion of its extent the uterine surface is

' T hope later to return to this question, and also to consider therewith the

question of the “critical period” in Perameles (J. Beard, ‘On Certain
Problems of Vertebrate Embryology,’ Jena, 1896).

THE PLACENTATION OF PERAMELES. 431

comparatively smooth and flat; but in a certain section on the
ant-mesometrial side of the uterus it is markedly folded and
very irregular in contour, owing to the presence of irregular
projections of the uterine epithelium. This folded area no
doubt represents the former allantoic placental area. The
projecting portions of the uterine epithelium just mentioned
are apparently eventually shed off into the uterine lumen, for
the lumen at this stage contains a detritus consisting of
maternal red blood-corpuscles, together with cellular elements
evidently derived from these irregular projections. The pre-
sence of these projections is readily intelligible when one
remembers how irregular was the surface of the allantoic
placental syncytium.

Genital Organs.—In this stage the lumen of the common
uterine canal is still continuous with the median pseudo-
vaginal passage. This latter, about its mid region, measures
‘7 mm. in long and ‘12 mm. in short diameter. It now appears
lined by a very delicate layer of connective-tissue endothelium,
outside which the tissue is very compact and vascular, but the
extravasated blood present in it in the preceding stage has now
disappeared. In its middle portion the passage contains an
irregular detritus consisting of red blood-corpuscles and
cellular elements.

Anteriorly, just behind the point of opening of the common
uterine canal into the passage, portions of allantoic stalks are
found still persistent in a degenerate condition, but with the
positions of the allantoic vascular trunks still recognisable.
The stalks are three in number,!—a larger one measuring in
diameter ‘4 mm. by 3: mm., and two smaller ones with a
diameter of ‘2 mm. each. In the region where the remains of
the allantoic stalks are found, the lumen of the passage is
almost completely obliterated, since the stalks are not only
closely surrounded externally by a loose layer derived from
the surrounding connective tissue, but are separated from

1 The genital organs reached me with only two young. It may be that the
larger and more degenerate stalk here described has persisted from a previous
parturition.

voL. 40, PART 3.—NEW SER. aa

432 JAS. P. HILL.

each other by delicate partitions derived from the latter. The
stalks now present quite a reticular appearance; the larger one
stains less deeply than the other two, and has undergone
marked fibrous degeneration. The nuclei are few in number,
and stain deeply and homogeneously. The lumina of the
allantoic vessels are occupied more or less completely by loose
branching cells.

In two other sets of genital organs, one froma P. obesula
with pouch-young measuring 4 cm., I have found similar per-
sistent remains of allantoic stalks in the upper portion of the
pseudo-vaginal passage in various stages of degeneration and
absorption. It is not necessary here to describe the appear-
ances in detail. Suffice it to say that the enucleated stalks,
closely invested by a connective-tissue sheath, undergo marked
fibroid degeneration, and eventually become invaded and
broken up by the ingrowth of the surrounding connective
tissue.

I may point out here that the existence of these remains
of the allantoic stalks, blocking up the pseudo-vaginal passage,
shows conclusively that the vesicular portion of the allantois
must be absorbed in utero, a view already maintained on
account of the presence of maternal leucocytes in it.

Concluding Remarks.

Before concluding this paper we may briefly inquire what
conclusions may legitimately be drawn from the fact of the
occurrence of an allantoic placenta among the Metatheria.

Has the allantoic placenta of Perameles been independently
evolved within the limits of the Marsupial order or is it
directly and genetically related to that of Eutheria through
the common ancestry of the Meta- and Eu-theria from an
earlier diphyodont protoplacental stock? In a previous paper
(18) in this Journal, on the tooth development of Perameles,
by Professor J. T. Wilson and myself, we incidentally touched
upon this question, and expressed our preference for the latter
of these two views; and I may here at once say that a much
fuller knowledge of the details of the placentation process in

THE PLAOENTATION OF PERAMELES. 433

Perameles has in no whit served to weaken our previously ex-
pressed opinion.

In view of the present very incomplete state of our know-
ledge regarding the condition of the foetal membranes in other
Australian polyprotodont Marsupials, especially in Dasyurus
and Myrmecobius, and even of the precise uterine changes in
Phascolarctus and other Diprotodonts whose foetal membranes
have been examined, it is impossible to decide finally between
these two views, which alone seem to us worthy of considera-
tion. In the concluding section of the paper just referred to
we presented in brief form a case in favour of the second
alternative. And in this summing up we dealt with the bear-
ings upon the case of facts relating to the dentition, the
placentation, and the mammary function. Here it is pro-
posed rather to treat shortly of those facts and considerations
which, in the opinion of the writer, tend to negative the first
alternative.

From the preceding account of the placentation phenomena
in Perameles I think we may justly conclude that the pro-
cesses of utero-gestation in that form are fundamentally the
same as those occurring in the more generalised Eutherians.
Such differences in detail as exist are, in my opinion, to be
regarded either as evidences of primitiveness of type on the
part of the Perameles placenta, or as physiological adaptations
such as Hubrecht has pointed out we may expect to find in
different types of placentation, in view of “‘the great youth of
the placenta as compared with the other chief components of
the organisation of a mammal” (9, p, 388).

I wish here more especially to lay emphasis on my convic-
tion that it is just as impossible to draw a hard and fast line
between the placentation phenomena as they occur in Pera-
meles and in the lower Eutherians, as it is to arbitrarily
mark off from each other the various types of placental forma-
tion occurring among the Eutheria themselves.

Now it seems on a priori grounds exceedingly improbable
that an allantoic placenta should have been twice indepen-
dently acquired, and in such a fundamentally similar manner

434, JAS. P. HILL.

within the limits of the mammalian class. Such would, in
our opinion, be a most remarkable instance of parallelism.

It is true that in the existence of a vascular syncytium
formed from the uterine epithelium, the placenta of Perameles
exhibits a modification of structural arrangement of a kind
occurring in no other mammal. But it cannot be held that
the existence of even such a unique modification gives support
to the view of the genetic independence of the placenta of
Perameles, any more than the existence of manifold and even
more marked structural differences in the types of placenta-
tion occurring among the Eutheria themselves witnesses to
their essential morphological diversity. In each of these cases
alike we find the real, or at least by far the most probable,
explanation in a differentiation of truly homologous parts due
primarily to physiological adaptation.

It is no doubt a tempting and easy solution of the problem
to regard the allantoic placenta of Perameles as a direct and
natural advance upon such a condition of foetal membranes as
occurs in Phascolarctus (alone, so far as is yet known, among
marsupials), where, as Caldwell (12) and Semon (8) have
shown, the vesicular allantois reaches and fuses with the dis-
coidal area of true chorion, develops a rich respiratory surface,
but forms no union with the uterine wall. And certainly,
if the placenta of Perameles be an independent acquisition
within the marsupial order, Phascolarctus would seem to
present the more primitive type of arrangement of fetal
membranes. As Semon (8) has pointed out, “ wir haben uns
nur vorzustellen, dass beim Phascolarctus—Typus im Bereich
der Athemfliche der Allantois eine innige Vereinigung der
Keimblasenwand mit den miitterlichen Geweben eintrat um
auf den Urtypus der Hihiillenanordnung der Placentalier zu
kommen.” At the same time he makes the further important
qualification, “‘ Deshalb, weil sich bei Phascolarctus in der
Anordnung seiner Eihillen primitivere Zustande erhalten
haben als bei den meisten anderen Marsupialieren, halte ich
ihn natiirlich nicht tberhaupt fiir ein besonders primitives
Beutelthier, oder gar fiir den Stammvater der tbrigen.” But,

THE PLACENTATION OF PERAMELES. 435

indeed, the evidence on all hands goes to show that the Dipro-
todontia represent a comparatively recent offshoot from a
primitive polyprotodont stem. And we are entirely unable to
accept the derivation of the Eutheria from a Perameles type
through a Phascolarctus type, as suggested by Semon (21,
p. 310). For it must ever be borne in mind that on the
strength of the evidence derived from a study of dentition
the whole marsupial order constitutes a well-marked natural
group, exhibiting like characteristics of degeneracy from the
typical and original mammalian condition. And, in this
group, Phascolarctus is distinguished not by less, but by an
even greater degree of retrogressive dental modification than
Perameles. It therefore seems unlikely that the former should
have retained unmodified more primitive embryonal nutritive
arrangements than the latter. So far, indeed, as the decision
of this problem can be shown to depend upon the question of
the primitiveness of the general structural organisation ex-
hibited by Phascolarctus and Perameles respectively, it can
hardly be denied that the evidence at our disposal is strongly
in favour of the latter to be regarded as representing a more
archaic marsupial type.

It is, of course, possible that in the remote past the imme-
diate promammalian (?) ancestors of the protoplacental stock
may have exhibited a condition of the foetal membranes some-
what resembling that of Phascolarctus; but we are unable to
avoid the conviction that as in dentition, so in its embryonic
appendages, Phascolarctus has shared largely in the general
marsupial decadence. And the fact that in all non-placental
marsupials, with the single exception of Phascolarctus, so far
as is known, the allantois never reaches the chorion, but
remains buried in the extra-embryonic splanchnocele as a
rudimentary structure, with no respiratory function, we con-
sider as indirect evidence in favour of our view. For, as
Selenka and Semon have pointed out, this condition is cer-
tainly to be regarded as a secondary one; and if this be so,
then the admitted existence of such a process of late secondary
reduction renders our view of still earlier coenogenetic sim-

436 JAS. P. HILL.

plification, to say the least, not wholly improbable. Thus, in
our view, it is unnecessary to trace the placental ancestry of
Eutheria back into the marsupial group. The occurrence
there of a true allantoic placenta, and its absence in the
majority of members of the order, do no doubt, at first sight,
suggest that in this group we must find the first beginnings of
the organ. But we believe that the explanation is to be found
in the fact that marsupials are, after all, a markedly specialised
group, and that in it conditions have obtained producing
placental disappearance, just as conditions (probably identical
in character) have determined the degeneration of other early
nutritional arrangements, i.e. the milk-teeth. We therefore
fall back upon the view that the Metatheria and Eutheria are
the divergent branches of a common ancestral stock, which
was not only diphyodont but also placental.

We may next inquire whether the facts and conclusions
detailed in the present paper have any bearing upon the
question of the condition of the foetal membranes in these
primitive Placentalia.

We believe that the facts of placentation in Perameles
most strikingly confirm and support the opinion of Balfour
(19), that in the primitive types of Placentalia both the allan-
toic and yolk-sac vessels may have been concerned in main-
taining a placental circulation. We have insisted on the
fundamental similarity of the placentation of Perameles with
that of the more generalised Eutheria; and if we select Eri-
naceus as representing a fairly generalised Eutherian type, we
find that here, according to Hubrecht’s account (9), just as in
Perameles, an extensive yolk-sac (omphalopleural) placental
counection is developed at an early stage, only to be replaced
later by the formation of the definitive allantoic placenta,
through the union of a large vesicular and vascular allantois
with the non-vascular chorion. Now if we leave aside the
trophoblastic differentiation in the one and the formation of a
maternal syncytium in the other, the type of placentation
occurring in these two generalised Metatherian and Eutherian
forms is an essentially similar one. This fact can in our

THE PLACENTATION OF PERAMELUES. 437

opinion only be regarded as conclusive in favour of the view
that from such a condition of foetal membranes as is common
to these two types, Perameles and Erinaceus, phylogenetic
speculation on the placenta must start. We would, therefore,
attach very special phylogenetic importance to the non-separa-
tion in marsupials of the vascular omphalopleure into yolk-sac
splanchnopleure and somatopleural chorion. ‘This non-sepa-
ration, ensuring, as Semon (8) has pointed out, the retention
of the vessels of the vascular area in a superficial position,
eminently suited for the performance of nutritive and respira-
tory functions, we can only regard as a physiological adapta-
tion, and as probably the first to have been adopted on the
initiation of uterine development in the common ancestors of
the Metatheria and Eutheria. This condition is not “ pro-
bably a purely marsupial modification,’ as Minot would have
us believe (6, p. 129), for it is undoubtedly also manifested in
the occurrence in certain lowly Eutherians of a temporary
yolk-sac placenta preceding the formation of the definitive
allantoic one. As Hubrecht has shown in Erinaceus (9), it is
only after the allantoic placenta has taken the place of the
omphalopleural that the mesoderm of the vascular omphalo-
pleure splits into splanchnic and somatic layers, and this
delaying of the splitting process Hubrecht (20) attributes to
the vital importance of the yolk-sac placenta.

It is not necessary to dilate on the significance of the dis-
coid form of the allantoic placenta of Perameles. “On the
Balfourian hypothesis,” as Professor G. B. Howes has pointed
out, the view that the discoidal type of allantoic placenta is
the most primitive ‘‘is by far the most natural one warranted
by the facts.” ?

University oF Sypney, N.S.W.;
April 16th, 1897.

1 © Nature,’ vol. xl, p. 420.

438

JAS. P. HILL.

List or PAPERS REFERRED TO.

. Hitt, J. P.—*‘ Preliminary Note on the Occurrence of a Placental Con-

nection in Perameles obesula, and on the Foetal Membranes of
certain Macropods,” ‘Proc. Linn. Soc.,’ New South Wales, vol. x,
(2nd ser.), part 4, 1895.

. SeLENKA, E.—“ Studien iiber Entwickelungsgeschichte der Thiere,’ IV

(1 and 2), Das Opossum (Didelphys virginiana) ; V (1), Beutel-
fuchs und Kanguruhratte (Phalangista et Hypsiprymnus). Wiesbaden,
1886-91.

. CaLpwELL, W. H.—* The Embryology of Monotremata and Marsupialia,”

Part I, ‘ Phil. Trans.,’ vol. clxxviii B., 1887.

. Minot, C. S.—‘* Uterus and Embryo: (1) Rabbit, (2) Man,” ‘ Journ.

of Morphology,’ vol. ii, 1889.

. Duvat, M.—‘‘ Le Placenta des Rongeurs,” ‘Journ. de |’Anat. et de la

Physiologie,’ tome xxv, 1889.

. Miyor, C. 8.—* A Theory of the Structure of the Placenta,” ‘ Anat.

Anz.,’ Bd. vi, 1891.

. Huprecut, A. A. W.—“ Studies in Mammalian Embryology. III. The

Placentation of the Shrew (Sorex vulgaris, L.),” ‘Quart. Journ.
Mier. Sci.,’ vol. xxxv, 1894.

. Semon, R.—‘‘ Die Embryonalhiillen der Monotremen und Marsupialer,”

‘Zoologische Forschungreisen in Australien und dem Malayischen
Archipel.,’ Bd. i.

. Husrecut, A. A. W.—“Studies in Mammalian Embryology. I. The

Placentation of Hrinaceus europeus, with Remarks on the
Phylogeny of the Placenta,” ‘ Quart. Journ. Micr. Sci.,’ vol xxx, 1890.

. Freiscumann, A.—‘‘ Embryologische Untersuchungen.” Heft 1. ‘ Un-

tersuchungen tber einheimische Raubtiere,’ Wiesbaden, 1889.

. Osporn, H. F.—‘‘ The Foetal Membranes of the Marsupials,” ‘ Journ. of

Morphology,’ vol. i, 1888.

. Catpwett, W. H.—“ On the Arrangement of the Embryonic Membranes

in Marsupial Animals,” ‘Quart. Journ. Micr. Sci.,’ vol. xxiv, 1884.

. Husrecut, A. A. W.—“Spolia Nemoris,” ‘ Quart. Journ. Mier. Sci.,’

vol. xxxvi, 1895.

. VernuHout, J. H.—‘ Bijdrage tot de Kennis der Placentatie van den mol

(Talpa europea, L.),’ Diss., Amersfoort, 1894.

. SrraHL, H.—“ Placenta und Hihaute,” ‘ Ergebn. der Anat. u. Entwickel-

ungsgesch. von Merkel u. Bonnet,’ Bd. i, 1891.

THE PLACENTATION OF PERAMELES. 439

16. Owrn, R.—‘ On the Anatomy of Vertebrates,’ vol. ii, 1868.
17. Srrruinc, E. C.—“On some Points in the Anatomy of the Female
Organs of Generation of the Kangaroo, especially in Relation to the

Acts of Impregnation and Parturition,” ‘Proc. Zool. Soc.,’ London,
1889.

18. Witson, J. T., and Hiu1, J. P.— Observations upon the Development
and Succession of the Teeth in Perameles ; together with a Contribu-
tion to the Discussion of the Homologies of the Teeth in Marsupial
Animals,” ‘ Quart. Journ. Mier. Sci.,’ vol. xxxix, 1697.

19. Batrour, F. M.—‘‘On the Evolution of the Placenta, and on the Possi-
bility of employing the Characters of the Placenta in the Classification
of the Mammalia,” ‘ Proc. Zool. Soc.,? London, 1881.

20. Husrecut, A. A. W.— Die Keimblase von Tarsius,” ‘ Festschrift fir
Carl Gegenbaur,’ Leipzig, 1896.

21. Sermon, R.—‘‘ Enstehung und Bedeutung der embryonalen Hiillen und
Anhangsorgane der Wirbelthiere,” ‘C. R. 3™* Congrés Int. de Zool.,’
Leyden, 1896.

EXPLANATION OF PLATES 29—33,

Illustrating Mr. Jas. P. Hill’s paper on “ The Placentation of
Perameles.”
(“Contributions to the Embryology of the Marsupialia,” I.)

All sections drawn were outlined by means of Zeiss’s camera lucida.

List oF Common REFERENCE LETTERS.

all. a. Allantoic artery. all. c. Allantoic cavity. dl. cap. Allantoic capil-
lary. all. cl. Allantoic canal. ail. ent, Allantoic entoderm. a//. mes. Allanto-
chorionic mesenchyme. ail. v. Allantoic vein. d¢/. omph. Bilaminar
omphalopleure. ec. m. Circular musculature. c@. HExtra-embryonic splanch-
nocele. ca@. w. Inner (ceelomic) wall of allantois. ch. ect. Chorionic ecto-
derm. ch. ect’. Isolated chorionic ectoderm cell. c. ¢. Interglandular con-
nective tissue. ect. Ectoderm. ex¢. Entoderm. ea. syx. Syncytium beyond
allantoic placental area. g/. Uterine gland. p. dewc. Polynuclear leucocytes.
pl. a. Allantoic placental area. pl. syn. Syncytium of allantoic placental area.
som. Somatic mesoderm of chorion. s. ¢. Sinus terminalis. syz. Syncytium.
syn. c. Capillary of syncytium. syz./. Syncytial lobule. vase. omph. Vascular

44,0) JAS. P. AIL.

omphalopleure. vi¢. a. Vitelline artery. vé¢. vo. Vitelline vein. y. spl. In-
vaginated yolk splanchnopleure.
N.B.—Unless otherwise stated, sections are transverse.

Fie. 1.—Wall of non-gravid uterus, Perameles. s. Serosa. m. Mucosa.
ep. Uterine epithelium. x 55.

Srace A.

Fie. 2.—Portion of the syncytium, showing the numerous nuclei in a
continuous protoplasmic layer. x 740.

Stace B.

Vie. 3.—Portion of left uterine wall (serosa not shown). m. c. Capillary
of corium. xX 55.

Fic. 4.—Portion of the uterine syncytium, showing the lobular character of
its deeper surface (syz. /.), the arrangement of the nuclei, and the presence
in it of maternal capillaries (syz.c.). 7. Leucocytes. m.c. Capillary of corium.
x 40.

Stace C.

Fic. 5.—7 mm. embryo, P. nasuta. all. s. Allantoic stalk. y.s. Yolk-
stalk. x about 8.

Fic. 6.—Section passing through the margin of the allantoic placental
area, and including the whole breadth of the vascular omphalopleure (vase.
omph.) on one side, and a portion of the bilaminar omphalopleure (d2/. omph.),
together with the adjacent syncytium (ew. syz.). y.s, Cavity of yolk-sac.
x 118.

Fie. 7.—Section of marginal portion of the allantoic placental area. x 140.

Fic. 8.—Portion of allantoic placental area centrally from Fig. 7, showing
the chorionic ectoderm (ch. ect.) as a continuous layer of enlarged cells ad-
herent to the syncytium (pi. syz.), as in Fig. 7. x 140.

Fie. 9.—Portion of the central region of the allantoic placental area, show-
ing degeneration of the chorionic ectoderm (ch. ect.) X 280.

Fic. 10.—Section of allantoic stalk. x 140.

Fic. 11.—Section passing through opening of allantoic canal (o. all. cl.)
into vesicular portion of allantois. x 118.

Fic. 12.—Portion of the bilaminar omphalopleure in section. x 220.

Stace D.

Fie. 13.—Right uterus opened up, showing embryo still partially enclosed
in its membranes and in relation to the allantoic placental area (pl. a.). For
description see text (pp. 411, 412). x about 62.

Fic. 14.—Section through the persistent portion of the proamnion. am.
amnion. xX 140.

THE PLACENTATION OF PERAMELES. 441

Fie. 15.—Section through the marginal portion of the allantoic placental
area (cf. with Fig. 7). x 150.

Fig. 16.—Portion of the central region of the allantoic placental area
(cf. with Fig. 8). x 150.

Fic. 17.—Portion of the allantoic placental area, showing the interlocking
of the allantoic capillaries (a//. cap.) with the vascular ridges of the syncytium.
xX 325.

Fie. 18.—Section passing through a gland opening in the allantoic placen-
tal area, occluded by an allantoic capillary (a//. cap.). x 230.

Kies. 19 and 20.—Portions of allantoic placental area, showing isolated
and greatly enlarged chorionic ectoderm cells (ch. ect’.). x 230.

Fic. 21.—Portion of allantoic placental area, showing an isolated multi-
nuclear chorionic ectoderm cell (ch. ect’.) x 2380.

Fig, 22.—Horizontal section of allantoic placental area, to show the net-
work formed by the allantoic capillaries (a//. cap.). x 225.

Fic. 23.—Section of allantoic stalk (cf. with Fig. 10). x 140.

Fic. 24.—Section passing through the sinus terminalis (s. ¢.), and including
the adjacent portions of the vascular (vase. omph.) and bilaminar omphalo-
pleure (di/, omph.), together with the syncytium (ez. syz.) in contact therewith.
x 240.

Stace H.

Fie. 25.—Genital organs from the ventral aspect. The left uterus (Z. wz.)
has been opened up, exposing the allantoic placental area (p/. a.), and the
allantoic stalk (a//. s.) has been traced back into the median pseudo-vaginal
passage (med. p.). 61. Bladder. f.¢. Fallopian tube. +, wf. Right uterus.
wu. 8. Urino-genital strand. v. cec. Vaginal ceca. xX 2.

Fic. 26.—Portion of the corium of the mucosa beneath the allantoic
placental area, showing the alterations in the connective tissue, uterine glands,
and blood-vessels. cc. ¢’. Hypertrophied interglandular connective tissue.
d. gl. Gland with its lumen occupied by the disorganised epithelium. m. v.
Vessel of corium with thickened walls. m. v'. Vessel with its lumen filled up
by the proliferated endothelium. m.v’. Vessel with its lumen completely
obliterated. ¢. p'. Thickened tunica propria around the uterine glands. x 325.

Fic. 27.—Portion of allantoic placental syncytium in section, to show
degenerative and absorptive change. jp. leuc. Polynuclear leucocytes. sp.
Spaces in syncytial protoplasm. x 550.

Fic. 28.—Section through the marginal portion of the allantoic placental
area (cf. with Figs. 7 and 15). al. mes’. Outer (placental) wall of allantois.
sp. Spaces in syncytial protoplasm. syz. 0. Lamellar outgrowth of the mar-
ginal portion of the placental syncytium. x 80.

4.42 JAS. PP. Hii.

Fie. 29.—Portion of the allantoic placental area, showing the syncytium
(pl. syn.) and the outer (placental) wall of allantois (a// mes'.). x 230.

Fie. 30.—Portion of the greatly degenerate syncytium outside the allantoic
placental area, showing the regeneration of the uterine epithelium. d. De-
tritus occupying gland lumen. ew. syu'. Degenerating syncytium. gl. ep.
Gland epithelium. g/. 0. Opening of gland into uterine lumen. 7. ep. Re-
generating uterine epithelium. x 230.

Fic. 31.—Superficial portion of the syncytium (ez. syz'.) and the regenerated
uterine epithelium (7. ep.). XX 520.

Figs. 32, 33, and 34.—Sections illustrating certain appearances found at
the margin of the allantoic placental area (cf. text, pp. 424, 425). all. m.
Margin of allantois. ad/. mes'. Outer (placental) wall of allantois. ch. ec".
Persistent chorionic ectoderm cells (?). syz.o. Lamellar outgrowth from
margin of allantoic placental syncytium. All x 230.

Fig. 35.—Section through the urino-genital strand of Fig. 25, showing the
two allantoic stalks (a//. s.) in the median pseudo-vaginal passage (med. p.).
c. bl. Blood coagulum. wr. Urethra. vag. 7. Lateral vaginal canal. x 18.

Fic. 86.—P. nasuta, new-born young. X nearly 8.

APPENDIX.

TABLE OF COMPARISON OF THE ORGANISATION OF THE EMBRYOS OF
Sraces C—H.

Form of
body

Limbs

|
4 en

|

12, rat be Stage C.
7mm.
from crown to rump.

Marked cervical flex-|Head

Facial region
definitely — estab-
lished. External
nares formed. Dis-
tinct hyomandibu-

ure.

lar groove and
precervical sinus
(Fig. 6).

indicated, the 3rd
the largest. Hind
limb still a flattened
bud. In both, plan-
tar surface still di-
rected mesially.

Notochord INotochord unconstric-

and

vertebral |

column

ted, invested by a
continuous mesen-
chymatous sheath.

P. obesula, Stage D.
8°75 mm.
from crown to rump.

still
bent, neck protu-
berance prominent.
External auditory
meatus and _trian-
gular ear pinna.
Snout now more
marked. Precervi-
cal sinus closed.

5 digits are now
distinct, lst and 5th
quite small, 3rd the
largest. Limb now
flexed at the elbow.
Plantar surface di-
rected somewhat
dorsally. Digits of;
hind limb © still
united, paddle-like,
plantar surface me-
sially directed.

and neural arches
laid down.

P. nasuta, new-born.
Stage E
14 mm. crown to rump.
Head length 6 mm.

strongly|Head raised but bent

at right angles with
trunk. Lips fused
to form ‘ Saug-
mund.” Prominent
snout. Positions of
eyes and ear pinne
just recognisable,
covered by epitri-
chium. Remains of
allantoic stalk at
navel (Fig. 36).

In fore-limb, 5 digits|In the fore-limb the|In fore-limb slender

recurved claws on
Qnd, 3rd, and 4th
digits, the third
digit the largest.
In hind limb digits
all indicated, but
not free from each
other; the 4th the
largest.

Cartilaginous centra|Marked cartilaginous

centra with trans-
verse processes and
neural arches, the
latter not yet united
above spinal cord.

4 deb

JAS. P. HILL.

|

P. obesula, Stage D.
8-75 mm.
from crown to rump.

P. nasuta, Stage C.
7 mm.
from crown to rump.

P, nasuta, new-born.

Stage E

14 mm. crown to rump.

Head length 6 mm.

Nervous
system

Eye

Har

Nose

Mouth

Distinct hemisphere|Mesial

anlagen. Dorsaland| wall thickened in
hippocampal region
but still no fissura
arcuata. Hypophy-
sis no longer com-
municates with the
oral cavity.
Indications of
*Sprossen.”

ventral spinal nerve-
roots united, Spinal
cord reaches tip of
tail, but is here ru-
dimentary. Ante-
rior commissure of
spinal cord distinct.
Anterior and poste-
rior white columns
laid down. Lateral
columns very thin.
Hypophysis _ still
connected with roof
of mouth by a stalk
with a narrow lu-

Nopigment in outer
wall of optic cup.
Lens with cavity.
Mesoderm has pen-
etrated into optic
cup.

rounded by con-
densed mesenchyme
Long ductus endo-
lymphaticus. Hva-
ginations for semi-
circular canals.

Slit-like nasal sac in|Nasal

Open communica-
tion with mouth.

Palate unformed.

men.
Optic stalk still open.|Optic stalk still with a

lumen. Outer wall
of optic cup pig-
mented. Ovalish

lens cavity. No
differentiation of

Auditory vesicle sur-|Anterior vertical semi-|Semicircular

hemisphere| Marked fissuraarcnata

along mesial hemi-
sphere wall. Cor-
pora striata deve-
loping. Hypophy-
sis dorso-ventrally
compressed.

** Sprossen ” not
very marked.

Retina deeply pig-

mented, no differen-
tiation into layers.
Hyelids united and
covered by epitri-
chium.

the retina into
layers.

canals
circular canal now| formed, but peri-
formed. Utriculus} lymphatic spaces

and sacculus still in
wide communica-
tion. Periotic cap-
sule condensed me-
senchyme.

cavities
Open directly into
mouth. Shallow-
grooved anlagen of
the organs of Jacob-
son. Solid anlagen
of lachrymal ducts.

shaped Zalnleisten
anteriorly. Taste-
buds differentiating
on tongne.

not yet differentia-
ted. Cartilaginous
periotic.

still| Adult form nearly es-

tablished.: Turbinal
projections arising.
Jacobson’s organ
formed, and its car-
tilage laid down.
Lachrymal duet
opens into nose.

Palate unclosed. Lens-|Palate formed, and

invests larynx pos-
teriorly. Teeth
anlagen. Tongue
grooved, and with
distinct taste-buds.

THE PLACENTATION OF PERAMELES.

Es sci dee Stage C.

from crown to rump.

Alimentary 'isophagus open. Se-
canal, &e. |

Heart and
vessels

Urino-
genital
system

parate dorsal and
ventral pancreas an-
lagen.
open to exterior.
Lateral thymus an-
lagen.
cavity opens into
gut.

Distinct septum supe-
rilus and septum
spurium. Com-

mencing division of

the ventricular ca-|

vity indicated by an
internal fold and
external —_ groove.
Transversely ex-
panded sinus veno-
sus, opening into
right auricular divi-
sion, Truncusaorte
undivided. Two an-
terior dorsal aorte.
Yolk - sac circula-
tion predominant.

Cloaca not!

Yolk - sac!

|

|

|

\Ventricular

P. obesula, Stage D.
8°75 mm.
from crown to rump.

exterior, still
mains of cloacal
membrane. Yolk-
sac no longer opens
into gut. Thymus
anlagen united me-
sially, but free pos-
teriorly ; they now
reach the pericar-
dium. Lungs now

re-

lobed, the right

with a ventro-mesial lobe.

4.45

P. nasuta, new-born.
Stage E
14 mm. crown to rump.
Head length 6 mm.

Cloaca just opened to/Pancreas completely

formed. Thymus
anlagen approxima-
ted in their mid-
regions. Lungs with
numerous simple
alveoli. Cartilagin-
ous rings round tra-
chea. Diaphragm
complete.

Bronchi have

branched out to form secondary bronchi.
Diaphragm still incomplete close to the

mesial line.

(septum _inferius)
almost reaches the
cushions of the au-
riculo - ventricular
ostium, likewise the
septum superius,
the auricles only
communicating be-
low its concave
edge. Sinus veno-
sus less extensive.
Inferior vena cava
established. Allan-
toic circulation pre-
dominant.

septum] Adult circulation. Po-

sition of allantoic
arteries still recog-
nisable in urachus.

Mesonephros of con-|Mesonephros now of|Mesonephros still of

siderable size, tu-
bules convoluted
and with distinct
glomeruli. | Wolf-
fian ducts open into
cloaca together with
allantoic canal. Dis-
tinct supra - renal
anlagen. Genital-
leisten small.

large size. Peri-
toneal funnels of the
Mullerian ducts laid
down. Anlagen of,
ureters as short out-
growths from the
posterior ends of|
the Wolffian ducts,
surrounded by con-
densed mesenchyme.
Genital-leisten dis-!
tinct, indifferent (?).

great size. Muller-
ian ducts laid down
for part of their
course. Permanent
kidney definitely es-
tablished and of
some size, with tu-
buli contorti. Ure-
ters openintocloaca
mesially to Wolffian
ducts. Projecting
supra-renalanlagen.
Prominent genital-
leisten, ¢ (?).

4.4.6 JAS.-P. HILL.

P. nasuta, Stage C. P. obesula, Stage D.
7 mm. | 8°75 mm.

from crown to rump. | from crown to rump.

|

|

P. nasuta, new-born.

Stage E.

14 mm. crown to rump.

Head length 6 mm.

Skin and |Skeleton in the me- Nucleated epitrichial/Thick epitrichial layer

skeleton | senchymatousstage. layer on epidermis.
No trace of hair,
claw anlagen on 3rd
and 4th fore digits.
Skeleton cartilagi-

nous.

on epidermis. Well-
marked hairanlagen
onsnout and cheeks.
No ossification in
cartilaginous skele-
ton, which is now
fully formed. Up-
per and lower jaw
ossifications.

Comparing the above three embryos with the tables and figures in Keibel’s
‘Normentafel zur Entwickelungsgeschichte des Schweines (Sus scrofa
domesticus),’ Jena, 1897, they correspond, as nearly as one can judge, as

follows :
Embryo C to No. 85, Fig. 24.

Embryo D barely to No. 91, Fig. 28.

Embryo E to No. 93.

GREEN PIGMENT OF INTESTINAL WALL OF CHMTOPTERUS. 447

On the Green Pigment of the Intestinal Wall
of the Annelid Chetopterus.

By

E. Ray Lankester, M.A., LL.D., F.R.S.,
Linacre Professor, and Fellow of Merton College, Oxford.

With Plates 34—37.

I. INTRODUCTION.

In the year 1864 I obtained specimens of Chetopterus
variopedatus, Renier (at that time called C. insignis by
Baird), when collecting at Herm in the Channel Islands.

My specimens preserved in alcohol gave to the spirit a
strong blackish-brown coloration, and the fluid was observed
to have a deep red fluorescence. I showed the coloured fluid
to Professor Stokes, of Cambridge, where in 1864 I was an
undergraduate, and he rapidly examined it with a direct-vision
spectroscope. He pointed out to me the remarkable absorp-
tion bands which the fluid caused in the spectrum of light
passed through it, and expressed the opinion that these were
similar to if not identical with those caused by some solutions
of chlorophyll. In view of the fact that the colouring matter
was soluble in alcohol and caused a red fluorescence, as well
as a banded absorption spectrum resembling that of chloro-
phyll, Professor Stokes was of the opinion that the source of
the colour was probably to be found in chlorophyll swallowed
by the marine worm which had been immersed in the spirit.

A year or two later I acquired a Sorby-Browning spectro-
scope of my own, and became familiar with the absorption
spectra of chlorophyll, of hemoglobin, of turacin, and many

VOL. 40, PART 3.—NEW SER. HH

448 BE. RAY LANKESTER.

other organic pigments. I satisfied myself that the pigment
derived from Chetopterus was in no way connected with the
lodgment of particles of green vegetable matter in its alimen-
tary canal, but was due to a strong and abundant blackish-
green substance formed in the actual walls of the middle
portion of the alimentary tract of the Chetopterus. I sup-
posed that this substance, on account of its solubility, fluores-
cence, and banded absorption spectrum, must be considered as
a ‘ variety’ or ‘ species’ of “ chlorophyll.”

In 1872 I obtained at Naples specimens of the Gephyrean
Bonellia viridis, and was extremely interested to find that
they—like Cheetopterus—imparted a strong greenish colour to
the alcohol in which they were placed, accompanied by fluores-
cence.

This coloured solution I found also gave a very powerful
series of absorption bands when examined with the spectro-
scope, and I erroneously concluded that this colouring matter
too—from Bonellia—must be considered as a “ chlorophylloid”
substance.

In the first and second editions of the English translation of
Sach’s ‘ Botany’ (Oxford, second edition, 1882, p. 767) a note
is published in which, on my authority, it is stated that “a
chlorophylloid substance” occurs in the intestinal wall of
Cheetopterus and in the integument of Bonellia.

I had observed that the absorption spectrum of the colouring
matter of Bonellia, though resembling that of chlorophyll, yet
differed considerably from it, and I proposed to myself to make
a more careful examination both of it and of the green pig-
ment from Cheetopterus, with the use of adequate measuring
apparatus and scale in fixing the position of the bands in their
absorption spectra.

I was prevented from carrying out my purpose by the loss
of my Chetopterus material and the difficulty of obtaining
more, and by the pressure of other pieces of work.

It fortunately occurred to me in 1875 to ask my friend Mr.
H. C. Sorby, F.R.S., to undertake the investigation of the
colouring matter of Bonellia, of which I had a certain quantity.

GREEN PIGMENT OF INTESTINAL WALL OF CHMTOPTERUS. 449

The result was the interesting and important paper by Mr.
Sorby published in this Journal in 1875, wherein he describes
the acid, neutral, and alkaline conditions of this pigment,
characterises it spectroscopically, and gives to it the name
* Bonellein.” I shall take the liberty of changing that name
in the present paper to “ Bonellin.”

Bonellin was subsequently studied by Krukenberg (‘ Vergl.
Physiol. Studien,’ ii, 1882), and by Schenck (Sitzb. k. Akad.
Wiss. Wien,’ 72, ii), who did not add anything material to
the observations published by Sorby. Krukenberg, indeed,
fell into some errors on the subject.

After Sorby’s demonstration of the independent and peculiar
nature of Bonellin, I was more than ever anxious to re-examine
the pigment of Chetopterus, but more than twenty years have
elapsed before I have carried out my intention.

In the meantime my friend and fellow-student Moseley
provided himself with a micro-spectroscope as part of his
equipment when he served as naturalist on the “ Challenger”
Expedition, and he published on his return in this Journal
(vol. xvii, 1877) a very interesting account of a number of
colouring matters obtained from marine animals and exa-
mined by him with the spectroscope. One of the most
remarkable of these was a pigment from the integument of
species of Pentacrinus, which occurs in these animals in such
abundance that the spirit in which they are first preserved
becomes deeply impregnated with it, and will yield the pig-
ment as a powder on evaporation.

Moseley gave the name Purple Pentacrinin to this substance.
He showed that its solution in alcohol was fiuorescent, and,
like Bonellin, could be changed in colour accordingly as the
fluid was rendered alkaline or acid. It gave a very definite
set of three absorption bands, the position of which were fixed
and recorded in a spectrum map by Moseley. The colour
transmitted by the acid solution is “an intense pink,” that by
the alkaline is bluish green, whilst the fluorescence is red.
A second equally striking colouring matter was called Antedonin
by Moseley, and found by him in deep-sea species of Antedon

450 E. RAY LANKESTER.

(not in the European forms), and also in a Holothurian. This
pigment also shows an acid and an alkaline condition, and can
be converted from one to the other an indefinite number of
times. It is, however, soluble in distilled water, as well as in
alcohol.

The Bonellin of Sorby and the Pentacrinin of Moseley
indicate the existence of a group of tegumentary pigments in
marine animals, which whilst they resemble chlorophyll in
their solubility in alcohol, non-solubility in water, in the
fluorescence of their solutions, and in their banded absorption
spectra, yet differ from chlorophyll or any known derivative of
that substance in the exact position and number of. their
absorption bands, and in their relative stability when exposed
to sunlight, but most characteristically in the fact that they
undergo a striking change of colour and in the position of their
absorption bands, accordingly as the solution is rendered acid
or alkaline, whilst the change from the acid to the alkaline
state and back again can be effected an indefinite number of
times without destruction of the pigment.

I may state at once that the pigment from the intestine of
Cheetopterus appears to be one of this class of bodies, and I
propose to speak of it as Cheetopterin.

Whilst I do not propose on the present occasion to attempt
a review or classification of animal pigments, I think it appro-
priate to point out that there are other green tegumentary
pigments among Gephyrea, Chetopoda, and Arthropoda which
have properties very different from those of the Bonellin group.

I shall only refer to three of these, and but briefly. In the
present number of this Journal Professor Herdman describes a
new green-coloured Thalassema, which he has been kind
enough to dedicate to me, At first one would naturally be
inclined to suppose that the green pigment of Professor
Herdman’s Th. Lankesteri must be identical in nature with

1 The Holothuria nigra of the Cornish coast imparts a magnificent
flame colour to the alcohol in which it is placed, and the solution has a
brilliant green fluorescence. The absorption spectrum has not been studied
but it gives, I believe, no detached bands.

GREEN PIGMENT OF INTESTINAL WALL OF CHATOPTERUS, 451

that of the allied Bonellia viridis. It is, however, quite
unlike it. In actual tint Th. Lankesteri is of a much
brighter green than Bonellia viridis—tending to what is
called apple-green, whilst Bonelliais rather to be described as
chrome-green. The colour of Th. Lankesteri is exactly the
same as that of Hamingia arctica. Iam able to state this
as I am, I believe, the only person who has seen and recorded
by a coloured drawing this Gephyrzean in a living condition.
I dredged it in company with the Rev. Dr. Norman, F.R.S.,
and Professor Bourne, F.R.S., of Madras, at the mouth of
Lervik Harbour, Stordoe, Norway, in 1882. Moreover, the
green pigment of both Thalassema Lankesteri and Ha-
mingia arctica differs from that of Bonellia in that it is
not soluble in alcohol.! According to Professor Herdman’s
observations, the pigment of Th. Lankesteri is slightly
soluble in formol. Whether this signifies more than that the
water holding the formaldehyde in solution takes up the
pigment, and would do so even were the formaldehyde
not present, seems doubtful. I have no observation as to
solubility in water in referencé to the green pigment of
Hamingia, but it appears to me highly probable that it is
identical with that of Thalassema Lankesteri. In addition
to their want of solubility in alcohol, these two pigments differ
from that of Bonellia in not yielding a series of detached
absorption bands. I determined this in the case of Ha-
mingia, and Professor Herdman has done so for his new Tha-
lassema. Thalassemin does not change its colour when acted on
by dilute acids, whilst Bonellin is changed to a rich violet tint.

A second instance of green tegumentary pigments differing
from the Bonellin group is presented by Idotea viridis, the
Isopod crustacean. The pigment is in this case insoluble in
water, alcohol, or benzine. It is of a brilliant grass-green
colour, and is not improbably similar in character and origin
to the green pigment situated in the skin of some Lepidopterous
larve, and other adult leaf-frequenting insects. It is time

1 See Professor Herdman’s paper, and accompanying notes by Professors
Sherrington and Noél Paton, and Miss Newbigin.

452 EK. RAY LANKESTER.

that an effort was made to arrive at a further knowledge of these
insoluble pigments.

A third case to which I wish to allude is the green colora-
tion of the blood of some Lepidopterous larve. My friend and
colleague Professor Poulton has shown (‘ Proc. Roy. Soc.,’
1885, and vol. liv, 1893) strong reason to suppose that the
green colour of the blood in these larve is determined by the
presence of chlorophyll or of etiolin in the food consumed by
them. He has shown that the blood gives an acid reaction ;
the suggestion made by him is that the green chlorophyll or its
immediate antecedent passes through the wall of the alimentary
canal from the digested food into the blood in a modified state,
which he calls metachlorophyll. Such a passage seems, on the
other hand, to be impossible without a very radical change in
the chlorophyli, which is itself neither diffusible nor soluble in
watery media. It is most desirable that the study of the green
pigment in the blood and skin of Lepidopterous larve should
be carried further. One circumstance which induces me to
allude to it here is that Professor Baldwin Spencer, of Mel-
bourne, in his fine memoir ‘on Pentastoma published: in this
Journal in 1893, vol. xxxiv, p. 31, made a suggestion similar to
that made by Professor Poulton in regard to the passage of chlo-
rophyll through the intestinal wall. Professor Spencer found
that the perivisceral fluid of Pentastoma was coloured blood-
red by hemoglobin, and he supposes that the hemoglobin has
passed from the cavity of the gut of the Pentastoma through its
wall into the perivisceral fluid. Some of the Nematoid parasites
of birds have a blood-red perivisceral fluid which has not been
examined spectroscopically. Possibly it also is due to hemo-
globin, and might throw further light on the question. It isa
noteworthy fact that the suggestion should be made from two
separate sources, that non-diffusible substances like chlorophyll
and hemoglobin pass through the wall of the alimentary canal
into the blood fluid unchanged or but little changed. There
is evidently something here worthy of further investigation.
With regard to the supposed occurrence of chlorophyll in the
blood of Lepidopterous larve, Professor Poulton’s spectro-

GREEN PIGMENT OF INTESTINAL WALL OF CHATOPTERUS. 453

scopic observations seem to prove conclusively that the green
pigment present in the blood is not chlorophyll. In order to
prove that it is a direct derivative of the chlorophyll taken in
with food into the alimentary canal, it seems to be necessary
to study the derivatives of chlorophyll, and to show that by
chemical processes a substance can be produced from chloro-
phyll having the absorption spectrum of Poulton’s metachloro-
phyll, which it has not; having the power of resisting the
destructive action of light, which it has not ; capable of diffusing
through a living membrane, and of existing in solution in an
acid albuminous fluid, which it is not ; and lastly of changing
to an opaque blackish brown pigment when simply exposed to
oxygen gas, which it is not.

Il. Description oF CH#TOPTERIN: ITS MopE or OccuR-
RENCE AND OPpTicaAL PROPERTIES.

Mode of Occurrence.—Dr, Blaxland Benham has kindly
furnished me with an account of the mode of occurrence of the
intestinal pigment of Chetopterus variopedatus from
observations made by him at my suggestion in Mr. Hornell’s
laboratory in Jersey in the summer of 1896, and on material
preserved by him in formol and brought to Oxford. The
drawings in Plate 34 are by Dr. Benham, who has also recorded
the spectra of Cheetopterin and has carried out similar observa-
tions on Bonellin in my laboratory at Oxford. I am greatly
indebted to him for his assistance in preparing this account of
the two pigments.

The body of this strange-looking Cheetopod is divided into
three regions, as shown in fig. 1. The dark green, almost
black-looking pigment is confined to the intestinal epithelium
of the middle region. It gives the whole of the inner surface
a black appearance, and can be seen through the transparent
tissues of the body-wall.

In a transverse section its disposition is seen to coincide
with that of the entire epithelial layer of the intestine, as
shown in fig. 2. In order to observe satisfactorily its natural

454. E. RAY LANKESTER.

position, the use of alcohol must be avoided since the pigment
is dissolved by that preservative. It is, however, insoluble in
formol.

Claparéde, in his ‘Annélides Sedentaires,’ has described
the pigmentation of the epithelium of this region of the in-
testine. A careful examination, by teazing fresh material and
also by sections of material preserved in formol, shows, ac-
cording to Benham’s observations, that the pigment occurs
solely in the form of spherical corpuscles varying in size
(fig. 4), and embedded in the protoplasm of the epithelial
cells (fig. 3). These granules are not dissolved by alcohol
entirely, but a colourless, oily-looking stroma, quite structure-
less and translucent, of the same shape as the original coloured
granule, is left in the cell-body.

Claparéde speaks of the pigment in the intestinal wall as
“hepatic” pigment. Joyeux-Laffuie (‘Archives de Zoologie
Expérim.,’ 1890) gives a detailed account of the distribution
of the pigment-bearing cells in the intestinal wall; he figures
the cells and terms them ‘cellules biliaires.”” Dr. Benham
distinguishes the elongated ciliated cells which contain the
green granules from other associated “ gland-cells,’ of which
there appear to be two varieties.

It is evident that the terms ‘ hepatic pigment” and “bile-
cells” are not open to the same objection when applied to
these cells of the enteric epithelium, as when applied, accord-
ing to the custom of writers of forty years ago, to the brown-
coloured tunic of the earthworm’s intestine, now often called
the ‘‘chloragogenous” tunic or cells. The pigmented cells of
the intestine of Chetopterus are really of enteric origin, as is
the hepatic gland in Vertebrates, whilst on the other hand the
chloragogenous tunic is part of the ceelomic epithelium.

It is impossible to suppose, in view of the fact that Cheeto-
pterus lives buried in the sand in a large parchment-like tube,
that the intestinal pigment can have any function as pigment.
On the other hand, it is not unlikely that it may eventually be
shown that this green fluorescent ‘‘ Chetopterin” is really re-
presentative of the biliverdin of Vertebrate bile.

GREEN PIGMENT OF INTESTINAL WALL OF CHATOPTERUS 4505

An extremely interesting comparison suggests itself with the
* entero-chlorophyll” described by Dr. MacMunn as occurring
in corpuscles in the livers (gastric glands) of Mollusca and
in other Invertebrata (‘ Proc. Roy. Soc., vol. xxxv, p. 370). Dr.
MacMunn was probably ill-advised in using the term ‘ chlo-
rophyll”’ in connection with the substance discovered by him.
I have no doubt that he was led by my own erroneous classifi-
cation! of several green pigments in animals under the chloro-
phyll group. It is not possible to come to a conclusion from a
comparison of the absorption spectrum assigned by Dr. Mac-
Munn to his entero-chlorophyll with that of Chetopterin, since
Dr. MacMunn’s pigment is very difficult to obtain free from
admixture with other substances. On the other hand, Cheto-
pterin can be obtained in a fairly pure condition, so far as the
admixture of other pigments is concerned, by isolating the
mid-region of the body of Chetopterus and preparing the
alcoholic solution from that region only. It is true that even
so the solution contains fatty matters and other impurities,
and that we have not yet obtained Chetopterin either as a
pure thoroughly cleansed powder or in the crystalline con-
dition.

An investigation of the chemical properties of Chetopterin
has been undertaken at my request by Miss Newbigin in
Professor Noél Paton’s laboratory, and there is reason to hope
that before long we shall obtain this body in a chemically pure
state, and learn something as to its chemical constitution and
properties, which cannot fail to throw light on its physiological
significance and possible relationship to MacMunn’s entero-
chlorophyll.

The determination of the characters of a body occurring in
such definite form in the enteric epithelium of one of the
simpler forms of animal life cannot but lead to a better under-
standing of the physiology of the alimentary canal, and of the

1 This error of course did not include the green pigments of Spongilla,
Hydra, and such ciliate Protozoa as Stentor. In them there is no doubt that

the green pigment is chlorophyll, and that it occurs, as in plants, in self-
propagating corpuscles.

456 E. RAY LANKESTER.

internal chemical activities of the celis, upon a knowledge of
which a true physiology must be based.

Colour and Absorption Spectra of Chetopterin.—
The freshly-prepared alcoholic solution of Chetopterin (as
obtained from a fresh specimen of the mid-region of the
animal’s body) is of a blackish green colour by transmitted
light (see P]. 35, fig. 6), and shows a powerful red fluorescence,
resembling in colour that of an alcoholic solution of chlorophyll.
This solution is found to be neutral in reaction. When
examined with the spectroscope it shows four detached absorp-
tion bands, the position and intensity of which are represented
in Dr. Benham’s drawing (PI. 34, fig. 5, uppermost spectrum),
but are more exactly shown in the valuable observations kindly
made for me by Professor Engelmann, and recorded in the two
charts on Pl. 36.

When the neutral solution of Chetopterin is rendered acid
by a very slight addition of HCl, it assumes a fine indigo-blue
colour, as shown in PI, 35, fig. 7. The absorption spectrum is
sul four-banded, but the position of all the bands is shifted,
notably of the two in the blue. Dr. Benham’s drawing in
Pl. 34, fig. 5, shows this; the exact position and intensity of
the absorption is shown by the two dotted lines in Professor
Engelmann’s chart, Pl. 36. Professor Engelmann finds in a
sufficiently thin layer of the coloured liquid a faint ‘ fifth ”
band at wave length 500, indicated by a dip and rapid rise in
the curve traced by the upper dotted line of his chart. Ina
layer of greater thickness (recorded by the lower dotted line)
the differentiation of this band from the absorption on either
side of it is (as in Dr. Benham’s drawing) inappreciable, ex-
cepting by the most careful measurement and comparison.

The acidulated solution may now be rendered alkaline by
addition of KHO or NaHO, when it assumes a bright lemon-
green colour (Pl. 35, fig. 8). The alkaline solution still
exhibits four detached absorption bands, but they are very
much in the same position as those of the neutral solution.
The difference in the colour of the neutral and the alkaline
solution is due, as is shown very clearly by Professor Engel-

GREEN PIGMENT OF INTESTINAL WALL OF CHATOPTERUS. 457

mann’s chart (PI. 36), not to a difference in the position of the
points of maximum absorption, but to a difference in the
position of the points of maximum luminosity, and consequently
of the area and graduation of the absorption around its maxima.
This is very difficult to represent or record by shaded drawings,
but is given with absolute precision by Professor Engelmann’s
beautiful method of observation and record, of which I will
give some explanation below. The alkaline solution can be
rendered again neutral or acid, and the process reversed and
repeated indefinitely.

III. Cotour anp ABSORPTION SPECTRA OF BONELLIN.

Thad been anxious to compare the absorption spectra of Bo-
nellin and Chetopterin for myself, and after I obtained a supply
of the latter was unable for some time to procure the former.

I heard, however, in 1896 that Bonellia was flourishing in
the beautiful healthy tanks of the Laboratoire Arago, erected and
directed by Professor Henri de Lacaze Duthiers at Banyuls-sur-
Mer, near Perpignan. I accordingly wrote to that distinguished
zoologist, stating my desire to examine living specimens of
Bonellia in Oxford. With a kindness and courtesy for which
he is universally known and beloved, Professor de Lacaze
Duthiers sent to me from Banyuls, by express parcels-service,
two bottles of sea water, containing each a magnificent speci-
men of Bonellia viridis, which arrived in Oxford in a
perfect condition of living vigour. I was thus able to examine
again the pigment Bonellin, and to satisfy myself as to the
position in which it occurs in the body of Bonellia. My best
thanks are due to Professor de Lacaze Duthiers, and are here
recorded, for his great kindness.

Greef has already correctly described the mode of occurrence
of the green pigment of Bonellia. It is distributed in the super-
ficial ectodermic epithelium in the form of very fine granules
which give the ectodermic cells a grass-green appearance. It
also occurs as fine granules in clusters of subepidermic cells
apparently belonging to the connective tissue,

458 E. RAY LANKESTER.

From the specimens received in Oxford, after examination of
the histological relations of the pigment, Dr. Benham prepared
an alcoholic solution which we proceeded to study by means of
the spectroscope, and the application of acids and alkalies. A
portion of the pigment in alcoholic solution was sent by me in
the spring of 1897 to Professor Engelmann, together with the
solution of Cheetopterin. I am thus able to give here a very
accurate record of the absorption spectra of Bonellin (Plate 37),
for the purpose of comparison with those of Chetopterin. It
will be seen at once that the two bodies differ entirely from
one another in colour and absorption phenomena, whilst
agreeing in solubility, fluorescence, and in the exhibition of
neutral, acid, and alkaline conditions.

The appearance of the absorption bands of alkaline and
acidulated alcoholic solutions of Bonellin, as seen with the
Sorby-Browning micro-spectroscope, are drawn by Dr. Benham
in Plate 35, fig. 5. The freshly prepared solution of Bonellin
is alkaline, of a deep chrome-green colour. It is in this condi-
tion that the pigment appears to exist in the skin of the
animal (Plate 35, fig. 11). When neutralised the solution
assumes a greyish-blue colour (Plate 35, fig. 9). The addition
of a small quantity of acid to the neutral solution, changes
the colour to a splendid violet (Plate 35, fig. 10). The ab-
sorption spectra of these three conditions of Bonellin have
been described by Sorby and after him by Krukenberg and
by Schenck (who erroneously regarded Bonellin as a form of
chlorophyll).

It will be found that the statements of these authors (cited
on p. 449) are at variance in minor details with one another,
and also with what I now place on record as the result of the
observations of Professor Engelmann.

The carefully neutralised alcoholic solution of Bonellin
exhibits but four marked or isolated maxima of absorption
(absorption bands), as shown by Professor Engelmann’s chart
(Plate 37). These are as different in position as they well can be
from the four bands of Chetopterin. There is no need to refer
any further to a possible relationship between these two bodies.

GREEN PIGMENT OF INTESTINAL WALL OF CHATOPTERUS. 459

The acid Bonellin also exhibits four, and only four absorption
bands, but these do not coincide with any of the four bands of
the neutral solution. The alkaline solution presents a six-
banded spectrum, and of these six it is remarkable that the
strongest, viz. the first and the sixth, coincide in position with
those of neutral Bonellin; whilst the remaining four are
similar to those of acid Bonellin, but all shifted a little towards
the red end of the spectrum.

I am not prepared to discuss here either Sorby’s slight
divergences from Engelmann’s record or Krukenberg’s theory
of Bonellin and Bonellidin. My principal object is to show
how widely Bonellin differs from Cheetopterin (though resem-
bling it in general characters), and to present an accurate
record of the absorption phenomena of neutral, acid, and
alkaline alcoholic solutions of the pigment as obtained from
fresh specimens of Bonellia.

It now only remains for me to give some explanation of the
method of observation and record of absorption spectra—intro-
duced by Professor Engelmann,—without which the reader
will not properly understand the value of Plates 86 and 37.

1V. MEASUREMENTS OF THE ABSORPTION SPECTRA OF CHA-
TOPTERIN AND BoNELLIN BY PROFESSOR ENGELMANN.

Professor Engelmann kindly offered last year (1896), when I
was on a visit to Utrecht, to apply his beautiful instrument
for the measurement of the absorption of the luminous spec-
trum by coloured media to Chetopterin and Bonellin. I was
very glad to avail myself of his kind offer, in order to procure
a more accurate record of the position and intensity of the
absorption bands given by those pigments than is possible with
the ordinary micro-spectroscope. The charts forwarded to me
by Professor Engelmann as a result of his examination of the
solutions which I sent to him arereproduced in Plates 36 and 37.

The instrument used by Professor Engelmann is described
by him in the ‘ Archives de Microscopie.’ It is applied to the
body of an ordinary microscope, and consists in an arrange-

4.60 FE. RAY LANKESTER.

ment by which the light from a powerful incandescent Jamp is
passed through two parallel slits, A and B, giving in the field
of view of the eye-piece spectroscope two spectra exactly
parallel to one another, and of exactly equal intensity of light.
A diaphragm is made to traverse the field by the turning of a
screw, so as to present for observation a narrow band only of
the juxtaposed spectra. The exact wave length of this strip
is given by a scale introduced. Thus a width of the spectrum
corresponding to a range of only some two or three millionths
of a millimetre in wave length and of measured position in
the spectrum can be examined, whilst there is on either side
absolute darkness. The portion of the strip of light thus
studied belonging to the light coming through slit A can be
compared, as to the amount of light present, with the identical
representative portion of the spectrum belonging to slit B.
Under the conditions so far stated, the intensity of illumina-
tion (amount of light) of each half of the strip (that belonging
to spectrum of slit A, and that belonging to spectrum of slit
B) are exactly and sensibly equal. If now there be placed in
front of slit A a coloured transparent body, some of the light
passing through that slit will be stopped. Suppose the travel-
ling eye-piece diaphragm is adjusted so as to present to the
observer a strip of each spectrum corresponding to a wave
length which is partially absorbed by the coloured medium
introduced before slit A, then the portion of the strip
belonging to slit A will be sensibly dimmer than that belong-
ing to slit B.

Now by a micrometer screw Engelmann can reduce the
width of slit B until the amount of light coming through that
slit is no greater than the amount coming through slit A, ob-
scured as it is by the coloured medium. The amount of
movement of the screw required to bring the light of slit B
down to the intensity of that of slit A furnishes the measure-
ment of the absorption due to the coloured medium for the
wave length under observation (isolated by the travelling eye-
piece diaphragm slit).

The micrometer screw is standardised so as to give readings in

GREEN PIGMENT OF INTESTINAL WALL OF CHATOPTERUS. 461

percentages of the total amount of light passing through the slit
when not obscured, and having an aperture of *2 millimetre.
The percentage is then written off on the chart by a dot
corresponding to the wave-length (right or left of vertical lines
of the chart), the percentage itself being given by the higher
or lower position of the dot in relation to the horizontal lines.
The following is the report kindly furnished to me by Prof.
Engelmann, together with the table of measurements and the
charts given in Plates 36 and 37. It will be seen that by the
present method it is easy to calculate the form of absorption
curve for a greater thickness of solution from the observation
of that of a less, and vice versa, and that this has enabled
Prof. Engelmann to apply a satisfactory test to the accuracy of
his observations, which it must be remembered depend upon a
very delicate comparative judgment of the light intensity of
the two adjacent strips of spectrum, one of which is gradually
darkened by the turning of the micrometer screw until it is
judged to be exactly equal in light intensity to the other.

Urrecnat; 25th April, 1897.

At last I am able to send you the results of the
quantitative colour-analysis of your Cheetopterin and Bonellin.
You will find them in the accompanying tables, and in graphic
form on the four charts of curves.

The measurements were carried out with my micro-spectro-
photometer (‘ Zeitschr. f. wissen. Mikroskopie,’ Sept., 1888';
‘Archives de Microsc., xxii, 1888, p. 82, pl. iv; ‘Onder-
zochingen gedann in het physiol. laborat. des Utrechtsche
Hoogeschool.,’ 3, xi, 1889, pp. 89—49) with the use of an
Auer’s incandescent lamp as the source of light. The slit width
of the spectrum apparatus was in all cases 0°2 mm. The light
intensity thus given without absorption is taken as = 100 for
all wave lengths. Measurements were made of the intensity
of the light for a large number of X, after passage through a
plane-parallel layer of coloured solution of 5 and also of 2mm.
in thickness.

The measure of this intensity was in every case the width of

462 E. RAY LANKESTER.

the comparison slit, by which, for a given wave length, an
apparently equal luminosity in the absorbed and in the com-
parison slits was given. The apparatus allows one easily to
read off a slit-width as small as 0°0005 mm. Every measure-
ment was five times repeated. The number set down on the
records is in each case the mean. ‘Their probable error is in
general barely more than 1 per cent. of the measured value;
only in the outer red and violet, and with very powerful
absorption, is the error greater on account of the diminished
intensity of light. The course of the absorption curve (as
drawn on the charts) will then in all essentials faithfully re-
present the fact.

An objective test for the criticism of the trustworthiness of
the measurement results is afforded by the comparison of the
numbers arrived at when two different thicknesses of the
absorbing layer are used. If for any wave length the intensity

of the original light is weakened to - per cent, by passage

through a layer of the thickness 1, then for the same wave
length the intensity (= 7) after passage through a layer of the

thickness 1 is given by the formula 2, = = Accordingly, if

we have measured the course of the absorption for a known
thickness of layer, we can reckon it also for every other possible
thickness of layer. I have carried out the calculation for neutral
Bonellin and neutral Chetopterin, and inserted the calculated
values in thick type in the tables. The agreement between
calculation and observation is amply sufficient, especially when
we consider that important alterations in the light intensity
value must be brought about by minute changes in the position
and breadth of the spectrum strip, the average light intensity
of which is being determined, in those parts of the spectrum
where there are sharp alterations in the absorption. The
greatest care was given on this account to the exact position and
borders of the spectrum strip, and every time it was carefully
determined whether the wave length scale was exactly in its
proper position. The breadth of the spectrum strips, separately

GREEN PIGMENT OF INTESTINAL WALL OF CHETOPTERUS. 4638

examined as to light intensity, was on the average equal to
4001 « wave length. For example, for determining the
intensity at AX = 600 uu, the spectrum strip lying between
wave lengths 0°597 mw and about 0°630 « was isolated by means
of the ocular screw diaphragm slit (accordingly the rest of the
spectrum shut off). In the outermost red, where the disper-
sion is too small, spectrum strips of 0:°01—0-02 breadth were
isolated.

The coloured solutions were examined in small glass chambers
of known height, which I had prepared for the purpose by
Zeiss. They are to be recommended also for merely qualita-
tive spectroscopic observations cn coloured solutions, since one
can work with a very small quantity of fluid (a few cubic
millimetres). I intend soon to describe them and explain their
use more fully. You can get them from Zeiss.

As you will observe, I have analysed a neutral as well as an
acid and alkaline solution. All three show characteristic
differences, and indeed the colours also appear different to the
eye. Itisa pity that neither Bonellin nor Chetopterin have
been prepared in a chemically pure state, and perhaps cannot
be. If they were, one could make exact determinations upon the
(clearly very great) influence of the solvent upon the concen-
tration of the solution.

voL. 40, PART 3.—NEW SER. II

464.

EH. RAY LANKESTER.

CHATOPTERIN.

\ = Wave lengths in pp (1 = 0°001 yp).
?» = Intensity of the perpendicularly falling light passing through a plane-
parallel layer of z mm. thickness, in percentages of the original
light intensity.

a. Neutral alcoholic
solution.

met et
700/625
680\44-2
670|29°5
655| 2-4
640\11-2
625/30°6
GO0|L45/47°3

iy
87°
73"
4.
18°
42
BT:
20°1/52-0
21-0/53-7,

580
570

~ §60)17°6/50°7)

535|12°8/43°2)

520|15°8)48°2|

500|10°9/42°2
480|14°3)45°4

460
44.0

9°8/39-2
1°6/15°5

420) O°7| 9°5

7
5
5
7
0
5

(2, caleu-

lated

(72:0)
(54:0)
(23:0)

(46:0)
(52:0)
(53:0)
(49:6)
(44:0)

(48:0)
(41:4)

(45:7)

(39°6)
(19:0)

(12:0)

b= a, made acid by
|

e=b, made alkaline

by NaHO.

I. Min.

II. Min.

ILI. Min.

IV. Min.

640

560
550
540
533
520
510
500
490
480
470
460
450
44.0

4.20

i

23-0 51-0
15-0445 IL. Min.

18°5 50°0

14:0 44°0 LLL. Min.

16:0 46:0
17:0.48°0,

15:0,44-0 1V. Min.

17-0 46°0.
17°5 48°5)

16:5 47°5 V.? Min.

17:0 500°
16:0 49-0
12:0 —|
9:5 45:0
3:5) —

P (24:0.

(too little light)

i ie

655.
640

610
590
575
560
590

540
520

510

500
480

600,

9-0
15-0,

§20 23° 054

21°5

15-0
18-0
175)
14-0

13°5

11-0
13°5

8°5

5°5
9°0

460 4°5

44.0

2°0

P

40:0
445

37°5

32°5
36°5

22°0
130

50

I. Min.

‘O. LL. Min.

LIT. Min.

IV. Min.

GREEN PIGMENT OF INTESTINAL WALL OF CHHTOPTERUS.

A = Wave lengths in pu (1 = 0:001 p).
én = Intensity of the perpendicularly falling
parallel layer of ~ mm. thickness,

light intensity.

BONELLIN.

4.65

light passing through a plane-
in percentages of the original

a. Neutral alcoholic

\}

solution. |

(7,calcu-|

lated !
feo |. tgs \frount,:) r.
74:2 89-8 | (88°8) “700
69°6 86:5) (86°5) 680
60:8 83-2 (82-0) 645
1-7 18-0| (19°6) | I. Min. | 630
AT5 79:2! (74-2) 620
504 75°2) (76 0) 613
53°7 78:2 | (78:0) 600
50°9 75°8| (76-3) | II. Min. 590
62°3. 80:3 | (829) 570
53-9 78-0 | (782) 559
50°5 75-8 | (761) | III. Min. | 545
52°0 | 77°7 | (77:0) 535
23°1 | 56°8 | (55°6) | LV. Min. | 515
45°8 71:0} (78 2) 500
| 480
49°6 73-0| (75:5) 460
| 4.40
50°2 73°5| (75'8) (433
| 430
| 425
| 420

|) =a, made acid

by HCl.

o
8 *
73°5
64-9
48°9
16°7
6°5
36°9
43°4.
38°2
39°3
aya
39°S
35°6
38°3
49°9
52°6
59-1
47°1
43°3 |
40°7
37°7 | V. Min.

I. Min.

II. Min.

TV Min:

Hil. Min.

c = 4, made

alkaline by NaHO.

7
> 7

|
635 | 10°6/I. Min.
625 23:3 |
614 20°3
| 605 35:2 |
| 595 140-5
565 | 39-0
550| 37:3
540 37°7
520 328°8

LI. Min.

ILL. Min.
11 V. Min.

V. Min.
| |

510 36-7
490 380°7) VI. Min.

470 45°8
450 55:0

466 E. RAY LANKESTER.

EXPLANATION OF PLATES 34—37,

Illustrating Professor Ray Lankester’s Memoir on ‘The
Green Pigment of the Intestinal Wall of the Annelid
Cheetopterus.”

PLATE 34.

Fie. 1.—Chetopterus variopedatus, drawn of the natural size, as
seen when removed from its tube. 4. Anterior or cephalic region. B. Mid-
region (in which the dark pigment occurs). C. Posterior region. a. Pinnule
(modified notopodium of the 11th segment). 4. Median sucker formed
by modification of the pair of notopodia of the 12th segment. c. “ Fans”
formed by fusion of right and left notopodia in segments 13, 14, 15. d.
Neuropodia of segments 12, 18, 14, 15. e. Right nephridiopore of the
15th segment. £. Notopodial cirrhi of posterior segments. g. Neuropodia
of posterior segments. 4%. Right phosphorescent gland at the base of the
right pinnule. iz¢. Pigment of the intestinal wall showing through the
integument.

Fic. 2.—Transverse section of the body of Chetopterus variopedatus
taken at the point marked éz¢. in fig. 1. (From a drawing by Dr. Benham.)
a. Dorsal musculature forming a median crest or ridge. 4. Transparent
integument. c. Connective tissue holding the gut-wall to the body-wall.
d. Cavity of the gut. e. Green pigmented epithelium of the gut. f. A
nephridium in section. g. Ventral musculature. 4. Nerve-cords.

Fic. 3.—Epithelial cells of the gut to show the position of the green
granules. From a section made from a specimen preserved in formol so as to
avoid the solution of the green granules which occurs when alcohol is
employed. a. Free surface of the epithelium. 4. Branched base of an
epithelial cell. ¢. Oval nucleus.

Fic. 4.—Some of the green granules detached from the epithelial cells and
more highly magnified, showing their varied size and spherical form. (Drawn
by Dr. Benham.)

Fic. 5.—Absorption spectra of Chetopterin and Bonellin as seen with
Sorby’s micro-spectroscope. (Drawn by Dr. Benham.) Besides the shading
or absorption in the form of bands of greater or less breadth, the position of
the chief Frauenhoffer lines is indicated, and the whole spectrum is divided
into thirty-five spaces, the divisions between which correspond to wave lengths
ranging from 400 millionths of a millimetre on the right (blue end) to 750

GREEN PIGMENT OF INTESTINAL WALL OF CHATOPTERUS. 467

millionths of a millimetre on the left (red end)—a division line being ruled at
every point corresponding to the position of a difference in wave length of
10 millionths of a millimetre. In this drawing the dispersion of the spectrum
as actually seen is represented, the intervals corresponding to 10 millionths
of a millimetre of wave-length becoming increasingly larger as we pass from
the red (wave lengths of 750—650 millionths) to the blue and violet (wave
lengths of 500 to 400 millionths). But in the charts given in the next two
plates, for which I am indebted to Professor Engelmann, the intervals
occupied by wave lengths differing by ten millionths of a millimetre are laid
down without reference to dispersion at equal distances from one
another. The charts in fact correspond to a pure spectrum, whilst the
drawing, fig. 5, represents the appearance given by a prism of small dis-
persion.

‘Vy

PLATE 35.

Fics. 6—8.— Representation of the colour of the neutral, acid, and alkaline
alcoholic solutions of Chetopterin, as seen by transmitted light.

Figs. 9—1]1.—Representation of the colour of the neutral, aeid, and
alkaline alcoholic solutions of Bonellin, as seen by transmitted light.

PLATE 36.

Charts prepared by Professor Engelmann showing the intensity of absorp-
tion in different parts of the spectrum of acidulated, alkaline, and neutral
alcoholic solutions of Chetopterin. The horizontal lines in the charts—
numbered in ten groups of ten—correspond to 100 units of light intensity.
The round dots indicate the successive parts of the spectrum observed and
measured. The position of the dot in vertical displacement records the per-
centage of light transmitted. Thus the highest horizontal = 100 per cent.,
the lowest 0 per cent. or complete absorption. The successive “dot points ”’
of observation are joined by oblique lines, giving thus a continuous but
irregular curve of absorption. The vertical lines as shown by lettering on
the chart correspond to millionths of a millimetre of wave length. The
position of the chief solar lines is also indicated by strong vertical lines on
the charts. The upper chart has the record of four distinct solutions. The
two records in dotted lines are those relating to experiments with an
acidulated solution—in the one case the light was passed through a thickness
of the solution amounting to 5 millimetres, in the other case only 2 millimetres
were used (of the same solution). [It is not possible in our present know-
edge of Chetopterin to say what percentage of pure Chetopterin was present
in the alcoholic solution.} The unbroken black lines are the records of
similar experiments with an alkaline alcoholic solution of Chetopterin. In

voL. 40, PART 3.—NEW SER. K K

468 E. RAY LANKESTER.

the lower charts two records are given of the absorption of two different
thicknesses (respectively 2 millimetres and 5 millimetres) of a carefully
neutralised alcoholic solution of Chetopterin. (For further details and the
comparison of the observed absorption of the smaller thickness of solution
with the theoretical value calculated from that given by the greater thickness,
the reader is referred to the text.)

PLATE 37.

Charts prepared by Dr. Engelmann of the observed absorption of the
spectrum by acidulated, alkaline, and neutralised alcoholic solutions of Bonellin.
Two thicknesses of each solution were made use of, and their absorption re-
corded. See explanation of Plate 36 and the fuller statements in the text.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 469

Materials for a Monograph of the Ascons.—I. On
the Origin and Growth of the Triradiate and
Quadriradiate Spicules in the Family Clath-
rinide.

By
E. A. Minchin, M.A.,
Fellow of Merton College, Oxford.

With Plates 38—42.

ConTENTs. PAGE
PREFACE - : : é : . 470
I. IntRopuctory : : : - 473
(a) Methods of Investigation : . ATA
(4) Anatomy and Histolozy of the Genus Clathrina Ree
(c) System and Nomenclature . 500

II. Descrirtrive—Observations upon the Hievelomment of the
Spicules : 502
(a) The Development of the Triradiate Systems , 504.

(4) The Origin of the Fourth or Gastral Ray of the Quadri-
radiate Spicules 517

(c) Some Observations on the Formation of the Monaxon
Spicules - 523
Appendix—(1) The Triradiates of Clathrina clathrus O25

(2) The Influence of the Nucleus on the Secretion
of Lime - 526
(3) The Granules of the Dermal Layer . 526

IiI. Histortcat—KFarlier Observations upon Calcareous and Rice
Spicules ‘. . 528
(a) Observations upon Spicule Formation : . 528
(1) Observations upon Siliceous Sponges : . 529
(2) Observations upon Calcareous Sponges . 532

(4) Observations upon the Structure and suapositien of the
Calcareous Spicules F 542

IV. Turoret1cat—On the Origin and Myalation of the Spicules o of
Calcareous Sponges 543

Appenpa.—A. On the Presence of an Axial Organic Filament in
the Spicule Rays 569

B. On Compound Spicular Systems behaving as Single
Crystals , 572
BIBLIOGRAPHY . - 579
DESCRIPTION OF THE PLATES : : ‘ . 582

voL. 40, PART 4,—NEW SER. Ek

4.70 E. A. MINCHIN.

PREFACE.

Tue Ascons are a group of calcareous sponges which pre-
sent many points of interest and importance to the zoologist.
As sponges they occupy the lowest systematic position, for
though the possession of a distinct skeleton is an advance
upon the condition found in the so-called Myxospongize—sup-
posing these not to be degenerate in this respect,—yet in
almost all other points of organisation the Ascons show a far
more generalised condition than is found in any other class of
sponges. Regarded also not merely as sponges, but from the
wider standpoint of the Metazoa generally, the Ascons claim
our attention as extremely primitive types, characterised by a
simplicity of structure scarcely equalled in any other form of
animal life above the Protozoa. Not only, therefore, are
many problems of sponge morphology and embryology to be
found reduced to their lowest terms in the Ascon type, but it
might be expected further that a careful study of the group
would shed much light upon biological questions of still wider
significance, since many facts of organisation and development
which, in other animals, lie deep and are difficuit to approach,
are in these forms almost upon the surface, as it were.

For these reasons I have devoted myself for some time past
to studies upon the structure, development, and classification
of the Ascons, in the hope that a complete knowledge, so far
as this is possible, of the group might form a material contri-
bution to zoology. It might, however, appear to some that
there would be but little of importance to discover in the
group which was not already known, seeing how voluminous
is the literature of the Ascons, and how many authors have
published studies and monographs upon them even in quite
recent times. So far is the subject from being exhausted,
however, that the most elementary facts in their organisation
still remain unknown or misrepresented, and it is uot too much
to say that the Ascons offer a field which, so far as results go,
is almost unexplored. If this statement seems to many
exaggerated, I can only hope to justify it by bringing forward

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 471

new facts and observations which will, I trust, bear me out in
what I have said.

Studies upon a group of animals naturally fall under four
main heads—systematic, structural, developmental, and phy-
siological, including under the last-named category rather
more than is meant by the term in our universities and medical
schools. The first of these lines of study is independent of
time and place, and can be conducted in any laboratory or
museum with material fresh or preserved; the second and
fourth can also be carried on at any time, given a good labora-
tory and aquarium and abundance of healthily living speci-
mens; the third alone, that is the embryology, must be studied
at the right season as well as in the right place, and the
observer must be prepared when the season comes to lay aside
other work for the time and devote himself to it. Thus,
although I have been occupied for some time past with the
histology of Ascons, the embryology has always been regarded
by me as having a first claim on my time, and in consequence
my histological studies have been disjointed, and are still very
far from completion. At present my embryological material
is nearly complete, and I hope before long to commence the
publication of studies upon it ; but as some time must elapse
before a large quantity of material, comprising the whole
developmental cycle of a number of species, can be worked
through, it seemed best to bring out at once an account of
some histological points in which my results approach to
finality. The formation of the spicules so characteristic of
these sponges has long engrossed my attention, and I trust
that the descriptions and figures to follow will be held to sub-
stantiate my claim to have made out the main points in the
-morphology of the triradiate and quadriradiate spicules in
the genera characterised by having the rays of the triradiate
systems meeting at equal angles,—that is to say, the genera
for which in my recent revision of the system I have used the
names Clathrina, Gray,and Ascandra, Haeckel.

Even with regard to the development of the spicules I am
well aware that I still have much todo. On the one hand, I

472 E. A. MINCHIN.

am not in a position as yet to give a detailed account either
of the monaxon spicules in Clathrina, or of the three- and
four-rayed spicules in Leucosolenia; on the other hand,
many minute details in the development of these spicules, such
as their exact relations to the secreting cells, and so forth—all
such details, in fact, as require thin sections for their elucida-
tion—remain for the present to be studied. But thin sections
of the developing spicules and their formative cells are, as a
matter of fact, very difficult to obtain. In the adult sponge
new spicules are formed amongst and between older ones,
which are often, and indeed usually, present in such masses
as to make it hopeless to obtain thin sections without previous
decalcification, which, by destroying the spicules, defeats the
purpose in view, while thick sections are scarcely more in-
structive for the minuter points than are surface views. In
the embryo, however, where stages can be obtained con-
taining only minute spicules, thin sections of material not
decalcified can be made without much difficulty. I have
preferred, therefore, to let the questions of minute detail
stand over for the present until I can return to them in the
course of my embryological studies, and rather to bring forward
my observations as they are than to wait until I can make them
exhaustive. Inthe present memoir I propose to deal with the
formation of the equiangular triradiates and quadriradiates
rather from the morphological than from the cytological point
of view,—that is to say, the number and arrangement of the
cells which build them up, and their origin within the sponge.
These are questions which for the most part can be studied
satisfactorily only in surface views.

Before, however, proceeding to describe the results of my
observations, I feel it my pleasant duty to say a few words of
thanks and acknowledgment to those who have assisted me in
my work and made it a possibility for me. In the first place
I must acknowledge my indebtedness to two “‘ pious founders,”
one of whom has been dead more than six hundred years,
while the other, I am glad to say, is alive and well. But for
my election to a fellowship on the Foundation of Walter de

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 473

Merton, it would, perhaps, have been impossible for me to
have engaged in these researches at all. To Professor de
Lacaze-Duthiers, on the other hand, founder of the marine
laboratories of Roscoff and Banyuls, I am under the greatest
obligation for his generous hospitality. On the frequent
occasions on which I have visited these stations I have been
made entirely free of all their resources, not only of laboratory
and aquarium, but even of lodging, and have been treated
with that uniform kindness and consideration which in France
is always shown, without regard to race or nation, to those
whom the fame of French scientific institutes attracts as
visitors and students. In this way I have had at my disposal
a rich and varied fauna, especially at Banyuls, the value of
which can scarcely be estimated.

I must further express my warm thanks to Professor Ray
Lankester for his kindness in giving me a place in his labora-
tory at Oxford, which has been, so to speak, my permanent
head-quarters, where the greater part of my studies have been
conducted. Iam also indebted to him for much help and kind
encouragement in his double capacity, both of Professor in
the University and Editor of this Journal. Finally, I have to
tender my thanks and gratitude to Professor R. Hertwig, of
Munich, for the kindness and hospitality which he extended
to me during a three months’ stay at Munich last year, making
me free of his laboratory and helping me in every way.

To each and all of those I have mentioned, whether of the
thirteenth or nineteenth centuries, I wish to acknowledge my
obligations and express my sincere thanks.

I. IntRopvuctToRrY.

Berore approaching the subject of spicule development a
few words must be said, first, as to the technique employed,
and secondly, upon the structure and histology of the sponge.
In order to understand properly the origin of the spicule-
forming cells, and to distinguish these cells from other tissue
elements, it is necessary to be acquainted with the forms and

4,74, E. A. MINCHIN.

characteristics of the cells composing the wall of the sponge.
I propose, therefore, to describe the histology in sufficient
detail for the aim in view, reserving fuller treatment of the
various issues for another paper.

(a) Methods of Investigation.—For studying the de-
velopment of the spicules, and the general arrangement and
characters of the cell elements compesing the body-wall of
an Ascon, no very elaborate histological technique is required ;
in fact, almost everything can be made out from surface views
of pieces of the body-wall fixed with osmic acid, stained with
picrocarmine, and laid out flat in glycerine. There is, how-
ever, one precaution, the importance of which cannot be too
strongly emphasised: the sponges should not be brought back
to the laboratory and there preserved, but should be fixed
quite fresh from their habitat immediately they are found,
care being taken to select fully expanded specimens.

My method of operating is to carry with me, when seeking for
the specimens, different tubes containing osmic acid solution,
distilled water, and picrocarmine, or other reagents as may
be required. I take 1 per cent. osmic solution, and dilute
it with an equal volume of sea water. The picrocarmine
employed is either Ranvier’s or Weigert’s, both obtained from
the well-known firm of Griibler at Leipzig. Ascons are very
easy to preserve, even in liquids of very feeble penetrating
powers, on account of the numerous cavities and channels,
resulting from their structure as systems of thin-walled tubes.
Hence quite large specimens can be fixed whole in osmic
without fear. After being five to ten minutes in osmic the
specimens are rinsed with water and placed in the picro-
carmine. In this stain they can be brought back to the
laboratory, but should not stay in it more than two hours; in
general one hour is sufficient. The specimens are then washed
with distilled water and transferred to glycerine or alcohol,
according as surface preparations or sections are required.
Asa rule I put one half of each specimen so preserved into
glycerine, the other half into alcohol.

This method preserves the cells, so far as shape, appearance,

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 475

and cytoplasm are concerned, most excellently, and has the
great advantage of not acting upon the spicules or corroding
them in any way ; though should it be required to remove the
spicules, the specimens can easily be decalcified by adding a
little hydrochloric acid either to the glycerine or to the alcohol,
without the stain being in the least affected thereby. The
nucleus is shown up very distinctly in the cells, and as a rule
nothing else but the nucleus is stained by the picrocarmine.
But in the case of granular nuclei the finer structure of the
framework and chromatin elements is not well shown by this
method, which gives an evenly stained nucleus, exhibiting
only traces of granulation; and hence for nuclear studies this
method requires to be supplemented by others. Vesicular
nuclei, on the other hand, are very well shown. The difficulty
with which we have to contend in Ascon histology is that all
methods which show good nuclear detail, require fixing fluids
which are more or less strongly acid, and such fluids act, of
course, upon the spicules. ‘The best preservative I have found
for the nuclear structures in these forms is Flemming’s fluid,
but its application causes the spicules to dissolve rapidly, with
abundant evolution of gas, as the result of which the tissues
may be mechanically injured. Moreover the collar-cells are
not nearly so well preserved by Flemming’s fluid as by the osmic-
picrocarmine method, at least as regards collar and flagellum.
Absolute alcohol is a preserving agent much employed for
sponges, but I have not found it very good for Ascons. The
general structure of the cells is not so well preserved by the
method already mentioned, while the nuclei often have their
structure altered and deformed by it. It has, however, the
advantage of preserving the spicules.

Since I propose in the present memoir to discuss only the
general histological structure as seen in preparations made by
the osmic-picrocarmine method, and to leave the discussion of
finer cytological details for a subsequent paper, I will not con-
sider further here the action of the various fixatives and stains.

Sponges preserved with osmic and picrocarmine are often
more or less macerated, especially if the osmic has been

4.76 E. A. MINCHIN.

allowed to act too long. The collar-cells become detached
from their position, and are found loose in the tubes. When
this process has gone too far the elements of the dermal layer
may become displaced also. For surface views slight macera-
tion is not always harmful, since it makes it easier to remove
the collar-cells, but it is certainly to be avoided in material
for sections. Hence care should be taken not to let the osmic
acid act any longer than is absolutely necessary for the fixa-
tion of the cells. For studying the dermal layer in surface
views it is generally desirable to remove the collar-cells as far
as possible. This can be effected by brushing the gastral
surface of the body-wall gently with a fine paint-brush when
the specimens are in glycerine. In the species without quad-
riradiate spicules the collar-cells can be removed easily in this
way without damaging any other elements. In the species
with gastral rays to the spicules it is not so easy to clear away
all the collar-cells, and the cells on the gastral rays are also
brushed off in the process.

As a medium for mounting the pieces of the body-wall for
surface views glycerine is greatly to be preferred to Canada
balsam ; cells can be focussed much more clearly at different
depths, and the refraction of the spicules interferes less with
the distinctness of the vision. Glycerine has, however, the
disadvantage that it acts slowly upon the spicules. After a
time they become corroded, and finally dissolved. The time
this takes varies greatly in different preparations; I have some
two years old which are only slightly corroded, while others
go quicker. As a rule the corrosion does not become marked
for at least six months, so that plenty of time is allowed for
studying the preparations, though it should not be deferred
too long. Asa matter of fact this corrosion takes place also,
as a rule, in preparations mounted in Canada balsam, though
much more slowly. It is very little use, I have found, trying
to isolate the elements of the dermal layer. Good pores and
epithelial cells may sometimes be obtained in this way, but the
spicule cells almost invariably become separated from their
spicules.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 477

(6) Anatomy and Histology of the Genus Clath-
rina.—The young sponge, as is well known, has the form
for which Haeckel proposed the term Olynthus, a form shaped
like an open vase or sac, fixed at the blind extremity, and
opening at the other by a large aperture termed the osculum.
The wall of this organism is pierced by numerous pores, by
which water can pass from the exterior into the large central
space, or gastral cavity so-called. The inner and outer sur-
faces of the body-wall may therefore be conveniently distin-
guished as the gastral and dermal surfaces respectively.

The body-wall is composed of two layers, an outer dermal
layer and an inner gastral layer. The former makes up the
greater part of the wall, consisting mainly of a structureless
jelly; it contains the skeleton, the pores, and the nutritive
wandering cells, and is covered at all surfaces where it is
exposed by a flat contractile epithelium. The gastral layer is
uniform in composition, and consists of the characteristic
collar-cells ; it lines the interior of the so-called gastral cavity,
but, besides being interrupted at the pores, does not reach the
extreme margin of the osculum. There is thus formed a rim
or collar surrounding the oscular opening, composed of the
dermal layer alone, and covered on both faces by the flat con-
tractile epithelium. For this region I propose the distinctive
term “ oscular rim.”

The olynthus is only a transitory stage in the life-history of
a calcareous sponge, and in the Ascons, in which the gastral
layer remains continuous and not restricted to “ ciliated
chambers,” the complicated form by which the full-grown
specimens are characterised is attained by growth at two
points. The olynthus increases in length and becomes tu-
bular by growth at the oscular rim, and at the same time the
surface of the wall is increased by the formation of blind out-
growths or diverticula from the sides of the tube. These
diverticula continue to grow in length, and repeatedly branch
and anastomose until a dense network of tubes, from which
new oscula may arise, is formed surrounding the original
osculum of the olynthus, which in its turn may have multi-

478 E. A. MINCHIN.

plied by constriction and subsequent fission of the tube in a
longitudinal direction. Thus arises the sponge form typical
of the Ascons, consisting of a dense network of hollow tubes
converging towards, and opening by, one or more oscula; and
we obtain either the large erect oscular tubes with com-
paratively small basal network characteristic of the genus
Leucosolenia, or the short and often insignificant oscula,
acting as vents for a greatly developed network of tubes, cha-
racteristic of the genus Clathrina, according as growth pre-
ponderates at the oscular rim, or at the ends of the diverticula.

Such being the general structure of the sponge, we may now
consider in more detail the composition of the body-wall, and
more especially of the dermal layer. The gastral layer is
uniform throughout, and consists only of collar-cells. For the
purposes of the present investigation it is not necessary to
discuss the details of the structure of the collar-cells, since
they take no share whatever in the formation of the spicules.
Only one point may be mentioned, sufficiently obvious when
attention is called to it, but which, if passed over in silence,
might lead to the impression that my figures or preparations
were at fault. The wall of the sponge is the wall of a hollow
cylinder, and is possessed of a certain thickness; hence the
gastral layer, which occupies a more internal position, has a
less extended surface than the more externally placed dermal
layer. As a consequence when the wall is laid out flat the
gastral surface is stretched, and the collar-cells tend to separate
slightly from one another here and there, producing cracks
and spaces, as can be seen in PI. 41, fig. 39, and PI. 42, fig. 50.
These spaces are not natural, but are the inevitable result of
flattening out a curved surface. In their natural position the
collar-cells are in close contact, their cell limits forming in
surface view a network of polygonal areas.

The dermal layer in the genus Clathrina is sharply
differentiated into an external contractile or neuro-muscular
layer, and a more internal connective-tissue layer, scattered
about in which are the pore-cells and the wandering cells.
The connective-tissue layer consists of spicules and their

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 4.79

formative cells, suspended in a structureless jelly, which
makes up the greater part of the body-wall. Before we can
discuss the relations of the connective-tissue layer and the
origin of its cells it is necessary to be acquainted with the
other elements of the dermal layer, namely, the neuro-muscular
layer, the pores, and the wandering cells, in at least sufficient
detail to enable us to distinguish them from one another, and
so to trace their history.

The Dermal Epithelium (Neuro-muscular Layer;
“Ectoderm” of Authors).—In the first place let me repeat
the statement that the dermal layer, wherever exposed, is
covered by flattened epithelium, at least in the normal and
expanded condition of the sponge. I wish to lay especial stress
upon this, even at the risk of offending by repetition, since
reference has been made to me as one of those who deny the
existence of the flat epithelium, although, as a matter of fact,
I have never done so, but, on the contrary, have repeatedly
affirmed its presence. Thus in 1892 (2, p. 490, foot-note),
when criticising Bidder’s statement that the external epithe-
lium—or ectoderm, as we all called it then—consisted in Ascons
of mushroom-shaped cells, I stated such cells to be of rare
occurrence in freshly preserved material of Clathrina cla-
thrus, ‘ the predominant form of the ectoderm being flattened
epithelium.’ Again, in the same year (3, p. 181) I began
my description of the “ectoderm” of the same species with
the following statement :—“This layer, the contractile layer
of the sponge, consists in the expanded state of flattened non-
ciliated cells.” In the face of these very definite statements I
am unable to understand how Lendenfeld is able to say (1894.
[1], p- 161), ‘ Neuerlich haben. . . Bidder und Minchin die
Behauptung aufgestellt, ... dass die Spongien tiberhaupt kein
Plattenepithel besassen.” And again in the same year (1894
[2], p. 508) he refers to me as one who, in company with
Bidder, has opposed Schulze’s views as to the nature of the
flat epithelium. It is evident that von Lendenfeld has taken
very little trouble to understand the views that he criticises.
To clear up any misconceptions which this careless reviewer

480 KE. A. MINCHIN.

may have been the means of spreading, I think it necessary to
reaffirm the views which I have held ever since I began to
study the histology of the Ascons, and which I still hold more
strongly than ever; namely, that in the normal expanded
condition of the sponge the dermal layer is limited at its free
surfaces by a flat epithelium, of the usual type of epithelia to
which this term is applied; that this flat epithelium, in the
family Clathrinide at least, is the contractile or neuro-
muscular layer of the sponge; and that when completely
contracted the epithelial cells alter in form, becoming mush-
room-shaped every where,—that is to say, except in the interior
of the oscular rim.

The cells of the dermal flat epithelium have a very charac-
teristic appearance and structure in all the species. Seen in
surface view (PI. 38, figs. 1, 3, and 10; Pl. 39, figs. 16 and 20;
Pl. 41, fig. 40, &c.), the nuclei are seen scattered about, each
in the centre of a patch of very granular protoplasm, in
which the nucleus is sometimes almost hidden. Cell outlines
can often be made out as very delicate bright lines dividing
the surface into polygonal areas, but it is very difficult
to discern them as a rule, on account of the refraction of
the underlying spicules. The nucleus of the epithelial cell
is spherical and relatively large, usually causing the surface
to bulge out, as seen in sections (PI. 41, fig. 41). It con-
tains a fine and more or less even reticular framework, with
chromatin at the nodes, and usually a small nucleolus placed
somewhat excentrically. Stained with picrocarmine after
fixation with osmic acid, it appears evenly stained or finely
granular, with a darker spot representing the nucleolus.
Its absolute size varies in different species, as is the case with
the nuclei of all the other tissue elements; Clathrina con-
torta is especially remarkable for the large size of its cells,
and the nuclei of its flat epithelium are nearly twice as large
as those of coriacea, as may be seen by comparing Pl. 41,
fig. 40, and Pl. 39, fig. 16. The relative size of the nuclei of
the various tissue elements in each species is a much more
constant character, and especially the proportion between the

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 481

nuclei of the flat epithelium and those of the collar-cells ;
speaking generally, the diameter of the latter is about four
fifths of that of the former (compare Pl. 41, figs. 89 and 40).

The cytoplasm of the epithelial cells is packed with granules
of a very characteristic kind, which easily distinguish the flat-
tened epithelium and its derivatives, the pore-cells and the
spicule-cells, from either the collar-cells or the wandering cells.
As a convenient type in which to study the granulation of the
epithelial cells, we may take Clathrina coriacea (figs. 1—
16). Seen in osmic-picrocarmine-glycerine preparations these
granules appear of rounded but rather irregular outline, and
vary both in size and depth of colour. Some are pale, while
others appear darker and even quite black. If one of the
paler granules be carefully focussed, it appears to have a dark
border surrounding a clear ‘central spot. On slightly raising
the focus the clear central spot disappears, and at a still
higher focus a dark central spot is seen surrounded by a clear
area. The darker granules, of which there are usually several
in each cell, do not show these changes distinctly, but appear
simply black. In this species the granules have, in prepara-
tions of the kind mentioned, a yellowish or yellowish-brown
shimmer, and impart a similar tint to the protoplasm of the
whole cell. In general the granules appear opaque and dul],—
in short, they may be described as having an almost chitinous
appearance. In life they are the elements to which the sponge
owes its colour, whatever it may be. If the sponge is white,
which is more usual, the granules appear black in transmitted
light, dull white in reflected light. In the numerous colour
varieties of C. coriacea the granules appear red, orange,
yellow, lilac, or whatever may be the colour of the sponge, in
reflected light; and in Clathrina clathrus they similarly
have a constant lemon-yellow colour.

The distribution of the granules varies slightly with the
condition of the cell, though they are always more or less con-
centrated round the nucleus. When the cell is fully expanded
the nucleus is superficial, with very few granules or none at
all over it, that is to say external to it, and the granules are

482 E. A. MINCHIN.

spread out in the cell, extending in irregular groups and rows
up to its extreme limits. When the cell is contracted, even
slightly, the nucleus lies much more internally, and is covered,
or even hidden, by the granules which are aggregated in the
centre of the cell. In the latter condition the epithelium as a
whole presents in surface views the appearance of compact
granular masses of protoplasm, each with a nucleus situated
more deeply, scattered about at considerable intervals one
from another.

The protoplasm of the epithelial cells, apart from the granula-
tions already described, is clear and very difficult to make out
in surface views when the cell is very expanded, except just
round the nucleus. It appears to have a distinct alveolar or
reticular structure. The characteristic yellowish tint of the
cells is similarly most distinct when they are more or less
contracted and their protoplasm concentrated; when they are
expanded and spread out they appear simply greyish.

The appearance of the epithelial cells and the relative
quantity of their granules vary slightly in the different species
which I have been able to examine, though in all the general
characteristics are such as have been described above.

Clathrina coriacea is a species remarkable for having
its dermal epithelium very granular, and the leathery appear-
ance from which it derives its name is due partly to this cir-
cumstance, partly to the very contracted condition in which it
is usually found when exposed at low tide, the condition in
which it usually figures in collections. In contorta (Pl. 41,
fig. 40) the granules are relatively smaller and fewer, and tend
to be oval or slightly elongated in form, with occasionally a
larger, more rounded one amongst them. The number of the
larger granules varies ; usually there are two or three in each
cell. The yellowish-brown tint is very distinct in this species.
In blanca and clathrus the granules are smaller and the
cells appear greyish, the yellowish tint being scarcely notice-
able except when contracted. C. cerebrum has epithelial
cells rather different in appearance from those hitherto de-
scribed. In this form the protoplasm is remarkably vacuo-

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 483

lated, and the granules, which are small, are lodged at the
nodes, giving the cells a characteristic marbled appearance,
with a yellowish-brown tint, rather more brownish than in the
other species. The epithelial cells of reticulum are very
similar to those of cerebrum, except that the granules are
larger and the vacuolation less marked, giving a more normal
appearance.

Finally I have examined at Banyuls a species apparently
new, with all three kinds of spicules, triradiate, quadriradiate,
and monaxon. I propose to refer to this species as Clathrina,
sp. dub., in the present memoir. The epithelial cells of this
species are very similar in their characters to those of
coriacea, but rather less granular (PI. 39, figs. 17—20).

Thus the dermal epithelium shows not inconsiderable varia-
tions in its general appearance in different species, but in all
the forms its main characteristics are similar—flattened
granular cells of a type easily recognised. One small point
remains for notice, as it may at first sight seem strange in the
figures. The flat epithelium forms the external covering of
the dermal layer, and the spicules are placed of course inter-
nally to it, lodged in the structureless jelly which makes up
the greater part of the dermal layer. In some species, where
the wall contains a great number of large spicules, those
placed more externally often cause the wall to bulge out to
such an extent that the spicules sometimes become almost
enveloped in the epithelium, which is seen to be tucked in
under them. Thus in Pl. 39, fig. 19, a surface view of Clath-
rina, sp. dub., from the inner or gastral side, the cells of
flat epithelium often appear on the upper side of the spicules
placed most deeply, although, being external, they should
really appear below all the spicules. Other epithelial cells
are seen here and there in side view. Similarly in fig. 20,
seen from the dermal aspect, some of the epithelial cells
appear almost hidden under the spicules. These appearances
are not difficult to understand, and need not detain us further.

The Pores.—The true nature of the pores was first
described by Bidder (1891, p. 631) and myself (1892, p. 266,

484. ‘EL A. MINCHIN.

fig. 21) independently as consisting each of a single perforated
cell, placing the internal gastral cavity in communication with
the outer world through the body-wall. We differed, how-
ever, in our views as to the origin of these pore-cells, Bidder
regarding them as derived from metamorphosed “ endodermal ”
collar-cells, while I described them as arising from immigrated
cells of the “ectodermal” flat epithelium. More recently
Dendy (1893, p. 214) has thrown doubt on these statements,
and is inclined to regard the pores as “ inter-cellular and
not intra-cellular in nature,” ‘simply gaps between cells.”
It is evident, therefore, that Dendy has never studied the
pores in surface views of material suitably preserved and
mounted, for a single glance at such a preparation would be
sufficient, I think, to dispel his doubts for ever. On the other
hand, so long as the matter is only studied in series of sec-
tions, chopped up in paraffin from material thrown into alcohol
and stained in borax carmine, it will be possible to doubt
anything.

No fact in the histology of Ascons is more easily demon-
strable, even to the tyro in these matters, than that each pore
is, in the expanded condition, a single perforated granular
cell, of a peculiar and easily recognisable type. Compare
especially figs. 10, Pl. 38, and 39, Pl. 41; also figs. 1, Pl. 38,
18 and 18a, 19 and 19a, and 20, Pl. 39, and figs. 49, 50, and
52, Pl. 42. Seen in surface view the pore has two openings, a
smaller external or dermal aperture (derm. ap.) on a level with
the flat epithelium, and a larger internal or gastral aperture
(gastr. ap.) on a level with the collar-cells. The dermal open-
ing perforates a thin membrane which is usually more or less
free from granules, and stretches like a tympanum across the
cavity of the pore-cell. The gastral opening is surrounded
by a thick and granular wall, smooth on the inner side, but
running out into points and processes on the outer side—a con-
figuration due, as can be seen at once from fig. 39 on PI. 41,
to the pressure of the adjacent collar-cells. The projecting
points fit into the interstices between the collar-cells, each
depression between two projections being the impression made

MATERIALS FOR A MONOGRAPH OF THE ASCONS, 485

by the body of one such cell. At one side of the pore is
lodged the nucleus in a thickening of the wall of the pore.
Sometimes the protoplasm surrounding the nucleus is clear,
but more often it is so full of granules as to obscure the view
of the nucleus (figs. 18, 19, 39). In its position the nucleus
is usually situated rather towards the gastral aperture, nearly
on a level with the collar-cells.

The minuter characters of the pore-cells are such as to
leave but little doubt, even if there were no further evidence
obtainable, as to their histological affinities. In all respects
the pore-cells resemble the cells of the flat epithelium, but
with the characters of the latter exaggerated. The nucleus
is similar in structure, but slightly larger, and usually paler in
its staining reactions (osmic and picrocarmine). The granules
are identical in character with those of the epithelium, but
are present in much greater number—the cell, as a whole,
being much bigger,—and attain, both on the average and
in individual instances, a much greater size. This is less
noticeable in coriacea, with its very granular epithelium,
than in such a species as contorta, where the epithelial
cells have fewer and smaller granules (compare Pl. 41,
figs. 389 and 40). In consequence of the great increase
in the size and number of the granules, the yellowish-
brown colour which they impart to the cells in osmic-picro-
carmine-glycerine preparations is very distinct, and picks
them out at once from the rest of the tissue elements, even
in those species in which this tint is scarcely discernible in
the epithelial cells. In life also the granules of the pore-cells
show the peculiarities described above for the epithelial cells,
being dull white when the sponge is white, and of the same
colour as the sponge when it is coloured. In short, apart
from the especial functions and consequent peculiar form of
the pore-cells, or porocytes, as they may be termed gene-
rally, we may say that while, on the one hand, all their
characters show them to belong to the same category as the
epithelial cells, on the other hand, those characters are modi-
fied to an extent and with a constancy sufficient to enable us

VOL. 40, PART 4,.—NEW SER. M M

486 E. A. MINCHIN.

to regard the pore-cells as constituting a distinct class of cell-
elements.

If the pore-cells always remained expanded, it is inconceiv-
able that any misconceptions should have arisen as to their
true nature. But unfortunately, perhaps, for the student of
sponge literature, the pore-cells, like the epithelium from which
they originally sprang, are very contractile. On the least
provocation the pores close up, and then present an appear-
ance which has led to much confusion. The external aper-
ture seems to be the part first affected, the opening disappear-
ing by contraction of the delicate membrane in which it is
situated, like the closing of an iris diaphragm. Hence pore-
cells occur commonly which show a widely open gastral
aperture, but no trace of the dermal opening (PI. 42, fig. 50).
Next, the gastral opening narrows itself, and finally closes
up, and the result is a large compact granular cell, of a
type very familiar to all students of Ascon histology, since in
all specimens not completely expanded these cells are the
largest and most conspicuous elements in the sponge. It is, in
fact, far easier to obtain preparations showing the pores closed
than to obtain them with the pores open. In order to show
the open pores, as in the preparations represented in Pl. 38,
fig. 10, and Pl. 41, fig. 39, it is necessary to preserve fully
expanded specimens of the sponge as soon as they are found.
If, on the other hand, the sponge be brought back to the
laboratory and there preserved, every pore will be found, as a
rule, completely closed, and no longer recognisable as a pore
except by a comparison with preparations showing pore-cells
in the expanded state. Pl. 41, fig. 38, though showing pore-
cells which have not as yet acquired an opening, may be taken
as representing the characteristic appearance of the closed
pores, and fig. 39 also shows one such cell. It is hardly
necessary to state that in all their cytological characters these
cells agree with what has been said above for the pore-cells,
and differ from them only in form; their compactness makes
their yellowish-brown tint even more obvious than is the case
when expanded.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 487

The closed pore-cells are a very obvious feature of contracted
Ascons, and with increased contraction of the sponge the pore-
cells go through a remarkable series of movements. Situated
at first in the dermal layer on a level with the inner ends of
the collar-cells, they push their way inwards between the latter.
If the contraction goes still further, as it does commonly in
the species without gastral rays, the collar-cells are forced one
over the other, and form a layer two or three cells deep;
during these changes, or before, the porocytes migrate com-
pletely inwards, and form a granular epithelium covering over
the collar-cells, and lining the gastral cavity. Finally in the
extreme state of contraction the tubes become solid, filled
completely by the collar-cells, in the centre of which the
pore-cells are to be found forming, as it were, a granular axis
to the tube. As the sponge expands again all these move-
ments are gone through in reversed order.

The contracted porocytes have attracted considerable atten-
tion, as might have been expected, from those who have
studied calcareous sponges, and unfortunately they have also
received very different interpretations. They seem to have been
first clearly described by Metschnikoff (1879) in Clathrina
clathrus and primordialis in specimens which, as is
evident from the figures and descriptions given of the
dermal epithelium, were in a state of complete contraction.
Metschnikoff described with his usual accuracy the appear-
ance of these cells, chiefly as seen in the living condition ;
their large size and compact form, their densely granular pro-
toplasm, and the colour of their granules, which were yellow
in clathrus, dark brown by transmitted light in primor-
dialis, like the perfectly similar granules in the “ ectoderm”
-(l. c. pp. 360, 361, pl. xxii, figs. 1, 2,5, and8). Nevertheless
Metschnikoff regarded them as mesoderm cells, which is the
more remarkable since he represented them quite correctly
lying between, not under, the collar-cells (I. c. fig. 2). He
believed them to give rise to the triradiate spicules, or rather
he confused the cell masses which form the spicules with these
cells, which, as we shall see, was an error.

488 E, A. MINCHIN.

Since Metschnikoff’s description of these cells they do not
seem to have been much noticed until comparatively recent
times. In the eighties we can refer only to Carter, who
observed them in coriacea (1884, p. 20, and pp. 21, 22), and
described their granules as “ spherical, translucent, and glairy,
glistening from refraction of light, of a faint yellow tinge,”’
appearing, ‘‘ when in situ among the spicules and spongozoa,
to be loosely grouped round a delicate nucleated cell respec-
tively, the ‘ Kern’ of Haeckel.” This description betrays a
strange confusion between cell and nucleus, as well as a com-
plete misapprehension on the part of the author as to the
meaning of the word ‘ Kern.” The author gives some
microchemical reactions observed, and mentions that in
the colour varieties of the sponges they are the seat of the
colours ; the possibility is suggested “that they grow into
the larger cells of the protoplasm (the ‘ Kerne’) from which
they appear to be derived, when they may fulfil other offices.”
It is evident that the author’s description includes dermal epi-
thelium as well as porocytes. Dendy figured and described
porocytes in “ Leucosolenia” (Clathrina) cavata (1891
[2], pp. 18, 19, and pl. vi, figs. 4 and 5), and considered them,
as well as those seen by Carter, as symbiotic alge. Bidder (1891)
was the first to recognise their true nature as closed pores in
clathrus and primordialis, though I cannot follow him in
the account he gives of their origin as “metamorphosed collar-
cells,” and am doubtful about the excretory function which he
ascribed to them, a view which he developed further in later
contributions (1892, &c.). Bidder proposed for these elements
the technical term “ Metschnikoff’s cells,” and pointed out
correctly that their granules only differed from those in the
cells of the dermal epithelium in being of larger size.

In the same year Lendenfeld figured and described these
cells in “Ascetta” spinosa (1891, fig. 22, p. 205), and
apparently also in primordialis (lL. c¢., fig. 23, pp. 201, 202)
and clathrus (p. 212). In the case of spinosa he inter-
preted these cells as “parasites or symbiontes of vegetable
nature,” as Dendy had done before him. In the case of pri-

MATERIALS FOR A MONOGRAPH OF THE ASCONS, 489

mordialis, Lendenfeld took the singular view that the cells in
question were the mother-cells of the choanocytes, apparently
for the sole reason that they were continuous with some sort of
coagulum which appeared in his preparations between the
collar-cells. Indeed, although I formerly believed that von
Lendenfeld’s supposed ‘ Kragenmutterzellen ” were in reality
the closed pores, I should be more inclined now to regard them
simply as parts of the coagulum which he figures, had he not
more recently (1894 [2], p. 508) asserted definitely that the
cells which Bidder and myself identify as closed pores are the
same as those which he regards as the mother-cells of the
choanocytes. In that case the statement which he twice
repeats (1891, p. 201, and 1894 [2], p. 508), that the proto-
plasm of these cells agrees in its characters with that of the
collar-cells, is absolutely astounding, even to those who are
acquainted with his writings on sponge histology; for if there
are in the sponge body, cells which differ more than others, in
every character in which cells can differ, from the collar-cells,
it is the porocytes (compare my Pl. 41, fig. 39). Looking
again at his figures, I am inclined to think that in spinosa
(l. c., fig. 22, a) he has represented true porocytes, which he
identifies as symbiotic vegetable bodies, but that the structures
which in primordialis he has figured as ‘ Kragenmutter-
zellen”’ (1. c., fig. 23) are simply broken-down cell remnants
in a very badly preserved preparation. The question is, in fact,
impossible to decide, for it is enough to glance at the figures
to which reference has been made to see that they are utterly
untrue to nature, and represent, if anything, the cell elements
in an advanced stage of putrescence and disintegration. It is
nothing short of ludicrous that such figures, and others in the
same memoir (figs. 34—37, for instance, showing myriads of
absolutely non-existent connective-tissue cells) should ever
have appeared in a serious scientific journal as representing
the histology of these sponges ; ‘‘histology” of this kind would
certainly not pass muster in any other group of animals. I
think, therefore, that it would be an entire waste of time to
attempt to discuss further what von Lendenfeld may or may

490 E. A. MINCHIN.

not have seen, or what his views may have been at one time
or another, as to the nature of these or other cell elements.
The last author who, so far as I am aware, has described
the porocytes is Topsent (1892). In the same year (1) I had
published some observations far from complete upon the
histology of “ Leucosolenia” coriacea, amongst which I
had figured and described the contracted porocytes in sec-
tion, but had wrongly interpreted them as the amcebocytes or
“‘cellules digestives pigmentées ” of Topsent. In criticising
my observations Topsent rightly pointed out my error, and
gave a good and accurate description of the porocytes, but
without himself arriving at a true understanding of their
nature. He terms them “ cellules sphéruleuses du méso-
derme,” in distinction to the ameebocytes or “ cellules granu-
leuses du mésoderme,” and considers that they represent
reserves of nutriment. He was guided to this interpretation
from the reactions of their granules. It is by no means im-
possible that the cells in question should combine the functions
of storing reserve nutriment with that of acting as pores, but
I certainly cannot agree with Topsent in regarding this as
their sole function. I feel quite convinced that if he had
compared them in the expanded state with the contracted con-
dition in which he figures them, he would have agreed with me
in my interpretation of them as closed pore-cells simply ; for
that his drawings do represent contracted material is shown
by internal evidence. The appearance of the “ ectoderme ” in
his figs. a and B as compact masses of granules at considerable
distances apart is exactly the appearance it shows when con-
tracted. Moreover in the surface view A not a single pore is
shown; but a piece of this size, if expanded, would contain at
least half a dozen pores (compare Pl. 38, fig. 10). Topsent
remarks, in his criticism on my work, that I had only been
moderately struck (“médiocrement frappé”’) by these very
conspicuous cells ; but that is simply because my material was
all preserved immediately when found, and as a result the
pores were found expanded, and described by me as such,
except in one specimen which was contracted when found, as

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 491

of course often occurs, and in which, after cutting sections, I
saw and figured the contracted pore-cells, though without
myself recognising their true nature. Had Topsent made and
studied similar preparations he would have been rather sur-
prised, I think, to have found his “cellules sphéruleuses ”
either completely absent or comparatively rare; here and there
of course one finds a contracted pore in a specimen which,
as a whole, is expanded, and one generally also finds, without
difficulty, every transition between expansion and contraction.
It is possible that then Topsent would have been just as
“moderately struck” by these cells as I was myself.

I have gone into the question of these porocytes rather
fully ; but for this the importance of these porocytes, as we
shall see, from the point of view of spicule formation, as well
as the great confusion and diversity of opinion in the litera-
ture, must be my excuse. It is, indeed, a tangled web to
unwind when we try to introduce harmony into the existing
descriptions of these structures, nor can we expect that it should
be otherwise until spongiologists have more generally recog-
nised the excessive contractility of the species of Clathrina,
and the ease and rapidity with which they alter in appearance.

Looking generally at the matter, we see that contracted
pore-cells, however interpreted, have been seen in a consider-
able number of-species, more particularly in those which are
characterised by a very granular dermal layer, or, like
clathrus, by the granules being coloured. Thus Metschnikoff
saw them in clathrus and primordialis, but overlooked
them in blanca, where they are much less conspicuous.
Primordialis is a species which resembles coriacea in
being very granular; and in the latter Carter, Topsent, and
myself noticed these cells. Dendy observed them in cavata,
a species which his figures show to possess a very granular
dermal layer, and the only species in which Lendenfeld has
seen them beyond doubt is spinosa, a species which I have
myself had opportunity of studying at Banyals, and which
very closely resembles contorta, if it is not identical with it.
The only distinctive feature of contorta is the possession of

492 E. A. MINCHIN.

monaxon spicules, and these, when present, vary very much in
size and number. To Bidder, however, belongs the credit of
having been the first to recognise the true nature of these cells
as contracted pores. In addition to the species already enume-
rated, I can vouch for their occurrence in blanca, cerebrum,
reticulum, contorta, and “Clathrina, sp. dub.,” and
since in all these species the pores have similar characteristics
and contract in the same way, I have but little doubt these
usually very conspicuous cells—~ Kornerzellen,” “ cellules
sphéruleuses,” “yellow granules,’’ as they have been so variously
termed—will be found in any species of Clathrina where they
are sought for under suitable conditions.

Origin of the Porocytes.—We have now considered in
some detail the structure and appearance of the pore-cells, both
when expanded and functioning as pores, and in the contracted
condition in which they most often figure in the literature.
The porocytes do not, however, form a separate layer, growing
and multiplying amongst themselves, but each porocyte arises
separately and independently from the dermal flat epithelium. I
have already described in a former communication (1892 [3]) the
more usual method by which the pores arise, either at any point
during the general growth and increase in size of the tubes or
at the ends of the blind diverticula, namely, by immigration of
a cell of the flat epithelium. Without going fully into the
matter at present, I will merely refer to Pl. 39, figs. 15 and
16, por. c., which represent two of these future porocytes,
in process of immigration. It will be seen that the cells in
question only differ from the rest of the flat epithelium in
their larger size and more numerous granules, and in the
possession of rather larger nuclei. Each cell shows a portion
still on the surface, and therefore visible at the higher focus,
which has the granules spread out, and is without definite
limits, and a deeper portion, visible at a lower focus, which
has already come into contact with the collar-cells, and in
consequence has a sharp outline showing the characteristic
bays and points. My reason for figuring these two cells is
that they show a feature often to be observed, namely, the

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 493

presence of vacuoles, containing bodies which, to judge from
their appearance, consist of calcareous matter. The near
proximity to one of them (fig. 15) of what seems to be a much
corroded fragment of a spicule, has led me to the belief that
these porocytes may exercise the additional function of re-
moving, or rather absorbing, pieces of broken spicules. In
so far as they perform this office they might be termed
“ scleroclasts.” I make this suggestion with all caution, but
there seems to me no inherent improbability in the porocytes,
especially at an early stage in their differentiation, exerting a
scleroclastic function.

A very important growing point in the sponge, as has been
said, is the oscular rim; and here, too, porocytes are formed in
great numbers, and in a manner differing in details from that
which I have mentioned above as the typical method. The
oscular rim is, in fact, a region where the origin of porocytes is
so easy to study that it is a marvel to me that it should not have
been described long ago. For reasons which will be apparent
when we come to study the formation of the fourth ray of the
quadriradiates I propose to go into the matter rather fully,
and as a type I will take contorta, where the facts are very
plain.

If a piece of the oscular rim of contorta be laid out flat
and examined from the dermal surface, flat epithelium of the
normal type is seen (PI. 41, fig. 40). If, on the other hand, the
oscular rim be examined from the gastral view, especially
rather near the limit of the collar-cells, an epithelium is seen
which at first sight appears very different (Pl. 41, fig. 38). The
cells are packed with granules which obscure and often hide
the large pale nucleus, and the whole cell has a most pro-
nounced yellowish-brown tint. In fact, after what we have
already seen we can identify the cells without hesitation as
porocytes of a most typicalkind. PI. 41, fig. 41, shows the two
types of epithelium in a decalcified section. On the lower
(outer) side we see the typical dermal epithelium, and on the
upper (inner) side we see the layer of granular porocytic
epithelium, for so we may term it without further preamble,

494, E. A. MINCHIN.

Had the drawing (fig. 41) been continued as far again to the
right, collar-cells would have appeared in the section.

If now we follow up the layer of granular epithelium lining
the oscular rim in the direction of the margin of the osculum,
it is found to lose its distinctive characters by degrees, he-
coming more and more similar to the epithelium of the dermal
surface. Pl. 41, fig. 42, shows an epithelial cell from about
halfway up the oscular rim, obviously intermediate in its
characters between the cells figured in figs. 38 and 40 respec-
tively. Finally, near the actual margin of the osculum the
epithelium is quite of the ordinary type, and is continuous
where it turns the edge with the general flat epithelium
covering the exterior. There is, in fact, a gradual transforma-
tion in the oscular rim of ordinary cells of the flat epithelium
into porocytes, the changes consisting of (1) increase in size
of the cells; (2) increase in the number and size of the
granules; (3) slight increase in the size of the nucleus, and
corresponding decrease in its staining powers.

If, on the other hand, we follow the porocytic epithelium
downwards instead of upwards, we find as we approach the
limits of the collar-cell layer (Pl. 41, fig. 38) the porocytes be-
coming more compact and definite in their outlines, and hence
less like epithelial cells. Lower still (figs. 38 and 39) we find
them lying between the collar-cells, and there becoming
gradually transformed into the familiar pores. In other
words, the collar-cell layer continually extends its upper limit
by proliferation of its cells, and as it does so the porocytes
become surrounded by and enclosed amongst the collar-cells.
As soon as this occurs they spread out greatly and form a large
cavity, open towards the gastral space, and by perforation of
the outer wall of this cavity they acquire a dermal aperture.
In fact, they go through just the changes which a contracted
pore cell goes through when it becomes expanded.

In all the Ascous of the genus Clathrina studied by me,
and apparently also in the species of Leucosolenia, cer-
tainly in Ascandra falcata, the same state of things is
to be found. I found the oscular rim in the species blanca,

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 495

coriacea, clathrus, primordialis, cerebrum, reti-
culum, lacunosa, spinosa, and contorta, besides ““Clath-
rina, sp. dub.” Yet Lendenfeld, after publishing a mono-
graph of the Calearea, not only had not seen this very obvious
structure, which Lieberkiihn described with great clearness in
1865 (see below), but he even characterised my temerity in
venturing to point out its existence as showing a total lack of
“ Disciplin.” Iam not quite sure that I know what “ Disci-
plin” means, but if it implies failure to see very obvious
things on account of preconceived notions, it is a quality
which I am very happy not to possess. Bidder also (1891)
described the oscula of cerebrum as “lined with collared cells
continuously up to the granular lip.” I am unable to agree
with this statement, though it is true that in cerebrum the
oscular rim is shorter than in any other species known to me.
I have sections through an osculum of cerebrum where the
upper limit of the collar-cell layer is only separated from the
extreme edge of the osculum by about five epithelial cells in
the section. But in general, even in this species, the oscular
rim is deeper than that. The shortness of the oscular rim in
cerebrum leads, however, to one interesting deviation from
the more usual condition, such as has just been described in
contorta. The transformation of the epithelial cells into
porocytes in the former does not take place wholly within the
oscular rim, but commences before the epithelium has turned
the edge, some way down on the exterior of the oscular tube.
In contrast to cerebrum, the specimens of reticulum which
I have examined were remarkable for the extreme length of
the oscular rim, the region devoid of collar-cells often extend-
ing down to the commencement of the tubes which converge
to open into the cavity of the oscular tube.

Before leaving this subject there is one point to notice.
Both the sieve membrane,! formerly described by me in

1 In some specimens of coriacea there is aring-like sphincter, such as I
described inclathrus. Ihave seen what is obviously a trausition between
the two structures, namely, a sieve membrane with a very large aperture in
the centre surrounded by smaller apertures peripherally. In some specimens

496 k. A. MINCHIN.

coriacea, aud the ring-like sphincter which I described in
clathrus, as well as similar structures in other species, are
made up of the porocytic epithelium of the interior of the
oscular rim—a strong proof of the contractile nature of these
cells. I may repeat an earlier statement of mine, namely, that
when contracted the epithelium lining the oscular rim never
assumes the mushroom form characteristic of the epithelial
cells of the exterior when in a similar condition.

Do the porocytes multiply amongst themselves? I have
stated above that they do not, which means I have never seen
them do so, at least not when fully differentiated ; while on the
other hand I have, I think I may say, abundant proof of their
origin individually from ordinary epithelial cells. But Pl. 41,
fig. 38, shows a cell of the porocytic epithelium of the oscular rim
containing two nuclei, proving that they may multiply in this
region. I have never found functional pore-cells multiplying
to form more pores, though it would be rash to affirm too
positively that they could never do so, but at least it is a rare
occurrence. To sum up briefly the results of this investigation
upon the pore-cells, we have found—

(1) That the porocytes form a definite and well-characterised
layer of cells. (2) That their characteristic features are the
same as those of the ordinary dermal epithelial cells, but greatly
exaggerated. (3) That they arise by modification and trans-
formation of ordinary dermal epithelial cells, which come to lie
amongst the collar-cells in two ways: at the oscular rim by
inclusion amongst the collar-cell layer as it grows upwards;
elsewhere by actual immigration from the surface into the

there is no trace of either sieve membrane or sphincter. In others again,
incrusting forms of opaque, leathery appearance, with the tubes forming a
network in one plane, the whole gastrail cavity is traversed by a network
running in all directions, not only at the oscular rim, but everywhere. This
network has the same structure as the sieve membrane, and its threads are
made up entirely of granular porocytes surrounding an axis apparently of
jelly. The specimens which show this network are very resistent and firm
even when expanded, and no doubt derive support from it. Compare the
similar network figured by Dendy in “ Leucosolenia” proxima (1891
[2], pl. viii, figs. 1, 2).

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 497

interior. (4) That the pore-cells are eminently contractile,
and when contracted have received very different interpreta-
tions from different authors.

Connective-tissue Layer.—This consists of the spicules
and their secreting cells embedded in the structureless jelly
which constitutes the greater part of the body-wall.

The triradiate spicules have each a single cell applied to the
extremity of each ray. These spicule-cells, as we may term
them shortly, have the protoplasm very clear and almost free
from granules. The nuclei are slightly smaller, and stain more
deeply than those of the flat epithelial cells, appearing oval in
profile view, circular in surface view (fig. 14).

I desire to correct the statement made by me in a former
paper, to the effect that each spicule in coriacea had a nucleus
at the extremity of each ray, and a fourth at the confluence of
the rays (1892 [1], p. 265). When a spicule ray crosses the
centre of another spicule an appearance may be produced of a
cell at the centre of the spicule, which is not, however, the
case. There are no other cells on the fully formed spicule
than the three at the extremities of the rays. What applies to
the triradiates applies also to the basal rays of the quadri-
radiates. The apical rays of the quadriradiates have, as we
shall see, a variable number of nuclei upon them. The large
mouaxons of many species of Clathrina are covered by a
number of cells, but exactly how many is a point very difficult
to determine.

As we shall describe fully below the origin of the spicule
cells, it is not necessary to enter intoa further discussion upon
them here.

Amcebocytes.—The wandering cells are very important
and conspicuous elements of the sponge body. They give rise,
as is well known, to the sexual elements, and their complete
history will, I believe, furnish some points of considerable
interest when worked out. At present, however, I am still far
from being able to give a complete account of them, and will
therefore content myself with describing them as they occur
ordinarily in the adult sponge.

498 E. A. MINCHIN.

In the first place, there occur always large cells of lobose
and irregular form, densely packed with refringent granules.
These cells seem to possess a nutrient and distributive function,
and are abundant in all parts of the sponge. A_ striking
point about them is that their appearance is very different in
different species, though fairly constant in individuals of the
same species. ‘lo such an extent is this the case, at least for
the species in which I have studied them, that it would be
easy to identify and distinguish preparations of blanca,
coriacea, clathrus, cerebrum, reticulum, contorta, and
my undetermined species by their wandering cells alone. With
these differences they exhibit certain constant points of struc-
ture common to all, which makes it easy to distinguish them
in preparations. In size they are normally a good deal smaller
than the porocytes. Their outlines are rounded, and their
form either compact or irregular, with short lobed processes,
never fine and pointed. Their nucleus when visible through the
granules is seen to be large and rather pale, with a vesicular
structure anda very distinct and large nucleus. Their granules
are large, very refringent, and of a very glassy appearance after
osmic and picrocarmine, quite different from the opaque dull
granules of the porocytes and the flat epithelium. As arule the
nucleus cau only be discerned with difficulty as an indistinct
patch of colour, if stained, through the mass of granules.

Of all the species mentioned, the wandering cells of this
type are most remarkable in clathrus, where they have a
peculiar greenish or greenish-yellow colour in osmic-picro-
carmine-glycerine preparations, so that the eye can easily pick
them out in the preparation, even with a low power of the
microscope. The granules are of moderate size, rounded or
oval in form, and of refringent, glittering appearance.

In coriacea (Pl. 38, fig. 10, am.c.') the wandering cells are
yellowish, but very distinct in colour as well as in general
appearance from the yellowish-brown porocytes. They con-
tain large pale granules which have almost the appearance of
vacuoles, and also small dark granules with a steely glitter.
The nucleus is very hard to see.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 499

In “Clathrina, sp. dub.,” the densely packed granules
of the amcebocytes are of moderate size and glassy appearance,
without any particular tint (Pl. 39, fig. 19, am. c.*), and the
nucleus is quite obscured by them.

In blanca, cerebrum, and reticulum also the wandering
cells are colourless, but clearer than in the other species, and
of very compact form. The nuclei are distinct. Of the three,
blanca has the least granular cells, reticulum the most
granular.

In contorta the granules are small and very numerous, of
rounded or oval shape, and the cells are often of very irregular
form (P1]. 41, figs. 48, 44).

Besides the granular and refringent amcebocytes just de-
scribed, there appear to be others constantly present, which
have clear, finely granulated protoplasm. In coriacea (Pl. 38,
fig. 10, am.c.*) the cells are found to vary very much in
size, but are distinguished by their nucleus, which has a very
distinct nucleolus, and is large in comparison to the cell body.
The cell itself is of irregular form, often with sharp processes.
In Clathrina, sp. dub., the clear wandering cells, so far as I
have studied them, are very similar to those of coriacea.
In contorta, on the other hand, these cells are similar
in their characters to the granular wandering cells, and only
differ in the minute size of their granules (PI. 41, figs.
46, 47).

It is by no means certain that these cells are really different
from the granular cells. I am inclined to think that in the
case of contorta, at any rate, the clear cells are simply those
in which the supply of nutriment is exhausted. Perhaps they
correspond to the two classes of cells which Fiedler (1888) has
distinguished as “ Fresszellen” and “ Nahrzellen.”’

Finally, I have to mention very peculiar elements which I
have found in all the species. These are very minute cells witha
small faintly staining nucleus. Sometimes they are of globular
shape, but more often elongated, with the nucleus at one
extremity ; the former may be the condition of rest, the latter
that of active locomotion (Pl. 39, figs. 17, 19, and Pl. 41, fig. 41,

500 BE. A. MINCHIN.

am.c.>). In some preparations of coriacea these minute
elements appear to be connected by a series of transitions with
the clear amcebocytes already described ; that is to say, the
latter by repeated division seem to break up into these exces-
sively minute cells. It is possible that this process represents
simply the method by which the wandering cells multiply. I
make this suggestion with all reserve, but am guided to it by
appearances, which I hope to describe in a future paper on the
embryonic development. For the present it is sufficient for
me to have described these cells simply for purposes of recog-
nition, in order to distinguish them from the cells of the
dermal layer proper.

Apart from the sexual cells, of which it is not my intention
to treat in the present memoir, we can recognise two con-
stituent layers in the sponge body, besides a class of cell
elements which seem, properly speaking, to belong to neither
layer. These are—

(1) The gastral layer, consisting of the collar-cells lining the
interior.

(2) The dermal layer, consisting of—

(i) The external neuro-muscular flat epithelium.

(ii) The internal connective-tissue layer, consisting of the
spicules and their formative cells.

(111) The porocytes scattered about more or less evenly in
the wall.

(3) The amcebocytes or ameeboid wandering cells met with
in all parts of the sponge.

(c) System and Nomenclature.—In a former paper
(1896 [2]) I have put forward the outlines of the classification
which I intend to adopt. I recognised four genera of Ascons,
one of which has been seen as yet only by Haeckel. For these
genera I employed the names Clathrina, Gray; Leuco-
solenia, Bowerbank; Ascandra, Haeckel; and Ascyssa,
Haeckel.

The genus Clathrina is characterised by its reticulate
form, equiangular triradiate systems, collar-cells with basal
nuclei, parenchymella Jarva, and ‘ protascetta”’ stage in the

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 901

development. Ascandra is similar in all its characters, but
has the gastral layer folded by reason of the great develop-
ment of the gastral rays of the quadriradiates. The genus
Leucosolenia, on the other hand, has an erect or arbores-
cent form, sagittal triradiate systems, collar-cells with ter-
minal nuclei, amphiblastula larva, and “ protascyssa” stage in
the development. Of the genus Ascyssa, all that can be said
at present is that the skeleton consists entirely of monaxon
spicules.

My classification has not yet been in the field long enough
for it to have received the criticisms which I hope it will call
forth. The only point which as yet has been raised is with
reference to the name Ascandra. lLendenfeld has raised
objections to my retention of this name, coined by Haeckel, in
quite a different sense, for the species falcata, one out of the
many species which Haeckel referred to this genus. I will
not revive the discussion, since Lendenfeld has admitted
(‘ Zool. Centralbl.,’ iv, No. 7, p. 231) that the course taken by
me was “vielleicht formal richtig, real aber jedenfalls un-
praktisch.” If the retention of the name Ascandra is for-
mally correct I desire no more, for I am quite of opinion that
in a question of nomenclature, which should be treated purely
as a matter of names, finality can be attained only by rigid
adherence to rule, even if inconvenient.!

I propose now to develop my scheme of classification further
by dividing the Ascons into two families, corresponding to their

1 The abstract of my classification given by Lendenfeld (I. c.) misrepresents
one point completely. In my diagnosis of Clathrina I gave as one of the
characters ‘principal spicules of the skeleton, equiangular triradiate sys-
tems,” having previously defined the term “principal spicules ” as those “of
which the general skeleton is composed, and which are found in all parts of
the sponge,” as distinguished from spicules forming “a special dermal or
other layer restricted to some region of the sponge colony.’ Lendenfeld
cites this point in my diagnosis as “spicules principally equiangular ”
(Nadeln hauptsachlich gleichwinkelig), which gives quite a wrong impression.
Of Leucosolenia I stated “ triradiate systems always sagittal,” exception
of course being assumed of those abnormal and irregular spicules which are to
be found in every specimen. ‘This, again, is quoted as “spicules principally
sagittal.”

VoL. 40, PART 4,—NEW SER. NN

502 E. A. MINCHIN.

affinities as represented graphically in my former paper in the
form of a genealogical tree. The first family would be the
Clathrinidz, and includes the genera Clathrina and
Ascandra. The second family would be the Leucoso-
leniide, to include Leucosolenia and probably Ascyssa.
Their systematic position is as follows:

Class CALCAREA.

Sub-class Homoca@ta.
Order Ascones.

I. Family Clathrinide.
Genera Clathrina and Ascandra.

Il. Family Leucosoleniide.
Genera Leucosolenia and (?) Ascyssa.

Il. Descriptive: OBSERVATIONS UPON THE DEVELOPMENT OF
THE SPICULES.

Preliminary Remarks.—For the sake of clearness it may
be permitted to anticipate in one point the results of the in-
vestigations of which the description follows, with reference,
namely, to the relations between the triradiate and quadri-
radiate spicules. As is well known, each triradiate spicule is
formed of a system of three rays, and each ray is placed tan-
gentially in the body-wall, entirely enveloped in the dermal
layer, and never projecting freely from it, neither towards the
exterior nor the interior. Typically one ray—the posterior
ray—points away from the osculum, and therefore downwards,
in an erect olynthus; while the other two, the lateral rays, slant
upwards. In the species we are about to consider the three
rays meet at equal angles, and the posterior ray is only marked
out from the lateral rays by its position in the sponge, and
sometimes also by its greater length (Cl. blanca). The three
rays do not, however, lie all exactly in the same plane; but, as
might be expected from their position in the wall of a cylinder,
a plane containing any two of the rays is met by the third at

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 503

an acute angle, so that if the spicule were placed on a plane
surface with its gastral aspect downwards, the tips of the rays
would rest on the plane surface, while the centre of the spicule
would be raised up from it.

The quadriradiate spicules consist of a basal system of three
rays, orientated in the sponge body exactly as are the rays of
the triradiate spicule, and a fourth, the gastral or apical ray,
which arises from the centre of the basal system, making in
the genera Clathrina and Ascandra equal angles with the
three basal rays, and projects free into the gastral cavity,
passing between the collar-cells.

Not only is the basal system of the quadriradiate spicule

similar in all its relations to the whole of a triradiate spicule,
but it also develops in exactly the same manner. In the
species investigated by me not the least difference was ob-
served between the development of those triradiate spicules
which remained as such and those which, by addition of a
fourth ray, became quadriradiates. This accounts for the fact
that in nearly all species of Ascons the triradiate spicules and
the basal systems of the quadriradiates are so exactly similar,
at least in the case of the spicules of the general skeleton,
and differ, if at all, only in size. This fact is even more strik-
ing in the species of Leucosolenia, in which the spicular
systems have a bilateral form, than in the Clathrinide. At
its first appearance it is not possible to predict to which class
the spicule is destined to belong, since neither the fourth ray
nor its secreting cell appears until the basal system has attained
a certain size. The fourth ray is, in fact, an adventitious
element, superadded to the basal system from a totally different
source. For this reason I have proposed in a former paper
the term triradiate system, to denote in a general way both
the triradiate spicules and the basal systems of the quadri-
radiates.

It will therefore couduce greatly both to brevity and to sim-
plicity to describe first the development of the triradiate
systems, both of those which remain triradiate spicules and
those which become the basal systems of quadriradiates, and

504 E. A. MINCHIN.

then to proceed to describe the origin and formation of the
gastral rays.

(a2) The Development of the Triradiate Systems.—
The origin of the triradiate systems can be traced back to
certain cells of the dermal flat epithelium, which have wan-
dered inwards and have come to lie between the flat epithelium
and the collared epithelium of the gastral layer. This is true
equally of the first spicules formed in the young sponge
after the fixation and metamorphosis of the larva, and of all
spicules formed during subsequent growth in the adult sponge.
The immigration of the spicule-secreting cells in the young
fixed embryos first causes the dermal layer to become differen-
tiated into a more internal connective-tissue layer and an
external contractile epithelium. In the adults, where the
connective-tissue layer is well established, the skeletogenous
cells migrate from the epithelium into it, so that the connec-
tive-tissue layer is continually being recruited, as it were, from
the dermal epithelium. .

In the present memoir I propose to deal more particularly
with the formation of the spicules in the adult, a few stages
from embryos being described for comparison. I have observed
the formation of the triradiate systems in the adults of Clath-
rina_coriacea, Cl. reticulum, Cl. cerebrum, Cl. con-
torta, and Cl., sp. dub., and in the embryos of Cl. cerebrum,
Cl. reticulum, and Ascandra falcata. Since in all these
cases the development proceeds in an essentially similar
manner, and differs only in points of detail, 1 have taken
coriacea as a type (figs. 1—16), in order to avoid unnecessary
multiplication of similar figures. The description to follow
refers, therefore, to coriacea, unless the contrary is stated.
The differences shown by other types will be mentioned in
their place, and where necessary illustrated by figures.

The mother-cells of the spicules can be found without diffi-
culty in actively growing parts of the sponge (figs. 1—3
and 10, Pl. 38, act. bl.). They are compact, irregularly rounded
cells, which in all the characters of their nucleus and cyto-
plasm show an agreement amounting to identity with the

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 505

cells of the epithelium. Like them they are packed with
large opaque granules, some appearing quite dark, others of a
much paler tint in preparations. The spicule mother-cells, in
fact, only differ from the dermal epithelial cells in their posi-
tion and shape. Instead of being spread out into thin plates
they form compact lumps, and hence acquire a rather denser
and darker appearance. From all other cell elements they
are easily distinguished. They are much smaller and less
opaque than the contracted pore-cells, and their nuclei are
slightly smaller and stain more deeply. On the other hand,
their granules and the structure of their nuclei distinguish
them at once from the various kinds of wandering cells,
from which they differ so markedly that it will be sufficient
to refer to P]. 38, fig. 10, a.m. c.!, am. c.?, and to the description
of the wandering cells given above.

I have just spoken of these cells as ‘spicule mother-cells,”’
but this term is to be avoided if it conveys the idea that one
such cell produces a whole spicule, for, as we shall see, each
of the cells in question is responsible for one ray only of the
spicule. The commonly used term “scleroblast” is perhaps
a sufficiently vague term, meaning only a cell which secretes a
spicule, and not committing us to any theory or opinion as to
whether the cell so termed secretes the whole or part of a
spicule. It is, however, better on the whole to invent a term
which connotes accurately the relations of these cells to the
future spicule, and they may be conveniently termed “ actino-
blasts.”

Sometimes these skeletogenous cells are found singly (P1.38,
fig. 10, act. 6/.), or at least too far from any others for their
connection with them to be apparent. More usually they are
met with in trios (Pl. 38, figs. 1—3), and, as a rule, in such
close contact that their apposed surfaces become more or less
flattened against one another by mutual pressure, giving a
figure like a trefoil. Trios may, however, be found in which
there are considerable intervals between the component cells
(fig. 1) ; and this fact, taken in connection with the not infre-
quent occurrence of isolated ‘ actinoblasts,’ show that in many

506 E. A. MINCHIN.

cases, at least, the cells approach one another after immigra-
tion to form the trios. The next stage in the development is
the division of each cell of the trio into two cells, giving rise
to a figure which may be termed the sextet (Pl. 38, fig. 4,
coriacea; Pl. 39, fig. 17, Clathrina, sp. dub.; and fig. 22,
embryo of falcata). The division of each actinoblast takes
place in such a way that the plane of cleavage is parallel to
the inner and outer surfaces of the body-wall, so that the two
daughter-cells are placed one above the other when seen in a
surface view of the body-wall. Hence the two cells arising
from the division of a single actinoblast may be termed the
inner and outer formative cells respectively. The whole sextet
is made up now of two superposed cell-trefoils, if the phrase
may be allowed; so that when examined in surface view, three
cells with their nuclei can be made out at the higher focus,
and three others, exactly similar to the first and imme-
diately below them, at a lower focus.

In spite of much searching I have not been able to make
out the manner in which the division of the nuclei takes place.
Indeed, though I have studied very carefully the actively
growing portions of the sponge, cell division in the tissue cells
generally remains a mystery to me. I have never seen any
karyokinesis, nor have I come across any stages of direct
nuclear division. It is not uncommon, however, to find stages
intermediate between the trio and sextet,—that is to say, two
nuclei in a cell as yet undivided. In Pl. 39, fig. 22, is figured
a stage from an embryo of Ascandra falcata in which one
cell of the trio has completely divided, while in the other two
the division has not gone beyond the nucleus. It, is, more-
over, impossible in surface views of a sextet to say for
certain whether the division is complete between the upper
and lower cells; the sextet shown in Pl. 39, fig. 17, is an
instance of this.

The sextet is now ready to secrete the spicule, and the first
preparation for this event appears to consist in a fusion of the
formative cells at the centre of the sextet (fig. 4); if not
actually fused, the cells are at least in such close contact that

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 507

their limits are indistinguishable towards the interior. The
young triradiate spicule appears in the central portion of the
sextet, and is so placed that each of its rays corresponds to
one of the three pairs of formative cells (PI. 38, figs. 5, 6,
coriacea; Pl. 40, figs. 31, 32, embryo of reticulum; and
Pl. 41, fig. 48, contorta). The exact relation of the spicule
ray to its two formative cells is difficult to determine in surface
views. So far as can be made out by carefully focussing the
different structures, the ray lies between the two nuclei. It
is, however, impossible to discover by this method exactly
which of the two formative cells is concerned with the first
appearance of the spicule ray. In the absence of direct evi-
dence on this point there are some grounds for believing it to
be the inner formative cell which secretes the minute ray at
its first appearance, both on account of the subsequent rela-
tions which we shall shortly describe between this cell and
the ray, and for another reason which may be mentioned here.
In many preparations made by the osmic acid and picro-
carmine method the spicule-secreting cells are remarkable for
the fact that their nuclei stain much more brightly than the
nuclei of other cells of the dermal layer. Not only is this the
case with the nuclei of separate cells, but it even occurs, as we
shall see, when a cell contains two nuclei, one of which is
destined to be the nucleus of a scleroblast (figs. 18, 18 a, 19,
19a). This interesting reaction to staining fluids is not the
rule in preparations made by this method, and I am unable to
state upon what it depends, or to give any directions for
obtaining this result; but in a series of preparations mounted
from a specimen of the sponge which I denote for the present
as Clathrina, sp. dub., the deeper colour of the nuclei in
the spicule-secreting cells was very obvious (see Pl. 39, figs.
17—21)."

Fig. 17 shows a sextet in which the cells are still perfectly
distinct from one another, and no sign of the spicule is to be

1 This colour reaction suggests a connection between the nucleus and the
secretion of the calcareous matter, for which we shall find further evidence
when discussing the growth of the adventitious gastral ray.

508 BE. A. MINCHIN.

found. It will be noticed that the three lower (inner) cells
are more deeply stained than the three upper (outer) ones. In
fact, the three lower nuclei are as deeply coloured as the
nucleus of the neighbouring spicule-cell, while the three upper
nuclei are scarcely at all deeper than the nuclei of the flat
epithelium.

The above facts are strong evidence for believing that it is
the inner formative cells which are concerned with the first

SEXT.

c

te
A Vente. coe
Oi Sco) Soe

Fie. A.—Very young triradiate system in its sextet. xX 1600. ru. EP. Flat
epithelium. sExt. Sextet.

appearance of the spicule rays, but the point cannot be con-
sidered as established with certainty.

A very important point with reference to the young spicule
is the fact that its three rays appear to be at first quite
separate from one another. In some cases the distinctness of
the rays is extremely obvious, especially in the very young
spicules of Cl. contorta (Pl. 41, fig. 48) ; in other cases it is
not so obvious, and fusion of the rays seems to take place very
early (Pl. 40, figs. 31, 32). The smallest spicule seen by me
was in a preparation of Cl. coriacea, and the three rays were
like three bacteria, about 14 w in length; they appeared to be
in contact at their inner ends, but the minuteness of the
spicule and the thickness of the cells rendered it impossible to
make out their exact relations (see woodcut, Fig. A). As the
rays increase in size their separation from one another becomes
more marked, and is generally quite distinct in spicules with

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 509

rays up to 5 win length (figs. 5 and 48), sometimes even in
much larger spicules (fig. 23). But in spicules with rays more
than about 5 «in length complete fusion has taken place, as
a rule, between the rays (fig. 6), and instead of three separate
spicules, we have a single spicular system.

The spicule continues to grow, and at the same time its
formative cells begin to shift their position. Hach ray, as we
have seen, is surrounded by two formative cells, one lying
internally to it, the other externally. At first the ray is small
in proportion to its formative cells, and completely embedded
in them; but it soon grows to a length far exceeding the
diameter of the cells. As it does so, the inner formative cell
remains at the apex of the ray, impaled, as it were, upon its
point, while the outer formative cell remains at the base of the
ray (Pl. 38, fig. 7). In consequence of these relations, the six
cells of the sextet become widely separated from one another,
at least as far as their principal mass is concerned. The exact
limits of the cells are very difficult to determine on account of
the refraction of the spicule. It is evident from the manner
in which the spicule grows that it must be covered everywhere
by a layer of protoplasm. But except in the vicinity of the
nucleus the investing layer of protoplasm cannot, as a rule, be
made out, so that the bodies of the formative cells appear dis-
tinct from one another and adherent to the rapidly growing
spicule. The spicule does not appear naked when exposed by
the separation of its formative cells, but it is covered every-
where by a delicate layer or sheath, very hard to see in freshly
made preparations, but very distinct in preparations kept some
time in glycerine, in which the spicule is gradually dissolved.
This is the first appearance of the well-known spicule sheath.
It becomes more distinct as the spicule grows in size, but can
scarcely be distinguished until the spicule ray has outgrown its
formative cells, and it appears as a structureless membrane
or layer intervening between the spicule and the secreting
protoplasm.

The result of these changes is a stage very commonly found
in the growing spicule, and very characteristic, PI. 38, fig. 8,

510 E. A. MINOHIN.

shows it for Cl. coriacea; Pl. 39, figs. 18, 19, and 21, for
Clathrina, sp. dub., figs. 23 and 24 for embryos of falcata,
and fig. 27 for embryos of cerebrum; Pl. 40, fig. 33, for
reticulum; Pl. 41, fig. 38, forcontorta. In coriacea when
the rays of the spicule have attained a length of about 20 p,
each ray has on it two cells, one at the base and one at the
apex. The basal cell appears as a more or less fusiform mass,
applied to the spicule ray, and containing the uucleus in
its thickest part. The surface of the cell is usually smooth
and rounded, without processes. The apical cell, on the other
hand, has a very irregular outline, and runs out into processes
in a way which gives the impression of its being amceboid and
motile. It is placed quite at the tip of the spicule ray, its
nucleus lying on a level with the point or even beyond it, and
the portion of the cell body in contact with the ray is relatively
very small. All the formative cells have now become much
denser through gradual absorption of the granules with which
at an earlier stage they were packed, and the diminution and
disappearance of the granules goes on pari passu with the
growth of the spicule. The apical formative cells appear rather
clearer and less granular than the basal cells, which is perhaps
partly due to their being more spread out and less compact.

The next event in the history of the spicule is the dis-
appearance of the apical formative cells. The period at which
this occurs is somewhat variable, not only in the case of dif-
ferent species, but in the spicules of the same individual, and
even for the rays of the same spicule. In my preliminary
account I stated for Cl. coriacea that the apical formative
cells, which I then wrongly identified with the outer cells of
the sextet, disappeared when the spicule rays had attained a
length of 10m or 15 yw. More extended observations have
shown me that this is far too low an estimate for the average
length of ray at the time when this event takes place. Al-
though the apical cells are often not to be found at this early
stage, they more frequently persist until the spicule ray has
attained a length of at least 20 u, and sometimes even to a
much later stage of growth,

MATERIALS FOR A MONOGRAPH OF THE ASCONS. aM |

In coriacea it is, in my experience, rare to find a spicule
with rays exceeding about 20 » in length which has apical
cells present on all the rays, but here and there a spicule ray
occurs which far exceeds this length, and still bears a formative
cell at its apex, while the other rays of the spicule may show
no trace of any such cell. In Pl. 38, fig. 9, is seen a spicule
with rays averaging about 20 in length, and only one of the
rays retains its apical formative cell. The spicule in fig. 10
has rays of about 45 in length, and still bears a formative
cell at the apex of one of its rays. Fig. 14 shows a ray of
about 60 uw in length, in which the apical cell appears to be in
the act of leaving the ray. This is the longest ray I have ever
seen bearing an apical cell in this species, and as a rule the
apical cells disappear long before the ray reaches so great a
length.

In other species, so far as my observations go, the apical
cells may persist for a much longer time,—that is to say, until
the spicule ray has attained a much greater size. This is
especially the case in Clathrina, sp. dub. (see P1. 39, fig. 19),
in which species the apical cell disappears relatively late in the
history of the spicule; in fact, it is usually found adherent to
the tip of the ray after the basal cell has begun to migrate
from the base outwards, and sometimes the basal cell may be
found to have approached fairly close to the apical cell before
the latter disappears.

In Clathrina contorta also the apical cells persist to a
late stage (Pl. 41, fig. 38). It seems asif there was some rela-
tion between the size to which the spicule is destined to grow
and the persistence of the apical cells. In the species charac-
terised by larger spicules the apical cells persist to a relatively
later period, as measured by the growth of the rays, than in
the species with smaller spicules. Unfortunately it is in
species of the former kind that the facts are most difficult to
ascertain, on account of the greater thickness of the wall and
the number of spicules crossing each other in all directions.

We see, then, that the period at which the apical formative
cell disappears varies very much, and that in all cases it

512 E. A. MINCHIN.

persists long enough to show that it plays an important part
in the growth of the spicule. Its disappearance is, indeed, a
very remarkable fact, taking place as it usually does at a time
when the sister formative cell is still at or near the base of the
ray, and in all cases so far removed from the apex that a con-
siderable stretch of the ray is left exposed and, to all appear-
ance, covered only by the sheath.

So far as the further growth of the spicule is concerned, the
apical cell no longer exists for us, but the question as to what
becomes of it is of considerable interest. Unfortunately this
question is one very difficult to answer positively. When the
apical cell disappears, it is not to be supposed that it has any
other fate than that of becoming merged in some class of
cells, amongst which it is, or rapidly becomes, indistinguish-
able in its characters and appearance. It is scarcely possible
that it should become a collar-cell or a wandering cell, both
on general grounds and because it is sufficiently distinct in all
its characters from cells of either class. It is also highly
improbable for the latter reason that it should become a pore
cell, for which office its inferior size alone would disqualify it,
so to speak. If all these possibilities are excluded, there
remain only two classes of cells into the ranks of which it
could be received; namely, the flat epithelium and the con-
nective-tissue layer. It is by no means impossible that after
leaving the apex of the spicule ray it should become an actino-
blast, and join with other cells like itself to form a spicule.
All the actinoblasts, however, which I have seen were exces-
sively granular, while the apical cells have few or no large
granules, since those which they originally possessed became
absorbed, as we have seen, during the growth of the ray. I
have seen no appearances which suggest that the actinoblasts
originate from the cast-off formative cells of spicules already
formed, but many which indicate an origin for them from the
flat epithelium by immigration. If, however, the actinoblasts
come from the epithelium, there seems no great difficulty in
supposing that their daughters, the formative cells, return
there. The one feature of the latter which at all marks them

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 513

off, namely, their poorness in the coarse granules characteristic
of the epithelium, is one that would not be very distinctive,
since the epithelial cells vary a good deal in this respect, and
it might be supposed that when once back in the epithelium
they would rapidly re-acquire the granules which they lost
when engaged in secreting the spicule.

While, however, it seems a reasonable supposition, indirectly
supported in many ways, that the apical cell after leaving the
spicule ray returns to the flat epithelium, direct proof of this
is, from the nature of the case, very difficult, if not impossible,
to bring forward. Appearances such as that shown in PI. 38,
fig. 18, seem to show that the apical cell wanders away from
the spicule by amoeboid movement. Its destination could
only be determined with certainty by keeping it under ob-
servation when living, and this is scarcely possible. The
matter has to be studied by observing and comparing different
stages in preparations of preserved material, and by this
method it is not possible to obtain more than circumstantial
evidence. It can only be said, therefore, that the balance of
evidence makes it a matter of extreme probability that the
apical formative cell returns to the surface after leaving the
spicule, and forms part of the flat epithelium from which it
indirectly took its origin.

The history of the spicule subsequent to the disappearance
of the apical cell is comparatively simple. The rays at this
period have a sharply conical form, whatever may be the form
of the complete spicule. Thus in coriacea the fully formed
spicule rays have the form seen in Pl. 38, fig. 14,—cylindrical,
or very nearly so, for their proximal half, and then tapering
gradually to a blunt point. Contrasting fig. 14 with the figures
preceding it, especially figs. 9, 10, and 12, we see that the
spicule ray is built up to its full thickness at the base before it
has attained its full length, and that the formative cell remains
at or near the base until it has done its work there. It then
migrates slowly towards the tip, building up the spicule as it
goes, till finally in the fully formed spicule we find the
definitive spicule cell persisting at the extreme tip of the

514 BE. A. MINCHIN.

spicule. Comparing figs. 11 and 14 with figs. 9 and 10, one
striking point brought out is the great activity of the originally
basal formative cell. The apical cell usually disappears at so
early a period that by far the greater bulk of the spicule ray
must be secreted by the basal cell alone. Further, a com-
parison of different stages makes it evident that the spicule
ray grows in length as well as in thickness after the apical cell
has gone, and while the remaining formative cell is still at the
base. This is conclusive evidence that the ray is enveloped
completely by a layer of protoplasm even when the formative
cell is quite at the base and the apex is apparently exposed,
although the refraction of the spicule makes it impossible to
distinguish clearly any such enveloping layer of protoplasm
except near the nucleus.

During the growth of the ray the formative cell undergoes
certain changes. First, the granules, which, as we have seen,
have all along been diminishing in number and size, continue
this process until they are at last completely absorbed. At the
same time the protoplasm also appears to diminish slightly in
quantity during the period of most active secretion, probably
on account of its being spread over the whole spicule ray ; as
a consequence, the nucleus becomes slightly compressed be-
tween the spicule and the surface of the cell, and has an oval
outline in side view and a circular outline in surface view.
As the result of these changes the condition is attained which
is characteristic of all fully formed spicules—a clear finely
granular cell, with a slightly compressed nucleus, adherent to
the extreme tip of each spicule ray.

The above account of the formation of the spicules is
based, as has been said, upon their development in coriacea,
but in the other species studied by me the process of spicule
growth is so similar in all essential features that I have no
hesitation in regarding it as the normal and typical mode
of development for the triradiate systems of the genera
Clathrina and Ascandra.

The differences in the development of the spicules in different
species affect only matters of detail, and are largely such as are

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 515

due to the variations in the characters of the dermal epithelium,
which have already been discussed above. In none of the
species examined by me are the cells of the flat epithelium
and the actinoblasts so granular as in coriacea (compare PI.
39, figs. 17, 19, 20, and PI. 41, fig. 40, with Pl. 38, figs. 4—7).
In consequence the granules, being fewer, are absorbed more
quickly, and the cells rapidly assume the clear, finely granu-
lated appearance characteristic of the older spicule cells
(P1. 39, figs. 18 —21; Pl. 40, fig. 33; and Pl. 41, fig. 38). We
have noticed already the difference in the period at which the
apical cells disappear. In coriacea the apical cells have not,
as a rule, absorbed all their granules before their flight, but in
contorta and the undetermined species they become perfectly
clear. The spicule formation in cerebrum and reticulum
does not call for any special remark, and beyond satisfying
myself that it follows in these species the typical plan I have
not studied it in great detail. It is indeed by no means an easy
matter to follow the details closely in thick-walled species.
The great mass of refractile spicules crossing each other in all
directions makes it very difficult to follow minute details.
For this reason I have not verified the scheme of development
for the spicules of the adult Ascandra falcata, but have done
so for the larve, as I shall proceed to describe.

The development of the spicules is not difficult to follow in
larve which have fixed themselves on cover-slips. Clathrina
blanca is rather an unsuitable species for this purpose, on
account of the compactness and consequent opacity of the
embryos, but cerebrum, reticulum, and falcata, on the
other hand, are favorable objects. The fixed embryos should
be preserved and stained with osmic acid and picrocarmine,
and mounted either in glycerine or Canada balsam. The
first spicules appear on the second day, i. e. about twenty-four
hours after fixation ; but new ones are continually being formed
during subsequent development and growth. The embryos
consist of a superficial layer of cubical or flattened dermal
epithelium, enclosing the gastral layer, which is represented
at first by a compact mass of cells, in which a cavity appears,

516 E. A. MINCHIN.

and grows until the gastral cells form a single layer round it.
To see the spicules and their cells one must focus the microscope
just below the dermal epithelium and external to the gastral
layer. In this way I was able to prove to satisfaction that the
triradiate spicules, from their first appearance onwards, are
formed just as in the adult. This is shown in Pl. 39, figs.
22—26, for falcata; fig. 27, for cerebrum; and Pl. 40,
figs. 31 and 32, for reticulum. On the other hand, my
former statement with regard to cerebrum (1896 [1], p. 51),
that the spicule-forming cells commence to secrete the spicule
when still on a level with the dermal epithelium, was not
confirmed by more extended investigations. This statement
was based upon the examination of living specimens, at a time
when material was scarce. Working a year later, with abun-
dance of material carefully preserved, I was unable to find a
single instance in which even the smallest spicules were not
covered by the layer of dermal epithelium; and I must therefore
retract my former statement, or at least restrict its applica-
tion, The origin of the spicule-secreting cells from the dermal
epithelium is very easily made out in the larva, both in surface
views and sections. I reserve completer proof of this point
for a work on the embryology. A point of difference between
the young and the adults is found in the relatively large
number of irregular spicules in the former. It is, if anything,
rather the exception to find the rays meeting at regular angles ;
I have even observed instances in which one ray was lacking
entirely, the two remaining rays meeting, perhaps, at an acute
angle. Later, when the young sponge assumes the adult
structure, the spicules formed are quite normal, and the irre-
gularity becomes less marked. In spite of their tendency to
variation, the first spicules can always be recognised plainly as
belonging to the triradiate type, and that only. Monaxons
and quadriradiates do not appear, as far as I have observed,
till after the osculum is formed and the sponge has begun
to grow.

To summarise the facts here brought forward with regard to
the origin of the triradiate systems, we can recognise a scheme

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 517

of development common to all, with a number of constant and
definite stages, as follows :

(1) Formation of “trios”’ by immigration of cells from the
flattened epithelium.

(2) From the trios arise the ‘‘ sextets”’ by division of each
cell into two.

(3) The spicule appears with each of its rays corresponding
to two sister-cells of the sextet, i.e. to two cells which have
arisen from the division of one of the cells of the trio.

(4) As the rays increase in length the inner formative cells
of the sextet remain at the apices, the outer formative cell at
the bases, of the rays.

(5) Disappearance of the apical formative cells.

(6) The basal formative cells, after building up the rays to
their full thickness at their bases, migrate slowly to the
extreme tips of the rays, where they remain adherent as the
definitive spicule-cell.

(6) The Origin of the Fourth or Gastral Ray of the
Quadriradiate Spicules.—It has already been stated that
the three basal rays of the quadriradiate spicule develop
exactly as do the triradiate spicules, and that the fourth ray is
an adventitious structure, derived from an entirely different
source. In fact, while the mother-cells of the triradiate
system are derived from the flat epithelium, that is to say,
from the dermal surface of the body-wall, the mother-cell of
the fourth ray is derived from the gastral surface of the body-
wall, and from the layer of cells which we have termed above
the porocytes.

Referring for a moment to what has already been stated
with reference to the origin of the porocytes, we have seen
that they originate in two ways in two different regions: (1)
at the blind ends of diverticula, as well as throughout the
sponge generally, from large granular cells of the flat epithe-
lium which migrate inwards through the dermal layer, and
come to lie between the collar-cells; (2) at the tip of the
osculum, from the layer of large granular epithelial cells
lining the interior of the oscular rim, which in their turn are

voL, 40, PART 4,—NEW SER. 00

518 BE. A. MINOHIN.

continuous at the oscular margin with the general flat epi-
thelium covering the exterior.

The mother-cells of the adventitious rays—the gastral
actinoblasts, as they may be termed—also originate in ways
which, though essentially similar, yet differ in details in the
two regions of the sponge body just mentioned. In the first
case a porocyte, which is actually performing the functions of
a pore, divides and gives off a cell which becomes a gastral
actinoblast. In the second case one of the granular epithelial
cells which line the oscular rim, and are destined to form
pores, gives rise to a gastral ray without ever having been
functional as a pore. Thus in all cases the gastral actino-
blast arises from a porocyte ; but while the commonest method
is for the porocyte to become a functional pore-cell, which in
its turn gives off an actinoblast, we have in a particular region
a layer of porocytes giving origin to functional pore-cells on
the one hand, and to actinoblasts on the other.

For demonstrating these facts Clathrina contorta is a
particularly favorable object, on account of the yellowish-
brown colour which characterises its pore-cells in preparations
made with osmic acid and picrocarmine. The porocytes are
in consequence very sharply marked off from other cells, and
it is very easy to trace them. It is only necessary to lay out
a piece of the sponge wall with the gastral surface uppermost,
and if the piece selected contains gastral rays in an early stage,
there is no difficulty in finding them. By the study of Cl.
contorta I was able to confirm and to extend observations
which I had made a year earlier upon Clathrina, sp. dub.

I will commence with the more usual mode of origin. PI.
42, fig. 49, shows in Cl. contorta a young triradiate system,
lying at a deeper level, below the collar-cell layer. Imme-
diately over the young spicule is a coarsely granular cell, the
gastral actinoblast, wedged in amongst the collar-cells and
still in continuity with a pore-cell close at hand. The pore-
cell and the actinoblast resemble each other exactly in the
characters of the nucleus and cytoplasm. The nucleus of the
actinoblast lies over the centre of the spicule, and as yet no

MATERIALS FOR A MONOGRABH OF THE ASCONS. 019

trace of the gastral ray is to be seen. Fig. 50 on the same
plate shows almost exactly the same state of things, except
that the rays of the triradiate system have grown slightly, and
at their junction a minute gastral ray has appeared in the
gastral actinoblast, the nucleus of which is now no longer
placed over the centre of the spicule, but to one side.

Fig. 51 shows a condition considerably in advance of the
foregoing ; the actinoblast is quite separate from the pore, ana
its nucleus has divided in two, while the ray secreted by it
has also grown considerably. In fig. 52 the condition is much
the same, except that the actinoblast still retains its connec-
tion with the pore.

In Clathrina, sp. dub. (PI. 39, figs. 18—21), I was able to
observe some important stages in the formation of the fourth
ray, which place the origin of the actinoblast beyond all doubt.
The preparations were mounted in the usual way after removal
of the collar-cells as described above, so that it was possible
to make out clearly the actinoblasts of the basal system. Fig.
18 shows a triradiate system with its formative cells, and ex-
tending over it is a pore-cell, perfectly typical and normal
except in possessing two nuclei. In fig. 18a the pore-cell is
drawn separately. One of the two nuclei is much deeper in
colour, and lies close to the minute gastral ray which has
already appeared at the junction of the rays of the triradiate
system. The pore-cell shows a commencing constriction,
dividing off the portion with the more deeply stained nucleus
as an actinoblast from the pore-cell proper with the paler
nucleus.

Figs. 19 and 19a show a similar state of things, except that
the actinoblast has nearly left the pore-cell, being connected
with it by a drawn-out neck of protoplasm which, as can be
’ seen from its indented outline, extended between the collar-
cells. Finally, figs. 20 and 21 show the actinoblast completely
cut off and secreting the growing gastral ray.

The stages figured in the two species enable us to construct
with certainty the following history for the first appearance of
the gastral ray. When a gastral ray is to be added to a tri-

520 E. A. MINCHIN.

radiate system the first sign of activity is the division of the
nucleus in a neighbouring pore-cell. An outgrowth from the
pore-cell containing one of the nuclei then extends over the
triradiate system, and in this extension of the cell the gastral
ray arises close to the nucleus. Finally the outgrowth be-
comes nipped off from the pore-cell to form the actinoblast.
We see that the gastral ray may make its appearance when
the actinoblast has become all but separate from the pore-cell
(Pl. 42, figs. 49 and 50), or long before this event takes place
(Pl. 39, figs. 18 and 18a). The amount of separation between
the two cells probably depends upon how far distant the pore-
cell originally was from the triradiate system.

The origin of the gastral rays in the oscular rim has not
been followed by me in the same detail, but only sufficiently to
make their origin clear, which I think is done by Pl. 41, fig. 38.
Here we see a triradiate system with its six cells lying under
the coarse granular epithelium of porocytes; one of the latter,
which already contains two nuclei, is engaged in the formation
of the fourth ray. The figure shows well the great difference
between the basal formative cells and the gastral actinoblast.
Pl. 41, fig. 41, shows an almost identical state of things in a
section of a decalcified specimen of the sponge. I have not
seen the earliest stages of the gastral actinoblasts in this
region, so cannot say whether they represent the whole of a
porocyte, or whether, as is more probable, a porocyte gives off
a cell to form a gastral ray, as in the cases already described.

Having, as I think, made sufficiently clear the origin of the
gastral actinoblasts, it only remains to follow their subsequent
history, which I have done in the case of Clathrina contorta,
Cl. reticulum, and Cl. cerebrum. Their behaviour in the
different species presents some points of interest, and is most
conveniently studied in sections, which should not be too thin
—10p at least.

To begin with contorta. We have seen that the nucleus
of the actinoblast divides into two (Pl. 41, figs. 88 and 41;
Pl. 42, figs. 51—53). The cell does not, however, divide
Each nucleus soon divides again (Pl. 42, fig. 54). Finally

MATERIALS FOR A MONOGRAPH OF THE ASOONS. 521

an exceedingly long and slender ray is formed, projecting
far into the gastral cavity, enveloped in a continuous mass of
protoplasm, a true plasmodium which contains four nuclei
scattered along the spicule ray (Pl. 42, fig. 55).

In cerebrum (PI. 40, figs. 28—30) there are two classes of
quadriradiates, distinguished by the characters of their gastral
rays. In one class the gastral ray is comparatively short, and
beset with spines near the extremity (fig. 29) ; these spicules
are very characteristic of the species, and have been noticed
by all the authors. The other class of quadriradiates, on the
other hand, seems to have been generally overlooked, though
it is the more abundant in many individuals; it has the gastral
ray long and tapering, often slightly curved at the tip, and
without spines (fig. 30). Typical examples of these two classes
are distinct enough, but nevertheless they are connected by
numerous intermediate forms. Thus in some spicules the
gastral rays are spiny like that represented in fig. 28, but
have the ray prolonged considerably beyond the spiny region ;
in others a long tapering ray like fig. 80 may have minute
rudiments of spines upon it.

In the young spicule the gastral ray is slender, straight, and
smooth (fig. 28). After the ray has attained its full thickness
at the base the actinoblast migrates towards the tip, leaving
the cylindrical basal portion apparently exposed. The nucleus
of the actinoblast does not divide, but in the spiny spicules
granules resembling chromatin make their appearance, scattered
about in the region of the spines (fig. 29). In the smooth
spicules similar granules are present, but instead of being
scattered they are collected together at the extreme tip of the
spicule ray, the nucleus being at the opposite extremity near
the basal limit of the actinoblast (fig. 30). I am unable at
present to explain these appearances, which were observed in
all the spicules examined, and both figs. 29 and 30 could have
been repeated, if necessary, to any extent. The impression
given is that granules of chromatin are budded off from the
nucleus to superintend, as it were, the centres of secretion
which give rise to the spines, though it cannot be said that

522 E. A. MINCHIN.

each such granule corresponds to a spine. In the smooth
spicules, on the other hand, the granules seem to be about to
form a second nucleus, and this is perhaps the manner in
which nuclear division takes place; but I have not succeeded
in finding an actinoblast containing two definite nuclei, but
only the state of things shown in fig. 30, one nucleus and a
bunch of granules. If this really represents nuclear division
it would be rather an unusual type, though essentially similar
to the manner in which, according to Fiedler (1888), the
nucleus of the ovum of Spongilla buds off little masses of
chromatin which represent the polar bodies.

In reticulum (Pl. 40, figs. 833—87) some quadriradiates
have the fourth ray comparatively short (fig. 35), while others
have it excessively long (fig. 37). The former have one
nucleus, the latter two, in the actinoblast. It is very easy to
find instances in which the nucleus has only recently divided
(figs. 33, 34, and 36), though I have not seen the actual process
of nuclear division, but this would probably be a favorable
object in which to study it. The interesting point to notice,
from the point of view of the growth of the spicule, is the
variation in the period at which nuclear division takes place.
We may assume that if the two nuclei are close together the
division must have taken place recently, since they afterwards
become so widely separated, as shown in fig. 37. In this way
we see that the division sometimes takes place at an extremely
early period (figs. 33 and 84), and sometimes only after the
ray has attained a considerable size, as in fig. 36, where the
small size of the two nuclei is a further proof of their recent
division. Thus it would seem as if a ray was sometimes
destined from the first to attain a great length, while at other
times an elongated ray arises by some change at a later period
in the secreting cell of a ray originally destined to be short.
From the large number of spicules that occur with gastral
rays similar to that drawn in fig. 35, it can hardly be doubted
that many never go beyond this stage, and are, so to speak,
adult spicules. Buta comparison of figs. 35,36, and 37 shows
how gastral rays of the short kind may be converted into rays of

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 523

the long kind by division of the nucleus after the actinoblast
has retreated from the base of the ray. On the other hand, in
fig. 33 the spicule ray is shown to be in its infancy from the
way in which the actinoblast extends down to the base, and
the presence of two nuclei gives every promise of a future
great development in size of the spicule ray.

The facts both in cerebrum and reticulum point to an
incipient differentiation of the quadriradiates into two classes,
already quite distinct in typical examples, but as yet connected
by transitions. In reticulum, at least, the spicules of one
class can probably become actually converted into spicules of
the other class. In view of the many instances amongst
Ascons and Calcarea generally of the spicules of a particular
kind—triradiates, quadriradiates, or monaxons—being differ-
entiated into two classes as a specific character, these facts are
not without interest.

There remains one fact to mention with regard to the
gastral actinoblasts, and that is the frequent occurrence in
them of rod-shaped or needle-like bodies of crystalline appear-
ance (Pl. 39, fig. 18, Pl. 40, fig. 30, and Pl. 41, fig. 41, z). I
am unable to state what may be their nature or significance.
That shown in fig. 41 was most distinct, and is interesting as
occurring in a decalcified specimen.

To sum up the facts observed with reference to the quadri-
radiate spicules, the fourth or gastral ray is an adven-
titious element superadded to the triradiate system,
and secreted by a mother-cell which is derived from
a porocyte. The nucleus of the secreting cell may
remain single, or divide into two or into four nuclei;
but in all cases the cell itself remains undivided,
forming a plasmodium-like investment to the spicule,
or at least to its terminal portion.

(c) Some Observations on the Formation of the
Monaxon Spicules.—In a former paper (1896) I described
the formation of the monaxon spicules in young stages of
Leucosolenia variabilis, and showed that they originated
each in a single cell of the flat epithelium. I studied at

524 EH. A. MINCHIN.

Banyuls the growth of the large monaxons of the species
termed by me Clathrina, sp. dub., but did not succeed in
finding the earliest stages. In the youngest stage found
the spicule bore three cells, two fusiform cells attached to
the shaft, and a branched cell at the extreme apex of one
end, apparently the proximal end. The two cells on the shaft
each closely resembled a basal formative cell on the ray of a
triradiate system, and the cell at the extreme tip resembled an
apical formative cell. A later stage had five cells, four on the
shaft and one at the apex. A still later stage had the apical
cell and a number along the shaft, but it was difficult to deter-
mine exactly how many on account of the great size of the
spicule (Pl. 39, fig. 20, spic. monaz.).

The stage with five cells is obviously to be derived from the
stage with three by division of each of the two cells applied to
the shaft in the earlier stage. This suggests that the stage
with three cells is derived in a similar manner from a stage
with one cell on the shaft and one at the apex; this would be
a state of things exactly comparable to what is found ordinarily
on the ray of a triradiate, with its basal and apical formative
cell. The huge monaxons would then be formed just as a
single ray of a triradiate system, with the difference that the
basal formative cell repeatedly divides to furnish a row of cells
which build up the spicule. In this connection we may refer
to the interesting lines of growth described by Ebner (1887,
Pl. 41, figs. 51—58) in the large monaxon spicules of Leu-
candra alcicornis and aspera, each system of lines being
probably referable to a separate secreting cell.

Many of the large monaxons in Clathrinide, on the other
hand, are almost certainly not true monaxons, but derived
by modification of a triradiate system (cf. Haeckel, 1872,
p. 350). In this way one can distinguish true or primary
monaxons from what may be termed secondary monaxons. A
good instance of the latter is to be found, probably, in the
large elbowed monaxons in the stalk of Clathrina lacunosa.
I have long had a belief that the large monaxons found in
Clathrinidz would turn out to be in all cases secondary

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 525

monaxons. My observations, so far as they go, have by no
means dispelled this suspicion.

Lendenfeld employed the presence or absence of monaxon
spicules to classify Ascons. If the term “ monaxon spicule”’
be employed without further qualification, it represents a
character which, if used for systematic purposes, yields, to my
mind, a very artificial classification. But I think it extremely
probable that primary monaxons will be found restricted to
the Leucosoleniddz, where they are the first spicules
formed, and that in Clathrinide, where the triradiate
systems appear first, the monaxons are secondary. This is,
however, at present a pure speculation. I hope at some future
time to be in a position either to prove or to demolish these
views.

Appendix.—A few points in connection with the formation
of the spicules or with their cells seem to call for special
notice before leaving the subject.

(1) The Triradiates of Clathrina clathrus.—In 1892
({3] p. 183) I described the spicules of this form as having on
their rays sometimes one cell, sometimes two, and sometimes
a cell with two nuclei close together. As this is rather a
different state of things from anything I have found in the
species investigated in this paper, I examined the point again,
and find my description perfectly correct. Looking now at a
series of drawings made by me at that time with the aid of the
camera lucida, I have come to the conclusion that the apparent
anomaly is due to the frequent persistence of the apical forma-
tive cell on the fully formed spicule rays of this species. When,
as is frequently the case, there are two nuclei in a single cell,
adherent to the tip of the spicule, I believe this to be brought
about by the basal formative cell having travelled to the apex
of the spicule ray, and there fused with the persistent apical
cell.

That the apical cell should persist in this way is rather
unusual, but may perhaps be connected with the cylindrical
form of the rays, which in this species terminate abruptly

526 E. A. MINCHIN.

or are even slightly clubbed—an uncommon form of spicule
ray. Whether this form of ray with persistent apical cell
is more primitive than the more pointed rays or not I am
unable to say, but it seems probable. Even in Cl. clathrus,
however, the apical cell by no means invariably persists. In
the majority of cases the spicule ray simply has a single
fusiform cell at or near its distal extremity, as in the forms
described above.

(2) The Influence of the Nucleus on the Secretion
of Lime.—It may be of interest to bring together the facts
observed which indicate a relation between the nucleus and the
secretion of the spicule. These are (a) the deeper stain, in
many cases, taken by the nuclei of the secreting cells; (5) the
fact that the nucleus places itself in the region of greatest
activity ; for instance, in the case of the rays of the triradiates
the nucleus of the basal cell remains near the base of the ray
till it is fully formed, and then moves along towards the tip,
superintending, as it were, the work to be done; (c) the
division of the nucleus where the secretion is carried on over a
considerable length of ray, as in the case of the long gastral
rays of reticulum and contorta, or the large monaxons of
Clathrina, sp. dub.; and (d) finally we may refer to the
chromatin-like bodies scattered over the spinous portion of the
gastral rays of cerebrum. So far as can be judged from
surface views, the minute sclerite always makes its appearance
in the immediate neighbourhood of the nucleus.

(3) The Granules of the Dermal Layer.—We have
seen that all the cells of the dermal layer contain very charac-
teristic granules, to which in the first place the colour of the
sponge is due. When the sponge is any other colour than
white, the colouring matter is contained in the granules of the
dermal layer, and is rapidly dissolved out of them by alcohol.
The granules attain their greatest development in the poro-
cytes, where they were clearly described by Metschnikoff ; but
they are also present in the flat epithelium, and usually in
great abundance. ‘They are always found in the young spicule-
secreting cells; but while in the case of the triradiate systems

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 527

they are absorbed more or less rapidly as the spicule grows, in
the case of the gastral actinoblasts they persist throughout in
undiminished size and quantity. As I am not acquainted with
any species with quadriradiates which shows colour varieties, I
am unable to state whether the granules in the gastral actino-
blasts would retain the vivid coloration which the granules
often show in some species,—coriacea, for instance. In the
latter I have seen an orange-red specimen which had a sieve
membrane across the osculum, and, as might have been ex-
pected, the sieve membrane was also of an orange-red colour.

I am not able at present to make any positive statements
about these granules, but will briefly notice some views that
have been put forward about them.

Bidder (1892 and elsewhere) regards the granules as excre-
tory, especially in the porocytes. They seem, however, too
constant an element in the cells of the dermal layer for
Bidder’s theory to be a complete explanation of them. There
is, moreover, a strong prima facie argument against the cells
of the flat epithelium or the porocytes being concerned in
excretion, as then the excreted products would immediately be
carried back into the sponge by the currents. On the other
hand, it must be confessed that the porocytes which line the
oscular rim would be in a particularly favorable situation for
exercising the function of excretion.

Topsent, on the other hand, as we have seen, regarded the
granules of the porocytes, or “ cellules sphéruleuses,” as repre-
senting reserve nutriment. It might be doubted if Topsent
would have come to this conclusion had he known that the
“cellules sphéruleuses” were simply contracted pore cells.
He seems, further, to have overlooked the fact that similar
granules occur in the cells of the flat epithelium.

It seems to me probable that the granules in question sub-
serve more than one function, but as a peculiarity of all the
cells in which they occur is their contractility, especially in
the case of the porocytes in which they are most abundant, I
have long thought they might have some connection with this
function. Biitschli has shown how, on the alveolar theory of

528 E. A. MINCHIN.

protoplasm, the granules might exercise a great influence on
internal movements of the protoplasmic framework, such as
those which result in contraction in one or another direction
(1892, pp. 207—209; English translation, pp. 323—327). The
distribution of the granules in the dermal layer favours this
view of their function, since they are most abundant in the
very contractile porocytes, of which any specially contractile
organs—sphincters or sieve membranes—are formed, while in
the spicule-secreting cells they rapidly disappear. An excep-
tion to this is apparently to be found in the case of the gastral
actinoblasts, which, however, according to Lieberkiihn (1865,
p. 737), are retractile, a fact that would explain the persist-
ence of the granules in them. It is noteworthy also that
while the dermal layer of the Clathrinide is remarkable
both for its richness in granules and for its contractility, the
corresponding layer in the non-contractile Leucosolenias is
very clear and free from large granules.

The theory that the granules aid in the contraction is, how-
ever, devoid of any actual basis of experiment or observation,
and is to be regarded merely as a possible hypothesis, which is
at least worth testing.

III. HistoricAL: EARLIER OBSERVATIONS UPON CALCAREOUS
AND OTHER SPICULES.

(2) Observations upon Spicule Formation.—Direct
observations on the origin and growth of the spicules in cal-
careous sponges are few and far between in the literature of the
group ; confident assumptions, on the other hand, with regard
to the way in which it is supposed a priori that the spicules
should and must arise, are to be found in abundance. In the
almost total absence of facts, the foundation for the current
beliefs concerning spicule formation in the Calcarea is to be
sought partly in deductions from theory, partly in analogies
from the well-established facts as to the origin of siliceous
spicules. It is best, therefore, to begin by a brief survey of
what has been ascertained in non-calcareous sponges.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 529

1. Observations upon Siliceous Sponges.—For the
spicules of Monaxonida the classic researches of Lieberkiihn
and Carter established the fact so far back as the fifties that
the spicules of Spongilla originated each within a single
cell. Ata still earlier date Carter had been of opinion that
spicules and horny skeleton were ‘ formed in the intercellular
substance, and not within the cells; 7’! but direct observations
caused him to abandon this opinion. Lieberkihn gave a full
and clear description of the formation of the spicules of
Spongilla in 1856 (pp. 407—409, figs. 17—30) ; and a year
later, but independently, Carter published an account con-
firming that given by Lieberkiihn (1857, p. 23, fig.8). Carter
afterwards extended his observations to marine sponges, and
described a similar origin for the spicules in Esperia ega-
gropila, Jnstn. (1874 [1], pp. 101—105, and [2] p. 456,
pl. xxi, fig. 26), and in Microciona armata, Bwk.
(1874 [2], p. 457, pl. xxi, fig. 27). Schmidt had already
(1864, p. 5, Taf. i, fig. 13) described the formation of the
spicules of Reniera, sp., and he is quoted by Sollas as
describing in detail their origin and growth in Esperia, ina
work which I have not been able to see.?__ Kolliker also figured
and described the monaxon spicule in its formative cell (1864,
p. 61, woodeut 15), and Keller described the spicule formation
in the embryo of Chalinula fertilis (1879, p. 334, pl. xix,
fig. 20). Sollas gave further instances of spicule development
in a series of papers dealing with sponges of Norway; the
bundles of trichites originating, each bundle, in one cell (1880,
pp. 141, 142, pl. vii, fig. 21); the formation of the sterras-
ters of Geodia Barretti, Bwk., each in a single cell (1880,
pp. 256 and 401, pl. xi, fig. 18, and pl. xvii, figs. 18—22) ;
and the monaxon spicules arising each in a cell (1882, p. 159,
pl. vii, figs. 12, 17, 18). It is not necessary to multiply these
instances further ; all observers are of one accord in describing
the monaxon spicules and microscleres of Demospongiz gene-

1 *Ann. and Mag. Nat. Hist.’ (2), iv, 1849, p. 95.

2 «Zoologische Ergebnisse der Nordenfahrt,’ p. 120; see Sollas (1888),
p. xlv.

530 E. A. MINCHIN.

rally, as arising each within one single mother-cell.
Sometimes even, as in the case of the bundles of hair-like
spicules known as trichites, more than one spicule may arise in
the scleroblast. In view of this pleasing unanimity with
regard to the first origin of the siliceous monaxon spicule, it is
to be regretted that so little seems to be known of the fate of
the primitive scleroblast. Does the mother-cell persist on the
fully formed spicule? Does the nucleus of the secreting cell
remain single, or does it ever divide? These are questions yet
to be answered. Carter discusses the point, and is of the
opinion that “ ornamental or subsidiary parts, such as the
spines, &c., are subsequently added, probably after the spicule
has left the mother-cell, and has got into the intercellular
sarcode, as shown by the central canal never extending into
them” (1875, p. 12). Here, then, we see the belief very clearly
expressed that the mother-cell is not responsible for the whole
spicule, even in simple forms ; and as no one would now ascribe
any skeletogenous function to the jelly or “sarcode” of
Carter, it must be supposed that other cells besides the original
mother-cell take part in aiding the growth of the spicule.
Ridley and Dendy are also of opinion that “it is pretty certain
that the larger forms at any rate become free from the parent
cell (silicoblast) before attaining their full size” (1887, p. xiv).
On the other hand, Maas found in Spongilla larve spicules of
1 mm. in length, with only one silicoblast nucleus, and thinks
‘that the whole duration of the growth only claims a single
cell, the more so as he has never seen spicules with cells
applied to them like an epithelium” (1890, p. 539). Delage
also (1892) figures many instances of spicules and their cells,
but never more than one cell to a spicule. But in the
“ shovels”? (anisochele) of the Esperia larva Maas found
four nuclei to each shovel, one on each side of the blade and
one on each side of the shaft (1892, p. 420, Pl. xxviii, fig. 19).
These shovels are united into rosettes, and this fact suggests,
according to Ridley and Dendy, that each such rosette origi-
nates within a single cell (1887, p. xx). From all this con-
flict of evidence and opinions the only thing clear is that the

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 531

subject is ripe for further investigations, and that at present
no certain conclusion can be drawn as to the fate of the
mother-cell.

Of the origin of the triaxon spicule of Hexactinellids we
know at present nothing whatever. But with regard to the
tetraxon siliceous spicule, the evidence is all in favour of an
origin within a single mother-cell. Schulze’s observations on
Plakina (1880) are, so far as I know, the earliest in this field.
The author figures in the plainest manner the tetractines in the
young sponge as enclosed each in a single cell (pl. xxii, fig. 29),
and the theory is advocated that the primitive siliceous spicules
were irregular thorny bodies, in which the rays first became
concentrated in a point, giving rise to irregular multiradiate
spicules; that then the number and direction of the rays became
fixed, so that there were formed hexactines and tetractines ; and
that from these forms arose triactines, diactines, and even
monactines, by reduction of the rays (1880, p. 445). A year
later Schulze figured and described in Corticium “a dis-
tinctly formed regular tetractine” “in a rounded mesoderm
cell containing a few pigment granules,” situated close to the
outer surface of the sponge; the spicule was “ still so com-
pletely embedded in the protoplasm of the cell that its tips
did not even project into the surrounding ground substance ”
(1881, pp. 426, 427, pl. xxi, fig. 10).

The next author to deal with the origin of the tetractines
is Sollas (1888). He says, “In the Choristida all the spicules,
both large and small, originate each in a single scleroblast,
which persists throughout the life of the spicule. The sclero-
blast in the case of the large spicules is a large granular cell,
extending all round the spicule, which it has formed as a
siliceous secretion ” (p. xlvi). Reference is then made to two
figures (pl. u1, fig. 20; pl. xi, fig. 10), both of which, however,
only show one arm of the spicule, the rhabdome, with a cell
upon it; but in Tribrachion Schmidtii the author figures
very clearly entire orthodizenes of various sizes, each with a
single scleroblast upon the rhabdome (pl. xvii, figs. 12 and
20). On the other hand, Sollas also figures three cases in

5382 E. A. MINCHIN.

which the shaft of a large spicule has a series of pyriform
cells along it, considered by the author to be derived from the
“collenchyma” (pl. iv, fig. 30; pl. v, fig. 6a; pl. viii, fig. 37 ;
compare p. 12). Sollas further states that in the Lithistida
the crepis of the desma is formed in a single scleroblast, but
that other cells are found on the arms of the desma. ‘ Hach of
the four depressions which occur about the centrum in the
angle between the arms of a tetracladine desma appears to be
occupied by a scleroblast, and others may possibly be distri-
buted along the sides of the arms” (p. xlvi). Reference is given
to figures (pl. xxx, figs. 20 and 21) showing the minute tetrac-
tines, each enclosed in a cell. Finally, Lendenfeld (1894 [1],
p. 166) expresses an opinion with regard to the tissues and
asters of Geodia and Ancorina, which is the same as that
which Sollas enunciated in more general terms. He believes
the spicules to arise within single cells as a refractile body near
the nucleus, afterwards embracing it. Some of the cells have
a process and are believed to form tricenes; others show a
radial striation and form asters. No figures are given in
support of these statements.

From a study of the literature of siliceous spicules, and
without going beyond the statements of the authors, one
would be justified, I think, in making the following generalisa-
tions :—(1) Every spicule originates within a single cell,
which perhaps in many cases receives assistance afterwards
from other cells, and certainly does so in the case of the
complicated desmas of the Lithistida. No evidence for
division taking place in the mother-cell or in its nucleus.
(2) The forms with few rays are derived from those with
many, by reduction and loss of some of the rays. This
applies in the first instance to the spicules of the Hexacti-
nellids and Tetractinellids, but it perhaps holds good also
for the phylogeny of the simple spicules of the Mo-
naxonida.

(2) Observationson Calcareous S ponges.—The earliest
recorded observations upon the relations of calcareous spicules
to the tissues of the sponge are those of Kolliker (1864) upon

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 933

a species from Villafranca which he termed Nardoa spon-
giosa.! Kélliker discovered the spicule sheath left behind
after the spicule is dissolved by acetic acid. On the rays of
the quadriradiates which project into the gastral cavity the
author observed “another problematic structure, namely, a
dark, granulated, conical mass, which envelops the calcareous
ray, and also, as it seemed to me, the sheath. Seen in surface
view these structures appear as round cells, and only profile
views make clear their true relations (pl. vii, fig. 10). In
some cases this dark granular mass is continued in reduced
quantity (verschmalert) over the portion of the calcareous ray
lodged in the epithelium ; but I was unfortunately unsuccessful
in discovering the true significance of this curious structure ”
(p. 65). It is easy to recognise in this account a description,
perfectly accurate as far as it goes, of the cells I have termed
the gastral actinoblasts, the secreting cells of the gastral rays.
The cells which the author mentions further on in his descrip-
tion as occurring on the walls of the canals near the ciliated
epithelium, “ rather larger rounded cells, singly or in groups,
the significance of which remained unknown to me” (p. 65),
were very probably porocytes.

1 Haeckel identifies Kolliker’s ‘ Nardoa spongiosa” with either
*“ Ascaltis Gegenbauri” or cerebrum. Iam inclined to dispute both
these identifications for the following reasons :—(1) Haeckel’s “Ascaltis
Gegenbauri” is figured by him as having the gastral layer folded like
Ascandra falcata; this is a point which certainly would not have escaped
Kolliker, who neither mentions nor figures any such peculiarity; (2) the
excellent figures of the sponge given by KoOlliker (pl. ix, figs. 6, 7) are
not from a specimen of Clathrina cerebrum; (3) there is a distinct
reference on p. 54 of Kdlliker to monaxons in his sponge, though he does not
mention them in the special description. From Kolliker’s figures of the

sponge aud of its long and slender gastral rays (pl. vii, fig. 10; and pl. ix,
fig. 8) I am inclined to identify it as contorta, which is one of the com-
monest Ascons in the Gulf of Lyons; or possibly as spinosa, Lendenfeld,
which, as I have stated above, I believe to be identical with contorta,
differing only in the total absence of the monaxon spicules so variable in size
and number in the true contorta. The gastral ray figured at pl. vii, fig. 10,
is not full-grown, as shown by the position of the actinoblast near the base.

As Kolliker had ten specimens of his sponge there may possibly have been
more than one species amongst them.

vou. 40, PART 4.—NEW SER. fle

5384 E. A. MINCHIN.

In the following year (1865) were published Lieberkiihn’s
observations upon “Grantia botryoides” = Leucoso-
lenia Lieberkihnii, O. S., of which it is not too much
to say that his descriptions attain a far higher degree of
truth and correctness in many points than some which have
appeared more than a quarter of a century later. He recog-
nises at the outset the fact of the body-wall being composed
of two distinct layers, ‘(a layer of contractile parenchyma
and a layer of ciliated cells which clothe the inner surface”
(p. 734). Here we have the two layers for which I have
revived Haeckel’s terms, dermal and gastral layer; in the
former “spherical, oval, and stellate corpuscles, at varying
distances from one another, can be distinguished in the trans-
parent homogeneous parenchyma” (p. 735). On the same
page we are told that just below the oscular margin the ciliated
lining ceases, and the aperture is surrounded by a simple layer
of jelly substance ; what better description could be given of the
oscular rim, of which it has so strangely fallen to my lot to main-
tain the existence and to point out the significance? Further
on (p. 738) the author repeats the observation, and gives it a
more general application; and on p. 742 he shows that a
similar region occurs in Sycons. With regard to the spicules
Lieberkiihn observes (p. 736) that on many of the projecting
gastral rays “a fine layer of the contractile substance [i. e. of
the dermal layer] can be seen pushing its way out between
the ciliated cells, and either covering the spicule completely
or partially in a fine layer, or enclosing only the root of
it in a thicker mass (Anhaufung).” Treatment with acetic acid
dissolves the spicule, ‘and the contractile substance remains
behind as a more or less thin-walled sheath.” Hereby the
author thinks that Kolliker’s observations, which he quotes,
receive an explanation ; and he further observes, “ In favor-
able cases the conical mass can even be traced through the
epithelium, and recognised in continuity with the contractile
substance, in many places very granular. Besides this, the
same forms of thickened sheaths of the contractile substance
can be found singly on the free outer surface [doubtless the

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 585

scleroblasts on the projecting monaxons]. These, as well as
those of the inner surface, are retractile”’—an observation
which, if true, would be of great interest.!

Not less remarkable than these early observations of Kolli-
ker and Lieberkiihn, which so nearly solved the question of
the origin of the gastral ray, is the fact that no further
observations upon the gastral rays are to be found in all the
voluminous literature of calcareous sponges until we come to
those of Dendy, to be mentioned in due course, published in
the present decade.

The next author to formulate opinions upon the formation
of calcareous spicules was Haeckel, but his views were purely
theoretical and devoid of any basis of fact or observation.
Haeckel regarded the dermal layer of the sponge as a syn-
cytium, formed by the fusion of cells originally distinct in the
embryo. He supposed the ground substance or sarcodine to
be made up of the fused protoplasm of the cells; and that in
this sarcodine, ‘‘ at once the outer covering of the body and its
‘skeletogenous layer,’ its contractile and its sensitive tissue ”’
(p. 164), the spicules arise by a sort of crystallisation, the
spicule sheath being formed by “a thickening and separation
from the sarcodine” (p. 167). Altogether a very clear and
logical theory, did it but harmonise with the facts.”

Schulze, in his classical memoir on Sycon raphanus
(1875), did not make any observations on the origin of the
spicules, but appears to have been largely of Haeckel’s opinion.

1 Haeckel, having never seen a scleroblast—incredible though this may
seem,—understands Lieberkiihn to mean that the sheath is retractile, and
denies that this is the case. I do not attach this significance to Lieberkiibn’s
statements. The scleroblast may quite well be retractile.

? It is remarkable that Haeckel should never have seen and figured the gastral
actinoblasts, especially after the clear descriptions of Kélliker and Lieberkiihn.
In such a figure as that of Ascaltis Gegenbauri (1872, pl. ix, fig. 7) we
are astonished to see nothing of these cells, usually so conspicuous. A closer
inspection, however, of the figure, to which reference has been made, reveals
a number of sperm masses in close proximity to the spicule ray, in such a way
as to provoke the suspicion that Haeckel confused two totally distinct kinds
of cell elements.

536 E. A. MINCHIN.

He believed the external protoplasm of the cells of the dermal
layer to be “ completely fused and modified into a homogeneous
and possibly contractile ground substance” (p. 252), and he
agreed with Haeckel in regarding the spicule sheath as a con-
densed layer of the ground substance closely surrounding the
spicule (p. 254). That Schulze at this period agreed with
Haeckel as to the mode of formation of the spicules may be
further inferred from the fact that in his work on the develop-
ment of Sycandra raphanus (1878), while figuring clearly
the monaxon spicules each in its secreting cell (pl. xix, figs.
10, 11), he yet stated in the text that the spicules arise in the
hyaline substance between the two layers.

With the year 1879 we come to the only important obser-
vations, as apart from theories, on the formation of calca-
reous spicules which have been made since the sixties, namely,
those of Metschnikoff. This author first demonstrated that
calcareous, like siliceous spicules, arise within cells, and not
free in the ground substance of the sponge body. Valuable
as were Metschnikofi’s results, they are nevertheless not free
from important errors of misinterpretation as regards the
details of the development, though based on accurate and
careful observation.

The specimens of Ascons which Metschnikoff investigated
seem to have been for the most part more or less contracted,
since in sections he found a dermal epithelium composed of
flask-shaped cells. We have already noticed his observations
on the porocytes, which he believed to be granular mesoderm
cells. In Clathrina primordialis he found these cells
giving rise to the skeleton. ‘The smallest calcareous spicules
are formed in the interior of such cells” (p. 361, and pl. xxii,
figs. 4,5, 8, and 12 s.). Thisisa most important statement in
view of what has been said above, namely, that both the poro-
cytes and the spicule-forming cells originate from the dermal
epithelium. The cell masses surrounding the smallest spicules
do in fact resemble closely in appearance the contracted poro-
cytes, aresemblance due to the possession of similar cytological
characters derived from a common origin. Hence it is not

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 537

difficult to understand how Metschnikoff should have confused
them in contracted specimens. The youngest spicules are
figured by Metschnikoff in the middle of rounded granular proto-
plasmic masses, which become trefoil-shaped as the spicule
grows. There can be no doubt from his description that he
regards each of these protoplasmic masses as a single cell. He
appears, however, to have seen them only in the fresh state,
and in his figures no nuclei are shown; had he applied suitable
reagents, he would of course have found each of these masses
to be six cells, and not one only. It is a little strange he
should not have seen any cell outlines, but they are not very
obvious until the spicule has begun to grow and to cause the
cells to separate. In any case Metschnikoff was the first to
point out that the spicules “always arise in the cell proto-
plasm, and not in the jelly-like ground substance.” He does
not figure cells on the rays of the spicules, but the cell a on
pl. xxii, fig. 5, is obviously a spicule-cell detached from the
spicule (compare his description, p. 861). On the whole, I
think I may fairly claim that Metschnikoff’s observations
confirm, or even forestall, my own, though his interpretation
of them requires modification.

Poléjaeff, in his work on the ‘Challenger’? Calcarea,’
observed cells which he interpreted as scleroblasts on some of
the spicules, monaxon as well as triradiate (1883, p. 32). The
example which he figures (pl. vi, fig. 3, c.) is certainly very
remarkable in appearance, the result, perhaps, as the author
suggests, of preservation in alcohol.

In 1885 Lendenfeld published a series of papers on Australian
calcareous sponges, which contain some observations on the
spicules and their origin. Of the spicules he states (1885 [2],
pp. 979, 980) that by prolonged action of gold-potassium
chloride the spicule splits into prisms “parallel to one
another, radiating from the axis.” “The radial structure
first makes its appearance in the interior, close to the inner
axis, which is a cylindrical chord of organic matter without
lime.” ‘The inner part, the part produced first, of the spicule
is softer, and contains more organic matter, whilst the outer

538 E, A. MINOHIN.

layers, the youngest part, is [sic] harder, and resists the action
of reagents; the whole spicule is composed of prisms formed
as cuticular productions by the cells clothing the spicule
from without” (p. 980). ‘The spicules first make their
appearance within cells, and the axial rod (not canal !) is part
thereof. The succeeding layers are cuticular productions of
endothel cells.” The author gives no pictorial illustrations
of his researches on spicule structures, but he has a figure
showing “ mesoderm cells’’ forming an “ endothel” “in the
the shape of a hollow tube” covering the rays of a triactine
of “Ascetta procumbens” (1885 [1], pl. Ixiii, fig. 3; and
description, p. 1146; also [2], p. 980). He is more successful
in his delineations of hard structures than of soft, since he
draws quite correctly the sharply conical form of the rays of
the young triradiate as compared with their more cylindrical
contour when fully formed.

Lendenfeld’s later statements with regard to calcareous
spicules are not much happier than his early efforts in this
direction, and a few quotations may serve to make clear
his standpoint. On p. 198 of his work on Adriatic sponges
(1891) we read, ‘On the surface of the spicules [of Clath-
rina primordialis] one observes not infrequently flattened
cells, sometimes provided with processes, which singly or
united in small groups partly envelop the spicule.” In
opposition to Metschnikoff the author finds that ‘ the irregu-
larly shaped lumps of protoplasm”? which envelop young
spicules, or are applied to them, are fairly transparent and free
from refractile granules (p. 199). ‘‘ The spicules probably
arise in cells, but it must be pointed out that even the
youngest of the large spicules which are to be found in the
sponge body are much longer than any known sponge cell
with exception of the ripe ova. The further growth of the
spicule takes place by means of numerous cells, which settle
on the surface of the young spicule and precipitate spicular
substance upon it..... The skeleton-forming elements are
always cells of the intermediate layer (Zwischenschicht) ”
(p. 383). These three sentences furnish a good instance of

MATERIALS FOR A MONOGRAPH OF ''HE ASCONS. 5389

this author’s habitual method of stating dogmatically, and as
if from his own observation, propositions which are the result
merely of imagination, and deduced a priori from, in this
instance at least, unsound premises.

In the present decade Dendy has put forward views on the
origin of calcareous spicules, which can best be made clear by
quotations from his works. “It is generally admitted,” he
says, “that both calcareous and siliceous spicules originate
within special mother-cells, but probably in both cases they
subsequently receive additional layers of the calcareous or
siliceous material from other cells. . . . There are probably
two kinds of calcoblasts, primary and secondary. The primary
calcoblasts are the mother-cells in which the spicules originate,
and the secondary calcoblasts are the cells which secrete
additional layers of calcareous matter around the spicule after
it has been formed” (1891 [1], p. 25 ; compare 1891 [2], p. 15).
He states his belief in the same passage that the calcoblast is
“an ordinary stellate mesodermal cell.”’ ‘‘ The spicule sheaths

. . are formed by a slight concentration of the structure-
less mesodermal jelly around the spicule” (1893, p. 223).
Supposed primary calcoblasts are figured on the uniaxial
spicules of Leucandra phillipensis, and it is stated that
“their characters certainly justify the assumption that they
are but slight modifications of ordinary stellate cells” (1893,
p- 225, figs. 44—47). Dendy quotes my former statements
(1892 [1], p. 265, fig. 15, a, 6), which I have stated above to
be erroneous, as to the presence of a fourth cell at the centre,
in addition to the three on the rays, of the triradiates of Cl.
coriacea; and considers that the cells on the rays represent
the secondary calcoblasts, “‘ for we can hardly suppose that
the spicule is originally formed by more than one cell” (1898,
p. 225). In so far as this remark applies to the triradiate
systems, I think the observations described above sufficiently
refute it, but in so far as Dendy wishes to express his conception
of what constitutes a true spicule in general, as distinguished
from compound spicular systems, I quite agree with him.

In considering Dendy’s views of spicule formation, it must

540 E. A. MINCHIN.

be borne in mind that his “‘ stellate mesodermal cells ” are the
same as my “spicule cells.” Dendy believes the mesoderm of
Calcarea to be packed with stellate connective-tissue cells, for
the most part independent of the spicules, and similar in nature
to the stellate cells in the jelly of a Medusa; I, on the other
hand, have been unable so far to find stellate connective cells
of this kind at all in Ascons, and am of opinion that what has
been taken for them is the spicule-cells, which in sections
are generally found separated from their spicules, as the me-
chanical result of section-cutting. Hence Dendy’s “stellate
cells,’ being identical with my “spicule-cells,’ we are agreed
in stating that they are “calcoblasts.” Where I must differ
from Dendy is with regard to the existence of a special mother-
cell for the triradiate systems. In the assumption of “ primary
calcoblasts,” of which the existence has certainly not been
demonstrated, we can trace the influence partly of an analogy
with siliceous spicules, partly of Metschnikoff’s descriptions
and figures.

Dendy considers that ‘‘ the calcoblasts, at any rate in the
case of large spicules, must be amceboid, for unless they be so
I cannot understand how the spicules can increase uniformly
in thickness” (1893, p. 225). I have shown, however, that
they do not increase uniformly in thickness, but are built up to
their full thickness, first at the base and afterwards at the tip.

The apical ray of the quadriradiate system, according to
Dendy, “is not naked, but clothed by an investing sheath of
plate-like nucleated cells,” which “resemble the cells of the
ectoderm” (1891 [2], p. 14, pl. vi, fig. 2). Dendy considers these
sheaths as mesodermal structures, though ‘‘ there is nothing
to prove that they are not endodermal ;”! probably, however,
they are “‘calcoblasts,”’ derived from the stellate connective-
tissue cells. The last thing probably with which Dendy would
have thought of connecting them is the “yellow granules,” as

1 Bidder, in his review of Dendy’s work, regards these cells as endodermal
(1891, p. 630), and has some curious remarks on the subject of the mesoderm.
This discussion seems to me a proof of how meaningless are the terms “ ecto-
derm,” ‘‘ endoderm,” and ‘‘ mesoderm,” when applied to these forms of life,

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 541

he termed the porocytes. Even in zoology truth is sometimes
stranger, or seems so, than fiction.

In 1895 I published a preliminary account of the origin of
the triradiate systems of Clathrina coriacea, and in 1896
I showed that the monaxon spicules in the young Leuco-
solenia variabilis are formed each by a cell of the flat
epithelium. The last author who has written upon calcareous
spicules is Déderlein in the present year. In Petrostoma
Schulzei the author finds young quadriradiates in the cortex.
“From them... by reduction (Verkummerung) of one of
the four arms arises the triradiate. This derivation of the
triradiate calcareous spicule seems to me much more natural
than the opposite view” (p. 21). The author does not clearly
state what he means by “ the opposite view,” and his remarks
might at first sight be supposed to refer either to Haeckel or
to my theory, published two years before (1895); but as he
makes no mention of my preliminary note on the origin of the
triradiate systems, the author’s words must be supposed to
have a more abstract reference, and to be intended simply as
an exposition in general terms of his views on the phylogeny
of calcareous sponge spicules.

Since Déderlein was only able to study dried specimens of
Petrostoma, we learn nothing from his description as to the
relation of the young spicules to the soft parts.

To sum up the results of our researches in the literature of
calcareous spicules: we might divide the authorities in this
field conveniently into two classes or groups, of which the
first would comprise Kolliker, Lieberkiihn, and Metschnikoft ;
the second all the remaining authors. From the former we
obtain valuable observations, but no theories ; the latter give
us, on the one hand, plenty of theories, which would be valuable
if they had any foundation of fact-; but of observations, on the
other hand, next to nothing. Until the publication of
Metschnikoff’s work (1879) the predominant authorities were
in favour of an origin for the calcareous spicules in the gela-
tinous ground substance of the body-wall. Metschnikoff having
destroyed this notion by demonstrating a cellular origin for

542 E. A. MINCHIN.

the spicules, a mode of formation analogous to that supposed
to occur in siliceous sponges, has been universally assumed.
Two notions have dominated all the writings on this subject :
(1) that spicules must arise each in a special mother-cell ;
(2) that their subsequent growth must be aided by other cells
of the body-wall. To these principles some authors add a
third, as obviously based on the analogy of siliceous sponges,
namely, that the tetraxon spicule must be the primary form,
from which the others arise by reduction of the number of
rays. I shall discuss the last of these three propositions more
fully in the theoretical portion of this paper, and hope to show
with regard to it what is sufficiently obvious in the case of the
first two, namely, that a priori reasoning, founded on pure
analogy, has led to conclusions hopelessly at variance with the
facts.

(5) Observations upon the Structure and Compo-
sition of the Calcareous Spicules.—The chemical and
physical nature of the calcareous spicules has been so tho-
roughly worked out by Haeckel (1872), Sollas (1885), and
Ebner (1887), that as I have nothing to add to the sub-
ject, I shall content myself with the references to these
authors, and a brief summary of the results reached by the
careful investigations of Ebner asa preliminary to a theoretical
discussion.

The most striking, and in many ways unexpected peculiarity
of these bodies, of which the triradiate and quadriradiate forms
are so clearly shown by their development to be compound
structures, built up from different sources, is the fact that
each spicule behaves optically as a single crystal. This was
first shown by Sollas and confirmed by Ebner, who further
denied the presence of organic matter in them, postulated by
Haeckel under the name of spiculin. ‘“ Each spicule behaves
as a single crystal individual, and an organic substance cannot
be demonstrated init. . . . The spicule by no means consists
of pure calcium carbonate in the form of calcite, though very
similar to it from the point of view of crystallography (in
krystallographischer Beziehung), but the spicule substance

MATERIALS FOR A MONOGRAPH OF THE ASCONS, 543

contains in addition considerable quantities of other inorganic
constituents, amongst which sodium, magnesium, and sulphuric
acid are demonstrated, together with, probably, water” (1887,
p- 132). “The spicules of calcareous sponges are mixed
crystal individuals, consisting principally of calcite, and con-
taining no organic substance ;' their outer form, not limited
by the surfaces of a true crystal, is conditioned by the specific
activity of a living organism, and their inner structure, though
completely crystalline, is in relation with the outer form by a
peculiar distribution of the constituent parts” (p. 134). “ The
calcite first secreted contains the greatest amount of impurities
(the central thread), and as the tip of the spicule continues to
grow a substance corresponding to the central thread is first
formed ”’ (p. 1383).

IV. THEORETICAL: ON THE ORIGIN AND EVOLUTION OF
CALCAREOUS SPONGE SPICULES.

The spicules of sponges, and especially of calcareous sponges,
are structures which seem in a certain sense to connect the
two worlds of living and dead matter. Composed, as a rule,
of simple inorganic materials, they show the geometrical regu-
larity of form which we associate with the mineral kingdom.
Formed, on the other hand, within a living body as a product
of its vital activity, they exhibit a diversity of character and a
progressive evolution of form such as we find only in living
organisms. As in the case of crystals, we can reduce them to
a few fundamental types, constructed according to simple
geometrical patterns. As in the case of living bodies, on the
other hand, we can distinguish in them the characters of
classes, orders, or families, until we come down to specific
characters, and finally to racial and individual variations.
Bodies which lie so close in many ways to the border-line
between the organic and the inorganic must excite an interest,
which is a sufficient justification for embarking upon a dis-
cussion as to their nature, and for considering the causes which

1 For further remarks on this point see Addendum A (infra, p. 569).

544 E. A. MINCHIN.

have been instrumental in producing the fundamental types
under which they occur.

Four theories have been put forward to explain the forms of
the spicules of Calcarea or of sponges generally, which we may
term respectively the Biocrystallisation Theory of Haeckel, the
Adaptation Theory of Schulze, the Mechanical Theory of
Sollas, and the Alveolar Theory of Dreyer.

Haeckel, whose views on the secretion of spicules we have
already noticed, regarded calcareous spicules as the products of
a process of “ biocrystallisation,” ‘‘i.e. a combination of the
erystallising activity of calcium carbonate and the organis-
ing activity of the sarcodine.”’ Calcareous spicules are
*‘biocrystals, i. e. form individuals, which represent a mean
between an inorganic crystal and an organic secretion.”
“Their first origin depends on a compromise between the
crystallising efforts of the calcium carbonate and the forma-
tive activity of the fused cells of the syncytium.” The
primitive secretion of calcium carbonate assumed a semicrys-
talline nature, and gave rise to bodies ‘“‘ which were utilised '
by natural selection as spicules for building up a skeleton,
and which afterwards, by the interaction of adaptation and
heredity, became modified in form and differentiated in very
various ways in the struggle for existence ” (1872, p. 377).!

Haeckel considered the primary forms of spicule in the
Calcarea to be the triradiates, and the monaxon or acerate forms.
He regarded the quadriradiate form as undoubtedly secondary,
and derived from the triradiate by addition of the gastral ray.
He discussed the relationship between the two primary forms,
and considered (p. 350) the three possible theories—(1) that the
monaxon and triradiate spicules are independent formations ;
(2) that the monaxon spicule has arisen from the triradiate ;
(3) that the triradiate spicule has arisen from the monaxons.

1 Schmidt at an earlier date (1870, p. 4) had attempted to explain the
regular tetraxon siliceous spicule by the action of crystallisation, but con-
sidered any such explanation inapplicable to calcareous spicules. Schmidt
evidently thought that in siliceous sponge spicules the silica was in the

crystalline form of quartz, whereas it is now known to be in the colloid form
of opal.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 9540

The last of these possibilities he sets aside at once, as
entirely improbable and unsupported by facts. The second is
regarded as undoubtedly true in some cases; but the first is
the true relationship between monaxons and triradiates in
general. The vast majority of monaxon spicules in Calcarea
are regarded as primary forms, fundamentally distinct from the
other primary form, the regular triradiate spicule. There are
thus two stems or lines of descent marked out in Calearea, the
first starting from the genus Ascetta, the second from the
genus Ascyssa. The two primary forms of spicule are traced
back by Haeckel to two distinct types of biocrystal. ‘The
fundamental form of all monaxon spicules is the absolutely
regular spindle, or a cylinder, on the basal surfaces of which
are seated two similar cones with curved mantle surfaces. The
original and fundamental form of all triradiates and quadri-
radiates is . . . the absolutely regular triradiate, which can be
considered as a hemiaxon form of the hexagonal crystalline
system, in which calcium carbonate crystallises as calcite”
(p. 377).

Schulze, in his great work on Hexactinellids (1887, pp.
497—504), put forward a general theory of the forms of
spicules, not only in Calcarea, but in sponges generally. He
first of all attempts to trace the main lines of descent in the
phylum of sponges, and recognises three main stems, each
leading back to a distinct ancestral form, characterised by a
peculiar type of skeleton. The three lines of descent are (1) the
Calcarea, with a calcareous skeleton; (2) the Tetraxonia, with
siliceous spicules typically of the tetraxon type, retained in
the modern Tetractinellids! and Lithistids; the Monaxonida
are considered as having arisen from the Tetraxonia by reduc-
tion of the rays of the spicule, and the Keratosa from the
Monaxonida by loss of the spicules and their replacement by
spongin; (3) the Triaxonia, with siliceous spicules of the
Triaxon type, represented by the modern Hexactinellids. Each
of these three types must have acquired its skeleton indepen-

1 Under the Tetractinellids, Schulze includes such forms as Plakina,
Corticium, &c., now usually separated in the division Carnosa.

546 E. A. MINCHIN.

dently of the other two, since there are no transitions between
them so far as the skeleton is concerned, and their common
ancestor can only have been a simple primitive form without a
skeleton.

Schuize next discusses what is the most primitive form of
spicule in each of the three great groups into which sponges
are thus separated, and tries to show that the form of the
primitive spicule is in each case a direct adaptation to the
simplest type of anatomical structure, as seen in the more
primitive or least specialised examples of the groups, or in
young specimens. It would be out of place here to discuss
at length his explanation of the origin of the triaxon and
tetraxon siliceous skeleton, and we will only consider in detail
his theory as applied to Calcarea.

Schulze does not consider any crystallisation theory as a
sufficient explanation of the form of a spicule as a whole: (1)
because symmetrical forms occur not only in the case of
spicules composed of a crystalline material such as calcite, but
also in the case of siliceous spicules composed of a colloid sub-
stance, namely, opal; (2) because the rays frequently deviate
from the typical angles, and are often markedly curved. He
considers even the fundamental types of the spicules to be de-
termined solely by the matrix in which they lie. In Calcarea
the most primitive group is that of the Ascons, represented by
the olynthus stage in the development of the higher forms,
and the most primitive type of spicule is the regular triradiate,
with three equal rays meeting at equal angles of 120° ina
plane. An Ascon or an olynthus has the form of a thin-
walled tube, open at the free end, with the wall perforated by
uniformly distributed pores. The triradiate spicules occur
embedded in the wall, their rays lying tangentially to the
surface, each spicule being typically so orientated that one ray
is directed away from the osculum, while the other two run
obliquely forwards, or rather upwards. In the angle between
any two rays a pore is situated,! and the regularity of the

1 Schulze suggests also an alternative arrangement, which to the best of
my belief, however, does not occur in nature, and need not be discussed.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 547

angles is to be explained in much the same manner as in the
well-known instance of the honeycomb. The triradiate systems
are, in fact, the form of spicule best fitted to support the porous
wall of the Ascons, and the anatomical structure and arrange-
ment of the soft parts show the utility of the specific structure
of the skeletal elements. A similar theory is worked out by
Schulze with great ingenuity and thoroughness for the primi-
tive tetraxon spicule of the Tetraxonia and the triaxon spicule
of Hexactinellids. In the case of the latter his view certainly
receives great support from the structure of the Cambrian
Protospongia, the oldest known fossil sponge.

Sollas, in his monograph of the Tetractinellida (1888), made
an attempt to explain the forms of spicules as the mechanical
result of forces acting upon them during their growth. After
claiming that the forms of bones or other skeletons in higher
animals are due to the effects of pressure and tension exerted
upon them during their growth by the surrounding tissues, he
thinks (p. lxxv) that ‘‘ were it possible to connect the special
forms of spicule with these forces an explanation would be
reached which would fulfil” the necessary conditions ; namely,
that a theory to explain the forms of spicules should be ‘ in-
dependent of the nature of the material, and capable of being
applied to all organisms in which spicular forms are developed.”
The principle of this explanation is ‘‘that all spicular struc-
tures tend to grow along lines of least resistance.” ‘‘ The
simplest form of spicule is a minute granule, generally more
or less spherical,” and from this as a starting-point Sollas
derives the various forms by the aid of his theory. We need
not follow here in detail the ingenious manner in which the
principle of “least resistance” is applied to explain the various
spicular forms of 'Tetractinellids and other sponges. For the
calcareous sponges he says (p. Ixxxi), “Let a scleroblast be
situated near the surface of a sponge, as it must be in the
Ascones ; the surface tension will here also lead to the growth
of three actines inclined at angles of 120° to each other, and
thus the triradiate spicule so common in the calcareous sponges
may have arisen.”

548 E. A. MINCHIN.

Dreyer, in his studies on skeletal formations (1892), is more
ambitious than either Haeckel or Schulze, since he tries to
find a theory to explain the forms of the spicules, not only in
sponges, but Radiolaria and Echinoderms as well. Like
Sollas, he invokes mechanical principles, but only to explain a
supposed universal type of primitive sclerite. He will have
nothing of either crystallisation or adaptation in explanation of
the forms of spicules. Crystallisation he rejects, with Schulze
and Sollas, as an agency which determines the forms, on the
ground of the occurrence of spicules similar in form but com-
posed of very different materials, such as lime or silica, or even
substances of organic nature. Hence he concludes that the
morphology of the sponge spicule is quite independent of the
nature of the materials composing it (p. 299). Adaptation he
considers inadequate, on the ground that natural selection is
“an externally regulative,” not an ‘internally formative ”
principle (p. 349).

Dreyer regards the tetraxon type of spicule as the primitive
form of skeletal element, not only for sponges, but also for
Rhizopods and Echinoderms. To explain this universally re-
curring type of spicule it is necessary to seek for some constant
cause; and this he finds in the vesicular structure of
living matter. All living bodies are built up of three orders
of vesicular elements, namely, (1) cells, (2) vacuoles, and (3)
the alveoli of the protoplasmic framework, according to
Biitschli’s theory of the ultimate structure of protoplasm, with
which he agrees (p. 350). The spicules being formed by living
bodies, and being therefore deposited in the midst of these
vesicular structures, have primitively the tetraxon form as a
direct mechanical result of vesicular tension (Blasenspannung).
Each such primitive tetraxon is, in fact, regarded by the author
as laid down in the nodes of the alveolar framework in such a
way that each arm lies in the interspace between three of the
four contiguous alveoli, which naturally touch each other at a
node.

The most unfortunate gap which at once presents itself in
this theory is the fact of its being totally inapplicable to the

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 549

triaxon spicules of the Hexactinellids ; and Dreyer is obliged
to leave them entirely out of consideration.

In considering these ‘different views, certain objections to
each one of them readily suggest themselves.

Haeckel’s theory is entirely borne out as regards the mo-
lecular structure of the spicules by Ebner’s investigations
(1887), and this author proposes to retain for them Haeckel’s
term ‘‘ biocrystals.” But crystallisation cannot be taken as an
adequate explanation of the external form of the spicules, at
least as they now occur. For in the case of the monaxons,
Haeckel’s “ absolutely regular spindle ” is a form which never
occurs, and though it might possibly have been the form under
which the spicules first appeared at a very remote period of the
earth’s history, yet Haeckel himself is of opinion that the
monaxon spicules as we know them are the results of natural
selection acting upon the primitive biocrystal ; in other words,
it is natural selection, and not crystallisation, which is the
factor which models the various forms assumed by the monaxon
spicules at the present day. A similar train of reasoning,
strengthened by additional arguments, applies to the case of
the triradiate spicules, and insuperable objections to Haeckel’s
theory in this instance are furnished, it seems to me, by the
observations upon the origin of these bodies set forth above by
me. ‘The case against Haeckel may be summed up as follows:

1. Ebner has already pointed out (1887, p. 134) that the
morphological axis of a triradiate spicule cannot possibly be
compared to a crystalline axis.

2. The occurrence in other sponges of forms of spicule even
more symmetrical and regular than those of Calcarea, com-
posed of non-crystalline materials, such as colloid silica or
spongin, has supplied such a weighty argument to all Haeckel’s
critics, that to attack his theory on these grounds may seem to
many like flogging a dead horse. Since, however, it is clear
that in other cases the form of the spicule is independent of
the nature of the material, it is reasonable to suppose that the
same may be true of calcareous spicules.

3. And, finally, my own investigations prove, to my mind,

vou. 40, pART 4,—NEW SER. QQ

550 E. A. MINCHIN.

that the triradiate spicule is not a simple and _ primitive
skeletal element, but is built up from three monaxon spicules ;
in fact, that it arises in just the manner which Haeckel dis-
missed without further discussion as improbable and un-
founded. It is, therefore, clearly impossible that the triradiate
system should owe its form, as a whole, to the action of
crystallisation.

While unable to accept biocrystallisation as an explanation
of the primitive forms of the calcareous sponge spicule, I fully
agree with the views Haeckel has put forward with regard to
their evolution. With Haeckel I consider that there are two
primitive types of spicule in the Calcarea, the monaxon and
the triradiate, which are independent of one another in the
form in which we now meet with them, though I shall try to
refer both to modifications of a yet simpler monaxon type.
With Haeckel I believe that these two forms of spicule are
associated with two lines of descent in the Calcarea, the one
starting from forms such as those constituting Haeckel’s genus
Ascetta, the other from his genus Ascyssa; and I may refer
to my own later attempts (1896) to give this theory further
support, and to introduce a classification of the Homoccela
modelled upon it. With Haeckel, finally, 1 am fully of the
opinion, of which I think the researches here brought forward
are a convincing proof, that the quadriradiate spicule is a
secondary form, derived, by the addition of an adventitious
ray, from the triradiate system.

The view advocated by Schulze—namely, that the funda-
mental forms of the spicules are to be explained by adaptation
to the primitive types of structure in sponges—is the one which
seems to me to contain the true solution of the problem, at
least in the case of calcareous sponges ; and I shall presently
try to show that my investigations at once confirm and extend
Schulze’s conclusions. The one defect, to my mind, in Schulze’s
presentation of the case is that he left out of consideration
the monaxon spicules, which are certainly of as primitive a
type as the triradiates. No scheme of evolution seems to me
complete which fails to indicate the relations of the monaxon

MATERIALS FOR A MONOGRAPH OF THE ASOONS. 551

spicules to the other forms. This gap I shall endeavour to
fill, but will first consider in more detail the views advocated
by Sollas and Dreyer.

Sollas’s theory was put forward to explain the forms of
spicules originating within a single cell, and his explanation of
the triradiate system was based upon the assumption that it
developed in this manner. Since, however, the triradiate sys-
tem owes its origin to three mother-cells, it is clear that Sollas’s
theory would not apply to it, still less to the quadriradiate. No
explanation based upon the principle of growth along lines of
least resistance would explain why the gastral actinoblast
should attach the monaxon spicule which it secretes to a tri-
radiate system. Sollas’s theory is therefore reduced at the outset
to an explanation of, if anything, the simple monaxon sclerite
from which the more complex forms are put together. In the
case of monaxon sclerites secreted by a single cell, it seems to
me highly probable that the mechanical principle involved by
Sollas does operate largely in producing the form of the
sclerite. But the principle is, after all, only an explanation of
how the scleroblast lays down the sclerite, and not of why it
has that form and no other. It is as if one should try to
*‘explain ” laughter by describing accurately the facial and
other muscles concerned, and the manner of their contraction.
If the lines of least resistance are such as to cause the secretion
of calcite to take an elongate, needle-like form while still
completely embedded in the scleroblast, this presupposes a
certain polarity in the cell; but to what is the polarity itself
due? I am unable to understand how the analogy of a bone,
supposed to owe its form to the tension of the muscles and
other tissues, can give us any help in the present case; for the
problem is to explain why the molecules of calcium carbonate
should be deposited more at the two opposite ends of the cell
than in the middle, and this can hardly be a matter simply of
pressure and tension on the sclerite itself.

I cannot, therefore, consider Sollas’s theory as more than ex-
planation, at most, of the mechanics of growth in the case of
the simplest skeletal elements, and not as a solution of the

552 E. A. MINCHIN.

problem why the sclerite should have the monaxon form, still
less of the triradiate and quadriradiate type.

Dreyer’s theory assumes the well-known theory of Butschli
(1892) as to the ultimate structure of protoplasm. Let me
begin, therefore, by declaring myself also a firm believer in
Biitschli’s views ; the translation published by me of his great
work may perhaps be taken as a sufficient guarantee of my
acquaintance with his theories. Biitschli regards protoplasm
as having the structure of a very fine emulsion or foam ;
droplets of a watery fluid or enchylema are suspended in a
denser and more viscid fluid which constitutes the alveolar
framework. The relations of these two parts are similar as
regards structure to those of an ordinary froth or lather, the
enchylema corresponding to the air, the alveolar framework to
the liquid, whatever it may be, of which the froth is composed.
In cross-section the alveolar framework would appear as a fine
network containing meshes of various sizes which were filled
originally with the enchylema. In addition to these two parts,
all known protoplasm shows a great number of granules or
microsomes, varying greatly in size, consistency, appearance,
and chemical reactions, but always contained in the alveolar
framework and lodged at the nodes of the reticulum. Proto-
plasm, therefore, consists of three constituent parts,—alveolar
framework, granules, and enchylema. The vacuoles, on the
other hand, which are so often a conspicuous feature of proto-
plasm, and which may contain various bodies, are not to be
regarded as another primary element, but as produced by the
fusion or running together of alveoli, just as in a froth a large
bubble may arise by the breaking down of smaller ones.

Dreyer assumes that the sclerites are first deposited at the
nodes of the alveolar framework, and that the supposed primi-
tive tetraxon type of all spicules is the mechanical result of
this fact. In the first place it may be pointed out that this
assumption remains to be proved, for Calcarea at any rate. It
is by no means impossible that the spicule at its first appear-
ance is deposited in something corresponding to a vacuole, and
if so, Dreyer’s explanation would no longer apply. We may

MATERIALS FOR A MONOGRAPH OF THE ASOONS. 553

pass over this point for the present, however, and assume with
Dreyer that the first appearance of the sclerite takes place in a
node of the framework ; in other words, that the minute sclerite
has the value of a granule, and is to be classed as such.
Now the granules commonly found in protoplasm are very
variable, as has been said,in nature and consistency ; but they
always present themselves, whatever their size, as rounded
bodies of a more or less spherical form. In spite of their posi-
tion at the nodes of a framework, where, as Dreyer assumes,
they would be subject to ‘‘ vesicular tension,” no granules
have been recorded, to my knowledge, which show any approach
to the supposed primitive tetraxon form. If the sclerites have,
as a result of their position at the nodes of the framework, a
peculiar form and shape which all the other granules occupy-
ing the same position conspicuously lack, it follows that there
must be some other cause to produce this anomalous result
than their position simply. If vesicular tension cannot in any
other instance cause the granules at the nodes to assume a
tetraxon form, why should it do so in the case of the sclerites ?
If now we try to find a distinction between the sclerites and
the other granules which would account for their difference of
form, we can find none but the fact that the sclerites have, ex
hy pothesi, a supporting or skeletal function. If, however,
their form be correlated not with their position, but with their
function, we are thrown back at once, as it seems to me, to
seek an explanation for the alleged primitive tetraxon, not in
the mechanical results of its position and origin, but in adap-
tation to its surroundings,—in fact, to just the theory
which Dreyer rejects.

In any case there seems no reason why the sclerite should
have had a tetraxon form, unless it had at the same time a
supporting function. It is extremely probable, however, that
the material of which skeletons of any kind are built up
appeared at first as mere concretions, or perhaps excretions,—in
fact, as products, and perhaps as waste products, of the meta-
bolism, before they were utilised for any function. If such a
view be admitted, there is then no reason at all why the

554 E. A. MINCHIN.

sclerites should not at their first appearance have been similar
in form to any other cell granules; that is to say, simply little
lumps of rounded or irregular, or perhaps of crystalline form.

Whatever may have been the course of evolution, the sclerites
must at first have been extremely minute, and no sooner did they
appear than there was at once an opening, so to speak, for the
action of adaptive and selective forces, related to the life of the
organism as a whole, as apart from the molecular or mechanical
forces acting upon the sclerite itself. Dreyer’s hypothetical
tetraxon sclerite, if it were really the fundamental spicular
form, could only be explained, it seems to me, in this way.
Natural selection, we will admit and proclaim, is not an
“internally formative” principle; it could not have caused
the first appearance of sclerites, which must be attributed
primarily to variations in the cell metabolism. But natural
selection is, we are told, an “ externally regulative ” force ;
and given a simple sclerite of whatever form, there is imme-
diately something upon which external regulation can be exer-
cised ; but there is no reason to postulate that the primitive
sclerite should be of the tetraxon rather than of any other
type.

In the foregoing discussion of Dreyer’s theory I have
endeavoured to follow up his conclusions entirely from his own
premises ; and in order not to confuse the argument, I have
purposely refrained from mentioning one constant accompani-
ment of the spicules of calcareous sponges which seems to me
to make Dreyer’s theory altogether inapplicable to that group
at least. I refer to the spicule sheath, which is nothing less
than an envelope of structureless organic material, inserted
between the spicule and its secreting cell.

All the authors who have expressed any opinion as to the
spicule sheath, namely, Haeckel, Schulze, and Dendy (see
above), agree in regarding it as a layer derived from the
general gelatinous ground substance condensed round the
spicule. But since the sheath appears between the spicule
and its secreting cell, it obviously cannot be derived from the
ground substance outside the cell. It may, however, be an

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 555

intra-cellular secretion of the cell of the same nature as the
intercellular secretion which produces the gelatinous matrix of
the dermal layer. In the absence of precise observations on
this point, my statements with regard to the spicule sheath are
to be taken as provisional ; but I think it highly probable that
the spicule sheath represents a matrix secreted by the cell in
which the spicule is subsequently deposited. I imagine to
myself the spicule mother-cell secreting a minute drop or
vacuole of the substance, whatever it may be, of which the
sheath is composed, and in this vacuole the sclerite appearing
as a minute concretion. The sclerite may have had originally
a perfectly crystalline form, as Haeckel supposes, or it may
have been a simple globular concretion like the crystalline
substances deposited in an organic matrix in Rainey’s* inter-
esting experiments. In either case it early assumed a definite
form in correlation with the functions imposed upon it, and as
it grew into a spicule, such as we are familiar with, the matrix
in which it was first deposited grew with it to form the
sheath.

If this interpretation of the spicule sheath be correct, it is
obvious that the calcareous sclerite cannot be regarded, in
sponges at least, as having the value of a granule deposited at
a node of the protoplasmic framework, but rather as a simple
concretion within a vacuole, not subject to any force of the
nature of vesicular tension. This representation of the mode
in which the spicule is deposited is of course purely specu-
lative, and lacks as yet any support from actual observation
in the case of sponges. There are, however, instances of

sclerites deposited in this manner in both animal and vege-
table cells.”

1 ‘British and Foreign Medico-Chirurg. Review,’ xx, 1857, pp. 451—476;
see also Harting, ‘ Quart. Journ. Micr. Sci.’ (n. s.), xii, 1872, pp. 118—123
and Ord, ibid., pp. 219—239, pls. xv, xvi.

* Compare, for instance, Leger’s observations on the clinorhombie crystals
of calcium oxalate in the cysts of Lithocystis Schneideri (‘ Ann.
Mag. Nat. Hist.’ [6], xviii, 1895, p. 479). Sclerites are figured in cells by
Semon (‘ Mitth. Zool. Stat. Neapel,’ vii, pl. ix, fig. 3) and Blochmann (‘ Die
Epithel-Frage bei Cestoden u. Trematoden,’ Hamburg, 1896), but neither

556 EH. A. MINCHIN.

For all these reasons I am most decidedly of opinion that
Dreyer’s theory of a primitive tetraxon spicule, produced by
the influence of vesicular tension upon a sclerite deposited at a
nodal point of the protoplasmic framework, must be rejected
for calcareous sponges at least. My own observations upon
the formation of the spicules seem to me to indicate clearly
that the primitive form of skeletal element in the
Calcarea was a rod-shaped or fusiform sclerite ; that
the triradiate system has arisen from a junction
of three such simple elements; and that the quadri-
radiate system, the last stage in the evolution, and
not the first, was further built up by the addition of
yet another monaxon sclerite to the triradiate
system. The primitive monaxon sclerite probably differed in
one point from the ordinary monaxon spicules that we are familiar
with in existing Calcarea. The latter always project free from
the surface of the sponge, only the proximal portion of the
spicule being embedded in the sponge wall. In the triradiate
systems, on the other hand, the rays are placed longitudinally,
and completely embedded in the body-wall, at least in all the
more primitive Ascons. Hence it is probable that the ances-
tral monaxon spicules lay tangentially in the body-wall, and
did not project from it. The spicules of this character under-
went two divergent courses of evolution. Some remained
single, but acquired a portion projecting free from the surface
of the dermal layer, becoming the existing monaxon spicules.
Others again acquired no such projecting portion, but, remain-
ing completely embedded in the body-wall, became united with
one another in trios to form the triradiate spicules as we know

author notices their relation to the protoplasmic structure. The formation of
sclerites in plants—the so-called raphides—offers the best instance to the
point. Vacuoles of mucilage arise in the cell protoplasm, and run together to
form a large vacuole in which the raphides are deposited as crystals. See
Haberlandt, ‘Physiologische Pflanzenanatomie (Leipzig, 1896), p. 449, figs.
190, 191; de Bary, ‘Comparative Anatomy of the Phanerogams and Ferns’
(Oxford, 1884), pp. 1837—139 ; Gardner and Ito, “On the Structure of the
Mucilage-secreting Cells, &c.,’’ ‘Annals of Botany,’ i (1887), pp. 27—54,
pls. iii, iv.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 557

them. In this way arose from a common source the two
primitive types of spicule which we find in the Calcarea.

Before proceeding to discuss the factors which may be sup-
posed to have brought about this course of evolution, I will
first examine all the objections which occur to me against
my theory of the origin of the triradiate and quadriradiate
system by fusion of skeletal elements primitively distinct.

1. A great objection, at first sight, to regarding the many
rayed calcareous spicules as compound bodies, is the well-
established fact that they behave optically as single crystals
(Sollas and Ebner, see above). ‘This fact certainly seems
strong evidence for regarding them as simple and uniform
bodies. But it may be pointed out that, whatever their
ancestral history, each ray does actually arise from a separate
mother-cell in the course of the development. It is just as
difficult to comprehend why the united products of a number
of independent cells should behave as a single crystal, as to
believe that a body of this kind arose ancestrally from a fusion
of separate elements. This is especially so in the case of the
quadriradiate systems, where the fourth ray has an origin
quite distinct from the basal system, and is even formed at a
later period. I am justified therefore, I think, in claiming
that my theory of a fusion of skeletal elements to form a
structure which behaves as a single crystal individual, does
not present greater difficulties than do the ascertained facts of
development.

2. It might be urged against the views here advocated, that
no other sponge spicules are known to arise by fusion of
separate elements. This is an argument which at most would
only excite a bias against the theory, but could not afford a
positive disproof of it. It is certainly true that, so far as is
known, the polyaxon spicules of other sponges arise always
each in a single cell. On the other hand, instances of secon-
dary deposit of skeletal material uniting primitively distinct
spicules are not wanting. We may cite the case of the
compound desmas of the Lithistida; the secondary siliceous

’ For further remarks on this point see Addendum B (infra, p. 572).

558 E. A, MINCHIN.

deposits uniting the spicules in the Dictyonina; and finally
the secondary deposits of calcite uniting the calcareous spicules
in the Pharetrones, of which Déderlein has so recently (1897)
described a living example in his Petrostoma Schulzei.
These examples are sufficient to show, were evidence required,
that spicules may become united together secondarily to form
compound structures.

3. A quite trivial difference, as it seems to me, can be pointed
out between the formation of a ray of a triradiate, and of a
primary monaxon spicule. While in the former the actinoblast
divides, in the latter, so far as is known, the mother-cell
remains single. But the history of the gastral ray shows
plainly that it depends on the size of the spicule to be formed,
whether or no the mother-cell remains undivided. And for
my part I much doubt whether, in the case of primary mon-
axons which attain a large size, the secreting cell always
remains single.

None of these objections are in any way a serious obstacle,
it seems to me, to regarding the triradiate and quadriradiate
systems not as true or primary spicules, but as compound
spicular systems. ‘To my mind it is essential to the definition
of a true spicule or primary spicular element that it should be
formed by a single mother-cell, or by the descendants of one
such cell. In this way each ray of a triradiate system repre-
sents a single spicule, homologous with a primary monaxon,
and the whole structure is a compound or secondary spicular
system. Examples of primary siliceous spicules are the ordi-
nary monaxons and tetraxons which arise each in a single cell,
while the desmas of Lithistida are instances of secondary
systems.

This view of the origin of the triradiate systems completes,
as it seems to me, and extends Schulze’s theory that they arose
as an adaptation to the structure of the body-wall of the
simplest Calcarea. The primitive monaxon sclerites lying in
the body-wall would soon take on an arrangement suited to its
structure, and came to lie probably one between every pair of
adjacent pores. Then, by union of the monaxons in trios, the

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 559

characteristic arrangement of the triradiate systems was brought
about which we may still observe in the olynthus of any cal-
careous sponge, or in the oscular tube of the Ascons (Fig. B).

from three monaxon spicules (sp.) lying between the pores (P.)

If it be asked why only three sclerites should have fused
together, and not two or four or any other number, it is impos-
sible to say more than that the triradiate arrangement represents
a compromise between too great flexibility on the one hand,
and too great brittleness on the other. Such a question would,
in fact, be as difficult to auswer off-hand as to say why insects
have only six legs, or mammals seven cervical vertebre. We
can see that a skeleton of monaxon sclerites would afford little
support to an erect oscular tube, while the union of many such
sclerites into a lattice-work would result, in the case of objects
of the size of a calcareous sponge, in an excessively fragile
structure.

The theory that the triradiate systems arose in this way, as
an adaptation to the structure of the body-wall, seems to me to
explain not only why they should be formed, as we see they
are formed, of three simple monaxons, but also their extra-
ordinary and striking regularity. It is the old problem of the

560 E. A. MINCHIN.

honeycomb over again. I can understand, however, that an
attempt might be made to explain this characteristic regularity
in a different manner. It might be urged, since each spicule
is formed by three cells of more or less equal size, closely
pressed together, that therefore the rays would naturally meet at
angles of 120° as the result of their being formed in the axis of
the cell; but after carefully weighing this theory I am obliged
to reject it as an adequate explanation, for the following
reasons. First, because the actinoblasts are never exactly equal
or perfectly regularly placed, nor are the rays formed exactly
in the axis of the cell, but almost always a little to one side or
the other; hence if that were the only factor at work, we
should rather expect irregularity to be the rule, and equality
between the angles to be a rare exception. Secondly, any such
explanation, even if accepted for the spicules of the Clathri-
nide, which have the three rays meeting at equal angles,
would fail completely in Leucosoleniz, where we find indeed
great regularity, but cf a different type. In Leucosolenia
the triradiate systems are of a pronounced bilateral type, with
two curved lateral rays, each meeting the straight posterior ray
at an angle much less than 120°, and enclosing anteriorly an
angle much greater. We may express it by saying that while
each lateral angle is, in a projection (120—vz)°, the anterior
angle is (120+2mn)°. Now if the orientation of the rays is to
be referred to the arrangement of the actinoblasts, the bilateral
form of the triradiate systems of Leucosolenia could only be
explained by a corresponding arrangement or difference in size
of the actinoblasts, which would bring us to an explanation
which would itself require to be explained.!

I cannot, therefore, regard the regularity of the triradiate
systems as due to the arrangement of the secreting cells. On
the other hand, the view that this, their most striking charac-
teristic, arises from adaptation to the structural requirements

1 So also Ludwig, discussing the calcareous bodies in the Holothurians,
refers the branching at angles of 120° to the arrangement of the cells
(“ Echinodermen,” I, in ‘ Bronn’s Thierreich,’ ii, 3, p.60). But what ex-
plains the arrangement of the cells?

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 561

of the sponge as a whole, receives further support from the
fact mentioned above, namely, that in the case of the triradiate
systems first secreted after the fixation of the larva, extreme
irregularity prevails. The sponge has not at this stage assumed
its cylindrical form, but is a compact mass of cells; the spicules
are formed in the superficial layer just as in the adult, and
were their regular form due to the arrangement of the secreting
cells, there is no reason why the rays should not at this early
stage meet at the same constant angles as in the adult. It is
not, however, until the young sponge has assumed its charac-
teristic tubular form that regularity becomes the rule, and not
the exception, amongst its spicules. I can only interpret these
facts by supposing that the forms of the spicules are influenced
by the requirements of the sponge as a whole, and not by the
arrangement of its cells, so that the spicules do not attain
their characteristic form until the sponge itself has the structure
to which that form is an adaptation. And this conclusion
seems to me still further supported by the above-mentioned
differences between Clathrina and Leucosolenia. Inthe
former the reticulate form of the sponge colony is associated
with the possession of equiangular triradiates, which in the
latter become bilateral in correlation with the upright form of
the sponge as a whole.

If the triradiate systems furnish a strong argument, as I
believe they do, for the influence of adaptation on the spicules,
the course of events in the growth of a quadriradiate system
greatly strengthens this impression. We have seen that after a
triradiate system has been laid down in the usual way, the
gastral ray is tacked on to it from quite a different source,
namely, by a porocyte. We have seen, further, that the poro-
cytes come from the dermal epithelium, and we know that in
Leuycosolenia each cell of this epithelium can secrete a
monaxon spicule. Hence there is no difficulty in supposing
that the porocyte in secreting a monaxon spicular element is
simply exercising a function which was primitively possessed
by every cell of the dermal layer. We can invoke the aid of
such well-known principles as heredity, atavism, and even

562 E. A. MINCHIN.

metabolism, to explain why the porocyte secretes the gastral
ray. But none of all these factors of evolution can explain
why the presence of a young triradiate system in the vicinity
should stimulate the porocyte to resume its long-forgotten
functions, nor, above all, why it should not only secrete a
monaxon spicule, but should go so far as to stick it on to the
triradiate system. The formation of the gastral ray is, indeed,
a most crucial test of the theories of spicular origin, and all
simply mechanical theories are at once, as it were, impaled
upon its relentless point.

We have in the foregoing pages considered some very
different hypotheses in our search for the principle which
determined and modelled the fundamental types of the cal-
careous sponge spicule. We have examined the rival, and
often conflicting claims of crystallisation on the one hand, and
of the mechanical effects that might be supposed to follow
from the ultimate structure of protoplasm, from the dynamics
of the cell, or from the number and arrangement of the secret-
ing cells on the other hand. None of these theories, however,
stand the test of a thorough analysis, or can supply more than
a part of the explanation. All the evidence which we have at
our disposal drives us to seek the principle which guided and
directed the evolution of the spicular forms in a process of

1 While the general method in which the spicules arise seems to me explica-
ble only by adaptation, there are characters in all of them which are certainly
difficult to explain in this way. There are the peculiarities which separate and
distinguish the species, and which consist in small details of relative size,
length, curvature, or sharpness of the rays. A peculiarity of this kind in the
quadriradiates of Leucosolenia complicata, Mont. (= Ascandra com-
plicata, H., + A. pinus, H.), is worth mentioning in this connection, since
it seems to have escaped notice hitherto. In this species the gastral ray does
not arise from the junction of the three basal rays, but from the straight
posterior (unpaired) ray near its base—a fact which would be difficult to
reconcile with the theory of a primitive tetraxon spicule. Specific characters
of this kind are difficult to explain by any principle which requires that they
should be of utility to the organism, and they seem rather to have arisen by
the perpetuation in some unexplained way of a sport or variation not in itself
“useful” to the whole sponge. This is, however, no place to raise the much-
disputed question of the utility of specific characters.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 563

adaptation to the needs of the whole organism, resulting in
the form best suited for the function which the sclerite has to
perform —the function, namely, of supporting the tissues in the
manner required by their structure and arrangement in the
primitive forms in which the spicules first arose.

In explaining the main types of calcareous spicules by
adaptation, we are referring them to a very wide-spread and
universal category of structures characteristic of living beings.
There is probably no living organism which does not possess
some special features, owing their characteristics or even their ex-
istence to the fact that they are of utility to the organism in its
efforts to maintain and perpetuate its existence in the way in
which its environment necessitates. If now it be asked how
this process of adaptation came about, we find ourselves con-
fronted with one of the most important and fundamental ques-
tions of evolution. The most commonly received explanation
of adaptation is that put forward by Darwin—namely, that it
is due to the natural selection of favorable varieties in the
struggle for existence, resulting in the survival of the fittest
types and the gradual elimination of the less perfectly
adapted.

Very recently, however, Delage (1895) has put forward a
very different view. Natural selection, according to him, is
not a factor of progress in evolution, but one which tends to
maintain the fixity of the species by elimination of harmful
variations. Adaptation, we are told, is the rock upon which
both Darwinian and Lamarckian hypotheses founder, since
natural selection is an inadequate explanation, and acquired
characters are not inherited (p. 839). There is no phylo-
genetic adaptation ; that is to say, species are not adapted to
their mode of life, but it is the individual which becomes
adapted by its reaction to its environment— auto-regula-
tion ” (p. 828). ‘ Phylogeny creates organs without regard to
function; ontogeny . . . adapts them to the necessary func-
tions :” in other words, “in phylogeny it is the organ which
makes the function ; in ontogeny it is the function which makes
the organ” (p. 831). The cause of this ontogenetic adapta-

564 E. A. MINCHIN.

tion is to be sought in “ functional excitation,” under the
influence of which “the individuals adapt themselves regu-
larly, without interruption, and in all their organs” (p. 828).

Space would not permit of a thorough discussion of all the
questions raised by these propositions, but it is a legitimate
task to discuss here how far the ontogeny and phylogeny of
the spicules support or refute Delage’s views, and to test them
by the facts of spicule development. At the outset some con-
sequences may be pointed out which follow from his theories.
If the adaptation be always ontogenetic, and due to a reaction
on the part of the individual organism to a functional excita-
tion, then an adaptive structure might be expected to arise
only when the conditions to which it is an adaptation have come
into operation. If, on the other hand, the adaptation is phylo-
genetic, acquired in the past history of the evolution of the
species, and fixed and handed on by heredity, there would be
nothing astonishing in finding that an adaptation shows itself
before it is functional or useful. A prophetic adaptation of
this kind would, however, be fatal, it seems to me, to the view
that adaptation is always ontogenetic.

To return now to our spicules. We believe that in phylo-
geny they arose by the union of three separate monaxons,
which first came to lie one between each pair of pores, and
then fused together in trios. A consideration of the structure
of the wall shows at once how, when union took place, they
met at quite regular angles of 120°. So far all is simple and
natural, and there is nothing to contradict the view that in
phylogeny the regular forms of the triradiate spicules were pro-
duced by “ auto-regulation ” in response to functional excita-
tion. The matter is, however, quite otherwise when we come
to the ontogeny. Whatever may have been the past history, the
three spicule rays are not laid down at the present day first as
three full-sized monaxon spicules, and then secondarily united
into one system. The supposition that they ever were formed
in this way is purely theoretical. They are, on the contrary,
united to form a triradiate system when still excessively
minute and enclosed within their secreting cells. This

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 565

being so, it is difficult to see how the functional excitation,
so feasible in phylogeny, can influence the ontogeny, since
the triradiate systems have their characteristic and definite
form long before they are able to exercise their function of
support. The adaptation of their form to the structure of
the sponge is therefore entirely prophetic. The only way in
which functional excitation could modify their form would
be to produce curvature of the rays at a late period of their
growth. Such a curvature is found normally in the spicule
rays of Leucosolenia, and exceptionally in Clathrina, and
is perhaps the result of a functional excitation exercised as
follows. If we look at a young triradiate system surrounded
by its six cells (compare especially P1. 38, fig. 8, Pl. 39, fig. 21,
and Pl. 41, fig. 38), it is impossible to avoid the idea that the
apical formative cell exerts in some way a directive function
in laying down the ray. Going ahead at the extreme tip of
the ray, it seems to mark out its future direction. Of the two
formative cells, the basal cell might be termed the secretive
cell, the apical cell the directive element. If now an apical
formative cell, acting in this way, were deflected from its path
and turned to one side by a pore or other tissue element, a
curved ray would result. I hope at some future time to be
able to bring forward observations on this point, and especially
to trace the history of abnormal spicules, and investigate the
causes of their abnormality. In any case the functional
adaptation could only produce, as has been said, a late curva-
ture of the rays, and could not influence the fundamental
angles at which they meet, for the latter are determined at a
very early period within the secreting cells. The apical for-
mative cell might represent a medium for auto-regulation
and ontogenetic adaptation, but in the case of spicules which
lose their apical cells at an early period, variations in the
curvature of the rays due to a functional stimulus would be
impossible. The spicule must, it would seem, in such cases
grow straight on, like the incisor of a rodent, which grows
into its brain if not worn down, and if a pore is in the way of
the spicule it is the pore which must shift its position.
vou. 40, part 4.—NEW SER. RR

566 E. A. MINCHIN.

The ontogeny of the spicules, therefore, points clearly to
their regular form being a phylogenetic adaptation, which has
become fixed and handed on by heredity, appearing in the
ontogeny as a prophetic adaptation. Nowhere is this fact
more evident than in the rim of a growing oscular tube of
an Ascon, since in this region we find triradiate systems
being laid down and even attaining to their full size in a
perfectly regular manner before any pores are formed. Ata
later period, as the gastral layer grows up, pores are formed
amongst and between them, but the spicules appear first, long
before the wall has the structure to which we find that their
form is an adaptation. This is true equally of the triradiate
systems of Clathrina with straight, and those of Leucoso-
lenia with curved rays, showing that neither form can be
explained as produced by functional excitation during the
ontogeny.

The facts already mentioned with regard to the development
of the embryo show that the hereditary impulse or tendency
does not come into full play until the sponge as a whole has the
structure to which the triradiate system is an adaptation, and
that the spicules first formed are very irregular in consequence.
But in spite of this influence exerted by the structure of the
whole organism, it is not possible to connect the regularity of
the spicules with any functional excitation in the ontogeny,
though this is a possible explanation of the phylogeny. But
at this stage of the inquiry we find ourselves confronted bya
dilemma. If functional excitation produced the adaptation in
phylogeny, then it is a case of an acquired character being
inherited. If, on the other hand, acquired characters be not
inherited, then the adaptation must have arisen from innate or
constitutional variations,which became gradually fixed, doubtless
by selection, as a constant hereditary tendency of the organism.

If, therefore, we wish to avoid the hypothesis, not as yet
established by a single clear instance, of the transmission of
acquired characters, we are brought back, it seems to me, to
the old familiar theory of natural selection, in order to explain
the constant recurrence and regular form of the triradiate

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 567

system, and for the operation of this principle the spicules
offer a splendid field. Few structures are more variable than
spicules. In every specimen a certain percentage of irregular,
or aS we say abnormal spicules are to be found, and sometimes
quite a high percentage of them.! Ascons, again, are for the
most part shore forms, and nothing gives a more vivid idea of
what the struggle for existence means than, after becoming
familiar with the fragile and delicate structure of these organ-
isms, to watch a stormy sea breaking over the rocks upon
which one knows that Ascons are actually living and growing.
The wonder is rather that any specimen ever survives. That
natural causes produce a high death-rate among these crea-
tures can be further deduced from the fact that a single
specimen produces an enormous number of larve, and as the
latter exercise but little choice as to the localities in which
they fix themselves, a very large proportion must be wiped out
of existence by the first storm that breaks over them. All
these arguments are perhaps somewhat trite, but they may
serve to emphasise the fact that the field is in every way a
favorable one for the operation of natural selection. There is
variation to supply the material for selection, and there is an
environment calculated in every way to favour the survival of
individuals distinguished by even slight improvements in the
supporting framework. Into the question of the origin of the
variations on the one hand, and of how the characters of the
fittest, after they have survived, are fixed and become per-
petuated on the other hand, this is no place to enter. If,
however, the application of the principle of natural selection
raises some difficulties in the present state of our knowledge,
especially as regards matters of detail, the same is true toa
much greater degree of any other theory. Many of the objec-
tions so often brought forward against the theory of natural
selection seem to me to amount simply to the following argu-
ment: there are many facts which natural selection cannot
explain, therefore natural selection explains nothing. For my

1 Compare also the facts brought forward above (pp. 521—528) with
regard to variation in the gastral rays.

568 E. A. MINCHIN.

part I find a great difficulty in explaining some facts by natural
selection, but I am none the less of opinion that, in other
cases, it is the only possible explanation.

SUMMARY.

1. The first appearance of a calcareous spicule or spicular
element, both ancestrally and in the actual development, was
probably a minute vacuole in a cell of the dermal layer, filled
with an organic substance perhaps identical with the inter-
cellular ground substance, within which the minute sclerite
appeared as a crystal or concretion.

2. The ancestral sclerite, though crystalline in structure,
soon assumed a non-crystalline form as a whole, as an adapta-
tion to its secondarily acquired function of support, and as it
grew in size the contents of the vacuole formed the spicule
sheath.

3. The ancestral form of spicule in the Calcarea was a
simple monaxon, placed tangentially and completely embedded
in the body-wall, lying between two adjacent pores.

4, From this ancestral spicule the forms of spicule now
occurring in the Calcarea arose as follows: (a) the primitive
monaxon acquired a distal portion projecting from the surface,
as in the existing primary monaxons; (4) groups consisting
each of three primitive monaxons became united by their con-
tiguous ends to form a single triradiate system; (c) tosome
of the triradiate systems thus formed a fourth ray was added,
secreted by the pore-cell, giving rise to the quadriradiate
system; (d) some of the triradiate systems, by loss of one ray
and placing of the other two in a straight line, or by loss of
two rays, perhaps became modified into secondary monaxon
spicules.

5. The power of secreting a monaxon sclerite was primi-
tively possessed by every cell of the dermal layer, and this
condition appears to be retained in Leucosolenia. In
Clathrina, on the other hand, all the skeletogenous cells
migrate inwards from the dermal epithelium, and form a con-

MATERIALS FOR A MONOGRAPH OF THE ASOCONS. 569

nective-tissue layer distinct in function from the contractile,
undifferentiated dermal epithelium. In Leucosolenia also
the actinoblasts of the triradiate systems form a deeper layer,
but the dermal epithelium secretes primary monaxons—at
least in the young form—and is non-contractile.

6. The forms of the spicules are the result of adaptation to
the requirements of the sponge as a whole, produced by the
action of natural selection upon variation in every direction.

ADDENDA.

A. On the Presence of an Axial Organic Filament
in the Spicule Rays.

(Note to p. 543.)

Since the above paper went to press I have succeeded in
demonstrating very clearly the existence of an axial organic
thread in the spicule rays by means of the following staining
mixture, recommended to me by my friend Dr. Ritchie, Lec-
turer on Pathology in the University of Oxford:

Nigrosin, 1 per cent. in H,O. 3 - 1 volume.
Picric acid, sat. sol. in H,O . i . 9 volumes.

The action of this mixture upon the sponge tissue is that, in
the first place, the calcite of the spicules is rapidly dissolved by
the picric acid, which also colours the cytoplasm of the cells
yellowish, while the nigrosin stains the ground substance a
faint blue and the spicule sheaths deep blue, and also colours
a very delicate filament in the axis of each spicule ray.

For showing the filaments, sections of material which has
been decalcified before embedding are almost useless ; on the
other hand, the filaments are seen very well in sections not
previously decalcified after treatment with the nigrosin mixture,
especially in fairly thick sections, in which it is possible to find
large pieces or whole rays of the spicules. I attribute this
difference to the fact that after decalcification the filaments are
quite unsupported within the spicule sheaths, and are readily

570 E. A. MINCHIN.

displaced from their position; even in the most successful
preparations the filaments have a more or less undulating
appearance (Fig. C), and show signs of displacement. The
processes which a piece of tissue goes through in its passage
from alcohol to paraffin are such as to set up a great many
diffusion currents in the tissues, and in consequence, if the
spicules are decalcified, the axial filaments are in nearly every
case displaced and stuck against the spicule sheaths, and
undistinguishable from them after staining.

The simplest method of demonstrating the axial filaments is
the following. Take any spirit specimen of an Ascon, and put
a small piece—a tube or part of one—first in water and then
in the staining mixture for about half an hour. Then wash
thoroughly in water, and pass through alcohol to clove oil.
In the latter medium the tube can be split open with needles,
and the collar-cells removed with a paint-brush, which generally
brings away the pore-cells as well. Then the piece of the wall
is laid out flat on a slide, and mounted in Canada balsam. I
have never failed to demonstrate the axial filaments in the
clearest manner in such preparations.

T also made the attempt to unmount and stain some of my
preparations mounted in glycerine, but without great success.
The stain in these cases acted feebly, and the delicate filaments
could only be made out with great difficulty. On the other
hand, the sheaths were well shown, and it could be seen that
they appear very early in the development of the spicule when
it is still quite enveloped in its cells. At first the sheath forms
an excessively thin layer, which gets thicker as the development
proceeds. In such a stage as that shown in Pl. 38, fig. 7, the
sheath is perfectly distinct when stained, a fact that leaves
no doubt as to its being, like the spicule, an intra-cellular
secretion.

The axial filament runs to the extreme tip of the spicule ray,
and joins the spicule sheath (woodcut, Fig. C), showing that
the spicules must be perforated at the tips of the rays. The
filaments are more distinct towards the tips of the rays than at
the bases, and the behaviour of the filaments at the centre of

MATERIALS FOR A MONOGRAPH OF THE ASOCONS. 571

the spicule is often very difficult to make out; but in coriacea,
which I studied carefully, I was able to see clearly the state of
things in a large number of instances. As the filaments ap-
proach the centre they swell out slightly and become less

Fic. C.—Tracing with the camera lucida of the spicule sheath (sp. sf.) and
axial filament (az. f.) of Clathrina coriacea, after treatment with
nigrosin and picric acid. x 800.

definite ; finally they end at a short distance apart, and when
seen with insufficient magnification appear independent of
each other, but with a high power they are seen to be con-
nected by three delicate curved filaments, forming a minute
triangle enclosed by curved lines, concave towards the ex-
terior.

This structure was seen by me in all the spicules in which I
could make out clearly the central portions of the filaments ;
but considerable variations were observable in the dimensions
of the central triangle, and in some cases it was so small that
the filaments appeared to touch each other. I believe that the
central triangle is easily explained by the development of the
spicule, and the fact that the rays are at first separate from
one another. The abrupt termination of the filaments before
they meet represents the distance between the rays at their
first origin, and the delicate connecting threads represent the
secondary union of the three sclerites. I have not had time
yet to make extended observations upon the union of the fila-
ments in different species, as I hope to do, but I have made
and examined a preparation of reticulum, a species which

572 E. A. MINCHIN.

seemed likely to yield interesting results, since it is a form in
which the rays unite at a very early period (see Addendum B,
infra). The preparation showed clearly filaments in the mon-
axons and in the gastral rays, but with regard to the latter I
was not able to make out clearly in surface view their relations
to the filaments in the basal triradiate system. In the tri-
radiates which I examined I could only find one instance
showing a distinct, but small, central triangle. In all the
other cases the filaments appeared to meet at the centre, but
closer observation showed that they were united by, or were in
contact with, a distinct central granule which stained a little
more deeply than the contiguous ends of the filaments. I
think this observation, so far as it goes, confirms my view of
the significance of the central triangle.

In any case I think these observations afford the most con-
clusive proof of the existence of an organic axial filament in
the calcareous sponge spicule. Haeckel (1872) and Sollas
(1885) affirmed the existence of this structure, while Ebner
(1887) denied it. Lendenfeld (1885) gave a most detailed and
circumstantial description of the axial filaments (see above,
p- 538), but I find it difficult to believe that he saw anything
of what he described; for when Ebner cast doubts upon the
existence of an organic axis to the spicule, Lendenfeld in his
very next paper on the subject (1891, p. 382) acquiesced in all
Ebner’s statements without a protest, not even so much as
mentioning that he or any one else had ever described such a
structure.

B. On Compound Spicular Systems behaving as
Single Crystals.

(Note to p. 557.)

With regard to the apparent paradox which is presented by
the triradiate and quadriradiate spicules, namely, the fact that
while on the one hand they are bodies of multiple origin, they
behave on the other hand as single crystals, I am indebted

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 573

for some valuable suggestions to my friend Mr. H. A. Miers,
Fellow of Magdalen College and Professor of Mineralogy in
the University of Oxford. Professor Miers pointed out to me
that there were two possibilities as regards the early develop-
ment of the spicules; either that the rays when first secreted
are of a non-crystalline nature, aud become crystalline as the
result of contact and fusion; or that the rays are crystalline
from the first. In the latter case there would be two further
alternatives to be considered, namely, that the rays might be
at first three separate crystals independently orientated, or
that the rays, even when distinct from each other, might
possess a molecular structure similarly orientated throughout
the whole system.

Since these possibilities can be tested without difficulty by
examining very young spicules between the nicols of a polar-
ising microscope, I have searched through my preparations for
the early stages of the development of the spicules. The pre-
parations which are figured in this paper, being mounted in
glycerine, and now for the most part more than two years old,
are so much corroded by the glycerine that the youngest stages
are almost all dissolved entirely ; but I have found a number of
young spicules in preparations mounted in Canada balsam,
which are scarcely or not at all corroded. Besides three
minute spicules in preparations of the adult coriacea, I have
examined an embryo of contorta, two embryos of cere-
brum, one of falcata, and one of reticulum. Of these
only the embryos of cerebrum showed slight traces of cor-
rosion ; in the other embryos the spicules did not seem in the
least damaged, though the preparations are eighteen months
old. Slight corrosion, it may be pointed out, would not affect
the crystallographic results, so long as the spicule be not com-
pletely dissolved; but in the case of the embryos of cere-
brum it affects the question of the separation of the rays.
Since the parts last deposited seem to be the first to be cor-
roded, it is possible that in my preparations of cerebrum
spicule rays which appear distinct had really become united
and then separated again by corrosion, If so, my results with

574 E. A. MINCHIN.

regard to the time at which the rays become united, as mea-
sured by the length of the rays, would require correction.

The results of my investigations, in which I have to acknow-
ledge much valuable assistance from Professor Miers, will be
found below in tabulatory form; but I will first briefly indicate
the conclusions to be drawn from them. The chief point of
interest which I have discovered is that the spicule rays
are not crystalline when first laid down; so long as
the rays are quite separate they will not light up when rotated
between crossed nicols.’ After union of the rays they gradually
become crystalline, the change appearing to start from the
centre, and in fact from the secondary deposit of calcite by
which the rays are united together. I have seen many appear-
ances which indicate that the rays are united by a disc or
globule of calcite deposited at the centre over the apposed
extremities of the rays. In some preparations of cerebrum,
viewed from the dermal surface, one sees at a lower focus
the interspaces between the rays; at a higher focus a circular
piece of calcite comes into view, sometimes so distinct that it
looks like the rudiment of another ray. This central portion
seems to be in many cases the part that first lights up between
crossed prisms. On the other hand, spicules may be found
which have the rays completely joined, but which fail to show
any crystalline properties; others, again, light up so feebly
between the prisms that it is very difficult to detect the feeble
illumination, except by observing the slight degree to which it
waxes and wanes as the stage is rotated. On the other hand,
amongst and between spicules which remain dark there
occur others scarcely differing in size, or even smaller, which
light up with the greatest brilliancy, thus supplying a valuable
contrast.

The next result which merits special notice is the fact that
the period in the development at which the rays assume a
crystalline character varies greatly in different species, in

1 As this result would also be produced if the optic axis of the erystal

coincided with that of the microscope, I have, where it was possible, tilted
the preparation, and found that the spicules still remained dark,

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 575

different individuals of the same species, and in the different
classes of spicules in the same individual. The embryo of
contorta shows the latter point very well. On the third day
the numerous spicules fall into two distinct classes: (1)
those which have the rays more or less equal in length, and
meeting at more or less equal angles ; (2) those which have
two larger rays placed in the same straight line, or nearly so,
while the third ray may be only slightly smaller than the other
two, or it may be very much smaller, or finally it may be
absent altogether, in which case the whole spicule is simply
spindle-shaped, and is a true monaxon spicule. It is evident
that the spicules of the second class are the young forms of the
large monaxons which characterise this species, and that these
spicules are secondary monaxons derived from two rays of the
primitive triradiate.

In the embryo of contorta the rays seem in all cases to
unite when about 4 or 5 wu in length, or even earlier. The
spicules of the first class, the true triradiates, remain non-
crystalline for a much longer period, and only light up between
prisms when the rays have reached a length of 16 or 17 yu.
The spicules of the second class, on the other hand, the future
monaxons, become crystalline much earlier. I have not found
any which remained perfectly dark, while those with rays over
8 uw in length (i. e. with the monaxon shaft over 16 in length)
light up with the greatest brilliancy.

In cerebrum also two classes of spicules are early distinct,
plainly corresponding to the two classes just described in
contorta. Those of the first class, the regular triradiates, are
relatively more abundant. In one embryo examined (A) they
become crystalline when the rays measure on the average
about 13—15 m; in another (B) this event takes place earlier,
when the rays are about 10 or 12 in length. ‘The spicules
of the second class can be recognised as the young forms of the
horn-like spicules peculiar to this species. In the adult, though
variable, they have typically two thick rays, placed nearly in a
straight line at their bases, but curving like horns towards the
tips, while the third ray, which lies in a different plane, is

576 E, A. MINCHIN.

usually much smaller, and may be reduced to a mere nodule,
or may even be quite absent. In the latter case, which is of
very frequent occurrence, the spicule becomes simply a curved
monaxon. Lendenfeld has figured what I take to be some
spicules of this class (1891, pl. viii, figs. 3, c, d, e), but I cannot
say that his figures represent the forms of these spicules as I
am familiar with them. Bidder terms them tripods (1891,
p. 627). I regard these spicules, which show every variation
between two extreme forms, the one triradiate, the other
secondarily monaxon, as the homologues of the monaxons of
contorta, and like them they become crystalline much earlier
than the regular triradiates, and can be picked out in the
embryo by the brilliance with which they light up between
crossed prisms. While the spicules become crystalline at a
relatively late period in contorta and cerebrum, in falcata
they do so much earlier, so far as my observations go; the
rays are separate and non-crystalline till they have a length of
about 3 pu, then they unite and show crystalline properties.
In reticulum, finally, I have not been able to find any
spicules, however small (2 m) which have the rays either
separate or non-crystalline. The smallest visible spicules
glitter like stars between the prisms.

The comparison of different species seems to show that the
triradiate spicules are the product of a long course of evolution
in a certain direction, the furthest point in which has been
attained by reticulum. If the view which I have taken of
the phylogeny of the spicules be the true one,—if, that is to
say, they arose primitively from fusion of three or four separate
monaxon spicules,—it seems to me highly probable that
the ancestral triradiate or quadriradiate spicules, when they
first appeared, would not have behaved optically as single
crystals; and that the fact that they do so now is a secondary
character, the result of their ontogeny. Instead of now being
formed as three (or four) distinct spicules which fuse together
when full grown, the relations of the rays are in a certain
sense predestined before they are secreted. ‘They arise in
close contact, and fuse when still excessively minute ; hence

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 577
the fact, at first puzzling, that a compound spicular system
behaves as a single crystal.

These speculations are impossible to test, unfortunately,
unless by some happy chance an Ascon or calcareous sponge
should be found in which monaxon spicules, laid down and
built up separately, are joined together after attaining their
full growth, to form triradiate systems. But it would be
interesting to examine between nicols the compound masses
formed by fusion of spicules in such a form as Petrostoma
(Déderlein, 1897), in order to see whether the masses in
question behave as single crystals or show a number of crystal-
line centres corresponding to the number of spicules which
have been united.

As regards the significance, from the point of view of crystal-
lography, of the fact that the spicules at their first appearance
are not crystalline, I think I had better leave the explanation
to those who are more conversant with the facts and theories of
crystallisation than myself. I may, however, point out that
since an organic axis to the spicule ray has now been demon-
strated, it is evident that the greater part of the spicule ray at
its first appearance would be organic in its composition ; a
fact which may perhaps furnish a clue to the mystery—if there
be a mystery.

Summary of Addendum B.

(1) The rays are non-crystalline so long as they are distinct
from one another.

(2) They may remain non-crystalline for some time after
union has taken place.

(3) The crystallisation appears to start from the secondary
deposit which unites the rays at the centre.

(4) With regard to the period at which the rays become
crystalline, the species contorta, cerebrum, falcata, and
reticulum form a diminishing series, the last-named being
the species in which crystallisation sets in earliest.

(5) Those triradiate systems which, by hypertrophy of two

578 E. A. MINCHIN.

rays and diminution of the third, become modified to form the
secondary monaxons, become crystalline much earlier than the
more regular triradiates, especially as regards the two rays
which are placed in the same straight line to form the shaft
of the monaxon spicule.

Synopsis of Observations.

The numbers in brackets denote the length in p of the rays of the spicules
observed. Sometimes the rays vary in length, and the two extremes are
given. In the case of spicules of the second class in contorta and cere-
brum the length given is half the length of the shaft formed by the two
rays placed in a straight line.

s. denotes that the rays were separate, 7. that they joined; while those
spicules too much corroded to judge of this point are marked c.

(1) Coriacea, adult.—3 spicules examined. 2 remained dark (43.,
5 8.); 1 lit up feebly (6 7.).

(2) Contorta embryo.—26 spicules examined.

Class 1.
Perfectly dark.—8 (4, 63, 7, 12, 18, 13, 16, 18).
Faintly illuminated.—1 (17).
Fairly bright.—4 (16, 17, 17, 17).

Class 2.—13 examined ; all lit up.
Faintly.—3 (3, 64, 63).
Distinctly.—2 (43, 73).
The rest brilliantly.

(3) Cerebrum.—2 embryos.
Embryo A.
Class 1.
Perfectly dark.—13 (9 ¢., 11 s., ll ¢., ll ¢,12¢, 1388, 1458. 14¢,
14 ¢., 15 s., 15 ¢., 15—17 s., 16s.).
Faint illumination in centre.—8 (137., 137., 147., 147., 147., 147.,
14—16 7., 15 7.).
Rays faintly illuminated.—12 (137., 183—157., 14s., 147., 147.
14—16 j., 14—17 7., 15 7., 157., 15 7., 16 7., 16 7.).
Distinctly illuminated.—12 (13 7., 13—147., 14—167., 15 7., 15 j.,
U5 9.5 15 J 169., 169., Liga lige, Le 9)
Brightly illuminated.—2 (15 j., 17 j.).

1 Even when the spicule is perfect, it is often very difficult to be certain
whether the three clearer lines usually seen at the junction of the rays indicate
real interspaces, or lines of recent union.

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 579

In addition 1 was observed (13 s.) of which 2 rays lit up; 1 remained
dark.
Class 2.—6 observed, all of which lit up; 3 distinctly (14j., 15 7.,
15 j.) and 3 brilliantly (13 7., 147., 15 7.).
Embryo B.
Class 1.
Perfectly dark.—3 (10 s., 1O—12 s., 12 7.).
Faint illumination in centre—9 (9s. 9—11j., 10s. 11j., 117,
12 7., L11—18 7. [?] 187., 147.).
Rays faintly illuminated.—2 (10 7., 11 /.).
In addition a triradiate was observed with unequal rays of 11, 9, and 6 p;
the first ray lit up distinctly, the second faintly, the third not at all,
Class 2.
Centre bright.—2 (12 7., 15 7.).
Illumination fairly bright.—1 (11 j.).
Illumination brilliant.—6 (8 j., 9—10 j., 10 7., 10 7., 11 j., 11 j.).
(4) Falcata, embryo.—8 spicules examined.
Perfectly dark.—3 (23 s., 3.s., 3s.).
The remaining five were slightly larger, and appeared to have the rays
united ; they all lit up distinctly.
(5) Reticulum, embryo.—All the spicules, even the minutest (rays less
than 2 »), seemed to have their rays united, and lit up brightly.

BIBLioGRAPHy.
(In Chronological Order.)
1856. Lizperxktun, N.—“Zur Entwickelungsgeschichte der Spongillen,”

‘Arch. f. Anat. u. Physiol.,’ 1856, pp. 399—414, Taf. xv.

1857. Canter, H. J.—“On the Ultimate Structure of Spongilla, &.,”
‘Ann. Mag. Nat. Hist.’ (2), xx, pp. 21—41, pl. i.

1864. KottiKer, A.—‘Icones Histologice,’ Abth. I, Leipzig (Wilhelm
Engelmann).

1864. Scumipt, O.—‘Spongien des adriatischen Meeres,’ i supplement,
Leipzig, 1864.

1865. Lizserxtun, N.—“ Beitrage zur Anatomie der Kalkspongien,”’ ‘ Arch.
f, Anat. u. Physiol.,” 1865, pp. 732—748, Taf, xix.

1870. Scumipt, O.—‘Grundziige einer Spongien-Fauna des atlantischen
Gebietes,’ Leipzig, 1870.

1872. Harcxet, E.—‘ Die Kalkschwamme,’ Berlin, 1872, 3 vols.

1874. (1) Carter, H. J.—“*On the Nature of the Seed-like Body of
Spongilla, &.,” ‘Ann. Mag. Nat. Hist.’ (4), xiv, pp. 97—111, pl. x.

580

1874.

1875.

1875.

1878.

1879.

1879.

1880.

1880.

1881.

1882.

1883.

1884.

1885.

1885.

1885.

1887.

HE. A. MINCHIN.

(2) Carter, H. J.—‘‘ Further Instances of the Sponge Spicule in its
Mother-cell,” ‘Aun. Mag. Nat. Hist.’ (4), xiv, pp. 456—458, pl. xxi,
figs. 26, 27.

Carter, H. J.—‘ Notes Introductory to the Study and Classification
of the Spongida,” ‘ Ann. Mag. Nat. Hist.’ (4), xvi, pp. 1—40, 126—
145, 177--200, pl. iii.

Scuuuze, F. E.—* Ueber den Bau und die Entwicklung von
Sycandra raphanus, Haeckel,” ‘ Zeitschr. f. wiss. Zool.,’ xxv
suppl., pp. 247—280, Taf. xviii— xxi.

Scuuuze, I. E.—‘ Untersuchungen tiber den Bau und die Entwick-
lung der Spongien. V.Die Metamorphose von Sycandra raphanus,”
‘Zeitschr. f. wiss. Zool.,’ xxxi, pp. 262—295, Taf. xvili, xix.

Keuier, C.—‘‘ Studien tiber Organisation und Entwicklung der
Chalineen,” ‘ Zeitschr. f. wiss. Zool.,’ xxxiii, pp. 317—349, Taf.
XVHI—Xx.

Merscuntkorr, E.—Spongiologische Studien,” ‘ Zeitschr. f. wiss.
Zool.,’ xxxil, pp. 8349—387, Taf. xx—xxiil.

Scuutze, F. E.—‘‘ Untersuchungen, &c. IX. Die Plakiniden,”
‘Zeitschr. f. wiss. Zool.,’ xxxiv, pp. 407—451, Taf. xx—xxil.

Sottas, W. J.—“The Sponge Fauna of Norway,” ‘Ann. Mag. Nat.
Hist.’ (5), v, pp. 130—144, pls. vi, vii; pp. 241—259, pls. x—xii ;
pp. 396—409, pl. xvii.

Scuuizy, F. E.—“ Untersuchungen, &c. X. Corticium candela-
brum,” ‘Zeitschr. f. wiss. Zool. xxxv, pp. 410—430, Taf. xxi.

Sottas, W. J.—‘ The Sponge Fauna of Norway,” ‘Ann. Mag. Nat.
Hist.’ (5), ix, pp. 141—165, pls. vi, vii; pp. 426—453, pl. xvii.

Pottsanrr, N.—‘ Report on the Calcarea, ‘ Challenger’ Reports,”
‘ Zool.,’ viii, pt. xxiv.

Carter, H. J.—‘‘On the Spongia coriacea of Montagu, &c.,”
‘Ann. Mag. Nat. Hist.’ (5), xiv, pp. 17—29, pl. i.

(1) LenpEenFexp, R. von. —“ A Monograph of the Australian Sponges,”
Parts 1, 2, 3, ‘Proc. Linn. Soc. N.S. Wales,’ ix, pp. 121—154;
pp. 3L0—344; pp. 1083—1150; pls. lix—lxvii.

(2) Lenpenretp, R. von.—“The Histology and Nervous System of
Calcareous Sponges,” ‘ Proc. Linn. Soc. N.S. Wales,’ ix, pp. 977—
983.

Sorttas, W. J.—“On the Physical Characters of Calcareous and
Siliceous Sponge Spicules and other Structures,” ‘Sci. Proc. R.
Dublin Soe.’ (n. s.), iv, pp. 374—392, pl. xv.

Epner, V. von.—‘ Ueber den feineren Bau der Skelettheile ats Kalk-
schwamme nebst Bemerkungen tiber Kalkskelete iiberhaupt,” ‘ Sitz-

1887

1887.

1888.

1888.

1890.

1891.

1891.

1891.

1891.

1892.

1892.

1892.

1892.

1892.

1892.

1892

MATERIALS FOR A MONOGRAPH OF THE ASOONS. 581

ber. d. k. Akad. d. Wiss. Wien,’ I Abth., Bd. xev, pp. 55—149,
Taf. i—iv.

Rivtey, S. O., and Denpy, A.—*‘ Report on the Monaxonida, ‘ Chal-
lenger’ Reports,” ‘ Zool.,’ vol. xx, part 59.

Scuutze, F. E.—* Report on the Hexactinellida, ‘Challenger’ Re-
ports,” ‘ Zool.,’ vol. xxi.

Frepier, K.—“ Uber Ei- und Samenbildung bei Spongilla fluviatilis,”
‘Zeitschr. f. wiss. Zool.,’ xlvii, pp. 85—128, Taf. xi, xii.

Sotxas, W. J.— Report on the Tetractinellida, ‘Challenger’ Reports,”
‘ Zool.,’ vol. xxv.

Maas, O.—“ Ueber die Entwicklung des Siisswasserschwammes,”
‘ Zeitschr. f. wiss. Zool.,’ 1, pp. 527—554, Taf. xxii, xxiii.

Bipper, G. P.—Review of “‘ A Monograph of the Victorian Sponges,’
by A. Dendy, ‘Quart. Journ. Micr. Sci.,’ n.s., xxxii, pp. 625—632.

(1) Denpy, A.—‘ Studies on the Comparative Anatomy of Sponges.
III. On the Anatomy of Grantia labyrinthica, Carter, and the
so-called Family Teichonide,” ‘Quart. Journ. Micr. Sci.,’ n.s.,
xxxil, pp. 1—89, pls. i—iv.

(2) Denny, A.—‘‘A Monograph of the Victorian Sponges,” part 1,
‘Trans. Roy. Soc. Victoria,’ III, i, pp. 1—81, pls. i—xi.

LENDENFELD, R. von.—‘‘ Die Spongien der Adria. I. Die Kalk-
schwamme,” ‘ Zeitschr. f. wiss. Zool.,’ liii, pp. 185—321; pp. 361—
433, Taf. vili—xv.

Brover, G. P.—“ Note on Excretion in Sponges,” ‘ Proc. Roy. Soc.,’
li, pp. 474—4.84, four figs.

Burtscuut, O.—‘ Untersuchungen uber mikroskopische Schiume und
das Protoplasma,’ Leipzig (Wilhelm Engelmann), 1892 ; translated as
‘Investigations on Microscopic Foams and on Protoplasm,’ London
(A. and C. Black), 1894.

Detace, Y.—* Embryogénie des Eponges, &.,” ‘Arch. Zool. exp. et
gén.’ (2), x, pp. 8345—498, pls. xiv—xxi.

>

Dreyer, F.—“ Die Principien der Gerustbildung bei Rhizopoden,
Spongien, und Kchinodermen,” ‘ Jenaische Zeitschr. f. Naturwiss.,’
xxvi (n. f. xix), pp. 204—468, Taf. xv—xxix.

Maas, O.—‘‘ Die Metamorphose von Esperia Lorenzi, O. S., &.,”
‘Mitth. Zool. Stat. Neapel,’ x, pp. 408—440, Taf. xxvii, xxviii.

(1) Mincuin, E. A.— Note on a Sieve-like Membrane across the
Oscula of a Species of Leucosolenia, &c.,” ‘Quart. Journ. Mier. Sci.,’
N. S., XXXili, pp. 251—272, pls. x, xi.

(2) Mincuiy, E. A.—“The Oscula and Anatomy of Leucosolenia

voL. 40, paRT 4.—NEW SER. Ss

582 E. A. MINCHIN.

clathrus,” ‘Quart. Journ. Mier. Sci.,’ n.s., xxxilil, pp. 477—495,
pl. xxix.

1892. (3) Mincutn, E. A.—‘‘ Some Points in the Histology of Leucoso-
lenia (Ascetta) clathrus, O.S%.,” ‘ Zool. Auzeiger,’ xv, pp. 180—
184.

1892. Torpsent, E.—‘ Notes histologiques au sujet de Leucosolenia
coriacea (Mont.), Bow,” ‘ Bull. Soc. Zool. France,’ xvii, pp. 125—
129.

1898. Denny, A.—‘ ‘Studies, &c. V. Observations on the Structure and
Classification of the Calcarea Heterocceela,” ‘Quart. Journ. Mier.
Sci.,’ n. s., xxxv, pp. 159—257, pls. x—xiv.

1894, (1) LenpEnFetp, R. von.— Die Tetractinelliden der Adria, &c.,”
‘Denkschr. d. k. Akad. d. Wiss. Wien, Math. naturw. Classe,’ Ixi,
pp. 91—204, Taf. i—viii.

1894. (2) LenpenFELD, R. von.— Ergebnisse neuerer Untersuchungen uber
Spongienepithelien,” ‘ Zool. Centralbl.,’ i, pp. 505—510.

1895. Detace, Y.—‘ La Structure du Protoplasma et les Théories sur |’Héré-
dité, &c.,’ Paris, Reinwald et Cie., 1895.

1896. (1) Mincuin, HE. A.—“ Note on the Larva, &., of Leucosolenia
variabilis, &.,” ‘Proc. Roy. Soc.,’ vol. lx, pp. 42—52.

1896. (2) Mincutn, E. A.—“ Suggestions for a Natural Classification of the
Asconide,”’ ‘Ann. Mag. Nat. Hist.’ (6), xviii, pp. 349—362.

1897. DoéprrieIn, L.—‘ Ueber die Lithonina, eine neue Gruppe von Kalk-
schwamme,” ‘ Zool. Jahrbiicher, Abth. f. Syst. Geogr. u. Biol. d.
Thiere,’ x, pp. 15—32, Taf, ii—vi.

EXPLANATION OF PLATES 38—42,

Illustrating Mr. E. A. Minchin’s paper on “ Materials for a
Monograph of the Ascons.”

All the figures are drawn with the camera lucida, using oc. 4 (Zeiss) and
obj. 2 mm. immersion (Leitz), and are magnified about 1400.

SIGNIFICANCE OF THE LETTERING.

act. bl. Actinoblast. am.c.1 Nutritive wandering cell (ameebocyte). am.c.?
Clear wandering cell. «am.c.? Minute wandering cell. apic. form. cell. Apical
formative cell of the spicule ray. das. form. cell. Basal formative cell of the
spicule ray. col. Collar. col.cell. Collar-cell. derm.ap. Dermal (external)
aperture of pore. ff. Flagellum. j.ep. Flatepithelium. gasér. act.b/. Gastral

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 583

actinoblast. gastr. ap. Gastral (internal) aperture of pore. gastr. ray. Gastral
ray of quadriradiate. x.fl.ep. Nuclei of flat epithelium. por.c. Porocyte.
seat, Sextet of spicule-forming cells. spice. Spicule. spic. cell. Definitive
spicule-cell. spic. monax. Monaxon spicule. sp.sk. Spicule sheath. ¢rirad.
syst. Triradiate system. «. Rod-like body in spicule-forming cells.

PLATE 38.
All the figures relate to Clathrina coriacea.

Fic. 1.—Surface view from the gastral side of the dermal layer to show
three cells of the flat epithelium that have migrated inwards (actinoblasts).
The drawing further shows five cells of the flat epithelium, two spicule rays,
one of them bearing a spicule-cell, and a pore-cell.

Fic, 2.—Surface view from the gastral side of dermal layer showing a
“trio ” of actinoblasts near a spicule ray. Three nuclei of the underlying
flat epithelium are figured to show the relations of the cells.

Fie. .3.—Surface view from dermal side of dermal layer showing a trio of
actinoblasts underneath a spicule ray. At a higher level are three cells of
the flat epithelium.

Fig. 4.—A “sextet ” showing three formative cells lying just above three
others. From the dermal aspect. -

Fig. 5.—A sextet in which the minute triradiate system has made its
appearance.

Fic. 6.—A slightly older triradiate system in its sextet of formative cells.
From the gastral aspect.

Fic. 7.—A further stage in the growth of the triradiate system seen from
the gastral side; the three inner formative cells have travelled to the apices
of the rays, while the three outer ones remain at the base.

Fic. 8.—Typical six-celled stage in the formation of the triradiate system,
seen from the dermal aspect; the three irregularly shaped inner formative
cells at the apices of the rays, the three spindle-shaped outer formative cells
at the bases of the rays.

Fie. 9.—Triradiate system in which two apical formative cells have dis-
appeared, one still remaining. Dermal aspect.

Fic. 10.—View of dermal layer after removal of collar-cells, showing
various cell elements from the dermal aspect. The drawing shows ten cells of
the flat epithelium (7. ep.), six pore-cells (por. c.), one actinoblast (act. d/.),
one young triradiate system bearing three basal formative cells and a single
apical formative cell (apic. form. cell) persistent on one ray; overlying this the
ray of a fully formed triradiate system, bearing a spicule-cell (spic. ced/), and
finally one granular (nutritive) wandering cell (am.c.!) and five non-granular
wandering cells (am. c.’).

584 BE. A. MINCHIN.

Fic. 1]1.—Triradiate system showing one ray fully formed and the other
not yet so. In the former ray the basal formative cell is at the apex of the
spicule as the definitive spicule-cell; in the latter it is still not far from the
base of the ray. .

Fic. 12.—Triradiate system not quite full-grown; the rays are conical in
form, and the formative cells still contain numerous granules. In the case of
one ray the formative cell is quite at the base; in the case of the other it
has begun to migrate towards the apex.

Fig. 13.—Triradiate system showing what is apparently an apical formative
cell just detached from one of the rays at an unusually late period of its
growth. The basal formative cell in this ray has travelled halfway to the
apex; that on one of the other rays is still at the base.

Fic. 14.—A fully formed spicule ray, showing the spicule cell at the
extreme edge.

PLATE 39.

Fies. 15 anp 16.—Clathrina coriacea.

Fig. 15.—A porocyte in process of immigration from the flat epithelium,
near a triradiate system. Dermal aspect. One portion of the cell has
definite limits and an indented outline, and lies at a deeper level, in
contact with the collar-cells. The other portion is spread out without
clearly defined limits, and is quite superficial. The cell shows a large
vacuole containing apparently calcareous matter, spicule remnants.
Near the cell lies what appears to be a fragment of a broken spicule.

Fig. 16.—Surface view of the dermal epithelium, with a porocyte in
process of immigration. For the latter compare description of last
figure.

Fies. 17—21.—Clathrina, sp. dub.

Fig. 17.—Surface view of dermal layer from dermal aspect, showing three
cells of flat epithelium, two spicule rays, one bearing a spicule-cell, two
minute wandering cells, and a sextet of spicule-forming cells. The
spicule has not as yet appeared.

Fig. 18.—A young quadriradiate surrounded by formative cells. Gastral
aspect. The rays of the basal system are surrounded by a sextet of
formative cells. The minute gastral ray is contained in a porocyte
containing two nuclei, one close to the ray in question.

Fig. 18a.—The porocyte and spicule of the last figure drawn separately
from the other cells.

Fig. 19.—View of dermal layer, gastral aspect; collar-cells removed.
The figure shows twelve cells of the flat epithelium; several spicules
and spicule rays, one of which shows a single spicule-cell at its

MATERIALS FOR A MONOGRAPH OF THE ASCONS. 585

extremity, while another still bears both its formative cells; a very
young quadriradiate, the basal rays enveloped in their sextet of
formative cells; the gastral ray contained ina gastral actinoblast which
is still continuous with a pore-cell; and finally, a granular wandering
cell, partly covered by a spicule, and two minute wandering cells.

Fig. 194.—The porocyte and gastral actinoblast of the last figure drawn
separately.

Fig. 20.—View of dermal layer, dermal aspect. The figure shows seven
cells of the flat epithelium, and a portion of another; one pore-cell ;
portions of three triradiate systems, one bearing a spicule-cell at the
extremity of a ray, aud the proximal extremity of a large monaxon
spicule, bearing two spicule-cells ; and finally a young quadriradiate
surrounded by the six formative cells of the basal system, and the
gastral actinoblast.

Fig. 21.—A young quadriradiate surrounded by the six cells of the basal
system, and the gastral actinoblast.

Fics. 22—26.—From embryos of Ascandra falcata.
Fig. 22.—A sextet from an embryo of the third day of fixation.
Fig. 23.—A young triradiate system from an embryo of the same date.
Fig. 24.—An older triradiate system from an embryo of the sixth day.
Fig. 25.—A ray of a triradiate, with its two formative cells, from an
embryo of the same date.

Fig. 26.—A fully formed ray, with its single cell, embryo of the same
date.

Fic. 27.—Young triradiate systems, with their formative cells, from
Clathrina cerebrum, embryo of third day of fixation. The overlying
nuclei of the flat epithelium are drawn in outline only.

PLATE 40.

Fics. 28—30.—From sections of Clathrina cerebrum, adult.

Fig. 28.—Young quadriradiate with gastral ray enveloped by its actino-
blast.

Fig. 29.—Fully formed spiny gastral ray with its actinoblast, con-
taining a nucleus and scattered granules of chromatin.

Fig. 30.—Fully formed smooth gastral ray with its actinoblast, con-
taining a nucleus at the lower extremity, and an aggregation of
chromatin granules near the apex.

Fics. 31 anp 32.—From embryos of Clathrina reticulum.

Fig. 31.—Young triradiate system surrounded by sextet of formative
cells; the overlying nuclei of dermal epithelium drawn in outline only.

Fig. 32.—An older triradiate system, with cells as in last figure.

586 E. A. MINCHIN.

Fics. 33—37.—From sections of adult Clathrina reticulum.

Fig. 33.—A young quadriradiate spicule with formative cells; the basal
rays, two of which appear in the section, have each two formative
cells. The gastral ray is completely enveloped in its actinoblast.

Fig. 34.—A slightly older gastral ray with its actinoblast.

Fig. 35.—Gastral ray apparently fully formed, the actinoblast with its
single nucleus having retreated to the extremity of the ray.

Fig. 36.—Gastral ray with actinoblast in which the nucleus has only
recently divided, at a late stage in the growth of the spicule.

Fig. 37.—Very long and slender gastral ray, with two nuclei in the
actinoblast.

PLATE 41.
All the figures refer to Clathrina contorta.

Fic. 38.—Surface view of the oscular rim in the region just above the
limit of the collar-cells. Gastral aspect. The preparation contains eight
porocytes, one with two nuclei; a quadriradiate spicule, each basal ray
bearing the usual two formative cells, and the gastral ray enveloped in an
actinoblast with two nuclei; and seven collar-cells, more or less completely
shown. ‘To show the identity of porocytes and gastral actinoblasts.

Fic. 39.—View of the gastral surface of the body-wall in a spot slightly
below that figured in the last figure, showing collar-cells and four poro-
cytes, three of them opened and functioning as pores. The collars of the
collar-cells are seen in optical section at higher focus as irregular circles, in
the centre of each of which the flagellum appears as a dot.

Fie. 40.—Surface view of dermal epithelium of oscular rim on the outer
side.

Fic. 41.—Section of oscular rim, showing epithelium as in Fig. 40 on the
lower side, and granular porocytic epithelium, as in Fig. 38, on the upper
side. The section being fairly thick, and being decalcified, shows the ground
substance full of spaces formerly occupied by spicule rays, to two of which
formative cells are adhering, and a gastral ray projects on the upper side
bearing an actinoblast with two nuclei. Note also four minute wandering
cells.

Fic. 42.—An epithelial cell from the inside of the oscular rim, near the
margin of the osculum, intermediate in its characters between the ordinary
dermal epithelial cells and the porocytes.

Fie. 43.—A granular wandering cell.

Fie. 44.—A granular wandering cell near two sexual cells (spermatogonia?).

Fie. 45.—Two sexual cells similar to those in the last figure.

Fic. 46.—A finely granular wandering cell (the fine granules only drawn in
part).

MATERIALS FOR A MONOGRAPH OF THE ASOCONS. 587

Fic. 47.—A finely granular wandering cell.
Fic. 48.—A minute triradiate system in its sextet of formative cells.

PLATE 42.
All the figures refer to Clathrina coutorta.

Fic. 49.—View of the gastral surface of the body-wall, showing, amongst
the collar-cells, a porocyte in continuity with a gastral actinoblast, which is
overlying a young triradiate system. The gastral ray has not made its appear-
ance. Only the gastral aperture of the pore has been figured, and the forma-
tive cells of the triradiate system have also been omitted.

Fic. 50.—A similar view of a slightly more advanced stage. The porocyte
has the dermal aperture closed; the minute gastral ray has now appeared.

Fic. 51.—A similar view of still older stage. The gastral actinoblast has
completely separated from the porocyte, and the nucleus of the former has
divided into two.

Fic. 52.—Slightly older but rather anomalous stage. The gastral actino-
blast has its nucleus divided into two, but is still in continuity with the poro-
cyte. Collar-cells not figured.

Fic. 53.—Young quadriradiate showing the gastral ray a little from the
side, enveloped in an actinoblast with two nuclei.

Fic. 54.—More advanced stage. The actinoblast now has four nuclei.

Fie. 55.—Fully formed gastral ray, from a section. The large actino-
blast is spread over the gastral ray like a plasmodium, and contains four
nuclei.

t
43 fa
Ay 4 ca)
-
v = Py bee al) ee 8!
‘ P
Ens’ gen aah Ses
= 1 ae aT i a -% ?
. . :
P } j ’
Urns PA Oe ae
} ‘
rr & " r a
r bs ec oper
«
Pa, 12am: ¢ ‘ ‘
et - i ‘J ‘
ni ¥
=F i ne 7 . .
a7 4 (Gs oF '
] P ij
:
4° pitta! :
7 y j Ath
; .
| an AS he . ”
: = =
r ‘ Ta ret d ':
y ee ” 7
~~
is
7 z
-
-
;
1 -
| >
; =
: r
a
.- r=
4
24
¢
: ‘
'
ae “
a z a
‘
«
:
.
FF
.
7 5
» .
ss | 2
=
a ui
7) ’ .
>, o ic)
e

CORN Wes

B= POz an
>
Peuh | ;
L -.-! ~ re) *
7 ? * a
: e =} j \
j 4
é eS
'
T {
j ‘
4 A’
a L 6
‘ .
ay
~
ie Se ww
=
* ¥
,
* 7 on

THE EARLY DEVELOPMENT OF AMPHIOXUS. 589

The Early Development of Amphioxus.
By
E. W. MacBride, M.A.,

Fellow of St. John’s College, Cambridge; Professor of Zoology in the McGill
University, Montreal.

With Plates 43—45.

Tue work which forms the subject of the present essay was
carried out in the Cambridge Zoological Laboratory during
the years 1895-7. The material on which my results are
founded consisted of a collection of embryos and larve which
had been obtained by Mr. Sedgwick and Dr. Willey during
their visits to Faro in 1890-91, and I have to express my
warmest thanks to Mr. Sedgwick for placing this valuable
material at my disposal.

It will, no doubt, be the opinion of many zoologists that a
fresh paper on the early development of Amphioxus needs
considerable justification, since it may appear superfluous to
attempt to improve on the admirably clear account given by
Hatschek (8 and 5) of this subject, an account which has be-
come incorporated in all text-books, and forms part of the
classical literature of zoology. That such an opinion, how-
ever, would not be well founded, and that on the contrary a
large number of doubtful points of great theoretical im-
portance remained to be cleared up, will be evident when we
briefly examine the present state of our knowledge on this
subject.

The foundation of our knowledge of the development of
Amphioxus was laid by Kowalevsky, whose two papers (6 and

590 E. W. MACBRIDE.

7) may be considered together. In these the segmentation of
the egg and the formation of the blastula and the process of
invagination are described. Kowalevsky states that the blasto-
pore is at first posterior in position, but becomes later shifted
on to the dorsal surface; he describes the formation of the
central nervous system and its appendix, the neurenteric canal.
He also describes the formation of the ccelomic pouches
(myotomes) as folds of the gut wall, and notes the fact that
the first pair have thinner walls and a wider cavity than the
rest, and communicate by a broader slit with the gut cavity.
He likewise gives an account of the larval life of Amphioxus,
and it is interesting to note that in his second paper (7) he
anticipated the results which Lankester and Willey obtained
later (9 and 13) as to the development of the gill-slits and the
formation of the atrial cavity. The two longitudinal ridges,
which by their union form the atrial cavity, were seen and
figured by him, and the cavities they contained also described.
His account of the gill-slits appeared, however, so extra-
ordinary to his contemporaries, that it was supposed he was
misled by pathological specimens, and it needed Dr. Willey’s
researches to convince zoologists of its accuracy.

A year or two after Kowalevsky’s second paper was pub-
lished Professor Hatschek undertook a re-investigation of the
subject, and the account given of the development of Am-
phioxus in Korschelt and Heider’s ‘ Lehrbuch der Vergleichen-
den Entwickelungsgeschichte’ is taken directly from Hatschek’s
paper. Hatschek confirmed in his first paper (8) Kowalevsky’s
account of the segmentation of the egg; but with regard to
the invagination he asserts that the blastopore is from the first
in its definitive dorsal position, or, what comes to the same
thing, that the closure of the blastopore is mainly effected by
the backward growth of the dorsal (anterior) lip.

This statement has been seized on with avidity by certain
workers in Vertebrate embryology as affording a possibility of
twisting the developmental history of Amphioxus into accord-
ance with theories which regarded the main developmental
process in Vertebrates as a concrescence of two distinct halves

THE BARLY DEVELOPMENT OF AMPHIOXUS. 591

along the mid-dorsal line, or, as it was sometimes expressed,
the meeting of the lips of a long slit-like blastopore.

Hatschek’s account of the origin of the mesodermal struc-
tures is curious and interesting. Like Kowalevsky, he finds
two longitudinal folds of the gut wall, but these he believes to
terminate posteriorly in two large “ pole” cells situated in the
lip of the blastopore. By cross-folds the successive pouches
are cut off from these longitudinal folds. Later two inde-
pendent outgrowths from the alimentary canal form the
“head cavities,” of which the right retains longest its com-
munication with the alimentary canal. The formation of a
continuous ventral body-cavity by the fusion of the lower
parts of the pouches (somites) is deduced from the fact that
the divisions between the somites can no longer be traced in
this part of the body. In a later paper (4) he describes in the
larva what he terms ‘‘a genuine kidney,” a tube lying in front
of the mouth. This structure will be referred to throughout
this paper as ‘ Hatschek’s nephridium.” In his last paper
(5) he gives an account of the derivation of the muscular and
skeletal tissue from the coelomic pouches, pointing out that
everything in Amphioxus is essentially epithelial in nature,
the connective tissue having no nuclei in it.

Lankester and Willey’s paper on the development of the
atrial chamber (9) confirms in most points Kowalevsky’s
statements, but the authors deny that the cavities in the
metapleural folds are ccelomic in nature as Kowalevsky had
imagined. The right atrial fold is described as extending
much further forward than the left. They found also that
Hatschek’s nephridium opens into the gut; a result which
Van Wijhe (11) announced as an independent discovery three
years later.

Willey (18) in 1891 published an account of the develop-
ment of the gill-slits and other organs appearing in the larval
stage. He describes the internal and external openings of the
club-shaped gland (an organ noticed already by Kowalevsky),
which he regards as the fellow of the first gill-slit. He states
that the oral hood which conceals the true mouth of Amphi-

592 E. W. MAOBRIDE.

oxus, and which was regarded by Lankester as a forward pro-
longation of the atrial folds, is a downgrowth of the upper
margin of the preoral pit as far as its upper part is con-
cerned, its lower part being formed beneath the mouth inde-
pendently.

Lwoff’s paper (10) deserves special attention not only
because it is the first systematic study of the embryology of
Amphioxus with modern methods, but also on account of the
disagreement of the results he arrived at with those of Hat-
schek. Lwoff maintains that before invagination commences
the cells constituting the blastula become sharply divided into
two sorts, endoderm and ectoderm cells; that invagination
first involves the endoderm cells, but that later the ectoderm
bends round the dorsal lip of the blastopore, and displacing
the endoderm forms the dorsal wall of the gut, and that the
whole of this ectodermal stage is employed in the formation
of the notochord, and in the production of mesoderm. Lwoff
believes that the mesodermal ‘‘ folds” are the mechanical result
of the pressure of the nerve-tube and notochord on the upper
wall of the gut; he asserts that the cavity of this fold dis-
appears, and that the cavities of the somites appearing later
have no connection with or relation to the enteric cavity ;
hence that Amphioxus is not an “ enterocelous animal.” He
finds that the pole-cells of Hatschek have no existence, therein
confirming Wilson (14) ; and that the blastopore is closed not
by the special growth of the dorsal lip, but from all sides.

It will thus be evident that the most interesting part of the
development of Amphioxus, viz. the formation of the primitive
germinal layers, was involved in great uncertainty, and it was
with a view of clearing up the questions thus raised, and also
from a feeling of dissatisfaction with the accounts given of the
nature and origin of such structures as Hatschek’s nephri-
dium, and the cavities in the atrial (metapleural) folds, &c.,
that I was led to undertake a re-investigation of the whole
subject. That the early development of Amphioxus is of -
great theoretical importance can, I think, be hardly denied ;
when we consider that in it we have the only instance of

THE EARLY DEVELOPMENT OF AMPHIOXUS. 593

Vertebrate development where, the yolk being evenly distributed,
its disturbing influence is negligible ; and when one recollects
the weary controversies which have been waged round the
meaning to be attached to the gastrulation and formation of
the layers in the heavily yolked eggs of the higher Verte-
brates, one must feel that a knowledge of the development of
Amphioxus alone could bring these questions to a definite issue.

The results I have arrived at differ considerably from any
obtained hitherto, and in claiming to have penetrated more
deeply into the developmental processes than Hatschek or
Kowalevsky I rely entirely on the higher grade of perfection
which methods of dealing with small organisms have reached
in the meantime.

All the embryos were embedded in the ordinary way in cel-
loidin; but after hardening the celloidin with chloroform I
adopted a plan of clearing on which really the whole success
of the work depended. This was as follows :—The hardened
celloidin was immersed for a minute or so in absolute alcohol,
in order to remove any traces of moisture which might be
present in the chloroform ; it was then placed in cedar oil and
left for a night in a warm place (the dish containing the cedar
oil being placed on the top of the thermostat). In the morn-
ing the celloidin had become so transparent as to be almost
invisible when looked at in the cedar oil. Little pieces of the
block containing the embryos could then be cut out and
examined with ease under a low power, and their exact orien-
tation determined. They were then embedded in paraffin, and
cut into series of sections from 4 to 5 yu thick, and stained on
the slide.

The material at my disposal was preserved in a variety of
ways, but except for the earliest stages my results are based
only on specimens preserved in osmic acid. I wish to lay
particular emphasis on this, as should any zoologist feel
inclined to work over the same ground with a view of testing
my results, and use such fixing reagents as corrosive subli-
mate or picro-sulphuric acid, he is foredoomed to failure.
After such fluids the gut becomes swollen and the body-walls

594 - E. W. MACBRIDE.

collapsed, so that it is impossible to make out anything of the
limits of the coelomic cavities. The greatest difficulty I have
found in dealing with Amphioxus larve is to stain the con-
nective tissue; and I have, in fact, only found this possible with
specimens preserved in osmic acid. Of course insufficient or
careless preservation in this fluid is valueless, and leads as
usual to maceration. For the stages up to the end of the
gastrulation, however, before any real differentiation of tissues
has taken place, almost any reagent gives fair results.

I do not intend in this paper to refer, except incidentally, to
those features in the development of Amphioxus which have
been satisfactorily worked out, and about which there is a
general consensus of opinion. I may, however, remind the
reader that the eggs are spawned in the evening about 7
o’clock ; segmentation takes place rapidly, so that by 11 p.m.
the blastula is complete and invagination has commenced ;
two hours later the gastrula stage is attained, and by 5 a.m.
the mesoderm has appeared, and the embryo which has been
for some time actually rotating within the egg-shell is hatched.
On the morning of the next day about 8 a.m. the embryo has
acquired the definite form of the larva, with a pointed snout, a
swollen pharyngeal region, and a very attenuated body, and
the mouth and first gill-slit are formed. No further develop-
ment has been attained with embryos reared in aquaria, and
in their natural habitat the rate of development is after this
very slow, extending over several months. I shall divide my
account of the subject into the following parts :—a. The gas-
trulation. 6. The formation of the mesoderm. c. The fate of
the celomic cavities. d. The origin of the atrial or “ epi-
pleural” folds.

1. The Gastrulation.

A specimen of the youngest embryos examined is repre-
sented in Pl. 43, fig. 1. It has the form of a regular sphere
bounded by one layer of cells of approximately equal size, and
is in fact a perfectly typical blastula. Although Hatschek
figures a difference in size of the cells in the two poles of the

THE EARLY DEVELOPMENT OF AMPHIOXUS. 595

egg as being observable in the earlier stages of segmentation,
I have been quite unable to find any trace of it in sections,
and in this respect confirm Lwoff (10). In agreement with
the latter author, I find when the first change ushering in
the process of invagination takes place, namely, a flattening
of one side of the blastula, that then for the first time a
differentiation of the cells composing the embryo into two
sorts become observable. Some of them, in fact, become
taller and slightly narrower than the rest. Lwoff lays great
emphasis on this phenomenon. He regards the more cylin-
drical cells as alone representing the endoderm of Inverte-
brates, any further portion of the blastula which may be
involved in the process of invagination being denominated as
ectoderm.

It is difficult to find words to adequately characterise the
artificiality and arbitrariness of such a view. The only
circumstances under which it could be maintained would be if
the supposed endoderm were sharply marked off from the
ectoderm, and if further there were a pause in the process of
gastrulation after the so-called endoderm had been invaginated,
but before the invagination of the ectoderm had commenced.
A glance at Pl. 43, figs. 2 and 3, will show that the taller cells
on one side shade imperceptibly into the shorter and rounder,
so that it is impossible to say where the one begins and the
other ends. On the other side, it is true, there is an abrupt
transition at the point marked 2 This point I have found by
careful comparison with one another of successive stages to
correspond to the dorsal (anterior) lip of the blastopore, the
very place where Lwoff supposes ectoderm to be invaginated.
This spot is easily recognisable in the earlier stages of invagi-
nation (Pl. 43, figs. 4, 5, and 6), but becomes less recognisable
in the later stages. Asa tentative explanation of it I may
suggest that here an active multiplication of cells takes place,
and that those which are added to the invaginated portion of
the blastula become laterally compressed and columnar,
whereas those added to the ectoderm remain stretched by the
internal turgidity of the fluid in the blastoccele or segmenta-

596 E. W. MACBRIDE.

tion cavity. As invagination proceeds, the future dorsal sur-
face of the embryo becomes recognisable by the close apposi-
tion of the layers of ectoderm and endoderm which subsists
here, whilst on the other side the outer and inner layers of the
gastrula diverge from one another (at the point marked o in
fig. 7, for example). A little later we are able to perceive
that the future dorsal surface has become definitely flattened,
this being the first preparation for the formation of the neural
plate, the rudiment of the central nervous system ; and as the
appearance of this structure enables us to determine the long
axis of the future animal, we are able to say that the blastopore
is at first posterior. At the same time the peculiar character
of the dorsal lip at this stage, with the abrupt transition from
ectoderm to endoderm, and the close parallelism of the two
layers enables us with certainty to identify it with the corre-
sponding part in earlier stages.

Thus I regard the gastrulation as a fairly uniform pushing
in of the under or flattened surface of the blastula, accom-
panied by division and multiplication of the cells, such multi-
plication being at first most active in the dorsal (future
anterior) lip of the blastopore. The blastopore, which is still
wide, becomes rapidly narrowed by the upgrowth of the
ventral lip (PI. 43, figs. 8, 9, and 10) : in contra-distinction to
what Hatschek (8) asserts, the dorsal lip remains relatively
stationary.

Coincidently with this increased activity of growth in the
ventral lip, a sharp, abrupt transition becomes now observable
in it from ectoderm to endoderm, a fact which supports the
explanation given above of a similar phenomenon observed in
the dorsal lip. In support of his view that there is an invagi-
nation of ectoderm round the dorsal lip of the blastopore,
Lwoff speaks of a frequent accumulation of cells just inside
the blastopore on the dorsal side, and figures two longitudinal
sections of gastrule in illustration of this poimt. I have to
definitely state that no such appearances are ever seen in
properly orientated sections, and that Lwoff has been misled
by his inability to distinguish between oblique and sagittal

THE EARLY DEVELOPMENT OF AMPHIOXUS. 597

sections, though I do not admit that such an accumulation,
did it exist, would prove his point.

If we examine a transverse section of a completed gastrula—
such a one, for instance, as PI]. 43, fig. 1l1—we find no difference
in character between the cells forming the dorsal wall of the
alimentary canal and those forming the ventral wall, such as
we should have the right to expect did Lwoff’s hypothesis in
any way correspond with the facts. Before leaving this
subject, however, it is but just to notice a statement of
Lwoft’s, that had he been dealing with the development of
Amphioxus alone, he should not have ventured to put forward
the hypothesis of an ectodermal origin of the dorsal wall of
the archenteron; but that as he found in other Vertebrates
that this dorsal wall was entirely used up in the formation
of the notochord and mesoderm, and did not take part in
the definitive wall of the alimentary canal, and was in some
cases apparently derived from ectoderm, he felt justified in
reading this interpretation into the developmental processes
of Amphioxus. Such an attitude of mind seems to me the
entire converse of the proper one to be adopted under these
circumstances. Quite apart from the superior value to be
attached to the significance of the processes in Amphioxus
owing to the primitive nature of the adult, it is one of the
best known facts of embryology that the presence of large
quantities of yolk clogs and utterly distorts the develop-
mental processes, and that we have to interpret the cases
where much yolk is present in the light of those where little
yolk is present, and not vice versa. Moreover, a very
simple and natural explanation can be suggested why in the
Vertebrate embryo the yolk should be confined to the ventral
wall of the archenteron. We know that many, if not
most, developmental processes are ultimately reducible to
processes of folding, such as would be rendered entirely im-
possible were the tissue in which they have to take place
clogged with yolk. Hence in the higher Vertebrates the pro-
cesses of invagination itself are profoundly modified; and, as
explained in detail in the careful work of Will (12) (who in

voL. 40, PART 4.—NEW SER. TT

598 E. W. MACBRIDE.

this confirms the ideas of Balfour!), the bulky ventral wall of
the archenteron can no longer be folded in, and the persistent
invagination of the yolkless dorsal wall has the appearance of
an independent ingrowth of the ectoderm.

A good illustration of the influence of yolk is seen in the
~ development of Molluscs. There it has been customary to
regard the four macromeres as representing the endoderm.
These cells are, however, much too heavily loaded with yolk
to give rise to any definitive tissue; from them the smaller
micromeres which are to form the ectoderm are budded off,
and from them later, in continuity with these, the cells which
give rise to the epithelium of the intestine. (See Balfour’s
‘Text-book of Comparative Embryology,’ vol. i, Dev. of
Mollusca.) It is significant that assertions of the share of the
ectoderm in the formation of the alimentary canal should have
been made principally in such cases (eggs of insects, cephalo-
pods, &c.), where the accumulation of yolk is so great as to
preclude any possibility of the yolk-bearing cells being directly
converted into permanent tissue.

2, The Formation of the Mesoderm.

Shortly after the gastrulation is complete the outline of the
alimentary canal, as seen in transverse section, ceases to be
round. Its dorsal wall becomes flattened, and is then drawn
out into two lateral angles, and almost immediately afterwards
a median hollow ridge is formed, the first suggestion of the
future notochord. The lateral angles mentioned are the ex-
pression of two longitudinal hollow ridges or folds of the gut
wall. Lwoff regarded them as the mechanical effect of the
downward pressure of the nerve-tube as it became folded into
the interior of the body, and likewise of the notochord when it
was formed. The absurdity of this view is well shown by
Pl. 43, fig. 12; there these two lateral ridges are plainly seen,
whilst the nerve-cord is still a flat plate, and the notochord is
barely indicated. That these hollow outgrowths (which we

1 «A Comparison of the Early Stages in the Development of Verte-
brates,” by F. M. Balfour, ‘ Quart. Journ. Mic. Sci.,’ 1875.

THE EARLY DEVELOPMENT OF AMPHIOXUS. 599

may term the ceelomic grooves) are due to an independent
process of folding, originating in the endoderm, is also shown
by the fact that the endoderm is no longer in close contact
with the ectoderm, a distinct slit-like blastoceele being ob-
servable in several places, which certainly would not be the
case if the endoderm merely passively followed the foldings of
the ectoderm.

Shortly after the appearance of the celomic grooves a fresh
pair of outgrowths from the alimentary canal make their
appearance anteriorly. These, which will be denominated
collar cavities, are shown in fig. 13. As will be seen, they
are situated slightly nearer the middle line than the coelomic
grooves(which in the figure are seen external to them, separated
by a very narrow fold of the gut wall), and their lumina are at
first excessively narrow (at least in preserved specimens), but
they soon enlarge, and their openings into the cavity of the
alimentary canal become clear (fig. 14, c). About the same
time the front part of the celomic grooves becomes constricted
off from the alimentary canal, and thus a definite “somite ”
or myomere is formed (fig. 14, 6). Behind, however, the
celomic grooves still open into the alimentary canal, as shown
in fig. 14, a. If we compare with such a series of sections
through older embryos, such as those figured in fig. 17,
d and e, we shall arrive at a clear comprehension as to how
the celomic grooves are converted into a series of somites.
We can always find at the hinder end of the embryo appear-
ances (like those represented in figs. 14, a, and 14, c) of the
gut wall being folded so as to produce a pair of ccelomic
grooves, and we can follow the walls of the fold constituting
the coelomic groove into the walls of the last somite (compare
fig. 17, d). The cavity of no other somite communicates with
the gut; it is only the last somite, for the time being,
whose cavity is in communication with the gut cavity through
the coelomic groove; and hence we see that as the embryo
grows in length, and the ccelomic groove with it, this latter
becomes progressively constricted from the gut. and divided
into somites at the same time, each new piece which is con-

600 E. W. MACBRIDE.

stricted off becoming formed into a somite, and nipped off
from the open part of the celomic groove, which then again
grows in length, and the process is repeated. The formation
of a somite, then, is essentially a process of obliterating the
cavity of the coelomic groove for a certain space, and the
so-called last somite is really the undifferentiated hinder
end of the celomic groove.

The entire independence of the collar cavities from the
coelomic grooves is emphasised by the fact that for some time
after the latter are shut off from the gut the collar cavities
still retain their openings into it. This is shown in the trans-
verse section (fig. 14, ¢) and in the longitudinal section (fig.
15, a). These collar cavities are the “ first protovertebre ”
of Kowalevsky, and in his paper (7) he notes the fact that
they communicate by a broader slit with the alimentary canal,
and retain this communication longer than the rest. Later,
it is true, the right collar cavity becomes completely shut off
from the gut, but the left retains its communication, as is
shown in Pl. 43, fig. 16.

Shortly after this period the embryo begins to diminish
rapidly in diameter, owing to the consumption of the yolk in
the endoderm cells, whilst at the same time it increases in
length, and the cavities in its interior diminish in size, owing
to the gradual shrinkage. Hence one requires now to have
specimens preserved in such a way as to give firmness and
resistance to the outer tissue, if one is to make out anything
of the internal anatomy at all. The yolk, which is present not
only in the ventral but also in the dorsal ectoderm and in the
walls of the ccelomic folds, acts whilst it endures, somewhat
like paraffin, in preventing too great shrinkage; after it has
gone nothing but osmic acid will give any help.

Just as the disappearance of the yolk is commencing, the
third division of the mesoderm, the head cavities, make their
appearance. The two head cavities really constitute the extreme
anterior end of the alimentary canal, which grows out into two
lateral horns. In Pl. 44, fig. 17, a, we see them still opening
widely into the gut; but in the next section (17, 6), taken

THE EARLY DEVELOPMENT OF AMPHIOXUS. 601

further back, we see them quite free from the gut, and so we
ean conclude that they are recurved. Hatschek (8) spoke of
the two head cavities as distinct outgrowths from the gut, and
further stated that the right one still communicated with the
gut after the left had been cut off. This is not a correct
account of what happens: the whole anterior part of the
alimentary canal becomes shut off from the hinder part, and its
two horns, which later become converted into the head cavities,
still communicate with one another after separation from the
gut has taken place (fig. 18, 5).

We have thus seen that the mesoderm originates as five
hollow outgrowths from the gut—an anterior median, viz. the
head cavity rudiment, and two pairs of lateral ones, viz. the
collar cavities and the coelomic grooves. Thus in the formation
of the primary layers Amphioxus is in fundamental agreement
with Balanoglossus, as described by Bateson (1). The main
difference between the two types is that whereas in Balano-
glossus the trunk cavity remains undivided, in Amphioxus it
becomes broken up into a series of segments, a difference
which we may plausibly correlate with the different modes of
life pursued by the two animals. PI. 45, figs. 25 and 26, are
two diagrams intended to make these relations clearer.

3. The Fate of the Celomic Cavities.

The two horns of the common head cavity rudiments rapidly
become separated from one another; the right now shows
itself as an irregular-shaped and thin-walled sac; the left, on
the other hand, is composed of cylindrical cells, and remains
small and round (fig. 18,@). The right soon after gets shifted
ventrally, and forms the greater part of the cavity of the
preoral snout during the whole of the larval period (compare
fig. 21, a). I have not been able to identify it in the adult,
and can only suppose that it becomes obliterated, the space
corresponding to it being apparently occupied by connective
tissue. The left, as is well known, acquires an opening to the
exterior, and constitutes the preoral ciliated pit (fig. 21, a),
which Hatschek first discovered. This preoral pit persists into

602 E. W. MACBRIDE.

the adult condition as a ciliated area on the inner side of the
preoral hood. By the time that the head cavities have com-
menced to appear, notochord and nerve-cord have become well
advanced in their development. The notochordal fold has
become completely shut off from the gut, and is quite solid in
front, though still a groove behind, whilst the neural plate has
passed through the stage of being covered by a flap growing
from the adjacent ectoderm (figs. 13, 14, and 16), and has
become converted into a tube (fig. 17, 6, c, and qd), still re-
taining an anterior opening, the neuropore (figs. 17,a@, and 18,6).
The collar cavities have now become large thin-walled sacs ;
the right extending by this time nearly to the mid-ventral
line; the left does not extend so far, but it still retains its
communication with the gut. This communication has by this
time become drawn into an exceedingly narrow tube (fig. 17, ¢,
neph.), and is in fact the rudiment of Hatschek’s ne-
phridium.

At the close of what we may term the embryonic development
—that is to say, at the end of the second day of development
—both collar cavities have undergone further changes. They
have extended forward at the sides of the notochord, above the
head cavities, to just beneath the neuropore (fig. 18, 6); behind
they have nearly reached the mid-ventral line, the left being
more obliquely directed, as it has to pass over the area where
the future mouth will be formed (fig. 19, 6). When they have
reached the ventral line they extend backwards to a con-
siderable distance behind the first gill-slit, forming the ventro-
lateral angles of the body, and giving to the transverse section
a squarish appearance ventrally, which contrasts strongly with
the rounded appearance behind the point they extend to
(compare Pl. 44, fig. 20, @ and c). The inner walls of the
dorsal portions of the collar cavities become like the corre-
sponding parts of the somites converted into longitudinal
muscles, and constitute in fact the first myotomes; but during
early larval life, at any rate, the persistent cavity of this “first
myotome” on the right side remains in open and obvious
communication with the ventral part of the collar cavity

THE EARLY DEVELOPMENT OF AMPHIOXUS. 608

(fig. 21, @). On the left side the front section of the ventral
collar cavity seems to become solid, and from it are apparently
derived the true oral tentacles, so well described by Lankester,
and the muscles moving them. ‘ Hatschek’s nephridium ”
has now become a horizontally placed tube, the openings of
which into the first myotome (collar cavity) on the left and
into the gut are easily seen (fig. 21,6 andc). It is curious
that this internal opening completely escaped the observation
of both Van Wyhe and Hatschek ; it is perfectly easy to see in
any good series of sections through a specimen preserved in
osmic acid. Hatschek (4) states that this nephridium persists
into the adult. I have, however, been able to find no trace
of it after the metamorphosis.

The ventral extensions of the collar cavities appear in the
larva to reach some distance behind the last gill-slit formed.
In any series of sections in this region we get quite similar
appearances to those represented in figs. 20, a, 6, and ec. The
natural conclusion to be drawn from this is that behind the
first gill-slit, as the gut grows and produces new gill-slits, the
collar cavities grow pari passu.

The ccelomic cavities in the somites we saw at their time of
formation to be exceedingly minute; in some cases it would be
more correct to say that there was no cavity at all, but only a
radiate arrangement of the cells round a virtual cavity. Such
a state of things might in the minds of some suggest that
there was something to be said for Lwoff’s position that
Amphioxus is not an enteroccelous animal, since its coelom
could not be traced into continuity with the alimentary cavity.
Such a position, however, appears to me to be quite untenable,

even apart from the fact that the collar cavity is at one time
in open and obvious communication with the enteric space.
What determines whether an animal is to be regarded as
enteroccelous or not is whether or not evidence is forthcoming
to show that the walls of the body-cavity have been derived
from those of the alimentary canal by a process of folding, for
it is the walls only to which we can attribute an objective
existence ; the presence or absence of a cavity at any moment

604 E. W. MACBRIDBE.,

of development is due to the relations of growth and pressure
subsisting between them. Noone who has seen well-preserved
sections can doubt that it is by a folding process that the
mesoderm is formed in Amphioxus.

Later the cavities of the somites enlarge, and the dorsal
portions of their inner walls are like the corresponding parts
of the collar cavities converted into longish muscles (Musc.,
fig. 20, a, 6, and c), and form all the myotomes except the
first. The ventral portions wedge themselves in between the
gut and the posterior ventral extension of the collar cavities
(V. Tr., figs. 19, c, 20, a, 6, and c). Whether in the living
condition these ventral portions are completely hollow, or
whether, as is suggested by the examination of sections,
they are partly represented by solid tongues of tissue, it is
impossible to settle. Of course in the adult the ventral por-
tions of the somites give rise to the dorsal ccelomic canals,
and to the canals running in the primary and secondary gill
bars (Lankester) ; the dorsal coelomic canals are clearly repre-
sented in the larva, but I have found it impossible to be
certain whether the rest of the ventral portions of the somites
are open spaces in the larva or not. Just at the close of
embryonic life the “ myotome” becomes separated from the
rest of the somite by a septum, and the ventral portions of the
somites acquire communication with each other, about the
region where the dorsal ccelomic canals afterwards appear.
This ventral fusion of the somites was inferred by Hatschek
from the fact that he could not trace the dividing lines
between the somites to the mid-ventral line. The specimens
of later larvae which were well enough preserved to rely on
showed the trunk ceelom (derived from the ventral fusion of
the somites) clearly only behind the gill-slits, where the
collar cavities were dying out. Elsewhere the extreme
difficulty of staining the connective tissue and peritoneal epi-
thelium made it impossible to be certain whether a narrow
slit-like cavity or only a wedge of tissue intervened between
the gut wall and the collar cavity.

THE EARLY DEVELOPMENT OF AMPHIOXUS. 605

4, The Origin of the Atrial Folds.

Kowalevsky (7) was the first to discover that the atrial
cavity was formed by the meeting in the mid-ventral line of
two long ridges or folds. These, which were more exactly
investigated by Lankester and Willey (9), are situated at the
ventro-lateral angles of the body, and the atrial cavity is at
first a small space situated in the middle line beneath the
pharynx (fig. 23, 6). Later the atrial cavity extends up at
the sides of the pharynx, and the origin of the folds becomes
consequently shifted up the body. This is the account of the
origin of the atrial cavity given by Lankester and Willey ; but
it must be remembered that as the dorsal limits of the atrial
cavity are from the beginning conterminous with those of the
gill-slits, the process might be more correctly described as a
great relative growth of the ventral region of the pharynx and
surrounding structures. Lankester (8) terms the folds which
actually wall in the atrial cavity ‘‘epipleural,” and the pro-
jecting angles after these folds have united ‘ metapleural.”
I shall use the term “atrial fold” to include the whole, of
which both are parts.

It must already have struck the reader that the posterior
ventral extensions of the collar cavities which I have described
above occupied precisely the region where the atrial folds
subsequently appeared ; hence it will not be surprising when
I state that the cavity of the atrial fold, termed by Lankester
and Willey “ pseudoceelic,” is nothing but the backward ex-
tension of the collar cavity. This I have succeeded in proving
for the right collar cavity (comp. fig. 22, a, b,and c) ; and since
the left collar cavity has precisely the course which Willey
describes for the oral hood and left atrial fold, no one will
doubt that this is the case for the left side also. From the
walls of these two collar cavities the ventral muscles of
Amphioxus are formed, and their lumen becomes occluded in
the centre (fig. 24), but remains at the sides as the “ meta-
pleural lymph canal.”

Lankester and Willey describe the atrial folds as first

606 E. W. MACBRIDE.

appearing behind, and then growing forward; but the first
recognisable trace of the future fold on the right side is an
epithelial thickening (fig. 22, a) in the anterior region of the
pharynx. This thickening, which later lines the outside of the
fold, is recognisable even at the end of the embryonic period.

It will be remembered that the name “ collar cavity ”? was
given to the coelomic pouches so denominated on account of
their general resemblance in mode of formation to the collar
cavities of Balanoglossus. The homology implied in this
name is borne out by the subsequent history of the sacs in
question; for (1) they remain distinct from the cavities
derived from the celomic grooves or trunk celom, and (2)
they swell out into ridges overhanging and protecting the
gill-slits, just as the hinder edge of the collar region does in
Balanoglossus, as Bateson (2) has pointed out,—only in that
animal, of course, at most two gill-slits are protected.

Résumé and Conclusions.

The most important points established in this paper are as
follow.

(1) The primitive gut or archenteron is formed in Amphi-
oxus by a typical process of embolic invagination, the endo-
derm being at first not sharply marked off from the ecto-
derm. The blastopore is at first posterior, but subsequently
becomes dorsal by the preponderant growth of the ventral lip
of the blastopore.

(2) The mesoderm originates in Amphioxus as a series of
true gut pouches, viz. one anterior unpaired pouch and two
pairs of lateral pouches. Of these the first divides to form
the two head cavities: the anterior pair give rise to the first
pair of myotomes, and in addition to two long canals extend-
ing back ventrally: the posterior pair are gradually separated
from the gut, and pari passu divided into a series of myo-
tomes. The whole process of mesoderm formation is therefore
referable to the type found in Balanoglossus, the main differ-

THE EARLY DEVELOPMENT OF AMPHIOXOS. 607

ence being that the pouch corresponding to the trunk ceelom
of Balanoglossus becomes segmented.

8. Hatschek’s nephridium is the persistent connection of
the left of the pair of collar-pouches with the gut.

4. The metapleural ‘‘lymph canals” found in the atrial
folds arethe persistent ventro-lateral extensions of the “collar-
pouches.”

The general conclusions which can, I think, be fairly de-
duced from the foregoing study are, in the first place, that all
attempts to explain the formation of the nervous system of
Vertebrates by the coalescence of the two halves of a nervous
ring lying in the lips of a long slit-like blastopore must be
given up, any appearances interpreted thus being due to
secondary disturbances introduced by increasing food-yolk ;
for of any such process no trace is observable in the simple
development of Amphioxus.

Secondly, that the theory of the descent of the Vertebrates
from a form somewhat lke Balanoglossus receives strong
support from the early developmental history of Amphioxus.
I say from a form somewhat like Balanoglossus advisedly, for
it may not be superfluous to lay stress on this point, that we
can no more suppose Vertebrates to be descended from Balano-
glossus than from Amphioxus. The main stem of the Verte-
brate phylum most probably continued throughout its whole
history to lead an active predatory existence, and Balano-
glossus and Amphioxus are to be regarded as degenerate off-
shoots from different levels of this stem. It is much more
probable that the Tornaria larva of Balanoglossus gives the
best idea of the remote ancestor of the Vertebrates, and in
this respect the condition of the nervous system in the larval
Amphioxus is of great interest. So far as we know central
nervous systems are generally developed in close connection
with prominent sense-organs. Now the Tornaria (in its later
stages) has two main nervous centres—(a) at the apex of the
preoral lobe, a sensory nervous plate with two eye-spots; (d)
a short nervous tube in the collar region. Of these, the
second has probably been developed in connection with a

608 E. W. MACBRIDE.

series of sensory tentacles such as Cephalodiscus possesses in
this region, and which probably correspond to the ambulacral
tube-feet or tentacles of Echinoderms. The first is lost in
Balanoglossus, owing no doubt to its burrowing life; but in
the free-living Vertebrate ancestor this would not have oc-
curred. As the preoral lobe became reduced in size (a
process which may have been connected with the giving up of
cilia as a means of progression and obtaining nutriment) the
two nervous centres of the Tornaria-like ancestor would be-
come approximated, and we should reach the condition which
we actually find in the Amphioxus larva, viz. a sense-plate
immediately followed by a nervous tube; for the part of
nervous system under the neuropore becomes pigmented, and
is sensitive to light. Figs. 27 and 28 are diagrammatic side
views of Balanoglossus and an Amphioxus larva, and are
intended to emphasise the immense diminution which the pra-
oral lobe has undergone in the latter.

If these conclusions are well founded, Amphioxus would
represent a more primitive offshoot from the Vertebrate stem
than Ascidians, for the larve of the latter possess a large
vesicular brain, which only retains a small pore leading into
the stomodeum. This deduction is, however, supported by
the fact that whereas the Ascidian larva possesses a long post-
anal muscular tail (a feature which has become more and more
accentuated in fishes), in the Amphioxus larva the anus is as
Hatschek pointed out, and as I can confirm, at the extreme
posterior end of the body on a vertical neurenteric canal, and
becomes only slowly and to a small extent shifted forwards
during development.

ZooLocicaL LABORATORY, CAMBRIDGE ;
August, 1897.

10.

i:

12.

13.

14.

THE EARLY DEVELOPMENT OF AMPHIOXUS. 609

List or PAPERS REFERRED TO IN THIS PAPER.

. Bateson.—‘‘The Early Stages in the Development of Balanoglossus

(sp.?),” ‘Quart. Journ. Mic. Sci.,’ 1884.

. Bateson, W.— The Later Stages in the Development of Balano-

glossus Kowalevskii,” ibid., vol. xxv, 1885.

. Hatscnex, B.— Studien iiber Entwickelung des Amphioxus,” ‘ Arb.

Zool. Inst. Wien,’ Bd. iv, 1881.

. Hatscuex, B.—‘ Mittheilungen titber Amphioxus,” ‘ Zool. Anz.,’ 7ten

Jahrgang, 1884.

. Hatscuex, B.—‘ Uber den Schichtenbau von Amphioxus,” ‘ Anat.

Anz.,’ 3ten Jahrgang, 1888.

. Kowatzvsxy, A.—‘ Entwickelungsgeschichte des Amphioxus lanceo-

latus,” ‘Mém. Acad. Impér. de St. Pétersbourg,’ tom. xi, 1867.

. Kowatevsky, A.—‘‘ Weitere Studien iiber die Entwickelungsgeschichte

des Amphioxus lanceolatus,” ‘Arch. f. mikr. Anat.,’ Bd. xiii,

1877.

. Lanxester, HE, Ray.— Contributions to the Knowledge of Amphi-

oxus lanceolatus, Yarrell,’ ‘Quart. Journ. Mic. Sci.,’ vol. xxix,
1889.

. Lanxester, HE. Ray, and Wittey, A.‘ The Development of the Atrial

Chamber of Amphioxus,” ibid., vol. xxxi, 1890.

Lworr, B.—‘ Die Bildung der primaren Keimblatter und die Entstehung
der Chorda und des Mesoderms bei den Wirbelthieren,’ Moskau,
1894.

Van WisuE, J. W.—“ Uber Amphioxus,” ‘Anat. Anz.,’ 8th Jahrgang,
1898.

Witt, L.—* Béitrage zur Entwickelungsgeschichte der Reptilien,” ‘ Zool.
Jahrbiicher,’ vi.

Wittzy, A.—‘‘ The Later Larval Development of Amphioxus,” ‘ Quart.
Journ. Mic. Sci.,’ vol. xxxii, 1891.

Witson, EH. B.—“ Amphioxus and the Mosaic Theory of Development,”
‘ Journal of Morphology,’ vol. viii, 1893.

610 E. W. MACBRIDE.

EXPLANATION OF PLATES 43—45,

Illustrating Mr. E. W. MacBride’s paper on ‘The Early
Development of Amphioxus.”

List or ABBREVIATIONS EMPLOYED.

Al, Alimentary canal. Ch. Notochord. Hnd. Endostyle. g/. Club-shaped
gland. Muse. Longitudinal muscles of the somites. N.c. Nerve-cord.
Neph. Persistent communication of the left collar cavity with gut, commonly
called “ Hatschek’s nephridium.” V. Coll. Ventral portion of collar cavity.
V. Tr, Trunk body-cavity produced by the fusion of the ventral portions of
the somites. «x marks the dorsal lip of the blastopore in gastrule.

The last four figures are diagrams; the outlines of all the rest lave been
drawn with the camera lucida.

PLATE 48.

(All the figures drawn with magnification obtained by Zeiss C, oc. 2.)

Fic. 1.—Section of young blastula.

Fig. 2.—Sagittal section of blastula which is just commencing to flatten.

Fic. 3.—Sagittal section of blastula which has flattened on one side.

Fic. 4.—Sagittal section of gastrula in which invagination is just com-
mencing.

Fig. 5.—Sagittal section of gastrula in which invagination is more advanced.

Fic. 6.—Sagittal section of gastrula in which invagination is still more
advanced.

Fic. 7.—Sagittal section of gastrula in which invagination is well advanced ;
the first trace of the flattening which marks the dorsal side is visible.

- Fig. 8.—Sagittal section of gastrula in which invagination is very advanced.

Fie. 9.-—Sagittal section of gastrula in which blastopore is becoming
narrowed.

Fic. 10.—Sagittal section of completed gastrula. The blastopore has been
shifted to the dorsal side (in consequence of a slight obliquity a piece of
the medullary fold is included in the section).

Fic. 11.—Transverse section of embryo about the age of that in Fig. 10
(7—8 hours from fertilisation).

Fic. 12.—Transverse section of embryo of about ten hours; first trace of
neural plate and ccelomic groove.

THE EARLY DEVELOPMENT OF AMPHIOXUS. 611

Fic. 13.—Transverse section of embryo of about ten hours; shows first
trace of collar cavity distinct from cclomic groove.

Fic. 14, a, 6, and c.—Three sections from a series through an embryo of
from ten to twelve hours in age.
a. Shows the celomic groove.
&. Shows the celomic groove closed off in front so as to form the
ccelom.
c. Shows the independent opening of the collar cavities.

Fic. 15, a and 6.—Two sections from a series cut parallel to the sagittal
longitudinal plane through an embryo of twelve to thirteen hours.
a. Shows the ccelomic groove and collar cavity opening into the archen-
teron.
6. Shows neurenteric canal and division of trunk ccelom into somites.

Fic. 16.—Transverse section of embryo of about thirteen hours, showing
the left collar cavity only, opening into the archenteron.

PLATE 44.

(All the figures on this plate are drawn under the magnification of a
Zeiss D, oc. 2.)

Fie. 17, a, 6, c, d, and e,—Five sections from a series through an embryo
of fourteen to fifteen hours.

a. Shows head cavities opening into gut and anterior pore of the nervous
system.

6. Shows head cavities constricted off from gut.

c. Shows downward extension of right collar cavity and persistent com-
munication of the left with the gut (Weps.) (Hatschek’s nephridium).

d. Shows the last somite, which is still continuous with the ccelomic
groove.

e. Shows the ccelomic groove.

Fie. 18, a, —Two sections from a series through an embryo of about
twenty hours.
a. Shows the right and left head cavities becoming differentiated the one
from the other, the collar cavities having shifted forward.
6. Shows that further back the two head cavities still communicate with
one another. The open neuropore is still seen.
Fie. 19, a, 6, and c.—Three sections from a series through an embryo of
about twenty-four hours.
a. Shows complete separation of right and left head cavities and the
ventral shift of the former.
6. Shows inequality of the two collar cavities ; the left retains a narrow
communication with the gut (NepA.).

612 E. W. MACBRIDE.

c. Shows backward continuation of the ventral part of the collar cavity
(V. Coll.). V. Tr. A solid wedge of tissue which represents the trunk
ccelom formed by the fusion of the ventral ends of the somites.

Fic. 20, a, 6, and c.—Turee sections from aseries through an embryo of the
same age as that figured in Fig. 19, showing the dying out of the ventral
extension of the collar cavity (Y. Coll.).

Fig. 21, a, 4, and c.—Three sections from a series through a pelagic larva
showing relations of the fully developed nephridium of Hatschek.

a. Shows the opening into the cavity of the first muscular somite (left
collar cavity).

6. Shows appearance of a section across the middle of the structure.

ec. Shows the opening into the pharynx.

PLATE 45.

Fig. 22, a, 6, and c.—The ventral portions of three sections through an old
pelagic larva, showing the formation of the atrial fold and the continuation of
the ventral part of the collar cavity into it. gl. Club-shaped gland.

Fig. 23, a and 4.—The ventral portions of two sections through a still
older pelagic larva, showing a further stage in the development of the atrial
folds.

Fig. 24.—Ventral portion of a section through a young adult Amphioxus,
showing the cavities of the metaplenral folds (the same as the collar cavities),
and the schizoceelic artefacts which appear to the outer side of them.

Fig. 25.—Diagram showing the origin of the various portions of the
cceelom from the gut in Amphioxus.

Fic. 26.—Similar diagram to show the origin of the ccelom in Balano-
glossus.

Fic. 27.—Diagrammatic lateral view of an Amphioxus larva, showing
mutual relationships of the head cavities, collar cavities, and trunk ccelom.

Fic. 28.—Similar diagram of Balanoglossus.

(Fig. 22, a, 6, and ¢ are drawn with Zeiss F, oc. 2; Figs. 23 and 24
with Zeiss D, oc. 2).

N.B.—In Figs. 17 to 20 the hypoblast and mesoblast are coloured red, and
the epiblast grey.

On the contrary, in Figs. 21 to 24 the epiblast is coloured red, the meso-
blast blue, and the hypoblast, including the notochord, grey.

ON DREPANIDOTANIA HEMIGNATHI, 613

On Drepanidotenia hemignathi, a New Species
of Tapeworm.

By

Arthur E. Shipley,

Fellow and Tutor of Christ’s College, Cambridge, and University Lecturer in
the Advanced Morphology of the Invertebrata.

With Plate 46.

Tue specimens of the above-named tapeworm, of which I
received but ten, are all small; they vary in length from
10 mm. to 22mm. The head is very small ; immediately be-
hind it, there being practically no neck, the body begins to
broaden out, and in some specimens the proglottides attain a
width of 2mm. The segmentation of the body commences
immediately behind the head, and is very well marked a little
further back. The posterior border of each segment overlaps
the succeeding one with a prominent edge or rim ; this is well
shown in longitudinal section (fig. 6). The number of seg-
ments varies from some fifty to sixty to over a hundred. The
measurements given above are about the average, but, as is
well known, tapeworms are extremely extensible animals, and
‘this to a great extent diminishes the value of figures quoted in
reference to their size. In some specimens the body is
stretched, and the length of the segments equals one half or
even two thirds of their breadth, but in the commoner forms
the segments are very short and broad, sometimes eight or ten
times as broad as long. They are flattened, as is seen in trans-
verse section, and sometimes, especially towards the posterior

vo. 40, PART 4,—NEW SER. UU

614 ARTHUR E. SHIPLEY.

end, the whole body is hollowed so that each segment is curved.
The most posterior segments, which are crowded with embryos
well advanced in their development, are rounder, less flat-
tened, longer, and they readily broke off.

I was not able to detect any genital pore on the exterior
even with the aid of powerful lenses, but sections (figs. 4 and
6) and stained mounted specimens show that it is on the same
side of the body in all the segments.

The head of the tapeworm bears four suckers, and in the
midst of them is the rostellum (fig. 9). The shape of the
head is very various: in some cases the suckers are, as it were,
hunched up and lying at each corner of a square, the lateral
diameter of which does not exceed the dorso-ventral (fig. 8) ;
in other specimens the head is not separated from the body by
a deep constriction, but is flattened and spread out (fig. 7), so
that the lateral suckers are separated from one another by a
space considerably wider than that which lies between the dorsal
and the ventral suckers.

The rostellum is minute and sunk in a pit (fig. 3); it bears
a wreath of ten hooks. In all the specimens which I cut into
sections, and I think in the others as well, the rostellum was
retracted, the points of the hooks folded in against the axis of
the rostellum, and not reaching so far forward as the mouth
of the pit. When the animal is fixed to the mucous mem-
brane of its host this rostellum is doubtless protruded from
its sheath, and the hooks are divaricated. Certain muscle-
fibres which run from the base of the rostellum, and lose them-
selves in the parenchyma, probably serve to retract it.

The hooks are slightly curved, and the projection which
corresponds with the inner fork of the more triradiate hooks
of other genera is hardly, if at all, marked (fig. 2). Measur-
ing in a straight line from the base to the tip the hooks are
18—23 in length, thus corresponding pretty closely with
those of Drepanidotenia tenuirostris, which, according
to Railliet,! measure 20 to 23 uw, and to those of D. lanceo-
lata, which measure 25 to 31m.

1 «Traité de Zoologie médicale et agricole,’ Paris, 1895.

ON DREPANIDOTHNIA HEMIGNATHI. 615

The four suckers present no peculiarities; they are deeply
cupped, with a small orifice to their lumen, but probably they
are capable of considerable change of form (fig. 9). They are
probably retracted by some muscle-fibres which cross one
another and run into the parenchyma.

The segmentation of the body begins immediately behind
the suckers; at first the segments are very short, but they
gradually increase in size throughout the first three quarters
of the length of the body. For the last quarter the seg-
ments are crowded with embryos; they become in this region
much narrower, more cylindrical in shape, and longer, and
are very easily broken off. The posterior free edge of the
segments of the anterior two thirds of the body is sharp, and
may overlap the segment behind, or may stand out clearly
from it.

The water-vascular system is weil developed; on each side
of the body are two longitudinal canals,—one, the ventral, much
bigger than the other, or dorsal. The lining of the former
seems to be a structureless cuticle with no cells especially
related to it, but the wall of the dorsal vessel is surrounded by
a number of small deeply stained cells (fig. 4). I did not see
any communication between the vessels of one side, but the
larger vessels communicate as usual, one with another, by a
transverse vessel running from side to side along the posterior
border of each segment. In the head the vessels all com-
municate. In some of the better preserved sections such
structures as are depicted in fig. 10 were seen: these may or
may not be flame-cells; they look rather like them. No
valves were seen in the course of the vessels.

The lateral nerve-cords are well marked, lying externally to
_ the ventral excretory canals; they fuse together in the head,
forming a ganglion which is indicated in fig. 3. No traces of
the nerve-ring described by Tower! as running round the
posterior end of each segment of Moniezia, or of the secon-
dary nerves described by the same observer, were to be seen.
But these, if present, probably require fresh material and

1 *Zool. Anz.,’ vol. xix, 1896, p. 323.

616 ARTHUR E. SHIPLEY.

special methods of preservation to make them manifest.
Special nerve-cells, described below, are scattered through the
parenchyma of the body.

The histology—at least in some specimens—could be fairly
well made out, and agrees roughly with what Blochmann has
described in Ligula monogramma.! The whole body is
covered by a cuticle, the outer fifth of which stains more
deeply than the remainder. Within this, with a high power,
a number of dots or knobs become visible (fig. 10). These
are the swollen terminations of certain strands or processes of
the ectoderm cells. The cells themselves, as Blochmann has
shown, lie removed to some distance from the cuticle they
secrete, but are in contact with it by means of the above-
mentioned processes ending in the knobs.

The ectoderm cells are not all at one level, but on the
whole form a fairly well-marked layer. Each cell is fusiform
in shape, and produced into two or three processes, which
project both peripherally and centrally. They contain large
and well-marked nuclei. Neither the cells nor their processes
are laterally in contact ; they are separated one from another
to varying extents by the intrusion of some of the paren-
chymatous network which makes up so much of the body of
a Cestode.

This parenchyma consists of a meshwork which permeates
everywhere the body of the tapeworm, surrounding all the
organs, and often, as is the case with the ectoderm and the
muscles, passing in between their constituent cells. In the
spaces of the meshwork there is believed to be a fluid. The
meshwork itself is secreted and nourished by certain large
star-shaped cells which are irregularly scattered through the
parenchyma, and which give off processes in all directions
(fig. 10).

Round the generative glands this parenchymatous network
becomes condensed, the spaces disappear, and it forms a close
sheath to the ovary, testis, &c. At the posterior end of each
segment it is also somewhat condensed, and in section presents

1 «Die Epithelfrage bei Cestoden und Trematoden,’ Hamburg, 1896.

ON DREPANIDOTENIA HEMIGNATHI. 617

the appearance of a well-marked double line, which is very
characteristic, and is well shown in fig. 6.

Scattered amongst the parenchyma are certain faintly stained
cells which seem to be bipolar, and which differ from the cells
of the parenchyma both in shape and in their powers of
absorbing the staining reagents. These I take to be nerve-
cells which are in communication with the nerve-fibres of the
lateral cords. The latter are entirely devoid of any nerve-
cells on their course.

Muscle-fibres are scattered through the substance of the
body, and one set of longitudinal muscles are most definitely
arranged. This layer is situated just below the epidermis in
the anterior part of the segment, but as the latter increases in
size posteriorly, the cylinder of muscle-fibres, which retains
the same diameter throughout, comes to lie more deeply in
the tissues. These muscles, like the nervous system and ex-
cretory canals, run from segment to segment; some of them,
if not all, end in the cuticle, where it is most bent in at the
posterior end of each segment. Laterally the fibres are not
in contact, being separated by considerable intervals. Their
regular arrangement is shown in fig. 6.

In the posterior segments, which are so ripe that the slightest
touch breaks them off, the parenchyma has undergone con-
siderable degeneration, the cells are less clear, and the spaces
of the meshwork are larger and more irregular.

The generative organs begin to arise very early in the series
of segments. Already in the eighth or tenth segment clusters
of cells are segregating, and their deep staining shows that
they belong to the gonads. In the sexually ripe segments the
ovary is centrally placed, and is supported on each side by a
lobe of the testis. From the latter a fine vas deferens leads
into an extensive vesicula seminalis, which is as a rule crowded
with spermatozoa; from this a muscular duct leads to the
unilateral genital pore. I was unable to make out the details
of the penis, and similarly I failed to detect any yolk-gland
amongst the female genitalia.

The vagina leads at once into a large receptaculum seminis,

618 ARTHUR E. SHIPLEY.

whose walls were strengthened by a series of cuticular-looking
rings, whose cut ends are shown in figs. 4 and 6. This com-
municates both with the oviduct and with the uterus. The
latter presents no special points of interest; in the posterior
segments it contains the typical three-hooked larve, each
segment containing at least one hundred and probably more.

SYSTEMATIC.

In his paper on tznias in birds, Dr. Fuhrmann! remarks
that of the 240 odd species of tapeworm described from avian
hosts, only twenty-one have been studied anatomically; the
remainder are but little more than names, and probably many
of the names are of doubtful validity.

A certain amount of order has been introduced into this mass
of material by the establishment of certain sub-groups, and by
the giving of a new generic name to the members of these sub-
divisions; thus in 1891 Blanchard and Railliet? established the
genus Davainea; in 1892 Railliet® suggested two new generic
names, Drepanidotenia and Dicranotenia, for certain
tapeworms inhabiting, for the most part, domestic birds.
These are characterised chiefly by the nature of the hooks.
In the following year Diamare* founded the genus Cotugnia,
in which the generative organs are double and have two
pores, but which is distinct from the genus Dipylidium
of Leuckart. All these genera are characteristic avian
tapeworms, and are, with but very few exceptions, confined to
birds.

There is little doubt that the tapeworm which I have
described above from the intestine of Hemignathus pro-
cerus corresponds with a Drepanidotenia of Railliet,° who
defines his genus as follows:

“‘ Tapeworms provided with a simple crown of uniform hooks,

1 © Rev. Suisse Zool.,’? tome iii, 1895-6, p. 433.

2 © Mem. Soc. Zool. France,’ tome iv, 1891, p. 420.

3 Thid., tome xvii, 1892, p. 115.

4 «Boll. Soc. Napoli,’ ser. 1, vol. vii, 1893, p. 9.

5 «Traité de Zoologie médicale et agricole,’ Paris, 1895, p. 298.

ON DREPANIDOTHNIA HEMIGNATHI. 619

which are usually few in number; the outer limb (manche)
of the forked base of the hooks is much longer than the inner
(garde), which is always slight ; the point is directed backwards
when the rostrum is withdrawn. The majority live in the
intestines of aquatic birds. Their larva is a Cysticercoid, and
is found encysted in the bodies of small fresh-water Crus-
tacea.”

Railliet describes eight species of Drepanidotenia; in one
of these the genital pores are on alternate sides of the body in
successive segments: the remaining seven species are unilateral
in this respect, but they fall into two groups,—one, with three
species, in which the number of hooks is eight ; and the other,
with four species, in which the number of hooks is ten.

It is to this latter group that we must add the tapeworm
from H. procerus. The four species D. anatina, D.
sinuosa, D. setigera, and D. tenuirostris differ inter se
in several respects, but perhaps the simplest way of deter-
mining the species is by measuring their hooks. Of these four
species, D. hemignathi most nearly resembles D. tenui-
rostris, which occurs in certain of the ducks; it differs, how-
ever, markedly in size, being when mature about + to ;4, the
length of the last named. It resembles D. tenuirostris in
the length of its hooks in the head, which in the latter are
20—238 yw, in the former are 18 to 23 «4; but whereas the hooks
of the embryo are about the same length in the new species,
i.e. about 20, in D. tenuirostris they are but 7p. The
neck is short, not long as in the last-named species, and the
eggs are small, about 40—50 in diameter, and spherical in
shape, not cylindrical as Krabbe! figures them, with a length
of 85 u. The hooks also differ in shape; those of D. tenui-
rostris have a much more strongly developed process corre-
sponding with the inner limb of the forked base than occurs in
D. hemignathi.

The new species, which I have named after its host, may be
characterised as follows :

1 « Danske Selsk. Skr.,’ Bd. viii, 1870, p. 249.

620 ARTHUR E. SHIPLEY.

Drepanidotenia hemignathi, n. sp.

Length 1—2°2 centimetres; breadth, in the middle of the
body, 2 millimetres. Head flattened and compressed, rostrum
with a crown of ten hooks; each hook 18—23 » in length, and
with but a slight trace of the inner limb of the forked base.
Neck short. The first segments are short, but they very soon
(eighth or tenth) show traces of reproductive organs. Genital
pore unilateral. The posterior limit of each segment is sharply
defined, and forms an angle of about 45 degrees with the
sides. Egg spherical, diameter about 40—50y. The three
pairs of embryonic hooks measure about 20 u each in length.

Habitat: Hemignathus procerus, Sandwich Islands, in
the intestine.

Note 1.—In a paper which I published in this Journal? last
year on Arhynchus hemignathi I stated that the parasites
were found “adhering lightly to the skin around the anus.”
I had this description from Mr. Perkins, and I understood it
to imply that the parasites were outside the body. In this I
find I was mistaken; and fearing that others may be under a
similar misapprehension, I am writing this note to say that
they occur inside the body-cavity in the angle where the rectum
joins the external skin.

Note 2.—Mr. Perkins has also given me two or three speci-
menus of a tapeworm from a Loxops, sp. This bird, like the
Hemignathus, is a member of the family Drepanidide, which
is confined to the Sandwich Islands. Unfortunately the
specimens are without their head, and I am unable to identify
them. They differ markedly from the Drepanidotenia de-
scribed above.

THE ZooLtocicAL LABORATORY,
CAMBRIDGE ;
July, 1897.

1 «Quart. Journ. Mier, Sci.,’ vol. xxxix, 1897, p. 207.

ON DREPANIDOT@NIA HEMIGNATHI. 621

EXPLANATION OF PLATE 46,

Illustrating Mr. Arthur E. Shipley’s paper “On Drepanido-
tenia hemignathi, a New Species of Tapeworm.”

List of Abbreviations.

cut. = Cuticle. dors. ex. can. = Dorsal excretory canal. ectod. = Kcto-
derm. at. nerve = Lateral nerve. dong. muses. = Longitudinal muscles.
Nerv. syst. = Junction of nerve-cords in head. ov. = Ovary. par. cell =
Parenchyma cell. Proc. of ectod. = Knob-like ends of ectoderm cells under
cuticle. rec. sem. = Receptaculum seminis. Ros. = Rostellum. ¢es. =
Testis. wt. = Uterus. ven. ex. can. = Ventral excretory canal. ves. sem. =
Vesicula seminalis.

Fie. 1.—A view of Drepanidotenia hemignathi, x 15. The dark
patches in the anterior two-thirds of the body are caused by the generative
organs; in the posterior third they represent the eggs in the uterus.

Fic. 2.—An isolated hook from the rostellum, x 500.

Fic. 3.—A longitudinal section through the head, x 100. ‘The rostellum,
ros., is retracted. The point of fusion of the two lateral nerves is shown at
nerv. syst. The section passes between the suckers.

Fic. 4.—A transverse section through a mature proglottis, x 70.

Fic. 5.—A longitudinal section, somewhat oblique, showing the regular
arrangement of the longitudinal muscles, x 50.

Fic. 6.—A longitudinal section through several mature proglottides, x 50.
This shows the transverse connection between the two ventral longitudinal
excretory canals and the transverse lines formed by the concentration of the
parenchyma at the posterior end of each proglottis.

Fic. 7.—A view of the head in an expanded, flattened-out state, x 60.

Fic, 8.—A view of another head in a contracted, bunched-up condition,
x 40.

Fic. 9.—A transverse section through the head, showing the ten hooks on
the rostellum and the four suckers.

Fic. 10.—A portion of a proglottis, highly magnified to show the minute
anatomy, X 450,

=
A al enn ha

el ;

rt
a
ye oa J
a
fa aS _ Ps
— re) ar
e

s- 2,
, -
F
i
">
4
‘
} ‘
4
P ‘
o VP
is
’
T *
‘
;
i
o
Se
- ‘
~
f i vt
'
’
/
‘
fF
jay vom?
A eis
Ly
®
4
! ’
)
bed 4
L
.
as
4 a

My % a a ; ;
4 ; 7 ' - ; o ae | r ‘
avon Pee dee la! meee a
' , pe ch en a ‘ee er ees
7 Ss iw - ‘ = =
‘hd or ay ‘ ‘ —— yi : ?
a oe tone : 2 * ae eres
ae * v ae 3 Fe
ay ip . : p' a ~ .
ae ws t
oh Pay 2S
ede ove U 24 Ey he pe stii ) 7
‘ +
n Ah j ! j } wv, ys wt =
: '
‘ 2 = a
we ’
(
. 2 = ‘
. 7 ee
\ P >» ke ae
%
bo '
. g ;
am ’
= te
» s*
' a6
“t , + > Tae

at f vi | om, il ts
“4 ie at * f wr ory ore f si
' 7 a ee | Pee
hes : ?, : we ete Aig
‘ ; a -- 3
: ' j
ah ies Ve SPR
: na _ : Algae ume inn ye eat r Ti
& A a?
rar ‘ PL ae | a al 4 ee ey ra | a is trys i * al h ;
a eh + , Ph “ chal Fils
‘ i = 1
© t
woe Prthd .
‘ Te }
’ 7 2" : - M o
; 2 6
e Sune
= a)
’ , = ¥
? i J pid
+ ‘ ry -_ =
in 4) F
. i rs. a ae i
a fate
a4 ‘i or A
; 7 > @& Se ee y
; a ee '

SPENGELIA, A NEW GENUS OF ENTEROPNEOSTA, 623

Spengelia, a New Genus of Enteropneusta.
By

Arthur Willey, D.Sc.,
Balfour Student of the University of Cambridge.

With Ptate 47.

Part I.—Draenosts.

From a rock-pool on the weather side of Lifu (Loyalty
Islands) last year I obtained a single specimen of an Ente-
ropneust living in company with Ptychodera flava,! which
has proved, on examination, to constitute the type of a very
distinct new genus.

Having been informed by Mr. J. P. Hill that he had
received two kinds of Enteropneusta from Funafuti, one of
which was Pt. flava, and the other a new species, I sent my
material to him for comparison. Mr. Hill saw at once that
my form was quite distinct from the Funafuti species, and he
had the goodness to leave it intact until my return to Sydney.

The genera of Enteropneusta, as defined by Spengel, fall
naturally into three groups, for which it is to be hoped
Professor Spengel will shortly create family names.

Group I, including the genus Ptychodera, briefly charac-
terised by the presence of an outer layer of circular muscles

1 Inaformer paper on Pt. flava I suggested to add the name “cale-
donica”’ until Eschscholtz’s species should be re-examined. Meanwhile Mr.
J. P. Hill has informed me that the same species occurs at Funafuti.

Under these circumstances it will be well to cancel the name “ caledonica,”’
and if the form from the Marshall Islands turns out to be different, then its
name, i. e. the name given by Eschscholtz, must be changed,

624 ARTHUR WILLEY.

in the integument of the trunk, the occurrence of dorsal roots
putting the fibrous layer of the collar nerve-cord in connection
with the fibrous layer of the epidermis, and the presence of
liver saccules, and of synapticula between the branchial bars.

Group IT, ineluding the genera Schizocardium and Glan-
diceps, characterised by the presence of an inner layer of
circular muscles (inside the longitudinal layer), and by the
occurrence of a long vermiform process extending forwards
from the anterior end of the notochord or proboscis czcum.!

Group III, including the genus Balanoglossus, charac-
terised by the absence of circular muscles in the integument
of the trunk, and by the absence of synapticula.

Spengelia belongs to the second of the above groups, but
exhibits features which render it a remarkably synthetic
genus.

The following table (compiled from Spengel) will suffice to
show the relation of Schizocardium and Glandiceps to one
another, and will assist in the appreciation of the characters
of Spengelia.

SCHIZOCARDIUM.

1. Ventral septum of proboscis ex-
tends to anterior end of vermi-
form process of notochord.

2. Pericardial auricles well deve-
loped.

3. Peripharyngeal spaces present.

4. Synapticula present.

5. (@sophageal portion of branchial
sac reduced.

6. Liver saccules present.

. Medial gonads absent.

. Anterior, unpaired, post-branchial

intestinal pores (Darmpforten)
present.

oo

GLANDICEPS.

Ventral septum of proboscis stops
short at base of vermiform process
of notochord.

Pericardial auricles rudimentary.

Peripharyngeal spaces absent.

Synapticula absent.

(Hsophageal portion of branchial sac
well developed.

Liver saccules absent.

Medial gonads present.

Ditto.

1 Jt would be desirable to translate the German word “ Hicheldarm” in
such a way as not to involve an abstruse morphological conception, which

some authors object to,

SPENGELIA, A NEW GENUS OF ENTEROPNEUSTA. 625

9. Posterior, paired, prehepatic in- Ditto.
testinal pores present.

10. Accessory genital pores present, Ditto.
which, when they occur laterad
from the main series, perforate
the longitudinal musculature.

Spengelia agrees with Glandiceps in the characters mentioned
in the above table under Nos. 1, 2, 3,5, 7, and probably 6.!
It only agrees specially with Schizocardium in the possession
of synapticula (No. 4 in above table), this being also a marked
Ptychoderoid feature.

A further most interesting reminiscence of the Ptychodera
type in the organisation of Spengelia is the occurrence of
vestigial roots arising from the dorsal side of the collar nerve-
cord.

SPENGELIA POROSA, nov. gen. et sp.

External Form. [See Pl. 47, fig. 1.]

The proboscis was longer than the collar, measuring, when
extended, 10°5 mm.? It was pear-shaped, and of a rich yellow
colour. The collar measured 6°25 mm., and was coloured a
rich orange, especially in the middle region, while the posterior
region of the collar was whitish yellow. The rest of the body
had a dull yellow colour,

The branchial region was 30 mm. in length, and in this
region the body was quite cylindrical and faintly annulated.
The post-branchial portion of the body, present in the ane
cimen, measured about 20 mm.

Apart from the absence of genital pleura, Spengelia was
readily distinguished from among the multitude of Ptycho-
dera flava by the length of the proboscis and the bright
orange-coloured collar.

But what at once distinguishes Spengelia from any other
Enteropneust hitherto described is the occurrence on each
side of the dorsal middle line of a series of deep dermal pits

’ Unfortunately the post-genital region was lacking from my specimen.

? After preservation in a picro-acetic mixture the proboscis measured
5°25 mm., and the collar 4 mm.

626 ARTHUR WILLEY.

in the post-branchial genital region. The mouth of each pit
is about 1 mm. in diameter.

At a glance the pits appear to be regularly paired, but a re-
examination has shown that they are not quite so regular as
represented in fig. 1.

In the fresh condition the sides of the genital region of the
body were occupied by elongated, somewhat pyriform bodies,
which caused definite ridges on the external surface. These
projections were caused by the gonads. The individual was a
mature male.

The dermal pits lie in the submedian line, in direct con-
tinuation from the branchial groove, and the most anterior
pits invade the posterior extremity of the branchial region, in
consequence of which a number of the outer pores of the
posterior gill-slits do not open near the surface of the body,
but deep down at the base of the dermal pits.

The last newly formed gill-slit on each side opens directly
from the gut into the base of a dermal pit; and at first I
thought they represented the intestinal pores (Darmpforten)
described by Spengel (discovered originally by Schimkewitsch).

Under some circumstances it might be very difficult to
distinguish true gill-slits from intestinal pores.

Spengelia differs from Schizocardium and Glandiceps in
the absence of anterior intestinal pores. I am unable to say
anything about the posterior intestinal pores, as my specimen
was incomplete.

Transverse sections show that the dermal pits of Spengelia
are remarkably deep, extending through more than half the
thickness of the body, and actually branching amongst the
gonads. They appear to serve for the irrigation of the gonads.
If anyone saw a single section passing through the middle of
a dermal pit, he would say that Spengelia possessed genital
pleuree. Perhaps these pits owe their origin to an incomplete
fusion of genital pleure with the body-wall; or they may
merely represent local depressions of the floor of a branchio-
genital groove.

Apart from their connection with the gut by means of the

SPENGELIA, A NEW GENUS OF ENTEROPNEUSTA. 627

posterior gill-slits, the dermal pits do not communicate with
the intestine, although they extend very near to the wall of
the latter.

INTERNAL STRUCTURE.

1. Vermiform Process of Notochord.—This is a long,
generally solid cord of cells, lying in the centre of the pro-
boscis, and surrounded by a stout limiting membrane, which
serves for the insertion of the median dorso-ventral muscles
of the proboscis. Its diameter is not quite equal throughout
its course. It agrees closely with the corresponding process
in Glandiceps, particularly in the fact that the ventral
septum of the proboscis does not accompany it, as it does in
Schizocardium. =

2. Collar Nerve-cord.—One of the most interesting an
generically important characters of Spengelia is the occurrence
of vestigial dorsal roots. They do not reach the epidermis,
nor do they contain fibres or “ Punktsubstanz.” Otherwise
their similarity to the roots of Ptychodera is complete. I have
seen two such roots in Spengelia. The anterior root is the
longer, and it runs obliquely, so that it appears in several
sections separate from the nerve-cord. It is mostly solid, but
contains a few minute disconnected cavities. The posterior
root is hollow and much shorter than the anterior root, so that
it does not appear in section separate from the nerve-cord.

3. Splanchnic Nerve-fibres.—A rather puzzling feature
in the anatomy of Spengelia is the occurrence of a layer of
nerve-fibres (Punktsubstanz) at the base of the epithelium of
the buccal or throat cavity. Anteriorly it is a thick layer, and
it becomes gradually thinner posteriorly. It may be traced as
a very thin layer for a long distance beyond the opening of
the notochord into the buccal cavity, and even at the base of
the epithelium forming the cesophageal portion of the branchial
sac. The occurrence of this well-defined layer of splanchnic
nerve-fibres round the throat and esophagus alone distin-
guishes Spengelia from all other Enteropneusta.

628 ARTHUR WILLEY.

4. Synapticula.—By preparing out a piece of the wall of
the branchial sac I became aware of the presence of synap-
ticula in Spengelia before seeing them in section (fig. 3).

5. Gonads and Genital Ducts.—In the branchial region
gonads occur both medially and laterally, that is on each side
of the branchial groove (fig. 2). Their ducts open at the lips
of the latter. In the post-branchial region where the dermal
pits occur, the genital pores are numerous, and are not confined
to the submedian line, so that several genital pores may be
seen in one section. Some genital pores open into the dermal
pits, while others open directly to the exterior near the dorso-
lateral margins of the body. But in Spengelia, contrary to
what obtains in other Enteropneusta (with exception of
Balanoglossus canadensis and of the mediad accessory
pores of Schizocardium brasiliense), the accessory genital
ducts and pores do not perforate the longitudinal musculature.
In the post-branchial region of Spengelia there is a very wide
interval between the dorsal longitudinal muscles and the
ventro-lateral longitudinal muscles, and all the genital pores
occur in this interval.

6. Miscellaneous.—With regard to other points, it is
only necessary to mention here that Spengelia agrees with
Schizocardium and Glaudiceps in having an unpaired asym-
metrical proboscis-pore. ‘The canal (Eichelpforte) leading to
the pore swells out into a large vesicle before discharging to
the exterior, and at the base of the vesicle there are muscle-
fibres presenting the appearance of a sphincter muscle. .

Spengelia further agrees with Glandiceps in the massive
development of chondroid tissue in the neck of the proboscis
and in the length of the posterior cornua of the proboscis
skeleton.

In the preserved condition the gill-pores were clearly visible
at the base of the branchial grooves, as they are in Glandi-
ceps talaboti.

7. Summary.—lIf it were not for the presence of vestigial
roots in the collar nerve-cord, Spengelia (apart from its
own peculiar features, e.g. dermal pits, splanchnic layer of

SPENGELIA, A NEW GENUS OF ENTEROPNEUSTA. 629

« Punktsubstanz,” accessory genital pores not perforating the
longitudinal musculature, &c.) might almost be defined as a
Glandiceps with synapticula. Spengel specially mentions
the absence of synapticula as indicating the primitive character
of Glandiceps. In a recent paper in the ‘ Quarterly Journal
of Microscopical Science’! dealing with impressions of
Ptychodera flava derived from an examination of fresh
material, I expressed the opinion that Ptychodera presented a
more primitive type of organisation than the other known
Enteropneusta. The discovery of this new genus, Spengelia,
which has so many points in common with Glandiceps (see
above), and yet which has synapticula between the branchial
bars and vestigial roots arising from the collar nerve-cord, goes
a long way to prove that Ptychodera is relatively primitive,
and that Glandiceps and Balanoglossus are derived forms.

I hope to supplement the above account by a second part
containing an illustrated description of the anatomy of Spen-
gelia. What has been said is enough to establish the genus.

My work, so far as it has gone on the preserved specimen,
has been carried out in Professor W. A. Haswell’s laboratory
at the University of Sydney. I made some sections of
Ptychodera flava for purposes of control, but in addition I
have had the advantage of examining Mr. J. P. Hill’s beautiful
preparations of three species of Ptychodera, besides having
his opinion on various points in my own preparations of Spen-
gelia, notably on the vestigial roots and the splanchnic layer
of ‘ Punktsubstanz.”

Sypney; May 10th, 1897.
1 Vol. 40, p. 165.

voL. 40, PART 4.—NEW SER. xx

680 ARTHUR WILLEY.

EXPLANATION OF PLATE 47,

Illustrating Mr. Arthur Willey’s paper, “ Spengelia, a New
Genus of Enteropneusta.”

Fic. 1.—Dorsal view of Spengelia porosa, after a sketch from the
living animal, showing the dermal pits behind the branchial grooves. p. Pro-
boscis. ¢. Collar. d. g. Median dorsal groove. 4r. g. Branchial or branchio-
genital groove. d.p. Dermal pits.

Fig. 2.—Macroscopic section through the mid-branchial region of Spengelia,
showing the medial gonads and the cesophageal groove. m.g. Medial gonad.
d.1.m. Dorsal longitudinal musculature. 47. g. Branchio-genital groove.
v. l,m. Ventro-lateral longitudinal musculature. 7. g. Lateral gonads. @.
(isophageal groove. 4r.s. Branchial sac.

Fic. 3.—Portion of branchial skeleton of Spengelia, to show synapticula.
From a preparation treated with caustic soda and mounted in glycerine.
Zeiss, 3 A, cam. luc, ¢. 4. Skeleton of tongue-bar. s. d. Skeleton of primary
or septal bar. sy. Synapticula.

ON A PRORHYNCHID TURBELLARIAN. 631

On a Prorhynchid Turbellarian from Deep Wells
in New Zealand.

By

William A. Haswell M.A., D.Sc., F.R.S.,
Professor of Biology, Sydney University.

With Plate 48.

In 1892'I announced briefly the discovery of this remarkable
Turbellarian, which was found by Dr. Charles Chilton in deep
wells in Canterbury, New Zealand, among other animals,
chiefly Crustacea, of which he has since published a valuable
account.” At that time the general examination which I had
made of the specimens had led me to the conclusion that the
new form found its nearest allies in the Alloiocela. The
series of lateral diverticula of the intestine, the complex struc-
ture of the pharynx, the entire absence of a body-cavity, and
other features led me to take this view. A more thorough
examination, however, with the aid of sections of better pre-
served specimens, has led to the result that, while having
certain points of affinity with the Alloiocela, the new form
finds its nearest allies by far in the family Prorhynchide,
and is in many respects closely related to the genus Pro-
rhynchus, while presenting some remarkable features in which
it differs not only from that genus, but apparently from all the
rest of the Rhabdoccela.

1 “Jottings from the Biological Laboratory of Sydney University, No. 17,
Three Zoological Novelties,’ ‘Proc. Linn. Soc. New South Wales,’ 2nd

series, vol. vii.
2 «Trans. Linn. Soe.’ (2), vol. vi, pp. 163—284.

6382 WILLIAM A. HASWELL.

Of the four specimens received three are sexually mature,
and agree with one another in all essential points; the fourth,
somewhat smaller than the rest, differs from them in the
rudimentary condition of the reproductive apparatus. In none
of them is the state of preservation very perfect in all respects,
and the following description is necessarily incomplete in a
good many points. I am also unable, owing to a good many
gaps in our Sydney scientific libraries, to attempt to deal in
any complete manner with the literature.

The total length of the largest specimen is 2°5 cm., and the
greatest breadth 4mm. No pigment is present in any part,
and there are no eyes. The body (fig. 1) is elongated, con-
siderably depressed, thick in the middle where the alimentary
canal and other organs lie, thin at the sides; these lateral thin
regions assuming the character of thin solid longitudinal
flanges. The anterior extremity is broad and truncate, the
antero-lateral angles produced into small compressed projec-
tions; the posterior portion of the body tapers to a blunt
point. The mouth is a wide aperture at the anterior end.
The openings of two ciliated sacs (c. s.) are situated on the
ventral surface, a little behind the antero-lateral angles. The
two excretory apertures (exv.) are situated on the ventral
surface, some distance apart, a little in front of the middle
of the body. The female generative aperture (?) is also
situated on the ventral surface in the middle line, some little
distance behind the excretory pores.

Integument, Integumentary Glands, Muscular
Layers, and Parenchyma.—The specimens are not in a
sufficiently good state of preservation to permit of a detailed
investigation of the structure of these parts, and the following
account is necessarily incomplete. The presence of vibratile
cilia on the surface cannot be determined with certainty ; but
elevations of the integument, the arrangement and structure
of which remain uncertain, bear, singly or in groups, long
straight cilia, probably of a non-motile and sensory character.
The cuticle (fig. 3, c.) is (002 mm. in thickness; it does not
appear completely homogeneous, but exhibits indications of

ON A PRORHYNCHID TURBELLARIAN. 633

structure. It is perforated by innumerable minute openings—
the openings of the ducts of the integumentary glands. In
the epidermis, which is about ‘01 mm. in thickness, nuclei,
though they doubtless exist, are not to be made out with cer-
tainty in any of the series of sections. The rounded clear
spaces, the presence of which is so characteristic of the
epidermis of the Turbellaria in general, occur abundantly. A
distinct, though thin basement membrane (0.) lies below the
epidermis. |

All over the surface there occur at fairly wide intervals (on
an average 0°5 mm. apart) remarkable unicellular glands (g/.),
which for the sake of distinction I will term the superficial
integumentary glands. These are pear-shaped when seen in
vertical section, but in transverse section often present a
stellate appearance, owing to the presence of processes given
off from the basal part; the total length, including the duct,
is‘O7 mm. The broad end lies altogether beneath the epidermis
and basement membrane, in the zone of ducts, to be presently
referred to, while from the narrow end a short duct pierces the
integument to open on the surface. The most striking feature
of these glands is the enormous relative size of the nucleus,
which almost completely fills the cell, so that sometimes only
a very thin layer of cytoplasm is distinguishable around it.
This cytoplasm, prolonged into the narrow neck of the cell, is
of a very finely granular character, without any appearance of
reticular or other structure; it has become stained by the
eosin in the sections, and the substance of the deeper or basal
part of the cell is much more intensely stained than the rest.

Immediately below the basement membrane is a thin layer
_of circularly arranged muscular fibres—the outer circular
layer (e.c.), and in immediate contact internally with this
again is a muscular layer of approximately equal thickness in
which the fibres run longitudinally, the outer longitudinal
layer (e./.). Between this and the inner layers of muscle is,
over all parts of the body, a broad stratum—the zone of ducts
—closely packed with the dilated portions of the ducts (dct.)
of the deep integumentary glands. These are so closely placed,

634 WILLIAM A. HASWELL.

and are so strongly coloured by the hematoxylin, that scarcely
any intermediate substance is distinguishable. It is in this
zone that the wide portions of the superficial unicellular glands
lie embedded. Here and there occurs a cell of stellate shape.
Below this zone is the inner circular layer of muscular fibres,
which is much thicker than the outer; and below this again
is the inner longitudinal layer, which is of great thickness,
and extends uniformly over both surfaces.

The deep integumentary glands lie below the muscular
layers, scattered through the parenchyma in all parts of the
body. Their ducts, which frequently branch and _ perhaps
anastomose, pass from them with a sinuous course, perforate
the muscular layers and the layers of the integument, and
open on the outer surface. Each gland consists of a single
cell of about ‘1 mm. diameter, of evenly rounded outline
without processes, except where the duct is given off. Enclos-
ing the cell is a distinct capsule, in which lies a flattened
nucleus as large as that of the cell itself. The nucleus of the
gland, situated towards the middle, always has a lobed outline,
and contains about six spherical nucleolar bodies of about equal
size. The cytoplasm exhibits a strongly marked reticulum,
the threads of which have a prevailingly radiate arrangement.
The interfibrillar substance appears clear and homogeneous in
the great majority of the glands, but in many the cell is full
of well-formed rhabdites. The ducts appear as tubes with
well-defined walls. When they enter the zone of ducts they
become dilated, contracting again as they pass through the
superficial layers. The ducts of the glands which contain
rhabdites also contain rhabdites, while in the interior of the
ducts of those glands in which no rhabdites occur is a reti-
culum similar to that of the cell itself. Special strands of
ducts of the deep integumentary glands run forwards to open
about the anterior margins of the body, and very many open
on the extreme lateral margin. The parenchyma completely
and closely fills up the interspaces between the organs, and a
celom, such as is described as being characteristic of the
Rhabdoceeles in general, is not in any way represented. The

ON A PRORHYNOCHID TURBELLARIAN. 635

parenchyma is apparently syncytial, with a coarse reticulum
continuous from cell to cell, and large nuclei. Strands of
muscular fibres (parenchyma muscle) pass through it, mainly
in a dorso-ventral direction.

No account is given by v. Graff! or by Vejdovsky? of the
integument in Prorhynchus, so that I have no means of
comparing it with that of the form under consideration. The
superficial integumentary glands apparently correspond to the
cells described by Vejdovsky (l.c., p. 144) as occurring on the
ventral surface of P. hygrophilus, and figured in fig. 76.

Digestive System.—The pharynx (fig. 1, pA.) is of great
relative size (about one third of the entire body in length) and
high degree of complexity. It is capable, as shown in one of
the specimens, of protrusion to some extent through the
aperture of the mouth, when the free margin appears irre-
gularly lobed. Its sheath extends back only a short distance
from its anterior end. Its form is that of a thick-walled
cylinder somewhat contracted posteriorly. The layers of
muscle in its wall (fig. 4) have the following arrangement,
taken from without inwards :—(1) External longitudinal, (2)
external circular, (8) internal circular, (4) internal longi-
tudinal. Between the external and internal circular layers is
a broad zone occupied only by glands, nerves, and radial and
arched fibres. The radiating fibres run from the inner surface
of the external longitudinal layer to the basement membrane ;
they divide the internal longitudinal layer (which lies imme-
diately beneath the epithelium and basement membrane) into
numerous very regularly arranged bundles. The arched fibres
are all directed in the transverse plane, so that they only
_ appear in transverse sections. They are arranged in bundles
running with a strong curvature from the external circular
layer to the deeper part of the internal circular. The pharyn-
geal unicellular glands lodged in the zone above referred to
send narrow ducts inwards to open into the internal cavity of

1 ‘Monographie der Turbellarien,’ I. “ Rhabdoccelida,” p. 264 (1882).

2 “Zur vergleichenden Anatomie der Turbellarein,” ‘ Zeitschr. f. wiss.
Zool.,’ lx (1891).

636 WILLIAM A. HASWELL.

the pharynx; the openings of the ducts are most abundant
round the anterior aperture. A complex system of nerves
extend through the wall of the pharynx.

Von Graff! refers to the pharynx of Prorhynchus as
belonging to the same type as that of the Plagiostomide,
and in his account of P. stagnalis states that it has the out-
ward form and probably also the internal structure of the
‘pharynx variabilis” of that family. He had, however, not
examined sections. Vejdovsky? states that the order of the
layers in P. hygrophilus is that laid down by v. Graff as
diagnostic of the “‘ pharynx doliiformis;”’ but the description
which he gives and his figure 92 do not bear out this state-
ment. The order of layers in the “ pharynx doliiformis,” as
given by v. Graff,? is, from without inwards, longitudinal,
circular, longitudinal, circular; whereas the order which
Vejdovsky describes and figures is longitudinal, circular, cir-
cular, longitudinal, as in the form now under consideration,
though all the layers are very feebly developed, being each
only about one fibre in thickness. This order of the layers
occurs in none of the described types—that in the “ pharynx
variabilis” being circular, longitudinal, circular, longitudinal,
—and would appear to be characteristic of the family Pro-
rhynchide.

The intestine extends to very near the posterior extremity
of the body. Anteriorly it gives off on the ventral side a short
and wide diverticulum, which passes forwards for a short dis-
tance below the posterior extremity of the pharynx. Laterally
it is divided throughout its entire extent by constrictions
brought about by ingrowths of the investing fibrous layer of
the intestinal wall. Of these there are about forty on each
side. On the right side they correspond very closely with the
windings of the ovary. The epithelium of the intestine consists
of extremely long narrow cells of very irregular shape, many
of which are filled with the large granules so characteristic of

1 Loe. cit., pp. 87 and 265.
2 Loc. cit., p. 148.
3 Loe. cit., p. 84.

ON A PRORHYNCHID TURBELLARIAN. 637

the intestinal cells of Turbellaria in general. The epithelium
is supported by a thin layer of a fibrous character, apparently
muscular, containing numerous spindle-shaped nuclei.

Nervous System.—The brain (fig. 1) consists of two large
ganglia situated near the anterior extremity of the body above
the mouth, and connected together by a thick commissure ;
the nerve-cells are mainly confined to the ganglia, to which
they form a thick investment. Given off from the brain on
each side is a large nerve passing to the corresponding ciliated
sac. Two main longitudinal trunks (figs. 1 and 2) are given
off on each side posteriorly. One of these, which is much the
smaller, runs in the lateral flange of the body.! It gives off
regularly arranged transverse branches, of which those on the
inner side join corresponding branches of the larger trunk,
thus giving rise to a number of commissures connecting the
two trunks together, while those on the outer side given off
opposite the commissures run outwards towards the lateral
margin, giving off branches to the integument. In addition
to the commissures which connect it with the smaller trunk,
the larger nerve-cord gives off on its inner side many trans-
verse branches running in the ventral wall of the body;
whether these form complete commissures was not ascertained.
The ciliated sacs are two deep excavations, each situated just
behind the corresponding anterior angle of the body, and
opening on the ventral surface near the lateral border by a
wide orifice. The cavity is lined with a layer of regularly
arranged, large, columnar cells, beset at their inner extremity
with long cilia. Internally and posteriorly the ciliated sac is
prolonged into a narrow cylindrical tube, which, after
receiving the ducts of a number of unicellular glands
similar to those that discharge on the outer surface, ends
blindly. In close contact with the columnar cells of the
sac, and probably in continuity with them, is a layer of
nerve-cells, processes from which go to form the nerve already
referred to.

1 The lateral cords have been made too thick in fig. 1, and made to run
too far back.

638 WILLIAM A. HASWELL.

Excretory System.—There are two excretory apertures
(fig. 1, ew.) situated on the ventral surface, just behind the
ventral diverticulum of the intestine, and immediately external
to the main nerve-cord. Each leads into a tubular sac with
thick walls, and lined with a thick layer in which nuclei are
not visible, though doubtless it is of the nature of an ecto-
dermal epithelium. Spaces similar to those in the epidermis
occur at intervals ; and in the interior of the sac is a quantity
of fibrillated substance, which may partly be the remnants of
cilia, though of this there is no positive evidence. From this
excretory sac runs dorsad a sinuous canal, which bifurcates to
form an anterior and a posterior longitudinal trunk. The
former is the smaller. It runs forwards, external and dorsal
to the larger nerve-cord, to the anterior extremity of the body,
where in the neighbourhood of the ciliated sac it becomes
somewhat dilated and much convoluted. Its continuation, or
a branch, passes transversely in front of the brain, but is not
traceable as far as the middle line. The posterior longitudinal
trunk is very wide, and, like the anterior, twists about in a
sinuous manner. Some distance back it bifurcates, the two
branches running back side by side for some distance. Both
anterior and posterior trunks give off numerous branches.
At some points they seem to unite for a short distance and
diverge again.

In the main the disposition of the parts of the excretory
system resembles that of the corresponding parts in Pro-
rhynchus, as described by Schultze, v. Graff, and others ; but
there are some important points of difference. The position
of the external openings is the same in both, and in both the
short vessel into which the aperture leads bifurcates to form
anterior and posterior trunks; but in Prorhynchus the
posterior trunk does not bifurcate, and an internal longitu-
dinal vessel running through the entire length of the body is
given off from a transverse commissural vessel situated far
forwards.

Reproductive Apparatus.—The organs of the two sexes
are both mature in three out of the fourspecimens. The male

ON A PRORHYNCHID TURBELLARIAN. 639

aperture is a small slit situated in a recess below the pharynx,
this recess opening on the exterior through the mouth. The
penis (figs. 1, 5, and 6) is a long and slender, pointed, chitinous
spine. Contained within the lumen of the penis is a finely
fibrillated material, which stains faintly with eosin. Investing
it is a layer of fine fibres, most of which take a spiral course.
This layer is continuous at the base of the penis with the wall
of the ejaculatory duct, while at the apex it is continuous with
the penis sheath ; between it and the penis on the one hand,
and the penis sheath on the other, are several large nuclei.
The sheath of the penis consists of an outer layer of circular
fibres, and an inner of longitudinal; the latter is a continuous
layer in the greater part of its extent, but divided towards the
apex of the penis into a number of distinct bundles. Sup-
porting the sheath are ten slender chitinous rods, which, where
the sheath is reflected in front to become continuous with the
investing layer of the penis, bend sharply backwards for some
distance.

The ejaculatory duct presents a slight bulbous dilatation at
the base of the penis. It is a thick-walled tube with a
muscular wall, composed mostly of circular fibres. The penis
sheath is continued over it for a short distance, but stops short
completely before reaching the vesicula seminalis. The vesi-
cula seminalis (fig. 1, v. s.) is a very large sac of long oval
form. Its walls are thick and muscular, the fibres taking for
the most part an oblique direction round the wall. In the
interior is a thick layer of a finely fibrillated substance bound-
ing a relatively narrow central lumen. At the junction of the
ejaculatory duct and vesicula open the ducts of a number of
_ unicellular glands, which probably are the prostate or granule
glands. From the vesicula runs a very greatly coiled narrow
duct, the vas deferens. This becomes somewhat thicker
posteriorly, and eventually is continued into a wide thin-
walled spirally twisted sac, the sperm reservoir (sp. 7.)—com-
pletely filled, in all three mature specimens, with ripe sperms.
From this a short, straight, narrow efferent duct leads
obliquely towards the left, where it enters the testis, and is

640 WILLIAM A. HASWELL.

continued as an exceedingly fine tube connecting together the
lobes of that organ.

The above account of the efferent part of the male apparatus
differs in certain important respects from the accounts that
have been published of the structure of those parts in Pro-
rhynchus. The exceptional position of the male aperture is
shared with that genus and certain of the Alloioccela. The
form of the penis corresponds, to some extent, with that of
Prorhynchus stagnalis as described and figured by v.
Graff! The chitinous lamelle of the penis sheath which he
describes obviously correspond to the chitinous rods referred
to above. But if Hallez’s statement be correct, that in Pro-
rhynchus stagnalis the sheath is continued back as the
wall of the true ejaculatory duct, which encloses an inner
duct continuous with the wall of the penis itself and passing
back to the “secretion reservoir,” then there is at least one
important point of difference.

The testis (figs. 1 and 2, Ze.) is a long narrow body extending
from a little behind the posterior end of the pharynx ou the
left side of the intestine between it and the main trunk of the
excretory system to near the posterior end of the body. It is
divided into a large number—over a hundred—of small lobes
connected together by the efferent duct and its branches.
Investing the lobes is a thin layer of connective-tissue fibres.

The female (fig. 1,9) aperture is situated in or near the
middle line on the ventral surface. It leads through a short
passage lined with ciliated cells into a rounded chamber
(vagina) lined with an epithelium, the cells of which are large
and similar in general shape to those of the epithelium of the
intestine, but much shorter. From this there runs forwards a
wide pyriform chamber, which may be provisionally termed
uterus (fig. 1, uf.). It has a wall composed of a fibrous layer
and a layer of large cells similar to those lining the vagina.
From the first-mentioned chamber the oviduct, a wide sinuous
tube with a lining of somewhat flattened granular cells, passes
backwards and to the right to become continuous with the

1 Loc. cit., p. 265, Taf. xv, fig. 19.

ON A PRORHYNCHID TURBELLARIAN. 641

vitello-ovary. The latter (fig. 1, ov.) is long and narrow,
thrown into numerous sinuosities, and occupies a position on
the right side of the intestine corresponding exactly to that
occupied on the left by the testis, though it does not extend so
far back as the latter, stopping short some distance in front of
the posterior end of the intestine. It is enclosed in a fibrous
layer similar to and continuous with that enclosing the intes-
tine, and is lined internally with a layer of flattened granular
cells. Its interior is occupied, for the most part, by vitelline
cells or follicle cells. These are large cells arranged, for the
most part, in an epithelium-like manner, but sometimes col-
lected into more irregular clumps. Each contains a large
nucleus, and each has a distinct enclosing membrane. Through-
out the greater part of the length of the vitello-ovary the
follicle cells are loaded with large rounded yolk-granules, but
in the posterior portion these granules are entirely absent, and
the cells are much smaller.

Here and there is an ovum (fig. 7). Hach of these is a large
rounded cell with a large nucleus and very fine-grained proto-
plasm, enclosed in a follicle of regularly disposed follicle cells.
The ova are smaller and more numerous in the posterior part
of the ovary than they are in front, and here the follicle cells
that enclose each are numerous, and have the appearance of a
columnar epithelium. Further forward, where the ova are
large, the follicle cells that surround them are much smaller
than the rest. Towards the posterior end of the vitello-ovary
there is, in two of the three mature specimens, a great mass
of sperms distending the cavity, the wall of which is here very
thin; there is thus formed from a portion of the ovary a dis-
tinct, though perhaps temporary, bursa seminalis. Further
back again the vitello-ovary resumes its normal character.
In the sexually immature specimen the penis is completely
developed, but the various parts of the male duct have not
yet become formed. The testis is represented by a narrow
duct connecting together a chain of small cavities. In these
and in the lumen of the connecting duct are a number of cells
of a peculiar character, having the appearance of malformed

642 WILLIAM A. HASWELL.

sperms of various stages. From this runs forwards an irre-
gular median channel without well-defined walls; this contains
bodies similar to those in the testis itself, and is traceable as
far forwards as the base of the penis. The female aperture is
present ; it leads into a cavity from which runs backwards on
the right side a wide irregular channel representing the ovary,
but containing no cells that are recognisable. as follicle cells
or ova.

The remarkable disposition of the reproductive organs de-
scribed above is one which is not paralleled, so far as I can
ascertain, in any other Turbellarian, and is perhaps sufficient
in itself, apart from the other points of difference, to render
necessary the separation of the present form from Pro-
rhynchus as at least generically distinct.

In Prorhynchus stagnalis, according to v. Graff,! the
testis is not sufficiently known; the female opening is in the
middle of the ventral surface; the ovary is an elongated body
in the posterior blind portion of which are numerous germs
of ova, while further forwards are the ripe ova surrounded by
yolk-cells. In P. sphyrocephalus the same author states
that a similar condition of things obtains.

None of the specimens of Prorhynchus hygrophilus
obtained by Vejdovsky had the testes developed. He assumes
that they were in a degenerate condition, and that Pro-
rhynchus is a proterandrous hermaphrodite. He adds,
however, that from the statements of v. Kennel and Braun it
may be looked upon as determined that the testes of Pro-
rhynchus are developed as small rounded follicles on both
sides of the intestine. Of these the latter author has described
two or three pairs, while v. Kennel characterises them as
vesicles sometimes in close contact with one another, some-
times separate, at first in a single series, later arranged in
several irregular rows extending not quite as far as the pos-
terior end of the body.

Vejdovsky had found only three such follicles, and these not
in a state of functional activity, in the form of longish round

1 Loc. cit., p. 266.

ON A PRORHYNOCHID TURBELLARIAN. 643

vesicles situated close to the sides of the intestine. The
vitello-ovary consists in its posterior portion of indifferent
reproductive cells; further forwards it contains a row of ova
surrounded by yolk-cells.

Impregnation.—Embedded in the parenchyma, close to the
ovary, uot far from its posterior end in the specimen, without a
bursa seminalis, is a chitinous tube, which is without doubt the
remains of the chitinous part of the penis of another individual.
This circumstance affords very strong evidence in favour of
the conclusion that the sperms of one individual are conveyed
into the interior of the ovary of another, to form the seminal
bursa by the penis piercing the body-wall and penetrating to
the ovary. The structure of the penis and vesicula seminalis
would of itself suggest such a mode of copulation; the whole
apparatus has exactly the form of a hypodermic syringe, with
a compressible ball instead of a cylinder and piston.

A mode of copulation in which a wound is inflicted in any
part of the body by the penis, and masses of sperms are dis-
charged into the wound, was long ago described by Lang? as
occurring in many Polycladida. But, so far as I am aware,
this is the first recorded instance of the direct injection of the
ovary with sperms through a perforation of the body-wall by
means of the penis.

In the incomplete condition of our knowledge of the genus
Prorhynchus it is somewhat difficult to give a diagnosis
which will clearly define the new form. Von Graff’s definition
of Prorhynchus is as follows :—‘ Prorhynchida mit Wim-
pergribchen, Mund am Vorderende des Korpers, ein chitinoses
copulationsorgan vorhanden, Korper fadenformig gestreckt.’’?
In all of these points, with the exception of the last, which is
not important, the new form agrees with Prorhynchus. But
it differs from it apparently in several material points not
included in v. Graff’s diagnosis. Such are the more complex

1 ‘Der Bau von Gunda segmentata, &c.,” ‘ Mitth. ans der Zool. Stat.
zu Neapel,’ Bd. iii, p. 222 (1882); ‘Die Polycladen,’ pp. 231 and 636
(1884).

2 Loc. cit., p. 263.

64.4 WILLIAM A. HASWELL.

pharynx, the bifurcate posterior vessel of the excretory system,
the rod-like chitinous supports of the penis sheath, the
laterally placed vitello-ovary, and the unpaired laterally
placed testis. In view, however, of the necessity for a more
thorough knowledge of the structure of tlie described species
of Prorhynchus in these and other respects, I think it best
to regard the New Zealand species provisionally as a member
of the genus Prorhynchus, and propose for it the name
P. putealis.

EXPLANATION OF PLATE 48,

Illustrating Mr. William A. Haswell’s paper “On a Pro-
rhynchid Turbellarian from Deep Wells in New Zealand.”

Fic. 1.—Semi-diagrammatic view of the organisation of Prorhynchus
putealis as seen from the ventral aspect. x 10. a./.v. Anterior longi-
tudinal trunk of excretory system. c@. Diverticulum at the anterior end of
the intestine. com. Commissure between the brain ganglia. c.s. Ciliated
sac. e. Hjaculatory duct. ex., ec. Excretory apertures. iné¢. Intestine.
1.a. Large longitudinal nerve-cord. /.2?. Smaller longitudinal nerve-cord.
od. Oviduct. ov. Ovary. ph. Pharynx. s. 6. Bursaseminalis. sp. d. Ante-
rior portion of sperm-duct. sp.d'. Posterior portion of sperm-duct. (e.
Testis. ut. Uterus. v.s. Vesicula seminalis. 9. Female aperture.

Fic. 2.—Semi-diagrammatic representation of a transverse section of the
body in the middle of the intestinal region, to show the relations of the
various organs. X 30. ex. HExcretory vessels. iz. Intestine. /.~'. Large
longitudinal nerve-cord. J/. 2. Smaller longitudinal nerve-cord. ov. Ovary.
te. Testis.

Fie. 3.—Section of the integument. 6. Basement membrane. c. Cuticle.
ep. Epidermis. e.c.m. External circular layer of muscle. e./.m. External
longitudinal muscular layer. dct. Ducts of deep integumentary glands. g/.
Superficial integumentary gland.

Fie. 4.—Portion of a transverse section of the pharynx. e.c. External
circular layer. ./. External longitudinal layer. ep. Epithelium. gi.
Glandular zone. i.c. Internal circular layer. ¢./. Internal longitudinal
layer. r. Radial fibres.

ON A PRORHYNCHID TURBELLARIAN. 645

Fic. 5.—Ventral view of the penis and its sheath reconstructed from
horizontal sections. c.m. Circular layer of muscle of penis sheath. ej.
Hjaculatory duct. ¢.m. Longitudinal layer of penis sheath. p. Chitinous
tube of penis. p'. Investing layer of penis. pr. Protractor muscles. r.
Chitinous rods. z.#. Line of section of fig. 6.

Fic. 6.—A transverse section of the penis and its sheath at about the line
marked z.x. in Fig. 5. Lettering as in Fig. 5. sh. Portion of the sheath
reflected on the penis at the anterior end.

Fic. 7.—Section of vitello-ovary behind the vitelline region. fol. Follicle
cells. iz. Investing fibrous layer. ov. Ovum.

VoL. 40, PART 4.—NEW SER. ip

«

ve

7
= _ 7 i _
a . ¥
H od p Ree ass
ee a a he
a ; a. a oo
ie i »
( Fe , ;
“—» a es 7 : vt
: “iy i 7
f
F 4
bd -

\

.
a J ee r
° .

1

if s,
bie’ qh} cares

<, ene eee

me ; { 7

eg rit Nee Fy + 7 Slee: Uy
F 9 ‘?) ai 6

A .

rt bak A te

= t an 'y

Lr

ste
[ihe tee as ey ~ nie:

pe

JPEG: f * Sats ae ut ape htern t ge oe

.
aad

c.4

i

Th,

a aT ha '
? elt , %) rt te Rae

ty “Hy

as ;

ay im dat

sna wa

4) <r

DEVELOPMENT OF THE ATRIAL CHAMBER IN AMPHIOXUS. 647

Note on the Development of the Atrial Chamber
in Amphioxus.

By
E. Ray Lankester, M.A., LL.D., F.R.S.

Owine to the promotion of my friend Professor MacBride
to a post in the University of Montreal in Canada, it has not
been possible for me to confer with him personally as to the
significance of certain passages in his interesting and valuable
memoir on “The Early Development of Amphioxus” published
in the present number of the Journal. I must therefore
take the passages in question as they stand, and deal with
them in print.

Professor MacBride has made the interesting discovery that
the lymph-canals in the metapleura of Amphioxus are pro-
longations of the pair of coelomic pouches which immediately
follow the head cavity (bifid in its later growth). He identifies
them with the collar cavities of Balanoglossus, and sees in
this relation a confirmation of the view, advanced originally
by Bateson, that the atrial chamber may be considered as
morphologically identical with the small region overhung by
the free posterior margin of the collar of Balanoglossus, and
the wall enclosing the atrial chamber with the free projecting
ring of the Balanoglossus collar. Whilst I by no means deny
that there is a relationship between the metapleura of Amphi-
oxus and the collar of Balanoglossus, it appears to me that
MacBride has been led by theoretical bias into a serious
misapprehension of the significance of the observations on the
development of the atrial chamber in Amphioxus published
by me in conjunction with Dr. Arthur Willey in 1890.

648 EB. RAY LANKESTER.

In referring to those researches Professor MacBride says
(p. 591), “ Lankester and Willey’s paper on the development
‘of the atrial chamber confirms in most points Kowalevsky’s
statements.” Later (p. 605) he says, ‘“ Kowalevsky was the
first to discover that the atrial cavity was formed by the
meeting in the mid-ventral line of two long ridges or folds.”
And further, “ Lankester terms the folds which actually wall
in the atrial cavity ‘epipleural,’ and the projecting angles
after these folds have united ‘ metapleural.’ I shall use the
term ‘atrial fold’ to include the whole, of which both are
parts.”

It appears to me that both in his references to Kowalevsky’s
observations and in his iteration of the word “folds ” and pro-
posal to call what I had called epipleur and metapleur by the
term “atrial fold,” Professor MacBride is in error, and that
his statements and use of terms are inconsistent with the
facts demonstrated by Willey and myself.

The region of the body of Amphioxus termed “ epipleur ”
by me (at a time when I accepted Rolph’s theoretical scheme of
its development based on Kowalevsky’s observations, and now
shown to be erroneous) includes the whole of the atrial wall
on each side from the level of the dorsal artery to the median
ventral raphe. It seems to me absolutely unjustifiable to
speak of this as a “ fold” since my paper of 1890, the more so
inasmuch as the erroneous view of its origin (that of Rolph)
was that it arose as a horizontal down-growing fold. To call
it and the metapleur resting on it “the atrial fold” at the
present day, as Professor MacBride does, is simply to ignore
Willey’s and my results, and to perpetuate error.

It seems that Professor MacBride has somehow forgotten
what Willey and I actually showed, since he declares that we
confirmed Kowalevsky, and further that Kowalevsky dis-
covered that the atrial cavity was formed by the meeting in
the mid-ventral line of two long folds.

Asa matter of fact, Willey and I did not confirm Kowa-
levsky. The whole point of our paper lay in a correction of a
misinterpretation made by that most distinguished observer,

DEVELOPMENT OF THE ATRIAL CHAMBER IN AMPHIOXUS. 649

So far from confirming the existence of “folds,” or admitting
anything which Professor MacBride can justifiably term
‘atrial folds,’ we showed that there are no atrial
folds. We showed that what Kowalevsky had mistaken for
“atrial folds” are really the metapleura, and that these do
not grow round and meet in the middle line, but that a very
small in-sinking is formed between them, and is covered in
by a minute horizontal growth right and left which we called
the “subatrial ridges or folds,” their union resulting in the
formation of what is, at first, a very narrow “subatrial floor”
lying between the two upstanding metapleura. We showed,
then, that “the two long folds ’ of Kowalevsky, quoted by
MacBride as though it were established that they are the
rudiments of the epipleura (as in Rolph’s scheme) do not form
the atrial cavity, but are really the metapleura. MacBride says
we confirmed Kowalevsky, and that Kowalevsky “discovered,”
the mode of formation of the atrial cavity. He did not do so,
but, on the contrary, was misled as to the nature of what he
thought to be coalescing folds. And Willey and I did not
confirm him, but discovered a totally different mode of forma-
tion.

As to the continuity of the cavity of the metapleura with the
celom, Willey and I showed that the space in the metapleura
is not at any time freely open above into the adjacent ccelomic
cavity, as was figured by Kowalevsky, who mistook the meta-
pleural cavity for the lateral division of the ccelom, in which
the gonads develop. We reproduced Kowalevsky’s figure, and
showed that it was defective. MacBride confirms us in deny-
ing a continuity of the lymph-space of the metapleur with the
adjacent celom. We suggested, with reserve and caution,
that the lymph-space of the metapleur might form as “ pseudo-
ccel,”’—that is, as an intercellular space,—having no relation
to enteric pouches. MacBride has now shown that the most
anterior pair of enteric pouches (the collar-pouches) are the
sources of the lymph-spaces in the metapleura. This is cer-
tainly altogether a different relation to that indicated by
Kowalevsky’s transverse section (fig. 1, p. 449, of Willey’s

650 E. RAY LANKESTER.

and my paper, ‘ Quart. Journ. Micr. Sci.,? 1890, vol. xxxi)
and in no way justifies MacBride’s implication that Kowalevsky
was right in considering the metapleural canals as coelomic in
nature, and that Willey and I were wrong in denying their
connection with celom.

I will conclude this note by a quotation from the paper by
Willey and myself above cited (p. 455),in which the difference
between the actual state of things made known by us and the
theory of “ atrial folds” so strangely resuscitated by MacBride,
is set forth.

“It is important to point out that the mode of formation of
the atrium as a narrow groove, which closes and sinks (as it
were) into the body of the Amphioxus, is really different in
important respects from the enclosure of a space by down-
growth of large folds, though ultimately, no doubt, the two
contrasted modes of formation come to the same thing, so far
as the more obvious morphological relations are concerned.
The mode of formation which really occurs in Amphioxus is
readily harmonised with the existence of the post-atrioporal
extension of the atrium, which gradually tapers to a fine
ceecal canal. It also gives us an essentially different view of
the region called “ epipleur”? by Lankester, and generally so
designated, from that which Rolph’s theory necessitated. That
portion of the epipleur into which the myotomes of the body-
wall extend is seen now to be no downgrowth, no extension or
fold. It is the original unchanged body-wall, which bounds
the sides of the animal’s body in front of the atripore, just as
much as it does behind. The only new growth in the atrial
region which takes part in the limitation of the surface is the
subatrial growth formed by the two little horizontal folds
which floor in the atrium when it is a mere canal. These in
the adult are represented by the region of longitudinally
pleated ventral wall between the two metapleura.”

meee x LTO VOL. 40;
NEW SERIES.

Actinotrocha, structure of, by Mas-
terman, 281

Albunea, new species of, by Garstang,
211

Amphioxus, note on the development
of its atrial chamber, by Lankester,
647

= the early development of,
by MacBride, 589

Arthropod head and Annelid pro-
stomium, by Goodrich, 247

Ascons, materials for a monograph of,

by E. A. Minchin, 469

Bdellostoma Stouti, development
of, by Bashford Dean, 269

Calappa, observations on, by Gar-
stang, 211

Cephalodiscus, structure of, by Mas-
terman, 281

Chetopterus, green pigment of its
intestinal wall, by Lankester, 447

Clathrinide, origin and growth of
their spicules, by Minchin, 469

Colleutt on the structure of Hy-
dractinia, 77

Crustacea, decapod, respiratory me-
chanism of, by Garstang, 211

Cunningham, J. T., on the ovarian
ova of marine fishes, 101

Dean, Bashford, on the development
of the Californian hag-fish, Bdel -
lostoma Stouti, 269

Diplochorda, the, by Masterman, 281

Drepanidotenia hemignathi, a
new Tapeworm, by Shipley, 613

Embiide, on the, by Grassi and San-
dias, 1

Enteropneusta, a new genus of, by
Willey, 623

Fishes, ovarian ova of, by Cunning-
ham, 101

Garstang on the respiratory mechan-
ism of Decapod Crustacea, &c.,
211

Gephyrean, new British, by Herd-
man, 367

Goodrich on the anatomy of Sternas-
pis, 233

#5 on the nephridia of the
Polycheta, 185

bs on the relation of the
Arthropod head to the Annelid
prostomium, 247

Grassi and Sandias on the Termites,
and on the Embiide, 1

Hamingias remarks on, by Herdman,

67

Haswell on a Prorhynchid Turbel-
larian from deep wells in New
Zealand, 631

Herdman on anew British Echinuroid
Gephyrean, 367

Hesione, nephridia of, by Goodrich,
185

Heteroplana, a new genus of Pla-
narians, by Willey, 203

Hydractinia, structure of, by Mar-
garet C. Colleutt, 77

Lankester on the development of the
atrial chamber of Amphioxus, 647
5 on the green pigment of
the intestinal wall of the Annelid
Chetopterus, 447

652

MacBride on the early development
of Amphioxus, 589

Marsupialia, embryology of, by J. P.
Hill, 385

Masterman on the Diplochorda,
Actinotrocha, and Cephalodiscus,
281

Minchin on the origin and growth of
the spicules in the Clathrinide
(Ascons, Calcispongiz), 469

Nautilus, on the adhesive tentacles,
pericardium, and spermatophores
of, by Arthur Willey, 207

mis tentacles and osphradia of,
by Willey, 197

Nephridia of the Polycheeta, by Good-
rich, 185

Nephthys, nephridia of, by Goodrich,
185

Ovary and ovarian ova of marine
fishes, by J. T. Cunningham, 101

Perameles, placentation of, by J. P.
Hill, 385

Pigment, green, of the intestinal wall
of Chetopterus, by Lankester, 447

Placentation of Perameles, by J. P.
Hill, 385

Planarians, anew genus of, by Willey,
203

Prorhynchid Turbellarian from deep
wells in New Zealand, by Haswell,
631

INDEX.

Protozoa, parasitic, in Termites, by.
Grassi and Sandias, 1
Ptychodera flava,

by Arthur
Willey, 165

Sandias, see Grassi

Shipley on Drepanidotenia he-
mignathi, a new Tapeworm, 613

Spengelia, a new genus of Entero-
pneusta, by Willey, 623

Sternaspis, the anatomy of, by Good-
rich, 233

Termites, Grassi and Sandias on
the, 1

Thalassema, remarks on, by Herd-
man, 367

Turbellarian, a Prorhynchid, from deep
He in New Zealand, by Haswell,

3

Tyrrhena, nephridia of, by Goodrich,

185

Willey on Heteroplana, a new genus
of Planarians, 203
» on Ptychodera flava, 165
», On Spengelia, a new genus of
Enteropneusta, 623
», on the adhesive
pericardium, and
of Nautilus, 207
», _ on the tentacles and osphradia
of Nautilus, 197

tentacles,
spermatophores

PRINTED BY ADLARD AND SON,
BARTHOLOMEW CLOSE, E.C., AND 20 HANOVER SQUARE, W.

r)
ve iy a,
y \ ¥) i
} oe eu '
P wil { ‘
fe : : aK fj
\ f Hi
) hd Le t by ENA
f i ‘ae rh) i 4
f i
\ ;
j °)
J !
; i
{
. i
iA

SNR Rea ON EROS at
til Met iL Aks yee}

NN

Brie ayy

. i te \ . y Aly ‘ ue ;
Moen Oi. iam aig Le Hite PCL ae Aa eo ARH bd "ot
(ULB STS” A Pn Hit
1 ae oe Hf PP any a 4) ; AA AN até uy
ON Ga aaa alts ay Mitre hi .
| OR a TN :
mn : CA ¥ ( ( wi if! ix Wire Ci y var ne f hy at, hie
10" 4 i) f ve Hl
1 : ear f dd od eek ad ee
Vea a POLO « 6 pe, rs Ne sisi | Aaalheg a eee Ao) A y 4% b}
[ LO aC Ape | UAT is Pi int i CA On,
a i? ayn! CT ia on ; ete kos PT
iv ORT Ne DADA i Patino e I Viera Fes) he
i . Ma f an DFAT, © WRN Tea waite! iF ’
hy ar A ative oy Ah -) | |
if SRL A RS ee Vio 8) Ol ia
h : : q By aah) ae? jet ee j i
; Miny i te Lay viet aed: ih ; te \AGay Ve ara “¥ ; ; m f
ys \ Nee MOR) aE dol ae : mi
; y Baad Veh ou en 5 et ee LA Oe ee } 4
4 bak ey y wie vs We! PY Pt i
ti ; ‘ ~; i
ha \ ‘
rr 7 A PAH ye Tl) taka ihe) AC i ” Lig sit ; ae
y vik Bt ‘ ‘ Thuy ies Paty ; Fey 297) in! a + iy aay ;
‘a ’ ’ iP
uy Wi : 7 -
AS vi Wen sAt sR, wa Fat ei ay P a al
iy } (ian ey Wie iat } bi te halla Shinde
Lay i fi ri) a
: e ve " ay | AAO UE. t eh 4 f TMA Mad A, ye
i oN na ar Saami 3 Ae ie vir jinn
By aan, a" lana? Wal es nl
i ij 5 J , ' rt . j
i : wy y at : Wy hy a9 My ay
Ay, ‘ wry Oh hy en ed oy ae ‘tha ADA ay
' DA ia | y) Oh) sien oy ae iinet
F aierun yy lad Oy oa ar Bay Ww Rahi a dy ae (
\ ee UT, (iM tiled hohe Ben hy, Y ha witht A, Celcrnnnd
, aM Nia Naolh” Ie
ij ba
j ’
- ) ;
iy | ‘

®
ie i > U
; ri a ay i
4 : ‘ t A ee s if ‘ef
i aa Ay The
i | HD 1 ie AN Lg ‘ D 1 f
re ; tae) hi We hs
ny he :
by , iy a vi iy iy
i a Tre oe a a aria: Beaks ky abla a ‘f ‘ ih 7
(Qe iyi ae he ane Nae Wah Ste Puaawy, yl
f A ; hh ? Ti 4
} ii a 4
j : ‘
6 Fi ~ a : ‘
AA), j ny uty SE
+ A) CL ee
a a Lt bs 4 ( PS 7 WT
TCAD RI ST RRR
ey afin NY ded xy a!
PO ea ee LO
Ay WT man ro Add yelp! : ce 7 We

TL