A CONTRIBUTION TO THE

EMBRYOLOGY AND PHYLOGENY

OF THE

PYCNOGONIDS

A DISSERTATION

PRESENTED \ > THE BOARD OF UNIVERSITY STUDIES FOR THE:
DEGREE OF DOCTOR OF PHILOSOPHY.

T. H. MORGAN,

Bruce Fellow, Johns Hopkim University.

Reprinted from Studies from the Biological Laboratory. Vol. V. Mo. 1.

BALTIMORE:

PRESS OF ISAAC FRIEDENWALD ).

1891.

A CONTRIBUTION TO THE EMBRYOLOGY AND
PHYLOGENY OF THE PYCNOGONIDS. By

T. H. MORGAN, ^ellow of Johns Hopkins University. With
Plates I, II, III, IV, V, VI, VII and VIII.

TABLE OF CONTENTS.

Part I.

Introduction 1

Embryology 3

Tanystylum and Phoxichilidium 4

Pallene empusa 8

External changes 8

Internal changes 15

Comparisons 22

Phylogeny 25

Criticism of Dohrn and Hoek 25

Conclusions from Embryological Data 28

The " Pantopod-larva " 33

Conclusions 34

Part II.
The Metamorphosis of Tanystylum 36

Part III.

The Structure and Development of the Eyes of Pycnogonids 49

The Anatomy of the Adult Eye 49

Histology of the Eye 56

The Development of the Eyes of Tanystylum 59

Comparison with the Arachnid Simple Eyes 63

Inversion of the Eye 65

General Consideration of the Simple Eye 69

Reference to Literature 71

Description of Plates 72

PART I.

In the year 1767 Karl Linne, in the twelfth edition of his
Systema Naturae, described under the name of Phalangium a
Pycnogonid, and here for the first time is the question raised
whether the group is to be ranged under the Arachnids or the
Crustacea. A hundred years elapsed and the problem remained

2 T. H. MORGAN.

unsolved ; the group was placed now here, now there, now
amongst the Crustacea and next amongst the Arachnids. Then
Prof. Dohrn attempted the solution from the standpoint of
Embryology, instituting a comparison " or even identity " be-
tween the Pycnogonid larva and the Nauplius, believing the
Pycnogonids to have diverged from the Crustacea at this point.
During the following t\venty:five years opinions once more vacil-
lated between Arachnidan and Crustacean affinities. Recently
Prof. Dohrn and Dr. Hoek have each independently mono-
graphed the group, placing the morphology of the order on a
very firm basis. Independently likewise they each reached the
conclusion that the group is to be placed neither with the Arach-
nids nor with the Crustacea, but that these three groups have
come down in parallel lines.

The early stages of the embryology of the Sea-Spiders have
been practically untouched, and before any final decision as to
the affinities of the group is to be made, these stages in the devel-
opment should be known and take equal rank with Comparative
Anatomy in disentangling the relationships of the group.

For many reasons the present paper attempts in no way to
give a complete answer from the embryological side. The very
great difficulties of a suitable technique had slowly to be over-
come, and the time at command prevented a detailed description
of the different organs arising from the germ-layers, so that
much remains that might be done.

In the summer of 1889 material for work was collected
at Wood's Holl, Mass. Through the courtesy of Prof. Mac-
Donald I was enabled to collect and study at the station of
the U. S. Fish Commission at this place. To Prof. MacDonald
I am also indebted for many other kindnesses extended during
my stay at Wood's Holl. Three genera of Pycnogonids, each
with a single species, are to be found at this place, viz. Pallene
empusa, Phoxichilidium maxillare, Smith (Anoplodactylus lentus,
Wilson), and Tanystylum orbiculare. During July, August and
September these are found with eggs. Pallene inhabits the
hydroids (Tubularia, Pennaria) on the piles of the wharves, and
is also common on the red sea-weeds below low-tide mark. The
hydroids or sea- weeds as soon as collected were brought into the
laboratory and worked over piece by piece. Each bunch was

THE PYCNOGONIDS. 3

in turn swished rapidly backward and forward in a dish con-
taining a small amount of water, so that the Pycnogonids were
shaken loose and could be easily picked out. The other genera
were more easily found, and on separating the masses of
hydroids, etc., could be readily seen clinging to the stems. The
males of Pallene carry on each pair of ovigerous legs a small
bunch of eggs. Each bunch contains from one or two to fifteen
or twenty eggs. The eggs of Phoxichilidium and Tanystylum
are individually much smaller than the last, but are very
numerous, so that the bunches are much larger, especially so in
the former. Phoxichilidium carries several bunches strung along
on the ovigerous legs of the male: the bunches are white and
very conspicuous against the purple color of the adult. Tany-
stylum has smaller bunches of eggs, with the individual eggs
larger than the former, and the masses are carried so that they
form a circle of clusters held against the ventral side of the male.

The adults with eggs were put into alcoholic picro-sulphuric
acid for several hours and then gradually carried through different
grades of alcohol of increasing strength. Other methods of
hardening gave far less satisfactory results, i. e. boiling water or
Flemming's solution.

To prepare the eggs and embryos for study they were passed
through absolute alcohol (one hour), turpentine (2-4 hours), soft
paraffine (one hour), hard paraffme (1-2 hours). They were cut
in paraffine, and fixed to the slide with albumen fixative ; then
back again through turpentine, absolute alcohol, 95 percent, 80 per
cent, 70 per cent alcohols to Kleinenberg's hematoxylin, where
they remain for a very long time (12-48 hours) ; then washed
fifteen minutes in acid alcohol and up again through the alcohols
to turpentine and into balsam. In Pallene each egg was in
many cases pricked with a very sharp needle before going into
absolute alcohol. It is necessary to do this under a dissecting
microscope. By these methods very excellent results were often
obtained and, after many failures of other methods, was found to
be the only satisfactory one. In Pallene the larger size of the
egg makes a study of the earlier stages much easier, but the other
genera have a much simpler development, and it seems better to
give first an account of these.

To Prof. W. K. Brooks I am greatly indebted for help and
suggestions during the work.

4 T.H.MORGAN.

Tanystylum and Phoxichilidium.

The eggs of Tanystylum measure .08 mm. in diameter (in pre-
served specimens), and those of Phoxichilidium .05 mm. In
both animals there is a regular segmentation. In Plate III,
Figure a is a surface view of an egg of Phoxichilidium divided into
two equal parts ; in Fig. h of an egg into four, but with the seg-
ments shifted around, and Fig. c of an egg into eight equal parts.
Similarly Figs, d and e for the two and four-cell stages of
Tanystylum. The eight-celled stage I did not obtain for this
species. Fig./" is a surface view of an egg of Tanystylum at about
the twenty-four-celled stage, and an optical section of an egg at
this stage shows each cell to run from the periphery to the center
of the egg, where they all come together at a point. Each cell
is thus pyramidal in shape and contains a single nucleus.

In Plate I, Fig. 8, there is a section of an egg at a later stage
than the last. This and the following sections are from eggs
which were cut in paraffine. As the section is a later stage
than the preceding surface view, the pyramids are narrower
than before. Fig. 11 shows a somewhat older stage for
Phoxichilidium. Yery soon after this — perhaps after the next
cleavage — a most important change takes place in the egg. I
have not seen the actual change in the living eggs, but serial
sections leave not the slightest doubt as to the process. Each
nucleus divides radially into two, and this is followed by division
of each pyramid into two parts in a plane at right angles to the
radius of the circle lying in that pyramid. This is shown by
Plate I, Fig. 9, for Tanystylum. Here it is seen that by a
process of multipolar delamination the egg divides into two
germ-layers — an outer peripheral circle of cells and an inner
mass of cells. These latter soon round off and leave no trace of
the pyramidal arrangement. In Fig. 9 one cell is still seen
running from the periphery to the center. This section does not
show a nucleus in the inner part. The inner cells I shall speak
of as the entoblast and the outer circle as the ectoblast. This
figure (9) is somewhat diagrammatic, inasmuch as only part of
the inner cells is shown, for usually they are more closely packed
together than here shown ; and in this case it is due to a part of
the section having broken away (redrawn in figure) and set free
some of the inner cells. In Fig. 10 is a more accurate drawing

THE PYCNOGONIDS. 5

of an egg of Tanystylurn immediately after del animation has
taken place.

Exactly similar changes were seen in the eggs of Phoxichi-
lidium, but the figures just given serve in every respect for both
species. The central as well as the peripheral cells continue to
divide, but soon the entoblast cells lose their well-defined
boundaries and the nuclei seem in part to disappear. The result
is that we have in the center of each egg a granular yolk with
scattered nuclei in it. Here and there is an indication of a cell
boundary. During this time the ectoblast cells have divided
tangentially and have become much smaller, yet at all times a
distinct boundary remains between ectoblast and entoblast. Fig.
12 for Phoxichilidium shows an embryo at this stage.

It is here seen that the ectoblast cells over one hemisphere are
somewhat higher than at the opposite, and I find this very con-
stant in sections of both species at this stage of development.
After this it becomes difficult to follow out the fate of the germ-
layers. The outer cells become smaller and flatter to form the
ectoblast and the inner cells arranged into organs, the most con-
spicuous of which is the digestive tract. There is a triangular
invagination to form the stomodseum, and the proboscis appears
between the first pair of appendages, which has now begun to
form. These appendages are very conspicuous in surface views,
where they project beyond the surface of the body. Between
them appears the slightly projecting proboscis, and about the
middle of the embryo are seen the second and third pairs of
appendages, which are small and inconspicuous. Dohrn has
given excellent figures of embryos at this stage, both in his
earlier paper and in his later monograph. Soon after this the
egg coverings swell up somewhat and the embryos finally break
out of the egg case, so that the appendages now can straighten
out.

In Plate IY, Fig. 9, is a surface view of a larva of Tanystylum
as seen from the ventral surface. Dohrn speaks of this as the
Pantopod-larva, and Hoek calls it a Protonymphon. Its general
characters are shown in the figure. There are three pairs of
appendages : the first pair chelate, with the movable daw working
outwards and downwards ; the second and third pairs have
distally a sharp spine, in the middle of the limb a large segment,

6 T. H. MORGAN.

and proximally a prolongation of the body which is to be
regarded as the basal segment of the appendage. Between the
first pair of appendages the proboscis projects forwards, with
the mouth opening at its distal end. Behind the base of the
proboscis is seen the pair of ventral ganglia. From these the
circumcesophageal commissure ascends upwards around the
oesophagus, and laterally are given off on each side a pair of
nerves to the second and third pairs of appendages. Dorsal to
these ganglia — four in all, the two pairs being fused — is seen the
outline of the mesenteron, which ends blindly behind. About
the middle of the embryo and just back of the pair of ganglia is
seen a pair of oblong masses, which are the beginnings of another
pair of ventral ganglia. In the basal joint of the first pair of
appendages is an opaque mass composed of large cells, from
which runs out a duct towards the base of the spine, and Dohrn
has traced it out to the very tip of the spine, where it opens to
the exterior. This organ is a gland, and its duct is the long
tube in the spine.

A dorsal view of the embryo shows the brain, just above and
a little in front of the ventral ganglia. On its upper surface are
a pair of small pigmented eyes. A transverse section through
the body of this Pantopod-larva is shown in Plate I, 13. It
passes through the center of the body and cuts below the large
ventral ganglionic mass, passing through the base of the third
pair of appendages. Above it cuts the posterior part of the
brain. In the center of the section is the digestive tract. This
is cut at a point where it is about to give off diverticula to the
first and third pairs of appendages. The first of these is
marked D1 and the second Dn.

Below the digestive tract is the second pair of ganglia, which
sends out nerves to the corresponding appendages. Above the
digestive tract the section at B cuts the posterior part of the
brain. The cavity of the body has a few scattered mesoblast
cells in it and a few bits of broken muscle fibers, These fibers
in the living embryo served to connect the mid-gut to the body-
walls. In the base of the legs are seen several cells, which at M
are arranged as though around a central cavity. This cavity does
not seem to connect with the mid-gut, and it seems very probable
that it represents the body-cavity in the legs, and that the sur-

THE PYCNOQONIDS. 7

rounding cells are of mesoblastic origin. In Pallene, as we shall
see later, there are similar cavities, which are undoubtedly meso-
blastic in origin. The mid-gut ends blindly behind, as is seen in
other sections.

In Fig. 14 is shown part of a section through a more anterior
part of the body. The section passes through the brain above
and the first pair of ganglia below. These two are connected by
the circumcesophageal ring. This commissure is composed of
both cells and fibers. In the next section to the one figured a
broad band of nerve fibers passes on each side from the brain to
the ventral ganglia, and in Fig. 14 a few of these fibers may also
be seen. In the middle of the commissure lies the cross-section*
of the oesophagus. Its lumen is triangular and its walls com-
posed of a layer of cells, which are clear around the periphery of
the triangle. Around the oesophagus are a few scattered meso-
blast cells. From the brain, B, there is part of its substance
which projects ventrally towards the oesophagus and is quite
conspicuous in sections of the Pantopod-larva in this region. On
each side of the brain are seen a pair of diverticula of the mid-
gut. These go to the first pair of appendages and are traceable
to D1 of Fig. 13. The first pair of appendages are, as Dohrn has
shown, innervated from the brain.

The section of the embryo following Fig. 13 gives a cross-sec-
tion of the pair of simple eyes. A part of this section is shown
by Fig. 15. In the middle line of this section above the posterior
part of the brain, B, the ectoblast is thickened beneath the
cuticle, and immediately on each side of the middle line appear
the pair of eyes. Each is seen in section to be composed of
three cells — two clear outer ones with large nuclei and an inner
much pigmented cell. Around the eyes the ectoblast is quite
thick, and on each side it sinks down slightly from the surface,
suggesting that the eye may be here increased in size.

In Fig. 16 is a section through the third pair of ventral ganglia
which we saw on the ventral side of the embryo posterior to the
fused pairs of first and second ganglia. The section shows that
at this place the epiblast is greatly thickened and the cells
columnar, with clear outer portions. In the ganglion on the left
of the section the cuticle seemed to dip down into the center of
the ganglion, where it becomes extremely thin. This was not

8 T. H. MORGAN.

so clearly seen on the right-hand ganglion, but traces of it were
seen here also. These structures I shall speak of as the Ventral
Organs ; and we shall come across them again in Pallene, where
they are worked out in greater detail.

Pallene empusa.

There is a great variability in the size of the eggs on the
ovigerous legs of the males. The average size of those which
seem to be normal eggs is .25 mm., but often bunches contain
one or more much smaller eggs, some of which seem to begin to
develop. These very small eggs are most probably immature
'and accidental. Although the eggs of Pallene are much larger
than in the preceding animals — having 125 times the volume —
yet the yolk is almost transparent, so that nuclear divisions of
the segmenting egg are easily seen from the surface. The adult
is remarkably transparent and consequently most difficult to see
as it rests quietly amongst the hydroids or sea-weeds. It seems
that the eggs have also adapted themselves for purposes of protec-
tion in thus becoming transparent. More or less time necessarily
elapses after the animals are collected before they can be exam-
ined, so that I have been unable to see the extrusion of the polar
bodies.

The nucleus of the egg and its accompanying protoplasm
is extremely large and situated near the center of the egg.
Each division of the nucleus is accompanied by a division of
its surrounding protoplasm, so that in surface views it is impos-
sible to separate the one from the other. In the figures of Plate
III the darker masses in each cell indicate the position of the
nucleus and protoplasm, but for the sake of brevity I shall speak
of this simply as the nucleus. In the unsegmented egg the
nucleus is seen to elongate and to divide into two halves, and
this is immediately followed by division of the egg itself into
two parts. The first furrow divides the egg into two unequal
segments. Fig. A, Plate III, shows an egg after the first divi-
sion. Each segment contains a single nucleus. (Hoek has made
a different observation from this — see infra.) After this first
segmentation the segments flatten somewhat and remain so
during the resting period. The second division came in about
three-quarters of an hour (to an hour) after the first. The

THE PYCNOGONIDS. 9

nucleus of the smaller segment — the micromere — divided first
and then the segment itself into equal parts. This was followed
five minutes later by the division of the nuclei of the large seg-
ment— the macromere — and then the macromere itself.

In Fig. B an egg is shown as seen from above after this second
division, which is indicated by 2 — 2. The planes of segmentation
of the micromere and macromere are drawn in this figure as
though they coincided, but this is not the rule. The plane of
segmentation of the micromere may be turned at as much as 30°
(or exceptionally almost 45°) to that of the macromeres. I think
it should nevertheless be considered as one and the same plane
and one and the same division, in spite of individual variations.

The segments again flatten together, and after an interval of
about an hour the third division commences. The plane of
division is at right angles to the last and is shown by Fig. C, in
which the egg is seen from above. Again the planes of micro-
mere and macromere do not quite coincide ; the numbers 2 — 2
indicate the second division and 3 — 3 give the plane of the
present (third) division. This plane is a zigzag line lying
between 3 — 3. If we examine the opposite pole, the lower, of
an egg at this stage, we find the large macromeres come together,
as shown in D, although there are many differences of detail in
this respect. So far the planes of division of micromere and
macromere have been supposed to coincide or to be referred to the
same division plane, but after this stage is reached (8 segments)
the planes cannot be considered identical. About an hour after
the last division the fourth rhythm comes on. This division plane
is shown for the macromeres in side view in figure E, by the line
4 — 4, and is seen to be parallel to the first plane of segmenta-
tion, 1 — 1. The four macromeres become eight. About the
same time — maybe five minutes before or after — the micromeres
divide. In Fig. E they are drawn after division has taken place,
and Fig. F shows how these cells are divided, the curved lines
indicating twin cells. As we see from the very nature of the
division of the macromeres, it is impossible to have their plane
of division to correspond with that of the micromeres, and the
division lines of the latter lie on a plane nearly at right angles
to the plane of division of the micromeres. We have now eight
macromeres and eight micromeres. After a resting period of
about an hour the fifth rhythm begins.

10 T. H. MORGAN.

The nuclei of the macromeres of Fig. E show by their elonga-
tion how they are to divide, and the plane of division lies between
them or at right angles to the last division plane. Fig. G shows
the egg after division into eight micromeres, but the micromeres
had not as yet divided and remain eight in number. Soon,
however, they do divide, but no definite plane of division for
all of the segments was discovered. There are now sixteen
macromeres and sixteen micromeres.

The next plane of segmentation of the macromeres — at the
sixth rhythm — comes in after another resting period of an hour.
This is shown by Fig. E at 6 — 6, 6 — 6. The two planes are at
right angles to the last two (5 — 5) and are parallel to the first
and fourth (1 — 1, 4 — 4). I did not see the sixteen micromeres
divide into thirty-two, nor could I trace them further. There
are thirty-two macromeres and sixteen micromeres, .or in all
forty-eight cells. After an hour's interval the macromeres lying
between the planes 1 — 1 and 4 — 4 in Fig. H were each seen to
have two nuclei, but this did not seem to be the case with the
macromeres of the lower pole. The egg had been under obser-
vation for mauy hours and did not develop further, so that this
last attempt at division in the upper macromeres may have no
special significance.

Besides the general facts of segmentation as just given, several
interesting variations in the method of segmentation of the egg
were clearly made out. In the example given above the micro-
meres divided first (at the second segmentation), giving two
micromeres and one macromere. In two other observations this
was also true, but in three other cases the macromere divided
live minutes before the micromere. And again in the last case
the two macromeres divided into four before the two micromeres
divided into four, so that the greater amount of yolk of the
macromere did not seem altogether to retard segmentation. In
several later stages (thirty-two macromeres and sixteen micro-
meres) the rhythm of macromeres and micromeres given above
did not so closely correspond, but the micromeres seemed to
drop behind.

If the accumulation of yolk has been a very recent event in the
egg of Pallene — and the variations in size may bear this inter-
pretation— these differences in the method of segmentation may

THE PYCNOGONIDS. 11

be due in part to the segments of the egg struggling to give
synchronous beats at each rhythm, but being hampered by the
presence of the yolk. If the first plane of segmentation of Pallene
corresponds with the first plane of division of Phoxichilidium and
Tanystylum, and & priori this seems most probable, is it not con-
ceivable that the acquisition of yolk to one-half of the egg might
cause great changes in the synchronism of segmentation of the
two parts ? and if so, it is easy to imagine that we have here an
egg in a period of variation, and that one or another of the above
changes may become fixed for the species.

Later stages of segmentation than those described are difficult
to follow or to distinguish between macromere and micromere
of the upper pole, but in general the cells around the upper pole
are smaller and more numerous than at the lower. Fig. J,
Plate III, shows the ventral (or macromere) hemisphere of an
egg at a later stage than any of the preceding. The outer ends
of the cells, those seen in the figure, are polygonal, and the nuclei
lie very near to the surface. The whole cell is pyramidal in
shape, with its apex at the center of the egg and its polygonal
base at the periphery of the egg^ with the nuclei in the base
of the pyramids. Each nucleus is still accompanied by a
surrounding mass of protoplasm, and it is without doubt this
protoplasm which is seen from surface views. At this time the
cells of the lower pole rarely divide, and remain for a long time
almost constant in size and number, but the cells at the upper
pole undergo rapid changes ; and our attention must now be
turned almost entirely to that region.

The next change which we can see in surface views is at the
upper pole, where a whitish opaque area is forming, and which may
be profitably compared to the similar formation in spiders' eggs and
called there the Primitive Cumulus. This is due to an accumu-
lation of cells at this point, where an invagination is about to
take place to form the stomodseum. Soon there are other opaque
areas formed at the surface, which are seen to occupy definite
positions and are the thickenings for the brain, ventral ganglia and
appendages. These are shown in Figs. I and II, Plate IV. In
Fig. I wTe have a surface view of the head region of a young
embryo. In the upper part of the figure are first the two oval
thickenings to form the brain or supra-cesophageal ganglion.

12 T. H. MORGAN.

At this stage they just touch on the middle line. Below this and
on the middle line is the thickening of the stomodseal invagina-
tion. In the center is a triangular-shaped cavity — the cavity
of the invagination. The base of the triangle is towards the
brain. On each side are the thickenings of the first pair of
appendages. In the lower part of these is a slight depression,
which indicates the line between the claw and the next segment
against which it works. Posterior to these various thickenings
appear two rows of opaque areas, the nerve ganglia. Three
pairs of these are seen in this figure, the first two being somewhat
smaller than the third. On each side of the last pair appear part
of the thickenings of the first pair of ambulatory legs (the fourth
pair of appendages).

Fig. II is a continuation of the last figure, and shows the ven-
tral and posterior part of the same embryo. The upper pair of
ganglia are the second pair, and were seen in Fig. I, and the
second pair of this is the third of the last figure. On each side of
this third pair of ganglia are again the first pair of walking legs.
Two more pairs of ganglia follow this third. On each side of the
fourth and fifth pairs of ganglia appear the second and third
pairs of walking legs. An embryo at a stage later than the last
is shown by Fig. III. Here the embryo has elongated in an
antero-posterior direction, so that the figure shows only the
ventral side of the animal. Fig. IY is a dorsal view of an embryo
at the same stage. These figures show that the region about the
stomodseum has grown forward and has the opening of the stomo-
dseum at its distal end. This outgrowth is seen in the figures as
a forward prolongation of the embryo on the middle line. On
each side of it are seen the first pair of appendages, which have
also grown forward.

On the ventral side of the embryo appear on each side of the
middle line five pairs of large ganglia (see Fig. III). On each
side of the last pairs appear the three pairs of walking legs, which
are longer than in the last figure, are bent, and stand out from
the surface of the embryo. Posterior to the last pair of ganglia
the embryo has a thickened mass of undifferentiated substance.
In a dorsal view (Fig. IY) we see the brain, and behind it the
large yolk mass of the embryo. The two halves of the brain
have come together, and each half is slightly bi-lobed. Around

THE PYCNOGONIDS. 13

the periphery of the yolk appear the six pairs of appendages,
and behind, the thickened posterior mass of the embryo.

Returning to the ventral view of the embryo (Fig. Ill), there
is seen a most interesting structure in the center of each ganglia.
Each is an invagination of the surface into a ganglion, and these
invaginations are elliptical in outline, with the long axis corres-
ponding to that of the embryo. These structures I shall call the
Ventral Organs, and, by means of serial sections, we shall later
study them in more detail. In the next stage, shown by Fig. V,
the embryo is more oval in outline. The appendages have also
elongated and become more bent. The posterior pair have
begun to grow forward between the more anterior pairs. The
most important change is in the fusion of the first two pairs of
ganglia, which now appear as one pair of rather large ganglia.
This pair of ganglia still shows its double structure by the pres-
ence of two pairs of ventral organs. The third pair of ganglia
also shows a pair of ventral organs, but the more posterior ganglia
are covered by the posterior appendages, so that these ganglia
cannot be seen from the surface.

After this stage is passed the appendages grow enormously in
length, and the embryo becomes flattened from side to side. A
figure of an embryo at this stage is shown in side view in Fig. VI.
The first appendage is chelate, and has a joint near its base, and
one also sees it has moved more dorsally to the proboscis. Beyond
and beneath this appendage appears the proboscis, which has
much elongated. Near the base of the appendage is seen part
of the brain. This figure also shows that the yolk mass is con-
tinuous along the center of each ambulatory limb. The three
pairs of legs are much bent, and each ends in a spine-like process.
The body is seen to end behind in a knob-like projection. The
ventral ganglia are shown between the bases of the legs, and the
height of each at this stage is about equal to its length. In older
embryos the yolk begins to disappear and the sides of the embryo
to become thicker. After this the embryo lengthens a great deal,
the appendages grow much longer and become segmented.
Another pair of appendages appears behind the third pair of
walking legs, and the knob-like projection at the end of the
embryo is pushed more dorsally, to form the rudimentary abdo-
men. Four pairs of eye spots appear over the posterior end of
the brain.

14 T. H. MORGAN.

After these changes have taken place we reach a stage shown
by Figs. VII and VIII. The first figure is a ventral view of the
embryo at a time when it is ready to leave the parent. The three
pairs of segmented walking legs have become straightened out at
the sides of the body. In the figures only the proximal ends
are shown. The fourth pair of walking legs appears at the pos-
terior end of the body. The first pair of appendages — the cheli-
cerge — are now attached quite dorsally to the proboscis, which
appears between and below them. Each has three segments,
including the terminal one, which acts together with the second
to form the pincers. On the sides of the body, just in front of the
first paif of ambulatory legs, are a pair of projections, one on each
side. These are the beginnings of the third pair of limbs — the
ovigerous legs. I have seen no traces of the second pair of
appendages in the ontogeny of Pallene. Five pairs of large
ganglia lie along the body, and in addition a small sixth pair.
The first pair, as we have seen, is double, and composed of the
first and second pairs. The next pair, the third, is between the
first pair of ambulatory legs; the fourth pair of ganglia lies
between the second pair of legs; the fifth near the base of the
third pair of legs, and the sixth, each of whose ganglia are smaller
than those in front, at the base of the fourth pair of legs. The
small, seventh pair of ganglia belongs to the rudimentary
abdomen, and does not lie in a plane with the more anterior
ones, but, like the sixth, is dorsal to the one in front of it, as
partially shown in the figures. The oesophagus is seen to start
at the distal end of the proboscis and to disappear in the body,
where it is not easily followed on account of its transparency.
The prolongation of the mid-gut into the legs is not shown in
the figures, except for the last pair of immature legs.

In Fig. VIII is a dorsal view of the same embryo. Four pairs
of large ventral ganglia are seen, and in addition the small rudi-
mentary ganglia of the abdomen. The fused first and second
pairs are hidden by the brain. On the surface of the brain are
seen four small pigmented eyes. The oesophagus is triangular
in shape in cross-section, with the broad base turned upwards,
and this base is seen from the surface. At the posterior end of
the animal the rudimentary abdomen is seen, and at its end is
the opening of the anus.

THE PYCNOGONIBS. 15

Internal Changes.

Sections of the early stages of segmentation give little infor-
mation in addition to what we see from surface views. They
show that each nucleus is contained in a separate segment, and
that many of these segments run to the center of the egg. Those
cells derived from the first micromere cannot reach to the center
of the sphere.

Figrl-, Plate I, is a section of an egg at the stage when there
are thirty-two micromeres. The upper part of the figure is
probably in the region of the micromeres. In the center of the
egg appears a clear cavity without yolk, but this is not constant
for all eggs. In some it is certainly absent, but I am unable to
say whether or not it is caused by hardening reagents. What
seems to be a similar cavity is found in the eggs of some spiders.
The whole egg is divided up into the segments in the form of
pyramids. Only some of these in this figure can be traced to the
center, but other sections show each pyramid of the macromeres
to have its apex at or near the center of the egg. The nuclei
are situated around the periphery of the section, and the proto-
plasm, which invariably accompanies each nucleus, sends out
fine pseudopodial filaments into the surrounding yolk.

Fig.-ffcis in the same stage as the last, and undoubtedly passes
through the micromeres at the upper pole. These micromeres
and their nuclei are seen to be smaller than the cells elsewhere,
and the pyramids are seen to fall short of the center of the egg.
There is no central cavity present in this egg.

Sections of eggs somewhat older than the last show that the
nuclei of the upper pole have rapidly increased, and that they have
migrated to the surface of the yolk, which loses its pyramidal
structure over the upper surface. The protoplasm surrounding
each nucleus comes into contact with that of surrounding nuclei,
though how closely such fusion is formed I cannot say. Such a
condition is shown in Fig. 3, where the protoplasm forms a cap
over the upper surface of the yolk. The nuclei in this section
are abnormally large, which is probably due to the hardening
reagents, but the section seems normal in other respects. The
lower area of the same egg shows three or four scattered nuclei
near the surface of the yolk. At this stage of development there
is a single layer of cells at the surface of the yolk. Between this

16 T. H. MORGAN.

stage and the next which I have figured there is a gap. In this
next section, Fig. 4, we see the number of nuclei at the upper
pole to be more numerous than before and much smaller. The
protoplasmic cap has become larger and its protoplasm is for
the most part without the amoeboid processes, of the last figure.
A very important change has taken place between these two
stages, viz. the formation of an inner layer of cells. Within the
area of the cap appear a few somewhat flattened cells in Fig. 4,
and these send out processes into the underlying yolk. Two of
these cells are shown in this figure. I have not made out any
definite arrangement for these cells, but they seem to lie only
under the upper surface of the embryo. Where these cells come
from or at what time they appear must in part be a matter of
conjecture, but much light is thrown on their possible origin by a
study of somewhat older stages. In Fig. 5 is a section of such a
stage. The number of nuclei in the outer germ-layer has nearly
doubled, and the area itself covers a much greater surface of the
egg than in the last stage. Under the blastoderm are seen five
inner cells with their pseudopodial extension. The larger num-
ber of these cells here than in the last figure is due in part to
the greater thickness of the sections. At the lower pole are
two nuclei. These belong to the outer germ-layer, although they
show (as did all the outer at an earlier stage) the protoplasmic
extensions into the yolk. This section itself throws little light
on the cells of the inner layer, but in other sections at a similar
stage I have found what seems to furnish the solution of the
problem. From such sections I have drawn Figs. 6 and 7.
These are taken from the periphery of the cap of cells at some
such point as P in Fig. 5. Fig. 6 shows two cells which have
just separated from the outer layer of cells, and we also see
beyond them a single mass of protoplasm with two nuclei which
presumably have just divided. I have not seen any karyoki-
netic figures in nuclei dividing in this direction, and it is possible
we have here a direct division, as Heider has recently shown in
Hydrophilus at a similar stage of development, although it is
equally as possible it is due to poor preserving agents. Again,
in Fig. 7, we see a single mass of protoplasm with two nuclei,
the result of division. These I believe to give a clue to the
origin of the inner cells of the preceding figures, and to point
out that they too have had a similar origin from the outer layer.

THE PYONOGONIDS. 17

Keeping before us the process of delanrination in Phoxichi-
lidium, etc., I think we may regard these inner nuclei of Pallene
to have come from the outer cells by delamination, and even
that we may push the comparison a step farther and consider
that each cell of the outer layer to have given rise at one time
in its history to an inner cell, and then that the outer cells con-
tinued to divide tangentially to form the blastoderm. The
reasons for such a belief are these : — In cross-section the number
of inner nuclei are slightly in excess of the peripheral nuclei of
the lower pole (see Figs. 4 and 5). As the outer cells of the
upper pole were at the beginning more numerous than the peri-
pheral cells of the lower pole, we ought to get, if the hypothesis
be true, exactly what we do find. Further, at the periphery of
the blastoderm, where the inner cells of the lower pole are added
on, we can always see such a method of multiplication taking-
place. The differences between this process and that in Phoxi-
chilidium are these : that multipolar delamination does not take
place simultaneously in all the cells at once, but in Pallene
slowly progresses as the cap of cells makes its way to the lower
pole of the egg, incorporating into itself as it goes the outer cells,
which cells, as they are added on first, give off an inner cell.

After this stage we pass to older embryos, where these two
layers — ectoblast and entoblast — begin to differentiate into
organs. The first to appear is the stomodseum, which results
from an invagination of ectoblast. This is shown by Fig. 17,
Plate II. Here it is seen the ectoblast at one point has pushed
inwards, and around the periphery of the invagination appear
several cells with branching and anastomosing pseudopodia.
These I believe to be the first appearance of the mesoblast. A
few of the cells drawn in the figure belong without doubt to the
entoblast, and at this stage it is difficult to separate the two. The
section passes longitudinally through the embryo, and just ventral
to the stomodseum there is a thickening of the ectoblast to form
the first ventral ganglion. Under the ectoblast are found much
branched entoblast cells, which are still comparatively few in
number. Whether the circumstomodasal mesoblast comes from
ectoblast or entoblast I am unable to say. Dorsal to the epiblast
the epidermis continues conspicuously for a short distance and
then becomes exceedingly thin.

18 T. H. MORGAN.

From this stage we pass to embryos of the age represented by-
Figs. I and II, Plate IV". The first figure, 18, is a cross-section
through the stomodseum of Fig. I (see line 18 of the figure). The
invagination is deeper than in the last section, and its lumen is
closed. The ectoblast nuclei of its walls are two layers thick. On
each side of this central invagination is what appears to be a lateral
invagination. These supposed invaginations have given me end-
less trouble, and even now I feel some uncertainty as to my
interpretation of them. On each side of these is seen the thick-
ened ectoblast of the first pair of appendages. The lateral in-
vaginations are, I believe, caused by the growth of these append-
ages, which tend to grow outwards as they increase in size ; but
are prevented from doing so by the egg coverings, and the result
is that the ectoblast becomes folded on itself to make room for
the growing appendages. This view is strengthened by the pres-
ence of somewhat similar invaginations at the side of the other
appendages. This part which is pushed in would correspond to
the dark furrow between each appendage of Fig. I and the cells
forming the stomodseum. Themesoblast around thestomodseum
and under the appendages has increased, and is now clearly dis-
tinguishable from the inner covering of the entoblast, which lies
only at the periphery of the yolk, and between it and the meso-
blast.

In Fig. 19 we have a section from the same series as the last,
but more anterior and through the region of the brain correspond-
ing to line 19 of Fig. I. The section is entirely in front of the
stomodseum, and cuts the two brain thickenings of the surface
view. Here again the ectoblast is seen to be distinctly folded,
and this folding (invagination !) is directly continuous with the
last. At first sight this seems nothing more than a forward con-
tinuation of the last groove, but it is not clear why the folding of the
appendage should exert any influence over that part of the brain.
Again, it might be interpreted as folding due to the brain, but I
can see no good reason why such thickenings (not outgrowths)
should produce the groove. There is one other possibility, viz.
that these may represent invaginations into the brain itself.
Against such a view is the absence in surface views of any such
invaginations, and also that the folds are directly continuous with
and quite similar to the groove between the stomodseum and the

THE PYCNOGONIDS. 19

appendages, so that I cannot hold this either as a true solution.
Beneath these ingrowths appears the pair of thickenings for the
brain, which are seen to be continuous across the middle line.
On the inner surface of these lobes are seen a few entoblast cells,
as shown.

The next figure, 20, passes through the middle of the brain,
but is farther forward than the last (see line 20, Fig. I, Plate III,
which indicates the plane of the section). The cells of the ecto-
blast seem to pass into those of the brain. Sections still farther
forward, where the two lobes are still cut, show the brain thick-
enings on each side and separated from each other. We see in
surface views of embryos at this stage that the ventral ganglia
have appeared, and Fig. 21 is a section through two of these —
the first pair. Not only do we see that the ectoblast has greatly
thickened in two places on each side the middle line, but the
outermost cells of each mass show certain peculiarities. They
are much elongated at right angles to the surface, and their outer
points come together at the middle point of the surface of the
ganglion. The left-hand ganglion in the figure is cut through
this central point, and the ganglion on the right a little to one
side. The nuclei lie in the inner parts of the cells, and the outer
parts of the cells are clear and transparent, while their inner
ends are more granular. This difference in the parts of the cells
is very constant through the later stages. The cells form a
structure which I have called the Ventral Organs.

In Fig. 22 is a section of a pair of ventral ganglia from an
embryo at stage III. Here there is in the middle of each ventral
ganglion a wide invagination lined by columnal cells with their
clear outer portions turned towards the cavity of the invagina-
tion. The nuclei in the cells are larger than those in other parts
of the ganglia, and are seen quite often in process of karyokinetic
division. The spindles of the dividing nuclei are in some at
right angles to the long axis of the cells, and. in others parallel to
the axis. A narrow connection of small ectodermal cells runs
across the middle line from ganglion to ganglion.

The next figure, 23, is from an embryo at about stage V. The
outer edges of the invagination have turned in until they have
nearly met above the cavity of the invagination. In other
respects the section is similar to the last.

20 T. H. MORGAN.

At the next stage, shown by Fig. 24, the arching over is com-
pleted and fusion has taken place, so that there is a cavity in
each ganglion. The cavity does not run through the whole
length of each ganglion, but lies only in its middle portion.
There is no communication between the cavities of different
ganglia. In this figure each ganglion has increased in size, and
the cells have become more numerous, and, further, the neighbor-
ing ganglia are connected by a cross commissure of fibers.
Further, the nuclei of the outer cells, those lining the cavity, are
now more like the nuclei of the ganglion cells. Later stages
show that the central cavity of each ganglion disappears,
although they seem to persist for quite a long time after they
have been enclosed by the ganglia. That this structure may
have represented an organ in some ancestral form I believe
possible, first, because it is distinctly marked, differentiated,
before any invagination takes place; secondly, because their
cells are histologically distinct from the surrounding ectoblast
and from the cells of the ganglion beneath, and lastly, from the
arrangement of the cells, which suggests a sense organ comparable
to the simple ectodermal organs of many animals. It also seems
probable that we have exactly similar structures in the Pantopod-
larva of the other Pycnogonids.

Returning to stage III we have a cross-section of the body
drawn in Fig. 25. The section passes through the middle of the
body in the plane of a pair of walking legs. The large mass of
yolk is still seen filling the upper part of the embryo. Over the
ventral surface of the yolk the entoblast cells form a continuous
covering and are indicated by JV in the figure. At the base of
the appendage the entoblast sends out a prolongation which is
filled with yolk, while the entoblast cells are here higher and
more closely packed than elsewhere. Into each of the six
walking legs is a similar protrusion of entoblast from the mid-
gut. Beyond these diverticula there is in each leg a definitely
marked cavity in the mesoblast. In Fig. 25 the surrounding
mesoblast is shown by M. As the prolongations of the mid-gut
push into the legs these cavities cannot any longer be made out,
and either become lost or are too complicated to follow out.
Besides these cavities, which would seem to be body-cavities, we
shall come to others later in development which seem to be

THE PYCNOGONIDS. 21

schizocoels. The surrounding mesoderm of these latter is irreg-
ular and quite different from that around the body-cavity proper.
The ectoderm over the appendages is thicker than elsewhere.
The ventral ganglia at this stage have in most cases the ventral
organs, although in this figure none are open and they have
closed rather early. The ectoblast of the dorsal surface is quite
thin and the mesoblast has not appeared in that part of the
animal.

In Fig. 26 is a drawing of a cross-section through the head of
an embryo at a stage a little later than III, but not so late as Y.
The section passes through the brain above and the second pair
of ganglia below. Between the brain and the ventral ganglia
the stomodseum is seen. Its lumen is seen to be triangular in
outline, with its base turned upwards. On each side of the
stomodseum appear cross-sections of the first pair of gut-diverti-
cula. A single row of entoblast cells surrounds the central
mass of yolk. Scattered mesoblast cells are found around and
between these different structures, leaving here and there
openings between the cells. Such schizocoels are shown by S, S
in the figure.

Going on to stage V we find that sections give little more
than was seen in the preceding figures. The prolongations of
the mid-gut into the legs have grown longer, and, as seen from
surface views, the first and second pairs of ganglia have fused.

In stage VI the yolk mass continues to fill the central and
upper part of the embryo, but it now begins to decrease in
quantity. This may be due in part to the fact that part of it
becomes digested and built up into the tissue of the embryo, and
in part to its extension into the legs.

A cross-section of an embryo at a stage a little older than YI,
in a plane of a pair of limbs, is shown in Fig. 27. The ventral
ganglia have increased enormously in size and show a large
amount of punkt-substance. In the middle of the section lies
the mid-gut with its contained yolk, which has decreased very
much as compared with preceding stages.

In Fig. 28 is drawn a section of another embryo, at about
the same stage as the last, but cuts the embryo between a pair
of legs. Above the large ganglion is seen the digestive tract,
which is completely separated from the body-wall by a series of

22 T. H. MORGAN.

schizocoels. The uppermost of these, lying in the middle line, is
the heart, H. At the sides of the mid-gut appear two well-
marked cavities, and a third between the nerve ganglion and
the outer wall. Also below these are two or more spaces, but
these latter, and perhaps some of the others, are artefacts. The
mesenteron is covered on its outer side by a distinct layer of
mesoblast cells. Cross-section of some of the legs lying below
the body is also shown in the figure.

Between stage VI and stage VII- VIII there is a considerable
gap. During this period the embryo has lengthened and the
fourth pair of walking legs has appeared. The rudimentary
abdomen has been pushed up dorsally and the proctodeum
invaginated until at stage VII it has opened into the mesen-
teron. The eyes are now seen, but at stage VI they are seen in
sections as four thickenings of the epiblasts above the brain. In
stage VII- VIII we find by sections that the yolk has almost
completely disappeared from the digestive tract. Cross-sections
also show the large nerve ganglia, the mid-gut with its diverti-
cula, and on the dorsal part of the mid-gut the tubular heart
running from the first walking leg to the third.

Fig. 29 is from a cross-section through the posterior part of
the embryo and cuts the mid-gut at its juncture with the procto-
deum. Below the digestive tract D is a pair of small ganglia,
V1, which are the ganglia of the rudimentary abdomen. On
account of the shifting of the abdomen to the dorsal side these
ganglia have been carried above the last ventral ganglia, which
are shown at V6. But this is not all, for this sixth pair of
ganglia has likewise been afiected by the shifting of the abdomen
and in turn lies above and back of the fifth pair of ganglia. In
the sixth pair of ganglia, as shown in the figure, we have evident
traces of the ventral organs. In the upper part of the section,
and on each side of the middle line, are the diverticula of the
fourth pair of walking legs. A few scattered mesoderm cells
appear in the cavity of the body. The embryo must leave the
male at about this time, for I have never found older embryos
attached and only exceptionally ones so old as these last figured.

Comparisons.
Comparing the embryology of the two types represented by
Phoxichilidium and Pallene, we find that most of the changes of

THE PYCNOGONIDS. 23

the latter may be explained by considering it an abbreviation
of the type represented by Phoxichilidium. An exact com-
parison of the segmentation of the two forms would be of
interest; but, in order that such a comparison should be of
value, the exact orientation of the segmentation plane would be
essential, and such observations are wanting.

It would seem a priori most probable that the first planes in
each must correspond, and that the unequal segmentation of the
e^g of Pallene has been caused by the greater accumulation of
yolk in that part of the egg which corresponds to the macromeres.
It is also probable that the smaller segments — the micromeres —
correspond either to the anterior or to the ventral part of the
embryo(?). Which of these is correct it is difficult to say. The
embryo differentiates earlier in what corresponds to the anterior
region of the adult than over the whole ventral surface, which
suggests that the smaller cells may have adapted themselves to
this early differentiation ; but it seems equally possible that this
differentiation may be due to phylogenetic laws in this particular
case rather than to any mechanical connection with the micro-
mere differentiation. So that for the present the question must
remain unsettled until by actual experiment (which would not
be difficult) the orientation of the segmentation planes be
determined.

There is an observation on the segmentation of the egg of
Pallene brevirostris by Hoek, which was mentioned in the pre-
vious description of segmentation of Pallene empusa. He says :
" Le fractionnement commence par le fractionnement du noyau,
et seulement apris que quartre noyau sont formes, un premier
fractionnement devise l'oeuf en une partie plus grand et une
autre beaucoup plus petite. Chaque partie contient deux
noyaux qui dans le plus petit segment sont plus rapproches
1'une de 1' autre que dans 1' autre segment."

I cannot believe this a correct observation in the light of what
I have observed over and over again in Pallene empusa. There
is a stage which corresponds exactly with the description, but
the egg itself has previously divided into two with a single
nucleus in each segment, and subsequently each of these nuclei
divides into two just before the segment itself divides, and at the
same time the first furrow becomes more distinct again as the

24 T. H. MORGAN.

segments round off to form the second furrow, and this seems
to be the stage which Hoek has described as the first segmenta-
tion. Nor does it seem probable that the differences of our
observations are due to their having been made on different
species, for in each case the egg is approximately the same size.

The changes which take place in the formation of the endo-
derm of Pallene may, I believe, be referred to the simpler
changes of Phoxichilidium, etc., and furnish in this respect an
excellent basis for further comparison with other forms having
much yolk present. In Phoxichilidium the pyramidal segments
divide into an outer and an inner cell, while in Pallene the
nuclei alone divide; and although delamination is still multi-
polar, it is not synchronous over the egg. A further comparison
of the mesenteron I am unable to give, owing to lack of com-
plete observations of the changes of Phoxichilidium. In both
types the oesophagus invaginates with a triangular lumen, and
in each the proctodeum forms later in development.

The few observations I have made on the ventral organs of
Tanystylum leave no doubt that it is the same structure as in
Pallene. Prof. Sedgwick has described in Peripatus paired
ventral organs which correspond in number and position with
the pairs of ventral ganglia. Comparing Fig. 21, Plate II, with
his figure for the ventral organs of the jaws of Peripatus (Studies
Morph. Lab., Camb., Vol. IV, Pt. I, Plate 10, Fig. 4), a striking
similarity is seen, and, in addition to this, he says in his account
of the ventral organs that they are slightly invaginated from the
surface. Whether these structures are in any way related it is
impossible to say, but it is worth while to call attention to the
close similarity both in position and structure between these
organs in the two groups.

A comparison of the appendages and their order of develop-
ment is of interest. Prof. Dohrn has most carefully and fully
worked out the transformations of the larval forms, and through
his skill we have a very thorough knowledge of the transforma-
tion of the embryo. According to his account the six-legged
larvae of the Pycnogonids with the Pantopod-development pass
into the adult condition by the body elongating behind the last
pair of appendages and the walking legs appearing from before
backwards, in much the same way as Prof. Claus believes the

THE PYCNOGONIDS. 25

typical Nauplius to pass into the adult. During the growth of
the walking legs the second pair of appendages of the Pantopod-
larva has increased a little in size, but the third pair loses its
spines and the whole appendage becomes a simple prolongation
of the embryo. At a later stage, when the larva has increased
in size, this third pair grows out again to form the ovigerous legs.
In Pallene we have seen that the first, fifth and sixth pairs of
appendages appear simultaneously in the embryo. The seventh
comes in very much later, and the third after the seventh*. The
second did not appear at all in the ontogeny, so that in the
young Pallene the only appendages in the young embryo that
correspond to those of the Pantopod-larva are those of the first
pair. The development of Pallene has become so much abbre-
viated that there is only a trace of the true Pantopod-larva
found in its ontogeny.

Phylogeny.

There is a general agreement that the Pycnogonids are to be
placed within the large group of Arthropods, but after this there
is the greatest divergence of opinion as to which group of Arthro-
pods they are most nearly allied. In general there are two
prominent categories to which all or nearly all of these theories
may be referred. One lot of workers believed in a Crustacean
relationship, and another lot placed the Pycnogonids amongst
the Arachnids. It is needless to give here the reasons assigned
for these opinions, as Prof. Dohrn has given in his monograph
of the Pantopoda of the Gulf of Naples a most complete and
exhaustive bibliography of the literature. The two most import-
ant views which we have at present are those of Dohrn and of
Hoek, but, as the latter agrees with much that the former has
given, we may consider the two together.

Prof. Dohrn's earlier view of the relationship of the Pantopod-
larva to the Nauplius he has completely given up in his more
recent papers. In an appendix to a paper in the Archives de
Zoologie experimental, Dr. Hoek gives a summary of Dohrn's
recent theory as to the phylogenetic position of the Pycnogonids,
and has at the same time contrasted his own views with Dohrn's,
so that a translation of a part of this review will best bring out
the resemblances and differences of their opinions.

26 T. H. MORGAN.

" For the Crustacea Dohrn rejects the Nauplius theory (of
Fritz Miiller and Claus) and adopts that of Hatcheck, who
believes the Crustacea to have descended from parents which had
the form of Phyllopods, just as these Phyllopods have in turn
descended from Annelids. From these same Annelids, according
to Dohrn, the Pycnogonids have come down. The number of
their segments was originally more numerous than we now see
them, the presence of a pair of rudimentary ganglia in connection
with the last pair of ventral ganglia allows us to add an eighth
pair of appendages, and, together with the first, all of these
appendages were originally homotypes. . . . They received diver-
ticula from the intestine . . . and each appendage enclosed
within itself a reproductive organ with a special genital opening.
(The so-called excretory organs of the palps and of the oviger-
ous legs are the rudimentary sexual organs of these appendages.)
The appendages were much shorter than we now see them, the
heart showed many openings, etc. The supposed ancestor that
Dohrn reconstructs might be very well compared to an Annelid.

" We also notice that Dohrn persists in his opinion, published
in his work of 1879, that the Pycnogonids have a parentage
neither with the Arachnids nor with the Crustacea (they have
developed by the side of the last and altogether independently).
In this I am in accordance with Dohrn. I was struck in the
first place by the very general presence in the Pycnogonids of a
characteristic larval form (the Protonymphon), and its presence
suggested to me the idea of their descent from an ancestor
resembling somewhat a larva which took its place by the side of
the hypothetical ancestor of the Crustacea — the Nauplius — and
this by the side of a third (the ancestorof the Annelids), and that
all three groups would have descended from a common ancestor.
I had tried to obtain in this way an explanation of the affinities
of the three groups of animals (Annelids, Pycnogonids, Crustacea).
Dohrn, on the contrary, took into account the larval forms, but
constructed a phylogeny comparing together those animals having
a considerable number of segments. It is true I do not wish to
deny that such a method of looking at it has not just as good a
basis of reason, only it appears to me at present that it can be
defended just as little, and certainly no better than the larval
theory.

THE PYCNOGONIDS. 27

" No one affirms that the intestinal diverticula which now in
Phoxichilidium and Nymphon penetrate into the proboscis were
originally [ancestrally] in the palps and ovigerous legs. The
homology of the glandular organs of appendages II and III with
parts of the sexual organs is not based on any observation. . . .
But since I believe there are certain things which lessen Dohrn's
arguments in favor of the Annelid theory, I believe that at bottom
and the contents of pages 82-115 of his work confirms this
opinion, the ideas of this author and of mine upon the phylogeny
as well for the Crustacea as for the Pycnogonids are not very
different. At first I saw only the weak side of Dohrn's theory,
that he does not give any explanation of the almost universal
presence in the ontogeny of these animals of a characteristic larval
form (thelarval Protonymphon, which for myself is a true primary
larva in the sense of Balfour). So that I cannot adopt a theory
which causes to arise a form of Arthropods with many segments
from an Annelid with the same number of appendages ; but such
is not the opinion of Dohrn, as clearly expressed in the following
paragraph : ' Thus the larva of the greater part of the Pycno-
gonids may be regarded, with a grain of salt, as a form much
like the ancestor; but if, on the other hand, the absence of an
anal opening, the pincers of the first pair of appendages, the long
claws with their accessory spines, the structure of the proboscis
with its triturating apparatus and its ganglia, and the form of
the cutaneous glands with their integumentary hairs, can only
be considered as having been acquired in a much later stage, and
transmitted to larval life where they are found at present ; but
what remains in the larva that we may regard in reality as
inherited from its original condition ? Nothing but the nervous
system, i. e. the brain and ganglia, an intestine, three pairs of
appendages of a form modified according to circumstances, and
two eyes! But these are attributes which one finds equally in
the larva of Annelids with three segments ! And if then we take
into consideration that at one time the body of the Pycnogonids,
as we believe, showed a great uniformity in the segments (and
the generative organs prove this), that on the one hand there was
a concentration and a differentiation, and on the other hand
a reduction in number of segments, these two conclusions lead us
to admit a direct descent of the Pycnogonids from ancestral

28 T. H. MORGAN.

Annelid forms which were homonomously segmented. Then
the larva of the Pantopods came from a larval form with Panta-
pod characters added on, but at the same time a larva which
never had an independent and mature existence.' "

Whatever Hoek's position may be with respect to the Panto-
pod-larva, he agrees with Dohrn in the most important part of
the latter's theory that the Pycnogonids have come down from
the Annelids independently of the other groups of Arthropods.
In examining the preceding account we may take it up for the
sake of clearness in two parts, the first pertaining to the ancestry
of the adults, leaving the larval form out of account, and the
second part where the meaning of the larval form will be
considered.

I believe that if the account I have given of the early stages
of development be even approximately correct, there is little or
no ground for a comparison between Crustacea and Pycno-
gonids— certainly not with any existing forms. The multipolar
delamination of the endoderm in the Pycnogonids has no
homologue amongst the Crustacea, nor is there any special
similarity in the formation of the organs. There seems to be
no trace of gastrulation like the Crustacean in the ontogeny of
the group. And if we have reason for rejecting a relationship
between the Pantopod-larva and the Nauplius — and I believe
with Dohrn that we have — there remains nothing in common to
the ontogeny of the two groups.

Nor are there any special affinities between the Insects and
Sea-Spiders, but there is one striking similarity between the
latter and Peripatus, which I have already spoken of; but an
isolated fact of this kind gives little ground for further com-
parisons. The ventral organs in the two groups present a
striking agreement, but there is no proof forthcoming as to a
real homology of the structures. The process of the formation
of the endoderm described by Heider and by Wheeler in Insects
shows a certain resemblance to multipolar delamination, but if
it be such it is a more complicated form than shown by the
Pycnogonids. With these two exceptions there seems to be
nothing else in common in the ontogeny.

We are then left to decide between an independent origin for
the Pycnogonids or a relationship with the Arachnids. Prof.

THE PYCNOGONIDS. 29

Dohrn has ably maintained the first theory, and the preceding
translation gave the conclusion he had reached. Dr. Hoek
likewise, as we have seen, holds this opinion, although not
agreeing as to details with Prof. Dohrn. On the other hand, a
study of the early stages of the embryology has brought to light
certain facts which for me point decidedly toward a community
of descent between Arachnids and Pycnogonids. The latter
show undoubted traces of degeneration, and we cannot derive
them from any existing animals. But I believe the Pycno-
gonids and the Arachnids have come down along the same line,
or, in other words, have had ancestors in common long after
those ancestors came from Annelid-like forefathers. The reasons
for such a belief are as follows :

1. The Pycnogonids form the endoderm by a process of multi-
polar delamination, which is shown in its simplest form in
Phoxichilidium and Tanystylum, and in a more modified con-
dition in Pallene. In no other group of the Triploblastica do
we find a similar phenomenon except in the Arachnids. The
description of the embryology of Chelifer which Metchnikoff has
given shows this process or something quite analogous to take
place in it. The segmentation is holoblastic, and at a later
stage the large cells containing yolk divide into an outer, more
protoplasmic layer of cells and the inner cells, which are very
granular. The outer form the ectoblast, and most probably the
inner, judging from his figures, form the entoblast.

In the Spiders the process is not so well marked, but if the
conception which Balfour had of the formation of the yolk-
nuclei be true, then we may make a direct comparison between
the two groups. He says : " It appears to me probable that at
the time when the superficial layer of protoplasm is segmented
off from the yolk below the nuclei undergo division, and that a
nucleus with surrounding protoplasm is left with each yolk
column." This description for the Spiders may be substituted,
word for word, for the process of delamination of Pallene.

2. The first trace of the embryo to appear in Pallene is a
round opaque area at the spot where the stomodseum invag-
inates. In Schimkewitsch's recent account for the Spiders he
shows that the primitive cumulus in them is the place where the
stomodseum invaginates. This is also true for Pallene, but here

30 T. H. MORGAN.

the stomodseum invaginates quite early and perhaps simul-
taneously with the early formation of mesoblast at this place.
Further, Schimkewitsch has called attention to the fact that the
stomodseum of Spiders in its earliest development is a triangular
invagination, and he compares it directly with the triangular
invagination of the oesophagus of the Pycnogonids.

3. The early formation of body-cavity surrounded by meso-
blast in the legs of Spiders has an exact parallel in Pallene and
Phoxichilidium. Yet, however tempting such a comparison
may be in this connection, it must be admitted I have not con-
clusively proved this to be true for the Pycnogonids, but only
exceedingly probable.

4. In both Arachnids and Pycnogonids we have well marked
diverticula from the mid-gut into the legs. In the Pycnogonids
these go into the chelicerse and the four pairs of walking legs,
and the same holds for the Spiders ; but from a comparison of
the appendages of the two groups we must suppose that the
second pair of Pycnogonid's appendages to have lost their diverti-
cula, and the last appendages either to have acquired diverticula,
or more probably inherited them together with the appendage.
In Chelifer the diverticula appear very early in development and
contain some of the yolk from the mid-gut. This is shown very
distinctly in Metchnikoff's figures for Chelifer, and in this respect
the embryo resembles closely the embryos of Pycnogonids.

5. In both Arachnids and Pycnogonids the first pair of ap-
pendages are chelate. This in itself would draw attention to
the similarities of the two groups, but we know further that in
both groups this first pair of chelate appendages is innervated
from the brain. These facts were considered by Balfour suffi-
ciently important to indicate alone a relationship between the
groups. He says: " The presence of chelate appendages inner-
vated in the adult by the supra oesophageal ganglia rather points
to a common Phylum for the Pycnogonida and Arachnida,
though, as shown above, all the appendages in the embryos of
true Arachnida are innervated by post-oral ganglia."

I have not been able to find any post-oral ganglia for Pallene,
but the first pair of appendages arises on the sides of the stomo-
dseum and later moves forward. In this respect it compares
closely with the Spiders, and the early innervation of this pair

THE PYCNOGONIDS. 31

from the brain itself may be regarded as a more abbreviated
condition than what was seen (by Balfour) in the Spiders.
Metehnikoff's figures for Chelifer show the first pair of append-
ages to arise above and on each side of the proboscis-like upper
lip. If future work verifies Metchnikoff's suggestion that this
proboscis (riisselformige, provisorische Oberlippe) is homolo-
gous, entirely or in part, to the proboscis of the Pycnogonids,
as his figure might indicate, then does the whole development of
Chelifer show remarkably close resemblances to the Pycnogonids.

6. The fourth pair of ambulatory legs — the seventh pair of
appendages — has been a stumbling block in the way of those
who have compared Pycnogonid with Arachnid. Semper and
Schimkewitsch have attemped to solve the difficulty by assum-
ing that the third pair of appendages of the Pycnogonids — the
ovigerous legs — are new structures, and then have called the
four pairs of walking legs homologous in the two groups. Prof.
Dohrn has shown the impossibility of dropping out in the count
the ovigerous legs, and has shown that this pair of appendages
are homonomous with the others. The two pairs of ventral organs
in the anterior larval ganglia we have seen point unmistakably
to the same conclusion and give the final proof, if such were
really necessary.

We are led then to a comparison of the appendages of the two
groups, beginning with the chelicerse and going back pair for pair,
which leaves one pair over for the Pycnogonids. Any comparison
between the two groups must take into account this extra pair
of legs. Balfour has suggested that this last segment and its
appendages to represent the first abdominal segment of the
Arachnids. The third pair of appendages of the Pycnogonids
have assumed a special function, and at this time we might sup-
pose that an additional pair to have been added on from the
abdominal segments. We also know that the embryos of Spiders
have rudimentary appendages on the abdomen, and, as we have
assumed the Pycnogonids to have come from the latter group,
not recently but remotely, when these appendages may have
been larger or even functional, we have some ground for a belief
of such an origin of the last segment.

There is another fact which may be of importance in this con-
nection. Not only is the seventh pair of ganglia carried dorsal-

32 T. H. MORGAN.

wards by the abdomen, but at the same time the sixth pair of
ganglia also is carried dorsalwards and above the fifth pair, from
which the connecting nerve fibers pass upwards to the sixth. At
the present time it is impossible to determine whether this is due
to a mechanical adjustment between the shifted abdomen and the
last pair of thoracic ganglia, or whether the sixth pair, belonging
properly to the abdomen, has taken part in the general shifting
of that structure.1

There are certain objections against this comparison which I
have attempted, and if they do not directly oppose yet do not sup-
port the hypothesis. First and most important is the uncertainty
of brain invaginations in Pycnogonids. These seem to be present
in all the Arachnids, and easily seen in the development of the
embryo. In the Pycnogonids I have not been able to find such
invaginations. We have seen in Fig. 19 that the groove might
possibly bear such an interpretation ; but even if this were true,
we would expect to find a much more pronounced involution, but
this does not seem to be the case. Again, the ventral organs,
which havebeen compared with those of Peripatus,lend nosupport
to the hypothesis. It is possible that the Pycnogonids have come
from the Arachnids, at a time when these latter have had such
organs in common with the ancestors of the Insects, and that
they have been fully retained in the Pycnogonids. Lastly, the
openings of the reproductive organs. Typically the ovaries and
testes of the Sea-Spiders are a pair of organs united posteriorly by
a cross commissure. They extend into the walking legs and
open on the second joint of these. There are many exceptions
to this, but they are regarded as secondary. The openings on
the legs have no homologue in the Arachnids, except perhaps in
Limulus, nor does it furnish any ground for comparison with
other Arthropods. But if we assume the group to have come
directly from the Annelids, we have no better ground here for a
comparison. We are greatly in need of observations on the
development of the sexual openings, and until we get such, the

1 If we assume with Hatchek a common descent for all Arthropods, and that
in the Insects we have several of the anterior segments about the mouth, sup-
pressed, we might assume that the Spiders have lost a third pair of appendages,
and the Pycnogonids retained it, and in this way bring into line the other
appendages of the groups.

THE PYCNOGONIDS. 33

question must remain an open one. It seems not improbable,
however, that the openings may be secondary and connected in
some way with the secondary presence of reproductive organs
in the appendages.

I hope to have shown that these three objections are of
negative value, at any rate so long as the present uncertainty
surrounds them, and that we have sufficient grounds for a com-
parison in the early stages of development and in some of the
important adult structures of the two groups.

The Pantopod-Larva.

There is a general resemblance between the Nauplius of the
Crustacea and the larva of the Pycnogonids, but the differences
become greater and greater the more closely we examine the
two forms. In each the body contains three pairs of append-
ages, but those in the Nauplius show biramous structures, while
none of the Pantopod-larval appendages show such structures.
Moreover, the first pair of appendages of the Pantopod-larva is
chelate and innervated from the brain. Other characteristics of
the larva are the well-marked proboscis with its triangular oeso-
phagus, the mid-gut sending out prolongations into some of the
appendages, and the absence of any anal opening.

Have we then any basis for the assumption that this Pantopod-
larva is a modified Trochophore of the Annelid ancestors % The
problem is very similar to that of a supposed relationship between
the Nauplius and the Trochophore, and in both cases an answer
exceedingly difficult to give. Prof. Dohrn believes that the
Pantopod-larva is to be regarded as a Trochophore with the
Pycnogonid characteristics reflected back upon it. Prof. Lang
believes in a similar process for the Nauplius. Without discus-
sing the latter, let us confine our attention to the possibilities of
the Pantopod-larva.

What characters have been "reflected" on Dohrn's theory, and
what remains after these are removed ? Prof. Dohrn has given
us the answer. "... The absence of an anal opening, the pincers
of the first pair of appendages, the long claws with the accessory
spine, the structure of the proboscis with its triturating appa-
ratus and its ganglia, the form of the tegumentary glands and
their characteristic hairs, can only be considered as having been

34 T. H. MORGAN,

acquired in a much later stage and transmitted to larval life."
That is to say, that almost every characteristic of the larva
has been handed back from the adult! And what remains?
" Nothing but the nervous system ... an intestine, three pairs
of appendages and two eyes." " But these are the attributes
which one finds equally in the larva of Annelids !"

There is one vital fact which has been left out of account, viz.
the presence of an anus in the Trochophore. How can we account
for its absence on the above hypothesis when we know the
Pantopod-larva to be a free-living larval form? And unless
some special reason for such a loss can be imagined, the very
basis of the comparison between Pantopod-larva and Trochophore
is gone! What we have done in the above process of subtrac-
tion is to have removed the most striking structures of the adult
from the larva and have left, not a Trochophore, but only the
framework of the Pycnogonid.

For two main reasons I am unable to believe in the phylogeny
given by Dohrn or by Hoek. First, because it seems to me there
are facts derived from the early stages of development which
point unmistakably to a relationship between Pycnogonids and
Arachnids; and in the second place I cannot believe any actual
homology to exist between the Fantopod-larva and the Trocho-
phore, nor any fair reasons to assume that the characteristics of
the Fycnogonids have been reflected upon a Trochophore.

If, then, we start with the assumption that there is a relationship
between the Sea-Spiders and Arachnids, we may examine into
the meaning of the larval form as a corollary to such a position.

I have stated that the Fycnogonids are degenerate, and prob-
ably not derivable directly from any existing group. If so, how
far down in the ancestral tree of the Arachnids have they arisen ?
The very great differences in the adult structure of the groups
indicate no very recent origin, but possibly they came in at a
time when the Arachnids had the first pair of appendages chelate,
and these were innervated from the supra-cesophageal ganglia, and
had coeca from the digestive tract into most or all of the append-
ages. After the divergence of the Pycnogonids as a group from
the general phylum of the Arachnids, the Pantopod-larva may
have developed.

The Pycnogonids have adapted themselves to a very special

THE PYCNOQONIDS. 35

habitat, and in general, what we may suppose to have happened,
was a decrease in the size and an increase in the number of eggs,
with the resulting early development of the free larva that we
find to-day. This larva represents the more anterior segments of
the adult, viz. that part containing the proboscis, the oesophagus,
the eyes, the chelae, and two post-oral appendages. Behind this
is an undeveloped part, which slowly grows in length as the
animal increases in size. The anus belongs to the last segment,
and does not appear until that segment is wholly or in part
developed.

If the Arachnids have come from Annelid ancestors with
many segments, we have a clue to the slight resemblance between
the Pantopod and the Trochophore. The former represents the
most anterior segments of the adult Sea-Spiders, and therefore
to some extent the anterior segments of the Annelids or of the
Trochophore. But at no time in the ontogeny of the Pycnogo-
nids have Trochophore and Pantopod-larva been transformed
the one into the other, as Dohrn believes.

Such seems to me the more probable view of the meaning of
the Pantopod-larva. This belief has grown out of my work on
the embryology of the group, and whether future work supports
or disproves such an hypothesis, it is hoped it may be useful, if
only as furnishing another point of view for looking at thephylo-
geny of the Pycnogonids, or may lead to a more complete study
of the embryology of the group.

36 T. H. MORGAN.

PART IL— THE METAMORPHOSIS OF
TANYSTYLUM.

The material for the second and third parts of the present
account was collected during the summer of 1890, while at the
Marine Biological Laboratory at Wood's Holl, Mass. I wish to
express my gratitude to the Director, Prof. C. O. Whitman, and
to the trustees of the laboratory, for the opportunity to carry on
investigations under the favorable conditions offered at the
marine station.

My object in giving the figures and description of the meta-
morphosis of Tanystylum is, in the first place, because I was
fortunate enough to obtain probably a complete series of the
larval stages of this Pycnogonid, and inasmuch as the remark-
able condition of the development, and adult condition of the
appendages of the Pycnogonids, offer in themselves a most interest-
ing field for study, I hope a contribution to the complete metamor-
phosis of one form may be found useful. Again, by means of
serial sections of each of these stages, I hope to have added a
little to our knowledge of the internal changes taking place
during development, as heretofore we have had only such knowl-
edge as could be gained by surface views. Lastly, the figures
give a series to which intelligible reference may be made when
speaking in Part III of the Development of the Eye.

ISTot much difficulty was experienced in getting excellent sec-
tions of the embryo during its metamorphosis. As described in
Part I, the embryos were killed in alcoholic picro-sulphuric
solution (35 per cent alcohol saturated with picric acid, to which
the usual amount of sulphuric acid was added). Before staining,
the appendages were cut off (only possible, however, in the later
stages) and the body stained overnight in hasmatoxylin ; after-
wards embedded in paraffine and cut into sections, which are
better if cut so thick as 10 to 15 p. The figures on Plate V
were all drawn to scale, showing the enormous increase of growth
during larval life. Figs. XXV and XXVI of Plate VI were
also drawn to the same scale.

I cannot be sure that the series of figures drawn on Plate V
and Figs. XIX, XX, XXI on Plate VI give the perfectly com-

THE PYCNOQONIDS. 3?

plete metamorphosis, although I believe the set must be almost
filled out especially for the later stages. Here I could procure
great numbers of individuals, and, comparing them and finding
no intermediate stages, I inferred none such existed. But in the
very young stages, before the appearance of the first pair of
walking legs, it is extremely difficult to know how complete the
series is ; but from the slight gradations in size and structures in
those figured, I think we shall not be far wrong in assuming an
almost or entirely complete series.

In Part I of the present paper I have figured and described
at length the structures of the Pantopod-larva of Tanystylum at
the time when it leaves the parent. I shall speak of an embryo
of this age as stage 1 (see Plate IY, Fig. IX). In examining the
hydroids for Sea-Spiders, I found one embryo a little larger in
size than this first stage, and having other slight differences which
plainly show it to be a stage in advance of the last, and probably
a stage immediately following stage 1, or after the first larval
skin is shed. Figures of this stage (2) are seen in X and XI of
Plate Y, showing the ventral and dorsal views respectively.
The embryo has much the same shape as stage 1, but there is
seen an important difference between the two stages. The older
larva has the posterior part of its body filled with a solid lobed
mass of cells, which contrasts strongly with the clear posterior part
of stage 1. Further, the second pair (really the third, as the first
is double) of ventral ganglia are much larger. The posterior
pair of lobes of the mid-gut project to the sides of the embryo,
and an examination of the periphery of the posterior border
shows a slight indentation on each side, indicating the beginning
of another pair of appendages. The first pair of appendages,
the mandibles, are three-jointed, and have, projecting forward
from the first, proximal joint, a strong spine, as in stage 1. The
second and third appendages are each two-jointed, with a strong
terminal, movable spine. The proboscis lies in the mid-ventral
line, and projects forward between and ventral to the first pair
of appendages. Fig. XI shows the dorsal view of the same
embryo. From the figure we see that the digestive tract of the
embryo is on each side three-lobed. The first pair of lobes pro-
jects into the first pair of appendages, the second pair of pouches
towards the base of the third pair of appendages, and the third

38 T. H. MORGAN.

pair of pouches projects into that portion of the posterior part
of the embryo from which the next pair of appendages arises. In
the anterior dorsal part of the larva are seen the first traces of a
pair of eyes ; the first pair of appendages is bent towards the
ventral surface.

This embryo was stained and cut, and from the series of sec-
tions the following facts obtained : There is a small brain
connected by means of a pair of commissures with the anterior
ventral ganglion. This first ventral ganglion is, as we saw in
the earlier larva (Part I), really a pair of fused ganglia. Behind
this lies the second ganglion (the third of the series). The
oesophagus, running back from the mouth through the proboscis,
opens into the mid-gut. The latter has a pair of prolongations
into the base of the first pair of appendages, and another pair of
pouches run to the base of the third pair of appendages. The
posterior part of the larva contains a backward continuation of
the digestive tract, which seems still to end blindly behind. The
cells lining the wall of the gut are now quite different from those
observed in stage 1, which were seen in Fig. 13, Plate 1. In
the present stage (2) they are much larger, are filled with large
round masses of yolk-like substance, so that the lumen of the gut
is almost obliterated. The yolk-like substance must have come
from the exterior, and indicates that the animal is feeding at this
stage and, perhaps, is supplied with abundance of food-material.
No solid extraneous matter was ever seen in the tract of either
larvae or adults of Sea-Spiders, so that the food is probably
obtained by sucking the juices of other animals — another simi-
larity to Arachnids ! — and probably from the hydroids, amongst
which they live. The series of sections did not show any traces
of a heart. A single pair of eyes was seen over the brain, and
to the sides of these a thickened mass of ectodermal cells.

Passing to what seems to be the next stage of development, we
have an embryo, shown by Fig. XII, Plate V, stage 3. There
is quite an increase in size, especially marked in the length of the
embryo. The first pair of appendages and the proboscis are
longer and larger than in the last. The second and third
pairs are as in the preceding stage ; but the embryo now shows
clearly external indications of a fourth pair of appendages.
The preserved larva was quite opaque, and nothing of the

THE PYCNOGONIDS. 39

internal organs could be made out from the surface, so that sec-
tions were again resorted to. Here it was found that the mid-
gut gave off the usual diverticula to the first and third pairs of
appendages and, besides, quite a large pair of pouches into the
undeveloped fourth pair of appendages. The sections show that
the gut ends blindly behind, and no trace of proctodeum was
found. The lumen of the digestive tract is almost entirely
obliterated by the large size of the cells, and these are completely
filled with spherules of yolk. Large, active ectoblast and meso-
blast cells, together with the mid-gut, fill up the posterior part
of the embryo. The fourth pair of appendages is quite far
advanced, as the ectoblast, by pushing in from behind, has par-
titioned off on each side a part of the posterior part of the body.
The ingrowth of ectoderm on each side is always two-layered, fur-
nishing a new wall to the outer sides of the body and to the
inner side of each new leg. The first (double) pair of ventral
ganglia is followed by a well-developed second pair, each con-
taining large cavities, and these are followed by a smaller fourth
pair of ganglia. There is a single pair of eyes.

The next embryo which I have found shows a considerable
increase in size, and I cannot be certain that it is the stage fol-
lowing the last. Still it is probable, as I have found five or six
of this size and none in between this and the last. The larva is
shown by Figs. XIII and XIY giving the ventral and dorsal
views of this stage (4th). The most marked external change is
in the great elongation of the body posterior to the third pair
of appendages, and in this region we can distinguish a central
middle lobe and a pair of lobes on either side of this closely
pressed against the middle part, the side lobes being the ' anlage'
of the fourth pair of appendages. Three pairs of ventral ganglia
were made out from surface views, but sections show also a small
forming fourth pair. The embryo shown by Fig. XIV is the
same, but from dorsal view. It shows a slight constriction in
the middle of the body at the base of the fourth pair of append-
ages, and the continuation of the body posteriorly between these
appendages is seen to be broader than the same region on the
ventral surface; and further the base of the middle posterior
lobe suddenly widens out, as seen in the figure. A single pair
of eyes is shown on the anterior dorsal surface. Sections of this

40 T. H. MORGAN.

embryo give these points : The third pair of ganglia still have
large cavities and a small fourth pair is in process of formation.
The mid-gut sends the usual diverticula to the first and third
pairs of appendages, also quite a large pair of pouches to the
fourth appendages, and lastly a short pair, in the region of the
third ganglia, runs out at right angles to the gut. These last
pouches pass into that part of the embryo corresponding to the
sudden widening at the base of the middle posterior lobe. Sec-
tions show that at this place the ectoblast pushes into the body
cavity from behind forwards, forming partitions which are the
beginnings of a new pair — the fifth pair — of appendages. Pos-
teriorly the mid-gut seems to be continuous with a proctodeum
which, I think, forms at this stage. It also seems probable that
the heart forms at this stage of development, but the sections I
have do not speak definitely on this point. I could not discover
any trace, as yet, of reproductive organs.

We come now to a stage which, I suppose on account of its
larger size, is easily found and not at all uncommon. I think it
is undoubtedly the stage formed immediately after the moulting
of stage 4. The small undeveloped fourth legs of stage 4 have
now, in stage 5, Fig. XY, spread out laterally after the shedding
of the cuticle, and have become segmented. The whole larva is
longer than in the last stage, and all the external organs are
also comparatively larger — the appendages, the proboscis, and
especially the posterior part of the body. The most marked
change is in the addition of the large fourth pair of appendages
— the first pair of walking legs of the adult. Each of these is at
this stage composed of six segments, of which the first three are
somewhat smaller than the following four distal segments. The
fourth segment of each limb, counting outwards from the body,
shows internally traces of division into two parts, although the
chitin covering the ectoblast is not divided. The spines pro-
jecting from the three distal segments of the legs must assist
greatly in clinging to the hydroids, and particularly the long
terminal hook with its two side hooks. Five pairs of ventral
ganglia are seen from the surface — the first, as we know, being
composed of two pairs. Other differences are easily gathered
from the figure. Sections show that the digestive tract opens
posteriorly to the exterior by means of a proctodeum. The first

THE PYCNOGONIDS. 41

pair of diverticula from the gut runs forward into the limb as far
as the glands of the mandibles, which are still retained.

Where these pouches join the digestive tract a second pair of
diverticula originate and passes to the base of the third pair of
appendages. A large pair of diverticula run out into the penul-
timate segment of the fourth pair of appendages, and behind
these is a pair of pouches, the fourth, into the fifth pair of appen-
dages, and, lastly, traces of a fifth pair of pouches is found.
In Plate VIII, Fig. 52, there is a cross-section through the pos-
terior part of this larva. It passes through the fifth pair of
appendages, which are still at the sides of the body, and below
the section passes through a pair of small forming ganglia. This
section shows the thick ectoblast, the scattered mesoblast cells,
especially abundant in the legs, and the large endodermal cells
filled with food particles. On each side of the ventral ganglia
are two other invaginations of the ectoderm. These push in
both above (not seen in this section) and below and from behind
— forwards as well, and form a partition wall in the body cavity,
which cuts off a pair of appendages. The invagination for the
ventral ganglia, shown in the figure on each side of the ventral
median line, are the ingrowths for the fifth and last pair of body
ganglia. There is a distinct proctodeum opening behind. In a
single section of this series I found a well-marked 'anlage' of
the sexual organs. It was on the dorsal surface of the digestive
tract, in the middle line, and in that part of the body corres-
ponding to the base of the first pair of walking legs. It was
thickest in the middle, and tapered away towards the sides of the
larva, but was confined entirely to the body. Dorsal to this
mass of cells of the reproductive organ was what seemed to be an
accumulation of blood corpuscles, and I fancied I could make
out traces of a large heart cavity. In another embryo of the
same age as the last I could find no traces of reproductive
organs. There are four eyes on the dorsal surface, and the first
pair are nearer together than the posterior pair, and on each side
of the first pair there is a thickened mass of ectoblast cells. We
saw there were five pairs of ventral ganglia, and it is worthy of
note that the embryo with a single pair of walking legs has formed
the ' anlage ' of all the other ventral ganglia to the appendages.

In stage 6, represented by Fig. XVI, we find a second pair

42 T.H.MORGAN.

of walking legs, and, moreover, the embryo has increased a great
deal in size. The first three pairs of appendages are essen-
tially as in the last stage. The fourth pair has now acquired its
adult number of segments, viz. eight. The two new segments
which have been added were interposed, the one between the
fifth and sixth (distal) of the preceding stage, and the other in
the middle of the fourth pair, in which we saw faint internal
traces of division in the last stage. The newly acquired fifth pair
has now six segments, three smaller proximal ones and three
distal.

In the posterior part of the animal the undeveloped sixth pair
of appendages appears on each side of the body, and in between
the bases of these appendages appear the small 4 anlage ' of a
seventh pair. Five pairs of ventral ganglia are seen, the pos-
terior larger than in the last stage. The posterior part of the
abdomen lies at this stage still in the same plane as the rest of
the body. The diverticula of the digestive tract are in the first
appendages as in the last stage, the glands in the mandibles are
still present, and the fifth pair of ventral ganglia still retain their
invagination cavities.

The * anlage' of the reproductive organs appears between the
first and second pairs of walking legs in the mid-dorsal wall of
the digestive tract, and is larger than in the last stage. In the
anterior part of the body there seems to be formed a large heart
cavity.

After the next moult {stage 7), Fig. XVII, we see an important
change has taken place in the appendages. The third pair has
almost completely disappeared, and only two small stumps indi-
cate the point at which their appendage last appeared. The
first and second pairs of appendages are unchanged in structure.
I found one embryo of this age (7) having still the third pair of
appendages, and for some reason the larva was quite appreciably
larger than that given for this stage. Another pair of walking
legs has been acquired — the third — and each of these has six
segments.

The second pair of walking legs (the fifth pair of appendages)
has now acquired its complete number of segments, eight, while
the first pair retains its previous number of eight. The new
segments which have appeared in the third pair of walking legs

THE PYCNOGONIDS. 43

were interpolated at similar places to those of the preceding leg,
viz. in the middle of the fourth and in between the fifth and
sixth. There is an undeveloped pair of appendages at the pos-
terior part of the larva, and the abdominal part of the body,
instead of lying between them, as previously, has now moved
dorsalwards, so that it projects from the posterior-dorsal part of
the embryo. The first and second pairs of ventral ganglia (the
first and second and the third in a true count) are in close con-
tact or partially fused. The reproductive organ has much
increased in size. In the region of the first pair of walking legs
it extends through a few sections as two entirely separated parts,
lying on each side of the middle line. In the region of the
second pair of walking legs the two lateral parts unite into a
single median structure which is almost as broad as the body,
and the organ ends soon after this as a median, thickening on
the mid-dorsal surface of the digestive tract. The glands of the
mandibles have almost entirely disappeared.

At the next stage of development we have the formation of
the seventh and last pair of appendages {stage 8), Fig. XVIII.
There is a great increase in size, and an important change takes
place in the second appendage. These lose their early larval
structure and probably their first function, and now enter
upon a new role. They are three-jointed and have lost the large
terminal spine. The tip of the appendages is covered with many
smaller spines, and sometimes I have seen one larger than the
others, which presumably represents the large terminal claw of
the preceding stage. The appendage is now three-jointed, having
acquired an additional segment.

The first pair of appendages do not differ from the preceding
stages. The third pair is. still in an undeveloped condition,
although somewhat larger than in the last stage. The fourth
and fifth and sixth pairs have each eight segments, but the
seventh pair, which has just been acquired, has only six segments.
All of the walking legs have increased greatly in length, and
each contains a long diverticulum from the digestive tract. Five
pairs of ventral ganglia are seen, although the first and second
are almost completely fused.

Sections show that the mid-gut sends out diverticula to the
first, fourth, fifth, sixth, and seventh pairs of appendages and a

44 T. H. MORGAN.

pouch to the base of the rudimentary third pair. This pair of
appendages is filled with embryonic mesoblast, and the ectoblast
covering the appendage is thrown into a series of crenations
beneath the cuticle, so that it would be much longer if extended,
and is in condition to expand rapidly at the moulting period.
This folding of the walls of the undeveloped appendage was
seen also in all of the pairs of walking legs at the stage pre-
ceding their extension, and makes intelligible the great increase
in length of the appendage when the next moult takes place. At
the point of union of the gut-diverticulse of the seventh appendage
to the central digestive tube there is formed in the angle a small
posterior pair of ventral ganglia. These were noc seen in sur-
face views, but lie above and behind the last pair of ganglia seen
in the figure (XVIII), and belong to the rudimentary abdomen.

The ' anlage' of the reproductive organs is divided anteriorly
into two parts, lying on each side of the dorsal side of the mid-
gut, and run as far forwards as the base of the first pair of
walking legs. Behind these divisions are united into a single
mass in the region of the second pair of walking legs, and dis-
appear at the anterior part of the third pair. There is a well-
developed heart lying dorsal to these reproductive cells.

In the next stage we have a change which causes the larva to
approximate much more nearly to the adult structure. The most
important part of the change consists in the almost total loss of
the first pair of appendages. See Figs. XIX and XX, Plate VI.
Looking at the animal from the ventral surface we see no trace
of this pair of appendages. Turning the animal over (XX) we
find two projections at the anterior dorsal border where the
first pair formerly arose. Each projection has two segments. A
change has also taken place in the .second pair, for these have
now doubled the number of segments, and, except in size, have
the adult structure. The third pair of appendages (XIX) has
increased greatly in size, and now these form a pair of prominent
projections on each side at the base of the proboscis. Each is
bent almost at right angles back onto itself, and lies closely
applied to the body in the space between the proboscis and the
first pair of walking legs. Externally they do not show divisions
into segments. The only point to note in the walking legs is
that the fourth pair have now eight segments. The first three

THE PYCNOGOMDS. 45

ganglia have completely telescoped, and this compound ganglia,
it will be remembered, innervates the second, third and fourth
pairs of appendages, as well as the proboscis. On the dorsal
surface (XX) is to be seen the two pairs of eyes equally dividing
the periphery of the cupola-like projection of the dorsal surface.
Sections do not give any additional facts worthy of note.

The last stage of larval life, or rather the last stages showing
external traces of immaturity, is shown by Fig. XXI, stage 10,
and although after this the larva undoubtedly undergoes several
moults, accompanied by increase in size before becoming sexually
mature, it did not seem necessary to figure such intermediate
stages. The only change which has taken place in this stage
XXI is in the appearance of segmentation of the third pair of
appendages, and the number of the segments of this pair is that
of the adult. The second pair of appendages has also increased
in length, reaching to the distal end of the proboscis. All of the
walking legs have the adult number of segments.

Having followed the larva through all these changes of growth,
it may not be out of place to say a few words about the adult
structures. A figure of the adult Tanystylum from the dorsal
surface is shown by Fig. XXII. The compactness of the body of
this species is exceptional and has been carried to such an extent
that the bases of the walking legs are almost in contact, so that
they radiate in all directions (except anteriorly) as from a cen-
tral point. In most other Pycnogonids there is not so great a
concentration of the body, and these undoubtedly come nearer to
the ancestral form, and, moreover, in some of these simpler forms
the rudimentary abdomen lies in the same plane as the body and
behind the attachment of the posterior pair of legs. The figure
for the anterior region of Phoxichilidium (Fig. XXI Y) shows a
form in which there is no such concentration. In Tanvstylum,
as seen in Fig. XXII, the abdomen has shifted far up on the back
of the animal, and is separated by only a short space from the base
of the prominence bearing the eyes. The adult still retains the
rudiments of the first pair of appendages. These have a large
basal segment and a smaller distal one covered with spines.
Both males and females develop the ovigerous legs, and the
males carry during the breeding season from one to three bunches
of eggs on each of these appendages, as shown in Fig. XXIII.

46 T. H. MORGAN.

(This is not drawn to the same scale as* XIV.) The bunches on
each leg are entirely separated from each other, while in Phox-
ichilidium (see Fig. XXIV) the bunches are larger than in the
last (but with the eggs individually smaller), and are stuck to
both of the ovigerous legs, which are thus bound firmly together
by the sticky masses of eggs.

Before leaving the metamorphosis of Tanystylum, I wish to
refer briefly to the development of the ventral ganglia. I had
previously shown in Pallene that each ventral ganglion was
formed by an invagination of the ectoblast, and I was led to
infer a similar formation in Tanystylum, and by inference for all
of the Pycnogonids. A study of the metamorphosis of Tany-
stylum has verified this beyond all doubt for that form, and I
think it will be legitimate to extend this to the whole group.

The earliest stages of the development of the ventral organs in
Tanystylum are shown in Fig. 52, Plate YIII, where the thickened
ectoblast pushes in at two points on each side of the mid-ventral
line. The nuclei of the cells of the invagination at this stage are
scarcely, if at all, larger than those in the rest of the ventral
ectoblast. A later stage in the formation of these ventral organs
is shown by Fig. 54. It is taken from the third pair of ganglia. In
this the cavities of invagination have closed, and the ganglion is
now cut off from the ventral ectoblast. There are many
ganglion-cells or nuclei forming the ganglion, and the nuclei in
the cells bounding the cavities are larger, and often er show karyo-
kinetic figures than do the other nuclei of the ganglion. The last
stage of development is shown by Fig. 53. Here the central
cavity is almost obliterated on the right side, but on the left side
the section cuts a little to one side of the central cavity, and only
the cells of the wall appear, and two of these are seen in process
of karyokinetic division. It is to be noted that the nuclei of the
ventral organs at this stage are seen to approximate in size to
those nuclei forming the substance of the ganglion, but stain
more deeply. If we examine a little later stage, the central cav-
ities of the ganglia have disappeared completely, the nuclei are
all of the same size, and the only indication of the ventral organs
is in the deeper staining of the cluster of nuclei at the spot in
each ganglion where the cavity was last seen. Ultimately this
difference also disappears.

THE PYCNOGONIDS. 47

In the first part of this paper I was rather inclined to believe
that the ventral organs had a phylogenetic significance, but still
I must believe their interpretation to be an open question. The
large size of the nuclei of these organs during their time of
greatest activity in forming, and the subsequent merging of these
nuclei (and cells) into ganglion nuclei (and cells), suggest that there
may be no ancestral meaning to the process, but that there may
be larval organs — something like neuroblasts — for the formation
of ganglia. And it has been suggested to me that these may fall
into the same category as the neuroblasts of Annelids, and may
present, by analogy, an intermediate step in the formation of
such cell-masses or teloblasts.

Another interesting problem might be discussed if there were
space, viz. the appearance and subsequent loss of the third pair
of appendages ; and, again, the curious law discovered in the
sequence of the walking legs, and its relation to the number of
segments. When a pair appears for the first time, the number
of segments is six ; at the next moult the number increases to
eight, by the interpolation of one and the division of one into
two, and this method of increase is common to all of these
appendages.

In thinking, too, over the remarkable series formed by the
different species of Sea-Spiders, where, as Dohrn has shown,
we can proceed from the simplest to the most complex, finding
amongst living forms all intermediate stages, one finds a most
suggestive field for speculation. Prof. E. B. Wilson has pointed
out to me the great perfection of this series, and I have since
been interested to find several other parallel cases in semi-
parasitic groups of animals, as, for instance, in the Copepods,
in the Cuninas, in the Barnacles and in the Sponges. It would
be interesting to discover the correlation, if any, between the
persistence of a perfect series of animals of a group and semi-
parasitism.

[There are two other figures, XXV and XXYI, which I have
added to Plate VI, but these are not directly connected with the
second part of this paper; nevertheless, as I only obtained them
after the figures to the first part were printed, I have determined
to add them now for the sake of completeness. The first figure
is of the larva of Phoxichilidium at a time when it leaves the

48 T. H. MORGAN.

parent. This larval form shows certain differences from that of
Tanystylum. In it the second and third pairs of legs are much
longer, as the distal spine is prolonged out into a long thread-
like continuation of the limb. In Tanystylum these spines were
thick and short and adapted for crawling amongst the hydroids,
while the long limbs of the larva of Phoxichilidium could hardly
be used in this way, and suggest rather that they are used for
catching to any objects they may come in contact with. The
first pair of appendages are thicker in Phoxichilidium, with
smaller claws, and the proboscis, instead of projecting forward
between the first appendages, is only raised slightly above the
level of the body at about its middle point. I have not
obtained later stages of this larva, but the accounts of other
species of this genus show that it becomes parasitic either inside
the digestive tract of a hydroid or buried in its substance, and
here it undergoes its metamorphosis until the adult structures
are almost reached.

The other figure (XXYI) is of a young Pallene after it has
left the parent, and this individual was found living freely
amongst the Pennaria. The figure shows a ventral view, and
the stage corresponds to one following that shown by Figs. VII
and VIII of Plate IV. It shows the third pair of appendages
beginning to develop, which, in Figs. YII and VIII of Plate
IV, only appeared as two slight projections from the sides of
the body. In the present case these appendages are folded
across the breast — if the term be applicable to Pycnogonids —
closely adhering to it in the region between the base of the pro-
boscis and the first pair of walking legs. The embryo has also
fully developed its fourth pair of walking legs, which, it will be
remembered, were in an undeveloped state in the previous stage.]

THE PYCNOGONIDS. 49

PART III.— THE STRUCTURE AND DEVELOPMENT
OF THE EYES OF PYCNOGONIDS.

In reaching a decision as to the systematic position of the
Pycnogonids, I hoped to test the conclusions I had reached from
a study of the embryology by taking another point of view.
Fortunately a most interesting field opened in a consideration of
the structure and development of the eyes. While enjoying the
opportunities of the Marine Biological Laboratory I was enabled
to procure and study an abundance of fresh material. The
advantages of using such material is now so well known as to
need only mention, and I am certain that I could not have
obtained nearly so satisfactory results from preserved animals.

The results obtained from a study of the eyes have added, I
believe, very much to the solution of the problem I had before
me. Hoek and Dohrn give the only figures of the Pycnogonid
eyes that I have found, but the results are confessedly incom-
plete, due in the one case to poor preservation of material, and
in the other to the difficulties of technique. Neither of these
accounts will be of any service to us in our study of the eyes,
and may be dismissed here. The subject will be treated in the
following order :

1. The Anatomy of the Adult Eyes.

2. The Histology of the Adult Eyes.

3. The Development of the Eyes.

4. Comparison with the Arachnid Simple Eyes.

5. Inversion of the Pycnogonid Eye.

6. General Considerations of the Arachnid Eye.

1. The Anatomy of the Adult Eyes.

The Sea-Spiders have in all four simple eyes, situated in the
middle of the first (compound) segment of the body, on a raised,
cupola-like projection. In some of the species the eyes are cer-
tainly degenerate, but in none are all traces of the eyes lost.
The position of the eyes on the body can be best shown by
reference to the figures. In Fig. XXII we see the dorsal surface
of Tanystylum, and on this, in the so-called first segment, raised

50 T. H. MORGAN.

on a dorsal projection, lie the four eyes equally dividing the
circumference. From the exterior each eye is seen surrounded
by pigment, generally of a yellowish-black color. The eyes
are extremely small and covered by thick chitin, rendering it
exceedingly difficult to get at the eyes either for sections or for
maceration.

Fig. XX IV shows a side view of the anterior end of Phoxichili-
dium. It shows the shape of the cupola-like projection bearing
the eyes, two of which are seen on side view. The eyes of this
species are much larger than the eyes of either of the other two
Sea-Spiders, and for this reason were less difficult to manage, and
the following account will apply to the eyes of Phoxichilidium,
unless otherwise stated. "Alcoholic picro-sulphuric" solution
gave by far the most satisfactory results. The body was cut
across near to the eyes, and the portion containing the eyes fell
immediately into the killing fluid. After thoroughly washing
out the picric. acid the eyes were stained in hematoxylin and
afterwards cut in paraffine.

We may start with a figure showing the whole eye removed
from the body, as seen in Fig. 30, Plate VII. The figure is of an
eye which had been sufficiently macerated to free it from the
surrounding tissue, removed in glycerine to a slide, and rolled
around until the innermost part of the eye was turned upwards.
The figure therefore shows the inner surface of the eye — that
part nearest to the center of the cupola.

The pigment was, in handling, to some extent removed from
this part of the surface, so that by focusing beneath this another
layer of the eye could be seen. The figure shows the distribution
of the optic nerve to the eye, which comes directly from the
brain beneath the eye. The most important part of this figure
is the arrangement of the parts of the eye on each side of a
median line — the raphe. This runs through almost the whole
inner length of the eye as a somewhat irregular line from above
downwards, and is formed by the elements of the visual part of
the eye coming in contact from the two sides. Indeed, in this
figure almost seventy-five pairs of these elements are seen
arranged along each side of the raphe.

The optic nerve, before it reaches the eye, breaks up into
several branches, and the branches enter the eye aloug its inner

THE PYCNOGONIDS. 51

surface and along a line corresponding approximately to that
of the raphe. The nerve branches disappear in the pigment
covering the inner surface of the eye, and cannot be followed
farther.

If we make a section in a horizontal plane through the cupola,
so that all of the eyes are cut transversely to their longest diam-
eter which runs from above downwards, we obtain a figure
shown in 31. The four eyes are about equally distant from each
other. In front of each the chitin is greatly thickened to form
a lens ; beneath this the corneal hypodermis is very much thick-
ened ; under the corneal hypodermis is a cup-shaped mass com-
posed of two layers — 1st, a light layer containing nuclei, and 2d,
a layer of pigment surrounding the first. The entrance of the
nerve is shown in some of the eyes at their inner, middle part.
If we examine one of these eyes under a higher power of the
microscope, we obtain such a figure as is shown in Fig. 32. The
chitinous lens is well shown here, and the great increase of its
thickness over the eye is very marked. The ectoderm of the
body, the so-called hypodermis, also undergoes a great change in
the region of the eye. The cells are much longer, have con-
spicuously large nuclei, and tapering away at their inner ends
into fine processes which run out to the two sides of the eye.
The whole arrangement of these corneal-hypodermal cells —
lentigen cells — is markedly bilateral, as seen in Fig. 32.

Immediately under the lentigen cells lies the middle layer of
the eye. This is the retinal part of the eye, or that part sensi-
tive to light. Immediately under the corneal hypodermis this
middle layer of the eye shows a meshwork of fiber-like pro-
cesses, which are probably interlacing and anastomosing nerve
fibrils. Passing inwardly, there next follows a zone containing
conspicuous nuclei and large vacuolated spaces. There next
comes a zone which contains slightly stainable, rounded bodies
slightly smaller than the nuclei. Lastly, in the bottom of this
cup-shaped middle layer is a portion of the middle layer which
stains more deeply than the rest of the middle layer, and
radiating through this from the bottom of the cup outwards
are several deeply stained (drawn black in the figure) rod-like
bodies or bacilli. Corresponding somewhat to the intervals
between the bacilli, the middle layer is indistinctly split up into

52 T. H. MORGAN.

bands running from without in, through what has been spoken of
as the zones. These zones of the middle layer are not of course
separated from each other, but pass insensibly into each other,
and are only of use in describing the structure of the middle layer.

The nuclei found in the middle layer lie largely to the right and
the left of the middle line of the eye, and through quite a wider
zone of this layer. Thus they show to a large extent a bilateral
arrangement. The inner layer is surrounded on its entire inner
surface by a pigmented layer — the inner and pigmented layer of
the eye. This layer is composed of pigment granules, both black
and yellow, and it is impossible in such sections to resolve the
layer into its constituent cells. The entrance of a part of the
optic nerve is seen in the figure. It disappears in the pigment
layer in the inner middle line, and cannot be traced farther, in
sections, than the pigment layer.

Fig. 33 is a drawing of a longitudinal section through the
same eye. It is therefore at right angles to the last figure and
along the line of the raphe. This section, however, is not
median through all its length, but at the outer part passes to one
side of this line; at the inner part, however, it cuts approxi-
mately the middle line. But this slight obliquity is really advan-
tageous to an understanding of the eye, as it passes more in the
plane which the principal retinal elements follow. The three
layers of the eye are clearly seen, although the distinction
between the outer and middle layers is not so definitely marked
as we saw in the cross-section, and is due here to the oblique
plane of the section. Under the corneal hypodermis lies the
large middle layer, in which we see the same structures as shown
in the cross-section of the eye. The nuclei are conspicuous and
numerous, and much more abundant here than in a section
passing through the exact middle line. There are large vacuoles
lying amongst the nuclei, but I have been unable to determine
whether these are artefacts or natural to the eye. But I can
scarcely believe them to be artificial products, as they seem to
be constantly found on this part of the eye, and show no evidence
of being due to tearing of the sections.

The layer of rounded bodies is not shown in the figure. The
form of the retinal elements is better seen in this section than in
the cross-section. At the inner part of the middle layer is a

THE PYCNOGONIDS. 53

series of deeply stained rods — the bacilli, which occur regularly
from the top to the bottom of the eye, and it was these which
we saw in Fig. 30 arranged along each side of the middle line.
These bacilli are continuous with the cell-walls of the retinal
cells, which pass outwardly from the bacilli as far as the zone of
nuclei. The third and inner layer of the eye covers completely
the middle layer as far forward as the cells of the corneal hypo-
dermis, but its ending in the region of the hypodermis was not
well shown in the section from which this figure was taken.
This inner layer is composed of variously shaped cells containing
much dark pigment, but the shapes of these cells can only be
determined by maceration.

It may be well to explain here that the cells composing the
middle layer correspond in number to the bacilli and cell bound-
aries shown in this ligure, and that they form a series of inverted
retinal elements, the length of which is almost as great as the
depth of the middle layer ; and, as we shall see later, there is a
nucleus to each of these elements.

There remains a third plane in which I have cut sections of the
eye, viz. that passing at right angles to the length of the retinal
rods. A few sections of this series are shown in Figs. 34, 35
and 36. The first of these (34) is the second section from the
inside, the first section having cut off a slice of pigment. We
see the periphery surrounded by pigment, and in the center we
see the arrangement of the outer ends of the retinal cells where
they come in contact from the two sides. The raphe is here
clearly seen to be formed by the contact of the bacilli of the
two sides of the eye, and is formed as a zig-zag line by the inter-
locking of the bacillar ends of the retinal cells. Between the
bacilli the granular protoplasm of the cell is seen, which runs
almost to the tip, so that the bacilli are not formed at the inner
ends of the cells, hut along the side walls at the inner end. It is
also not to be overlooked that only the superficial layer of retinal
elements interlock at the surface, forming the raphe ; and, while
there are many other elements in the interior of the eye, these
do not seem to come to the inner surface, but lie just beneath
these superficial cells. At one point in Fig. 34, viz. in the lower
left-hand portion, the section passes a little below the superficial
bacilli, bringing to light the transversely cut ends of two of the
interior retinal elements.

54 T. H. MORGAN.

A section of the eye somewhat deeper — that is, more outwardly
— than the last is shown in Fig. 35. Along the middle line the
central elements are cut transversely, showing the size in cross-
section of the outer ends of these elements. On account of the
bending of the outer retinal cells around the surface of the eye,
they are cut now farther down than in Fig. 34, where their walls
are thinner. Other elements lying between the central and
superficial ones are cut somewhat obliquely, and the walls of
some of these are seen to thicken, and to join the bacilli as they
approach the middle line. In the lower part of the figure one
sees that the bacilli show a double structure, so that one-half of
the thickness belongs to one cell and the other half to the neigh-
boring one ; so that they are really double, but usually the sepa-
ration into two parts is not seen, owing to the close fusion
between the cells. This figure also shows the entrance of a part
of the optic nerve, disappearing in the pigment layer.

Fig. 36 passes almost through the middle of the eye, but prob-
ably somewhat nearer to its inner half. The most interesting
structures shown in the figure are the small rounded stainable
bodies in the central elements. These lie in the outer ends of
the cells found in the middle of the section. Around these the
protoplasm seems absent or thinner, so that each has sometimes
a halo about it. In other cells they are not seen, although they
show a clear portion in the protoplasm, and probably the bodies
are cut at a different level. The interlacing of the cells them-
selves in the central part of the eye should be noticed. They are
longer from side to side, giving them in cross-section a lanceolate
shape. At the sides of the section lie many scattered nuclei.
Lastly, the nerve fiber is seen entering the eye along the middle
line. The sections following this one as we pass to the outer
part of the eye add little or nothing to what we have gained
from the previous figures, and did not seem worth drawing.
The vacuoles in the outer part of the inner layer are very con-
spicuous in some of these sections, and one also finds a few
nuclei almost, or quite, in the middle line, but at all times they
show a marked tendency to bilateral arrangement.

Such is the gross anatomy of the eye as gathered from sections,
and a great many sections were made, prepared by many dif-
ferent methods, to obtain even this meagre information. As a

THE PYCNOGONIDS. 55

check to the results obtained in Phoxichilidium, the less prom-
ising eyes of Pallene and Tanystylum were studied. It is not
necessary for our purposes to give a full account of the structure
of the eyes of these forms, as they showed in all essential points
a close agreement with Phoxichilidium. To call attention to a
few of the differences may not be out of place, so I have given
two longitudinal sections of the eye of Pallene and Tanystylum.

Fig. 37 is a longitudinal section through the eye of Pallene
empusa. The most noticeable difference, when compared with
the eye of Phoxichilidium, is in the absence here of any well-
marked lens. The chitin is somewhat thicker over the eye than
elsewhere, but not sufficiently so to form a definite lens. The
three layers of the eye are clearly seen. The corneal hypo-
dermis is not so well developed as in Phoxichilidium, and per-
haps this is correlated with the non-development of the lens.
The middle layer is essentially similar to that of Phoxichilidium,
with its nuclei, vacuoles and rods. The bacilli are, however,
relatively more abundant and closer together, and each bacillus
is relatively longer. The inner pigmented layer is less well
developed and forms a narrow yellowish zone around the inner
part of the eye. Cross-sections add nothing to our knowledge,
and sections at right angles to the raphe show that it, the raphe,
is essentially as in Phoxichilidium.

In Tanystylum we find the same structures seen in the other
two as shown in Pig. 38. The chitin over the eye is thicker
than elsewhere, and forms a slightly developed lens, being
intermediate in structure between Pallene and Phoxichilidium.
The corneal hypodermis is much as in Pallene. The outer part
of the middle layer shows few nuclei, which indicates it is near
the median plane of the eye. The bacilli are fewer than either
those of Pallene or Phoxichilidium. Fig. 39 is a part of a section
across the raphe, and, passing through it, shows the arrange-
ment of the cells on the two sides. It was impossible to trace in
these eyes the nerve fibers after they disappear in the pigment
layer. They seem, however, to run around the sides of the eye,
probably innervating the pigment cells in their course, and
enter the middle layer from the sides and at its outer part
beneath the corneal hypodermis.

56 T. H. MORGAN.

Hi8tologt of the Eye.

The following account of the histology is confined almost
entirely to a description of the middle and inner layer of the eye
of Phoxichilidinm. The shape of the cells of the outer layer is
sufficiently shown in Fig. 32. They are seen to be columnar,
and have conspicuous nuclei in the middle part of the cell. The
cells themselves show a bilateral arrangement, the smallest being
on each side of the middle line and thence towards the sides,
increase in length and breadth for a time, when they again
begin to get smaller, passing into the ectoderm of the body at
the point where the pigmented layer touches the surface.

To obtain a knowledge of the middle and inner layers careful
macerations were necessary, and as the results obtained by the
different methods varied with the reagents used, it may be better
to mention those found most useful. The easiest method to
break up the eye into its constituent elements is maceration in a
modification of Haller's fluid. The following proportions were
used : Acetic acid (glacial) 5 parts, water 10 parts, glycerine 5
parts. The cupola bearing the eyes was cut off and quickly
immersed in the fluid, remaining there from a half to an hour,
washed in water, and teased out sometimes in water, sometimes
in glycerine. The method has one disadvantage in that it causes
the bacilli to swell up and become transparent, so that one may
be misled — as I was at first — by the apparent absence of bacilli
at the ends of the cells. The method has the great advantage in
that it separates the individual elements completely from one
another without breaking them up into pieces. Figs. 40, 41, 42,
44 and 45 were obtained in this way. A less successful reagent
was that of sulphuric acid and sea-water. It has the advantage
that the bacilli do not change their shape, but it does not easily
separate the cells. Osmic acid was not very successful, probably
on account of the difficulty of penetration.

After the macerated pieces were put under a cover-slip a little
gentian violet was run under, which stained the cells and the
nuclei, rendering them easier to study. If one is successful the
middle layer breaks up into a great number of elements like
those shown in Figs. 40, 41 and 42. Each of these elements —
cells — is a somewhat flattened body with one clear end — the inner
— and a more protoplasmic middle portion containing a nucleus

THE PYGNOGONIDS. 57

which stains deeply with gentian violet (40). Beyond this there
is a slenderer region which continues outwardly (upwards in
the figure) into a protoplasmic syncytium. This syncytium
lies on the outer part of the middle layer, under the corneal
hypodermis. The cell thus obtained shows no trace of the
bacilli, but their absence is due to the macerating fluid, which
causes them to swell at the sides of the cell.

Fig. 40 h was the most successful maceration obtained. Parts
of three cells — retinal elements — are seen each passing at their
outer ends into a common syncytial mass, as in the former case.
The upper part of the figure shows a fine fibril running also into
the fused protoplasmic part, and this fiber could be traced out-
wards into one of the nerve branches running to the eye, and
undoubtedly represents the ultimate nerve fibril which passes to
the retinal elements. The distribution of the nerve to the outer
part of the middle layer, and the bacillus present at the inner
ends of the retinal elements, show conclusively that the retina is
"inverted"; so that light entering the eye must, before it can
penetrate to the sensory end of the retinal elements — the bacillar
end— pass through the layer of nerve fibrils and fused proto-
plasmic layer of the eye.

Fig. 41 a and h shows to what an extent some, if not all, of the
retinal elements are flattened. In a the spatula-like end of the
element is seen, and this shape is due to the swelling of the bacilli
at the sides of the retinal cell. In b the same element is shown
as rolled over on its side, and is seen to be much narrower
in this diameter than in the former. This accords fairly well
with the shape of the elements seen in the section of Fig. 36,
Plate VII.

In Fig. 42 three of the retinal elements are seen still in con-
tact— two lying above a third, the inner end of which shows
through the transparent ends of the others. Two have spatula-
shaped ends and the third is for most of its length broader than
the others. Each has a nucleus nearer the outer half where the
protoplasm swells out the cell, and beyond this the elements
taper away to join, no doubt, in the syncytium at the outer part
of the eye.

In Fig. 43 is a portion of the inner end of a retinal cell which
was obtained by sulphuric acid and sea-water, and the figure is

53 T. H. MORGAN.

more highly magnified than in the preceding cases. The proto-
plasm runs down to the end of the cell, but on each side are the
rod-like bodies. On the left side of this element is a piece of a
rod of a neighboring cell ; the bacilli being formed by the fusion
of a pair of these rods of adjoining cells. Further, the end of the
retinal element has been so twisted that one also gets a side view
of the rod, showing it to be thinner through than broad. This
preparation is unsatisfactory, in that it is broken off at the point
where the cell-wall is continuous with the bacillus.

In Fig. 44 we see a copy of a part of the superficially lying
retinal elements of the eye of Pallene. It shows how the cells
lie side by side in the superficial part of the middle layer, but at
one point one of the under cells comes out to the surface. The
nuclei are placed in the upper (outer) part of each cell, and below
these cells abut against those on the other side and form the
raphe (not seen here). All of the middle layer cells seem to be
very much alike, and although I have examined quite a number
of macerations, 1 have never obtained any decided departure
from those given above. On the other hand, I cannot deny that
other cells may exist in this layer, for, on account of the small
size of the eyes and the minute size of the retinal cells, it would
be very easy to overlook any unimportant elements, but I think
there can be no doubt but that what we have called the retinal
elements form the greatest part of the middle layer.

If we next examine the cells of the inner layer — the pigment
cells — we find them showing a very great diversity of shape.
They are easily macerated apart by Haller's fluid (modified as
above). A few of these cells are shown in Figs. 45 a, b, c, d,
illustrating the main varieties.

There seems to be a more or less definite distribution for these
different kinds of cells, and we may refer them, 1 believe, to
different parts of the posterior layer. Those pigment cells in the
innermost part of the eye, lying over and at the sides of the
raphe, are shown in Fig. 45 a. These are short, somewhat
cylindrical cells, about as high as wide, and fitting tightly into
one another. Ten of them form the mass in this figure, and in
some of them the nucleus can be seen, and in all the cell boun-
daries made out. The cells found over the sides of the eye show
great differences in size and shape, as seen in Fig. 45 b, c and d.

THE PYGNOGONIDS. 59

In d the cells are drawn out at one side, becoming longer in one
diameter, and this process seems continuous into a nerve fibril,
though the process could only be traced for a short distance, and
I have not succeeded in tracing it actually into such a fibril.
In c two cells are shown which have fused together at the base,
and from the point of union a process runs out which presumably
is continuous into a nerve fibril. I think one must be impressed
by the close similarity between two such cells shown in c and
the similarly arranged retinal cells of Fig. 40. Now, although
I have no actual proof that the cells d lie nearer to the cells «,
and that the cells c are farther around to the sides of the eye, yet
from what I have seen I think it exceedingly probable that in
passing from the innermost cells of the pigment layer to the
outermost (those nearest the hypodermis) we go through a series
of changes as shown by a, d, c. Finally, in Fig. 45, b, there is
seen a long, slightly pigmented cell resembling much in shape a
retinal element. I cannot affirm positively that this cell belongs
to the pigment layer, but it seems exceedingly probable that it
does. If so we find quite a good series connecting the most
modified pigment cell on the one hand with the retinal elements
on the other ; and, further, each of those pigment cells seems to
connect by quite a large process with a nerve fibril. The
important bearing of this arrangement, etc., will be discussed in
another place.

The Development of the Eyes of Tantsttlum.

Instead of beginning with the earliest stages of the eye, which
are the most difficult to understand clearly, let us start at the
other end, when the eye has not quite reached maturity, and
work thence backwards.

Fig. 51, Plate VIII, is from a longitudinal section through the
posterior eye of a larva which has just acquired four pairs of
walking legs. (See Fig. XVIII, Plate V.) In this the ecto-
derm has, in hardening, separated somewhat from its chitinous
covering, but there is no special thickening of the chitin over the
eye to form a lens. The three layers of the eye which were
found in the adult are here easily recognized. The outer layer,
which is formed but slightly modified hypodermal cells, con-
tains large nuclei, which are arranged in a single layer. The

60 T. H. MORGAN.

arrangement of the elements of the middle layer is not well
shown in the figure, but other sections show that they are essen-
tially as in the adult, and that small rods, bacilli, are developed
along the sides of the inner ends. The eye also shows in other
sections a distinct raphe. The inner layer is not so much pig-
mented as in the adult. The essential interest of this figure lies
in the relationship it reveals between the first, the outer, and the
second, the middle, layers of the eye. If we follow from above
downwards the layer of corneal hypodermis, we find that when
it reaches the lower angle of the eye it suddenly turns inwards
and becomes directly continuous into the middle layer of the eye,
and that the large nuclei of the outer can be traced continuously
into the nuclei of the middle layer at this point.

This figure does not show the relationship between the pig-
ment layer and the hypodermis, but others do, and it may be
stated that just below the eye the pigment layer becomes con-
tinuous into the hypodermis, and at the upper part of the eye, as
seen in this figure, the middle and inner layers are in close con-
tact. We may interpret this condition of the eye by referring it
to an invagination of the ectoderm, the opening, if it exists,
being at the lower corner of the middle layer ; that the two
walls of the invagination are of different thicknesses, etc., the
middle layer being one and the pigment layer the other; and,
finally, the whole invagination being pushed over to one side
— dorsalwards. I think this interpretation of the conditions is
not forced, and is the explanation of the conditions shown not
only by this but by other figures.

Figs. 49 and 50 are from a larva having three pairs of walking
legs. (See Fig. XVII, Plate Y.) These two figures belong to
the same series, and are neighboring sections of one of the pos-
terior eyes. The animal is about ready to moult, so that we find
two distinct chitinous coats, the one closely adhering the other
over the latter and not directly connected to it. Fig. 49 is some-
what to one side of the middle line, especially in the upper part
of the eye, but near to it in the lower part. The corneal-hypo-
dermal cells at the lower corner of the eye pass into the cells of
the middle layer. The outlines of the cells are better seen than
in the previous section, and small rod-like bodies are seen at
their inner ends. The middle layer is smaller than in the pre-

THE PYCXOGONIDS. 61

ceding older stage. A narrow zone of pigment cells forms the
inner layer of the eye, and at the lower corner of the eye it passes
directly into the hypodermis below the eye, while at the upper
corner of the eye the pigment layer is never found intimately
connected with the hypodermis, but is in close contact with the
middle layer of the eye. Fig. 50 cuts the upper part of the eye
in the middle line, but not so much so the lower. In the upper
corner of the middle layer there lies a cell which is shorter and
broader than the retinal cells farther down, and very often the
second cell from the top shows a somewhat similar difference in
shape. In fact, these two cells seem to be present in all the
larval eyes, and the meaning of this we shall see later.

Fig. 48 shows the posterior eye of a still younger larva. (See
Fig. XYI, Plate Y.) Here again we find practically the same
conditions as in the preceding figures, but the eye is smaller.
The corneal hypodermis passes at the lower corner into the
middle layer of the eye. The pigment zone behind the eye is
narrow, and the third layer is seen to continue into the hypo-
dermis below the eye. At this place the cells of the hypodermis
are very distinctly seen passing into the pigment layer of the
eye. The middle layer is separated very distinctly from the
corneal hypodermis by an elongated cavity, probably caused by
shrinkage.

Between this eye and the next earlier stage there is quite a
difference in size and structure. This next stage is shown in
Fig. 47, which is from a larva having a single pair of walking
legs. (See Fig. XY, Plate Y.) Both posterior eyes are shown
in this figure (47). They are not, as in all older eyes, raised above
the general surface of the body. Each eye is still seen to be
three-layered, though this is not so apparent as in older stages.
Both the outer and the inner layers are quite narrow. The
middle layer is seen to be formed of two quite large clear cells
with very large nuclei. These form by far the most conspicuous
part of the eye in cross-section, but I have not determined defi-
nitely whether there are only these two cells, as seen in cross-
section, or whether there are a few more present. There is, it
must be confessed, not much evidence of invagination of the eye
at this stage, and for this reason it seemed better to begin with
older stages, which could be easily interpreted as such. How-

62 T. H. MORGAN.

ever, in the eye on the left side of the figure (47), the continua-
tion of the corneal hypodermis with the middle layer is seen, and
in the right eye the pigment layer is intimately connected with
the hypodermis on the outer side of the eye ; so that the arrange-
ment is essentially as in older stages. I could not determine
definitely as to the number of cells forming the pigment layer,
but in cross-section one can discover not more than two nuclei,
so that the number must be exceedingly small. Besides the pair
of eyes seen in this figure, there is an anterior pair resembling
the above both in structure and size. There is one difference
shown in the sections of the anterior eyes, viz. that there is a
thickening of hypodermal cells on each side just beyond the
outer corner of the eyes, but of this I shall speak later. More-
over, the anterior eyes are nearer together than the posterior pair.

Fig. 46 is from a larva without any walking legs, at an age
shown by Fig. XIV, Plate V. The larva has but a single pair
of eyes, which correspond to the anterior of the two pairs of older
larvse. We find the eyes have essentially the same structure as
in the one last described, but are much nearer together than
those of the last figure (a posterior pair). Each is composed of
three layers. Of these the middle is much larger, and is com-
posed of two large transparent cells with large nuclei. The con-
tinuity of the corneal hypodermis with the middle layer was not
clearly made out, nor that of the pigment layer with the ecto-
derm below the eye ; but I think we may fairly push our inter-
pretation even to this, but on account of the extreme smallness
of the eye and the difficulty of making very thin sections, such
connections would not be readily discovered. The posterior
layer of the eye seems to be formed of but a single large flat-
tened, pigmented cell, showing a clear nucleus, but it cannot be
affirmed with anything like certainty that this is the only cell.

On each side of the anterior eyes the ectoderm, as shown in Fig.
46, is a greatly thickened mass, indeed forming two bodies about
the size of the two eyes themselves. I am at a loss to interpret
these thickenings. At first I believed they might be connected
with the sudden increase in the size of the eye, and the very great
change in size in the eye between stages XV and XYI of Plate
Y might seem to support such an hypothesis. But, on the other
hand, the posterior eyes also increase greatly in size at the same

THE PYCNOGONIDS. 63

time, and these have the ectoblast very slightly thicker at the
edge of the eye. Possibly these may represent rudimentary
eyes which were placed on each side of the first pair, or else
have shifted position from some other place ; at any rate, they
are of interest and very constant in position, and furnish very
little ground for speculation.

The structure of the eye of the Pantopod-larva was described
in Part I. This description is correct, I believe, as far as it
goes, but at the time I had little knowledge of the adult eye, and
hence overlooked an extremely narrow layer of ectoblast between
the two large cells and the chitin. I have since re-examined the
eye of this larva, and can assert that in some sections this layer
could be made out, and in others a thin membrane-like body
separated the clear cells (middle layer) from the chitin, and this,
presumably, is the same ectoblast. Unfortunately, the section
from which the figure of Plate I was made had only a very
narrow ectoderm layer, and was overlooked. We may affirm,
then, that the eye at all stages of development is a three-layered
structure.

Comparison with the Arachnid Simple Eyes.

We are now in a position to make, if possible, a practical
application of the preceding. Does a study of the structure and
development of the Pycnogonid eye throw any light upon the
relationship of the Sea-Spiders to other groups % I think there
can be no doubt of this, and that it furnishes a most satisfactory
verification of the relationship pointed out in the preceding
sections.

It seems absolutely impossible to bring the eyes into the same
category with the eyes of the Annelids, which is a decided gain
for our position, and it is par excellence an Arthropod eye, and
even more, as I hope to show it must be referred to a particular
group of Arthropods — the Arachnids. The absence of any very
close resemblance to the eyes of Annelids would not be in itself-
a final argument against Dohrn's position, inasmuch as the eyes
may have been much modified during the transition from
Annelid to Pycnogonid ; but this can scarcely be the case, for
the eyes seem to show traces of degeneration in many, if not in
all (as shown, perhaps most strikingly, by the large vacuoles, and

64 T. H. MORGAN.

the slight development of lens in some forms), and this could not
be accounted for had the group first acquired an eye so different
from the Annelids, and afterward, without change of habitat,
degenerated. Further, we find a remarkable agreement with the
Arachnid eye, and it seems much more probable that this (even
alone, but we can consider it now in connection with many other
peculiarities, both in structure and development, common to the
two groups) has a phylogenetic meaning, and forces one to
abandon the theory advanced by Dohrn.

First 1 must state that the principal papers which I have used
in the following comparison are those by Lankaster and Bourne
on the Simple Eyes of Limulus, Locy on the Development of the
Spider's Eye, and Mark on the same subject, Parker on the Eyes
of Scorpions, and Laurie on the Development of the Eye of the
Scorpion. Of the more speculative papers of Patten and Watase
I shall speak separately.

The adult eyes of the Sea-Spiders show three distinct layers —
the corneal hypodermis, the retinal or middle layer, and the pos-
terior or pigment layer. It has been demonstrated beyond doubt
that the Arachnid median eyes are composed of three layers,
and, further, that these are more apparent in the embryo than in
the adult Arachnid. Further, the arrangement of the layers is
essentially the same, for we find a corneal hypodermis, a middle
sensory layer, and a posterior narrow layer.

The middle layer in the embryo Spiders (where the eyes have
been studied in most detail) contains at the inner ends of the
retinal cells clear (stainable) rod-like bodies, or bacilli, which
seem to be identical with the similar bacilli of Pycnogonids.
The bacilli in the Pycnogonids are at the inner ends of the cells —
in other words, the retina is inverted — and in the embryo Spiders
we find that similar rods are also at the inner ends of the cells —
that is, the retinal elements are here also inverted. A secondary
shifting, or change, takes place in the Spider's eyes, so that in
the adult the eyes and retina become righted ; but how this later
change comes about is at present, I believe, unknown.

Lastly, the eyes of the Pycnogonids are formed by a process of
invagination of the same nature as the invagination in the
Arachnids (Locy, Parker, Laurie, Brooks). This invagination,
instead of being an open cup (as in Insects), turns to one side

THE PYCNOGONIDS. 65

and comes to lie under another part of the hypodermis, which
forms the outer layer of the eye, and the wall of the invaginated
tissue lying outmost forms the middle layer, and the inner the
inner layer of the adult eye. The invagination of the cells to
form the eye does not take place all at once in the eye of the
Sea-Spiders, but is a gradual process, so that a few cells at a time
are added to the eye, and this is connected with the presence of
a small, free larval form having the beginnings of the eyes. In
the Arachnids, on the other hand, the process of invagination is
more marked, in that all the cells are invaginated at once.
These similarities in development and adult structure, it seems
to me, point unmistakably to a community of descent between
the Pycnogonids and the Arachnids, and further, they point to
the same conclusion which we reached in studying the embry-
ology of the group, that the Sea-Spiders show in their adult
structures close resemblances to the embryos of the Arachnids,
inasmuch as the inversion of the eye of the Pycnogonid agrees
with the inversion of the embryo Spider, and the well-defined
three-layered condition of the latter in the embryo is seen fully
developed in the eye of the adult Pycnogonids.

Inversion of the Eye.

The early inversion of the eye of the Spiders has been noticed
by Locy, and Mark has extended this discovery to the solution
of certain theoretical problems. The latter suggested that the
process of invagination of the Spider's eye repeats the method by
which such an eye arose, that is to say, if we start with a simple
cup-shaped invagination and suppose the walls on *One side to
have increased greatly, so that they ultimately filled up the cen-
tral cavity, rolling over into this cavity, the inversion of the
eye was explained. The opposite wall of the cup remaining thin
and developing pigment, became the inner laj^er of the eye, and
the general hypodermis of the body formed the secreting cells of
the lens or the outer layer. Patten has suggested that the
Arachnid eye is but a later stage of development of that seen
in the Insect eye. The walls of the optic cup turn in from the
two sides, the elements at the sides becoming first horizontal, then
turning into the cavity of the vesicle, the retinal cells at the
bottom of the vesicle forming a posterior (inner) layer to the

66 T. H. MORGAN.

eye. Patten's suggestion was at the time, I think, little more
than a suggestion of a possibility, and no structures in the
Arachnid eye were given to support it.

Mark, on the other hand, has complicated his hypothesis by
certain additions, both of which I believe one would, a priori,
rather reject, if possible. First, that the one-sided development
he supposes to have taken place to have been the result of a
change of position in the lens, so that part of the eye was placed
to better advantage than the rest, a supposition not founded on
any facts, and it seems to me highly improbable. Secondly, the
inverted elements of the embryo right themselves (as they are so
found in the adult) by a complete . change in the parts of each
element, so that the rods which are at first secreted at one end of
the cell — the inner — subsequently are formed at the opposite end
of the element — the outer — and that the nerve fibrils were also
transferred from one end of the cell to the other. To be sure we
do not yet know how the eye does right itself, yet it seems most
improbable it has done so by such an unusual method, and it
would be much easier to believe that the elements shift around
so that their inner ends become the outer.

I have examined carefully the inverted eye of the Fycnogonids
to see if it would throw any light upon the early inversion of the
eyes of Arachnids. In the first place, I am inclined to believe
that this inversion of the Spider's eye cannot be explained as a
purely embryonic phenomenon. From what has been given in
the preceding sections, I am ready to believe that this inversion
of the embryo Spider's eye is to be explained as a phylogenetic
process, and that the adult Pycnogonid's eye represents approx-
imately the condition of the eye of the ancestral Arachnid (or,
at any rate, the ancestral Spider).

Let us now see if there is anything in the Pycnogonid eye to
determine whether this primitive inversion was caused by a one-
sided development of the eye, as Mark has assumed (for Spiders),
or whether we have 'any evidence pointing in other directions.
I have been much impressed by the bilateral arrangement of the
eye in the Pycnogonids, and it has seemed to me probable that
this might give a clue to the solution we are in need of. The
corneal hypodermal cells show a marked bilateral structure in
the arrangement of the cells on each side of the middle line.

THE PYCXOGONIDS. 67

The middle layer (and this is more to the point) shows a strong
bilateral arrangement — first, in respect to the nuclei which lie
largely on the two sides, and, secondly, in the arrangement of
the retinal elements to the right and left of a median raphe.
The third layer has a bilateral arrangement, the cells along the
middle being different from those along the sides of the eye, and
those at the sides approximating in structure to the retinal
elements.

It does not seem to me that it is possible to interpret this
bilaterality on Mark's hypothesis, because it furnishes not the
slightest reason why an eye so derived should show a two-sided
arrangement, nor is it likely to be explained as a secondary
arrangement.

On the other hand, if we assume the inversion to have arisen
by a turning in of two sides of an optic cup — a process already
shown in the Insects' simple eyes — we see clearly why the eye
should present so perfectly a bilateral plan. To explain this
more in detail, we may start with a simple invagination, forming
a thick-walled cup with nerve fibers entering the inner ends of
the cells. The early stages of development of Insects' simple
eyes show such a condition. Next, the lips of the invagination
touch and fuse, forming an almost continuous layer over the eye,
giving the corneal hypodermis. The bilateral arrangements of
this layer in the Pycnogonids may be the expression of this union
from the two sides. At the same time that this occurs, the more
horizontally-lying retinal elements in the upper part turn into
the cavity of the cup, so that their morphologically outer ends
now come to point inwards. The nerve fibers which ran up to
the back part of the eye continue to do so, and, running from
their point of contact around the outer wall of the cup, supply
the elements with fibrils. This is practically the condition of
the simple eyes of Insects. Such an eye may be said to be
(imperfectly) three-layered at this stage, as we may now count
the partially inverted elements as a middle layer. The elements
lying at the base of the cup are in a disadvantageous position for
vision, being covered by a sensory layer (the inverted fibers)
lying between them and the light. So these elements of the
bottom of the cup and at the sides degenerate and become a pig-
ment layer at the back of the eye. JSTow we have reached the
type of eye found in the Sea-Spiders.

68 T. H. MORGAN.

The evidence that the eyes of the Sea-Spiders have evolved
along some such line as this is found in the structure of the eye
as existing to-day. The median raphe of the middle layer indi-
cates the points of contact between the retinal elements from the
two sides. These, as they became inverted, came into contact in
the middle line. The bilateral arrangement of the nuclei follows
almost as a necessity of this two-sided inversion. These would
be crowded together when the retinal elements turn in, and few
or none along the middle line. Patten shows on the outer ends
of the retinal elements of Insect eyes a layer of what he calls
" secondary nuclei," which, so far as I can judge, are identical
with the round stainable bodies found at the inner ends of the
inverted elements of the Pycnogonid eye. One sees at once that
by the inversion of the lateral elements of the cup the bodies
must be found in exactly the same position as they are found in
the Pycnogonid eye.

If further evidence were needed to prove the two-sided inver-
sion, it would, I think, be amply found in the conditions existing
in the pigment layer of the eye. The cells immediately at the
back (at the base of the cup) are cylindrical cells, such as would
result from a shortening of the retinal elements. If one draws a
diagram of the process I have sketched, it becomes evident that
as we progress around the sides of the eye the pigment cells
must pass insensibly into the retinal cells ; and such, indeed, we
found previously to be the case. Thus the retinal cells here
have not suffered so great a change as those at the back of the
eye; but, developing pigment, they furnish a dark background
for the retinal cells, but still retain largely their ancestral shape.
Finally, the large nerve process running from these pigment cells
is explained by their ancestral function as fibrils to the retinal
elements.

To sum up, I believe all the layers of the Pycnogonid eye give
abundant evidence that the eye has developed by the turning in
of two sides of a primitive optic vesicle, and that the simple eyes
of Insects furnish all the intermediate stages, both in develop-
ment and adult structure, between a simple cup-like invagination
and the three-layered condition of the Pycnogonid eye.1

'This and several other facts (as multipolar delamination, etc.) point
perhaps to a genetic relationship of Insects to Arachnids, but I cannot go into
these comparisons at present.

THE PYGNOGONIDS. 69

I am inclined to push this a step further and attempt to show
that the median Arachnid eyes have developed along a similar
path, and this must be a necessary corollary to the position to
which I have assigned the Pycnogonids, and believe, as 1 have
before said, that the inversion of the Spider's eye is its Pycno-
gonid stage of evolution and the righting of it a secondary and
later change. I also believe there is no evidence that in the
evolution of the Spider's eye a change of position in the lens has
taken place.

General Consideration of the Simple Eyes.

Before leaving the subject there are two points to which I
must refer — the meaning of the invagination of Arachnid (and
Pycnogonid) eye, and the explanation of the grouping together
of the retinal cells into bundles or retinulse.

It will be at once seen, with respect to the first point, that I
have placed more faith in the evidence furnished by the adult
eye of the Pycnogonids and Spiders than in that given by the
ontogeny of those eyes, while Mark has utilized the ontogeny
as furnishing a basis for his considerations.

It seems to me that the development of the eye, as we find it
in the Pycnogonids, is entirely an ontogenetic process, i. e., it
represents here an abbreviated condition. Instead of recapitu-
lating all the early stages, forming first a cup, then turning in
the walls, it performs the whole process at one time and thus
complicates the matter. In the Pycnogonids the presence of the
eyes in all the larval stages, presumably functional in each, must
have changed the original process to a very great extent.1 Yet
through all these stages we have been able to recognize the
three-layered condition of the eyes. We must believe, from
what has been said of the growth of the eye, that in the Pycno-
gonids the invagination (of the Arachnid type) has been retarded,
so that, one end of the invagination being formed, the in turned
and inverted cells have functioned as larval eyes, and as the
animal increased in size the invagination kept pace, adding more
and more cells to the layers of the eye, so that all of the stages
had presumably functional eyes, and at the same time the larva
retained the original type of Arachnid invagination.

1 We do not know whether the ancestral Arachnids had free-living larva1
or not.

70 T. H. MORGAN.

It may not be out of place to say a few words about the forma-
tion of bundles of elements in the eyes of many Arachnids,
as the preceding may have some bearing on the question. In
several of the Arachnid median eyes the retinal elements group
themselves into bundles of a varying number of cells — the
retinulse. Watase has advanced the idea that these bundles
represent each an invagination — a pit — of ectoderm, and repre-
sent the process of formation of the eye elements — ommatidia.
He homologizes the retinulse of the simple eyes with the reti-
nulse of the compound eye ; in the latter he also believes they
represent ancestral invaginations. I do not care to carry my
comparisons into the compound eye, but shall speak only of the
simple eyes, in which, I think, there is not sufficient evidence
that these bundles each represent an invagination.

It seems most probable that these groups of cells forming
retinulse represent secondary changes in the eye. If the Arach-
nid eyes have, as I believe, come along some such lines as have
been just sketched, then it is evident that the eyes have several
times undergone profound changes of which the median Arachnid
eye to-day is the outcome. The Insects' eyes show no such com-
plicated groupings of cells as do the Arachnid eyes. The Pycno-
gonid eyes show no trace of such bundles of cells. And by no
means do all the Arachnid eyes show this, as I have satisfied
myself in the Spiders. So, unless one had preconceived ideas to
apply to the simple eyes, I see no reason for supposing such a com-
bination of cells to represent an invagination. On such a theory
the clear central part of the bundles — the center being formed by
the union of the inner parts of the cells forming the group —
represents the original chitin secretion of the hypodermal cells.
I can find no evidence to support this assumption, although I
have examined carefully, both by sections and macerations, the
retinula-bundles of several Arachnids. The clear central part —
the rhabdome — is not a secretion from a cell, hut only a clear
protoj)lasmic (?) jyortion of it, and is included in the general wall
of the cell. Nor is there any evidence that it is of a chitinous
nature, for to micro-chemical reagents it acts entirely differently.
Further, an apparently similar substance is formed at the sides
of part of the retinal elements of both Insects and Pycnogonids,
and although in the former the cells are in the proper position,

THE PYCNOGONIBS. 71

this clear stain able portion is at the sides and not at the ends,
for a thick protoplasm lies between the rods. Finally, we find
similar transparent and perhaps more solid parts of the retinal
elements of other eyes belonging to quite different groups, as in
the Ascidians and in the Mollusks (see Pecten), and certainly
in the latter group no one will imagine them chitinous or any
homologue of chitin.

LIST OF WORKS REFERRED TO.

1. Balfour. Notes on the Development of the Araneina. Quart.
Journ. Micr. Sci., vol. XX, 1880.

2. Balfour. Elements of Embryology. 1880.

3. Clans. Crustaceen-Systems. Wien, 1876.

4. Dohrn, Ant. Neue TJntersuchungen liber Pycnogoniden.
Mitth. a. d. Zool. Stat. Neapel, Band I, 1879.

5. Dohrn, Ant. Die Pantopoden des Golfes von Neapel. Fauna
und Flora des Golfes von Neapel. Band III, 1881.

6. Hatcheck. Beitrage zur Entwicklnngsgeschichte der Lepi-
dopteren. Jena, Zeit. Naturwiss., No. 11, 1877.

7. Heider, K. Ueber die Anlage der Keimbliitter von Hydro-
philus piceus. Abh. d. K. Preuss. Akad. d. Wiss. Berlin, 1885.

8. Midler, Fritz. Fur Darwin. Leipzig, 1864.

9. Metschnikoff, E. Entwicklnngsgeschichte des Chelifer. Zeit.
f. wiss. Zool., XXI, 1871.

10. Schimkewitsch. Zur Entwicklungsgeschichte der Araneen.
Zool. Anz., VII, 1884.

11. Schimkewitsch. Etude sur le developpement des Araignees.
Archiv de Biologie, 6, 1887.

12. Sedgwick, A. A Monograph of the Development of Peripatus
Capensis. Stud. Morph. Lab., Camb., IV, Pt. I, 1888.

13. Semper. Ueber Pycnogonida und ihre in Hydroiden schma-
rotzenden Larven-formen. Arb. a. d. Zool. Zoot. Inst. Wiirzburg, 1,
1874.

14. Wheeler. The Embryology of Blatta Germanica and Dory-
phora Decemlineata. Jour. Morph., Ill, 1889.

15. Wilson, E. B. The Pycnogonida of New England and Adjacent
Waters. U. S. Commissioner's Report, Pt. VI, 1878.

72 T. H. MORGAN.

REFERENCE LETTERS FOR PLATES.

B. Brain.

D. Mid-gut, mesenteron.

D1, D", Dll\ DIV, etc. Diverticula of mid-gut.

E. Ectoblast.
N. Entoblast.
M. Mesoblast.

P. Periphery of blastoderm.

S. Schizocoels.

Vl~\ Ventral ganglia 1 — 7.

H. Heart.

L. Legs.

0. Opening ventral organs.

G. Central cavity of segmenting egg.

DESCRIPTION OF PLATES.

Plate I.

Figure 1. Section of segmented egg and membranes of Pallene
empusa.

Fig. 2. Section of another egg of Pallene at same stage to show
central cavity G.

Fig. 3. Section through upper pole of egg of Pallene. First
formation of blastoderm.

Fig. 4. Section through upper pole of egg of Pallene. Later
stage than last. Inner cells — entoblast — are present.

Fig. 5. Section through egg of Pallene. Later stage than last.
Cap of cells — blastoderm — covers more of egg than in last.

Fig. 6. Section through periphery of blastoderm, at P of Fig. 5.

Fig. 7. Section in same region as last.

Fig. 8. Section of segmenting egg of Tanystylum orbiculare, to
show pyramidal shape of cells.

Fig. 9. Section of egg of Tanystylum, to show process of multi-
polar delamination.

Fig. 10. Same, with delaminated entoblast.

Fig. 11. Section of segmenting egg of Phoxichilidium maxillare.

Fig. 12. Same, to show degeneration of endoderm.

Fig. 13. Cross-section of the Pantopod -larva of Tanystylum
orbiculare.

Fig. 14. Circumcesophageal ring of same.

THE PYCNOGONIDS. 73

Fig. 15. Eyes of same.

Fig. 16. Ventral organs of same. Early stage of third pair of
ganglia.

Plate II.— Sections of Pallene empusa.

Fig. 17. Longitudinal section through stomodaeum of young
embryo. First origin of mesoblast at M.

Fig. 18. Cross-section through stomodaeum of older stage. See
line 18, Fig. I, Plate IV.

Fig. 19. Section anterior to last. See line 19, Fig. I, Plate IV.

Fig. 20. Section anterior to Fig. 19. See line 20, Fig. I, Plate IV.

Fig. 21. Cross-section of ventral organs. See line 21, Fig. I,
Plate IV.

Fig. 22. Cross-section of ventral organs at stage III.

Fig. 23. Cross-section of ventral organs at stage V.

Fig. 24. Cross-section of ventral organs at stage VI.

Fig. 25. Cross-section of embryo at stage III.

Fig. 26. Cross-section through brain and second pair of ventral
ganglia of stage III.

Fig. 27. Cross-section of embryo in the plane of a pair of legs,
about stage VI.

Fig. 28. Cross-section of embryo in the plane between a pair of
walking legs, about stage VI.

Fig. 29. Cross-section through the rudimentary abdomen and
the body of an embryo, stages VII-VIII.

Plate III.

A-J, Pallene empusa. a-c, Phoxichilidium maxillare. d-f, Tany-

stylum orbiculare.

Fig. A. Segmenting egg, side view, 2-celled stage.

Fig. B. Segmenting egg, seen from above, 4-celled stage.

Fig. C. Segmenting egg, seen from above, 8-celled stage.

Fig. D. Segmenting egg, seen from below, 8-celled stage.

Fig. E. Segmenting egg, side view, 16-celled stage.

Fig. F. Segmenting egg, surface views of micromeres, 16-celled
stage.

Fig. Gr. Segmenting egg, side view, 16 macromeres, 8 micro-
meres.

Fig. H. Segmenting egg, side view, 24 macromeres, 16 micro-
meres.

Fig. J. Segmented egg, lower pole, base of pyramidal cells shows
at surface.

74 T. H. MORGAN.

Fig. a. Phoxichilidium, 2-celled stage.

Fig. b. Phoxichilidium, 4-celled stage.

Fig. c. Phoxichilidium, 8-celled stage.

Fig. d. Tanystylum orbiculare, 2-celled stage.

Fig. e. Tanystylum orbiculare, 4-celled stage.

Fig. f. Tanystylum orbiculare, 32 (?) -celled stage.

Plate IV.
I-VIII, Pallene. IX, Tanystylum.

Fig. I. Surface view of young embryo, to show anterior region.

Fig. II. Surface view of ventral region of same.

Fig. III. Ventral view of embryo with four pairs ventral ganglia,
first and second separate.

Fig. IV. Dorsal view of embryo about same age (a little older
perhaps) as last.

Fig. V. Ventral view of embryo. First and second ventral
ganglia have fused.

Fig. VI. Side view of embryo, showing rudimentary abdomen
behind as a continuation of the body and in same plane as the body.

Fig. VII. Ventral view of embryo at the age when it leaves the
parent. Fourth pair of walking legs seen behind; third pair of
appendages beginning just in front of first pair of walking legs.

Fig. VIII. Dorsal view of last, to show brain and four eye spots.
In both VII and VIII only the proximal ends of the legs are shown.

Fig. IX. The Pantopod-larva (Protonymphon of Hoek) of Tany-
stylum orbiculare.

Plate V. — Metamorphosis of Tanystylum. Zeiss 4Ai

Fig. X, stage 2. Young larva after leaving parent and after first (?)
moult. Ventral view.

Fig. XI, stage 2. Dorsal view of last.

Fig. XII, stage 3. Ventral view.

Fig. XIII, stage 4. Ventral view, showing three pairs of ventral
ganglia, the first really composed of two.

Fig. XIV, stage 4. Dorsal view of last. One pair of eyes — the
anterior.

Fig. XV, stage 5. Ventral view with first pair of walking legs,
each with six segments.

Fig. XVI, stage 6. Ventral view with two pairs walking legs, the
first with eight segments, the second with six.

Fig. XVII, stage 7. Ventral view with three pairs walking legs,

THE PYCNOGONIDS. 75

the first with eight segments, the second with eight, the third with
six.

Fig. XVIII, stage 8. Ventral view with four pairs walking legs,
the first with eight segments, the second with eight, the third with
eight, the fourth with six.

Plate VI. — Larv^ and Adult of Tanystylum. Larva of
Phoxichilidium. Larva of Pallene.

Fig. XIX, stage 9. Tanystylum. Ventral view. Showing re-
newed growth of third pair of appendages. Zeiss 4 A h

Fig. XX, stage 9. Dorsal view of last. Showing rudiments of
first appendages. Zeiss 4AJ.

Fig. XXI, stage 10. Tanystylum. Ventral view. Showing seg-
mentation of third appendages, and fusion of first, second and third
ventral ganglia. Zeiss 4 A h

Fig. XXII. Adult Tanystylum. Dorsal view. Proportionately
less magnified.

Fig. XXIII. Ovigerous leg and eggs of Tanystylum. Zeiss 2 A h

Fig. XXIV. Adult Phoxichilidium. Side view of male with eggs.
X30.

Fig. XXV. Pantopod-larva of Phoxichilidium.

Fig. XXVI. Larval Pallene. Ventral view. Showing the unde-
veloped third pair of appendages.

Plate VII.— Eyes of Phoxichilidium, 30-36. Pallene, 37.
Tanystylum, 38-39. Zeiss 4 D I, except Fig. 31.

Fig. 30. Eye of Phoxichilidium, viewed on the inner side, from
a macerated preparation. Showing distribution of nerve to eye and
bilateral arrangement of elements along the raphe.

Fig. 31. Section across cupola, containing the eyes which are cut
into " cross sections " (horizontal with reference to body).

Fig. 32. Cross-section of eye as in 31, but more magnified.

Fig. 33. Longitudinal section of eye, parallel to raphe.

Fig. 34. Section in plane at right angles to other two. Showing
bacillar ends of the elements of the retina, forming by their union
the raphe.

Fig. 35. Section in same plane, but further out than last.

Fig. 36. Section in same plane, but nearer the center of the eye.

Fig. 37. Longitudinal section of eye of Pallene.

Fig. 38. Longitudinal section of eye of Tanystylum.

Fig. 39. Inner view of cells forming raphe of Tanystylum.

76 T. H. MORGAN.

Plate VIII. — Macerations of Eye of Phoxichilidium. De-
velopment of Eye of Tanystylum and Section of Ganglia
of same, etc.

Fig. 40. Retinal elements of eye of Phoxichilidium obtained by

maceration in Haller's fluid.

Fig. 41. Others of same.

Fig. 42. Collection of three retinal elements, from maceration.

Fig. 43. End of cell showing bacilli, from a maceration in sul-
phuric acid and sea-water.

Fig. 44. Retinal cells of outer surface of middle layer of Pallene.

Fig. 45. Pigment cells of inner layer of Phoxichilidium.

Fig. 46. Eye of Tanystylum-larva. Longitudinal section of
anterior and only pair. Stage 4. Zeiss 4 F f .

Fig. 47. Same of older larva. From posterior pair. Stage 5.

Fig. 48. Same of one posterior eye of older larva. Stage 6.

Fig. 49. Same of one posterior eye of older larva. Stage 7.

Fig. 50. Section in same series as last.

Fig. 51. Same of older larva. Stage 8. Zeiss 4 F f .

Fig. 52. Cross-section through posterior part of body of larval
Tanystylum, showing beginning of a pair of ganglia, etc. Zeiss 4 D % .

Fig. 53. Pair of ganglia with ventral organs disappearing.

Fig. 54. Pair of ganglia with cavities of ventral organs. Zeiss 4
Df.

VOL.V PLATK 1

^P

'DIEC FROM BIOL. LAB. VOL.V

PLATE III

I) ( |

^

zx

! Vi

c.

/rtwn *• Co. Z./M.. B*lLi.

STUDIES PROM BIOL. LAB. VOL.V

AV/

;?\

1

*

M

T H.MORGAN,

A. Hotnf* r« r.u

STUDIES FROM BIOL LAB. VOL.V

J

\ n

W

.

-*,■■■'

STUDIES FROM BIOL. LAB. VOL.V

a

r\i

fk

A. Hot* & Co. Litk. R-Ji»